key: cord-329855-pr7g6ivu
authors: Kalfaoglu, Bahire; Almeida-Santos, José; Adele Tye, Chanidapa; Satou, Yorifumi; Ono, Masahiro
title: T-cell hyperactivation and paralysis in severe COVID-19 infection revealed by single-cell analysis
date: 2020-05-30
journal: bioRxiv
DOI: 10.1101/2020.05.26.115923
sha: 
doc_id: 329855
cord_uid: pr7g6ivu

Severe COVID-19 patients can show respiratory failure, T-cell reduction, and cytokine release syndrome (CRS), which can be fatal in both young and aged patients and is a major concern of the pandemic. However, the pathogenetic mechanisms of CRS in COVID-19 are poorly understood. Here we show single cell-level mechanisms for T-cell dysregulation in severe SARS-CoV-2 infection, and thereby demonstrate the mechanisms underlying T-cell hyperactivation and paralysis in severe COVID-19 patients. By in silico sorting CD4+ T-cells from a single cell RNA-seq dataset, we found that CD4+ T-cells were highly activated and showed unique differentiation pathways in the lung of severe COVID-19 patients. Notably, those T-cells in severe COVID-19 patients highly expressed immunoregulatory receptors and CD25, whilst repressing the expression of the transcription factor FOXP3 and interestingly, both the differentiation of regulatory T-cells (Tregs) and Th17 was inhibited. Meanwhile, highly activated CD4+ T-cells express PD-1 alongside macrophages that express PD-1 ligands in severe patients, suggesting that PD-1-mediated immunoregulation was partially operating. Furthermore, we show that CD25+ hyperactivated T-cells differentiate into multiple helper T-cell lineages, showing multifaceted effector T-cells with Th1 and Th2 characteristics. Lastly, we show that CD4+ T-cells, particularly CD25-expressing hyperactivated T-cells, produce the protease Furin, which facilitates the viral entry of SARS-CoV-2. Collectively, CD4+ T-cells from severe COVID-19 patients are hyperactivated and FOXP3-mediated negative feedback mechanisms are impaired in the lung, while activated CD4+ T-cells continue to promote further viral infection through the production of Furin. Therefore, our study proposes a new model of T-cell hyperactivation and paralysis that drives pulmonary damage, systemic CRS and organ failure in severe COVID-19 patients.

Negative regulatory mechanisms of T-cell activation control inflammation in cancer, autoimmunity, and infections thus preventing excessive and prolonged inflammation which can induce tissue destruction, or immunopathology. Immune checkpoints such as CTLA-4 and PD-1 are well known examples of T-cell intrinsic regulators. Upon recognizing antigens, T-cells are activated and start to express PD-1 and CTLA-4, which in turn suppresses the two major signalling pathways for T-cells: T-cell receptor (TCR) signalling and CD28 costimulation. 1 In addition, the transcription factor Foxp3 can be induced in activated T-cells, especially in humans, and plays a key role as an inducible negative regulator during inflammation. 2 COVID-19 is caused by the coronavirus SARS-CoV-2, which is closely related to the severe acute respiratory syndrome coronavirus (SARS-CoV). The major symptoms such as cough and diarrhoea in mild to moderate patients can be understood through the type of tissues that can be infected by SARS-CoV-2. SARS-CoV-2 binds to angiotensin-converting enzyme 2 (ACE2) on the surface of human cells through its spike (S) protein. Viral entry is enhanced by the type II transmembrane serine protease TMPRSS2, which cleaves a part of S protein and thereby exposes the fusion domain of the S-protein. 3, 4 SARS-CoV-2 establishes infections through epithelial cells in the upper and lower airways, which express ACE2 and TMPRSS2. 5 In addition, the pro-protein convertase Furin activates the S-protein of SARS and SARS-CoV-2. 6, 7 Intriguingly, T-cell specific deletion of Furin results in the impairment of effector T-cells and regulatory T-cells (Tregs) and leads to the development of age-related autoimmunity, which is accompanied by increased serum IFN-g, IL-4, IL-6, IL-13, and IgE. 8 In addition, Furin is preferentially expressed by Th1 cells and is critical for their IFN-g production. 9 As evidenced in a parasite infection model, Furin-deficient CD4 + T cells are skewed towards a Th2 phenotype. 10 It is poorly understood how SARS-CoV-2 induces severe infection in a minority of patients, who develop respiratory distress and multiorgan failure. These severe patients show elevated serum cytokines, respiratory failure, haemophagocytosis, elevated ferritin, Ddimer, and soluble CD25 (IL-2R a chain, sCD25), which are characteristic features of secondary haemophagocytic lymphohistiocytosis (sHLH)-like conditions or cytokine release syndrome (CRS). In fact, severe COVID-19 patients have elevated levels of prototypic CRS cytokines from innate immune cells including IL-6, TNF-a, and IL-10. 11, 12 Recently McGonagle et al. proposed that activated macrophages drive immune reactions that induce diffuse pulmonary intravascular coagulopathy, or immunothrombosis, in severe COVID -19 patients. 13 While this may explain the unique pulmonary pathology of severe COVID-19 patients, the underlying molecular mechanisms are poorly understood. Importantly, CRS in severe COVID-19 patients may be more systemic and involve a wide range of T-cells. Firstly, circulating T-cells are severely reduced in severe SARS-CoV-2 infections for unknown reasons 12, 14 . Intriguingly, severe COVID-19 patients show elevated serum IL-2 and soluble CD25 (IL-2 receptor a chain). 11, 12 Since IL-2 is a potent growth factor for CD25-expressing activated T-cells 15 , the elevation of both IL-2 and CD25 indicates that a positive feedback loop for T-cell activation is established and overrunning in severe patients.

These collectively highlight the roles of T-cells in the pathogenesis of severe SARS-CoV-2 infection, although the pathogenetic mechanisms are largely unknown.

In this study, we analysed the transcriptomes of CD4 + T-cells from a single cell RNA-seq dataset from a recent study 16 and thereby investigated the gene regulation dynamics during SARS-CoV-2 infection. We show that SARS-CoV-2 induces multiple activation and differentiation processes in CD4 + T-cells in a unique manner. We identify defects in Foxp3mediated negative regulation, which accelerates T-cell activation and death. In addition, by analysing multiple transcriptome datasets, we propose the possibility that those abnormally activated T-cells enhance viral entry through the production of Furin in severe COVID-19 patients.

We recently showed that, using scRNA-seq analysis, melanoma-infiltrating T-cells are activated in situ and differentiate into either FOXP3 high activated Tregs or PD-1 high T follicular helper (Tfh)-like effector T-cells 17 . We hypothesised that those mechanisms for T-cell activation and differentiation in inflammatory sites are altered in COVID-19 patients. To address this issue, we analysed the scRNA-seq data from bronchoalveolar lavage (BAL) fluids of mild and severe COVID-19 patients. 16 Firstly, we performed in silico sorting of CD4 + T-cells and analysed their transcriptomes (Fig.   1a ). We applied a dimensional reduction to the CD4 + T-cell data using Uniform Manifold Approximation and Projection (UMAP) and Principal Component Analysis ( Supplementary   Fig. 1a) . As expected, most T-cells were from either mild or severe COVID-19 patients.

Notably, clusters 1, 2, 3, 5, 6, and 7 did not contain any cells from healthy controls (HC) ( Fig.   1b and 1c) , indicating that these cells uniquely differentiated during the infection, regardless of whether it was mild or severe disease.

Differential gene expression analysis showed that in comparison to mild patients, CD4 + Tcells from severe COVID-19 patients expressed higher levels of the AP-1 genes FOS, FOSB, and JUN, the activation marker MKI67 (Ki67), Th2-related genes IL4R and MAF, and chemokines including CCL2, CCL3, CCL4, CCL7, CCL8, and CXCL8 (Fig. 1d, Supplementary Fig.   1b ). These suggest that CD4 + T-cells in severe COVID-19 patients are highly activated in the (Fig. 1d, Supplementary Fig. 1b) . These were further confirmed by pathway analysis, which identified interleukin, JAK-STAT, and MAPK signalling pathways as significantly enriched pathways (Fig. 1e) .

On the other hand, CD4 + T-cells from severe patients showed decreased expression of interferon-induced genes including IFIT1, IFIT2, IFIT3, and IFITM1 (Fig. 1d) . Pathway analysis also showed that CD4 + T-cells from severe patients expressed lower levels of the genes related to interferon downstream pathways (Fig. 1e) , suggesting that type-I interferons are suppressed in severe patients. Notably, CD4 + T-cells in severe patients showed lower expression of the TNF superfamily ligands TNFSF10 (TRAIL) and TNFSF14 (LIGHT) and the surface protein SLAMF1 and KLRB1, all of which have roles in viral infections [18] [19] [20] [21] .

Next, we performed a pseudotime analysis in the UMAP space, identifying two major trajectories of T-cells, originating in Cluster 0 (Fig. 2a, 1b) . Pseudotime 1 (t1) involved Clusters 0, 2, 1, 7, 5, 6, and 3, showing a longer trajectory, while pseudotime 2 (t2) involved Clusters 0, 8, and 4. Interestingly, the cells in the origin showed high expression of IL-7 receptor (IL7R), which is a marker of naïve-like T-cells in tissues. 17 The expression of IL7R was gradually downregulated across the two pseudotime trajectories (t1 and t2, Fig. 2b) .

Intriguingly, T-cells at the ends of both trajectories included MKI67 (Ki-67) + T-cells and some cells were NR4A1 + or NR4A3 + (Fig. 2c, Supplementary Fig S1c) , which indicated activation and cognate antigen signalling 22 . These analyses support that the trajectories successfully captured two major pathways for T-cells to be activated during the infection.

Interestingly, well-known immunoregulatory genes including IL2RA, CTLA4, TNFRSF18, and TNFRSF4 were more expressed in T-cells across pseudotime 1 than pseudotime 2 (Fig. 2d,   Supplementary Fig. 1b) . Although these genes are often associated with Tregs, FOXP3 was not induced in both of these trajectories, and thus most of the T-cells did not become Tregs.

T-cells in pseudotime 2 showed modest increase of CTLA4 and TNFRSF18 only towards the end of the trajectory (Fig. 2d) .

Furthermore, IL2RA was significantly upregulated in CD4 + T-cells from severe COVID-19 patients compared to mildly affected patients (Fig. 2e) . Since CD25 (IL2RA) is a key marker for Tregs and activated T-cells 23 , we asked if CD25-expressing T-cells in COVID-19 patients were Tregs. Intriguingly, the percentage of FOXP3 + cells in IL2RA + CD4 + T-cells was significantly reduced in severe patients compared to mild patients (15.3% vs 48.6%, Fig. 2f ).

This indicates that FOXP3 transcription is repressed in CD25 + T-cells further suggesting that IL2RA + T-cells are activated T-cells or 'ex-Tregs' (i.e. effector T-cells that were used to express FOXP3 but more recently downregulated FOXP3 expression 24, 25 ), rather than functional Tregs.

Since T-cells in the late phase of pseudotime 1 upregulated IL2RA and MKI67, we asked if expanded TCR clones expressed more IL2RA. Expanded TCR clones in severe patients had more IL2RA + T-cells than those in mild patients (14% and 28% cells in mild and severe patients expressed IL2RA, respectively; p < 0.001, Fig. 2g ), confirming that IL2RA + T-cells are associated with the severe phenotype. However, no significant difference was observed between expanded and non-expanded TCR clones in severe patients. Notably, IL2 transcription was not induced in CD4 + T-cells in severe patients (Fig. 2h) , suggesting that CD25 + activated T-cells in severe patients die, at least partly, by cytokine deprivation.

Given that T-cells from severe patients dominated in the last part of pseudotime 1 (i.e. Clusters 3 and 6), these findings indicate that T-cells become more activated and vigorously proliferate in severe COVID-19 patients than mild patients. These CD25 + activated T-cells are likely to be short-lived and do not initiate FOXP3 transcription in severe COVID-19 patients, while they can differentiate into Tregs in moderate infections.

Firstly, we hypothesized that FOXP3 transcription was actively repressed by cytokines in the microenvironments in severe COVID-19 patients. FOXP3 transcription is activated by IL-2 and TGF-b signalling and is repressed by IL-6 and IL-12 signalling. 26 In fact, some macrophages from severe COVID-19 patients expressed TGFB1 and IL6 (Fig. 2i) , as shown by Liao et al. 16 However, CD4 + T-cells did not increase Th17-associated genes including RORC, IL17A, and IL17F, and the expression of CCR6, a marker for Th17 cells, was significantly reduced in severe COVID-19 patients ( Fig. 2j and Supplementary Fig. 1d ). This suggests that the differentiation of both Tregs and Th17 is inhibited.

Th17 differentiating T-cells express both Foxp3 and RORg-t before they mature. 27 In addition, FOXP3 intermediate CD45RA -T-cells express RORg-t and Th17 cytokines. 28 Together with the scRNA-seq analysis results above, these support the model that activated T-cells show differentiation arrest or preferentially die before becoming Tregs or Th17 cells in severe COVID-19 patients. Importantly, IL2RA expression was significantly increased in severe COVID-19 patients in comparison to mild patients, whilst very few T-cells expressed IL2 (Fig. 2h) .

PD-1 is another key immunoregulatory molecule for suppressing immune responses during viral infection. 29 However, PD-1 may play multiple roles in CD4 + T-cells, as PD-1 is a marker for Tfh. In fact, PD-1 high BCL6 high Tfh-like T-cells are a major effector population in melanoma tissues. 17 Thus we asked if PD-1-expressing T-cells show Tfh differentiation and/or if PD-1expressing T-cells can succumb to PD-1 ligand-mediated inhibition in COVID-19 patients.

However, in SARS-CoV-2 infection, BCL6 was not induced in the major activation and differentiation pathway, pseudotime 1, indicating that those activated T-cells did not differentiate into Tfh. Comparatively, T-cells in pseudotime 2 showed some upregulation, although this was statistically not significant (Fig. 3a) . This suggests that Tfh differentiation was suppressed in COVID-19 patients. PDCD1 was highly upregulated in both pseudotime 1 and 2, suggesting that these cells are vulnerable to PD-1 ligand-mediated suppression.

Interestingly, macrophages from severe COVID-19 patients expressed higher levels of CD274 (PD-L1) yet the expression of PDCD1LG2 (PD-L2) was not significantly different between mild and severe patients (Fig. 3b) . These results indicate that PD-1-mediated T-cell regulation was at least partially operating in severe COVID-19 patients. Meanwhile, TBX21 and GATA3 expression is induced in T-cells across pseudotime 1, suggesting that these T-cells may differentiate into Th1 and Th2 cells (Fig. 3c) .

T-cells in the late phase of pseudotime 1 upregulated the expression of cytokines including IFNG and IL10, which are Th1 and Th2 cytokines, respectively (Fig. 3d) . In addition, IL21 (a Th2 and Th17 cytokine) was upregulated in some cells in pseudotime 1 whereas IL-32 was highly sustained in both of the pseudotime trajectories. These indicate that differentiation processes for T-cell lineages are simultaneously induced in activated T-cells from the lung of COVID-19 patients.

Accordingly, we hypothesized that CD25-expressing activated T-cells preferentially differentiate into effector T-cells in severe COVID-19 patients, instead of their most frequent fate as Tregs in a normal setting. 23 In order to test this hypothesis and reveal dynamics of each T-cell differentiation pathway, we analysed co-regulated genes across pseudotime 1, obtaining 4 gene modules by a hierarchical clustering (Fig. 3e) .

Heatmap analysis of pseudotime 1 successfully captured the pseudo-temporal order of gene expression: genes in modules II and IL7R are firstly activated, followed by genes in module IV (apart from IL7R), subsequently by genes in module I, and lastly genes in module III alongside module II genes again (Fig. 3e) . Reasonably, module II contained the AP-1 transcription factors FOS and JUN, suggesting that T-cells that highly express these genes have been recently activated. In pseudotime 1, these FOS + JUN + T-cells were followed by Tcells with high expression of genes in module IV, which contained CD40LG and FASLG (Fig.   3e) . These CD40LG + FASLG + T-cells are considered to activate CD40-expressing macrophages and dendritic cells as well as inducing apoptosis of FAS-expressing cells by providing CD40 signalling and Fas signalling upon contact.

CD40LG + FASLG + T-cells are followed by T-cells that highly expressed genes in module I, which include the Th1 transcription factor TBX21, the Th2 transcription factor GATA3, and FOXP3. In addition, these T-cells upregulated the immediate early genes EGR1 and NFATC2 and the activation markers CD38 and PDCD1 (Fig. 3e) . These collectively suggest that those (GITR), all of which were found to be upregulated (Fig. 3e) . On the other hand, 

In order to further understand why CD25-expressing T-cells failed to differentiate into effector Tregs, we hypothesized that CD25-expressing activated T-cells are more likely to differentiate into multiple effector T-cell subsets in severe COVID-19 patients than mildly affected individuals. In fact, IL2RA + CD4 + T-cells from severe COVID-19 patients expressed Th1, Th2, and IL-10 signature genes more frequently (Fig. 3f) whereas Th17 differentiation was suppressed in IL2RA + T-cells from both groups.

Although IL-10 has been classically regarded as a Th2 cytokine, Th1 cells can produce IL-10. 30 In fact, IL4, IL5, IL12A, and IL13, were not detected in any T-cells analysed in the dataset (data not shown). Thus, we asked if Th2 differentiation was diverted into IL-10 producing immunoregulatory T-cells (Tr1), which differentiate by IL-10 signalling and produce IL-10 and thereby suppress immune responses particularly in mucosal tissues. 31 . Surprisingly, however, higher frequencies of IL2RA + T-cells expressed Th1 and Th2 signature genes, concomitantly expressing IL10 signature genes (Fig. 3g) . This strongly supports that IL-10

producing T-cells are not immunoregulatory but Th1-Th2 multifaceted effector T-cells.

IL10RA and IL10RB were expressed in activated T-cells in both of the trajectories (Supplementary Fig. 1c) , and the frequency of IL10RA + cells was significantly higher in IL10 + cells than IL10-cells (67% vs 49%, p = 0.003, Fig. 3h) . This suggests that a positive feedback loop for IL-10 expression promoted the differentiation of the multifaceted effector T-cells in an autocrine manner 32, 33 .

Intriguingly, FURIN expression was increased in CD4 + T-cells from severe COVID-19 patients (Fig. 1d) . Furin was previously associated with Treg functions in a knockout study 8 , although the underlying mechanisms were not clear. In addition, it was not known if Furin was specifically expressed in Tregs amongst CD4 + T-cells. Recently we showed that the majority of Treg-type genes are in fact regulated by TCR signalling 17, 34 , since Tregs receive infrequent-yet-regular TCR signals in vivo 22 . We hypothesise that T-cells produce Furin upon activation, which can enhance the viral entry of SARS-CoV-2 into lung epithelial cells during inflammation.

Firstly, we analysed the microarray data of various CD4 + T-cell populations from mice 35 . In line with our previous observations 17 , all the antigen-experienced and activated T-cell populations including Tregs, memory-phenotype T-cells, and tissue-infiltrating effector Tcells showed higher expression of FURIN than naïve T-cell populations (Fig. 4a) .

Next, we asked if human naïve and memory CD4 + T cells can express FURIN upon receiving TCR signals. We addressed this question using the time course RNA-seq analysis of CD45RA + CD45RO − CD4 + naïve T-cells and CD45RA − CD45RO + CD4 + memory T-cells which were obtained from 4 individuals and activated by anti-CD3 and anti-CD28 antibodies 36 .

sustained over 2 weeks in the culture (Fig. 4b) .

Memory T-cells also showed higher expression of FURIN over the time course (p = 0.059), with a significant difference at its peak (24 hours, p = 0.004). Thus, FURIN expression is induced by TCR signals in human and mouse T-cells.

In SARS-CoV-2 infection, 36% of CD4 + T-cells and 56% of IL2RA + CD4 + T-cells from severe patients expressed FURIN, while in mild patients only 11% and 30% of those cells, respectively, expressed FURIN (Fig. 4c) . Importantly, FURIN was significantly induced in CD4 + T-cells in pseudotime 1, particularly when T-cells upregulated CD25, CTLA-4, and TNFRSF molecules, but not in pseudotime 2 (Fig. 4d) . These collectively support that FURIN expression is induced in highly activated non-regulatory CD25 + CD4 + T-cells in severe COVID-

Our study has shown that CD4 + T-cells in severe COVID-19 patients have dysregulated activation and differentiation mechanisms. The most remarkable defect was the decoupling of Treg-type activation and FOXP3 expression, which normally occurs simultaneously to sustain the effector Treg population while inflammation is resolved. 34 This Treg-type setting is further enhanced upon activation, when Tregs begin to show the effector Treg phenotype, further upregulating the expression of the immune checkpoint molecules. Importantly, Tregs need to sustain FOXP3 transcription in a persistent manner across time, 38 otherwise they can downregulate FOXP3 expression and become effector Tcells. 39, 40 Since IL-2 signalling enhances FOXP3 transcription, CD25 + T-cells are likely to differentiate into FOXP3 + Tregs in normal situations. 23 However, in severe COVID-19 patients, those CD25 + T-cells are considered to be vigorously proliferating, whilst becoming multifaceted effector T-cells or dying, instead of maturing into FOXP3 + Tregs. Accordingly, we propose to define the unique activation status of CD25 + FOXP3-T-cells as hyperactivated T-cells (Fig. 5) .

CD25 expression occurs mostly in CD4 + T-cells, and therefore, these CD25 + hyperactivated Tcells are likely to be the source of the elevated serum soluble CD25 in severe COVID-19 patients. These hyperactivated CD25 + T-cells produce Furin, which can further enhance SARS-CoV-2 viral entry into pulmonary epithelial cells.

The risk factors for the development of severe COVID-19 include age, obesity, cardiovascular diseases, diabetes, and the use of corticosteroids. 41, 42 These diseases are associated with dysregulated hormonal and metabolic environments that can dysregulate the homeostasis of CD25 + T-cells and FOXP3-expressing Tregs. Thus, it is imperative to investigate if genes and metabolites associated with the disease conditions have any roles in promoting the differentiation of hyperactivated T-cells. Previous reports showed that Tregs accumulated in atherosclerotic lesions, 43 and FOXP3 expression was reduced in CD25 + CD4 + T-cells from patients with prior myocardial infarctions. 44 In addition, T-cells in patients with obesity may show different responses to T-cell activation. Intriguingly, leptin a key hormone produced by adipose tissue, is thought to prevent CD25 + CD4 + T-cell proliferation but is relatively deficient in obese patients. 45, 46 Furthermore, the function of Tregs is impaired in type-1 diabetes patients. 47 In addition, FOXP3 transcription is transiently activated in T-cells of severe COVID-19 patients but may be repressed due to their unique metabolic states. Tcell activation is dependent on glycolysis, which converts glucose to pyruvate, and the tricarboxylic acid (TCA) cycle, which activates oxidative phosphorylation (OXPHOS) and generates ATP in mitochondria. 48 Treg differentiation is more dependent on OXPHOS and can be inhibited by glycolysis. 49 Importantly, our pathway analysis suggested that these metabolic pathways were altered in severe COVID-19 patients, although further studies on metabolism are required. Furthermore, the hypoxic environment in the lung of severe COVID-19 patients may activate HIF-1a, which mediates aerobic glycolysis, and thereby promotes the degradation of FOXP3 proteins. 50 The reduction of FOXP3 proteins may result in the abrogation of the FOXP3 autoregulatory transcriptional loop thus blocking Treg differentiation. 38 CD25 + hyperactivated T-cells also expressed PD-1, and PD-L1 expression in macrophages was increased in severe COVID-19 patients. This clearly shows that the PD-1 system is not able to control hyperactivated T-cells. This may be due to the status of macrophages and other antigen-presenting cells because CD80 on these cells disrupts the PD-1 -PD-L1 interaction and thereby abrogates PD-1-mediated suppression. 51 In addition, PD-L1 expression on lung epithelial cells may play a role in regulating PD-1-expressing T-cells, as shown in other viruses including Influenza Virus and Respiratory Syncytial Virus. 52, 53 Hyperactivated T-cells differentiated into multifaceted Th1-Th2 cells with IL-10 expression.

While IL-10 may serve as a growth factor for these cells through their IL-10 receptors, other cytokines in the microenvironment may drive the expression of both Th1 and Th2 (Fig. 5) .

In conclusion, our study demonstrates that SARS-CoV-2 drives hyperactivation of CD4+ Tcells and immune paralysis to promote the pathogenesis of disease and thus life-threatening symptoms in severely affected individuals. Therefore, therapeutic approaches to inhibit Tcell hyperactivation and paralysis may need to be developed for severe COVID-19 patients

The single-cell-RNA-seq data from COVID-19 patients and healthy individuals was obtained from GSE145926. 16 The microarray data of murine T-cell subpopulation were from the Immunological Genome Project (GSE15907 35 ). The RNA-seq data in GSE73213 36 was used for time course analysis of naïve and memory CD4 + T-cells.

We used h5 files of the scRNA-seq dataset (GSE145926 16 ) which were aligned to the human genome (GRCh38) using Cell Ranger, by importing them into the CRAN package Seurat 3.0. 57 Single cells with high mitochondrial gene expression (higher than 5%) were excluded from further analyses. In silico sorting of CD4 + T-cells was performed by identifying them as the single cells CD4 and CD3E, because no other methods, including the Bioconductor package singleR, reliably identified CD4 + T-cells. The TCR-seq data of GSE145926 16 was used to validate the in silico sorting and also for analysing gene expression in expanded clones.

Macrophages were similarly identified by the ITGAM expression and lack of PAX5, CD19 and CD3E expressions.

PCA was applied on the scaled data followed by a K-nearest neighbor clustering in the PCA space. UMAP was performed on clustered data using the first PCA axes. Differentially 

The enrichment of biological pathways in the gene lists was tested by the Bioconductor package clusterProfiler, 58 using the Reactome database through the Bioconductor package ReactomePA, and pathways with false discovery rate < 0.01 and q-value < 0.1 were considered significant.

Trajectories were identified using the Bioconductor package slingshot, assuming that the cluster that show the highest expression of IL7R and CCR7 is the origin. The CRAN package ggplot2 was used to apply a generalised additive model of the CRAN package gam to each gene expression data. Genes that were differentially expressed across pseudotime was obtained by applying the generalised additive model to the dataset using gam, performing ANOVA for nonparametric effects and thereby testing if each gene expression is significantly changed across each pseudotime (p-value < 0.05).

The enrichment of cytokine-expressing single cell T-cells was tested using a chi-square test.

The time course data of FURIN expression was analysed by one-way ANOVA with Tukey's honest significant difference test. (b) In severe COVID-19 patients, hyperactivated macrophages 13 may present antigens to CD4+ T-cells, which are activated and differentiate into CD25+ IL10R+ early activated T-cells which produce IL-10 rather than IL-2. FOXP3 transcription remains to be suppressed due to this and other unidentified mechanisms such as metabolism, while cytokines such as IL-10 further enhance the activation of CD25+ T-cells, resulting in the generation of CD25+

hyperactivated T-cells that express immune checkpoints, multiple effector T-cell cytokines, and Furin. The multifaceted Th differentiation may lead to unfocused T-cell responses and thereby paralyse the T-cell system. In addition, Furin can activate the S-protein of SARS-CoV-

Stimulatory and Inhibitory Co-signals in Autoimmunity

From stability to dynamics: understanding molecular mechanisms of regulatory T cells through Foxp3 transcriptional dynamics

SARS-CoV-2 Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically Proven Protease Inhibitor

Evidence that TMPRSS2 Activates the Severe Acute Respiratory Syndrome Coronavirus Spike Protein for Membrane Fusion and Reduces Viral Control by the Humoral Immune Response

SARS-CoV-2 entry factors are highly expressed in nasal epithelial cells together with innate immune genes

The spike glycoprotein of the new coronavirus 2019-nCoV contains a furin-like cleavage site absent in CoV of the same clade

Cleavage Site in the Spike Protein of SARS-CoV-2 Is Essential for Infection of Human Lung Cells

T-cell-expressed proprotein convertase furin is essential for maintenance of peripheral immune tolerance

Proprotein convertase furin is preferentially expressed in T helper 1 cells and regulates interferon gamma

Proprotein convertase FURIN constrains Th2 differentiation and is critical for host resistance against Toxoplasma gondii

Clinical features of patients infected with 2019 novel coronavirus in Wuhan

Clinical and immunological features of severe and moderate coronavirus disease 2019

Immune mechanisms of pulmonary intravascular coagulopathy in COVID-19 pneumonia

Reduction and Functional Exhaustion of T Cells in Patients with Coronavirus Disease 2019 (COVID-19). medRxiv

Structure and function of the interleukin 2 receptor: affinity conversion model

Single-cell landscape of bronchoalveolar immune cells in patients with COVID-19

Elucidating T Cell Activation-Dependent Mechanisms for Bifurcation of Regulatory and Effector T Cell Differentiation by Multidimensional and Single-Cell Analysis

Role of Tumor Necrosis Factor-Related Apoptosis-Inducing Ligand in Immune Response to Influenza Virus Infection in Mice

TNF Superfamily Networks: bidirectional and interference pathways of the herpesvirus entry mediator (TNFSF14)

The Measles Virus Receptor SLAMF1 Can Mediate Particle Endocytosis

CD161intCD8+ T cells: a novel population of highly functional, memory CD8+ T cells enriched within the gut

A timer for analyzing temporally dynamic changes in transcription during differentiation in vivo

Controversies concerning thymus-derived regulatory T cells: fundamental issues and a new perspective

Self-antigen-driven activation induces instability of regulatory T cells during an inflammatory autoimmune response

Plasticity of Foxp3(+) T cells reflects promiscuous Foxp3 expression in conventional T cells but not reprogramming of regulatory T cells

Signal transducer and activator of transcription 4 limits the development of adaptive regulatory T cells

TGF-beta-induced Foxp3 inhibits T(H)17 cell differentiation by antagonizing RORgammat function

Functional delineation and differentiation dynamics of human CD4+ T cells expressing the FoxP3 transcription factor

The PD-1/PD-L1 Axis and Virus Infections: A Delicate Balance

Interleukin-10 production by effector T cells: Th1 cells show self control

The Biology of T Regulatory Type 1 Cells and Their Therapeutic Application in Immune-Mediated Diseases

IL-10 Receptor Signaling Is Essential for T<sub>R</sub>1 Cell Function In Vivo

IL-10: The Master Regulator of Immunity to Infection

Control of regulatory T-cell differentiation and function by T-cell receptor signalling and Foxp3 transcription factor complexes

Immunological Genome Project, C. The Immunological Genome Project: networks of gene expression in immune cells

Promoter H3K4 methylation dynamically reinforces activation-induced pathways in human CD4 T cells

Foxp3 controls regulatory T-cell function by interacting with AML1/Runx1

A temporally dynamic Foxp3 autoregulatory transcriptional circuit controls the effector Treg programme

Plasticity of Foxp3(+) T cells reflects promiscuous Foxp3 expression in conventional T cells but not reprogramming of regulatory T cells

Self-antigen-driven activation induces instability of regulatory T cells during an inflammatory autoimmune response

Risk Factors Associated With Acute Respiratory Distress Syndrome and Death in Patients With Coronavirus Disease

Letter to the Editor: Obesity as a risk factor for greater severity of COVID-19 in patients with metabolic associated fatty liver disease

Statins induce the accumulation of regulatory T cells in atherosclerotic plaque

Regulatory T cells and IL-10 levels are reduced in patients with vulnerable coronary plaques

Drug Insight: the role of leptin in human physiology and pathophysiology-emerging clinical applications

A Key Role of Leptin in the Control of Regulatory T Cell Proliferation

Are Regulatory T Cells Defective in Type 1 Diabetes and Can We Fix Them?

Metabolic Pathways in Immune Cell Activation and Quiescence

Sugars and fat -A healthy way to generate functional regulatory T cells

Control of T(H)17/T(reg) balance by hypoxia-inducible factor 1

Restriction of PD-1 function by <em>cis</em>-PD-L1/CD80 interactions is required for optimal T cell responses

Local Blockade of Epithelial PDL-1 in the Airways Enhances T Cell Function and Viral Clearance during Influenza Virus Infection

Expression of Programmed Death-1 Ligand (PD-L) 1, PD-L2, B7-H3, and Inducible Costimulator Ligand on Human Respiratory Tract Epithelial Cells and Regulation by Respiratory Syncytial Virus and Type 1 and 2 Cytokines

SnapShot: effector and memory T cell differentiation

Tissue-Derived Hedgehog Proteins Modulate Th Differentiation and Disease

Foxp3-Deficient Regulatory T Cells Do Not Revert into Conventional Effector CD4<sup>+</sup> T Cells but Constitute a Unique Cell Subset

Comprehensive Integration of Single-Cell Data

clusterProfiler: an R Package for Comparing Biological Themes Among Gene Clusters

MO is supported by MRC project grant (MR/S000208/1) and is currently a Visiting Associate