key: cord-0841116-yrfcqf4b authors: Fan, X.; Chi, X.; Ma, W.; Zhong, S.; Dong, Y.; Zhou, W.; Ding, W.; Fan, H.; Yin, C.; Zuo, Z.; Yang, Y.; Zhang, M.; Ma, Q.; Liu, J.; Fang, T.; Wu, Q.; Chen, W.; Wang, X. title: Single-cell RNA-seq and V(D)J profiling of immune cells in COVID-19 patients date: 2020-05-27 journal: nan DOI: 10.1101/2020.05.24.20101238 sha: f85cd535c65ed6ee4bc85647bfa0648815c0fab0 doc_id: 841116 cord_uid: yrfcqf4b Coronavirus disease 2019 (COVID-19) has caused over 220,000 deaths so far and is still an ongoing global health problem. However, the immunopathological changes of key types of immune cells during and after virus infection remain unclear. Here, we enriched CD3+ and CD19+ lymphocytes from peripheral blood mononuclear cells of COVID-19 patients (severe patients and recovered patients at early or late stages) and healthy people (SARS-CoV-2 negative) and revealed transcriptional profiles and changes in these lymphocytes by comprehensive single-cell transcriptome and V(D)J recombination analyses. We found that although the T lymphocytes were decreased in the blood of patients with virus infection, the remaining T cells still highly expressed inflammatory genes and persisted for a while after recovery in patients. We also observed the potential transition from effector CD8 T cells to central memory T cells in recovered patients at the late stage. Among B lymphocytes, we analyzed the expansion trajectory of a subtype of plasma cells in severe COVID-19 patients and traced the source as atypical memory B cells (AMBCs). Additional BCR and TCR analyses revealed a high level of clonal expansion in patients with severe COVID-19, especially of B lymphocytes, and the clonally expanded B cells highly expressed genes related to inflammatory responses and lymphocyte activation. V-J gene usage and clonal types of higher frequency in COVID-19 patients were also summarized. Taken together, our results provide crucial insights into the immune response against patients with severe COVID-19 and recovered patients and valuable information for the development of vaccines and therapeutic strategies. We first performed unbiased clustering of the single-cell mRNA profiles and 102 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. . https://doi.org/10.1101/2020.05.24.20101238 doi: medRxiv preprint identified 38 clusters, which could be categorized into 18 cell types (Fig. 1a-c, S1a) . 103 There were 9 classes of T lymphocytes (CD3E+) identified, including 4 subtypes of 104 CD4+ T cells (11,961 cells) and 5 subtypes of CD8+ T cells (17,683 cells) (Fig. 1a, b) . 105 Interestingly, we observed that the percentage of T cells was decreased in SARS-CoV- (2,284 cells) and plasmablasts (698 cells) (Fig 1a-c) . In addition to T and B lymphocytes, 110 we also picked up some other types of PBMCs, including monocytes (7,508 CD14+ (Fig. S3b) . Accordingly, the majority of B lymphocytes in RE samples were 127 B cells (Fig S3b) . The majority of plasma B cells (85.4%), which are responsible for 128 antibody production in an effective immune response, in patients with severe clinical 129 features (Fig 1d, e, S3b ). To further investigate how T cells responded to SARS-CoV-2 virus infection, we first 131 compared the T cell subclusters (Fig. S4a) . We found that the 15 subclusters of T cells 132 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. . https://doi.org/10.1101/2020.05.24.20101238 doi: medRxiv preprint could be grouped into 5 modules based on the highly variable genes in the T cell 133 transcriptomic profiles rather than on patient groups (Fig. S4a, b) . Then, we analyzed 134 the differentially expressed genes (DEGs) of T cells from different patient groups (Fig. 135 2a). We found that inflammatory genes, including IFNG (interferon gamma), CD160, 136 S100A8 and GZMA, were highly expressed by T cells from patients with severe 137 infection, indicating that T cells were highly activated to participate in the response 138 against virus infection in these patients ( Fig. 2a-d) . Accordingly, gene ontology (GO) 139 analysis of the DEGs suggested that cytokine production and immune response-related 140 leukocyte activation may occur in COVID-19 patients with severe symptoms (Fig. 2b ). Interestingly, T cells from RE patients highly expressed RNF125, CXCR4, and PELI1, 142 indicative of T cell activation still existing even after recovery at early time, which is 143 consistent with the GO analysis results. Since CD8+ T cells, also known as cytotoxic T cells, play essential roles in 145 recognizing, binding and killing cells when infected by viruses 11 , we determined the 146 differentiation trajectories of the CD8+ T cells by monocle analysis 12 (Fig. 2e, S5a) . We activation. This indicated that naïve CD8 T cells from recovered patients may be at 156 different states with specific transcriptome profiles ( Fig S4c) . Additionally, most central 157 memory CD8 T cells were from healthy samples (41.4%), while most effector memory 158 CD8 T cells (57.6%) were from RL samples, indicating that recovered patients probably 159 have a recent memory of immune responses induced by SARS-CoV-2 infection. Next, 160 we further analyzed the DEGs of effector CD8 T cells based on patient groups (Fig. 2f, 161 g). The effector CD8 T cells from severe samples highly expressed PTGDR, which has 162 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. . https://doi.org/10.1101/2020.05.24.20101238 doi: medRxiv preprint been identified as a mediator of allergic airway inflammation 13 , and GZMK , XCL2, 163 which were highly expressed in activated T cells, indicating that effector CD8 T cells 164 were highly active in disease conditions. Additionally, CXCR4 and RNF125, which play 165 roles in T cell migration, maintenance and activation, were expressed by CD8 T cells 166 from the RE group but not by those from the RL group (Fig. 2f, g) , suggesting that 167 inflammatory responses are still active when SARS-CoV-2 virus is eliminated. We also 168 found that CCR7 and SELL were relatively highly expressed by effector CD8 T cells 169 from RL samples compared to those from other groups, indicating that these T cells Although we used magnetic beads to enrich CD3+ and CD19+ lymphocytes, we still 184 captured 8,285 monocyte, which were categorized as classical CD14+ monocyte and 185 non-classical FCGR3A (CD16)+ monocyte (Fig 2i) . The subclusters of monocytes 186 showed distinct gene expression patterns that correlated with the status of SARS-CoV-187 2 infection (Fig. 2i, S5b) . In CD14+ monocyte, almost all cluster 22 cells (1,328/1,334) 188 were from severe samples with high expression of KLF6 and IL1R2 (Fig 2i) . As (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. TGFBI signals were decreased in severe samples (Fig. 2j) . In B cells, we found that the composition of each cluster differed in terms of infection 217 status (Fig. 2l) . To further explore the gene expression differences of B cells in the 218 various COVID-19-related states, we next analyzed the DEGs and GO terms of virus-219 infected and healthy people (Fig2n, o, S6b). We found that genes playing roles in the 220 response to viruses, the regulation of cytokine production, and apoptosis were enriched 221 in patients with severe conditions (Fig. 2o ). In addition, GO analysis indicated that the 222 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. (Fig. 3d ). In addition to 256 these highly used pairs, IGLV2-23::IGLJ6 was a unique combination in severe COVID-257 19 patients (Fig 3d) . Additionally, the TCR pairings with the highest frequencies in (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. . https://doi.org/10.1101/2020.05.24.20101238 doi: medRxiv preprint and MX1, indicating that these cells were activated (Fig. 4d) . Importantly, we traced the (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. . https://doi.org/10.1101/2020.05.24.20101238 doi: medRxiv preprint The filtered gene-cell matrix was normalized to 10 4 molecules per cell for sequencing (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. o. Representative GO terms of group-specific genes in n. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. . https://doi.org/10.1101/2020.05.24.20101238 doi: medRxiv preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 27, 2020. . https://doi.org/10.1101/2020.05.24.20101238 doi: medRxiv preprint Umap: Uniform manifold approximation and 580 projection for dimension reduction From Louvain to Leiden: guaranteeing 582 well-connected communities transcriptomic data across different conditions, technologies, and species Metascape provides a biologist-oriented resource for the analysis of systems-587 level datasets Single-cell mRNA quantification and differential analysis with Census The dynamics and regulators of cell fate decisions are revealed by 591 pseudotemporal ordering of single cells Dynamic interaction networks in a hierarchically organized tissue Single-cell transcriptomic profiling of the aging mouse brain CellPhoneDB: 598 inferring cell-cell communication from combined expression of multi-subunit ligand-receptor 599 complexes Molecular classification and comparative taxonomics of foveal and 601 peripheral cells in primate retina Models and estimators linking individual-based and sample-based 603 rarefaction, extrapolation and comparison of assemblages VDJtools: unifying post-analysis of T cell receptor repertoires VDJdb in 2019: database extension