BRIEF COMMUNICATION https://doi.org/10.1038/s41591-020-0901-9 Single-cell landscape of bronchoalveolar immune cells in patients with COVID-19 Mingfeng Liao1,6, Yang Liu1,6, Jing Yuan2,6, Yanling Wen1, Gang Xu1, Juanjuan Zhao1, Lin Cheng1, Jinxiu Li2, Xin Wang1, Fuxiang Wang2, Lei Liu1,3 ✉ , Ido Amit 4 ✉ , Shuye Zhang 5 ✉ and Zheng Zhang 1,3 ✉ Respiratory immune characteristics associated with The macrophage compartments differed largely in both compo- Coronavirus Disease 2019 (COVID-19) severity are cur- sition and expression of FCN1, SPP1 and FABP4 in different cell rently unclear. We characterized bronchoalveolar lavage fluid groups (Fig. 1c and Extended Data Fig. 2d). FABP4 was prefer- immune cells from patients with varying severity of COVID-19 entially expressed by controls and patients with moderate infec- and from healthy people by using single-cell RNA sequencing. tion, while FCN1 and SPP1 were highly expressed by patients Proinflammatory monocyte-derived macrophages were abun- with severe/critical infection (Fig. 1d). We conducted differen- dant in the bronchoalveolar lavage fluid from patients with tially expressed gene (DEG) analysis (Extended Data Fig. 2e), gene severe COVID-9. Moderate cases were characterized by the ontology (GO) analysis and gene set enrichment analysis (GSEA) presence of highly clonally expanded CD8+ T cells. This atlas (Extended Data Fig. 2f) between cell groups. Group 1 expressed of the bronchoalveolar immune microenvironment suggests the peripheral monocyte-like markers S100A8, FCN1 and CD14, potential mechanisms underlying pathogenesis and recovery and group 2 expressed high levels of the chemokines CCL2, CCL3, in COVID-19. CXCL10 and other genes. Both groups had gene expression patterns The underlying immune responses to severe acute respira- suggestive of classic M1-like macrophages. Group 3 expressed the tory syndrome (SARS)-CoV2 remain unclear. Abnormal immune immunoregulatory genes A2M, GPR183 and CCL13 and the pro- responses have been reported in coronavirus animal infection fibrotic genes TREM2, TGFB1 and SPP1, suggestive of alternative models1,2 and lymphopenia is a prominent feature of severe M2-like macrophages, a reparative but profibrotic subset5. Group COVID-19 (ref. 3). Bronchoalveolar lavage fluid (BALF) contains 4 expressed the alveolar macrophage (AM) genes FABP4, APOC1, microenvironment information on bronchioles and lung alveoli. MARCO and genes associated with lipid metabolic functions6 and Here, we performed single-cell RNA sequencing (scRNA-seq) on was enriched in BALFs of controls and moderate cases but not in BALF cells from three patients with moderate COVID-19 (M1– severe cases (Extended Data Fig. 2f,g). Using SCENIC, we found M3), six patients with severe/critical infection (S1–S6), three enhanced expression of the transcription factor genes NFκB, STAT1, healthy controls (HC1–HC3) and a publicly available BALF (HC4)4 STAT2 and interferon (IFN) regulatory factors in group 1 and 2, sample (Supplementary Tables 1 and 2). Clustering analysis showed M2-promoting TFEB, NR1H3, PPARA and CREB1 activities in 31 distinct clusters composed of macrophages (CD68), neutrophils group 3 and AM-specific PPARG and CEBPB in group 4 macro- (FCGR3B), myeloid dendritic cells (mDCs) (CD1C, CLEC9A), plas- phages (Extended Data Fig. 2h and Methods). These findings sug- macytoid dendritic cells (pDCs) (LILRA4), natural killer (NK) cells gest that a highly proinflammatory macrophage microenvironment (KLRD1), T cells (CD3D), B cells (MS4A1), plasma cells (IGHG4) is present in the lungs of patients with severe COVID-19, which and epithelial cells (TPPP3, KRT18) (Fig. 1a), identified by signature is consistent with previous knowledge of macrophage populations genes (Extended Data Fig. 1a,b). Major cell types, including mDCs, during steady-state, inflammation and recovery7. pDCs, mast cells, NK cells, T cells and B cells, contained cells from Memory T cell responses have been observed in patients with most samples, whereas macrophages showed specific enrichment in SARS8 and are necessary for resolving SARS-CoV infection in mice9,10. different groups (Extended Data Fig. 1c–e). BALFs of patients with We identified six major clusters of T and NK lymphocytes (Extended severe/critical COVID-19 infection contained higher proportions Data Fig. 3a) based on expression of canonical genes (Extended of macrophages and neutrophils and lower proportions of mDCs, Data Fig. 3b,c). Lower CD8+ T cell and higher proliferating T cell pDCs and T cells than those with moderate infection (Fig. 1b and proportions were observed in patients with severe/critical infection Extended Data Fig. 1f). than in patients with moderate infection (Extended Data Fig. 3d,e). We re-clustered 20 macrophage subclusters to further dissect their These CD8+ T cells expressed higher levels of GZMA, GZMK and heterogeneity (Extended Data Fig. 2a). According to recent classi- FASLG and the tissue-residence markers ITGA1, CXCR6 and JAML fication criteria4 and typical markers (Extended Data Fig. 2b), the (Extended Data Fig. 3f), with upregulation of genes involved in acti- macrophages were grouped by FCN1, SPP1 and FABP4 expression vation, migration and cytokine-related pathways in moderate cases patterns including the FCN1hi group 1, FCN1loSPP1+ group 2, SPP1+ and translation initiation, cell homeostasis and nucleoside metabolic group 3 and FABP4+ group 4 macrophages (Extended Data Fig. 2c). pathways in severe cases (Extended Data Fig. 3g). 1Institute for Hepatology, National Clinical Research Center for Infectious Disease, Shenzhen Third People’s Hospital, Shenzhen, China. 2Department for Infectious Diseases, Shenzhen Third People’s Hospital, Shenzhen, China. 3The Second Affiliated Hospital, School of Medicine, Southern University of Science and Technology, Shenzhen, China. 4Department of Immunology, Weizmann Institute of Science, Rehovot, Israel. 5Shanghai Public Health Clinical Center, Fudan University, Shanghai, China. 6These authors contributed equally: Mingfeng Liao, Yang Liu, Jing Yuan. ✉e-mail: [email protected]; [email protected]; [email protected]; [email protected] 842 NATURE MEDICINE | VOL 26 | JUNE 2020 | 842–844 | www.nature.com/naturemedicine NATURE MEDICINE BRIEF COMMUNICATION a b Macrophages Neutrophil mDC pDC 0.37 0.14 0.04 0.94 0.0019 31 Epithelial 0.25 0.15 0.21 10 25 0.11 0.3 0.27 0.68 0.0022 13 (13,16,25,28,31) 1.0 0.06 0.010 28 16 0.2 23 0 16 0.8 0.04 Plasma (19) 0.005 Macrophages 1 17 2 18 0.6 0.1 0.02 5 (0,1,2,3,4,5,7,8,10, 19 3 19 Percentage 26 Mast (30) 5 11,12,18,21,22,23,26) 4 20 0.4 0 0 30 0 2 21 5 21 6 22 HC MS HC MS HC MS HC MS 18 22 7 23 –3 –3 –4 6 T Median: 0.861 0.622 0.71000.0010.032 0.028 0.0250.010 1 × 10 9 × 10 5 × 10 11 14 8 24 0 4 3 7 9 25 UMAP_2 (6,9,14) 24 10 26 T NK B Plasma 10 9 11 27 0.34 0.74 0.32 1 12 0.6 0.95 0.12 0 17 12 28 0.04 0.15 0.03 0.22 0.4 0.34 13 29 0.12 0.1 0.3 0.074 mDC (20) NK (17) 14 30 0.09 8 20 0.4 0.3 –5 15 15 31 0.02 27 B (27) 0.06 0.2 Neutrophil (15) 0.2 29 pDC (29) 0.03 0.01 0.1 Percentage 0 0 0 0 –5 0 5 10 HC MS HC MS HC MS HC MS UMAP_1 Median: 0.088 0.304 0.1270.012 0.0420.016 0.003 0.010 0.003 00.003 0.007 c HC MS d 4 7.5 3 FCN1 2 1 5.0 Expression 0 Group 1 6 2.5 Group 2 SPP1 4 UMAP_2 Group 3 2 0.0 Group 4 Expression 0 –2.5 6 4 FABP4 2 –4 04 –4 04 –4 04 Expression 0 UMAP_1 M1 M2 M3 S1 S2 S3 S4 S5 S6 HC1HC2 HC3 HC4 Fig. 1 | Dysregulated bronchoalveolar immune landscapes in patients with severe COVID-19 infection. a, UMAP presentation of major cell types and associated clusters in BALFs (n = 13). b, The proportions of the major BALF immune cell types among healthy controls (HC) and patients with moderate (M) and severe (S) COVID-19 infection (using two-sided Student’s t-test for pairwise comparisons). c, UMAP projection of four macrophage groups among controls (n = 4) and patients (moderate, n = 3; severe, n = 6). d, Expression of FCN1, SPP1 and FABP4 by BALF macrophages from each sample. We further assessed T cell clonal expansion using single-cell heterogeneous, whereas a larger proportion of CD8+ T cell effectors T cell receptor sequencing (scTCR-seq) of seven T cell clusters from with tissue-resident and highly expanded features were present in BALF from patients with COVID-19 (Fig. 2a and Extended Data BALFs from patients with moderate infection. Fig. 4a). TCR clonotype was obtainable from 66.9% (26.6–87.3%) of Finally, we measured several cytokines and chemokines in T cells in each patient and from 60.9% (45.2–84.2%) of cells in each BALF of patients with COVID-19 (Supplementary Tables 1 and 4). T cell cluster (Supplementary Table 2 and Extended Data Fig. 4b). Compared to patients with moderate COVID-19 infection, patients On the uniform manifold approximation and projection (UMAP), with severe/critical infection had much higher levels of inflam- projections of clonally expanded T cells showed an aggregative matory cytokines, particularly interleukin (IL)-8, IL-6 and IL-1β, distribution, indicating transcriptional homogeneity in patients in their BALFs (Extended Data Fig. 5a). Three patients (S1, S8 with moderate infection versus a diffusive distribution indicating and S9) with critical infection displayed persistently high levels transcriptional heterogeneity in patients with severe/critical infec- of cytokines.