Single-cell analysis of two severe COVID-19 patients reveals a monocyte-associated and tocilizumab-responding cytokine storm

Abstract

Several studies show that the immunosuppressive drugs targeting the interleukin-6 (IL-6) receptor, including tocilizumab, ameliorate lethal inflammatory responses in COVID-19 patients infected with SARS-CoV-2. Here, by employing single-cell analysis of the immune cell composition of two severe-stage COVID-19 patients prior to and following tocilizumab-induced remission, we identify a monocyte subpopulation that contributes to the inflammatory cytokine storms. Furthermore, although tocilizumab treatment attenuates the inflammation, immune cells, including plasma B cells and CD8⁺ T cells, still exhibit robust humoral and cellular antiviral immune responses. Thus, in addition to providing a high-dimensional dataset on the immune cell distribution at multiple stages of the COVID-19, our work also provides insights into the therapeutic effects of tocilizumab, and identifies potential target cell populations for treating COVID-19-related cytokine storms.

Fig. 1. An atlas of peripheral immune cells in severe COVID-19 patients.

a Flowchart depicting the overall design of the study. Blood draws from patient P1 were performed at 2 time points (day 1 and day 5) and from P2 at 3 time points (day 1, day 5 and day 7). P1 on day 1 and P2 on day 1 and day 5 were positive based on a nucleic acid test of a throat swab specimen. P1 on day 5 and P2 on day 7 were negative based on a nucleic acid test of a throat swab specimen. Patients on day 1 were at the severe stage, and patients were in the remission stage on day 5 (P1 and P2); the day 7 blood draw for P2 (still in the remission stage) was performed based on a positive nucleic acid test on day 5. Note that the samples on day 1 were collected within 12 hours of tocilizumab treatment. b-d UMAP representations of integrated single-cell transcriptomes of 69,237 PBMCs, with 13,239 cells derived from our COVID-19 patients and 55,998 derived from the 10X Genomics official website. Cells are colour-coded by clusters (b), disease state (c), and sample origin (d). Dotted circles represent cell types with a > 5% proportion within PBMCs in (b), and clusters significantly enriched in patients versus controls are shown in (c, d). Mono, monocyte; NK, natural killer cells; mDCs, myeloid dendritic cells; pDCs, plasmacytoid dendritic cells. e Violin plots of selected marker genes (upper row) for multiple cell subpopulations. The left column presents the cell subtypes identified based on combinations of marker genes.

Fig. 2. A unique monocyte subpopulation contributes to inflammatory storms in severe-stage COVID-19 patients.

a UMAP plot showing three clusters of CD14⁺ monocytes and 1 cluster of CD16⁺ monocytes. Cells are colour-coded by clusters. b Bar plot of the proportion of monocytes in cluster 9 at the severe and remission stages and in healthy control individuals. Source data are provided as a Source Data file. c Heatmap of differentially expressed genes (DEGs) in monocytes from the pairwise comparison between the severe-stage patients, remission-stage patients, and healthy control individuals. d UMAP plots showing the expression of selected cytokines in all monocyte clusters. e, f Box plots of the average expression of genes involved in the signalling pathways regulation of acute inflammatory response (e) and cell chemotaxis (f) in monocytes from the severe stage (n = 912 cells) and remission stage (n = 678 cells) and in healthy control individuals (n = 9719 cells). Centre line, median; box limits, upper and lower quartiles; whiskers, 1.5x interquartile range; points, outliers; P values were calculated using two-sided Wilcoxon rank-sum tests. Source data are provided as a Source Data file. g Heatmap of the area under the curve (AUC) scores of expression regulation by transcription factors (TFs), as estimated using SCENIC. The top-ranked TFs showing the highest difference in expression regulation estimates in monocytes from severe-stage COVID-19 patients are shown. h UMAP plots showing the expression of the ATF3, NFIL3, and HIVEP2 genes in monocytes (top) and the AUC of the estimated regulon activity of the corresponding TFs, predicting the degree of expression regulation of their target genes (bottom).

Fig. 3. The monocyte-centric molecular interactions of peripheral immune cells in severe-stage COVID-19 patients.

a Dot plot of the predicted interactions between monocytes and the indicated immune cell types in the severe and remission stages and in healthy control individuals. P values are indicated by the circle sizes, as shown in the scale on the right (permutation test). The means of the average expression level of interacting molecule 1 in cluster 1 and interacting molecule 2 in cluster 2 are indicated by the colour. Assays were carried out at the mRNA level but were extrapolated to protein interactions. b Summary illustration depicting the potential cytokine/receptor interactions between monocytes and other types of peripheral immune cells in the severe and remission stages and in healthy control individuals. Bolded lines indicate predicted enriched cytokine/receptor interactions between monocytes and other immune cell types.

Fig. 4. Enhanced humoral and cell-mediated immunity in severe COVID-19 patients.

a UMAP representations of B cell and plasma B cell clusters from the severe and remission stages and in healthy control individuals. b Bar plot of the proportions of plasma B cells in the B cell lineage from the severe and remission stages and in healthy control individuals. Source data are provided as a Source Data file. c UMAP representations of CD8⁺ T cell subtypes (left) and the distribution of cells from the severe and remission stages and in healthy control individuals in each subtype (right). d Dot plot of the expression of the CCR7, PRDM1, and MKI67 genes in all CD8⁺ T cell subtypes. e Heatmap of differentially expressed genes in effector CD8⁺ T cells based on pairwise comparisons between severe-stage patients, remission-stage patients, and healthy control individuals. f, g, Bar plots of GO terms enriched in effector CD8⁺ T cells from the severe stage (f) or the severe and remission stages (g). P values were calculated using a hypergeometric test and the Benjamini–Hochberg correction algorithm (i.e. multi-test adjustments) in Metascape. h, i, Box plots of the average expression of genes involved in the signalling pathways associated with cell chemotaxis (h) and regulation of cell killing (i) in effector CD8⁺ T cells from the severe stage (n = 1128 cells) and remission stage (n = 1964 cells) and in healthy control individuals (n = 1715 cells). Centre line, median; box limits, upper and lower quartiles; whiskers, 1.5x interquartile range; points, outliers; P values were calculated using two-sided Wilcoxon rank-sum tests. Source data are provided as a Source Data file.