Single-cell RNA-sequencing highlights a curtailed NK cell function in convalescent COVID-19 pregnant women

Abstract

Introduction: During gestation the immune system undergoes dramatic remodelling to protect the maternal-fetal dyad from infections whilst also preventing fetal rejection. We investigated how SARS-CoV-2 modifies the immune landscape during infection and in recovered pregnant women.

Methods: We immunophenotyped our two independent geographical cohorts using a 14-colour flow cytometry panel (surface and intracellular staining). We estimated cytokines and SARS-CoV-2 IgG antibodies in validation cohort using a multiplexd flow cytometry panel. Single-cell RNA sequencing (scRNA-seq) was performed using a Chromium Single Cell 3' Gel Bead Chip and Library Kit from 10x Genomics (Drop-seq method). Furthermore, we estimated the cytotoxic functions of natural killer (NK) cells by flow cytometry using surface and intracellular staining.

Results: Using two independent geographical cohorts, we identified that NK cells had a sustained reduction during active infection and after recovery. Further, scRNA-seq data revealed that infection with SARS-CoV-2 rewired the gene expression profile of NK, monocytes, CD4⁺, CD8⁺ effector T cells and antibody producing B cells in convalescent pregnant women. Several gene pathways associated with cytotoxic function, interferon signalling type I & II, and pro- and anti-inflammatory functions in NK and CD8⁺ cytotoxic T cells were attenuated in recovered pregnant patients compared with healthy pregnancies. We validated our scRNA-seq of NK cells from convalescent pregnant women and confirmed that NK cells had diminished levels of cytotoxic proteins; perforin, CD122 and granzyme B.

Discussion: Overall, our study uncovers that SARS-CoV-2 infection deranges the adaptive immune response in pregnant women even after recovery and may contribute to post-COVID19 sequalae of symptoms.

Keywords: COVID-19; NK cells; PBMCs; ScRNA-seq; cytotoxic cells; immunophenotyping and cytokines; pregnancy.

Figure 1

Immunophenotyping of PBMCs from pregnant SARS-CoV-2 infected and recovered patients. (a) Malaysian Cohort 1: Description of PBMCs used for immunophenotyping collected from pregnant healthy controls (Preg-HC, blue), SARS-CoV-2 infected (Preg-INF, orange) and COVID-19 recovered (Preg-R, pink). (b) Unsupervised clustering of immune cells based on 14-color flow cytometry panel. Each PBMC sample from Preg-HC, Preg-INF, and Preg-R groups were concatenated bioinformatically using the FlowJo software. Two major clusters for lymphocytes (pink) and monocytes (blue) population were identified and presented on a UMAP plot. The combined cell population is shown in gray. (c) UMAP plot characterising 7 identified immune subsets; monocytes (cyan), CD4⁺ T cells (navy), CD8⁺ T cells (red), NKT cells (lilac), CD19⁺ B cells (green), NK cells (rose), and Tregs (yellow). (d) Overlay of UMAP to show the different immune cell landscape; Preg-HC (blue), Preg-INF (orange), and Preg-R (pink) patient groups. The combined cell population is shown in gray. (e) Delineation of all CD8⁺ T subsets. UMAP overlay showing the central memory (blue), early (red), intermediate (orange), late effector (bright green), and Naïve (dark green) CD8⁺ T cells. The combined cell population is shown in rose. (f) UMAP overlay depicting the various CD8⁺ T cells based on the different patient groups. Preg-HC (cyan), Preg-R (pink), and Preg-D (orange). The combined cell population is shown in gray. (g) Original FACS plots of cytotoxic CD8⁺ T cells. Identification of Naïve, memory, and effector memory CD8⁺ T cells based on CD45RA and CCR7 markers. FACS plots show the CD45RA⁺CCR7⁺ naïve, CD45RA^-CCR7⁺ CM cells, CD45RA^+highCCR7^- late EM, CD45RA^+midCCR7⁺ intermediate EM, CD45RA^-CCR7^- early EM cells in the different patient groups. (h) The percentage of Naïve (upper left), intermediate (EM II, right) and late EM III (bottom left) cells are shown as violin plots with each dot represents an individual sample. P-values show the significance among Preg-HC, Preg-INF and Preg-R groups and compared using Kruskal–Wallis test, adjusted for Dunn’s multiple comparisons test. P ≤0.05 considered significant (*p ≤0.05).

Figure 2

Increased effector CD4⁺ T cells and NKT cells in infected and recovered pregnant women. (a) UMAP analysis of total CD4⁺ T cells into different CD4⁺ T subsets including Naïve T cells (green), central memory T cells (CM; blue), early effector T cells (EMI; red), late effector cells (EMII; orange), and Tregs (yellow). The combined cell population is shown in gray. (b) UMAP for different CD4⁺ T subsets showing Preg-HC (cyan), Preg-R (pink), and Preg-INF (orange). The combined cell population is shown in gray. (c) Original FACS plots identifies CD4⁺ T cells subsets based on CCR7 and CD45RA markers. Classification of naïve, memory and effector memory CD4⁺ T cells based on CD45RA and CCR7 markers. FACS plots show the CD45RA⁺CCR7⁺ Naïve, CD45RA^-CCR7⁺ CM cells, CD45RA^+highCCR7^- late EM, CD45RA^+midCCR7⁺ intermediate EM, CD45RA^-CCR7^- early EM cells. (d) The percentage of Naïve, early, and late EM cells are shown in violin plots. P-values show the significance among Preg-HC, Preg-INF, and Preg-R groups and compared using Kruskal–Wallis test, adjusted for Dunn’s multiple comparisons test. P ≤0.05 considered significant (*p ≤0.05). (e) UMAP plots showing the distribution of NKT cells (blue) in the CD3⁺ T cell compartment. (f) UMAP displaying the overlay with different patient groups. Preg-HC (cyan), Preg-R (pink), and Preg-INF (orange) for NKT cells. Combined samples are shown in gray. (g) Original FACS plots presenting the expression of CD56 on CD3⁺CD8⁺ T cells with antibody markers staining for CD8 and CD56. (h) The percentage of NKT cells displayed by violin plots among Preg-HC, Preg-INF, and Preg-R groups. Kruskal–Wallis test, adjusted for Dunn’s multiple comparisons test was used to determine significance (*p ≤0.05).

Figure 3

Validation of immunophenotyping studies and dysregulated humoral immune response in German cohort 2. (a) German Cohort 2: description of cohort and experimental plan using matched PBMCs and serum from pregnant healthy controls (Preg-HC) and COVID-19 recovered (Preg-R) pregnant women. (b) Reduced percentage of CD14⁺ monocytes and significantly increased CD14⁺CD16⁺ in Preg-R patients. Box plots show median, interquartile range (IQR), and the whiskers corresponding to the highest and lowest points within 1.5 times of IQR. Each dot represents an individual sample. Wilcoxon rank-sum test was used for p-value significance to compare pregnant healthy control (Preg-HC) and COVID-19 recovered (Preg-R). P ≤0.05 considered significant (*p ≤0.05). (c) Box and whisker plot representing the percentage of Naïve, CM, EM I, and EM II CD8⁺ T cells in Preg-HC and Preg-R patients. CD8⁺ Naïve and CD8⁺ CM T cells were lower in Preg-R. CD8⁺ EM I and CD8⁺ EM II T cells were significantly increased Preg-R. Wilcoxon rank-sum test was used for p-value significance. P ≤0.05 considered significant (*p ≤0.05). (d) Box and whisker plot representing NKT cells in Preg-R compared with Preg-HC. (e) Box and whisker’s plot showing reduced CD56⁺ NK cells and CD56⁺CD16⁺ NK II cells in Preg-R. Increased levels of CD56^-HLA-DR⁺ lymphoid cells were significantly increased in Preg-R. Wilcoxon rank-sum test was used for p-value significance. P ≤0.05 considered significant (*p ≤0.05). (f) High levels of Nucleocapsid IgG antibody levels in serum of Preg-R group compared to Preg-HC (left graph). Wilcoxon rank-sum test was used for p-value significance. P ≤0.05 considered significant (*p ≤0.05). Spike S1 (middle graph) and RBD IgG concentration (ug/mL; right graph) at different collection time-points (CT) points (CT1, 0 day; CT2, 39 days; and CT3, 89 days post infection) during their pregnancy. Decreased levels of Spike S1 and RBD IgG antibodies 89 days post first collection time point in Preg-R group. Each dot represents an individual. Kruskal–Wallis test, adjusted for Dunn’s multiple comparisons test. P ≤0.05 considered significant (*p ≤0.05). (g) Examination of IL-10, MCP-1, and IL-8 levels in the serum of recovered women up to 89 days post infection. Each dot represents an individual on a box-whisker plot. CT1, 0 day; CT2, 39 days; and CT3, 89 days post infection-during their pregnancy Kruskal–Wallis test, adjusted for Dunn’s multiple comparisons test. P ≤0.05 considered significant (*p ≤0.05, **p ≤0.01, ***p ≤0.001, ****p ≤0.0001).

Figure 4

Single cell atlas and immune cell composition of healthy controls and recovered COVID-19 pregnant women. (a) Integrated UMAP (UMAP-CCA) of 30,394 cells derived from PBMCs. Sc-RNA-seq data were analyzed using Seurat pipeline. UMAP plots shown the different immune cell subsets based on distinct gene expression. Cell types are color-coded as shown in UMAP. (b) Dot plots show the expression level of canonical cell markers used to assign cell subset identification for major cell type including CD4⁺ T, CD8⁺ T, and B cells. (c) Feature plots show the key canonical markers used for identification of individual cell types. Green intensity shows increasing expression. (d) Percentage of major cell types in Preg-HC and Preg-R patients. (e) UMAP analysis of individual Preg-HC (blue) and Preg-R (orange) patient groups which highlight a reduced NK I cell population.

Figure 5

Activated classical (c)Monocytes in recovered pregnant patient group. (a) UMAP subplots for cMonocytes of Preg-HC and Preg-R patients. (b) Differentially expressed genes (DEGs) analysis in cMonocytes in Preg-R vs Preg-HC. Volcano plots showed the significantly upregulated and downregulated genes. (c) Dot plot shows the top 10 highly upregulated and downregulated genes in cMonocytes. (d) GSEA pathway analysis based on differentially expressed genes in cMonocytes obtained from Preg-R vs Preg-HC based on a bubble plot (left; activated and right; supressed). (e) GSEA plots show selected pathways (cytokine-mediated signaling pathway, interleukin-1 beta production, and response to type I interferon).

Figure 6

Reduced cytotoxic functions of NKT and NK cells in recovered pregnant women. (a) UMAP plots show the NKT (upper panel) and NK I (lower panel) cells from pregnant healthy controls and recovered patients. (b) Differential gene expression analysis of NKT, NK I, and CD8⁺ TEMRA cells using volcano plots. (c) GSEA pathway enrichment analysis for NK (I) Selected activated and suppressed pathways are shown on GSEA bubble plots. (d) KEGG pathway analysis of NK I cells. Most significantly pathways are shown on the GSEA plot. (e) Dot plot represents the selected cytotoxic function expressing genes in NKT (C3), NK I (C7), NK II (C8), and NK III (C18) cells. (f) Violin plots show significantly regulated genes related with cytotoxic functions in NK I cells. Each represents one cell. (g) Dot plot representing genes related with cytotoxic functions in CD8⁺ TEMRA (C2) cells.

Figure 7

Dysregulated Type I and type II IFNs signalling in recovered pregnant women. (a) Schema of Type I and II signaling pathways. Dot plot showing Type I IFNs signalling genes (IFNAR1, IFNAR2, TYK2, JAK1, STAT1, IRF-9, ISG20, and OASL) and their corresponding expression level. Type II IFNs signaling genes (IFNG, IL32, JAK2, TNF, and TGFBI). (b) GSEA plot for antiviral innate immune response. (c) Antiviral innate immune response and expressed genes in different immune cell subsets using dotplot.

Figure 8

Validation of NK cell functions molecules by flowcytometry. (a) FACS plots represents the expression of perforin, granzyme B, IL 10 and IFN-γ in CD3^-CD19^-CD56⁺NK cells in Preg-HC and Preg-R groups. (b) Box plots represent the statistical significance of perforin, granzyme B, IL 10 levels in NK cells. (c) FACS histogram plot presents the expression of CD122 on CD56⁺CD16⁺ classical NK cells. Preg-HC and Preg-R samples overlayed on each other for identification of CD122 expression for comparative purpose and expression was normalized to mode (left FACS plot). Box and Whisker plot represents the median expression of CD112 marker. Wilcoxon rank-sum test was used for p value significance. P ≤0.05 considered significant (*p ≤0.05).