Early NK-cell and T-cell dysfunction marks progression to severe dengue in patients with obesity and healthy weight

Abstract

Dengue is a mosquito-borne virus infection affecting half of the world's population for which therapies are lacking. The role of T and NK-cells in protection/immunopathogenesis remains unclear for dengue. We performed a longitudinal phenotypic, functional and transcriptional analyses of T and NK-cells in 124 dengue patients using flow cytometry and single-cell RNA-sequencing. We show that T/NK-cell signatures early in infection discriminate patients who develop severe dengue (SD) from those who do not. These signatures are exacerbated in patients with overweight/obesity compared to healthy weight patients, supporting their increased susceptibility to SD. In SD, CD4⁺/CD8⁺ T-cells and NK-cells display increased co-inhibitory receptor expression and decreased cytotoxic potential compared to non-SD. Using transcriptional and proteomics approaches we show decreased type-I Interferon responses in SD, suggesting defective innate immunity may underlie NK/T-cell dysfunction. We propose that dysfunctional T and NK-cell signatures underpin dengue pathogenesis and may represent novel targets for immunomodulatory therapy in dengue.

Fig. 1. Distinct T-cell and NK-cell profiles in SD.

a Study design showing the two time points (TP) of disease at the indicated time of illness onset (TP1: day 3–5; TP2: day 6–9 shown with a grey and red circle, respectively) and the typical course of a dengue infection with blood viraemia declining around day 5 and severe disease manifestations starting around day 4–5 of illness onset. Created in BioRender. Gregorova, M. (2025) https://BioRender.com/xezlm7c. Flow cytometry analyses of T-cells and NK-cells shown here were performed at TP1 and TP2. b Frequency of immune subsets at TP1 [in grey, N = 42: non-SD (N = 31); SD (N = 11); HW (N = 21); OW/OB (N = 21)] and TP2 [in red, N = 104: non-SD (N = 74); SD (N = 27); HW (N = 49); OW/OB (N = 52)]. The middle line in each box represents the median with IQR. c Log₂ ratio of Median Fluorescence Intensity (MFI) values of selected markers in total NK, CD8⁺ and CD4⁺ T-cell subsets between TP1 (N = 41) and TP2 (N = 41) in non-SD (N = 30) and SD (N = 11) patients. d–i Log₂ ratio of mean abundances/PD-1⁺ MFI of cell subsets between SD and non-SD patients at TP1 (N = 42) (d–f) and TP2 (N = 104) (g–i) in CD8⁺ T, CD4⁺ T-cells, and NK-cells. Red/blue bars indicate significance (red bars: p.adj. < 0.05; blue bars: *p.adj < 0.05; **p.adj < 0.01; ***p.adj < 0.001) calculated using Wilcoxon rank-sum tests. j–m Linear discriminant analysis (LDA) of T-cell and NK-cell flow cytometry data is shown at TP1 (j, l; N = 42) and TP2 (k, m; N = 84). Data points represent individual patients, and symbols/colours indicate disease severity and BMI groups as follows. j, k Dengue (D), dengue with warning signs (DWS), and severe dengue (SD) are indicated in blue, orange and red circles, respectively. l–m: Healthy weight (HW), overweight (OW), and obesity (OB) are indicated in green, yellow and orange circles, respectively. Ellipses represent 95% confidence intervals. LD1 and LD2 were derived using all features shown in d–i. Source data are provided as a Source Data file.

Fig. 2. Increased PD-1⁺ CD4⁺ T-cells in SD.

a Percentage of CD4⁺ T-cells co-expressing CD38 and HLA-DR. b PD-1 expression in CD4⁺ T-cells expressed as Median Fluorescence Intensity (MFI). c, d Percentage of CD4⁺ T-cells co-expressing Ki-67 and CLA (c) and Ki-67 and GzmB (d). e, f Percentages of CD38⁺ HLA-DR⁺ and Ki67⁺ CLA⁺ CD4⁺ T-cells and PD-1 MFI levels in CD4⁺ T-cells at TP1 (e) and TP2 (f). a–f TP1, N = 42; TP2: N = 104. PD-1 staining is shown for two representative patients with non-severe dengue (non-SD) or severe dengue (SD). Data for patients with non-severe dengue (non-SD), severe dengue (SD), healthy weight (HW) and overweight/obesity (OW/OB) are shown in blue, red, green and orange circles, respectively. g Correlation of CD4⁺ T-cell subsets with clinical parameters/biomarkers at TP2 (N = 104), using Spearman’s rank correlation test with FDR correction. h UMAP plots with FlowSOM clusters (1, 6, 7, 9–11) visualised in non-SD and SD patient groups at TP1 (N = 42) and TP2 (N = 84) and expression levels of PD-1, CD38, CD69, and GzmB. i Stacked bar chart showing the frequency (y-axis) of each cluster in the patient groups with the bubble graph representing MFI levels (colour scale) and cell frequencies (size). The clusters are indicated on the x-axis; arrows highlight selected clusters on the top of the graph. j Frequency of the FlowSOM clusters showed in (h, i) (1, 6, 7, 9–11) at TP1 (N = 42) and TP2 (N = 84) within non-SD and SD patient groups. The middle line in each box represents the median with IQR. Error bars represent max/min value ± 1.5*IQR. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 calculated by one-way ANOVA with Benjamini–Hochberg correction. Source data are provided as a Source Data file.

Fig. 3. Altered PD-1 and GzmB expression in CD8⁺ T-cells of SD patients.

a–e Percentage of CD8⁺ T-cells expressing CD38 and HLA-DR (a), GzmB (b) and Ki-67 and CLA (c). PD-1 expression levels in CD8⁺ T-cells expressed as Median Fluorescence Intensity (MFI) (d). e Expression levels of GzmB (MFI) in PD-1⁺ CD8⁺ T-cells. Data is shown at TP1 (N = 42) and TP2 (N = 104). f, g Percentages of CD38⁺ HLA-DR⁺ and GzmB⁺ CD8⁺ T-cells and PD-1 MFI levels in CD8⁺ T-cells at TP1 (f; N = 42) and TP2 (g; N = 104). PD-1 staining is shown for two representative patients with non-SD or SD. Data for patients with non-severe dengue (non-SD), severe dengue (SD), healthy weight (HW) and overweight/obesity (OW/OB) are shown in blue, red, green and orange circles, respectively. h Correlation of CD8⁺ T-cell subsets with clinical parameters/biomarkers at TP2 (N = 104), using Spearman’s rank correlation test with Benjamini–Hochberg correction. i, j Single correlations of CD8⁺ T-cells expressing PD-1 with plasma levels of angiopoietin-2 (i) and VCAM-1(j) at TP1 (N = 42) and TP2 (N = 104). k UMAP plots with FlowSOM clusters (4, 6, 7, 10, 11, 14, 16) visualised in non-SD and SD patient groups at TP1 (N = 42) and TP2 (N = 84) and expression levels of PD-1, CD69, GzmB, and perforin. l Stacked bar chart showing the frequency (y-axis) of each cluster in the patient groups, with the bubble graph representing MFI levels (colour scale) and cell frequencies (size). The clusters are indicated on the x-axis; arrows highlight selected clusters on the top of the graph. m Frequency of the FlowSOM clusters shown in (k, l) (4, 6, 7, 10, 11, 14, 16) at TP1 (N = 42) and TP2 (N = 84) within non-SD and SD patient groups. The middle line in each box represents the median with IQR. Error bars represent max/min value ± 1.5*IQR. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.00001 by one-way ANOVA with Benjamini–Hochberg correction. Source data are provided as a Source Data file.

Fig. 4. Altered DENV NS3-specific T-cells in SD.

a Representative flow cytometry plots showing IFN-γ, TNF-α, MIP-1β, and CD107a production by CD4⁺ and CD8⁺ T-cells from a non-SD patient (day 6 of illness onset) after stimulation with NS3 DENV2 peptides, PMA/ionomycin, or unstimulated (DMSO). All data shown here were obtained in PBMCs of patients with DENV2 infection. b CD4⁺ and CD8⁺ T-cell responses shown as log10 of cytokine⁺ (IFN-γ and/or TNF-α and/or IL-2 and/or MIP-1β) and/or CD107a ⁺ cells at TP1 (N = 24) and TP2 (N = 37) after stimulation with NS3 DENV2 peptides. c CD4⁺ and CD8⁺ T-cell responses shown as cytokine⁺ and/or CD107a⁺ at TP2 (N = 37) after stimulation with NS3 DENV1–4 peptide pools. Data for patients with non-severe dengue (non-SD) and severe dengue (SD) are shown in blue and red circles, respectively. d, e Production of each cytokine and CD107a is shown for CD4⁺ (d) and CD8⁺ (e) T-cells in HW or OW/OB patients with non-SD and SD at TP1 (N = 24) and TP2 (N = 37) following NS3 DENV2 peptide stimulation. Data for patients with healthy weight (HW) and overweight/obesity (OW/OB) are shown in green and orange circles, respectively. f Cytokine⁺ and/or CD107a⁺ T-cells were divided into three groups based on their properties: degranulation only (CD107a), degranulation and cytokine production (IFN-γ and/or TNF-α and/or CD107a), and cytokine production only (IFN-γ and/or TNF-α). Pie charts show the average percentages of the cytokine⁺ and/or CD107a⁺ cells in total responding T-cells from non-SD and SD patients. g Representative histogram and boxplots showing PD-1 expression in DENV2-specific CD4⁺ and CD8⁺ T-cells at TP1 (N = 24) and TP2 (N = 37). h Gating strategy and representative flow cytometry plots of CD95 and PD-1 expressing T-cells. CD95 MFI levels in PD-1⁺ and PD-1⁻ CD4⁺ and CD8⁺ T-cells are shown for SD patients (TP1, N = 8; TP2, N = 7). The middle line in each box represents the median with IQR. Error bars represent max/min value ± 1.5*IQR. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 by one-way ANOVA with Benjamini–Hochberg correction. Source data are provided as a Source Data file.

Fig. 5. Elevated T-cell co-inhibitory receptors in SD.

a, b Percentage of co-inhibitory expression in CD4⁺ (a) and CD8⁺ (b) T-cells at TP1 in N = 10 patients with non-severe (non-SD; blue circles) or severe dengue (SD; red circles). c, d Correlation of the indicated markers with frequencies of TIM-3⁺ CD4⁺ T-cells (c) and LAG-3⁺ CD8⁺ T-cells (d). e Pie charts showing the number of co-inhibitory receptors simultaneously expressed by CD4⁺ and CD8⁺ T-cells in non-SD and SD patients at TP1 (N = 11). The different shades of grey represent the expression of 1 to 5 co-inhibitory receptors, and the outer arcs indicate the specific co-inhibitory receptors expressed as defined by Boolean gating. f, g UMAP plots showing selected FlowSOM clusters of CD4⁺ (f) and CD8⁺ T-cells (g) at TP1 in non-SD and SD patients (N = 11). The expression level of each marker is shown in heatmap (Z-score) and UMAP plots; the frequency of each cluster is shown for non-SD and SD patients. h Expression levels of the indicated metabolic markers are shown as Median Fluorescence Intensity (MFI) in PD-1⁺ CD4⁺ and CD8⁺ T-cells of patients with healthy weight (HW; green circles) and overweight/obesity (OW/OB; orange circles) experiencing non-SD and SD (day 5–8); N = 15. i Expression levels of the indicated metabolic markers shown as MFI in PD-1⁺ and PD-1⁻ activated (HLA-DR⁺ CD38⁺) CD4⁺ and CD8⁺ T-cells from non-SD and SD patients (N = 17). j Percentages of PD-1⁺ and PD-L1⁺ CD8⁺ T-cells in patient PBMCs prior to anti-PD-1/PDL-1 blockade (N = 9). k Frequency of GzmB⁺ perforin⁺ CD8⁺ T-cells after stimulation with NS3 DENV2 peptides in the presence of anti-PD-1/PDL-1 blocking antibodies or isotype controls. l Pie charts showing the number of functions simultaneously expressed by CD8⁺ T-cells following NS3 DENV2 peptide stimulation with or without blocking antibodies. The different shades of grey represent the expression of 1–4 functions, the outer arcs indicate the specific functions as defined by Boolean gating. The middle line in each box represents the median with IQR. Error bars represent max/min value ± 1.5*IQR. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 calculated by one-way ANOVA with Benjamini–Hochberg correction. Source data are provided as a Source Data file.

Fig. 6. Impaired NK-cell responses in SD.

a Gating strategy and representative flow cytometry plots of CD57 and NKG2C staining in non-severe (non-SD) and severe dengue (SD) patients at TP1. b Frequency of Ki-67⁺ NK-cells with differential expression of CD57 and NKG2C in non-SD (N = 11) and SD (N = 12) patients. c, d Expression of activating (c) and inhibitory (d) receptors in Ki-67⁺ NK-cells from non-SD and SD patients (c N = 23; d N = 15). e Correlation of NK-cell markers with clinical parameters/biomarkers using Spearman’s rank correlation test with Benjamini–Hochberg correction. f–h UMAP and phenograph cluster analyses of NK-cells in non-SD and SD patients, with clusters 1, 6, 8 and 10 highlighted in colour (f). The expression levels of each marker are shown in the heatmap (z-score) and UMAP (g), and the frequency of each cluster is shown in non-SD and SD patients (h). f–h N = 23 (non-SD, N = 11; SD, N = 12). i–k UMAP and phenograph cluster analyses of NK-cells in non-SD and SD patients, with clusters 1, 3 and 8 highlighted in colour (i). The expression levels of each marker are shown in the heatmap (z-score) and UMAP (j), and the frequency of each cluster is shown in non-SD and SD patients (k); i, k N = 15 (non-SD, N = 6; SD, N = 9). l, m Linear discriminant analysis at TP1 (N = 23); LD1 and LD2 were derived using all features shown in (b–d, h, k). Data points represent individual patients with dengue (D; blue circles), dengue with warning signs (DWS; orange circles), and SD (red circles) (l), and patients with healthy weight (HW; green circles), overweight (OW; yellow circles), and obesity (OB; orange circles). m Ellipses represent 95% confidence intervals. n–q NK-cell expression of cytokines (IFN-γ, TNF-α, MIP-1β) and CD107a after stimulation of PBMCs with K562, IL-12 + IL-18, or both (N = 19). Data are shown as fold change from unstimulated PBMCs. The middle line in each box represents the median with IQR. Error bars represent max/min value ± 1.5*IQR, *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001 calculated by Wilcoxon test or one-way ANOVA with Benjamini–Hochberg correction. Source data are provided as a Source Data file.

Fig. 7. Impaired type-I IFN responses in SD.

ScRNA-seq (BD Rhapsody) analyses of PBMCs from 24 patients with non-severe dengue (non-SD; N = 12) and severe dengue (SD; N = 12) are shown in (a–c). a UMAP showing the manually gated immune cells coloured by cell types. b Gene Ontology (GO) Over-representation analysis of genes significantly downregulated (all cell types combined) in SD vs non-SD patients. P-values were adjusted for multiple comparisons using the Benjamini–Hochberg method. c Type-I IFN signalling module scores are shown for the different cell types. Single-cell scores were generated using genes from the “type I interferon-mediated signalling pathway” GO term (GO:0060337). Scores were summed for cells from an individual patient and cell type. Box plots show the median (line) and interquartile range (box). The lower and upper whiskers extend to the smallest and largest values within 1.5 times the interquartile range from the first and third quartiles, respectively. Two-sided Wilcoxon rank-sum test with Benjamini–Hochberg correction for multiple comparisons. *p < 0.05, **p < 0.01. d–f TMT mass spectrometry proteomics analyses is shown for PBMCs from 11 patients with non-SD and SD (N = 6 and 5, respectively). Group comparisons were performed with two-sided Welch’s t-test (unequal variances) between non-SD and SD patients. The p-values were adjusted by the permutation-based FDR procedure implemented in Perseus v2.0.7.0 (default setting). d Volcano plot showing the differentially expressed proteins in SD versus non-SD patients. Highlighted are interferon-stimulated gene (ISG) products with adjusted p < 0.05, log2FC < −1 (2-fold decrease). e Heatmap displaying the 54 ISG products that are differentially expressed (adjusted p < 0.05) in SD versus non-SD patients. f GO over-representation analyses of significantly downregulated proteins [adjusted p < 0.05, log2FC < −0.58 (1.5-fold decrease)] in SD vs non-SD patients. In (b and f), Significant non-redundant gene ontology terms and associated Benjamini–Hochberg adjusted p-values are shown. Count = number of differentially expressed genes (DEGs)/proteins associated with each term. GeneRatio = fraction of DEGs in the gene set. Source data are provided as a Source Data file.