Glycans on non-structural protein 1 prevent premature T-cell mediated dengue virus clearance

Abstract

Non-structural protein 1 (NS1) of dengue virus (DENV) harbours two conserved N-glycosylation sites at positions 130 and 207, whose biological roles have remained elusive. Using a clinically relevant mouse model of severe dengue, we showed that DENV that lacked N207 glycans on NS1 was significantly attenuated, and this phenotype was dominant over wild-type virulent DENV. Mice infected with this mutant exhibited accelerated viral clearance, milder lymphopenia and more functional DENV-specific CD8⁺ T cells. Bulk and single-cell RNA sequencing, cytokine measurements and immune-phenotyping revealed blunted innate inflammatory responses early post-infection, which correlated with reduced PD-L1 expression on innate immune cells and reduced PD-1⁺ T-cells in mice infected with de-glycosylated DENV. PD-1 blockade demonstrated the involvement of premature T-cell apoptosis through the PD-L1/PD-1 axis in DENV pathogenesis. Collectively, our findings support that N207-de-glycosylated NS1 inhibits early inflammatory responses, which restricts PD-L1 upregulation on innate immune cells, which in turn limits PD-L1/PD-1 mediated T-cell apoptosis. Our study uncovers a novel immune evasion strategy and identifies PD-L1/PD-1 as a novel mechanism of dengue immunopathogenesis.

Keywords: Dengue; Glycosylation; NS1; PD-1; T-cell Apoptosis.

Figure 1. In vitro and in vivo fitness of N130Q and T209L DENV mutants.

(A) Viral titers in culture supernatants. C6/36, BHK-21, and Vero cells were infected at MOI 0.1 with D2Y98P (WT), N130Q or T209L DENV mutants (n = 6; 3 technical replicates per biological replicate; 2 biological repeats). Viral titers in the culture supernatants were measured by plaque assay. The detection limit was set at 10² PFU/mL. (B) sNS1 concentration in culture supernatants was determined by ELISA (n = 6; 3 technical replicates per biological replicate; 2 biological repeats). (C–G) IFNAR^−/− mice were infected with 10exp6 PFU of each virus via the subcutaneous (sc.) route. (C–E) Survival, body weight profile and clinical scores (n = 10); 0—healthy, 1—ruffled fur, 2—hunched back, 3—lethargy, 4—limb paralysis, 5—mice displaying 30% weight loss (euthanasia). (F, G) Viral titers in blood (F) and perfused organs (G) were determined by plaque assay (n = 4–9). (H) Systemic sNS1 levels were measured by ELISA (n = 5). Data shown in panels (C) to (G) are a combination of 2 biological repeats. All graphs were expressed as mean ± SD. Data analysis were performed using Log-rank (Mantel–Cox) test (E) and two-way ANOVA Tukey’s multiple comparisons test (F–H). Source data are available online for this figure.

Figure 2. Structural and biochemical features of purified WT and T209L NS1 proteins.

(A) Simulated 3D structure of monomeric WT and T209L NS1 proteins using AlphaFold2 software. (B) Detection of multimeric T209L and WT sNS1 by native PAGE western blot. (C) Lipid composition of WT and T209L sNS1 proteins determined by mass spectrometry. (D) Representative snapshots from MD simulations of WT and NS1 mutants. Cartoon representation from side (left) and top (right) views. NS1 protein is shown in blue. Lipids are shown in liquorice representation (cyan-carbon, blue-nitrogen, red-oxygen) while glycans are shown as red spheres. The graphs show last 100 ns averaged number of contacts between lipids and glycans as well as NS1:glycan–lipid buried SASA. The buried solvent accessible surface area (SASA) between NS1:glycans and lipid cargo was calculated as a sum of NS1:glycans and lipids before subtracting (NS1:glycan)–lipid SASA. Data shown are representative of at least two independent technical repeats. Data were expressed as mean ± SD. Data analysis were performed using two-way ANOVA Sidak’s multiple comparisons test (C). Source data are available online for this figure.

Figure 3. In vivo sNS1 depletion.

(A) Experimental design of sNS1 depletion in infected IFNAR^−/− mice. (B) Systemic sNS1 concentrations measured by ELISA (n = 9). (C) Viremia titers measured by plaque assay (n = 10). (D) Body weight profile (n = 9). (E) Clinical scores as described in the Legend of Fig. 1 (n = 10). Data shown are a combination of two biological repeats. All graphs were expressed as mean ± SD. Data analysis were performed using two-way ANOVA Tukey’s multiple comparisons test (B, C). Source data are available online for this figure.

Figure 4. Exogenous administration of sNS1 and co-infection experiments.

(A–D) Exogenous administration of sNS1 in vivo. (A) Experimental design. (B) Systemic sNS1 concentrations measured by ELISA (n = 5 from one biological repeat). (C) Viremia titers measured by plaque assay (n = 10). (D) Clinical scores as described in the legend of Fig. 1 (n = 10). (E–I) Co-infection with WT DENV and T209L DENV. IFNAR^−/− mice were sc. infected with either WT or T209L mutant (10⁶ PFU per mouse) or were co-infected with WT and T209L NS1 mutant viruses (5 × 10⁵ PFU of each virus). (E) Experimental design. (F) Viremia titers were measured by plaque assay (n = 10). (G) Survival rate (n = 10), (H) Body weight profile (n = 10). (I) Clinical scores as described in the legend of Fig. 1 (n = 10). 0—healthy, 1—ruffled fur, 2—hunched back, 3— lethargy, 4—limb paralysis, 5—mice displaying 30% weight loss (euthanasia). Data shown in panels (C–I) are a combination of 2 biological repeats. All graphs were expressed as mean ± SD. Data analysis were performed using two-way ANOVA Tukey’s multiple comparisons test (C, F) and Log-rank (Mantel–Cox) test (G). Source data are available online for this figure.

Figure 5. Role of T cells during infection with T209L DENV infection.

(A) FACS analysis of naive and activated CD4⁺ and CD8⁺ T cell subsets in WT- or T209L DENV-infected IFNAR^−/− mice at day 5 p.i. (n = 8–9) (B) IFNγ-ELISPOT (n = 10 from two biological repeats) and granzyme-ELISPOT (n = 4 from one biological repeat) of splenic CD8⁺ T cells purified from infected mice at day 3 and 5 p.i. and restimulated with a DENV-specific CD8 immunodominant peptide. (C–G) CD8⁺ T cell depletion in IFNAR^−/− mice infected with T209L NS1 mutant. Data shown are a combination of two biological repeats. (C) Experimental design. (D) Survival rate (n = 10). (E) Body weight profile (n = 10). (F) Clinical scores as described in the legend of Fig. 1 (n = 10). (G) Viremia titers were determined by plaque assay (n = 10). All graphs were expressed as mean ± SD. Data analysis were performed using two-way ANOVA Tukey’s multiple comparisons test (A, G), nonparametric Multiple Mann–Whitney t test (B), and Log-rank (Mantel–Cox) test (G). Source data are available online for this figure.

Figure 6. Inflammatory responses in response to infection with WT or T209L DENV.

Bulk (A, B) and single cell (C, D) RNA sequencing on white blood cells harvested at day 3 p.i. from mice infected with WT DENV, T209L DENV or left uninfected (n = 3, one biological repeat). (A) Innate immune responses and inflammatory responses (Reactome). (B) Expression levels of genes related to neutrophil activation and degranulation. (C) Annotated UMAP. (D) TNF-α and STAT1 gene expression. (E) Systemic cytokine profile measured by ELISA in mice infected with WT or T209L DENV mutant at day 3 and 5 p.i. (n = 10 from two biological repeats). (F) THP-1 macrophages were infected with WT, T209L DENV or a mix of both (1:1). TNFα levels in culture supernatants were measured by ELISA (n = 6 from two biological repeats). All graphs were expressed as mean ± SD. Data analysis were performed using two-way ANOVA Tukey’s multiple comparisons test (E, F). Source data are available online for this figure.

Figure 7. PD-L1 expression on innate immune cells and PD-1 blocking experiment.

(A) Flow cytometric detection of PD-L1⁺ circulating and splenic total CD45⁺ cells, activated neutrophils, activated monocytes, activated macrophages and activated dendritic cells harvested at day 3 p.i. from WT-, T209L- and co-infected IFNAR^−/− mice (n = 4 from one biological repeat). Absolute counts and mean fluorescence intensity (MFI) of PD-L1 expression are shown. (B–G) PD-1 blockade experiment. (B) Experimental design. (C) Survival rate (n = 10). (D) Body weight profile (n = 8). (E) Clinical scores as described in the legend of Fig. 1 (n = 10). (F) Viremia titers were measured by plaque assay (n = 10). (G) CD4⁺ and CD8⁺ T cell counts in blood and spleen harvested at day 5 p.i. were analysed by flow cytometry (n = 9–10). Data shown in (C–G) are a combination of two biological repeats. All graphs were expressed as mean ± SD. Data analysis were performed using two-way ANOVA Tukey’s multiple comparisons test (A, F, G) and Log-rank (Mantel–Cox) test (C). Source data are available online for this figure.

Figure EV1. Plaque morphology of WT and de-glycosylated NS1 DENV mutants.

The viruses were amplified in C6/36 cells and plaqued on BHK-21 cells. Data shown are representative of at least 2 independent biological repeats.

Figure EV2. Infection profile of WT and T209L DENV in IFNAR-muMT mice.

(A) Body weight profile (n = 9). (B) Viremia titers were determined by plaque assay (n = 4–9). (C) Clinical scores as described in the legend of Fig. 1 (n = 9). Data shown are a combination of 2 independent repeats. All graphs were expressed as mean ± SD. Data analysis were performed using two-way ANOVA Sidak’s multiple comparisons test (B).

Figure EV3

Relative abundance of markers of immune cell clusters used for single cell RNAseq data analysis.

Figure EV4. PD-L1 (CD274) gene expression in neutrophils, monocytes and dendritic cells in DENV-infected mice.

The expression of PD-L1 gene in monocytes, neutrophils and dendritic cell subsets in WT- and T209L DENV-infected IFNAR^−/− mice was determined by single cell RNAseq.

Figure EV5. PD-1 expression on T cells.

Flow cytometry analysis of circulating and splenic T cells harvested at day 3 p.i. from WT- or T209L NS1 DENV-infected IFNAR^−/− mice (n = 5). Percentages of PD-1⁺ T cells and mean fluorescence intensity (MFI) of PD-1 expression on T cells. Data shown are from one biological repeat. All graphs were expressed as mean ± SD. Data analysis were performed using two-way ANOVA Tukey’s multiple comparisons test.