Recapitulating dengue virus infection with human pluripotent stem cell-derived liver organoids for antiviral screening

Abstract

Dengue virus (DENV) poses a major global health threat, affecting an estimated 100 to 400 million people annually. The infection and pathogenesis remain incompletely understood, and no antiviral drug is currently approved for DENV treatment. Here, we develop a human pluripotent stem cell (hPSC)-derived liver organoid (hPLO) model to characterize DENV infection and screen for antivirals. The hPLOs, containing various liver cell types, are highly susceptible to DENV-2 infection, resulting in severe cell death and morphological changes that mimic the pathology observed in severe dengue cases. Single-cell RNA sequencing (scRNA-seq) of DENV-2 infected hPLOs reveals that proliferating hepatocyte-like cells are the primary target cells of DENV-2, with significant mitochondrial damage and alterations in cell-type composition. Further drug screening in hPLOs identifies oxyresveratrol (Oresveratrol, ORES) and omaveloxolone (RTA 408) as potent anti-DENV candidates. These compounds enhance resistance to DENV-2 infection by activating the NRF2 pathway, reducing oxidative stress, and preserving mitochondrial function. The efficacy of ORES and RTA 408 is further validated in the established AG6 mouse model. Our study not only establishes hPLOs as a valuable platform for studying DENV infection and pathogenesis, but also highlights the vital role of NRF2-mediated mitochondrial function for antiviral development.

Fig. 1. Generation and characterization of hPLOs.

a Schematic illustration of the experimental workflow for generating hPLOs. Created with MedPeer (medpeer.cn). b Representative bright-field images of hPLO cultures from day 12 to day 18, Scale bars, 200 μm. This was independently repeated six times with similar results. c Representative bright-field images demonstrating hPLOs morphology on days 25, 50, and 110. Scale bars, 200 μm. This was independently repeated six times with similar results. d and e Representative whole-mount immunofluorescence images of Day 25 hPLOs showing hepatocyte markers (AFP, ALB, HNF4A) (d), cholangiocyte markers (KRT7, CK19), and the proliferation marker KI67 (e). Scale bars, 50 μm. This was independently repeated three times with similar results. f Representative immunofluorescence staining of Day 50 hPLOs showing hepatic markers (TF, HNF4A, AFP, ALB). Scale bars, 100 μm. This was independently repeated three times with similar results. g–k Heatmap showing differentially expressed genes across differentiation stages. Source data are provided as a Source Data file. l Representative whole-mount immunofluorescence images showing markers for hepatic stellate cells (PDGFRB) in Day 25 hPLOs. Scale bars, 100 μm. This was independently repeated three times with similar results. m Representative whole-mount immunofluorescence images displaying hepatic markers in hPLOs derived from different hPSC lineages (H1, H9). Scale bars, 50 μm. This was independently repeated three times with similar results. n UMAP visualization of scRNA-seq data from the hPLOs. Proli Hepatocyte-like represents the proliferating hepatocyte-like cells. o Dot plot showing the expression of various cell-type markers (hepatic, stellate, cholangiocyte, and proliferation) in hPLOs. Proli Hep represents the proliferating hepatocyte-like cells; Hep represents the hepatocyte-like cells; Cho represents the cholangiocyte-like cells; Ste represents the hepatic stellate-like cells. The size of each circle represents the percentage of cells expressing each gene, while color intensity reflects average gene expression levels.

Fig. 2. Functionality characterization of hPLOs.

a Representative whole-mount immunofluorescence images showing the tight junction marker ZO-1 in Day 25 hPLOs. Scale bars, 50 μm. This was independently repeated three times with similar results. b Representative TEM images visualizing the ultrastructure of a bile duct in Day 25 hPLOs. Scale bars, 1 μm. This was independently repeated three times with similar results. c–f Representative images showing the functional characterization of hPLOs: CDFDA staining in Day 25 hPLOs (c), Scale bars, 50 μm; Rhodamine 123 transport assay in Day 25 hPLOs (d), Scale bars, 200 μm; ICG uptake and release in Day 50 hPLOs (e), Scale bars, 200 μm; PAS staining in Day 25 and Day 50 hPLOs (f), Scale bars, 50 μm. These experiments were independently repeated three times with similar results. g–i Quantitative analysis of human albumin (g), urea secretion (h) and AAT (i) in the cell culture supernatants, analyzed by ELISA, n = 3. Values and error bars reflect mean ± SD. Values for primary human hepatocytes (PHHs) were obtained from previously published studies. Source data are provided as a Source Data file. j and k Assessment of CYP3A4 and CYP2C9 activity of hPLOs by fluorescence-based assays, n = 3. Values and error bars reflect mean ± SD. Values for primary human hepatocytes (PHHs) were obtained from previously published studies. Source data are provided as a Source Data file.

Fig. 3. hPLOs support productive DENV-2 infection.

a Schematic overview of the experimental design showing DENV-2 infection of hPLO fragments (1 PFU per fragment). Created with MedPeer (medpeer.cn). b Morphology of DENV-2 and MOCK infected hPLOs (Day 25) on days 0, 2, 4, 6 and 8. Scale bars, 500 μm. MOCK indicates inactivated DENV-2. c Quantification of healthy (viable) hPLOs (Day 25) in both infected and MOCK groups. Right panel: Representative image showing the healthy (blue arrows) and unhealthy (red arrows) hPLOs. MOCK indicates inactivated DENV-2. n = 3. Values and error bars reflect mean ± SD. Source data are provided as a Source Data file. d Representative Live/Dead staining images of DENV-2 and MOCK infected hPLOs (Day 25) on day 4 post inoculation. Scale bars, 100 μm. This was independently repeated three times with similar results. e Quantification of viral RNA in the culture medium of hPLOs. n = 3 independent biological replicates. Values and error bars reflect mean ± SD. Source data are provided as a Source Data file. f Quantification of viral titers in culture supernatants. n = 3 independent biological replicates. Values and error bars reflect mean ± SD. Source data are provided as a Source Data file. g–j Whole-mount immunofluorescence images of MOCK and DENV-2 groups indicating the DENV-2 marker (g), Scale bars, 100 μm; hepatocyte and DENV-2 markers, cholangiocyte and DENV-2 markers, proliferation and DENV-2 markers, (h–j), Scale bars, 50 μm. These experiments were independently repeated three times with similar results.

Fig. 4. Single-cell transcriptome profiles of DENV-2 infected hPLOs.

a, b UMAP visualization of scRNA-seq data from MOCK and DENV-2 infected hPLOs. c Density plots showing the distribution of cell compositions in MOCK and DENV-2 infected hPLOs on days 1, 2, and 8 post infection. High relative cell density is shown as light red. d Quantification of the percentage of each cell type in MOCK and DENV-2 infected hPLOs. e–g, Trend plot of the percentage of each cell type in MOCK and DENV-2 infected hPLOs on days 1, 2, and 8 post infection. h UMAP visualization of DENV-2 distribution in DENV-2 infected hPLOs. i Expression levels of DENV-2 RNA in all cell types at the indicated time points. j Quantification of the percentage of each cell type infected with DENV-2 on days 1, 2, and 8 post infection.

Fig. 5. DENV-2 pathogenesis in hPLOs.

a Scatter plot showing DEGs in hepatocyte-like cells in DENV-2 infected hPLOs. b, c Volcano plots showing DEGs in proliferating hepatocyte-like cells (b), cholangiocyte-like cells (c) in DENV-2 infected hPLOs compared to the MOCK group on day 8. Statistical significance was assessed using the Likelihood-ratio test, with thresholds set at adjusted p < 0.01, log₂FC ≥ 0.26, and gene expression detected in ≥10% of cells in at least one group. d GO enrichment analysis of signaling pathways in DENV-2 infected proliferating hepatocyte-like cells compared with the MOCK group on day 8. Red bars: functions enriched by upregulated genes; Green bars: functions enriched by downregulated genes. Statistical significance was assessed using the hypergeometric test. Color intensity reflects enrichment p-values. e and f, Network plot showing the functional association of DEGs in cholangiocyte-like cells (e) and proliferating hepatocyte-like cells (f). Network nodes are colored by gene module category. g–l Box plots showing expression levels of Inflammatory response (g), Interferon alpha response (h), Interferon gamma response (i), Mitochondrial fragmentation involved in apoptotic process (j), Apoptotic mitochondrial changes (k), Mitochondrial electron transport cytochrome c to oxygen (l) scores across cell types and time points. Time points are indicated by different colors. Horizontal lines represent median values, with whiskers extending to the farthest data point within a maximum of 1.5× interquartile range. n = 6 independent biological replicates per group. Two-sided Dunn’s (Bonferroni) test was used for analysis, and a p value  <  0.01 was considered significant. *p value < 0.01; **p value < 0.001; ***p value < 0.0001, ns indicates no statistical significance. The exact p values are provided in the Source Data file.

Fig. 6. Screening and evaluating the efficacy of anti-DENV drugs using hPLOs.

a Schematic overview of the experimental protocol for DENV-2 infection and drug screening of hPLOs. Created with MedPeer (medpeer.cn). b Representative bright-field images of MOCK-, infected- and drug-treated (1 μM 7DMA and 1 μM JNJ1802) hPLOs. Scale bars, 500 μm. c Bar chart showing the percentage of healthy hPLOs in MOCK-, infected- and drug-treated groups. MOCK indicates inactivated DENV-2 group. n = 3. Values and error bars reflect mean ± SD. Statistical significance was assessed using an unpaired two-tailed t-test. The exact p values: p = 0.0093 (DMSO vs 7DMA), p = 0.0150 (DMSO vs JNJ1802), p = 0.0004 (MOCK vs 7DMA), p = 0.0030 (MOCK vs JNJ1802). Source data are provided as a Source Data file. d The molecular structure of ORES. e Representative bright-field image of 10 μM ORES-treated hPLOs. Scale bars, 500 μm. f Bar chart showing the percentage of healthy hPLOs in MOCK-, infected- and 10 μM ORES-treated groups. MOCK indicates inactivated DENV-2 group. n = 3. Values and error bars reflect mean ± SD. Statistical significance was assessed using an unpaired two-tailed t-test. The exact p values: p = 0.0487. Source data are provided as a Source Data file. g, h Quantification of viral RNA in hPLOs (g), and in culture medium (h) treated with 10 μM ORES. n = 3. Values and error bars reflect mean ± SD. Statistical significance was assessed using an unpaired two-tailed t-test. The exact p values: p < 0.0001. Source data are provided as a Source Data file. i The molecular structure of RTA 408. j Representative bright-field image of 0.2 μM RTA 408-treated hPLOs. Scale bars, 500 μm. k Bar chart showing the percentage of healthy hPLOs in MOCK-, infected- and 0.2 μM RTA 408-treated groups. MOCK indicates inactivated DENV−2 group. n = 3. Values and error bars reflect mean ± SD. Statistical significance was assessed using an unpaired two-tailed t-test. The exact p values: p = 0.0038. Source data are provided as a Source Data file. l, m Quantification of viral RNA in hPLOs (l), and in culture medium (m) treated with 0.2 μM RTA 408. n = 3. Values and error bars reflect mean ± SD. Statistical significance was assessed using an unpaired two-tailed t-test. The exact p values: p = 0.0026 (l), p = 0.0001 (m). Source data are provided as a Source Data file. n Heatmap showing the expression of Nrf2 downstream target genes in MOCK-, infected- and drug-treated hPLOs. n = 3. Source data are provided as a Source Data file. o Glycogen storage was shown by PAS staining in the MOCK-, infected-, ORES-, and RTA 408- treated hPLOs. MOCK indicates inactivated DENV-2 group. Scale bars, 100 μm. p, q Assessment of CYP3A4, LDH activity of MOCK-, infected- and drug-treated groups by fluorescence-based assays, n = 3. Values and error bars reflect mean ± SD. Statistical significance was assessed using an unpaired two-tailed t-test. The exact p values: p = 0.0116 (DENV-2 vs ORES, CYP3A4), p = 0.0224 (DENV-2 vs RTA 408, CYP3A4); p = 0.0027 (DENV-2 vs ORES, LDH), p = 0.0058 (DENV-2 vs RTA 408, LDH). Source data are provided as a Source Data file. r, s Quantitative analysis of human ALB (r), urea secretion (s), in cell culture supernatants as analyzed by ELISA, n = 3. Values and error bars reflect mean ± SD. Statistical significance was assessed using an unpaired two-tailed t-test. The exact p values: p = 0.0201 (DENV-2 vs ORES, ALB), p = 0.0451 (DENV-2 vs RTA 408, ALB); p = 0.0051 (DENV-2 vs ORES, Urea), p = 0.0456 (DENV-2 vs RTA 408, Urea). Source data are provided as a Source Data file.

Fig. 7. Efficacy of identified anti-DENV drugs in Huh-7 cells and AG6 mice.

a Schematic overview of the experimental design depicting DENV-2 infection in Huh-7 cells. Created with MedPeer (medpeer.cn). b Quantification of viral RNA in the culture medium from ORES-treated and RTA 408-treated cells. n = 3. Values and error bars reflect mean ± SD. Statistical significance was assessed using an unpaired two-tailed t-test. The exact p values: p < 0.0001. c Schematic overview of the experimental design for DENV-2 infection in AG6 mice. Created with MedPeer (medpeer.cn). d, e Quantification of viral RNA in the liver (d) and serum (e) of DENV-2 infected AG6 mice treated with vehicle, ORES and RTA 408. Vehicle group: n = 4. ORES group: n = 5. RTA 408 group: n = 4. Values and error bars reflect mean ± SD. Statistical significance was assessed using an unpaired two-tailed t-test. The exact p values: p = 0.0204 (Vehicle vs ORES, liver), p = 0.0140 (Vehicle vs RTA 408, liver); p = 0.0440 (Vehicle vs ORES, serum), p = 0.0234 (Vehicle vs RTA 408, serum).