A human iPSC-array-based GWAS identifies a virus susceptibility locus in the NDUFA4 gene and functional variants

Abstract

Population-based studies to identify disease-associated risk alleles typically require samples from a large number of individuals. Here, we report a human-induced pluripotent stem cell (hiPSC)-based screening strategy to link human genetics with viral infectivity. A genome-wide association study (GWAS) identified a cluster of single-nucleotide polymorphisms (SNPs) in a cis-regulatory region of the NDUFA4 gene, which was associated with susceptibility to Zika virus (ZIKV) infection. Loss of NDUFA4 led to decreased sensitivity to ZIKV, dengue virus, and SARS-CoV-2 infection. Isogenic hiPSC lines carrying non-risk alleles of SNPs or deletion of the cis-regulatory region lower sensitivity to viral infection. Mechanistic studies indicated that loss/reduction of NDUFA4 causes mitochondrial stress, which leads to the leakage of mtDNA and thereby upregulation of type I interferon signaling. This study provides proof-of-principle for the application of iPSC arrays in GWAS and identifies NDUFA4 as a previously unknown susceptibility locus for viral infection.

Keywords: Dengue Virus; NDUFA4; SARS-CoV-2; genome-wide association study; iPSC array; isogenic hiPSC lines; mtDNA; risk allele; single-nucleotide polymorphism; type I interferon.

Graphical abstract

Figure 1

An iPSC-array-based screen of ZIKV infection (A) Scheme of the iPSC screening. (B) The percentage of ZIKV-E positive cells of all iPSC lines upon ZIKV^PR infection (ZIKV^PR, Multiplicity of infection [MOI] = 1). Three biological replicates (each replicate includes one well of 96-well plate) were used to calculate the infectivity. (C and D) Representative confocal images (C) and the quantification (D) of ZIKV-E staining of permissive cell lines iPSC #1, iPSC #41, and iPSC #57 and low-permissive cell lines iPSC #15, iPSC #17, and iPSC #19 at 72 hpi (ZIKV^PR, MOI = 1). Scale bar, 50 μm. (E) qRT-PCR analysis of (+) and (−) ZIKV vRNA strands of permissive cell lines iPSC #1, iPSC #41, and iPSC #57 and low-permissive cell lines iPSC #15, iPSC #17, and iPSC #19 at 72 hpi (ZIKV^PR, MOI = 1). The data was normalized to actin beta (ACTB). (F) Multiple step growth curve of ZIKV in the supernatant of permissive cell lines iPSC #1, iPSC #41, and iPSC #57 and low-permissive cell lines iPSC #15, iPSC #17, and iPSC #19 (ZIKV^PR, MOI = 1). (G and H) Representative confocal images (G) and the quantification (H) of ZIKV-E staining in cerebral organoids derived from permissive cell lines iPSC #1, iPSC #41, and iPSC #57 and low-permissive cell lines iPSC #15, iPSC #17, and iPSC #19 (ZIKV^PR, 3 × 10⁶ plaque-forming unit [PFU]/mL). Cerebral organoids were age-matched and collected at day 20, then infected with ZIKV for 24 h. After removal of virus-containing medium, organoids were maintained in organoid medium for an additional 3 days. Scale bar, 50 μm. (I) qRT-PCR analysis of (+) and (−) ZIKV vRNA strands in cerebral organoids derived from permissive cell lines iPSC #1, iPSC #41, and iPSC #57 and low-permissive cell lines iPSC #15, iPSC #17, and iPSC #19 (ZIKV^PR, 3 × 10⁶ PFU/mL). Cerebral organoids were age-matched and collected at day 20, then infected with ZIKV for 24 h. After removal of virus-containing medium, organoids were maintained in organoid medium for an additional 3 days. The value was normalized to ACTB. Data are representative of at least three independent experiments. Data are shown as mean ± SD. p values were calculated by unpaired two-tailed Student’s t test; ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001. See also Figure S1.

Figure 2

Loss or reduction of NDUFA4 impairs ZIKV replication in vitro and in vivo (A) Manhattan plot with highlighted risk locus. (B) qRT-PCR analysis of NDUFA4 mRNA expression levels in permissive iPSC lines iPSC #1, iPSC #41, and iPSC #57 and low-permissive cell lines iPSC #15, iPSC #17, and iPSC #19. The value was normalized to ACTB. (C and D) Western blotting analysis (C) and the quantification (D) of NDUFA4 protein expression levels in permissive iPSC lines iPSC #1, iPSC #41, and iPSC #57 and low-permissive cell lines iPSC #15, iPSC #17, and iPSC #19. β-Actin was used as a loading control. (E) Western blotting analysis of NDUFA4 expression levels in WT or NDUFA4^−/− hiPSCs. β-Actin was used as a loading control. (F and G) Representative confocal images (F) and the quantification (G) of ZIKV-E staining in ZIKV-infected WT or NDUFA4^−/− hiPSCs at 72 hpi (ZIKV^PR, MOI = 1). Scale bar, 50 μm. (H) qRT-PCR analysis of (+) and (−) ZIKV vRNA strands in ZIKV-infected WT or NDUFA4^−/− hiPSCs at 72 hpi (ZIKV^PR, MOI = 1). The value was normalized to ACTB. (I) Multiple step growth curve of ZIKV virus in the supernatant of ZIKV-infected WT or NDUFA4^−/− hiPSCs (ZIKV^PR, MOI = 1). (J) Luciferase activity of WT or NDUFA4^−/− hiPSCs at 24 h after transfection with ZIKV replicon. (K and L) Representative confocal images (K) and the quantification (L) of ZIKV-E staining in cerebral organoids derived from WT or NDUFA4^−/− hiPSCs (ZIKV^PR, 3 × 10⁶ PFU/mL). Cerebral organoids were age-matched and collected at day 20, then infected with ZIKV for 24 h. After removal of virus-containing medium, organoids were maintained in organoid medium for an additional 3 days. Scale bar, 50 μm. (M) qRT-PCR analysis of (+) and (−) ZIKV vRNA strands in cerebral organoids derived from WT or NDUFA4^−/− hiPSCs at 72 hpi (ZIKV^PR, 3 × 10⁶ PFU/mL). Cerebral organoids were age-matched and collected at day 20, then infected with ZIKV for 24 h. After removal of virus-containing medium, organoids were maintained in organoid medium for an additional 3 days. The value was normalized to ACTB. (N and O) Representative images (N) and the quantification (O) of ZIKV-E staining in ZIKV-infected NDUFA4^−/−-oeCTRL and NDUFA4^−/−-oeNDUFA4 hiPSCs at 72 hpi (ZIKV^PR, MOI = 1). Scale bar, 50 μm. (P) qRT-PCR analysis of (+) and (−) ZIKV vRNA strands in ZIKV-infected NDUFA4^−/−-oeCTRL and NDUFA4^−/−-oeNDUFA4 hiPSCs at 72 hpi (ZIKV^PR, MOI = 1). The value was normalized to ACTB. Data are representative of at least three independent experiments. Data are shown as mean ± SD. For comparison of permissive and low-permissive lines, p values were calculated by unpaired two-tailed Student’s t test. For comparison of WT and KO lines, p values were calculated by two-way ANOVA analysis. For comparison of control and overexpression groups, p values were calculated by unpaired two-tailed Student’s t test; ^∗p < 0.05, ^∗∗∗p < 0.001. See also Figure S2.

Figure 3

Risk alleles of rs917172 and rs12386620 cause increased sensitivity to ZIKV infection (A) Relative luciferase activity of rs917172 (risk: G, non-risk: T) and rs12386620 (risk: C, non-risk: T) in 293T cells. (B) Relative NDUFA4 mRNA expression in hiPSCs carrying risk (G/G; C/C) and non-risk (T/T; T/T) alleles. The value was normalized to ACTB. (C) Western blotting analysis of NDUFA4 protein expression levels in hiPSCs carrying risk (G/G; C/C) and non-risk (T/T; T/T) alleles. β-Actin was used as a loading control. (D and E) Representative confocal images (D) and the quantification (E) of ZIKV-E staining in ZIKV-infected hiPSCs carrying risk (G/G; C/C) and non-risk (T/T; T/T) alleles at 72 hpi (ZIKV^PR, MOI = 1). Scale bar, 50 μm. (F) qRT-PCR analysis of (+) and (−) ZIKV vRNA strands in ZIKV-infected hiPSCs carrying risk (G/G; C/C) and non-risk (T/T; T/T) alleles at 72 hpi (ZIKV^PR, MOI = 1). The value was normalized to ACTB. (G) Multiple step growth curve of ZIKV in the supernatant of ZIKV-infected hiPSCs carrying risk (G/G; C/C) and non-risk (T/T; T/T) alleles (ZIKV^PR, MOI = 1). (H and I) Representative confocal images (H) and the quantification (I) of ZIKV-E staining in cerebral organoids derived from hiPSCs carrying risk (G/G; C/C) and non-risk (T/T; T/T) alleles (ZIKV^PR, 3 × 10⁶ PFU/mL). Cerebral organoids were age-matched and collected at day 20, then infected with ZIKV for 24 h. After removal of virus-containing medium, organoids were maintained in organoid medium for an additional 3 days. Scale bar = 50 μm. (J) qRT-PCR analysis of (+) and (−) ZIKV vRNA strands of cerebral organoids derived from hiPSCs carrying risk (G/G; C/C) and non-risk (T/T; T/T) alleles (ZIKV^PR, 3 × 10⁶ PFU/mL). Cerebral organoids were age-matched and collected at day 20, then infected with ZIKV for 24 h. After removal of virus-containing medium, organoids were maintained in organoid medium for an additional 3 days. The value was normalized to ACTB. (K) The percentage of IgG⁺ or IgM⁺ patients carrying risk or non-risk allele of rs917172 and rs12386620. Data are representative of at least three independent experiments. Data are shown as mean ± SD. For comparison with more than two samples, p values were calculated by two-way ANOVA analysis. For comparison with two samples, p values were calculated by unpaired two-tailed Student’s t test; ^∗p < 0.05, ^∗∗p < 0.01 and ^∗∗∗p < 0.001. See also Figure S3.

Figure 4

Deletion of a cis-regulatory region of NDUFA4 decreased the infection of ZIKV infection (A) qRT-PCR analysis of NDUFA4 mRNA expression levels in WT_Δ and NDUFA4^Δ hiPSCs. The value was normalized to ACTB. (B) Western blotting analysis of NDUFA4 protein expression levels in WT_Δ and NDUFA4^Δ hiPSCs. β-Actin was used as a loading control. (C and D) Representative confocal images (C) and the quantification (D) of ZIKV-E staining in ZIKV-infected WT_Δ and NDUFA4^Δ hiPSCs at 72 hpi (ZIKV^PR, MOI = 1). Scale bar, 50 μm. (E) qRT-PCR analysis of (+) and (−) ZIKV vRNA strands in ZIKV-infected WT_Δ and NDUFA4^Δ hiPSCs at 72 hpi (ZIKV^PR, MOI = 1). The value was normalized to ACTB. (F) Multiple step growth curve of ZIKV in the supernatant of ZIKV-infected WT_Δ and NDUFA4^Δ hiPSCs (ZIKV^PR, MOI = 1). (G and H) Representative images (G) and the quantification (H) of ZIKV-E staining in cerebral organoids derived from WT_Δ and NDUFA4^Δ hiPSCs (ZIKV^PR, 3 × 10⁶ PFU/mL). Cerebral organoids were age-matched and collected at day 20, then infected with ZIKV for 24 h. After removal of virus-containing medium, organoids were maintained in organoid medium for an additional 3 days. Scale bar, 50 μm. (I) qRT-PCR analysis of (+) and (−) ZIKV vRNA strands in cerebral organoids derived from WT_Δ and NDUFA4^Δ hiPSCs (ZIKV^PR, 3 × 10⁶ PFU/mL). Cerebral organoids were age-matched and collected at day 20, then infected with ZIKV for 24 h. After removal of virus-containing medium, organoids were maintained in organoid medium for an additional 3 days. The value was normalized to ACTB. Data are representative of at least three independent experiments. Data are shown as mean ± SD. p values were calculated by two-way ANOVA analysis; ^∗p < 0.05, ^∗∗p < 0.01 and ^∗∗∗p < 0.001. See also Figure S4.

Figure 5

NDUFA4 is associated with DENV and SARS-CoV-2 infection (A and B) Representative confocal images (A) and the quantification (B) of DENV NS3 staining in ZIKV-infected WT or NDUFA4^−/− hiPSCs at 72 hpi (DENV, MOI = 1). Scale bar, 50 μm. (C and D) Representative confocal images (C) and the quantification (D) of DENV NS3 staining in hiPSCs carrying risk (G/G; C/C) or non-risk (T/T; T/T) alleles at 72 hpi (DENV, MOI = 1). Scale bar, 50 μm. (E and F) Representative confocal images (E) and the quantification (F) of DENV NS3 staining in DENV-infected WT_Δ or NDUFA4^Δ hiPSCs at 72 hpi (DENV, MOI = 1). Scale bar, 50 μm. (G and H) Representative confocal images (G) and the quantification (H) of DENV NS3 staining in DENV-infected permissive cell lines iPSC #1, iPSC #41, and iPSC #57 or low-permissive cell lines iPSC #15, iPSC #17, and iPSC #19 at 72 hpi (DENV, MOI = 1). Scale bar, 50 μm. (I and J) Representative confocal images (I) and the quantification (J) of SARS-N in FOXJ1⁺ ciliated cells in airway organoids derived from WT or NDUFA4^−/− hiPSCs at 24 hpi (SARS-CoV-2, MOI = 0.1). (K) Relative expression of viral subgenomic RNA (N) transcription in SARS-CoV-2-infected airway organoids derived from WT or NDUFA4^−/− hiPSCs at 24 hpi (SARS-CoV-2, MOI = 0.1). The value was normalized to ACTB. (L) Viral titers of SARS-CoV-2 infected airway organoids derived from WT or NDUFA4^−/− hiPSCs at 24 hpi (SARS-CoV-2, MOI = 0.05). (M and N) Representative confocal images (M) and the quantification (N) of SARS-N in FOXJ1⁺ ciliated cells in airway organoids derived from permissive cell lines iPSC #1, iPSC #41, or iPSC #57 and low-permissive cell lines iPSC #15, iPSC #17, and iPSC #19 at 24 hpi (SARS-CoV-2, MOI = 0.1). (O) Relative expression level of viral subgenomic RNA (N) transcripts in SARS-CoV-2-infected airway organoids derived from permissive cell lines iPSC #1, iPSC #41, and iPSC #57 or low-permissive cell lines: iPSC #15, iPSC #17, and iPSC #19 at 24 hpi (SARS-CoV-2, MOI = 0.1). The value was normalized to ACTB. (P) Viral titers of SARS-CoV-2-infected airway organoids derived from permissive cell lines iPSC #1, iPSC #41, and iPSC #57 or low-permissive cell lines iPSC #15, iPSC #17, and iPSC #19 at 24 hpi (SARS-CoV-2, MOI = 0.05). Data are representative of at least three independent experiments. Data are shown as mean ± SD. For permissive and low-permissive cell lines, p values were calculated by unpaired two-tailed Student’s t test. For other figure panels, p values were calculated by two-way ANOVA analysis; ^∗∗p < 0.01, and ^∗∗∗p < 0.001. See also Figure S5.

Figure 6

Loss or reduction of NDUFA4 causes mitochondrial DNA leakage (A) Correlation of pathways analyzed by Gene Ontology analysis of hiPSCs. Five pathways are enriched in all 3 groups: WT versus NDUFA4^−/−, risk (G/G; C/C) versus non-risk (T/T; T/T), and WT_Δ versus NDUFA4^Δ. (B) Heatmap of mitochondrial stress signature genes in 3 groups of hiPSCs. 3 groups: WT versus NDUFA4^−/−, risk (G/G; C/C) versus non-risk (T/T; T/T), and WT_Δ versus NDUFA4^Δ. (C and D) Representative electron microscopy images (C) and the quantification of mitochondrial size (D) of iPSC lines with mock or ZIKV infection at 48 hpi (ZIKV^PR, MOI = 1). 4 groups: WT versus NDUFA4^−/−, risk (G/G; C/C) versus non-risk (T/T; T/T), WT_Δ versus NDUFA4^Δ, and iPSC #1 versus iPSC #19. Scale bars, 500 nm. (E) Representative confocal images of hiPSCs at 48 hpi with mock or ZIKV infection stained with anti-HSP60 or anti-DNA antibodies (ZIKV^PR, MOI = 1). 4 groups: WT versus NDUFA4^−/−, risk (G/G; C/C) versus non-risk (T/T; T/T), WT_Δ versus NDUFA4^Δ, and iPSC #1 versus iPSC #19. Scale bar, 10 μm. (F) Quantification of HSP60 staining intensity and nucleoid area in (E) in mock or infected conditions (ZIKV^PR, MOI = 1). (G) qPCR analysis of mitochondrial DNA leakage in cytoplasm after ZIKV infection in hiPSCs at 48 hpi (ZIKV^PR, MOI = 1). 4 groups: WT versus NDUFA4^−/−, risk (G/G; C/C) versus non-risk (T/T; T/T), WT_Δ versus NDUFA4^Δ, and iPSC #1 versus iPSC #19. Data are representative of at least three independent experiments. Data are shown as mean ± SD. p values were calculated by unpaired two-tailed Student’s t test; ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001. See also Figure S6.

Figure 7

Loss or reduction of NDUFA4 triggers type I interferon signaling (A) Gene Set Enrichment Analysis of type I interferon signal pathway in 4 groups of hiPSCs. 4 groups: WT versus NDUFA4^−/−, risk (G/G; C/C) versus non-risk (T/T; T/T), WT_Δ versus NDUFA4^Δ, and iPSC #1 versus iPSC #19. (B) qRT-PCR analysis of ISG15 and IRF7 mRNA expression levels with mock or ZIKV infection at 72 hpi (ZIKV^PR, MOI = 1). 4 groups: WT versus NDUFA4^−/−, risk (G/G; C/C) versus non-risk (T/T; T/T), WT_Δ versus NDUFA4^Δ, and iPSC #1 versus iPSC #19. The value was normalized to ACTB. (C and D) Representative confocal images (C) and the quantification (D) of ZIKV-E staining of permissive cell lines iPSC #1, iPSC #41, and iPSC #57 and low-permissive cell lines iPSC #15, iPSC #17, and iPSC #19 treated with potassium cyanide (KCN) at 72 hpi (ZIKV^PR, MOI = 1). Scale bar, 50 μm. (E and F) Representative confocal images (E) and the quantification (F) of ZIKV-E staining of permissive cell lines iPSC #1, iPSC #41, and iPSC #57 and low-permissive cell lines: iPSC #15, iPSC #17, and iPSC #19 treated with blocking antibodies of IFNAR at 72 hpi (ZIKV^PR, MOI = 1). Scale bar, 50 μm. Data are representative of at least three independent experiments. Data are shown as mean ± SD. p values were calculated by unpaired two-tailed Student’s t test; ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001. See also Figure S7.