Human inherited complete STAT2 deficiency underlies inflammatory viral diseases

Abstract

STAT2 is a transcription factor activated by type I and III IFNs. We report 23 patients with loss-of-function variants causing autosomal recessive (AR) complete STAT2 deficiency. Both cells transfected with mutant STAT2 alleles and the patients' cells displayed impaired expression of IFN-stimulated genes and impaired control of in vitro viral infections. Clinical manifestations from early childhood onward included severe adverse reaction to live attenuated viral vaccines (LAV) and severe viral infections, particularly critical influenza pneumonia, critical COVID-19 pneumonia, and herpes simplex virus type 1 (HSV-1) encephalitis. The patients displayed various types of hyperinflammation, often triggered by viral infection or after LAV administration, which probably attested to unresolved viral infection in the absence of STAT2-dependent types I and III IFN immunity. Transcriptomic analysis revealed that circulating monocytes, neutrophils, and CD8+ memory T cells contributed to this inflammation. Several patients died from viral infection or heart failure during a febrile illness with no identified etiology. Notably, the highest mortality occurred during early childhood. These findings show that AR complete STAT2 deficiency underlay severe viral diseases and substantially impacts survival.

Keywords: Genetic diseases; Immunology; Influenza; Innate immunity.

Figure 1. STAT2 variants in 10 kindreds with severe viral infections.

(A) Pedigrees of the 10 STAT2-deficient kindreds. Double lines connecting parents indicate consanguinity. Filled symbols indicate individuals with biallelic mutations, and half-filled symbols indicate carriers of heterozygous mutations. M, mutated allele; ?, unknown genotype. (B) Schematic illustration of the STAT2 gene with 22 coding exons and of the STAT2 protein with its domains. N, N-terminal domain; CC, coiled-coil domain; DBD, DNA-binding domain; L, linker domain; SH2, Scr homology 2 domain; P-Y690, tyrosine phosphorylation site; TAD, transcriptional activation domain. All previously reported and new STAT2 variants are indicated. (C) Population genetics of homozygous coding missense and pLOF STAT2 mutations from gnomAD and in-house cohorts. The patients’ variants are private and shown in red, whereas the 8 variants detected in gnomAD are shown in black. CADD, combined annotation-dependent depletion; MSC, mutation significance cutoff.

Figure 2. Overall survival of 23 patients with STAT2 deficiency.

Ticks on the graph represent censored data.

Figure 3. Impact of STAT2 variants on protein production and type I IFN signaling.

(A) Immunoblot of STAT2 and phosphorylated STAT2 in HEK293T cells transfected with an untagged pCMV6 expression vector containing either the WT STAT2 cDNA or 1 of the variant cDNAs in basal conditions (–) or after pretreatment with 10,000 U/mL IFN-α2A for 30 minutes (+). One representative blot from 3 experiments performed is shown. NT, not transduced; EV, empty vector. (B) Transcription levels for IFIT1, IFI27, RSAD2, and USP18 assessed by RT-qPCR on U6A fibrosarcoma cells transfected with empty vector, WT STAT2, one of the mutated alleles, or one of the homozygous variants found in gnomAD after pretreatment with 10,000 U/mL of IFN-α2B for 6 hours. The mean (n = 3) and SEM are shown. Results are normalized relative to WT unstimulated conditions. (C) Immunoblot of STAT2 and phosphorylated STAT2 in HEK293T cells transfected with an untagged pCMV6 expression vector containing either the WT STAT2 or one of the homozygous variants found in gnomAD in basal conditions (–) or after pretreatment with 10,000 U/mL IFN-α2A for 30 minutes (+). A representative blot from 2 experiments performed is shown.

Figure 4. STAT2 deficiency impairs cellular responses to type I IFN in the basal state without affecting leukocyte subsets.

scRNA-Seq was performed on PBMCs from a STAT2-deficient patient (P10) and controls. (A) Leukocyte subsets identified by clustering analysis. (B) Relative abundance of cell types among PBMCs. (C) Pseudobulk PCA. (D and E) GSEA. Genes ranked based on fold change differences in expression in the STAT2-deficient patient (D) or the IFNAR2-deficient patient (E) relative to healthy pediatric controls were projected onto the hallmark gene sets (http://www.gsea-msigdb.org/gsea/msigdb/genesets.jsp?collection=H). Six immune-related gene sets were chosen for visualization. NES, normalized enrichment score. (F) Normalized pseudobulk read counts for GSEA leading-edge genes for the hallmark IFN-α response gene set common to the STAT2-deficient patient and pediatric controls and to the IFNAR2-deficient patient and pediatric controls, as shown in D and E. Representative results for CD8EM cells and classical monocytes are shown.

Figure 5. ISG expression and intercellular interactions between CD8EM cells and classical monocytes are weak in STAT2-deficient patients.

scRNA-Seq was performed on PBMCs from a STAT2-deficient patient (P10) and controls. (A) Mean single-cell expression levels for representative IFN-stimulated genes (logarithmic scale). (B) CellChat analysis showing the crude number of predicted cell-cell interactions.

Figure 6. IFN-α fails to induce ISG expression in STAT2 deficient leukocytes.

(A) Leukocyte subsets identified by clustering analysis. PBMCs were incubated for 6 hours with or without IFN-α2b (1000 IU/mL). Variations due to batch and stimulation were integrated with Harmony (see Supplemental Methods). (B) Relative abundance of the identified cell types among PBMCs after 6 hours of incubation without (circles) and with (squares) IFN-α2b. (C) Pseudobulk PCA for leukocyte subsets. (D–F) GSEA in IFN-α2b–stimulated PBMCs from healthy controls and STAT2- and IFNAR2-deficient patients. Ranking of gene differential expression projected onto the hallmark gene sets. Immune-related gene sets were chosen for visualization. (D) IFNα2b-treated versus nonstimulated for healthy controls.(E) Relative fold-change difference in expression in the STAT2-deficient patient relative to controls for the expression changes induced by IFNα2b stimulation versus nonstimulation. (F) Relative fold-change difference in expression in the IFNAR2-deficient patient relative to controls for differences in expression induced by IFNα2b stimulation versus nonstimulation. (G) Representative volcano plots. Genes significantly up- and downregulated (FDR-adjusted P value < 0.05 and |log₂FC| > 1) are shown in red and blue, respectively. (H) WGCNA. Three modules of coexpressed genes (Mod3/15/23) were induced by IFN-α2b in control cells. These modules were used for the heatmap analysis. Heatmaps show batch-corrected Z-transformed normalized pseudobulk read counts.

Figure 7. STAT2-deficient cells do not upregulate USP18.

(A) Immunoblot of STAT1, phospho-STAT1, STAT2, phospho-STAT2, and USP18 in EBV-LCL cells derived from either a healthy control (HC) or a patient with complete IFNAR1, IFNAR2, STAT1, STAT2, or IRF9 deficiency after pretreatment with 10,000 U/mL IFN-α2A for 1, 6, 24, or 48 hours. One representative blot from 3 experiments performed is shown. (B) Transcription levels for IFI27, IFIT1, RSAD2, and USP18 assessed by RT-qPCR on EBV-LCL cells derived from either a healthy control or a patient with complete IFNAR1, IFNAR2, STAT1, STAT2, or IRF9 deficiency after pretreatment with 10,000 U/mL IFN-α2A for 1, 6, 24 or 48 hours. The mean (n = 3) and SEM are shown. Results are normalized relative to unstimulated healthy control conditions.