Human inherited complete STAT2 deficiency underlies inflammatory viral diseases

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.


Introduction
Human type I and III IFNs operate in almost all cell types (1,2). Type I IFNs (IFN-α/β) are thought to be secreted by most, if not all, of the more than 400 discernable cell types of the human body (1). The secretion of type III IFNs (IFN-λ1-4) seems to be restricted to epithelial cells of barrier organs and plasmacytoid and conventional dendritic cells (3). The main signaling pathway downstream from the type I (IFNAR) and type III IFN receptors (IFNLR) involves STAT1 and STAT2 (4,5). Once activated, these 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 STAT2dependent 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.
One 19-year-old patient (P11) developed severe COVID-19 pneumonia with involvement of 75% of the lungs, requiring noninvasive ventilation. She had not been vaccinated against SARS-CoV-2. She was treated with steroids, antibiotics, and the anti-SARS-CoV-2 monoclonal antibodies casirivimab and imdevimab and recovered. P10 had mild COVID-19 at the age of 17 years and was treated with the anti-SARS-CoV-2 monoclonal antibody sotrovimab on day 4 (D4) after symptom onset; she had previously received 3 doses of anti-SARS-CoV-2 vaccine (Comirnaty). Viremia was undetectable on D4, prior to administration of sotrovimab, and the patient recovered without complications. P1, P2, and P16 also had mild COVID-19 manifesting as a febrile upper respiratory tract infection at the age of 40, 14, and 3 years, respec-(P18); and chickenpox in 1 case (P16). P16 and P18 also presented secondary HLH a few days after inoculation with MMR and VZV vaccines. Human herpesvirus 6 (HHV6) viremia was detected in these patients at the time. PCR for measles, mumps, and rubella was not performed in P16 (although VZV PCR was positive on a cutaneous lesion), but the vaccine-strain mumps virus was detected in cerebrospinal fluid from P18. Diagnosis in the other patients was based on clinical presentation and timing relative to LAV vaccination; it was also confirmed by PCR in P8 (PCR on throat swab positive for measles, mumps, and rubella and PCR on cerebrospinal fluid positive for mumps), P11 (PCR on throat swab positive for mumps and rubella), and P15 (PCR on urine positive for mumps and IgM against mumps virus detected in serological tests).
Inflammatory manifestations. Systemic inflammation, manifesting as an increase in serum proinflammatory cytokine concentrations and multiple organ dysfunction, or (atypical) Kawasaki disease, was diagnosed or suspected in 6 patients (after LAV vaccination in 5), and inflammation and cytopenia compatible with secondary HLH were present in 2 patients. P7 had a first episode of meningoencephalitis and opsoclonus-myoclonus 1 month after vaccination with MMR and sustained substantial neurological damage following a second episode of meningoencephalitis, with seizures and opsoclonus-myoclonus triggered by an undetermined infection with fever and diarrhea at the age of 2.5 years. P11 was diagnosed with atypical Kawasaki disease after MMR vac-cination. Five patients (P8, P9, P10, P16, P18) had transient neutropenia, anemia, and thrombocytopenia, with or without coagulopathy, coinciding with systemic inflammation and multiple organ involvement during febrile illness, LAV administration, or viral infections, indicative of secondary HLH, although diagnosis based on HLH criteria was confirmed in only 2 patients (P16, P18). The determination of a panel of proinflammatory cytokines in serum samples from P10 outside clinical events (n = 2) and during mild COVID-19 (n = 1, 4 days from symptom onset) showed an increase in IL-6 levels during infection (Supplemental Figure 1A).
Immunological phenotype. Five of 12 patients (P8, P9, P10, P11, P17) had T and/or B cell lymphopenia during acute illness, with documented normalization upon resolution of the illness in 4 cases (data not shown). We also performed in-depth immunophenotyping by mass cytometry on PBMCs derived from P10 outside of infectious episodes at the age of 17 years. No notable differences with respect to a healthy control were detected (Supplemental Figure 1B). Mitogen-induced T cell proliferation was normal in the 5 patients tested. Total IgG and IgM levels were normal or high in all 11 patients tested (during Ig supplementation in 4 cases). Partial IgA deficiency was detected in 4 of the 11 patients tested, all under the age of 4 years, which is in line with findings in the general population of this age. Specific postvaccination antibody levels were normal in all 7 patients tested for tetanus and in both patients tested for pneumococcus. P10 mounted normal responses to the mRNA vaccine against SARS-CoV-2 (Supplemental Table 2).
Outcome and treatment. Eight patients died in early childhood (range: 2 months to 7 years). Six patients died from heart failure in the context of a febrile illness without identified viral or other infectious etiology (P3, P9, P12, P15, sibling 2, sibling 3). In 2 cases, the clinical diagnosis before death was tonsillitis (P12, P15), but no microbiological samples were obtained. One 5-year-old child died from fulminant hepatitis of possible but unproven viral etiology (sibling 4), and one 2-year-old child died from HSE (sibling 1). In P3, postmortem examination revealed enterocolitis, interstitial pneumonia, and cerebral cortical edema with neuronal apoptosis, consistent with overwhelming viral illness, and in sibling 4, the results were consistent with overwhelming infection. One patient was on cotrimoxazole prophylaxis at the time of death, whereas the other patients were not on any medication (in P3, P9, P15, and all siblings, death preceded the genetic diagnosis). Three patients (P7, P10, P18) received high-dose intravenous immunoglobulins (0.8 g/kg to 2 g/kg) during systemic inflammation or Kawasaki disease, resulting in a rapid resolution of symptoms. Several patients were treated empirically with antibiotics, antiviral drugs (acyclovir), and antifungal agents during febrile illnesses with multiple organ involvement, cytopenia, and/or coagulopathy, although no bacterial or fungal infections were proven, and 2 patients received corticosteroids (P7, P10). Fifteen patients are still alive at this writing (median age: 13 years; range: 5-40 years). Four of these patients are on acyclovir prophylaxis (aged 8-20 years), and 6 are on intravenous or subcutaneous immunoglobulins (aged 5-13 years), combined with cotrimoxazole prophylaxis in 2 patients. None of these patients has undergone HSCT. Overall survival was assessed (Figure 2), and mortality was found to be 35% in early childhood.
Expression and function of STAT2 mutant alleles. We studied the effect of the STAT2 alleles on STAT2 protein production and were predicted to receive weaker signals from a wide range of cell types, with differential expression of the galectin-and MHC-Irelated pathways (Supplemental Figure 2, A-D). These data confirm that STAT2 deficiency impairs cellular responses to type I IFN in the basal state without affecting leukocyte subsets per se. They also highlight the upregulation of the TNF/NF-κB signature.
IFN-α fails to induce ISG expression in STAT2-deficient leukocytes. We investigated STAT2-dependent cellular responses to IFN-α by performing scRNA-Seq in unstimulated or IFN-α2Bstimulated PBMCs from healthy controls, a STAT2-deficient patient, and an IFNAR2-deficient patient. Stimulation with IFN-α2B did not affect the distribution of the 17 leukocyte subsets identified in clustering analysis ( Figure 6, A and B). Pseudobulk PCA showed weaker transcriptional responses to IFN-α2B stimulation across all cell types analyzed in STAT2-and IFNAR2-deficient cells relative to healthy controls ( Figure 6C). GSEA with the hallmark signature gene sets revealed a significant impairment of the induction of type I ISGs by IFN-α2B across multiple cell types in the STAT2-and IFNAR2-deficient patients (Figure 6, D-G). Unlike IFNAR2-deficient cells, STAT2-deficient cells displayed a minimal residual response to type I IFN, which was most pronounced in classical and nonclassical monocytes (Figure 6, D-G). Weighted gene coexpression network analysis (WGCNA) (41) identified 3 modules of genes induced by IFN-α2B in control cells, but not in STAT2-and IFNAR2-deficient cells ( Figure 6H). Module 3, which included USP18, MX1, ISG15, OAS1-2-3, IL15, IRF7, and IRF9, was induced across all lymphoid and myeloid leukocyte subsets analyzed, whereas modules 15 and 23, containing JAK2, TRIM25-38-69, CXCL9-10-11, and UNC93B1, were more prominently induced in myeloid cells. Overall, STAT2 deficiency was found to impair cellular responses to type I IFNs profoundly across leukocytes ( Figure 6H).
Impaired antiviral immunity to HSV-1 in STAT2-deficient SV40 fibroblasts. We report the first case, to our knowledge, of HSV-1 encephalitis (HSE) in STAT2 deficiency (sibling 1). We infected SV40 fibroblasts from healthy controls and from STAT2-, IFNAR1-, STAT1-, and IRF9-deficient patients with HSV-1 in vitro. We observed at least a 10 7 -fold increase in HSV-1 titers relative to healthy cells in STAT2-deficient cells 48 hours after infection (Supplemental Figure 3A). Similar results were obtained for IFNAR1-, STAT1-, and IRF9-deficient cells. Pretreatment with IFN-α2B for 16 hours before HSV-1 infection significantly decreased viral replication levels relative to untreated cells 72 hours after infection in healthy control cells, but not in STAT2-, IFNAR1-, STAT1-, and IRF9-deficient cells (Supplemental Figure 3A). These data confirm the greater susceptibility to HSV-1 infection of nonhematopoietic STAT2-deficient cells in vitro.
Lack of ubiquitin-specific peptidase 18 upregulation in patientderived STAT2-deficient LCLs stimulated with IFN-α. Ubiquitinspecific peptidase 18 (USP18) is an ISG that is recruited to IFNAR2 by STAT2 and downregulates the response to type I IFNs through steric hindrance, preventing JAK1 from binding to IFNAR2 (6-8, 42, 43). Impaired USP18 upregulation may, therefore, theoretically contribute to the hyperinflammation observed in individuals with STAT2 deficiency, although hyperinflammation has been documented in patients with other inborn errors of type I IFN immunity ( (Ex5-In19), was studied in an EBV-transformed lymphoblastoid cell line (EBV-LCL) obtained from P19. Immunoblotting demonstrated a complete absence of the STAT2 protein (C612X, v628fs14X, R330X, Q274X, del[5′US-In8]/del[Ex5-In19]) or a truncated protein (R667X, R510X, delEx13) in this overexpression system and an absence of phosphorylation of the Y-690 residue of STAT2 after stimulation with IFN-α2A for all alleles ( Figure  3A). We then introduced the STAT2 alleles into STAT2-deficient fibrosarcoma cells and performed quantitative reverse-transcriptase PCR (RT-qPCR) to evaluate upregulation of the ISGs IFIT1, IFI27, RSAD2, and USP18 after stimulation with IFN-α2A. We evaluated the upregulation of MX1, RSAD2, and USP18 in LCLs from P19. ISG induction was suppressed by all variant alleles, but was normal for the WT allele ( Figure 3B). We also tested the missense variants found in the homozygous state in the general population for their expression and phosphorylation of STAT2 protein and for the upregulation of ISGs after IFN-α2A treatment. These were normal relative to WT STAT2 (Figure 3, B and C). We concluded that the patients' alleles resulted in a loss of expression of full-length protein and a loss of function of STAT2, at least in this experimental setting.
Single-cell RNA-Seq reveals impaired basal ISG expression in STAT2-deficient leukocytes. We then investigated the impact of STAT2 deficiency on leukocyte subsets, their transcriptomic profile, and their predicted molecular interactions by single-cell RNA sequencing (scRNA-Seq) on PBMCs from P10 and a patient with IFNAR2 deficiency in clinical remission compared with healthy pediatric controls (37). Clustering analysis identified 24 different leukocyte subsets, none of which was altered in the STAT2-deficient patient (Figure 4, A and B). Pseudobulk principal component analysis (PCA) and gene-set enrichment analysis (GSEA) (38, 39) revealed a distinctive transcriptional pattern across multiple leukocyte subsets that was common to STAT2-and IFNAR2deficient patients ( Figure 4C). ISGs, which are mostly induced by type I and/or type II IFNs, were downregulated in all 24 leukocyte subsets in the STAT2-and IFNAR2-deficient patients, with MX1, IRF9, USP18, and ISG15 all downregulated in CD8 + effector memory T cells (CD8EM cells) and classical monocytes (Figure 4, D-F, and Figure 5, A and B). The single-cell expression of MX1, IRF9, STAT1, IRF1, ICAM1, ISG15, and USP18 was also downregulated ( Figure 4F). In contrast, a significant upregulation of genes involved in TNF/NF-κB signaling was observed in 8 leukocyte subsets in STAT2-and IFNAR2-deficient patients ( Figure  4, D and E). Intercellular communication analysis with CellChat (40) indicated that intercellular interaction between CD8EM cells and classical monocytes was weaker in STAT2-and IFNAR2deficient patients ( Figure 5B and Supplemental Figure 2A). CD8 + central memory T cells (CD8CM cells) and CD8EM cells were the only cell types for which the classical monocytes of STAT2-and IFNAR2-deficient patients were predicted to provide weaker signals. In contrast, STAT2-and IFNAR2-deficient CD8EM cells tein levels was observed after IFN-α2A stimulation in STAT1-, STAT2-, IFNAR1-, IFNAR2-, or IRF9-deficient LCLs ( Figure 7A and Supplemental Figure 3B). Moreover, phosphorylated STAT1 (phospho-STAT1) and phospho-STAT2 levels returned to baseline within 24 hours in healthy control LCLs, whereas they remained high for up to 24 hours after stimulation with IFN-α2A in STAT2-the activation of the JAK/STAT pathway downstream from IFNAR in EBV-derived LCLs from a healthy control and patients with complete STAT2, STAT1, IRF9, IFNAR1, or IFNAR2 deficiency at various time points after stimulation with IFN-α2A. In the healthy control, USP18 expression was detected from 6 to 48 hours after IFN-α2A stimulation. However, no upregulation of USP18 pro-  potentially inducing the expression of genes harboring γ-activating sequences (GAS), thereby driving IFN-γ-like inflammatory responses (44). However, HLH also occurs in AR complete STAT1 deficiency (11). Nevertheless, we measured STAT1 production, STAT1 phosphorylation, and ISG induction in LCLs from a healthy control and patients with complete IFNAR1, IFNAR2, STAT1, STAT2 (P10 and P19), or IRF9 deficiency after stimulation with IFN-α2A for 1, 6, 24, or 48 hours. Phospho-STAT1 levels in STAT2-and IRF9-deficient LCLs and phospho-STAT2 levels in STAT1-and IRF9-deficient LCLs were similar to those of the control 1 hour after stimulation, but phosphorylation levels remained higher at the later time points than in the healthy control. We observed higher levels of baseline STAT1 phosphorylation in STAT2-and IRF9-deficient LCLs. STAT2-, STAT1-, IFNAR1-, IFNAR2-, and IRF9-deficient LCLs displayed a strong impairment of the induction of the ISRE-dependent ISGs IFI27, IFIT1, RSAD2, and USP18 by RT-qPCR analyses 1, 6, 24, and 48 and IRF9-deficient LCLs. RT-qPCR showed a strong impairment of the induction of ISGs, including USP18, in STAT2-, STAT1-, IFNAR1-, IFNAR2-, and IRF9-deficient LCLs at all time points after stimulation with IFN-α2A ( Figure 7B). These findings suggest that low levels of USP18 expression in response to IFN-α2A stimulation may contribute to the inflammatory phenotype in patients with STAT2 deficiency and in patients with STAT1 or IRF9 deficiency, but such mechanisms clearly cannot operate in patients with IFNAR1 or IFNAR2 deficiency. The phenotypic similarity of IFNAR1 and IFNAR2 deficiencies and STAT2 deficiency, in terms of inflammatory episodes, and the lack of type I IFN-induced ISG expression in STAT2-deficient LCLs suggest that the impaired induction of USP18 makes no major contribution to hyperinflammation in the absence of STAT2 (20,32,33).
Prolonged STAT1 phosphorylation in patient-derived STAT2deficient LCLs upon stimulation with IFN-α. STAT1 homodimers can operate downstream from IFNAR in the absence of STAT2, notypic alterations in P10, the observed enrichment in these cell types probably reflects their (hyper)activation rather than their accumulation in peripheral blood. Neutrophils were the only cells for which the positive enrichment score in the STAT2 patient was much higher during acute mild COVID-19 than in the basal state or in a healthy control (Supplemental Figure 4B). This patient did not experience severe inflammation at the time of COVID-19, but our data nevertheless highlight an enhancement of inflammatory cascades (TNF/NF-κB and IL-6/JAK/STAT3) in STAT2 deficiency, both in the basal state and during acute COVID-19, mediated predominantly by neutrophils.

Discussion
We describe inherited complete STAT2 deficiency in 23 patients of distant ancestries from 10 families in 7 countries. The predominant clinical presentation was disseminated infection after inoculation with LAV, especially the measles vaccine (clinical infection in 70% of the patients receiving any LAV). Disseminated disease after LAV vaccination for MMR has been reported in patients with other defects of type I IFN immunity, such as complete deficiencies of STAT1, IFNAR1, IFNAR2, and IRF9, but not deficiency of IL-10RB, a subunit of IFNLR (11,20,(22)(23)(24)(45)(46)(47)(48)(49). The penetrance of disseminated disease following MMR in these patients was high but incomplete, except for patients with complete AR STAT1 deficiency. Patients with deficiencies of IFNAR1 or IFNAR2 are also susceptible to disease caused by the YFV LAV (24,27). None of the patients with STAT2 deficiency reported here had been vaccinated against YFV. YFV vaccination is contraindicated for individuals with STAT1, STAT2, IRF9, IFNAR1, or IFNAR2 deficiencies. Live attenuated VZV vaccination has also caused disease in patients with STAT2 deficiency and is therefore also contraindicated. None of the known patients with an inborn error of type I/III IFN immunity had received the oral live attenuated poliovirus vaccine. It is probably safer to vaccinate these individuals with the inactivated poliovirus vaccine. Inborn errors of type I/ III IFN immunity should be considered in any patient with adverse reactions to MMR, YFV, or VZV LAV. The presence of neutralizing anti-type I IFN autoantibodies should also be investigated (27).
Susceptibility to severe natural viral diseases was noted in 43% of STAT2-deficient patients (10/23). A high prevalence of influenza pneumonia was observed (6/7 or 86% of patients with PCR-proven influenza A). AR IRF9 deficiency, AR IRF7 deficiency, AD TLR3, and AR IFNAR2 deficiency also underlie influenza pneumonia (22,23,33,(50)(51)(52)(53). All these defects affect both type I and type III IFN immunity through different mechanisms: TLR3 induces the transcription of type I and III IFNs; IRF7 is required for their amplification; and IRF9 is the DNA-binding partner in ISGF3 (54). AR STAT2 deficiency is the fifth monogenic etiology of critical influenza to be described (17). It may not be coincidental that these 4 etiologies impair both type I and type III IFN immunity, whereas critical influenza pneumonia has only been reported in 1 of the 29 known patients with IFNAR1 or IFNAR2 deficiency (19)(20)(21)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33) and in none of the more than 30 patients with IL10RB deficiency (55)(56)(57)(58). Despite the high prevalence of herpesvirus infections in the general population and in STAT2-deficient patients, no life-threatening herpesvirus infections were reported in this cohort, besides 1 case of fatal HSE. No major anomalies in hours after stimulation with IFN-α2A ( Figure 7B and Supplemental Figure 3C). We then assessed the upregulation of the classical IFN-γ-induced ISGs IRF1 and ICAM1, which are GAS-dependent, in response to IFN-α2A and IFN-γ (Supplemental Figure 3, D and E). IRF1 expression was induced in the healthy control and in IRF9-deficient patient cells, but was induced to a much lesser extent in STAT2-deficient patient cells following stimulation with IFN-α2A, whereas ICAM1 expression was not upregulated. Thus, the prolonged STAT1 phosphorylation observed in STAT2deficient LCLs after IFN-α2A stimulation was not accompanied by increased expression of IRF1 and ICAM1, 2 typical IFN-γdependent ISGs, at least in these experimental conditions. We also found no significant difference in the expression of SOCS1, SOCS3, and CIITA between the tested cell lines at any time point after stimulation with IFN-α2A (data not shown).
Enhanced inflammatory response in the basal state and during COVID-19 in individuals with STAT2 deficiency. We next analyzed the differential gene expression in whole blood from a STAT2deficient patient (P10) during acute mild COVID-19 due to PCR-confirmed natural SARS-CoV-2 infection (D4 after symptom onset) or in the basal state, 5 months after infection, relative to a healthy control matched with the patient for age and sex. GSEA against hallmark gene sets based on differentially expressed gene (DEG) ranking on bulk RNA-Seq revealed an enrichment in genes relating to inflammatory responses, including TNF signaling via NF-κB and the IL-6/JAK/STAT3 pathway (Supplemental Figure  4A) on D4 after symptom onset. Moreover, in the basal state, the STAT2-deficient patient displayed enhanced expression of the inflammatory signature at the transcriptome level relative to the control. Absolute cell-type deconvolution analysis demonstrated higher positive enrichment score for classical monocytes, basophils, CD8EM cells, and non-Vδ2 TCR-γδ T cells in the STAT2deficient patient than in the control, both during acute COVID-19 illness and outside of infectious/inflammatory episodes (Supplemental Figure 4B). Given the absence of major immunophe- STAT2, IFNAR1, IFNAR2, STAT1, IRF9, IRF7, and IL10RB deficiencies and patients with autosomal dominant or AR TLR3 deficiency is presented in Table 4.
The sample size is too limited to draw definitive conclusions on the nature and range of viral infections that cause life-threat-the immunophenotype or in antibody responses were detected in STAT2-deficient patients, in agreement with recent work evidencing normal antibody response to mRNA SARS-CoV-2 vaccines in the absence of type I IFN signaling (59). An overview of the infectious and immunologic phenotype of patients with AR complete , 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. viruses, including HSV-1, VZV, CMV, and HHV6 (16,17,22). The severity of STAT1 deficiency may be due to the additional deficit of type II IFNs, which does not in itself underlie any particular viral illnesses (10,12), but may aggravate deficits of type I and type III IFNs. The phenotype of STAT1 and STAT2 deficiency is in part reflected in mice, in that both Stat1 -/and Stat2 -/mice display increased susceptibility to viral infection with increased mortality and indications of inflammation, although formal comparisons are challenging (2). The incomplete penetrance of life-threatening viral diseases in patients with STAT2 deficiency accounts for 15 of the 23 patients being still alive at ages of 3 to 38 years (median: ening disease in STAT2 deficiency. We noted a predominance of severe infections with MMR vaccine, influenza, and herpesviruses. Moreover, most patients had infections with several other common viruses, including SARS-CoV-2, some without severe consequences. The diverse ancestries/countries of these patients suggest that these findings are genetically robust and largely independent of the viruses encountered. Our descriptions are reminiscent of those for AR IRF9 deficiency and IFNAR1 or IFNAR2 deficiencies (19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33). They contrast with the findings for patients with complete STAT1 deficiency, who have an almost completely penetrant phenotype of severe infections caused by multiple Table 4. Infectious and immunologic phenotype of patients with complete AR STAT2, IFNAR1, IFNAR2, STAT1, IRF9, IRF7, and IL10RB  deficiencies and with autosomal dominant or AR TLR3  with naturally occurring viruses. In conclusion, STAT2 deficiency underlies severe viral diseases characterized by excessive inflammation due to impaired responses to type I IFN in the initial phase of infection. Half of the patients described in this cohort have survived into teenage years or adulthood. The range of viral susceptibility remains to be determined.

Methods
For complete information on Methods, see Supplemental Methods. Data availability. Raw data generated from next-generation sequencing in this study were deposited in the NCBI Sequence Read Archive (SRA BioProject PRJNA936917, PRJNA818002, PRJNA856671, PRJNA898284, PRJNA924565, PRJNA856671). See also Supplemental Methods.
Statistics. Descriptive statistics were used (e.g., mean and SEM in plots, as described in the figure legends). No measure of significance has been used in this paper. Pseudobulk principal component analysis and gene-set enrichment analysis were used for analyzing scRNA-Seq data.
Study approval. Patients from previous reports were recruited by contacting physicians, and new patients were recruited via international calls and from the Human Genetics of Infectious Diseases Laboratory cohort of patients with severe viral infections. All patients or their legal guardians gave consent for participation in the study. The study was approved by the Ethical Research Committee of Leuven University Hospitals (protocol number S60905). The data were collected via a case record form. first, the timing of most LAVs and incidence of primary viral infections is in childhood; second, 6 of the patients received immunoglobulins after initial presentation; third, further functional development of T and B cell function could provide protection. Six STAT2-deficient patients died of heart failure in the context of febrile illness without identified pathogen. Whether this points to a failure to detect the disease-causing pathogen, to inflammatory cardiomyopathy/viral myocarditis, or to a hitherto unknown role for STAT2 besides host defense is unknown. Ten of the 15 surviving patients receive antiviral prophylaxis (acyclovir or immunoglobulins), and 5 patients have remained well as of this writing in the absence of prophylaxis.
Systemic inflammation and HLH have been reported in several children with defective IFN responses (11,20,25,26,31,60). Systemic inflammation was detected in 7 of 23 STAT2deficient patients and was mostly triggered by exposure to LAV (n = 5). HLH has been described in patients with complete STAT1, IFNAR1, or IFNAR2 deficiency, whereas only prolonged fever has been reported in patients with the rarer IRF9 deficiency. We propose a 2-phase model in which deficiencies of type I IFN immunity in the early stages of a viral infection lead to uncontrolled viral replication, with the activation and/or recruitment of tissular and circulating leukocytes leading to tissue and systemic inflammation. This 2-phase model of overwhelming inflammation is the core pathogenic mechanism for both hypoxemic COVID-19 pneumonia and multisystem inflammatory syndrome in children (MIS-C) (52,(61)(62)(63)(64). Our data reveal a TNF/NF-κB and JAK/ STAT3 signaling signature and roles for circulating monocytes, neutrophils, and CD8EM cells in inflammation in patients with STAT2 or related deficiencies. Given the overall decrease in signaling proximal and distal to IFNAR in STAT2 deficiency, the observed lack of USP18 expression and prolonged STAT1 phosphorylation probably play a minor role in the hyperinflammation observed in patients with AR STAT2 deficiency, particularly as similar inflammatory episodes occur in patients with STAT1, IFNAR1, and IFNAR2 deficiencies. Nevertheless, more data are needed to document this model further.
There is no standard treatment for AR STAT2 deficiency. Based on the results in patients with AR STAT1 complete deficiency, which is invariably lethal if not treated by HSCT, HSCT may be beneficial and possibly even curative in STAT2-deficient patients (11). However, it remains unclear whether the morbidity and mortality associated with STAT2 deficiency outweigh the intrinsic risks of HSCT. STAT2 deficiency-related mortality is strictly due to infectious or inflammatory complications, so early diagnosis and early intervention in the event of an acute infection could improve prognosis substantially. Antiviral medication should be initiated, if indicated, and a role for exogenous IFN-γ has been explored in the acute phase of infection in a patient with IFNAR1 deficiency with severe COVID-19 pneumonia (30). The early administration of anti-SARS-CoV-2 combined monoclonal antibody therapy in an IRF9-deficient patient resulted in the rapid clearance of viremia (65). High-dose immunoglobulins (0.8-2 g/ kg) and corticosteroids can be used to manage inflammatory complications of viral infection (16). Moreover, regular immunoglobulin replacement therapy or acyclovir may prevent severe infections