Multisystem inflammation and susceptibility to viral infections in human ZNFX1 deficiency

Abstract

Background: Recognition of viral nucleic acids is one of the primary triggers for a type I interferon-mediated antiviral immune response. Inborn errors of type I interferon immunity can be associated with increased inflammation and/or increased susceptibility to viral infections as a result of dysbalanced interferon production. NFX1-type zinc finger-containing 1 (ZNFX1) is an interferon-stimulated double-stranded RNA sensor that restricts the replication of RNA viruses in mice. The role of ZNFX1 in the human immune response is not known.

Objective: We studied 15 patients from 8 families with an autosomal recessive immunodeficiency characterized by severe infections by both RNA and DNA viruses and virally triggered inflammatory episodes with hemophagocytic lymphohistiocytosis-like disease, early-onset seizures, and renal and lung disease.

Methods: Whole exome sequencing was performed on 13 patients from 8 families. We investigated the transcriptome, posttranscriptional regulation of interferon-stimulated genes (ISGs) and predisposition to viral infections in primary cells from patients and controls stimulated with synthetic double-stranded nucleic acids.

Results: Deleterious homozygous and compound heterozygous ZNFX1 variants were identified in all 13 patients. Stimulation of patient-derived primary cells with synthetic double-stranded nucleic acids was associated with a deregulated pattern of expression of ISGs and alterations in the half-life of the mRNA of ISGs and also associated with poorer clearance of viral infections by monocytes.

Conclusion: ZNFX1 is an important regulator of the response to double-stranded nucleic acids stimuli following viral infections. ZNFX1 deficiency predisposes to severe viral infections and a multisystem inflammatory disease.

Keywords: HLH-like disease; ZNFX1; brain calcification; interstitial lung disease; leukoencephalopathy; susceptibility to viral infections; thrombotic microangiopathy; type I interferon; virally induced hepatitis.

FIG 1.

Severe viral infections and inflammatory disease in patients with ZNFX1 deficiency. A, Kaplan-Meier survival curve for patients; dashes indicate ages of patients who are alive. B, Overall inflammatory organ involvement with or without a proven link to infections; number of patients affected. C, May-Gruenwald-Giemsa staining (light microscopy; magnification, ×1000) of a bone marrow aspirate from P5.2. A macrophage with engulfed leukocytes is shown: its nucleus is indicated by an ar, and the engulfed leukocytes are indicated by an arrow. D, Computed tomography image of the brain of P1.2 at the age of 15 years showing calcification of the basal ganglia and white matter abnormalities (white ars). E, A high-resolution computed tomography image of the lungs P1.2 at the age of 9 years and 11 months, showing bilateral diffuse ground glass attenuation, subpleural thickening, and septal thickening. F, Jones staining of a kidney biopsy specimen, highlighting TMA lesions in P5.2. The arrow indicates a small arteriole with endothelial cell swelling and a fibrin/red blood microthrombus obliterating the lumen. Two glomeruli with capillary lumen dilatation and red blood cell stasis are indicated by asterisks. Acute tubular lesions with epithelial cell necrosis, lumen debris, and interstitial hemorrhage are observed (scale bar = 50 μm). ARDS, Acute respiratory distress syndrome; MOF, multiorgan failure; MPGN, membranoproliferative glomerulonephritis.

FIG 2.

Regression of white matter changes in the brain of patient P4.2 following HSCT. Axial fluid–attenuated inversion recovery. Magnetic resonance images at the ages of 32 months (A and D), 42 months (B and E), and 8 years (C and F) demonstrating an initial increase in periventricular and deep white matter changes 6 months after HSCT (B and E) and then marked regression seen at last follow-up (5 years after HSCT) (C and F).

FIG 3.

Biallelic ZNFX1 variants lead to the loss of protein expression in response to stimulation by intracellular nucleic acids. A, The pedigrees of the 8 families. Patients carrying homozygous or compound heterozygous deleterious variants in ZNFX1 are indicated by solid symbols. Healthy individuals carrying heterozygous variants are indicated by dotted symbols. Affected persons with an unknown genotype are indicated by open red symbols, whereas unaffected individuals are indicated by open diamonds. Circles indicate females, and squares indicate males. Slashes over symbols indicate deceased patients. N/A (meaning not available) indicates that sequencing was not performed. B, Predicted domains and identified variants in the ZNFX1 amino acid sequence. The 11 deleterious variants identified are indicated by arrows. The domain homologous to the RNA helicase Aquarius (Protein Data Bank identifier 4PJ3) is highlighted in orange, with an insert shown in yellow. C, A ribbon diagram of a homology model of ZNFX1 (183-1255), based on the structural template RNA helicase Aquarius (Protein Data Bank identifier 4PJ3) is shown. Locations of the 4 missense variants within this domain are shown as teal spheres in the present study. D, A protein immunoblot for ZNFX1 in dermal fibroblasts from a healthy donor (control [CTRL]) and from P5.1, P3.2, and P2.1 under resting conditions and 24 hours after transfection with the nucleic acids poly(dA:dT) or poly(I:C). β-Actin was used as a loading control.

FIG 4.

Biallelic defects in ZNFX1 deregulate ISGs’ expression and protection against viral infections in response to treatment with nucleic acids. A and B, Flow cytometry analysis of monocytes from P2.1 and a healthy control (CTRL) pretreated for 12 hours with different concentrations of LyoVec-poly(I:C) and subsequently infected with VSV–green fluorescent protein (GFP) for 5 hours. Representative plots of a single experiment showing VSV-GFP signal versus area of side scattered signal (SSC-A) (A) and mean percentage of VSV-GFP-positive monocytes relative to the unstimulated condition (no LyoVec-poly(I:C)) for 4 repeats (B). Error bars refer to the SD (n = 4). P values were calculated by using 2-way ANOVA and the Sidak multiple comparisons test. C, Transcriptomic analysis results for selected ISGs involved in antiviral responses are summarized in a heat map showing the mean difference in fold induction of ISGs expression from resting conditions in dermal fibroblasts from 4 patients (P1.2, P2.1, P3.2, and P5.2) over that in dermal fibroblasts from 4 different age-matched, healthy controls. Three different stimulations were used: 18 hours of intracellular poly(I:C) (LyoVec Poly (I:C)), 6 hours of soluble poly(I:C) (Poly (I:C)), or 18 hours of transfected poly(dA:dT) (LyoVec Poly (dA:dT)). D-F, The same data were used to study the activity of canonic double-stranded nucleic acids sensing pathways according to the Kyoto Encyclopedia of Genes and Genomes. Colored highlights indicate the rate of fold induction of gene expression in patients over that in the controls: red highlights the indicated increase, blue highlights the indicated decrease, and white boxes indicate no difference. Results from stimulation with LyoVec-poly(I:C) is shown in (D), with soluble poly(I:C) in (E) and LyoVec-poly(dA:dT) in (F). CASP8, Caspase 8.

FIG 5.

Increased ISG expression in response to transfected poly(dA:dT) in biallelic defects in ZNFX1 is associated with increased mRNA stability. A, The mRNA expression levels of OAS1, OAS2, and MX1 (representative ISGs) by skin fibroblasts from P1.2, P2.1, P3.2, and P5.2 (red squares) and 4 healthy controls (CTRLs) (black circles) at baseline (zero hours) and at different time points (6, 12, 18, 24. and 30 hours) after stimulation with transfection reagent–complexed poly(dA:dT). B, Mean values of mRNA stability of representative ISG mRNAs in fibroblasts from 4 healthy controls and 4 patients (P1.2, P2.1, P3.2, and P5.2). Gene transcription was inhibited by the addition of 5,6-dichlorobenzimidazole 1-β-D-ribofuranoside (DRB) 18 hours after transfection with LyoVec-poly(dA:dT).qPCR was performed at the indicated time points after the addition of DRB. The amount of mRNA at each time point was normalized against ribosomal 18S RNA and represented relative to the amount at the time of DRB addition (time zero). The half-life (t^1/2) of each mRNA (red for P1.2 and black for CTRL) was calculated by using nonlinear regression analysis. A representative result of 3 independent experiments is shown. Concentrations of IFN-β (C) and CXCL10 (D) in the supernatant of dermal fibroblasts from 3 healthy controls (CTRL [black bars]) and 3 patients (P1.2, P3.2, P5.2 [red bars]) following 18 hours of stimulation with poly(I:C)LyoVec or poly(dA:dT)LyoVec. Fibroblasts were transfected with plasmids expressing ZNFX1 or green fluorescent protein. Shown is the mean of 3 repeats for each of the 3 samples(n = 9), with error bars showing the SD. P values were calculated by using ordinary 1-way ANOVA as follows: ns = 0.12; *P = .033l; **P = .002; and ***P < .001.