Human intracellular ISG15 prevents interferon-α/β over-amplification and auto-inflammation

Abstract

Intracellular ISG15 is an interferon (IFN)-α/β-inducible ubiquitin-like modifier which can covalently bind other proteins in a process called ISGylation; it is an effector of IFN-α/β-dependent antiviral immunity in mice. We previously published a study describing humans with inherited ISG15 deficiency but without unusually severe viral diseases. We showed that these patients were prone to mycobacterial disease and that human ISG15 was non-redundant as an extracellular IFN-γ-inducing molecule. We show here that ISG15-deficient patients also display unanticipated cellular, immunological and clinical signs of enhanced IFN-α/β immunity, reminiscent of the Mendelian autoinflammatory interferonopathies Aicardi-Goutières syndrome and spondyloenchondrodysplasia. We further show that an absence of intracellular ISG15 in the patients' cells prevents the accumulation of USP18, a potent negative regulator of IFN-α/β signalling, resulting in the enhancement and amplification of IFN-α/β responses. Human ISG15, therefore, is not only redundant for antiviral immunity, but is a key negative regulator of IFN-α/β immunity. In humans, intracellular ISG15 is IFN-α/β-inducible not to serve as a substrate for ISGylation-dependent antiviral immunity, but to ensure USP18-dependent regulation of IFN-α/β and prevention of IFN-α/β-dependent autoinflammation.

Extended Data Figure 1. Mutations in ISG15-deficient individuals, allele characterization and serum IFN-α concentrations

a, Sanger sequencing of ISG15 exon 2 from genomic DNA in kindred C, with the variants highlighted. b, The wild type (WT) and three mutant alleles (G55X, L114fs, E127X- ISG15) were inserted into an expression vector and used to transfect HEK293T cells. Other HEK293T cells were mock-transfected (mock) or left untransfected (not transf.). The cell lysates isolated were subjected to western blotting, with recombinant human (Rh)-ISG15 used as a control. c, d, Plasma samples from P1, P2, P3, P5, P6 and the mother and brother of P5/6 were used in cytopathic protection assays, to measure antiviral activity (c) and blocking antibodies against IFN-α were used to assess specificity (d) (experiment was performed one time).

Extended Data Figure 2. A form of ISG15 that cannot be conjugated rescues the phenotype of ISG15-deficient cells

a, Lentiviral particles containing luciferase, wild-type (WT) ISG15-RFP or ISG15(ΔGG)-RFP genes were used to transduce hTert-immortalized fibroblasts from C1, a STAT1^−/− subject, P1, P2 and P3. RFP-positive cells were obtained by sorting and were cultured for a few weeks. The cells were then treated with 1,000 IU of IFN-α2b for 12 h, washed with PBS and left to rest for 36 h, after which relative mRNA levels for IFIT1 were determined. b, The experimental setting described in a was used in the presence or absence of vehicle control, anti-ISG15 antibodies or control IgG for luciferase and wild-type ISG15–RFP-transduced C1 and P1 hTert-immortalized fibroblasts (showing representative experiments with technical replicates and s.e.m., out of 3 performed).

Extended Data Figure 3. Prolonged IFN signalling, low USP18, and high interferon-stimulated-gene-encoded protein levels in patient-derived cells and in ISG15-silenced human fibrosarcoma HLLR1-1.4 cells

a, Left panels, SV40-immortalized fibroblasts from two controls (C10 and C12) and two ISG15-deficient patients (P1 and P2) were stimulated with IFN-β (500 pM) for 4 to 36 h. Cell lysates (30 µg) were analysed with the indicated antibodies. Right, EBV-transformed B cells from control (C3) and patient P1 were stimulated with IFN-β for 8 to 24 h. Cell lysates (30 µg) were analysed with the indicated antibodies. b, HLLR1-1.4 cells were transfected with control siRNA or ISG15 siRNA. One day post-transfection, IFN-β (500 pM) was added for various amounts of time. Cell lysates (30 µg) were analysed with the indicated antibodies (MxA and MX1 are used synonymously). c, WISH cells were stimulated and lysates analysed as described in b. d, HLLR1-1.4 cells were transfected with control siRNA, USP18 siRNA and UbcH8 (also known as UBE2E2) siRNA (left) or control siRNA, USP18 siRNA and HERC5 siRNA (right). One day post-transfection, cells were left untreated (naive) or were primed for 8 h with IFN-β (500 pM). Cells were washed and left to rest for 16 h before being pulsed for 30 min with 100pM IFN-α2 or IFN-β. Cell lysates (30 µg) were analysed with the indicated antibodies.

Extended Data Figure 4. ISG15 controls the stability of the USP18 protein, but not of other interferon-stimulated-gene products

a, HLLR1-1.4 cells were transfected with either control siRNA or ISG15 siRNA. One day post-transfection, cells were stimulated with IFN-β (500 pM) for 6 h. Cycloheximide (CHX, 20 µgml⁻¹) was then added for various time periods, from 30 min to 5 h. Cell lysates (30 µg) were analysed with the indicated antibodies. b, As in a, with additional controls, cells treated with IFN only. Several interferon-stimulated genes were analysed. c, hTert-immortalized fibroblasts from patient P1 transduced with lentiviral particles expressing RFP and luciferase and wild-type ISG15 (LV ISG15 WT) were stimulated with IFN-β (500 pM) for 6 h. CHX was then added for the indicated times. Cell lysates (15 µg) were analysed by western blotting. d, HEK293T cells were transfected with USP18, HA–ubiquitin and Flag–ISG15 as indicated. Two days later, cells were lysed in modified RIPA buffer, USP18 was immunoprecipitated (IP) and analysed with anti-USP18 antibodies. Left panels, cell lysates (30 µg) were analysed by western blot with the indicated antibodies. Right panels, the immunoprecipitates were gel separated and transferred onto a membrane. The membrane was cut into two parts above the 50 kDa marker, both of which were blotted with anti-USP18 antibodies. The top part was exposed for 2 min, the bottom part for 20 s. Asterisk indicates IgG heavy chain. e, HEK293T cells were transfected with 1 µg of the USP18 construct alone or with 1 µg of HA– ubiquitin, in the presence or absence of Flag-tagged ISG15, either wild type or a mutant form of ISG15 that cannot be conjugated as it lacks the two carboxyterminal glycine residues (Flag–ISG15(ΔGG)). Two days later, cells were lysed in modified RIPA buffer, USP18 was immunoprecipitated and analysed with anti-HA or anti-USP18 antibodies. Asterisk indicates IgG heavy chain. f, hTert-immortalized fibroblasts from patient P3 were transfected with control siRNA or SKP2 siRNA. We added IFN-β (500 pM) 24 h later and the cells were incubated for the indicated times. Cell lysates were analysed with the indicated antibodies and USP18 levels were determined as a function of actin levels.

Extended Data Figure 5. Autoantibody development in ISG15-deficient individuals

a, b, Serum samples from ISG15-deficient, SLE and AGS patients were evaluated for the presence of IgG and IgA autoantibodies in a blinded experiment. Values for the negative control samples for each antigen were averaged and ratios of each sample to the mean for the negative controls plus 2 standard deviations were calculated, with values greater than 1 considered positive. A heat map of the ratio values was generated with MultiExperiment Viewer software (MeV, DFCI Boston, MA), with values coded as follows: 0, blue; 1, black; 5, yellow.

Figure 1. Familial segregation of the ISG15 allele and CT scans for the affected families

a, Familial segregation in a family from Turkey (Kindred A), a family from Iran (Kindred B) (previously reported) and a family from China (Kindred C). Asterisks denote stop codons; E denotes unknown genotype; fs, frameshift; WT, wild type. b, Graphical representation of the proISG15 protein, with the LRLRGG motif required for substrate ISGylation and the eight-amino-acid sequence (black) cleaved to yield ISG15, and the putative proteins synthesized in the patients. c, Axial view cerebral CT scans of P1, P2, P3, P4, P5, P6 and the healthy mother of P4, P5 and P6 (ISG15^−/+).

Figure 2. High levels of interferon-stimulated gene expression in ISG15-deficient individuals

a, Relative mRNA levels for IFI27, IFI44L, IFIT1, ISG15, RSAD2 and SIGLEC1 in peripheral blood from patients (n = 3) or controls (C) (n = 24) or in peripheral blood mononuclear cells (PBMCs) from family members (wild type and heterozygous for ISG15 deficiency) (n = 2) and patients (n = 2), as assessed by RT–qPCR, comparison done with unpaired t-tests. ***P < 0.0001; horizontal bars represent means; RQ defined in reference to C1 unstimulated condition. b, hTert-immortalized fibroblasts from C1, C16, C18, a NEMO^−/− subject (as a negative control due to known hyporesponsiveness), P1 and P2 were treated with the indicated doses of IFN-α2b for 12 h, washed with PBS and left to rest for 36 h, after which relative mRNA levels were assessed. b shows a representative experiment of three performed. GUS is used as a housekeeping control gene. NS, not stimulated. c, The same experimental procedure as in b was followed, but the mRNA was used for a microarray experiment, for C1, C6, C16, P1, P2 and P3 cells, with green indicating relative upregulation and red indicating downregulation of the probe concerned. d, In the same experimental setup, we used lentiviral particles containing luciferase–RFP (red fluorescent protein) or wild-type ISG15–RFP genes to transduce hTert-immortalized fibroblasts and then assessed mRNA levels for IFIT1 and MX1 by RT–qPCR. Panel d shows one representative experiment of three performed, where (−) denotes not transduced conditions.

Figure 3

a, hTert-immortalized fibroblasts from a control (C18) and patient P3 were treated with 100pM IFN-α2 for 0.5 to 36 h. Cell lysates were analysed by western blot for levels of phosphorylated STAT (pSTAT) proteins and proteins encoded by interferon-stimulated genes. b, hTert-immortalized fibroblasts from controls (C1, C18) and patients P1 and P2 were treated with 100pM IFN-α2 for 12 h, washed and left to rest for 24 or 36 h. Protein levels were assessed by western blot. c, HLLR1-1.4 cells were transfected with control short interfering RNA (siRNA) or with siRNA targeting USP18, ISG15 or both. One day later, cells were left untreated (naive) or were primed for 8 h with 500 pM IFN-β, washed and left to rest for 16 h, and then restimulated for 30 min with 100 pM IFN-α2or IFN-β. Lysates were analysed with the indicated antibodies. d, HLLR1-1.4 cells were transfected with control siRNA, ISG15 siRNA or UBE1L siRNA. One day later, IFN-β (500 pM) was added for various periods of time. Lysates were analysed as indicated. e, hTert-immortalized fibroblasts from P3 transduced with lentiviral particles expressing RFP and luciferase (LV), wild-type ISG15 (LV ISG15 WT) or the IS15(ΔGG) mutant (LV ISG15(ΔGG)) were stimulated with IFN-β (500 pM) for 8 to 36 h and protein levels assessed by western blot.

Figure 4. Free ISG15 stabilizes USP18 by preventing SKP2-dependent ubiquitination

a, hTert-immortalized fibroblasts from control (C18) and patient P3 were left untreated (−) or treated for 6 h with IFN-β (500 pM). Cycloheximide (CHX) was added for an additional 0.5 to 5 h. Lysates were analysed as indicated. b, hTert fibroblasts from P3 transduced with lentiviral particles expressing RFP and luciferase (LV) conjugated to wild-type ISG15 (LV ISG15 WT) or the non-conjugatable ISG15(ΔGG) mutant (LV ISG15(ΔGG)) were processed as in a. c, HEK293T cells were transfected with USP18 expression vector alone or in combination with various amounts of ISG15 (Flag–ISG15), either wild-type or ΔGG. Two days later, lysates were analysed as indicated. Ratios of USP18 or ISG15 to AKT are shown below gels. d, HEK293T cells were cotransfected with USP18 and haemagglutinin (HA)–ubiquitin in the presence of wild-type ISG15 (Flag–ISG15). Two days later, lysates were subjected to immunoprecipitation (IP) with anti-USP18 antibodies. Lysates (left panels) were analysed with antibodies against HA and actin. Immunoprecipitates (right panels) were analysed with antibodies against HA and USP18. Asterisk indicates IgG heavy chain. e, HEK293T cells were transfected with USP18, Flag–SKP2 and Flag–ISG15 expression vectors, as indicated. Two days later, USP18 was immunoprecipitated. Lysates (left panels) and immunoprecipitates (right panels) were analysed with antibodies against SKP2, USP18 and ISG15. Arrowheads indicate two endogenous SKP2 isoforms; arrow indicates ectopic Flag–SKP2; asterisk, background band (lanes 6, 7 and 9, 10). f, HLLR1-1.4 cells were transfected with control siRNA or SKP2 siRNA and 24 h later IFN-β (500 pM) was added and cells were incubated for various times. Lysates were analysed as indicated. USP18 and ISG15 levels were quantified relative to actin levels.