Case Report: Novel SAVI-Causing Variants in STING1 Expand the Clinical Disease Spectrum and Suggest a Refined Model of STING Activation

Abstract

Gain-of-function mutations in STING1 cause the monogenic interferonopathy, SAVI, which presents with early-onset systemic inflammation, cold-induced vasculopathy and/or interstitial lung disease. We identified 5 patients (3 kindreds) with predominantly peripheral vascular disease who harbor 3 novel STING1 variants, p.H72N, p.F153V, and p.G158A. The latter two were predicted by a previous cryo-EM structure model to cause STING autoactivation. The p.H72N variant in exon 3, however, is the first SAVI-causing variant in the transmembrane linker region. Mutations of p.H72 into either charged residues or hydrophobic residues all led to dramatic loss of cGAMP response, while amino acid changes to residues with polar side chains were able to maintain the wild type status. Structural modeling of these novel mutations suggests a reconciled model of STING activation, which indicates that STING dimers can oligomerize in both open and closed states which would obliviate a high-energy 180° rotation of the ligand-binding head for STING activation, thus refining existing models of STING activation. Quantitative comparison showed that an overall lower autoactivating potential of the disease-causing mutations was associated with less severe lung disease, more severe peripheral vascular disease and the absence of a robust interferon signature in whole blood. Our findings are important in understanding genotype-phenotype correlation, designing targeted STING inhibitors and in dissecting differentially activated pathways downstream of different STING mutations.

Keywords: SAVI; STING; autoinflammatory disease; interferonopathy; pediatrics; type I interferon; whole exome sequencing.

Figure 1

Three novel variants in STING1 cause SAVI. (A–D) Erythematous lesions of the hands and tapered fingers in Patient 1 (A, B), Patient 2 (C) and Patient 4 (D). (E–G) Erythematous rash on the feet and tapered toes in Patient 1 (E), with similar lesions and tissue loss of the feet in Patient 2 (F) and Patient 4 (G). Patient number is depicted on the bottom right corner of each picture. (H) Pedigree charts. Solid symbols represent affected individuals; open symbols, unaffected individuals. P1-P5: patients, WT: wild type. (I) Exon locations of all SAVI-causing STING1 variants. Open boxes represent non-coding exons, solid boxes represent coding exons, while lines represent introns. Previously reported variants are in black and the three novel variants in red. Transcript diagram was obtained from ENSEMBL (www.ensembl.org). (J) 28-gene Interferon (IFN) scores from whole blood (WB) and peripheral blood mononuclear cells (PBMCs). Solid symbols indicate WB and open symbols indicate PBMCs. Healthy control (HC) samples included 19 WB and 10 PBMCs. SAVI patients with N154S mutation (n=3) in black, family members with H72N mutation (n=3) in red, and patient with G158A mutation (n=1) in blue. Paired PBMC and WB IFN scores from the same time point are connected by a line. Each patient with SAVI is represented by a different symbol.

Figure 2

H72N, G158A, F153V variants lead to autoactivation. (A) Novel STING variants lead to strong STING autoactivation as measured by IFNB1 reporter activation in 293T cells. 50 pg of various STING constructs were transfected. The three novel SAVI-causing variants are labeled in blue, and other H72 variants are labeled in green. The SAVI-causing variants are shown in the order of autoactivating potential. *An asterisk above the white bar designating 0 μg/ml cGAMP indicates significant change compared (p<0.0001) to wildtype (WT) non-treated, by ordinary one-way ANOVA Dunnett’s multiple comparisons tests. An asterisk above the black bar designating 20 μg/ml cGAMP indicates significant change (p<0.003) of 20 μg/ml cGAMP treated cells compared to the respective non-treated cells; by two-way ANOVA, Bonferroni’s multiple comparisons test. Data are presented as mean±SD of duplicates. Similar results were obtained in independent experiments. (B) Ribbon structural model of apo STING dimer, with the two subunits colored in yellow and green. Location of SAVI-causing variants are highlighted with spheres, with the three novel variants in red, variants in the connector helix loop region in blue, p.G166E variants in the ligand-binding pocket in cyan, and variants in the polymer interface in purple. The major mutation region is enlarged on the right, showing how the F153 phenyl ring buttress the polymer interface. (C) Structure model of H72 compared to N72 indicates increased dimer flexibility of p.H72N mutation, which confers gain-of-function. The two STING monomers forming a dimer are in green and gold/purple respectively.

Figure 3

A hypothetical STING activation model reconciles current structural models. (A) Wild type apo STING forms a dimer in open conformation, with C-terminal tail blocking the polymer interface and preventing polymerization. cGAMP binding leads to a 180° rotation and a closed dimer, which also changes the polymer interface conformation and subsequently loses C-terminal tail blocking. This leads to STING polymerization via side-by-side packing and TBK1 recruitment. (B) SAVI mutant G158A “pushes” the formation of a closed dimer, which mimics ligand-bound STING and confers autoactivation in the absence of ligand binding. (C) Some other SAVI mutants like R284G, R284S, H72N, F153V directly affect the polymer interface conformation, which leads to release of C-terminal tail binding. The STING dimer remains in the open state, but the polymer interfaces are free of C-terminal tail blocking, which allows polymerization and autoactivation.