The patient, available family members, and age-matched controls were included in this study. The project was approved by our Institutional Review Board and conducted according to the Declaration of Helsinki Principles. Informed consent was obtained from the parents to participate in this study.

Due to the financial issue, whole-exome sequencing was not performed. Instead, targeted next-generation sequencing, including library construction, capture, and sequencing, was carried out at MyGenostics Gene Technologies (Beijing, China) using genomic DNA isolated from the whole blood of the patient and his parents. The panel was designed to cover the genes known to cause immune diseases.

All structure models were visualized using the molecular graphics program PyMol. The STAT3 structure (PDB ID: 6TLC) was obtained from the Research Collaboratory for Structural Bioinformatics Protein Data Bank (http://www.rcsb.org/). All heteroatoms were removed, and chain A was kept as a STAT3 wild type (WT) monomer. The STAT3 P330R structure was mutated using the PyMol mutagenesis tool. Hydrophobic surfaces were colored based on a normalized hydrophobicity scale using the Color_h script (https://pymolwiki.org/index.php/Color_h). The electrostatic potential was calculated using the PyMol vacuum electrostatics program.

Molecular-dynamics (MD) simulations were carried out using Gromacs 5.0.7 on the STAT3^WT and STAT3^P330R monomer structures. The Amber ff14SB force field was used to model the protein, and the TIP3P water model was used to describe the solvent. Then, the system was energy minimized and equilibrated using first NPT and then NVT, followed by a 50 ns equilibrium simulation.

The patient was hospitalized in the inpatient ward and treated with steroid, hydroxychloroquine and antimicrobial agents. Peripheral blood was collected and peripheral blood mononuclear cells (PBMCs) were isolated by density centrifugation. The cells were resuspended in RPMI 1640 supplemented with penicillin, streptomycin, and L-glutamine along with 10% fetal calf serum. PBMCs were then stimulated for 15 min with 20 ng/ml IL-6 (Peprotech) or not. Following stimulation, cells were immediately fixed with PhosFlow Fix Buffer I (BD Biosciences) for 10 min at 37°C, then permeabilized using BD Phosflow Perm Buffer III (BD Biosciences) following the manufacturer’s instruction. Cells were then stained with anti-STAT3 (pY705) or isotype-matched control antibodies (BD Biosciences) for 1 hour. Samples were collected with FACSCanto II (BD Biosciences) and analyzed with FlowJo Software (TreeStar).

The human STAT3 coding sequence was cloned from cDNA isolated from the whole blood using reverse transcription PCR. The primer pair was forward: 5’-cttggtaccgagctcggatccgccaccATGGCCCAATGGAATCAGCTA-3’; reverse: 5’-gaagggccctctagactcgagCATGGGGGAGGTAGCGCA-3’. The PCR product was then ligated into the pcDNA3.1-3×Flag-C vector between the BamHI and XhoI restriction sites using ClonExpress II One Step Cloning Kit (Vazyme). The mutation was introduced using Mut Express II Fast Mutagenesis Kit V2 (Vazyme) following the manufacturer’s instructions. The mutagenesis primer pair was forward 5’-CCTGCATGCgCATGCATCCTGACCGGCCCCTCGTC-3’; reverse 5’-AGGATGCATGcGCATGCAGGGCTGCCGCTCCACCAC-3’.

Human embryonic kidney 293 (HEK293) cells (ATCC) were transfected with the WT or mutant STAT3 plasmid or empty vector. The transfected cells were stimulated with IL-6 (20 ng/mL) for 15, 30, or 60 min. Total STAT3 protein (Cell Signaling Technology), phospho-STAT3 (Y705) (Cell Signaling Technology) and GAPDH (Proteintech) were measured by western blot. For dephosphorylation experiments, the transfected HEK293 cells were stimulated with IL-6 (20 ng/mL) for 30min and then harvested at different times (15, 30, 60, or 120 min) after IL-6 withdrawal.

HEK293 cells were transiently co-transfected with indicated STAT3 expression plasmid and a STAT3-responsive firefly luciferase reporter plasmid (Beyotime) and a renilla luciferase reporter plasmid (Beyotime) using Lipofectamine 3000 reagent (Invitrogen) according to the manufacturer’s protocol. Twenty-four hours after the transfection, the cells were not or treated with 20 ng/mL IL-6 for 16 hours. The luciferase activity was measured using the Dual-Glo Luciferase Assay System (Promega).

PBMCs were stimulated with 50 ng/ml phorbol myristate acetate (PMA) plus 500 ng/ml ionomycin for five hours in the presence of 1 µl/ml GolgiStop (BD Biosciences). The cells were then stained with anti-CD3 (BioLegend) and anti-CD8 (BD Biosciences) monoclonal antibodies for 30 min. Cells were fixed with Fixation and Permeabilization Solution (BD Biosciences) for 30 min at 4°C and permeabilized with Perm/Wash Buffer (BD Biosciences) for 20 min at room temperature. Cells were then stained for 30 min with anti-IL-17A antibody (BioLegend). Samples were collected with FACSCanto II and analyzed with FlowJo Software.

For Treg cells, PBMCs were first stained with anti-CD3 antibody (BioLegend), then fixed and permeabilized. Cells were then stained with anti-FOXP3 antibody (BD Biosciences) for 1 hour before washing and acquisition. The T lymphocyte subpopulations were detected with the antibodies against CD3 (BioLegend), CD4 (BioLegend), CD8 (BioLegend), CD45RA (BD Biosciences) and CD27 (BD Biosciences). Cells acquisition was performed using a FACSCanto II flow cytometer, and the data were analyzed using FlowJo software.

HEK293 cells were transfected with the WT or mutant STAT3 plasmid or empty vector. Forty-eight hours after the transfection, the cells were incubated for 4 hours in the presence of ruxolitinib (Abmole Bioscience), tofacitinib (Abmole Bioscience), stattic (Abmole Bioscience), or dimethyl sulfoxide (Sigma-Aldrich) before being activated with IL-6 (20 ng/mL) or not.

The literature search was conducted in the PubMed, Web of Science, and Cochrane Library databases for relevant articles published until January 11, 2022, to identify studies reporting STAT3 GOF germline mutations-related ILDs. All the search terms were Medical Subject Heading terms, including STAT3 terms (“STAT3”, “signal transducer and activator of transcription 3”) and gain of function terms (“gain of function”, “activating germline mutation” and “activating mutation”). Additional studies were manually screened from citations of relevant articles. All selected articles were reviewed by abstract or full-text. Review articles, duplicate articles, and meeting abstracts were excluded.

Results were analyzed with the Student t-test using GraphPad Prism software, and p < 0.05 was considered statistically significant.

The patient was a 1-year-old boy of non-consanguineous Chinese parents. He was born at full-term with uneventful perinatal history. There was no family history of lung disease or primary immunodeficiency. He had no adverse reactions to vaccination with bacille Calmette-Guérin and hepatitis B. At 4 months of age, he was brought to the hospital for eczematous dermatitis ( Figure 1A ), mild cough, and a lump behind the left ear. His respiratory symptoms gradually worsened, and he developed respiratory distress and hypoxemia with possible cytomegalovirus (CMV) infection as the triggering event. Bronchoalveolar lavage and blood tests for CMV DNA were positive. He also suffered from repeated respiratory infections with positive sputum cultures of Acinetobacter baumanni, Klebsiella pneumonia, Pseudomonas aeringinosa and Stentophomonas maltophilia, and even twice respiratory failures requiring intubation and mechanical ventilation. Multiple courses of antibiotics have been required since then. He responded to the antibiotics with decreased inflammatory markers over 2-3 weeks, but fever recurred while on broad-spectrum antibiotics. At 6 months old, chest computed tomography suggested interstitial lung disease with diffuse ground-glass opacity ( Figure 1B ). The histology of lung biopsy showed chronic interstitial lung disease with diffuse thickening of alveolar membranes and hyperplastic type 2 pneumocytes ( Figure 1C ). The alveolar membranes contained some stromal cells and mild to moderate infiltrate of lymphocytes, some plasma cells, and macrophages. There was no proteinosis nor significant established interstitial fibrosis. Microbiological investigations of Lung biopsy including bacterial culture, fungal culture, acid-fast bacillus staining and PCR, and CMV were negative. Bronchoalveolar lavage pneumocystis pneumonia staining, blood tests for CMV, Epstein-Barr virus, human immunodeficiency virus, β-glucan and QuantiFERON-TB were negative. He was treated empirically with prednisolone (1.8/mg/kg/day) with gradual tapering according to the disease status. Hydroxychloroquine (6 mg/kg/day) was also added. The respiratory symptoms were partially improved. When chest symptoms were relatively stable, he also exhibited recurrent diarrhea with elevated white blood cells and neutrophils ( Table 1 ) and one episode of stool culture positive for Clostridium difficile. At the age of 11months, he experienced increasing diarrhea, poor feeding, and septic shock with metabolic acidosis and elevated inflammatory markers. Unfortunately, he eventually died of respiratory failure at 1 year old.

Based on the clinical phenotype, congenital lung disease was considered for the patient. Genetic tests were done for the known related genes including SFTPC, SFTPB, ABCA3, CSF2RA, CSF2RB, NKX2.1, FOXF1 and SLC7A7, but no meaningful mutation was revealed. Therefore, the next-generation sequencing was performed for the patient and his parents. No severe combined immunodeficiency-related genetic mutation was found. Instead, a de novo heterozygous missense mutation of the STAT3 gene (c.989C>G, p.P330R) was identified in the patient ( Figures 2A, B ). The mutation was not found in the gnomAD (v2.1.1), ClinVar or Human Gene Mutation Database. This mutation was located within the DNA-binding domain (DBD), resulting in a proline to arginine substitution at position 330 ( Figure 2C ). The arginine residue was not observed at this position in homologous sequences ( Figure 2D ), suggesting highly conserved among different species.

Structurally, the proline is very rigid and therefore may induce a required exceptional backbone conformation at this position. To evaluate the impacts of this mutation on STAT3 structure, in silico studies were performed. The mutant residue lay on the surface of the STAT3 protein ( Figure 2E ) and decreased the hydrophobic interactions in this region ( Figure 2F ). We next performed the MD simulation of the WT and P330R STAT3 to estimate the conformational changes. The root-mean-square deviation (RMSD) for the P330R system showed higher fluctuations than the WT system ( Figure 2G ), suggesting an unstable protein structure. Beyond that, according to the electrostatic potential calculation, the charge changed from neutral (P) to positive (R) ( Figure 2H ), which may increase the interaction with negatively charged nucleic acids. These results strongly indicated that the mutation might alter the STAT3 structure and function.

The functional alterations of mutant STAT3 were investigated. STAT3 can be activated via phosphorylation upon different cytokine stimulations, such as IL-6. It was demonstrated that, upon IL-6 stimulation, the patient cells have much higher phosphorylation of STAT3 compared to normal controls. In addition, the mutation also activated STAT3 under the steady-state ( Figure 3A ). Next, we further performed functional validation by transfecting either WT or mutant constructs into HEK293 cells. Like the findings in patient cells, the phosphorylated STAT3 of mutant STAT3-transfected cells was significantly enhanced than that of the WT after IL-6 stimulation. The STAT3^P330R mutant also displayed high phosphorylation in the basal condition ( Figure 3B ), suggesting constitutive activation. Besides, the P330R mutation also impaired the dephosphorylation of phosphorylated STAT3 on IL-6 deprivation ( Figure 3C ).

To detect the transcriptional activity of the P330R mutation, a STAT3-driven luciferase reporter assay was performed. We observed enhanced transcriptional activity of STAT3^P330R under the steady-state ( Figure 3D ), further suggesting constitutive activation of mutant STAT3. The mutant also showed a severalfold increase in the luciferase activity compared with STAT3^WT after being stimulated with IL-6 ( Figure 3E ). Collectively, these results suggested that the STAT3^P330R represented a GOF mutation leading to both increased phosphorylation and transcriptional activity of STAT3.

STAT3 plays an irreplaceable role in regulating the immune signaling pathways, including inhibiting Tregs and enhancing Th17 cell determination (4). Given this, we investigated whether the P330R mutation would cause Th17 and Treg cells dysregulation. PMA/ionomycin-activated PBMCs of the patient showed increased IL-17A levels in the T cells compared to the normal controls ( Figures 4A, B ), suggesting the over-activation of the Th17 pathway. However, the percentage of Treg cells was within a normal range in the patient ( Figure 4C ). The levels of immunoglobulin G, A and M were not reduced. Lymphocyte subpopulation analysis indicated decreased NK cell percentage but normal T and B cell levels in the patient’s PBMCs ( Table 1 ). The proportions of effector memory (EM) CD4⁺ T, EM CD8⁺ T, naive CD4⁺ T and naive CD8⁺ T cells were also normal ( Figure 4D ). Taken together, these data suggested that the STAT3^P330R gain-of-function mutation would contribute to activating the Th17 pathway.

STAT3 is activated by JAK-mediated phosphorylation, so targeting upstream JAK kinases could be a potential approach to inhibit STAT3 activation. We thus performed functional experiments to determine whether JAK inhibitors could suppress STAT3^P330R activation. It was found that both ruxolitinib and tofacitinib significantly reduced the phosphorylation of mutant STAT3 induced by IL-6 stimulation ( Figure 5A ). Given the broad suppressive effect of JAK inhibitors, we sought to test the ability of a direct STAT3 selective inhibitor (Stattic) in pharmacologic inhibition of STAT3 GOF mutation. Similar to JAK inhibitors, the STAT3 inhibitor also markedly suppressed the phosphorylation of mutant STAT3 in the steady-state, as well as in IL-6-stimulated condition ( Figures 5B, C ). In addition, the inhibitory effect was dose-dependent. The more amount of inhibitor used, the more inhibition of STAT3^P330R phosphorylation seen. These data indicated that selective inhibition of STAT3 represents a potential therapeutic approach for STAT3 GOF syndrome.

Here we summarized current literature on the clinical features ( Table 2 ) and treatments ( Table 3 ) of STAT3 GOF patients with ILD. Twenty patients were included, with 9 males, 10 females, and 1 gender not mentioned. In addition to ILD, all the patients also have other autoimmune disorders or lymphoproliferation. Eleven of 20 (55%) ILD patients had mutations in the DBD ( Figure 1C ). The median age at onset of ILD was 3 years based on the available information. One patient developed ILD at 3 months old, and prior to this, he was diagnosed with type I diabetes at 3 weeks of age. The most frequent lung disease was lymphocytic interstitial pneumonia (7/20, 35%), followed by desquamative interstitial pneumonia (3/20, 15%), nonspecific interstitial pneumonia (3/20, 15%) and interstitial pneumonia (3/20, 15%). The most common lung symptoms were dyspnea or tachypnea (7/20, 35%) and infections (6/20, 30%). Asthma was the primary manifestation before the diagnosis of ILD.

As for the treatment, seven patients received targeted therapies: one received ruxolitinib only, two received tofacitinib only, and four received tocilizumab combined with ruxolitinib or tofacitinib. Only one patient died of respiratory failure, and a significant improvement was observed in the other six patients. Besides, the JAKs inhibitor was effective in patients who failed tocilizumab monotherapy.