The Role of Interferon-γ in Autoimmune Polyendocrine Syndrome Type 1

Abstract

Background: Autoimmune polyendocrine syndrome type 1 (APS-1) is a life-threatening, autosomal recessive syndrome caused by autoimmune regulator (AIRE) deficiency. In APS-1, self-reactive T cells escape thymic negative selection, infiltrate organs, and drive autoimmune injury. The effector mechanisms governing T-cell-mediated damage in APS-1 remain poorly understood.

Methods: We examined whether APS-1 could be classified as a disease mediated by interferon-γ. We first assessed patients with APS-1 who were participating in a prospective natural history study and evaluated mRNA and protein expression in blood and tissues. We then examined the pathogenic role of interferon-γ using Aire^-/-Ifng^-/- mice and Aire^-/- mice treated with the Janus kinase (JAK) inhibitor ruxolitinib. On the basis of our findings, we used ruxolitinib to treat five patients with APS-1 and assessed clinical, immunologic, histologic, transcriptional, and autoantibody responses.

Results: Patients with APS-1 had enhanced interferon-γ responses in blood and in all examined autoimmunity-affected tissues. Aire^-/- mice had selectively increased interferon-γ production by T cells and enhanced interferon-γ, phosphorylated signal transducer and activator of transcription 1 (pSTAT1), and CXCL9 signals in multiple organs. Ifng ablation or ruxolitinib-induced JAK-STAT blockade in Aire^-/- mice normalized interferon-γ responses and averted T-cell infiltration and damage in organs. Ruxolitinib treatment of five patients with APS-1 led to decreased levels of T-cell-derived interferon-γ, normalized interferon-γ and CXCL9 levels, and remission of alopecia, oral candidiasis, nail dystrophy, gastritis, enteritis, arthritis, Sjögren's-like syndrome, urticaria, and thyroiditis. No serious adverse effects from ruxolitinib were identified in these patients.

Conclusions: Our findings indicate that APS-1, which is caused by AIRE deficiency, is characterized by excessive, multiorgan interferon-γ-mediated responses. JAK inhibition with ruxolitinib in five patients showed promising results. (Funded by the National Institute of Allergy and Infectious Diseases and others.).

Figure 1 (facing page).. Interferon-γ–Mediated Responses in Blood from Patients with APS-1.

Panels A through C show proximity extension assay–based surveys of 180 proteins, performed with serum from 35 patients with autoimmune polyendocrine syndrome type 1 (APS-1) and 40 healthy donors. Panel A is a heat map showing expression of each of the 58 differentially expressed proteins (DEPs) in each person. Red and blue indicate high and low expression, respectively; expression (expressed as the z score) ranges from −4 to 4. Panel B is a volcano plot comparing the serum proteins detected in patients with APS-1 as compared with healthy donors. Red data points are DEPs — that is, proteins with a log₂ ratio of expression (patients:healthy donors) of greater than 0.4 or less than −0.4 (i.e., falling outside the area of the graph delineated by the two gray vertical lines) at a false discovery rate (FDR) of less than 0.05 (i.e., falling above the gray horizontal line). Select interferon-γ–regulated genes are labeled. Panel C shows the top five enriched Molecular Signatures Database “hallmark” gene sets in DEPs from Panel A. Colors indicate the –log₁₀ FDR value, and the size of the data point indicates the gene count. The x axis shows the factor by which expression was higher in patients (e.g., 30 indicates 30 times as high in patients as in healthy donors). NF-κB denotes nuclear factor κB, and TNF-α tumor necrosis factor α. Panel D shows levels of interferon-γ and interferon-γ–inducible chemokines CXCL9 and CXCL10 in the serum of 28 to 86 patients with APS-1 and 20 to 47 healthy donors. Panel E shows expression of 15 interferon-γ–regulated genes in peripheral-blood mononuclear cells (PBMCs) of 20 patients with APS-1 and 8 healthy donors, measured by NanoString and expressed as z scores. Long horizontal bars indicate the means, and I bars indicate the standard errors. The numbers above each graph are median differences and 95% confidence intervals derived from the Mann–Whitney test. The confidence intervals were not adjusted for multiplicity and should not be used in place of hypothesis testing.

Figure 2 (facing page).. T-Cell Infiltration and Interferon-γ Pathway Signals in Multiple Autoimmunity-Affected Tissues of Patients with APS-1.

Panel A shows representative images of hematoxylin and eosin (H–E) staining, immunohistochemical staining with CD4 (brown) and CD8 (red), interferon-γ in situ hybridization staining, phosphorylated signal transducer and activator of transcription 1 (pSTAT1) immunohistochemical staining, and in situ hybridization staining for the interferon-γ–inducible chemokine CXCL9 in endobronchial biopsy tissue of a patient with APS-1 with autoimmune pneumonitis, liver tissue of a patient with APS-1 with autoimmune hepatitis, stomach tissue of a patient with APS-1 with autoimmune gastritis, and ileal tissue of a patient with APS-1 with autoimmune enteritis. Scale bars in Panel A are as follows: 20 μm for interferon-γ images of lung, liver, and stomach and for the pSTAT1 image of liver; 50 μm for the H–E image of lung, CD4 and CD8 images of lung and liver, interferon-γ image of ileum, pSTAT1 image of lung, and CXCL9 images of lung and liver; 100 μm for H–E images of liver and ileum, CD4 and CD8 images of ileum, pSTAT1 images of stomach and ileum, and CXCL9 image of ileum; and 200 μm for H–E, CD4 and CD8, and CXCL9 images of stomach. Panel B shows representative images of in situ hybridization staining for interferon-γ and CXCL9 in stomach, duodenal, and ileal tissues of healthy donors. Scale bars in Panel B indicate 100 μm. In both panels, insets show magnified images of the area delineated by white rectangles.

Figure 3.. Survival and Organ Damage among Aire^−/−Ifng^−/− and Aire^−/− Mice with or without Ruxolitinib Treatment.

The graph in Panel A shows survival among Aire^−/− and Aire^−/−Ifng^−/− mice (20 per group). Representative hematoxylin and eosin–stained lung sections and pathology scores of the indicated tissues in 10-week-old Aire^−/− and Aire^−/−Ifng^−/− mice (27 to 31 per group) are also shown; higher pathology scores indicate more severe pathologic changes. The graph in Panel B shows survival among 8 untreated (control) and 26 ruxolitinib-treated Aire^−/− mice. Representative hematoxylin and eosin–stained lung sections and pathology scores of the indicated tissues in 10-week-old Aire^−/− control and ruxolitinib-treated mice (8 to 10 per group) are also shown. The scale bars in both panels indicate 1 mm. In both panels, the hazard ratios and 95% confidence intervals in the Kaplan–Meier curves were derived from the Mantel–Haenszel test, the T bars in the bar graphs indicate the standard errors, and the numbers above each bar in the bar graphs are median differences and 95% confidence intervals derived from the Mann–Whitney test. Confidence intervals were not adjusted for multiplicity and should not be used in place of hypothesis testing.

Figure 4.. Ruxolitinib-Induced Remission of Autoimmune Manifestations in Patients 1 and 2.

Panel A shows photographs of hair on the scalp of Patient 1 before and after ruxolitinib treatment. The graph shows serum levels of interferon-γ before and after treatment. Panel B shows the macroscopic appearance of gastric mucosa during endoscopy in Patient 2. The graph shows serum levels of interferon-γ before and after treatment.

Figure 5.. Ruxolitinib-Induced Remission of Autoimmune Manifestations in Patients 3, 4, and 5.

Panel A shows photographs of hair on the scalp of Patient 3 before and after ruxolitinib treatment. The graph shows the patient’s change in weight after treatment. Panel B shows graphs of the number of bowel movements per day and body weight before and after treatment in Patient 4. Panel C shows graphs of body weight and serum levels of interferon-γ before and after treatment in Patient 5.