Haploinsufficiency for Human ABCF1 Underlies Gastrointestinal Autoimmunity

While studying familial celiac disease, we identified four individuals with a rare heterozygous loss-of-function variant in the highly conserved, early-evolved ABCF1 gene. In addition, the 246 individuals heterozygous for ABCF1 loss-of-function variants in the All of Us, UK Biobank, and BioMe cohorts had a higher risk of autoimmune diseases, including 10% with inflammatory bowel disease. Peripheral blood mononuclear cells (PBMC) from ABCF1-deficient patients had high levels of constitutive interferon-stimulated gene expression and enhanced cellular responses to type I interferon and cytokines. We show here that ABCF1 plays an essential role in regulating the JAK–STAT pathway in humans and mice. ABCF1 binds to ribosomal proteins and is involved in liquid–liquid phase separation via its low-complexity domain.Activation of the stress responses can lead to the trafficking of both STAT1 and ABCF1 into stress granules, and this sequestration eventually inhibits STAT1 phosphorylation. Finally, Abcf1-deficient mice displayed more severe inflammatory responses and tissue damage in response to dextran sulfate sodium challenge. Gastrointestinal inflammation in Abcf1-deficient mice is driven primarily by increased infiltration of myeloid cells, which can be treated with tofacitinib, a JAK kinase inhibitor. We have, thus, discovered a new genetic etiology of autoimmunity and gastrointestinal inflammation, and suggest a targeted treatment for individuals with ABCF1 haploinsufficiency and chronic inflammation.Figure 1.A rare pLoF variant of ABCF1 disrupts cellular responses to IFN-α and multiple pathways in a family including multiple individuals with celiac disease. a. Pedigree of a family containing three patients (P1, P2, and P3) diagnosed with CeD in childhood. The three patients were found to be heterozygous for an ABCF1 mutation (E243X). The mother is also a carrier but has a mild phenotype. HLA-DQ genotype is indicated. b. Sanger sequencing electropherogram. c. A phenotype-association study was performed on 246 individuals heterozygous for an ABCF1 pLoF variant found in the UK Biobank, All of Us, and BioMe databases. Results of RNA sequencing on PBMCs from two wild-type healthy controls (HCs) and four patients (Pts) (d–h). d.ABCF1 mRNA level. e. Constitutive and IFN-α-stimulated mRNA levels for CXCL10, GBP1, RSAD2, and IFIT2. f. Heatmap of ISGs differentially expressed in the presence and absence of IFN-α stimulation. In the bubble plot on the right, circle size indicates the fold-change difference, and colors indicate the adjusted p-value. g. Volcano plot of genes displaying differential constitutive expression between patients and controls. h. Gene set enrichment analysis (GSEA) of genes displaying differential constitutive expression (at least a two-fold change and adjusted p value < 0.001). Unstim.: unstimulated; IFN-α: 40 ng/mL interferon-alpha; LPS: 1 μg/mL lipopolysaccharide. TPM: Transcripts per million. Log2FC: log2 fold-change, padj: p-values adjusted by the false discovery rate (FDR) procedure. P values were calculated in unpaired one-tailed Student’s t tests (d and e). ns, not significant; **p < 0.01; and ****p < 0.0001.