Somatic mosaicism and common genetic variation contribute to the risk of very-early-onset inflammatory bowel disease

Abstract

Very-early-onset inflammatory bowel disease (VEO-IBD) is a heterogeneous phenotype associated with a spectrum of rare Mendelian disorders. Here, we perform whole-exome-sequencing and genome-wide genotyping in 145 patients (median age-at-diagnosis of 3.5 years), in whom no Mendelian disorders were clinically suspected. In five patients we detect a primary immunodeficiency or enteropathy, with clinical consequences (XIAP, CYBA, SH2D1A, PCSK1). We also present a case study of a VEO-IBD patient with a mosaic de novo, pathogenic allele in CYBB. The mutation is present in ~70% of phagocytes and sufficient to result in defective bacterial handling but not life-threatening infections. Finally, we show that VEO-IBD patients have, on average, higher IBD polygenic risk scores than population controls (99 patients and 18,780 controls; P < 4 × 10^-10), and replicate this finding in an independent cohort of VEO-IBD cases and controls (117 patients and 2,603 controls; P < 5 × 10^-10). This discovery indicates that a polygenic component operates in VEO-IBD pathogenesis.

Fig. 1. Analysis of CYBB mosaicism in a male patient.

a Pedigree structure for the family of the male patient with the mosaic hemizygous mutation in CYBB (chrX:37,663,371A/G; p.W380X). b Sanger sequencing of the chrX:37,663,371 CYBB mutation site in the patient and unaffected relatives (sister and mother). c p91-phox protein expression (the gene product encoded by CYBB) analysed by flow cytometry assay (FACS). Control is a healthy donor. d Measurement of oxidative burst in neutrophils and monocytes using the dihydrorhodamine-1,2,3 (DHR) assay. Obtained from the patient and a healthy donor (control). e Defective bacterial handling in monocyte derived macrophages with the CYBB mosaicism. Intracellular survival of Salmonella typhimurium was quantified using the agar plate technique. Results show three technical replicates. Obtained from the patient and a healthy donor (control). f, g Quantification of mutant read proportion at chrX:37,663,371 using the IGV browser. PBMCs were sorted into immune cell subsets (Supplementary Figs. 6B,C) and compared with buccal swabs and hair follicles, as well as with healthy donor immune cells and a HEK293T cell line as technical control. h FACS sorting strategy for DHR-high and DHR-low populations following DHR staining and PMA stimulation (Supplementary Fig. 6A). i Quantification of mutant reads at chrX:37,663,371 following sorting based on DHR for control DHR-high, patient DHR-high, and patient DHR-low neutrophils (Supplementary Fig. 6A). j Gentamicin protection assay on neutrophils for control DHR-high, patient DHR-high, and patient DHR-low populations (Supplementary Fig. 6A). Briefly, neutrophils were infected at a MOI 1:10 for 45 min with Salmonella enterica serovar typhimurium and subsequently treated with gentamicin for 45 min. Neutrophils were then lysed and plated on LB agar plates for CFU counting on the following day. ***p < 0.001, Mann–Whitney U-test.

Fig. 2. Functional validation of pathogenic variants identified in monogenic IBD genes.

a Defective MDP response in a patient with hemizygous XIAP p.R222X. MDP (muramyl dipeptide) induced intracellular TNF response was determined using FACS. b Absent SAP staining (gene product of SH2D1A) as indicated by C-terminal antibody in a patient with hemizygous SH2D1A p.R75X. Measured with fluorescence-activated cell sorting (FACS). c Defecting ROS production in neutrophils from patients with homozygous CYBA p.S118N variants. Dihydrorhodamine-1,2,3 (DHR) flow cytometry assay (FACS) was performed to measure NADPH oxidase activity in response to PMA, E. coli particles and formylpeptide.

Fig. 3. Distribution of CD and UC risk scores in VEO-IBD, CD, UC cases and healthy controls.

The CD score was calculated using 147 CD risk alleles and the UC score using 119 UC risk alleles. Both scores were generated for a discovery cohort comprising 99 VEO-IBD cases (VEO-COLORS), 7578 CD cases, 6318 UC cases, 18,780 UK population controls (from the UKIBDGC), all of European ancestry. The replication cohort comprised 117 VEO-IBD cases (VEO Toronto) and 2603 population controls (from the NIDDK Genetics Consortium). The CD and UC risk scores did not significantly differ between the two VEO-IBD cohorts (CD P = 0.98; UC P = 0.64). The Student’s t-test was used in group comparisons.