Deletion polymorphism upstream of IRGM associated with altered IRGM expression and Crohn's disease

Abstract

Following recent success in genome-wide association studies, a critical focus of human genetics is to understand how genetic variation at implicated loci influences cellular and disease processes. Crohn's disease (CD) is associated with SNPs around IRGM, but coding-sequence variation has been excluded as a source of this association. We identified a common, 20-kb deletion polymorphism, immediately upstream of IRGM and in perfect linkage disequilibrium (r2 = 1.0) with the most strongly CD-associated SNP, that causes IRGM to segregate in the population with two distinct upstream sequences. The deletion (CD risk) and reference (CD protective) haplotypes of IRGM showed distinct expression patterns. Manipulation of IRGM expression levels modulated cellular autophagy of internalized bacteria, a process implicated in CD. These results suggest that the CD association at IRGM arises from an alteration in IRGM regulation that affects the efficacy of autophagy and identify a common deletion polymorphism as a likely causal variant.

Figure 1

A common, 20-kb deletion polymorphism upstream of IRGM. (a) Hybridization of DNA from 90 HapMap YRI samples to the Affymetrix SNP 6.0 array reveals a correlated pattern of variation in intensity across six copy-number probes spanning the region upstream of IRGM, suggesting the existence of a common copy-number polymorphism. (Darker shades of red represent reduced hybridization intensity.) Quantitative PCR indicated that the copy-number-variable region was an insertion/deletion (Supplementary Fig. 1). (b) Mapping of the deletion breakpoints by microarray and PCR. Red indicates evidence for deletion; black indicates evidence to the contrary; blue arrows indicate locations of flanking PCR primers able to generate a PCR product in individuals with at least one deletion allele. (c) Sequence of the reference and deletion alleles. The CD-associated SNP rs13361189 is indicated in boldface red type. Physical coordinates are on the reference human genome (build 35/36); allele frequencies are for extended HapMap analysis panels, including 540 samples. (d) CD association of typed and imputed polymorphisms in the IRGM region in the NIDDK IBDGC genome scan,. Blue trace indicates recombination map. The deletion polymorphism and rs13361189 are indicated by red arrows. rs13361189 (which is in perfect linkage disequilibrium with the deletion, r² = 1.0) was also the most strongly associated SNP in the combined WTCCC genome scan and replication study.

Figure 2

Differential expression of IRGM from the deletion (CD risk) and reference (CD protective) haplotypes. (a) The exonic SNP rs10065172, which is in strong linkage disequilibrium with the 20-kb insertion/deletion polymorphism upstream of IRGM, can be used to distinguish IRGM transcripts that arise from the deletion haplotype from IRGM transcripts that arise from the reference haplotype. This makes it possible to measure the relative expression of the two haplotypes in heterozygotes by measuring the relative abundance of the two rs10065172 alleles in cDNA. (Yellow and green rectangles indicate transcribed sequence; green rectangles indicate protein-coding sequence.) (b–g) Differential expression of IRGM from the two structural haplotypes in heterozygotes. The relative abundance of the C and T alleles of rs10065172 in cDNA (colored circles) and genomic DNA (black squares) from human tissues and cell lines were measured. Gray squares indicate control measurements for genomic DNA from 48 HapMap YRI individuals, identifying the three reference genotype classes CC, CT and TT. Genomic DNA from all other samples was heterozygous for the C and T alleles (black squares); HeLa genomic DNA (b,c) showed a 2:1 allelic ratio (CCT) reflecting HeLa’s triploid 5q karyotypic status. In e, cDNA and genomic DNA from lymphoblastoid cell lines from ten different heterozygous individuals were analyzed: the ten different colors represent the ten individuals; technical replicates from the same cell line are connected by line segments. HeLa, SNU182 and the lymphoblastoid cell lines expressed the C allele of rs10065172 more strongly than the T allele (b–e); HCT116 and primary bronchial cells expressed the T allele much more strongly than the C allele (f,g).

Figure 3

IRGM expression levels affect the autophagy of Salmonella typhimurium in human epithelial cells. (a) siRNA constructs directed at IRGM reduced endogenous IRGM transcripts by six- to eightfold in HeLa cells. Cells were transfected with control (siControl) or IRGM-targeted (siIRGM1, siIRGM2) siRNA duplexes and assayed 48 h later by quantitative RT-PCR. Error bars, s.d. (b) IRGM knockdown reduces the efficiency of anti-bacterial autophagy. HeLa cells stably expressing LC3-GFP (a marker for autophagic vesicles) were transfected with control or IRGM-directed siRNA oligos, then infected with S. typhimurium after 48 h. The percentage of bacteria encapsulated in LC3-positive vesicles was determined by microscopy. Reductions in IRGM expression resulted in a significant loss of anti-bacterial autophagy compared to control cells. (c) Microscopic examination reveals a decreased number of autophagically encapsulated bacteria in IRGM knockdown cells. Control siRNA-treated (siControl) and IRGM siRNA-treated (siIRGM1, siIRGM2) cells were infected with S. typhimurium SL1344 and imaged by confocal microscopy. Control cells show numerous bacteria (red in merged image) surrounded by LC3-GFP membranes (green in merged image). Such bacteria-containing autophagosomes (indicated by dashed boxes) were almost completely absent in IRGM-deficient cells. The actin cytoskeleton was visualized using phalloidin (blue in merge). Images are flat projections of confocal z-stacks. Scale bars, 10 μm. (d) HeLa cells stably expressing LC3-GFP were transfected with increasing amounts of an IRGM expression construct and infected 24 h later with S. typhimurium. The percentage of bacteria found within LC3-positive vesicles increased with increasing amounts of exogenous IRGM. Protein expression levels were confirmed by blotting lysates for Flag-IRGM expression, using an antibody to Flag to detect exogenous IRGM and actin as a loading control (lower panel). Error bars in b and d, s.e.m.