Genome-wide
association studies of type 1 diabetes (T1D) have reported numerous new T1D
risk loci. These loci now require
detailed mapping to identify candidate genes and causal variants for functional
studies. We used ImmunoChip,
a custom high density genotyping array, to genotype 16,086 T1D case and control
samples and 2,670 families for 154,939 single nucleotide polymorphisms (SNPs)
from 186 loci previously implicated in autoimmunity. Of previously reported non-MHC T1D associated
regions, 36 were replicated (P ≤ 3.23 × 10-7) with most (29/36)
attaining genome-wide significance (P ≤ 5 x 10-8). Three novel T1D
loci were identified: 1q32.1, containing CAMSAP2,
GPR25 and C1orf106; 4q32.3,
between CPE and TLL1; and 10p11.22, a region 3´ of NRP1, and 5´ of ITGB1. A hypothesis testing procedure was performed
to determine which SNPs in each non-MHC locus were likely to be causal. Of 16,991 candidate SNPs tested, the causal
variant hypothesis was not rejected for 388 SNPs. The positions of these SNPs suggested
possible roles in disease pathogenesis.
For 27 of the 39 T1D-associated loci, the most significant SNP was
located within the transcription unit of a gene. For four loci, the most associated SNP is a
non-synonymous coding variant. For 3 more
loci, the most associated SNP was in a non-coding RNA. A majority of the putative causal SNPs
overlapped promoter histone marks, enhancer histone marks, DNase I sensitive
sites, predicted protein binding sites or other motifs compiled in ENCODE. In summary, ImmunoChip genotyping has
clarified the complex genetic risk in T1D, identifying putative causal genes.
These genes are located in both previously disease-associated loci and in 3
novel T1D risk loci. Despite the
tissue-specific autoimmune response that characterizes T1D, the majority of T1D
risk variants localize in or near genes whose products act upon the immune
system, rather than the target organ.