NEW YORK (GenomeWeb) – A team from Brigham and Women's Hospital, the Broad Institute, and elsewhere has focused in on a handful of suspected causal variants in rheumatoid arthritis and type 1 diabetes, using follow-up assays to explore the functional significance of at least some of these variants.
As they reported online this week in Nature Genetics, the researchers began by fine mapping 76 loci, based on data for tens of thousands of individuals with or without the autoimmune conditions. They also sequenced histone H3K4me3 regulatory sequences in and around these sites in nearly 600 individuals.
In addition to showing that that they could impute common variants with high accuracy, the researchers narrowed in on small sets of suspected causal variants at five rheumatoid arthritis-associated loci and 10 loci linked to type 1 diabetes. From there, the group used functional assays to begin characterizing a subset of the sites, including a SNP in CD28-CTLA4 and small insertions or deletions in the MEG3 and TNFAIP3 genes.
The authors noted that while prior genome-wide association studies implicated more than 150 loci — and functions involving CD4+ T autoimmune cell pathways — in the autoimmune conditions, the set of causal variants that actually lead to disease has not been as well characterized in the past.
"Pinpointing [causal variants] will enable mechanistic investigation to identify the specific genes, regulatory structures, and genetic mechanisms central to autoimmunity," senior author Soumya Raychaudhuri, a data sciences, genetics, and medical and population genetics researcher affiliated with BWH, the Broad, the University of Manchester, and Harvard, and his colleagues wrote.
For their fine-mapping step, Raychaudhuri and his colleagues used the ImmunoChip to genotype 11,474 arthritis cases and 15,870 controls, as well as 9,334 individuals with type 1 diabetes and 11,111 without. They noted that 46 of the 76 loci subjected to fine mapping had been linked to rheumatoid arthritis, while 49 were associated with type 1 diabetes.
To fill in any gaps in sequence data in these regions, the team also used Illumina MiSeq technology to sequence 799 H3K4me3 regulatory regions apiece in 568 of the ImmunoChip-genotyped individuals, bringing in data from the 1000 Genomes Project and other prior studies until it was possible to impute common variants at the sites in question with roughly 89 percent accuracy.
From there, the researchers narrowed in on 20 sites linked to rheumatoid arthritis, identifying five loci with proposed causal variants, along with 34 type 1 diabetes-linked loci (10 with variants that appeared to be causal). The sites each contained five or fewer causal variant candidates, they noted, which were prioritized based on their predicted protein-coding or regulatory consequences.
Using new and existing functional data — including from promoter capture Hi-C, electrophoretic mobility shift assays, chromatin immunoprecipitation sequencing, and other approaches — the team went on to explore the potential effects of some of these proposed causal variants in each disease.
Among other limitations cited, the study's authors cautioned that the specific genes accounting for disease susceptibility still need to be determined, since specific variants may have ties to more than one gene and, relatively few of the variants shortlisted in the analysis fell in linkage disequilibrium with documented quantitative trait loci.
Even so, they concluded, "we believe that the combination of statistical evidence with functional follow-up is a powerful way to prioritize potentially causal variants."