Evidence for association of bipolar disorder to haplotypes in the 22q12.3 region near the genes stargazin, IFT27 and parvalbumin

Abstract

We have previously reported genome-wide significant linkage of bipolar disorder to a region on 22q12.3 near the marker D22S278. Towards identifying the susceptibility gene, we have conducted a fine-mapping association study of the region in two independent family samples, an independent case-control sample and a genome-wide association dataset. Two hundred SNPs were first examined in a 5 Mb region surrounding the D22S278 marker in a sample of 169 families and analyzed using PLINK. The peak of association was a haplotype near the genes stargazin (CACNG2), intraflagellar transport protein homolog 27 (IFT27) and parvalbumin (PVALB; P = 4.69 × 10(-4)). This peak overlapped a significant haplotype in a family based association study of a second independent sample of 294 families (P = 1.42 × 10(-5)). Analysis of the combined family sample yielded statistically significant evidence of association to a rare three SNP haplotype in the gene IFT27 (P = 8.89 × 10(-6)). Twelve SNPs comprising these haplotypes were genotyped in an independent sample of 574 bipolar I cases and 550 controls. Statistically significant association was found for a haplotype window that overlapped the region from the first two family samples (P = 3.43 × 10(-4)). However, analyses of the two family samples using the program LAMP, found no evidence for association in this region, but did yield significant evidence for association to a haplotype 3' of CACNG2 (P = 1.76 × 10(-6)). Furthermore, no evidence for association was found in a large genome-wide association dataset. The replication of association to overlapping haplotypes in three independent datasets suggests the presence of a bipolar disorder susceptibility gene in this region.

FIG. 1. TDT analysis of family study phases

The top three panels illustrate the significance of single SNP and haplotype analysis as –log(P) for the two independent family samples and their merged results. Black circles denote the single SNP analyses. The haplotypes are colored by number of SNPs used in the window (5 SNPs—red, 4 SNPs—orange, 3SNPs—green, 2SNPs—blue). UCSC genome browser gene track is provided for reference. The NCBI 36 map is used for positions.

Fig. 2. Association analysis of the case control replication sample

Significance levels as −log(P) of the 12 SNPs genotyped in the case control replication sample are illustrated as black circles. Black lines denote the significant individual haplotype window incorporating all haplotypes in the window. The significant haplotypes from the three family sample analyses are also illustrated here for comparison: family study phase 1 is red, family study phase 2 is orange, the merged family sample is purple. The most significant haplotype from the LAMP analysis is illustrated in green. Gene locations and the LD structure in the HapMap Caucasian sample are below for reference. Positions are based on the NCBI 36 map.