Association of the GABRB2 polymorphisms with schizophrenia in Korean population

Abstract

The gamma-aminobutyric acid (GABA) A receptor is a multisubunit chloride channel that mediates the fastest inhibitory synaptic transmission in the central nervous system. Disturbances in GABAergic system have been observed in schizophrenics, and GABRB2, which encodes for GABA A receptor beta 2 subunit in 5q34 region, has been suggested for its possible relationship in schizophrenia. We did case-control association study of 51 single nucleotide polymorphisms (SNPs) in GABRB2 gene to examine the association of GABRB2 gene and schizophrenia in Korean population. `Illumina� HumanHap550-Duo Genotyping BeadChip` was used to genotype 21 schizophrenics (M=13, F=8) and 90 normal controls (M=45, F=45). Association analysis was done by PLINK (MGH & MIT Borad institute). Total 8 SNPs in GABRB2 gene were significant under the allele association test (p<0.001), and their odds ratio were greater than 3.0 suggesting strong association of GABRB2 with schizophrenia. Moreover, 7 SNPs in GABRB2 gene were significant under genotype association test (dominant model, p<0.001). However, the association between SNPs and schizophrenia failed to reach the significance level (α=0.05) after multiple testing correction by FDR method. We also conducted haplotype association analysis and 5 haplotype block structures were reconstructed in GABRB2 gene. A haplotype (GGCTGGCTCCCGTCC) in block 3 was significantly different between schizophrenia patients and controls (p=0.0001). Another haplotype (GACC) in block 5 was also significantly differnet in schizophrenia group (p=0.0005). This study indicates that we need to deal with the false positives more seriously, especially in genome-wide association study, because most of the significant SNPs cannot pass the significance level after multiple testing corrections. Validating the significant SNPs with a different genotyping technology as well as increasing statistical power of association study by collecting more case/control samples is strongly recommended.