Association study of polymorphisms in the GluR6 kainate receptor gene (GRIK2) with schizophrenia

Based on the fact that phencyclidine (PCP), a noncompetitive antagonist of the NMDA receptor, induces phenotypes closely similar to schizophrenia in human and in animal models, a glutamatergic dysfunction hypothesis has been proposed for the pathogenesis of schizophrenia (Luby et al., 1959, Javitt and Zukin, 1991, Mohn et al., 1999). Since genetic factors are known to significantly increase the risk for the disease (λ_s=10) (McGuffin et al., 1995), the genes involved in the glutamatergic transmission system are strong candidates for schizophrenia-susceptibility genes. Two genes encoding major subunits of the NMDA receptor, which is a primary target of PCP, GRIN1 and GRIN2B, have shown no significant association with schizophrenia (Nishiguchi et al., 2000, Sakurai et al., 2000, Tani et al., 2002). Genes encoding other types of glutamate receptors, such as AMPA receptors, kainate receptors and metabotropic glutamate receptors, are also good candidates for the susceptibility genes because they probably cross-talk at the presynapse and regulate the activity of the NMDA receptor, which is primarily postsynaptic (Meador-Woodruff and Healy, 2000). In fact, an antagonist of AMPA/kainate receptors, LY293558, and an agonist of metabotropic glutamate receptors, LY354740, have been reported to reverse the psychotic effects of noncompetitive antagonists of the NMDA receptor such as PCP and ketamine (Moghaddam et al., 1997, Moghaddam and Adams, 1998). To examine the glutamatergic hypothesis for schizophrenia pathology, we studied polymorphisms in three glutamate receptor genes, GRM2, GRIK1 and GRIN1, although none of the three genes showed a significant association with the disease (Joo et al., 2001, Shibata et al., 2001, Tani et al., 2002).

The GluR6 kainate receptor gene GRIK2 consists of 17 exons spanning over 700 kb and has been mapped to 6q16.3–q21 (Paschen et al., 1994, Barbon et al., 2001). Recent studies have shown that kainate receptors, GluR5 and GluR6, are involved in neuronal plasticity (Bortolotto et al., 1999, Schmitz et al., 2001, Contractor et al., 2001). Multiple genome-wide scanning studies mapped a schizophrenia susceptibility locus in the close vicinity, 6q21–q22.3 (Cao et al., 1997, Kaufmann et al., 1998, Martinez et al., 1999). In addition, a significant decrease of GRIK2 mRNA expression level has been reported in schizophrenic brains (Porter et al., 1997). These lines of evidence suggest that GRIK2 is a strong candidate for the susceptibility locus for schizophrenia. No association of a microsatellite in the 3′ untranslated region of GRIK2 with the disease has been reported (Chen et al., 1996). However, the possibility that the susceptibility locus lies in other parts of the gene needs to be examined with more polymorphisms since the size of the gene (>700 kb) far exceeds the average size of linkage disequilibrium (LD) reported on the human genome (60 kb) (Reich et al., 2001).

In this study we tested for an association of schizophrenia with GRIK2 using 15 common SNPs evenly distributed throughout the gene with an average interval of 53.4 kb. Assuming that the size of LD in the region is approximately 60 kb (Reich et al., 2001), the LDs among the 15 SNPs are expected to cover the entire GRIK2 region including all the coding sequences. We compared allele and genotype frequencies between patients with schizophrenia and controls. To enhance the detection power of association, we compared pairwise haplotype frequencies deduced from genotype frequencies of the 15 SNPs.