INTRODUCTION

The human 5-hydroxytryptamine 7 (5-HT7) serotonin receptor is a G-protein-coupled receptor (Hoyer et al, 2002). These receptors are expressed in discrete areas of the brain as well as in both vascular and gastrointestinal smooth muscle (Bard et al, 1993; Lovenberg et al, 1993), and the recent availability of selective antagonists and knockout mice strains has increased our knowledge about their functional roles. The 5-HT7 receptors are now known to have an important role in modulating hippocampal neuronal functions such as learning and memory, disturbances in which are thought to be fundamental in schizophrenia (Hedlund and Sutcliffe, 2004; Thomas and Hagan, 2004). 5-HT7 receptor knockout mice have also exhibited a specific impairment in contextual fear conditioning, which is associated with hippocampus-dependent learning, and reduced ability for long-term potentiation in the CA1 region of the hippocampus (Roberts et al, 2004).

Other in vivo and in vitro studies suggest that 5-HT7 receptors can mediate the effects of 5-HT on hippocampus glucocorticoid receptor expression (Weaver et al, 2001; Laplante et al, 2002; Beique et al, 2004). Glucocorticoid has been tentatively associated with neurotoxicity, and considering the neurodevelopmental hypothesis of schizophrenia might have some role in this disease (Cotter and Pariante, 2002).

Neuropsychopharmacologically, 5-HT7 receptors show high affinity for a number of antidepressants and typical and atypical antipsychotics. For example, clozapine, which is a classical atypical antipsychotic drug and which has been shown to be effective in treating schizophrenia (Safferman et al, 1991; Kane et al, 2001), exhibits moderate potency as a 5-HT7 receptor antagonist. This finding suggests that an interaction between and/or overactivity of receptor systems may be involved in the pathophysiology of schizophrenia, although the pharmacological basis of the unique actions of clozapine is not well understood. Other lines of evidence also support the association of 5-HT7 receptors to schizophrenia; A postmortem study showed decreased expression of 5-HT7 receptors in the dorsolateral prefrontal cortex of schizophrenics (East et al, 2002). Furthermore, genomewide linkage studies of schizophrenia have shown a linkage in 10q22 (Mowry et al, 2000; Fallin et al, 2003) (OMIM: SCZD11,%608078) close to the location of the human 5-HT7 receptor gene (HTR7: 10q21–24).

Here, we evaluate whether HTR7 is associated with schizophrenia in a sample including 383 unrelated Japanese schizophrenia patients and 351 unrelated controls. A common hypothesis about allelic architecture proposes that the greatest genetic risk for common, complex diseases including schizophrenia is from disease loci that have one common variant (common disease–common variants hypothesis: CD–CV hypothesis) (Chakravarti, 1999). In the light of this CD–CV hypothesis, association analysis using linkage disequilibrium (LD) mapping would seem to be for a reasonable approach to narrow down the number of potential susceptibility genes or variants for schizophrenia. We therefore performed (1) LD mapping in HTR7 and selection of ‘haplotype-tagging (ht) SNPs’, (2) an association analysis using ‘htSNPs’, and (3) a systematic mutation search to detect actual susceptibility variants capable of modifying the function of HTR7.

MATERIALS AND METHODS

Subjects

A total of 383 patients with schizophrenia (200 male and 183 female; mean age±standard deviation (SD), 42.6±14.5 years) participated in the present study. The subjects for the mutation search were 48 schizophrenic patients who were also among the 383 patients in the association analysis (27 male and 21 female; mean age±SD, 39.5±14.7 years). A total of 351 healthy volunteers (193 male and 158 female; mean age±SD, 33.5±13.1 years) were recruited as control subjects. The subjects for initial ‘LD mapping’ were 96 controls who were also subjects in the association analysis.

The characterization details and psychiatric assessment of these subjects were identical to those published elsewhere (Suzuki et al, 2003; Ikeda et al, 2005). The patients were diagnosed according to DSM-IV criteria with the consensus of at least two experienced psychiatrists on the basis of unstructured interviews and review of medical records. All healthy controls were also psychiatrically screened based on unstructured interviews. All subjects were unrelated to each other and ethnically Japanese, and were individually matched for gender and geographical origin.

After the study had been described, written informed consent was obtained from each subject. This study was approved by the Ethics Committee at Fujita Health University and Nagoya University Graduate School of Medicine.

SNP Inclusion for LD Mapping

We included all SNPs from the dbSNP database (URL; http://www.ncbi.nlm.nih.gov/SNP/) and Celera Discovery System database (URL; http://www.celeradiscoverysystem.com/) for LD mapping (Table 1 and Figure 1).

Table 1 Summary of SNPs in HTR7
Full size table
Figure 1
figure 1

Genomic structure of HTR7 with SNPs used in LD mapping. Exon D is an alternative splice exon. Numbers in parentheses represent minor allele frequencies (MAFs) of 96 controls.

Full size image

Selection of ‘htSNPs’

‘LD blocks’ were first determined using criteria based on 95% confidential bounds on D′ values (Gabriel et al, 2002), with Haploview version 2.05 software (Barrett et al, 2004). ‘htSNPs’ were then selected within each ‘LD block’ for 90% haplotype coverage using SNPtagger software (Ke and Cardon, 2003). This strategy for association analyses after initial LD mapping and ‘htSNPs’ selection is considered to be reasonable on the basis of descriptions given in other papers (Kamatani et al, 2004; van den Oord and Neale, 2004).

SNP Genotyping

Genomic DNA was extracted from peripheral blood of all subjects. For rapid genotyping of SNPs, we used TaqMan assays (SNP1, SNP3, SNP4, SNP5, and SNP7), polymerase chain reaction (PCR)-restriction fragment length polymorphism (RFLP) assays (SNP6), primer extension with denaturing high performance liquid chromatography (dHPLC) (SNP2) and direct sequencing (SNP8) (Table 1). TaqMan probes and Universal PCR Master Mix were from Applied Biosystems (CA). A 5-μl total reaction volume was used and, after PCR, the allelic specific fluorescence was measured on an ABI PRISM 7900 Sequence Detector System (Applied Biosystems, CA). Methods for RFLP assays, primer extension, and direct sequencing were described in greater detail previously (Suzuki et al, 2003).

Dual-Luciferase Reporter Assay

Fragments of 262 bp including SNP2 were PCR amplified. Genomic DNAs with identified genotypes were used as templates, and PCR products of both genotypes were cloned into pGL3-promoter Vector (Promega, WI). These Vectors with both alleles and the Renilla luciferase vector, phRL-TK vector, were transiently transfected into Chinese hamster ovary (CHO) cells using Lipofectamine 2000™ (Invitrogen, CA). After 48 h, cell extracts were prepared and assayed for firefly luciferase activity (LAF) and Renilla luciferase activity (LAR) as described by the manufacturer (PikkaGene Dual SeaPansy™ Luminescence Kit, Toyo Ink, Japan) on a Fluoroskan Ascent FL (Thermo Labsystems, Finland). All experiments were repeated at least three times. To correct for transfection efficiency, the relative luciferase activity (RLA) was calculated as: RLA=LAF/LAR. The RLA of mutants (RLAmt) was shown by % ratio of the RLA of wildtype (RLAwt). Means and standard errors were calculated and an unpaired two-tailed t-test was performed using the software package SPSS 10.0J (SPSS Japan Inc., Japan). Further details about primer pairs and culture conditions for the cells are available on request.

Mutation Search

We performed dHPLC analysis, details of which can be seen in a previous paper (Suzuki et al, 2003). Primer pairs were designed using information from the GenBank sequence (accession number: NM-030059.11) and Erdman's paper (Erdmann et al, 1996), as well as from 14 amplified regions that covered all the exons, alternative splice exon (exon D), and introns including the branch sites and 1000 bp upstream from initial exon of HTR7.

Statistical Analysis

Genotype deviation from the Hardy–Weinberg equilibrium (HWE) was evaluated by χ2 test (SAS/genetics, release 8.2, SAS Institute Japan Inc., Tokyo, Japan).

Marker–trait association analysis was performed with the COCAPHASE 2.403 program (Dudbridge, 2003). This program performs log-likelihood ratio tests under a log-linear model for global P-value. To estimate haplotype frequencies, an expectation–maximization (EM) algorithm was used (Zhao et al, 2000). Rare haplotypes found in less than 3% of both cases and controls were excluded from association analyses due to the limitation of the EM algorithm. For haplotypic analyses, we calculated global P-values in multi-SNP haplotype systems. In both haplotypic and single marker allelewise analyses, we performed the permutation procedure that is a tool in the COCAPHASE program in a 1- to 3-marker sliding window fashion. We emphasized the permutation P-values rather than individual global P-values, because this permutation method is considered to be appropriate for these analyses (Dudbridge, 2003) (ie Bonferroni correction for multiple testing is considered to be too conservative to apply to genetic association analyses (Nyholt, 2001)). Ten thousand permutations were performed in each permutation procedure, and ORs (ORs) of risk alleles or haplotypes were calculated when the most frequent alleles or haplotypes had ORs of 1. To detect risk haplotypes, we performed the individual procedures in this program. The significance level for all statistical tests was 0.05.

RESULTS

LD Mapping

We genotyped eight SNPs for 96 controls. After testing for deviation from HWE (all SNPs were in HWE), we evaluated pairwise LD matrices between each SNP, and selected ‘htSNPs’ from LD blocks. Consequently, we determined three ‘LD blocks’ and four ‘htSNPs’ (SNP1, SNP2, SNP5, and SNP7) (Table 2).

Table 2 Pairwise Linkage Disequilibrium (LD) Matrices and ‘LD Blocks’ in HTR7
Full size table

Association Analysis

We expanded genotyping of these four ‘htSNPs’ for all the schizophrenia subjects and the remaining 255 controls. In this step, genotype distributions were again in HWE.

Almost all global P-values except 1-marker global P-values of SNP1 and SNP7 were associated with schizophrenia. After adjustment by permutation procedure, we found a significant association of all permutation P-values with schizophrenia (Table 3). An assessment of the components of these significant associations indicated SNP2 and SNP5 as possible key components for schizophrenia (‘A’ allele of SNP2, case=23.5%, control=17.7%, and P-value=0.00571; ‘A’ allele of SNP5, case=35.8%, control=28.9%, and P-value=0.00504; Table 3). The ORs for the significant risk genotypes are modest; OR=1.43 (95% CI: 1.11–1.85) for SNP2, OR=1.37 (95% CI: 1.10–1.71) for SNP5.

Table 3 Association Analyses of Four ‘ht SNPs’ in HTR7
Full size table

To detect the risk haplotypes, individual haplotypic analyses were performed from positive global P-values (Table 4). The most significant haplotype overtransmitted in schizophrenia was a combination of SNP2 and SNP5 (‘A–A’ haplotype, case=23%, control=16%, and P-value=0.0015); however, the estimated haplotype frequencies of each risk haplotype were nearly equal to the minor allele frequency (MAF) of SNP2 (case=23.5%, control=17.7%).

Table 4 Individual Haplotypic Analyses from Positive Permutation Analyses of Schizophrenia
Full size table

Dual-Luciferase Reporter Assay of SNP2

Taken together with the results of the LD mapping (SNP2 was in strong LD with SNP5), we hypothesized that SNP2, a possible promoter SNP (533 bp from initial exon), may be a susceptibility SNP for schizophrenia. To confirm the functional relevance of SNP2, we developed a dual-luciferase reporter assay. However, the results from this assay did not provide evidence that SNP2 modifies transcriptional activity of HTR7 (P-value=0.782; RLAwt=100±58.2, RLAmt=86.2±45.2).

Mutation Search

To detect the actual susceptibility variants, we performed a systematic mutation search in all exons and introns including branch sites and 1000 bp upstream from the initial exon. However, we could not find any functional mutations in these regions.

DISCUSSION

In this study, two htSNPs (a possible promoter SNP (SNP2) and an intronic SNP (SNP5)) and haplotypes showed an association with schizophrenia in a Japanese population. However, the functional relevance of SNP2 could not be confirmed with the dual-luciferase reporter assay.

The method of using ‘htSNPs’ in LD association analyses is more sensitive and powerful than that of using randomly selected SNPs. Here, selection of ‘htSNPs’ resulted in the fitting of another criteria, ‘LD-selected htSNPs’, at a relatively stringent r2 threshold (r2>0.8) (Carlson et al, 2004). This method of ‘LD-selected SNPs’ reflects the evolutionary relationships of haplotypes. Therefore, our ‘htSNPs’ could identify disease associations with either specific haplotypes or with clades of related haplotypes (Carlson et al, 2004). Given this LD pattern of HTR7, the actual susceptibility variants may exist anywhere in HTR7.

Although our sample size in this mutation search was sufficient to detect common variants with more than 5% MAF at 95% power (Collins and Schwartz, 2002), we failed to find the actual susceptibility variants through a systematic mutation search. Since the actual promoter region in HTR7 has not been determined, the regions targeted in our mutation search might not have been adequate. In addition, we did not search the conserved noncoding sequence that plays a role in gene regulation. Further investigation will be required for conclusive results.

An additional point deserves attention in interpreting the results: The positive association with schizophrenia could be due to type I error, possibly because of population stratification or unmatched age samples. Replication study with genomic control or a family-based population will be required.

In conclusion, we found the first, and a significant, association of SNPs in HTR7 with Japanese schizophrenia patients after consideration of problems in statistical genetics. These results support the supposition that HTR7 is a schizophrenia-susceptibility gene.