Introduction

The dopamine hypothesis is the oldest neurochemical theory of the pathophysiology of psychosis (Tost et al., 2010). The contemporary pathophysiological model of the dopamine hypothesis assumes that psychotic symptoms are triggered by a dysregulation of dopaminergic activity in the brain (Kapur et al., 2005; Tost et al., 2010). The dysregulation of the dopamine system has been one of the most enduring and plausible explanations for the pathophysiology of schizophrenia (Abi-Dargham and Moore, 2003; Kapur et al., 2005; Seeman et al., 2006; Di Forti et al., 2007; Talkowski et al., 2008; Tost et al., 2010). A large body of genetic association studies has reported an association of dopamine receptor genes (DRDs) with schizophrenia. However, a single susceptibility gene probably does not exist, even though a few specific variants in DRDs have been suggested to play small but significant roles in individual susceptibility to schizophrenia (Dubertret et al., 1998; Glatt et al., 2003; Jonsson et al., 2003; Owen et al., 2004; Talkowski et al., 2008; Saiz et al., 2010; Srivastava et al., 2010).

Although the theoretical advantages of an association design in detecting the modest effects expected in a complex disease are apparent (Risch, 2000), most previous association studies of schizophrenia, including those on DRDs, have only examined one or a few variants in one or a few genes of interest. Considering the most likely mode of inheritance of schizophrenia is via multiple genes, each with a small effect, a limited numbers of polymorphisms and a relatively small sample size for investigation can be significant limitations. Association designs have been suggested as an alternative, with a dense set of single nucleotide polymorphisms (SNPs) covering the regions of interest in multiple genes (Chowdari et al., 2007, Hong et al., 2010). However, the expense, the time-consuming nature, and the labor involved in such large-scale genotyping may be restrictive factors for individual laboratories (Chowdari et al., 2007).

Many researchers have demonstrated the efficacy of DNA pooling and associated technologies that accurately and reliably estimate SNP or microsatellite allele frequencies in a pooled DNA sample as an initial screen for genetic associations (Mohlke et al., 2002; Sham et al., 2002; Werner et al., 2002; Schnack et al., 2004). Two- and three-stage association tests have been carried out using many polymorphisms and DNA pooling techniques to find susceptibility loci for schizophrenia (Hoogendoorn et al., 2005; McGhee et al., 2005; Chowdari et al., 2007; Gunnell et al., 2007; Zaharieva et al., 2008). Hoogendoorn et al. performed multi-stage analyses of associations on 18 microsatellite markers in 12 dopaminergic genes including DRDs (Hoogendoorn et al., 2005). They were able to identify one marker in DRD5 from pooled allele frequency comparisons, but its significance was lost in analyses with an extended sample. Chowdari et al. also performed multi-stage association tests on 69 SNPs in ACSL6 and SIRT5 for schizophrenia, which comprised analyses of allele frequencies estimated by DNA pooling, individually genotyped pools, and extended sample genotyping (Chowdari et al., 2007). An association study using pooled DNA genotyping could identify susceptibility genes for schizophrenia by screening the 5q31-32 linkage region (Zaharieva et al., 2008). Although several studies did not yield convincing evidence for associations between schizophrenia and the genes of interest, the multi-stage strategy proved to be straightforward and particularly helpful for researchers as a rapid and efficient tool in large-scale association studies.

We conducted a comprehensive three-step analysis to investigate the association between five dopamine receptor genes (DRD1, 2, 3, 4, and 5) and schizophrenia in a Korean population. The scheme for analysis was as follows. Stage 1: SNPs in DRDs were selected from the dbSNP, and allele frequencies were estimated using pooled DNA analysis using Matrix-assisted laser desorption time-of-flight mass spectrometry (MALDI-TOF MS). Stage 2: We genotyped the pooled sample individually for screened SNPs and conducted genetic association analyses. Stage 3: Genetic association analyses for suggestive SNPs were performed with an extended sample.

Results

Results of Stage 1 review of SNPs assays tested for association by using DNA pools

Of a total of 142 SNPs, 88 (62%) SNPs in the control sample had minor allele frequencies of 1.5-48.5% and were thus designated as polymorphic. We removed markers with poor quality data and spectra with signal-to-noise ratios below 3.5 or with a peak height below 1.0 intensity unit. We also removed SNPs for which fewer than three of six possible observations remained for any pool. The median standard deviations of estimated allele frequencies in pools of controls and cases were 0.0127 and 0.0138, respectively.

As detailed in Table 1, 10 of 88 SNPs exhibited significantly different frequencies between case (n = 150) and control (n = 150) groups at the 5% level of significance. We identified two SNPs at DRD1, three SNPs at DRD2, and five SNPs at DRD4. These SNPs with a pre-set case-control difference were then genotyped individually among the samples composing the pools using MALDI-TOF MS analysis.

Table 1 Results of association testing of 10 SNPs in DRDs using pools of DNA *Genomic sequence (www.ncbi.nlm.gov, build 127). Cases include 117 individuals with schizophrenia, 16 with schizophrenia spectrum disorders, and 17 with bipolar I disorder. SZ: schizophrenia, CTR: control, HWE: Hardy-Weinberg equilibrium
Full size table

Results of Stage 2 genetic association analyses for suggested SNPs using individual samples from DNA pools

The samples constituting the DNA pools were genotyped individually for 10 SNPs with suggestive associations in Stage 1. When we evaluated the Hardy-Weinberg equilibrium (HWE) in the control samples, rs4245149 (P = 0.035), rs936461 (P = 0.004), and rs1800955 (P = 0.022) deviated from equilibrium. Because we had used a pooled sample of patients with schizophrenia (n = 117) and other major psychoses (n = 33), we compared the samples in two ways: all cases vs. controls and schizophrenic patients vs. controls. When comparing all cases with controls, allele-wise analyses for a single marker revealed two significant associations at rs1799914 of DRD1 (chi-square = 3.983, P = 0.046, OR = 1.643, CI: 1.006-2.684) and rs752306 of DRD4 (chi-square = 5.713, P = 0.017, OR = 0.575, CI: 0.364-0.909). Analysis of the schizophrenia sample at two SNPs revealed results similar to those found using all cases (rs1799914 of DRD1, chi-square = 3.651, P = 0.056, OR = 1.675, CI: 0.983-2.853; rs752306 of DRD4, chi-square = 4.745, P = 0.029, OR = 0.587, CI: 0.362-0.951), though rs1799914 did not show an association at the nominal level of significance (Table 1).

A linkage-disequilibrium (LD) test was carried out on a pair-wise basis for all SNPs in the 117 patients with schizophrenia and the 150 controls. Little evidence for linkage disequilibrium between SNPs in individual genes was observed (Table 2).

Table 2 Pair-wise LD analyses in DRDs using samples of individuals with schizophrenia and controls LD statistics are presented according to r2 values. Upper and lower diagonals represent the results for patients with schizophrenia (N = 117) and those for controls (N = 150), respectively
Full size table

Results of Stage 3 genetic association study with suggested SNPs in the extended sample

The SNPs rs1799914 of DRD1 and rs752306 of DRD4 were assayed in an extended sample of 270 patients with schizophrenia and 350 control individuals using the TaqMan genotyping method. The concordance rates of genotypes between MALDI-TOF MS and the TaqMan method were 100% in each SNP when we checked the genotypes of 267 overlapped individuals (117 schizophrenia and 150 controls). As shown in Table 3, the extended sample did not show associations at rs752306 of DRD4 (allele wise comparison, chi-square = 0.525, P = 0.469, OR = 1.122, CI: 0.821-1.534). rs1799914 of DRD1 revealed a significant association with a lower P-value than the previous stage's association result (allele-wise comparison, chi-square = 5.695, P = 0.017, OR = 1.540, CI: 1.078-2.199).

Table 3 Results of association testing of two SNPs suggesting an association with schizophrenia based on previous stages in the extended sample *indicates P-value of exact test. SZ: schizophrenia, CTR: control, HWE: Hardy-Weinberg equilibrium
Full size table

Discussion

Pooled DNA analysis is an economical method for large-scale genetic association studies and is particularly useful for rapid screening of large numbers of SNPs (Sham et al., 2002; McCarthy et al., 2008). Applying this strategy, we performed a multi-stage genetic association study of the SNPs in DRD1-5, looking at the suggestive case-control differences from the pooled analysis and then genotyping them in individual samples for confirmation, followed by a replicate analysis with extended samples. The 10 most studied SNPs were not in LD for each gene. This means that those SNPs in some way represent different regions of each gene. Finally, we were able to find evidence for an association (P = 0.017) between rs17799914 of DRD1 and schizophrenia. However, the possibility of a type I error in our result should be noted, as the P-value was not adjusted for multiple testing.

The functional significance of the DRD1 receptor for the pathophysiology of schizophrenia has been postulated. Insufficiency of dopamine transmission in the prefrontal cortex contributes to the cognitive deficits observed in patients with schizophrenia, and robust empirical evidence suggests a central role of prefrontal cortex DRD1 receptors in working memory and executive function (Okubo et al., 1997; Abi-Dargham et al., 2002; Rybakowski et al., 2005). Also, altered dopamine neural networks resulting from abnormal interaction of DRD1 and DRD2 in the prefrontal cortex is thought to be linked to the development of psychotic symptoms (Abi-Dargham and Moore, 2003; Tost et al., 2010).

The sequence of DRD1 has been determined and mapped to chromosome 5q35.1 (Grandy et al., 1990), and several studies have suggested that the region is linked to schizophrenia (Sklar et al., 2004, Zheng et al., 2006, Escamilla et al., 2007). As mentioned earlier, a study using pooled DNA genotyping reported that several polymorphisms close to the region were associated with schizophrenia (Zaharieva et al., 2008). With respect to association studies involving DRD1 and schizophrenia, variants of rs4532 (-48A/G, recognized by a DdeI restriction cut site) have most often been studied for DRD1. Although the SNP has not yet shown a clear association with schizophrenia (Cichon et al., 1994; Liu et al., 1995; Dmitrzak-Weglarz et al., 2006; Zhang et al., 2010), a recent meta-analysis reported that the odds ratio of rs4532 in DRD1 reached 1.18 (95%CI: 1.01-1.38) (Allen et al., 2008). Moreover, rs4532 of DRD1 was reported to have an association with performance on tests (e.g., the Wisconsin Card Sorting Test) in patients with schizophrenia (Rybakowski et al., 2005). Very recently, a study using a Spanish population found an association between two individual SNPs (rs11746641 and rs11749676) and haplotypes containing the SNPs in DRD1 and schizophrenia, especially among men (Hoenicka et al., 2010). These findings support the relationship between DRD1 and the pathophysiology of schizophrenia.

The associations of the most suggestively associated SNPs on DRDs in Stage 1 failed to be replicated in the Stage 2 analyses. The significant association of rs752306 on DRD4 in Stage 2 was also not replicable in the Stage 3 analyses. These findings suggest the possibility of negative association between other DRDs and schizophrenia. Generally, lack of clear association results between DRDs and schizophrenia in various studies may be due to genetic heterogeneity, variability in the loci analyzed, and the power and sample size. Therefore, it would be overstated to conclude that DRDs were not associated with a risk of schizophrenia, though our analyses show no support for associations with most DRDs.

Despite the negative association results at Stage 2 in our analyses, some of the suggestively associated SNPs in Stage 1 have been studied previously in various neuropsychiatric disorders as well as schizophrenia. Since an association between rs1800955 (-521CT) of DRD4 and schizophrenia was first reported by Okuyama et al. (1999), this polymorphism has become one of the most intensively studied SNPs of DRD4 not only for schizophrenia (Mitsuyasu et al., 2001; Jonsson et al., 2003; Xing et al., 2007), but also for other psychiatric illness such as personality disorders (Okuyama et al., 2000; Mitsuyasu et al., 2001; Joyce et al., 2003) and ADHD (Kirley et al., 2004; Bellgrove et al., 2005; Yang et al., 2008). Overall, results are still inconsistent despite extensive investigations, and recent meta-analyses for associations between the SNP and schizophrenia showed significant effects with odds ratios of 1.15 to 1.22 (Jonsson et al., 2003; Allen et al., 2008).

Arinami et al. isolated and investigated SNP rs1799978 (-241 AG) in the 5-UTR' region of DRD2 for any relation with schizophrenia (Arinami et al., 1997). While they found no association between rs1799978 of DRD2 and schizophrenia, similar to our present result, this SNP has been given particular attention recently for its relationship to the phenotype of a differential treatment response within a patient group with schizophrenia (Lencz et al., 2006; Xing et al., 2007), suggesting the possibility of an association between this SNP and sub-phenotypes of schizophrenia.

Several important limitations of this study should be noted. Only known SNPs were selected from a public database without novel SNPs, thus limiting the possible variants of loci in other introns and the variable number of tandem repeat (VNTR) polymorphisms across the DRDs. Another limitation is the possibility of false negative associations among the remaining 78 SNPs during the pooled analysis. The difference of allele frequencies in SNPs between the pooling stage and the subsequent individual genotyping might represent intrinsic limitations of the DNA pooling approach, such as reduced power and difficulties ensuring equimolar representation of samples (McCarthy et al., 2008).

The case sample set in the pooling stage was composed of patients with non-homogenous diagnostic phenotypes including bipolar I disorder (n = 17, 11%), which might cause confusion in the screening results. However, most cases were patients with schizophrenia (n = 117, 78%) and schizophrenia spectrum disorders (n = 16, 11%). A number of studies have shown that schizophrenia and bipolar I disorder have overlapping genetic backgrounds, clinical phenotypes, and treatment effects by common therapeutic agents (Hill et al., 2008; O'Donovan et al., 2009). Taken together, these findings suggest that the diagnostic phenotype of our case sample set may be regarded as representative of schizophrenia and of major psychosis. An additional limitation is the deviation of several SNPs from HWE. In the extended control sample, rs1799914 of DRD1 also deviated according to the nominal level of significance (P = 0.028). However, the degree of disequilibrium was not tremendous, and genotyping error was less likely because of the perfect concordance shown in the re-genotyping of 10% of all samples and between different the genotypic methods in overlapping samples. It is possible that the deviation from HWE in these SNPs occurred by chance as a result of the limited sample size. Therefore, it is unlikely that our significant finding was substantially based on such deviations in the control sample.

In conclusion, we report a possible association between rs17799914 of DRD1 and schizophrenia. Considering the functional significance of DRDs in the hypothesis that the dysregulation of the dopamine system is relevant to the pathophysiology of psychosis, DRDs may be implicated in the disease itself or specific behavioral phenotypes seen in schizophrenia and other psychiatric disorders, though our analyses show no support for associations with most DRDs. As case-control association studies are vulnerable to methodological artifacts that arise from population stratification, further studies with other ethnic groups are needed to support our results. We performed a multi-stage analytic study to comprehensively examine the genetic association of schizophrenia with the five DRDs. Our findings supported the efficacy of a large-scale multi-stage association study method using SNP information from a public database for finding genes linked to schizophrenia.

Methods

Subjects

Written informed consent was obtained from each subject prior to the study. This study was approved by the ethics committee at Seoul National University Hospital.

For the DNA pooling study, blood samples from a total of 150 patients with psychosis and 150 normal controls were randomly selected from our patient sample set with a major psychotic disorder and control sample set, respectively. Of the 150 patients, 117 met the Diagnostic and Statistical Manual of Mental Disorder-4th Edition (DSM-IV) criteria for schizophrenia, and 16 patients had schizophrenia spectrum disorders, including schizoaffective disorder or delusional disorder. The others met criteria for bipolar I disorder (n = 17), and all bipolar I patients had exhibited psychotic features at least once in their lifetimes.

For the extended sample set, a total of 270 patients with schizophrenia (DSM-IV), which included 117 from the pooled set, were recruited for this study. Consensus diagnostic meetings of more than two psychiatrists were held regularly to evaluate the participants' final diagnoses. Most of the patients (196) were comprehensively interviewed by trained nurses using the Korean-translated version of the Diagnostic Interview for Genetic Studies (DIGS) (Nurnberger et al., 1994; Joo et al., 2004). Exclusion criteria based on additional information garnered from medical records included subjects younger than 18, a history of any kind of organic brain abnormality, an alcohol problem, drug abuse, or other physical conditions presenting with psychiatric symptoms. A total of 350 unrelated Korean normal controls, which included 150 from the pooled set of the subjects, were recruited randomly from healthy volunteers following a brief interview by a psychiatrist. Subjects with past history or current evidence of organic mental disorders, abuse of illegal substances, and any medical conditions that might give rise to mental symptoms were excluded. All subjects with schizophrenia and healthy controls in this study were ethnically Korean.

SNP selection

One hundred forty-two putative SNPs for dopamine receptor genes were selected based on the information of the dbSNP (http://www.ncbi.nlm.nih.gov/SNP/). The function and heterozygosity of SNPs were considered mainly for the inclusion. Coding or regulatory SNPs showing a minor-allele frequency of more than 1% among Asians have been selected. Furthermore, those SNPs within 2 kb upstream of the transcription start site, the first intron and all exonic regions of five genes, DRD1-5, were selected. Among the selected 142 SNPs of DRDs, eleven SNPs were missense variants (DRD1: rs5330, rs5331, DRD2: rs1800496, rs1110977, DRD3: none, DRD4: rs1800443, DRD5: rs6282, rs2227843, rs2227852, rs2227847, rs2227842, rs2227851). The first intron sequence for each gene has been investigated since DNA methylation has become an issue in the pathophysiology of psychiatric diseases including schizophrenia and in the mechanism of action of psychotropics (Newton and Duman, 2006; Muir and McKechanie, 2009). The methylation of CpG islands is often involved in tissue-specific gene expression and has been identified in Intron 1 of DRD4 and DRD2 (Kamakura et al., 1997; Zhang, 2007). These genes span 125.12 kb; information on the exonic structure and transcript details of each gene are available on the AceView website (http://www.ncbi.nlm.nih.gov/IEB/Research/Acembly/). We then amplified and screened these SNPs for heterozygosity in a pooled DNA sample of 300 individuals (pools of 150 cases and 150 controls each) based on SNPs with a minor allele frequency of 1.5-48.5% using MALDI-TOF MS (Sequenom, San Diego, CA). For each case- and control-based pool and each SNP, six allele frequency estimates were calculated, derived from three separate PCR amplifications and two separate panels.

DNA extraction and pool construction

Genomic DNA was extracted from subjects' blood samples. DNA was measured with Fluoroskan Ascent (Thermo Labsystems, Franklin, MA) together with Pico green reagents and kits (P-7589, Molecular Probes, Eugene, OR). The selected DNA samples were diluted to a standard concentration, and individual aliquots of DNA were transferred into a single tube by using the Hamilton ML2200 and Vivace automated pipetting stations to ensure that a constant amount of each DNA sample was transferred to the pool. The pool was then mixed gently and requantitated before further dilution to a working concentration of 5 ng/µl.

PCR, MassEXTEND (SEQUENOM, San Diego) reactions

For all assays of allelotyping or genotyping processed in this stage, the same conditions were used. The MALDI-TOF MS assay was performed under standard conditions set forth by Buetow et al. (2001) with little modification. PCR amplification was performed in a total volume of 5 µl containing Genomic DNA (25 ng for pooled and 2.5 ng for individual), 0.5 units of Taq polymerase (HotStarTaq, Qiagen, Valencia, CA), 25 µmol of each dNTP (Promega, Madison, WI), 50 pmol of gene-specific PCR primer 1 and 2 carrying a universal sequence tag, and 10 pmol of universal sequence primer using 384-well microtiter plates. Cycling was performed in a Thermocycler (model 9600; Perkin-Elmer; Branchburg, NJ) at 95℃ for 15 min, followed by 45 cycles of 95℃ for 20 s, 56℃ for 30 s, 72℃ for 30 s, and a final extension of 3 min at 72℃.

A shrimp alkaline phosphatase (SAP) reaction was performed to dephosphorylate unincorporated dNTPs. The temperature conditions were 37℃ for 10 min followed by the inactivation of SAP at 85℃ for 5 min.

Subsequently, homogenous MassEXTEND reactions (hME™, SEQUENOM) were performed by using triple terminator mixes. Assays were grouped according to SNP-specific requirements on the termination mixes (ddACG, ddACT, ddAGT, and ddCGT, respectively). The final reaction volume of 9 µl comprised 1 unit of Thermosequenase (Amersham Pharmacia, Piscataway, NJ), 50 µM of the respective termination mix (ddTTP, ddATP, ddCTP, ddGTP), and 20 pmol of the test-specific MassEXTEND primer. A universal temperature program was applied, comprising an initial denaturation at 95℃ for 2 min followed by 40 cycles of 95℃ for 5 s, 52℃ for 5 s, 72℃ for 5 s, and finally 72℃ for 1 min. Primer extension products were resolubilized from the solid support by applying an ammonium hydroxide solution to the beads. For the subsequent MALDI-TOF MS analysis, a 15-nl sample was transferred from a 384-well microtiter plate serially onto patches of a 384-element silicon chip preloaded with 3-hydroxy-picolinic-acid (3-HPA) (SpectroCHIP, SEQUENOM).

MALDI-TOF MS analysis and allele frequency determination

Application of the analyte solution to the preloaded matrix patches of the silicon chip dissolved the matrix. After the solution evaporated, homogenous crystals of matrix and analyte formed. An automated analysis of these samples was performed on a SEQUENOM-Bruker MassARRAY mass spectrometer by scanning the 384 elements with a 337-nm laser pulse. Twenty shots were summed per element to yield the final spectra.

Mass spectra were processed by proprietary software (SPECTROTYPER, SEQUENOM) using baseline correction, peak identification, and peak area calculation algorithms. Normalized peak areas were computed as individual peak areas divided by the sum of the total peak area.

Genotype assays for individual samples of extension

After suggestive SNPs for association with schizophrenia were identified from analyses using the individual genotyping method (Stage 2), we performed individual genotyping with the extended sample (270 schizophrenia sample and 350 controls) for suggestive SNPs by TaqMan™ assays (Applied Biosystems, Foster City, CA). A total reaction volume of 5 µl included 50 ng of template DNA, 2.5 µl 2 × TaqMans™ Universal PCR Master Mix, and 0.5 µl 20 × SNP Genotyping Assay Mix (including primers and fluorescently labeled probes). PCR cycling was performed using an ABI Prisms 7900HT SDS (Applied Biosystems) under the following conditions: 95℃ for 15 minutes, followed by 40 cycles of 95℃ for 15 seconds and 60℃ for 1 min. The PCR was followed by allelic discrimination using the ABI Prisms 7900HT SDS. A confirmatory process was adopted; specifically, 10% of all samples were regenotyped by the same method.

Statistical analysis in individually genotyped samples

Allele and genotype associations for single markers were evaluated with contingency chi-square tests and exact tests. The pair-wise LD statistic r2 was calculated with unphased genotype data, and the general pattern of LD was investigated in both the schizophrenia group and the control group. All the genetic data analyses including the HWE evaluation were performed by PowerMarker Version 3.25 (Liu and Muse, 2005). The odds ratio and confidence interval in comparison with allele frequencies were obtained using SPSS Version 12.0 (SPSS, Inc., Chicago, IL).