Introduction

Schizophrenia (F20, International Classification of Diseases—10th Revision (ICD X)) is a severe mental illness characterized by the alteration of higher functions, deterioration of behavior, cognition, emotions, motivation and perception, and marked by socio-occupational dysfunction. A wide variety of positive (auditory hallucinations, paranoid delusions), negative (affective flattening, anhedonia, alogia) and cognitive (declined attention, memory) symptoms comprise its diverse clinical heterogeneity, which, together with the lack of decisive laboratory tests, set obstacles for unanimous diagnosis.1 Typically, the onset in male patients is in young adulthood, between 15 and 25 years, and in women it emerges about a decade later.2 The lifetime incidence of schizophrenia is estimated to be about 1% worldwide, with untypical rates among certain groups of the society.2 In DSM-IV-TR (Diagnostic and Statistical Manual of Mental Disorders, 4th edition, text revision, 2000), the criteria of schizoaffective disorder (F25, ICD X) include both characteristic schizophrenia symptoms (generally hallucinations or delusions) and affective symptoms (manic, depressive or mixed), which can be observed during the same episode of the illness, although during some periods the psychotic symptoms can be observed independently from prominent mood disorder features. Similar to schizophrenia, the onset of schizoaffective disorder is also usually in early adulthood, but it is more common in women than in men.3

The high heritability of schizophrenia was estimated to be about 80% and its complex (non-Mendelian) mode of transmission was suspected through the abundance of data from family, twin and adoption studies.1, 4 It was also suggested that the disorder is a result of the complicated additive interaction between multiple genes, each having a small effect.2, 5, 6 The multifactorial polygenic threshold model, sustained by a number of linkage and association studies, proposes that both genetic background and environmental factors, including infections, stress and trauma, are necessary for triggering schizophrenia.4, 5 The case–control association studies have been proven to be a more appropriate method for detection of genes with small effect, compared with linkage studies, as they allow recognition of relatively small differences if the sample size and the selection criteria are adequate.1, 7

Several hypotheses concerning the etiology of schizophrenia have been proposed, and most of them have focused on the alterations during neuronal development.8 According to the developmental model, various prenatal and perinatal noxae cause formation of inappropriate synaptic contacts, irregular cell signaling and migration during the fetal period in brain areas important for development of integration in the society. Such maldevelopment does not interfere with the basic brain functioning in the early years, but expresses itself when the subject is stressed by demands for integration in young adulthood.9, 10

The biochemical models are based on alterations in the neurotransmitter pathways supported by the effect of some of the antipsychotic medicaments that block certain receptors in the brain. The classical dopaminergic hypothesis asserts that the excessive dopaminergic signaling causes the psychotic symptoms, and is a result of increased sensitivity and density of dopamine D2 receptors in some brain regions.2, 11 Similar are the serotonin, glutamate and other metabolic hypotheses, but none of them can sufficiently explain the complex symptoms of schizophrenia, and it is still unclear whether the affection of those circuits is a consequence of the treatment and years of persistency, rather than a cause for the disease.1, 2

Despite the incontestable significance of the genetic background of schizophrenia and great efforts in a large number of studies to clarify it, none of the candidates suspected to be involved in the pathogenesis was conclusively confirmed. To date, there are still no causative genes that are commonly accepted, even though some candidates were replicated in different studies and seemed to be promising.12 Population heterogeneity, phenotypic heterogeneity, interweaving diagnostic criteria and misclassification, overestimated risk in the initial study, small sample size, influence of multiple loci (causal heterogeneity) and genetic–environmental interactions are among the indicated causes for the failure of replication of association studies.5, 7, 12, 13 Some of the strongest causative candidates to date are considered to be DISC1, DTNBP1, COMT, NRG1 and RGS4.

The aim of this study was to research the association between the common genetic variants in candidate genes, recently reported in the literature, and schizophrenia in an independent collective group from Bulgarian patients with schizophrenia or schizoaffective disorder, who differ in their genetic background from patients analyzed in other studies.

Materials and methods

Subjects

In this study, we have collected 811 unrelated Caucasians from Bulgaria, namely 198 patients with schizophrenia, 57 with schizoaffective disorder and 556 unrelated healthy control individuals. Samples from patients with schizophrenia (paranoid form) or schizoaffective disorder were recruited at four regional University psychiatric clinics—1st Psychiatric Dispensary of Sofia, Psychiatric Dispensary of Plovdiv, Psychiatric Dispensary of Blagoevgrad, and Psychiatric Dispensary of Radnevo. The patients were examined and diagnosed, according to the criteria of the DSM-IV, by two psychiatrists independently. The psychiatrists conducted interviews and provided written information to the participants. A written informed consent form was signed before the collection of blood samples. The identity of the patients was known only by the interviewing psychiatrists and by the head of the corresponding psychiatric clinic. All samples and accompanying information (age, gender, severity and length of affection) were identified only by a unique code. All patients were on antipsychotic therapy. The mean age of the patients was 45.3 years (range: 23–81; s.d.±10.9) and the mean age of disease onset was 26.9 years (range: 13–52; s.d.±8.4). Of the 255 affected patients, 124 (49%) were males and 131 (51%) were females.

The control samples were collected from healthy volunteers in Bulgaria, who have provided written informed consents as well. This group included 289 (52%) males and 267 (48%) females, with an average age of 50.5 years (range: 18–86; s.d.±16.0). Each individual of the control group was interviewed by a physician so that any psychiatric disorder could be excluded, and data on their general medical history and information about their age, gender and ethnos were obtained. Ethical approvals for conducting the study were obtained from the Ethical Committees in the SNP Research Center (present name: Center for Genomic Medicine), Yokohama Institute, The Institutes of Physical and Chemical Research (RIKEN, Yokohama, Japan) and the Local Medical Ethic Committees from each of the participating institutions.

DNA sampling

DNA was extracted from the peripheral venous blood (30 ml) and was stored in the DNA Bank at the Department of Medical Genetics (Medical Faculty, Medical University, Sofia, Bulgaria). Each tube was marked with a unique identification code. The DNAs were extracted by a standard phenol–chloroform procedure and stored at −20 °C before their transportation to Japan, where the genotyping was performed.

Genotyping

A total of 64 single nucleotide polymorphisms (SNPs) from 25 genes were determined with commercially available pre-designed TaqMan probes and primers by means of the GeneAmp PCR System 9700 from Applied Biosystems (Foster City, CA, USA), according to the manufacturer’s protocol. The genotypes were determined by an ABI PRISM 7900HT Sequence Detection System (Applied Biosystems). The remaining 119 SNPs were genotyped by a combined Multiplex PCR and Invader assay, as described elsewhere.14

Gene and SNP selection

A total of 59 candidate genes from 18 chromosomes were selected if they met at least one of the following criteria: evidence for positive association with schizophrenia from previous case–control or family-based studies; positional candidates—located in some of the most promising candidate regions from genome-wide scans and large-scale studies of schizophrenia; or functional candidates—plausible involvement in the disease pathogenesis consistent with the neurodevelopment and biochemical (neurotransmitter and cell signaling) etiological models. Some candidate genes were excluded when most of their previously reported markers were already examined in Bulgarian trios in earlier studies.

We have selected 202 candidate SNPs from 59 candidate genes according to the public information in NCBI (Builds 35 and 36), the Schizophrenia Research Forum (http://www.schizophreniaforum.org) and the literatures. Most of the SNPs had minor allele frequencies of >0.05, although exceptions were made for non-synonymous SNPs with an unknown frequency from genes of interest as well as from six SNPs with minor allele frequencies of <0.05 showing an association in Caucasian studies. For some genes (DISC1, DTNBP1, RGS4, GABRB2, ERBB4 and ERBB3), we have selected additional non-synonymous SNPs, and polymorphisms at the 5′ and 3′ untranslated regions with presumptive effect over the transcriptional level. Detailed information about genotyped SNPs is illustrated in Table 1.

Table 1 Candidate genes and SNPs successfully genotyped in the first screening for association with schizophrenia
Full size table

Statistical analyses

Genotype and allele frequencies were evaluated by using the two-tailed Fisher’s exact test in allelic, dominant- and recessive-inheritance models. Deviations from the Hardy–Weinberg equilibrium were evaluated by the HAPLOVIEW 4.0 program (Broad Institute of MIT and Harvard).

Results

This study was performed in two subsequent screenings. In the first screening, all selected candidate SNPs were genotyped in 185 cases and 184 controls collected in three major psychiatric centers in Bulgaria. During the first screening, additional samples for verification were collected from the dispensary in Radnevo. All SNPs showing a P-value of <0.05 in the two-tailed Fisher’s exact test in the first screening proceeded to the second screening. Three SNPs (two from DISC1 and one from DTNBP1 genes) showed deviation from the Hardy–Weinberg equilibrium in the case group and were excluded from further analysis. Six SNPs (two from the ERBB3 gene and four from the ERBB4 gene) were found to be monomorphic in this population.

The genotyping in the initial set of samples identified 21 SNPs as possible candidates for susceptibility for schizophrenia/schizoaffective disorder with a P-value of <0.05 by Fisher’s exact test. Subsequently, these 21 variants were genotyped in an additional 70 case samples from Bulgarian patients with schizophrenia and 372 Bulgarian control samples. Two polymorphisms—rs2255340 from DISC1 gene and rs909706 from DTNBP1 gene—were included in the follow-up screening, because some trend for association was considered in the initial genotyping and because those two candidates were suggested to be important functional candidates in previous studies. Probably because of the small sample size of the additional independent case group, only one SNP (rs6277) from the DRD2 gene revealed a positive association (P=0.00067, Table 2). The SNP remained significantly associated with schizophrenia after the adjustment for multiple testing, according to strict Bonferroni’s correction. These data, together with the negligible discrepancies of the allele and genotype frequencies with previous studies in Caucasian populations: Northern European populations living in Australia, Spain, Finland and Russia,15, 16, 17, 18 support the association of DRD2 with schizophrenia.

Table 2 Genotyping results for 21 SNPs after the first and the second screening
Full size table

Discussion

We have performed a case–control association study for schizophrenia in the Bulgarian population using 183 SNPs present in 59 candidate genes, the implication of which in the disease etiology had been previously reported. We have successfully assessed 180 SNPs in a group of 185 cases and 184 controls, and selected 21 SNPs from 15 genes to be possible candidates for the association (P<0.05). Later, we managed to obtain additional sets and performed adjustments for multiple testing, according to strict Bonferroni’s correction. Finally, we implied that the marker, rs6277 (DRD2), preserved statistical significance for association with schizophrenia/schizoaffective disorder. We could replicate only rs6277 (corrected P-value of 0.014). However, the previously reported markers were not replicated, either because of the limitations in the number of patients and controls (lack of the power) or because of false-positive results in papers reported earlier.

Neurotransmitters have been considered to be involved in the development of schizophrenia in the past. Dysfunction of the dopaminergic pathway in its pathogenesis has been extensively discussed in the literature for the last few decades. The classical dopaminergic hypothesis remains one of the most commonly acknowledged hypotheses, although it still remains unclear whether the observed alterations are causal or result from long-term treatment with antipsychotic medications. It postulates that psychotic symptoms are a consequence of dopamine excess in specific brain regions (particularly in the mesolimbic and nigrostriatal areas).2 Postmortem and in vivo neuroimaging studies indicate increased dopamine D2 receptor density and binding affinity in the striatum of schizophrenic patients.11, 19, 20 Enhanced expression of DRD2 in the caudate nucleus was implicated in the cognitive dysfunction of affected individuals.2 Furthermore, common antipsychotic medications are capable of affecting the positive symptoms of the disease2 as a result of antagonistic or partially agonistic interactions with the dopamine D2 receptor, whereas dopamine agonists, such as amphetamines, provoke schizophrenia-like psychosis.2, 11, 21

The dopamine receptor genes comprise a large superfamily that encodes G-protein-coupled receptors important for regulation of some higher functions, such as locomotion, cognition and emotions.22, 23, 24 Five human dopamine receptors are described and classified into two major groups, dopamine 1-like (D1) and dopamine 2-like (D2), in accordance with their transcriptional and pharmacological properties.25 The dopamine D2 receptors (DRD2) are abundant on the postsynaptic membranes in the mesocortical pathway,26 implicated in the control of the feelings of reward and pleasure. At least two different isoforms of the protein exist as a result of alternative splicing of exon 6.23, 27

Several case–control association studies have been performed to evaluate the likely contribution of variants in the DRD2 gene to schizophrenia susceptibility. Positive results for the association between the A allele of the TaqI A polymorphism (rs1800497) and schizophrenia were discovered by case–control studies in few independent groups, where an overtransmission of certain haplotypes, including the TaqI A2 allele, was observed.28, 29 After Itokawa et al.30 focused on the implication of the Ser311Cys polymorphism, Arinami et al.31 were the first to describe a positive association between this non-synonymous SNP of the Dopamine D2 receptor gene and schizophrenia. However, over 20 subsequent studies attempted to elucidate this variant’s contribution to the disorder and most of them failed to reveal that correlation. Comprehensive meta-analyses suggested a weak but positive effect of the Cys allele as the risk factor for susceptibility to schizophrenia (P=0.007)32, 33 In another study, Arinami et al.34 reported a positive association for another DRD2 gene polymorphism, −141C (Ins/Del), which became one of the most commonly tested in the quest for susceptibility variants. In an attempt to replicate these data, several case–control studies and meta-analyses were conducted, with controversial results.2, 35

Our results in the Bulgarian sample group (Table 2) are consistent with the previous reports by other researchers on the positive association between the synonymous C957T (rs6277) polymorphism and schizophrenia, in different Caucasian populations.15, 16, 17, 18 In our study, the G allele (allele 2), which corresponds to the C allele in the single nucleotide polymorphism database (dbSNP) and the literature, was identified as the risk allele. After the initial study by Lawford et al.,17 reporting association of rs6277 in the Australian (Caucasian) population, three subsequent studies confirmed significantly higher frequencies of the C allele and the C/C genotype in patients with schizophrenia who were of European descent (in the Finnish,15 Spanish16 and Russian18 populations). However, a case–control study using subjects from India failed to confirm the positive relation with schizophrenia, although some tendency for association was seen.36 Furthermore, the results of two meta-analyses, including the data from all five studies18 or those from Caucasian populations,37 found a nominally significant association of the C allele and the C/C genotype with schizophrenia with pooled allelic odds ratios (ORs) of 1.42 (95% confidence interval (CI): 1.26–1.61)18 and 1.45 (95% CI: 1.21–1.73],37 and summary OR for the genotype of 1.60 (95%CI 1.32–1.95).18 As all positive associations were found in populations from European descent, ancestry-specific effect must be taken into consideration, and the genetic effect of this variant across populations with different ancestral descents must be evaluated.37

Our findings have provided additional evidence for the importance of variants in the DRD2 genes for the vulnerability to schizophrenia. Despite rs6277 being a synonymous polymorphism causing no amino acid substitution (Pro319Pro), it was reported to cause some effect on the gene expression in an in vitro assay, suggesting that the 957T variant significantly affects the stability and the expression level of the DRD2 mRNA.15, 17, 23

Recent in vivo study in healthy volunteers discovered a correlation between the C957T polymorphic variants and the intrasynaptic dopamine levels and D2 receptor-binding affinity in the striatum. In support of the dopaminergic hypothesis for the etiology of schizophrenia, the highest binding potential of the receptor was found in the C/C homozygotes, whereas the T/T homozygotes showed the lowest affinity.15, 17, 38 Furthermore, C957T showed a population attributable risk for schizophrenia of 24% and an attributable risk in patients with schizophrenia of 42%.17

In summary, our findings further suggest that DRD2 may play an important, but most probably not independent, role in the pathogenesis of schizophrenia in the Bulgarian population. Previous genetic and functional studies point out this gene as a possible candidate for implication in schizophrenia etiology. Although the association of NOTCH4, DISC1, DTNBP1, RGS4 and other highly putative candidate genes was not confirmed after the statistical evaluation, we observed some tendency for association with schizophrenia in the Bulgarian population. Hence, these genes should be further evaluated using a larger set of samples.