Associations between a locus downstream DRD1 gene and cerebrospinal fluid dopamine metabolite concentrations in psychosis

Dopamine activity, mediated by the catecholaminergic neurotransmitter dopamine, is prominent in the human brain and has been implicated in schizophrenia. Dopamine targets five different receptors and is then degraded to its major metabolite homovanillic acid (HVA). We hypothesized that genes encoding dopamine receptors may be associated with cerebrospinal fluid (CSF) HVA concentrations in patients with psychotic disorder. We searched for association between 67 single nucleotide polymorphisms (SNPs) in the five dopamine receptor genes i.e., DRD1, DRD2, DRD3, DRD4 and DRD5, and the CSF HVA concentrations in 74 patients with psychotic disorder. Nominally associated SNPs were also tested in 111 healthy controls. We identified a locus, located downstream DRD1 gene, where four SNPs, rs11747728, rs11742274, rs265974 and rs11747886, showed association with CSF HVA concentrations in psychotic patients. The associations between rs11747728, which is a regulatory region variant, and rs11742274 with HVA remained significant after correction for multiple testing. These associations were restricted to psychotic patients and were absent in healthy controls. The results suggest that the DRD1 gene is implicated in the pathophysiology of psychosis and support the dopamine hypothesis of schizophrenia.


Introduction
Dopaminergic innervation is prominent in the central nervous system and is critically implicated in many central and peripheral functions [5]. The catecholaminergic neurotransmitter dopamine is released from the presynaptic terminals of dopaminergic neurons and exerts its action by targeting five different G-protein-coupled receptors located both pre-and postsynaptically [5]. Dopamine is then degraded to its major metabolite homovanillic acid (HVA) by catechol-O-methyltransferase and monoamine oxidase.
Dopaminergic dysfunction is considered to be implicated in various mental disorders, mainly schizophrenia [5,9,22]. Schizophrenia affects approximately 1% of the world's population with a heritability up to 80% [37]. Many dopaminergic gene variants have been associated with the disorder, however the results have been ambiguous and difficult to replicate until recently, when a genome-wide significant association was found between the dopamine receptor D2 (DRD2) gene and schizophrenia [1]. The association between dopaminergic gene variants and measurable biological markers may be more robust and consistent than the associations between gene variants and the disorder itself. Moreover, this approach can shed further light to the understanding of genotype-phenotype associations.
We consider the cerebrospinal fluid (CSF) HVA to be a relevant measurable marker as it reflects the dopamine turnover rate in the central nervous system (CNS), it is partially genetically determined and it has been associated with schizophrenia. Concentrations of HVA in ventricular, cisternal, and lumbar CSF show a craniocaudal gradient [25,28]. Studies in postmortem human brains have shown that CSF HVA reflects brain HVA concentrations [36,42]. Studies in human twins and other primates have shown that HVA concentrations are partially under genetic influence [30,32]. HVA concentrations are significantly lower in drug-free patients with schizophrenia relative to controls [6,43]. Moreover, quetiapine and olanzapine administration was found to be associated with a significant increase in CSF HVA [26,33].
In the present study, we have searched for association between DRD1, DRD2, DRD3, DRD4 and DRD5 single nucleotide polymorphisms (SNPs) and dopamine turnover rate in the CNS, as reflected by the CSF concentrations of the major dopamine metabolite HVA, in patients with psychosis.

Subjects
The subjects of the present study have been investigated as previously described [4]. Patients with psychotic disorder were recruited among inpatients at four psychiatric university clinics in Stockholm County between 1973 and 1987 and were asked to participate in pharmacological and/or biological research projects [8]. The participants were observed for at least 48 h without any antipsychotic medication and CSF samples (12.5 ml) were drawn by lumbar puncture.
Three to 34 years after the first investigation, patients were asked to participate in genetic research studies and blood was drawn for genotyping. Patients were then asked to participate in a diagnostic structured interview [35] and permit the researchers to retrieve their medical records. Available records were scrutinized by researchers to obtain a life-time diagnosis according to DSM-III-R and DSM-IV. In 2010, hospital discharge diagnoses were also obtained from the Swedish psychiatric inpatient register, a register covering all inpatient hospitalizations in Sweden since 1973. For each hospitalization the diagnosis was recorded according to the International Classification of Diseases, 8th, 9th or 10th revisions.
The majority of the participants had experienced several hospitalizations, but only one diagnosis was given per participant, following a diagnostic hierarchy [14,41]. The final diagnoses, used in the present study, were based on the Swedish psychiatric inpatient register, as it was not possible to retrieve all medical records and several of the patients were not willing to participate in a diagnostic interview.
Analyses in healthy controls were conducted for SNPs that were nominally associated with HVA concentrations in psychotic patients, in order to evaluate whether the effects of the associated SNPs were restricted to patients with psychosis. CSF samples were drawn by lumbar puncture from unrelated healthy Caucasians between 1973 and 1987. Eight to 20 years after the first investigation, the subjects were interviewed to re-assess the absence of psychiatric morbidity [18] and whole blood was drawn for genotyping.
The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the Karolinska University Hospital. Written informed consent was obtained from all the participating subjects.

CSF monoamine metabolite concentrations
CSF samples (12.5 ml) were drawn by lumbar puncture with the patients and controls in a sitting or recumbent position between 8 and 9 a.m, after at least 8 h of bed-rest and absence of food intake or smoking. HVA concentrations were measured by mass fragmentography with deuterium-labeled internal standards [38].

DNA analysis
Genomic DNA was extracted from whole blood. Totally, 67 SNPs were genotyped in DRD1 (n = 17), DRD2 (n = 23), DRD3 (n = 18), DRD4 (n = 7) and DRD5 (n = 2). These SNPs were either candidate SNPs reported to be associated with mental disorders, mainly schizophrenia, enzyme function or monoamine metabolite concentrations, or tag-SNPs selected using HapMap to cover the genes of interest with an r 2 threshold of 0.8. The genotyping was performed using the Illumina BeadStation 500GX and the 768-plex Illumina Golden Gate assay (Illumina Inc., San Diego, CA, USA) [15].

Statistical analysis
The associations between SNPs and HVA CSF monoamine metabolite concentrations were tested with multiple linear regression (STATA 12.1), where concentration was modeled as a linear function of the allele count and three to five covariates. In the analyses of psychotic patients, back-length, gender, age at lumbar puncture, diagnosis (i.e., schizophrenia spectrum psychosis or other psychosis) and use of antipsychotics were included as covariates. Back-length was defined as the distance between the external occipital protuberance and the point of needle insertion. Antipsychotic treatment was considered as present if the patient had taken antipsychotics during a three-week period prior to the lumbar puncture. In the analyses of healthy controls, back-length, gender and age at lumbar puncture were included as covariates. Hardy-Weinberg (HW) equilibrium was tested using exact significance as implemented in STATA 12.1. Normality of residuals was checked graphically with STATA 12.1. Adjustments for multiple testing were performed using Bonferroni correction taking into account the number of tests conducted (␣= 0.05/67 = 7.5 × 10 −4 ).

Results
Psychotic patients (n = 74, 45 men and 29 women) participated in the present study. The mean age of disease onset ± standard Table 1 Minor allele frequencies (MAF), exact significance of testing for Hardy-Weinberg equilibrium (HWE) and p-values (P) from multiple linear regressions of single nucleotide polymorphisms (SNPs) nominally associated with homovanillic acid (HVA) concentrations in the cerebrospinal fluid of psychotic patients and the corresponding association statistics among healthy controls. deviation was 27.6 ± 7.8 years, whereas their mean age ± standard deviation was 30.4 ± 7.2 years at lumbar puncture. Twenty-six of the patients were treated with antipsychotics at the time of lumbar puncture, whereas thirty-six patients were free from antipsychotic medication since three weeks or longer. Sixty-four patients were diagnosed with schizophrenia spectrum disorder (schizophrenia n = 60, schizoaffective disorder n = 4) and ten with other psychosis (psychosis not otherwise specified n = 7, delusional disorder n = 1, bipolar disorder n = 1, alcohol induced psychotic disorder n = 1). In order to evaluate how the Swedish in-patient resister-based diagnoses conformed to other diagnostic tools, separate analyses were conducted. Evaluations originating from the medical records in 52 of the patients resulted in a diagnosis of a psychotic disorder in 98% of these individuals. Of 44 patients participating in a diagnostic interview 91% displayed a psychotic disorder [35]. These results are in accordance with previous reports showing that the registerbased diagnoses of schizophrenia spectrum psychosis have a high validity, with 85% to 94% of the patients displaying these diagnoses when diagnostic evaluations using information from medical records and a structured clinical interview were made [41].
In the 74 psychotic patients, 67 SNPs in five genes encoding dopamine receptors were selected and genotyped. The results are illustrated in Supplementary Table S1 (see Supplementary Table  S1 in the online version DOI: 10.1016/j.neulet.2016.03.005). The minor allele frequencies for the selected markers ranged from 2% to 50%. In psychotic patients, the mean (standard deviation) concentration of HVA was 178.6 (79.3) nmol/L. Six of the investigated polymorphisms were found to be nominally associated with CSF HVA concentrations and none of these SNPs showed departure from Hardy Weinberg equilibrium (p-value < 0.05) ( Table 1). The residuals of the nominal associations were approximately normally distributed.
The six SNPs that were associated with HVA concentrations in psychotic patients were tested in 111 healthy Caucasians (63 men and 48 women). Their mean ages ± standard deviation were 28.4 ± 7.5 years at lumbar puncture. In healthy controls, the mean (standard deviation) concentration of HVA was 167.5 (68.4) nmol/L. The minor allele frequency for the six selected markers ranged from 12% to 39%. Departure from Hardy Weinberg equilibrium (p < 0.05) was found in one of the SNPs analyzed. The residuals were approximately normally distributed. No associations were found in healthy individuals between the selected SNPs and HVA (Table 1).
Preliminary analysis excluding the SNPs showed that HVA concentrations were not associated with antipsychotic treatment in psychotic patients. Our independent variables, i.e., the SNPs, are not expected to be associated with the presence or absence of antipsychotic treatment and moreover we have included the use of antipsychotics as a covariate in our analyses. Thus, the use of antipsychotics in a cluster of patients should not confound our analyses.

Discussion
To our knowledge, there is only one previously published study searching for associations between gene variants and CSF monoamine metabolite concentrations, including HVA, in patients with psychosis. In that study, nominal associations between genes encoding enzymes implicated in the monoamine metabolism and CSF monoamine metabolite concentrations were found [4]. No previous studies have searched for associations between dopamine receptor gene variants and CSF monoamine metabolite concentrations in psychotic patients.

DRD1
Almost all independent studies and one meta-analysis failed to show evidence of association between DRD1 gene variants and schizophrenia (http://www.szgene.org). DRD1 has been associated with bipolar disorder [11,34], attention-deficit/hyperactivity disorder [7] and autism [17]. DRD1 rs4532 has been associated with treatment response to antipsychotic drugs [29] and tardive dyskinesia in patients with schizophrenia [21]. In the present study, rs4532 was not associated with CSF HVA concentrations.
We identified four SNPs, i.e., rs11747728, rs11742274, rs265974 and rs11747886, located in a region 3.5-8 kbp downstream DRD1, that were associated with CSF HVA concentrations in psychotic patients. The associations between rs11747728 (p value = 0.0004) and rs11742274 (p value = 0.0005) with CSF HVA remained significant after correction for multiple testing. The SNP with the lowest p value, rs11747728, is a regulatory region variant, located 4 kbp downstream DRD1. These associations were restricted to psychotic patients and were absent in healthy controls. These SNPs have not been ascribed any association with schizophrenia or other mental disorders.
D1 receptors are expressed at a high level of density in various regions in CNS [5] and the lack of association between DRD1 and schizophrenia leads to other approaches, such as the investigation of intermediate phenotypes, in order to implicate DRD1 in the disease processes. A recent study displayed that the expression of DRD1 mRNA was decreased in dorsolateral prefrontal cortex of patients with schizophrenia compared to controls [19]. The DRD1 density in different brain regions in patients with schizophrenia and healthy controls has also been studied with positron emission tomography scan with some studies finding an increased DRD1 availability in prefrontal cortex in patients relative to controls [2], mainly in the case of drug-naïve patients [3]. This increase may rep-resent a compensatory upregulation secondary to a mesocortical dopamine function deficiency. The identification of a locus downstream DRD1 affecting the major degradation product of dopamine supports the hypothesis of the implication of DRD1 in psychosis and generally the dopamine hypothesis of schizophrenia.

DRD2
DRD2 gene variations have been extensively investigated for associations with schizophrenia, with numerous positive studies, three positive meta-analyses (www.szgene.org) and a positive genome-wide association study [1]. DRD2 has also been associated with mood disorders, substance use disorders [44], Tourette's syndrome and post-traumatic stress disorder [27].
The most robust association found was between DRD2 rs6277 and schizophrenia with a positive meta-analysis with an odds ratio of 1.29 (www.szgene.org). Rs6277 has been included as a candidate SNP in the present study and has not been found to be associated with CSF HVA concentrations in psychotic patients. Another interesting result is the identification of a SNP, i.e., rs2514218, located 47 kbp downstream DRD2 found to be genome-wide significantly associated with schizophrenia [1]. Rs2514218 was not included in the present study.
In the present study, DRD2 rs2234689 was nominally associated with CSF HVA concentrations in psychotic patients. DRD2 rs2234689 is a downstream gene variant and has not been ascribed any functionality or association with mental disorders.

DRD3
DRD3 gene variants have been associated with schizophrenia in many independent studies. However, all meta-analyses conducted failed to confirm significant associations (www.szgene.org). DRD3 SNPs have also been associated with other mental disorders such as autism [10] and unipolar depression [12].
In the present study, no DRD3 SNP was associated with CSF HVA concentrations in patients with psychotic disorder.

DRD4
DRD4 gene variants have been associated with schizophrenia in some independent studies and two meta-analyses (www.szgene. org), as well as with other mental disorders, mainly ADHD [31].
In the present study, DRD4 rs3758653 was nominally associated with CSF HVA concentrations in patients with psychosis. Rs3758653 is an upstream gene variant previously reported to be associated with DNA methylation across the DRD4 promoter region in both lymphoblastoid cell lines and post-mortem brain tissue [13]. Rs3758653 has been also reported to be associated with Alzheimer's disease [23].

DRD5
DRD5 gene variants have been associated with schizophrenia (www.szgene.org) and ADHD [20] in independent studies. In the present study, no DRD5 SNP was associated with CSF HVA concentrations in psychotic patients.

Limitations
The present study suffers from some limitations. First, being a genetic study the numbers of participants is small. Moreover, the assumption that a Bonferroni correction for the total number of tests conducted is a sufficient correction for a genetic study may be considered as a limitation, as there is no sufficient prior evidence to limit the analysis to the specific genes. However, as we have not tested the whole genome and have used a strong a priori hypothesis, i.e., that the selected dopamine-related genes may be regarded likely to influence dopamine metabolite concentrations in patients with psychosis, we have not applied the standard threshold for genome-wide significance but a Bonferroni correction taking into account the total number of tests conducted. The results must be seen as tentative, and there is need for independent replication. Also, several of the patients had been medicated with antipsychotics drugs. However, from a statistical point of view, the various use of antipsychotics among the patients should not confound the results as the presence of antipsychotics has not been found to be associated with HVA concentrations in the present sample and is not expected to be associated with the investigated SNPs.

Conclusions
In psychotic patients, we found two significant and two nominal associations between SNPs located at a locus downstream DRD1 and the CNS dopamine turnover rate, as reflected by the CSF concentration of HVA. These associations were present in patients with psychotic disorder and absent in healthy controls. The present study suggests that the DRD1 gene is implicated in the pathophysiology of schizophrenia and supports the dopamine hypothesis of schizophrenia.