PRODH Polymorphisms, Cortical Volumes and Thickness in Schizophrenia

Schizophrenia is a neurodevelopmental disorder with high heritability. Several lines of evidence indicate that the PRODH gene may be related to the disorder. Therefore, our study investigates the effects of 12 polymorphisms of PRODH on schizophrenia and its phenotypes. To further evaluate the roles of the associated variants in the disorder, we have conducted magnetic resonance imaging (MRI) scans to assess cortical volumes and thicknesses. A total of 192 patients were evaluated using the Structured Clinical Interview for DSM-IV (SCID), Positive and Negative Syndrome Scale (PANSS), Calgary Depression Scale, Global Assessment of Functioning (GAF) and Clinical Global Impression (CGI) instruments. The study included 179 controls paired by age and gender. The samples were genotyped using the real-time polymerase chain reaction (PCR), restriction fragment length polymorphism (RFLP)-PCR and Sanger sequencing methods. A sample of 138 patients and 34 healthy controls underwent MRI scans. One polymorphism was associated with schizophrenia (rs2904552), with the G-allele more frequent in patients than in controls. This polymorphism is likely functional, as predicted by PolyPhen and SIFT, but it was not associated with brain morphology in our study. In summary, we report a functional PRODH variant associated with schizophrenia that may have a neurochemical impact, altering brain function, but is not responsible for the cortical reductions found in the disorder.


Introduction
Schizophrenia is currently considered to be a neurodevelopmental disease that arises from a complex phase-specific interaction of genetic and environmental factors [1][2][3]. The prevalence of schizophrenia in the general population is estimated to be 0.3%-1.6% [4,5]. The high heritability (,64%-83%) of schizophrenia [6,7] may be due to a combination of multiple common alleles, each one with a small to moderate effect, and some rare alleles with much larger effect sizes [8].
One of the strongest known genetic risk factors for schizophrenia is the 22q11 deletion [9]. Up to one-third of patients with 22q11 deletion syndrome (22q11DS), also known as DiGeorge or velocardiofacial syndrome, develop schizophrenia or schizoaffective disorder [10]. Moreover, while the prevalence of the chromosome 22q11 deletion in the general population is one in 4000 individuals, its frequency is estimated to be approximately 1% in adult patients with schizophrenia [11][12][13][14]. Therefore, the 22q11DS deletion has been suggested as a genetic subtype of schizophrenia, and the genes located in this region might therefore contribute to susceptibility to the disease.
One of the genes commonly deleted in 22q11DS is the proline dehydrogenase (PRODH) gene, which encodes the proline dehydrogenase enzyme that catalyzes the first step in proline catabolism [15]. The PRODH gene, consisting of 14 coding exons and located in a low copy repeat sequence, is widely expressed in the brain and in other tissues [16,17]. The resulting enzyme, also known as proline oxidase (POX), is a mitochondrial tumor suppressor [18] that inhibits proliferation and induces apoptosis [19]. High proline concentrations appear to activate NMDA and AMPA receptors and may therefore act as a neuromodulator [20][21][22]. Proline is also a precursor of the neurotransmitter glutamate [23], which seems to be involved in schizophrenia pathophysiology. Moreover, whereas high levels of plasma proline are found on a relatively common basis in patients with 22q11DS [24,25], some evidence suggests that hyperprolinemia may lead to neurocognitive dysfunction in animal models [26,27] and may be involved in the cognitive and psychiatric features of 22q11DS [25].
Patients with 22q11.2 deletion syndrome present abnormalities in brain structure, such as a global brain volumetric reduction that includes several cortical regions, the cerebellum and the hippocampus [28]. Similarly, patients with schizophrenia consistently present brain volumetric reductions compared to healthy controls, specifically due to grey matter reductions of the anterior cingulate, frontal (particularly medial and inferior) and temporal lobes, hippocampus/amygdala, thalamus, and insula that may be magnified over time [29]. A widespread cortical thickness reduction has also been found in the frontal, parietal, temporal and occipital regions of patients with schizophrenia [30].
Only two previous studies have investigated the role of PRODH in brain volume. The first, employing a sample of 92 healthy controls, observed a decrease in the striatal grey matter volumes of carriers of the functional risk haplotype, which contains the rs4819756, rs2870983, and rs450046 minor alleles (haplotype: GCC) [31]. The other study investigated two other PRODH single nucleotide polymorphisms (SNPs -rs2008720 and rs372055) with respect to brain anatomy in a sample of 51 patients with schizophrenia. While no association was found for rs2008720, significant differences in grey matter (GM) and total brain volume were revealed for rs372055, with C carriers showing increased total GM and total brain volume in comparison with patients carrying the TT genotype [32].
However, the associations of other important functional PRODH SNPs with brain morphology have not been investigated, and neither previous study investigated the different roles of these polymorphisms in both normal and pathological conditions. In the present study, we aimed to associate 12 PRODH polymorphisms, and especially functional variants, with schizophrenia and other clinical variables such as treatment resistance, age of onset, and positive and negative symptoms. To further understand the impacts of the studied polymorphisms, we conducted MRI scans in a subsample of patients.

Ethics Statement
This study was approved by the Research Ethics Committee of UNIFESP [CEP No. 1737/06]. Written informed consent was obtained from all recruited participants or their caregivers, and the clinical and laboratory investigations were strictly conducted according to the principles expressed in the Declaration of Helsinki. All invited subjects agreed to enroll in this study. If the patients were within a stable period according to their clinical features and could understand the consent forms, they doublesigned the documents themselves. The next of kin, caretakers or guardians consented on the behalf of those participants whose capacity to consent was compromised. Other clinical variables were assessed using the Positive and Negative Syndrome Scale (PANSS); Calgary Depression Scale [33]; Global Assessment of Functioning (GAF); and Clinical Global Impression (CGI) [34] instruments. All available information, including medical records, were used to assess remission, which was defined by mild severity (score of 3 or lower) in eight items of the PANSS maintained for the 6 months before the assessment, as previously proposed [35]. The illness duration was directly assessed with the patient, and when in doubt, a companion or informant was questioned and/or the information was recovered from the available clinical records. Symptom clusters (Negative, Positive, Disorganization, Excited and Anxiety/Depression) were classified according to the PANSS ratings [35]. Treatment-resistant (TR) status was defined, in accordance with the International Psychopharmacological Criteria (IPAP) [www. ipap.org], as a failure to respond to 4-to 6-week trials of monotherapy with two different antipsychotics at adequate doses (equivalent to 5 mg of risperidone or 400 mg of chlorpromazine). A 1.5 Mb microdeletion in 22q11.2 was detected in one patient with schizophrenia using multiplex ligation-dependent probe amplification [11]. Therefore, this subject was excluded from the analyses, as this deletion could be a confounding genotype factor.

Subjects
A total of 179 healthy subjects with no family history of severe psychiatric illness were recruited by UNIFESP. These subjects were also investigated using a modified version of SCID screening to exclude any with a previous illness episode with psychotic features or a family history of similar diseases.
Considering that the Brazilian population is admixed, the population structure was verified for a sample of our study population (Table 1).

DNA Isolation and Population Structure Analysis
Whole blood was collected into EDTA tubes and genomic DNA isolation was performed using the Gentra Puregene Kit (Qiagen, Maryland, USA) according to the manufacturer's protocol.
The population genetic ancestry structure was assessed by a panel of 48 ancestry-informative insertion-deletion polymorphisms, previously described and validated in the Brazilian population [36], analyzed in three multiplex PCR assays followed by capillary electrophoresis.

Genetic Analysis
The characteristics of the study population are described in Table 1. First, the Hardy-Weinberg equilibrium was verified for all SNPs using the chi-square test. Logistic regression was then performed to associate each genotype with the diagnosis of schizophrenia, using age as a covariate. The Bonferroni correction for multiple comparisons was performed. The linkage disequilibrium and haplotypes were assessed using Haploview [37]. The associations between the haplotypes and the disease were investigated using the chi-square test.
Statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS version 15.0), and the level of significance was set at 0.05.

In silico Analysis
For the SNPs associated with schizophrenia, we tested the predictive values using SIFT (http://sift.jcvi.org/) and PolyPhen (http://genetics.bwh.harvard.edu/pph/) and analyzed the amino acid conservation using the UCSC genome browser (http:// genome.ucsc.edu). Both SIFT and PolyPhen predict whether an amino acid substitution affects protein function. SIFT prediction is based on the degree of conservation of amino acid residues in sequence alignments [38], while PolyPhen uses straightforward physical and comparative considerations [39]. with an eight-channel head coil. A series of exploratory sagittal images (nine to eleven slices of 5 mm with 1 mm spacing) was performed to evaluate the image quality and the positioning of the head of each subject. T1 images were acquired sequentially using a pulse sequence (SPGR) with the following parameters: TR = 2000 ms, TE = 3.42 ms, matrix size = 256 x 256, FOV = 245 mm, flip angle = 15 degrees, NEX = 1, 1.0 mm slice thickness with no gaps, yielding 192 slices.

Imaging Preprocessing -FreeSurfer
The MRI scans were analyzed using the FreeSurfer software package (version 5.0 http://surfer.nmr.mgh.harvard.edu) based on established processing steps (recon-all pipeline). The cortical volumes and thickness information of the brain regions were estimated using the automated parcellation of the Desikan-Killiany Atlas [40] and exported to SPSS for further analysis.

Imaging Group Analysis
First, a general linear model (GLM) was used to assess the association of each cortical and subcortical volume and thickness with schizophrenia. Then, all measures were inserted as dependent variables, and each SNP, schizophrenia diagnosis and the interaction between them were included as fixed factors. The total intracranial volume (TIV), gender and age were included as covariates. Statistical significance was set a priori at a two-tailed alpha of 0.05.
For the haplotype analyses, only polymorphisms in Hardy-Weinberg equilibrium and with a minor allele frequency .5% were included (rs16983466, rs2238731, rs2904552, rs2870983 and rs372055). The results of the linkage disequilibrium test are described in Figure S1. One haplotype block composed of three variants, and five composed of two variants, were analyzed (Table  S2). Concerning the haplotype composed by three variants, a significant association was detected between the A/G/C haplotype (rs2904552/rs2238731/rs16983466) and schizophrenia (p = 0.001). Analyzing the combinations of two variants, only those haplotypes constructed with rs2904552 SNP remained significant (p,0.05- Table S2).

Association between PRODH Variants and Schizophrenia Phenotypes
We analyzed the association of age at onset, TR, GAF, CGI, Calgary, Negative, Positive, Disorganized, Excited and Anxiety/ Depression clusters with each PRODH polymorphism, but no significant associations were detected.

In silico Analysis
To predict the likely pathogenicity of rs2904552, we conducted an in silico analysis. This SNP seems to have a functional role, as it was classified as ''probably damaging,'' with a score of 0.996, using PolyPhen and as ''damaging,'' with a score of 0.04, using SIFT. Data were also available for five orthologous species (three mammals, one frog and one fish), and the residues showed evolutionary conservation in all these organisms. Furthermore, this variation was reported in a previous study to decrease the enzyme activity [41].

Association between MRI and Schizophrenia
For the cortical and subcortical volumes, 21 regions were associated with schizophrenia after Bonferroni correction and are described in Table S3. In terms of cortical thickness, seven regions differed between the cases and controls (Table S4).

Association between PRODH Variants and MRI
In the comparison of all PRODH genotypes with all brain volumes and cortical thicknesses, only one association remained significant after Bonferroni correction: the interaction between rs2238731 (V427M) and schizophrenia was associated with right pars opercularis volume (p = 0.00020; effect size = 0.081; power = 0.965). Controls carrying the GA genotype had an increased volume (6261.506144.95 mm 3 ) in this region compared to carriers of the GG genotype (4193.376628.09 mm 3 ). However, in schizophrenia patients, GA genotype carriers had a smaller volume (3796.686527.12 mm 3 ) than did GG carriers (3993.416733.26 mm 3 ).

Discussion
In this study, one PRODH variant (rs2904552 or R431H) and its haplotypes were associated with schizophrenia, with the GG genotype present at a higher frequency among patients than among healthy controls. To further understand the role of this SNP, we investigated its role in brain volumes and cortical thicknesses.
Different lines of evidence suggest an important role for PRODH in schizophrenia: 1) it is located at 22q11.2, the deletion of which is one of the strongest genetic risk factors for schizophrenia; 2) PRODH deletion missense mutations have been described in patients with schizophrenia [41,42]; 3) an association between schizoaffective disorders and moderate hyperprolinemia was previously been detected [43]; 4) it is widely expressed in the brain; 5) mice with PRODH deficiency show elevated plasma and brain proline levels, locally decreased glutamate and c-aminobutyric acid (GABA) levels and a deficit in sensorimotor gating [17,44]; 6) the PRODH enzyme is rate limiting in the conversion of proline to glutamate [41], which is important for signal transduction between neurons and for synaptic plasticity [45] and seems to be deregulated in schizophrenia [46].
Indeed, a recent metabolomics study reported abnormalities related to proline metabolism in schizophrenia [47]. Furthermore, Savio et al. (2012) reported that long-term exposure to proline induced behavioral changes in adult zebrafish that were reversed by the acute administration of atypical antipsychotics [48]. In this vein, several studies have investigated peripheral proline levels as a risk factor for schizophrenia (reporting no association) [49,50] or the positive associations between hyperprolinemia and schizoaffective disorder [43] and schizophrenia in women [51]. Despite these mixed findings, the data still support a functional role for PRODH variants and hyperprolinemia in the pathophysiology of schizophrenia [31].
Along with its role in proline catabolism, PRODH is also known as a key enzyme in controlling homeostasis, providing energy and shuttling redox potential between cellular compartments and the production of reactive oxygen species [52].
PRODH rs2904552 is a nonsynonymous mutation that causes the substitution of an arginine for a histidine in codon 431. This change seems to promote a functional effect, according to our in silico findings, and also results in a moderate decrease (30-70%) of PRODH enzyme activity [41]. Also, another study has detected this polymorphism in combination with L441P in a subject with high levels of proline [42]. Two other studies have investigated this mutation; however, neither of them reported a positive association with schizophrenia [43,53]. As we detected an association between the major G-allele and schizophrenia, we suggest that the minor A-allele might be a protective factor and may thus prevent the onset of the disease by altering the enzyme activity in brain.

MRI
Our results support earlier findings of reduced total, left and right hemisphere gray matter volumes, corpora callosa, frontal, parietal and temporal lobes, and hippocampi, all of which are brain regions involved in the pathophysiology of schizophrenia [29]. We also found a significant association between the interaction of rs2238731 and schizophrenia diagnosis with the right pars opercularis. Controls carrying the GA genotype had an increased volume in this region compared to GG genotype carriers; however, patients carrying the GA genotype had a smaller volume than did GG patients. This result should be further verified in a larger sample due to the small sample of the GA genotype carriers in the both the control (N = 2) and patient (N = 19) groups.

Limitations
Some limitations of this study should be noted. First, we should have analyzed plasma proline levels. Considering that rs2904552 variant may be affecting PRODH enzyme, its substrate might be dysregulated. For this measurement, gender, acute alcohol use and medication, such as valproic acid, may be taken into consideration [43,54,55]. Furthermore, the duration of illness and medications may have been confounding factors for the neuroimaging analysis.

Conclusions
This study represents the first report of the effects of PRODH polymorphisms on brain morphology in both patients and healthy controls, testing for the interaction between this SNP and a diagnosis of schizophrenia. Only one SNP (rs2238731) seemed to be significantly associated with right pas opercularis volume, but this finding should be confirmed in larger samples. Therefore, considering the functional effect of rs2904552 and the importance of proline levels in schizophrenia, this SNP may act by altering neurochemistry and is not responsible for the deficits in brain structure found in schizophrenia.
In summary, this study indicates that the PRODH rs2904552 polymorphism may be a genetic risk factor for developing schizophrenia, and further studies should address how disturbance of the proline pathway could affect brain function.