Access the full text.
Sign up today, get DeepDyve free for 14 days.
Background: Genome-wide significant associations of schizophrenia with eight SNPs in the CNNM2, MIR137, PCGEM1, TRIM26, CSMD1, MMP16, NT5C2 and CCDC68 genes have been identified in a recent mega-analysis of genome-wide association studies. To date, the role of these SNPs on gray matter (GM) volumes remains unclear. Methods: After performing quality control for minor-allele frequency > 5% using a JPT HapMap sample and our sample, a genotyping call rate > 95% and Hardy-Weinberg equilibrium testing (p > 0.01), five of eight SNPs were eligible for analysis. We used a comprehensive voxel-based morphometry (VBM) technique to investigate the effects of these five SNPs on GM volumes between major-allele homozygotes and minor-allele carriers in Japanese patients with schizophrenia (n = 173) and healthy subjects (n = 449). Results: The rs7914558 risk variant at CNNM2 was associated with voxel-based GM volumes in the bilateral inferior frontal gyri (right T = 4.96, p = 0.0088, left T = 4.66, p = 0.031). These peak voxels, which were affected by the variant, existed in the orbital region of the inferior frontal gyri. Individuals with the risk G/G genotype of rs7914558 had smaller GM volumes in the bilateral inferior frontal gyri than carriers of the non-risk A-allele. Although several effects of the genotype and the genotype-diagnosis interaction of other SNPs on GM volumes were observed in the exploratory VBM analyses, these effects did not remain after the FWE-correction for multiple tests (p > 0.05). Conclusions: Our findings suggest that the genetic variant in the CNNM2 gene could be implicated in the pathogenesis of schizophrenia through the GM volumetric vulnerability of the orbital regions in the inferior frontal gyri. Keywords: Schizophrenia, Genome-wide association study, Voxel-based morphometry, Cyclin M2 (CNNM2), Inferior frontal gyrus Background accesses tens of thousands of DNA samples from patients Schizophrenia is a common and complex psychiatric dis- and controls can be a powerful tool for identifying common order with a lifetime risk of approximately 1%. This risk factors for complex diseases, such as schizophrenia. To disorder has a strong genetic component; indeed, the esti- date, GWASs on schizophrenia have identified several mated heritability is 81% . Multiple genetic variants that genome-wide significant associated variants located in the have a small effect have been implicated in the pathogen- zinc finger protein 804A (ZNF804A), neurogranin (NRGN), esis of schizophrenia . A genome-wide association study transcription factor 4 (TCF4) genes and a major histocom- (GWAS) of single-nucleotide polymorphisms (SNPs) that patibility complex (MHC) region [3,4]. Subsequently, the influences of these SNPs in the genes on brain function and * Correspondence: firstname.lastname@example.org structure have been reported . We have found that the Department of Psychiatry, Osaka University Graduate School of Medicine, genome-wide supported variant of the NRGN gene is asso- Osaka, Japan Core Research for Evolutionary Science and Technology of the Japan ciated with the brain morphology of the anterior cingulate Science and Technology Agency, Saitama, Japan cortex in patients with schizophrenia . Full list of author information is available at the end of the article © 2013 Ohi et al.; licensee BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Ohi et al. Behavioral and Brain Functions 2013, 9:40 Page 2 of 9 http://www.behavioralandbrainfunctions.com/content/9/1/40 Recently, a combining analysis of a mega-analysis of (VBM) technique, we tested the hypothesis that these risk GWAS data from 17 separate studies (9,394 cases and variants would be associated with GM volumes. 12,462 controls) and the replication data (8,442 cases and 21,397 controls) of European ancestry have found genome- Material and methods wide significant associations of schizophrenia with eight Subjects SNPs [rs7914558 (cyclin M2; CNNM2), rs1625579 (micro- VBM analyses were conducted on 173 patients with RNA 137, MIR137), rs17662626 (PCGEM1, prostate-specific schizophrenia (59.0% males, 102 males and 71 females, transcript, PCGEM1), rs2021722 (tripartite motif containing mean age ± SD, range 36.0 ± 12.3 years) and 449 healthy 26, TRIM26), rs10503253 (CUB and Sushi multiple do- subjects (47.7% males, 214 males and 235 females, mean mains 1, CSMD1), rs7004633 (matrix metallopeptidase age ± SD, range 35.4 ± 12.8 years). All subjects were bio- 16, MMP16), rs11191580 (5′-nucleotidase, cytosolic II, logically unrelated within the second-degree of relationship NT5C2) and rs12966547 (coiled-coil domain containing and were of Japanese descent [25,26]. The subjects were 68, CCDC68)] from the five new (1p21.3, 2q32.3, 8p23.2, excluded if they had neurological or medical conditions 8q21.3 and 10q24.32-q24.33) and two previously reported that could potentially affect the central nervous system, (6p21.32-p22.1 and 18q21.2) loci . In studies exploring such as atypical headache, head trauma with loss of con- brain activation during a sentence completion task, indi- sciousness, chronic lung disease, kidney disease, chronic viduals without the risk allele of the MIR137 genotype had hepatic disease, thyroid disease, active cancer, cerebrovas- significantly greater activation in the posterior right medial cular disease, epilepsy, seizures, substance-related disorders frontal gyrus than at-risk individuals . To our know- or mental retardation. Patients were recruited from the ledge, however, no study has investigated the effects of Osaka University Hospital. Each patient with schizophrenia these SNPs on gray matter (GM) volumes. had been diagnosed by at least two trained psychiatrists ac- Many attempts have been made to minimize the clinical cording to the criteria from the Diagnostic and Statistical and genetic heterogeneity in studies of schizophrenia. One Manual of Mental Disorders, Fourth Edition (DSM-IV) strategy for gene discovery uses neurobiological quantita- based on the Structured Clinical Interview for DSM-IV tive traits (QT) as intermediate phenotypes that reflect the (SCID). Controls were recruited through local advertise- underlying genetic vulnerability better than diagnostic ments at Osaka University. Psychiatrically, medically and categorization, such as schizophrenia [9,10]. This strategy neurologically, the healthy subjects were evaluated using has the potential to reduce clinical and genetic heterogen- the non-patient version of the SCID to exclude individuals eity . Structural GM volumes indicate substantial who had current or past contact with psychiatric services heritability rates ranging from moderate (40–70%) to high or who had received psychiatric medication. Current symp- (70–95%) in the frontal and temporal brain regions toms of schizophrenia were evaluated using the positive [12,13]. Mata analyses of brain morphological studies in and negative syndrome scale (PANSS) . Mean age and individuals with first-episode schizophrenia and neuro- handedness did not differ significantly between the study leptic naive schizophrenia as well as chronic patients group and the controls (p > 0.38), while the female ratio, with schizophrenia revealed reduction of GM volume years of education, estimated premorbid intelligence quo- in frontal, striato-limbic and temporal regions were tient (IQ) and total gray matter volumes were significantly present in the early stage of schizophrenia and were un- lower in the patients with schizophrenia compared to the related to the effects of neuroleptic treatment, chron- controls (p < 0.016) (Additional file 1: Table S1). All partici- icity and duration of illness [14-16]. Some studies have pants provided written informed consent after the study shown that abnormalities in GM volumes are inter- procedures had been fully explained. This study was per- mediate phenotypes that bridge the gap between the formed in accordance with the World Medical Associa- genotype and diagnostic categorization [11,17]. Charac- tion’s Declaration of Helsinki and was approved by the terizing the functional effects of novel and poorly Research Ethical Committee of Osaka University. understood genetic variants on the intermediate phe- notypes provides important insights into the neural SNP selection and SNP genotyping mechanisms by which the variants increase the risk We selected eight SNPs, rs7914558, rs1625579, rs17662626, for schizophrenia . Our research group has had a rs2021722, rs10503253, rs7004633, rs11191580 and long-standing interest in the effects of genetic vari- rs12966547, from a previous mega analysis of GWASs ants (i.e., COMT, DISC1, PACAP, BDNF, APOE, AKT1 . Of these eight SNPs, rs17662626 in the PCGEM1 gene and NRGN) on brain structure in psychiatric disor- (2q32.3) was not polymorphic in the samples obtained ders [6,19-24]. In this study, we examined the impact from the HapMap Japanese in Tokyo (JPT) project. Be- of the genome-wide supported variants on the GM cause the other seven SNPs were common genetic variants volumes of patients with schizophrenia and healthy sub- in the HapMap JPT samples (minor allele frequency > 5%), jects. Using a comprehensive voxel-based morphometry we focused on these SNPs. Venous blood was collected Ohi et al. Behavioral and Brain Functions 2013, 9:40 Page 3 of 9 http://www.behavioralandbrainfunctions.com/content/9/1/40 from the subjects, and genomic DNA was extracted from GM volumes between minor allele carriers and major allele whole blood according to standard procedures. These homozygotes. SNPs were genotyped using the TaqMan 5′-exonucle- ase allelic discrimination assay (Assay ID: rs7914558: Magnetic resonance imaging procedure C__31978821_10, rs1625579: C___8946584_20, rs2021722: All magnetic resonance imaging (MRI) studies were C__11690541_10, rs10503253: C___1503810_20, rs7004633: performed on a 1.5 T GE Signa EXCITE system. A three- C__29048976_10, rs11191580: C__31656012_10 and dimensional volumetric acquisition of a T1-weighted gradi- rs12966547: C____152930_10, Applied Biosystems, Foster ent echo sequence produced a gapless series of 124 sagittal City, CA, USA), as previously described [19,22]. Detailed in- sections using a spoiled gradient recalled acquisition in the formation on the PCR conditions is available upon request. steady state (SPGR) sequence (TE/TR, 4.2/12.6 ms; flip The genotyping call rates were 98.2% (rs7914558), 97.9% angle, 15°; acquisition matrix, 256 × 256; 1NEX, FOV, 24 × (rs1625579), 83.6% (rs2021722), 97.9% (rs10503253), 97.9% 24 cm; slice thickness, 1.4 mm). Subjects with MRI abnor- (rs7004633), 99.8% (rs11191580) and 97.6% (rs12966547). malities, such as infarcts, hemorrhages or brain tumors, No deviation from the Hardy-Weinberg equilibrium (HWE) were screened out prior to including this study as part of in the examined SNPs was detected in the patients or in con- routine clinical diagnosis and treatment. Therefore, there trols (p > 0.01), with the exception of rs2021722. A significant were no gross abnormalities in any of the subjects. Each deviation from HWE in the rs2021722 was found in both image was visually examined to eliminate images with mo- -17 -36 the patients (p=1.02×10 ) and controls (p= 2.05 × 10 ) tion or metal artifacts, and the anterior commissure- with a relative excess of CC homozygotes, TT homozygotes posterior commissure line was adjusted. The MRI images and undetermined subjects. According to the dbSNP data- were processed using the VBM8 toolbox in Statistical Para- base (National Center for Biotechnology Information), the metric Mapping 8 (SPM8; http://www.fil.ion.ucl.ac.uk/spm/ SNP is shown as a tri-allelic variant with T/C/A. A number software/spm8/) running on MATLAB R2013a (Math- of genome-wide significant variants within MHC (6p21.32- Works, Natick, MA, USA) according to the VBM8-Toolbox p22.1), including rs2021722, have been identified . How- Manual (http://dbm.neuro.uni-jena.de/vbm8/VBM8-Manual. ever, the MHC region has been excluded from further ana- pdf). The T1 images were normalized and segmented into lysis. Analyzing the region is difficult because of its high GM, white matter (WM) and cerebrospinal fluid (CSF) using linkage disequilibrium (LD) and ethnic heterogeneity. Minor the VBM8 toolbox with defaults for the extended options. allele frequencies of rs1625579 were under 5% in our patients The modulated non-linear only (i.e., with no affine com- (3.2%) and controls (2.5%). Therefore, in this study, we ex- ponent) option was selected to create volumetric GM par- cluded these SNPs rs2021722 and rs1625579 from the VBM titions. Finally, the images were smoothed with an 8-mm analyses. Genotype and allele distributions for each SNP in- full-width, half-maximum isotropic Gaussian kernel. cluded in the VBM analyses between the patients with schizophrenia and the controls are shown in Table 1. All risk Statistical analyses alleles were defined based on the previous GWAS : In genetic association analysis, we performed power calcula- rs7914558 (major G-allele), rs10503253 (minor A-allele), tions using the Power Calculator for Two-Stage Association rs7004633 (minor G-allele), rs11191580 (major T-allele) and Studies (http://www.sph.umich.edu/csg/abecasis/CaTS/) . rs12966547 (minor G-allele). To increase the statistical The power estimate was based on an allele frequency of 0.53 power and decrease type I errors, homozygotes and the het- at rs7914558, a prevalence of 0.01, an alpha level of 0.05, and erozygotes for the minor allele groups were combined and assuming varying degrees of odds ratio using a multiplicative treated as minor-allele carriers. In this study, we contrasted model. In brain morphological analyses, we performed power Table 1 Genotype and allele distributions for each SNP between the patients with schizophrenia and healthy subjects Risk Genotype frequencies Risk allele Allelic OR allele +/+ +/−−/− +/+ +/−−/− frequencies p value (95% CI) SNP IDs Gene Chr +/− SCZ (n = 173) CON (n = 449) SCZ CON (χ2) rs10503253 CSMD1 8p23.2 A/C 0.06 0.46 0.49 0.09 0.46 0.45 0.29 0.32 0.22 (1.5) 1.19 (0.90-1.56) rs7004633 MMP16 8q21.3 G/A 0.03 0.37 0.60 0.06 0.36 0.57 0.21 0.25 0.24 (1.4) 1.20 (0.89-1.61) rs7914558 CNNM2 10q24.32 G/A 0.23 0.53 0.24 0.29 0.49 0.22 0.49 0.53 0.23 (1.4) 1.16 (0.91-1.49) rs11191580 NT5C2 10q24.33 T/C 0.49 0.42 0.09 0.50 0.42 0.07 0.70 0.72 0.55 (0.4) 1.09 (0.83-1.43) rs12966547 CCDC68 18q21 G/A 0.15 0.44 0.41 0.17 0.44 0.39 0.37 0.39 0.54 (0.4) 1.09 (0.84-1.41) Abbreviations: Chr Chromosome; SCZ patients with schizophrenia; CON healthy controls; +, risk allele; -, non-risk allele; OR odds ratio. For alleles, the first allele is the risk allele. All risk alleles are represented based on the previous GWAS . Ohi et al. Behavioral and Brain Functions 2013, 9:40 Page 4 of 9 http://www.behavioralandbrainfunctions.com/content/9/1/40 calculations using the G*Power Version 3.1.5 . The clusters in bilateral inferiror frontal gyri at the lenient uncor- power estimate was based on an alpha level of 0.05, a power rected threshold of p < 0.001 and cluster sizes > 100. The of 0.80 and assuming varying degrees of effect size using extraction of these relative GM volumes were performed after t tests. Standardized effect sizes were calculated using including confounding factors such as age, sex and education Cohen’s d method (http:www.uccs.edu/faculty/lbecker). years and modulated by total brain volumes in the VBM ana- Statistical analyses of the demographic variables were lyses. Anatomic localization was performed according to both performed using PASW Statistics 18.0 software (SPSS the MNI coordinates and Talairach coordinates, which were Japan Inc., Tokyo, Japan). Based on the assumption that obtained from M. Brett’s transformations (http://imaging. most of demographic variables, such as age and education mrc-cbu.cam.ac.uk/imaging/MniTalairach) and presented as years, were not fitted to a normality distribution with the Talairach coordinates. Kolmogorov-Smirnov test (p < 0.05), differences in clinical characteristics between patients and controls or between Results genotypes were analyzed using the non-parametric Mann– Our study size of 173 cases and 449 controls had suffi- Whitney U-test for continuous variables, such as age and cient power (>80%) to detect a genetic effect at odds years of education, and χ tests for categorical variables, ratio of 1.43 or greater for rs7914558 when the allele such as gender and handedness, as shown in Additional frequency was 0.53. Unfortunately, in our sample sizes, file 1: Table S1-S3. The presence of HWE was examined there was no allelic association with schizophrenia for using the χ test for goodness-of-fit via SNPAlyze V5.1.1 any of the five SNPs [rs7914558 (CNNM2), rs10503253 Pro software (DYNACOM, Yokohama, Japan). The allelic (CSMD1), rs7004633 (MMP16), rs11191580 (NT5C2) distributions of each SNP between patients and controls and rs12966547 (CCDC68)] (p > 0.22, Table 1). were analyzed using χ tests with the SNPAlyze software. We investigated the effects of diagnosis (cases and con- The significance levels for HWE and other statistical tests trols), genotype (major-allele homozygotes and minor- were set at two-tailed p-values, p <0.01 and p <0.05, allele carriers) and their interaction of five SNPs on GM respectively. volumes in a comprehensive exploratory VBM analysis. We performed a comprehensive exploratory whole brain The effects of diagnosis between patients with schizophre- search using the SPM8 statistical tools to examine the ef- nia and healthy subjects were found in all analyses of the fects of the diagnosis, the genotype and their interaction of present study (p < 0.05). Patients with schizophrenia each SNP on GM volume in patients with schizophrenia showed smaller GM volumes compared with healthy sub- and healthy subjects. As two-way ANOVA can simultan- jects primarily in the frontal and temporal lobes, including eously investigate these effects only in one model, these ef- the bilateral inferior frontal gyri, which was consistent with fects on GM volume were statistically assessed using a full previous studies [14,30]. We found significant effects for factorial model for a 2 × 2 ANOVA with diagnosis (cases the risk-allele homozygotes of rs7914558 on decreased and controls) and genotype status (major-allele homozy- GM volume in the bilateral inferior frontal gyri (right, T = gotes and minor-allele carriers) as independent variables in 4.96, p = 0.0088; left, T =4.66, p = 0.031), as shown in SPM8. Gray matter volumes are correlated with age, and Table 2 and the regions based on the hot color map in gender and years of education differed significantly be- Figure 1. To compare the effects of the genotype in both tween the patient and the control groups (Additional file 1: the patients with schizophrenia and healthy subjects, we Table S1). Therefore, age, gender and years of education extracted the means and SD for relative GM volumes from were included as covariates of no interest into the analyses nominal clusters in bilateral inferior frontal gyri and to control for confounding variables. We contrasted GM the extracted GM volumes were shown in Figure 2 and volumes between the diagnostic groups (smaller volume Additional file 1: Table S2. The risk G-allele homozygotes region in patients with schizophrenia compared with of the CNNM2 polymorphism had smaller GM volumes in healthy subjects), the genotype groups (smaller or larger the bilateral inferior frontal gyri compared to the non-risk volume region in minor-allele carriers relative to major- allele carriers. The inferior frontal gyrus can be subdivided allele homozygotes) or the genotype-diagnosis inter- into three macroanatomical structures: the orbital (Brod- action. Non-sphericity estimation was used. We applied mann area: BA47), opercular (BA44) and triangular (BA45) a voxel-level height threshold of p <0.001 (uncorrected for parts. In the present study, the peak GM regions affected multiple comparisons) and clusters of more than 100 con- by the CNNM2 genotype existed in the orbital parts of the tiguous voxels were considered for the exploratory VBM bilateral inferior frontal gyri. When the two genotype analyses. And then we applied family-wise error (FWE)cor- groups (G-allele homozygotes and A-allele carriers) were rection for multiple testing to avoid type I errors at the divided into three genotype groups (individuals with G/G whole brain level. Eventually, the significance level was set genotype, G/A genotype and A/A genotype) and we per- at p <0.05 (FWE- ). To obtain a cluster as large as formed an additional VBM analysis using a multiple regres- corrected possible, we extracted relative GM volumes from nominal sion model, the number of risk G-allele was significantly Ohi et al. Behavioral and Brain Functions 2013, 9:40 Page 5 of 9 http://www.behavioralandbrainfunctions.com/content/9/1/40 Table 2 Effects of the CNNM2 genotype and genotype-diagnosis interaction on GM volumes p values (peak) Talairach coordinates Brain regions R/L BA CS T FWE x y z Non-risk minor allele carrier > Risk major allele homozygote Inferior frontal Gyrus R 11/47 1306 4.96 0.0088 22 31 −20 Inferior frontal Gyrus L 47 437 4.66 0.031 −22 18 −22 Middle frontal Gyrus L 11 667 4.33 0.11 −24 38 −18 Posterior cingulate L 29 248 3.73 0.61 −7 −48 11 Non-risk minor allele carrier < Risk major allele homozygote no suprathreshold clusters Genotype-diagnosis interaction Superior temporal Gyrus L 22 598 4.23 0.16 −46 −20 1 Abbreviations: R right; L left; BA Brodmann area; CS Cluster size; FWE family-wise error. All regions shown have nominal association at a voxel-level height threshold of p < 0.001 and a minimum clusters extent of 100 voxels. Significant uncorrected results (FWE- p < 0.05) are shown in bold face. corrected related to smaller GM volumes of the right inferior frontal including chlorpromazine equivalent of total antipsychotics gyrus (T = 4.67, p = 0.029) but not the left inferior frontal (mg/day), duration of illness or PANSS scores as covariant gyrus (T =3.64, p = 0.70) in total subjects. for VBM analyses only in patients. The genotype effect on As shown in Additional file 1: Table S1, patients with these regions did not change even after controlling for schizophrenia participated in this study were chronic and these factors, suggesting that there was no potential clinical the symptoms were moderately stable. There was no dif- impact on our outcomes. ference in demographic information, such as duration of In the exploratory VBM analyses, we also found nominal illness or PANSS scores, between risk G-allele homozy- effects of the risk-allele homozygotes of rs7914558 on de- gotes and non-risk A-allele carriers of rs7914558 (p >0.20, creased GM volume in the left middle frontal gyrus and Additional file 1: Table S3). However, to find whether there the left posterior cingulate and a nominal genotype- was potential clinical impact on our outcomes, we add- diagnosis interaction with the GM volume in the left su- itionally investigated the genotype effect on GM volume perior temporal gyrus ( p < 0.001, Table 2). uncorrected Figure 1 Effects of the rs7914558 polymorphism at the CNNM2 gene on GM volumes. There were effects of the risk-allele carriers of rs7914558 at CNNM2 on decreased GM regions (red areas shown on the hot color map). There was no effect of the genotype on increased GM regions (blue area shown on the winter color map). Each color map shows the t values corresponding to the color in the figure. Ohi et al. Behavioral and Brain Functions 2013, 9:40 Page 6 of 9 http://www.behavioralandbrainfunctions.com/content/9/1/40 Figure 2 The impacts of the CNNM2 genotype on GM volume of the bilateral inferior frontal gyri. (A) Anatomical localizations are displayed on coronal, sagittal, and axial sections of a normal MRI spatially normalized into the Montreal Neurological Institute template (p <0.05). The most significant cluster of the genotype effect was in the right inferior frontal gyrus. The region is shown as a cross-hairline. The color bars show the t values corresponding to the color in the figure. (B, C) Each column shows the relative GM volumes extracted from a nominal cluster at p < 0.001 and cluster size > 100 in the right inferior frontal gyrus (peak Talairach coordinates; 22, 31, -20) (B) and in the left inferior uncorrected frontal gyrus (peak Talairach coordinates; -22, 18, -22) (C). Error bars represent the standard error. Additionally, we found several marginal effects of genotypes and a large sample size of at least 3400 patients and 3400 and genotype-diagnosis interactions of other SNPs on GM controls is needed. On the other hand, in brain morpho- volumesinthe exploratoryanalyses( p <0.001, logical analyses, our sample size had sufficient power uncorrected Additional file 1: Table S4 and Figure S1-S4). However, (>80%) to detect a genotype effect on GM volumes at these effects of genotypes and genotype-diagnosis interac- medium effect size (Cohen’s d) of 0.25 or greater. The ob- tions on GM regions did not survive after the FWE-correc- served effect sizes on the inferior frontal gyri were tion for multiple tests at the whole brain level (p >0.05). medium to large (0.29-1.00). These findings suggest that the genome-wide supported variant of schizophrenia had Discussion larger effect on GM volumes of the inferior frontal gyri To date, it remains unclear whether the genome-wide than diagnostic status, and support that GM volumes ab- significant risk variants for schizophrenia in a mega- normalities were prominent intermediate phenotypes analysis of GWASs influenced GM volumes. This study bridging the gap between a susceptibility genetic variant is the first to identify the GM morphology associated and diagnostic categorization. with genome-wide risk variants using a comprehensive Rs7914558 is located in intron1 of the CNNM2 gene VBM technique. Of the five genetic variants investigated (also known as ACDP2) on chromosome 10q24.32. The in this study, we found influences of the CNNM2 geno- CNNM2 gene spans 160.3 kb of genomic DNA and con- type on the bilateral inferior frontal gyri at the whole tains eight exons. This gene belongs to a member of the brain level. GM volumes in the bilateral inferior frontal ancient conserved domain-containing protein family be- gyri, particularly the orbital region, in the risk G-allele cause the protein shares a domain conserved in a large homozygotes of CNNM2 polymorphism were smaller number of species ranging from bacteria to human . than those observed in the non-risk allele carriers. Members of this protein family contain a sequence motif In genetic association analysis, our sample size had suffi- that is present in the cyclin box, a cyclic nucleotide- cient power (>80%) to detect a genetic effect with odds ra- monophosphate (cNMP)-binding domain. The CNNM2 tio of 1.43 or greater for rs7914558. However, previous gene has a ubiquitous expression pattern in humans . large GWAS has reported the genetic effect with low odds In particular, the level of CNNM2 expression in the ratio of 1.10 for the SNP . To detect such a small gen- brain is moderate to high (http://www.ebi.ac.uk/gxa/ etic effect, our sample size had insufficient power (12%), experiment/E-MTAB-37/ENSG00000148842). Ohi et al. Behavioral and Brain Functions 2013, 9:40 Page 7 of 9 http://www.behavioralandbrainfunctions.com/content/9/1/40 However, whether the expression level of this gene in the of the inferior frontal gyri. Further research is needed to in- brains of patients with schizophrenia is lower or higher vestigate how a possible relationship between the CNNM2 than that in healthy subjects is unknown. The encoded gene and hypofunction of the NMDA receptor would result protein CNNM2 plays an important role in magnesium in decreased GM volumes of the inferior frontal gyri. homeostasis by modulating Mg2+ concentration. The There was no significant effect of the other four variants CNNM2 mRNA is upregulated when there is a defi- on any GM volumes. There are several possible reasons ciency of magnesium in the brain . CNNM2 medi- for the absence of an association. A false negative associ- ated Mg2 + −sensitiveNa+currentswereblocked by ation cannot be excluded in our study because we applied increased extracellular Mg2+ concentrations . We astrict FWE correction for multiple comparisons at the assessed the effect of the rs7914558 genotype on whole brain level (p < 0.05). In the Additional file 1: Figure CNNM2 expression using bioinformatics data (http:// S1-S4, the regions shown at the more lenient uncorrected www.sanger.ac.uk/resources/software/genevar/) to threshold of p < 0.001 may be helpful in further studies. examine whether the rs7914558 genotype might be an Interestingly, many of the effects of these genome-wide expression quantitative trait loci (eQTL). In silico ana- significant variants at the lenient level involved decreased lysis showed that the CNNM2 gene expression of the GM volumes, including the medial, middle and inferior high-risk G genotype of rs7914558 was significantly fontal gyri. Reduced GM volumes in these regions have lower than that of the non-risk genotype in the com- been repeatedly demonstrated in imaging studies of bined lymphoblast-derived HapMap CEU and YRI sam- schizophrenia [14,30]. Another interpretation is that the ples (r = −.23, t = −2.59, p = 0.011). The low expression effect of these variants was not sufficiently sensitive to the of this gene resulted increased Mg2+ levels. Increased morphological vulnerability of the GM volumes. The effect extracellular Mg2+ concentrations caused a decrease in of these variants may be preferable in identifying genotype- theactivityofthe glutamate N-methyl-D-aspartate related vulnerability on other intermediate phenotypes, (NMDA) receptor . These findings suggest that the such as cognitive functions and personality traits. There- CNNM2 gene mayplayanimportant role in thehypo- fore, further research is needed to confirm whether the ef- function of the NMDA receptor, which is implicated in fects of these variants could be related to the susceptibility the pathophysiology of schizophrenia. of cognitive functioning. The inferior frontal gyrus has a multifunctional role in There were several limitations to this study. Because a human behavior, interpersonal interactions and communi- number of statistical analyses, including the effects of diag- cation . The inferior frontal gyrus consists primarily of nosis, genotypes and their interaction on GM volumes, the heteromodal association neocortex, which is a major were performed, a correction for multiple testing should site of involvement in schizophrenia . Several studies be considered. However, a consensus on how to correct have reported that the relative GM in patients with schizo- such multiple testing on study combining brain imaging phrenia was significantly reduced in the bilateral inferior and genetics has not been reached in this research field. frontal areas [38-40]. The inferior frontal gyrus can be To control type I errors, we applied the strict FWE correc- subdivided into three macroanatomical parts: orbital, tion for all VBM analyses, while we did not perform any opercular and triangular. The orbital region is one of the correction on the genetic modality. The existence of a false major regions of the social brain that connects to the orbi- positive association cannot be excluded as an explanation tofrontal cortex , while the opercular and triangular for our results, although we were quite careful to match regions form Broca’s area, which is an important region ethnicity and correct for multiple testing. Further investi- for speech-language production . We found that the gations of other samples with much larger sample sizes CNNM2 genotype affects brain volumes in the orbital re- and/or with different ethnicities and/or in relatives of gions of the inferior frontal gyri. Functionally, the orbital those with schizophrenia are needed to confirm our region is thought to be involved in the processing of em- findings. It is unclear whether our results are directly/ pathy  and sentence comprehension [43,44] in the left indirectly linked to the CNNM2 polymorphism rs7914558, hemisphere and decision-making cognition  and fine to other polymorphisms in high LD with this variant or to movement control  in the right hemisphere. Social interactions between this variant and other variants. To functions are widely impaired in patients with schizophre- determine whether rs7914558 is the most strongly asso- nia [45-49]. It is still unclear whether and to what extent ciated variant for schizophrenia and brain structure in the effects of CNNM2 polymorphism on GM structure this gene, an extensive search for other functional vari- observed here might be associated with an increased risk ants at this locus is needed. Additionally, as with other for schizophrenia. We suggest that the CNNM2 variant risk variants for schizophrenia, clarifying the biological may play a role in the social cognition and social function- role of this SNP through in vitro and in vivo studies is ing impairments noted in patients with schizophrenia important to improve the understanding of the patho- through GM volumetric vulnerability of the orbital regions physiology of schizophrenia. Ohi et al. Behavioral and Brain Functions 2013, 9:40 Page 8 of 9 http://www.behavioralandbrainfunctions.com/content/9/1/40 Conclusions Received: 23 July 2013 Accepted: 21 October 2013 Published: 25 October 2013 We found that a genome-wide supported variant of CNNM2 could be associated with GM morphological vul- nerability of the bilateral inferior frontal gyri. These results References 1. Sullivan PF, Kendler KS, Neale MC: Schizophrenia as a complex trait: suggestthatthere maybepossible deleterious effects of the evidence from a meta-analysis of twin studies. Arch Gen Psychiatry 2003, risk G-allele at CNNM2 in the inferior frontal gyri, which 60:1187–1192. may, at least partially, represent the mechanism by which 2. Sun J, Kuo PH, Riley BP, Kendler KS, Zhao Z: Candidate genes for schizophrenia: a survey of association studies and gene ranking. Am J CNNM2 increases the risk for schizophrenia. Further re- Med Genet B Neuropsychiatr Genet 2008, 147B:1173–1181. search will be needed to clarify the function of the risk 3. Stefansson H, Ophoff RA, Steinberg S, Andreassen OA, Cichon S, Rujescu D, CNNM2 variant on the pathophysiology of schizophrenia. Werge T, Pietilainen OP, Mors O, Mortensen PB, Sigurdsson E, Gustafsson O, Nyegaard M, Tuulio-Henriksson A, Ingason A, Hansen T, Suvisaari J, Lonnqvist J, Paunio T, Borglum AD, Hartmann A, Fink-Jensen A, Nordentoft Additional file M, Hougaard D, Norgaard-Pedersen B, Bottcher Y, Olesen J, Breuer R, Moller HJ, Giegling I, et al: Common variants conferring risk of schizophrenia. Additional file 1: Table S1. Demographic information for patients with Nature 2009, 460:744–747. schizophrenia and healthy subjects. Table S2: Effects of the rs7914558 4. O’Donovan MC, Craddock N, Norton N, Williams H, Peirce T, Moskvina V, genotype on extracted relative GM volumes of the bilateral inferior frontal gyri. Nikolov I, Hamshere M, Carroll L, Georgieva L, Dwyer S, Holmans P, Marchini Table S3: Demographic information for risk G-allele homozygotes and non- JL, Spencer CC, Howie B, Leung HT, Hartmann AM, Moller HJ, Morris DW, risk A-allele carriers of rs7914558. Table S4: Impacts of each genetic variant on Shi Y, Feng G, Hoffmann P, Propping P, Vasilescu C, Maier W, Rietschel M, GM volumes in the exploratory VBM analyses. Figure S1: Effects of the Zammit S, Schumacher J, Quinn EM, Schulze TG, et al: Identification of loci rs10503253 polymorphism (CSMD1)on GMvolumes. Figure S2: Effects of the associated with schizophrenia by genome-wide association and follow- rs7004633 polymorphism (MMP16)on GMvolumes. Figure S3: Effects of the up. Nat Genet 2008, 40:1053–1055. rs11191580 polymorphism (NT5C2)on GM volumes. Figure S4: Effects of the 5. Rose EJ, Donohoe G: Brain vs behavior: an effect size comparison of rs12966547 polymorphism (CCDC68)on GMvolumes. neuroimaging and cognitive studies of genetic risk for schizophrenia. Schizophr Bull 2013, 39:518–526. 6. Ohi K, Hashimoto R, Yasuda Y, Nemoto K, Ohnishi T, Fukumoto M, Abbreviations Yamamori H, Umeda-Yano S, Okada T, Iwase M, Kazui H, Takeda M: Impact GM: Gray matter; VBM: Voxel-based morphometry; GWAS: Genome-wide of the genome wide supported NRGN gene on anterior cingulate association study; SNP: Single-nucleotide polymorphism; CNNM2: Cyclin M2; morphology in schizophrenia. PLoS One 2012, 7:e29780. QT: Quantitative traits; DSM-IV: Diagnostic and statistical manual of mental 7. Ripke S, Sanders AR, Kendler KS, Levinson DF, Sklar P, Holmans PA, Lin DY, disorders fourth edition; SCID: Structured clinical interview for DSM-IV; Duan J, Ophoff RA, Andreassen OA, Scolnick E, Cichon S, St Clair D, Corvin HWE: Hardy-Weinberg equilibrium; LD: Linkage disequilibrium; MRI: Magnetic A, Gurling H, Werge T, Rujescu D, Blackwood DH, Pato CN, Malhotra AK, resonance imaging; FWE: Family-wise error. Purcell S, Dudbridge F, Neale BM, Rossin L, Visscher PM, Posthuma D, Ruderfer DM, Fanous A, Stefansson H, Steinberg S, et al: Genome-wide Competing interests association study identifies five new schizophrenia loci. Nat Genet 2011, The authors declare that they have no competing interests. 43:969–976. 8. Whalley HC, Papmeyer M, Romaniuk L, Sprooten E, Johnstone EC, Hall J, Authors’ contributions Lawrie SM, Evans KL, Blumberg HP, Sussmann JE, McIntosh AM: Impact of a RH supervised the entire project, collected the data, wrote the manuscript, was microRNA MIR137 susceptibility variant on brain function in people at critically involved in the design, analysis and interpretation of the data and was high genetic risk of schizophrenia or bipolar disorder. responsible for performing the literature review. KO was critically involved in the Neuropsychopharmacology 2012, 37:2720–2729. collection and analysis of the data, and contributed to the editing of the final 9. Meyer-Lindenberg A, Weinberger DR: Intermediate phenotypes and manuscript and contributed intellectually to the interpretation of the data. HY genetic mechanisms of psychiatric disorders. Nat Rev Neurosci 2006, and SU took part in genotyping. HY, YY, MF, MF, YW, MI, HK and MT contribute 7:818–827. with sample collection and gave comments to the manuscript. All authors 10. Tan HY, Callicott JH, Weinberger DR: Intermediate phenotypes in contributed to and have approved the final manuscript. schizophrenia genetics redux: is it a no brainer? Mol Psychiatry 2008, 13:233–238. 11. Potkin SG, Turner JA, Guffanti G, Lakatos A, Torri F, Keator DB, Macciardi F: Acknowledgments Genome-wide strategies for discovering genetic influences on cognition We would like to thank all of the individuals who participated in this study. and cognitive disorders: methodological considerations. Cogn This work was supported by research grants from the Japanese Ministry of Neuropsychiatry 2009, 14:391–418. Health, Labor and Welfare (H22-seishin-ippan-001); KAKENHI, 22390225-Grant- 12. Kaymaz N, van Os J: Heritability of structural brain traits an in-Aid for Scientific Research (B), 23659565-Grant-in-Aid for Challenging endophenotype approach to deconstruct schizophrenia. Int Rev Neurobiol Exploratory Research and Grant-in-Aid for Scientific Research on Innovative 2009, 89:85–130. Areas (Comprehensive Brain Science Network) from the Japanese Ministry of 13. Rijsdijsk FV, Viding E, De Brito S, Forgiarini M, Mechelli A, Jones AP, McCrory E: Education, Culture, Sports, Science and Technology (MEXT); CREST of JST; Heritable variations in gray matter concentration as a potential and the Japan Foundation for Neuroscience and Mental Health. endophenotype for psychopathic traits. Arch Gen Psychiatry 2010, 67:406–413. 14. Chan RC, Di X, McAlonan GM, Gong QY: Brain anatomical abnormalities in Author details high-risk individuals, first-episode, and chronic schizophrenia: an Department of Psychiatry, Osaka University Graduate School of Medicine, activation likelihood estimation meta-analysis of illness progression. Osaka, Japan. National Hospital Organization, Yamato Mental-Medical Schizophr Bull 2011, 37:177–188. Center, Nara, Japan. Core Research for Evolutionary Science and Technology 15. Leung M, Cheung C, Yu K, Yip B, Sham P, Li Q, Chua S, McAlonan G: Gray of the Japan Science and Technology Agency, Saitama, Japan. Molecular matter in first-episode schizophrenia before and after antipsychotic drug Research Center for Children’s Mental Development, United Graduate School treatment. Anatomical likelihood estimation meta-analyses with sample of Child Development, Osaka University, Osaka, Japan. Department of size weighting. Schizophr Bull 2011, 37:199–211. Molecular Neuropsychiatry, Osaka University Graduate School of Medicine, Osaka, Japan. Biofunctional Imaging, Immunology Frontier Research Center, 16. Ren W, Lui S, Deng W, Li F, Li M, Huang X, Wang Y, Li T, Sweeney JA, Gong Osaka University, Osaka, Japan. Department of Radiology, Osaka University Q: Anatomical and functional brain abnormalities in drug-naive first- Graduate School of Medicine, Osaka, Japan. episode schizophrenia. Am J Psychiatry. in press. Ohi et al. Behavioral and Brain Functions 2013, 9:40 Page 9 of 9 http://www.behavioralandbrainfunctions.com/content/9/1/40 17. Goldman AL, Pezawas L, Mattay VS, Fischl B, Verchinski BA, Chen Q, Dermitzakis ET: Population genomics of human gene expression. Weinberger DR, Meyer-Lindenberg A: Widespread reductions of cortical Nat Genet 2007, 39:1217–1224. thickness in schizophrenia and spectrum disorders and evidence of 35. Kochlamazashvili G, Bukalo O, Senkov O, Salmen B, Gerardy-Schahn R, Engel heritability. Arch Gen Psychiatry 2009, 66:467–477. AK, Schachner M, Dityatev A: Restoration of synaptic plasticity and 18. Meyer-Lindenberg A: Imaging genetics of schizophrenia. Dialogues Clin learning in young and aged NCAM-deficient mice by enhancing Neurosci 2010, 12:449–456. neurotransmission mediated by GluN2A-containing NMDA receptors. 19. Hashimoto R, Hashimoto H, Shintani N, Chiba S, Hattori S, Okada T, J Neurosci 2012, 32:2263–2275. Nakajima M, Tanaka K, Kawagishi N, Nemoto K, Mori T, Ohnishi T, Noguchi 36. Liakakis G, Nickel J, Seitz RJ: Diversity of the inferior frontal gyrus–a meta- H, Hori H, Suzuki T, Iwata N, Ozaki N, Nakabayashi T, Saitoh O, Kosuga A, analysis of neuroimaging studies. Behav Brain Res 2011, 225:341–347. Tatsumi M, Kamijima K, Weinberger DR, Kunugi H, Baba A: Pituitary 37. Ross CA, Pearlson GD: Schizophrenia, the heteromodal association adenylate cyclase-activating polypeptide is associated with neocortex and development: potential for a neurogenetic approach. schizophrenia. Mol Psychiatry 2007, 12:1026–1032. Trends Neurosci 1996, 19:171–176. 20. Hashimoto R, Hirata Y, Asada T, Yamashita F, Nemoto K, Mori T, Moriguchi Y, 38. Buchanan RW, Vladar K, Barta PE, Pearlson GD: Structural evaluation of the Kunugi H, Arima K, Ohnishi T: Effect of the brain-derived neurotrophic prefrontal cortex in schizophrenia. Am J Psychiatry 1998, 155:1049–1055. factor and the apolipoprotein E polymorphisms on disease progression 39. Yamasue H, Iwanami A, Hirayasu Y, Yamada H, Abe O, Kuroki N, Fukuda R, in preclinical Alzheimer’s disease. Genes Brain Behav 2009, 8:43–52. Tsujii K, Aoki S, Ohtomo K, Kato N, Kasai K: Localized volume reduction in prefrontal, temporolimbic, and paralimbic regions in schizophrenia: an 21. Hashimoto R, Moriguchi Y, Yamashita F, Mori T, Nemoto K, Okada T, Hori H, MRI parcellation study. Psychiatry Res 2004, 131:195–207. Noguchi H, Kunugi H, Ohnishi T: Dose-dependent effect of the Val66Met 40. Suzuki M, Nohara S, Hagino H, Kurokawa K, Yotsutsuji T, Kawasaki Y, polymorphism of the brain-derived neurotrophic factor gene on Takahashi T, Matsui M, Watanabe N, Seto H, Kurachi M: Regional changes memory-related hippocampal activity. Neurosci Res 2008, 61:360–367. in brain gray and white matter in patients with schizophrenia 22. Hashimoto R, Numakawa T, Ohnishi T, Kumamaru E, Yagasaki Y, Ishimoto T, demonstrated with voxel-based analysis of MRI. Schizophr Res 2002, Mori T, Nemoto K, Adachi N, Izumi A, Chiba S, Noguchi H, Suzuki T, Iwata N, 55:41–54. Ozaki N, Taguchi T, Kamiya A, Kosuga A, Tatsumi M, Kamijima K, Weinberger 41. Rogers RD, Owen AM, Middleton HC, Williams EJ, Pickard JD, Sahakian BJ, DR, Sawa A, Kunugi H: Impact of the DISC1 Ser704Cys polymorphism on Robbins TW: Choosing between small, likely rewards and large, unlikely risk for major depression, brain morphology and ERK signaling. Hum Mol rewards activates inferior and orbital prefrontal cortex. J Neurosci 1999, Genet 2006, 15:3024–3033. 19:9029–9038. 23. Ohi K, Hashimoto R, Yasuda Y, Fukumoto M, Nemoto K, Ohnishi T, 42. Amunts K, Schleicher A, Burgel U, Mohlberg H, Uylings HB, Zilles K: Broca’s Yamamori H, Takahashi H, Iike N, Kamino K, Yoshida T, Azechi M, Ikezawa K, region revisited: cytoarchitecture and intersubject variability. J Comp Tanimukai H, Tagami S, Morihara T, Okochi M, Tanaka T, Kudo T, Iwase M, Neurol 1999, 412:319–341. Kazui H, Takeda M: The AKT1 gene is associated with attention and brain 43. Sakai KL, Nauchi A, Tatsuno Y, Hirano K, Muraishi Y, Kimura M, Bostwick M, morphology in schizophrenia. World J Biol Psychiatry 2013, 14:100–113. Yusa N: Distinct roles of left inferior frontal regions that explain 24. Ohnishi T, Hashimoto R, Mori T, Nemoto K, Moriguchi Y, Iida H, Noguchi H, individual differences in second language acquisition. Hum Brain Mapp Nakabayashi T, Hori H, Ohmori M, Tsukue R, Anami K, Hirabayashi N, Harada 2009, 30:2440–2452. S, Arima K, Saitoh O, Kunugi H: The association between the Val158Met 44. Sakai KL: Language acquisition and brain development. Science 2005, polymorphism of the catechol-O-methyl transferase gene and 310:815–819. morphological abnormalities of the brain in chronic schizophrenia. 45. Bragado-Jimenez MD, Taylor PJ: Empathy, schizophrenia and violence: a Brain 2006, 129:399–410. systematic review. Schizophr Res 2012, 141:83–90. 25. Hashimoto R, Ohi K, Yasuda Y, Fukumoto M, Iwase M, Iike N, Azechi M, 46. Sevy S, Burdick KE, Visweswaraiah H, Abdelmessih S, Lukin M, Yechiam E, Ikezawa K, Takaya M, Takahashi H, Yamamori H, Okochi T, Tanimukai H, Bechara A: Iowa gambling task in schizophrenia: a review and new data Tagami S, Morihara T, Okochi M, Tanaka T, Kudo T, Kazui H, Iwata N, Takeda in patients with schizophrenia and co-occurring cannabis use disorders. M: The impact of a genome-wide supported psychosis variant in the Schizophr Res 2007, 92:74–84. ZNF804A gene on memory function in schizophrenia. Am J Med Genet B 47. Billeke P, Aboitiz F: Social cognition in schizophrenia: from social stimuli Neuropsychiatr Genet 2010, 153B:1459–1464. processing to social engagement. Front Psychiatry 2013, 4:4. 26. Ohi K, Hashimoto R, Yasuda Y, Kiribayashi M, Iike N, Yoshida T, Azechi M, 48. Kuperberg GR, Kreher DA, Ditman T: What can event-related potentials tell Ikezawa K, Takahashi H, Morihara T, Ishii R, Tagami S, Iwase M, Okochi M, us about language, and perhaps even thought, in schizophrenia? Int J Kamino K, Kazui H, Tanaka T, Kudo T, Takeda M: TATA box-binding protein Psychophysiol 2010, 75:66–76. gene is associated with risk for schizophrenia, age at onset and 49. Vrtunski PB, Simpson DM, Weiss KM, Davis GC: Abnormalities of fine motor prefrontal function. Genes Brain Behav 2009, 8:473–480. control in schizophrenia. Psychiatry Res 1986, 18:275–284. 27. Lindenmayer JP, Bernstein-Hyman R, Grochowski S: A new five factor model of schizophrenia. Psychiatr Q 1994, 65:299–322. doi:10.1186/1744-9081-9-40 28. Skol AD, Scott LJ, Abecasis GR, Boehnke M: Joint analysis is more efficient Cite this article as: Ohi et al.: The impact of the genome-wide than replication-based analysis for two-stage genome-wide association supported variant in the cyclin M2 gene on gray matter morphology in studies. Nat Genet 2006, 38:209–213. schizophrenia. Behavioral and Brain Functions 2013 9:40. 29. Faul F, Erdfelder E, Lang AG, Buchner A: G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 2007, 39:175–191. 30. Ellison-Wright I, Bullmore E: Anatomy of bipolar disorder and schizophrenia: a meta-analysis. Schizophr Res 2010, 117:1–12. Submit your next manuscript to BioMed Central 31. Wang CY, Shi JD, Yang P, Kumar PG, Li QZ, Run QG, Su YC, Scott HS, Kao KJ, and take full advantage of: She JX: Molecular cloning and characterization of a novel gene family of four ancient conserved domain proteins (ACDP). Gene 2003, 306:37–44. • Convenient online submission 32. Goytain A, Quamme GA: Functional characterization of ACDP2 (ancient conserved domain protein), a divalent metal transporter. • Thorough peer review Physiol Genomics 2005, 22:382–389. • No space constraints or color ﬁgure charges 33. Stuiver M, Lainez S, Will C, Terryn S, Gunzel D, Debaix H, Sommer K, Kopplin K, • Immediate publication on acceptance Thumfart J, Kampik NB, Querfeld U, Willnow TE, Nemec V, Wagner CA, Hoenderop JG, Devuyst O, Knoers NV, Bindels RJ, Meij IC, Muller D: CNNM2, • Inclusion in PubMed, CAS, Scopus and Google Scholar encoding a basolateral protein required for renal Mg2+ handling, is • Research which is freely available for redistribution mutated in dominant hypomagnesemia. Am J Hum Genet 2011, 88:333–343. 34. Stranger BE, Nica AC, Forrest MS, Dimas A, Bird CP, Beazley C, Ingle CE, Submit your manuscript at Dunning M, Flicek P, Koller D, Montgomery S, Tavare S, Deloukas P, www.biomedcentral.com/submit
Behavioral and Brain Functions – Springer Journals
Published: Oct 25, 2013
Access the full text.
Sign up today, get DeepDyve free for 14 days.