RESEARCH ARTICLE Neuropsychiatric Genetics Analysis of Neurogranin (NRGN) in Schizophrenia Rhodri Ll. Smith,* Deborah Knight, Hywel Williams, Sarah Dwyer, Alex Richards, George Kirov, Michael C. O’Donovan, and Michael J. Owen MRC Centre for Neuropsychiatric Genetics and Genomics, Department of Psychological Medicine, Henry Wellcome Building, School of Medicine, Cardiff University, Cardiff CF14 4XN, United Kingdom Received 22 December 2010; Accepted 11 March 2011 A recent study reported a genome-wide signiﬁcant association between schizophrenia and rs12807809—a SNP located approximately 3 kbp upstream of the neurogranin gene (NRGN). We sought to determine if (a) NRGN contains common exonic variants or variants affecting expression (eQTLs) that could account for the association with rs12807809 and (b) there exist rare non-synonymous highly penetrant variants that could potentially confer high risk of schizophrenia. We sequenced all four exons of NRGN in a screening set of 14 individuals but found no novel common polymorphisms. We additionally sequenced the coding exons in up to 1,113 individuals (699 cases) but this revealed only a singleton-coding variant in exon 2 (G246T leading to Gly-55 ! Val amino acid change) in which prediction of function analysis suggested is likely to be benign. Finally, analysis of a brain expression dataset of at least 130 individuals did not identify any eQTLs that were correlated with associated SNP rs12807809 following correction for multiple testing. Ó 2011 Wiley-Liss, Inc. Key words: neurogranin; sequencing; GWAS; schizophrenia; case-control INTRODUCTION Schizophrenia is a psychiatric disorder with a complex etiology involving both genetic and environmental factors. Affecting approximately 1% of the population, the disorder is characterized by hallucinations, delusions, and cognitive deﬁcits thought to arise at least in part from defective neurodevelopment [Weinberger, 1995]. Neurogranin (NRGN), a postsynaptic protein kinase substrate, functions primarily as a regulator of calmodulin (CaM) activity in the postsynaptic compartment [Huang, 1993]. NRGN is abundantly expressed in structures that are involved in learning and memory, such as dendritic spines [Watson, 1992] and CA1 pyramidial neurones of the hippocampus. During embryonic development, maternal thyroid hormones, and retinoids regulate NRGN expression which is maximal at the time of cortical synapse formation and dendritic growth [I~ niguez, 1993]. Such ﬁndings suggest a key role for NRGN in neurodevelopment [I~ niguez, 1993] and cognition [Pak, 2000; Huang, 2006, 2007] both of which are thought to be relevant to the etiology of schizophrenia [Goldman-Rakic, 1994]. The possible involvement of NRGN in schizophrenia was originally investigated in a case-control and a family-based candidate gene association study of Portuguese subjects. Signiﬁcant associa- Ó 2011 Wiley-Liss, Inc. How to Cite this Article: Smith RLl, Knight D, Williams H, Dwyer S, Richards A, Kirov G, O’Donovan MC, Owen MJ. 2011. Analysis of Neurogranin (NRGN) in Schizophrenia. Am J Med Genet Part B 156:532–535. tion was detected between SNP rs7113041 (Fig. 1) and affected status in males [Ruano, 2008]. More recently, the SGENE-plus [Stefansson, 2009] consortium reported genome-wide signiﬁcant evidence for association (P ¼ 2.3 109; OR ¼ 1.15) between schizophrenia and a marker (rs12807809; Fig. 1) just upstream of NRGN in a combined sample of 12,965 cases and 34,598 controls derived from three large genome-wide association studies (GWAS) as well as additional replication samples. Thus, currently, the NRGN locus is one of only four loci (ZNF804A [O’Donovan, 2008], TCF4 [Stefansson, 2009], NRGN [Stefansson, 2009], and the wider MHC region [Stefansson, 2009]) that are implicated in schizophrenia at genome-wide levels of signiﬁcance. In the present study, we sought to determine whether NRGN contains common exonic variants, or variants that inﬂuence expression of NRGN or neighboring genes, that might account for the association between rs12807809 and schizophrenia. Second, we attempted to determine if rare non-synonymous variants exist at the locus that might confer high risk of the disorder. The hypothesis that rare variants contribute to schizophrenia is supported by the high incidence of schizophrenia in individuals carrying rare deletions of 22q11.2 [Murphy, 1999] as well as a number of other rare copy number variants (CNVs) [The International Schizophrenia Consortium, 2008; Kirov, 2009a,b; Stefansson, 2008; Walsh, 2008; Xu, 2008], although it should be noted that with respect to the Grant sponsor: Medical Research Council. Rhodri Ll Smith, Deborah Knight are joint ﬁrst authors *Correspondence to: Dr. Rhodri Ll. Smith, BSc(Hons), MSc, PhD, Henry Wellcome Building, Heath Hospital, Cardiff University, Cardiff, Glamorgan, UK. E-mail: firstname.lastname@example.org Published online 2 May 2011 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/ajmg.b.31191 532 SMITH ET AL. schizophrenia phenotype, these CNVs are not highly penetrant [Vassos, 2010]. With the exception of CNVs and major structural re -arrangements, there have been no systematic scans of the genomes of people with schizophrenia for rare single base (or small insertion deletion) variants. The reason for this is that rare variant detection relies mainly on sequencing. This has only very recently become technically feasible at a genome-wide level, but remains very expensive. In the absence of few clear targets for sequencing from unambiguous linkage ﬁndings, sequencing approaches to rare mutation detection have therefore largely focussed on candidate genes, an approach that has met with some notable successes in nonpsychiatric phenotypes [Cohen, 2004; Fearnhead, 2004; Ahituv, 2007; Ji, 2008]. Clearly, as one of only four loci that are implicated in schizophrenia at genome-wide levels of signiﬁcance, NRGN is a pre-eminent candidate gene for schizophrenia, although it should be noted that GWAS studies implicate genomic locations, not genes, and it is always possible that the association at this locus points to an as yet unknown functional element. METHODS AND MATERIALS NRGN is approximately 7.4 kbp in length and is located on chromosome 11q24.2. It has four exons with exons 1 and 2 encoding the protein whilst exons 3 and 4 remain untranslated (Fig. 1). The gene has two known reference transcripts which differ by three bases; however, both encode the same 78 amino acid protein. MUTATION SCREENING We initially sequenced all four exons of NRGN in 14 unrelated Caucasian subjects from the UK diagnosed with schizophrenia, each of whom had at least one affected sib. Details of case recruitment and diagnosis are given elsewhere [O’Donovan, 2008]. Briefly, all cases were white, born in the British Isles, and had a consensus diagnosis of schizophrenia according to DSM-IV criteria made by two independent raters following semi-structured interviews using the Present State Examination [Wing, 1974] or the Schedules for Clinical Assessment in Neuropsychiatry (SCAN) [Wing, 1990] and 533 a review of case records. Cases with substance-induced psychotic disorder or psychosis due to a general medical condition were excluded. Multicenter and Local Research Ethics Committee approval were obtained, and all subjects gave written informed consent to participate. To further investigate the gene for rare variants, we sequenced the coding exons of the gene in a minimum of 652 Bulgarians cases diagnosed with schizophrenia according to DSM-IV and 386 healthy Bulgarian controls. Cases were recruited from psychiatric inpatient and outpatient services. All had a history of hospitalization for schizophrenia. Each proband was interviewed with an abbreviated Bulgarian version of SCAN (Schedules for Clinical Assessment in Neuropsychiatry) that contains the sections on psychosis and affective disorders. Consensus best-estimate diagnoses were made according to DSM-IV criteria based on interview and hospital notes. Ethics committee approval was obtained from all regions where patients were recruited. All subjects provided written informed consent for participation in genetic association studies. The sample for which sequence data were obtained for the coding exons has power of 80% to detect a neutral allele with a population frequency of >0.0007. Assuming an allele with an effect size (OR) of 10 or of 2, the respective population frequencies for which we have 80% power are >1.3 103 and >0.03. Power to detect alleles in the non-coding exons with a MAF > 0.05 and MAF > 0.1 was 80% and 95%, respectively. All exons were ampliﬁed by PCR using speciﬁed forward and reverse primers. PCR products were subsequently cleaned using Ampure (Agencourt Bioscience Corporation, Beverly, MA). Sequencing by PCR was achieved using BigDye Chemistry (Applied Biosystems, Foster City, CA) in both the forward and reverse directions for the common variants, and in the forward direction for rare variant detection. Products were then cleaned using CleanSEQ (Agencourt Bioscience Corporation) and ﬁnally sequenced on a 3100 capillary sequencer (Applied Biosystems). Exonic regions were screened for sequence variation using NovoSNP (Version 2.1.6) software [Weckx, 2005]. Sequencing reactions that failed or were of poor quality were removed from the analysis. Traces with a NovoSNP score >5 were manually checked. FIG. 1. A schematic of the neurogranin gene highlighting the location of both SNPs previously showing evidence of association with schizophrenia. AMERICAN JOURNAL OF MEDICAL GENETICS PART B 534 eQTL Analysis eQTL Analysis NRGN expression was analyzed within two eQTL datasets derived from brain tissue. The ﬁrst, published by Myers and Gibbs  consists of gene expression proﬁles generated using RNA extracted from the frontal or temporal cortex of 184 neuropathologically normal human brains. Genotyping and expression analysis was performed using the Affymetrix GeneChip Human Mapping 500 K Array Set and Illumina HumanRefseq-8 Expression BeadChip platforms. The second eQTL dataset [Gibbs and van der Brug, 2011] analyzed 143 neurologically normal subjects of European ancestry. Frozen tissue samples were obtained from four brain regions (cerebellum, pons, frontal, and temporal cortices) for each subject. Genotyping was performed using Inﬁnium HumanHap550 BeadChips (Illumina, San Diego, CA). Expression analysis was performed using Illumina HumanRef-8 Expression BeadChips. Genotype data were used as presented in the original publications [Myers and Gibbs, 2007; Gibbs and van der Brug, 2011]. The expression data were normalized and log transformed as described in the original articles [Myers and Gibbs, 2007; Gibbs and van der Brug, 2011]. Samples with over 10% missing data were removed from the Myers et al. dataset, leaving 161 samples. In the Gibbs et al. dataset, samples where over 50% of the probes had a present/absent call P-value of >0.05 were removed. This left 143 samples in the temporal cortex dataset, 133 samples in the frontal cortex dataset, 130 samples in the cerebellum dataset, and 134 samples in the pons dataset. Expression levels were then corrected for clinical and biological factors that might inﬂuence expression, using linear regression in the statistical package R [Ihaka and Gentleman, 1996]. In the Myers et al. dataset, these factors were sex, age at death, cortex region, postmortem interval, institute source, hybridization batch, and ENO2 expression. For the Gibbs et al. dataset, these factors were sex, age at death, postmortem interval, institute source, hybridization batch and ENO2 expression. ENO2 expression was included as a factor because it is a reliable neuronal marker [Marangos and Schmechel, 1987], and so including it will correct for differing proportions of neurons between samples. eQTLs were tested by linear regression of normalized NRGN expression level on SNP genotypes (coded as the number of minor alleles at each SNP: 0, 1, or 2). This analysis was performed using the—assoc option in the software package PLINK [Purcell and Neale, 2007]. In the expression dataset, we identiﬁed two SNPs which were at least weakly correlated (r2 > 0.2) with rs12807809, the variant associated with schizophrenia in the earlier study [Stefansson, 2009]. Neither of these nor rs12807809 was associated with NRGN expression (P > 0.05). Moreover, given that association formally points to a region rather than a gene, we investigated whether expression of any transcripts within 1 Mb of this SNP were associated with schizophrenia. However, rs12807809 was not associated at a nominal level of signiﬁcance with any of the transcripts following bonferroni correction (data not shown) in the region although data was available from one of the Gibbs et al. dataset only. RESULTS In the screen set in which all exons were sequenced (n ¼ 14), analysis of exons 1–4 revealed no variants, which effectively precludes the existence of common exonic variation in this gene (95% power to detect variants with MAF 1%). In the larger sample in which the coding region was screened, we observed only a single variant, G > T at position 246, in exon 2. The variant is nonsynonymous, and is predicted to change amino acid residue 55 of NRGN from a glycine to a valine. No other polymorphisms were detected in our sequence data, including rs61910621 which is listed in dbSNP build 131. DISCUSSION The NRGN locus is one of a small number of loci that have been associated with schizophrenia at a genome-wide signiﬁcant level of support [Ruano, 2008], and given its function, the NRGN gene itself is a likely source of the association signal. Our study aimed to identify if common exonic variants exist within the gene, or common eQTLs in the vicinity of the gene, might account for the association. We also sought to establish if there are rare nonsynonymous variants that contribute to the disease. Observing the latter would ﬁrmly establish NRGN as the relevant functional unit responsible for the association at this locus. However, we found only a single non-synonymous variant and no synonymous variants within this gene. Given high power (>0.99) to detect variants at a frequency >0.01, our data effectively exclude a role in NRGN for even fairly rare coding susceptibility variants in schizophrenia. However, it should be noted that risk alleles or haplotypes inﬂuencing expression could in principal be exclusive to a particular developmental stage or anatomical region and thus would not have been detected by our study. We did identify a single occurrence non-synonymous variant in a case which results in a glycine to valine change. The nsSNP identiﬁed in the current study alters the amino acid sequence at position 55, which lies just outside the IQ domain of NRGN (encoded by amino acids 22–47) that is important for CaM binding [Prichard, 1999]. PolyPhen (http://genetics.bwh.harvard.edu/ pph/), a prediction of function program, suggests that this variant is likely to be benign. Given the singleton nature of the observation, we can therefore not implicate this variant as of likely relevance to schizophrenia either statistically or on the basis of likely functionality, although predictive algorithms do not preclude an effect. Clearly, our analysis cannot discount a role for ultra rare or unique high penetrance variants in this gene in schizophrenia; doing so would require sequence analysis of every person in the world with the disorder. However, our data do show if such variants exist, they must have very low population frequencies, and will be extremely difﬁcult to implicate in the disorder. Based upon the power analysis for a dominant allele with a relative risk of 10, our study puts an upper boundary of the cumulative population frequency of all such rare variants within NRGN exonic sequence at 0.0001. If such a set of variants exist at just below this level, it will be necessary to sequence approximately 8,000 cases and SMITH ET AL. 8,000 controls to demonstrate an association at a very modest P ¼ 0.05. In summary, our study provides strong evidence that common exonic variation does not account for the genome-wide signiﬁcant association between schizophrenia and variation at NRGN, and, with the caveats above, suggests that eQTLs at this locus cannot explain it either. The source of the association signal therefore remains to be explained. We also ﬁnd no evidence that rare but fairly highly penetrant coding variants at this locus are involved in schizophrenia. No study can exclude the existence of private or ultra rare variants, but if these exist, in the absence of multiple families showing unequivocal linkage to variants at this locus, documenting them will require sequencing studies at least as large as those deployed in GWAS studies. ACKNOWLEDGMENTS The authors are grateful to all individuals who have participated in, or helped with this research. The research was supported by the Medical Research Council (UK). 535 Ji W, et al. 2008. Rare independent mutations in renal salt handling genes contribute to blood pressure variation. Nat Genet 40(5):592–599. Kirov G, et al. 2009a. Support for the involvement of large copy number variants in the pathogenesis of schizophrenia. Hum Mol Genet 18(8): 1497–1503. Kirov G, et al. 2009b. Neurexin 1 (NRXN1) deletions in schizophrenia. Schizophr Bull 35(5):851–854. Marangos PJ, Schmechel DE. 1987. Neuron speciﬁc enolase, a clinically useful marker for neurons and neuroendocrine cells. Annu Rev Neurosci 10:269–295. Murphy KC, et al., 1999. High rates of schizophrenia in adults with velocardio-facial syndrome. Arch Gen Psychiatry 56:940–945. Myers AJ, Gibbs JR, et al. 2007. A survey of genetic human cortical gene expression. Nat Genet 39(12):1494–1499. O’Donovan M, et al. 2008. Identiﬁcation of loci associated with Schizophrenia by genome-wide association and follow-up. Nat Genet 40(9): 1053–1055. Pak JH, et al. . Involvement of Neurogranin in the modulation of calcium/ calmodulin-dependent protein kinase II, synaptic plasticity and spatial learning: A study with knockout mice. PNAS 97(21):11232–11237. Prichard L, et al. 1999. Interactions between neurogranin and calmodulin in vivo. J Biol Chem 274(12):7689–7694. REFERENCES Purcell S, Neale B, et al. 2007. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 81(3):559–575. Ahituv N, et al. 2007. Medical sequencing at the extremes of human body mass. Am J Hum Genet 80(4):779–791. Ruano D, et al. 2008. Association of the gene encoding neurogranin with schizophrenia in males. J Psychiatr Res 42:125–133. Cohen JC, et al. 2004. Multiple rare alleles contribute to low plasma levels of HDL cholesterol. Science 305(5685):869–872. Stefansson H, et al. 2008. Large recurrent microdeletions associated with schizophrenia. Nature 455:232–236. Fearnhead NS, et al. 2004. Multiple rare variants in different genes account for multifactorial inherited susceptibility to colorectal adenomas. Proc Natl Acad Sci USA 101(45):15992–15997. Stefansson H, et al. 2009. Common variants conferring risk of schizophrenia. Nature 460:744–747. Gibbs JR, van der Brug MP, et al. 2011. Abundant quantitative trait Loci exist for DNA methylation and gene expression in human brain. PLoS Genet 6(5):e1000952. Goldman-Rakic PS. 1994. Working memory dysfunction in schizophrenia. J Neuropsychiatr Clin Neurosci 6:348–357. Huang KP, et al. 1993. Characterisation of a 7.5-kDa protein kinase c substrate (RC3 protein, neurogranin) from rat brain. Arch Biochem Biophys 305:570–580. Huang FL, et al. 2006. Environmental enrichment enhances Neurogranin expression and hippocampal learning and memory but fails to rescue the impairments of Neurogranin null mutant mice. J Neurosci 26(23): 6230–6237. Huang FL, et al. 2007. Long-term enrichment enhances the cognitive behaviour of the aging neurogranin null mice without affecting their hippocampal LTP. Learn Mem 14:512–519. Ihaka R, Gentleman R. 1996. R: A language for data analysis and graphics. J Comput Graph Stat 5(3):299–314. I~ niguez MA, et al. 1993. Thyroid hormone regulation of RC3, a brainspeciﬁc gene encoding a protein kinase-C substrate. Endocrinology 133:467–473. The International Schizophrenia Consortium. 2008. Rare chromosomal deletions and duplications increase risk of schizophrenia. Nature 455: 237–241. Vassos E, et al. 2010. Penetrance for copy number variants associated with schizphrenia. Hum Mol Genet 19(17):3477–3481. Walsh T, et al. 2008. Rare structural variants disrupt multiple genes in neurodevelopmental pathways in schizophrenia. Science 320:539–543. Watson JB, et al. 1992. Localisation of the protein kinase c phosphorylation/ calmodulin binding substrate RC3 in dendritic spines of neostriatal neurons. PNAS 89:8581–8585. Weckx S, et al. 2005. NovoSNP, a novel computational tool for sequence variation discovery. Genome Res 15:436–442. Weinberger DR. 1995. From neuropathology to neurodevelopment. Lancet 346:552–557. Wing. JK. 1974. The measurment and classiﬁcation of psychiatric illness. Cambridge: Cambridge University Press. Wing JK, et al. 1990. SCAN. Schedules for clinical assessment in neuropsychiatry. Arch Gen Psychiatry 47(6):589–593. Xu B, et al. 2008. Strong association of de novo copy number mutations with sporadic schizophrenia. Nat Genet 40(7):880–885.