|Home | About | Journals | Submit | Contact Us | Français|
A missense polymorphism in the NRG1 gene, Val > Leu in exon 11, was reported to increase the risk of schizophrenia in selected families from the Central Valley region of Costa Rica (CVCR). The present study investigated the relationship between three NRG1 genetic variants, rs6994992, rs3924999, and Val > Leu missense polymorphism in exon 11, in cases and selected controls from an isolated population from the CVCR. Isolated populations can have less genetic heterogeneity and increase power to detect risk variants in candidate genes. Subjects with bipolar disorder (BD, n = 358), schizophrenia (SZ, n = 273), or unrelated controls (CO, n = 479) were genotyped for three NRG1 variants. The NRG1 promoter polymorphism (rs6994992) was related to altered expression of NRG1 Type IV in other studies. The expression of NRG1 type IV in the dorsolateral prefrontal cortex (DLPFC) and the effect of the rs6994992 genotype on expression were explored in a postmortem cohort of BD, SZ, major depressive disorder (MDD) cases, and controls. The missense polymorphism Val > Leu in exon 11 was not significantly associated with schizophrenia as previously reported in a family sample from this population, the minor allele frequency is 4%, thus our sample size is not large enough to detect an association. We observed however an association of rs6994992 with NRG1 type IV expression in DLPFC and a significantly decreased expression in MDD compared to controls. The present results while negative do not rule out a genetic association of these SNPs with BD and SZ in CVCR, perhaps due to small risk effects that we were unable to detect and potential intergenic epistasis. The previous genetic relationship between expression of a putative brain specific isoform of NRG1 type IV and SNP variation was replicated in postmortem samples in our preliminary study.
Schizophrenia (SZ) is a debilitating brain disease characterized by delusions, hallucinations, decreased emotional affect, paranoia, and motor deficiencies (Liddle et al. 1994). Though the exact neurobiology of schizophrenia is not wholly understood, family, twin, and adoption studies have demonstrated that schizophrenia is a complex disease with a significant genetic component. First-degree biological relatives of patients with schizophrenia have an estimated 10% risk of developing the disease, compared to 1% for the general population (Gottesman and Erlenmeyer-Kimling 2001). In twin studies, concordance rates of 41% – 65% have been seen in monozygotic twins compared to 0% – 28% in dizygotic schizophrenic twins, suggesting heritability as high as 85% (Tsuang et al. 2001). In light of this evidence, much effort has been directed toward the discovery of genes that increase the risk of schizophrenia.
Schizophrenia linkage to chromosome 8p has been identified in multiple studies (DeLisi et al. 2002; Kaufmann et al. 1998; Levinson et al. 1996; Pulver et al. 1995; Shaw et al. 1998). Neuregulin 1 (NRG1) is located at 8p12 and is involved in neurodevelopment, regulation of glutamate, and synaptic plasticity (Tosato et al. 2005). Stefansson et al (Stefansson et al. 2002) first reported an association between NRG1 and schizophrenia in an Icelandic population via a haplotype (HAPICE) consisting of five SNPs (SNP8NRG221132, SNP8NRG221533, SNP8NRG241930, SNP8NRG243177, SNP8NRG433E1006) and two microsatellite markers (478B14-848, 420M91395) located at the 5′ end of the gene that doubled the risk for the disorder. Since this initial report, confirmatory studies of NRG1 association have been reported in different populations in Scotland (Stefansson et al. 2003), China (Li et al. 2004), Hungary (Keri et al. 2009), Japan (Fukui et al. 2006), Sweden (Alaerts et al. 2009), and a second Scottish cohort (Thomson et al. 2007), though the associated haplotype varies between studies. Negative studies of association have also been reported in Japan (Ikeda et al. 2008; Iwata et al. 2004), Ireland (Thiselton et al. 2004), Denmark (Ingason et al. 2006), Spain (Rosa et al. 2007), and the United States (Crowley et al. 2008).
Variation in NRG1 has been found to be associated with various biological indices that are known to underlie schizophrenia. For example, brain imaging measurements of the anterior cingulate (Wang et al. 2009), prefrontal functioning (Mechelli et al. 2008; Mechelli et al. 2009), overall white matter integrity (Zuliani et al. 2011), excitatory synapse development (Ting et al. 2011), GABA interneuron dysfunction (Ting et al. 2011), and immune system dysregulation (Marballi et al. 2010; Shibuya et al. 2010)have been linked to genetic variation in NRG1. These observations contribute to the validity of the association of NRG1 with schizophrenia and its likeliness to contribute to the underlining basis for the development of this illness.
A novel missense mutation present in NRG1 (Val > Leu in exon 11) that increased the risk of schizophrenia in individuals from Costa Rica was reported in a family based association analysis (Walss-Bass et al. 2006a). Our research group has been independently studying the genetics of schizophrenia in families and unrelated individuals from the same region of the Central Valley of Costa Rica (CVCR) (Bertisch et al. 2009; Cooper-Casey et al. 2005; DeLisi et al. 2002); therefore we attempted to replicate the missense association with the Val > Leu polymorphism in exon 11 in this study. We do not believe that we have any overlap in samples with those reported in the Walss-Bass et al. (2006) paper (Walss-Bass et al. 2006a), although we have not made a formal comparison due to potential IRB ramifications. We also selected rs3924999 for analysis as it has been linked to lower prepulse inhibition, an endophenotype of schizophrenia (Hong et al. 2008), and for its association with schizophrenia in a Chinese Han cohort (Zhang et al. 2009). In addition rs6994992 was selected based on its inclusion in the HAPICE risk haplotype (Stefansson et al. 2002), and based on evidence that the T/T genotype is associated with decreased activation of frontal and temporal lobe regions and increased risk of psychosis (Hall et al. 2006). This SNP was also selected for investigation for its function in promoting expression of the NRG1 type IV transcript in postmortem tissue (Law et al. 2006);(Shamir and Buonanno 2010). Therefore, we studied the expression of NRG1 type IV in postmortem brain sample and association with genotypes of rs6994992. Though there is NRG1 intragenic epistasis between 5′ and 3′ markers (Nicodemus et al. 2010) in functional imaging studies, our group selected the present markers to attempt replication of implicated SNPs that were also functionally relevant to the CVCR collection. The SNPs chosen for this study were selected primarily based on conclusions published in the literature when designing the study. As associations between schizophrenia and the HAPICE haplotype have been both supported and refuted across varying populations, we were interested in adding the CVCR results to the pool of association data. rs6994992 was also selected based on reports of its function in promoting expression of the NRG1 type IV transcript in postmortem tissue. Finally, cost of the materials needed to test a wider range of markers was also a factor in determining which SNPs to investigate. As this was a preliminary association study to test our samples with these three implicated SNPs, the present data can be added to meta-analysis using SNPs in NRG1.
In addition to schizophrenia and control subjects, we have included bipolar subjects in the genotyping and brain gene expression as well. Bipolar disorder (BD) illness affects approximately 0.8 – 1.6% of the population (Kessler et al. 1997) and is characterized by cyclical episodes of mania and depression, with a return to normal state between episodes (Berns and Nemeroff 2003). There is a significant genetic component to BD based upon twin studies; BD has an estimated heritability as high as 93% (Kieseppa et al. 2004). Though schizophrenia and bipolar disorder have been historically categorized as divergent psychopathologies, there is a growing body of evidence suggesting that causative commonalities exist between the two disorders (Badner and Gershon 2002; Berrettini 2003). To date, multiple studies have shown schizophrenia implicated loci and genes having a positive association to BD (Craddock et al. 2006; Harrison and Weinberger 2005), including 8p12 (Park et al. 2004) and NRG1 (Green et al. 2005; Prata et al. 2009). NRG1 in particular has also been shown to possibly play a role in bipolar psychopathology, pointing to a common involvement of this cell-cell interaction and growth involved protein in both bipolar disorder and schizophrenia (Georgieva et al. 2008; Thomson et al. 2007).
A population sample from the central valley of Costa Rica was chosen for this study due to its geographical isolation and genetic homogeneity. The Costa Rican genome is comprised of largely European and Amerindian ancestry (Morera et al. 2003), as a result of six waves of Spanish colonization and admixture with the indigenous Amerindians (Carlos, 1997). It is estimated that fewer than 1000 families gave rise to the three million residents of the CVCR (Mathews et al. 2004), and due to mountainous boundaries and dense lowland jungle, the CVCR population has remained isolated from immigration and emigration for 500 years. Consequently, the CVCR region presents a homogenous population perfect for studying genes underlying complex genetic disorders. Chromosomal areas of interest have already been identified in this cohort for both schizophrenia (Cooper-Casey et al. 2005; DeLisi et al. 2002; Walss-Bass et al. 2006b) and bipolar disorder (Freimer et al. 1996).
While large association studies have shown the relative value of identifying common variants that contribute statistically significant associations, there are usually small relative risks for disease attributable to an individual SNP (e.g. (Purcell et al. 2009; Ruderfer et al. 2011). However, family based studies in relatively isolated populations can offer knowledge about regions of interest that might contain rare variants in pathways that have etiological relevance to schizophrenia. Thus, the CVCR collection offers a resource for exploring the effects of genes that have been largely implicated in multiple studies, and perhaps can increase the association signal by increased genetic homogeneity that is lacking in larger association studies. Thus, as sample sizes for schizophrenia and bipolar disorder are projected to be larger in the future, the relative risks attributable to a single variant will most likely be decreasing due to genetic heterogeneity. With a modest sample size pursued in the present study, there is adequate power in pursuing biologically based genes on an a priori basis. The purpose of this study is to examine NRG1 SNPs previously associated with SZ in a geographically isolated, relatively homogenous population from the CVCR.
Subjects were recruited with the approval of the Ministry of Health of Costa Rica and the ethics committee for the Hospital Nacional Psiquiatrico and by the Institutional Review Board at the University of California at Irvine. Unrelated individuals with diagnoses of schizophrenia (n = 273) and bipolar disorder (n = 358) whose four grandparents were of Spanish descent were obtained by screening patients admitted to the National Psychiatric Hospital of Costa Rica as previously described(DeLisi et al. 2001; DeLisi et al. 2002). Control subjects (n = 479) with the same ancestral criteria were recruited from large companies via questionnaires administered by the companies’ health services department. Controls were selected if no family history of schizophrenia, bipolar disorder, suicide or hospitalization for psychiatric reasons were present, and if self-reports of psychosis, diagnosis of schizophrenia, bipolar disorder, use of medications for depression or psychiatric conditions, and suicide attempts were negative. Interviews of affected subjects were conducted with the Diagnostic Interview for Genetic Studies (Nurnberger et al. 1994) using a translated Spanish DIGS 2 (DeLisi et al. 2001) and diagnoses were based upon DSM-IV criteria (Association 1994) by a consensus of two independent local psychiatrists. A third independent psychiatrist made a final diagnosis, based on family history, medical records, and a summary of personal interviews. All participants gave written informed consent for participation.
Blood samples were collected from participants in Costa Rica and sent to the laboratory at UC Irvine. DNA was isolated from these samples via 10% SDS and Proteinase K digestion, phenol-chloroform extraction, followed by a sodium acetate precipitation (Bell et al. 1981). Purity and concentration were assessed by 260 nm and 280 nm absorbances on the SpectraMax Plus spectrophotometer (Molecular Devices, Sunnyvale, CA) and aliquots were diluted to a working concentration of 2.5ng/μl.
DNA samples were genotyped for NRG1 SNPs rs3994992 and rs6994992 using pre-validated TaqMan 5′-allele discrimination assays (Applied Biosystems, Foster City, CA). The third polymorphism G/T in exon 11 (Val > Leu) was genotyped with a custom TaqMan assay using the following reference sequence in the Walss-Bass study (Walss-Bass et al. 2006a): GAACATGGACAATGTCATGCAGCATGCCCACTGTTTGGTTGTAGTCAGTCCTGGCAAGTGGAAGTGACCTGTGATGACATCTGCTCTCATCCCTTTCCAGAGGCGGAGGAGCTGTACCAGAAGAAGTGCTGACCATAACCGGCATCTGCATCGCCCTCCTTGTGGTCGGCATCATGTGT[G/T]TGGTGGCCTACTGCAAAACCAAGTAAACCTTCTTTCTCCATGCCTTTCTCTCTCCTTCATGCAGAGACAGCTTAGATGGCCAGGGCTTTGCAGAATCTGAGCTCCACAGCCTAGTCTTGGGG. The Walss-Bass assay was performed on an ABI Prism 7000 Sequence Detection system, in a total reaction of 25 μl (4 μl DNA, 12.5 μl TaqMan Universal PCR Master Mix, No AmpErase UNG, 1.25μl 40× TaqMan assay, 7.25 μl H2O) using the following amplification conditions: denaturation at 95 °C for 10 minutes, followed by 40 cycles at 95 °C for 15 seconds and at 55 °C for 1 minute. Individual PCR reactions for rs3924999 and rs6994992 were carried out on an ABI Prism 7900 Sequence Detection system in a total reaction of 12 μl (4 μl DNA, 5 μl TaqMan Universal PCR Master Mix, No AmpErase UNG, 0.5μl 40× TaqMan assay, 2.5 μl H2O) using the following amplification protocol: denaturation at 95 °C for 10 minutes, followed by 50 cycles at 92 °C for 15 seconds and at 58 °C for 1.5 minutes. The genotype of each sample was determined by measuring allelic-specific fluorescence using SDS 2.3 software for allelic discrimination (Applied Biosystems).
Genotype and allelic statistical analyses were performed using Yates corrected χ2 for continuity and Fisher’s Exact Test for analyses that contained low cell numbers. Deviation from Hardy–Weinberg equilibrium (HWE) was tested using on line calculator (http://ihg.gsf.de/cgi-bin/hw/hwa1.pl).
Brain samples were obtained at the UC Irvine/UC Davis Brain Repository through a uniform process approved by the Institutional Review Board. Postmortem diagnoses were made through an extensive review of multiple sources of information including the medical examiner’s conclusions, coroner’s investigation, medical and psychiatric records, toxicology results, interviews of the decedents’ next-of-kin and a neuropathological examination as previously described (Tomita et al. 2004; Vawter et al. 2006). The human brain dissection and freezing protocol were performed as previously described (Jones et al. 1992; Vawter et al. 2006) and brains were stored in −80°C freezers until further dissected. RNA was extracted from 100 mg samples of dissected brain using a standard Trizol (Life Technologies, Carlsbad, California) procedure. Integrity of total RNA was evaluated via 18S and 28S ratios and RNA integrity numbers (RIN) using the 2100 Agilent Bioanalyzer (Agilent Technologies, Santa Clara, CA). Only samples with pH values higher than 6.5 were included in the analysis. cDNA was synthesized with oligo dT primers using the Superscript procedure (Invitrogen, Carlsbad CA). Expression levels of NRG1 IV were interrogated by TaqMan expression assay (ABI) using the same primer and probe sequence provided in the Law et al. study in which differences in gene expression for this subtype were observed (Law et al. 2006): forward 5′GCTCCGGCAGCAGCAT3′; reverse 5′GAACCTGCAGCCGATTCCT3′; internal Probe 5′ FAM ACCACAGCCTTGCCT-MGB-3′. Since the NRG1 E187 exon has high homology with other genes (Steinthorsdottir et al. 2004), a TaqMan probe provided specificity for amplification of the NRG1 gene as the probe spanned the junction of the E187 - Ig1 exons (Law et al. 2006).
After amplification, the PCR product was run on a 2% agarose gel, the amplicon’s specificity was confirmed by sequencing using the same primers used in the TaqMan assay. We compared the NRG1 type IV TaqMan expression assay results for diagnoses groups (Table 1) to controls by an analysis of covariance taking into account age, sex, and RNA quality. The TaqMan expression was normalized with GAPDH. Brain gene expression levels were compared in bipolar disorder (n = 10), control (n = 22), MDD (n = 13) for a total of 45 postmortem subjects that had genotypes for NRG1 rs69949992.
The genotype counts for each SNP are shown in Table 2. The recessive and dominant association tests for three NRG1 SNPs genotyped were not significant with either BD or SZ (Table 3). There was a trend for SNP (Val > Leu exon 11) in BD cases only to not be in Hardy Weinberg equilibrium (nominal p = 0.043).
Two brain regions (hippocampus and DLPFC) were analyzed for expression differences of NRG1 Type IV. For the hippocampus results, there were no statistically significant differences in expression for SZ, BD, or MDD cases compared to controls; however, in the DLPFC, the MDD cases (n = 12) showed decreased expression (p = 0.004) compared to controls (n = 22) when pH, age, RIN, and gender are included in the ANCOVA model. The RIN factor was significant (p = 0.05). The hippocampus showed very low levels of amplification indicating low expression of intact poly-adenylated mRNA.
We next investigated the rs6994992 genotype effect on NRG1 type IV expression in hippocampus and DLPFC. The genotype effect was significant (p = 0.040) only in DLPFC; however there were no homozygous T carriers in the SZ group. We did confirm that NRG1 type IV expression was increased in the TT compared to the CC group in the DLPFC and that the effect was significant in the direction previously reported as the TT genotype showed a significantly increased expression (p = 0.024) of 15.3 fold compared to CC genotype (Figure 1). The CT group also showed a significant difference compared to the TT group (p = 0.023); again, the direction supported the dominant effect of the T allele increasing expression of the NRG1 type IV expression. The CT group did not show a significant difference from the CC group (p = 0.83). Interestingly, the MDD subjects showed a decreased expression compared to controls for both the CC and CT genotype group comparisons (p = 0.009, p = 0.018), the TT genotypes were not different comparing expression of NRG1 type IV between MDD and controls (p = 0.21).
Within the isolated CVCR population, this study failed to find an association with schizophrenia or bipolar disorder testing three NRG1 SNPs. Thus while our study had 80% – 86% power for SZ and BD, respectively to find association for two of the more common SNPs (Skol et al. 2007) with risk ratio of ≥ 1.37, it was not adequately powered for the rarer exon 11 missense SNP. This minor allele frequency requires a larger sample to definitively test for association. Though there are positive reports for NRG1 as both a schizophrenia and bipolar susceptibility gene for one or more of these SNPs, this hypothesis has not been consistently proven in the literature as shown in the introduction. Although we tested the same missense mutation previously associated in a family sample in CVCR (Walls-Bass, 2006), we did not replicate these findings in a larger case - control analysis from the same CVCR population, perhaps due to this rare SNP frequency of 4%.
Additionally, we confirmed a prior report of an association between higher expression of NRG1 type IV and the rs6994992 T/T genotype in the DLPFC as previously reported (Law et al. 2006) (Shamir and Buonanno 2010). We also report the preliminary finding of decreased NRG1 type IV expression in MDD in the DLPFC. Although previous studies have not found an association between NRG1 SNPs and MDD in a large sample of European ancestry (Schosser et al. 2010), there was decreased NRG1-alpha protein in MDD and SZ in the prefrontal cortex (Bertram et al. 2007). Since NRG1 has strong pleiotropic effects related to growth factor signaling in the brain, the findings of decreased NRG1 expression in prefrontal cortex could lend support to the overall growth factor hypothesis of depression (Evans et al. 2004).
Two caveats to the postmortem findings are that we cannot rule out the effect of antidepressant medications without animal studies as a potential cause of this decrease in MDD. The decreased expression in DLPFC of NRG1 type IV in MDD requires additional postmortem studies in part based upon small sample size, for validation, but is consistent with a reported decrease of NRG1 protein in MDD frontal cortex. In conclusion, while the present findings do not support association of the NRG1 variants in the CVCR population with schizophrenia or bipolar disorder, the tested SNPs might have some intergenic epistasis or be of a smaller effect size that we do not have power to detect. The present findings continue to show that there is a robust effect of genetic variation on NRG1 type IV expression which is thought to be brain-specific and with multiple impacts on brain function.
Support was received from the William Lion Penzner Foundation, Pritzker Neuropsychiatric Disorders Research Consortium for support of the collection and genotyping of the Costa Rica samples and the NIMH Conte Center funding P50 MH60398 for brain sample collection. We appreciate the UC Irvine Davis staff for contributions to brain tissue acquisition, and the Costa Rica staff for subject ascertainment.
Role of the funding source
The authors are members of the Pritzker Neuropsychiatric Disorders Research Consortium, which is supported by the Pritzker Neuropsychiatric Disorders Research Fund L.L.C. A shared intellectual property agreement exists between this philanthropic fund and the University of Michigan, Stanford University, the Weill Medical College of Cornell University, HudsonAlpha Institute of Biotechnology, the Universities of California at Davis, and at Irvine, to encourage the development of appropriate findings for research and clinical applications.
Conflict of interest
All authors have no conflicts of interest to declare.
ContributorsAuthors EM, BR, WEB, LED, WB, and MPV conceived and designed the study. EM, BR, LED, WB, and MPV carried out the computational analyses and candidate gene selection. EM, BR, AM, LED conducted subject screening. EM, BR, carried out the genotyping. EM and MPV performed the statistical analysis of the genotyping and expression data. RMM, HA, SJW, JB, EGJ, AS, WEB provided the guidance and additional support on this project. EM, BR, WEB, LED, WB, and MPV wrote the first draft of the paper, all authors revised the current paper. LED, AM, and WB recruited, diagnosed, and gathered patients and controls. BR, EM, LED, AM, WB, MPV contributed to the collection and preparation of control DNA samples.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errorsmaybe discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.