|Home | About | Journals | Submit | Contact Us | Français|
Background: Chromosome 22q11 deletion syndrome (22q11DS) increases the risk of development of schizophrenia more than 10 times compared with that of the general population, indicating that haploinsufficiency of a subset of the more than 20 genes contained in the 22q11DS region could increase the risk of schizophrenia. In the present study, we screened for genes located in the 22q11DS region that are expressed at lower levels in postmortem prefrontal cortex of patients with schizophrenia than in those of controls. Methods: Gene expression was screened by Illumina Human-6 Expression BeadChip arrays and confirmed by real-time reverse transcription-polymerase chain reaction assays and Western blot analysis. Results: Expression of GNB1L was lower in patients with schizophrenia than in control subjects in both Australian (10 schizophrenia cases and 10 controls) and Japanese (43 schizophrenia cases and 11 controls) brain samples. TBX1 could not be evaluated due to its low expression levels. Expression levels of the other genes were not significantly lower in patients with schizophrenia than in control subjects. Association analysis of tag single-nucleotide polymorphisms in the GNB1L gene region did not confirm excess homozygosity in 1918 Japanese schizophrenia cases and 1909 Japanese controls. Haloperidol treatment for 50 weeks increased Gnb1l gene expression in prefrontal cortex of mice. Conclusions: Taken together with the impaired prepulse inhibition observed in heterozygous Gnb1l knockout mice reported by the previous study, the present findings support assertions that GNB1L is one of the genes in the 22q11DS region responsible for increasing the risk of schizophrenia.
Schizophrenia, a devastating mental disorder that affects approximately 1% of the world's population, is a genetically complex disorder. The multifactorial polygenic model has received the most support as the mode of inheritance that underlies the familial distribution of schizophrenia; therefore, a variety of genetic, environmental, and stochastic factors are likely involved in the etiology. However, it is also possible that specific genes play major roles in susceptibility to schizophrenia. Genes involved in 22q11.2 deletion syndrome (22q11DS) substantially increases susceptibility to schizophrenia. 22q11DS is associated with several diagnostic labels including DiGeorge syndrome, velocardiofacial (or Shprintzen) syndrome (VCFS), conotruncal anomaly face, Cayler syndrome, and Opitz GBBB syndrome. Schizophrenia is a late manifestation in approximately 30% of 22q11DS cases, which is comparable to the risk to offspring of 2 parents with schizophrenia. The 22q11 deletion is detected relatively frequently in patients with schizophrenia; a number of studies have shown that 22q11DS schizophrenia is a true genetic subtype of schizophrenia1,2.
Although the deleted region is approximately 3 Mbp in most patients with 22q11DS, the critical region is approximately 1.5 Mbp.3,4 Less than 30 genes are located in the 22q11DS region. Studies of 22q11DS patients without the common chromosomal deletion suggested that the TBX1 is a major contributor to the conotruncal malformations of 22q11DS.5 One of the mutations in the TBX1 was found to be a loss-of-function mutation.6 Mice heterozygous for a null mutation in Tbx1 develop conotruncal defects.7 Deletion of one copy of the Tbx1 affects the development of the fourth pharyngeal arch arteries, whereas the homozygous mutation severely disrupts the pharyngeal arch artery system.8 The contribution of the TBX1 haploinsufficiency to psychiatric disease was suggested by the identification of a family with VCFS in a mother and her 2 sons. These 3 patients all had a null mutation of the TBX1, and one of the sons was diagnosed with Asperger syndrome after psychiatric assessment.9
Contribution of genes in the 22q11DS region to susceptibility to schizophrenia has been examined mainly by genetic association studies. Associations between schizophrenia and nucleotide variations in the ZNF74,10 DGCR, 11 DGCR14,12 PRODH,13 ZDHHC8,14 COMT,15–18 and CLDN519,20 genes have been reported. These associations, however, have not been confirmed in other populations19–22 or by meta-analyses.19–24
Studies of genetically engineered mice have provided supporting evidence for roles of the genes located in the human 22q11DS region in schizophrenia. Prodh knockout mice exhibited deficits in learning and responses to psychomimetic drugs.25 Observation of overlapping loci across 5 heterozygous mice strains with different deletion sites revealed that a 300-kb locus, which contains the Gnb1l, Tbx1, Gp1bb, and Sept5 genes, is crucial for impaired sensorimotor gating measured by prepulse inhibition test (PPI).9 In that study, the authors speculated that the GP1BB was unlikely to be related to schizophrenia because it is expressed only in platelets. The GP1BB causes Bernard-Soulier disease, which has no associated psychiatric disorders. The Sept5 heterozygous knockout mice did not show impaired PPI. Gnb1l or Tbx1 heterozygous knockout mice showed reduced PPI.9 Therefore, the authors concluded that the Tbx1 and Gnb1l are strong candidates for psychiatric disease in patients with 22q11DS.9 In another study, however, Tbx1 heterozygous knockout mice showed normal locomotor activity, habituation, nesting, and locomotor responses to amphetamine.25
Recently, Williams et al26 reported associations between polymorphisms in the GNB1L gene region and schizophrenia in the United Kingdom, German, and Bulgarian population. They found excess homozygosity at rs5746832 and rs2269726 in male schizophrenia subjects and that the markers associated with male schizophrenia were related with cis-acting changes in GNB1L expression. These mouse and human studies indicated a correlation between GNB1L gene expression and psychosis.
The working hypothesis of the present study was that genes in the 22q11DS region involved in the susceptibility to schizophrenia were likely to be expressed at lower levels in patients with schizophrenia than in control subjects. We performed a scan of expressional changes of the genes in the 22q11DS region in schizophrenic and control prefrontal cortex and found that the GNB1L gene was compatible with our hypothesis.
Brain specimens were from individuals of European descent Australian and Japanese. Australian sample comprised 10 schizophrenic patients and 10 age- and gender-matched controls (Supplementary Table S1). The diagnosis of schizophrenia was made according to the Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV criteria (American Psychiatric Association 1994) by a psychiatrist and a senior psychologist. Control subjects had no known history of psychiatric illness. Tissue blocks were cut from gray matter in an area of the prefrontal cortex referred to as Brodmann's area 9 (BA9). Japanese samples of BA9 gray matter from Japanese brain specimens consisted of 6 schizophrenic patients and 11 age- and gender-matched controls (Supplementary Table S1). In addition, postmortem brains of 37 deceased Japanese patients with schizophrenia were also analyzed (Supplementary Table S1). The Japanese subjects met the DSM-III-R criteria for schizophrenia. The study was approved by the Ethics Committees of Central Sydney Area Health Service, University of Sydney, Niigata University, University of Tsukuba, Tokyo Metropolitan Matsuzawa Hospital, and Tokyo Institute of Psychiatry.
Total RNA was extracted from brain tissues with ISOGEN Reagent (Nippon Gene Co, Tokyo, Japan). The RNA quality was checked using a Nanodrop ND-1000 spectrophotometer (LMS, Tokyo, Japan) to have an OD 260/280 ratio of 1.8–2 and an OD 260/230 of 1.8 or greater. Microarrays were used to screen for differential gene expression between Australian schizophrenic patients and controls. In brief, 500 ng of total RNAs were reverse transcribed to synthesize first- and second-strand complementary DNA (cDNA), purified with spin columns, then in vitro transcripted to synthesize biotin-labeled complementary RNA (cRNA). A total of 1500 ng of biotin-labeled cRNA was hybridized on Sentrix® Human-6 Expression BeadChip (Illumina Inc., San Diego, CA) at 55°C for 18 h. The hybridized BeadChip was washed and labeled with streptavidin-Cy3, then scanned with an Illumina BeadStation 500 System (Illumina Inc). Scanned image was imported into BeadStudio (Illumina Inc) for analysis. Forty-six thousand transcripts can be analyzed by a single BeadChip.
Expression of the GSCL, HIRA, SEPT5, GNB1L, TBX1, and CDC45L genes was analyzed by TaqMan Real-time polymerase chain reaction (PCR) system (Applied Biosystems, Foster City, CA). From RNA, cDNA was synthesized with Revertra Ace (Toyobo, Tokyo, Japan) and oligo dT primer. Expression of these 6 genes was analyzed with an ABI PRISM 7900 HT Sequence Detection System (Applied Biosystems), with the TaqMan gene expression assays for GSCL (Hs00232019_m1), HIRA (Hs00983699_m1), SEPT5 (Hs00160237_m1), GNB1L (Hs00223722_m1), TBX1 (Hs00271949_m1), and CDC45L genes (Hs00185895_m1) and normalized to expression of Human GAPDH Control Reagents (Applied Biosystems). GNB1L expression was analyzed in Australian samples and replicated the analysis in Japanese subjects.
Protein was extracted from prefrontal cortex tissues with Laemmli Buffer. Western blotting method was used to compare GNB1L protein levels between schizophrenics and controls. Each of 2 μg protein was run on Pro-Pure™ SPRINT NEXT GEL (Amresco, Solon, OH) and transferrd to BioTrace™ PVDF (Nihon Pall Ltd, Tokyo, Japan). Polyclonal antibodies against the human GNB1L protein (OTTHUMP00000028644) were generated by injecting rabbits with the following peptide: CAGSKDQRISLWSLYPRA (MBL, Nagoya, Japan). Mouse polyclonal antibody against beta-actin (Sigma Aldrich Japan, Tokyo, Japan) was also used for normalization purpose. The bound primary antibodies were detected with goat anti-rabbit or anti-mouse IgG antibody HRP conjugate (MBL) and Immobilon™ Western, Chemiluminescent HRP Substrate (Millipore, Billerica, MA) on X-film (Fujifilm Medical, Tokyo, Japan). The signals of GNB1L or beta-actin of each subject on X-films were quantitated by computer software, ImageJ 1.40g (http://rsb.info.nih.gov/ij/), and GNB1L protein levels were normalized to beta-actin and compared.
The subjects comprised 1918 unrelated Japanese patients with schizophrenia (1055 men, 863 women; mean age ± standard deviation [SD], 48.9 ± 14.5 years) diagnosed according to DSM-IV with consensus from at least 2 experienced psychiatrists and 1909 mentally healthy unrelated Japanese control subjects (1012 men, 893 women; mean age ± SD, 49.0 ± 14.3 years) of whom the first- and second-degree relatives were free of psychosis as self-reported by the subjects. The association analysis was approved by the Ethics Committees of the University of Tsukuba, Niigata University, Fujita Health University, Nagoya University, Okayama University, and Teikyo University, National Center of Neurology and Psychiatry, University of Tokyo, and all participants provided written informed consent. DNAs were extracted from these blood samples and the same brain tissues used for gene expression analysis. The tag single-nucleotide polymorphisms (SNPs) comprising rs5746832, rs5746834, rs2269726, rs748806, rs29807124, rs5993835, rs13057609, rs4819523, rs2073765, rs7286924, rs10372, rs3788304, and rs11704083 at the GNB1L gene region were selected by Haploview program using HapMap Project Japanese data set (http://www.hapmap.org/), as the previously reported schizophrenia-associated SNPs, rs5746832 and rs2269726, were forced included. The TaqMan reaction was performed in a final volume of 3 μl consisting of 2.5 ng genomic DNA and Universal Master Mix (EUROGENTEC, Seraing, Belgium), and genotying was performed with an ABI PRISM 7900HT Sequence Detection System (Applied Biosystems).
Genotyping quality control consisted of ≥98% successful calls. We confirmed concordance among repeat genotyping in ≈10% of genotypes.
The correlations between GNB1L expression and 13 SNPs, rs5746832, rs5746834, rs2269726, rs748806, rs29807124, rs5993835, rs13057609, rs4819523, rs2073765, rs7286924, rs10372, rs3788304, and rs11704083, were examined in Australian and Japanese brain tissues, respectively.
Mice treated with haloperidol were studied to examine the effects of antipsychotic treatments on Gnb1l gene expression. Thirty-nine C57/BJ6 male mice (age, 8 weeks; weight, 20–25 g) were housed under 10 h/14 h light/dark conditions with normal food and water ad libitum, where groups of 5 or 6 mice were housed separately, and 0.5 mg/kg haloperidol or saline was injected intraperitoneally once each day for 4 weeks or for 50 weeks. The dosage of haloperidol was at maximum clinically used, and 4 or 50 weeks for treatment term correspond to several years or half a lifetime in human terms, respectively. We used extreme but likely condition to clear up the effect of the medication. We determined the dosage of haloperidol according to the previous studies.27–31 Mice were sacrificed 4 h after the last injection to obtain brain tissues.
The prefrontal cortex was taken, and RNA was extracted with RNeasy kit (Qiagen, K.K., Tokyo, Japan). A cDNA was synthesized with Revertra Ace (Toyobo) and oligo dT primer. Expression of Gnb1l was analyzed by TaqMan real-time polymerase chain reaction (PCR) with an ABI PRISM 7900 HT Sequence Detection System (Applied Biosystems), with the TaqMan gene expression assay for Gnb1l (Mm00499153_m1). Expression of Gnb1l was normalized to that of rodent Gapdh with Rodent Gapdh Control Reagents (Applied Biosystems).
All animal procedures were performed according to protocols approved by the Animal Care and Use committee of University of Tsukuba.
Microarray analysis was performed with GeneSpring software version 7.3.1 (Silicon Genetics, Redwood, CA). The mean background noise level was first corrected in each sample, and then per-chip normalization was applied to eliminate systematic differences between chips. Two-tailed Student's t-test was used to examine the difference between schizophrenic patients and controls. In real-time PCR experiments, GAPDH or Gapdh was used as an internal control, and measurement of threshold cycle (Ct) was performed in triplicate. Data were collected and analyzed with Sequence Detector Software version 2.1 (Applied Biosystems) and the standard curve method. Relative gene expression was calculated as the ratio of expression of the target gene to the internal control (GAPDH or Gapdh). Correlations of GNB1L gene expressions and 2 quality parameters, postmortem interval (PMI) and pH, of brain samples were analyzed with analysis of variance (ANOVA) one-way tests by JMP computer software version 5.1. The density of images reflecting GNB1L protein levels was also compared between schizophrenics and controls with the Wilcoxon test implemented in JMP computer software version 5.1. Deviation from Hardy-Weinberg equilibrium (HWE), allelic associations, and linkage disequilibrium (LD) between SNPs were evaluated with Haploview software version 3.11. A nominal association was defined when the given P value for allelic or genotypic tests was less than 5% (uncorrected P < .05). If a nominal significant association was found in the analysis, permutation test was also performed with Haploview software version 3.11. Correlations of GNB1L gene expressions and either protein expression or genotypes of the tag SNPs were analyzed with ANOVA one-way tests by JMP computer software version 5.1.
Human-6 Expression BeadChip demonstrated that GSCL (GI_48885362-S) and TBX1 (GI_18104949-I) of 28 genes located in the 22q11DS region were expressed at lower levels in schizophrenic brains than in the control brains in the Australian samples (P < .05) (Supplementary Table S2). However, the signals of these transcripts were low, and reliable confidence was not obtained from any subject. Expression of CDC45L (GI_34335230-S) tended to be lower in schizophrenic brains than in control brains (P = .07). Data of GNB1L were not available in this platform (Supplementary Table S2).
We used real-time PCR experiments to evaluate expression of the 3 genes that were potentially underexpressed in schizophrenia prefrontal cortex by microarray and GNB1L, which was not assessed by the microarray in the Australian and Japanese brain samples. The difference in gene expression between the schizophrenia and control groups was not confirmed for CDC45L. In addition, because the reliability of HIRA and SEPT5 was not sufficient due to weakly expressed sequences in the array screening, we reexamined expression levels of these genes by real-time PCR method and did not find significant differences in gene expression between the schizophrenia and control groups. Expressions of TBX1 and GSCL were too low to obtain reliable signals with the TaqMan gene expression assay (Hs00271949_m1 and Hs00232019_m1, respectively). Relative expression of GNB1L was significantly lower in Australian schizophrenic prefrontal brains than in Australian control brains (average ratio = 0.57, P < .001) and in Japanese patients with schizophrenia than in control subjects (average ratio = 0.53, P < .0001) (figure 1A). No difference in GNB1L expression was observed between the Japanese and Australian schizophrenic patient groups (data not shown). GNB1L expression was not significantly correlated with pH of the brain tissue samples overall (figure 2), neither with gender (P = .62) nor PMI (F = 0.61, P = .44). Western blotting analysis also demonstrated the lower levels of GNB1L protein in brains of the schizophrenia sample than in those of the control sample from each ethnic group (approximate average ratio = 0.75, P = .027 in Australian sample and approximate average ratio = 0.69, P = .033 in Japanese sample) (figure 1B). There is a significant correlation between gene and protein expression observed in our samples (F = 4.7, P = .037).
There were no significant associations of tag SNPs at the GNB1L gene studied in the present study with schizophrenia in our Japanese case-control sample (table 1). Also no significant differences were found in distributions of homozygotes and heterozygotes between schizophrenics and controls (table 1). Williams et al26 reported male-specific associations of rs5746832 and rs2269726 with schizophrenia and correlation between those markers and the gene expression. However, such male-specific associations of rs5746832 and rs2269726 were not observed in our sample (table 1).
There was a nominally significant correlation between rs5748832 and GNB1L expressions in whole subjects (P = .014) and in Japanese (P = .028), but not in Australian (P = .66) (table 2). An allele of rs5748832 is correlated with high GNB1L expression in this study, while the previous study showed the opposite direction of correlation.26
Significant deviation from HWE in the genotypic distributions was observed at rs4819523 in the control group. Lower proportions of heterozygotes than those expected by HWE seemed to cause these deviations. Although genotype errors, chance findings, or actual structural variations in some subjects might have potentially caused these deviations, we could not determine which was most likely to cause these HWE deviations.
Gnb1l expression in mice was examined to exclude the possibility that reduced GNB1L expression was the effects of chronic treatment with antipsychotic drugs. The patients whose brains were examined in the present study had received long-term medication of typical antipsychotic drugs; therefore, we chose haloperidol as a representative antipsychotic drug. As a result, while Gnb1l gene expression in prefrontal cortex of mice treated with haloperidol for 4 weeks was not changed, the expression was higher in those treated with haloperidol for 50 weeks than in those with saline injected (P = .02) as shown in figure 3.
In the present study, we hypothesized that haploinsufficiency of some genes in the 22q11DS region might increase the susceptibility to schizophrenia not only in patients with 22q11DS but also in the those without 22q11DS and that such genes would be expressed at lower levels in the brains of schizophrenic patients than in control subjects. GNB1L appears to meet this hypothesis. Reduced GNB1L gene expression was detected in both mRNA and protein levels in Australian and Japanese subjects, suggesting that lower GNB1L gene expression produces lower GNB1L protein levels which underlie schizophrenia across ethnicities. Treatment of mice with haloperidol indicated that the reduction of GNB1L expression is not likely a consequence of antipsychotic medication treatment, though the possibility of reduction of GNB1L expression by other antipsychotic drugs remains. The present study did not provide evidence of whether TBX1 expression is altered significantly in schizophrenic brains because the signals detected by Illumina's Sentrix® Human-6 Expression BeadChip or TaqMan assay were very weak. Paylor et al9 mapped PPI deficits in a panel of mouse mutants and found that PPI was impaired by either haploinsufficiency of Tbx1 or Gnb1l. The present study of human brains confirms that GNB1L is an important candidate for susceptibility to schizophrenia.
There is little information about the function of GNB1L. GNB1L expression is relatively low in adult brain but is high in fetal brain. GNB1L encodes a guanine nucleotide–binding protein (G protein), beta polypeptide 1–like, which is a member of the WD repeat protein family. WD repeats are minimally conserved regions of approximately 40 amino acids typically bracketed by Gly-His and Trp-Asp (GH-WD) that may facilitate formation of heterotrimeric or multiprotein complexes. Members of this family are involved in a variety of cellular processes, including cell cycle progression, signal transduction, apoptosis, and gene regulation. GNB1L contains 6 WD repeats.32 GNB1L shows homology to the human guanine nucleotide–binding protein β subunit (GNB1). GNB1 functions in G-protein–coupled receptor protein signaling pathways and intracellular signaling cascade.
Williams et al26 reported excess homozygosity at rs5746832 and rs2269726 in male schizophrenia subjects and that the markers associated with male schizophrenia were related with cis-acting changes in GNB1L expression. Firstly in the present study, we failed to confirm the association in our Japanese case-control population. Secondly, we found a nominally significant correlation between rs5746832 and GNB1L expression in the Japanese brain samples, but failed to find it in our limited number of the Australian samples. Further, the association between allele and gene expression in our Japanese samples was in the opposite direction from that reported in the Caucasian samples. It might be due to possible differences in LD block between haplotype phases across rs5746832 and harboring potential cis-acting variations of the gene between 2 ethnic populations. Even if such cis-acting variations are present, diagnosis has tremendous effect on the gene expression, in comparison to that of the SNP. The power of the present study to replicate the findings of excess homozygosity in male subjects is greater than 90% assuming the odd ratio of greater than 1.5 found in UK populations by Williams et al.26 However, if the odd ratio assumes 1.3 observed in a German population by them, the power drops to 0.65. Although the gene frequencies of rs5746832 and rs2269726 were significantly different between Caucasian and Japanese populations, the frequencies of homozygotes were almost the same between 2 populations. Because of small sample size, we did not attempt allele-specific expression analysis in our brain sample. Therefore, we could not conclude whether lower GNB1L gene/protein expression in schizophrenia was due to cis-acting differences by genetic polymorphisms in this locus or not in this study.
The present study showed that reduced expression of GNB1L may be involved in the pathophysiology of schizophrenia; however, it does not exclude the possibility that other genes in the 22q11DS region contribute to the susceptibility to schizophrenia. The array used in the present study did not examine all isoforms of the genes in the 22q11DS region. In addition, the reliability of weakly expressed sequences in the array screening is not sufficient. Therefore, we reexamined expression levels of the genes, which reliable data (greater than 0.96 confidence) was produced by the array in no subjects, by real-time PCR method. The study is also limited by the areas and ages of the brains examined. We examined only adult postmortem prefrontal cortex. Differential gene expression in other brain regions or during other developmental stages may also influence the susceptibility to schizophrenia.
The consortium data of the Stanley Medical Research Institute showed no significant differences (P > .05) in the following gene expression levels in postmortem prefrontal cortex between patients with schizophrenia and controls: DGCR6, PRODH, DGCR2, STK22B, DGCR14, CLTCL1, CLTCL1, HIRA, UFD1L, CDC45L, CLDN5, TBX1, FLJ21125, TXNRD2, COMT, ARVCF, DKFZp761P1121, DGCR8, HTF9C, RANBP1, and ZDHHC8. The expression of RTN4R might be potentially reduced (P = .02). No data were available for GSCL, MRPL40, SEPT5, GP1BB, and GNB1L (http://www.stanleyresearch.org/brain/menu.asp).
A trans-acting effect on expression of the disease gene may also be expected to modulate disease susceptibility. Large-scale studies in humans have indicated that a significant proportion of the heritable variance in gene expression is attributable to trans-acting polymorphism.33,34 As one of the examples, recent study reported that microRNAs regulate gene expression posttranscriptionally.35 Even for schizophrenia, Bray et al36 indicated that the reduction in DTNBP1 expression in schizophrenia is likely to result in part from trans-acting risk factors. Such trans-acting factors that regulate GNB1L gene expression, however, have not been identified.
In conclusion, the present study further supports the role of GNB1L in the pathophysiology of schizophrenia.
Grant-in-Aid for Scientific Research on Priority Areas; Research on Pathomechanisms of Brain Disorders from the Ministry of Education, Culture, Sports, Science and Technology of Japan (20390098 and 20023006); Japan Science and Technology.
Australian human brain tissues were received from the NSW Tissue Resource Centre, which is supported by The University of Sydney, Neuroscience Institute of Schizophrenia and Allied Disorders National Institute of Alcohol Abuse and Alcoholism and NSW Department of Health.