Biowing.com.cn

European Journal of Human Genetics (2005) 13, 807–814& 2005 Nature Publishing Group All rights reserved 1018-4813/05 $30.00 An association study of the N-methyl-D-aspartatereceptor NR1 subunit gene (GRIN1) and NR2Bsubunit gene (GRIN2B) in schizophrenia withuniversal DNA microarray Shengying Qin1,2,6, Xu Zhao1,2,6, Yuxi Pan1,2, Jianhua Liu3, Guoyin Feng4, Jingchun Fu5,Jiying Bao5, Zhizhou Zhang1,2 and Lin He*,1,2 1Bio-X Life Science Research Center, Shanghai Jiao Tong University, Shanghai 230030, PR China; 2Institute forNutritional Sciences, SIBS, Chinese Academy of Sciences, 320 Yue Yang Road, Shanghai 200031, PR China; 3School ofLife Science & Biotechnology, Shanghai Jiao Tong University, Shanghai, PR China; 4Shanghai Mental Health Center,Shanghai, PR China; 5Inner Mongolia Mental Health Center, Huhehaote, PR China Dysfunction of the N-methyl-D-aspartate (NMDA) receptors has been implicated in the etiology ofschizophrenia based on psychotomimetic properties of several antagonists and on observation of geneticanimal models. To conduct association analysis of the NMDA receptors in the Chinese population, weexamined 16 reported SNPs across the NMDA receptor NR1 subunit gene (GRIN1) and NR2B subunit gene(GRIN2B), five of which were identified in the Chinese population. In this study, we combined universalDNA microarray and ligase detection reaction (LDR) for the purposes of association analysis, an approachwe considered to be highly specific as well as offering a potentially high throughput of SNP genotyping.
The association study was performed using 253 Chinese patients with schizophrenia and 140 Chinesecontrol subjects. No significant frequency differences were found in the analysis of the alleles but somewere found in the haplotypes of the GRIN2B gene. The interactions between the GRIN1 and GRIN2B geneswere evaluated using the multifactor-dimensionality reduction (MDR) method, which showed a significantgenetic interaction between the G1001C in the GRIN1 gene and the T4197C and T5988C polymorphisms inthe GRIN2B gene. These findings suggest that the combined effects of the polymorphisms in the GRIN1 andGRIN2B genes might be involved in the etiology of schizophrenia.
European Journal of Human Genetics (2005) 13, 807 – 814. doi:10.1038/sj.ejhg.5201418Published online 20 April 2005 Keywords: schizophrenia; universal DNA microarray; multifactor-dimensionality reduction (MDR) transmission, plasticity and excitoxicity. Functional NMDA As one type of the glutamate receptors in the glutamatergic receptors are mainly composed of an NMDA receptor NR1 excitatory neurotransmitter system, N-methyl-D-aspartate subunit (GRIN1) and one of the four NMDA receptor NR2 (NMDA) receptors play crucial roles in excitatory synaptic subunits (GRIN2A, GRIN2B, GRIN2C and GRIN2D).1 – 3 Forseveral years, there has been considerable interest in thepossible role of the NMDA receptors in schizophrenia *Correspondence: Dr L He, Bio-X Life Science Research Center, ShanghaiJiao Tong University, Hao Ran Building, PO Box 501, Shanghai 200030, PR following the discovery that several noncompetitive China. Tel: þ 86 21 64717740; Fax: þ 86 21 62822491; antagonists of the NMDA receptors, such as phencyclidine (PCP), ketamine and MK-801, were found to cause 6These authors contributed equally to this work.
psychotic conditions mimicking schizophrenia in humans Received 30 August 2004; revised 20 January 2005; accepted 3 March2005 and laboratory animal models.4 – 8 The involvement of the Association study in schizophrenia with microarray NMDA receptors is also supported by genetic animal SD ¼ 12.05) served as controls. The participants recruited in models: for instance, knockdown mice expressing only this study all provided written informed consent.
5% of normal levels of the essential GRIN1 subunit surviveto adulthood but display behavioral abnormalities, such as elevated motor activity and stereotypy and deficiencies in social and sexual interactions, which can be ameliorated by We examined 11 SNPs (C112T, C113T, G716A, A750G, treatment with haloperidol or clozapine.9 – 15 G1001C, A290G, G301A, G308A, A1970G, G2108A and Mutation screening and association analysis for the G6435A) in the GRIN1 gene and five SNPs (G366C, NMDA receptors have been performed by several groups C2664T, C3538T, T4197C and T5988C) in the GRIN2B focusing mainly on the GRIN1 and GRIN2B genes.16 – 23 The gene, which had been identified in previous studies.16,18,20 association of the variants in GRIN1 and GRIN2B with The primer sets were designed to amplify the SNP-contain- schizophrenia has been established in several reports,18,19 ing regions. The NCBI number of the SNP sites, corre- but not replicated in others.16,22 Until now, GRIN1 and sponding primer sequences and the sizes of the amplicons GRIN2B have always been studied individually. However, are presented in Supplementary material 1. The PCRs were the functional NMDA receptors are heterodimer complexes carried out on the PTC-100 DNA engine (MJ Research, USA) that involve gene – gene interactions arising from interac- in a total volume of 15 ml containing 10 ng of genomic tions between receptor subunits and probably have a DNA, 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 – 3.0 mM synergistic effect on the pathogenesis of schizophrenia.
MgCl2, 200 mM dNTP, 1 mM of each primer and 0.25 U Taq There has been accumulating evidence supporting the DNA polymerase. The cycling protocol consisted of hypothesis that gene to gene interactions play an im- denaturation at 951C for 1 min, followed by 30 – 35 cycles portant role in susceptibility to common human dis- at 951C for 30 s, 50 – 651C for 1 min, 721C for 1 min, and a eases.24 In the light of this evidence, the present study final extension at 721C for 10 min. Subsequently, the PCR has focused on the GRIN1 and GRIN2B genes as a pilot products were sequenced for SNP examination using ABI PRISM 3100 DNA Sequencer (Applied Biosystems, USA).
DNA microarray technology has been advocated as one For each SNP, 40 subjects were examined in both sense and of the most powerful approaches for highly parallel, large- scale SNP analysis. Several approaches for SNP genotypingwith DNA microarray have been designed.25 – 29 Recently, the universal DNA microarray method mediated by ligase In the universal microarray approach, five probes including detection reaction (LDR) has been demonstrated to be a two discriminating probes, one common probe and two highly specific and sensitive assay for SNP detection.29 – 32 corresponding zip-codes, which are coupled to the slides at We combined this approach with the slide self-activating known locations, were designed for each SNP as described method28,33 to establish a high-discrimination and cost- in a previous study.29 In addition, the positive and negative effective technology with significant promise for future marker probes were designed for result evaluation. The designed probes are summarized in Supplementary materi- In the present study, we first examined the SNPs al 1. The discriminating probe contains a zip-code identified in previous studies across the GRIN1 and GRIN2B complement sequence on the 50-end and the discriminat- genes in the Chinese population, and then performed SNP ing base on the 30-end. The common probe is phosphory- genotyping using the universal microarray method to lated on the 50-end and contains a fluorescent label (Cy3) evaluate the single locus and haplotype associations. The on the 30-end. The two probes are hybridized adjacent to gene – gene interactions were examined with the multi- one another on the template DNA and the nick between factor-dimensionality reduction method (MDR), which can the two probes can be sealed by the ligase only if there is identify evidence for high-order gene to gene interactions perfect complementarity at the junction. The ligated in the absence of any statistically significant independent products are hybridized to the zip-codes array. Only the main effects with relatively small samples.34 – 36 spots corresponding to the SNP type are seen to givestronger fluorescent signals (Figure 1).
The standard microscope glass slides were first pre- cleaned with 1 M NaOH overnight and washed several times with distilled water. After a neutralization step in 1% HCl and an extensive washing step in distilled water, the All subjects were Han Chinese in origin. The 253 patients pretreated glass slides were immersed in 2% 3-aminopro- with schizophrenia (mean age of onset 25.45 years, mean pyl-trimethoxysilane (Sigma, USA) for 15 min followed by age at present 38.43 years, SD ¼ 10.9) were diagnosed five washing steps in acetone. Subsequently, slides were according to DSM-IV by two independent clinicians. The baked at 1101C for 45 min, and then the slide surface was 140 unrelated healthy Han Chinese (mean age 43.37 years, activated with 1,4-phenylene diisothiocyanate (Sigma, Association study in schizophrenia with microarrayS Qin et al Strategy for the universal microarray mediated by ligase detection reaction. The discriminating probe containing zip-code complement sequence (czipa or czipb) and the common probe containing a fluorescent label can be ligated only if there is perfect complementarity at the junction.
When hybridized to the universal array containing the zip-codes (zipa and zipb), the correspondent spots can be seen to give a stronger signal.
USA) according to Guo et al33 and Beier and Hoheisel.37 librium was measured between each of two locus and D0 The 50-terminal amino-modified zip-codes were dissolved was the normalized linkage disequilibrim statistics.38 in a 400 mM sodium carbonate buffer (pH 9.0) to a final Differences in allelic and separate haplotypic distributions concentration of 20 mM and spotted using a GMS 417 were estimated by w2 test using EpiInfo 6.0 and the phasing Arrayer (Genetic MicroSystems, USA) under a designed of case and control haplotype was performed separately.
program. The slides were deactivated in 10% ammonium The limit of significance was set to 0.05. The odds ratios hydroxide solution. Finally, they were rinsed five times in (OR) and their 95% confidence intervals were estimated by distilled water and centrifuged for 1 min at 1000 r.p.m.
EpiInfo 6.0 for the effects of alleles. Difference in overall The target DNA sequences were amplified using a haplotype distribution was assessed by w2 test using multiplex PCR method. After the completion of the CLUMP1.639 with 300 000 simulations.
amplification, 1 ml of Proteinase K (20 mg/ml) was added, The gene – gene interactions were analyzed by MDR then heated at 701C for 10 min and quenched at 941C for software. There are six general steps involved in imple- 15 min. The ligation reaction for each subject was carried menting the MDR method for case – control study. The first out in a final volume of 20 ml containing 20 mM Tris-HCl step of MDR involves partitioning the data into some (pH 7.6), 25 mM potassium acetate, 10 mM magnesium number of equal parts for cross-validation. With 10-fold acetate, 10 mM DTT, 1 mM NAD, 0.1% Triton X-100, 10 ml of cross-validation in our study, the data are randomly Multi-PCR product, 1 pmol of each discriminating oligo, divided into 10 equal parts, and the MDR model is 1 pmol of each common probe and 0.5 ml of 40 U/ml Taq developed on 9/10 of the data (training set) and then DNA ligase (New England Biolabs, USA). The LDR was tested on 1/10 of the remaining data (testing set). In step 2, performed with 40 cycles of 941C for 30 s and 631C for a set of n genetic factors is selected from the pool of all factors. In step 3, the n factors and their multifactor classes Following the LDR reaction, the solution was diluted to or cells are represented in n-dimensional space. For obtain 65 ml of hybridization mixture containing 5 Â SSC example, for two polymorphisms, each with three geno- and 0.1 mg/ml salmon sperm DNA. The mixture was then types, there are nine two-locus genotype combinations.
quenched at 941C for 2 min and chilled on ice immedi- Then, the ratio of the number of cases to the number of ately. The area of the slide containing the oligo array was controls is evaluated within each multifactor cell. In step 4, sealed with Gene Frame slide chambers (ABgene, UK) and each multifactor cell in n-dimensional space is labelled as the mixture was then introduced. Hybridization was high risk if the ratio of cases to controls meets or exceeds carried out at 651C for 3 h. After removal of the chamber, the threshold (T) of 1.0 and low risk if the threshold is not the slide was washed for 2 min in 1 Â SSC, 0.1% SDS and exceeded. In one recent report, it has been observed in then for 1 min in 0.05 Â SSC by gently shaking at room simulation studies that a threshold of 1.0 is optimal at this temperature. Finally, the slide was spun at 1000 r.p.m. for stage of MDR modelling for unbalanced data sets.40 In this 4 min and imaged on a GMS418 scanner (Genetic Micro- way, a model for cases and controls is formed by pooling Systems, USA). 10 mm resolution of 16-bit TIFF images was those cells labelled high risk into one group and those cells analyzed using the Genepix 2.0 software (Axon Instru- labelled low risk into another group. This reduces the n- ments, USA). The average signals after subtraction of local dimensional model to one dimension. In step 5, all background were used for calculating the signal ratios of possible combinations of n factors are evaluated sequen- spot pairs corresponding to different alleles.
tially for their ability to classify affected and unaffectedindividuals in the training set and the best n-factor model is selected. In step 6, the independent test set from the Haplotypes were generated using the program Arlequin cross-validation is used to estimate the prediction error of (http://anthropologie.unige.ch/arlequin). Haplotype fre- the best model selected in step five. The prediction error is quencies, linkage disequilibrium and Hardy – Weinberg the proportion of subjects for whom an incorrect predic- equilibrium were performed by Arlequin. Linkage disequi- tion was made. Steps 1 to 6 are repeated 10 times with the Association study in schizophrenia with microarray data split into 10 different training and testing sets. The that were difficult to sequence could be genotyped easily number of times a particular set of factors is identified in by the microarray procedure. The DNA targets quality each possible 9/10 of the subjects is defined as cross- required in this microarray procedure was not so strict as in the sequencing method. As shown in Figure 2, the same Once MDR identifies the best combination of factors, the quality of the DNA target, which was genotyped only final step is to determine which multifactor levels (eg approximately by sequencing, could be discriminated genotypes) are high risk and which are low risk using the clearly by the microarray procedure.
entire data set. This final evaluation is conducted with a In the present study, 16 reported polymorphisms (11 in threshold ratio that is determined by the ratio of the the GRIN1 gene and 5 in the GRIN2B gene) were selected number of affected individuals in the dataset divided by for examination in the Chinese population containing 20 the number of unaffected individuals in the dataset. Theprevious user-defined threshold ratio (T) is not used in thisfinal evaluation. Also, the MDR program can display thenumber of cases and controls for each multilocus –genotype combination. To make it clear, we showed thisas the percentage of it in the total subjects. Since the MDRprogram randomly shuffles the order of the observations inthe dataset using a random seed supplied by the user beforedividing the data into a training set and a testing set, weran the analysis 10 times using 10 different randomnumber seeds and the results were averaged in order toavoid spurious results due to chance divisions of the data.
This MDR procedure can be carried out for each possible model size. In this study, we considered two-locus interac-tions through four-locus interactions. Single best modelsare selected from among each of the two-factor, three-factor and four-factor combinations. Among this set of bestmultifactor models, the combination of factors thatminimizes the prediction error and maximizes the cross-validation consistency is selected as the final best one.
Finally, the statistical significance of the result wasdetermined by comparing the average prediction errorfrom the observed data with the distribution of averageprediction errors under the null hypothesis of no associa-tions derived empirically from 1000 permutations. Ourpermutation testing strategy was to randomize the case andcontrol labels in the original dataset 1000 times to create aset of permuted datasets. MDR can then be run on eachpermuted data set and the maximum cross-validationconsistency and minimum prediction error identified foreach were saved and used to create an empirical distribu-tion for estimation of a P-value The null hypothesis wasrejected when the upper-tail Monte Carlo P-value derivedfrom the permutation test was p0.05.
Characteristics evaluation of the DNA microarray procedure. Each column of the array corresponds to a polymorphism After the genotyping procedure, it was possible to validate allele and contains five-spot replicates. Deposition scheme: 1 – 1001G;2 – 1001C; 3 – 366G; 4 – 366C; 5 – 2664C; 6 – 2664T; 7 – 4197T; 8 – differences in sensitivity and specificity by direct sequen- 4197C; 9 – 5988T; 10 – 5988C; 11 – negative marker; 12 – positive cing. We randomly selected 40 samples to be genotyped marker. (a), (b) and (c) are the scanning results that, respectively, using both methods. The result showed that the two indicate that the five polymorphisms are all heterozygous, homo- methods were completely consistent. It is clear that this geneous and miscellaneous. (d) and (e) are the genotyping resultcomparisons between the microarrray and the sequencing method microarray procedure demonstrates high specificity in with the same quality template. The marked part of the picture is the relation to genotype (Figure 2). Moreover, some regions genotyping result of the C2664T locus of the same sample.
Association study in schizophrenia with microarrayS Qin et al Allele distribution of the five polymorphisms in the GRIN1 and GRIN2B genes aThe number of schizophrenics is 253.
patients and 20 controls. One variant in the GRIN1 gene Estimation of linkage disequilibrium of the four (G1001C) and four variants in the GRIN2B gene (G366C, C2664T, T4197C and T5988C) were observed. Three of them (G366C, C2664T and T4197C) were in coding sequence, but were synonymous mutations. The fiveidentified SNPs were genotyped in 253 Chinese patients with schizophrenia and 140 Chinese control subjects. The distribution of these five SNPs was in Hardy – Weinberg equilibrium (data not shown). The data for allele frequen- cies are shown in Table 1. No significant differences in the For each pair of SNPs, the standardized D0 is shown. D0 values X0.7 allele frequencies were observed between the patients andcontrols. Linkage disequilibrium between each pair ofGRIN2B SNPs is shown in Table 2. Three of the alleles instrong LD (C2664T, T4197C and T5988C) were used to the null hypothesis of no association, it is highly unlikely generate haplotypes. The haplotype frequencies of the that a prediction error X30.772% will be observed for the GRIN2B gene are shown in Table 3. We found that the two-locus model and a prediction error X30.754% will be frequency of the CCT haplotype (the C allele of the observed for the three-locus model. As the MDR procedure C2664T, the C allele of the T4197C, and the T allele of the prefers the best model that has the minimum prediction T5988C) was higher in controls (0.089) than in patients error, the three-locus model, which included the G1001C (0.022) (P ¼ 0.000146 after correction by the number of polymorphism in the GRIN1 gene and the T4197C and tests performed, w2 ¼ 18.79, OR ¼ 0.23). The difference of T5899C polymorphisms in the GRIN2B gene, was regarded CCC was nominally significant estimated by the w2 test as the best model. Table 5 summarizes the three-locus – (P ¼ 0.02, w2 ¼ 5.40, OR ¼ 1.54) using EpiInfo 6.0 and was genotype combinations associated with high risk and with not significant after the correction by the number of tests low risk, along with the corresponding distribution of cases performed (P ¼ 0.2). The overall haplotype distribution was and of controls, for each multilocus – genotype combina- (P ¼ 0.000031, w2 ¼ 23.60). With the MDR method, thegene – gene interactions were examined, and the results foreach number of factors considered are summarized inTable 4. All the models had the same cross-validation consistency of 10.0 and the permutation testing showed The microarray procedure used in this study was demon- that the prediction errors of the two-locus and three-locus strated to be reliable by comparing with sequencing models are both significant at the 0.001 level. Thus, under method in the examination of 40 subjects. The universal Association study in schizophrenia with microarray Frequencies of estimated haplotypes composed of the three polymorphisms of the GRIN2B gene aThe haplotype frequencies defined by C2664T, T4197C and T5988C were determined. Only haplotypes with a frequency greater than 2% are listed.
bValues in parentheses are P-values corrected by the number of tests performed. P ¼ 0.000031 (w2 ¼ 23.60) for difference in overall haplotypedistribution.
Results of the MDR analysis of the dataset SNPs included in the best candidate model *Po0.001 based on 1000 permutations.
Distribution of high-risk and low-risk genotypes in the best three-locus model microarray platform can be used for any group of SNPs and selection acting against molecular diversity during human only the LDR primers require to be designed. Compared evolution. Of all the identified variants in both GRIN1 and with other gene-specific arrays, the microarray procedure GRIN2B genes, three are located in protein coding gives researchers freedom to detect specified polymorphic sequences, but only as synonymous mutations. Moreover, loci based on the same universal array and avoids the need one polymorphism G1001C seemed to have functional to optimize the operation condition each time. Further- significance. It is located in the promoter region of GRIN1 more, this approach can readily distinguish the poly- and could alter a consensus sequence for the p50 subunit of morphism types of insertions and deletions in repeat the transcription factor NF-kB.19 The association analysis of G1001C polymorphism in the GRIN1 gene has been On examination of the SNPs, only one of the 11 reported done by several groups,19,21 – 23 but only Begni et al19 SNPs existed in the GRIN1 gene in our sample, which is detected a significant result in the Italian population. In perhaps due to ethnic differences. The lower rate of the previous association study for the four polymorphisms nonsynonymous SNPs in this gene, which has also been (G366C, C2664T, T4197C and T5988C) in the GRIN2B reported by other groups,16,17 may be mainly due to gene, Ohtsuki et al18 observed a significant difference in Association study in schizophrenia with microarrayS Qin et al the allele frequency of the G366C polymorphism between interaction between the G1001C in the GRIN1 gene and patients and control individuals in the Japanese popula- the T4197C and T5988C polymorphisms in the GRIN2B tion. Also, their haplotype analysis showed a borderline gene. To our knowledge, this is the first report of the significant result for the overall haplotype distribution.18 interaction between the two subunit genes of the NMDA However, the significant result of G366C was not replicated receptors associated with schizophrenia.
in the analysis on a sample of Caucasian patients.16 In our In conclusion, using the universal microarray procedure study, no significant frequency differences of the five mediated by LDR combined with effective MDR statistical alleles (G1001C, G366C, C2664T, T4197C, and T5988C) analysis, our study provided the possibility that the between schizophrenic patients and controls were ob- combinational effects of the polymorphisms in the GRIN1 served. However, in the low-frequency haplotype (CCT) and GRIN2B genes are associated with schizophrenia in the and the overall haplotype distribution of the GRIN2B gene, Chinese population. However, as case – control studies are we did find evidence of association with schizophrenia.
susceptible to positive and negative artefacts from un- Moreover, all the three SNPs contained in the haplotype of known population stratification, it is necessary to validate our study are also included in the significant haplotype of or replicate our association results in other ethnic groups.
the previous study by Ohtsuki et al, which furthersupported possible association of the haplotype in GRIN2Bgene with schizophrenia. Since this association has been found in a relatively small sample size (253 cases and 140 This work was supported by grants from the national 973 and 863 controls) involving a low-frequency haplotype, it is programs, the National Natural Science Foundation of China, the susceptible to a false-positive result and further study with Shanghai Municipal Commission for Science and Technology, and theKey Project of the Ministry of Education (No. 03066). We are grateful to the Office of Technology Transfer & Enterprise Development of the There is a growing awareness that the complex disorders Vanderbilt University for providing the MDR software and to Dr Lance may be genetically attributable to complex interactions among multiple genes. Many studies have observed thiscombination effect in a number of complex disorders.34,40 –44 The two subunits of the NMDA receptors have different functions. The GRIN1 subunit possesses the characteristic 1 Hollmann M, Heinemann S: Cloned glutamate receptors. Annu ion channel properties and the specificity of the NMDA 2 Nakanishi S, Masu M: Molecular diversity and functions of receptors arises from the GRIN2 subunits, which display glutamate receptors. Annu Rev Biophys Biomol Struct 1994; 23: different regional and temporal expression patterns.1 – 3 We hypothesized that gene – gene interactions across the two 3 Moriyoshi K, Masu M, Ishii T, Shigemoto R, Mizuno N, Nakanishi S: Molecular cloning and characterization of the rat NMDA subunit genes probably contributes to the pathogenesis of receptor. Nature 1991; 354: 31 – 37.
schizophrenia although the specific SNPs alone do not 4 Luby ED, Cohen BD, Rosenbaum G, Gottlieb JS, Kelley R: Study of show any association with the disorder. Using the MDR a new schizophrenomimetic drug; sernyl. AMA Arch Neurol method, we were able to detect interactions in the absence 5 Javitt DC, Zukin SR: Recent advances in the phencyclidine model of main effects, where logistic regression and other of schizophrenia. Am J Psychiatry 1991; 148: 1301 – 1308.
approaches may not be able to do so.34 Application of 6 Ellison G: The N-methyl-D-aspartate antagonists phencyclidine, MDR to case – control data sets has routinely yielded ketamine and dizocilpine as both behavioral and anatomicalmodels of the dementias. Brain Res Brain Res Rev 1995; 20: 250 – statistically significant interactions in various common human diseases, such as sporadic breast cancer, essential 7 Malhotra AK, Pinals DA, Adler CM et al: Ketamine-induced hypertension, atrial fibrillation and type II diabetes exacerbation of psychotic symptoms and cognitive impairment mellitus.34,40,43,44 In our MDR analysis, although the prediction errors of the two-locus and three-locus models 8 Abi-Saab WM, D’Souza DC, Moghaddam B, Krystal JH: The had the same level of significance, the three-locus model NMDA antagonist model for schizophrenia: promise and pitfalls.
had a lower prediction error (30.754%) than that in the Pharmacopsychiatry 1998; 31 (Suppl 2): 104 – 109.
two-locus model. Furthermore, the three loci in the three- 9 Li Y, Erzurumlu RS, Chen C, Jhaveri S, Tonegawa S: Whisker- related neuronal patterns fail to develop in the trigeminal locus model were located within the two different subunit brainstem nuclei of NMDAR1 knockout mice. Cell 1994; 76: genes, which may display the gene to gene interactions associated with schizophrenia and the two-locus model are 10 Forrest D, Yuzaki M, Soares HD et al: Targeted disruption of NMDA receptor 1 gene abolishes NMDA response and results in included in the three-locus model. Thus, we selected the neonatal death. Neuron 1994; 13: 325 – 338.
three-locus model as the best model. However, the MDR 11 Ikeda K, Araki K, Takayama C et al: Reduced spontaneous activity program could not statistically compare the two models to of mice defective in the epsilon 4 subunit of the NMDA receptor confirm that the three-locus model is more significantly channel. Brain Res Mol Brain Res 1995; 33: 61 – 71.
12 Sakimura K, Kutsuwada T, Ito I et al: Reduced hippocampal LTP associated than the two-locus model, which is a limitation and spatial learning in mice lacking NMDA receptor epsilon 1 of the present study. The best model showed a genetic subunit. Nature 1995; 373: 151 – 155.
Association study in schizophrenia with microarray 13 Ebralidze AK, Rossi DJ, Tonegawa S, Slater NT: Modification of primer elongation reaction on oligonucleotide microarrays.
NMDA receptor channels and synaptic transmission by targeted disruption of the NR2C gene. J Neurosci 1996; 16: 5014 – 5025.
29 Gerry NP, Witowski NE, Day J, Hammer RP, Barany G, Barany F: 14 Kutsuwada T, Sakimura K, Manabe T et al: Impairment of suckling Universal DNA microarray method for multiplex detection of low response, trigeminal neuronal pattern formation, and hippocam- abundance point mutations. J Mol Biol 1999; 292: 251 – 262.
pal LTD in NMDA receptor epsilon 2 subunit mutant mice.
30 Consolandi C, Busti E, Pera C et al: Detection of HLA polymorphisms by ligase detection reaction and a universal array 15 Das S, Sasaki YF, Rothe T et al: Increased NMDA current and spine format: a pilot study for low resolution genotyping. Hum density in mice lacking the NMDA receptor subunit NR3A. Nature 31 Favis R, Day JP, Gerry NP, Phelan C, Narod S, Barany F: Universal 16 Williams NM, Bowen T, Spurlock G et al: Determination of the DNA array detection of small insertions and deletions in BRCA1 genomic structure and mutation screening in schizophrenic and BRCA2. Nat Biotechnol 2000; 18: 561 – 564.
individuals for five subunits of the N-methyl-D-aspartate gluta- 32 Busti E, Bordoni R, Castiglioni B et al: Bacterial discrimination by mate receptor. Mol Psychiatry 2002; 7: 508 – 514.
means of a universal array approach mediated by LDR (ligase 17 Sakurai K, Toru M, Yamakawa-Kobayashi K, Arinami T: Mutation detection reaction). BMC Microbiol 2002; 2: 27.
analysis of the N-methyl-D-aspartate receptor NR1 subunit gene 33 Guo Z, Guilfoyle RA, Thiel AJ, Wang R, Smith LM: Direct (GRIN1) in schizophrenia. Neurosci Lett 2000; 296: 168 – 170.
fluorescence analysis of genetic polymorphisms by hybridization 18 Ohtsuki T, Sakurai K, Dou H, Toru M, Yamakawa-Kobayashi K, with oligonucleotide arrays on glass supports. Nucleic Acids Res Arinami T: Mutation analysis of the NMDAR2B (GRIN2B) gene in schizophrenia. Mol Psychiatry 2001; 6: 211 – 216.
34 Ritchie MD, Hahn LW, Roodi N et al: Multifactor-dimensionality 19 Begni S, Moraschi S, Bignotti S et al: Association between the reduction reveals high-order interactions among estrogen-meta- G1001C polymorphism in the GRIN1 gene promoter region and bolism genes in sporadic breast cancer. Am J Hum Genet 2001; 69: schizophrenia. Biol Psychiatry 2003; 53: 617 – 619.
20 Rice SR, Niu N, Berman DB, Heston LL, Sobell JL: Identification of 35 Hahn LW, Ritchie MD, Moore JH: Multifactor dimensionality single nucleotide polymorphisms (SNPs) and other sequence reduction software for detecting gene-gene and gene-environ- changes and estimation of nucleotide diversity in coding and ment interactions. Bioinformatics 2003; 19: 376 – 382.
flanking regions of the NMDAR1 receptor gene in schizophrenic 36 Ritchie MD, Hahn LW, Moore JH: Power of multifactor dimen- patients. Mol Psychiatry 2001; 6: 274 – 284.
sionality reduction for detecting gene – gene interactions in the 21 Tani A, Kikuta R, Itoh K et al: Polymorphism analysis of the presence of genotyping error, missing data, phenocopy, and upstream region of the human N-methyl-D-aspartate receptor genetic heterogeneity. Genet Epidemiol 2003; 24: 150 – 157.
subunit NR1 gene (GRIN1): implications for schizophrenia.
37 Beier M, Hoheisel JD: Versatile derivatisation of solid support media for covalent bonding on DNA-microchips. Nucleic Acids Res 22 Martucci L, Wong AH, Trakalo J et al: N-methyl-D-aspartate receptor NR1 subunit gene (GRIN1) in schizophrenia: TDT and 38 Jorde LB: Linkage disequilibrium and the search for complex case-control analyses. Am J Med Genet 2003; 119B: 24 – 27.
disease genes. Genome Res 2000; 10: 1435 – 1444.
23 Hung CC, Chen HY, Chen CH: Systematic mutation analysis of 39 Sham PC, Curtis D: Monte Carlo tests for associations between the human glutamate receptor, ionotropic, N-methyl-D-aspartate disease and alleles at highly polymorphic loci. Ann Hum Genet 1 gene(GRIN1) in schizophrenic patients. Psychiatr Genet 2002; 40 Cho YM, Ritchie MD, Moore JH et al: Multifactor-dimensionality 24 Moore JH: The ubiquitous nature of epistasis in determining reduction shows a two-locus interaction associated with Type 2 susceptibility to common human diseases. Hum Hered 2003; 56: diabetes mellitus. Diabetologia 2004; 47: 549 – 554.
41 Xu Q, Jia YB, Zhang BY et al: Association study of an SNP 25 Pastinen T, Kurg A, Metspalu A, Peltonen L, Syvanen AC: combination pattern in the dopaminergic pathway in paranoid Minisequencing: a specific tool for DNA analysis and diagnostics schizophrenia: a novel strategy for complex disorders. Mol on oligonucleotide arrays. Genome Res 1997; 7: 606 – 614.
26 Southern EM, Maskos U, Elder JK: Analyzing and comparing 42 Carrasquillo MM, McCallion AS, Puffenberger EG, Kashuk CS, nucleic acid sequences by hybridization to arrays of oligonucleo- Nouri N, Chakravarti A: Genome-wide association study and tides: evaluation using experimental models. Genomics 1992; 13: mouse model identify interaction between RET and EDNRB pathways in Hirschsprung disease. Nat Genet 2002; 32: 237 – 244.
27 Gunderson KL, Huang XC, Morris MS, Lipshutz RJ, Lockhart DJ, 43 Moore JH, Williams SM: New strategies for identifying gene – gene Chee MS: Mutation detection by ligation to complete n-mer DNA interactions in hypertension. Ann Med 2002; 34: 88 – 95.
arrays. Genome Res 1998; 8: 1142 – 1153.
44 Tsai CT, Lai LP, Lin JL et al: Renin – angiotensin system gene 28 Erdogan F, Kirchner R, Mann W, Ropers HH, Nuber UA: Detection polymorphisms and atrial fibrillation. Circulation 2004; 109: of mitochondrial single nucleotide polymorphisms using a Supplementary Information accompanies the paper on European Journal of Human Genetics website (http://www.nature.com/ejhg)

Source: http://www.biowing.com.cn/Upload/PicFiles/2008.10.18_11.10.40_8270.pdf