No association between glutathione-synthesis-related genes and Japanese schizophrenia


Tohru Ohnuma, MD, PhD, Department of Psychiatry, Juntendo University, School of Medicine, 2-1-1 Hongo, Bunkyo-ku, Tokyo 113-8421, Japan. Email:


Aims:  Schizophrenia is a major psychiatric disorder with complex genetic, environmental, and psychological causes, and oxidative stress may be involved in the pathogenesis of the disease. Glutathione (GSH), one of the main cellular non-protein antioxidants and redox regulators, and altered GSH levels have been reported in various regions in patients with schizophrenia. Three enzymes are responsible for GSH synthesis: glutamate cysteine ligase modifier (GCLM), glutamate cysteine ligase catalytic subunit (GCLC), and glutathione synthetase (GSS). Previously, positive associations between GCLM and schizophrenia were reported in Europeans, but not in the Japanese population. Thus, in this study, we investigated the association between the GSH synthesis genes (GCLM, GCLC, and GSS) and schizophrenia in Japanese individuals.

Methods:  Seventeen single-nucleotide polymorphisms (SNP) in GCLM, GCLC, and GSS were genotyped in 358 patients with schizophrenia and in 359 controls.

Results:  No SNP showed a significant association between their allelic or genotypic frequencies and schizophrenia. Case–control haplotype association analysis using windows of two or three SNP showed no significant associations with schizophrenia. The case–control haplotype analyses based on the ascertained linkage disequilibrium blocks also showed no significant associations in any genes with schizophrenia.

Conclusions:  The three primary GSH synthesis genes do not have an apparent degree of association with schizophrenia in the Japanese population.

SCHIZOPHRENIA IS A major psychiatric disorder with complex genetic, environmental, and psychological causes. Hyperfunction of dopaminergic1 and hypofunction of glutamatergic (N-methyl-d-aspartate [NMDA] glutamate receptor) neurotransmission2 has been shown to be involved in the pathophysiology of the disease. Several studies have also suggested that oxidative stress may be involved in the pathogenesis of the disease.3–5 Oxidative stress is related to dopaminergic and glutamatergic neurotransmission. Excess dopamine may be produced in the brains of patients with schizophrenia, and auto-oxidation that produces dopamine quinine may occur.6,7 Further, NMDA receptors are also involved in the pathophysiology of the disease, because NMDA is a substrate that is protected from oxidation by modulation of the redox-sensitive site in the nervous system.8,9

Glutathione (GSH), one of the main cellular non-protein antioxidants and redox regulators, protects nervous tissue from reactive oxygen species.6 Altered GSH levels have been reported in various regions in patients with schizophrenia. Decreases have been reported in peripheral blood,10 the medial prefrontal cortex as measured by magnetic resonance spectroscopy (MRS),11 the caudate region in post-mortem brains,12 and cerebrospinal fluid.11 GSH is synthesized from glutamatergic amino acids, such as glutamate, glycine, and cysteine. Among these, disrupted neurotransmission of glutamate and glycine have been reported to be involved in the pathophysiology of schizophrenia. Indeed, the levels of these substrates may be altered in the brain13 and peripheral tissues14,15 of patients with schizophrenia. Thus, altered/disrupted GSH synthesis may be important in the pathophysiology of schizophrenia, and investigation of an association between genes related to GSH synthesis and schizophrenia is warranted.

Two enzymes are responsible for GSH synthesis. The first and rate-limiting enzyme is glutamate cysteine ligase (GCL),16 which is composed of two subunits, GCL modifier (GCLM)17 and GCL catalytic subunit (GCLC).18 The second enzyme is glutathione synthetase (GSS).16 Interestingly, positive associations between the gene encoding GCLM and schizophrenia were previously reported in European19 and Chinese populations,20 but not in the Japanese population.21,22 For GCLC, trinucleotide repeat (TNR) polymorphisms situated at the 5′-untranslated region of GCLC showed significant associations with European patients with schizophrenia,23 whereas GSS did not.19 However, to the best of our knowledge, the latter two genes have not been investigated in the Japanese population. Thus, in the current study, based on the hypothesis that disrupted GSH synthesis resulting in altered oxidative stress may be involved in schizophrenia, we investigated the association between the primary GSH synthesis genes (GCLM, GCLC, and GSS) and schizophrenia by performing case–control analysis in a Japanese population.



This case–control genetic association study was performed using 358 Japanese patients with unrelated schizophrenia (122 paranoid, 135 disorganized, and 101 catatonic type patients; 162 men, 196 women; mean age 39.6 years, SD ± 13.8) that met the DSM-IV diagnosis of schizophrenia according to structured clinical interviews, and 359 healthy controls (179 men, 180 women; mean age 41.6 years, SD ± 15.3). Healthy controls did not meet current or past criteria for any Axis 1 disorder. All participants met the following criteria: (i) no systemic or neurological disease; (ii) no prior head trauma with loss of consciousness; and (iii) no lifetime history of alcohol or substance dependency. Patients and controls were recruited from two geographic regions of Japan: Saitama and Tokyo. The mean age of the schizophrenic group was not significantly different from that of the control group (Student's t-test, t = 1.81, P > 0.05). The gender distribution between the two groups was not significantly different (χ2 = 1.72, P > 0.05). The ethics committee of the Juntendo University School of Medicine approved this study. All individuals gave written informed consent prior to participating in the study.


Genomic DNA was extracted from peripheral white blood cells using a QIAamp DNA Blood Maxi kit (Qiagen, Courtaboeuf, France). All of the single-nucleotide polymorphisms (SNP) that were investigated were typed using TaqMan technology (Assay-by-Design) on an ABI7500 system (Applied Biosystems, Foster City, CA, USA). All probes and primers were designed by the Assay-by-Design service of Applied Biosystems. Standard polymerase chain reactions (PCR) were carried out using the TaqMan Universal PCR Master Mix reagent kit in a 10-µL volume. For the selection of SNP, tag SNP for each gene (r2 > 0.8, minor allele frequencies >0.05) were chosen from The International HapMap Project database (release 21a of phase II, January 2007) ( with successful TaqMan probe design. For GCLM, a further additional polymorphism, rs2301022 from the reconfirmed positive schizophrenia association,19 was also included. Four SNP that could be used as comprehensive tag SNP of GCLM (1p22.1, 22.4 kb) were chosen. The distance between each SNP, genomic structure, and location of the investigated SNP are shown in Figure 1. The following SNP were investigated: rs41303970, rs2301022, rs718875, and rs7549683 (‘rs’ number of each SNP is the ID of the US National Center for Biotechnology Information SNP cluster from the dbSNP database; rs41303970 is located in the 5′ region of the gene. rs2301022 and rs718875 are intronic tag SNP, and rs2301022 showed a positive association with schizophrenia in a previous study.19 rs7549683 is located in the 3′ untranslated region. For GCLC (6p12, 47.7 kb), the following nine intronic SNP were chosen: rs534957, rs2397147, rs502862, rs3799695, rs1555906, rs761141, rs553822, rs542914, and rs670548. For GSS (20q11.2, 27.4 kb), four intronic SNP were chosen: rs6060127, rs6142264, rs2273684, and rs2236270. The confirmatory direct DNA sequencing methods for the SNP showed the deviations from Hardy–Weinberg equilibrium (HWE) were performed to check for typing errors by the TaqMan method. PCR primers of direct sequencing for five SNP were as follows and either forward (F)/reverse (R) primers were used for sequencing: rs3799695; TGAACCCCCAACACTGTAAC(F), AGAACTGGAAGGGACATTGG(R), rs670548; TTTAGCGGGAACGCATGCTGT(F), GGATGGGAAAGGCCATGGGGG(R), rs718875; TCGGAAGTGCTTTGGCTCAAATCGT(F), GGTTTCAAGAAGAGGGTGAATAC(R), rs7549683; TCAAACTATTCAGGTTACAGACACA(F), TCCTTAATTCAGGGCCGACA(R), rs2236270; CAGCACATCTCTGGAAACAGTG(F), TCCCTCAAGCCTTTTGCTAC(R). Detailed information on the PCR conditions is available upon request.

Figure 1.

Genomic structure of human (a) GCLM, (b) GCLC and (c) GSS, including the location of the single-nucleotide polymorphisms (SNP). Exons are denoted by boxes, untranslated regions are in white, and translated regions are in black. Distances between SNP are shown as base pairs.


For the case–control association study, genetic statistical analysis, HWE testing, and differences in genotypic/allelic frequencies were all done using SNPAlyze V7.0 Pro (Dynacom, Yokohama, Japan). Linkage disequilibrium (LD), denoted as D', was calculated from the haplotype frequency using the expectation–maximization algorithm. The LD block was also identified using SNPAlyze software when D' was greater than 0.75. Case–control haplotype analysis was also performed using SNPAlyze software. Permutation analysis was used to determine empirical significance and to calculate the P-values based on 10 000 replications. The global P-values represent the overall significance using the χ2-test when the observed versus expected frequencies of all haplotypes are considered together. All P-values reported are two-tailed. Statistical significance was defined as P < 0.05. We performed power calculations using the Power Calculator for Two Stage Association Studies ( Power was calculated under prevalence of 0.01 using an additive or a multiplicative model, based on allelic frequencies of the associated markers ranging from 0.045 (rs6142264) to 0.442 (rs1555906) and odds ratios ranging from 1.012 (rs542914) to 1.174 (rs502862) for the SNP investigated in this study with an alpha level of 0.05. Results of power analysis showed the power ranging from 5% (rs718875, rs534957, rs761141, rs542914, and rs670548) to 24% (rs502862).


A total of 17 SNP in GCLM, GCLC, and GSS were genotyped in 358 patients with schizophrenia and in 359 controls. Two SNP, rs3799695 and rs670548 in GCLC, showed the Japanese-specific features in their minor allele frequency (MAF); although MAF for both SNP was in the range of 17–19%, the homozygous genotypes of the minor allele were null. Thus, although these SNP may show a deviation from HWE in these current Japanese samples (P < 0.05), the genotypic features from these SNP corresponded well with the genotypic frequencies in Japanese samples from the HapMap database ( In addition, we confirmed the SNP showed the deviations from HWE in either cases or controls (P ≤ 0.05; rs3799695, rs670548, rs718875, 7549683 and rs2236270) of some randomly chosen subjects with direct DNA sequencing to check for typing errors by the TaqMan method. All genotypes determined by direct sequencing methods were in agreement with the genotypes of the TaqMan methods for these SNP. No SNP showed a significant association between their allelic or genotypic frequencies and schizophrenia (Table 1). Case–control haplotype association analysis using windows of two or three SNP in the three genes showed no significant associations with schizophrenia (Table 2).

Table 1.  Distribution and statistical analysis of the GCLM, GCLC and GSS gene polymorphisms
 Genotype frequency (%)P-valueHWE*Allele frequency (%)χ2P-valueOdds ratio (95%CI)
  • *

    P < 0.05.

  • HWE, Hardy–Weinberg equilibrium P-value.

rs41303970AAGAGG0.0360.322AG0.0660.0801.037 (0.787–1.366 )
Schizophrenia18 (5.0)88 (24.5)253 (70.5)120 (16.8)596 (83.2)
Controls7 (2.0)106 (29.6)245 (68.4)124 (17.3)594 (82.7)
rs2301022CCCTTT0.7470.460CT0.1130.7371.042 (0.755–1.220 )
Schizophrenia204 (57.0)133 (37.2)21 (5.9)541 (75.6)175 (24.4)
Controls197 (54.9)143 (40.0)19 (5.3)537 (74.8)181 (25.2)
rs718875CCCTTT0.0560.498CT0.0350.8521.027 (0.780–1.352)
Schizophrenia7 (2.0)107 (29.9)244 (68.2)121 (16.9)595 (83.1)
Controls17 (4.7)90 (25.1)252 (70.2)124 (17.3)594 (82.7)
rs7549683GGGTTT0.0910.103GT0.2400.6251.065 (0.729–1.209 )
Schizophrenia224 (62.6)120 (33.5)14 (3.9)568 (79.3)148 (20.7)
Controls229 (63.8)104 (29.0)26 (7.2)562 (78.3)156 (21.7)
rs534957CCCGGG0.9400.140CG0.0160.8991.014 (0.816–1.260)
Schizophrenia47 (13.1)155 (43.3)156 (43.6)249 (34.8)467 (65.2)
Controls50 (13.9)152 (42.3)157 (43.7)252 (35.1)466 (64.9)
rs2397147CCCTTT0.7120.507CT0.2850.5941.069 (0.732–1.195)
Schizophrenia18 (5.0)126 (35.2)214 (59.8)162 (22.6)554 (77.4)
Controls17 (4.7)137 (38.2)205 (57.1)171 (23.8)547 (76.2)
rs502862CCCTTT0.3070.861CT1.5580.2121.174 (0.662–1.096)
Schizophrenia20 (5.6)124 (34.6)214 (59.8)164 (22.9)552 (77.1)
Controls12 (3.3)121 (33.7)226 (63.0)145 (20.2)573 (79.8)
rs3799695AAAGGG0.408<0.05*AG0.5330.4651.106 (0.691–1.184)
Schizophrenia234 (65.4)124 (34.6)0 (0)592 (82.7)124 (17.3)
Controls224 (62.4)135 (37.6)0 (0)583 (81.2)135 (18.8)
rs1555906AAAGGG0.7950.291AG0.4690.4941.075 (0.755–1.145)
Schizophrenia79 (22.1)171 (47.8)108 (30.2)329 (45.9)387 (54.1)
Controls74 (20.6)169 (47.1)116 (32.3)317 (44.2)401 (55.8)
rs761141CCCTTT0.2850.676CT0.0090.9261.013 (0.779–1.316)
Schizophrenia237 (66.2)103 (28.8)18 (5.0)577 (80.6)139 (19.4)
Controls232 (64.6)116 (32.3)11 (3.0)580 (80.8)138 (19.2)
rs553822CCCTTT0.3760.930CT0.8870.3461.121 (0.703–1.132)
Schizophrenia27 (7.5)135 (37.7)196 (54.7)189 (26.4)527 (73.6)
Controls18 (5.0)138 (38.4)203 (56.5)174 (24.2)544 (75.8)
rs542914AAACCC0.5400.930AC0.8530.9261.012 (0.782–1.310)
Schizophrenia17 (4.7)111 (31.0)230 (64.2)571 (79.7)145 (20.3)
Controls12 (3.3)120 (33.4)227 (63.2)574 (79.9)144 (20.1)
rs670548AAAGGG0.854<0.05*AG0.0270.8691.024 (0.776–1.349)
Schizophrenia0 (0)122 (34.1)236 (65.9)122 (17.0)594 (83.0)
Controls0 (0)120 (33.4)239 (66.6)120 (16.7)598 (83.3)
rs6060127CCCGGG0.480.770CG1.4230.2321.167 (0.905–1.505)
Schizophrenia17 (4.7)126 (35.2)215 (60.1)160 (22.3)556 (77.7)
Controls13 (3.6)116 (32.3)230 (64.1)142 (19.8)576 (80.2)
rs6142264CCCTTT0.6960.465CT0.1460.7031.105 (0.542–1.512)
Schizophrenia0 (0)29 (8.1)329 (91.9)29 (4.1)687 (95.9)
Controls0 (0)32 (8.9)327 (91.1)32 (4.5)686 (95.5)
rs2273684GGGTTT0.3960.108GT0.9590.3271.131 (0.690–1.132)
Schizophrenia15 (4.2)125 (34.9)218 (60.9)155 (21.6)561 (78.4)
Controls14 (3.9)143 (39.8)202 (56.3)171 (23.8)547 (76.2)
rs2236270GGGTTT0.0570.171GT0.0430.8361.032 (0.719–1.306)
Schizophrenia264 (73.7)89 (24.9)5 (1.4)617 (86.2)99 (13.8)
Controls271 (75.5)74 (20.6)14 (3.9)616 (85.8)102 (14.2)
Table 2.  Results of two and three single-nucleotide polymorphisms (SNP)-based haplotype analyses
 2 SNP-based haplotype P-value* 3 SNP-based haplotype P-value*
  • *

    P-value represents global P-value.

 rs2301022        0.941      
 rs718875         0.870     
 rs2397147        0.868      
 rs502862         0.500     
 rs3799695          0.691    
 rs1555906           0.839   
 rs761141            0.953  
 rs553822             0.165 
 rs542914              0.802
 rs6142264         0.459     
 rs2273684          0.327    

D' > 0.75 was assumed to represent a strong LD. The LD between all four SNP in GCLM and GSS and the SNP rs3799695–rs542914 in GCLC were considered strong, suggesting that these SNP form a single haplotype block in controls (Table 3). Based on these results, case–control haplotype analyses were performed for these LD blocks (minor haplotypes with frequencies less than 3% in either the schizophrenia cases or controls were omitted). No significant association was observed between any of the major haplotypes investigated in any genes and schizophrenia (all P > 0.03) with the strict Bonferroni correction (for individual haplotype analysis, four comparisons were performed; thus, P < 0.0125 was considered significant.).

Table 3.  Results of linkage disequilibrium (D'/r2 value) between the single-nucleotide polymorphisms (SNP) Thumbnail image of


In this study, we focused on and performed genetic case–control analysis to look for an association between genes related to GSH synthesis (GCLM, GCLC, GSS) and Japanese patients with schizophrenia. All genes investigated here failed to show an association with schizophrenia for each individual SNP, for two-/three-window haplotypes, and following haplotype block analysis. For GCLM, three case–control studies from various geographic regions in Japan to date, including a study with large enough statistical power, failed to show an association with schizophrenia.21 Thus, GCLM alone cannot be considered a candidate gene for Japanese schizophrenia. Although SNP rs2301022 showed an association with European patients with schizophrenia,19 it was not a significant SNP in Chinese patients with strict correction for multiple tests20 or Japanese patients21,22 with schizophrenia. The difference in the importance of GCLM between Japanese, Chinese, and European populations may be due to ethnic differences. Indeed, genetic differences, such as the LD pattern, have been reported among ethnic groups, even between Chinese and Japanese ( populations, and the frequency of rs2301022 is significantly different between Japanese and Caucasian populations.21,22 For GCLC and GSS, this is the first case–control study using tag SNP that looked for a genetic association with Japanese schizophrenia. Similar to European individuals, these genes did not show an association with Japanese schizophrenia. Taken together, results from this study and previous studies in Japanese populations suggest that the primary genes related to GSH synthesis (GCLM, GCLC, GSS) appear not to be associated with schizophrenia. For GCLC, however, there is an interesting report that may indicate a genetic association with schizophrenia, because a TNR situated just 10 base pairs upstream of the start codon of GCLC showed a significant association with European patients with schizophrenia.23 In addition, individuals with TNR polymorphisms associated with schizophrenia showed decreased GCL expression, GLCL protein activity, and GSH content, which suggests that decreased resistance to oxidative stress may be involved in schizophrenia.23 Thus, it will be worthwhile to search for an association between these TNR and Japanese schizophrenia in our future study. Considering these current results, altered GSH levels in some tissues from patients with schizophrenia may not be due to genetic impairments in these primary genes for GSH synthesis, although the TNR in GCLC may be a possible genetic risk. Altered and especially decreased GSH levels in schizophrenia may contribute to increased activity of enzymes involved in GSH degradation, such as GSH-transferase (GSTM1, GSTM5), GSH-peroxidase (GPX1, GPX5), and γ-glutamyl transpeptidase (GGT). For some of these enzymes, however, no genetic associations were found with Japanese schizophrenia.22

In conclusion, this study is the first study to perform a genetic case–control study focusing on the three primary GSH synthesis genes (GCLM, GCLC, and GSS) simultaneously using tag SNP in Japanese individuals. Our results failed to show any associations between schizophrenia and these genes. A limitation of our study is that the number of participants was relatively small, and thus the statistical power was not adequate to produce robust results, possibly showing a type II error (false negative). However, we currently conclude that these genes do not have an apparent degree of association with Japanese schizophrenia.


Funding for this study was provided by the Juntendo Institute of Mental Health 2008. Authors T. Ohnuma and R. Hanzawa contributed to the experimental work, interpretation of the data, and writing of the paper. Authors Y. Nagai and N. Shibata contributed to the experimental work and the statistical analysis. Authors H. Maeshima, H. Baba, T. Hatano, Y. Takebayashi, Y. Hotta and M. Kitazawa contributed to the collection of samples. H. Arai contributed to the conception and design of the study. All authors contributed to the conceptualization, design, and writing of this manuscript.