Failure to support a genetic contribution of AKT1 polymorphisms and altered AKT signaling in schizophrenia

Authors


Address correspondence and reprint requests to Takeo Yoshikawa MD PhD, Laboratory for Molecular Psychiatry, RIKEN Brain Science Institute, 2–1 Hirosawa, Wako-city, Saitama 351–0198, Japan.
E-mail: takeo@brain.riken.go.jp

Abstract

The protein kinase v-akt murine thymoma viral oncogene homolog (AKT) gene family comprises three human homologs that phosphorylate and inactivate glycogen synthase kinase 3β (GSK3β). Studies have reported the genetic association of AKT1 with schizophrenia. Additionally, decreased AKT1 protein expression and the reduced phosphorylation of GSK3β were reported in this disease, leading to a new theory of attenuated AKT1-GSK3β signaling in schizophrenia pathogenesis. We have evaluated this theory by performing both genetic and protein expression analyses. A family based association test of AKT1 did not show association with schizophrenia in Japanese subjects. The expression levels of total AKT, AKT1 and phosphorylated GSK3β detected in the schizophrenic brains from two different brain banks also failed to support the theory. In addition, no attenuated AKT-GSK3β signaling was observed in the lymphocytes from Japanese schizophrenics, contrasting with previous findings. Importantly, we found that the level of phosphorylated GSK3β at Ser9 tended to be inversely correlated with postmortem intervals, and that the phosphorylation levels of AKT were inversely correlated with brain pH, issues not assessed in the previous study. These data introduce a note of caution when estimating the phosphorylation levels of GSK3β and AKT in postmortem brains. Collectively, this study failed to support reduced signaling of the AKT-GSK3β molecular cascade in schizophrenia.

Abbreviations used
BA

Brodmann's area

ETDT

extended transmission disequilibrium test

GSK3β

glycogen synthase kinase 3β

LD

linkage disequilibrium

NSW

TRC, New South Wales Tissue Resource Center

PBS-T

phosphate-buffered saline with 0.05% Tween 20

PDT

pedigree disequilibrium test

PMI

postmortem interval

PP2A

protein phosphatase 2A

SDS

sodium dodecyl sulfate

SNP

single nucleotide polymorphism

Schizophrenia is a major psychiatric disease with a worldwide prevalence of approximately 1%. Although the pathogenesis of schizophrenia is largely unknown, the neurodevelopmental hypothesis of schizophrenia is widely accepted. It suggests that pathophysiological changes in schizophrenia start in the early neurodevelopmental period (Weinberger 1996). Based on this, the expression of neurotrophic and transcription factors, protein kinases and other molecules associated with neural development have been studied, with some proteins and transcripts showing altered expression levels in the postmortem brains of schizophrenics (Barbeau et al. 1995; Impagnatiello et al. 1998; Takahashi et al. 2000; Ilia et al. 2002; Ohnuma et al. 2003; Aoki-Suzuki et al. 2005).

AKT (also known as protein kinase B) defines a family of closely related, highly conserved cellular homologs with protein kinase activity. Three family members, AKT1, AKT2 and AKT3, have been identified in humans. They play an important role in neuronal survival and differentiation (Conti et al. 2001; Dudek et al. 1997). Moreover, these enzymes mediate intracellular signaling for axon elongation and branching (Markus et al. 2002). Phosphorylated AKT inactivates glycogen synthase kinase 3β (GSK3β) through phosphorylation at the Ser9 site (Cross et al. 1995). GSK3β mediates apoptotic signals either by inhibiting transcription factors or by degrading β-catenin. Perturbation of this GSK3β pathway may mediate the pathogenesis of neurodegenerative disorders and schizophrenia (Kozlovsky et al. 2002; De Ferrari et al. 2003). Kozlovsky et al. (2000, 2004) reported decreased expression levels of GSK3β protein and mRNA in postmortem brains from schizophrenic patients. Beaulieu et al. (2004) showed a functional link between the AKT-GSK3β signaling pathway and increased dopaminergic neurotransmission, another putative etiologic factor in schizophrenia.

Emamian et al. (2004) demonstrated that protein expression of AKT1 and phosphorylation levels of GSK3β at Ser9 were reduced in postmortem brains and Epstein–Barr virus-transformed lymphocytes from schizophrenic patients. They showed that AKT1 polymorphisms were associated with schizophrenia, and that subjects with a core risk haplotype had a lower expression of AKT1 in lymphocytes relative to subjects with a common haplotype. In mouse experiments, treatment with the typical neuroleptic haloperidol resulted in the up-regulation of phosphorylated AKT1 (active) and phosphorylated GSK3β at Ser9 (inactive) in the brain.

Tau is a microtubule-associated protein and another target of GSK3β (Morishima-Kawashima et al. 1995). Beffert et al. (2002) reported that reelin signaling, another system implicated in schizophrenia, suppressed tau protein phosphorylation through the activation of AKT and the inactivation of GSK3β. Therefore, impairment of AKT activity in vivo would be reflected in the enhanced levels of phosphorylated tau protein. In this study we aimed to extend previous reports by re-examining the AKT-GSKβ cascade and tau phosphorylation levels in schizophrenia and other mental disorders, performing protein expression analyses in postmortem brains and lymphocytes obtained from patients, and by undertaking a genetic association study between AKT1 and schizophrenia.

Materials and methods

Subjects for the family based association study

Families with a genetic predisposition for schizophrenia for the family based association study were recruited from a geographic area located in central Japan. The probands, both in- and outpatients, were followed up by hospital doctors for at least 6 months. The sample consisted of 124 families with 376 members, of whom 163 suffered from schizophrenia. These included: (i) 80 independent and complete trios (schizophrenic offspring and their parents); (ii) 15 probands with one affected parent; (iii) 13 probands with affected siblings; (iv) 30 probands with discordant siblings (for detailed information see Yamada et al. 2004). Consensual diagnosis was made according to the DSM-IV (1994) criteria by at least two experienced psychiatrists on the basis of direct interviews, available medical records and information from hospital staff and relatives. None of the patients had additional Axis-I disorders, as defined by DSM-IV, and none of the present family members suffered from neurodegenerative disorders, including Parkinson's and Alzheimer's diseases. The present study was approved by the ethics committee of RIKEN. All subjects gave informed written consent to participate in the study after the provision and explanation of study protocols and purposes.

SNP genotyping of AKT1

Genomic DNA was isolated from blood samples using a standard method. We genotyped five single nucleotide polymorphisms (SNPs) at the AKT1 locus described in the original report (Emamian et al. 2004). AKT1 consists of 16 exons, and SNP1 (rs3803300), SNP2 (rs1130214), SNP3 (rs3730358), SNP4 (rs2498799) and SNP5 (rs2494732) are located in intron 2, intron 3, intron 5, exon 11 and intron 13, respectively. Assays-by-Design SNP genotyping products and TaqMan assay methods (Applied Biosystems, Foster City, CA, USA) were used to score the SNPs. Genotypes were determined using an ABI7900 sequence detection system and sds v2.2 software (Applied Biosystems). Each marker was checked for allele-inheritance inconsistency within a pedigree with pedcheck software, http://watson.hgen.pitt.edu/register (O'Connell and Weeks 1998), and either conflicts or flagged alleles were resolved by re-genotyping.

Statistical analyses of genetic association

Transmission distortions in the family panel were evaluated using the pedigree disequilibrium test (PDT) (Martin et al. 2000) (pdt Analysis Program v5.1; http://www.chg.duke.edu/software/pdt.html) and extended transmission disequilibrium test (ETDT) (Sham and Curtis 1995; http://www.mds.qmw.ac.uk/statgen/dcurtis/software.html). Transmit software (Clayton 1999; http://watson.hgen.pitt.edu/docs/transmit.html) was run as a global test of haplotype transmission. Genomic linkage disequilibrium (LD) patterns retained in the Japanese population were determined by pairwise LD examination of markers within AKT1 in 186 unrelated individuals from our schizophrenic pedigree panel. The standardized disequilibrium coefficient (D′) and the squared correlation coefficient (r2) were calculated with cocaphase software (http://www.mrc-bsu.cam.ac.uk/personal/frank/software/unphased/) (Dudbridge 2003).

Samples for protein analyses

The postmortem brain samples of Brodmann's area (BA) 9 (frontal cortex) and anterior hippocampus were obtained from the New South Wales Tissue Resource Centre (NSW TRC), University of Sydney, Australia. The samples were divided into two cohorts. The first cohort comprised 10 pairs of BA9 samples, and the second cohort consisted of eiht pairs of BA9 and 14 pairs of hippocampus samples. Each pair consisted of a schizophrenic patient (including one schizoaffective patient, pair no. 7 in Table 1) and a non-psychiatric control subject matched for age and sex. Detailed demographic data is shown in Table 1. Clinical information was collected in a standardized manner for use with the Diagnostic Instrument for Brain Studies (DIBS) (Keks et al. 1999).

Table 1.   Demographic data of NSW samples
Pair No.Schizophrenic patientsControl subjectsProvided brain region
SexAge
(years)
PMI
(h)
pHEthnicitySexAge
(years)
PMI
(h)
pHEthnicity BA9* HPC*
  • *

    BA9 samples are divided into two cohorts. Pairs No.1–10 comprise the first cohort and No.11–18 the second cohort. All hippocampal samples are included in the second cohort. Aus, Australian; BA, Brodmann's area; Cauc, Caucasian; F, female; HPC, hippocampus; M, male; PMI, postmortem interval.

1M57486.7Cauc/AusM56246.5Cauc/AusYesYes
2M4021.56.2Cauc/AusM3721Cauc/AusYesYes
3F58196.1Cauc/AusF56236.7Cauc/AusYesYes
4M32266.2Cauc/AusM37246.4Cauc/AusYesYes
5M30246.6Cauc/AusM34216.7Cauc/AusYesYes
6M57186.6Cauc/AusM53166.8Cauc/AusYesYes
7F61176.4Cauc/AusF60216.8Cauc/AusYesYes
8M5296.1Cauc/AusM57205.9Cauc/AusYesYes
9F56396.6Cauc/AusF52105.8Cauc/AusYesYes
10M44276.6Cauc/AusM44506.6Cauc/AusYesYes
11F67276.2Cauc/AusF689Cauc/AusYesYes
12M6756.4No infoM69166.6Cauc/AusYesNo
13M57366.4Cauc/AusM56376.8Cauc/AusYesNo
14M50366.2Cauc/AusM50196.3Cauc/AusNoYes
15F66136.5Cauc/AusF78376.5Cauc/AusYesYes
16M51216.0Cauc/AusM57186.6Cauc/AusYesNo
17M27336.3Cauc/AusM2828Cauc/AusYesNo
18F67216.4Cauc/AusF6664.5Cauc/AusYesYes
Mean
SD
 52.2
12.8
24.5
10.9
6.36
0.21
  53.2
13.2
22.2
10.7
6.37
0.60
   

An independent set of postmortem brain samples of BA6 (frontal cortex) from 60 subjects were obtained from the Stanley Medical Research Institute: the set comprised patients with schizophrenia, bipolar disorder, severe depression and non-psychiatric comparison subjects (Table 2). Two senior psychiatrists established DSM-IV diagnoses using information from all available medical records and from family interviews. Details regarding subject selection, diagnostic procedures and tissue processing are described elsewhere (Torrey et al. 2000). Data from BA6 were collected by an investigator blind to diagnosis. All samples were stored at −80°C until use.

Table 2.   Demographic summary of Stanley brain samples
 SchizophreniaBipolar disorderDepressionControl
  • *

    Values are indicated as mean ± SD;

  • **

    **Lifetime neuroleptic dose in fluphenazine equivalent dose.

Sex and numberM = 8, F = 5M = 7, F = 6M = 9, F = 5M = 9, F = 6
Age (years)*43.2 ± 13.340.3 ± 11.347.6 ± 8.748.1 ± 10.7
Antipsychotics*,**376 000 ± 65 00019 000 ± 23 000  
pH*5.92 ± 0.266.21 ± 0.236.19 ± 0.226.27 ± 0.24
PMI (h)*27.7 ± 13.433.1 ± 16.926.1 ± 9.623.7 ± 9.9

The lymphocytes isolated from peripheral blood were obtained from schizophrenia, bipolar disorder and non-psychiatric control subjects. Each disease set comprised seven patients and these were compared with seven control subjects recruited from central Japan. Lymphocytes were transformed by the Epstein–Barr virus in a standard manner. We cultured immortalized lymphocytes in RPMI1640 media (SIGMA, St Louis, MO, USA) supplemented with 20% fetal bovine serum (Invitrogen, Carlsbad, CA, USA) and penicillin-streptomycin (Invitrogen). We used confluent cultures of lymphoblastocytes in our experiments.

Western blot analysis

Either postmortem brain tissue or lymphocytes were homogenized in ice-cold lysis buffer [20 mm Tris-HCl (pH 7.5), 150 mm NaCl and 0.5 mm EGTA] containing protease inhibitors (0.5 mm phenylmethylsulfonyl fluoride, 2 µg/mL aprotinin and 10 µg/mL leupeptin) and phosphatase inhibitors (5 mm NaF, 50 mm Na3VO4 and 1 µm okadaic acid). Lysates were ultracentrifuged at 4°C for 20 min at 1.1 × 105 g. The supernatants were assayed for total protein concentration by Lowry's method, and concentrations were equalized with lysis buffer. Supernatants were diluted with either 2 × or 3 × sodium dodecyl sulfate (SDS) sample buffer and boiled for 3 min. Samples were separated on 12% SDS-polyacrylamide gels and transferred to polyvinylidene difluoride membranes. The membranes were incubated for at least 1 h in blocking solution containing 5% skimmed milk in phosphate-buffered saline with 0.05% Tween 20 (PBS-T) and then incubated overnight at 4°C with primary antibodies in the blocking solution. After that, the membranes were washed in PBS-T.

As conventional chemiluminescent western blotting was not sensitive enough to detect phospho-GSK3β, an alkaline phosphatase-conjugated secondary antibody (Promega, Madison, WI, USA) was used for detecting all proteins of interest in the first NSW TRC cohort and GSK3β and phospho-GSK3β in the Stanley bank cohort. The signals were developed using nitro-blue tetrazolium chloride-5-bromo-4-chloro-3′-indolylphosphatase p-toluidine salt reagent (Nacalai tesque, Kyoto, Japan) and signal intensities were quantified using NIH Image software (http://rsb.info.nih.gov/nih-image).

The other proteins of interest in the Stanley bank cohort were detected using conventional chemiluminescence, probing with either horseradish peroxydase-conjugated anti-mouse or rabbit IgG as a secondary antibody, and then visualizing the signal with either the ECL western blotting detection system (GE Healthcare Bio-Sciences, Piscataway, NJ, USA) or SuperSignal reagent (Pierce, Rockford, IL, USA), according to the manufacturer's instructions. Chemiluminescent signals were detected by an image analyzer LAS-1000 (Fujifilm, Tokyo, Japan) and quantified by Image Gauge.

In the second NSW TRC cohort we used standard-sized gels (15 cm × 13.5 cm) to load larger quantities of proteins (20 µg of protein equivalents in each well), and all antibodies, except the antibodies for tubulin and GSK3β, were diluted in Can Get Signal reagent (Toyobo, Osaka, Japan), which can enhance the antigen–antibody reaction. The signals were quantified as stated above.

Antibodies used for western blot analysis were: anti-Akt (1 : 1000; Cell Signaling, Danvers, MA, US), recognizing all AKT isoforms; anti-Akt1 (1 : 500; Upstate Biotechnology, Lake Placid, NY, USA), specific for the AKT1 isoform; anti-phospho-AKT (Ser473; 1 : 1000; Cell Signaling); anti-GSK3β (1 : 2500; BD Biosciences, San Jose, CA, US); anti-phospho-GSK3β (Ser9; 1 : 1000; Cell Signaling), anti-Tau (pS199; 1 : 2500; Biosource International, Camarillo, CA, USA), anti-Tau (pS396; 1 : 2500; Biosource International); anti-α-tubulin (1 : 25 000; Sigma). Tau-C, a purified rabbit polyclonal antibody against the C-terminal of tau protein, was used to detect whole tau (Ishiguro et al. 1995; Sato et al. 2002).

Statistical analyses for expression comparisons

The statistical significance of expression levels among groups was calculated by either the Wilcoxon signed-rank test (for paired samples) or the Mann–Whitney U-test, two-tailed (for unpaired samples). The statistical significance of correlation was evaluated by the Spearman's rank correlation test. We excluded one sample pair (pair no. 8 in Table 1) from the first NSW TRC cohort from the Wilcoxon signed-rank test because of an outlier with extremely low tubulin expression. However, we included this pair in the correlation test.

Results

Association analysis between AKT1 and Japanese families with a predisposition to schizophrenia

Linkage disequilibrium examination showed strong LD between SNP1 and 2, and SNP3, 4 and 5 (Table 3). Emamian et al. (2004) reported marginally significant transmission distortion of one SNP, SNP3 (p = 0.05), which was in strong LD with SNP4 (D′ = 0.90), as seen in our samples, and detected significant, albeit modest, global haplotypic associations (p = 0.02–0.05) with those haplotypes that combined SNP3 with neighboring SNPs. We examined the same five SNPs in Japanese families with a predisposition to schizophrenia, but none of the SNPs exhibited significant transmission disequilibrium either by PDT (for all families) or by ETDT (for 80 independent and complete trios) (Table 4). Moreover, haplotype transmission analysis found no SNP-based haplotypes that were preferentially transmitted to schizophrenics (Table 4).

Table 3.   Pairwise marker-to-marker linkage disequilibrium (LD) statistics for the AKT1 locus
 SNP1SNP2SNP3SNP4SNP5
  1. For each pair of markers, standardized D is shown above the diagonal and r2 below the diagonal. D > 0.8 and r2 > 0.1 are shown in gray.

SNP10.9560.2860.340*0.218
SNP20.2180.0780.1450.376*
SNP30.0100.0030.8971.000
SNP40.0730.0030.0941.000
SNP50.0340.0150.0550.457
Table 4.   Results of the family based association study between AKT1 and schizophrenia
 PDT p-value*ETDT
p-value
Haplotype transmission (global p-value**)
SUMAVE1SNP2SNPs3SNPs4SNPs5SNPs
  • *

    The PDT program computes two statistical measures, PDT-SUM and PDT-AVE. Briefly, PDT-SUM gives more weight to larger families, whereas PDT-AVE places equal weight on all families.

  • **

    p-values for multiallele testing. ETDT, extended transmission disequilibrium test; PDT, pedigree disequilibrium test.

SNP10.870.810.300.300.52   
SNP20.760.630.880.890.290.330.54 
SNP30.450.280.300.220.290.500.780.70
SNP40.540.320.130.110.260.41  
SNP50.930.690.350.29    

AKT signaling in lymphocytes from patients with schizophrenia and bipolar disorder

Emamian et al. (2004) reported that expression levels of AKT1 and GSK3β phosphorylated at Ser9 were both dramatically decreased to less than 50% in cultured lymphocytes from schizophrenic patients. We performed western blotting to examine these potential alterations of AKT signaling in cultured lymphocytes obtained from both schizophrenic patients and patients with bipolar disorder. β-Actin was used as an internal control because the expression of tubulin (used in the study of Emamian et al. 2004) in lymphocytes varied widely among the samples (data not shown). The lymphocyte expression of AKT and GSK3β was unaltered in schizophrenia and bipolar disorder samples relative to controls (Figs 1a, b and e). Furthermore, no significant differences were found in the phosphorylation ratios of GSK3β (p-Ser9-GSK3β/total GSK3β) between psychiatric patients and controls (Fig. 1d). The phosphorylation level of AKT (p-Ser473-AKT/total AKT) was also not altered in schizophrenic patients (p = 0.90), but it was significantly increased in patients with bipolar disorder compared with controls (p = 0.038) (Fig. 1c).

Figure 1.

 Western blot analysis of protein expression and phosphorylation in cultured lymphocytes. (a) Relative expression levels of AKT against β-actin. (b) Relative expression levels of GSK3β against β-actin. (c) Phosphorylated fractions of AKT (p-Ser473-AKT) relative to total AKT. (d) Phosphorylated fractions of GSK3β (p-Ser9-GSK3β) relative to total GSK3β. The y-axis indicates the ratio of mean protein density from each disease group to the mean density value of corresponding control subjects. The mean value of each group is represented by a horizontal bar. Blots detected by chemiluminescence are shown in (e). Bp, bipolar disorder; Ct, control; Sz, schizophrenia; *p < 0.05; ns, not-significant.

AKT signaling in brain samples from the NSW TRC

Frontal cortex (BA9) postmortem samples from the first NSW TRC cohort were examined by western blotting to evaluate AKT signaling in schizophrenia samples. Tubulin was used as a marker because its expression remained stable in all brain samples. There were no significant differences in the expression levels and phosphorylation ratios of AKT, GSK3β(Fig. 2) and tau (Figs 3a and b) between the schizophrenia and control cases.

Figure 2.

 Western blot analysis of protein expression and phosphorylation status in postmortem brains (BA9) of the first cohort from the NSW TRC (paired samples). (a) Relative expression levels of AKT against tubulin. (b) Relative expression levels of GSK3β against tubulin. (c) Phosphorylated fractions of AKT (p-Ser473-AKT) against total AKT. (d) Phosphorylated fractions of GSK3β (p-Ser9-GSK3β) against total GSK3β. Each age- and sex-matched pair is connected by a line, and the mean value of each group is indicated as a horizontal bar connected with dotted lines between the groups; the p-values are also shown. Blots were visualized by alkaline phosphatase staining (e). Ct, control; Sz, schizophrenia.

Figure 3.

 Expression of phosphorylated tau and its correlation with phosphorylated AKT and GSKβ, and the effects of confounding factors on the expression of phosphorylated AKT and GSKβ, examined in postmortem brains (BA9) of the first NSW TRC cohort (paired samples). (a) Phosphorylated fractions of tau (p-Ser199-tau) against total tau (tauC). Each age- and sex-matched pair is connected by a line and the mean value of each group is indicated by a horizontal bar connected by dotted lines between groups; the p-value is also shown. (b) Phosphorylated fractions of tau (p-Ser396-tau) against total tau (tauC) are shown as in (a). (c) Correlation between the ratios of phosphorylated tau (p-Ser199-tau/TauC) and phosphorylated AKT (p-Ser473-AKT/AKT), examined in the combined schizophrenic and control samples. Spearman's correlation coefficient (r) and the p-value are shown. (d) Correlation between the ratio of phosphorylated tau (p-Ser396-tau/TauC) and phosphorylated AKT. (e) Effect of sample pH on the phosphorylation level of AKT (p-Ser473-AKT/AKT) in the combined schizophrenia and control samples. Spearman's correlation coefficient (r) and the p-value are shown. (f) Effect of postmortem interval (PMI) on the phosphorylation level of GSKβ (p-Ser9-GSKβ/GSKβ).

Given the fact that phosphorylated and therefore active AKT is a negative regulator of GSK3β, a direct phosporylator of tau, we speculated that an inverse correlation might exist between phosphorylation levels of AKT and tau. However, we found to the contrary significant positive, rather than negative, correlations between Ser473-phosphorylated AKT and Ser199-phosphorylated tau (Fig. 3c), and between Ser473-phosphorylated AKT and Ser396-phosphorylated tau (Fig. 3d). These results suggest that other, more salient, signal transduction pathways determine the phosphorylation status of AKT and/or tau in postmortem brains. AKT and tau are both dephosphorylated by protein phosphatase 2A (PP2A) (Goedert et al. 1992; Andjelkovic et al. 1996; Beaulieu et al. 2005). We previously reported that hypothermia induced by reduced glucose metabolism leads to tau hyperphosphorylation through the inactivation of PP2A (Planel et al. 2004). Hypoglycemia-hypothermia is one of the major agonal medical consequences. The precise mechanism for current findings on the phosphorylation status of AKT and tau is not known, but the involvement of factors relevant to terminal conditions, including phosphatase activities, should be further examined. In this regard, Li et al. (2004) reported a relationship between terminal medical conditions and the tissue pH of postmortem brains. We therefore examined the effects of pH and postmortem interval (PMI) on the phosphorylation level of proteins in brain samples. A significant inverse correlation was found between pH and AKT phosphorylation levels (Fig. 3e). An inverse correlation was also found between pH and tau phosphorylation at Ser199 (r = −0.51, p = 0.025, data not shown), and between pH and tau phosphorylation at Ser396 (r =−0.22, p = 0.36, data not shown), albeit not significant. For the effects of PMI on protein phosphorylation, GSK3β showed a tendency for inverse correlation with PMI (Fig. 3f). The phosphorylation levels of AKT and tau did not correlate with PMI (data not shown). Collectively, these data raise the point that the phosphorylation levels of AKT, GSK3β and tau in brain samples are largely dependent on the terminal medical state of the patient and the storage conditions after death.

We used anti-pan-AKT antibody to detect all subtypes of AKT proteins in the above experiments. Next, we focused on AKT1 subspecies using AKT1-specific antibodies in the second cohort of frontal cortex and hippocampus samples from the NSW TRC. No significant differences between schizophrenics and controls were observed in the expression levels of AKT1, GSKβ or phosphorylated GSK3β(Fig. 4).

Figure 4.

 Western blot analysis of protein expression and phosphorylation in postmortem brains (BA9, a, b and c; hippocampus, d, e and f) of the second NSW TRC cohort (paired samples). Panels (a) and (d) show the relative expression levels of AKT against tubulin. Panels (b) and (e) show the relative expression levels of GSK3β against tubulin. Panels (c) and (f) show the phosphorylated fractions of GSK3β (p-Ser9-GSK3β) against total GSK3β. Each age- and sex-matched pair is connected by a line and the mean value of each group is indicated by a horizontal bar connected by dotted lines between groups; the p-values are also shown. Blots detected by chemiluminescence are shown in (g). Ct, control; Sz, schizophrenia.

AKT signaling in brain samples from the Stanley cohort

Emamian et al. (2004) analyzed frontal cortex samples (the anatomical subregions were not specified) obtained from the Stanley foundation and reported decreased AKT1 levels and reduced phosphorylated GSKβ in schizophrenics compared with controls. Therefore, we assessed the samples from the same resource. We also included the frontal cortex of individuals with bipolar disorder and major depression as disease control groups, in addition to schizophrenia. The samples from schizophrenic and bipolar patients showed slightly, but not significantly, higher expression levels of AKT than in control subjects, whereas the expression in major depression samples did not differ from controls (Fig. 5a). No differences were found in the total GSK3β and the phosphorylation levels of GSK3β at Ser9 and tau at Ser199 between the three major psychiatric patients and control subjects (Figs 5b, c and d).

Figure 5.

 Western blot analysis of protein expression and phosphorylation in postmortem brains (BA6) from the Stanley cohort. (a) Relative expression levels of AKT against tubulin. (b) Relative expression levels of GSK3β against tubulin. (c) Phosphorylated fractions of GSK3β (p-Ser9-GSK3β) against total GSK3β. (d) Phosphorylated fractions of tau (p-Ser119-tau) against total tau (tauC). The mean value of each group is indicated by a horizontal bar; the p-values are also shown. Blots were visualized by chemiluminescence (AKT, GSK3β, tauC, p-Ser199-tau and tubulin) and alkaline phosphatase staining (GSK3β and p-Ser9-GSK3β in the bottom two lines) (e). Bp, bipolar disorder; Ct, control; Dp, major depression; Sz, schizophrenia.

Discussion

In contrast to the report of Emamian et al. (2004), we were unable to detect an association between AKT1 polymorphisms and schizophrenia in Japanese families with a predisposition to schizophrenia. Several other research groups have attempted replication studies, but the results vary among groups. Schwab et al. (2005) performed a family based association study using 79 Caucasian schizophrenia sib-pair families, an ethnic group similar to the original study. They investigated five SNPs used in the original study plus an additional two SNPs in the neighborhood of SNP3, and detected significant association with SNP3 (p = 0.002) and haplotypes (p = 0.0013) spanning the SNP3 locus. Their results replicate the association in Caucasians. In Japanese samples, Ikeda et al. (2004) reported a significant association between SNP5, but not SNP3, and haplotypes including SNP5 of AKT1 and schizophrenia in case-control samples. Ohtsuki et al. (2004) also examined Japanese case-control samples, but could find no significant associations. In Taiwanese subjects, Liu et al. (2006) showed no significant association of the five SNPs with schizophrenia in a family based study. It is possible that AKT1 polymorphisms confer a variable disease risk across different populations, with a strong contribution in Caucasians that falls to either equivocal or weak in East Asians. We have previously reported an association between AKT1 and bipolar pedigrees in Caucasian samples (Toyota et al. 2003).

Our protein analyses failed to support the theory of decreased AKT expression in schizophrenia proposed by Emamian et al. (2004), as the expression levels of total AKT in schizophrenic brains from both the NSW TRC and the Stanley Bank were unchanged, and AKT1 levels in the second cohort of NSW TRC were also unaltered. One possible reason for the discrepancy may be a difference in the brain regions examined. We examined BA6 from the Stanley foundation and BA9 from the NSW TRC; both are areas of the frontal cortex. Emamian et al. (2004) did not specify the precise anatomical region used in their study and may have analyzed a different subarea of the frontal cortex. Our lymphocytes from Japanese schizophrenics did not display altered AKT levels. It is possible that reduced AKT expression in schizophrenia is not robust enough to be reliably detected across ethnic populations, anatomical regions and detailed experimental procedures including the selection of internal control probes.

Our examination of GSKβ from the brains and lymphocytes of schizophrenic subjects did not confirm the findings of Emamian et al. (2004). Importantly, it sounds a strong note of caution in the assessment of AKT brain phosphorylation levels, GSK3β and the downstream target tau. It is clear that the phosphorylation status of these proteins are affected by numerous factors that are difficult to control in postmortem specimens, such as temperature, the period of brain dissection and preservation, PMI and pH. Indeed, several lines of evidence have suggested that the phosphorylation status of GSK3β in postmortem brains does not reflect that of the living brain. For example, Li et al. (2005) found that in the mouse brain, approximately 90% of both phospho-Ser-GSK3α and β were dephosphorylated within 2 min of decapitation. This is in keeping with our failure to detect phosphorylated GSK3β at Ser9 in brain samples using standard chemiluminescent methods. Furthermore, a number of reports have shown that GSK3β phosphorylation status and function are controlled by circadian rhythms (Martinek et al. 2001; Iitaka et al. 2005).

Before the report of Emamian et al. (2004), GSK3β was implicated in the neurodevelopmental disturbances of schizophrenia. Taking the opposite stand, Kozlovsky et al. (2002) postulated that reduced activity of GSK3β could contribute to the pathogenesis of schizophrenia, based on their findings that schizophrenic postmortem brains showed reduced expression levels of GSK3β mRNA and protein. On the other hand, Beasley et al. (2002) did not observe a reduction of GSK3β protein levels in schizophrenia.

In conclusion, although the newly proposed theory of reduced signaling in the AKT-GSK3β molecular cascade may explain some aspects of schizophrenia pathology, more evidence is required, particularly regarding the in vivo phosphorylation levels of constituent proteins directly linked to the functional status of the signaling cascade.

Acknowledgements

This work was supported by research grants from RIKEN Brain Science Institute and a Grant-in-Aid for Scientific Research (Japanese Ministry of Education, Culture, Sports, Science and Technology). The NSW TRC tissues were received from the Australian Brain Donor Programs NSW Tissue Resource Centre, which is supported by the University of Sydney, National Health and Medical Research Council of Australia, Neuroscience Institute of Schizophrenia and Allied Disorders, National Institute of Alcohol Abuse and Alcoholism and NSW Department of Health. Postmortem brains were donated by The Stanley Medical Research Institute's Brain Collection, courtesy of Drs Knable, Torrey, Webster, Weis and Yolken.

Ancillary