Selective reduction of a PDZ protein, SAP-97, in the prefrontal cortex of patients with chronic schizophrenia


Address correspondence and reprint requests to Hiroyuki Nawa, Department of Molecular Biology, Brain Research Institute, Niigata University, Asahimachi-dori 1–757, Niigata 951–8585, Japan. E-mail:


Many postsynaptic density proteins carrying postsynaptic density-95/discs large/zone occludens-1 (PDZ) domain(s) interact with glutamate receptors to control receptor dynamics and synaptic plasticity. Here we examined the expression of PDZ proteins, synapse-associated protein (SAP) 97, postsynaptic density (PSD)-95, chapsyn-110, GRIP1 and SAP102, in post-mortem brains of schizophrenic patients and control subjects, and evaluated their contribution to schizophrenic pathology. Among these PDZ proteins, SAP97 exhibited the most marked change: SAP97 protein levels were decreased to less than half that of the control levels specifically in the prefrontal cortex of schizophrenic patients. In parallel, its binding partner, GluR1, similarly decreased in the same brain region. The correlation between SAP97 and GluR1 levels in control subjects was, however, altered in schizophrenic patients. SAP102 levels were also significantly reduced in the hippocampus of schizophrenic patients, but this reduction was correlated with sample storage time and post-mortem interval. There were no changes in the levels of the other PDZ proteins in any of the regions examined. In addition, neuroleptic treatment failed to mimic the SAP97 change. These findings suggest that a phenotypic loss of SAP97 is associated with the postsynaptic impairment in prefrontal excitatory circuits of schizophrenic patients.

Abbreviations used

AMPA receptor-binding protein


α-amino-3-hydroxy-5-methyl-4-isoxazoleproprionic acid


brain-derived neurotrophic factor




long-term potentiation


neuron-specific enolase


polyacrylamide gel electrophoresis


PSD-95/discs large/zone occludens-1


polyvinylidene difluoride


postsynaptic density


synapse-associated protein


sodium dodecyl sulfate–polyacrylamide gel electrophoresis

Dysfunction of glutamatergic neurotransmission and plasticity is implicated in the pathophysiology of schizophrenia (Weickert and Kleinman 1998; Cull-Candy et al. 2001). Recent studies demonstrate that synaptic function and plasticity of glutamate receptors are regulated by a family of postsynaptic molecules containing a common motif, called the PDZ (postsynaptic density-95/discs large/zone occludens-1) domain (O'Brien et al. 1998; Nagano et al. 1998; Fanning and Anderson 1999). The PDZ domain refers to a stretch of 80–90 amino acid residues that was initially identified in three proteins, postsynaptic density (PSD)-95, Dlg and ZO-1. Among various postsynaptic PDZ proteins, PSD-95, also called synapse-associated protein (SAP) 90, was first reported to bind to the C-terminal motif of the NR2 subunits of the NMDA-type glutamate receptor through this domain (Cho et al. 1992; Kornau et al. 1995). Other postsynaptic proteins sharing structural similarity with PSD-95 were subsequently identified: SAP102 (Muller et al. 1996), PSD-93/chapsyn-110 (Brenman et al. 1996; Kim et al. 1996) and SAP97 (Muller et al. 1995). Similarly, a group of PDZ molecules interacting with α-amino-3-hydroxy-5-methyl-4-isoxzolepropionic acid (AMPA)-type glutamate receptors has been identified. SAP97 associates with the C-terminal motif of glutamate receptor 1 (GluR1)(Leonard et al. 1998; Sans et al. 2001) and GRIP, AMPA receptor-binding protein (ABP), stargazin and Pick1 bind to the C-terminal motif of GluR2/3 (Dong et al. 1997; Srivastava et al. 1998; Dong et al. 1999; Xia et al. 1999; Chen et al. 2000). The PDZ domain of PSD-95 and other proteins also interacts with a variety of other signaling molecules at synapses (O'Brien et al. 1998; Nagano et al. 1998; Fanning and Anderson 1999). The latest gene-targeting experiments revealed that the interaction between glutamate receptors and the PDZ proteins regulates receptor functions as well as synaptic plasticity: Mice lacking the gene for PSD-95 exhibit abnormal long-term potentiation (LTP) and learning deficits (Migaud et al. 1998). Neurons carrying mutated glutamate receptor carboxyl termini have abnormal synaptic plasticity and impaired learning behaviors (Sprengel et al. 1998; Hayashi et al. 2000). These findings suggest that the postsynaptic glutamate receptors cannot work effectively unless properly arranged both spatially and temporally by these PDZ proteins. Therefore, impaired expression of PDZ proteins might be involved in various brain diseases.

Several studies on post-mortem brain revealed a reduction in the AMPA and NMDA receptor levels in the corticolimbic regions of schizophrenic patients (Harrison et al. 1991; Eastwood et al. 1997; Kerwin and Harrison 1997). The association between decreased glutamate receptor levels and schizophrenia has also been examined pharmacologically and genetically. Administration of non-competitive NMDA receptor channel antagonists, phencyclidine and ketamine, to humans provokes various psychotic symptoms such as those observed in patients with schizophrenia, associated with a reduced regional cerebral blood flow in corticolimbic regions (Castren et al. 1993; Lahti et al. 1995). In agreement, NMDA receptor knockout mice have impaired social interaction behavior that normalizes after administration of atypical neuroleptics (Mohn et al. 1999). Such changes might be due to the altered synaptic connectivity of this region. Although this hypothesis is controversial, these findings support the idea that hypoglutamatergic function of the prefrontal cortex contributes to the etiology or pathophysiology of schizophrenia.

PDZ proteins markedly influence postsynaptic glutamate receptor function (Migaud et al. 1998; Sprengel et al. 1998; Hayashi et al. 2000). Both PDZ proteins and glutamate receptors interact to regulate and maintain their expression and localization (Fanning and Anderson 1999; Nagano et al. 1998; O'Brien et al. 1998). We recently reported that brain-derived neurotrophic factor (BDNF) up-regulates expression of PDZ proteins including SAP97 (Jourdi et al. 2001). Given the impaired signaling of BDNF/TrkB in schizophrenic patients (Takahashi et al. 2000), the above findings raise the question of whether schizophrenia also perturb the expression of PDZ proteins through the abnormal BDNF/TrkB signaling. In the present study, we quantitatively measured protein levels of the five PDZ proteins and examined the correlation of the PDZ protein levels with the levels of their interacting glutamate receptors.

Materials and methods

Human tissue sampling

Post-mortem brain tissue was collected from 34 patients with chronic schizophrenia (23 men, 11 women; mean age: 62.0 years; SD 14.7) and from 41 controls (26 men, 15 women, mean age: 61.4 years; SD 12.8) who had no history of neuropsychiatric disorders (Table 1). The 34 schizophrenic patients had all taken typical antipsychotics for prolonged periods. The post-mortem diagnoses of schizophrenia, based on the reported psychiatric symptoms, corresponded to the DSM-IV categories (American Psychiatric Association 1994). In each case, the left cerebral hemisphere was fixed in formalin for diagnostic examination and the right hemisphere was frozen at −80°C. All pathologic analyses were performed at the authors' institutes. Protein samples were randomly taken from post-mortem brains that did not exhibit neurodegenerative abnormalities with conventional pathologic staining (Wakabayashi et al. 1999). Conclusions drawn from post-mortem brain samples require careful consideration, however, because cause of death, post-mortem delay after heart cessation, autolysis during storage after death, etc., could all impact on the findings (Harrison 1999). In an attempt to control for these factors, as well as for age-associated neuronal degeneration, the quality of the samples was evaluated by measuring levels of neuron-specific enolase (NSE) as an internal control. NSE is often used as a molecular marker for live neurons (Suzuki et al. 2001). Accordingly, brain samples that contained less than half the average level of NSE were excluded prior to examinations.

Table 1.  Autopsy and clinical data

All anatomic orientations were based on the Human Brain Atlas. The dorsolateral prefrontal cortex (approximately BA46) and the occipital cortex (approximately BA 17) were sampled. Hippocampal tissue including the CA regions and dentate gyrus, but not the surrounding entorhinal and parahippocampal areas, was also harvested. The families of the control and schizophrenic subjects provided written informed consent for the use of brain tissues. All experiments were performed under the authorization of the Matsuzawa Hospital Ethics Committee and the Niigata University Ethics Committee.

Neuroleptic treatment of animals

Male Wistar rats (7 weeks old, initial weight 200–220 g) were housed on a 12-h light–dark cycle with free access to food and water. Different groups of rats were treated intraperitoneally with either haloperidol (0.5 mg/kg, intraperitoneally) or vehicle for 2 weeks. Haloperidol was dissolved in a minimal amount of acetic acid, diluted with distilled water, and pH-balanced with NaOH to give a final solution of 0.5 mg/mL, pH 5.5. Brains were harvested rapidly from decapitated rats, and each brain region was dissected on ice. All animal experiments were performed according to the Animal Use and Care Committee Guidelines of Niigata University.

Western blot analysis

Relative levels of PDZ proteins and glutamate receptors were determined by western blot analysis using methods similar to those described previously (Wakabayashi et al. 1999). Brain tissues were homogenized with tissue lysis buffer containing 200 mm Tris buffer and 2% sodium dodecyl sulfate (SDS). After centrifugation at 10 000 × g for 10 min, supernatants were boiled and protein concentrations were determined using a Micro BCA kit (Pierce Chemical, Rockland, IL, USA). As a stab gel could not accommodate all of the above samples, protein samples (50 µg/lane) prepared from schizophrenic patients and control subjects, were alternately allocated to lanes and separated in two 1.0-mm thick, 7.5% polyacrylamide slab gels, together with prestained molecular weight markers (New England BioLabs, Beverly, MA, USA) as well as with internal positive controls of standard human brain extract (50 µg/lane) and rat hippocampal protein (50 μg/lane). Protein was denatured in Laemmli sample loading buffer [10% glycerol, 2% SDS, 20 mm dithiothreitol (DTT), 63 mm Tris, and 0.01 mg/mL bromophenol blue (pH 6.8)] at 90°C and separated by SDS–polyacrylamide gel electrophoresis (PAGE). The protein was transferred onto polyvinylidene difluoride (PVDF) protein sequencing membrane (BioRad, Hercules, CA, USA) or a nitrocellulose membrane (Schleicher and Schull, Dassel, Germany) by electrophoresis in transfer buffer [25 mm Tris, 19.2 mm glycine (pH 8.5), and 20% methanol]. Completion of protein transfer was confirmed by the transfer of the prestained markers. The membrane was rinsed with Tris-buffered saline containing Tween-20 (TBST; 10 mm Tris, 150 mm NaCl, and 0.1% Tween-20, pH 8.0) and treated with blocking solution (TBST plus 5% non fat dry milk). Immunoblots were treated with various rabbit affinity-purified antibodies or mouse monoclonal antibodies: anti-SAP102 antibody (0.7 µg/mL), anti-PSD-95 antibody (0.3 µg/mL), anti-chapsyn-110 antibody (0.5 µg/mL) (made by M. Watanabe; Hokkaido University, Japan), anti-GRIP1 antibody (3 µg/mL) (Upstate Biotechnology, Waltham, MA), anti-SAP97 antibody (2 µg/mL) (Wakabayashi et al. 1999), anti-glutamate receptor 1 (GluR1) antibody (0.5 µg/mL) (Chemicon, Temecula, CA, USA), and anti-NSE antibody (0.5 µg/mL) (Chemicon). After extensive washing with TBST, the membrane was reacted with anti-rabbit/mouse immunoglobulin conjugated with peroxidase (Chemicon). The immunoreactivity on the membrane was visualized by a chemiluminescence reaction combined with film exposure (ECL kit; Amersham Pharmacia Biotech, Piscataway, NJ, USA). All the above procedures were performed simultaneously for multiple immunoblots to minimize technical variation between blots. Densitometric quantitation of the band intensities was performed within the range of linear exposure of the film. Small variations in transfer efficacy among immunoblots were normalized by a mean intensity of the target PDZ protein in the above internal standards. All the blotting was repeated and differences in PDZ proteins were confirmed by an independent set of immunoblotting. Specificity of the antibodies had been tested previously (Wakabayashi et al. 1999; Iwakura et al. 2001; Suzuki et al. 2001).

Statistical analysis

Values with a standard distribution were examined using the chi-square test. Protein levels that did not fit a standard distribution were examined using the Mann–Whitney U-test to compare individual PDZ proteins. Spearman's rank correlation test was used to examine relative protein levels. A p-value of less than 0.05 was considered to be statistically significant. All conditional values of the samples are presented as mean ± SD and experimental data from post-mortem samples or rat tissue are presented as mean ± SE.


Post-mortem brain samples were collected from 34 chronic schizophrenia patients (classified by DSM-IV) (Table 1). Control subjects who had no history of neuropsychiatric disorders were mostly matched for age, sex, and post-mortem interval (Table 2). Prefrontal cortex was analyzed for differences between the two groups in the protein levels of the five PDZ proteins, SAP97, PSD-95, chapsyn-110, GRIP1 and SAP102. Neuronal content of a cytoplasmic enzyme, NSE, was monitored in parallel as an internal control. The distribution of these protein levels did not fit a standard distribution and therefore Mann–Whitney non-parametric analysis was mainly used for individual comparisons of PDZ proteins. SAP97 levels were significantly decreased in the schizophrenic group (U25,14 = 249, Z = 2.16, p = 0.03) (Fig. 1). The levels of SAP102 in the schizophrenic group tended to decrease (p = 0.09), but those of all the other PDZ proteins were similar in comparison with controls. Even if the parametric analysis of Welch's t-test was applied to the comparison for GRIP1 levels, the difference failed to reach a statistical significance range (p = 0.08). In the prefrontal cortex, there was no significant correlation between SAP97 and storage time (rs = −0.27, Z = −1.69, p = 0.09 for control plus schizophrenic groups, n = 39), although storage time was significantly different between the groups.

Table 2.  Profiles of patients and normal control subjects for the protein samples
  1. Data indicate the mean ± SD of age (year), storage time (month) and PMI (hour). FC, prefrontal cortex; HIP, hippocampus; OC, occipital cortex. **Significantly different from controls (p < 0.01. Mann–Whitney U-test). Control subjects died with bleeding (5), heart failure (5), cerebral infarct (3), myocardial infarct (3), pneumonia (2), gastric cancer (2) and others (21).

  n 25 14
 FCAge 62.8 ± 13.5 62.5 ± 12.6
  PMI  5.2 ± 3.6  7.0 ± 6.6
  storage time107.4 ± 66.7112.8 ± 63.7
  n 16 21
ProteinHIPAge 64.5 ± 10.8 64.4 ± 13.5
  PMI  3.7 ± 1.4 15.1 ± 10.2**
  storage time 76.1 ± 60.2145.9 ± 49.8**
  n 20 13
 OCAge 58.8 ± 12.1 64.5 ± 15.0
  PMI  6.6 ± 5.0 14.2 ± 9.6
  storage time143.8 ± 58.4172.0 ± 8.6
Figure 1.

PDZ protein levels in the prefrontal cortex of schizophrenic patients. Bar graphs show the variability and differences in the SAP97, PSD-95, chapsyn-110, GRIP1, and SAP102 protein levels between schizophrenic and control subjects. Western blot analysis of the brain homogenates from the dorsolateral prefrontal cortex was performed with anti-SAP97, anti-PSD-95, anti-chapsyn-110, anti-GRIP1, and anti-SAP102 antibodies (see details in Materials and methods). A neuronal cytoplasmic marker, NSE, was examined as a control. Results of densitometric measurements are shown on the bottom as means ± SEM of values obtained (b). Only the difference in SAP97 levels was statistically significant (n = 25 for control, n = 14 for patients; *p < 0.05). (a) Blots with the 10 highest NSE intensity levels in each group, were selected as examples and arranged from the left in the order of SAP97, SAP102, and NSE levels for display. SCZ, schizophrenics; CON, controls.

PDZ protein levels were also analyzed in the hippocampus and occipital cortex (Fig. 2). In the hippocampus, there was no significant decrease in SAP97 levels (U16,21 = 148, Z = −0.61, p = 0.54). The SAP102 levels, however, were significantly lower in the schizophrenic group (U16,21 = 238, Z = 2.15, p = 0.032). Expression of the other PDZ proteins was not changed in this brain region. However, SAP102 levels in the hippocampus were positively correlated with storage time (rs = −0.36, Z = −2.16, p = 0.03 for control plus schizophrenic groups, n = 37) as well as post-mortem interval (rs = −0.45, Z = −2.68, p = 0.007 for control plus schizophrenic groups, n = 37). Thus, the difference in storage time might be responsible for the changes in SAP102. In the occipital cortex there were no significant differences in the PDZ protein levels between groups (Fig. 3).

Figure 2.

PDZ protein levels in the hippocampus. Bar graphs display hippocampal SAP97, PSD-95, chapsyn-110, GRIP1, and SAP102 protein levels between the schizophrenic and control subjects. Protein samples from all hippocampal subregions containing CA1-4 regions and dentate gyrus were subjected to immunoblotting with anti-SAP97, anti-PSD-95, anti-chapsyn-110, anti-GRIP1, and anti-SAP102 antibodies. A neuronal cytoplasmic marker, NSE, was used as a control. Results of densitometric measurements are shown on the bottom as means ± SEM of values obtained (b). SAP102 levels significantly reduced (n = 16 for control, n = 21 for patients; *p < 0.05). (a) Blots with the 10 highest NSE intensity levels in each group were chosen as examples and arranged from the left in the order of signal intensities for SAP97, SAP102, and NSE for display. SCZ, schizophrenics; CON, controls.

Figure 3.

PDZ protein levels in the occipital cortex. SAP97, PSD-95, chapsyn-110, GRIP1, and SAP102 protein levels, as well as those of NSE in the occipital cortex were determined by western blotting of schizophrenic and control brains. Results of densitometric measurements are shown as means ± SEM of values obtained. There were no statistically significant differences in PDZ protein levels (n = 20 for control, n = 13 for patients). SCZ, schizophrenics; CON, controls.

SAP97 interacts with the AMPA receptor subunit, GluR1 (Kornau et al. 1995; Srivastava et al. 1998; Xia et al. 1999). To examine the influence of the decrease in SAP97 on this glutamate receptor subunit, we performed western blotting on GluR1 in the prefrontal cortex where SAP97 levels were altered in schizophrenic brains. The Mann–Whitney U-test revealed that GluR1 protein levels were reduced in the prefrontal cortex of schizophrenic patients (U25,14 = 253, Z = 2.28, p = 0.022) (Fig. 4). In contrast, AMPA receptor protein levels did not significantly change in the occipital cortex (U20,13 = 83, Z = −1.73, p = 0.083), indicating the regional specificity of the changes. Consistent with reported interaction between GluR1 and SAP97, Spearman's rank correlation revealed a positive correlation between GluR1 levels and SAP97 levels in the prefrontal cortex of patients (rs = 0.63, Z = 2.29, p = 0.02, n = 14) as well as in controls (rs = 0.44, Z = 2.14, p = 0.03, n = 25) (Fig. 5). In the schizophrenic subjects, there was a sharper increase of SAP97 against GluR1 than in the control group. In addition, the correlation line only for the schizophrenic group passes near the origin. The occipital cortex exhibited a similar trend in the correlation between GluR1 and SAP97, although the intercept of the ordinate was similar for both groups. GluR1 protein in the prefrontal cortex of schizophrenic patients appeared to have a proportional correlation with SAP97 because of their mutual interaction. In contrast, GluR1 levels in the prefrontal cortex of controls, as well as in the occipital cortex of the both groups, did not so depend on SAP97 levels, presumably involving an unidentified partner molecule of GluR1. Although it is unknown how this illness alters the molecular interaction of GluR1 protein, these results indicate that molecular interaction between the AMPA receptor and SAP97 qualitatively differed between these brain regions of schizophrenic patients.

Figure 4.

Decrease in GluR1 protein in the prefrontal cortex of schizophrenic patients. The AMPA receptor subunit GluR1, which binds to SAP97, was measured by western blotting in the same protein samples from the prefrontal and occipital cortices. Results of densitometric measurements are shown as means ± SEM of values obtained and compared between the groups by Mann–Whitney U-test (b) in the prefrontal cortex (PFC) and in the occipital cortex (OC) (*p < 0.05). (a) The same blotted lanes with the 10 highest NSE intensity levels were selected and arranged from the left in the order of GluR1 intensities for display. SCZ, schizophrenics; CON, controls.

Figure 5.

Correlation between the AMPA receptor subunit and SAP97 protein levels in the prefrontal and occipital cortex. GluR1 and SAP97 protein levels in the prefrontal cortex (PFC) and occipital cortex (OC) are plotted on the ordinate and the abscissa, respectively. Their correlation was analyzed using Spearman's rank correlation test. They were both positively correlated in the prefrontal cortex (rs = 0.44, p = 0.032, Z = 2.14, n = 25 for control and rs = 0.64, p = 0.022, Z = 2.29, n = 14 for schizophrenia) and in the occipital cortex of control subjects (rs = 0.61, p = 0.007, Z = 2.66, n = 20), but not in this region of schizophrenic subjects (rs = 0.53, p = 0.068, Z = 1.83, n = 13). The best fit line for the correlation in the control group is shown with a broken line (y = 0.204x + 91.2 for the prefrontal cortex and y = 0.272x + 82.9 for the occipital cortex) and that in the schizophrenic group with a thick line (y = 1.17x + 14.9 for the prefrontal cortex and y = 1.04x + 90.6 for the occipital cortex).

Abnormal PDZ protein levels correlated with schizophrenia, but neuroleptic treatment might be a potential cause of the changes. Therefore, to assess any influence of haloperidol, a typical neuroleptic, which was preferentially administered to the patients, the drug was given to adult rats for 2 weeks (Fig. 6). SAP97 and SAP102 protein levels as well as those of NSE were measured in two brain regions, hippocampus and frontal cortex. Haloperidol had no effect on the PDZ protein levels nor on NSE in the two brain regions examined. Therefore, the alteration in the PDZ protein levels in schizophrenia does not seem to be a consequence of neuroleptic treatment.

Figure 6.

Effects of haloperidol treatment on SAP97 and SAP102 protein levels in rat frontal cortex and hippocampus. (a) SAP97 and SAP102 protein levels were measured in the frontal cortex (FCTX) and hippocampus (HIP) of the right hemisphere of rats that were treated with haloperidol (HPD) (1.5 mg/kg/day) or vehicle (CON) for 14 days. (b) No significant difference was observed by Mann–Whitney U-test between haloperidol-treated and control rats (n = 5 rats each).


More than 20 PDZ proteins have been identified in the central nervous system thus far. The PDZ proteins are implicated in transmitter receptor dynamics and activity at individual synapses (Nagano et al. 1998; O'Brien et al. 1998; Weickert and Kleinman 1998; Fanning and Anderson 1999; Scannevin and Huganir 2000). Therefore, impaired expression of the PDZ proteins might result in abnormal postsynaptic function. The present study attempted to evaluate the contribution of the PDZ proteins to schizophrenic pathology. Each PDZ protein binds to neurotransmitter receptor(s) as well as to various signaling molecules to play particular postsynaptic roles. For example, Pick 1 and GRIP1, which both interact with the C-terminal terminal of the AMPA receptor subunits, GluR2/3, are involved in their internalization and synaptic stabilization (Dong et al. 1999; Dev et al. 1999; Chung et al. 2000; Daw et al. 2000; El-Husseini et al. 2000; Osten et al. 2000). SAP97 interacts with the AMPA receptor subunit GluR1 and stabilizes receptor protein expression at postsynaptic sites (Leonard et al. 1998; Jourdi et al. 2001; Sans et al. 2001). Among the various PDZ proteins, levels of the PDZ protein SAP97 were specifically decreased in the prefrontal cortex of schizophrenics (59%, p < 0.05) compared to those in controls in the present experiment. In parallel, GluR1 protein levels were significantly reduced in the same brain region. SAP97 protein levels positively correlated with AMPA-type glutamate receptor levels but their correlation qualitatively differed in schizophrenic group. Therefore, the SAP97 reduction in the prefrontal cortex is likely to contribute to the impaired expression of the AMPA receptor protein and resultant glutamatergic dysfunction in this brain region. These findings lend support to the hypothesis and previous reports of hypofunction of the prefrontal cortex in schizophrenic patients, which presumably involves impaired excitatory neurotransmission (Duncan et al. 1999; Bressan and Pilowsky 2000).

To control for the use of various post-mortem samples as well as neurochemical changes induced by drug treatment, we paid attention to quality of post-mortem samples (Harrison 1999). The use of samples from schizophrenic patients with a long storage time or with a long post-mortem interval was unavoidable because of the limited number of samples available. SAP102 levels, but not SAP97 levels, were correlated with storage time and post-mortem interval. Thus, the reduction in SAP102 protein levels in the hippocampus might be due to its degradative change during post-mortem interval and storage. Possible influences of drug treatment were assessed in animal experiments: Treatment of a typical neuroleptic failed to alter concentrations of various PDZ proteins in rats. This finding is also supported by independent animal data demonstrating that chronic treatment of rats with haloperidol does not decrease, but rather increases, AMPA receptor levels (Meador-Woodruff et al. 1996; Sokolov 1998).

Previous post-mortem studies also report a decrease in AMPA receptor subunit protein levels in the prefrontal cortex of schizophrenic patients (Eastwood et al. 1997; Kerwin and Harrison 1997). As discussed above, the PDZ protein, SAP97, is implicated in the expression of AMPA receptor subunits as well as in the stabilization of the AMPA receptor complexes (Jourdi et al. 2001; Sans et al. 2001). In agreement, our preliminary study revealed that other AMPA receptor subunits, GluR2/3, were also decreased in the same prefrontal samples (58% reduction, U25,14 = 249, Z = 2.16, p = 0.03). We previously found the down-regulation of the high-affinity BDNF receptor, TrkB, in corticolimbic regions of schizophrenic patients. As BDNF/TrkB signaling is a potent regulator for the expression of various PDZ protein (Jourdi et al. 2001; Sato et al. 2001), the impaired BDNF/TrkB signaling of schizophrenic patients might be responsible for the SAP97 protein decrease in the prefrontal cortex of schizophrenic patients. Interestingly, a major psychotomimetic drug, phencyclidine, acutely alters SAP97 gene expression, suggesting a potential link between psychotic symptoms and the PDZ protein SAP97 (Linden et al. 2001). Accordingly, the impaired protein expression of SAP97 and GluR1 in the prefrontal cortex might be associated with the hypofunction of the prefrontal cortex in schizophrenic patients revealed by the above neurochemical studies as well as by neuroimaging studies (Weinberger et al. 1986; Selemon et al. 1995; Andreasen et al. 1997; Duncan et al. 1999; Bressan and Pilowsky 2000).

One of the presynaptic markers, synaptophysin, is reported to decrease in the prefrontal cortex of schizophrenic patients (Eastwood et al. 1995; Honer et al. 1999). Not all presynaptic markers, however, appear to decline; protein levels of synapsin II are maintained in this region in schizophrenic patients (Karson et al. 1999; Imai et al. 2001). Similarly, we observed no significant alteration in other PDZ molecules, most of which are localized at postsynaptic sites. Therefore, selective reduction of SAP97 might reflect a difference in susceptibility of synaptic loss accompanying schizophrenia or its phenotypic down-regulation, although a definitive answer awaits ultrastructural analysis. Taken together, the postsynaptic abnormality observed in the prefrontal cortex of schizophrenic patients might provide a molecular basis for both the structural and neurochemical impairments in schizophrenia, although this etiologic argument is assumptive and remains to be examined experimentally. Future studies are needed to examine how postsynaptic dysfunction might contribute to the pathophysiology of schizophrenia.


We thank Mrs E. Higuchi for technical assistance. This work was supported by the Japanese Society for the Promotion of Science (RFTF-96L00203) and Grant-in-Aid for Creative Scientific Research and for Basic Science Research.