Abnormal Expression of Brain-Derived Neurotrophic Factor and Its Receptor in the Corticolimbic System of Schizophrenic Patients

Home , Neurochemical

Molecular Psychiatry (2000) 5, 293–300  2000 Macmillan Publishers Ltd All rights reserved 1359-4184/00 $15.00 www.nature.com/mp ORIGINAL RESEARCH ARTICLE Abnormal expression of brain-derived neurotrophic factor and its receptor in the corticolimbic system of schizophrenic patients M Takahashi1,2, O Shirakawa3, K Toyooka1, N Kitamura3, T Hashimoto3, K Maeda3, S Koizumi4, K Wakabayashi5, H Takahashi5, T Someya2 and H Nawa1

1Molecular Neurobiology, Brain Research Institute, Niigata University, 951–8585, Japan; 2Department of Psychiatry, Niigata University School of Medicine, 951–8510, Japan; 3Department of Psychiatry and Neurology, Kobe University School of Medicine, 650–0017, Japan; 4Novartis Pharma, Takarazuka, Hyogo Pref, 665–8666, Japan; 5Pathology & Brain Disease Research Center, Brain Research Institute, Niigata University, 951–8585, Japan

Previous neuropathological studies have revealed that the corticolimbic system of schizophrenic patients expresses abnormal levels of various synaptic molecules, which are known to be influenced by the neuronal differentiation factors, neurotrophins. Therefore, we determined levels of neurotrophins and their receptors in the postmortem brains of schizophrenic patients and control subjects in relation to molecular impairments in schizophrenia. Among the neurotrophins examined, levels of brain-derived neurotrophic factor (BDNF) were elevated specifically in the anterior cingulate cortex and hippocampus of schizophrenic patients, but levels of nerve growth factors and neurotrophin-3 showed no change in any of the regions examined. In parallel, the expressions of TrkB receptor and calbindin-D, which are both influenced by BDNF, were reduced significantly in the hippocampus or the prefrontal cortex. How- ever, neuroleptic treatment did not appear to mimic the neurotrophic change. Neither with- drawal of drug treatment in patients nor chronic administration of haloperidol to rats altered levels of BDNF. These findings suggest that neurotrophic abnormality is associated with the corticolimbic structures of schizophrenic patients and might provide the molecular substrate for pathological manifestations of the illness. Molecular Psychiatry (2000) 5, 293–300. Keywords: neurotrophin; BDNF; TrkB; calbindin-D; development; haloperidol; postmortem brain

Introduction to the aberrant synaptic transmission of the mature brain, both of which seem to be comprised in the brain Abnormal structural and neurochemical development, dysfunction of schizophrenics. To assess potential con- resulting in abnormal neural circuitry and function, tribution of BDNF to such pathological phenotypes of has been implicated in schizophrenia.1,2 Recently, a schizophrenic patients, we quantitated levels of BDNF, family of neurotrophic factors, comprising nerve together with other neurotrophins, NGF and NT-3, growth factor (NGF), BDNF, neurotrophin-3 (NT-3) and using two-site enzyme immunoassay (EIA)16,17 and neurotrophin-4/5, have been suggested to be potent attempted to correlate these findings with the regulators of neuronal phenotypic differentiation and expressions of BDNF receptors (TrkB) and calbindin D, synaptic plasticity in mammals.3–6 In addition, there is which may reflect the activity of BDNF in patients.18,20 growing evidence that these neurotrophins are involved in the cerebral structural organization such as the formation of the ocular dominance column, the Materials and methods whisker representation pads and neocortical layers.7–9 Human tissue sampling Together with the findings that BDNF regulates pheno- Brain tissue was collected posthumously from 18 typic expressions of brain neurons,10–15 these obser- patients with chronic schizophrenia (10 men, eight vations imply that abnormal signals of neurotrophins women; mean age: 61.6 years; SD: 15.2) and from 36 could lead to the malformation of cortical circuitry or controls (23 men, 13 women, mean age: 59.4 years; SD: 12.9) who had no history of neuropsychiatric disorders (Table 1). All the schizophrenics had taken typical Correspondence: H Nawa, Department of Molecular Biology, antipsychotics for prolonged periods, but six of them Brain Research Institute, Niigata University, Asahimachi-dori 1– had been off medication for at least 3 months before 757, Niigata 951–8585, Japan. Present address of N Kitamura, Kobe City Central Hospital, death. The diagnoses of schizophrenia corresponded to Kobe, Japan. the DSM-IV categories (American Psychiatric Associ- Received 6 October 1999; revised and accepted 6 December 1999 ation, 1994). The prefrontal cortex was sampled from Neurotrophins and schizophrenia M Takahashi et al 294 Table 1 Autopsy and clinical data

Sex/age Cause of death Subtype Duration (year) Medication PMI PFC ACC HIP OCC (years) (DSM- dose − IV)illness medication (mg day 1)

Schizophrenia S-1 M/23 Suffocation 295.1 1 1 Lev (40), 23 * * * Tri (30) S-2 F/40 Drowning 295.1 19 16 Hal (2.25), 24 * * Chl (12.5) S-3 F/44 Heart failure 295.1 15 15 Thr (30) 17 * * S-4 M/49 Heart failure 295.3 26 11 Hal (4.5) 5 * S-5 M/53 Heart failure 295.1 36 18 Off-drug 5 **** S-6 M/58 Gastric cancer 295.9 31 4 Hal (3) 19 * S-7 M/61 Heart failure 295.1 38 27 Chl (40) 9 * * S-8 F/63 Heart failure 295.9 10 10 Off-drug 4 * * * S-9 F/63 Gastric cancer 295.1 35 23 Hal (4) 6 * S-10 F/65 Panperitonitis 295.3 16 9 Tht (60) 15 **** S-11 M/66 Gastric cancer 295.1 45 35 Off-drug 9 **** S-12 M/70 Heart failure 295.3 35 26 Thr (75), 22 **** Oxy (60) S-13 F/72 Pneumonia 295.3 34 21 Chl (100) 3 **** S-14 F/72 Pneumonia 295.9 37 21 Off-drug 18 * * * S-15 F/73 Myocardial infarction 295.3 27 27 Off-drug 16 **** S-16 M/74 Unknown 295.3 21 21 Hal (2.25), 7 * Lev (100) S-17 M/80 Heart failure 295.1 29 25 Pro (30) 36 * * * S-18 M/83 Decrepitude 295.1 54 15 Off-drug 5 * * * Control C-1 M/35 Bleeding WND None 17 * * C-2 M/42 Bleeding WND None 4 * * C-3 F/42 Fresh muscle necrosis WND None 6.5 * * C-4 F/44 Gorham’s disease WND None 6 **** C-5 M/45 Cardiac hypertrophy WND None 11.5 * C-6 M/45 Cerebral infarct WND None 2 * * C-7 M/47 Myositis WND None 2 * * C-8 F/47 Bleeding WND None 10 * * C-9 F/48 Myasthenia gravis WND None 3 * * * C-10 M/49 Falling WND None 2 * * * C-11 F/49 Myopathy WND None 3 **** C-15 F/50 Castleman’s disease WND None 3 * C-13 M/51 Pontine hemorrhage WND None 5 * * C-14 M/55 Suffocation WND None 6 * * C-15 F/57 Heart failure WND None 2 * * C-16 M/59 Pulmonary tuberculosis WND None 9 * C-17 M/59 Myeloma WND None 2 **** C-18 M/59 Cerebral hemorrhage WND None 4 * * C-19 F/59 Myopathy WND None 3 * * * C-20 M/61 Amyloid neuropathy WND None 7 * C-21 M/61 Lung cancer WND None 2 * C-22 M/62 Bleeding WND None 2 * C-23 M/65 Pulmonary infarction WND None 6 * * * C-24 F/65 Heart failure WND None 4 * * * C-25 M/68 Cerebral infarct WND None 3 **** C-26 F/69 Heart failure WND None 22 * * C-27 M/70 Heart failure WND None 3 * * C-28 M/70 Gastric cancer WND None 5 * C-29 M/72 Myasthenia gravis WND None 3 * * * C-30 M/73 Cerebral infarct WND None 3 * * C-31 F/73 Neuropathy WND None 4 **** C-32 F/76 Retroperitoneal malignant WND None 5 **** tumor

Continued

Molecular Psychiatry Neurotrophins and schizophrenia M Takahashi et al 295 Table 1 Continued

Sex/age Cause of death Subtype Duration (year) Medication PMI PFC ACC HIP OCC (years) (DSM- dose − IV)illness medication (mg day 1)

C-33 M/77 Gastric cancer WND None 6 * * * C-34 M/81 Cardiac tamponade WND None 6 * C-35 F/82 Renal failure WND None 10 * * C-36 M/84 Decrepitude WND None 13 * * *

M, male; F, female. 295–1, disorganized type; 295–3, paranoid type; 295–9, undifferentiated type; WND, without neuropsychiatric disease. Lev, levomepromazine; Tri, trimipramine; Hal, haloperidol; Chl, chlorpromazine; Thr, thioridazine; Tht, thiotixene; Oxy, oxy- pertine; Pro, propericiazine; Off-drug, no neuroleptic treatment for at least 3 months before death. PMI, postmortem interval (h). PFC, prefrontal cortex; ACC, anterior cingulate cortex; HIP, hippocampus; OCC, occipital cortex. Asterisk indicates available subjects in each region. the dorsolateral prefrontal cortex (BA46) and the Enzyme immunoassay anterior cingulate from the area BA24. The hippocam- Neurotrophin levels were measured by two-site pal tissue taken did not include the surrounding enzyme immunoassay (EIA) as described pre- entorhinal and parahippocampal areas. All the anatom- viously.16,17 In brief, tissue extracts (10–100 ␮l; ical orientation was based on the Human Brain Atlas.19 duplicate) or neurotrophin standards (1–300 pg; The families of the control and schizophrenic subjects triplicate) were loaded into the wells of EIA titer plates provided written informed consent for the use of brain that had been coated with 100 ng well−1 anti-BDNF, tissues as materials for the investigations. For determi- anti-NGF or anti-NT-3 antibody (primary antibody). nation of neurotrophin concentrations, tissues were Biotinylated anti-BDNF, anti-NGF or anti-NT-3 anti- homogenized in 20–50 vol homogenization buffer body (secondary antibody; 10–30 ng ml−1) was added [50 mM Tris, 0.3 M NaCl, 0.3% Triton X-100, 1% bov- to the wells, and then bound biotinylated secondary ine serum albumin (BSA; Corn Fraction V), pH 7.5] antibodies were detected using streptavidin-␤-galacto- containing protease inhibitors [aprotinin (200 kalli- sidase (1:10 000; Sigma). The activity of ␤-galactosidase krein units ml−1), 0.1 mM phenylmethanesulfonyl flu- retained in each well was measured by incubation with oride, 0.1 mM benzethonium chloride and 1 mM 200 mM 4-methylumbelliferyl-␤-d-galactosidase benzamidine (Sigma, St Louis, MO, USA)].20 Homogen- (MUG, Sigma). The amount of the fluorescent product ates were centrifuged for 20 min at 14 000 × g at 4°C, was monitored by a Fluorolite-1000 fluorometer and supernatants were stored at −80°C until use. The (Dynatech Laboratories, Chantilly, VA, USA) with exci- protein contents of samples were determined using a tation at 364 nm and emission at 448 nm. An average Micro BCA kit (Pierce, Rockland, IL, USA) with bovine of two measurements of a sample was standardized by serum albumin as a standard. All the experiments were its protein concentration. done under the authorization of the Niigata University Ethics Committee. Table 2 Statistical analysis of patient population

Neuroleptic treatment of animals Brain area Cases Control Schizophrenia Male Wistar rats (initial weight 210–270 g) were housed in groups of three or four on a 12-h light-dark ACC n 22 14 cycle with free access to food and water. Different Age 62.2 ± 12.6 61.8 ± 16.6 groups of rats were treated orally with either haloperi- PMI 5.3 ± 4.6 14.7 ± 9.5** dol (target dose, 1.5 mg kg−1 day−1) or vehicle as drink- PFC n 25 9 ing water for 28 days. Haloperidol was dissolved in a Age 60.2 ± 14.1 61.9 ± 15.9 minimal amount of acetic acid, diluted with distilled PMI 5.1 ± 3.6 12.8 ± 7.7* water, and pH-balanced with NaOH to give a final HIP n 13 13 ± ± drinking solution of 0.025 mg ml−1, pH 6.5. This treat- Age 61.6 11.0 64.0 13.6 PMI 3.5 ± 1.5 13.1 ± 9.8** ment regimen resulted in plasma haloperidol concen- ± −1 = OCC n 23 13 trations of 4.8 0.8 ng ml (n 6) after 28 days, equiv- Age 58.8 ± 13.3 64.5 ± 15.0 alent to the therapeutic range for haloperidol in PMI 6.8 ± 5.2 14.2 ± 9.6* humans. Brains were removed rapidly from decapi- tated rats, and each brain region was dissected on ice. Data indicate the mean ± SD of age and PMI for the cases All the animal experiments were done according to the listed in Table 1. Animal Use and Care Committee guidelines of the Nii- *Significantly different from controls (ANOVA, P Ͻ 0.05). gata University. **Significantly different from controls (ANOVA, P Ͻ 0.01).

Molecular Psychiatry Neurotrophins and schizophrenia M Takahashi et al 296

Figure 1 Neurotrophin levels in the brain of schizophrenic patients. Scatterplots display the variability and differences in the levels of BDNF, NGF and NT-3 between the schizophrenic and control subjects. Signiﬁcantly different from controls (**P Ͻ 0.01, ANOVA with age as a covariate,***P Ͻ 0.001, one-way ANOVA). Clear diamonds, neuroleptic-withdrawn schizophrenics; solid diamonds, neuroleptic-treated schizophrenics; clear circles, controls; SCZ, schizophrenics; CON, controls; PFC, prefrontal cortex; ACC, anterior cingulate cortex; HIP, hippocampus; OCC, occipital cortex. Horizontal solid or dashed line represents a mean of values. Note: A sample, whose raw measurements varied signiﬁcantly in a duplicated EIA, was not adopted as data. This is the reason why the number of samples differs among neurotrophins.

Molecular Psychiatry Neurotrophins and schizophrenia M Takahashi et al 297 Western blot analysis subjects. Thus, the difference in PMI appeared not to Relative levels of TrkB, calbindin-D and NSE were contribute to the BDNF increase (see details in determined by Western blot analysis using methods Discussion). However, age was a contributor to the similar to those described previously.11 Protein variances of the BDNF levels in the frontal and anterior samples for Western blotting were prepared from the cingulate cortices of schizophrenics (frontal, ␤ =−0.79, same tissue extracts for the two-site EIAs by denaturing P = 0.02; anterior cingulate, ␤ =−0.52, P = 0.04), them with 2% sodium dodecyl sulfate (SDS), separated increasing age being associated with a reduction in by SDS-polyacrylamide gel electrophoresis (PAGE), BDNF levels, but not in the controls (frontal, ␤ =−0.34, and transferred to a nitrocellulose membrane P = 0.10; anterior cingulate, ␤ =−0.12, P = 0.66). Age (Schleicher and Schull, Dassel, Germany) by electro- was therefore included as a covariate in analysis of phoresis. The membrane was probed with anti-TrkB variance (ANOVA) in the above analysis. Even though antibody (Novaris Japan, 1:1000), anti-calbindin-D the BDNF abnormality might indeed be correlated with antibody (Sigma, 1:1000) or anti-NSE antibody schizophrenia, neuroleptic treatment was a potential (Chemicon, Temecula, CA, USA; 1:2000). After exten- cause of the abnormality, and needed to be evaluated sive washing, the immunoreactivity on the membrane in the present study. To assess any influence of halo- was detected with anti-rabbit/mouse immunoglobulin peridol, a typical neuroleptic, which was preferentially conjugated to horseradish peroxidase followed by administered to the patients, the drug was given chron- chemiluminescence reaction (ECL kit; Amersham, UK). ically to adult rats (Figure 2). BDNF levels were meas- Densitometric quantitation of the band intensities was ured in the three brain regions, hippocampus, cingu- done, and each value was expressed as a percentage of late cortex and frontal cortex. Haloperidol had no effect the standard samples. on BDNF levels in all the brain regions examined. Chronic application of BDNF has been reported to Statistical analysis down-regulate its high-affinity receptor TrkB in brain The potential confounding influences of age, sex and neurons and desensitize further BDNF signaling.18 To PMI upon neurotrophin and other protein levels were determine levels of TrkB protein, we carried out West- investigated using multiple regression. Changes in neu- ern blot analysis for the hippocampus as well as for rotrophin and other protein abundance associated with the prefrontal cortex. Expression of TrkB protein was schizophrenia were then investigated by analysis of detected as a specific 145-kDa band, consistent with variance (ANOVA) with these factors as covariates the apparent molecular size of non-truncated TrkB where necessary and their correlation was examined by receptors. TrkB immunoreactivity in the hippocampus the Pearson’s product moment (using SPSS 6.1J, SPSS was significantly lower in the schizophrenics than in Japan Inc, Tokyo, Japan). P Ͻ 0.05 was considered stat- the controls (43.9% decrease, P Ͻ 0.05, ANOVA) istically significant. All values are presented as (Figure 3). Consistent with this finding, correlation means ± SEM. analysis by the Pearson’s product moment method

Results Postmortem samples of these brain regions, and the occipital cortex as a control, were obtained randomly from 18 chronic schizophrenics (classified by DSM-IV), and from 36 age-matched controls who had no history of neuropsychiatric disorders (Table 1). There was a sampling difference of postmortem intervals (PMI) between the two groups, but all the other indices were similar (Table 2). We analyzed the difference in neurotrophin levels between the two groups and found that BDNF levels were significantly increased in the hippocampus and anterior cingulate cortex, but not in the frontal cortex or occipital cortex, of the schizophrenics in comparison with controls (Figure 1a). These differences were also shown to be significant by Friedman Figure 2 Influence of haloperidol on BDNF levels in rat nonparametric analysis (data not shown). NGF levels brain. Effects of chronic haloperidol treatments on BDNF tended to be increased in the hippocampus of the expression in adult rats. Protein levels of BDNF were meas- schizophrenics but not to a significant extent (P = 0.07, ured in the prefrontal cortex (PFC), cingulate cortex (CC) and ANOVA) (Figure 1b). NT-3 levels did not vary signifi- hippocampus (HIP) of the right hemisphere of rats that had been treated with haloperidol (1.5 mg kg−1 day−1) (diamonds) cantly between the two groups, although those in the or vehicle (circles) for 28 days. Haloperidol concentrations in hippocampus and anterior cingulate cortex were too their plasma were monitored and ascertained that those low for EIA detection, and not subjected to statistical reached the therapeutic range needed in humans (see analysis (Figure 1c). Multiple regression analysis Materials and Methods). No significant difference was revealed that PMI did not affect neurotrophin levels in observed by ANOVA between haloperidol-treated and control any brain region in either the schizophrenics or control rats (n = 6 each).

Molecular Psychiatry Neurotrophins and schizophrenia M Takahashi et al 298

Figure 3 Impaired expressions of TrkB receptors and calbindin-D. Western blot analysis of the brain homogenates from the hippocampus of schizophrenics and controls, which were used in EIA. Protein was separated by SDS/PAGE on a 7.5% gel, transferred to a nitrocellulose membrane, and incubated with anti-TrkB, anti-calbindin D and anti-NSE antibodies, respectively. The immunoreactive bands from each of the subjects were arranged from the left in the order of their BDNF levels, with arrows showing the positions to which size markers migrated. Results of densitometric measurements are shown on the right side as means with SEM of values obtained. The same results were obtained with another Western blot (data not shown). *Signiﬁcantly different from controls (P Ͻ 0.05, ANOVA). SCZ, schizophrenics; CON, controls. Note: A sample in a control group for TrkB and that for calbindin D as well as that in a schizophrenic group for calbindin D were deleted from analysis and display because of dirt on the blots.

revealed that TrkB immunoreactivity was correlated TrkB receptors, confirmed the impairment of the inversely with the level of BDNF when all the values BDNF/TrkB activity in schizophrenia. were plotted (r =−0.44, P Ͻ 0.05, F[1, 23 = 5.58]) (Figure 4). However, similar reduction in TrkB levels Discussion was found in the prefrontal cortex as well. The prefrontal area of schizophrenic patients contained a sig- Using two-site EIAs, we demonstrated that BDNF lev- nificant lower level (43 ± 23%) of TrkB protein, com- els were significantly elevated in the hippocampus pared to the control (100 ± 18%) (P Ͻ 0.05, F[1, (2.3-fold increase, P Ͻ 0.001) and anterior cingulate 16] = 8.00), although the statistical correlation between cortex (1.7-fold increase, P Ͻ 0.01) of schizophrenics TrkB and BDNF levels was not apparent in this region. compared to those in controls. Previous pathological To investigate further the deficit in the BDNF/TrkB examinations have indicated that several changes in activity in schizophrenics, we measured the protein molecular markers appear to represent the effects of levels of calbindin-D, whose expression is maintained neuroleptic treatment rather than schizophrenic path- and up-regulated by BDNF11,20 Immunoblots revealed ology itself.21,22 Thus, one of the major concerns in the the presence of a specific 28-kDa band, corresponding present postmortem study had been the possibility that to the authentic size of calbindin-D (Figure 3). Levels the BDNF increase in schizophrenic patients might of calbindin-D in the hippocampus of schizophrenics reflect neurochemical changes that were induced by were significantly decreased in comparison with con- medical treatment. However, there was no clear dis- trols (37.9% decrease, P Ͻ 0.05). Expression of neuron- sociation of BDNF distribution between neuroleptic- specific enolase (NSE) was measured in parallel as a treated and neuroleptic-withdrawn patients in the control neuronal marker. There was no significant dif- present study (see Figure 1a). Moreover, chronic treat- ference in the levels of NSE, suggesting the specificity ment of a typical neuroleptic failed to alter BDNF proof the phenotypic changes. The observed decline in tein concentrations in the hippocampus of rats. This calbindin-D protein, as well as the down-regulation of view is also supported by independent animal data

Molecular Psychiatry Neurotrophins and schizophrenia M Takahashi et al 299 the hippocampal region evaluated by positron emission tomography varies considerably in schizophrenic patients depending on their symptom profiles,26,27 pro- viding no consistent findings about functional activi- ties in this brain region. Neurochemical studies have rather suggested hypofunction of glutamatergic transmission both in schizophrenic models and in patients.28 In agreement, the psychostimulants, phen- cyclidine and ketamine, both of which block excitatory NMDA-type glutamate channels, decrease BDNF mRNA expression in the hippocampus of rats and reduce regional cerebral blood flow in this region of the patients, respectively.29,30 These observations suggest that upregulation of BDNF in schizophrenia can- not simply be explained by local neuronal excitability. Figure 4 Negative correlation between levels of BDNF and The activation of TrkB enhances phenotypic differ- TrkB in the hippocampus. BDNF levels determined by EIA entiation of pyramidal and non-pyramidal neurons, were correlated inversely with levels of TrkB protein meas- inducing expression of a GABA-synthesizing enzyme ured by immunoblotting in the hippocampus of the same (GAD67),11 neuropeptides,10,11 calbindin-D.10,18 synap- individuals (n = 23, r =−0.44, P Ͻ 0.05). Clear diamonds, neu- tophysin13 and the AMPA receptor subunits.15 Interest- roleptic-withdrawn schizophrenics; solid diamonds, neuro- ingly, reduced expression of all these molecules is leptic-treated schizophrenics; clear circles, controls. often associated with schizophrenia31–37 and may be ascribed to the down-regulation of TrkB receptors. Recently, another phenotypic marker of GABAergic neurons has been investigated:38 decreased expression of reelin has been found in schizophrenic patients.39 demonstrating that chronic treatment of rats with halo- In fact, these pathological changes quite resemble the peridol does not increase, but rather decreases the level phenomena seen in a reeler mouse.40–42 This mouse of BDNF mRNA in the limbic system.45 These obser- carries a mutated gene for reelin and displays vations indicate that the increase in BDNF protein in cortical/hippocampal dysplasia.43,44 It is especially the hippocampus and anterior cingulate cortex is likely noteworthy that overexpression of BDNF in transgenic to represent a molecular pathological feature of the dis- mice suppresses expression of reelin during brain ease itself, rather than an effect of neuroleptics. development and induces aberrant cortical lami- There was a difference in PMI between the two nation.9 Such a scenario may be important in the brain groups, but several lines of evidence indicate that our malformation found in schizophrenic patients. Taken finding is unlikely to be the result of prolonged PMI in together, the BDNF/TrkB abnormality observed in schizophrenics. Multiple regression analysis revealed schizophrenic patients might provide a molecular basis that PMI did not affect neurotrophin levels in any brain for both the structural and phenotypic impairments region in either the schizophrenics or control subjects. seen in this disease, although this etiological argument Moreover, it is quite unlikely that the longer PMI for is assumptive and remains to be examined experimen- schizophrenic samples would have led to a region-spe- tally. Future studies should provide an answer to the cific increase in levels of BDNF, but not in those of question of when neurotrophic dysfunction appears other neurotrophins. Moreover, it has been demon- and leads to neurochemical or anatomical defects in strated that PMI (up to 6 h) has no influence on measur- the corticolimbic structures and how it contributes to able neurotrophin contents.23 Accordingly, expression the emergence of schizophrenic symptoms. of high-affinity BDNF receptor, TrkB, is reduced in the corticolimbic structures of schizophrenics and correlated inversely with BDNF levels in their hippocam- Acknowledgements pus. Although it is uncertain whether the reduction of We thank Sumitomo Pharmaceuticals for supplying the TrkB receptor resulted from the enhanced release of human recombinant BDNF, Dr R Kuwano for anti-cal- BDNF in patients or reflected a substantial defect of its bindin-D antiserum, Dr N Nishino and Mr H Jourdi for regulation in this disease, this evidence for TrkB recep- advice and Mrs K Hatano for technical assistance. This tor down-regulation in schizophrenia confirms that work was supported by the Japanese Society for the neurotrophic signaling of BDNF is indeed impaired Promotion of Science (RFTF-96L00203) and Grant-in- in patients. Aid for Scientific Research (MESSC). The expression of BDNF is influenced by various hormones, cytokines and neurotransmitters, among which the strongest positive regulator is suggested to References 24,25 be an excitatory neurotransmitter, glutamate. Is 1 Weinberger DR. Implication of normal brain development for the hyperexcitability responsible for the local increase of pathogenesis of schizophrenia. Arch Gen Psychiatry 1987; 44: BDNF in schizophrenic patients? The neural activity of 660–669.

Molecular Psychiatry Neurotrophins and schizophrenia M Takahashi et al 300 2 Waddington JL. Schizophrenia: developmental neuroscience and enhance the expression of BDNF mRNA in cultured cerebellar pathobiology. Lancet 1993; 341: 531–536. granule cells. Mol Brain Res 1993; 18: 201–208. 3 Stoop R, Poo MM. Synaptic modulation by neurotrophic factors. 26 Tamminga CA, Thaker GK, Buchanan R, Kirkpatrick B, Alphs LD, Prog Brain Res 1996; 109: 359–364. Chase TN et al. Limbic system abnormalities identified in schizo- 4 Lewin GR, Barde YA. Physiology of the neurotrophins. Ann Rev phrenia using positron emission tomography with fluorodeoxyglu- Neurosci 1996; 19: 289–317. cose and neocortical alterations with deficit syndrome. Arch Gen 5 Schuman EM. Neurotrophin regulation of synaptic transmission. Psychiatry 1992; 49: 522–530. Curr Opin Neurobiol 1999; 9: 105–109. 27 Liddle PF, Friston KJ, Frith CD, Hirsch SR, Jones T, Frackowiak 6 Jankowsky JL, Patterson PH. Cytokine and growth factor involve- RSJ. Pattern of cerebral blood flow in schizophrenia. Br J Psychiatry ment in long term potentiation. Molec Cell Neurosci 1999; 14: 1982; 160: 179–186. 273–286. 28 Olney JW, Farber NB. Glutamate receptor dysfunction and schizo- 7 Cabelli RJ, Hohn A, Shatz CJ. Inhibition of ocular dominance col- phrenia. Arch Gen Psychiatry 1995; 52: 998–1007. umn formation by infusion of NT-4/5 or BDNF. Science 1995; 267: 29 Lahti AC, Holcomb HH, Medoff DR, Tamminga CA. Ketamine acti- 1662–1666. vates psychosis and alters limbic blood flow in schizophrenia. Neu- 8 Prakash N, Cohen CS, Frostig RD. Rapid and opposite effects of roreport 1995; 6: 869–872. BDNF and NGF on the functional organization of the adult cortex 30 Castren E, da Penha Berzaghi M, Lindholm D, Thoenen H. Differen- in vivo. Nature 1996; 381: 702–706. tial effects of MK-801 on brain-derived neurotrophic factor mRNA 9 Ringstedt T, Linnarsson S, Wagner J, Lendahl U, Kokaia Z, Arenas levels in different regions of the rat brain. Exp Neurol 1993; 122: E et al. BDNF regulates reelin expression and Cajal–Retzius cell 244–252. development in the cerebral cortex. Neuron 1998; 21: 305–315. 31 Harrison PJ, McLaughlin D, Kerwin RW. Decreased hippocampal 10 Nawa H, Pelleymounter MA, Carnahan J. Intraventricular adminis- expression of a glutamate receptor gene in schizophrenia. Lancet tration of BDNF increases neuropeptide expression in newborn rat 1991; 337: 450–452. brain. J Neurosci 1994; 14: 3751–3765. 32 Eastwood SL, McDonald B, Burnet PW, Beckwith JP, Kerwin RW, 11 Mizuno K, Carnahan J, Nawa H. Brain-derived neurotrophic factor Harrison PJ. Decreased expression of mRNAs encoding non-NMDA promotes differentiation of striatal GABAergic neurons. Dev Biol glutamate receptors GluR1 and GluR2 in medial temporal lobe neu- 1994; 165: 243–256. rons in schizophrenia. Mol Brain Res 1995; 29: 211–223. 12 Ahmed S, Reynolds BA, Weiss S. BDNF enhances the differen- 33 Akbarian S, Kim JJ, Potkin SG, Hagman JO, Tafazzoli A, Bunney tiation but not the survival of CNS stem cell-derived neuronal pre- WE Jr et al. Gene expression for glutamic acid decarboxylase is cursors. J Neurosci 1995; 15: 5765–5778. reduced without loss of neurons in prefrontal cortex of schizo- 13 Takei N, Sasaoka K, Inoue K, Takahashi M, Endo Y, Hatanaka H. phrenics. Arch Gen Psychiatry 1995; 52: 258–266. Brain-derived neurotrophic factor increases the stimulation-evoked 34 Eastwood SL, Burnet PW, Harrison PJ. Altered synaptophysin release of glutamate and the levels of exocytosis-associated proteins expression as a marker of synaptic pathology in schizophrenia. in cultured cortical neurons from embryonic rats. J Neurochem Neuroscience 1995; 66: 309–319. 1997; 68: 370–375. 35 Glantz LA, Lewis DA. Reduction of synaptophysin immunoreactiv- 14 Marty S, Berninger B, Carroll P, Thoenen H. GABAergic stimu- ity in the prefrontal cortex of subjects with schizophrenia. Regional lation regulates the phenotype of hippocampal interneurons and diagnostic specificity. Arch Gen Psychiatry 1997; 54: 660–669. through the regulation of brain-derived neurotrophic factor. Neu- 36 Gabriel SM, Davidson M, Haroutunian V, Powchik P, Bierer LM, ron 1996; 16: 565–570. Purohit DP et al. Neuropeptide deficits in schizophrenia vs Alzhei- 15 Narisawa-Saito M, Carnahan J, Araki K, Yamaguchi T, Nawa H. mer’s disease cerebral cortex. Biol Psychiatry 1996; 39: 82–91. Brain-derived neurotrophic factor regulates the expression of 37 Nemeroff CB, Youngblood WW, Manberg PJ, Prange AJ-Jr, Kizer AMPA receptor proteins in neocortical neurons. Neuroscience JS. Regional brain concentrations of neuropeptides in Huntington’s 1999; 88: 1009–1014. chorea and schizophrenia. Science 1983; 221: 972–975. 16 Korsching S, Thoenen H. Two-site enzyme immunoassay for nerve 38 Pesold C, Impagnatiello F, Pisu MG, Uzunov DP, Costa E, Guidotti growth factor. Meth Enzymol 1987; 147: 167–185. A et al. Reelin is preferentially expressed in neurons synthesizing 17 Nawa H, Carnahan J, Gall C. BDNF protein measured by a novel gamma-aminobutyric acid in cortex and hippocampus of adult rats. enzyme immunoassay in normal brain and after seizure: partial dis- Proc Natl Acad Sci USA 1998; 95: 3221–3226. agreement with mRNA levels. Eur J Neurosci 1995; 7: 1527–1535. 39 Impagnatiello F, Guidotti AR, Pesold C, Dwivedi Y, Caruncho H, 18 Frank L, Ventimiglia R, Anderson K, Lindsay RM, Rudge JS. BDNF Pisu MG et al. A decrease of reelin expression as a putative vulner- down-regulates neurotrophin responsiveness, TrkB protein and ability factor in schizophrenia. Proc Natl Acad Sci USA 1998; 22: TrkB mRNA levels in cultured rat hippocampal neurons. Eur J Neu- 15718–15723. rosci 1996; 8: 1220–1230. 40 Arnold SE, Hyman BT, Van HG, Damasio AR. Some cytoarchitectu- 19 Nieuwenhunhuys, R, Voogd J, van Huijzen CHR. The Human Cen- ral abnormalities of the entorhinal cortex in schizophrenia. Arch tral Nervous System—A Synopsis and Atlas. Springer-Verlag, Gen Psychiatry 1991; 48: 625–632. Berlin, 1988. 41 Conrad AJ, Abebe T, Austin R, Forsythe S, Scheibel AB. Hippocam- 20 Pappas IS, Parnavelas JG. Neurotrophins and basic fibroblast pal pyramidal cell disarray in schizophrenia as a bilateral phenom- growth factor induce the differentiation of calbindin-containing enon. Arch Gen Psychiatry 1991; 48: 413–417. neurons in the cerebral cortex. Exp Neurol 1997; 144: 302–314. 42 Akbarian S, Kim JJ, Potkin SG, Hetrick WP, Bunney WE Jr, Jones 21 Sokolov BP. Expression of NMDAR1, GluR1, GluR7, and KA1 glut- EG. Maldistribution of interstitial neurons in prefrontal white mat- amate receptor mRNAs is decreased in frontal cortex of ‘neurolep- ter of the brains of schizophrenic patients. Arch Gen Psychiatry tic-free’ schizophrenics: evidence on reversible up-regulation by 1996; 53: 425–436. typical neuroleptics. J Neurochem 1998; 71: 2454–2464. 43 D’Arcangelo G, Miao GG, Chen SC, Soares HD, Morgan JI, Curran 22 McAllister CG, van Kammen DP, Rehn TJ, Miller AL, Gurklis J, T. A protein related to extracellular matrix proteins deleted in the Kelley ME et al. Increases in CSF levels of interleukin-2 in schizo- mouse mutant reeler. Nature 1995; 374: 719–723. phrenia: effects of recurrence of psychosis and medication status. 44 Del Rio JA, Heimrich B, Borrell V, Forster E, Drakew A, Alcantara Am J Psychiatry 1995; 152: 1291–1297. S et al. A role for Cajal–Retzius cells and reelin in the development 23 Narisawa-Saito M, Wakabayashi K, Tsuji S, Takahashi H, Nawa H. of hippocampal connections. Nature 1997; 385: 70–74. Regional specificity of alterations in NGF, BDNF and NT-3 levels in Alzheimer’s disease. Neuroreport 1996; 7: 2925–2928. 24 Zafra F, Hengerer B, Leibrock J, Thoenen H, Lindholm D. Activity Reference added in proof dependent regulation of BDNF and NGF mRNAs in the rat hippo- 45 Chlan-Fourney J, Li X-M, Juorio AV, Ashe P, Boulton AA. Differen- campus is mediated by non-NMDA glutamate receptors. EMBO J tial regulation of brain derived neurotrophic factor (BDNF) mRNA 1990; 9: 3545–3550. in rat brain by chronic haloperidol, risperidone and clozapine treat- 25 Bessho Y, Nakanishi S, Nawa H. Glutamate receptor agonists ment. Soc Neurosci Abs 1998; 24: 295.4.

Molecular Psychiatry