Prominent Synaptic and Metabolic Abnormalities Revealed By

Molecular Psychiatry (2008) 13, 1102–1117 & 2008 Nature Publishing Group All rights reserved 1359-4184/08 $30.00 www.nature.com/mp ORIGINAL ARTICLE Prominent synaptic and metabolic abnormalities revealed by proteomic analysis of the dorsolateral prefrontal cortex in schizophrenia and bipolar disorder K Pennington1,2, CL Beasley3, P Dicker4, A Fagan5, J English1, CM Pariante6, R Wait7, MJ Dunn2 and DR Cotter1 1Department of Psychiatry, Royal College of Surgeons in Ireland, Dublin, Ireland; 2Proteome Research Centre, UCD Conway Institute of Biomolecular and Biomedical Research, University College Dublin, Dublin, Ireland; 3Department of Psychiatry, University of British Columbia, Vancouver, BC, Canada; 4Department of Molecular and Cellular Therapeutics, Royal College of Surgeons, Dublin, Ireland; 5School of Medicine and Medical Science, UCD Conway Institute of Biomolecular and Biomedical Research, University College Dublin, Dublin, Ireland; 6King’s College, Section of Stress, Psychiatry and Immunology, Division of Psychological Medicine, Institute of Psychiatry, London, UK and 7Kennedy Institute of Rheumatology Division, Imperial College, London, UK

There is evidence for both similarity and distinction in the presentation and molecular characterization of schizophrenia and bipolar disorder. In this study, we characterized protein abnormalities in the dorsolateral prefrontal cortex in schizophrenia and bipolar disorder using two-dimensional gel electrophoresis. Tissue samples were obtained from 35 individuals with schizophrenia, 35 with bipolar disorder and 35 controls. Eleven protein spots in schizophrenia and 48 in bipolar disorder were found to be differentially expressed (P < 0.01) in comparison to controls, with 7 additional spots found to be altered in both diseases. Using mass spectrometry, 15 schizophrenia-associated proteins and 51 bipolar disorder-associated proteins were identified. The functional groups most affected included synaptic proteins (7 of the 15) in schizophrenia and metabolic or mitochondrial-associated proteins (25 of the 51) in bipolar disorder. Six of seven synaptic-associated proteins abnormally expressed in bipolar disorder were isoforms of the septin family, while two septin protein spots were also significantly differentially expressed in schizophrenia. This finding represented the largest number of abnormalities from one protein family. All septin protein spots were upregulated in disease in comparison to controls. This study provides further characterization of the synaptic pathology present in schizophrenia and of the metabolic dysfunction observed in bipolar disorder. In addition, our study has provided strong evidence implicating the septin protein family of proteins in psychiatric disorders for the first time. Molecular Psychiatry (2008) 13, 1102–1117; doi:10.1038/sj.mp.4002098; published online 16 October 2007 Keywords: two-dimensional gel electrophoresis; schizophrenia; bipolar disorder; septin; synaptic; metabolic

Introduction decrease in neuronal size,10,11 reduced pyramidal neuron dendritic field12,13 and decreased dendritic The dorsolateral prefrontal cortex (dlPFC) (Brodmann spine density14,15 with similar reductions in neuronal areas 9, 10, 42, 45 and 46) has primary functions in size observed in BPD.16 Efforts to determine the working memory, willed action and decision making.1 neurochemical and molecular basis for these cellular Support for dysfunction in this brain region at the alterations have revealed abnormalities in neurotrans- imaging, cellular and molecular level in schizo- mission, signal transduction, inhibitory interneuron phrenia2–5 and bipolar disorder6–8 is very strong. function and glial cells.17 In a review of 100 To investigate these findings at the cellular and neurochemical markers assessed in multiple brain protein level, a large number of neuropathological areas in the Stanley Foundation Brain Consortium, it studies investigating schizophrenia have focused on was found that developmental/synaptic, GABAergic this brain region. Cytoarchitectural studies have and markers involved in signal transduction were all found evidence for an increase in neuronal density,9 reduced in the schizophrenia group in the prefrontal cortex.18 Correspondence: Professor DR Cotter, Department of Psychiatry, Global gene and protein profiling studies are Royal College of Surgeons in Ireland, Dublin, Ireland. E-mail: [email protected] powerful methods to reveal molecular mechanisms Received 22 February 2007; revised 16 July 2007; accepted 31 July of dysfunction and specific candidate genes/proteins, 2007; published online 16 October 2007 which were previously unknown or overlooked. Proteomics of psychiatric disease in prefrontal cortex K Pennington et al 1103 Several microarray investigations of mRNA expres- analysis; Table 1 provides the demographic informa- sion levels in schizophrenia and control subjects have tion of the 96 cases included in the study. All brains now been carried out using post-mortem tissue from underwent clinical neuropathological examination the prefrontal cortex.19–23 In the first study of this and none demonstrated evidence of neurodegenera- kind, many genes encoding for presynaptic proteins tive changes or other pathological lesions. Frozen were found to be abnormally expressed.23 Indeed, tissue was available from one hemisphere of each synaptic disturbance in schizophrenia is well de- brain, with approximately equal numbers sampled in scribed, involving both the pre- and postsynaptic a random manner. Investigators were blinded to group machinery, and is considered a characteristic feature identity until after completion of the data collection of this complex disorder.24–27 stage. However, mRNA may not be the most reliable method to use in order to gain insight into the Tissue dissection and protein solubilization underlying pathophysiology of schizophrenia, where A 100–150 mg portion of cortical grey matter was abnormalities in gene function may or may not affect dissected from the fresh, frozen brain tissue of each of downstream protein expression.28,29 To approach this the 105 cases of dlPFC. Each sample was solubilized issue, proteomic techniques have been used, on a using a pestle in 1.5 ml Eppendorf tubes with 1 ml of complementary basis, to investigate the underlying solubilization buffer, a lysis buffer based on that molecular abnormalities in the cortex in schizophre- originally used by O’Farrell,38 containing 9.5 M urea, nia.30–33 These studies have identified a number 2% (w/v) CHAPS, 0.8% (v/v) Pharmalyte, pH 3–10, of abnormally expressed proteins, including those 1% (w/v) DTT and 150 ml Mini Complete Protease involved in oxidative stress and mitochondrial Inhibitor Cocktail (Roche Molecular Biochemicals, function. Indianapolis, IN, USA), and centrifuged at 13 000 g at It is possible that some of the findings of altered 20 1C for 5 min. metabolism and mitochondrial function may be related to confounding factors present when working Separation of proteins by two-dimensional gel with post-mortem tissue such as brain pH and post- electrophoresis mortem interval (PMI).34 For example, genes relating Two-dimensional gel electrophoresis was carried out to energy metabolism and proteolytic activities are as described previously.33 Briefly, proteins were decreased at low brain pH.35 Given these findings, separated in the first dimension according to their and those of others that have found significant pI values by isoelectric focusing using pH 4–7 18 cm alterations at the protein level with varying PMI immobilized pH gradient strips (GE Healthcare, times,36,37 high-throughput studies of protein and Amersham, UK). Samples were diluted in rehydration gene expression must take these post-mortem factors buffer (8 M urea, 0.5% w/v CHAPS, 0.2% w/v DTT, into consideration in their data analysis.33 0.2% w/v Pharmalyte, pH 3–10) and a protein load of The aim of the present study was to further 100 mg for analytical gels and 400 mg for preparative characterize differential protein expression in the gels was loaded onto the strips using in-gel rehydra- cortical grey matter in post-mortem brain tissue taken tion.39,40 Immobilized pH gradient strips were left to from the dlPFC (Brodmann’s area 9) of 105 patients rehydrate overnight and subsequently focused for from control, schizophrenia or bipolar patient groups 75 000 V h. Following isoelectric focusing, strips were using a non-hypothesis gel-based approach. This is equilibrated in 6 M urea containing 30% w/v glycerol, the most extensive proteomic study of its type 2% (w/v) SDS and 0.01% (w/v) Bromophenol blue, in psychiatric disorders to date, including the lar- with the addition of 1% (w/v) DTT for 15 min, gest samples of schizophrenia and bipolar disorder followed by the same buffer without DTT, but with subjects. the addition of 4.8% (w/v) iodoacetamide for 15 min.41 Second-dimension SDS-polyacrylamide gel electrophoresis was performed using 12% polyacry- Materials and methods lamide gels run overnight at 20 mA per gel at 15 1C Samples (Hoefer DALT system, GE Healthcare). Following Blocks of dlPFC grey matter from Brodmann area 9 electrophoresis, gels were fixed overnight in a (superior frontal gyrus) were obtained from the methanol/acetic acid solution and visualized with second Stanley Brain Collection (www.stanleyre- the Owl Silver Stain kit (Insight Biotechnology, search.org). This sample consists of 105 subjects (35 London, UK). For preparative gels, the PlusOne Silver controls, 35 schizophrenics and 35 bipolar disorder). Staining kit (Amersham Biosciences, Uppsala, Swe- Diagnoses were made according to the Diagnostic and den) was used with modifications to allow sub- Statistical Manual of Mental Disorders IV criteria. sequent use with mass spectrometric analysis.42 Detailed case summaries are provided on demographic, clinical and histological information (Table Image analysis 1). Due to missing demographic data including one Silver-stained gels were scanned at 100 mm resolution case of CADASIL and additional technical problems using a Personal SI laser densitometer (Molecular (poor protein separation and matching with other Dynamics, Sunnyvale, CA, USA). Protein spots were gels), nine samples were removed from the final analysed quantitatively using the image analysis

Molecular Psychiatry Proteomics of psychiatric disease in prefrontal cortex K Pennington et al 1104 Table 1 Demographic, histological and clinical details of the 97/105 cases used for analysing protein expression by 2D electrophoresis in the dlPFC region of the frontal cortex

Variable Unaffected Schizophrenic Bipolar controls disorder

Number of cases 30 34 32

Demographics Age at death in years 44.0 (8.13) 42.5 (8.6) 45.1 (11.0) Gender (female:male) 7:23 9:25 15:17*

Histological Mean PMI and range (h) 29.2 (13.1) 31.4 (15.8) 37.8 (18.4)* Mean pH and range 6.59 (0.26) 6.47 (0.24) 6.42 (0.31)* Side of brain (left and right) 13L, 17R 16L, 18R 17L, 15R Refridgerator interval (h) 3.6 (2.7) 6.0 (4.2)** 9.9 (10.7)** Brain weight (g) 1456.5 (155.9) 1438.6 (107.1) 1400 (141.5)

Clinical Causes of death 0 suicide 7 suicide 14 suicide 29 cardiac 14 cardiac 7 cardiac 1 other 10 other 11 other Mean lifetime alcohol abuse (1 = low, 5 = high) 0.7 2.2 3.0 Treated with antipsychotics at the time of death (no:yes) 30:0 2:32 14:18 Treated with antidepressants at the time of death (no:yes) 30:0 26:8 9:23 Treated with mood stabilizers at the time of death (no:yes) 30:0 23:11 15:17

Abbreviations: dlPFC, dorsolateral prefrontal cortex; PMI, post-mortem interval; 2D, two dimensional. Mean scores are given with standard deviations. Disease group details were found to be significantly different from the control group following independent t-test (disease versus control groups). *P < 0.05; **P < 0.01.

system Progenesis V2003.2 (Nonlinear Dynamics, regression analysis) was carried out excluding disease Newcastle, UK). Following automatic and subsequent status from the model. The variables examined using manual editing, warping and matching procedures, this model were antipsychotic, mood stabilizer and the normalized volume of each protein, an integration antidepressant status at the time of death and lifetime of optical density and area, was measured objectively alcohol abuse and fluphenazine equivalent dose. as a percentage of the total volume of all the detected To determine if our proteomic findings converged spots and the data exported for analysis. with mRNA data from the same brain regions (dlPFC), additional post hoc analyses were carried out using Statistical analysis three of the microarray data sets (Study ID numbers 1, The primary analysis was to detect proteins differing 3 and 7) from the same tissue that is available online in expression between the control group and either from the www.stanleygenomics.org website. From the the schizophrenia or bipolar disorder groups. Since 100 probes, which corresponded to the proteins listed brain pH, refrigerator interval, PMI and age are known in Table 2 independent t-tests were carried out to considerably influence protein expression in the between the disease groups and controls. brain,36,37 an analysis of covariance (ANCOVA) on normalized spot volumes was performed. Estimated Protein identification by liquid differences between controls and schizophrenia or chromatography-tandem mass spectrometry bipolar disorder were obtained from the model as All protein spots found to be significantly differen- differences in least squares means (linear contrasts). tially expressed were subsequently cut from the

Prior to analysis, the data were log10 transformed to preparative gels using a 1.5 mm pipette tip and placed reduce distributional skew and to improve the normal in water. Spot plugs were destained and proteins were approximation for valid P-values. Significance testing digested with trypsin as described previously.42 was performed at the 1% level. All statistical tests All samples were run on a Finnigan LTQ mass were carried out using SAS V9.1 software. spectrometer connected to a Surveyor chromato- Secondary analysis of the effects of each of the graphy system incorporating an auto-sampler. Tryptic covariates included in the primary analysis was also peptides were resuspended in 0.1% formic acid and carried out using ANCOVA. In the case of potential were separated by means of a modular CapaLC system confounding variables, which covary with disease (Finnigan) connected directly to the source of the status, a secondary multivariate analysis (multiple LTQ. Each sample was loaded onto a Biobasic C18

Molecular Psychiatry Table 2 Lists all significantly (P < 0.01) differentially expressed protein spots identified by MS

Spot no. Protein SWISS- Fold changes in ANCOVA P-values No. of Theoretical Functional SMRI array Covariate PROT protein density peptides pI/mass KDa ontology database accession (adjusted following matched/% expression no. ANCOVA) sequence changes SCZ & BPD SCZ BPD SCZ BPD coverage

1 Dihydropyrimidinase- Q14195 1.47 1.55 < 0.0001 < 0.0001 2/2.9 6.04/62.0 Synaptic/vesicle/CNS *,^, þ ,#,1 related protein-3 development 2 1.73 1.86 0.0002 < 0.0001 5/13.9 1 3 D-3 Phosphoglycerate O43175 1.67 2.25 0.005 < 0.0001 3/4.5 6.7/56.5 Metabolism/CNS BPD7 ^,D dehydrogenase development 4 Septin-5 Q9NVA2 1.71 1.75 0.0003 0.0002 1/4.6 6.6/42.8 Synaptic/exocytosis/ BPD&SCZ3 *.#,D cell division 5 1.57 1.77 0.004 0.001 1/5.2 * 6 T-Complex 1, alpha P17987 1.63 2.33 0.0096 < 0.0001 2/3.8 6.0/60.3 Protein folding BPD3,7 ^,1,D subunit

SCZ SCZ SCZ

7 6-Phosphoglu- O95336 1.28 0.008 4/20.9 6.0/27.5 Metabolism—other ^ conolactonase 8 Dynamin 1 Q05193 1.33 0.0055 2/3.8 6.00/97.3 Synaptic/vesicle 7b Endoplasmatic P30040 1.28 0.008 4/8.1 7.17/29.0 Protein synthesis reticulum protein ERp29 cortex prefrontal Pennington in K disease psychiatric of Proteomics 9 Gamma enolase P09104 À3.06 0.0002 2/6.7 4.8/29.1 Energy metabolism 1 10 Glial fibrillary acidic P14136 1.79 0.008 35/66.1 4.25/50.0 Cytoskeletal protein, astrocyte tal et 11 Guanine deaminase Q9Y2T3 À3.29 0.0002 2/9.1 5.4/51.0 Metabolism—other 12 Neurofilament triplet P07196 À1.5 0.002 17/22.1 4.6/61.4 Cytoskeletal/synaptic BPD3 L protein 13 Protein c2orf32 Q96F85 À2.31 0.008 1/13.4 7.7/18.6 Unknown 14 Protein kinase C and Q9BY11 À1.88 0.003 3/5.4 5.2/51.0 Synaptic/endocytosis casein kinase in neurons protein 1

BPD BPD BPD

15 4-Trimethylamino- P49189 1.46 0.009 2/3.2 6.00/53.5 Metabolism—other butyraldehyde dehydrogenase 17 Alcohol dehydrogenase P14550 2.27 0.0004 4/15.1 6.00/35.4 Metabolism—other [NADP þ ] 18 Aldehyde dehydrogenase P14550 2.11 0.001 5/28.7 6.6/56.4 Metabolism—other * 1A1 oeua Psychiatry Molecular 19 Aldehyde dehy- P49419 1.99 0.009 1/2.7 6.43/55.2 Metabolism—other SCZ7BPD3 D drogenase family 7 member A1 1105 1106 oeua Psychiatry Molecular

Table 2 Continued

Spot no. Protein SWISS- Fold changes in ANCOVA P-values No. of Theoretical Functional SMRI Covariate PROT protein density peptides pI/mass KDa ontology array accession (adjusted following matched/% database no. ANCOVA) sequence expression coverage changes BPD BPD BPD

20 Alpha internexin Q16352 À1.23 0.0004 2/3.4 4.25/55.3 CNS development/ SCZ7 þ cytoskeletal cortex prefrontal in disease psychiatric of Proteomics 21 Alpha tubulin P68363 1.73 0.0065 2/7.1 7.1/56.3 Cytoskeletal SCZ7 *,^ 22 1.93 0.0029 7/4.24 23 ATP synthase beta P06576 1.33 0.0072 4/12.3 5.3/56.5 Energy metabolism D chain 24 ATP synthase delta P30049 À1.36 0.0049 1/2.46 5.13/17.5 Energy metabolism þ chain 25 ATP synthase gamma P36542 1.92 0.0002 1/13.4 9.2/33.0 Energy metabolism BPD3,7 chain (splice isoforms

of liver) Pennington K 26 ATPase (transitional O14983 1.25 0.0084 2/0.4 5.1/110.3 Energy metabolism ER) 27 Beta tubulin Q9H4B7 1.54 0.0016 2/6.3 4.8/49.6 Cytoskeletal

28 Cadherin-24 precursor Q86UP0 1.57 0.0027 2/1.2 4.6/87.8 Other *,^,D al et 29 (i) Collagen alpha-1 (ii) P02458 1.56 0.0047 3/1.1 8.7/134.5 Cytoskeletal D chain precursor 19b Cytosol aminopeptidase P28838 1.99 0.009 6/12.94 6.00/52.6 Metabolism—other 30 Cytosolic acyl coenzyme O00154 1.64 0.01 2/5.79 8.33/102.3 Metabolism—other þ A thioester hydrolase 31 D-3-Phosphoglycerate O43175 2.19 0.0002 2/4.89 6.00/56.6 CNS development/ BPD7 dehydrogenase energy metabolism 31b Dihydrolipoyl P09622 2.19 0.0002 4/5.7 8.33/54.1 Energy metabolism dehydrogenase 32 Gamma enolase P09104 1.96 0.0064 2/6.7 4.9/47.1 Energy metabolism þ 33 Gamma synuclein O76070 À1.44 0.0046 2/19.7 5.0/13.3 CNS development BPD1 34 Inorganic pyrophos- Q15181 À1.23 0.0069 10/19.5 5.5/32.7 Metabolism—other BPD & SCZ3 phate 35 Isocitrate dehydro- P50213 1.62 0.0034 1/6.3 6.9/39.6 Energy metabolism BPD7 genase subunit alpha 36 Kinesin light chain 2 Q9H0B6 1.63 0.0061 2/1.6 6.72/68.9 Cytoskeletal *,^,D 29b Mitochondrial O75439 1.56 0.005 2/4.7 6.00/54.3 Energy metabolism D processing subunit peptidase beta subunit 20b Neurofilament triplet M P07197 À1.23 0.0004 10/6.2 4.25/102.2 Cytoskeletal þ protein 37 NG,NG-dimethy- O94760 À1.21 0.0065 22/21.5 5.13/31.0 Cell signalling/ * D larginine demethylamino- metabolism-other hydrolase 1 38 Peroxiredoxin 2 P32119 À1.18 0.004 7/43.9 5.9/21.9 Energy metabolism * 39 Peroxiredoxin 6 P30041 1.39 0.0068 4/28.3 6.3/24.9 Energy metabolism ^,D 29c Probable ubiquitin O75592 1.56 0.004 2/0.3 6.7/510 Protein synthesis SCZ7 D ligase protein mycbp2 Table 2 Continued

40 Pyridoxal phosphate Q8TCD6 1.60 0.0049 1/5.8 6.3/27.8 Metabolism—other phosphatase 41 Secernin 1 Q12765 1.85 0.0035 2/7.7 4.66/46.4 Exocytosis/proteolysis SCZ7 ^ 42 Septin 5 Q99719 1.45 0.007 7/12.5 6.7/49.4 Exocytosis/cell BPD & *,D division/synaptic SCZ3 43 Septin 6 Q14141 2.11 0.0002 2/5.8 6.6/42.8 Synaptic/cell division BPD & *,K,D SCZ7 44 Septin 11 Q9NVA2 1.68 0.0023 2/3.5 6.7/49.4 Cell division BPD1 * 36b 1.63 0.006 5/13.8 D 24b Serotransferrin P02787 À1.36 0.0049 4/8.3 6.00/76.9 Blood regulation 16 Serum albumin P02768 1.80 0.0016 1/2.5 5.92/69.4 Blood regulation SCZ1 *,^,D 28b Serum albumin P02768 1.57 0.00027 5/7.9 6.00/69.3 Blood regulation SCZ1 45 Stress 70 protein P38646 1.96 0.0047 2/6 6.00/73.6 Protein folding/energy SCZ3 *,D metabolism 46 Succinyl CoA ligase Q9P2R7 2.11 0.002 2/2.2 7.6/50.3 Energy metabolism * beta chain 47 T-Complex 1, beta P78371 1.66 0.002 20/54.4 6.4/57.3 Protein metabolism/ BPD3 * subunit protein folding 48 T-Complex 1, zeta Q92526 2.19 0.0006 8/14.7 6.6/57.9 Protein folding/ * cortex prefrontal Pennington in K disease psychiatric of Proteomics subunit protein metabolism 49 Ubiquinol-cyto- Q9UDW1 À1.21 0.001 2/6.5 9.5/71.8 Energy metabolism SCZ1 þ chrome C reductase complex al et 20c Ubiquinol-cyto- P31930 À1.23 0.0004 1/0.3 6.00/52.6 Energy metabolism BPD3 þ chrome C reductase complex, core protein 20d Vacuolar ATP synthase P21281 À1.23 0.0004 17/18 4.25/56/5 Energy metabolism þ subunit B, brain isoforms

Abbreviations: ANCOVA, analysis of covariance; CNS, central nervous system; ER, endoplasmic reticulum; SMRI, Stanley Medical Research Institute. Relationships between potential confounding factors and protein expression were explored for all significant proteins using ANCOVA. Proteins found to be significantly expressed in the same tissue at the transcript level in three of the six data sets available at www.stanleygenomics.org are also reported. Data relating to direction of change in these transcript studies were not available publicly on the website and are therefore not presented here. SMRI array database study ID numbers 1, 3 and 7 refer to those listed on the website. Proteins that were found to be predicted by age (*), brain pH (^), post-mortem interval ( þ ), refridgerator interval (#), antipsychotic dose (mg per lifetime) (D) and either antidepressant (K) or antipsychotic (1) medication treatment at the time of death with significance P < 0.01 are listed in the final column. oeua Psychiatry Molecular 1107 Proteomics of psychiatric disease in prefrontal cortex K Pennington et al 1108 Picofrit column (100 mm length, 75 mm ID) at a flow rate of 30 nl minÀ1. The samples were then eluted from the C18 Picofrit column by an increasing acetoni- trile gradient. The mass spectrometer was operated in positive ion mode with a capillary temperature of 200 1C, a capillary voltage of 46 V, a tube lens voltage of 140 V and with a potential of 1800 V applied to the frit. All data were acquired with the mass spectrometer operating in the automatic data-dependent switching mode. A zoom scan was performed on the five most intense ions to determine charge state prior to tandem mass spectrometry analysis.

Database searching The raw mass spectral data were analysed using Proline, a proteomics analysis platform. Thermo RAW files were converted into mzXML format. X! Tandem release 2005.02.01 was used to analyse these files with a static iodoacetamide modification, with Uni- prot/SwissProt release 7.6 used as the FASTA search database. The results were stored in a MySQL database and were interrogated using Proline’s web- based proteomics toolset. Figure 1 A representative two-dimensional gel indicating the protein spots found to be significantly differentially expressed in schizophrenia and/or bipolar disorder and Further investigation of septin abnormalities subsequently identified with mass spectrometry. Spot Changes in septin 5 expression were remarkable, with numbers relate to the proteins listed in Table 2. two protein spots significantly increased in expression in both disorders and a further form of the same protein found to be increased in bipolar disorder when compared to control cases. To confirm the presence of multiple protein spots of septin 5 on the with ANCOVA (correcting for age, brain pH, refrig- two-dimensional (2D) gel and to further investigate erator interval and PMI). septin 5 protein spots, a 2D blot was carried out. A 2D Of these 101 spots, 35 were excluded because they gel (100 mg protein load) was run as described could not be visually confirmed or were not present previously and transferred onto Hybond nitrocellu- in the preparative gel used for identification of each lose membranes (GE Healthcare) using a semi-dry protein spot. Thus a total of 66 spots of interest were 2 apparatus (GE Healthcare) for 2 h at 0.8 mA per cm . visually confirmed and available for identification by The membrane was subsequently blocked with 5% mass spectrometry. Of these, 11 were in the schizo- non-fat dried milk (NFDM) in phosphate-buffered phrenia group, 48 in the bipolar disorder group and 7 saline (PBS) for 1 h. The septin 5 antibody (Chemicon, spots were abnormally expressed in both disorders in CA, USA; MAB5358) was applied at 1/1000 in 5% comparison to control patients. NFDM/PBS and incubated overnight at 4 1C. Follow- ing two washes in 5% NFDM/PBS, the horseradish Mass spectrometry peroxidase-conjugated anti-mouse secondary anti- Of the 66 protein spots visually confirmed as being body (Dako, Glostrup, Denmark) was added at a differentially expressed protein spots in the schizo- concentration of 1/2000 and left for 1 h at room phrenia or bipolar disease groups, 63 proteins were temperature. Two further washes in 5% NFDM/PBS identified using liquid chromatography-tandem mass and then PBS were carried out before incubating for spectrometry (see Figure 1). Nine proteins were 5 min in Supersignal chemiluminescent substrate identified that were found to be differentially ex- (Pierce, IL, USA). Blots were then exposed to X-ray pressed in schizophrenia, 45 proteins were identified film and developed. that were significantly associated with bipolar disorder and 6 proteins were identified that were significant in both disorders. Results The identities of these protein spots, their 2D gel electrophoresis fold change alteration in volume density on the 2D A total of 1944 individual spots were visualized gel, the theoretical pI and MW and the peptide across the group of 105 dlPFC gels (pH 4–7) (Figure 1). sequences matched for each protein are shown in Statistical analyses were carried out on all of these Table 2. The proteins are listed alphabetically with spots and subsequently 101 differentially expressed their functional ontology and any predictive relation- significant (P < 0.01) spots (20 schizophrenia, 69 ship with any of the major confounding factors bipolar disorder, 12 both disorders) were identified evaluated.

Molecular Psychiatry Proteomics of psychiatric disease in prefrontal cortex K Pennington et al 1109

Figure 3 Two-dimensional (2D) western blot of septin 5 showing five isoforms of the protein. Protein spots identified as differentially expressed in primary analysis (numbers 4, 5 and 42) are listed, in addition to those others identified by 2D blot (numbers 50–51). See Table 3 for fold differences between disease and control groups for these spots (corresponding numbers are used).

Table 2). Additionally, another form of septin 5 and three further septin family proteins were found to be significantly increased in bipolar disorder (P < 0.01; septin 6 and two forms of septin 11). To further examine the forms of septin 5 proteins observed on the 2D gels we performed a 2D western blot using an antibody-specific for septin 5. Five spots were identified, which could result from differences in post-translational modifications, possibly including phosphorylations. Figure 3 shows all five septin 5 protein spots detected by the 2D western blot, and identifies the positions of the three septin 5 spots that were first found to be abnormal in the primary analysis (spots 4, 5 and 42). We then went back to Figure 2 Visual representation of the cell functions the primary analysis and re-examined the data for the affected in schizophrenia and bipolar disorder in the two newly identified spots (spots 50 and 51). dorsolateral prefrontal cortex grey matter. Each protein As can be seen from Table 3, there is increased found to be differentially expressed and identified using expression of all five forms of septin 5 in bipolar mass spectrometry was grouped into one or more ontolo- disorder in comparison to controls (at P < 0.05). For gical categories. Up- and downregulation is indicated by its septin 5 there are also three forms, which are position in either minus or plus on the graphs axis. significantly increased in expression in the schizophrenia sample. Septin 11 is upregulated in two forms in bipolar disorder, with one of these increased Functional analysis of protein findings in schizophrenia. Septin 6 is altered only in bipolar To evaluate the differences in overall protein expression disorder where it shows a 2.3-fold increased expres- in both schizophrenia and bipolar disorder, proteins sion. Post hoc ANCOVA analysis of these septin differentially expressed in both disorders were grouped family proteins combined showed them to be highly according to functional ontology. This grouping was significant both for bipolar disorder (P < 0.0001) and based on literature searching, information provided for schizophrenia (P < 0.0001). each protein at www.ebi.ac.uk/ego, www.expasy.org, http://harvester.embl.de/html/protein_lists.html and www.hprd.org. Figure 2 illustrates the broad distribu- Findings at the transcript level tion of protein expression abnormality seen in each Out of 100 probes (corresponding to 23 out of the 52 group and the differences between them. proteins reported in Table 2) 31 were significantly abnormally expressed at the transcript level (P < 0.05) Confirmation of protein changes in the Stanley Medical Research Institute array Two of the six proteins found to be differentially database as well as at the protein level. The results expressed in both schizophrenia and bipolar disorder from these analyses are listed in a separate column in were forms of septin 5 (P < 0.01) (see Figure 1 and Table 2.

Molecular Psychiatry Proteomics of psychiatric disease in prefrontal cortex K Pennington et al 1110 Table 3 Eight forms of septin 5, 6 and 11 as confirmed by entially expressed protein spots) than in schizophre- mass spectrometry and in septin 5 additionally by 2D blot nia (18 differentially expressed protein spots), although seven of these were differentially expressed Septin Spot Mean ANCOVA adjusted fold changes in both. The clearest point of distinction was a pre- isoforms no. and P-values dominance of metabolic-associated protein changes in bipolar disorder, and of synaptic-associated protein SCZ BPD changes in schizophrenia. The following sections will focus on the possible roles of these proteins in Septin 5 50 1.34 (NS) 1.45 (0.045) synaptic and metabolic dysfunction in schizophrenia 51 1.29 (NS) 1.68 (0.032) and bipolar disorder. 4 1.71 (0.0003) 1.75 (0.0002) 42 1.28 (0.049) 1.47 (0.007) 5 1.57 (0.004) 1.77 (0.001) Evidence for synaptic dysfunction in psychiatric Septin 6 43 1.28 (NS) 2.11 (0.0002) disorders Septin 11 36b 1.48 (0.015) 1.68 (0.0023) Expression changes in many synaptic-associated 44 1.19 (NS) 1.63 (0.006) proteins such as synaptophysin, complexin I and II, All septins 1.39 ( < 0.0001) 1.67 ( < 0.0001) synaptosomal-associated protein (SNAP)-25 and syntaxin (for example Eastwood et al.,43 Sawada et al.,44 Abbreviation: 2D, two-dimensional. Owen et al.45 and Frankle et al.46) have been found in The altered expression and significance levels of each of schizophrenia. In addition, these protein changes are these protein spots in both schizophrenia and bipolar in keeping with synaptic-associated roles for many of disorder are listed. Protein spots 4, 5 and 42 were revealed the recently described candidate genes for schizo- by primary analysis (significance P < 0.01) and spots 50 and phrenia.47 Consequently, our finding of substantial 51 were revealed to be significant (P < 0.05) after 2D blot analysis. synaptic-associated protein pathology in schizophrenia is in keeping with these previous findings. Additionally, we have been able to extend the literature by increasing our knowledge of the identity of some of the synaptic-associated proteins involved Discussion in this disease process. In the present study, the This study investigated the global protein expression expression of five protein spots that are associated levels of 1944 protein spots found in the pH range 4–7 with synaptic function were increased (two forms of in 97 2D gel images of human post-mortem brain septin 5 and dihydropyrimidinase-related protein-3 tissue taken from schizophrenia, bipolar disorder and (DRP-3), one form of dynamin I (DYN-I)) in schizo- control cases. This is the largest proteomic study to phrenia and two proteins were decreased (protein date that has investigated dlPFC grey matter in both kinase C and casein kinase in neurons 1 (PACSIN 1) schizophrenia and bipolar disorder. Our findings and neurofilament triplet L). Synaptic dysfunction in demonstrate novel and robust alterations in the septin bipolar disorder was also implicated and is consistent family of proteins in both disorders. The findings also with the literature.48–51 In bipolar disorder, six extend and elaborate on the evidence already im- proteins with synaptic-associated function were plicating synaptic and metabolic pathology in these found to be increased (three forms of septin 5 and disorders. septin 6 and two forms of DRP-3). Functionally grouping the proteins revealed rela- The majority of the synaptic proteins found to be tively different disease profiles in schizophrenia and altered in the schizophrenia group in the present bipolar disorder (see Figure 2). Functionally, 7 of 15 study were increased and not decreased as shown in of the proteins found to be significantly differentially Figure 2. However, this apparent increase in synaptic- expressed in schizophrenia are associated with associated protein expression is largely due to the synaptic function and 4 of 15 with metabolic or number of septin protein changes in schizophrenia, mitochondrial function. A smaller number of proteins all of which were increased (see Table 3). Otherwise, also fell into the categories of protein synthesis, we found both increases and decreases in expression protein folding, cytoskeleton and brain development. of synaptic-associated proteins (septin 5, DRP-3 and In marked contrast, in bipolar disorder, 25 of 51 of the dynamin were increased and neurofilament triplet L significantly differentially expressed proteins were (NF-L) and PACSIN 1 were decreased). The first two associated with metabolic and mitochondrial func- of these proteins are implicated in the present study tion, 9 of 51 cytoskeletal and 6 of 51 synaptic for the first time in schizophrenia, whereas DYN-I, associated. Other categories included proteolysis, which is involved in vesicle endocytosis, has been protein folding, protein synthesis, brain develop- found to be upregulated in schizophrenia in ment, cell signaling, cell division and blood regula- two previous proteomic studies.31,32 NF-L expression tion. It is also of interest to note that the majority of has previously been reported to be reduced in proteins abnormally expressed in bipolar disorder schizophrenia. were upregulated rather than downregulated in The proteins associated with synaptic function that comparison with controls. Protein expression changes were found to be most significantly altered in both were more prominent in bipolar disorder (55 differ- disorders were two forms of DRP-3, which were found

Molecular Psychiatry Proteomics of psychiatric disease in prefrontal cortex K Pennington et al 1111 to have increased expression in schizophrenia and recent paper of reduced NF-L protein expression in bipolar disorder (P < 0.0002; 1.5- to 1.9-fold change). the dlPFC in elderly schizophrenic patients.65 Inter- This protein (also known as Ulip6/Unc-33like pro- estingly, this protein also interacts with 14-3-3 gamma tein/CRMP5) was first characterized in 1996 and which has been implicated in schizophrenia pre- found to have significant homology to Unc-33, a viously66,67 and which also interacts with the dis- protein important for neuronal development and rupted in schizophrenia 1 protein strongly implicated neurite outgrowth.52 Since then, the protein has been in the disease.68,69 shown to interact with an integral membrane glycine transporter located in axonal terminals and found to Postsynaptic density and glutamatergic system be localized to synaptic regions of rat neurons.53 implicated by findings Given this evidence for a role for DRP-3 in synaptic Several proteins that were differentially expressed in function, it has been suggested that this protein may this investigation, namely PACSIN, NF-L, DYN-I and participate in the regulation of glycine transporter 2 septin 5, are involved in the function of the pre- and/ recycling, a process important in the modulation of or postsynaptic density, and through these in NMDA- neurotransmitter reuptake at the nerve terminals.53 regulated glutamatergic neurotransmission, strongly DRP-3 is also involved in glutamatergic function, and implicated in the pathophysiology of schizophre- a degraded form of the protein occurs in the presence nia.70,71 In addition, many of the susceptibility genes of glutamate excitotoxicity (N-methyl-D-aspartate for schizophrenia to date have been shown to play a (NMDA)) and oxidative stress.54 The functional role role in glutamate function.47 Proteins such as dis- of DRP-3 with the glutamatergic system is yet to be rupted in schizophrenia 1, SNAP-25, 14-3-3 gamma, revealed, but given the evidence for glutamatergic synaptojanin 1, complexins I and II and synataxin dysfunction in schizophrenia47 it is interesting that have all previously been implicated in schizophrenia an increase in DRP-3 could be indicative of a hypo- pathology and are known to directly interact with one glutamatergic state. or more of the proteins found to have altered PACSIN 1 was found to be reduced in schizophre- expression in this study (see Figure 4). Alpha nia and is one of a family of cytoplasmic phospho- internexin is the only protein listed in Figure 4, proteins, which are thought to have a role in synaptic which was shown to be selectively altered in bipolar vesicle formation and transport and more generally in disorder and this protein interestingly has been recruiting interacting proteins to sites of endocytosis. shown to be involved in NMDA receptor function- PACSIN 1 is specifically localized in neurons and has ing.57 These findings thus support evidence for been found to colocalize with dynamin and bind to glutamatergic dysfunction in mood disorders72,73 as synaptojanin.55 Interestingly, DYN-I was significantly well as in schizophrenia. increased in schizophrenia in comparison to control patients (P = 0.0055) and is localized to presynaptic nerve terminals in the brain.56 Additionally, DYN-I has also been found to interact with NMDA receptor 157 and SNAP-23,58 closely related to SNAP-25, proteins which have been previously implicated in the pathogenesis of schizophrenia.59–62 DYN-I is a GTPase essential for the recovery of synaptic vesicles from the neuronal membrane once the contents have been released63 and has been hypothesized to have evolved to have a primary role in synaptic vesicle recycling.64 Indeed, Yamoshita et al. (2005) have suggested that DYN-I is indispensable for vesicle endocytosis, specifically at the fast ‘kiss-and-run’ form of synaptic communication. In support of our findings are those of Prabakaran et al. (2004), who found that DYN-I and II were increased in the grey matter but decreased in the white matter of schizo- Figure 4 Cartoon illustrating the interactive protein part- phrenics in comparison to controls, while a recent ners of the NMDA receptors 1 and 2, which have previously proteomics study of the anterior cingulate cortex in been implicated in schizophrenia (outlined). Proteins schizophrenia also reported an increase in dynamin 1 implicated in this study for the first time are underlined. in schizophrenia.32 Those proteins implicated previously and also shown to be NF-L is a neuronal-specific intermediate filament abnormally expressed in this study are outlined underlined. protein with a cytoskeletal and synaptic function Alpha internexin was found in bipolar disorder only, all others were found in schizophrenia and/or bipolar known to interact with NMDA receptors in the disorder. Refer to Table 2 for detailed information on each postsynaptic density. We found this protein to be of the proteins found to be abnormally expressed in this significantly decreased in expression in schizophre- study. Interactive information was obtained from the nia in comparison with control patients. This is in human protein reference database (www.hprd.org). NMDA, agreement and further confirms findings reported in a N-methyl-D-aspartate.

Molecular Psychiatry Proteomics of psychiatric disease in prefrontal cortex K Pennington et al 1112 Evidence for septin family contribution to and the microtubule-associated protein.78–80 Septin 5 schizophrenia and bipolar disorder (also known as CDCrel-1 and Peanut-like 1) appears Five forms of septin 5 present on the 2D gels analysed particularly interesting as it has been previously in this study were significantly increased in expres- found to colocalize with GABAergic vesicles.81 In a sion in bipolar disorder, with three of these signifi- study investigating the distribution of septin 5 in the cant also in schizophrenia in comparison to controls adult human brain, differential centrifugation re- (see Table 3). In addition, two forms of septin 11 and vealed it to associate predominantly with neurons one of septin 6 were found to be significantly and to co-purify with SNAP-25 and synaptophysin increased in bipolar disorder. While the functional marked synaptosomes.82 Further work is needed to roles of the septin family of proteins within the brain evaluate the expression pattern of this protein in the are unclear at present, there is a growing literature dlPFC in normal human brain. indicating that they have important, diverse and Little is known about the function of septin 11 as it developmental stage-specific functions within the has only recently been sequenced. It has been found brain.74 to be highly expressed in the brain along with septins With regard to schizophrenia, it is of interest that 3, 4, 5, 7 and 8.74 Septins 2 and 4 have previously the gene for septin 5 is located in chromosomal region been associated with Alzheimer’s disease due to their 22q11.2, parts of which are deleted in velo–cardio– localization in neurofibrillary tangles83 and septin 5 is facial syndrome and whose sufferers have been found a known binding partner of the Parkinson’s disease- to have a much higher rate of schizophrenia than the associated protein parkin.84 In addition, a review of general population (30%).75 A recent molecular the septin family of proteins and their distribution in examination of the deletion breakpoints in adults mouse brains81 found that the majority of the septin with 22q11 deletion syndrome found that the septin 5 family were localized around synaptic vesicles near gene is located within the region deleted in all the the terminals in unmyelinated axons with the highest patients suffering this disorder, including all those expression within the brain seen in the cerebral with schizophrenia (n = 44 in total).76 While the cortex, hippocampus, cerebellum, thalamus and mechanism by which septin 5 may be involved in striatum. These data and the other work discussed the pathogenesis of schizophrenia is unknown, it is previously indicate that members of the septin family clear that there is now intriguing evidence for its of proteins may function in synaptic vesicle transport, involvement. There is an apparent contradiction in a fusion or recycling events in the human brain. protein with increased expression in schizophrenia Abnormalities in the expression of these proteins (septin 5) being located within a deletion region could contribute in part to the synaptic dysfunction associated with the disorder. Of interest in this regard observed in schizophrenia and bipolar disorder. is a recent study that describes an increase in the proportion of one form of catechol O-methyltransfer- Evidence for metabolic dysfunction in bipolar disorder ase in comparison with total catechol O-methyltrans- and schizophrenia ferase protein expression, the gene for which is also Changes in mitochondrial and metabolic func- located on 22q11, in schizophrenia and bipolar tion have been implicated in schizophrenia previou- disorder compared to controls77—for like septin 5, sly.31–33,85–88 The results of this study extend this catechol O-methyltransferase is located in the 22q11 abnormality to bipolar disorder, whereby almost half deletion region. In addition, at the mRNA level, of all significantly differentially expressed proteins previous work by the same group has found no identified were found to have a mitochondrial or differences in the catechol O-methyltransferase tran- other metabolic function. In the schizophrenia group, script.77 This work, with our own, indicates that the this figure was a quarter, suggesting that metabolic presence of candidate genes within a deletion region dysfunction in this area is relatively more prominent associated with the disorder does not necessarily in bipolar disorder than schizophrenia. In a previous imply reduced cortical mRNA and protein expression proteomic study of white and grey matter in the levels within the group of schizophrenic subjects as a dlPFC in schizophrenia, 50% of differentially ex- whole. Rather, the data suggest that altered expression pressed identified proteins were found to be altered of these candidate genes are vulnerability factor to in this functional group.31 The finding of abnormal disease regardless of the direction of expression metabolic-associated proteins supports previous work changes. suggesting a ‘mitochondrial hypothesis’ of the dis- Functionally, septins are GTP-binding proteins order89 and also previous imaging studies which have often found in association with actin and microtubule found an increased metabolism, for example,90 in filaments, and decreased septin expression has been bipolar subjects in comparison with controls. found to lead to increased microtubule stability.78 Microarray work has revealed decreases in meta- Functionally, it has also been found that septin 5 bolic-associated transcripts in bipolar disorder.91,92 binds to presynaptic SNARE complex proteins and However, decreased metabolic-associated transcripts inhibits exocytosis.79,80 Indeed, many septins have have been associated with decreased brain pH,34 a been shown to have synaptic-associated binding common confounding factor in psychiatric post- partners which have been previously implicated in mortem brain research. Due to this latter finding, it schizophrenia, for example syntaxin, complexin II is important to interpret with care the data that is

Molecular Psychiatry Proteomics of psychiatric disease in prefrontal cortex K Pennington et al 1113 obtained from the Stanley array data set, which is not the transcript level data, we carried out parametric corrected for brain pH, as the mean values in the t-tests on all the probes that related to our signifi- schizophrenia and bipolar groups are lower than in cantly differentially expressed proteins of interest the control subjects. In addition to this, and in (n = 84) in both the schizophrenia and bipolar dis- support of our data, in a sub-group of medication- order groups in comparison with control subjects. free bipolar disorder patients, mitochondrial gene More specifically, we chose three of the data sets transcripts were shown to be increased in expression available on this Stanley Medical Research Institute in comparison with controls.93 genomic database (Study IDs; 1, 3 and 7) as these used Some of the findings associated with mitochondrial the HGU133A Affymetrix platform and hence had dysfunction are in agreement with the findings in comparable probes. From these data sets, 29 probes other psychiatric diseases in a previous proteomic relating to 23 specific proteins were found to be study from our group that focused on the anterior significantly differentially expressed at P < 0.05 in one cingulate cortex in a different brain series.33 Specifi- or more of the studies whose data sets were examined. cally, the increase in ATP synthase beta in bipolar More details about the proteins we found to be disorder was also found in patients with major abnormally expressed are given in Table 2. Of depression in the anterior cingulate cortex study, particular interest was the finding of significant and succinyl CoA ligase and alpha tubulin were alteration of septin 11 in two of the studies in bipolar found to be increased in bipolar disorder in both disorder (1 and 7) and two septin 6 probes in both studies. Proteins also found to be abnormally ex- schizophrenia and bipolar disorder in the Kato data pressed in a recent proteomic study of the dlPFC in set only. There was also significance seen in one of schizophrenia are D-3 phosphoglycerate dehydrogen- the septin 5 probes in both schizophrenia and bipolar ase, ubiquinol cytochrome c reductase, peroxiredoxin disorder in the Bahn data set; however, it should be 2, gamma enolase and aldehyde dehydrogenase.31 noted that this probe is not specific to septin 5 alone. However, in this study, these proteins were all found In total, 4 of the schizophrenia-associated and 19 of to be decreased in schizophrenia in comparison to the bipolar-associated proteins in this study were also controls, whereas in our study, with the exception of found to be abnormally expressed at the transcript peroxiredoxin 2, we found increases in these proteins level in one or both of these disorders in three of the in bipolar disorder in comparison to controls. We did microarray studies carried out on the same tissue. observe a decrease (À3.06-fold) in gamma enolase in Unfortunately, direction of change of transcript schizophrenia versus controls, which has also been expression could not be calculated directly from the found previously.31 This divergence could be inter- data provided on the website. We were therefore preted as providing further evidence for differential unable to ascertain whether gene and protein expres- disease expression of metabolic proteins, with mostly sion changes were in the same direction. Nonetheless, increases observed in bipolar disorder and decreases given the fact the microarray data sets were not in schizophrenia. However, it could also be explained adjusted for covariates, validation of differential by our accounting for the effect of potentially expression at the transcript level is encouraging and confounding factors such as age and PMI in the provides further evidence of abnormalities in these statistical analysis, factors that were not taken into proteins in these disorders. consideration in the study by Prabakaran et al. (2004). Therefore, this study extends the findings of Methodological issues mitochondrial and metabolic dysfunction in schizo- In proteomic studies of human post-mortem material phrenia and suggests that there may be greater it is useful to know what proteins might be altered dysfunction in this area in bipolar disorder. Although due to the influence of confounders. Other proteomic these findings are in contrast with a recent microarray study designs range from incorporating almost all study which found no evidence for alterations in the covariates (though notably not PMI) into their dlPFC, compared to 393 abnormally expressed gene ANCOVA analysis30 to none.31 In the present study transcripts in the orbitofrontal cortex,94 this study we adjusted for age, brain pH, PMI and refrigerator does support our findings of metabolic dysfunction in interval with a secondary analysis of the effects of bipolar disorder. As proteomic studies act to reveal these factors plus alcohol abuse and medication only potential candidate proteins involved in a status at time of death on protein expression in the disease process, further work examining mitochon- whole-brain series. The results of these analyses drial protein complexes in these disorders is needed suggest that many proteins shown to be altered in to confirm and extend these findings. expression by disease are also related to some of the confounding factors that exist within these patients Protein to transcript level: comparisons using the (see Table 2). Additionally, the importance of ac- Stanley genomics database counting for the effects of tissue pH on mRNA Tissue from the dlPFC (BA46) from the same samples expression in microarray work is now acknow- as those used in this study has been used previously ledged,34 indicating that these are important factors to in six microarray studies whose data sets are available consider when analysing high-throughput data sets. through the public online database www.stanleygeno- Some of the findings revealed in this study may mics.org.95 To investigate how our findings relate to appear to contradict previous research, for example

Molecular Psychiatry Proteomics of psychiatric disease in prefrontal cortex K Pennington et al 1114 an upregulation of glial fibrillary acidic protein family of proteins in major psychiatric disorders. (GFAP).17,30 More specifically, reduced levels of Based on our present findings, future work will seek phosphorylated GFAP have been reported in the to confirm and extend the involvement of septin PFC and hippocampus,96 and a downregulation of proteins, and also to characterize more fully the four forms of GFAP were also reported in another synaptic proteome in these disorders. proteomic study of the dlPFC.30 However, an increased protein expression of GFAP in lamina II, III, IV and VI in BA9 of the dlPFC has also been Acknowledgments reported.97 This previously reported increase in GFAP was found to correlate with antipsychotic intake; Post-mortem brains were donated by the Stanley however, post hoc analysis of this in our data set did Foundation Brain Bank Consortium courtesy of Drs not support this finding (Spearman’s rank 0.221). We Llewellyn B Bigelow, Juraj Cervenak, Mary M Her- consider that the reason for the inconsistency of these man, Thomas M Hyde, Joel Kleinman, Jose D Paltan, results may relate to the presence of GFAP on 2D gels Robert M Post, E Fuller Torrey, Maree J Webster and as multiple protein spots representing post-transla- Robert Yolken. We also acknowledge the contribution tional modifications and/or different isoforms.98 Non- of the SMRI in providing access to the Stanley significant trend reductions in other GFAP spots Genomic Database. We thank Professor Peter Hall, could have been present in the schizophrenia sample. Department of Pathology, Queens University Belfast, The investigation of the functional states of the for his useful and insightful discussions. In addition, proteins (phosphorylation and glycosylation states) we thank Lance Hudson for technical assistance in 2- found to be abnormally expressed is important to DE image analysis. Access to and use of MS determine the functional relevance of these different instrumentation of Conway Institute is gratefully findings. acknowledged and we thank Dr Niaobh O’Donoghue, Another issue concerns the use of crude brain Kaspar Pedersen and Kieran Wynne for their techni- tissue in all proteomic studies of psychiatric dis- cal assistance in mass spectrometry. Additionally, we orders undertaken to date, including our own. The thank Matt Sullivan and the proteomics informatics proteome assessed in such samples is obviously group ( http://proteomics.ucd.ie) for use of their restricted to the proteins that are separated on a 2D Proline software. This work was funded by a Well- gel, and these are often highly abundant with lower come Trust project award to DRC, The Stanley molecular weight. Additionally, for a small number of Medical Research Institute and Science Foundation spots, several proteins were identified within them, Ireland under Grant no. 04/RPI/B499 to MJD. RW therefore the quantitative data relating to these are received financial support from ARC and MRC. less certain and further work is required to validate these findings (see Table 2). In relation to our findings of altered septin expression in schizophrenia and References bipolar disorder, we feel that is worth noting the very 1 Frith C, Dolan R. The role of the prefrontal cortex in higher large number of septin proteins revealed to be cognitive functions. Brain Res Cogn Brain Res 1996; 5: 175–181. abnormal in this study, the consistent direction of 2 Goldman-Rakic PS, Selemon LD. Functional and anatomical this change and the consistency with the microarray aspects of prefrontal pathology in schizophrenia. Schizophr Bull data (as discussed above). We consider that this 1997; 23: 437–458. 3 Goldstein JM, Goodman JM, Seidman LJ, Kennedy DN, Makris N, represents robust internal confirmation of these Lee H et al. Cortical abnormalities in schizophrenia identified by changes. We propose to confirm these finding further structural magnetic resonance imaging. Arch Gen Psychiatry 1999; using 2D differential in gel electrophoresis in an 56: 537–547. independent brain series and also using subcellular 4 Sanfilipo M, Lafargue T, Rusinek H, Arena L, Loneragan C, Lautin enrichments for synaptic and membrane-associated A et al. Volumetric measure of the frontal and temporal lobe regions in schizophrenia: relationship to negative symptoms. Arch proteins. Gen Psychiatry 2000; 57: 471–480. 5 Pantelis C, Velakoulis D, McGorry PD, Wood SJ, Suckling J, Phillips LJ et al. Neuroanatomical abnormalities before and after Conclusions onset of psychosis: a cross-sectional and longitudinal MRI comparison. Lancet 2003; 361: 281–288. This human post-mortem proteomic study has re- 6 Drevets WC, Price JL, Simpson Jr JR, Todd RD, Reich T, Vannier M vealed many alterations in protein expression in the et al. Subgenual prefrontal cortex abnormalities in mood dis- dlPFC in both schizophrenia and bipolar disorder orders. Nature 1997; 386: 824–827. patients in comparison to controls. Almost half of the 7 Strakowski SM, Adler CM, DelBello MP. Volumetric MRI studies proteins found to be abnormally expressed in schizo- of mood disorders: do they distinguish unipolar and bipolar disorder? Bipolar Disord 2002; 4: 80–88. phrenia were found to be synaptic-associated and 8 Sheline YI. Neuroimaging studies of mood disorder effects on the many of these proteins have functions that involve brain. Biol Psychiatry 2003; 54: 338–352. the presynaptic machinery and NMDA receptor 9 Selemon LD, Rajkowska G, Goldman-Rakic PS. Elevated neuronal activity. In bipolar disorder, metabolic abnormality density in prefrontal area 46 in brains from schizophrenic patients: application of a three-dimensional, stereologic counting method. was the largest area of cellular dysfunction observed J Comp Neurol 1998; 392: 402–412. in this study. A particularly interesting insight gained 10 Pierri JN, Volk CL, Auh S, Sampson A, Lewis DA. Decreased somal from this study is the involvement of the septin size of deep layer 3 pyramidal neurons in the prefrontal cortex of

Molecular Psychiatry Proteomics of psychiatric disease in prefrontal cortex K Pennington et al 1115 subjects with schizophrenia. Arch Gen Psychiatry 2001; 58: 31 Prabakaran S, Swatton JE, Ryan MM, Huffaker SJ, Huang JT, Griffin 466–473. JL et al. Mitochondrial dysfunction in schizophrenia: evidence for 11 Rajkowska G, Selemon LD, Goldman-Rakic PS. Neuronal and glial compromised brain metabolism and oxidative stress. Mol Psy- somal size in the prefrontal cortex: a postmortem morphometric chiatry 2004; 9: 684–697, 643. study of schizophrenia and Huntington disease. Arch Gen 32 Clark D, Dedova I, Cordwell S, Matsumoto I. A proteome analysis Psychiatry 1998; 55: 215–224. of the anterior cingulate cortex gray matter in schizophrenia. Mol 12 Broadbelt K, Byne W, Jones LB. Evidence for a decrease in basilar Psychiatry 2006; 11: 459–470, 423. dendrites of pyramidal cells in schizophrenic medial prefrontal 33 Beasley CL, Pennington K, Behan A, Wait R, Dunn MJ, Cotter D. cortex. Schizophr Res 2002; 58: 75–81. Proteomic analysis of the anterior cingulate cortex in the major 13 Black JE, Kodish IM, Grossman AW, Klintsova AY, Orlovskaya D, psychiatric disorders: evidence for disease-associated changes. Vostrikov V et al. Pathology of layer V pyramidal neurons in the Proteomics 2006; 6: 3414–3425. prefrontal cortex of patients with schizophrenia. Am J Psychiatry 34 Vawter MP, Tomita H, Meng F, Bolstad B, Li J, Evans S et al. 2004; 161: 742–744. Mitochondrial-related gene expression changes are sensitive to 14 Rosoklija G, Toomayan G, Ellis SP, Keilp J, Mann JJ, Latov N et al. agonal-pH state: implications for brain disorders. Mol Psychiatry Structural abnormalities of subicular dendrites in subjects with 2006; 11: 615, 663–679. schizophrenia and mood disorders: preliminary findings. Arch 35 Li JZ, Vawter MP, Walsh DM, Tomita H, Evans SJ, Choudary PV et Gen Psychiatry 2000; 57: 349–356. al. Systematic changes in gene expression in postmortem human 15 Kolluri N, Sun Z, Sampson AR, Lewis DA. Lamina-specific brains associated with tissue pH and terminal medical conditions. reductions in dendritic spine density in the prefrontal cortex Hum Mol Genet 2004; 13: 609–616. of subjects with schizophrenia. Am J Psychiatry 2005; 162: 36 Fountoulakis M, Hardmeier R, Hoger H, Lubec G. Postmortem 1200–1202. changes in the level of brain proteins. Exp Neurol 2001; 167: 16 Cotter D, Mackay D, Chana G, Beasley C, Landau S, Everall IP. 86–94. Reduced neuronal size and glial cell density in area 9 of the 37 Franzen B, Yang Y, Sunnemark D, Wickman M, Ottervald J, dorsolateral prefrontal cortex in subjects with major depressive Oppermann M et al. Dihydropyrimidinase related protein-2 as a disorder. Cereb Cortex 2002; 12: 386–394. biomarker for temperature and time dependent post mortem 17 Knable MB, Torrey EF, Webster MJ, Bartko JJ. Multivariate analysis changes in the mouse brain proteome. Proteomics 2003; 3: of prefrontal cortical data from the Stanley Foundation Neuro- 1920–1929. pathology Consortium. Brain Res Bull 2001; 55: 651–659. 38 O’Farrell PH. High resolution two-dimensional electrophoresis of 18 Torrey EF, Barci BM, Webster MJ, Bartko JJ, Meador-Woodruff JH, proteins. J Biol Chem 1975; 250: 4007–4021. Knable MB. Neurochemical markers for schizophrenia, bipolar 39 Rabilloud T, Valette C, Lawrence JJ. Sample application by in-gel disorder, and major depression in postmortem brains. Biol rehydration improves the resolution of two-dimensional electro- Psychiatry 2005; 57: 252–260. phoresis with immobilized pH gradients in the first dimension. 19 Hakak Y, Walker JR, Li C, Wong WH, Davis KL, Buxbaum JD et al. Electrophoresis 1994; 15: 1552–1558. Genome-wide expression analysis reveals dysregulation of myeli- 40 Sanchez JC, Rouge V, Pisteur M, Ravier F, Tonella L, Moosmayer M nation-related genes in chronic schizophrenia. Proc Natl Acad Sci et al. Improved and simplified in-gel sample application using USA 2001; 98: 4746–4751. reswelling of dry immobilized pH gradients. Electrophoresis 1997; 20 Vawter MP, Crook JM, Hyde TM, Kleinman JE, Weinberger DR, 18: 324–327. Becker KG et al. Microarray analysis of gene expression in the 41 Gorg A, Postel W, Gunther S, Weser J, Strahler JR, Hanash SM et al. prefrontal cortex in schizophrenia: a preliminary study. Schizophr Approach to stationary two-dimensional pattern: influence of Res 2002; 58: 11–20. focusing time and immobiline/carrier ampholytes concentrations. 21 Glatt SJ, Everall IP, Kremen WS, Corbeil J, Sasik R, Khanlou N et al. Electrophoresis 1988; 9: 37–46. Comparative gene expression analysis of blood and brain provides 42 Yan JX, Wait R, Berkelman T, Harry RA, Westbrook JA, Wheeler CH concurrent validation of SELENBP1 up-regulation in schizophre- et al. A modified silver staining protocol for visualization of nia. Proc Natl Acad Sci USA 2005; 102: 15533–15538. proteins compatible with matrix-assisted laser desorption/ioniza- 22 Vawter MP, Barrett T, Cheadle C, Sokolov BP, Wood III WH, tion and electrospray ionization-mass spectrometry. Electrophor- Donovan DM et al. Application of cDNA microarrays to examine esis 2000; 21: 3666–3672. gene expression differences in schizophrenia. Brain Res Bull 2001; 43 Eastwood SL, Burnet PW, Harrison PJ. Decreased hippocampal 55: 641–650. expression of the susceptibility gene PPP3CC and other 23 Mirnics K, Middleton FA, Marquez A, Lewis DA, Levitt P. calcineurin subunits in schizophrenia. Biol Psychiatry 2005; 57: Molecular characterization of schizophrenia viewed by microarray 702–710. analysis of gene expression in prefrontal cortex. Neuron 2000; 28: 44 Sawada K, Barr AM, Nakamura M, Arima K, Young CE, Dwork AJ 53–67. et al. Hippocampal complexin proteins and cognitive dysfunction 24 Eastwood SL, Burnet PW, Harrison PJ. Decreased hippocampal in schizophrenia. Arch Gen Psychiatry 2005; 62: 263–272. expression of the susceptibility gene PPP3CC and other calcineur- 45 Owen MJ, O’Donovan MC, Harrison PJ. Schizophrenia: a genetic in subunits in schizophrenia. Biol Psychiatry 2005; 57: disorder of the synapse? BMJ 2005; 330: 158–159. 702–710. 46 Frankle WG, Lerma J, Laruelle M. The synaptic hypothesis of 25 Sawada K, Barr AM, Nakamura M, Arima K, Young CE, Dwork AJ schizophrenia. Neuron 2003; 39: 205–216. et al. Hippocampal complexin proteins and cognitive dysfunction 47 Harrison PJ, Weinberger DR. Schizophrenia genes, gene expres- in schizophrenia. Arch Gen Psychiatry 2005; 62: 263–272. sion, and neuropathology: on the matter of their convergence. 26 Owen MJ, O’Donovan MC, Harrison PJ. Schizophrenia: a genetic Mol Psychiatry 2005; 10: 40–68. disorder of the synapse? BMJ 2005; 330: 158–159. 48 Eastwood SL, Harrison PJ. Hippocampal synaptic pathology in 27 Frankle WG, Lerma J, Laruelle M. The synaptic hypothesis of schizophrenia, bipolar disorder and major depression: a study of schizophrenia. Neuron 2003; 39: 205–216. complexin mRNAs. Mol Psychiatry 2000; 5: 425–432. 28 Anderson L, Seilhamer J. A comparison of selected mRNA 49 Eastwood SL, Harrison PJ. Synaptic pathology in the anterior and protein abundances in human liver. Electrophoresis 1997; cingulate cortex in schizophrenia and mood disorders. A review 18: 533–537. and a Western blot study of synaptophysin, GAP-43 and the 29 Gygi SP, Rochon Y, Franza BR, Aebersold R. Correlation between complexins. Brain Res Bull 2001; 55: 569–578. protein and mRNA abundance in yeast. Mol Cell Biol 1999; 19: 50 Saito T, Guan F, Papolos DF, Lau S, Klein M, Fann CS et al. 1720–1730. Mutation analysis of SYNJ1: a possible candidate gene for 30 Johnston-Wilson NL, Sims CD, Hofmann JP, Anderson L, Shore chromosome 21q22-linked bipolar disorder. Mol Psychiatry 2001; AD, Torrey EF et al. Disease-specific alterations in frontal cortex 6: 387–395. brain proteins in schizophrenia, bipolar disorder, and major 51 Beasley CL, Honer WG, Bergmann K, Falkai P, Lutjohann D, Bayer depressive disorder. The Stanley Neuropathology Consortium. TA. Reductions in cholesterol and synaptic markers in association Mol Psychiatry 2000; 5: 142–149. cortex in mood disorders. Bipolar Disord 2005; 7: 449–455.

Molecular Psychiatry Proteomics of psychiatric disease in prefrontal cortex K Pennington et al 1116 52 Byk T, Dobransky T, Cifuentes-Diaz C, Sobel A. Identification and 73 Toro C, Deakin JF. NMDA receptor subunit NRI and postsynaptic molecular characterization of Unc-33-like phosphoprotein (Ulip), protein PSD-95 in hippocampus and orbitofrontal cortex in a putative mammalian homolog of the axonal guidance-associated schizophrenia and mood disorder. Schizophr Res 2005; 80: unc-33 gene product. J Neurosci 1996; 16: 688–701. 323–330. 53 Horiuchi M, Loebrich S, Brandstaetter JH, Kneussel M, Betz H. 74 Hall PA, Jung K, Hillan KJ, Russell SE. Expression profiling the Cellular localization and subcellular distribution of Unc-33-like human septin gene family. J Pathol 2005; 206: 269–278. protein 6, a brain-specific protein of the collapsin response 75 Murphy KC, Jones LA, Owen MJ. High rates of schizophrenia in mediator protein family that interacts with the neuronal glycine adults with velo–cardio–facial syndrome. Arch Gen Psychiatry transporter 2. J Neurochem 2005; 94: 307–315. 1999; 56: 940–945. 54 Kowara R, Chen Q, Milliken M, Chakravarthy B. Calpain-mediated 76 Weksberg R, Stachon AC, Squire JA, Moldovan L, Bayani J, truncation of dihydropyrimidinase-like 3 protein (DPYSL3) Meyri S et al. Molecular characterization of deletion breakpoints in response to NMDA and H2O2 toxicity. J Neurochem 2005; 95: in adults with 22q11 deletion syndrome. Hum Genet 2007; 120: 466–474. 837–845. 55 Modregger J, Ritter B, Witter B, Paulsson M, Plomann M. All three 77 Tunbridge EM, Weinberger DR, Harrison PJ. A novel protein PACSIN isoforms bind to endocytic proteins and inhibit endocy- isoform of catechol O-methyltransferase (COMT): brain expression tosis. J Cell Sci 2000; 113(Part 24): 4511–4521. analysis in schizophrenia and bipolar disorder and effect of 56 Praefcke GJ, McMahon HT. The dynamin superfamily: universal Val158Met genotype. Mol Psychiatry 2006; 11: 116–117. membrane tubulation and fission molecules? Nat Rev Mol Cell Biol 78 Kremer BE, Haystead T, Macara IG. Mammalian septins 2004; 5: 133–147. regulate microtubule stability through interaction with the 57 Husi H, Ward MA, Choudhary JS, Blackstock WP, Grant SG. microtubule-binding protein MAP4. Mol Biol Cell 2005; 16: Proteomic analysis of NMDA receptor-adhesion protein signaling 4648–4659. complexes. Nat Neurosci 2000; 3: 661–669. 79 Beites CL, Xie H, Bowser R, Trimble WS. The septin CDCrel-1 58 Predescu SA, Predescu DN, Timblin BK, Stan RV, Malik AB. binds syntaxin and inhibits exocytosis. Nat Neurosci 1999; 2: Intersectin regulates fission and internalization of caveolae in 434–439. endothelial cells. Mol Biol Cell 2003; 14: 4997–5010. 80 Beites CL, Campbell KA, Trimble WS. The septin Sept5/CDCrel-1 59 Fatemi SH, Earle JA, Stary JM, Lee S, Sedgewick J. Altered levels competes with alpha-SNAP for binding to the SNARE complex. of the synaptosomal associated protein SNAP-25 in hippocampus Biochem J 2005; 385: 347–353. of subjects with mood disorders and schizophrenia. NeuroReport 81 Kinoshita A, Noda M, Kinoshita M. Differential localization of 2001; 12: 3257–3262. septins in the mouse brain. J Comp Neurol 2000; 428: 223–239. 60 Thompson PM, Egbufoama S, Vawter MP. SNAP-25 reduction in 82 Caltagarone J, Rhodes J, Honer WG, Bowser R. Localization of a the hippocampus of patients with schizophrenia. Prog Neuropsy- novel septin protein, hCDCrel-1, in neurons of human brain. chopharmacol Biol Psychiatry 2003; 27: 411–417. NeuroReport 1998; 9: 2907–2912. 61 Sokolov BP. Expression of NMDAR1, GluR1, GluR7, and KA1 83 Kinoshita A, Kinoshita M, Akiyama H, Tomimoto H, Akiguchi I, glutamate receptor mRNAs is decreased in frontal cortex of Kumar S et al. Identification of septins in neurofibrillary tangles in ‘‘neuroleptic-free’’ schizophrenics: evidence on reversible Alzheimer’s disease. Am J Pathol 1998; 153: 1551–1560. up-regulation by typical neuroleptics. J Neurochem 1998; 71: 84 Choi P, Snyder H, Petrucelli L, Theisler C, Chong M, Zhang Y et al. 2454–2464. SEPT5_v2 is a parkin-binding protein. Brain Res Mol Brain Res 62 Le Corre S, Harper CG, Lopez P, Ward P, Catts S. Increased levels of 2003; 117: 179–189. expression of an NMDARI splice variant in the superior temporal 85 Whatley SA, Curti D, Das GF, Ferrier IN, Jones S, Taylor C et al. gyrus in schizophrenia. NeuroReport 2000; 11: 983–986. Superoxide, neuroleptics and the ubiquinone and cytochrome b5 63 Robinson PJ, Sontag JM, Liu JP, Fykse EM, Slaughter C, McMahon reductases in brain and lymphocytes from normals and schizo- H et al. Dynamin GTPase regulated by protein kinase C phrenic patients. Mol Psychiatry 1998; 3: 227–237. phosphorylation in nerve terminals. Nature 1993; 365: 163–166. 86 Prince JA, Blennow K, Gottfries CG, Karlsson I, Oreland L. 64 Sontag JM, Fykse EM, Ushkaryov Y, Liu JP, Robinson PJ, Sudhof Mitochondrial function is differentially altered in the basal ganglia TC. Differential expression and regulation of multiple dynamins. of chronic schizophrenics. Neuropsychopharmacology 1999; 21: J Biol Chem 1994; 269: 4547–4554. 372–379. 65 Kristiansen LV, Beneyto M, Haroutunian V, Meador-Woodruff JH. 87 Maurer I, Zierz S, Moller H. Evidence for a mitochondrial Changes in NMDA receptor subunits and interacting PSD proteins oxidative phosphorylation defect in brains from patients with in dorsolateral prefrontal and anterior cingulate cortex indicate schizophrenia. Schizophr Res 2001; 48: 125–136. abnormal regional expression in schizophrenia. Mol Psychiatry 88 Marchbanks RM, Ryan M, Day IN, Owen M, McGuffin P, Whatley 2006; 11: 737–747, 705. SA. A mitochondrial DNA sequence variant associated with 66 Toyooka K, Muratake T, Tanaka T, Igarashi S, Watanabe H, schizophrenia and oxidative stress. Schizophr Res 2003; 65:33– Takeuchi H et al. 14-3-3 protein eta chain gene (YWHAH) 38. polymorphism and its genetic association with schizophrenia. 89 Kato T, Kato N. Mitochondrial dysfunction in bipolar disorder. Am J Med Genet 1999; 88: 164–167. Bipolar Disord 2000; 2: 180–190. 67 Middleton FA, Peng L, Lewis DA, Levitt P, Mirnics K. Altered 90 Mah L, Zarate Jr CA, Singh J, Duan YF, Luckenbaugh DA, Manji expression of 14-3-3 genes in the prefrontal cortex of subjects with HK et al. Regional cerebral glucose metabolic abnormalities in schizophrenia. Neuropsychopharmacology 2005; 30: 974–983. bipolar II depression. Biol Psychiatry 2007; 61: 765–775. 68 Hennah W, Tuulio-Henriksson A, Paunio T, Ekelund J, Varilo T, 91 Sun X, Wang JF, Tseng M, Young LT. Downregulation in Partonen T et al. A haplotype within the DISC1 gene is associated components of the mitochondrial electron transport chain in the with visual memory functions in families with a high density of postmortem frontal cortex of subjects with bipolar disorder. schizophrenia. Mol Psychiatry 2005; 10: 1097–1103. J Psychiatry Neurosci 2006; 31: 189–196. 69 Thomson PA, Wray NR, Millar JK, Evans KL, Hellard SL, Condie A 92 Bezchlibnyk YB, Wang JF, McQueen GM, Young LT. Gene et al. Association between the TRAX/DISC locus and both bipolar expression differences in bipolar disorder revealed by cDNA array disorder and schizophrenia in the Scottish population. Mol analysis of post-mortem frontal cortex. J Neurochem 2001; 79: Psychiatry 2005; 10: 657–668, 616. 826–834. 70 Goff DC, Coyle JT. The emerging role of glutamate in the 93 Iwamoto K, Bundo M, Kato T. Altered expression of mitochondria- pathophysiology and treatment of schizophrenia. Am J Psychiatry related genes in postmortem brains of patients with bipolar 2001; 158: 1367–1377. disorder or schizophrenia, as revealed by large-scale DNA 71 Collier DA, Li T. The genetics of schizophrenia: glutamate not microarray analysis. Hum Mol Genet 2005; 14: 241–253. dopamine? Eur J Pharmacol 2003; 480: 177–184. 94 Ryan MM, Lockstone HE, Huffaker SJ, Wayland MT, Webster MJ, 72 Zarate Jr CA, Singh J, Manji HK. Cellular plasticity cascades: Bahn S. Gene expression analysis of bipolar disorder reveals targets for the development of novel therapeutics for bipolar downregulation of the ubiquitin cycle and alterations in synaptic disorder. Biol Psychiatry 2006; 59: 1006–1020. genes. Mol Psychiatry 2006; 11: 965–978.

Molecular Psychiatry Proteomics of psychiatric disease in prefrontal cortex K Pennington et al 1117 95 Higgs BW, Elashoff M, Richman S, Barci B. An online database for 97 Toro CT, Hallak JE, Dunham JS, Deakin JF. Glial fibrillary acidic brain disease research. BMC Genomics 2006; 7:70. protein and glutamine synthetase in subregions of prefrontal 96 Webster MJ, Knable MB, Johnston-Wilson N, Nagata K, cortex in schizophrenia and mood disorder. Neurosci Lett 2006; Inagaki M, Yolken RH. Immunohistochemical localization of 404: 276–281. phosphorylated glial fibrillary acidic protein in the prefrontal 98 Inagaki M, Gonda Y, Nishizawa K, Kitamura S, Sato C, Ando S et cortex and hippocampus from patients with schizophrenia, al. Phosphorylation sites linked to glial filament disassembly in bipolar disorder, and depression. Brain Behav Immun 2001; 15: vitro locate in a non-alpha-helical head domain. J Biol Chem 1990; 388–400. 265: 4722–4729.

Molecular Psychiatry