Molecular Psychiatry (2009) 14, 601–613 & 2009 Nature Publishing Group All rights reserved 1359-4184/09 $32.00 www.nature.com/mp ORIGINAL ARTICLE Proteomic analysis of membrane microdomain-associated in the dorsolateral prefrontal cortex in schizophrenia and bipolar disorder reveals alterations in LAMP, STXBP1 and BASP1 expression A´ T Behan1, C Byrne2, MJ Dunn3, G Cagney2 and DR Cotter1 1Department of Psychiatry, Royal College of Surgeons in Ireland, RCSI ERC, Smurfit Building, Beaumont Hospital, Dublin 9, Ireland; 2School of Biomolecular and Biomedical Science, Conway Institute for Biomolecular and Biomedical Science, University College Dublin, Belfield, Dublin 4, Ireland and 3School of Medicine, UCD Conway Institute, University College Dublin, Dublin 4, Ireland

The dorsolateral prefrontal cortex (dlpfc) is strongly implicated in the pathogenesis of schizophrenia (SCZ) and bipolar disorder (BPD) and, within this region, abnormalities in glutamatergic neurotransmission and synaptic function have been described. Proteins associated with these functions are enriched in membrane microdomains (MM). In the current study, we used two complementary proteomic methods, two-dimensional difference gel electrophoresis and one-dimensional sodium dodecyl sulphate polyacrylamide gel electro- phoresis followed by reverse phase-liquid chromatography-tandem mass spectrometry (RP- LC-MS/MS) (gel separation liquid chromatography-tandem mass spectrometry (GeLC-MS/MS)) to assess protein expression in MM in pooled samples of dlpfc from SCZ, BPD and control cases (n = 10 per group) from the Stanley Foundation series. We identified 16 proteins altered in one/both disorders using proteomic methods. We selected three proteins with roles in synaptic function (-binding protein 1 (STXBP1), brain abundant membrane-attached signal protein 1 (BASP1) and limbic system-associated membrane protein (LAMP)) for validation by western blotting. This revealed significantly increased expression of these proteins in SCZ (STXBP1 (24% difference; P < 0.001), BASP1 (40% difference; P < 0.05) and LAMP (22% difference; P < 0.01)) and BPD (STXBP1 (31% difference; P < 0.001), BASP1 (23% difference; P < 0.01) and LAMP (20% difference; P < 0.01)) in the Stanley brain series (n = 20 per group). Further validation in dlpfc from the Harvard brain subseries (n = 10 per group) confirmed increased protein expression in SCZ of STXBP1 (18% difference; P < 0.0001), BASP1 (14% difference; P < 0.0001) but not LAMP (20% difference; P = 0.14). No significant differences in STXBP1, BASP1 or LAMP protein expression in BPD dlpfc were observed. This study, through proteomic assessments of MM in dlpfc and validation in two brain series, strongly implicates LAMP, STXBP1 and BASP1 in SCZ and supports the view of a neuritic and synaptic dysfunction in the neuropathology of SCZ. Molecular Psychiatry (2009) 14, 601–613; doi:10.1038/mp.2008.7; published online 12 February 2008 Keywords: 2D DIGE; schizophrenia; bipolar disorder; membrane microdomain; LAMP; STXBP1 and BASP1

Introduction density8,9 and size,10 glial cell density10 and dendritic spine density11 in the dlpfc in these disorders (for There is converging evidence that a fundamental review see 6Harrison and Weinberger, 2005). Func- pathology of schizophrenia (SCZ) and bipolar dis- tional imaging and neurochemical and cytoarchitec- order (BPD) involves a dysfunction of synaptic tural studies reveal patterns of aberrant activation and transmission, neural connectivity and GABA-ergic1 2,3 alterations to synaptic transmission, in particular to and glutamatergic activity. Dorsolateral prefrontal 12 13 4,5 6 7 GABA-ergic and glutamatergic neurotransmission cortex (dlpfc) is implicated in SCZ and in BPD in the schizophrenic brain. These findings strongly with cytoarchitectural studies of this region in SCZ suggest that changes in synaptic transmission and reporting an array of alterations in cortical neuronal neural connectivity are a core feature of SCZ. Studies by Beasley et al.14 and others15,16 have Correspondence: Professor DR Cotter, Department of Psychiatry, investigated the proteome of the anterior cingulate Royal College of Surgeons in Ireland, Beaumont, Dublin, Ireland. E-mail: [email protected] cortex and dlpfc in the SCZ and BPD brain and Received 24 June 2007; revised 21 November 2007; accepted 27 revealed prominent synaptic and metabolic abnorm- November 2007; published online 12 February 2008 alities in both disorders. Most recently, our group Proteomics of MM-associated proteins in psychiatric disease in PFC A´T Behan et al 602 undertook a proteomic analysis of the dlpfc17 and tion. This collection contained 105 (35 control further characterized the synaptic and synaptic-asso- (CON), 35 SCZ, 35 BPD) dissected from the frontal ciated pathology in SCZ by demonstrating alterations superior gyrus of Brodmann Area 9. Frozen grey in the septin family of proteins, protein kinase C and matter dlpfc blocks were also obtained from the casein kinase in neurons 1 (PACSIN1) and neurofila- McLean 66 series of the Harvard Tissue Resource ment triplet L in SCZ. These findings complement the Centre (www.brainbank.mclean.org) for the purpose literature reporting changes in protein kinase C18 and of validation studies. The latter collection contained synaptic proteins19 in SCZ, although some negative 66 brains (22 CON, 22 SCZ, 22 BPD). Frozen grey studies have been published.6,20 These findings, matter dlpfc blocks were obtained from the Stanley together with the known functions of candidate Foundation Brain Tissue Array Collection. This susceptibility ,6 add to a growing body of collection contained 105 brains (35 control (CON), evidence that describes and characterizes the synaptic 35 SCZ, 35 BPD) dissected from the frontal superior and N-methyl D-aspartate receptor dysfunction in gyrus of Brodmann Area 9. Frozen grey matter dlpfc SCZ. In the current investigation, our aim was to blocks were also obtained from the McLean 66 series focus in more detail on these functions. As proteomic of the Harvard Brain Tissue Resource Centre studies of whole brain tissue are biased towards the (www.brainbank.mclean.org) for the purpose of vali- identification of the more abundant proteins, and dation studies. The latter collection contained 66 exclude many of the proteins that are of interest to us, brains (22 CON, 22 SCZ, 22 BPD) dissected from the we chose to use enrichment techniques that enable the inferior frontal gyrus of Brodmann Area 45. Diagnoses targeting of the less abundant and more functionally for SCZ and BPD in the Stanley brain series were specific subproteomes.21 made according to the Diagnostic and Statistical Membrane microdomains (MM) are specialized Manual of Mental Disorders (DSM-IV) criteria. Diag- structures within the plasma membrane that have an noses for SCZ were made according to Feighner’s altered lipid composition, links to the cytoskeleton criteria and according to the DSM-IV criteria for BPD and, more importantly, contain a vast pool of in the Harvard series. For both brain series, exposure receptors, trafficking proteins and to antipsychotic drugs, presented as chlorpromazine signal transduction proteins in the brain.22 In addi- (CPZE) equivalents, was estimated from knowledge of tion, they contain soluble NSF attachment receptor prescribed antipsychotic drugs at time of death, chart (SNARE) proteins and complexes, as well as impor- note review and discussion with family members tant elements in the control of regulated .23 regarding compliance. CPZE equivalents represent Proteins such as N-methyl D-aspartate receptors, the approximate dose equivalent to 100 mg of CPZE GABA receptors and glutamate transporters are pre- (relative potency), and are used as an indicator of sent in MM and are implicated in SCZ pathology.22 relative antipsychotic potencies of antipsychotic Previous proteomic studies have identified proteins drugs. associated with MM in a variety of cell types,24–28 Due to the limited availability of the amounts of with one study identifying MM proteins within tissue samples, we were obliged to pool samples from chemical synapses from the rat brain.29 No expression the 10/10/10 Stanley brain series into one representa- study of MM proteins in the human cortex proteome tive pooled sample for each group to obtain sufficient has yet been carried out, although some preliminary MM protein for both proteomic analyses. From our work has been undertaken.30 Stanley series, a 10/10/10 series of the best-matched The aim of this study was to assess differential brains for the shortest post-mortem delay, highest pH, protein expression of MM-associated proteins in age and gender was selected for the pooled study. For pooled grey matter dlpfc samples from control, SCZ our validation study, an extended 20/20/20 series and BPD subjects. Our strategy involved the use of two from the Stanley series and a 10/10/10 series from the complementary proteomic methods, two-dimensional Harvard brain series were selected. See Table 1 for difference gel electrophoresis (2D DIGE) and one- summaries relating to age, gender, post-mortem delay, dimensional sodium dodecyl sulphate polyacryl- tissue pH, brain side, brain weight and cause of death amide gel electrophoresis (1D SDS-PAGE), followed for each of these subseries. by RP-LC-MS/MS (GeLC-MS/MS) on pooled samples from each group to first identify interesting candidate Study design proteins and then to validate the findings, first, in the same brain series and then in another independent Step 1: identification of ‘candidate’ proteins. Due to brain series. Our validation was focussed on proteins the limited availability of the amounts of tissue involved in neuroplasticity that showed expression samples, we were obliged to pool samples from the changes in both SCZ and in BPD. 10/10/10 Stanley brain series into one representative pooled sample for each group to obtain sufficient MM protein for both proteomic analyses. Two Materials and methods complementary proteomic analyses (2D DIGE and Samples GeLC-MS/MS) were viewed as exploratory and as Frozen grey matter dlpfc blocks were obtained from representing a method of identifying ‘candidate’ the Stanley Foundation Brain Tissue Array Collec- proteins of interest. ‘Candidate’ proteins were

Molecular Psychiatry Proteomics of MM-associated proteins in psychiatric disease in PFC A´T Behan et al 603 Table 1 Demographical information for the Stanley 20/20/20 brain series and the Harvard McLean66 10/10/10 brain series used for analysing protein expression in the dlpfc region of the frontal cortex

Stanley Harvard

CON SCZ BPD CON SCZ BPD

Number of cases 20 20 20 10 10 10 Age 44.1 (7.6) 46.8 (11.0) 41.0 (6.7) 54.4 (14.6) 54.3 (18.7) 53.5 (16.7) Gender (male, female) 14M:6F 11M:9F 14M:6F 6M:4F 7M:3F 6M:4F Mean PMI and range 24.0 (7.6) 28.3 (8.1) 25.0 (7.0) 19.1 (4.4) 21.9 (3.9) 15.6 (6.0) Mean pH and range 6.5 (0.2) 6.6 (0.1) 6.5 (0.2) 6.3 (0.3) 6.4 (0.2) 6.5 (0.3) Side of brain (left, right) 10L, 10R 12L, 8R 10L, 10R 7L, 3R 7R, 3L 5L, 5R Brain weight (g) 1434.1 (162.9) 1449.0 (111.7) 1394.1 (153.1) 1381.0 (163.9) 1373.4 (215.9) 1264.4 (112.8) Causes of death 0 suicide 5 suicide 7 suicide 0 suicide 2 suicide 4 suicide 18 cardiac 6 cardiac 7 cardiac 5 cardiac 2 cardiac 1 cardiac 2 other 9 other 6 other 5 other 6 other 5 other

Abbreviations: BPD, bipolar disorder; CON, control; dlpfc, dorsolateral prefrontal cortex; PMI, post-mortem interval; SCZ, schizophrenia. Mean scores are given with standard deviations. The 10/10/10 Stanley subsample used as the pooled samples for our proteomic studies are contained within the 20/20/20 brain series. defined as proteins that demonstrated altered homogenization buffer to fit 4 ml Sorvall tubes expression in the same direction for both 2D DIGE and centrifuged for 90 min at 4 1C at 100 000 g. The and GeLC-MS/MS in SCZ and/or BPD. resultant membrane pellet was resuspended in 4 ml of Tris-buffered saline (1 Â TBS) in a Sorvall tube. Step 2: validation study. It was decided that validation experiments should focus on the most Isolation of the detergent-resistant membrane fraction interesting of these proteins. Building on our The resuspended membrane pellet was centrifuged findings14,17 and those of others15,16 of synaptic for 90 min at 4 1C at 100 000 g, and the pellet was changes in both SCZ and BPD, selection criteria for resuspended in 2 ml of lysis buffer (1 Â TBS contain- these were that (i) proteins demonstrated altered ing 2 m EDTA and 2% (vol/vol) Triton X-100). expression in both BPD and SCZ, and (ii) proteins The sample was sonicated as described above and were involved in synaptic and synaptic plasticity- then mixed at 4 1C for 90 min on a fixed-speed, associated functions. Proteins that met these criteria fixed-angle rotator. After mixing, an aliquot (0.5 ml) would be validated using western blotting in of the solubilized membranes was diluted with an individual cases in an extended Stanley Foundation equal volume of 80% (wt/vol) sucrose in TBS Array series with the validation then extended to the containing 2 mM EDTA, 2% (vol/vol) Triton X-100 Harvard Brain Tissue Resource Centre McLean 66 and 1:100 dilution of a protease inhibitor cocktail brain series. (Sigma-Aldrich, Dublin, Ireland). This sample was pipetted in as a 40% part (bottom—1 ml) of a Preparation of the total membrane fraction from dlpfc discontinuous sucrose gradient consisting of 30% Total membrane fractions were isolated as previously (wt/vol) sucrose (middle—2 ml) and 5% (wt/vol) described31 with some minor modifications as sucrose (top—1 ml), both solutions made up in detailed below. Grey matter dlpfc (50 mg) was TBS containing 2 mM EDTA and 1:100 protease dissected from each case and pooled within the inhibitor cocktail. The sucrose gradient was centri- necessary group (CON/SCZ/BPD) to form three fuged at 100 000 g for 24 h at 4 1C in a Sorvall RC- pooled samples of 500 mg dlpfc for CON, SCZ and M120EX ultracentrifuge. Ten fractions (400 ml) BPD groups. Grey matter dlpfc (500 mg) for each were harvested from the top to the bottom of the group was homogenized in 5 ml of 150 mM NaCl, centrifuge tube at 4 1C and precipitated with 80% 32 20 mM Na2HPO4,2mM NaH2PO4,1mM EDTA, and (v/v) acetone as described previously. The result- 20% (vol/vol) glycerol, pH 7.4. Homogenization was ing pellets from the 10 Â 400 ml fractions were achieved by 30 passes of a Dounce homogenizer resuspended in urea/thiourea buffer (7 M urea, 2 M equipped with a glass pestle, followed by sonication thiourea, 4% 3-[(3-cholamidopropyl)dimethylammo- (30% maximum power pulsed for 30% of the time) nio]-1-propanesulphonate, 30 mM Tris/HCl (pH 6.8)) for 5 Â 1-min periods with a 1-min cooling period and assayed for protein concentration using the between each round of sonication. After sonication, Bradford assay. For MS analysis, fractions 3–6 the samples were centrifuged at 5000 g at 4 1C (detergent-resistant membrane pellets B30 mg) were for 20 min, following which the supernatant was pooled for our two proteomic approaches (see next removed, made up to the correct volume with section).

Molecular Psychiatry Proteomics of MM-associated proteins in psychiatric disease in PFC A´T Behan et al 604 Determination of flotillin-1 and transferrin receptor above and fixed for a minimum of 1 h in a methanol/ distribution acetic acid/water solution (4:1:5 v/v/v). The resulting The methods for SDS-PAGE and western blotting 2D protein profiles were visualized using the PlusOne were undertaken as described previously with the silver staining kit (Amersham Biosciences) with following modifications.33 For western blot analysis, slight modifications to ensure compatibility with gels were transferred to nitrocellulose membranes and subsequent MS analysis.38 Spots of interest were probed with specific antibodies as described pre- picked using a manual spot picker and pooled from viously.34 Mouse monoclonal antibodies to flotillin-1 the four preparative gels. Spot plugs were destained39 (clone 18, 1:500 dilution) and transferrin receptor and proteins were digested with trypsin as previously (anti-CD71, 1:200 dilution) were purchased from BD described.35 Biosciences (Oxford, UK) and Chemicon (Hampshire, UK) respectively. LC-MS/MS. Tryptic peptides of the selected 2D gel spots were analysed using Thermo linear quadrupole Step 1 ion trap mass spectrometer as described previously.36 All MS/MS spectra were sequence database searched 2D DIGE. Pooled protein samples (30 mg) from the using TurboSEQUEST. The MS/MS spectra were three disease groups were run in triplicate in a two searched and identified against the non-redundant CyDye DIGE experiment. An internal standard was International Protein Index database (v.3.14). prepared with equal fractions of all pooled samples and bulk-labelled with 240 pmol Cy3 per 30 mgof GeLC-MS/MS. The MM-enriched samples were protein. A 30 mg aliquot of each of the nine samples reduced, solubilized in 2  Laemlli buffer and was labelled with 240 pmol of Cy5, according to the separated on 4–20% gradient precast acrylamide manufacturer’s protocol. After labelling, Cy3- and Cy5- gels (7 cm) (Pierce, MSC Ltd., Dublin, Ireland). Gels labelled proteins were combined and diluted 1:1 with were stained with colloidal Coomassie Brilliant Blue dilution buffer (9.5 M urea, 2% 3-[(3-cholamidopropyl) G-250 (Biorad, Hemel-Hampstead, Herts, UK) and dimethylammonio]-1-propanesulphonate, 2% dithio- each lane was cut into 10 bands (6 mm  6 mm) and threitol, 1.6% Pharmalyte, pH 3–10) and loaded onto digested in-gel with trypsin according to the method pH 3–10 24 cm immobilized pH gradient strips (GE of Shevchenko et al.39,40 The resulting mixtures were HealthCare, Cork, Ireland) using in-gel rehydration as resuspended in 1% formic acid and analysed by previously described.35 Strips were left to rehydrate nano-electrospray liquid chromatography mass overnight and subsequently focussed for 75 000 Vh at spectrometry. A high-performance liquid chromato- 20 1C. Following the IEF dimension, strips were graphy instrument (Dionex, UK) was interfaced with a equilibrated in 6 M urea containing 30% (w/v) Thermo linear quadrupole ion trap mass spectrometer glycerol, 2% (w/v) SDS and 0.01% (w/v) Bromo- (ThermoFinnigan, California, USA). Chromatography phenol blue, with the addition of 1% dithiothreitol buffer solutions (buffer A, 5% acetonitrile and 0.1% for 15 min. Strips were then incubated for a second formic acid; buffer B, 80% acetonitrile and 0.1% 15 min in the same buffer without dithiothreitol, but formic acid) were used to deliver a 60 min gradient with the addition of 4.8% (w/v) iodoacetamide. The (35 min to 45% buffer B, 10 min to 90%, hold 2D PAGE was performed according to standard 10 min, 3 min to 5%, hold for 15 min). A flow rate of protocols described by Focking et al.36 Second- 2 mlminÀ1 was used at the electrospray source. Spectra dimension SDS-PAGE was performed using 12% were searched using the SEQUEST algorithm as polyacrylamide gels run overnight at 20 mA per gel described previously against the International Protein at 15 1C (Ettan-DALT System; GE Healthcare) and was Index database (v.3.14). The probability-based evalua- terminated when the bromophenol dye front had tion program, Protein Prophet, was used for filtering migrated off the end of the gels. identifications.41

2D DIGE image analysis. Scanning of the gels with Analysis of proteomic data. CyDye-labelled proteins was performed as described previously36 on a Typhoon 9410 Image scanner 2D DIGE: decyder analysis Average spot intensity (Amersham Biosciences, Amersham, UK). Using the values per spot were calculated across the triplicate software package DeCyder v6.5, intra-gel spot gels for each pooled group and spots that detection and quantification was performed using demonstrated an average spot intensity X1.2-fold the differential in-gel analysis mode whereas images change between the three groups were defined as from different gels were matched using the biological altered in expression. These spots were excised from variance analysis mode. Protein spots with an average preparative gels and identified by tandem mass expression level X1.2-fold change (threshold based spectrometry. on recent publications17,37) across triplicate gels were defined as being altered. GeLC-MS/MS: ion spectral count For crude quanti- fication at the peptide level, spectral counts (the Spot picking and processing. Preparative 2D gels number of tandem MS events leading to a productive (4 Â 100 mg preparative gels) were run as described protein identification) for each peptide spectrum

Molecular Psychiatry Proteomics of MM-associated proteins in psychiatric disease in PFC A´T Behan et al 605 were summed to give a total peptide count for each of BPD and SCZ) exposed to these drugs, with those protein. The sum of peptide counts for each disease patients who were not exposed (Student’s t-test). In group was calculated and tabulated as fold changes. addition, correlation analysis between protein Proteins that (i) have an ion spectral count of greater expression data in the Stanley and Harvard brain than 5 in at least one of the two groups (that is CON vs series and the dose of prescribed neuroleptic drugs BPD or CON vs SCZ) and (ii) a fold change of greater (CPZE equivalents), age, post-mortem interval and than 1.2 were defined as altered for the purpose of tissue pH was assessed. The effects on protein this investigation. expression of gender in the patient and control sample, and death by suicide within the patient Step 2 groups, were investigated (Student’s t-test). The mean coefficient of variation was determined for candidate Validation of proteins using western blotting. The proteins in both brain series. To control for inter-gel levels of these ‘candidate’ proteins outlined in the variation, the mean covariance of the internal Study design section were quantified by western standards run on each gel was determined. All blotting in the extended 20 CON/20 SCZ/20 BPD parametric statistical analyses were carried out in brain series from the Stanley and, separately, the 10/ SPSS13 for Windows (SPSS). Deviations from 10/10 Harvard brain collection. All were individually normality were assessed using Shapiro–Wilks test analysed in triplicate using whole-cell lysates as and non-parametric analyses based on ranks described previously.34 Equal amounts of denatured (Kruskal–Wallis and Wilcoxon rank sum tests). The homogenates (30 mg per lane) were loaded and influence of outliers on our results was investigated resolved on 10% SDS-PAGE gels as described using the Grubb’s test. before.34 Gels were blotted onto nitrocellulose mem- branes, blocked and incubated overnight at 4 1Cwith polyclonal antibodies against syntaxin-binding protein Results 1 (STXBP1) (1:1000, Synaptic Systems, Goettingen, Germany), brain abundant membrane-attached signal Flotillin-1 and transferrin receptor distribution protein 1 (BASP1) (1:750, Abcam, Cambridge, UK) or The position of the MM-associated proteins within monoclonal limbic system-associated membrane the gradient was determined by immunoblotting for protein (LAMP) (1:500, a kind gift from Dr Pat Levitt, flotillin-1, a known MM marker42,43 and showed Vanderbilt University, Tennessee), followed by flotillin-1 to be enriched in fractions 3–6, represent- incubation with the appropriate species-specific ing the 5–30% interface of the sucrose gradient horse radish peroxidase-linked secondary antibody (Figure 1). In keeping with previous reports that the (1:2000 dilution, DAKO, Dublin, Ireland). Blots were transferrin receptor is present in non-detergent-resis- incubated with ECL reagent (Amersham Biosciences), tant membrane fractions and is solubilized under exposed to X-ray film and developed accordingly. The the above conditions,44 this receptor was found in optical density of the protein bands was measured the soluble membrane fractions only (fraction 10; using Adobe Photoshop Image Acquisition Software. Figure 1). Band densities were normalized (i) against the optical density of the MemCode stained lane (total protein) for that sample (Pierce) and (ii) against an internal 5% 40% standard run on each gel (used to reduce gel-to-gel variation). The mean relative expression for all the three proteins was calculated. 1 2 3 4 5 6 7 8 9 10

Statistical analysis of western blotting data. Using the mean relative expression values from our western Flotillin 1 blotting analysis, analysis of variance (ANOVA) was run using diagnoses as factors. If ANOVA indicated a significant effect of diagnosis, post hoc analyses Transferrin R using the independent samples t-test of individual disease groups compared with controls were under- MM SOL taken. Protein expression was deemed significantly different from control levels if the group mean Figure 1 Western blotting was performed for the MM- differed from control mean group value at the 5% specific marker, flotillin-1 and membrane-soluble marker, level. Post-mortem interval and brain pH were transferrin receptor on each of the 10 fractions removed not significantly different between groups and were following sucrose density-gradient centrifugation. MM represents the MM-enriched fractions and SOL represents not included in the primary analysis. The effects of the soluble fractions. Fractions 3–6 were positive for the exposure to antipsychotic drugs, antidepressant drugs MM marker flotillin-1. DRM fractions 3–6 did not contain and mood stabilizers on LAMP, BASP1 and STXBP1 transferrin receptor, which was localized in the soluble protein expression were assessed by (i) comparing membrane fraction. DRM, detergent-resistant membrane; protein levels among those patients (combined group MM, membrane microdomains.

Molecular Psychiatry Proteomics of MM-associated proteins in psychiatric disease in PFC A´T Behan et al 606 Step 1 MS proteins such as spectral counts, their presence/ absence in MM to date and fold changes for the 2D DIGE and MALDI-TOF. A total of 1496 individual ‘candidate’ proteins of interest identified. Of the 56 spots were visualized across nine gels of MM- proteins identified by GeLC/MS-MS, 64.3% (36 of 56) associated proteins (pH 3–10) (Figure 2). Average were previously associated with MM in brain tissue values per spot were calculated across the triplicate and/or other cell types (see Supplementary Table S3). gels for each pooled group and the spots that The proteins were rich in channel, vesicle and demonstrated a X1.2-fold change between the three cytoskeletal proteins, and proteins involved in groups were selected for identification. Of these 101 neurological processes. The proteins were monitored selected spots, 96 spots of interest were visually across the pooled sample series using spectral counts confirmed on the preparative gel and identified by as a semiquantitative measure of protein abundance. mass spectrometry. Of these 96 spots, a total of 75 spots were identified as 51 proteins across disease groups using LC-MS/MS. Of the 75 spots charac- Selection of MM proteins for validation. As detailed terized by our DIGE proteomic study, 51 proteins were in the Study design above, we applied a series of identified and 56.8% (29 of 51) of these were criteria to our selection of MM proteins for validation. previously associated with MM in brain tissue and/ This yielded 16 proteins that were defined as or other cell types (see Supplementary Table S1). ‘candidate’ proteins on the basis of them showing a Identities of all protein spots, their presence/absence fold change in the same direction in disease groups by in MM to date, their fold change, their theoretical both proteomic methods. These proteins (Table 2) isoelectric point (pI) and mass and the peptide could be broadly classified into synaptic, metabolic, sequences matched for each protein are shown in signalling and cytoskeletal functional groups and 12 of these 16 have been previously associated with MM Supplementary Table S1. A representative 2D spot 45,51,46–50,52 map for MM proteins in the dfplc is outlined in in brain tissue (Table 2). Three proteins Figure 2. were differentially expressed in both SCZ and BPD, and involved in synaptic-associated functions (italics GeLC-MS/MS. Fifty-six proteins were identified in Table 2). These three proteins, LAMP, BASP1 and following 1D SDS-PAGE and RP-LS-MS/MS from STXBP1, were selected for validation. 494 peptides (B9 unique peptides per protein). Supplementary Table S2 outlines the X!Tandem Step 2 scores and peptide sequences for identified peptides. Supplementary Table S3 contains Validation by western blotting. For our validation, information for all proteins identified by GeLC-MS/ each protein migrated with a single band of the predicted molecular weight, that is, STXBP1 (66 kDa), BASP1 (40 kDa) and LAMP (60 kDa). Representative western blots for subjects in the three diagnostic groups are shown in Figure 3. Quantitative variability within these gels was assessed by looking at the variation in the optical density for the internal standard across all gels. We found that covariance for the internal standard across the gels was < 5% in both brain series.

Protein expression in the Stanley array subseries. The quantitative results are illustrated in a scatter plot distribution in Figure 4 containing mean values and s.d. for each group. ANOVA revealed a significant

effect of diagnosis for each of STXBP1 (F2,56 = 9.41; P < 0.001), BASP1 (F2,56 = 4.56; P < 0.01) and LAMP (F2,56 = 5.77; P < 0.01) in the extended Stanley 20/20/ 20 series. Planned post hoc comparisons showed increased levels of expression in SCZ compared to controls for STXBP1 (1.41±0.23; 24% difference;

t37 = 3.98; P < 0.001), BASP1 (1.95±1.16; 40% differ- ence; t37 = 2.82; P < 0.05) and LAMP (1.35±0.30; 22% difference; t37 = À2.76; P < 0.01) and increased expression of STXBP1 (1.55±0.49; 31% difference;

t38 = À3.72; P < 0.001), BASP1 (1.57±0.67; 23% differ- Figure 2 Representative 2D gel for MM-associated proteins ence; t38 = 3.35; P < 0.01) and LAMP (1.31±0.33; 20% from the dlpfc in the human brain. Spots highlighted and difference; t38 = 2.82; P = P < 0.01) in BPD in the numbered refer to the protein spot numbers listed in Table 2. Stanley series.

Molecular Psychiatry Table 2 List of all 16 proteins altered by both 2D-DIGE (LC-MS/MS) and GeLC-MS/MS (validated proteins are shown in italics)

Protein Accession Present Direction of change Average ratio Spot Theoretical Observed Functional group no. in MM (ion spectral count) (DIGE) no. pI/mass pI/mass

SCZ BPD SCZ BPD

Synaptic Limbic system-associated membrane Q13449 n 2.60 2.60 1.64 1.36 1 553/37393 5.15/46000 Cell adhesion molecule protein (LAMP) activity/synaptic plasticity 1.67 1.29 2 8.40/46000 Brain acid soluble protein 1 P80723 y 1.50 2.25 1.20 1.32 3 4.64/22562 7.50/55500 Synaptic plasticity/ (BASF)45 transcription regulator activity 1.64 4 7.40/48000 1.67 5 7.99/41500 Syntaxin-binding protein 146 P61764 y 1.30 1.00 1.46 1.45 6 6.50/6756S 6.20/33500 Neurotransmitter transporter activity/synaptic plasticity Contactin47 Q12860 y À1.40 — À1.35 — 7 5.62/ 5.40/87000 Synaptic plasticity 113320 Neural cell adhesion molecule PI 3592 y 1.50 — 1.35 — 8 4.77/83770 5.00/75000 Cell adhesion molecule (NCAM)48 activity/synaptic plasticity 1.55 9 4.82/70000 Prohibitin49 P35232 y À2.40 — À1.25 — 10 5.57/29804 6.03/30000 Synaptic signalling Metabolic Vacuolar ATP synthase subunit B P212S1 n 3.20 — 1.35 — 11 4.89/40329 5.90/12750 ATPase activity Vacuolar ATP synthase subunit A50 P38606 y À2.75 — À1.41 — 12 5.22/18360 5.50/22500 Transporter activity rtoiso Mascae rtisi sciti ies nPFC in disease psychiatric in A proteins MM-associated of Proteomics Voltage-dependent anion-selective P21796 y À1.90 À2.42 À1.51 13 6.32/38092 8.45/38000 Voltage-gated ion channel ´ channel protein 149 activity Behan T À1.36 14 8.30/38500

NADH-ubiquinone oxidoreductase Q9UI09 n 9.00 — 1.43 — 15 5.77/13327 6.01/13500 Oxidoreductase activity al et (13 kDa)-B subunit Creatine kinase B chain51 PI 2277 y 3.75 — 1.48 — 16 5.34/42644 6.35/46500 Catalytic activity: phosphotransferase 17 6.28/46500 Signalling Haemoglobin delta chain51 P02042 y À4.50 — À1.51 — 18 7.84/16055 7.80/30000 Transporter activity Guanine nucleotide-binding protein P62873 y 1.63 1.37 1.34 1.24 20 5.6/37245 6.05/39500 GTPase activity G (I)52 Cytoskeletal Beta actin51 P60709 y — À1.92 — À1.46 19 5.29/41737 5.40/44100 Structural constituent of cytoskeleton Tubulin, beta chain51 P07437 y 1.21 1.20 1.42 À1.12 21 4.78/49670 4.8/17000 Structural constituent of cytoskeleton 1.22 22 6.7/42000 Tubulin, alpha chain Q71U36 y 1.90 — 1.39 — 23 4.94/50135 5.2/59000 Structural constituent of cytoskeleton

Abbreviations: BPD, bipolar disorder; pI, isoelectric point; MM, membrane microdomains; SCZ, schizophrenia. oeua Psychiatry Molecular Protein identities are listed in the table, along with its Swiss-Prot accession number, presence/absence in MM to date, the direction of change for ion spectral count, the average ratio calculated for each protein and the functional group for each protein. Also listed for the DIGE results are the spot IDs (see Figure 2), and theoretical and observed pI/molecular weights for each protein. All proteins found in brain MM to date have been referenced within the table. 607 Proteomics of MM-associated proteins in psychiatric disease in PFC A´T Behan et al 608 parametric analyses based on ranks (Kruskal–Wallis and Wilcoxon rank sum tests), which gave P-values on a similar scale (data not shown). Using the Grubb’s test, no outliers were detected in the Harvard series while two outliers were detected (P < 0.01) in the Stanley series: one in the BPD group for STXBP1 (outlier value = 3.26) and one in the control group for BASP1 (outlier value = 2.32). The affected compari- sons were re-analysed with these outliers excluded and this showed no significant impact upon the results (data not shown).

Potentially confounding effects of drugs and post- mortem factors. In the Stanley subsample, there were no significant relationships between expres- sion values for LAMP, STXBP1 or BASP1 and brain pH, post-mortem interval, gender or age. In the Stanley subsample, subjects exposed to neuro- leptics, antidepressant drugs or mood stabilizers did not differ significantly in the protein expression levels of LAMP, BASP1 or STXBP1 compared to those patients who were not exposed. In the extended Stanley 20/20/20 subsample used for validating our proteomic findings, subjects (n = 40) that died from suicide (n = 12) did not differ significantly in Figure 3 Representative western blots demonstrating the protein expression levels of LAMP (t38 = 0.43; banding patterns for STXBP1, BASP1 and LAMP proteins P = 0.78), BASP1 (t38 = 1.11; P = 0.28) or STXBP1 in the dlpfc. Each blot is representative of n = 3 experiments (t38 = 0.28; P = 0.67) compared to those subjects who and molecular weight is demarcated in kDa according to the died of natural causes (n = 28) within our psychiatric molecular weight marker. All blots were normalized prior to groups. quantitative comparisons. BASP1, brain abundant mem- In the Harvard brain series, there were no signifi- brane-attached signal protein 1; BPD, bipolar disorder; cant relationships between expression values for CON, control; dlpfc, dorsolateral prefrontal cortex; LAMP, LAMP, STXBP1 or BASP1 and brain pH, post-mortem limbic system-associated membrane protein; SCZ, schizo- interval, gender or age. In the Harvard subsample, phrenia; STXBP1, syntaxin-binding protein 1. subjects exposed to neuroleptic or to antidepressant drugs did not differ significantly in protein expres- sion levels of LAMP, BASP1 or STXBP1 compared to those patients who were not exposed. Subjects Protein expression in the Harvard subseries. Analysis exposed to mood stabilizers (n = 9) showed reduced of variance also revealed a significant effect of expression of BASP1 (P = 0.01) and STXBP1

diagnosis for each of STXBP1 (F2,27 = 6.79; P < 0.01) (P < 0.001) compared to those who were not (n = 7). and BASP1 (F2,27 = 9.99; P < 0.001). ANOVA did not In the Harvard 10/10/10 subsample, subjects (n = 20) reveal a significant effect of diagnosis for LAMP that died from suicide (n = 6) did not differ signifi-

(F2,27 = 1.31; P = 0.29) in the Harvard brain series. cantly in the protein expression levels of LAMP

Planned post hoc comparisons showed increased (t18 = À0.48; P = 0.64), BASP1 (t18 = 0.77; P = 0.45) or levels of expression in SCZ for STXBP1 (1.05±0.05; STXBP1 (t18 = 0.28; P = 0.78) compared to those sub- 18% difference; t18 = À5.12; P = P < 0.0001), BASP1 jects who died of natural causes (n = 14) within our (1.10±0.04; 14% difference; t18 = À6.91; P = psychiatric groups. There was no information avail- P < 0.0001) but not LAMP (2.06±0.48; 20% able on four Harvard subjects from the disease groups.

difference; t18 = À1.53; P = 0.14). Planned post hoc No significant correlations were observed between comparisons showed no significant changes in the neuroleptic drug exposure, expressed as CPZEs and expression of STXBP1 (0.89±0.16; 4% difference; protein expression for LAMP, BASP1 and STXBP1 in

t18 = 0.60; P = 0.56), BASP1 (0.93±0.14; 2% the Stanley subsample series. Overall, there was no difference; t18 = 0.05; P = 0.82) and LAMP (2.11± correlation between the levels of the three proteins in 0.84; 22% difference; t18 = 1.32; P = 0.20) in BPD. the Stanley series while, in the complete Harvard Deviations from normality were detected (P < 0.01) series, levels of STXBP1 and BASP1 protein expres- for STXBP1/BASP1 in the Stanley series. However, sion correlated positively with one another (r = 0.93; deviations from normality do not necessarily invali- P < 0.0001). Within individual disease groups for the date the standard statistical procedures used in this Harvard series, STXBP1 and BASP1 expression paper (ANOVA, t-test). The robustness of our results correlated quite strongly with one another within using ANOVA and t-tests was confirmed using non- the SCZ (r = 0.93; P < 0.001) and BPD (r = 0.90;

Molecular Psychiatry Proteomics of MM-associated proteins in psychiatric disease in PFC A´T Behan et al 609 STANLEY HARVARD

ab1.20

3.00 1.10

S S 1.00 T T X 2.00 X B B 0.90 P P 1 1 0.80 1.00 0.70

0.00 155 ± 0.49 1.07 ± 0.30 1.41 ± 0.230.60 0.89 ± 0.16 0.92 ± 0.06 1.05 ± 0.05 BPD CTRL SCZ BPD CTRL SCZ c 6.00 d 20

5.00 10 B 4.00 B A A 30 S S P 3.00 P 1 1 40 2.00

1.00 0.80

0.00 1.57 ± 0.64 1.19 ± 0.34 1.95 ± 1.01 0.70 0.93 ± 0.14 0.94 ± 0.06 1.10 ± 0.04

BPD CTRL SCZ BPD CTRL SCZ

e 2.50 f 3.50

3.00 2.00 L L 2.50 A A M M P 1.50 P 2.00

1.50 1.00 1.00

0.50 1.31 ± 0.33 1.05 ± 0.24 1.35 ± 0.30 0.50 2.11 ± 0.84 1.66 ± 0.67 2.06 ± 0.48

BPD CTRL SCZ BPD CTRL SCZ

*p<0.05; **p<0.01; ***p<0.0001 Figure 4 Protein expression changes of STXBP1 (a and b), BASP1(c and d) and LAMP (e and f) in the Stanley and Harvard series (10 CON/10 SCZ/10 BPD). *P < 0.05; **P < 0.01, ***P < 0.0001. P-values were identified by analysis of variance using diagnoses as contrasts. BASP1, brain abundant membrane-attached signal protein 1; BPD, bipolar disorder; LAMP, limbic system-associated membrane protein; SCZ, schizophrenia; STXBP1, syntaxin-binding protein 1.

P < 0.0001) groups. This correlation was also observed experiments to confirm the protein changes in whole for STXBP1 and BASP1 within the SCZ group tissue samples. Our strongest findings in this study (r = 0.51; P < 0.05) in the Stanley series. were for STXBP1 and BASP1 changes in SCZ and less strong for STXBP1 and BASP1 changes in BPD and for LAMP changes in both SCZ and BPD. Our findings Discussion provide novel insights into the neuropathology of Studies to date suggest that MM are necessary for (i) these disorders and implicate synaptic and neuritic release and (ii) the recruitment of abnormalities in these disorders. functionally linked proteins that underlie neurotrans- STXBP1, a known MM protein,23 was significantly mitter vesicle dynamics and exocytosis. Synaptic increased in SCZ in the Stanley and Harvard brain dysfunction in SCZ is described,19 but as yet is series. STXBP1 was also significantly increased in incompletely characterized. Our primary aim in the BPD in the Stanley subsample, but not altered in BPD current investigation was to investigate synaptic in the Harvard brain series. STXBP1 is a presynaptic function and neuronal plasticity in SCZ and BPD. protein that binds to the closed conformation of For this reason, we undertook a proteomic analysis of syntaxin and thus inhibits its assembly into the SCZ and BPD disorder dlpfc tissue samples enriched SNARE complex.53 Release from STXBP1 and activa- with MM. We then undertook detailed validation tion of the open conformation of syntaxin is key to

Molecular Psychiatry Proteomics of MM-associated proteins in psychiatric disease in PFC A´T Behan et al 610 synaptic membrane fusion, triggering the release of implicated in SCZ70,71 by microarray studies,72 which into the synaptic gap.54 Thus, our have shown a significant increase in its expression in observation of increased STXBP1 could suggest a the DLPFC in SCZ.72 reduced syntaxin availability for SNARE complex Neurite formation is a crucial stage in the develop- assembly and a consequent reduction in neurotrans- ment and plasticity of the neuron. We have shown mitter release. This, however, must be interpreted LAMP, STXBP1 and BASP1 to be overexpressed in with some caution as the role of STXBP1 in the SCZ and BPD and these proteins have prominent regulation of syntaxin and neurotransmitter release roles in neurite formation and outgrowth.64,68 is incompletely characterized.55 Several studies STXBP1, together with syntaxin 1, is involved in the have examined syntaxin expression in SCZ and regulation of SNARE-dependent membrane fusion for have demonstrated an increase in the mRNA level membrane expansion during neurite formation.73 in the temporal cortex56 and in protein levels in These results suggest that an altered circuitry, as a the cingulate cortex.57,58 However, no changes were result of defective proteins that affect neurite forma- reported in total syntaxin protein expression in tion and outgrowth, may be a component of the the parietal cortex,57 temporal cortex,57 frontal pathology of both SCZ and BPD. Several genes such as cortex57,59,60 or the cerebellum61 in SCZ. Future NRG-1, Akt1 and DISC have been shown to confer studies would need to examine (i) the relationship susceptibility for SCZ and proteins encoded by these of STXBP1 with syntaxin and SNARE complex genes are implicated in neurite formation in cellular formation as well as (ii) the expression of both and animal models of neurodevelopment.74 In addi- proteins across several brain regions in SCZ. tion, the synaptic proteins VAMP-2 and SNAP-25 BASP1 demonstrated significantly increased levels have demonstrated expression changes in SCZ and of expression in SCZ in the Stanley and Harvard BPD and are also involved in neurite elongation and subsample series and significantly increased levels in sprouting.75 Further expression studies of proteins BPD in the Stanley subsample series when compared involved in neurite formation in these diseases are to controls. A single previous proteome study of un- required to strengthen a hypothesis such as this. It is enriched SCZ dlpfc tissue concurs with our results in worth noting that previous expression studies in SCZ demonstrating increased BASP1 in SCZ.16 Microarray and BPD generally reported a tendency towards a studies by Fienberg’s group observed a significant reduced expression in SCZ and BPD,76,77 with increase in BASP1 expression in the of the increases in some synaptic-associated proteins, such SCZ brain.62 BASP1 has been found previously in as SNAP-2560 and conflicting reports regarding other MM in the brain.63 It is a cytoskeleton-associated, such as GAP-43 and synaptophysin.37,78–82 calmodulin-binding protein that is widely and A strength of the current investigation is the use of abundantly expressed during brain development, two complementary proteomic approaches, 2D DIGE maintained in selected brain structures in the adult and GeLC-MS/MS to identify candidate disease- and reinduced during nerve regeneration.64 BASP1 is associated proteins within the MM proteome in SCZ functionally related to GAP-43. Both proteins play a and BPD. The use of these methods has allowed us to critical role in regulating neuritic outgrowth as well identify and then to proceed to validate important as the actin cytoskeleton and BASP1 can functionally new disease-associated proteins within a functionally substitute for GAP-43 in vivo.64 Our results for BASP1 important but low-abundance proteome. Another expression complement recent findings by Prabakar- major strength of the investigation is the extensive an et al.16 who also demonstrated a > 2-fold increase validation that we undertook to confirm the disease- in BASP1 expression in the white matter of schizo- associated protein changes. This involved under- phrenic brains. Interestingly, reductions in GAP-43 taking validation experiments not just in a subsample and protein expression have been reported in the of the Stanley Foundation Brain series, but also in a SCZ and BPD brain.65–67 subsample of the Harvard McLean 66 Brain series. We Limbic system-associated membrane protein is confirmed protein changes in two independent brain characterized in this paper for the first time as a series for STXBP1 and BASP1, which provides very MM-associated protein. LAMP is a glycoprotein strong evidence that the proteins STXBP1 and BASP1 expressed on the surface of somata and proximal are robust and worthy of further investigation in SCZ. dendrites of neurons in cortical and subcortical We did not replicate our findings for the three regions of the limbic system.68,69 We found that LAMP proteins in BPD in the Harvard brain series. One demonstrated significantly increased levels of expres- possible explanation for this relates to the fact that sion in SCZ and BPD in the Stanley subsample our tissue samples were from different regions of the compared to controls. In the Harvard subsample, dlpfc and that region-specific differences in the significant changes for both SCZ and BPD groups expression of STXBP1, BASP1 and LAMP within were not observed but percentage changes were of the the dlpfc could explain our loss of significance in the same order as the Stanley subsample series. As for Harvard brain series. Further work is required to BASP1, LAMP is involved in mediating neuritic characterize the potential region-specific expression outgrowth and depending on the neuronal popula- patterns of STXBP1, BASP1 and LAMP, in addition to tion, it can enhance or inhibit neurite outgrowth.70 clarifying the presence and nature of these protein LAMP is expressed in limbic structures and has been changes in BPD and the expression of the LAMP

Molecular Psychiatry Proteomics of MM-associated proteins in psychiatric disease in PFC A´T Behan et al 611 protein in SCZ. As our validation criteria investigated Trifluoperazine decreases synaptophysin expression those synaptic proteins altered in both SCZ and BPD, within the hippocampal CA1 subregion, while halo- we did not pursue those equally interesting synaptic peridol and CPZE increase SNAP-25 expression proteins NCAM, prohibitin and contactin. It is worth throughout the hippocampus. Haloperidol treatment noting that our NCAM data complement previous has also been shown to upregulate synaptophysin findings that show NCAM to be increased in SCZ.83 mRNA in the striatum and frontoparietal cortex.85 Future work by our group will examine the nature There is no animal literature assessing the influence and extent of changes of these latter proteins in SCZ of psychotropic medications on the expression of and BPD. LAMP, STXBP1 or BASP1 specifically within the A potential limitation of the study is the use of mammalian brain. Future studies should aim to pooled samples. However, it is increasingly accepted clarify in detail the influence of psychotropic agents, that validation of proteomic data is absolutely critical particularly mood stabilizers, on synaptic- and neuri- to acceptance and interpretation of findings. Conse- tic-associated proteins that are also involved in quently, it can be argued that where clinical samples neuronal plasticity. are limited and precious, it is an efficient and valid To conclude, our observed changes in STXBP1, strategy to undertake proteomic analysis of pooled BASP1 and LAMP protein expression provide strong clinical samples. This may allow identification of support for a substantial neurite pathology in the potential ‘candidate’ proteins, provided that exten- dlpfc in SCZ. sive validation is undertaken on such proteins. Certainly in our case, our strategy has successfully identified proteins that, on validation, were con- Acknowledgments firmed as differentially expressed. In addition, we Post-mortem brains were donated by the (i) Stanley have validated candidate MM proteins, using whole Foundation Brain Bank Consortium, courtesy of Drs and un-enriched dlpfc grey matter. As a result, we can Llewellyn B Bigelow, Maree J Webster and staff and conclude only that these MM proteins are differen- (ii) Harvard Brain Tissue Resource Centre, courtesy of tially expressed in the cortex as a whole, and not Drs Francine Benes, George Tejada, David J Ennulat, necessarily in the MM specifically. Nonetheless, our Louis Fernandez and staff. We would like to thank findings of elevation of these proteins, regardless of Professor Patrick Levitt and Dr Aurea Pimenta, their association with MM, provide evidence for Vanderbilt Kennedy Center for Research on Human synaptic and neuritic dysfunctions. Further studies Development, Vanderbilt University, Tennessee, for would be required to investigate if STXBP1, BASP1 the kind gift of the LAMP antibody. In addition, we and LAMP are present within MM only or in both the thank Drs Patricia Maguire and Martina Foy for their MM fraction and the rest of the plasma membrane. technical assistance in membrane microdomain iso- All post-mortem brain investigations are potentially lation. Access to and use of MS instrumentation of confounded by the effects of duration of post-mortem Conway Institute is gratefully acknowledged and we interval and tissue pH. Therefore, we were careful to thank Dr Niaobh O’Donoghue, Kasper Pedersen and select our patient subsamples such that they were Kieran Wynne for their technical assistance in mass closely matched for these variables. Exposure to spectrometry. We thank Patrick Dicker for his ex- psychotropic medications can also influence protein pertise in statistical analysis. In addition, we thank expression differentially between disease and control Matt Sullivan and the proteomics informatics group groups. To account for these potential effects we (http://proteomics.ucd.ie) for use of their Proline examined, in both the Stanley and the Harvard brain software. This work was funded by the Health series, the influence of neuroleptic agents, mood Research Board, The Wellcome Trust and Science stabilizers and antidepressant medications, pre- Foundation Ireland. scribed at the time of death, on protein expression. In the Stanley subsample, no significant differences in the protein expression of LAMP, STXBP1 or BASP1 References was found between patients, in the SCZ and the BPD groups, who were prescribed these drugs compared 1 Guidotti A, Auta J, Davis JM, Dong E, Grayson DR, Veldic M et al. GABAergic dysfunction in schizophrenia: new treatment with those who were not. However, in the Harvard strategies on the horizon. Psychopharmacology (Berl) 2005; 180: subsample, significant reductions in BASP1 and 191–205. STXBP1 protein expression were observed among 2 Beneyto M, Kristiansen LV, Oni-Orisan A, McCullumsmith RE, those exposed to mood stabilizers. As the direction of Meador-Woodruff JH. Abnormal glutamate receptor expression in this change was opposite to that associated with SCZ the medial temporal lobe in schizophrenia and mood disorders. Neuropsychopharmacology 2007; 32: 1888–1902. and BPD, this data and indeed the data as a whole 3 Beneyto M, Meador-Woodruff JH. Lamina-specific abnormalities of provide no evidence that our observations of in- AMPA receptor trafficking and signaling molecule transcripts in creases in protein expression of BASP1, STXBP1 or the prefrontal cortex in schizophrenia. Synapse 2006; 60: 585–598. LAMP in these disorders are a consequence of drug 4 Goldman-Rakic PS, Selemon LD. Functional and anatomical aspects of prefrontal pathology in schizophrenia. Schizophr Bull treatments. To date, expression of several synaptic 1997; 23: 437–458. proteins following atypical antipsychotic drug treat- 5 Sanfilipo M, Lafargue T, Rusinek H, Arena L, Loneragan C, ment in the rat hippocampus has been investigated.84 Lautin A et al. Volumetric measure of the frontal and temporal

Molecular Psychiatry Proteomics of MM-associated proteins in psychiatric disease in PFC A´T Behan et al 612 lobe regions in schizophrenia: relationship to negative symptoms. 27 Nebl T, Pestonjamasp KN, Leszyk JD, Crowley JL, Oh SW, Luna EJ. Arch Gen Psychiatry 2000; 57: 471–480. Proteomic analysis of a detergent-resistant membrane skeleton 6 Harrison PJ, Weinberger DR. Schizophrenia genes, gene expres- from neutrophil plasma membranes. J Biol Chem 2002; 277: sion, and neuropathology: on the matter of their convergence. 43399–43409. Mol Psychiatry 2005; 10: 40–68; image 5. 28 Blonder J, Hale ML, Lucas DA, Schaefer CF, Yu LR, Conrads TP 7 Brambilla P, Glahn DC, Balestrieri M, Soares JC. Magnetic et al. Proteomic analysis of detergent-resistant membrane rafts. resonance findings in bipolar disorder. Psychiatr Clin North Am Electrophoresis 2004; 25: 1307–1318. 2005; 28: 443–467. 29 Jia J, Lamer S, Schu¨ emann M, Schmidt M, Krause E, Haucke V. 8 Rajkowska G, Halaris A, Selemon LD. Reductions in neuronal and Quantitative proteomic analysis of detergent-resistant membranes glial density characterize the dorsolateral prefrontal cortex in from chemical synapses: evidence for cholesterol as spatial bipolar disorder. Biol Psychiatry 2001; 49: 741–752. organizer of synaptic vesicle cycling. Mol Cell Proteomics 2006; 9 Rajkowska G, Miguel-Hidalgo JJ, Makkos Z, Meltzer H, 5: 2060–2071. Overholser J, Stockmeier C. Layer-specific reductions in GFAP- 30 Martosella J, Zolotarjova N, Liu H, Moyer SC, Perkins PD, Boyes reactive astroglia in the dorsolateral prefrontal cortex in schizo- BE. High recovery HPLC separation of lipid rafts for membrane phrenia. Schizophr Res 2002; 57: 127–138. proteome analysis. J Proteome Res 2006; 5: 1301–1312. 10 Cotter D, Mackay D, Chana G, Beasley C, Landau S, Everall IP. 31 Parkin ET, Hussain I, Karran EH, Turner AJ, Hooper NM. Reduced neuronal size and glial cell density in area 9 of the Characterization of detergent-insoluble complexes containing the dorsolateral prefrontal cortex in subjects with major depressive familial Alzheimer’s disease-associated presenilins. J Neurochem disorder. Cereb Cortex 2002; 12: 386–394. 1999; 72: 1534–1543. 11 Black JE, Kodish IM, Grossman AW, Klintsova AY, Orlovskaya D, 32 Jiang L, He L, Fountoulakis M. Comparison of protein precipita- Vostrikov V et al. Pathology of layer V pyramidal neurons in the tion methods for sample preparation prior to proteomic analysis. prefrontal cortex of patients with schizophrenia. Am J Psychiatry J Chromatogr A 2004; 1023: 317–320. 2004; 161: 742–744. 33 Maguire PB, Wynne KJ, Harney DF, O’Donoghue NM, Stephens G, 12 Lewis DA, Hashimoto T. Deciphering the disease process of Fitzgerald DJ. Identification of the phosphotyrosine proteome from schizophrenia: the contribution of cortical gaba neurons. Int Rev thrombin activated platelets. Proteomics 2002; 2: 642–648. Neurobiol 2007; 78: 109–131. 34 Behan A, Doyle S, Farrell M. Adaptive responses to mitochondrial 13 Lewis DA, Moghaddam B. Cognitive dysfunction in schizophre- dysfunction in the rho degrees Namalwa cell. Mitochondrion nia: convergence of gamma-aminobutyric acid and glutamate 2005; 5: 173–193. alterations. Arch Neurol 2006; 63: 1372–1376. 35 Sanchez JC, Rouge V, Pisteur M, Ravier F, Tonella L, Moosmayer M 14 Beasley CL, Pennington K, Behan A, Wait R, Dunn MJ, Cotter D. et al. Improved and simplified in-gel sample application using Proteomic analysis of the anterior cingulate cortex in the major reswelling of dry immobilized pH gradients. Electrophoresis 1997; psychiatric disorders: evidence for disease-associated changes. 18: 324–327. Proteomics 2006; 6: 3414–3425. 36 Focking M, Boersema PJ, O’Donoghue N, Lubec G, Pennington SR, 15 Clark D, Dedova I, Cordwell S, Matsumoto I. A proteome analysis Cotter DR et al. 2-D DIGE as a quantitative tool for investigating the of the anterior cingulate cortex gray matter in schizophrenia. HUPO Brain Proteome Project mouse series. Proteomics 2006; 6: Mol Psychiatry 2006; 11: 423, 459–470. 4914–4931. 16 Prabakaran S, Swatton JE, Ryan MM, Huffaker SJ, Huang JT, 37 Beasley CL, Honer WG, Bergmann K, Falkai P, Lu¨tjohann D, Bayer TA. Griffin JL et al. Mitochondrial dysfunction in schizophrenia: Reductions in cholesterol and synaptic markers in association cortex evidence for compromised brain metabolism and oxidative stress. in mood disorders. Bipolar Disord 2005; 7: 449–455. Mol Psychiatry 2004; 9: 643, 684–697. 38 Yan JX, Wait R, Berkelman T, Harry RA, Westbrook JA, Wheeler CH 17 Pennington K, Beasley CL, Dicker P, Fagan A, English J, et al. A modified silver staining protocol for visualization of Pariante CM et al. Prominent synaptic and metabolic abnormal- proteins compatible with matrix-assisted laser desorption/ioniza- ities revealed by proteomic analysis of the dorsolateral prefrontal tion and electrospray ionization-mass spectrometry. Electrophor- cortex in schizophrenia and bipolar disorder. Mol Psychiatry 2007; esis 2000; 21: 3666–3672. Oct 16 e-pub ahead of print. 39 Shevchenko A, Wilm M, Vorm O, Mann M. Mass spectrometric 18 Dean B, Opeskin K, Pavey G, Hill C, Keks N. Changes in sequencing of proteins silver-stained polyacrylamide gels. Anal protein kinase C and adenylate cyclase in the temporal lobe Chem 1996; 68: 850–858. from subjects with schizophrenia. J Neural Transm 1997; 104: 40 Coppinger JA, Cagney G, Toomey S, Kislinger T, Belton O, 1371–1381. McRedmond JP et al. Characterization of the proteins released 19 Owen MJ, O’Donovan MC, Harrison PJ. Schizophrenia: a genetic from activated platelets leads to localization of novel platelet disorder of the synapse? BMJ 2005; 330: 158–159. proteins in human atherosclerotic lesions. Blood 2004; 103: 20 Opeskin K, Dean B, Pavey G, Hill C, Keks N, Copolov D. Neither 2096–2104. protein kinase C nor adenylate cyclase are altered in the striatum 41 Nesvizhskii AI, Keller A, Kolker E, Aebersold R. A statistical from subjects with schizophrenia. Schizophr Res 1996; 22: model for identifying proteins by tandem mass spectrometry. Anal 159–164. Chem 2003; 75: 4646–4658. 21 Pennington K, Cotter D, Dunn MJ. The role of proteomics in 42 Kokubo H, Helms JB, Ohno-Iwashita Y, Shimada Y, Horikoshi Y, investigating psychiatric disorders. Br J Psychiatry 2005; 187:4–6. Yamaguchi H. Ultrastructural localization of flotillin-1 to choles- 22 Allen JA, Halverson-Tamboli RA, Rasenick MM. Lipid raft terol-rich membrane microdomains, rafts, in rat brain tissue. microdomains and neurotransmitter signalling. Nat Rev Neurosci Brain Res 2003; 965: 83–90. 2007; 8: 128–140. 43 Slaughter N, Laux I, Tu X, Whitelegge J, Zhu X, Effros R et al. The 23 Gil C, Soler-Jover A, Blasi J, Aguilera J. Synaptic proteins flotillins are integral membrane proteins in lipid rafts that contain and SNARE complexes are localized in lipid rafts from rat TCR-associated signaling components: implications for T-cell brain synaptosomes. Biochem Biophys Res Commun 2005; 329: activation. Clin Immunol 2003; 108: 138–151. 117–124. 44 Harder T, Scheiffele P, Verkade P, Simons K. Lipid domain 24 Li N, Mak A, Richards DP, Naber C, Keller BO, Li L et al. Monocyte structure of the plasma membrane revealed by patching of lipid rafts contain proteins implicated in vesicular trafficking and membrane components. J Cell Biol 1998; 141: 929–942. phagosome formation. Proteomics 2003; 3: 536–548. 45 Khan TK, Yang B, Thompson NL, Maekawa S, Epand RM, 25 Bini L, Pacini S, Liberatori S, Valensin S, Pellegrini M, Raggiaschi Jacobson K. Binding of NAP-22, a calmodulin-binding neuronal R et al. Extensive temporally regulated reorganization of the lipid protein, to raft-like domains in model membranes. Biochemistry raft proteome following T-cell antigen receptor triggering. Biochem 2003; 42: 4780–4786. J 2003; 369: 301–309. 46 Gil C, Cubi R, Blasi J, Aguilera J. Synaptic proteins associate with 26 Foster LJ, De Hoog CL, Mann M. Unbiased quantitative proteomics a sub-set of lipid rafts when isolated from nerve endings at of lipid rafts reveals high specificity for signaling factors. Proc Natl physiological temperature. Biochem Biophys Res Commun 2006; Acad Sci USA 2003; 100: 5813–5818. 348: 1334–1342.

Molecular Psychiatry Proteomics of MM-associated proteins in psychiatric disease in PFC A´T Behan et al 613 47 Funatsu N, Miyata S, Kumanogoh H, Shigeta M, Hamada K, cortex of patients with schizophrenia. Cereb Cortex 2001; 11: Endo Y et al. Characterization of a novel rat brain glycosylphos- 136–147. phatidylinositol-anchored protein (Kilon), a member of the IgLON 67 Eastwood SL, Harrison PJ. Hippocampal and cortical growth- cell adhesion molecule family. J Biol Chem 1999; 274: 8224–8230. associated protein-43 messenger RNA in schizophrenia. Neuro- 48 Niethammer P, Delling M, Sytnyk V, Dityatev A, Fukami K, science 1998; 86: 437–448. Schachner M. Cosignaling of NCAM via lipid rafts and the 68 Horton HL, Levitt P. A unique membrane protein is expressed FGF receptor is required for neuritogenesis. J Cell Biol 2002; 157: on early developing limbic system and cortical targets. 521–532. J Neurosci 1988; 8: 4653–4661. 49 Kim KB, Lee JW, Lee CS, Kim BW, Choo HJ, Jung SY et al. 69 Zacco A, Cooper V, Chantler PD, Fisher-Hyland S, Horton HL, Oxidation–reduction respiratory chains and ATP synthase com- Levitt P. Isolation, biochemical characterization and ultrastruc- plex are localized in detergent-resistant lipid rafts. Proteomics tural analysis of the limbic system-associated membrane protein 2006; 6: 2444–2453. (LAMP), a protein expressed by neurons comprising functional 50 Raimondo F, Ceppi P, Guidi K, Masserini M, Foletti C, Pitto M. neural circuits. J Neurosci 1990; 10: 73–90. Proteomics of plasma membrane microdomains. Expert Rev 70 Pimenta AF, Levitt P. Characterization of the genomic structure of Proteomics 2005; 2: 793–807. the mouse limbic system-associated membrane protein (Lsamp) 51 Behan A´T, Wynne M, Clarke K, Sullivan MM, Cotter DM, gene. Genomics 2004; 83: 790–801. Maguire PM. Analysis of membrane microdomain-associated 71 Harrison PJ. The neuropathology of schizophrenia. A critical proteins in the insular cortex of post-mortem human brain. review of the data and their interpretation. Brain 1999; 122(Part 4): Proteomics Clin Appl 2007; 1: 1324–1331. 593–624. 52 Quinton TM, Kim S, Jin J, Kunapuli SP. Lipid rafts are required in 72 Iwamoto K, Bundo M, Kato T. Altered expression of mitochondria- Galpha(i) signaling downstream of the P2Y12 receptor during related genes in postmortem brains of patients with bipolar ADP-mediated platelet activation. J Thromb Haemost 2005; 3: disorder or schizophrenia, as revealed by large-scale DNA 1036–1041. microarray analysis. Hum Mol Genet 2005; 14: 241–253. 53 Yang B, Steegmaier M, Gonzalez Jr LC, Scheller RH. nSec1 binds a 73 Steiner P, Sarria JC, Huni B, Marsault R, Catsicas S, Hirling H. closed conformation of syntaxin1A. J Cell Biol 2000; 148: 247–252. Overexpression of neuronal Sec1 enhances axonal branching in 54 Mitchell SJ, Ryan TA. Munc18-dependent regulation of synaptic hippocampal neurons. Neuroscience 2002; 113: 893–905. vesicle exocytosis by syntaxin-1A in hippocampal neurons. 74 Bellon A. New genes associated with schizophrenia in neurite Neuropharmacology 2005; 48: 372–380. formation: a review of cell culture experiments. Mol Psychiatry 55 Rickman C, Medine CN, Bergmann A, Duncan RR. Functionally 2007; 12: 620–629. and spatially distinct modes of munc18-syntaxin 1 interaction. 75 Shirasu M, Kimura K, Kataoka M, Takahashi M, Okajima S, J Biol Chem 2007; 282: 12097–12103. Kawaguchi S et al. VAMP-2 promotes neurite elongation and 56 Sokolov BP, Tcherepanov AA, Haroutunian V, Davis KL. Levels of SNAP-25A increases neurite sprouting in PC12 cells. Neurosci Res mRNAs encoding synaptic vesicle and synaptic plasma membrane 2000; 37: 265–275. proteins in the temporal cortex of elderly schizophrenic patients. 76 Knable MB, Barci BM, Bartko JJ, Webster MJ, Torrey EF. Molecular Biol Psychiatry 2000; 48: 184–196. abnormalities in the major psychiatric illnesses: classification and 57 Gabriel SM, Haroutunian V, Powchik P, Honer WG, Davidson M, regression tree (CRT) analysis of post-mortem prefrontal markers. Davies P et al. Increased concentrations of presynaptic proteins in Mol Psychiatry 2002; 7: 392–404. the cingulate cortex of subjects with schizophrenia. Arch Gen 77 McCullumsmith RE, Kristiansen LV, Beneyto M, Scarr E, Dean B, Psychiatry 1997; 54: 559–566. Meador-Woodruff JH. Decreased NR1, NR2A, and SAP102 tran- 58 Honer WG, Falkai P, Young C, Wang T, Xie J, Bonner J et al. script expression in the hippocampus in bipolar disorder. Brain Cingulate cortex synaptic terminal proteins and neural cell Res 2007; 1127: 108–118. adhesion molecule in schizophrenia. Neuroscience 1997; 78: 78 Blennow K, Bogdanovic N, Gottfries CG, Davidsson P. The growth- 99–110. associated protein GAP-43 is increased in the hippocampus and in 59 Halim ND, Weickert CS, McClintock BW, Hyde TM, Weinberger DR, the gyrus cinguli in schizophrenia. J Mol Neurosci 1999; 13: 101–109. Kleinman JE et al. Presynaptic proteins in the prefrontal cortex of 79 Perrone-Bizzozero NI et al. Levels of the growth-associated patients with schizophrenia and rats with abnormal prefrontal protein GAP-43 are selectively increased in association development. Mol Psychiatry 2003; 8: 797–810. cortices in schizophrenia. Proc Natl Acad Sci USA 1996; 93: 60 Honer WG, Falkai P, Bayer TA, Xie J, Hu L, Li HY et al. 14182–14187. Abnormalities of SNARE mechanism proteins in anterior frontal 80 Sower AC, Bird ED, Perrone-Bizzozero NI. Increased levels of cortex in severe mental illness. Cereb Cortex 2002; 12: 349–356. GAP-43 protein in schizophrenic brain tissues demonstrated by a 61 Mukaetova-Ladinska EB, Hurt J, Honer WG, Harrington CR, novel immunodetection method. Mol Chem Neuropathol 1995; 24: Wischik CM. Loss of synaptic but not cytoskeletal proteins in 1–11. the cerebellum of chronic schizophrenics. Neurosci Lett 2002; 317: 81 Scarr E, Gray L, Keriakous D, Robinson PJ, Dean B. Increased 161–165. levels of SNAP-25 and synaptophysin in the dorsolateral 62 Hakak Y, Walker JR, Li C, Wong WH, Davis KL, Buxbaum JD et al. prefrontal cortex in bipolar I disorder. Bipolar Disord 2006; 8: Genome-wide expression analysis reveals dysregulation of myeli- 133–143. nation-related genes in chronic schizophrenia. Proc Natl Acad Sci 82 Chambers JS, Thomas D, Saland L, Neve RL, Perrone-Bizzozero NI. USA 2001; 98: 4746–4751. Growth-associated protein 43 (GAP-43) and synaptophysin altera- 63 Maekawa S, Iino S, Miyata S. Molecular characterization of the tions in the dentate gyrus of patients with schizophrenia. detergent-insoluble cholesterol-rich membrane microdomain (raft) Prog Neuropsychopharmacol Biol Psychiatry 2005; 29: 283–290. of the central nervous system. Biochim Biophys Acta 2003; 1610: 83 Vawter MP. Dysregulation of the neural cell adhesion molecule 261–270. and neuropsychiatric disorders. Eur J Pharmacol 2000; 405: 64 Frey D, Laux T, Xu L, Schneider C, Caroni P. Shared and unique 385–395. roles of CAP23 and GAP43 in actin regulation, neurite outgrowth, 84 Barr AM, Young CE, Phillips AG, Honer WG. Selective effects of and anatomical plasticity. J Cell Biol 2000; 149: 1443–1454. typical antipsychotic drugs on SNAP-25 and synaptophysin in the 65 Eastwood SL, Harrison PJ. Synaptic pathology in the anterior hippocampal trisynaptic pathway. Int J Neuropsychopharmacol cingulate cortex in schizophrenia and mood disorders. A review 2006; 9: 457–463. and a Western blot study of synaptophysin, GAP-43 and the 85 Eastwood SL, Heffernan J, Harrison PJ. Chronic haloperidol complexins. Brain Res Bull 2001; 55: 569–578. treatment differentially affects the expression of synaptic and 66 Weickert CS, Webster MJ, Hyde TM, Herman MM, Bachus SE, neuronal plasticity-associated genes. Mol Psychiatry 1997; 2: Bali G et al. Reduced GAP-43 mRNA in dorsolateral prefrontal 322–329.

Supplementary Information accompanies the paper on the Molecular Psychiatry website (http://www.nature.com/mp)

Molecular Psychiatry