Cytosolic protein interactions of the schizophrenia susceptibility gene dysbindin

Authors

  • Carri-Lyn R. Mead,

    1. Michael Smith Genome Sciences Centre, British Columbia Cancer Research Centre, Vancouver, British Columbia, Canada
    Search for more papers by this author
  • Michael A. Kuzyk,

    1. GenoLogics Life Sciences Software, Victoria, British Columbia, Canada
    Search for more papers by this author
  • Annie Moradian,

    1. Michael Smith Genome Sciences Centre, British Columbia Cancer Research Centre, Vancouver, British Columbia, Canada
    Search for more papers by this author
  • Gary M. Wilson,

    1. Michael Smith Genome Sciences Centre, British Columbia Cancer Research Centre, Vancouver, British Columbia, Canada
    Search for more papers by this author
  • Robert A. Holt,

    1. Michael Smith Genome Sciences Centre, British Columbia Cancer Research Centre, Vancouver, British Columbia, Canada
    2. Department of Psychiatry, University of British Columbia, Wesbrook Mall, Vancouver, British Columbia, Canada
    3. Department of Molecular Biology and Biochemistry, Simon Fraser University, University Drive, Burnaby, British Columbia, Canada
    Search for more papers by this author
  • Gregg B. Morin

    1. Michael Smith Genome Sciences Centre, British Columbia Cancer Research Centre, Vancouver, British Columbia, Canada
    2. Department of Medical Genetics, University of British Columbia, Health Sciences Mall, Vancouver, British Columbia, Canada
    Search for more papers by this author

Address correspondence and reprint requests to Gregg B. Morin, PhD and Robert A. Holt, PhD, Michael Smith Genome Sciences Centre, British Columbia Cancer Research Centre, 675 West 10th Avenue, Vancouver, BC V5Z 1L3, Canada. E-mail: gmorin@bcgsc.ca and rholt@bcgsc.ca

Abstract

J. Neurochem. (2010) 113, 1491–1503.

Abstract

Using immunoprecipitation, mass spectrometry, and western blot analysis we investigated cytosolic protein interactions of the schizophrenia susceptibility gene dysbindin in mammalian cells. We identified novel interactions with members of the exocyst, dynactin and chaperonin containing T-complex protein complexes, and we confirmed interactions reported previously with all members of the biogenesis of lysosome-related organelles complex-1 and the adaptor-related protein complex 3. We report interactions between dysbindin and the exocyst and dynactin complex that confirm a link between two important schizophrenia susceptibility genes: dysbindin and disrupted-in-schizophrenia-1. To expand upon this network of interacting proteins we also investigated protein interactions for members of the exocyst and dynactin complexes in mammalian cells. Our results are consistent with the notion that impairment of aspects of the synaptic vesicle life cycle may be a pathogenic mechanism in schizophrenia.

Abbreviations used:
ACTR1A

alpha-centractin

AP3

adaptor-related protein complex-3

BLOC

biogenesis of lysosome-related organelles complex

CCT

chaperonin containing T-complex protein

DCTN

dynactin

DISC1

disrupted-in-schizophrenia-1

DTNBP1

dysbindin

ESI-MS/MS

electrospray-tandem mass spectrometry

EXOC

exocyst

GAD

genetic association database

GFP

green fluorescent protein

HEK293

human embryonic kidney

SDS

sodium dodecyl sulfate

SDS–PAGE

SDS–polyacrylamide gel electrophoresis

SNARE

soluble N-ethylmaleimide-sensitive fusion-attachment receptor

STXBP1

syntaxin binding partner-1

Schizophrenia is a complex psychiatric disorder with a strong genetic component and unknown etiology (Harrison and Owen 2003; Sullivan et al. 2003). A number of genes have been identified as schizophrenia susceptibility genes including dysbindin (DTNBP1), neuregulin-1 (NRG1), catechol-O-methyltransferase (COMT), and disrupted-in-schizophrenia-1 (DISC1), among others (Harrison and Weinberger 2005). DTNBP1 is one of the most robust schizophrenia susceptibility genes identified to date, located within one of the most consistently replicated schizophrenia linkage regions (6p22.3) (Antonarakis et al. 1995; Moises et al. 1995; Straub et al. 1995, 2002b; Turecki et al. 1997; Schwab et al. 2000; Hallmayer et al. 2005; Maziade et al. 2005) and having one of the most replicated schizophrenia association findings (Straub et al. 2002a; Schwab et al. 2003; Tang et al. 2003; Van Den Bogaert et al. 2003, van den Oord et al. 2003; Funke et al. 2004; Kirov et al. 2004; Numakawa et al. 2004; Williams et al. 2004; Li et al. 2005). The DTNBP1 knockout mouse (sandy) shows increased dopamine turnover in specific brain regions (Chagnon et al. 2008), and schizophrenia patients have been found to have decreased expression of DTNBP1 at pre-synaptic glutamatergic terminals in the hippocampus (Talbot et al. 2004; Weickert et al. 2008) and dorsolateral prefrontal cortex (Weickert et al. 2004). Defects in neurosecretion and vesicular morphology in neuroendocrine cells and hippocampal synapses have been identified at the single vesicle level in sandy mice and implicate DTNBP1 functions in the regulation of exocytosis and vesicle biogenesis in endocrine cells and neurons (Chen et al. 2008).

Dysbindin was initially discovered as an interacting protein of dystrobrevin proteins (DTNA, DTNB) (Benson et al. 2001). Mutations in the DTNBP1 gene have been shown to cause Hermansky–Pudlak syndrome type 7, one of eight known human Hermansky–Pudlak syndrome types caused by defects in intracellular protein trafficking resulting from the dysfunction of lysosome-related organelles. DTNBP1 is a member of the biogenesis of lysosomal related complex-1 (BLOC1) (Li et al. 2003), which is known to interact with the BLOC2 and adapter-related protein complex-3 (AP3) complexes (Di Pietro et al. 2006), and functions in organelle biogenesis and the protein transport pathway (Setty et al. 2007). BLOC1 subunits are implicated in synaptic mechanisms (Salazar et al. 2006); Pallidin (PLDN) is involved in mediating vesicle docking and fusion (Huang et al. 1999) and disruption of soluble N-ethylmaleimide-sensitive fusion-attachment receptor (SNARE)-associated protein (SNAPIN) causes defective secretion of neurotransmitters in mice (Tian et al. 2005). Recently, Hikita et al. (2009) identified syntaxin binding partner-1 (STXBP1 or Munc18-1) as a DTNBP1 interacting protein in an analysis of DTNBP1 membrane interactions. STXBP1 has been implicated in the exocytosis of synaptic vesicles (Verhage et al. 2000; Shen et al. 2007; Toonen and Verhage 2007).

Disrupted-in-schizophrenia-1 was initially identified at a site for a balanced translocation (1 : 11) (q42.1;q14.3) that co-segregates with schizophrenia and other psychiatric disorders in a large Scottish pedigree (St Clair et al. 1990; Blackwood et al. 2001). Several studies provide evidence of DISC1 association with schizophrenia (Hodgkinson et al. 2004; Callicott et al. 2005; Cannon et al. 2005; Thomson et al. 2005; Zhang et al. 2006; Hennah et al. 2008). DISC1 has been shown to interact with multiple proteins including nuclear protein distribution nudE-like 1 (NDEL1) (Morris et al. 2003; Ozeki et al. 2003), platelet-activating factor acetylhydrolase IB subunit alpha (PAFAH1B1, also called lissencephaly-1) (Brandon et al. 2004), cAMP-specific 3′,5′-cyclic phosphodiesterase 4B (PDE4B) (Millar et al. 2005), growth factor receptor-bound protein 2 (GRB2) (Shinoda et al. 2007; Taya et al. 2007), fasciculation and elongation protein zeta-1 (FEZ1) (Miyoshi et al. 2003), DISC1-binding zinc-finger protein (DBZ) (Hattori et al. 2007), and pericentrin (PCNT or kendrin) (Miyoshi et al. 2004). PDE4B and FEZ1 have also had positive schizophrenia association results (Yamada et al. 2004; Pickard et al. 2007). Through these interactions it is postulated that DISC1 is involved in brain development including neuronal migration, neurite outgrowth and neural maturation through interaction with several cytoskeletal proteins (Matsuzaki and Tohyama 2007).

Defective synaptic transmission and neurotransmitter release is hypothesized to be a pathogenic mechanism in schizophrenia (Frankle et al. 2003; Harrison and Weinberger 2005). Defects in pre-synaptic vesicle proteins have been associated with schizophrenia (Mirnics et al. 2000; Honer and Young 2004; Camargo et al. 2007) and the schizophrenic brain has been shown to possess reduced levels of mRNA and/or proteins involved in synaptic vesicle fusion (Honer et al. 2002; Mukaetova-Ladinska et al. 2002; Halim et al. 2003; Knable et al. 2004). Our understanding of the vesicle life cycle involved in neurotransmitter release remains incomplete. For example, the disruption of DTNBP1 in sandy mice causes defects in neurosecretion and vesicle morphology in neuroendocrine cells and hippocampal synapses at the single vesicle level, including larger vesicle size, slower quantal vesicle release, lower release probability, and smaller total population of the readily releasable vesicle pool (Chen et al. 2008). However, the detailed mechanisms for how these DTNBP1-associated phenomena may contribute to neurotransmitter release or schizophrenia pathogenesis are unknown.

We reasoned that a better understanding of DTNBP1 interacting proteins would facilitate the investigation of DTNBP1 pathways and functions and potentially provide insight into schizophrenia etiology. Using proteomic techniques we have identified DTNBP1 interacting proteins with vesicle trafficking components. Intriguingly, these data also identify potential common links for two of the best schizophrenia susceptibility genes, DTNBP1 and DISC1, to portions of the vesicle trafficking system.

Materials and methods

Cloning

cDNAs for two isoforms of DTNBP1: version 1 (NM_032122) and version 3 (NM_183041), and exocyst component 4 (EXOC4), exocyst component 3 (EXOC3), AP3 subunit beta-1 (AP3B1), AP3 subunit beta-2 (AP3B2), dynactin subunit 2 (DCTN2), and alpha-centractin (ACTR1A, a member of the dynactin complex) were procured (IMAGE clones 4139934, 6183004, 5590332, 3914400, 7939584, and 3347881 respectively (American Type Culture Collection–ATCC, Manassas, VA, USA). The open reading frames were amplified (Table S7 for PCR primer sets) and ligated into the V954 donor vector of the Creator Splice system (Colwill et al. 2006). The cDNA containing cassettes were transferred into mammalian expression acceptor vectors for fusion to both N- and C-terminal 3XFLAG tags (V180 and V181) by Cre-lox recombination. DTNBP1 version 1 was also transferred to green fluorescent protein (GFP)-fusion vectors (V4 and V6). Completed constructs were sequence verified prior to use.

Antibodies

Antibodies were anti-DTNBP1 (goat, sc-46931, Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-GFP (rabbit, sc-8334, Santa Cruz Biotechnology), anti-FLAG (mouse, M2, Sigma-Aldrich, St. Louis, MO, USA), anti-DISC1 (goat, ab41985, Abcam Inc, Cambridge, MA, USA), and fluorescent secondary antibodies (IR-700, Roche, Basel, Switzerland).

Cell culture

Human embryonic kidney (HEK293, ATCC) and X57 (mouse striatal, obtained from Marian DiFiglia at Massachusetts General Hospital, Boston, MA, USA) cells were maintained at 37°C and 5% CO atmosphere in Dulbecco’s modified eagle’s medium with glutamax (Invitrogen, Carlsbad, CA, USA) supplemented with 10% added fetal bovine serum (Invitrogen).

Immunoprecipitation, protein complex preparation and mass spectrometry

The 3XFLAG-tagged cDNA expression vectors were transfected into HEK293 (for DTNBP1) or X57 (for exocyst and dynactin complex proteins) cells using lipofectamine 2000 reagent (Invitrogen) in Optimem reduced serum media (Invitrogen). Cells were separately transfected with the parent acceptor vector for the negative control sample. Approximately 48 h after transfection, confluent cells were harvested using chilled (4°C) phosphate buffered solution (1 mM phosphate, 155 mM NaCl, 3 mM Na2HPO4, pH 7.4) and the pellets were stored at −80°C. Cell pellets were resuspended in lysis/wash buffer with protease inhibitors (Tris-buffered saline (20 mM Tris-base, 100 mM NaCl, pH 7.4), 1 mM EDTA, 1% NP-40, 10 μg/mL Leupeptin, 10 μg/mL Aprotinin, 10 μg/mL Pepstatin, 1 mM AEBSF, 2 mM Na3VO4, 10 mM β-Glycerophosphate) and forced through a 20 gauge syringe thrice, rocked for 30 min at 4°C. The cell debris removed by centrifugation (15 min at 15 000 × g) and the supernatant filtered through a 45 μM membrane. The extracts were immunoprecipitated with anti-FLAG M2 agarose resin (Sigma-Aldrich) and proteins released with FLAG peptide, as described previously (Ewing et al. 2007).

The immunoprecipitate eluates, from the 3XFLAG-tagged cDNA and empty vector control transfections, were lyophilized and rehydrated in loading buffer [Tris-Cl/sodium dodecyl sulfate (SDS; 58mM Tris-Cl, 0.05% SDS, pH 6.8), 5% glycerol, 1.7% SDS, 0.1 M dithiothreitol, 0.03 μM bromophenol blue] and heated at 85°C for 10 min. The eluates were separated by one-dimensional SDS–polyacrylamide gel electrophoresis (PAGE) using a 4–12% gradient NuPAGE gel (Invitrogen) with NuPAGE MES running buffer (Invitrogen) for ∼ 1.5 h at 150 V. The gel was then stained in colloidal Coomassie (20% ethanol, 1.6% phosphoric acid, 8% ammonium sulfate, 0.08% Coomassie Brilliant Blue G-250) and de-stained with distilled H2O. The entirety of each lane was excised and divided into 16 pieces; each slice was finely diced and transferred to a 96-well plate. Automated in-gel dehydration, alkylation, trypsin digestion, and extraction were performed (Progest; Genomic Solutions, Ann Arbor, MI, USA). The extracts were lyophilized and re-suspended in 3% acetonitrile, 1.5% formic acid.

HPLC-electrospray-tandem mass spectroscopy (ESI-MS/MS) was performed on a 4000 QTrap mass spectrometer (Applied Biosystems/Sciex, Foster City, CA, USA), except for three DTNBP1 immunoprecipitations which were performed on a LCQ Deca (Thermo-Fisher, Waltham, MA, USA), both instruments were coupled to a Agilent 1100 Nano-HPLC (Agilent, SantaClara, CA, USA) using a nano-ESI interface. For the 4000 QTrap, samples were desalted using an on-line trap column (Zorbax, 300SB-C18, 5μm, 5 × 0.3 mm, Agilent Technologies, Santa Clara, CA, USA) and chromatography coupled to electrospray was performed on a 75 μm × 150 mm reverse phase column (Jupiter 4μ Proteo 90 A, Phenomenex, Torrance, CA, USA) using Buffer A (5% HPLC grade acetonitrile (Fisher, Ottawa, Canada), 0.1% formic acid (Fluka, Sigma-Aldrich) and Buffer B (90% acetonitrile, 0.1% formic acid) in a linear gradient of 0–20% B for 37 min, 20–39% B for 16 min, and 39–90% B for 8 min at a flow rate of 300 nL/min. ESI-MS/MS was performed with ESI at 1850 V, interface heater at 150°C, at 4.7 × 10−5 torr with nitrogen (99.999%, Praxair, Danbury, CT, USA) for nebulizer gas (0.5 mL/min) and curtain gas (2 L/min). Data were collected using a 400–1600 m/z Enhanced MS scan followed by an Enhanced Resolution scan to select the top five +2 and +3 ions for collisional-induced dissociation and a final Enhanced Product Ion MS scan. For the LCQ Deca, the chromatography was performed with flow splitting to produce a 0.2 μL/min rate. The samples were washed on a trap column (Jupiter 4μ, Proteo 90 A, Phenomenex) using buffer A (5% HPLC grade acetonitrile, 0.1% formic acid and 0.02% tri-fluoro acetic acid) for 5 min. Chromatography was then performed (column: 75 μm × 100 mm; Jupiter 4μ, Proteo 90 A, Phenomenex) using buffer B (90% acetonitrile, 0.1% formic acid, 0.02% tri-fluoro acetic acid) in a linear gradient of 0–35% B for 37 min, 35–65% for 8 min and 65–100% for 2 min. The capillary was at 160°C and ESI at 1800 V. The MS was operated at 1–1.9 × 10−5 torr in data-dependent acquisition mode with a full scan MS (400–1400 m/z) followed by MS/MS of the three most intense precursor ions using collisional-induced dissociation (helium). The MS/MS spectra emanating from the gel slices for each lane were concatenated and searched against the Ensembl human or mouse databases, as appropriate, using the Mascot (Matrix Science, Boston, MA, USA) and X!Tandem (http://www.thegpm.org/TANDEM/) search engines. Search parameters were 0.3 Da and 0.4 Da for precursor and product ion mass tolerance, respectively; trypsin digestion; one missed cleavage; oxidation (methionine); deamidation (asparagine, glutamine); phosphorylation (serine, threonine, and tyrosine), and carbamidomethylation of cysteine. The raw data are available through the PRIDE database (http://www.ebi.ac.uk/pride/).

Identification of candidate protein interactors

Using in house SpecterWeb software we processed the identified proteins from the MS/MS analyzes by subtracting: (i) proteins found in more than one negative empty vector control sample, (ii) common non-specific binding proteins (heat shock, transcriptional and translational machinery, keratins, and protein arginine N-methyltransferase 5 (PRMT5), (iii) proteins found in only 1 experimental sample, and (iv) proteins without two observations each having X!Tandem log(E) scores ≤ −3. The candidate interacting proteins were then subdivided into two tiers, where first tier proteins were those having a minimum of two unique peptides in each of at least two replicates, as identified by X!Tandem. The average X!Tandem log(E) score was calculated for each candidate protein across the experiments where it was observed. The list of candidate interacting proteins and the analysis data for the DTNBP1, dynactin and exocyst experiments are reported in Tables S1–S3, and Figures S5–S7, respectively.

Validation of protein interactions through immunoprecipitation-western analysis

The 3XFLAG-tagged constructs were co-transfected with GFP-tagged DTNBP1 (N-tagged for X57 experiments, C-tagged for HEK293 experiments) into HEK293 or X57 cells as described above. Anti-FLAG immunoprecipitation and separation by one-dimensional SDS-PAGE is described above. The negative controls were cells transfected with the parent acceptor vectors. An aliquot of the input lysate without immunoprecipitation was used as a positive control. The proteins were electro-transferred to nitrocellulose for ∼16 h at 100 mA in towbin buffer (0.25 M Tris-base, 2 M glycine, pH 8.5, 20% methanol). The nitrocellulose was blocked using NuPAGE Odyssey blocking buffer (Invitrogen) before being probed using the appropriate antibody (1 : 200 anti-GFP antibody, or 1 : 500 anti-DISC1 antibody) for ∼ 16 h at 4°C in blocking buffer. The filter was washed (thrice, 5 min each) with tris-buffered-saline-Tween buffer (Tris-buffered saline, 0.01% Tween-20) and then probed with the appropriate secondary antibody for 30 min at 22°C. The nitrocellulose was then washed with tris-buffered-saline-Tween buffer (thrice, 5 min each) and imaged using an Odyssey system (LI-COR Biosciences, Lincoln, NE, USA).

Linkage and association with schizophrenia analysis

The genetic association database (GAD, http://geneticassociationdb.nih.gov/) and the psychiatric genetics evidence project linkage database (http://slep.unc.edu/evidence/) were used to determine whether any of the protein interaction sets were over-represented with proteins having previous linkage or association evidence of involvement with schizophrenia. A query of the GAD database using the human genome organization gene nomenclature committee names showed few of the proteins have been investigated in schizophrenia association studies, thus GAD association data did not contribute to the schizophrenia over-representation analysis. Therefore, only the linkage database was queried with the cytogenetic location associated with each protein to perform the over-representation analysis. For AP3, BLOC1, chaperonin containing T-complex protein (CCT), dynactin, and exocyst all complex members were included in the analysis if any members of the complex had been identified in the protein interaction set. A Chi-squared test was used to determine the significance of finding the number of schizophrenia linked cytogenetic regions observed for each protein interaction set, given the number of cytogenetic regions across the entire human genome with evidence of schizophrenia linkage.

Results

DTNBP1 interacts with vesicle trafficking complexes: exocyst, dynactin, AP3, CCT and cytoskeletal components

To obtain high confidence protein interaction partners for DTNBP1 we used a co-immunoprecipitation comparative mass spectrometry (MS) analysis procedure where peptides from immunoprecipitated DTNBP1 protein complexes were compared with peptides found in protein complexes from control immunoprecipitates. HEK293 cells were transfected with FLAG epitope-tagged versions of DTNBP1 isoform 1 and isoform 3 or an empty vector control. Protein complexes from cytosolic fractions of the experimental and control cells were immunoprecipitated by an anti-FLAG antibody and size fractionated using one-dimensional SDS-PAGE. Rather than excise only individual visually observable bands and to obtain high sensitivity, accuracy and coverage we processed the entire lanes and compared the peptide (protein) composition between the experimental and control samples using the MS data. Seven independent experiments were performed; two experiments each for N- and C-tagged DTNBP1 isoform 1 and N-tagged isoform 3, and one experiment for C-tagged isoform 3, which were compared with six empty vector control experiments. In Figure S1, we show a representative gel image of a control and experimental sample and in Figure S2 representative ion chromatograms, MS spectra and MS/MS fragment assignments of two peptides for each of six candidate interacting proteins between the experimental gel slice and the cognate control gel slice. This demonstrates the observation of an assigned peptide in the experimental sample and its absence in the control sample. From these experiments we identified 83 unique candidate DTNBP1-interacting proteins that met score quality thresholds (see Methods) and were observed in at least two independent experiments (Table 1, Fig. 1, Table S1). Fourteen of the proteins were previously identified as DTNBP1 interacting proteins, including all seven members of BLOC1 (Li et al. 2003), all six members of the AP3 complex, and DNA-dependent protein kinase catalytic subunit (PRKDC) (Oyama et al. 2009) that previously had been shown to interact with either BLOC1 (Di Pietro et al. 2006) or DTNBP1 itself (Hikita et al. 2009). The remaining 68 novel DTNBP1 interacting proteins include several members of the exocyst and dynactin complexes, and components of the cytoskeleton, including actin and tubulin and the CCT complex. These interactions support a role for DTNBP1 in vesicle trafficking.

Table 1.   Protein interactions identified through immunoprecipitation-mass spectrometry. This table is a summary of the DTNBP1, exocyst (EXOC), and dynactin (DCTN) protein interaction partners
DTNBP1 interacting proteinsExocyst interacting proteinsDynactin interacting proteins
HEK293 cellsX57 cellsX57 cells
83 interacting proteins56 interacting proteins31 interacting proteins
  1. These candidate interacting proteins were identified from the immunoprecipitation comparative mass spectrometry data using a number of filtering and quality criteria (see Methods and Tables S1–S3). Results are grouped by complex or interest. Proteins shown in bold were used in validation experiments.

BLOC1 complex
CNODTNBP1     
BLOC1S1MUTED     
BLOC1S2PLDN     
BLOC1S3SNAPIN     
Dynactin complex
ACTR1ADCTN1 CAPZB ACTR1ADCTN3
CAPZA1DCTN2   CAPZBDCTN4
CAPZBDCTN3   DCTN1DCTN5
DNHD1L    DCTN2DCTN6
Exocyst complex
EXOC3  EXOC1EXOC5  
EXOC4  EXOC2EXOC6  
EXOC6B  EXOC3EXOC6B  
   EXOC4EXOC7  
Chaperonin containing TCP1 complex (CCT)
CCT1CCT5 CCT3 CCT2CCT7
CCT3CCT8 CCT8 CCT3CCT8
CCT4    CCT6A 
Tubulin/Actin associated proteins
ACTA1MYO6 ACTBL2TUBB2CACTR10TUBA1A
ACTA2MYH10 MTAP1ATUBB3ACTR1BTUBB2B
ACTG1SPTBN1 TTNTUBB6KLHL2TUBB3
CFL1TUBA1B TUBA1A PDCL3TWF1
KIF2ATUBA1C     
MAPTTUBB2B     
MYO1DTUBB4     
AP3 complex
AP3B1AP3M1     
AP3B2AP3S1     
AP3D1AP3S2     
Brain/neuron associated
   ENO1PRPH1CDH15 
   GDF7PHGDH  
Vesicular transport/trafficking associated/transporter
AP2A1KCTD17 ABCF2LRP1CLCN1 
COG5SCYL2 ARF3RAB1BKCTD2 
COG7SLC1A5 ARF4SEC61A1  
KCNMA1SLC27A4 GAPDHSLC16A1  
   HERC1SLC16A3  
   KCTD2UNC80  
Other proteins
AFG3L2IFIT1QPCTLARID1BPDIA3CADPFKP
BCAS4IGF2BP3SDCCAG3CADPFKPFEM1ASSR1
BTBD7MTDHSTAT1CAND1PKM2GAPDHSTPI1
C11orf48NT5C2TMCO1DDOSTPPM1BMSH5 
C17orf59PCBP1TMEM33DHCR7PRPSAP1  
CDIPTPGAM5TOP1GALK1PRPSAP2  
CSNK1A1PHGDHUPF0402GAPDHSRPN2  
CSNK1A1LPKM2YTHDC2HSD17B12SSR1  
CYP4F8PPM1BYWHAELDHATPI1  
FASNPRKDCZNF281MAGEE2YWHAQ  
HERC5PTPLAD1 MESTYWHAZ  
Figure 1.

 Overall protein interaction network of the DTNBP1, exocyst and dynactin complexes identified through immunoprecipitation-mass spectrometry. This protein interaction network shows the complete set of protein interactions (described in Tables S1–S3) for each of the five tagged immunoprecipitated ‘bait’ proteins (DTNBP1, ACTR1A, DCTN2, EXOC3, and EXOC4) identified in our immunoprecipitation comparative mass spectrometry data. Each circle denotes a protein. Proteins are clustered into complexes where applicable (grey circles represent proteins that are not clustered into a complex). Blue circles designate bait proteins, pink circles denote proteins clustered by complex. Orange lines are interactions which were validated in immunoprecipitation–western blot experiments (Figs 2 and 3). The number of proteins in the actin and tubulin associated cluster may be misleading: the mass spectrometry data strongly support interactions with actin and tubulin proteins but it is unclear which paralogs are the predominant partners as some of the assigned peptides are common to multiple tubulin or actin proteins. The inset figure in the top left shows a high level view of the protein interaction network and extends the network to include disrupted-in-schizophrenia-1 (DISC1). Asterisks denote novel interactions; dark blue circles indicate schizophrenia susceptibility genes.

Six DTNBP1 interactions were chosen for validation, including interactions with two members each of the dynactin (ACTR1A, DCTN2), exocyst (EXOC3, EXOC4), and AP3 (AP3B1, AP3B2) complexes (Table 1). Epitope-tagged versions of these proteins and DTNBP1 were co-expressed in HEK293 and X57 cells. Immunoprecipitation and western blotting confirmed the interaction of DTNBP1 with the six proteins in all 12 experiments with both HEK293 and X57 cells (Fig. 2). Until recently (Hikita et al. 2009), the interaction of DTNBP1 with individual members of the AP3 complex had not been shown. A yeast two hybrid experiment identified a candidate interaction between DTNBP1 and EXOC4 (Camargo et al. 2007), however, this interaction had not previously been investigated in mammalian cells. The average log(E) scores for most of the 68 novel DTNBP1 interacting proteins were within the range of scores for the 18 known and/or validated DTNBP1 interacting proteins (Figure S5, and Table S1). This indicates that most of the proteins identified are likely to be true DTNBP1 interaction partners.

Figure 2.

 Validation of protein interactions between DTNBP1 and the exocyst, dynactin, and AP3 complexes. Validation of protein interactions between DTNBP1 and the exocyst, dynactin, and AP3 complexes were performed in X57 and HEK293 cells through immunoprecipitation of 3XFLAG-tagged versions of AP3B1, AP3B2, ACTR1A, DCTN2, EXOC3, and EXOC4 with an anti-FLAG antibody followed by western blot for green fluorescent protein (GFP)-tagged DTNBP1 using a GFP antibody. (a) DTNBP1 – AP3 interaction validations in X57 and HEK293 cells; (b) DTNBP1 – dynactin interaction validations in X57 and HEK293 cells; (c) DTNBP1 – exocyst interaction validations in X57 and HEK293 cells. Arrows and boxes indicate the DTNBP1-GFP band, which is the expected size according to the positive control. The positive controls are the cell lysates without anti-FLAG immunoprecipitation and indicate the expected size of DTNBP1-GFP (68 kDa); positive control lysate amounts were not matched with experimental lanes. Additional bands are believed to be unspecific recognition of the GFP antibody or because of proteolytic cleavage and/or protein modifications.

The exocyst and dynactin complexes interact with the CCT complex and vesicular transport, trafficking, and transporter associated proteins

To extend the DTNBP1-associated protein interaction network we performed co-immunoprecipitation comparative mass spectrometry experiments in X57 cells for two exocyst complex proteins (EXOC3, EXOC4) and two dynactin complex proteins (ACTR1A, DCTN2). Two experiments were performed for each of these four ‘bait’ proteins and compared with two empty vector controls; Figure S3 shows a representative gel image. The results identified 56 and 31 unique protein identifications for the exocyst and dynactin complexes, respectively, after processing for background, quality, and experimental replication (see Methods). Of these, 48 and 23 were novel interactions, respectively (Table 1, and Tables S2 and S3). Similar to the DTNBP1 data, the average log(E) scores for the novel exocyst and dynactin interacting proteins over-lapped with the scores for their known interacting proteins (Figures S6 and S7). In addition, 47 and 15 of the interactions in the exocyst and dynactin data were identified by both bait proteins, respectively (Table 1, and Tables S2 and S3), and thus represent independent corroboration of these candidate protein-protein interactions.

DISC1 interacts with the exocyst and dynactin complexes

A yeast-2-hybrid study identified high quality interactions between DISC1 and each of EXOC1, EXOC7, and DCTN2 (Camargo et al. 2007), however, these were not investigated in mammalian cells. To further investigate DISC1 interactions with the exocyst and dynactin complexes, we expressed epitope-tagged EXOC3 and ACTR1A in HEK293 cells, immunoprecipitated the tagged EXOC3 and ACTR1A complexes and identified DISC1 proteins that co-immunoprecipitated through a western blot probed with an anti-DISC1 antibody. The DISC1 protein is expressed in various isoforms ranging from 70–105 kDa (Ishizuka et al. 2006). The approximately 80 kDa size we observed for endogenous DISC1 that co-immunoprecipitated with EXOC3 and ACTR1A (Fig. 3) was consistent with other studies that have interrogated endogenous DISC1 and found an abundant isoform of DISC1 to be approximately 75–80 kDa in human brain, HEK293 cells, HeLa (cervical carcinoma) cells, and SH-SY5Y (neuroblastoma) cells (Miyoshi et al. 2003; James et al. 2004; Ishizuka et al. 2006; Nakata et al. 2009). 5′-rapid amplification of cDNA ends experiments have also shown that a high fraction of DISC1 mRNAs in human brain express isoforms of DISC1 consistent with our observed size of DISC1 (Nakata et al. 2009).

Figure 3.

 Validation of the DISC1 interaction with the dynactin and exocyst complexes. The DISC1 interaction with members of the dynactin and exocyst complexes were validated in HEK293 cells through immunoprecipitation of 3XFLAG-tagged alpha-centractin (ACTR1A) (a) and of 3XFLAG-tagged EXOC3 (b) followed by western blot for native DISC1 using a polyclonal antibody. The approximately 80 kDa DISC1 isoform identified in this experiment (arrow) is consistent with the predominant 75–80 kDa isoform of DISC1 in human brain, HEK293, HeLa (cervical carcinoma) cells, and SH-SY5Y (neuroblastoma) cells (Miyoshi et al. 2003; James et al. 2004; Sawamura et al. 2005; Ishizuka et al. 2006).

Genes encoding DTNBP1 and dynactin interaction networks are significantly over-represented within chromosomal regions linked to schizophrenia

We used whole genome linkage data to determine if the genes that encode members of our DTNBP1, exocyst complex, and dynactin complex protein interaction sets tend to be located in cytogenetic regions linked to schizophrenia. Genetic linkage data were obtained from the Psychiatric Genetics Evidence Project Linkage Database (http://slep.unc.edu/evidence/) (Tables S4–S6). This database compiles findings from manual reviews of 144 papers in psychiatric genetics across a variety of disorders and includes studies on genome wide linkage and association, copy number variation, and gene expression. It shows that 366 of a total of 826 cytogenetic regions have been found to have strong or suggestive linkage to schizophrenia in whole genome scans. The DTNBP1 and the dynactin protein interaction networks (including all core complex members) comprise 96 and 34 total proteins, respectively, with 55 and 24 of these in schizophrenia linked cytogenetic regions (Chi-squared test; p = 0.017 and 0.006, respectively). In the 71 member exocyst network 34 genes were in schizophrenia linked cytogenetic regions, however this result was not significant (Chi-squared test; p = 0.619).

Discussion

It has been suggested that compromised neurotransmission as a result of aberrations in synaptic trafficking of endosomal vesicles and their neurotransmitter related cargoes may contribute to the etiology of schizophrenia (Ryder and Faundez 2009). DTNBP1 is linked genetically to schizophrenia and has an essential role in synaptic vesicle trafficking and homeostatic modulation of neurotransmission (Dickman and Davis 2009). Here, we investigate DTNBP1 protein interactions to identify additional proteins, pathways, and functions of possible relevance to schizophrenia. Our data significantly expand the overall protein interaction network for DTNBP1 and show that it is involved throughout the vesicle life cycle and vesicle trafficking system, from vesicle biogenesis and cargo sorting (BLOC1, DTNBP1, AP3) to vesicle trafficking (dynactin, tubulin/actin proteins), to membrane targeting and vesicle tethering (exocyst). We confirmed that DISC1 also interacts with the exocyst and dynactin complexes, indicating that both of these schizophrenia-implicated proteins may function through common vesicle trafficking processes (Fig. 1).

Our results include human cytosolic protein interactions for DTNBP1 obtained from endogenous cellular complexes and complement a recent study that focused on membrane-associated protein interactions of DTNBP1 (Hikita et al. 2009). Our results are consistent with previous studies that show interactions between DTNBP1 and members of the AP3 complex (Di Pietro et al. 2006; Hikita et al. 2009). As we focused on cytosolic interactions of DTNBP1 while Hikita et al. 2009 investigated membrane interactions of DTNBP1, there were a number of membrane or membrane-associated proteins they observed that were not replicated in our analysis, including STXBP1, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), ADP-ribosylation factor 1 (ARF1), ADP-ribosylation factor GTPase-activating protein 1 (ARFGAP), and Arf-GAP with SH3 domain, ANK repeat and PH domain-containing protein 1 (ASAP1). Conversely, most of the proteins found in our study were not found in the Hikita et al. 2009 study, including any members of BLOC1, of which DTNBP1 is known to be a member of (Li et al. 2003), nor members of the exocyst and dynactin complexes. These differences are not surprising given the different experimental approaches, but further highlight that DTNBP1 plays a role in many aspects of the vesicle lifecycle, both within the cytoplasm as well as at the membrane surface. In this way, our results complement the findings in Hikita et al. 2009 and implicate DTNBP1 in vesicle associated protein trafficking processes throughout the cell (Hikita et al. 2009).

Our data indicate DTNBP1 interacts with several actin and tubulin proteins, and with members of the CCT and dynactin complexes, which are fundamental components of the cytoskeleton and microtubule matrix. The CCT complex is a large multi-subunit complex that mediates protein folding for a variety of proteins, including several actin, tubulin and cell cycle regulator proteins (Frydman et al. 1994) and plays an important role in the assembly of the cytoskeleton and in cell division (Grantham et al. 2006). The 11 subunit dynactin complex (Schroer 2004) recruits and links the dynein motor protein to vesicles and to microtubules to facilitate cargo transport along the cytoskeletal matrix (Schroer and Sheetz 1991). In neuronal axons, the dynactin complex is involved in retrograde as well as anterograde transport (Schnapp and Reese 1989; Hafezparast et al. 2003). Mutations in the dynactin complex subunits can cause defects in axonal transport, for instance, individuals with mutations in the DCTN1 subunit suffer from motor neuron disease (Puls et al. 2003). The function of the dynactin complex may vary based on location, but in all cases it is thought to facilitate or regulate dynein or kinesin II targeting and recruitment. Our data also show the exocyst complex contacts the cytoskeletal transport system. In fact, the three proteins in common between DTNBP1 and the exocyst and dynactin complexes are CCT proteins (CCT3 and CCT8) and F-actin-capping protein subunit beta (CAPZB), an actin-capping protein and a member of the dynactin complex (Figure S4). We have also verified a DISC1 interaction with the dynactin complex. DISC1 has been shown to play an important role in microtubule dynamics through its interactions with pericentrin (PCNT) (Shimizu et al. 2008), and its interaction with the dynactin-dynein accessory components PAFAH1B1 and NDEL1 are thought to stabilize the motor assembly on the nuclear surface, the centrosome, and the cell cortex (Morris et al. 2003; Ozeki et al. 2003; Brandon et al. 2004; Liang et al. 2004). Taken together, our data indicate that cytoskeletal interactions may be important for the vesicle transport functions of DTNBP1, DISC1 and the exocyst complex.

Our data indicate DTNBP1 makes contact with several members of the exocyst complex. We also verified in mammalian cells an interaction of DISC1 with the exocyst complex. The exocyst complex is composed of eight subunits, and through SNARE protein recruitment, is thought to integrate signals from several different signaling pathways to determine the location, timing and number of secretory vesicles docked with the plasma membrane (Munson and Novick 2006). In animal neurons, the exocyst complex is required for neurite branching and synaptogenesis, but not synaptic vesicle release at mature synapses (Murthy et al. 2003; Lalli and Hall 2005). We also reproduced the interaction of DTNBP1 with the AP3 complex. AP3 is one of four adaptor protein complexes (AP1-4) that act as scaffolds, coordinating membrane lipids, membrane protein sorting signals, components of the vesicle fusion machinery, and additional components of the vesicle formation apparatus (Boehm et al. 2001; Robinson 2004; Sorkin 2004). The AP3 complex is involved in the generation of synaptic vesicles in neurons. Mutations in the AP3 complex δ subunit AP3D1 result in the mocha mouse, affect spontaneous and evoked release at hippocampal mossy fiber synapses (Scheuber et al. 2006), and show a selective increase in the content of synaptic vesicle cargoes. Our data potentially extend the involvement of DTNBP1 and DISC1 into vesicle tethering through the exocyst complex and the regulation of active vesicle transport through the dynactin complex. While our data show both DTNBP1 and DISC1 interact with components of the dynactin and exocyst complexes it is not known if DTNBP1 and DISC1 reside simultaneously in common protein complexes. Overall, our data indicate that DTNBP1 is involved throughout the vesicle transport process in vesicle generation, transport and membrane tethering through interactions with the BLOC1, AP3, exocyst, and dynactin complexes and that DISC1 is also involved in the vesicle lifecycle at the transport and membrane tethering stages.

Defects in any of the many factors that coordinately regulate neurotransmitter vesicle cycling could contribute to the etiology of schizophrenia (Ryder and Faundez 2009). DTNBP1 over-expression induces expression of the SNARE protein synaptosomal-associated protein 25 (SNAP25) which is involved in mediating vesicle docking and fusion at the cellular membrane, and synapsin-1 (SYN1) which is involved in regulating neurotransmitter release as well as increasing glutamatergic release (Numakawa et al. 2004). Defects in BLOC1 indirectly cause redistribution of cargo to the cell surface (Setty et al. 2007). AP3 and BLOC1 are linked to the fusion machinery involved in synaptic vesicle secretion that is hypothesized to be involved in schizophrenia; brain tissue from schizophrenia patients has reduced levels of DTNBP1 in hippocampal mossy fibers (Talbot et al. 2004), a phenotype that is also found in the AP3 deficient mocha brain (Salazar et al. 2006). Ablation of DTNBP1 expression in mouse and rat model systems results in the diversion of dopamine and glutamate receptors from lysosomes to the cell surface (Iizuka et al. 2007; Ji et al. 2009; Tang et al. 2009). Our data is consistent with the possibility that disruption of a DTNBP1-exocyst interaction may contribute to the diversion of glutamate and dopamine receptor containing vesicles from lysosomes to the plasma membrane. Overall, our results show DTNBP1 and DISC1 make multiple contacts with vesicle trafficking machinery. Thus, the aberrant function of DTNBP1 or DISC1, or the proteins in their interaction networks provide multiple sites for impairment of synaptic vesicle biogenesis. Our finding that genes for the proteins in our DTNBP1 and dynactin protein interaction networks are over-represented in schizophrenia linked cytogenetic regions implies that the interaction networks of DTNBP1 and DISC1 as a whole should be considered in the pathology of schizophrenia.

Defining the protein interaction networks for schizophrenia susceptibility genes like DTNBP1 will help understand their function, but may also illuminate pathways where individual components may make small contributions to disease etiology, but a large contribution when taken together. An example of this is the epistatic genetic effect in schizophrenia of the non-DTNBP1 BLOC1 subunits BLOC1S3 and MUTED (Morris et al. 2008). While members of the dynactin and exocyst complexes have not been considered potential candidate schizophrenia genes in the past, their functions and their interaction with two of the best schizophrenia susceptibility genes strongly supports future genetic investigation of these and other members of the DTNBP1, exocyst, dynactin, and DISC1 interaction networks.

Acknowledgements

This work was funded by the Mind Foundation of British Columbia, the Michael Smith Foundation for Health Research, The Canadian Institutes for Health Research and the British Columbia Cancer Agency.

Ancillary