Identification and Verification of Novel Rodent Postsynaptic Density Proteins
2004; Elsevier BV; Volume: 3; Issue: 9 Linguagem: Inglês
10.1074/mcp.m400045-mcp200
ISSN1535-9484
AutoresBryen A. Jordan, Brian D. Fernholz, Muriel Boussac, Chong‐Feng Xu, Gabriela Grigorean, Edward B. Ziff, Thomas A. Neubert,
Tópico(s)Ion channel regulation and function
ResumoThe postsynaptic density (PSD) is a cellular structure specialized in receiving and transducing synaptic information. Here we describe the identification of 452 proteins isolated from biochemically purified PSD fractions of rat and mouse brains using nanoflow HPLC coupled to electrospray tandem mass spectrometry (LC-MS/MS). Fluorescence microscopy and Western blotting were used to verify that many of the novel proteins identified exhibit subcellular distributions consistent with those of PSD-localized proteins. In addition to identifying most previously described PSD components, we also detected proteins involved in signaling to the nucleus as well as regulators of ADP-ribosylation factor signaling, ubiquitination, RNA trafficking, and protein translation. These results suggest new mechanisms by which the PSD helps regulate synaptic strength and transmission. The postsynaptic density (PSD) is a cellular structure specialized in receiving and transducing synaptic information. Here we describe the identification of 452 proteins isolated from biochemically purified PSD fractions of rat and mouse brains using nanoflow HPLC coupled to electrospray tandem mass spectrometry (LC-MS/MS). Fluorescence microscopy and Western blotting were used to verify that many of the novel proteins identified exhibit subcellular distributions consistent with those of PSD-localized proteins. In addition to identifying most previously described PSD components, we also detected proteins involved in signaling to the nucleus as well as regulators of ADP-ribosylation factor signaling, ubiquitination, RNA trafficking, and protein translation. These results suggest new mechanisms by which the PSD helps regulate synaptic strength and transmission. Neurons are highly polarized cells, specializing in the reception of numerous, independent signal inputs and rapid integration of these inputs into an electrochemical response. The major sites of signal input are synapses, which are highly ordered cell junctions formed between two neurons and are typically unidirectional in fast excitatory chemical neurotransmission in the mammalian CNS. The response to neurotransmitter (NT) 1The abbreviations used are: NT, neurotransmitter; PSD, postsynaptic density; ARF, ADP-ribosylation factor; LTP, long-term potentiation; EM, electron microscopy; 2DE, two-dimensional electrophoresis; E-value, expectation value; NLS, nuclear localization signal; MOCA, modifier of cell adhesion; CYLD, cylindromatosis turban tumor syndrome protein; GEF, guanine exchange factor; GAP; GTPase-activating protein, ARVCF, armadillo-repeat-velo-cardio-facial protein; eGFP, enhanced green fluorescent protein. 1The abbreviations used are: NT, neurotransmitter; PSD, postsynaptic density; ARF, ADP-ribosylation factor; LTP, long-term potentiation; EM, electron microscopy; 2DE, two-dimensional electrophoresis; E-value, expectation value; NLS, nuclear localization signal; MOCA, modifier of cell adhesion; CYLD, cylindromatosis turban tumor syndrome protein; GEF, guanine exchange factor; GAP; GTPase-activating protein, ARVCF, armadillo-repeat-velo-cardio-facial protein; eGFP, enhanced green fluorescent protein. release at the synapse is provided by a protein matrix of NT receptors and supporting proteins collectively known as the postsynaptic density (PSD) (for review, see Refs. 1Ziff E.B. Enlightening the postsynaptic density..Neuron. 1997; 19: 1163-1174Google Scholar, 2Kennedy M.B. Signal-processing machines at the postsynaptic density..Science. 2000; 290: 750-754Google Scholar, 3Yamauchi T. Molecular constituents and phosphorylation-dependent regulation of the post-synaptic density..Mass Spectrom. Rev. 2002; 21: 266-286Google Scholar). The PSD has several proposed functions including: signal amplification, cytoskeletal anchorage, biochemical signaling regulation, and NT receptor clustering (1Ziff E.B. Enlightening the postsynaptic density..Neuron. 1997; 19: 1163-1174Google Scholar, 4Banker G. Churchill L. Cotman C.W. Proteins of the postsynaptic density..J. Cell Biol. 1974; 63: 456-465Google Scholar, 5Blomberg F. Cohen R.S. Siekevitz P. The structure of postsynaptic densities isolated from dog cerebral cortex. II. Characterization and arrangement of some of the major proteins within the structure..J. Cell Biol. 1977; 74: 204-225Google Scholar, 6Kennedy M.B. The postsynaptic density..Curr. Opin. Neurobiol. 1993; 3: 732-737Google Scholar). Changes in size and composition of the PSD correlate with changes in synaptic strength (7Dosemeci A. Tao-Cheng J.H. Vinade L. Winters C.A. Pozzo-Miller L. Reese T.S. Glutamate-induced transient modification of the postsynaptic density..Proc. Natl. Acad. Sci. U S A. 2001; 98: 10428-10432Google Scholar, 8Kasai H. Matsuzaki M. Noguchi J. Yasumatsu N. Nakahara H. Structure-stability-function relationships of dendritic spines..Trends Neurosci. 2003; 26: 360-368Google Scholar), including alterations that are stably maintained such as long-term potentiation (LTP), a physiologically relevant increase in synaptic efficacy and a model for learning and memory (9Milner B. Squire L.R. Kandel E.R. Cognitive neuroscience and the study of memory..Neuron. 1998; 20: 445-468Google Scholar, 10Miller S. Mayford M. Cellular and molecular mechanisms of memory: The LTP connection..Curr. Opin. Genet. Dev. 1999; 9: 333-337Google Scholar). Therefore, an understanding of the protein composition of the PSD is a prerequisite for modeling the molecular interactions regulating synaptic strength. The structure of PSDs purified from rodent brains using gradient centrifugation and Triton X-100 extraction has been shown by electron microscopy (EM) to be virtually identical to the “in vivo” PSD structure (4Banker G. Churchill L. Cotman C.W. Proteins of the postsynaptic density..J. Cell Biol. 1974; 63: 456-465Google Scholar, 11Cohen R.S. Blomberg F. Berzins K. Siekevitz P. The structure of postsynaptic densities isolated from dog cerebral cortex. I. Overall morphology and protein composition..J. Cell Biol. 1977; 74: 181-203Google Scholar). Gel electrophoresis, enzymatic activity assays, and EM experiments have demonstrated that this procedure yields a highly pure, membrane-free PSD fraction (11Cohen R.S. Blomberg F. Berzins K. Siekevitz P. The structure of postsynaptic densities isolated from dog cerebral cortex. I. Overall morphology and protein composition..J. Cell Biol. 1977; 74: 181-203Google Scholar, 12Carlin R.K. Grab D.J. Cohen R.S. Siekevitz P. Isolation and characterization of postsynaptic densities from various brain regions: Enrichment of different types of postsynaptic densities..J. Cell Biol. 1980; 86: 831-845Google Scholar). Recent proteomic studies have investigated the composition of the PSD by SDS-PAGE or two-dimensional gel electrophoresis (2DE) coupled with MS (13Walikonis R.S. Jensen O.N. Mann M. Provance Jr., D.W. Mercer J.A. Kennedy M.B. Identification of proteins in the postsynaptic density fraction by mass spectrometry..J. Neurosci. 2000; 20: 4069-4080Google Scholar, 14Satoh K. Takeuchi M. Oda Y. Deguchi-Tawarada M. Sakamoto Y. Matsubara K. Nagasu T. Takai Y. Identification of activity-regulated proteins in the postsynaptic density fraction..Genes Cells. 2002; 7: 187-197Google Scholar, 15Yoshimura Y. Shinkawa T. Taoka M. Kobayashi K. Isobe T. Yamauchi T. Identification of protein substrates of Ca 2+/calmodulin-dependent protein kinase II in the postsynaptic density by protein sequencing and mass spectrometry..Biochem. Biophys. Res. Commun. 2002; 290: 948-954Google Scholar, 16Li K.W. Hornshaw M.P. Van Der Schors R.C. Watson R. Tate S. Casetta B. Jimenez C.R. Gouwenberg Y. Gundelfinger E.D. Smalla K.H. Smit A.B. Proteomics analysis of rat brain postsynaptic density: Implications of the diverse protein functional groups for the integration of synaptic physiology..J. Biol. Chem. 2004; 279: 987-1002Google Scholar). Li et al. (16Li K.W. Hornshaw M.P. Van Der Schors R.C. Watson R. Tate S. Casetta B. Jimenez C.R. Gouwenberg Y. Gundelfinger E.D. Smalla K.H. Smit A.B. Proteomics analysis of rat brain postsynaptic density: Implications of the diverse protein functional groups for the integration of synaptic physiology..J. Biol. Chem. 2004; 279: 987-1002Google Scholar) also performed shotgun proteomics using cysteine-containing peptides selected using ICAT techniques. However, each of these investigations identified less than one-third of previously described and biochemically confirmed PSD components, pointing to limitations in the techniques used. A recent paper by Yoshimura et al. (17Yoshimura Y. Yamauchi Y. Shinkawa T. Taoka M. Donai H. Takahashi N. Isobe T. Yamauchi T. Molecular constituents of the postsynaptic density fraction revealed by proteomic analysis using multidimensional liquid chromatography-tandem mass spectrometry..J. Neurochem. 2004; 88: 759-768Google Scholar) reports the identification by mass spectrometry of 492 proteins in the PSD, which suggests that the PSD is more complex than previously thought. However, the study was of PSDs from a subset of whole brain, and no attempt was made to confirm the localization of these proteins by independent means. In our study, we have taken advantage of increased sensitivity in protein identification afforded by the use of SDS-PAGE to fractionate proteins followed by in-gel tryptic digestion and nanoflow LC-MS/MS (for review, see Ref. 18Wu C.C. Yates 3rd., J.R. The application of mass spectrometry to membrane proteomics..Nat. Biotechnol. 2003; 21: 262-267Google Scholar). LC-MS/MS provides separation in a second dimension without the loss of hydrophobic or basic proteins as with 2DE (19Santoni V. Molloy M. Rabilloud T. Membrane proteins and proteomics: Un amour impossible?.Electrophoresis. 2000; 21: 1054-1070Google Scholar). We report the identification of 452 proteins in PSDs isolated from whole brain using stringent statistical criteria for validation of MS-based matches. These proteins include over 90% of published, biochemically confirmed PSD components, in addition to 307 proteins not previously shown to be in the PSD and including 75 previously uncharacterized proteins. We have expressed 16 of the novel proteins as recombinant fluorescent proteins in neurons and confirmed their localization in dendritic spines. Furthermore, subcellular fractions from our PSD purification probed with antisera against 18 additional novel proteins demonstrate that all of these proteins are present and many are enriched in the PSD fraction. Western blots of known pre- and postsynaptic proteins confirm the purity of the biochemically prepared PSD fraction used in this study. Taken together, these experiments validate the protein identifications obtained by MS. This analysis of the fundamental constituents of the PSD provides new insights into its multiple functions including protein translation, trafficking, and turnover and increases our understanding of the molecular components of learning and memory. Our protocol is based on that of Cohen and Carlin (11Cohen R.S. Blomberg F. Berzins K. Siekevitz P. The structure of postsynaptic densities isolated from dog cerebral cortex. I. Overall morphology and protein composition..J. Cell Biol. 1977; 74: 181-203Google Scholar, 12Carlin R.K. Grab D.J. Cohen R.S. Siekevitz P. Isolation and characterization of postsynaptic densities from various brain regions: Enrichment of different types of postsynaptic densities..J. Cell Biol. 1980; 86: 831-845Google Scholar). Rats or mice were euthanized by CO 2 in compliance with New York University Medical Center’s Institutional Animal Care and Use Committee, and whole brains were rapidly removed and placed in ice-cold solution A (0.32 m sucrose, 1 mm NaHCO 3, 1 mm MgCl 2, 0.5 mm CaCl 2, 0.1 mm PMSF (Sigma, St. Louis, MO) and 1× Complete Protease Inhibitors (Roche Applied Science, Indianapolis, IN). The brains were subjected to dounce homogenization in 40 ml of solution A per 10 g of wet brain tissue. The homogenates were diluted to 10% weight/volume with solution A and centrifuged at 1,400 × g for 10 min. The supernatant solution was saved, and the pellet was rehomogenized in 10% solution A per 10 g of initial weight and subjected to a centrifugation at 710 × g for 10 min. Supernates were pooled and subjected to a second centrifugation at 710 × g for 10 min. The supernates were then spun at 30,000 × g for 15 min to obtain a crude P2 fraction. The pellet was resuspended in solution B (0.32 m sucrose, 1 mm NaHCO 3) using 24 ml of solution B per 10 g of starting material. This homogenate was layered on top of a 10-ml 0.85 m sucrose, 10-ml 1 m sucrose, and 10-ml 1.2 m sucrose step sucrose gradient and centrifuged at 82,500 × g for 2 h. Purified synaptosomes were collected at the 1 m and 1.2 m sucrose interface by syringe aspiration. If the “short” procedure was used, we proceeded directly to the Triton extraction section. If the “long” method was used, the synaptosomes were resuspended with 4 volumes of solution B and collected by centrifugation at 48,200 × g for 20 min. We hypotonically lysed synaptosomes by resuspending the pellet in 10 ml of 6 mm Tris pH 8.1 per gram of starting material and rocking at 4 °C for 45 min. The lysed synaptosomes were then collected by centrifugation at 48,200 × g for 20 min followed by the Triton extraction method below. Synaptosomes were diluted to 60 ml of Solution B per 10 g of initial starting material and an equal volume of 1% Triton X-100, 0.32 m sucrose, 12 mm Tris pH 8.1, was added and rocked at 4 °C for 15 min. These lysed synaptosomes were centrifuged at 32,800 × g for 25 min, and the pellet was resuspended in 2.5 ml of solution B per 10 g of starting material. The suspension was layered on a second sucrose step gradient of 3 ml of 1 m sucrose, 3 ml of 1.5 m sucrose, and 4 ml of 2 m sucrose and centrifuged at 200,000 × g for 2 h. The PSD fraction was collected at the 1.5 m and 2 m sucrose interface and diluted to 6 ml with solution B. An equal volume of 1% Triton X-100, 150 mm KCl was added to obtain the “two-Triton” PSD fraction. These purified PSDs were then collected by centrifugation at 200,000 × g for 20 min. PSDs were solubilized using either an IEF rehydration buffer (9 m urea, 2 m thiourea, 4% CHAPS, 0.5% Triton X-100 in the rat “short” version) or a 2% SDS-Tris buffer in the “long” and mouse “short” protocols. The purity of the fractions was assessed by Western blots for CaMKII, NMDA receptor 1, synaptophysin, synaptic vesicle 2, and PSD-95. We isolated PSDs from rats using the “long” and “short” procedure, and from mice, we obtained PSDs using the “short” procedure. PSDs obtained from rats using the “short” procedure were solubilized using the IEF buffer described above containing urea. Proteins were separated by SDS-PAGE on 10% Criterion gels (Bio-Rad, Hercules, CA). After electrophoresis the proteins were visualized by Coomassie blue staining. Twenty-five (rat PSD “long” and “short” protocols) to 50 (mouse PSD “short” protocol) evenly spaced gel bands were excised from each lane without consideration of staining intensity, destained, and the proteins digested in-gel with trypsin under a tissue culture hood to minimize contamination (20Shevchenko A. Wilm M. Vorm O. Mann M. Mass spectrometric sequencing of proteins silver-stained polyacrylamide gels..Anal. Chem. 1996; 68: 850-858Google Scholar). The resulting peptides were extracted and dried under vacuum, then resuspended in 5–10 μl of 0.1% TFA. The peptide mixtures were analyzed using nanoflow LC/ESI-MS-MS. The peptides were loaded onto a 0.3 × 1-mm C18 nano-precolumn (LC Packings, Sunnyvale, CA), then washed 5 min with 2% ACN in 0.1% formic acid at a flow rate of 20 μl/min. After washing, flow was reversed through the precolumn and the peptides eluted with a gradient of 2–81% ACN in 0.1% formic acid. The gradient was delivered over 80 or 120 min by a CapLC (Waters, MA) HPLC system at a flow rate of 200 nl/min, obtained by a 15:1 precolumn flow split, through a 75-μm × 15-cm fused silica capillary C18 HPLC column (LC Packings PepMap) to a fused silica distal end-coated tip nano-electrospray needle (New Objective, Woburn, MA). The Q-TOF 1 (Micromass, Manchester, United Kingdom) data acquisition involved MS survey scans and automatic data-dependent MS/MS acquisitions, which were invoked after selected ions met preset parameters of minimum signal intensity of 8 counts per second, ion charge state 2+, 3+, or 4+, and appropriate retention time. Survey scans of 1 s were followed by CID of the four most intense ions for up to 11 s each, or until 5,000 total MS/MS ion counts per precursor peptide were achieved. The raw MS data were subsequently processed using manufacturer-supplied ProteinLynx software, which generated DTA files based on each MS/MS spectrum. Control experiments demonstrate the sensitivity of this system is better than 100 fmol of each protein in the gel, and 10 fmol of each peptide injected onto the HPLC columns. cDNA was obtained from Kazusa Research Institute (21Ohara O. Nagase T. Ishikawa K. Nakajima D. Ohira M. Seki N. Nomura N. Construction and characterization of human brain cDNA libraries suitable for analysis of cDNA clones encoding relatively large proteins..DNA Res. 1997; 4: 53-59Google Scholar) (“KIAA” prefix) or cloned from a rat brain cDNA library constructed with Superscript 3 (Invitrogen, Carlsbad, CA) and using PCR with Deep Vent (New England Biolabs, Beverly, MA) or Accuprime (Invitrogen) and gene specific primers. PCR products were digested and ligated into pEGFP vectors: N1 or N3 and C1 or C3 (Clontech, Palo Alto, CA) and pcDNA3.1/myc-His(-)A (Invitrogen). Dissociated neuronal cultures were prepared as previously described (22Osten P. Khatri L. Perez J.L. Kohr G. Giese G. Daly C. Schulz T.W. Wensky A. Lee L.M. Ziff E.B. Mutagenesis reveals a role for ABP/GRIP binding to GluR2 in synaptic surface accumulation of the AMPA receptor..Neuron. 2000; 27: 313-325Google Scholar) with the following modification: an equal amount of neocortex was included with the hippocampus. Three- to 4-week-old neurons were transfected with 4 μg of DNA with either a modified CaPO 4 (23Passafaro M. Piech V. Sheng M. Subunit-specific temporal and spatial patterns of AMPA receptor exocytosis in hippocampal neurons..Nat. Neurosci. 2001; 4: 917-926Google Scholar) or LipofectAMINE 2000 (Invitrogen) according to the manufacturer’s instructions. Cells were mounted with Vectashield mounting media, which contained 4′,6′-diamidino-2-phenylindole (DAPI) to visualize nuclei (Vector Labs, Burlingame, CA). Enhanced green fluorescent protein (eGFP) was visualized using an Axiovert 220M fluorescence microscope (Zeiss, Oberkochen, Germany). Ten to 20 μg of PSD were loaded onto 4–12% gradient preformed criterion gels (Bio-Rad), transferred and blotted using a variety of antibodies generously donated by Catherine Chew (LASP-1) (24Chew C.S. Parente Jr., J.A. Chen X. Chaponnier C. Cameron R.S. The LIM and SH3 domain-containing protein, lasp-1, may link the cAMP signaling pathway with dynamic membrane restructuring activities in ion transporting epithelia..J. Cell Sci. 2000; 113: 2035-2045Google Scholar), Phillip J Coates (prohibitin and BAP37), Matt Welch (ARP2, ARP3, and ARP2/3sub5), Francis Castets (zinedin), Karl Matter (guanine exchange factor (GEF)-LFC), Katherine Wilson (barrier to autointegration-1), AB Reynolds (ARVCF and catenin p120), Gary Silverman (squamous cell carcinoma antigen-1), Michael Greenberg (ephexin N-GEF), Sharon Eden (NCK-associated protein 1-NAP125 and Wave1), Jordi Perez i Tur (LGN-1), Orly Reiner (doublecortin-like kinase), and Massimo Pietropaolo (islet cell autoantigen p69). Approximately 40,000 uninterpreted mass spectra were acquired and used for protein identification as follows. The raw MS data were processed using Micromass ProteinLynx 3.5 software. Sequential MS/MS scans with the same precursor ions were combined before charge state deconvolution by MaxEnt 3 software (start mass 700, peak width auto, 1 ensemble member, 20 iterations, data compressed). Background was subtracted (polynomial order 10, 10% below curve removed), peaks were smoothed (2 channel window, 1 smooth, Savitzky Golay model), and centroided (minimum peak width at half height 4, centroid top 80%). From these data, DTA files based on each MS/MS spectrum were produced, merged into a text file, and used for database searching. Each query was searched using Mascot version 1.9.05 (25) (Matrix Science, London, United Kingdom) with permutations of the following search parameters: missed cleavages (0–3), modifications (unmodified, oxidation (M), deamidation (NQ), phosphorylation (ST), phosphorylation (Y), carbamylation (K), carbamylation (N-term)), taxonomy (mammalia, mus musculus, rattus), peptide tolerance in Daltons (0.25, 0.5, 1, 1.4), and MS/MS tolerance in Daltons (0.15, 0.25, 0.3, 0.4, 0.5). Unmodified parameters were as follows: peptide-charge (2+ and 3+), enzyme (trypsin), database (NCBI nonredundant August 16, 2003 with 1502194 sequences; 484011957 residues), instrument (ESI-QUAD-TOF), and monoisotopic masses were used. Peptides were searched with the highest number of potential modifications and with mass tolerances several times greater than the accuracy of the mass spectrometer so as to maximize the number of identified peptides. Peptides were then subjected to a series of increasingly stringent filters as variable modifications were reduced, and mass accuracy tolerances were reduced to approximate the accuracy of the mass spectrometer. The peptides were retained from the most stringent filter they passed in a manner similar to the “step analysis” algorithm of DTASelect and Contrast allows for Sequest search results (26Tabb D.L. McDonald W.H. Yates 3rd., J.R. DTASelect and Contrast: Tools for assembling and comparing protein identifications from shotgun proteomics..J. Proteome. Res. 2002; 1: 21-26Google Scholar). This analysis allowed for the optimal search condition to be retained for each peptide and overcomes the computational limitations of Mascot. This resulted in the maximum number of peptides to be identified from the 40,000 initial MS/MS spectra. Values of p were calculated using the equation given by Mascot software to convert scores to probabilities for each query, x: p(x)=10(−Score/10)(Eq. 1) where Score is the score given to the peptide by Mascot. This p value represents the probability that the sequence assignment is random, but does not reflect the database size, i.e. the number of potential matches for a given peptide m/z value within the peptide tolerance range. The integer Qmatch reflects the virtual database size within this range, and thus the overall probability this match is a false positive can be represented as follows: p′(x)=Qmatch∗p(x)(Eq. 2) To assess the probability of correct protein identifications based on these peptide matches, the number of MS/MS spectra used in the search must be taken into account. To do this, each protein identifications is given an expectation value (E-value), E, dependent upon the number of queries (MS/MS spectra), q, and p′(x) for each unique matching peptide. For proteins with only one peptide match: E=q∗p′(x)(Eq. 3) and for proteins with y peptides where y > 1: E=q(0.5)∗Π(p′(x)), where x=1 to y(Eq. 4) which reflects the low probability of two or more peptides matching to one protein randomly (27Fenyo D. Beavis R.C. A method for assessing the statistical significance of mass spectrometry-based protein identifications using general scoring schemes..Anal. Chem. 2003; 75: 768-774Google Scholar). Searches were integrated using custom software written in Java 1.3 (Sun Microsystems, Santa Clara, CA). Briefly, the Mascot search that yielded the lowest p′(x) for each peptide was used for all 40,000 queries. Peptides with p′(x) > 0.5 or Score < 15 or that were not at least 4 aa in sequence length were immediately discarded. Up to three top-scoring peptide matches were considered as long as each score was no less than 63% of the top-ranking peptide’s score, but only rarely did any second-ranking peptide pass these filters. The protein list was then constructed using a simple matrix matching proteins to any corresponding peptide identified by Mascot for all peptides that passed the filter. Any proteins without at least one peptide with p′(x) < 0.05, the threshold score determined by Mascot, were removed. All proteins were subsequently manually inspected for redundancy within clusters generated by grouping any proteins having at least 50% of the identified peptides in common. As a follow up, a second clustering was generated with NCBI BLASTCLUST (28Dondoshansky I. Wolf Y. BLASTCLUST. 2.2.6 Ed. National Institutes of Health, Bethesda, MD2000Google Scholar) (minimum homology length = 0.0, percent identical residues = 80%) and manually inspected to remove orthologs that could not be clustered by common peptide analysis. The cutoff used for inclusion in the list was a maximum E-value of 1.0. To test the validity of our statistical model for evaluating the validity of protein identifications, a random NCBI nonredundant database was generated to determine how well our E-value cutoff of 1 was able to prevent false positives. Each protein FASTA entry was rearranged such that the original sequence was destroyed, but the amino acid frequency and the protein length were unchanged for each entry. PSDs were purified from rat and mouse whole brains by density centrifugation using the “long” and “short” methods of Cohen et al. (11Cohen R.S. Blomberg F. Berzins K. Siekevitz P. The structure of postsynaptic densities isolated from dog cerebral cortex. I. Overall morphology and protein composition..J. Cell Biol. 1977; 74: 181-203Google Scholar) and Carlin et al. (12Carlin R.K. Grab D.J. Cohen R.S. Siekevitz P. Isolation and characterization of postsynaptic densities from various brain regions: Enrichment of different types of postsynaptic densities..J. Cell Biol. 1980; 86: 831-845Google Scholar), respectively. While subtle differences in staining intensities of individual proteins could be observed between Coomassie-stained SDS-PAGE gels of mouse and rat PSDs isolated by either method, the overall protein profiles appeared to be similar (Fig. 1). Each gel lane was subsequently cut into 25–50 gel slices, digested with trypsin, and analyzed for peptide content by LC-MS/MS. In total, ∼40,000 MS spectra were obtained from ∼120 separate LC-MS/MS experiments. These spectra were processed by Micromass MassLynx software and used to search the nonredundant NCBI database with in-house Mascot search software (25Perkins D.N. Pappin D.J. Creasy D.M. Cottrell J.S. Probability-based protein identification by searching sequence databases using mass spectrometry data..Electrophoresis. 1999; 20: 3551-3567Google Scholar). The results from a variety of Mascot search conditions, including appropriate variable modifications, a range of tolerances, and multiple taxonomies, were integrated using custom software. This method allowed us to find the optimal search condition for each peptide in a manner similar to that of DTASelect software (26Tabb D.L. McDonald W.H. Yates 3rd., J.R. DTASelect and Contrast: Tools for assembling and comparing protein identifications from shotgun proteomics..J. Proteome. Res. 2002; 1: 21-26Google Scholar). The compiled results were then clustered by custom software and BLASTCLUST (28Dondoshansky I. Wolf Y. BLASTCLUST. 2.2.6 Ed. National Institutes of Health, Bethesda, MD2000Google Scholar), which allowed orthologs and database redundancies to be collapsed manually from 4,418 initially identified proteins to ∼750. The identified proteins have been reported with their calculated E-values, i.e. the number of times they would be expected to match any target randomly during a database search. We have determined an E-value cutoff mathematically (see “Experimental Procedures”) and tested this value empirically by searching our entire query list against a randomized NCBI database using parameters similar to those used for our peptide identifications and integrated as described. The lowest E-value observed for a protein identification using the randomized database was 1.3, providing strong empirical corroboration for our mathematical E-value cutoff of 1. This result suggests, statistically, that we have made no false-positive protein identifications. Some proteins were identified on the basis of MS/MS spectra that appeared reasonable when inspected manually, but were excluded from our high-confidence list of 452 proteins (Table I) because they had E-values greater than 1. Moreover, several proteins that failed our rigorous statistical test have been shown by us (see below) and by others to localize to the PSD. Therefore, we have included all identified proteins in our supplemental data (Supplemental Table I). We have compared the proteins identified in this study to previously known and biochemically confirmed proteins of the PSD listed in several reviews (1Ziff E.B. Enlightening the postsynaptic density..Neuron. 1997; 19: 1163-1174Google Scholar, 2Kennedy M.B. Signal-processing machines at the postsynaptic density..Science. 2000; 290: 750-754Google Scholar, 3Yamauchi T. Molecular constituents and phosphorylation-dependent regulation of the post-synaptic density..Mass Spectrom. Rev. 2002; 21: 266-286Google Scholar, 29Walsh M.J. Kuruc N. The postsynaptic density: Constituent and associated proteins characterized by electrophoresis, immunoblotting, and peptide sequencing..J. Neurochem. 1992; 59: 667-678Google Scholar). We were unable to compare our set to the recent publication by Yoshimura et al. (17Yoshimura Y. Yamauchi Y. Shinkawa T. Taoka M. Donai H. Takahashi N. Isobe T. Yamauchi T. Molecular constituents of the postsynaptic density fraction revealed by proteomic analysis using multidimensional liquid chromatography-ta
Referência(s)