Neuropeptidomics Strategies for Specific and Sensitive Identification of Endogenous Peptides

Abstract

A new approach using targeted sequence collections has been developed for identifying endogenous peptides. This approach enables a fast, specific, and sensitive identification of endogenous peptides. Three different sequence collections were constituted in this study to mimic the peptidomic samples: SwePep precursors, SwePep peptides, and SwePep predicted. The searches for neuropeptides performed against these three sequence collections were compared with searches performed against the entire mouse proteome, which is commonly used to identify neuropeptides. These four sequence collections were searched with both Mascot and X! Tandem. Evaluation of the sequence collections was achieved using a set of manually identified and previously verified peptides. By using the three new sequence collections, which more accurately mimic the sample, 3 times as many peptides were significantly identified, with a false-positive rate below 1%, in comparison with the mouse proteome. The new sequence collections were also used to identify previously uncharacterized peptides from brain tissue; 27 previously uncharacterized peptides and potentially bioactive neuropeptides were identified. These novel peptides are cleaved from the peptide precursors at sites that are characteristic for prohormone convertases, and some of them have post-translational modifications that are characteristic for neuropeptides. The targeted protein sequence collections for different species are publicly available for download from SwePep. A new approach using targeted sequence collections has been developed for identifying endogenous peptides. This approach enables a fast, specific, and sensitive identification of endogenous peptides. Three different sequence collections were constituted in this study to mimic the peptidomic samples: SwePep precursors, SwePep peptides, and SwePep predicted. The searches for neuropeptides performed against these three sequence collections were compared with searches performed against the entire mouse proteome, which is commonly used to identify neuropeptides. These four sequence collections were searched with both Mascot and X! Tandem. Evaluation of the sequence collections was achieved using a set of manually identified and previously verified peptides. By using the three new sequence collections, which more accurately mimic the sample, 3 times as many peptides were significantly identified, with a false-positive rate below 1%, in comparison with the mouse proteome. The new sequence collections were also used to identify previously uncharacterized peptides from brain tissue; 27 previously uncharacterized peptides and potentially bioactive neuropeptides were identified. These novel peptides are cleaved from the peptide precursors at sites that are characteristic for prohormone convertases, and some of them have post-translational modifications that are characteristic for neuropeptides. The targeted protein sequence collections for different species are publicly available for download from SwePep. Neuropeptidomics is the technology approach for detailed analysis of endogenous peptides from the brain and the central nervous system (1Svensson M. Sköld K. Nilsson A. Fälth M. Nydahl K. Svenningsson P. Andrén P. Neuropeptidomics: MS applied to the discovery of novel peptides from the brain.Anal. Chem. 2007; 79: 14-21Crossref Scopus (52) Google Scholar, 2Che F.Y. Lim J. Pan H. Biswas R. Fricker L.D. Quantitative neuropeptidomics of microwave-irradiated mouse brain and pituitary.Mol. Cell. Proteomics. 2005; 4: 1391-1405Abstract Full Text Full Text PDF PubMed Scopus (146) Google Scholar, 3Clynen E. Baggerman G. Veelaert D. Cerstiaens A. Van der Horst D. Harthoorn L. Derua R. Waelkens E. De Loof A. Schoofs L. Peptidomics of the pars intercerebralis-corpus cardiacum complex of the migratory locust, Locusta migratoria.Eur. J. Biochem. 2001; 268: 1929-1939Crossref PubMed Scopus (141) Google Scholar, 4Skold K. Svensson M. Kaplan A. Bjorkesten L. Astrom J. Andren P.E. A neuroproteomic approach to targeting neuropeptides in the brain.Proteomics. 2002; 2: 447-454Crossref PubMed Scopus (104) Google Scholar, 5Svensson M. Skold K. Svenningsson P. Andren P.E. Peptidomics-based discovery of novel neuropeptides.J. Proteome Res. 2003; 2: 213-219Crossref PubMed Scopus (202) Google Scholar, 6Verhaert P. Uttenweiler-Joseph S. de Vries M. Loboda A. Ens W. Standing K.G. Matrix-assisted laser desorption/ionization quadrupole time-of-flight mass spectrometry: an elegant tool for peptidomics.Proteomics. 2001; 1: 118-131Crossref PubMed Scopus (111) Google Scholar). In contrast to proteomics, which is focused on studying proteins (>10 kDa) and their interactions, peptidomics is focused on studying endogenous peptides (<10 kDa), such as peptide hormones and neuropeptides. Neuropeptides are involved in many physiological processes including pain, hunger, and growth (7Hokfelt T. Broberger C. Xu Z.Q. Sergeyev V. Ubink R. Diez M. Neuropeptides—an overview.Neuropharmacology. 2000; 39: 1337-1356Crossref PubMed Scopus (471) Google Scholar). They often function as messengers, and some of them coexist with and complement the classical neurotransmitters (8Hokfelt T. Millhorn D. Seroogy K. Tsuruo Y. Ceccatelli S. Lindh B. Meister B. Melander T. Schalling M. Bartfai T. Terenius L. Coexistence of peptides with classical neurotransmitters.Experientia. 1987; 43: 768-780Crossref PubMed Scopus (305) Google Scholar).MS is a powerful tool utilized for thorough analytical profiling of a large number of neuropeptides (5Svensson M. Skold K. Svenningsson P. Andren P.E. Peptidomics-based discovery of novel neuropeptides.J. Proteome Res. 2003; 2: 213-219Crossref PubMed Scopus (202) Google Scholar). The MS methodology in combination with either ESI (9Fenn J.B. Mann M. Meng C.K. Wong S.F. Whitehouse C.M. Electrospray ionization for mass spectrometry of large biomolecules.Science. 1989; 246: 64-71Crossref PubMed Scopus (6273) Google Scholar, 10Yamashita M. Fenn J.B. Electrospray ion source. Another variation on the free-jet theme.J. Phys. Chem. 1984; 88: 4451-4459Crossref Scopus (1759) Google Scholar) or MALDI (11Karas M. Hillenkamp F. Laser desorption ionization of proteins with molecular masses exceeding 10,000 daltons.Anal. Chem. 1988; 60: 2299-2301Crossref PubMed Scopus (4792) Google Scholar) permits sensitive detection of peptide changes in complex mixtures of hundreds of different peptides simultaneously (5Svensson M. Skold K. Svenningsson P. Andren P.E. Peptidomics-based discovery of novel neuropeptides.J. Proteome Res. 2003; 2: 213-219Crossref PubMed Scopus (202) Google Scholar). The resolution and specificity of a neuropeptide analysis is further enhanced by coupling MS to LC or other high resolution separation techniques.Neuropeptidomics MS experiments, aimed at understanding the healthy and diseased mammalian brain, generate a large amount of data. To efficiently analyze these large datasets, reliable tools for automatic identification are needed. Such tools should be fast, yield few false peptide identifications (false positives), and leave few correct peptides unidentified (false negatives). So far, the main focus of the proteomics field has been on developing tools for identification of proteins, which are typically digested with trypsin, i.e. an enzyme with high specificity (12Olsen J.V. Ong S.E. Mann M. Trypsin cleaves exclusively C-terminal to arginine and lysine residues.Mol. Cell. Proteomics. 2004; 3: 608-614Abstract Full Text Full Text PDF PubMed Scopus (856) Google Scholar), limiting the search space of possible peptides. In contrast, endogenous peptide precursors are often processed by several enzymes (13Steiner D.F. The proprotein convertases.Curr. Opin. Chem. Biol. 1998; 2: 31-39Crossref PubMed Scopus (577) Google Scholar), and some of these have unknown specificity, making it difficult to accurately predict the sequence of mature endogenous peptides. Therefore, when searching for endogenous peptides, the entire proteome is often cleaved assuming an enzyme with no specificity (i.e. cleaving between any pair of amino acids). This creates a very large search space and yields poor results because only peptides that have strong experimental support can be identified. In a typical peptidomics experiment many hundreds of peptides are detected (5Svensson M. Skold K. Svenningsson P. Andren P.E. Peptidomics-based discovery of novel neuropeptides.J. Proteome Res. 2003; 2: 213-219Crossref PubMed Scopus (202) Google Scholar), but about an order of magnitude less are identified confidently.Many bioactive endogenous peptides are post-translationally modified, and it is common that a peptide contains more than one modification, further complicating the identification process. Important peptide modifications include acetylation, amidation, phosphorylation, and sulfation (7Hokfelt T. Broberger C. Xu Z.Q. Sergeyev V. Ubink R. Diez M. Neuropeptides—an overview.Neuropharmacology. 2000; 39: 1337-1356Crossref PubMed Scopus (471) Google Scholar). Approximately 300 different modifications have so far been reported for proteins (14Jensen O.N. Modification-specific proteomics: characterization of post-translational modifications by mass spectrometry.Curr. Opin. Chem. Biol. 2004; 8: 33-41Crossref PubMed Scopus (472) Google Scholar, 15Jensen O.N. Interpreting the protein language using proteomics.Nat. Rev. 2006; 7: 391-403Crossref Scopus (385) Google Scholar, 16Larsen M.R. Trelle M.B. Thingholm T.E. Jensen O.N. Analysis of posttranslational modifications of proteins by tandem mass spectrometry.BioTechniques. 2006; 40: 790-798Crossref PubMed Scopus (155) Google Scholar, 17Mann M. Jensen O.N. Proteomic analysis of post-translational modifications.Nat. Biotechnol. 2003; 21: 255-261Crossref PubMed Scopus (1597) Google Scholar). For example, 30% of the mammalian proteins are believed to be phosphorylated at one time or another (18Mann M. Ong S.E. Gronborg M. Steen H. Jensen O.N. Pandey A. Analysis of protein phosphorylation using mass spectrometry: deciphering the phosphoproteome.Trends Biotechnol. 2002; 20: 261-268Abstract Full Text Full Text PDF PubMed Scopus (789) Google Scholar). The C-terminal amidation, a common neuropeptide modification, seems to modify 50% of all bioactive peptides (19Eipper B.A. Milgram S.L. Husten E.J. Yun H.Y. Mains R.E. Peptidylglycine α-amidating monooxygenase: a multifunctional protein with catalytic, processing, and routing domains.Protein Sci. 1993; 2: 489-497Crossref PubMed Scopus (227) Google Scholar, 20Fricker L.D. Neuropeptide-processing enzymes: applications for drug discovery.AAPS J. 2005; 7: E449-E455Crossref PubMed Scopus (86) Google Scholar). Briefly the unknown specificity of the processing enzymes and the numbers of possible modifications make the identification of endogenous peptides difficult. Another difficulty stems from the less informative and inadequately understood fragmentation patterns for endogenous peptides compared with that of tryptic peptides.The aim of this study was to investigate how to optimize the identification process for endogenous peptides analyzed by tandem mass spectrometry by improving the sequence collections used by the search engines. During this study, several previously uncharacterized peptides were discovered from mouse brain tissue. Some of these peptides are potential novel neuropeptides as they are processed from proteins, known to contain neuropeptides, at sites that are characteristic for neuropeptides. Identifying novel neuropeptides is important for the understanding of the biochemical processes in the mammalian brain. This study demonstrates the importance of using optimized sequence collections when identifying endogenous peptides.EXPERIMENTAL PROCEDURESSequence CollectionsSwePep is a database constructed for endogenous peptides and mass spectrometry (21Falth M. Skold K. Norrman M. Svensson M. Fenyo D. Andren P.E. SwePep, a database designed for endogenous peptides and mass spectrometry.Mol. Cell. Proteomics. 2006; 5: 998-1005Abstract Full Text Full Text PDF PubMed Scopus (114) Google Scholar). This is a relatively new database specifically designed to speed up the identification process of endogenous peptides from complex tissue samples utilizing mass spectrometry. To create sequence collections that mimic the mouse peptidome rather than the mouse proteome, sequence information about peptides and their precursors were extracted from SwePep (updated February 15, 2006, containing 4,180 non-redundant peptide sequences). Four sequence collections were used in this study: 1) SwePep precursors, 2) SwePep peptides, 3) SwePep predicted, and 4) mouse proteome. These sequence collections are available for download from www.swepep.org.SwePep Precursor—The SwePep precursor sequence collection includes the sequences from the mouse peptide precursor proteins annotated in SwePep. Many precursor proteins, such as pro-opiomelanocortin, contain several known endogenous peptides (22Wilkinson C.W. Roles of acetylation and other post-translational modifications in melanocortin function and interactions with endorphins.Peptides. 2006; 27: 453-471Crossref PubMed Scopus (48) Google Scholar) and a number of possible cleavage sites for endogenous peptides. Therefore this sequence collection should contain many of the endogenous peptides despite its moderate size of 123 protein sequences with a total number of 23,601 amino acid residues. Using unspecific cleavage and a maximum peptide length of 50 amino acid residues 4,406,615 peptides were derived from this sequence collection.SwePep Peptides—The SwePep peptide sequence collection contains the sequences of the endogenous peptides annotated in SwePep from Mus musculus. It is constituted of 245 sequences and 6,776 amino acid residues. When using unspecific cleavage and a maximum peptide length of 50 amino acid residues this sequence collection generates 1,142,680 peptides.SwePep Predicted—Endogenous neuropeptides are processed in many steps to become active peptides. Predominantly they are cleaved from their precursor at the C terminus of two basic amino acids, separated by 0, 2, 4, or 6 other residues, by endopeptidases such as prohormone convertase 1 (PC1/3) 1The abbreviation used is: PC, prohormone convertase. and PC2 (13Steiner D.F. The proprotein convertases.Curr. Opin. Chem. Biol. 1998; 2: 31-39Crossref PubMed Scopus (577) Google Scholar, 23Seidah N.G. Chretien M. Eukaryotic protein processing: endoproteolysis of precursor proteins.Curr. Opin. Biotechnol. 1997; 8: 602-607Crossref PubMed Scopus (240) Google Scholar). The basic residues at the C terminus are then removed by carboxypeptidase E (24Zhou A. Webb G. Zhu X. Steiner D.F. Proteolytic processing in the secretory pathway.J. Biol. Chem. 1999; 274: 20745-20748Abstract Full Text Full Text PDF PubMed Scopus (409) Google Scholar). In the last step, the peptide may be modified. Important modifications on neuropeptides include C-terminal amidation and N-terminal acetylation (7Hokfelt T. Broberger C. Xu Z.Q. Sergeyev V. Ubink R. Diez M. Neuropeptides—an overview.Neuropharmacology. 2000; 39: 1337-1356Crossref PubMed Scopus (471) Google Scholar).By using the existing neuropeptide processing knowledge, possible peptide sequences were predicted from the mouse proteome (International Protein Index (IPI) mouse version 3.15, www.ebi.ac.uk/IPI/IPImouse.html) according to the following template: (K/R)Xm(K/R)↓Xk(K/R)Xn(K/R)↓ where m and n = 0, 2, 4, 6, X is any amino acid, and k = 3–50. Residues in bold signify amino acids that are not part of the final (detected) sequence. The C-terminal basic residues (Xk↓(K/R)Xn(K/R)) were removed, and the sequences Xk were stored in the SwePep predicted sequence collection.It is possible to define digestion rules for the search engines so that the theoretical digest of the proteome is performed at dibasic sites on the fly, but the SwePep predicted sequence collection speeds up the search, and it can be curated to include special cases and peptides from more than one type of cleavage.The SwePep predicted sequence collection was developed as a complement to the SwePep precursor and SwePep peptide sequence collections for identification of uncharacterized peptides and peptides from precursors not known to contain endogenous peptides. Peptides identified from the SwePep predicted sequence collection are likely to be biologically active because this collection only contains peptide sequences that have the specific cleavage pattern for neuropeptides. The SwePep predicted collection is constituted of precleaved sequences, and the searches are performed without any cleavage, i.e. the tandem mass spectra are directly matched against the sequences in the sequence collection. There are 3,413,034 predicted peptide sequences with 83,182,326 amino acid residues in this sequence collection. When using X! Tandem and its refinement function (25Craig R. Beavis R.C. A method for reducing the time required to match protein sequences with tandem mass spectra.Rapid Commun. Mass Spectrom. 2003; 17: 2310-2316Crossref PubMed Scopus (395) Google Scholar) this sequence collection generates 15,499,268 peptides with a maximum peptide length of 50 amino acid residues.Mouse Proteome—To compare this new identification approach with the commonly used identification approach, a sequence collection constituted of the whole mouse proteome (IPI mouse version 3.15, www.ebi.ac.uk/IPI/IPImouse.html) was searched using unspecific cleavage. The sequence collection of the mouse proteome consists of 68,222 protein sequences with a total number of 27,668,712 amino acid residues. When using unspecific cleavage and a maximum peptide length of 50 amino acid residues 250,809,615 peptides are generated.Search EnginesThis study was performed using two different search engines, X! Tandem (26Craig R. Beavis R.C. TANDEM: matching proteins with tandem mass spectra.Bioinformatics (Oxf.). 2004; 20: 1466-1467Crossref PubMed Scopus (1967) Google Scholar) and Mascot (27Perkins 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: 3551-3567Crossref PubMed Scopus (6711) Google Scholar), for searching the four sequence collection s described above.Search parameters were as follows. The SwePep peptides sequence collection, the SwePep precursor sequence collection, and mouse proteome sequence collection were searched using unspecific cleavage, and the precleaved SwePep predicted sequence collection was searched using no cleavage. The databases were searched using a peptide mass tolerance of ±2 Da and a fragment mass tolerance of ±0.7 Da. The first dataset was searched with a number of possible post-translational modifications (N-terminal acetylation, N-terminal pyroglutamic acid of glutamine, C-terminal amidation, deamidation of asparagine and glutamine, and oxidation of methionine). A full specification of search parameters is presented in the supplemental data. For X! Tandem the refinement function was used to allow unspecific cleavage of a precursor if one or more peptides have been identified from it (25Craig R. Beavis R.C. A method for reducing the time required to match protein sequences with tandem mass spectra.Rapid Commun. Mass Spectrom. 2003; 17: 2310-2316Crossref PubMed Scopus (395) Google Scholar).Mass Spectrometry DatasetsTwo different MS datasets were used for searching the sequence collections. One set contained 86 tandem mass spectra with manually identified peptides in the mass range from 500 to 3500 Da and with charge states 1, 2, 3, or 4. All tandem mass spectra were manually evaluated, and the peptides were unambiguously identified. Because this dataset was manually composed of spectra with known identities it does not reflect a typical collection of tandem mass spectra from an LC-MS analysis of a peptidomic sample. Therefore, a second dataset was evaluated. This dataset was obtained by analyzing a peptidomic sample from mouse hypothalamus with nanoflow capillary LC-ESI-MS/MS and contained 2,867 tandem mass spectra.Sample Preparation and Mass Spectrometry Analysis—The brain tissue was suspended in cold extraction solution (0.25% acetic acid) and homogenized by microtip sonication (Vibra cell 750, Sonics & Materials Inc., Newtown, CT) to a concentration of 0.2 mg of tissue/μl as described previously (4Skold K. Svensson M. Kaplan A. Bjorkesten L. Astrom J. Andren P.E. A neuroproteomic approach to targeting neuropeptides in the brain.Proteomics. 2002; 2: 447-454Crossref PubMed Scopus (104) Google Scholar, 5Svensson M. Skold K. Svenningsson P. Andren P.E. Peptidomics-based discovery of novel neuropeptides.J. Proteome Res. 2003; 2: 213-219Crossref PubMed Scopus (202) Google Scholar). Briefly the suspension was centrifuged at 20,000 × g for 30 min at 4 °C. The protein- and peptide-containing supernatant was transferred to a centrifugal filter device (Microcon YM-10, Millipore, Bedford, MA) with a molecular mass limit of 10,000 Da and centrifuged at 14,000 × g for 45 min at 4 °C. Finally the peptide filtrate was frozen and stored at −80 °C until analysis.Five microliters of peptide filtrate (equivalent to 1.0 mg of brain tissue) was desalted on a nano-precolumn (LC Packings, Amsterdam, The Netherlands) at 10 μl/min using a nano-LC system (Ettan MDLC, GE Healthcare). The filtrate was then separated using a fused silica capillary column (75-μm inner diameter, 15-cm length, NAN75-15-03-C18PM; LC Packings) by an isocratic flow of buffer A (0.25% acetic acid in water) for 35 min and eluted during a 60-min gradient from buffer A to B (35% acetonitrile in 0.25% acetic acid). The eluted peptides were analyzed by a linear trap quadrupole ion trap mass spectrometer (Thermo Electron, San Jose, CA). The spray voltage was 1.8 kV, the capillary temperature was 160 °C, and 35 units of collision energy were used to obtain fragment spectra. Four MS/MS spectra of the most intense peaks were obtained following each full-scan mass spectrum (Xcalibur 1.4 SR1). The dynamic exclusion feature was enabled to obtain MS/MS spectra on co-eluting peptides. Raw linear trap quadrupole data were converted to dta files by Xcalibur 1.4 SR1 and assembled by an in-house developed script to Mascot generic files.Verification of important in the identification process is to the from the search engines. One to this is to the of false identifications J. Fenyo D. A for the of mass protein identification Chem. 2000; PubMed Scopus Google Scholar, D. Beavis R.C. A method for the of mass protein identifications using Chem. 2003; PubMed Scopus Google Scholar). This is often achieved by an or by searching a sequence collection, the number of a and this number by the number of from the targeted sequence collection search J. Evaluation of with tandem mass spectrometry for protein the Proteome Res. 2003; 2: PubMed Scopus Google Scholar, R.E. an for database search Mass Spectrom. 2002; PubMed Scopus Google Scholar). A commonly used for a protein to be identified is that at two peptides have to be identified from that protein with a In contrast, each processed peptide may have a and it is important to obtain high data to the the for endogenous peptide identification have to be more false-positive rate for the first dataset in this study, containing known peptide was by the number of false identified peptides with the number of For the second dataset the false-positive rate was by searching the sequence by the search was used as the first to the search All peptides with a below the were manually verified or to the the of the peptides were only the the second for each tandem mass spectrum was to the to that the first and second not have a that is to each i.e. the for the first and second were was used for the correct the a approach to and identify a large number of endogenous peptides. Many of the identified peptides previously uncharacterized and novel processed of protein To identify more of the peptide in mouse brain are both experimental and that have to be automatic identification process is for endogenous peptides that has high specificity and To the samples are searched against sequence collections specifically designed to mimic the of the The sample was performed to protein (4Skold K. Svensson M. Kaplan A. Bjorkesten L. Astrom J. Andren P.E. A neuroproteomic approach to targeting neuropeptides in the brain.Proteomics. 2002; 2: 447-454Crossref PubMed Scopus (104) Google Scholar, 5Svensson M. Skold K. Svenningsson P. Andren P.E. Peptidomics-based discovery of novel neuropeptides.J. Proteome Res. 2003; 2: 213-219Crossref PubMed Scopus (202) Google Scholar) by of the enzymes in the tissue (5Svensson M. Skold K. Svenningsson P. Andren P.E. Peptidomics-based discovery of novel neuropeptides.J. Proteome Res. 2003; 2: 213-219Crossref PubMed Scopus (202) Google Scholar, L. neuropeptidomics by and and extraction with Proteome Res. 2006; 5: PubMed Scopus Google Scholar) to the endogenous peptide of the sample. Because the sample contains endogenous peptides so should the sequence collections. In this study, the four different sequence collections were using two search engines, Mascot and X! Tandem. The in-house constituted sequence collections were by searching two different datasets against Peptides 1 constituted 86 tandem mass spectra. 1 the search obtained by searching 1 against the different sequences collections. The numbers of identified peptides were obtained when searching the SwePep peptide sequence collection. The number of peptides were identified searching the mouse proteome. This is to the that the for a identification with the size of the sequence collection because the that the is J. Fenyo D. The of protein identification results as a function of the number of protein sequences Proteome Res. 2004; 3: PubMed Scopus Google Scholar). For example, the for a identification in Mascot from to when from the to the sequence collections false-positive the of the as a function of the of the number of peptide sequences in the sequence collection for a few peptides identified by X! Tandem. This that the number of false as the size of the sequence collection For 1, the only false was by Mascot when the SwePep predicted sequence collection was When searching these more targeted sequences collections, 3 times as many peptides were identified compared with searching the mouse proteome. The search of the mouse proteome not any identities that were not identified when searching the more targeted sequence collections. Another with searching large sequence collections using unspecific cleavage is that it is if the search includes a number of different post-translational of the as a function of the of the number of peptide sequences in the sequence collection for two of the peptides identified by X! Tandem. To a as a the of the should be less than all of the tandem mass spectra in 1 were identified in this This on the that many of the for identifying peptides and proteins are designed for tryptic peptides and that the fragmentation pattern of endogenous peptides from the fragmentation pattern of peptides digested with a