The Implications of Proteolytic Background for Shotgun Proteomics

The analysis by liquid chromatography coupled to tandem mass spectrometry of complex peptide mixtures, generated by proteolysis of protein samples, is the main proteomics method used today. The approach is based on the assumption that each protein present in a sample reproducibly and predictably generates a relatively small number of peptides that can be identified by mass spectrometry. In this study this assumption was examined by a targeted peptide sequencing strategy using inclusion lists to trigger peptide fragmentation attempts. It was found that the number of peptides observed from a single protein is at least one order of magnitude greater than previously assumed. This unexpected complexity of proteomics samples implies substantial technical challenges, explains some perplexing results in the proteomics literature, and prompts the need for developing alternative experimental strategies for the rapid and comprehensive analysis of proteomes. The analysis by liquid chromatography coupled to tandem mass spectrometry of complex peptide mixtures, generated by proteolysis of protein samples, is the main proteomics method used today. The approach is based on the assumption that each protein present in a sample reproducibly and predictably generates a relatively small number of peptides that can be identified by mass spectrometry. In this study this assumption was examined by a targeted peptide sequencing strategy using inclusion lists to trigger peptide fragmentation attempts. It was found that the number of peptides observed from a single protein is at least one order of magnitude greater than previously assumed. This unexpected complexity of proteomics samples implies substantial technical challenges, explains some perplexing results in the proteomics literature, and prompts the need for developing alternative experimental strategies for the rapid and comprehensive analysis of proteomes. Proteomics aims at analyzing in one single experiment the complete proteome of a cell type, a tissue, or a species. The studies in which the proteins constituting a proteome are identified and quantified are particularly informative and relevant from a biological and biomedical point of view. At present this is most frequently attempted by a shotgun, tandem mass spectrometry-based strategy (1Washburn M.P. Wolters D. Yates III, J.R. Large-scale analysis of the yeast proteome by multidimensional protein identification technology.Nat. Biotechnol. 2001; 19: 242-247Crossref PubMed Scopus (4029) Google Scholar). Although various implementations of this strategy differ in the specifics, they all share the same operating steps (2Aebersold R. Mann M. Mass spectrometry-based proteomics.Nature. 2003; 422: 198-207Crossref PubMed Scopus (5484) Google Scholar): the proteins present in a sample are initially digested in solution with a highly specific protease, typically trypsin. The resulting peptide mixtures are subjected to one-, two-, or three-dimensional fractionation, and the peptides eluting from the last separation step, typically reverse-phase chromatography, are analyzed by MS/MS. Lastly the MS/MS spectra collected are assigned to peptide sequences using a suite of software tools (3Sadygov R.G. Cociorva D. Yates III, J.R. Large-scale database searching using tandem mass spectra: looking up the answer in the back of the book.Nat. Methods. 2004; 1: 195-202Crossref PubMed Scopus (327) Google Scholar, 4Nesvizhskii A.I. Protein identification by tandem mass spectrometry and sequence database searching.Methods Mol. Biol. 2006; 367: 87-120Google Scholar). Shotgun proteomics is founded on the assumption that each protein present in a sample reproducibly generates a relatively small number of peptides, the boundaries of which conform to the cleavage specificity of the protease used. Trypsin cleaves at the C termini of arginine and lysine residues, and based on the occurrence of these two amino acids in proteins, an average of 10 peptides is expected for a stretch of a hundred residues. The validity of this assumption is critical to the success of proteomics experiments for several reasons. First, the number of peptides present in a protein digest determines the number of MS/MS cycles minimally required to fully analyze the sample. Because the MS-MS/MS duty cycle is given for a particular mass spectrometer (better than 1 Hz for modern instruments), in theory the number of peptides to be analyzed in the sample relates to the minimal duration of a proteomics experiment. Second, the sample complexity determines the practical dynamic range of proteome analyses. The nominal dynamic range of a mass spectrometer (at best 3–4 orders of magnitude) can in practice be significantly reduced by the fact that automated precursor ion selection (data-dependent acquisition (DDA) 1The abbreviation used is: DDA, data-dependent acquisition. primarily focuses on the most intense MS signals. In the case of very complex samples, only the highest intensity ions are fragmented, whereas ions of lower intensity pass through the system unselected even though their signal is well within the nominal dynamic range of the instrument. Third, database searches are often constrained to full tryptic peptides to restrict the searching time. Unspecific cleavages will thus produce peptides not anticipated by the search parameters, leading to misassignments or missed identifications. Despite the development of peptide separation systems with higher peak capacity (5Wolters D.A. Washburn M.P. Yates III, J.R. An automated multidimensional protein identification technology for shotgun proteomics.Anal. Chem. 2001; 73: 5683-5690Crossref PubMed Scopus (1544) Google Scholar, 6Plumb R.S. Rainville P. Smith B.W. Johnson K.A. Castro-Perez J. Wilson I.D. Nicholson J.K. Generation of ultrahigh peak capacity LC separations via elevated temperatures and high linear mobile-phase velocities.Anal. Chem. 2006; 78: 7278-7283Crossref PubMed Scopus (74) Google Scholar, 7Luo Q. Shen Y. Hixson K.K. Zhao R. Yang F. Moore R.J. Mottaz H.M. Smith R.D. Preparation of 20-μm-i.d. silica-based monolithic columns and their performance for proteomics analyses.Anal. Chem. 2005; 77: 5028-5035Crossref PubMed Scopus (149) Google Scholar) and of tandem mass spectrometers with faster acquisition rate, the proteome of any species has yet to be fully mapped. The complete analysis of even moderately complex samples such as isolated organelles (8Andersen J.S. Lam Y.W. Leung A.K. Ong S.E. Lyon C.E. Lamond A.I. Mann M. Nucleolar proteome dynamics.Nature. 2005; 433: 77-83Crossref PubMed Scopus (927) Google Scholar) or macromolecular complexes (9Ranish J.A. Yi E.C. Leslie D.M. Purvine S.O. Goodlett D.R. Eng J. Aebersold R. The study of macromolecular complexes by quantitative proteomics.Nat. Genet. 2003; 33: 349-355Crossref PubMed Scopus (310) Google Scholar) has required enormous efforts. All these considerations suggest that the comprehensive analysis of a complex sample is more difficult than initially anticipated, and one of the reasons for this could be a degree of complexity resulting from proteolysis of the protein sample that is higher than expected. To test this hypothesis and to assess the number of peptides actually generated by proteolysis of a protein, an in-depth characterization of the products of tryptic digestion of well defined proteins was carried out. Five pure bovine standard proteins, β-lactoglobulin, carbonic anhydrase, serum albumin, transferrin, and β-casein, were subjected individually to tryptic digestion, and the resulting peptide mixtures were extensively characterized by LC/MS/MS. To maximize the number of peptides identified in the proteolytic digests, a targeted MS/MS sequencing strategy (10Domon B. Broder S. Implications of new proteomics strategies for biology and medicine.J. Proteome Res. 2004; 3: 253-260Crossref PubMed Scopus (60) Google Scholar, 11Domon B. Picotti P. Ossola R. Stahl-Zeng J. Aebersold R. Identification and quantification of biomarkers in serum samples.in: Proceedings of the 53rd ASMS Conference on Mass Spectrometry, San Antonio, Texas, June 5–9, 2005. American Society for Mass Spectrometry, Santa Fe, NM2005Google Scholar, 12Picotti P. Lee H. Domon B. Aebersold R. Novel approach to identify low abundance biomarkers in serum.in: Proceedings of the 54th ASMS Conference on Mass Spectrometry, Seattle, Washington, May 28–June 1, 2006. American Society for Mass Spectrometry, Santa Fe, NM2006Google Scholar) was applied and was compared with the intensity-driven data-dependent acquisition method. The targeted approach is based on inclusion lists of precursor ions to trigger collision-induced dissociation. In this approach, the samples were initially analyzed in a high accuracy mass spectrometer in full scan mode. Data were then extensively processed off-line to extract and inventory monoisotopic ions of all the peptide ions observed. The samples were then subjected to MS/MS sequencing multiple times using inclusion lists to trigger fragmentation of the ions of interest, present in full MS scan regardless of their intensity, with retention time and charge state as the only constraints. The approach was shown to be a robust and effective means to sequence low abundance ions and revealed a number of peptides produced from proteolysis of a protein that is at least 10 times higher then previously assumed. Porcine trypsin (modified, sequencing grade) was purchased from Promega (Madison, WI). The standard bovine proteins β-lactoglobulin, β-casein, serum albumin, transferrin, and carbonic anhydrase and DTT were obtained from Sigma. Recombinant His-tagged human Pax-8 protein was overexpressed in Escherichia coli (BL21–3DE, Stratagene, Heidelberg, Germany) and purified by means of a HiTrap chelating nickel column (GE Healthcare). Tris(2-carboxyethyl)phosphine and iodoacetamide were purchased from Fluka (Buchs, Switzerland). HPLC-grade water and acetonitrile were purchased from Riedel-de Haën (Seelze, Germany). Each protein was solubilized in 0.1 m ammonium bicarbonate buffer containing 8 m urea at a final concentration of 3 mg/ml. After tris(2-carboxyethyl)phosphine reduction and iodoacetamide alkylation of cysteine residues, the solution was diluted to final 1 m urea, the pH was adjusted to 8.0, and the proteins were digested with trypsin at 37 °C. The digestion was repeated on the same set of proteins, after HPLC purification, to eliminate protein degradation products and other contaminants potentially present in the commercial protein preparations. Briefly proteins were loaded onto a macroporous reverse-phase C18 column (mRP-C18, 4.6 × 50 mm, Agilent Technologies, Waldbronn, Germany). Elution was carried out with a linear gradient of containing 0.1 and from to in at a of The was by at and containing the protein species were collected at the peak and then subjected to trypsin digestion with the range of to to and times to were used for each protein In the case of β-lactoglobulin, digestion was with trypsin on with trypsin from and with the from The digestion was by with to a final pH of The peptide mixtures were by or with a based on ammonium bicarbonate buffer by columns to trypsin was used. peptide samples were on a to in and samples were resulting from the same of the protein were analyzed on a mass spectrometer San with a ion separations of peptides were on an Agilent system Germany) with a with a C18 were loaded on the column from a Agilent and with a linear gradient of containing at a of gradient from to acetonitrile in 50 was used. each peptide sample a standard on the most intense ions MS scan was MS/MS spectra were in the linear ion each the at nominal resulting in an cycle time of state was fragmentation of and ions and ions of charge of ion was set to trigger an MS/MS of all the was on the resulting from the analysis of single protein tryptic to a of intensity of of The ions were an inclusion for The were by intensity and multiple lists containing up to ions Each digest was inclusion selection of precursor ions for MS/MS a number of times to the number of of peptide such as charge state and retention were used as for MS/MS attempts. low intensity the acquisition time was to the of MS/MS spectra to MS/MS up to MS/MS were the bovine for database using and trypsin were to the bovine protein ion was set to and was set to MS/MS on the tryptic digest of human Pax-8 were the human Data were of as a with of cysteine as a In the standard two missed cleavages and one were Data were with the sequence of the protein of Each peptide was subjected to and only high were were for as a at the or or residues, with of cysteine as a in this case only full tryptic cleavage and two missed each peptide the in the experimental and the monoisotopic was The high accuracy of was used as an and To assess the number of peptides generated by proteolysis of a protein, an in-depth characterization of the products of tryptic digestion of well characterized Five pure bovine standard proteins, β-lactoglobulin, carbonic anhydrase, serum albumin, transferrin, and by were subjected individually to tryptic digestion, and the resulting peptide mixtures were extensively characterized by LC/MS/MS. To a of the of the proteolysis by specific experimental the were repeated using in which critical such as the to the of commercial digestion, and peptide were was repeated on the same set of proteins subjected to an step, by reverse-phase HPLC or the to eliminate low abundance degradation products potentially present in the commercial protein preparations. digestion was on the human Pax-8 protein in coli to the of multiple in the To maximize the number of peptides identified in the proteolytic digests, a targeted MS/MS sequencing strategy was applied and compared with the The targeted approach is based on an inclusion of precursor ions that are within a and trigger a collision-induced dissociation. The method is in the sample of is analyzed in in a high mass accuracy mass the analysis is to the ion reproducibly to the sample. The are processed and analyzed in to extract all observed monoisotopic ions to peptides with the charge retention and The sample is analyzed in MS/MS times using inclusion lists containing the ions of to trigger collision-induced with retention time and charge state as constraints. time and are adjusted to the of the and of inclusion lists are based on the acquisition software inclusion capacity and on the sequencing ion spectra by targeted sequencing are then assigned to peptide sequences by database and the are Because most of the of database search to peptide sequences with tryptic tryptic digestion products be in the database search the for a more in this the of was not used to and lower Each peptide was subjected to and only high were In this MS/MS with a sequence that not using were to the were by a in the precursor ion mass lower than using the accuracy of mass as an of peptide in the with a low are in 3 the results of this of analysis for a bovine tryptic is a small protein, of amino with a of In digestion fully tryptic peptides containing more than amino peptides were identified using the data-dependent analysis with automated selection of highly intense precursor ions for the targeted peptide sequencing approach peptide of which not observed by using the automated method. the peptides were from the of bovine β-lactoglobulin, whereas sequences to of and In peptides were trypsin products or of of the fully tryptic peptides of of were identified by and targeted The approach an capacity of low abundance digestion peptides containing missed amino and a number of sequences containing one 3 the of tryptic sequences with the cleavage at the This is to the fact that peptides a highly at the C high and were more is the of peptides with times This that the various of a peptide were actually present in the sample to separation and were not produced to ion of a of observed of the same tryptic The used in 3 of the set of peptides from observed in analyses. Although in number than of peptide most of the tryptic peptides to low abundance signals. they to than of the ion only the peptides The of and fully tryptic peptides identified the set of protein the this was not significantly by the digestion the only a small in the occurrence of tryptic peptides was observed at higher to compared with the number of peptides identified protein with the mass of the in the proteolysis was observed digestion was on proteins purified by reverse-phase HPLC or as well as on the protein Pax-8 in of occurrence of were obtained was used as an This the in-depth characterization of the of tryptic in of expected products and digestion The of a targeted peptide sequencing strategy a of the number of peptide compared with In particular sequencing of low abundance peptides, which a low of for collision-induced in an automated mode. This revealed an unexpected complexity of tryptic given by a substantial number of peptides with a The of trypsin specificity was by Mann and Ong S.E. Mann M. Trypsin cleaves to arginine and lysine 2004; 3: PubMed Scopus Google Scholar) in an analysis of a proteome The study a of products identified in standard proteomics experiments on complex the sample used in that study was and was analyzed by automated selection of precursor thus and the high intensity the of specificity of trypsin for and is with the in this study in as as the high intensity precursor ions are The present that the low intensity precursor ions that very were not analyzed in the study Ong S.E. Mann M. Trypsin cleaves to arginine and lysine 2004; 3: PubMed Scopus Google Scholar) are for peptides with tryptic or a of trypsin specificity has previously and in the B. of Scholar, B. Data Scholar) database by on proteolytic cleavages the for a on the reasons for trypsin tryptic peptides in the proteomics to as low abundance of tryptic In comprehensive studies the of unexpected species in tryptic digests, and peptides R.J. M. analysis of a multidimensional liquid chromatography mass spectrometry on a mass of the is by search 2005; PubMed Scopus Google Scholar). In full trypsin specificity was as a database search the identification of tryptic the of tryptic peptides, even as search be by database search or of which to full tryptic peptides, a higher the for a on analysis these reasons peptides with tryptic or are in the proteomics In this the number of peptides identified in the digest using a targeted sequencing method was one order of magnitude greater than The complexity of such be even greater one that the peptides identified were only present in to be by the mass spectrometer and within the dynamic range of the peptides could potentially be found in the by searching MS/MS single or by using sequencing tools F. sequencing Proteome Res. 2005; PubMed Scopus Google Scholar). a range of or as a of sample D. S. Identification of by search of mass Biotechnol. 2005; PubMed Scopus Google the complexity of tryptic is to an a database search was on the MS/MS of the bovine digest in which peptide (at the or or was on fully tryptic peptides peptides were identified as a of the with from urea This number is to cleavage is for the used shotgun strategy and for proteomics strategies based on such as the mass and time method R.D. Shen Y. An mass strategy for quantitative and proteome PubMed Scopus Google Scholar). The approach on mass and retention times to identify peptides as a for MS/MS analysis that the number of peptides of a given mass within a of a is at least times higher peptides are compared with fully tryptic peptides The results of this not only that based on trypsin specificity the sample complexity that the on which the shotgun and mass strategies are founded are peptides produced of single proteins were to be of relatively low abundance of magnitude in the digestion compared with the specific Although this is not a for the analysis of single proteins or mixtures, for a complex biological an number of low abundance peptides is expected to a proteolytic in mass the range of protein in biological samples The human and 1: PubMed Scopus Google the proteolytic resulting from will to the signal of tryptic peptides from lower abundance This hypothesis was by analyzing a of a human serum digest for H. Aebersold R. Identification and quantification of using and mass Biotechnol. 2003; PubMed Scopus Google Scholar) in which the digest previously characterized was in at the concentration of by the tryptic peptides to be present in the sample from the analysis were not in mass whereas several peptides from highly serum proteins were In to the abundance of tryptic peptides in proteomics a proteomics was with to the yeast and F. A.I. P. Eng J. S. J. Aebersold R. The Res. 2006; PubMed Scopus Google Scholar, Eng J.K. H. A.I. B. Lee H. Yi E.C. Ossola R. Aebersold R. 2005; PubMed Scopus Google Scholar, J.A. A.I. J.S. P. Eng J. F. M. B. Lee H. B. Aebersold R. of the proteome with Biol. 2006; PubMed Scopus Google of the number of peptides and for yeast and to tryptic The higher obtained for serum the of a high number of serum with and the number of spectra in the are high that database search and systems to and unexpected digestion This is by the fact that is a of proteomics experiments in with a highly tryptic peptides will to be observed in samples and in the whereas each low abundance peptide is to a lower number of as in the shown in the same one a potentially high of peptide in in the database has to this by using and database searches F. A.I. P. Eng J. S. J. Aebersold R. The Res. 2006; PubMed Scopus Google Scholar, Eng J.K. H. A.I. B. Lee H. Yi E.C. Ossola R. Aebersold R. 2005; PubMed Scopus Google Scholar, J.A. A.I. J.S. P. Eng J. F. M. B. Lee H. B. Aebersold R. of the proteome with Biol. 2006; PubMed Scopus Google Scholar). This was by a of MS/MS spectra in to tryptic peptides, most of which were the in the database of such a high number of tryptic peptides is a of their in tryptic these that the with purified single proteins to the analysis of more complex protein This is a that the of shotgun proteomics to low abundance proteins in very complex biological the of MS/MS spectra collected in with M. R. H. R. R.J. J.S. Aebersold R. Eng J. H. J.S. P. J. J. R. Zhao R. M. P. H. of the Proteome results from the with and multiple a of proteins and a 2005; PubMed Scopus Google the comprehensive of complex such as the human serum proteome by shotgun approach a relatively small such as the yeast proteome not yet fully S. J.S. analysis of protein in 2003; PubMed Scopus Google Scholar). based on of highly proteins M. S. R. D. Recombinant for the of proteins from human 2006; PubMed Scopus Google Scholar) or at the protein or peptide J. Lee H. A.I. D. S. M. Aebersold R. peptide separation and identification for mass spectrometry based proteomics via Proteome Res. 2006; PubMed Scopus Google Scholar, M. M. P. D. Aebersold R. F. J.S. for tandem mass spectrometry shotgun proteomics through of Proteome Res. 2005; PubMed Scopus Google Scholar, M. P. D. F. J.S. protein by peptide and to proteome analysis of human 2005; PubMed Scopus Google Scholar, B. F. J.S. in proteome 2003; 3: PubMed Scopus Google Scholar) the identification of low abundance proteins not eliminate the of the shotgun of targeted sequencing in which the of peptide ions and their sequencing are and in two experiments as in this is a approach that more in-depth characterization of peptide mixtures to fully a complex proteome such as the human serum alternative strategies will to be for rapid and comprehensive proteome is to to a B. Aebersold R. Mass spectrometry and protein 2006; PubMed Scopus Google Scholar) the from proteomics Eng J.K. H. A.I. B. Lee H. Yi E.C. Ossola R. Aebersold R. 2005; PubMed Scopus Google Scholar, B. M. P. Aebersold R. with peptide Mol. Biol. 2005; PubMed Scopus Google or to for or proteins in a sample. will be used to peptides B. M. P. Aebersold R. with peptide Mol. Biol. 2005; PubMed Scopus Google Scholar, P. M. Lee H. D. J. B. R. B. Aebersold R. of peptides for quantitative proteomics.Nat. Biotechnol. PubMed Scopus Google Scholar) and their fragmentation to for of proteins by multiple mass multiple for 2006; PubMed Scopus Google Scholar, B. Stahl-Zeng J. Aebersold R. Novel strategy for rapid and quantification of biomarkers in serum.in: Proceedings of the 54th ASMS Conference on Mass Spectrometry, Seattle, Washington, May 28–June 1, 2006. American Society for Mass Spectrometry, Santa Fe, NM2006Google Scholar). In such on of ions specific to a peptide are and This is a very to in complex samples at a and specificity that is difficult to by shotgun proteomics and for and for an of the Pax-8 protein and Ossola for a serum peptide sample. P. P. for and with