Nature Methods 2, 955 - 957 (2005)
Published online: 18 November 2005; | doi:10.1038/nmeth811
Positional proteomics: selective recovery and analysis of N-terminal proteolytic peptidesLucy McDonald1, Duncan H L Robertson1, Jane L Hurst2
& Robert J Beynon11 Protein Function Group, Faculty of Veterinary Science, University of Liverpool, Crown Street, Liverpool L69 7ZJ, UK. 2 Mammalian Behavior and Evolution Group, Faculty of Veterinary Science, University of Liverpool, Crown Street, Liverpool L69 7ZJ, UK.
Correspondence should be addressed to Robert J Beynon r.beynon@liv.ac.uk Bottom-up proteomics is the analysis of peptides derived from single proteins or protein mixtures, and because each protein generates tens of peptides, there is scope for controlled reduction in complexity. We report here a new strategy for selective isolation of the N-terminal peptides of a protein mixture, yielding positionally defined peptides. The method is tolerant of several fragmentation methods, and the databases that must be searched are substantially less complex.
Bottom-up proteomics operates at the level proteolytic peptides, generated from single proteins or from complex mixtures of proteins1. These peptides, generated by exhaustive proteolysis to limit peptides in vitro, are then analyzed by various mass spectrometric methods. Mass spectrometric analysis yields either the masses of a formally connected set of peptides that were all derived from a single protein (peptide mass fingerprinting) or, by tandem mass spectrometry, sequence-derived information that can identify the parent protein of a single peptide2,
3. It can be argued that more peptides are analyzed than strictly necessary, and comprehensive proteomic analysis should focus on the minimal number of peptides that are required for protein identification. Methods such as ICAT4 implicitly adopt this principle, in as much as the selective chemistry recovers only those peptides that contain at least one cysteine residue. Cysteine-mediated peptide recovery, however, is likely to abstract more than one peptide for each protein, and it is not possible to target the recovered peptide(s) positionally, as cysteine residues can occur anywhere in the protein sequence.
Positionally defined peptides would yield a substantial information gain in protein identification strategies. Most obviously, the two positional locations within every protein are the extreme ends—the N-terminal and the C-terminal peptides. Methods for recovery of C-terminal peptides have been reported, predominantly based on the ability of a catalytically disabled trypsin, anhydrotrypsin, to selectively bind peptides that terminate in a lysine or arginine residue5,
6. There are several reports that indicate routes to selective recovery of N-terminal peptides, including specific N-terminal sequencing by mass spectrometry of gel-separated and blotted proteins7, selective modification of N-terminal serine or threonine residues8, modification of the hydrophobicity of a peptide mixture to preferentially expose N-terminal peptides by diagonal chromatography9,
10 and selective capture of all non–N-terminal peptides by amine scavenging beads11,
12. We report here a new approach to selective recovery of the N-terminal-most peptides of a complex protein mixture. The method, based on subtractive removal of internal peptides, is not reliant on any particular endopeptidase cleavage—a flexibility that can compensate for the limitations in N-terminal peptide size distributions. Moreover, in contrast to other approaches7,
11,
12, protein N termini that are naturally acetylated are automatically included in the analyte set. Indeed, if stable isotope–labeled acetic anhydride was used, it would be possible to identify and discriminate between naturally and chemically acetylated peptides. In brief, all available amino groups are blocked by acetylation. Subsequently, proteolysis generates new peptides, and all but the N-terminal peptide (whether naturally or artificially acetylated) expose a new amino group that is subsequently biotinylated. These biotinylated internal peptides are removed by recovery onto immobilized avidin or strepatividin, leaving behind the set of N-terminal peptides (Fig. 1 and Supplementary Methods online, which contains a protocol for N-terminal peptide recovery and a description of the analysis of protein databases).
 | | |
One of our major interests is in proteome dynamics in skeletal muscle13,
14,
15. The tryptic digest of the soluble protein fraction of mouse skeletal muscle contains peptides derived from a large number of proteins, and a matrix-assisted laser desorption/ionization–time-of-flight (MALDI-ToF) spectrum on an instrument of medium-level performance (resolution 12,000 FHWM (full width at half maximum)) yielded a detailed but complex mass spectrum (Fig. 2a). Owing to the complexity of the peptide mixture, we were unable to identify any N-terminal peptides in the spectrum. We passed the N-acetylated, trypsin-digested, biotinylated mixture over immobilized streptavidin. The unbound eluate gave a much simpler mass spectrum (Fig. 2b), and we were able to assign the highest intensity signals to true N-terminal peptides, confirmed by tandem mass spectrometry (Supplementary Fig. 1 online). To test the method with a more complex mixture, we applied the same protocol to the soluble proteins of mouse liver. After purification of N-terminal peptides, the MALDI-ToF spectrum remained complex. By liquid chromatography–tandem mass spectrometry, and even without optimized separation or mass spectrometric analysis, many peptides (over 90) could immediately be assigned as N termini of mouse proteins (Supplementary Fig. 2 and Supplementary Table 1 online). As predicted, all terminated at C-terminal arginine residues. Moreover, the data were consistent with known or inferred N-terminal processing (removal of initiator methionine, loss of signal peptide or propeptide) but in other cases have provided new information on N-terminal processing of liver proteins. All identifications were from a search of the entire database of mouse proteins rather than a restricted N-terminal database—there were virtually no peptides identified as internal sequences.
| | |
An analysis of extracted N-terminal peptides from mouse entries in Swissprot (Supplementary Fig. 3 online) confirmed that over 85% of all proteins yielded an informative N-terminal peptide with trypsin digest, and that this value rose to almost 90% if we used two endopeptidases (trypsin and endopeptidase Glu-C). Thus, a substantial fraction of proteins in a proteome can be uniquely identified simply by the mass of the N-terminal peptide, using one or multiple endopeptidase digests. But the complexity of most N-terminal peptide preparations would require liquid chromatography–electrospray tandem mass spectrometry or liquid chromatography–MALDI tandem mass spectrometry. Even partial sequence data considerably enhance identification, and remove the need for multiple proteolytic digests—the residual 4% unidentifiable proteins represent sequences in the database that are either replicated entries or which represent paralogous proteins. There has recently been discussion about the uncertainty of 'one hit wonders' in proteomics, and we conjecture that part of the uncertainty relates to the lack of information about the location of the peptide in the parent proteins. A positional proteomics strategy anchors the peptides at a precise location within the parent protein, greatly reducing the search space for identification algorithms. As an average protein might be predicted to yield 50 tryptic peptides, the approximate reduction in search space is also 50-fold. Further, selective isolation and partial sequencing of N- and C-terminal peptides would allow virtually full length PCR-amplification of the cDNA corresponding to an expressed protein sequence.
Note: Supplementary information is available on the Nature Methods website.
Received 23 August 2005; Accepted 3 October 2005; Published online: 18 November 2005.
REFERENCES
-
Bogdanov, B. & Smith, R.D. Mass Spectrom. Rev. 24, 168–200 (2004). | ISI |
-
Standing, K.G. Curr. Opin. Struct. Biol. 13, 595–601 (2003). | Article | PubMed | ISI | ChemPort |
-
Steen, H. & Mann, M. Nat. Rev. Mol. Cell Biol. 5, 699–711 (2004). | Article | PubMed | ISI | ChemPort |
-
Smolka, M.B., Zhou, H., Purkayastha, S. & Aebersold, R. Anal. Biochem. 297, 25–31 (2001). | Article | PubMed | ISI | ChemPort |
-
Kasai, K. J. Chromatogr. 597, 3–18 (1992). | Article | PubMed | ISI | ChemPort |
-
Sechi, S. & Chait, B.T. Anal. Chem. 72, 3374–3378 (2000). | Article | PubMed | ISI | ChemPort |
-
Yamaguchi, M. et al. Anal. Chem. 77, 645–651 (2005). | Article | PubMed | ISI | ChemPort |
-
Chelius, D. & Shaler, T.A. Bioconjug. Chem. 14, 205–211 (2003). | Article | PubMed | ISI | ChemPort |
-
Gevaert, K. et al. Nat. Biotechnol. 21, 566–569 (2003). | Article | PubMed | ISI | ChemPort |
-
Martens, L. et al. Proteomics 5, 3193–3204 (2005). | Article | PubMed | ISI | ChemPort |
-
Kuhn, K. et al. J. Proteome Res. 2, 598–609 (2003). | Article | PubMed | ISI | ChemPort |
-
Kuhn, K. et al. Proteomics 5, 2364–2368 (2005). | Article | PubMed | ISI | ChemPort |
-
Doherty, M.K., Whitehead, C., McCormack, H., Gaskell, S.J. & Beynon, R.J. Proteomics 5, 522–533 (2005). | Article | PubMed | ISI | ChemPort |
-
Doherty, M.K. et al. Proteomics 4, 2082–2093 (2004). | Article | PubMed | ISI | ChemPort |
-
Hayter, J.R., Robertson, D.H., Gaskell, S.J. & Beynon, R.J. Mol. Cell. Proteomics 2, 85–95 (2003). | PubMed | ISI | ChemPort |
Acknowledgments
Supported by grants to R.J.B. & J.L.H. from the Natural Environment Research Council and the Biotechnology and Biological Sciences Research Council. We are grateful to M. Doherty for assistance with the mass spectrometry.
Competing interests statement:
The authors declare that they have no competing financial interests. |
- and 
-amino groups are acetylated before proteolysis (trypsin in the figures, but potentially any other fragmentation method), which is followed by biotinylation of proteolytically exposed 
