Tackling trypsin weakness: large peptides harvested by SEC for secondary digestion

Tryptic troubles: sequence deficiencies limit proteome coverage

The bottom-up proteomics technique remains the most popular process for identifying proteins in complex biological samples. It is typified by digesting the mixture of extracted proteins with trypsin to produce a series of peptides which are identified by mass spectrometry. The peptide sequences lead to the identities of the proteins, sometimes one unique peptide being sufficient to pinpoint a particular protein.

However, it is not quite as straightforward as it appears. Trypsin attacks the proteins adjacent to lysine and arginine residues, so proteins with a lot of these produce many short peptides upon digestion. Small proteins of six residues or less are less likely to have a unique sequence for protein identification and do not bind to the conventional reversed-phase columns used in HPLC separation before mass spectrometric analysis.

Conversely, proteins with low numbers of lysine and arginine residues will produce large peptides that are not particularly amenable to mass spectrometry. Trypsin is also less effective for membrane proteins, which are poorly soluble in the trypsin solution, again tending to lead to large peptides upon cleavage.

Despite these drawbacks, trypsin is still regarded as the best enzyme for carving proteins into manageable portions.

In a new report, the digestion behaviour of trypsin has been characterised by scientists in Switzerland, who took a protein and broke it down theoretically (a technique referred to as in silico) and looked at the peptides that were produced.

Theoretical trypsin digestion: lost sequences in high-mass peptides

Manfredo Quadroni and colleagues from the University of Lausanne and the Swiss Institute of Bioinformatics, Geneva, took the entire proteome of the yeast Saccharomyces pombe and generated all of the possible theoretical peptides. They found that more than 55% of them were classified as small, with molecular masses lower than 800, but they covered only about 19% of the proteome. The medium-sized peptides from 800-3000 Da represented 39% of peptides and 57.4% of the total proteome.

However, the large peptides of molecular mass greater than 3000 Da represented only 5.4% of the total but covered more than 23% of the proteome sequence. It is these large peptides, containing so much information on the proteins, which are not processed well by LC/MS.

There are several reasons for this, including poor chromatographic separation, lower ionisation efficiency, complicated tandem mass spectra and the higher chance of unforeseen post-translational modifications.

A very similar theoretical peptide distribution was obtained for the in silico tryptic analyses of Saccharomyces cerevisiae and the human proteome.

Assisting trypsin: high-mass peptides isolated for further digestion

One way to circumvent this high-mass tryptic peptide problem was investigated by the Swiss researchers. Instead of ignoring them and losing the inherent sequence information that is locked in, they chose to isolate the peptide fraction containing the large peptides by size exclusion chromatography and subject them to secondary enzymatic analysis.

The concept was illustrated with the in vitro digestion of the soluble protein fraction of S. pombe. Each of the SEC fractions was analysed by LC-tandem MS and 218 peptides were identified with high confidence. Their tryptic peptides covered 20.5% of the total sequences of those peptides.

The high-mass peptide fraction from SEC was then digested with a second reagent: endoproteinase Glu-C, which cleaves peptide bonds that are C-terminal to glutamic acid residues, endoproteinase Asp-N, which cleaves peptide bonds N-terminal to aspartic acid residues, and chymotrypsin which cleaves the C-terminal side of the aromatic residues tyrosine, phenylalanine and tryptophan. Formic acid was also used, cleaving at aspartic acid-proline bonds.

All four secondary agents led to an increase in sequence coverage compared with trypsin alone, increasing the coverage from 20.5 to 32%. Each of the proteases identified new proteins that tryptic digestion did not, with a combined total of 75 new proteins.

When the vacuolar membrane protein fraction of S. cerevisiae was treated the same way, the secondary digestions produced a 50% relative increase in sequence coverage.

Clearly, the procedure has a beneficial effect on proteome discovery and the team demonstrated its general applicability by analysis of the phosphoproteome of S. pombe and a human melanoma cell line. In this case, the phosphopeptides from the secondary digestion were enriched on a titania column before analysis.

Using four enzymes instead of trypsin alone revealed a large number of new phosphosites and novel proteins in both systems, many of which had not been annotated in the standard reference databases.

SEC has been used before in proteomics studies but generally to fractionate mixtures of intact proteins, rather than the tryptic peptides. This is the first report of the use of SEC in bottom-up proteome analysis, said Quadroni, who recommended that it be used as a first step following digestion to recover the fraction containing the high-mass peptides. The remaining fractions should be pooled and processed by conventional techniques.

A small change in workflow provided a notable increase in coverage of those peptide sequences which are normally invisible to trypsin and to the identification of novel proteins and phosphorylation sites. This modified procedure should be generally applicable to most bottom-up proteomics studies.

The views represented in this article are solely those of the author and do not necessarily represent those of John Wiley and Sons, Ltd.