Targeting proline in (phospho)proteomics

Abstract

Mass spectrometry-based proteomics experiments typically start with the digestion of proteins using trypsin, chosen because of its high specificity, availability, and ease of use. It has become apparent that the sole use of trypsin may impose certain limits on our ability to grasp the full proteome, missing out particular sites of post-translational modifications, protein segments, or even subsets of proteins. To tackle this problem, alternative proteases have been introduced and shown to lead to an increase in the detectable (phospho)proteome. Here, we argue that there may be further room for improvement and explore the protease EndoPro. For optimal peptide identification rates, we explored multiple peptide fragmentation techniques (HCD, ETD, and EThcD) and employed Byonic as search algorithm. We obtain peptide IDs for about 40% of the MS2 spectra (66% for trypsin). EndoPro cleaves with high specificity at the C-terminal site of Pro and Ala residues and displays activity in a broad pH range, where we focused on its performance at pH = 2 and 5.5. The proteome coverage of EndoPro at these two pH values is rather distinct, and also complementary to the coverage obtained with trypsin. As about 40% of mammalian protein phosphorylations are proline-directed, we also explored the performance of EndoPro in phosphoproteomics. EndoPro extends the coverable phosphoproteome substantially, whereby both the, at pH = 2 and 5.5, acquired phosphoproteomes are complementary to each other and to the phosphoproteome obtained using trypsin. Hence, EndoPro is a powerful tool to exploit in (phospho)proteomics applications.

Keywords: (phospho)proteomics; EndoPro; mass spectrometry; proline effect; protease.

Figure 1

Characterization of EndoPro cleavage specificity. (A) Overview of amino acids after which was cleaved by EndoPro (n = 4, purple) and trypsin (n = 4, orange), based on a nonspecific search, revealing a high specificity of 84.6% A/P and 89.6% R/K for EndoPro and trypsin, respectively. Only amino acids with a cleavage frequency of 1% or higher were included. Data are represented as mean percentage of total cleavages per protease ± SEM. (B) An iceLogo showing the differences between the EndoPro cleavage site environment (17 032 unique environments from nonspecific search) and the human proteome, illustrating a disfavor for R/K on the +2 position and a reluctance to cleave between proline residues. (C) Overlap of unique proteins identified by EndoPro or trypsin using a semispecific search. Although the sizes of the identified proteomes are roughly equal, the overlap between the two is only 35%.

Figure 2

Comparison of peptide characteristics in EndoPro and tryptic digests. (A) Peptide length distribution of identified unique peptides following digestion with trypsin or EndoPro. All four EndoPro conditions probed here reveal a similar distribution, exhibiting a long tail toward peptides with more than 50 amino acids, which was not observed for tryptic peptides. (B) Charge distribution of all unique peptides identified following the different digestion conditions, where digestion with EndoPro results in more highly charged peptides (z ≥ 4). (C) Amino acid content of the peptides identified in the EndoPro digests under various digestion conditions. With increase in pH and digestion duration, negatively charged amino acids are more frequently observed and the A/P content of the peptides is reduced. (D) Cleavage specificity of the identified peptides. Digestion with EndoPro yields highly specific proline and alanine C‐terminal peptides, especially at pH = 2, with a Pro/Ala specificity close to that of trypsin for Arg/Lys. (E, F) Location of Asp on peptides digested ON with EndoPro at (E) pH = 2 and (F) pH = 5.5. At pH = 5.5, the negatively charged amino acid is disfavored at the C terminus of the generated peptides. This was not observed for peptides produced at pH = 2, indicating that two distinct sets of peptides are formed at these pH values.

Figure 3

Highly complementary protein identifications observed by using EndoPro or trypsin. (A) Overview of the overlap in proteins identified by using the different proteases and varying digestion conditions as listed in Table 2, illustrating how complementarity increases when cleaving with EndoPro at different conditions. The smallest overlap, 35%, is observed between EndoPro and trypsin. (B, C) Reproducibility of (B) Trypsin and (C) EndoPro technical replicate analyses, revealing a robust overlap of around 65%. (D) When comparing all unique protein groups identified in at least three out of four technical replicates, 30% of the proteins that are reproducibly identified using EndoPro are not identified in tryptic lysates.

Figure 4

Proteome Characteristics. (A) Comparison of the sequence coverage achieved by using trypsin and EndoPro (the latter under 4 different digestion conditions) for in total 380 selected proteins. Only these 380 proteins showing at least 50% more sequence coverage in one of the datasets were considered in B–E. For clarity, proteins for which the two proteases performed comparably were not included. Black indicates no coverage of a protein in a certain condition. (B) Comparison of the arginine and/or lysine content, which is significantly higher in EndoPro peptides. (C–E) Comparison of the proline content (C), isoelectric point (D), and molecular weight (E) of proteins identified using EndoPro (at 4 different conditions) or trypsin. Notably, as shown in (E) EndoPro favors smaller proteins; trypsin shows a bias for larger proteins. Significance was determined using one‐way ANOVA, with α = 0.05. **P < 0.01, and ***P < 0.001; error bars represent SEM.

Figure 5

EndoPro is highly complementary to trypsin in the identification of site‐specific phosphorylation events. (A) Comparison of identified unique phosphoproteins between EndoPro and trypsin, revealing a 37% overlap. (B) Overlap in identified unique phosphosites on 1652 phosphoproteins identified by both proteases, indicating that on these shared phosphoproteins, only 30% of the phosphosites could be identified by both proteases. (C) Heatmap displaying phosphosite spectral count scores of 13 762 phosphosites from low (1) to high (> 10), revealing that EndoPro is highly complementary to trypsin in identification of phosphosites. Black indicated not identified. (D) Global kinase classification analysis of all identified phosphopeptides, dividing them into 4 categories: proline‐directed, acidophilic, basophilic, or other. Although in all analyses the SP/TP motif encompasses over 50% of the detected sites, short digestion with EndoPro results in a further increase of this proline‐directed motif to about 70%.

Figure 6

Amino acid length and localization of phosphorylation sites on the identified phosphopeptides. (A) Localization of the phosphorylation on unique phosphopeptides from EndoPro, showing the highly preferred phosphorylation on the second to last amino acid on the peptide (i.e., Ser‐Pro or Thr‐Pro), and the disfavor for phosphorylation on the penultimate N‐terminal amino acid on the EndoPro peptides. (B) Localization of phosphorylation on unique phosphopeptides following trypsin digestion at pH = 8.5, revealing a strong disfavor for phosphorylation on the ultimate and penultimate N‐terminal amino acids on the peptides, and preferential phosphorylation on the third amino acid of the identified phosphopeptides.