Quantitative proteomics reveals the kinetics of trypsin-catalyzed protein digestion

Transcription

1 Anal Bioanal Chem (2014) 406: DOI /s RESEARCH PAPER Quantitative proteomics reveals the kinetics of trypsin-catalyzed protein digestion Yanbo Pan & Kai Cheng & Jiawei Mao & Fangjie Liu & Jing Liu & Mingliang Ye & Hanfa Zou Received: 9 June 2014 /Revised: 14 July 2014 /Accepted: 25 July 2014 /Published online: 19 August 2014 # Springer-Verlag Berlin Heidelberg 2014 Abstract Trypsin is the popular protease to digest proteins into peptides in shotgun proteomics, but few studies have attempted to systematically investigate the kinetics of trypsin-catalyzed protein digestion in proteome samples. In this study, we applied quantitative proteomics via triplex stable isotope dimethyl labeling to investigate the kinetics of trypsin-catalyzed cleavage. It was found that trypsin cleaves the C-terminal to lysine (K) and arginine (R) residues with higher rates for R. And the cleavage sites surrounded by neutral residues could be quickly cut, while those with neighboring charged residues (D/E/K/R) or proline residue (P) could be slowly cut. In a proteome sample, a huge number of proteins with different physical chemical properties coexists. If any type of protein could be preferably digested, then limited digestion could be applied to reduce the sample complexity. However, we found that protein abundance and other physicochemical properties, such as molecular weight (Mw), grand average of hydropathicity (GRAVY), aliphatic index, and isoelectric point (pi) have no notable correlation with digestion priority of proteins. Keywords Trypsin. Protein digestion. Kinetics. Stable isotope dimethyl labeling. Mass spectrometry Electronic supplementary material The online version of this article (doi: /s ) contains supplementary material, which is available to authorized users. Y. Pan: K. Cheng : J. Mao : F. Liu : J. Liu : M. Ye (*) : H. Zou (*) Key Lab of Separation Sciences for Analytical Chemistry, National Chromatographic Research and Analysis Center, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian , China mingliang@dicp.ac.cn hanfazou@dicp.ac.cn Y. Pan: K. Cheng : J. Mao : F. Liu : J. Liu University of Chinese Academy of Sciences, Beijing , China Introduction Shotgun proteomics, i.e., bottom-up proteomics, is a powerful strategy for identifying proteins from complex protein mixture [1]. It relies on enzymatic digestion of proteins into peptides prior to liquid chromatography-coupled tandem mass spectrometry (LC-MS/MS) analysis [2]. A variety of proteases including trypsin, Lys-C, Glu-C, etc. are applied to digest proteins in proteome research. Among these proteases, trypsin is the most commonly used enzyme. This is mainly attributed to the facts that tryptic peptides have highly basic residues at the C-termini of peptides, and they are in the preferred mass range for effective fragmentation by MS/MS. Thus, the fragmentation of tryptic peptides generally leads to a series of y- ion series and makes tandem mass spectra more easily interpretable [3]. Trypsin catalyzes the hydrolysis of peptide bonds immediately after lysine/arginine (K/R) residues in proteins. There are many potential trypsin cleavage sites in proteins, while not all these sites could be cut during a trypsin digestion. These sites are called missed cleavage sites [4 6]. Clearly, the missed cleavage sites have very slow kinetics to be hydrolyzed by trypsin. Though the trypsin digestion plays an important role in proteome analysis, systematic studying on the kinetics of trypsin-catalyzed reactions in proteome samples was not reported. In a proteome sample, huge number of proteins with different physical chemical properties coexists. If any type of proteins could be preferably digested, then limited digestion could be applied to reduce the sample complexity. However, no one has attempted to investigate the digestion priority of different proteins in a typical proteome study. To better understand the trypsin-catalyzed digestion process, kinetics study of this important process is required. There are numerous studies on kinetics analysis of trypsin-catalyzed hydrolysis of lysine or arginine derivatives [7, 8]. In these studies, a single substrate with only one cleavage site was

2 6248 Y. Pan et al. incubated with trypsin for determination of kinetics constants. The kinetics constants for peptides can also be determined in the same way when mass spectrometry was applied to monitor the reactions [9]. For these studies, one enzymatic reaction can only determine the kinetics constants for one substrate. In theory, the cleavage priority of cleavage sites in all proteins in proteome samples could be determined by synthesizing peptides centered with these sites. However, this approach is expensive and time-consuming. Though a single protein was also reported to study trypsin digestion kinetics, these studies cannot reflect the kinetics of the cleavage sites [10, 11]. Walmsley et al. [12] compared peptide abundances in 2- and 18-h human serum albumin (HSA) digests using label-free quantification and principal components analysis (PCA) and found that many cleavage sites showed variable digestion kinetics patterns. However, systematic investigation of the kinetics of the cleavage sites on proteins in complex proteome samples is still needed. Quantitative proteomics, especially stable isotope labeling, is a powerful tool for determining the different peptide amount between different proteome samples in high throughput. Recently, we have demonstrated in a preliminary study that quantitative proteomics could be a powerful tool to study the kinetics and digestion priority of trypsin digestion [13]. Triplex stable isotope dimethyl-labeling approach enable the simultaneous analysis of three samples in one time, and it is a reliable, a cost-effective, and an undemanding procedure that can be easily automated and applied in high-throughput proteomics experiments [14]. So, we employed the triplex stable isotope dimethyl-labeling approach to investigate the kinetics behavior of trypsin digestion in detail in this study. Proteome samples were digested with three different times, and then the resultant digests were labeled with stable isotope dimethyl labels, respectively. After quantitative proteomics analysis, 10,483 unique peptides from 2,270 proteins were quantified. Based on the abundance changes of the generated peptides during this time course study, four types of cleavage sites, i.e., very fast, fast, slow, and very slow, were determined. This enabled the investigation of cut priority of cleavage sites surrounded with different residues. It was found that the cleavage sites surrounded by neutral residues could be quickly cut, while those with neighboring charged residues (D/E/K/R) or proline residue (P) could be slowly cut. Because the quantified peptides could be classified into early and later generated peptides, the digestion priority of proteins with different physicochemical properties can also be investigated. In general, the results show that protein abundance and other physicochemical properties, such as molecular weight (Mw), grand average of hydropathicity (GRAVY), aliphatic index, and isoelectric point (pi) has no notable influence on the digestion priority of proteins. Experimental section Reagents and chemicals All the water used in this experiment was prepared using a Milli-Q system (Millipore, Bedford, MA). Formic acid (FA) was provided by Fluka (Buchs, Germany). Acetonitrile (ACN, HPLC grade) was purchased from Merck (Darmstadt, Germany). All the other chemicals and reagents were purchased from Sigma (St. Louis, MO). Fused silica capillaries with 75 μm i.d. were obtained from Polymicro Technologies (Phoenix, AZ). Cell growth and lysis The HeLa cells were grown according to Bian et al. [15]. The cell pellets were softly homogenized in a cold lysis buffer containing 8 M urea, 50 mm triethyl ammonium bicarbonate (TEAB; ph=8.0), 2 % protease cocktail (v/v), 1 % Triton X-100 (v/v), 65 mm dithiothreitol (DTT), 1 mm EDTA, 1 mm EDGA, 1 mm PMSF, 1 mm NaF, and 1 mm Na 3 VO 4, sonicated for 400 W 120 s, and centrifuged at 23,000g for 1 h. The supernatant containing the total cell proteins was precipitated with five volumes of cold acetone/ ethanol/acetic acid (v/v/v=50/50/0.1) at 20 C. Protein precipitant was centrifuged at 15,000g for 30 min. The pellet was washed separately with acetone and 75 % ethanol, then lyophilized to dryness, and stored at 80 C. Protein digestion and stable isotope dimethyl labeling Proteins (1 mg, the protein concentration was determined by Bradford assay) were denatured and reduced in 1 ml of 8 M urea, 100 mm TEAB, ph 8.0, and 10 mm DTT at 56 C for 40 min and then alkylated by 20 mm IAA in the darkness at room temperature for 30 min. The sample was diluted to 8 ml (1 M urea) with 100 mm TEAB, ph 8.0. Trypsin (Sigma) was added at a 40:1 protein/protease mass ratio along with CaCl 2 to 1 mm for digestion at 37 C; after digested for 1, 4, and 18 h, three same aliquots (100 μg) of samples were removed from the tube. To prevent further digestion, the trypsin was inhibited by addition of aprotinin (final concentration, 2.0 μg/ ml). Then, the three aliquots above were desalted by solidphase extraction (SPE) column, lyophilized, and labeled with light, intermediate, and heavy dimethyl, respectively. For the triplex stable isotope dimethyl labeling [14, 16], 50 μlofch 2 O(4%,v/v)CD 2 O(4%,v/v)and 13 CD 2 O(4%, v/v) were added into the sample solutions, respectively, and then 50 μl of freshly prepared NaBH 3 CN (0.6 M), NaBH 3 CN (0.6 M), and NaBD 3 CN (0.6 M) were added subsequently. The resultant mixture was incubated for 1 h at room temperature. Then, 10 μl of ammonia (25 %) and 25 μloffawere added to consume the excess labeling reagents and to acidify

3 Quantitative proteomics reveals the kinetics of protein digestion 6249 the sample. After mixing in a ratio of 1:1:1 on the basis of the total peptide amount, the labeled peptide mixture was desalted by the SPE column. Dry the samples by vacuum centrifugation and stored at 80 C until used. Nano LC-MS/MS analysis For 2D strong cation exchange (SCX)-RP LC-MS/MS analysis with the LTQ-Orbitrap mass spectrometer (Velos, Thermo Fisher Scientific), a capillary monolithic column (5 cm 200 μm ID) with phosphate functional groups was applied as an SCX trap column in the first dimension. The sample loading and analysis procedures were as follows: the peptide samples were first dissolved in 0.1 % (v/v) formicacidin water, and then loaded onto the monolith SCX trap column; the trap column was equilibrated with 0.1 % (v/v)formicacid in water for 10 min [17]. After that, it was directly connected to an RP analytical column in tandem by a union. Then, six gradient elution steps were applied to gradually elute peptides from the SCX trap column to the RP analytical column with ammonium acetate solution concentrations of 50, 150, 250, 350, 500, and 1,000 mm, respectively. After each elution step, a subsequent RP LC-MS/MS was executed in 150-min gradient time with 0.1 % (v/v)formicacidinacetonitrilefrom5to 35 % (v/v); a capillary column was first manually pulled to a fine point as spray tip, and then packed with C18 AQ beads (3 μm, 120 Å, Michrom Bio Resources). All MS and MS/MS spectra were acquired in the data-dependent mode with the 20 most intense ions fragmented by CID. multiplied by 1,000 and divided by its amino acid frequency of occurrence in the database to form the normalized peptide sequences [20]. The pi, GRAVY, and aliphatic index of proteins and peptides were calculated according to ExPASy ( Results and discussion Time course study of trypsin digestion of proteome sample Time course investigation of trypsin digestion of proteome samples was performed with quantitative proteomics at three time points. As shown in Fig. 1, the lysate of HeLa cells (1 mg) was firstly subjected to trypsin digestion, and the same aliquots (100 μg) of the digestion were removed at time points of 0.5, 2, and 18 h, respectively. The trypsin inhibitor aprotinin were added to the three-time course digestion immediately to avoid further digestion. Then, quantitative proteomics using triplex stable isotope dimethyl labeling was applied to monitor the abundance variation of generated peptides at different time points. The digests from the three time points were labeled with light (0.5-h digestion), intermediate (2-h digestion), and heavy dimethyl (18-h digestion), respectively. The labeled Data analysis Protein quantification was performed using MaxQuant (version , [18]. The raw files were searched against UniProt database of human downloaded from (released on 12/11/ 2013); carbamidomethylation on cysteine was set as a fixed modification, and oxidation on methionine was set as variable modifications. Peptides were searched using fully tryptic cleavage constraints and up to four missed cleavage sites were allowed; the mass tolerances for the precursor ions and fragment ions were set to 6 ppm and 0.5 Da, respectively. For quantification, stable isotope dimethyl labeling, different quantification modes integrated into MaxQuant were selected, respectively. The other settings were the same to the conventional search. Sequence logos were automatically generated by the WebLogo ( logo.cgi) [19]. The raw sequences for WebLogo analysis were centered at the cleavage site and extended 13 residues (±6 residues). The N- or C-terminal sequences that could not be extended were excluded. In order to eliminate the influence of the relative occurrence of different amino acids in the proteome, the raw sequences for WebLogo analysis were Fig. 1 Experimental scheme for investigating the kinetics of trypsincatalyzed protein digestion by quantitative proteomics

4 6250 Y. Pan et al. peptides from the above three aliquots (10 μg each) were combined and analyzed with 2D RP LC-MS/MS. The acquired raw files from two technical replication runs were processed using the MaxQuant platform. To keep only the highly reliable quantified results, more strict criteria were applied to filter the data, the peptide should be quantified with RSD <50 % in both runs [16, 17, 21], which led to the quantification of 10,483 unique peptides from 2,270 proteins. It was found that 42.6 % (11,128/23,346) peptides have at least one missed cleavage sites. In a control experiment where the digestion was performed as in the general proteomics experiment, it was found that 25.8 % (814/3,157) peptides have more than one missed cleavage sites. The high frequency of missed cleavage sites observed for the peptides quantified in this time course study indicated that the digestion in the initial digestion stage is not complete. The distributions of log2 ratios M/L, H/M, and H/L were given in Fig. S1 in the Electronic Supplementary Material; the percentages of peptides with log2 ratios M/L, H/M, and H/L out of [ 1, 1] were 61.6, 34.0, and 47.5 %, respectively. In quantitative proteomics experiments, the percentages are usually less than 2 % if the same amount of sample was used [22]. Clearly, the concentration of many peptides changed significantly during the different digestion time which indicated that these peptides were generated at different speed. To directly reflect the change of peptide abundance during the time course, the peak areas of the three isotopic peaks were normalized by that of the medium one. Therefore, the abundances of the generated peptides were represented as L/M, M/M, and H/M. The log2 ratios out of [ 1, 1] were considered as significant change and between [ 1, 1] were considered as unchanged. Except the log2 ratios L/M, M/M, and H/M, log2 ratios H/L were employed to remove the uncorrected ones. Because there are three time points, in theory, the dynamic change of the peptide concentration during the digestion can be clustered into nine types (Fig. 2 and Electronic Supplementary Material (ESM) Table S1). Cluster 1: The concentration of these peptides does not change significantly for all the three time points. Cluster 2: the peptide concentrations do not change much from 0.5 to 2 h, but decreased in 18 h. Cluster 3: the peptide concentration decreased from 0.5 to 2 h, but do not change much in 18 h. Cluster 4: the peptide concentrations decreased during all the digestion steps. The peptides from above four clusters have their peak concentrations at the time point of 1 h, indicating that these peptides were mainly generated in the first digestion step, and were classified as the early generated peptides (Table 1). Except cluster 1, the concentrations of other peptides decreased with further digestion, indicating that these peptides were gradually degraded. We compared the percentage of peptides with missed cleavage sites for these four clusters. It was found that 32.7%ofpeptidesincluster1havemissedcleavagesites, while 64.8, 90.0, and 91.3 % peptides in other three clusters have at least one missed cleavage sites (Fig. 3a). These data illustrated that the peptides in clusters 2, 3, and 4 were degraded with further digestion probably because the missed cleavage sites were slow ones which were cut afterwards. The rest of the peptides have their peak concentrations at other time points (Fig. 2). Cluster 5: the concentration of these peptides increased from 0.5 to 2 h, but remains unchanged afterwards. Cluster 6: the peptide concentration increased from 0.5 to 2 h, and then decreased. The peptides in above two clusters have their peak concentrations at 2 h, and they were mainly generated at the second digestion stage. Cluster 7: the peptides with concentration increased all the time. Cluster 8: the abundance of these peptides does not change from 0.5 to 2 h, but increased in 18 h. These peptides have their peak concentrations at 18 h, and they were mainly generated at the last digestion stage. The last cluster 9: the abundance of these peptides decreased from 0.5 to 2 h, and then increased from 2 to 18 h. It is hard to explain the abundance change for these peptides. Because only 0.25 % peptides (21/8,334) belong to this cluster, these peptides were not considered seriously in this study. Because the peptides in the last four clusters were mainly generated in the last two digestion step, they were classified as the late generated peptides (Table 1). It can be seen from Fig. 3a that the peptides in clusters 2, 3, 4, 6, and 9 have higher frequency of missed cleavage sites. Interestingly, the peptides in these clusters decreased in abundance at least in one interval of the three time points as evidenced in Fig. 2. And if the peptides in the cluster did not decrease in abundance in any of the intervals (clusters 1, 5, 7, 8), their frequencies of having missed cleavage were much lower. This is because the peptides with missed cleavage sites tend to be further cut during the digestion, which led to the decrease of their abundance. The high consistent of the frequency of missed cleavage sites with the abundance change during the time course indicated that these peptides were accurately quantified. Investigation of the cut priority of cleavage sites surrounded with different residues Each protein has many trypsin cleavage sites. After trypsin digestion, proteins will be digested into many peptides. The above quantitative proteomics study indicated that these peptides are not generated at the same speed. The reason that some peptides were generated earlier and some peptide generated later is that the trypsin-catalyzed hydrolysis rates are different for different cleavage sites. Though this quantitative proteomics cannot directly determine the kinetics constants for this enzymatic reaction, it can reveal the cut priority of the cleavage sites. The sequence surrounding the cleavage site can be described as P 4 -P 3 -P 2 -P 1 -P 1 -P 2 -P 3 -P 4 [23], where cleavage occurs between P 1 and P 1. For trypsin-catalyzed cleavage, all P 1 positions are either K or R. It is of interest to

5 Quantitative proteomics reveals the kinetics of protein digestion 6251 Fig. 2 Clusters of the peptides according to their abundance change during the time course investigate which types of residues surrounding the cleavage sites affect their digestion kinetics. Except for the peptides generated from protein N-/C-termini, the majority of peptides are generated by two trypsin cleavages. Thus, for each identified peptide, it typically has two terminal cleavage sites. Take a quantified peptide, FIDTTSKFGHGR, as the example. The N-terminal residue (F) on this peptide is the P 1 residue for the N-terminal cleavage site, while the C-terminal residue (R) is the P 1 residue. To extract the residues surrounding the cleavage sites, the identified peptides should be mapped to their parent proteins. For above the peptide, the sequences surrounding the two cleavage sites were determined to be DLK.FID and HGR.FQT after mapping to their parent protein Table 1 The numbers of the early and late generated peptides Early generated peptides Late generated peptides Cluster 1 (2,156) Cluster 5 (2,932) Cluster 2 (1,179) Cluster 6 (1,692) Cluster 3 (189) Cluster 7 (6) Cluster 4 (126) Cluster 8 (12) sequence, respectively. In this way, the residues around the peptide terminal cleavage sites could be determined. It is well known that K/R with a neighboring P residue on the C-terminal side (P 1 position) and K/R with an aspartic acid (D) or glutamic acid (E) residue on either the N- or C-terminal side (P 2 or P 1 position) were difficult to cut [3 6, 24]. We first investigate if there were notable differences in distribution of P on P 1 position and D/E on (P 2 and P 1 position) for the terminal cleavage sites on above nine clusters of peptides. As shown in Fig. 3b, the percentages of P residue on this position were all <0.15 %, which is far less than the nature P residue frequency of the human proteome (about 6.3 % as shown in ESM Fig. S2) [25, 26] and the frequency of P followed by K/R in database (5.7 %). It indicated that K/R followed with P was not likely cut by trypsin. This is consistent to Keil rules [27], the commonly accepted rule for a trypsin cut site is K/R.P. The percentages of negative charged amino acid residues D/E (P 2 or P 1 position) were less than the native frequency of the human proteome ( 13 %) for clusters 1 4, while the percentages for clusters 5 8 were higher than the native frequency of the human proteome. Based on the quantified ratios, the peptides in clusters 1 4 were generated earlier than those in clusters 5 8 during the digestion.

6 6252 Y. Pan et al. Fig. 3 Percentages of peptides (a) with missed cleavage sites; (b) with D/E (on P 2 and P 1 position), K/R (on P 1 position), and P (on P 1 position) for terminal cleavage sites; and (c) with D/E, K/R, and P (their positions were as in (b)) for missed cleavage sites in nine clusters

7 Quantitative proteomics reveals the kinetics of protein digestion 6253 Fig. 4 Sequence logos of four cleavage site types with different kinetics (very fast, fast, slow, and very slow sites). a All sequences, b the sequences only consider K as the cleavage site, and c the sequences only consider R as the cleavage site. The frequencies of amino acids in the peptides were normalized by their occurrence frequency in the proteome database The high frequency of acidic residues in late generated peptides means that the trypsin-catalyzed reaction is slow when K/R with neighboring D/E. This is also consistent with the fact that K/R with neighboring D/E is likely to be missed cleavage. In addition to the cleavage sites revealed from the peptide termini, there were missed cleavage sites on some of the identified peptides. Still, take the identified peptide FIDTTSKFGHGR as an example. It has one missed cleavage site. The sequence centered with the cleavage site was TSK.FGH. We then compared the distributions of P on P 1 position and D/E on P 2 or P 1 position for the missed cleavage sites on above nine clusters of peptides. As shown in Fig. 3c,the percentages for P were far higher than the N- and C-terminal cleavage sites, and similar with or slightly higher than the nature P amino acid composition of the human proteome. This confirmed that K/R followed by P was not likely cut by trypsin. The percentages of negative charged amino acids D/E (>18.4 %) were all higher than the native composition of the human proteome (D/E about 12 %) for all clusters. These percentages are higher than those of cleavage sites revealed by the terminal sites of the early generated peptides (clusters 1 4 in Fig.3b) while are quite similar with those of the terminal sites for the late generated peptides (clusters 5 8inFig.3b). This is not surprising since the missed cleavage sites were relatively slow. As shown above, the cleavage sites could be revealed by the quantified peptides with either terminal sites or missed cleavage sites. To investigate the cut priority of cleavage sites, they must be sorted in the order of their kinetics. Depending on when the sites got cut, the cleavage sites are classified into four types, i.e., very fast, fast, slow, and very slow sites (for details, see the Supplementary Note and Table S2 in the ESM). The fast cleavage sites (5,942) were the sites got cut in the first digestion step. Both C- and N-terminal sites for the early generated peptides (clusters 1 4) are fast cleavage sites because they were generated in the first digestion step. The slow cleavage sites (105) were the sites got cut in the second digestion step. The missed cleavage sites in peptides of cluster 3 containing one missed cleavage site belongs to this class because the concentrations of peptides in this cluster do not change much from 0.5 to 2 h, but decreased in 18 h, indicating that these peptides got cut in the second digestion step. The slower cleavage sites (1,161) were the sites got cut in the third digestion step. Based on the change of peptide concentration, the missed cleavage sites on peptides of cluster 2 and cluster 6 with one missed cleavage site belong to this type. The slowest cleavage sites (1,682) were the sites cannot be cut at any digestion steps. They are the missed cleavage sites on peptides from cluster 1, cluster 5, and cluster 8.

8 6254 Y. Pan et al. The sequence logos for the normalized peptide sequences centered with above four types of cleavage sites were generated by the WebLogo and are shown in Fig. 4. It is obvious that the cleavage sites K/R surrounded by neutral residues could be quickly cut, while those with neighboring charged residues (D/E/K/R) or P could be slowly cut (Fig. 4a). For the two types of fast cleavage sites, i.e., very fast and fast sites, R residues on P 1 position account for 55.3 and 44.2 % of all sites, respectively. While for the two types of slow cleavage sites, the R sites account for less than 25 % (ESM Fig. S3). This indicated that trypsin cleaves the C-terminal to K and R residues with higher rates for R. We are curious if there is any difference in the effects of surrounding residues on the kinetics of cleavage sites K and R. For this purpose, we generated the sequence logos for cleavage sites K and R separately (Fig. 4b, c). In general, they are quite similar but there are some differences. To examine the effects of surrounding residues on the kinetics of cleavage sites in detail, we compared the distribution of the neighboring residues (ESM Fig. S4). An interesting phenomenon is that the K/R following P (P 2 position) tends to be very fast cleavage sites, the special configuration of proline and the small side chain probably allow trypsin access more easily the cleavage sites. The presence of D/E on P 2,P 1,andP 2 positions of cleavage sites K make the kinetics slow, and the P 2 position is more sensitive to D/E; the similar situation was observed in cleavage sites R. Cleavage sites R with alkaline amino acid R on P 2 position were more difficult to be cut than acidic amino acids, the opposite situation occurred at the cleavage sites K. Investigation of the digestion priority of proteins with different physicochemical properties It is of interest to investigate if the peptides generated at different time points have correlation with different types of proteins. The quantified peptides were classified into two types: (1) the early generated peptides, these peptides are mainly generated at the first digestion stage (clusters 1 4) and (2) the late generated peptides, these peptides are mainly generated at the second and third digestion stages (clusters 5 8). There were 3,650 and 4,642 unique peptides that correspond to 1,397 and 1,346 proteins for the above two types of peptides, respectively. If a specific type of proteins is digested earlier during a digestion, then more peptides should present in the pool of the early generated peptides. Therefore, the protein digestion priority could be judged according to the distribution of early and late generated peptides across different physicochemical properties of their parent proteins. We have investigated the digestion priority of proteins relative to their abundances with a much smaller dataset [13]. In this study, we further investigated this correlation with much bigger dataset. The quantified proteins were classified into 16 bins according to their spectra counts, which approximately represented their abundances. Then, the distributions of early and late generated peptides across the spectra counts were investigated (Fig. 5). The overall distributions between the two types of peptides are quite similar, suggesting that the digestion priority of individual proteins is almost independent of their abundances. However, there is a subtle difference between these two distributions. The percentages for early generated peptides are slightly higher than those for late generated peptides in the low spectra count range, while the difference is opposite in the high spectra count range but the difference can be ignored when the spectra counts were larger than 32. If the protein spectra count accurately reflects the protein abundance [28 30], then the low abundance proteins are digested slightly earlier than the high abundance ones in general. We then investigated if the early/late generated peptides preferably derived from proteins with some types of physicochemical properties. We first investigated the digestion priority for proteins with different sizes. The Mw for the majority of the identified proteins were in the range of 30,000 to 100,000 Da. If the proteins are not denatured, then trypsin may have difficulty to access the cleavage sites buried inside big proteins, and so, the more late generated peptides should be derived from the big proteins. However, it was found that the distributions of the two types of peptides across the Mw of proteins are also very similar (ESM Fig. S5), indicating the digestion priority of proteins does not depend on their sizes. This is not surprising since the proteins are denatured, and so, Fig. 5 The distribution of the early generated peptides and the late generated peptides across the protein spectra counts. The percentages on the Y-axis are the percentages of the early or late generated peptides within each bin of log2 (spectra counts) of the proteins they derived from

9 Quantitative proteomics reveals the kinetics of protein digestion 6255 the accessibility of trypsin to the cleavage sites on the proteins with different sizes is similar. As trypsin cut the sites surrounded with neutral residues with high rate, it is of interest to investigate if the digestion priority depends on their hydrophobicity. The GRAVY value for a protein is calculated as the sum of hydropathicity values of all the amino acid residues divided by the number of residues in the sequence [31]. The aliphatic index of a protein is defined as the relative volume occupied by aliphatic side chains (alanine, valine, isoleucine, and leucine). An increase in the aliphatic index increases the thermostability of globular proteins [32]. Both GRAVY and aliphatic index reflect the hydrophobicity of proteins. It can be found that the distributions of early and late generated peptides across either GRAVY value or aliphatic index have no notable differences (ESM Fig. S6 and S7), indicating that there are no much differences in the digestion priority of proteins with different hydrophobicity. The GRAVY values of most proteins were <0, showing that the digested proteins were relatively hydrophilic because of the lysis buffer solution used here. Finally, we investigated the dependence of digestion priority on protein s pi values and no dependence was observed either (ESM Fig. S8). Above data indicated that digestion priority does not depend on the physicochemical properties of proteins investigated in general. Conclusions To study the kinetics of trypsin-catalyzed protein digestion, quantitative proteomics was applied to monitor the dynamics of the generated peptides from trypsin digestion of a proteome sample in a time course study. According to the dynamic change of peptide abundance, the peptides were divided into two types: the early generated peptides and the late generated peptides. In general, the trypsin-catalyzed digestion priority of individual proteins in a proteome sample is independent of their abundances and other physicochemical properties, such as Mw, GRAVY, aliphatic index, and pi. Thus, selective enrichment or depletion of specific type of proteins via limited digestion is likely impossible. The data also indicate that the priority order of cleavage depends largely on the kinetics properties of the cleavage sites, i.e., the residues surrounding the cleavage sites in proteins. The high consistency of the slow cleavage sites with the reported missed cleavage sites indicated that the quantitative proteomics approach is a good approach to compare the kinetics of trypsin-catalyzed cleavage. Acknowledgments This work was supported by the China State Key Basic Research Program Grant (2013CB911202, 2012CB910101, and 2012CB910604), the Creative Research Group Project of NSFC ( ), the National Natural Science Foundation of China ( , , , and ), National Key Special Program on Infection diseases (2012ZX ), and Analytical Method Innovation Program of MOST (2012IM030900). References 1. Hunt DF, Yates JR, Shabanowitz J, Winston S, Hauer CR (1986) Protein sequencing by tandem mass spectrometry. Proc Natl Acad Sci U S A 83(17): Wu C, Tran JC, Zamdborg L, Durbin KR, Li M, Ahlf DR, Early BP, Thomas PM, Sweedler JV, Kelleher NL (2012) A protease for middle-down proteomics. Nat Methods 9(8): Olsen JV, Ong S-E, Mann M (2004) Trypsin cleaves exclusively C- terminal to arginine and lysine residues. Mol Cell Proteomics 3(6): Siepen JA, Keevil E-J, Knight D, Hubbard SJ (2007) Prediction of missed cleavage sites in tryptic peptides aids protein identification in proteomics. J Proteome Res 6(1): Lawless C, Hubbard SJ (2012) Prediction of missed proteolytic cleavages for the selection of surrogate peptides for quantitative proteomics. Omics 16(9): Gershon PD (2013) Cleaved and missed sites for trypsin, Lys-C and Lys-N can be predicted with high confidence on the basis of sequence context. J Proteome Res 13(2): Wang S-S, Carpenter FH (1968) Kinetic studies at high ph of the trypsin-catalyzed hydrolysis of N α -benzoyl derivatives of L- arginamide, L-lysinamide, and S-2-aminoethyl-L-cysteinamide and related compounds. J Biol Chem 243(13): Simpson B, Haard N (1984) Purification and characterization of trypsin from the Greenland cod (Gadus ogac). 1. Kinetic and thermodynamic characteristics. Can J Biochem Cell Biol 62(9): Caprioli RM, Smith L (1986) Determination of K m and V max for tryptic peptide hydrolysis using fast atom bombardment mass spectrometry. Anal Chem 58(6): Fraser D, Powell RE (1950) The kinetics of trypsin digestion. J Biol Chem 187: Halsey JF, Harrington WF (1973) Substructure of paramyosin. Correlation of helix stability, trypsin digestion kinetics, and amino acid composition. Biochemistry 12(4): Walmsley SJ, Rudnick PA, Liang Y, Dong Q, Stein SE, Nesvizhskii AI (2013) Comprehensive analysis of protein digestion using six trypsins reveals the origin of trypsin as a significant source of variability in proteomics. J Proteome Res 12(12): Ye M, Pan Y, Cheng K, Zou H (2014) Protein digestion priority is independent of protein abundances. Nat Methods 11(3): Boersema PJ, Raijmakers R, Lemeer S, Mohammed S, Heck AJR (2009) Multiplex peptide stable isotope dimethyl labeling for quantitative proteomics. Nat Protoc 4(4): Bian Y, Ye M, Song C, Cheng K, Wang C, Wei X, Zhu J, Chen R, Wang F, Zou H (2012) Improve the coverage for the analysis of phosphoproteome of HeLa cells by a tandem digestion approach. J Proteome Res 11(5): Song C, Wang F, Ye M, Cheng K, Chen R, Zhu J, Tan Y, Wang H, Figeys D, Zou H (2011) Improvement of the quantification accuracy and throughput for phosphoproteome analysis by a pseudo triplex stable isotope dimethyl labeling approach. Anal Chem 83(20): WangF,ChenR,ZhuJ,SunD,SongC,WuY,YeM,WangL,ZouH (2010) A fully automated system with online sample loading, isotope dimethyl labeling and multidimensional separation for highthroughput quantitative proteome analysis. Anal Chem 82(7): Cox J, Mann M (2008) MaxQuant enables high peptide identification rates, individualized ppb-range mass accuracies and proteome-wide protein quantification. Nat Biotechnol 26(12): Crooks GE, Hon G, Chandonia J-M, Brenner SE (2004) WebLogo: a sequence logo generator. Genome Res 14(6): Rodriguez J, Gupta N, Smith RD, Pevzner PA (2007) Does trypsin cut before proline? J Proteome Res 7(1):

10 6256 Y. Pan et al. 21. Yang S, Nie A, Zhang L, Yan G, Yao J, Xie L, Lu H, Yang P (2012) A novel quantitative proteomics workflow by isobaric terminal labeling. J Proteome 75(18): Li Z, Adams RM, Chourey K, Hurst GB, Hettich RL, Pan C (2012) Systematic comparison of label-free, metabolic labeling, and isobaric chemical labeling for quantitative proteomics on LTQ Orbitrap velos. J Proteome Res 11(3): Schechter I, Berger A (1967) On the size of the active site in proteases. I Papain. Biochem Biophys Res Commun 27(2): Thiede B, Lamer S, Mattow J, Siejak F, Dimmler C, Rudel T, Jungblut PR (2000) Analysis of missed cleavage sites, tryptophan oxidation and N-terminal pyroglutamylation after in-gel tryptic digestion. Rapid Commun Mass Spectrom 14(6): Switzar L, Giera M, Niessen WM (2013) Protein digestion: an overview of the available techniques and recent developments. J Proteome Res 12(3): Apweiler R, Biswas M, Fleischmann W, Kanapin A, Karavidopoulou Y, Kersey P, Kriventseva EV, Mittard V, Mulder N, Phan I (2001) Proteome analysis database: online application of InterPro and CluSTr for the functional classification of proteins in whole genomes. Nucleic Acids Res 29(1): Keil B (1992) Specificity of proteolysis. Springer, Berlin 28. Liu H, Sadygov RG, Yates JR (2004) A model for random sampling and estimation of relative protein abundance in shotgun proteomics. Anal Chem 76(14): Old WM, Meyer-Arendt K, Aveline-Wolf L, Pierce KG, Mendoza A, Sevinsky JR, Resing KA, Ahn NG (2005) Comparison of label-free methods for quantifying human proteins by shotgun proteomics. Mol Cell Proteomics 4(10): Ning K, Fermin D, Nesvizhskii AI (2012) Comparative analysis of different label-free mass spectrometry based protein abundance estimates and their correlation with RNA-Seq gene expression data. J Proteome Res 11(4): Kyte J, Doolittle RF (1982) A simple method for displaying the hydropathic character of a protein. J Mol Biol 157(1): Atsushi I (1980) Thermostability and aliphatic index of globular proteins. J Biochem 88(6):