J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 1 4 5 4 – 1 4 6 2

Ava i l ab l e on l i ne a t www.sc i enced i r ec t . com

www.e l sev i e r . com/ loca te / j p ro t

Systematic and quantitative comparison of digest efficiencyand specificity reveals the impact of trypsin quality onMS-based proteomics

Julia Maria Burkharta, Cornelia Schumbrutzkia, Stefanie Wortelkampa,Albert Sickmanna, b, René Peiman Zahedia,⁎aLeibniz-Institut für Analytische Wissenschaften - ISAS - e.V., Dortmund, GermanybMedizinisches Proteom-Center (MPC), Ruhr-Universität, Bochum, Germany

A R T I C L E I N F O

Abbreviations: 1-DE, one dimensional gelGu-HCl, guanidinium hydrochloride; FA, form⁎ Corresponding author at: Leibniz-Institut für

Tel.: +49 231 1392 4143; fax: +49 231 1392 485E-mail address: rene.zahedi@isas.de (R.P.

1874-3919/$ – see front matter © 2011 Elseviedoi:10.1016/j.jprot.2011.11.016

A B S T R A C T

Article history:Received 5 September 2011Accepted 14 November 2011Available online 30 November 2011

Trypsin is the most frequently used proteolytic enzyme in mass spectrometry-based prote-omics. Beside its good availability, it also offers some major advantages such as an optimalaverage peptide length of ~14 amino acids, and typically the presence of at least two de-fined positive charges at the N-terminus as well as the C-terminal Arg/Lys, rendering trypticpeptides well suited for CID-based LC-MS/MS. Here, we conducted a systematic study of dif-ferent types of commercially available trypsin in order to qualitatively and quantitativelycompare cleavage specificity, efficiency as well as reproducibility and the potential impacton quantitation and proteome coverage. We present a straightforward strategy applied tocomplex digests of human platelets, comprising (1) digest controls using a monolithic col-umn HPLC-setup, (2) SCX enrichment of semitryptic/nonspecific peptides, (3) targetedMRM analysis of corresponding full cleavage/missed cleavage peptide pairs as well as (4)LC-MS analyses of complete digests with a three-step data interpretation. Thus, differencesin digest performance can be readily assessed, rendering these procedures extremely ben-eficial to quality control not only the trypsin of choice, but also to effectively compare aswell as optimize different digestion conditions and to evaluate the reproducibility of a ded-icated digest protocol for all kinds of quantitative proteome studies.

© 2011 Elsevier B.V. All rights reserved.

Keywords:Trypsindigestionquantitative proteomicsquality controldigest controlSCX

1. Introduction

In the last decade, due to their enormous potential for bio-medical and biochemical research, quantitative strategieshave become increasingly important in proteomic research.For a plethora of biological questions various chemical [1–6]and metabolic labeling [7] as well as label free techniques

electrophoresis; IAA, iodic acid; MRM, multiple rAnalytische Wissenscha0.Zahedi).

r B.V. All rights reserved

[8–12] have been established. In particular improvementsin technical instrumentation [13,14] and software algo-rithms [15,16] have led to increased reproducibility for reli-able and robust quantitative results. However, despitestable and sensitive LC-MS systems and elaborate software,quantitative analyses may fail and yield unreliable resultsin case of inefficient and irreproducible sample preparation.

oacetamide; SCX, strong cation exchange; BCA, bicichinonic acid;eaction monitoring.ften – ISAS - e.V., Otto-Hahn-Str. 6b, D-44227 Dortmund, Germany.

.

1455J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 1 4 5 4 – 1 4 6 2

Generally, sample preparation has to be optimized separate-ly for each specimen with regard to sampling, lysis, fraction-ation and finally proteolytic cleavage. To this end trypsin,which was the first proteolytic enzyme ever discovered – byWilhelm Kühne in 1876 in pancreatic juice [17] – is still themost commonly-used cleavage agent in protein chemistry,and particularly in mass spectrometry-based proteomics. Itcleaves C-terminal to Lys and Arg residues and, in the litera-ture, its specificity has been reported to range from almost100% [18] to significantly lower (~75% based signal intensi-ties) [19], while showing also transpeptidase activity [20].One of its main advantages is the generation of peptideswith an average length of ~14 amino acids (based on an in-silico digestion of the human Uniprot database) usually con-taining at least two defined positive charges: at the N-terminus and the C-terminal Arg/Lys, respectively, render-ing tryptic peptides perfectly suited for typical CID-basedLC-MS/MS analyses. Beside minor side-activities [20,21],trypsin specificity and efficiency can be further influencedby enzyme contamination [22], inappropriate storage [23]as well as inappropriate digestion conditions (temperature,pH, time, presence of protease inhibitors or detergents,etc.) [24].

Since differential studies, in particular, require an optimaland robust digestion of proteins – to improve not only inter-run reproducibility but also overall sequence coverage –proteolytic digestion is one of the limiting steps in peptide-centric quantitative proteomics. Various vendors offertrypsins that can strongly differ in price, activity and sideactivity. Especially for large amounts of protein or extensivesample cohorts, the use of less expensive trypsins might bepreferred to reduce experimental costs. Yet the actual choiceof trypsin - as well as inappropriate storage, shipment ordigestion conditions - can have severe consequences on theoutcome of a proteomics study.

Initiated by an incident in our laboratory, when a newbatch of an inexpensive trypsin which for many years hasbeen successfully utilized for large-scale digests, suddenlyyielded an unexpected but reproducible decrease in digestefficiency as well as specificity, we conducted a systematicstudy comparing six different trypsins (see Table 1 andFig. 1) from different vendors and price ranges.

The main goal was to develop a strategy to assess differ-ences in efficiency, specificity and reproducibility of trypticcleavage and the resulting impact on proteome coverage aswell as quantitation. Hence, our trypsin evaluation comprises(a) digest controls using a monolithic column setup [25], (b)

Table 1 – Overview of the evaluated trypsins.

no. manufacturer ordering no. side activity

1 Promega V5111 none

2 Sigma T-8658 none3 Sigma T-1426 chymotrypsin4 Sigma T-0303 chymotrypsin5 Fluka No longer commercially available6 Sigma 93614 chymotrypsin

global proteome analyses combined with digest-specific datainterpretation, (c) specific analyses of fractions enrichedin semitryptic peptides and (d) quantitative analyses of repro-ducibility and yield of missed cleavage occurrence, based onselected reaction monitoring of specific peptide pairs such asDFGSFDK/DFGSFDKFK (from human superoxide dismutase)and ITIADCGQLE/KITIADCGQLE (from human peptidylprolylisomerase A-like, isoform CRA_c ) in a complex digest ofhuman platelets.

In accordance with Picotti et al. [19], our data indicate, that,in general, all trypsins generate a certain amount of semitryp-tic cleavages, non-specific cleavages and missed cleavages.Consequently, for the trypsins evaluated here, clear differ-ences with respect to specificity, efficiency and reproducibilitycan be detected: interestingly, the prices do not fully correlatewith performance.

Based on our findings, we recommend conducting some ofthese straightforward experiments (a) to evaluate, optimizeand compare digestion conditions in general and (b) to regu-larly monitor the performance of existing or new trypsinstocks before using them for any kind of important/limited/expensive samples.

2. Material and methods

Different trypsins were acquired from Promega (Mannheim,Germany) and Sigma Aldrich (Steinheim, Germany); namelyPromega sequencing grade (Trypsin 1), Sigma T-8658 (Trypsin2), Sigma T-1426 (Trypsin 3), Sigma T-0303 (Trypsin 4), Sigma93614 (Trypsin 6). Finally, we assayed a trypsin from Flukawhich was stored at −30 °C for more than 7 years and whichis no longer commercially available (Trypsin 5). The numbersassigned here will be used throughout the rest of themanuscript.

Ammonium bicarbonate (NH4HCO3), iodoacetamide(IAA), guanidinium hydrochloride (Gu-HCl) and formalde-hyde were purchased from Sigma-Aldrich (Steinheim, Ger-many). All equipment for one-dimensional electrophoresis(1-DE) was obtained from NuPAGE®, Invitrogen (Karlsruhe,Germany). High purity chemicals for silver staining, sodiumdi-hydrogen phosphate (NaH2PO4), di-sodium hydrogenphosphate (Na2HPO4), sodium chloride (NaCl), and calciumchloride (CaCl2) were purchased from Merck (Darmstadt,Germany). Dithiothretiol (DTT) was acquired from RocheDiagnostics (Mannheim, Germany), bicichinonic acid (BCA)assay from Thermo Scientific and Spec C18AR tips from

according to manufacturer remarks

sequencing grade modifiedTrypsin,TPCK treatedsuitable for protein sequencing

≤0.1 units/mg protein TPCK treated≤1.0 units/mg protein

≤0.2% used as a cell-detachment agent(trypsinization of tissue fragments)

Fig. 1 – Overview of the analytical strategy for systematic trypsin evaluation. The same digestion conditions were applied intriplicate to human platelets using the different trypsins. Afterwards, for all digest replicates, efficiency, specificity as well asreproducibility were analyzed by a multi-pronged approach comprising (a) separation on a monolithic column HPLC-setup,LC-MS analyses (b) of complete lysates as well as (c) SCX fractions enriched in semitryptic / non-specific peptides, bothcombined with a three-step database search strategy (trypsin/semitrypsin/non-specific). In addition, selected peptide pairscontaining the same sequence with and without missed cleavage (such as DFGSFDK/DFGSFDKFK) were monitored in all digestreplicates by MRM. Finally, potential contamination of the various trypsins was checked by 1-DE with subsequent LC-MS/MSanalysis of the in-gel digested gel bands.

1456 J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 1 4 5 4 – 1 4 6 2

Agilent (Darmstadt, Germany). All solvents for HPLC wereobtained from Biosolve (Valkenswaard, The Netherlands).

2.1. Platelet isolation and purification

Human platelets from apheresis concentrates were isolatedand purified according to the protocol in Moebius et al. [26].Platelets were resuspended and lyzed in 800 μl 2 M Gu-HCl,50 mM Na2HPO4, pH 7.8. Protein concentration was deter-mined using the BCA assay. Disulfide bonds were reducedwith 10 mM DTT for 30 min at 56 °C. Afterwards free sulfhy-dryl groups were carbamidomethylated using 30 mM IAA for30 min at room temperature in the dark.

2.2. Digest and digest control

Aliquots of 500 μg patelet lysate were treated identicallyusing different trypsins (fixed protease:protein ratio of 1:30).All digests were performed independently in triplicate.Therefore, Gu-HCl was diluted to a concentration of 0.2 Musing 50 mM NH4HCO3, and ACN as well as CaCl2, wereadded to give final concentrations of 5% and 1 mM, respec-tively. Samples were incubated at 37 °C for 8 h while gentlyshaking (350 rpm) on a Thermomixer comfort (Eppendorf,Hamburg, Germany).

Digests were controlled using monolithic column separa-tion (PepSwift monolithic PS-DVB PL-CAP200-PM, Dionex) onan inert Ultimate 3000 HPLC (Dionex, Germering, Germany)by direct injection of 0.1 μg sample. Therefore, a binary gradi-ent (solvent A: 0.1% TFA, solvent B: 0.08% TFA, 84% ACN) rang-ing from 5-12% B in 5 min and then from 12-50% B in 15 min at

a flow rate of 2.2 μl/min and at 60 °C, was applied. UV traceswere acquired at 214 nm.

2.3. SCX chromatography

Prior to strong cation exchange chromatography (SCX) sam-ples were desalted using Spec C18AR tips according to themanufacturer's instructions and dried under vacuum. SCXwas performed using a 150 mm×1 mm PolySULFOETHYLcolumn (200 Å pore size, 5 μm particle size, PolyLC, USA) incombination with an inert Ultimate 3000 HPLC system (Dio-nex, Germering, Germany). Therefore, 100 μg of desaltedsample was resuspended in 5 mM NaH2PO4,pH 2.7 (SCX buff-er A), each. Peptides were eluted at a flow rate of 50 μL/minwith increasing salt content (SCX buffer B: 5 mM Na2HPO4,500 mM NaCl, 15% ACN, pH 2.7). One-minute fractionswere collected and subsequently combined to yield threefractions covering charge states +1 to +2 (SCX 1, SCX 2 andSCX 3), as depicted in Fig. 2. For all eighteen digest protocolsand replicates, 0.6%, 0.75% and 0.2% (v/v), respectively offractions 1, 2 and 3 were analyzed by nano-LC-MS/MS on anLTQ-Orbitrap XL.

2.4. nano-LC-MS/MS

nano-LC-MS/MS was performed on LTQ-Orbitrap XL or LTQ-Orbitrap Velos mass spectrometers (Thermo Fisher Scientific,Bremen, Germany) coupled to Ultimate 3000 Rapid SeparationLiquid Chromatography (RSLC) systems (Dionex, Germering,Germany). Briefly, peptides were preconcentrated on areversed-phase (RP) trapping column (Acclaim PepMap,

Fig. 2 – SCX fractionation of trypsin digests. Already from the UV traces, a substantial increase in the fraction of singly chargedpeptides in SCX 1 is clearly detectable for all trypsins compared to Trypsin 1. Based on triplicate analyses for each trypsin andeach fraction the median proportion as well as corresponding standard deviations (SD) of fully tryptic peptide identificationsrelated to the total number of peptide identifications (tryptic/semitryptic/non-specific) are given. Using Trypsin 1, fractions SCX1 and SCX 2, which are potentially enriched in semitryptic and non-specifically cleaved peptides, contain 84.2±1.4% and 82.8±1.5% fully-tryptic peptides, whereas using Trypsin 6 even fraction SCX 3, which should be enriched in fully-tryptic peptides,contains a share of only 53.2±22.8%. These results strongly indicate a reduced specificity and efficiency of Trypsin 6.SG=sequencing grade.

1457J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 1 4 5 4 – 1 4 6 2

75 μm×2 cm C18, 100 Å, Dionex) in 0.1% TFA followed by RPseparation (Acclaim PepMap RSLC 75 μm×15 cm, 2 μm, 100 Å,Dionex) using a binary gradient (solvent A: 0.1% FA, solventB: 0.1% FA, 84% ACN) from 5% to 50% B at a flow rate of300 nL/min in 50 min, 90 min or 180 min.

MS survey scans were acquired within the Orbitrap fromm/z 300 to 2000 at a resolution of 60,000 using the polysilox-ane ion at m/z 445.120030 as lock mass [27]. The ten (OrbitrapXL: five) most intense signals were subjected to collision in-duced dissociation (CID) in the ion trap taking into account adynamic exclusion of 12 s. CID spectra were acquired with anormalized CE of 35%, a default charge state of 2 and an acti-vation time of 30 ms. AGC target values were set to 106 forOrbitrap MS and 104 for ion trap MSn scans.

Alternatively, fragmentation of the three most intense sig-nals was conducted in the higher energy collision dissociation(HCD) cell of the LTQ-Orbitrap Velos. HCD spectra were ac-quired with a normalized CE of 35%, a default charge state of2 and an activation time of 0.1 ms with a resolution of 7,500.Orbitrap AGC target values were set to 106 for MS and 2*105

for MSn.

2.5. Data interpretation

Data interpretation was accomplished with the help of Prote-ome Discoverer 1.2 (Thermo Scientific). Therefore, all datawere searched with Mascot 2.3.2 (Matrix Science) against thehuman Uniprot database (November 04, 2010; 20,322sequences) using the following settings: (1) trypsin with amaximum of two missed cleavages, (2) carbamidomethyla-tion of cysteine as fixed and (3) phosphorylation of Ser/Thr/Tyr as well as oxidation of Met as variable modifications. Toaccount for charge-reducing peptide modifications, SCX

fractions corresponding to charge state +1 were searchedwith these additional dynamic modifications: (4) acetylationof protein N-termini and (5) N-terminal Glu→pyroGlu aswell as (6) N-terminal Gln→pyroGlu conversion.

In general, to maximize the number of identificationsand to account for non-specific cleavage, database searcheswere conducted in three successive steps: (i) all MS/MSspectra were searched using trypsin as protease. Subse-quently, all spectra not identified thus, were (ii) researchedwith semitryptic cleavage. (iii) Finally, to account for the po-tential occurrence of totally non-specific cleavage events,all spectra not identified in the semitryptic search werethen re-searched using “none” as protease setting. After da-tabase searching, the following filter criteria were appliedto all results: (7) high confidence corresponding to an FDR<1%, (8) ≤4 ppm mass deviation, (9) peptide length >6 and<22 amino acids, (10) at least two unique peptides perprotein.

2.6. MRM

Multiple reactionmonitoring was conducted using a TSQ Van-tage mass spectrometer coupled to an inert Ultimate 3000RSLC nano system as described above, in conjunction withthe Pinpoint software (both from Thermo Scientific). Briefly,Orbitrap Velos HCD data from entire platelet lysates wereused to select sequences which could be detected with andwithout missed cleavage sites. For these peptides, the 4–5most intense ions were chosen to build a specific and sched-uledMRMmethod, using 1.5 s as fixed cycle time and a Q1 res-olution of 0.2. Finally, all digest replicates were analyzedsuccessively with this method and data interpretation wasconducted using Pinpoint.

1458 J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 1 4 5 4 – 1 4 6 2

2.7. Trypsin purity analysis by 1-DE

1 μg of each trypsinwere loaded on a precast 1-DE gel (NUPAGE,Invitrogen). After separation, bands were visualized by silverstaining according to Blum et al. [28] and in-gel digestion of ex-cised gel bands was accomplished as described previously [29].Peptide extracts were analyzed on an LTQ Obritrap XL (49 mingradient, fivemost intense ions for CID) and data interpretationwas done as described above, whereas here the entire Uniprotdatabase (November 04, 2010; 522,019 sequences) was used.

3. Results

To quantitatively compare the different trypsins, we moni-tored efficiency as well as specificity of tryptic digests gener-ated from human platelets using a multi-pronged approach,summarized in Fig. 1. (1) All digests were analyzed using amonolithic column setup allowing for parallel separationand sensitive detection of peptides as well as proteins withina single run, as depicted in Fig. 3. Thus, different proteolyticdigests can be readily and thoroughly compared, which isalso an important prerequisite for any kind of quantitativeproteomics workflow. (2) 1 μg aliquots of each sample weredirectly analyzed by LC-MS/MS in conjunction with a three-step database search as described in the experimental sec-tion. (3) In addition, all digests were separated by SCX at pH2.7, and three pooled fractions covering charge states +1 to+2 were analyzed by nano-LC-MS/MS. Indeed, fraction +1 isof particular interest, since here a large share of the semi-tryptic peptides that usually contain only one positivecharge (at the N-terminus) will be enriched in addition toother charge-reduced peptide species (phosphorylated pep-tides, sialylated glycopeptides, N-terminally acetylated pep-tides, etc.) [30,31]. Since peptides which are charge reduced

Fig. 3 – Monolithic separation of non-digested (black) and digestedof proteolytic digests in general, and to compare the reproducibilitylab all protein digests are monitored via mononlithic column-base

due to post-translational modifications are usually presentsubstoichiometrically, UV signal intensity in the +1 chargestate area of the SCX gradient directly correlates with de-creased digest specificity, as depicted in Fig. 2. (4) Moreover,corresponding peptide pairs (same sequence with 0 and 1missed cleavages) were relatively quantified in all digestsamples by MRM to assess the level of reproducibility aswell as cleavage efficiency for all trypsins. (5) Finally, all eval-uated trypsins were analyzed by 1-DE and subsequent LC-MSanalyses of excised gel bands.

As illustrated in Fig. 3, a monolithic column setup can beutilized for evaluation of digest efficiency and reproducibili-ty. From these separations it can be readily concluded thatmissed cleavage peptide patterns vary substantially betweenthe different trypsins. Whereas Trypsin 6 generated a largeshare of non- or incompletely digested species, indicated byintense peaks at late retention times, Trypsin 1 providedthe highest reproducibility.

To avoid accompanying changes in peptide patterns,for direct nano-LC-MS/MS analyses, platelet digests were ana-lyzed without prior desalting. As illustrated in Fig. 4, to evalu-ate cleavage specificity, we searched LC-MS/MS data threetimes using different protease settings: (1) all MS/MS spectraagainst “Trypsin”, (2) MS/MS not-assigned withTrypsin, against “semitryptic”, (3) MS/MS spectra neitherassigned with Trypsin nor semitryptic-settings, against“none” for non-specific digestion.

(1) Trypsin: Except for Trypsin 6 (with 16,224 spectra) andTrypsin 1 (with 18,592 spectra), the number of MS/MSspectra searched was ~20,000 in most cases. The num-ber of peptide identifications after filtering varied sig-nificantly with Trypsin 1 yielding the highest amountof 7,630 identified peptides (41% of searches), followedby Trypsin 2 with 6,276 peptides (32%) whereas Trypsin

samples. (A) Thismethod can be used tomonitor the efficiencyof digest replicates as is illustrated in (B). Consequently, in ourd HPLC separation prior to further processing.

Fig. 4 – Direct LC-MS analyses of digest replicates using 1 μg of human platelet digest, each. Presented are median values andstandard deviations derived from three replicates. (A) All MS/MS spectra were searched with trypsin as protease usingProteome Discoverer 1.2. Whereas in most cases the number of searches is comparable, the number of identified peptidesremained after stringent filtering (<4 ppmmass deviation, >6 and <22 amino acids, <1% FDR, at least two peptides per protein)differs substantially, thus already indicating an increased occurrence of semi- and non-tryptic cleavages in Trypsins 2–6compared to Trypsin 1, where 41% of MS/MS searches could bematched to fully tryptic events. (B) Consistently, the distributionof peptide as well as protein scores for all six trypsins shows a similar trend, confirming that diminished digestion efficiencyresults not only in a lower abundance for certain peptides, but rather leads to a global reduction of peptide abundance. (C) AllMS/MS spectra which could not be assigned in (A) were then searched with semitryptic as protease setting. Note that thenumber of searches i.e. for Trypsin 1 (5,003) is significantly lower than one would expect from a first glance at the tryptic data(18,592 searches minus 7,630 peptides) in (A). This can be attributed to the stringent post-search filtering: here, 2,627 peptideswere indeed identified as tryptic, however did not meet the criteria. Despite the highest number of searches (9,921), Trypsin 6yields ~50% fewer identifications than Trypsins 4 and 5. (D) All MS/MS spectra that did notmatch the tryptic nor the semitrypticsearches, were finally searched nonspecific. Here, Trypsin 6 yields the highest number of identifications passing the filteringcriteria, with 241 peptides corresponding to 42 proteins, whereas with Trypsin 1 only 2 proteins were identified.

1459J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 1 4 5 4 – 1 4 6 2

6 yielded only 1,399 identified peptides (8%). On the pro-tein level, when taking into account only identificationswith at least two different peptides, Trypsin 1 yielded709 proteins, followed by Trypsin 2 with 612, whereasTrypsin 1 only identified 188 proteins.

(2) Semitryptic: As depicted in Fig. 4c, afore unassignedspectra show an inverse pattern after semitrypticsearch: Whereas Trypsin 1 now yields the lowest num-ber of searches (5,003) in conjunction with significantlylower numbers of semitryptic IDs (322 peptides and 41proteins), Trypsin 6 yields 9,921 searches; with the high-est numbers of identified peptides and proteins around1,200 and 200, respectively for Trypsin 4 and Trypsin 5.

(3) Nonspecific: As demonstrated in Fig. 4d, again the num-ber of searches increases from Trypsin 1 with 3,745 toTrypsin 6 with 6,789, with the number of identified pep-tides (15 for Trypsin 1 to 241 for Trypsin 6) as well asproteins showing the same trends (2 for Trypsin 1 and42 for Trypsin 6).

These results strongly indicate that even when using thesame conditions and the same samples for digestion, cleardifferences can be reproducibly detected between varioustrypsins with the proportion of semitryptic and non-specificpeptides ranging from 4% (Trypsin 1) to 43% of the identifiedpeptides (Trypsin 6).

1460 J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 1 4 5 4 – 1 4 6 2

Moreover, for tryptic searches, average protein and peptidescores correlate with the number of peptide identifications, ascan be seen from Fig. 4b. Thus, digests providing a highernumber of peptide IDs concurrently provide higher scores forthose as well, clearly illustrating considerable differences inprotease performance which already affect qualitative proteo-mic studies. In accordance with these results, clear differ-ences in the charge state-dependant intensity distributioncan be observed as illustrated in Supplemental Fig. 1.

As can be concluded from the non-tryptic cleavage sitesdetected in these analyses, all trypsins have chymotrypticside activity, however to highly varying degree. This can be at-tributed to the autoproteolytic generation of pseudotrypsinhaving Chymotrypsin-like activity [21], which on the otherhand can be compensated by methylation as well as by addi-tion of L-1-tosylamido-2-phenylethyl chloromethyl ketone(TPCK) as Chymotrypsin inhibitor [32].

To assess the extent of direct Chymotrypsin and other pro-tein contamination, we separated 1 μg of Trypsins 1–6 on a 1-DE gel followed by silver staining, excised all visualized bandsand analyzed them by LC-MS subsequent to in-gel digestion.However, we could only identify minor Chymotrypsin con-tamination in Trypsin 6, which also revealed the highest de-gree of chymotryptic side activity.

To further examine the extent of semitryptic cleavage, weseparated all digest replicates using SCX at pH 2.7 and foreach analyzed three distinct fractions by LC-MS/MS, labeledas SCX 1, SCX 2 and SCX 3 in Fig. 2. Strong differences in theUV traces around the area where singly charged peptidesare expected to elute (SCX 1 and SCX 2) already indicate a sig-nificant increase in semitryptic and non-specific cleavageevents for certain Trypsins. In all three fractions, for Trypsin1 the share of fully-tryptic peptides accounts for 84% (SCX 1)to 96% (SCX 2), demonstrating a high specificity even in theSCX 1 and SCX 2 fractions enriched with singly charged pep-tides, whereas the values for all other trypsins are consider-ably lower. Yet, for Trypsin 2 values of 66% and 69%,

Fig. 5 – Summary of MRM results. Here, specific sequences whicmissed cleavage, such as DFGSFDK and DFGSFDKFK were utilizecircles reflect the median areas for all monitored transitions withpeptide areas for all peptides withmissed cleavage and peptideshighest MRM signals, the lowest standard deviation of 24%, whemissed cleavages, corresponding to the overall best specificity, r

respectively can be achieved in fractions SCX 1 and SCX 2.In concordance with the previous results, again Trypsin 6has the lowest performance with only 30% and 37% of fully-tryptic peptides, respectively in fractions SCX 1 and SCX 2,and only 53% even in fraction SCX 3, which usually wouldbe assumed to be enriched in fully-tryptic peptides.

From these results it is straightforward to conclude thatTrypsin 1 yields the highest number of tryptic peptides andidentified proteins while concurrently yielding the lowestnumber of semitryptic and non-specific peptides. However,we wondered whether this might be at the expense of moreextensivemissed cleavages. Therefore, to assess (a) the gener-al impact of the different trypsins on quantitative proteomicstudies and (b) the extent of missed cleavages for the differenttrypsins, we conducted a targeted MRM screening. Hence,based on LTQ Obitrap Velos HCD data from the analyzed di-gests, we selected pairs of peptides with and without missedcleavages that share same sequence parts such as DFGSFDK/DFGSFDKFK and ITIADCGQLE/KITIADCGQLE. As these repre-sent sequences with a generally higher tendency of missedcleavage, we used them as a direct measure of cleavage effi-ciency as well as reproducibility (full list given in supplemen-tal table 1). However, due to different MS response of the twocorresponding partners of a pair, relative inter-pair quantita-tion is not feasible without the use of stable-isotope-labeledinternal standard peptides [33]. We therefore conducted rela-tive quantitation of the single peptides between all digest rep-licates to assess whether peptides with missed cleavages areoverrepresented in certain trypsin samples. Fig. 5 summarizesthe data obtained from these analyses, illustrating thatTrypsin 1 yielded not only the highest signal intensities ingeneral, but also the lowest standard deviations (24%) andthe lowest share of signal intensity contributed to missedcleavage peptides (18% of the total intensity). For all trypsinsevaluated in this study, 0.1-0.2% of the identified peptides de-rived from K-P and R-P bond cleavages. However we refrainedfrom SRM-based quantitation, since the corresponding peptide

h were detected in the platelet digests with and withoutd to set up a specific MRM method. The overall sizes of thein the respective digest replicates, comprising the sum of

withoutmissed cleavage. Thus, Trypsin 1 generally yields thereas 82% of the total area is attributed to peptides withouteproducibility as well as cleavage efficiency.

1461J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 1 4 5 4 – 1 4 6 2

sequences were identified in many different variants includingmissed cleavages and oxidized methionines rendering relativequantitation unfocused.

4. Discussion and conclusion

Reproducible, efficient and specific digestion of protein samplesis among the key factors for successful quantitation in proteo-mics. Especially in complex samples such as cell lysates, slightchanges in cleavage performance can dramatically alter peptidepatterns. Hence, non-specific or inefficient cleavage can sub-stantially increase the number of peptide features while con-currently reducing their average abundances. Thus, not onlywill identification and consequently quantitation rates suffer,but this will also lead, for instance, to complications in thealignment of peptide maps in label-free experiments.

In our group, for many years we successfully utilized an in-expensive trypsin for the digestion of large sample amounts orfor proof-of-principle experiments. However, as could be seenfrom some of our established digest control workflows also de-scribed here, at once a new stock of this trypsin reproduciblygenerated lower amounts of fully tryptic peptideswhile concur-rently increasing the proportion of semitryptic and non-specificpeptides. This dramatic reduction in performance stimulatedus to conduct a systematic comparison of different trypsins,qualitatively and quantitatively evaluating the occurrence ofsemitryptic, non-specific as well as missed cleavage events ina complex digest of human platelets.

In summary, our results are in accordance with Picotti et al.[19] who systematically analyzed a tryptic digest of a single pro-tein, ß-lactoglobulin and reported that fully-tryptic peptidesaccounted for ~75% of the total peptide intensity while manylow intensity peptide signals could be attributed to non-trypticcleavage events; all trypsins evaluated here generated a sub-stantial proportion of not fully-tryptic peptides. In 2004 Olsenet al. reported a specificity of ~100% for trypsin analyzingthe mouse liver proteome [18] – however sample complexity,MS duty cylce and dynamic range in conjunction with data-dependant acquisition might have lead to undersamplingan consequently slight overestimation of enzyme specificitybased on the generally higher intensities of fully-trypticevents, as also reported by Picotti et al.

In the present study, LC-MS/MS analyses of completedigests as well as SCX fractions enriched in semitryptic pep-tides revealed considerable differences in protease perfor-mance which do not fully correlate with price: The mostexpensive trypsin evaluated in this study was only 2nd bestin performance. In all experiments, Sequencing GradePromega Trypsin showed the best performance, however dueto its very low price Sigma T-1426 (~1/1,000th of the cost ofPromega per mg of enzyme) might be an alternative for con-ducting preliminary proof-of-principle experiments or forteaching purposes, whereas Sigma 93614 revealed a consis-tently poor performance and should not be used for any typeof proteomics experiment.

Using stable isotope-labeled internal standard peptides,Proc et al. [24] clearly demonstrated that for specific peptidesthe final yield strongly depends on digest conditions andthat for MRM-based assays of selected peptides this can

have an enormous impact on detection limits and quantita-tion. Using our approach, even without stable isotope-labeled peptides, digest performance can be readily moni-tored without much effort.

Independent of the results presented here, we in generalstrongly recommend the following procedures as routineworkflows for digestion and trypsin control, as they will sub-stantially help to improve the quality as well as the coverageof quantitative studies.

(1) Monolithic column-based separations as routine digestquality control as it is a robust, sensitive, fast and repro-ducible method, which requires only small amounts ofsample (~0.1 μg) and furthermore is compatible with alarge variety of substances that nevertheless might in-terfere with typical LC-MS/MS analyses.

(2) Targeted LC-MS/MS analyses of SCX-enriched semitryp-tic fractions in conjunction with the three-step data-base search strategy, to evaluate cleavage specificity.

(3) Targeted MRM analyses of selected peptide pairs toassess the degree and also reproducibility of missed-cleavage occurrence.

5. Conflict of interest statement

The authors have declared no conflict of interest.

Supporting information

All mgf files of the nano-LC-MS/MS analyses are publiclyavailable via the PRIDE repository (http://www.ebi.ac.uk/pride/), accession number 18826.

Supplementary materials related to this article can befound online at doi:10.1016/j.jprot.2011.11.016.

Acknowledgment

The financial support by the Ministerium für Innovation,Wissenschaft und Forschung des Landes Nordrhein-Westfalen and by the Bundesministerium für Bildung undForschung (MedSys Project SARA, 31P5800) is gratefullyacknowledged.

R E F E R E N C E S

[1] Chakraborty A, Regnier FE. Global internal standardtechnology for comparative proteomics. J Chromatogr A2002;949:173–84.

[2] Dayon L, Hainard A, Licker V, Turck N, Kuhn K, HochstrasserDF, et al. Relative quantification of proteins in humancerebrospinal fluids by MS/MS using 6-plex isobaric tags. AnalChem 2008;80:2921–31.

[3] Fauq AH, Kache R, Khan MA, Vega IE. Synthesis ofacid-cleavable light isotope-coded affinity tags (ICAT-L) forpotential use in proteomic expression profiling analysis.Bioconjug Chem 2006;17:248–54.

1462 J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 1 4 5 4 – 1 4 6 2

[4] Gygi SP, Rist B, Gerber SA, Turecek F, Gelb MH, AebersoldR. Quantitative analysis of complex protein mixturesusing isotope-coded affinity tags. Nat Biotechnol 1999;17:994–9.

[5] Ross PL, Huang YN, Marchese JN, Williamson B, Parker K,Hattan S, et al. Multiplexed protein quantitation insaccharomyces cerevisiae using amine-reactive isobarictagging reagents. Mol Cell Proteomics 2004;3:1154–69.

[6] Schmidt A, Kellermann J, Lottspeich F. A novel strategy forquantitative proteomics using isotope-coded protein labels.Proteomics 2005;5:4–15.

[7] Ong SE, Blagoev B, Kratchmarova I, Kristensen DB, Steen H,Pandey A, et al. Stable isotope labeling by amino acids incell culture, SILAC, as a simple and accurate approach toexpression proteomics. Mol Cell Proteomics 2002;1:376–86.

[8] Zhang G, Neubert TA. Automated comparative proteomicsbased on multiplex tandem mass spectrometry and stableisotope labeling. Mol Cell Proteomics 2006;5:401–11.

[9] Wang W, Zhou H, Lin H, Roy S, Shaler TA, Hill LR, et al.Quantification of proteins and metabolites by massspectrometry without isotopic labeling or spiked standards.Anal Chem 2003;75:4818–26.

[10] Vogel C, Marcotte EM. Calculating absolute and relativeprotein abundance from mass spectrometry-based proteinexpression data. Nat Protoc 2008;3:1444–51.

[11] Lange V, Picotti P, Domon B, Aebersold R. Selected reactionmonitoring for quantitative proteomics: a tutorial. Mol SystBiol 2008;4:222.

[12] Ishihama Y, Oda Y, Tabata T, Sato T, Nagasu T, Rappsilber J,et al. Exponentially modified protein abundance index(emPAI) for estimation of absolute protein amount inproteomics by the number of sequenced peptides per protein.Mol Cell Proteomics 2005;4:1265–72.

[13] Zubarev R, Mann M. On the proper use of mass accuracy inproteomics. Mol Cell Proteomics 2007;6:377–81.

[14] Kocher T, Swart R, Mechtler K. Ultra-high-pressure RPLChyphenated to an LTQ-Orbitrap Velos reveals a linear relationbetween peak capacity and number of identified peptides.Anal Chem 2011;83:2699–704.

[15] Cox J, Mann M. MaxQuant enables high peptide identificationrates, individualized p.p.b.-range mass accuracies andproteome-wide protein quantification. Nat Biotechnol2008;26:1367–72.

[16] Sturm M, Bertsch A, Gropl C, Hildebrandt A, Hussong R,Lange E, et al. OpenMS - an open-source softwareframework for mass spectrometry. BMC Bioinformatics2008;9:163.

[17] Kühne WF. Über die Verdauung der Eiweisstoffe durch denPankreassaft. Virchows Arch 1876;39.

[18] Olsen JV, Ong SE, Mann M. Trypsin cleaves exclusivelyC-terminal to arginine and lysine residues. Mol CellProteomics 2004;3:608–14.

[19] Picotti P, Aebersold R, Domon B. The implications ofproteolytic background for shotgun proteomics. Mol CellProteomics 2007;6:1589–98.

[20] Schaefer H, Chamrad DC, Marcus K, Reidegeld KA, Bluggel M,Meyer HE. Tryptic transpeptidation products observed inproteome analysis by liquid chromatography-tandem massspectrometry. Proteomics 2005;5:846–52.

[21] Keil-Dlouha VV, Zylber N, Imhoff J, Tong N, Keil B. Proteolyticactivity of pseudotrypsin. FEBS Lett 1971;16:291–5.

[22] Chamrad I, Strouhal O, Rehulka P, Lenobel R, Sebela M.Microscale affinity purification of trypsin reducesbackground peptides in matrix-assisted laserdesorption/ionization mass spectrometry of protein digests.J Proteomics 2011;74(7):948–57.

[23] Kunitz M, Northrop JH. Inactivation of Crystalline Trypsin.J Gen Physiol 1934;17(4):591–615.

[24] Proc JL, Kuzyk MA, Hardie DB, Yang J, Smith DS, Jackson AM,et al. A quantitative study of the effects of chaotropic agents,surfactants, and solvents on the digestion efficiency of humanplasma proteins by trypsin. J Proteome Res 2010;9:5422–37.

[25] Ivanov AR, Zang L, Karger BL. Low-attomole electrosprayionization MS and MS/MS analysis of protein tryptic digestsusing 20-microm-i.d. polystyrene-divinylbenzene monolithiccapillary columns. Anal Chem 2003;75:5306–16.

[26] Moebius J, Zahedi RP, Lewandrowski U, Berger C, Walter U,Sickmann A. The human platelet membrane proteomereveals several new potential membrane proteins. Mol CellProteomics 2005;4:1754–61.

[27] Olsen JV, de Godoy LM, Li G, Macek B, Mortensen P, Pesch R,et al. Parts per million mass accuracy on an Orbitrap massspectrometer via lock mass injection into a C-trap. Mol CellProteomics 2005;4:2010–21.

[28] Blum H, Beier H, Gross HJ. Improved silver staining of plantproteins, RNA and DNA in polyacrylamide gels.Electrophoresis 1987;8:93–9.

[29] Winkler C, Denker K, Wortelkamp S, Sickmann A. Silver- andCoomassie-staining protocols: detection limits andcompatibility with ESI MS. Electrophoresis 2007;28:2095–9.

[30] Mohammed S, Heck Jr A. Strong cation exchange (SCX) basedanalytical methods for the targeted analysis of proteinpost-translational modifications. Curr Opin Biotechnol 2011;22:9–16.

[31] Lewandrowski U, Zahedi RP, Moebius J, Walter U, SickmannA. Enhanced N-glycosylation site analysis ofsialoglycopeptides by strong cation exchangeprefractionation applied to platelet plasma membranes. MolCell Proteomics 2007;6:1933–41.

[32] Kostka V, Carpenter FH. Inhibition of Chymotrypsin Activity inCrystalline Trypsin Preparations. J Biol Chem 1964;239:1799–803.

[33] Gerber SA, Rush J, Stemman O, Kirschner MW, Gygi SP. Absolutequantification of proteins and phosphoproteins from cell lysatesby tandemMS. Proc Natl Acad Sci USA 2003;100:6940–5.