Impact of Digestion Conditions on Phosphoproteomics - Journal of

Apr 14, 2014 - High-throughput phosphoproteomics reveals in vivo insulin signaling dynamics. Sean J Humphrey , S Babak Azimifar , Matthias Mann...
1 downloads 0 Views 2MB Size
Article pubs.acs.org/jpr

Impact of Digestion Conditions on Phosphoproteomics Clarissa Dickhut, Ingo Feldmann, Jörg Lambert, and René P. Zahedi* Leibniz-Institut für Analytische Wissenschaften - ISAS - e.V., Dortmund 44227, Germany ABSTRACT: In the past few years, the focus of phosphoproteomics has shifted from merely qualitative to quantitative and targeted studies. Tryptic digestion is a critical step that directly affects quantification and that can be impaired by phosphorylation. Therefore, we systematically characterized the digestion efficiency of 19 nonmodified and phosphorylated model peptides. Whereas we quantified a strong reduction of tryptic cleavage within phosphorylated PKA motifs (R)-R-XpS/pT and also R-X-X-pT sequences, (R)-R-X-pY sequences were almost unaffected. Structural prediction implied the formation of salt bridges between R/K cleavage sites and phosphoamino acids pS/pT as the main reason for impaired tryptic digestion. We evaluated different conditions to optimize the digestion of such “resistant” phosphopeptides, yielding a substantial improvement of digestion efficiency. We performed a quantitative large-scale phosphoproteomic analysis of human platelets to validate our findings in a complex biological sample. Here, increasing trypsin concentrations up to a trypsin to peptide ratio of 1:10 led to a significant gain (i) in the overall number of phosphorylation sites (up to 9%) and (ii) in the intensities of individual phosphopeptides, thereby improving the sensitivity of phosphopeptide quantification. Still, for certain sequences, the negative impact of phosphorylation on digestion efficiency will further complicate the analysis of phosphorylation stoichiometry. KEYWORDS: phosphoproteomics, digestion, sample preparation



residues close to cleavage sites8−11 as well as PTMs12,13 are known to impair proteolysis. The interfering effect of phosphoserine (pS) and phosphothreonine (pT) residues nearby cleavage sites was first proposed by Lehmann and co-workers in 2001,12 observing incomplete tryptic digestion of protein kinase A (PKA) within the phosphorylated PKA consensus sequence R-X-pS. Later, they demonstrated that the majority of missed cleavage sites in a previously published large-scale phosphopeptide data set14 could be attributed to either negatively charged residues of aspartic and glutamic acid or pS and pT in close proximity to the proteolytic cleavage site.13 To gain closer insights into the impact of phosphorylation on tryptic digestion, we systematically characterized the digestion efficiency of model peptide sequences that are known to be prone to incomplete digestion. Hence, we selected two peptide sequences, containing the PKA consensus sequence (R)-R-X-S, and synthesized the respective pS/pT/pY variants of these peptides. In addition, we changed the position of the phosphoamino acids in relation to the cleavage site, leading to the synthesis of 19 different peptide sequences with internal tryptic cleavage sites. All synthesized peptide sequences are summarized in Table 1. Using these peptides, different digestion conditions, comprising changes in buffer composition, trypsin to peptide ratios, and incubation times, were evaluated by LC−MS.

INTRODUCTION

Protein phosphorylation is one of the most common posttranslational modifications (PTMs), regulating the localization and activity of proteins as well as protein−protein interactions in numerous cellular processes. Due to the development of specific phosphopeptide enrichment techniques and highly sensitive MS instruments, phosphoproteomics has enabled researchers to gain a comprehensive view on the dynamics of protein phosphorylation and phosphorylationbased signaling networks.1,2 An important though often underestimated aspect in proteomic studies is sample preparation prior to LC−MS analysis.3 Especially for quantitative and targeted proteomics, a thorough and efficient sample preparation, in regard to cell lysis, protein purification, and proteolytic digestion is essential for the reliability of the obtained findings. Incomplete digestion and unspecific hydrolysis of proteins may lead to an underestimation of protein abundance, impairing not only the quantification of a single protein but also the reproducibility and accuracy of the entire study.4 Due to its high cleavage specificity, trypsin is the commonly used proteolytic enzyme in MS-based proteomics, cleaving peptides carboxyterminal of the amino acids lysine and arginine. Thereby, mainly doubly charged peptides with an average length of around 14 amino acids are generated, which are well suited for MS analysis.4−6 However, various factors such as the tertiary structure of a protein,7 linear sequence motifs such as adjacent basic amino acids or negatively charged © 2014 American Chemical Society

Received: November 30, 2013 Published: April 14, 2014 2761

dx.doi.org/10.1021/pr401181y | J. Proteome Res. 2014, 13, 2761−2770

Journal of Proteome Research

Article

Table 1. Summary of the Synthesized Peptidesa peptide sequenceb

position of phosphoamino acid

AAAR|LSLTDPLVAER AAAR|LpSLTDPLVAER AAAR|LpTLTDPLVAER AAAR|LpYLTDPLVAER AAAR|LSLpSDPLVAER AAAR|LSLpTDPLVAER AAAR|LSLpYDPLVAER LSLTDPLVAER|AGTDESR LSLTDPLVAER|AGpTDESR LSLpSDPLVAER|AGTDESR LSLpTDPLVAER|AGTDESR LSLpYDPLVAER|AGTDESR LpSLTDPLVAER|AGTDESR LpTLTDPLVAER|AGTDESR LpYLTDPLVAER|AGTDESR QQR|R|GSLPEISNLR QQR|R|GpSLPEISNLR QQR|R|GpTLPEISNLR QQR|R|GpYLPEISNLR

P2′ P2′ P2′ P4′ P4′ P4′ P3′ P7 P7 P7 P9 P9 P9 P2′/P3′ P2′/P3′ P2′/P3′

(NaCl) from Merck (Darmstadt, Germany). Bicinchoninic acid (BCA) assay was acquired from Thermo Scientific (Schwerte, Germany), iTRAQ Reagents Multiplex Kit (4plex) from AB Sciex (Darmstadt, Germany), titanium dioxide beads (Titansphere, 5 μm) from GL Sciences (Tokyo, Japan), and Oligo R3 bulk media from Life technologies (Darmstadt, Germany). All UPLC-grade solvents were obtained from Biosolve (Valkenswaard, The Netherlands). Synthesis and Purification of Model Peptides

Model peptides were synthesized in-house using a Syro I synthesis unit (MultiSynTech, Witten, Germany) and Fmoc chemistry. For peptides containing phosphoamino acids the respective phosphoamino acid building blocks were employed. To evaluate the inhibitory effects of phosphoamino acids close to cleavage sites, 19 model peptides containing internal missed cleavage sites were synthesized. Most sequences were adopted from the C-terminal sequence 187AAARLSLTDPLVAERAGTDES206 of the human platelet glycoprotein 1b (GP1b, P13224), which contains the internal PKA consensus motif R-L-S and is prone to missed cleavage with trypsin.18,19 Based on the respective missed-cleavage phosphopeptides AAAR|LpSLTDPLVAER and LpSLTDPLVAER|AGTDESR (pS indicating a phosphoserine, whereas | indicates a potential cleavage site), peptide variants containing pT and pY instead of pS were synthesized. In addition, the position of the phosphoamino acid was shifted within the sequences. Furthermore, the artificial peptide sequence QQR|R|GSLPEISNLR (based on the PKA human sequence logo, as obtained from www.phosphosite.org,20 April 2012) containing two tryptic cleavage sites was synthesized in phosphorylated (phosphorylation at P2′/P3′) and nonphosphorylated form (summarized in Table 1). All synthetic peptides were purified by RP-HPLC on an U3000 HPLC system using the U3000 fraction collector. Peptides were separated on a Luna 10 μm C18 column (20 × 250 mm, Phenomenex, Aschaffenburg, Germany) at a flow rate of 20 mL/min using a binary gradient of solvent A (0.1% TFA) and solvent B (0.1% TFA, 84% ACN) from 5% to 55% B in 25 min. The collected fractions were controlled using an FT-ICRMS (Thermo Scientific, Bremen, Germany) equipped with an TriVersa Nanomate (Advion, Ithaca, USA) for direct infusion of the samples into the mass spectrometer. Lyophilized peptides were dissolved in water and used for subsequent digestion studies.

a

Nineteen missed-cleavage model peptides were synthesized to evaluate the impact of phosphorylation on tryptic digestion in a semiquantitative approach. b| indicates a potential cleavage site.

In accordance with the work of the Lehmann group,12,13 our results indicate that pS and pT residues in proximity to cleavage sites strongly hamper trypsin digestion efficiency, whereas pY had a smaller effect. Increasing trypsin concentration from an enzyme to peptide ratio of 1:50, as widely used in (phospho-) proteomic studies,15,16 to a ratio of 1:20, as well as the addition of acetonitrile (ACN) and trifluoroethanol (TFE), significantly improved digestion efficiency of “resistant” phosphopeptides, which contain a pS or pT at positions P2′ or P3′ (nomenclature according to Schechter and Berger;17 see Figure 1). We could confirm these results in a large-scale phosphoproteomic analysis of human platelets, yielding an improved overall performance of the quantification of phosphopeptides, as a consequence of an improved sensitivity and reproducibility.



MATERIALS

Materials and Reagents

Guanidine hydrochloride (GuHCl), iodoacetamide (IAA), calcium chloride (CaCl2), ammonium bicarbonate (ABC), triethylammonium bicarbonate (TEAB), and TFE were purchased from Sigma-Aldrich (Steinheim, Germany). Trypsin Gold, MS grade, was obtained from Promega (Mannheim, Germany). Dithiothreitol (DTT) and benzonase were acquired from Roche Diagnostics (Mannheim, Germany). Sodium dodecyl sulfate (SDS) was purchased from Roth (Karlsruhe, Germany), tris(hydroxymethyl)-aminomethane (Tris) from Applichem (Darmstadt, Germany), and sodium chloride

Tryptic Digestion of Model Peptides

To characterize the effect of phosphorylation on proteolytic digestion, 1 μg of each purified model peptide was digested in 20 μL of digestion buffer (1 mM CaCl2 and 50 mM ABC, pH 7.8) for 16 h at 37 °C using trypsin to peptide ratio of 1:50 (w/ w) (standard conditions). On the basis of the obtained results, different strategies to improve the digestion of resistant phosphopeptides were evaluated: (a) the addition of organic solvents to a final

Figure 1. Nomenclature of enzymatic cleavage according to Schechter and Berger. The peptide sequence is cleaved between position P1 and P1′. 2762

dx.doi.org/10.1021/pr401181y | J. Proteome Res. 2014, 13, 2761−2770

Journal of Proteome Research

Article

peptides were separated on a self-packed TSKgel Amide80 column (250 μm × 15 cm, Tosoh Bioscience, Stuttgart, Germany) using a binary gradient (solvent A: 98% ACN, 0.1% TFA; solvent B: 0.1% TFA) from 1% to 40% B at a flow rate of 4 μL/min in 35 min, followed by 15 min at 80% B. As phosphopeptides elute in the later part of the gradient, fractions were collected every 3 min starting at 20% B. For LC−MS analysis, each fraction was dried and dissolved in 15 μL of 0.1% TFA.

concentration of 10% ACN and 5% TFE, respectively, employing (b) increased trypsin concentrations and (c) different incubation times of 4 h, 8 and 16 h. After digestion, trypsin activity was quenched by the addition of TFA to a final concentration of 1%. All digests were performed independently in duplicate or triplicate. Digestion efficiency was controlled by nano-LC−MS/MS. Preparation of Platelet Samples

Platelets were prepared from human blood as described previously.21 Briefly, platelets were lysed in 1% SDS, 150 mM NaCl, 50 mM Tris (pH 7.8), followed by a determination of the protein concentration using the BCA assay. Disulfide bonds were reduced with 10 mM DTT for 30 min at 56 °C. Afterward, free sulfhydryl groups were alkylated using 30 mM IAA for 30 min at RT in the dark. For ethanol precipitation, the sample was diluted 10-fold with cold (−40 °C) ethanol and incubated for 1 h at −40 °C, followed by centrifugation at 20,000g for 30 min at 4 °C. After removal of the supernatant, the pellet was dissolved in 4 M GuHCl and diluted to a final concentration of 0.2 M GuHCl using 50 mM TEAB. Different conditions for tryptic digestion were evaluated in triplicate: (i) To test the influence of different trypsin to peptide ratios, we digested 100 μg of platelets using trypsin to peptide ratios of 1:10, 1:20, 1:50, and 1:100 (w/w), respectively. (ii) Furthermore, we digested 20 μg of platelets in the presence of 5% TFE, using trypsin to peptide ratios of 1:50 and 1:20. Digestion efficiency was controlled by monolithic RP separation as described previously.4 In the case of (i), 25 μg per triplicate was pooled to generate a multiplexed sample of 75 μg for each trypsin ratio. The four resulting samples were labeled with different 4-plex iTRAQ reagents according to the manufacturer’s instructions. Moreover, the remainder of the 1:10, 1:20, 1:50, and 1:100 samples (75 μg × 12 samples) were used to perform a quantitative label-free phosphopeptide analysis. In the case of (ii), triplicates were pooled after digestion to generate three multiplexed samples (60 μg each), which were used for an iTRAQ-based approach. For all approaches, phosphopeptides were enriched using titanium dioxide, based on a protocol published by Larsen and co-workers.22 First, all samples were dried under vacuum, resuspended in 1 mL of TiO2 loading buffer (80% ACN, 5% TFA, 1 M glycolic acid), and incubated with TiO2 beads for 10 min in two incubation steps, using a peptide to bead ratio of 1:6 in the first and a ratio of 1:3 in the second step. Afterward, beads of both incubation steps were combined in one microfuge tube. Washing and elution steps were conducted as described previously.22 To remove residual nonphosphorylated peptides, TiO2 eluates from the first enrichment procedure were dried under vacuum, dissolved in 70% ACN/1% TFA and incubated with another round of TiO2 beads, using the same peptide to bead ratio as in the first procedure. Afterward, beads were washed with 50% ACN/0.1% TFA and bound phosphopeptides were eluted by incubating with 1% NH4OH for 10 min. Subsequently, eluates were acidified using formic acid (pH < 2) and desalted with Oligo R3 microcolumns.23 The nonlabeled phosphopeptide samples were dried completely, dissolved in 15 μL of 0.1% TFA and directly measured by LC− MS. The enriched iTRAQ samples were further fractionated by hydrophilic interaction liquid chromatography (HILIC) using a nano RSLC HPLC system (Thermo Scientific). Briefly,

Nano-LC−MS/MS

Model peptides were analyzed on a LTQ-Orbitrap XL (Thermo Scientific), whereas the analysis of the nonlabeled platelet samples was performed on an Orbitrap Elite (Thermo Scientific). Both MS systems were online coupled to a nano RSLC HPLC (Thermo Scientific). LC−MS of the iTRAQlabeled samples was conducted on a LTQ-Orbitrap Velos mass spectrometer (Thermo Scientific) online coupled to an U3000 HPLC system (Thermo Scientific). For the nano RSLC systems, samples were loaded onto a trap column (Acclaim C18 PepMap100, 100 μm × 2 cm, Thermo Scientific) at a flow rate of 20 μL/min 0.1% TFA, followed by separation on a RP main column (Acclaim C18 PepMap100, 75 μm × 15 or 50 cm, Thermo Scientific) using a binary gradient of solvent A (0.1% FA) and solvent B (0.1% FA, 84% ACN). Gradients increased linearly from 5% to 50% B at a flow rate of 300 nL/min in 50 min (model peptides) or from 3% to 45% B at a flow rate of 250 nL/min in 90 min (label-free analysis). For the U3000 system, samples were loaded onto a selfpacked trap column (Kinetex 2.6 μm C18, 100 μm × 2 cm, Phenomenex) at a flow rate of 8 μL/min and subsequently separated on a self-packed main column (Kinetex 2.6 μm C18, 75 μm × 30 cm, Phenomenex) from 3% to 45% B at a flow rate of 230 nL/min in 90 min. The LTQ-Orbitrap XL and the Orbitrap Elite were operated in data-dependent acquisition mode acquiring full MS survey scans at R = 60,000, followed by collision-induced dissociation (CID) of the 5 (model peptides) or 15 (label-free analysis) most abundant ions with a normalized collision energy of 35. Target values and maximum fill times of the LTQ-Orbitrap XL were set to 5 × 105 ions and 500 ms for MS and 1 × 104 ions and 200 ms for MS/MS, whereas 1 × 106 ions and 100 ms for MS and 1 × 104 ions and 100 ms for MS/MS were used on the Orbitrap Elite. On the LTQ-Orbitrap Velos (iTRAQ samples), fragmentation of the 5 most intense signals was conducted using higherenergy collision dissociation (HCD), with a normalized collision energy of 50. A vial with 10% NH4OH solution was placed in front of the ion source for charge state reduction of iTRAQ-labeled peptides, as described by Thingholm et al.24 MS scans were acquired at R = 30,000, a target value of 1 × 106 and a maximum fill time of 100 ms, followed by MS/MS scans at R = 7500, a target value of 5 × 104 and a maximum fill time of 200 ms. For all instruments, only precursor ions with charge states between +2 and +5 were selected for fragmentation. Data Interpretation

Model peptide raw data were searched against an in-house database containing ragged versions of the synthetic peptides as well as potential byproducts (391 sequences; 78,389 residues) using Mascot 2.4.1 (Matrix Science). As a precise determination of digestion efficiency is either expensive (using stable 2763

dx.doi.org/10.1021/pr401181y | J. Proteome Res. 2014, 13, 2761−2770

Journal of Proteome Research

Article

Figure 2. Schematic of the model peptide digestion workflow. In-house synthesized and purified peptides containing internal cleavage sites were processed using different digestion conditions. Tryptic digests were analyzed by LC−MS, and semiquantitative data interpretation strategies (ηpsm, ηxic) were used to approximate digestion efficiency.

with a phosphoRS site probability ≥75% for all phosphorylation sites were considered in the data interpretation. Furthermore, for the label-free analysis, only phosphopeptides and phosphorylation sites detected in at least two of three replicates were included in the results, to ensure a high reproducibility of the findings. To compensate for experimental/systematic errors in the iTRAQ-labeled phosphopeptide samples, we determined normalization factors on the PSM level based on the LC−MS analysis of the nonenriched samples.27 3D peptide structures were predicted using the PEP-FOLD de novo peptide structure prediction tool (http://bioserv.rpbs. univ-paris-diderot.fr/PEP-FOLD,28), using only the best scoring model per sequence. The generated models were illustrated using the OpenAstexViewer 3.0 (http:// openastexviewer.net/web).

isotope-labeled synthetic standards) or laborious (analyzing all samples prior and after digestion using label-free quantification), we used two semiquantitative strategies for an approximation of digestion efficiency. First, we used a spectral counting approach, comparing the number of peptide spectrum matches (PSM, ηpsm) for both the fully tryptic and the missedcleavage form of the respective peptide (eq 1). Second, we used the respective peptide areas, as derived from the extracted ion chromatograms (XIC, ηxic) (eq 2). ηpsm =

no. fully trypitc PSM no. (fully tryptic PSM + missed − cleavage PSM) (1)

ηxic =

area fully tryptic peptide area (fully tryptic peptide + missed − cleavage peptide)



(2)

For the platelet samples, data were searched against the human Uniprot database (August 2012; 20,232 target sequences) using Mascot 2.4.1 and Proteome Discoverer (version 1.3, Thermo Scientific), including the reporter ions quantifier node (only for the iTRAQ-labeled samples) and the precursor ions area detector node (only for the nonlabeled samples) as well as the Percolator25 and phosphoRS26 (version 3.1) nodes for both strategies. All workflows employed the following search parameters: trypsin cleavage specificity with a maximum of 2 missed cleavages, phosphorylation of S/T/Y and oxidation of M as variable modifications, carbamidomethylation of C as fixed modification, mass tolerances of 10 ppm for MS, 0.5 Da (CID) and 0.05 Da (HCD) for MS/MS, respectively. Additionally, iTRAQ 4-plex labeling on K, Y, and N-term was set as a fixed modification for the stable isotope labeled samples. High confidence corresponding to a FDR 99% efficiency). The nonmodified sequence QQR|R|GSLPEISNLR was cleaved approximately 1:1, with a slight preference for the missedcleavage form R|GSLPEISNLR (57%) whereas the corresponding phosphorylated variants had a clear preference for R| GSLPEISNLR. Thus, less than 3% of the peptide sequences QQR|R|GpSLPEISNLR and QQR|R|GpTLPEISNLR were digested to the corresponding fully tryptic peptides GpSLPEISNLR and GpTLPEISNLR, whereas the residual amount of both peptides was processed to the missed-cleavage forms RGpSLPEISNLR and R|GpTLPEISNLR. In accordance with our previous results, pY at P2′ (QQR|R|GpYLPEISNLR) had a reduced influence on tryptic digestion. However, compared to the nonmodified peptide variant QQR|R|GSLPEISNLR, the presence of pY led to an increase of the missed-cleavage form R|GpYLPEISNLR to 80%, compared to 57% for the nonmodified sequence. This result could indicate a slight sequence dependent inhibitory effect of pY at P2′, which is more pronounced in the presence of consecutive cleavage sites and thus leads to a preferential cleavage most distant from pY. As it can be expected that the impaired digestion efficiency of phosphopeptides containing R-X-pS/pT is due to structural changes, we used a de novo peptide structure prediction tool (PEP-FOLD) to modulate the 3D structures of the nonmodified and phosphorylated forms of AAAR|LSLTDPLVAER and QQR|R|GSLPEISNLR (AAAR|LTLTDPLVAER and QQR|R|GTLPEISNLR, respectively). As PEP-FOLD is limited to the 20 standard amino acids, we substituted aspartate for pS and pT to mimic the effect of phosphorylation at P2′.29 Here, phosphorylation leads to conformational changes such that the positively charged arginine and the negatively charged aspartate residues are in close proximity and can form salt bridges (see Figure 4). In summary, our analyses demonstrate that under standard conditions peptide sequences containing phosphorylated PKA sites (R)-R-X-pS/pT are resistant to tryptic cleavage, as on average less than 10% of these peptides were detected in their fully tryptic form. Notably, the presence of pY (R-X-pY) as well as phosphorylation at P3′ (R-X-X-pT) had a reduced inhibitory effect on tryptic digestion. Our results confirm that the resistance of such phosphorylated sequences to tryptic cleavage might be caused by an impaired accessibility of the tryptic cleavage site due to a potential electrostatic interaction of arginine/lysine and the phosphoamino acid.

Figure 3. Digestion efficiency of the synthesized model peptide sequences. Each missed-cleavage peptide, based on either a sequence from human GP1b (A) or an artificial sequence obtained from phosphosite.org (B), was digested using a trypsin to peptide ratio of 1:50 (in triplicate). Digestion efficiency was evaluated by comparing the number of PSM (violet) and the peptide areas (red) for the missed-cleavage and the fully tryptic forms of the respective peptides.

peptide sequence was digested to the corresponding fully tryptic form, compared to 93% digestion efficiency of the nonphosphorylated variant. To analyze whether the distance between the phosphoamino acid and the cleavage site has an effect, we synthesized peptide sequences phosphorylated at P3′ (LSLTDPLVAER|AGpTDESR) and P4′ (AAAR|LSLpTDPLVAER, and accordingly pS and pY). Whereas phosphorylation at P3′ still hampered tryptic digestion, resulting in only partial digestion of the missed-cleavage peptide (∼50%), peptide sequences phosphorylated at P4′ were completely processed to their fully tryptic form. Furthermore, we included nonmodified sequences and peptide sequences with phosphoamino acids at a greater distance from the cleavage site (P7 and P9) as digestion controls. Almost complete digestion (93%) was monitored for the nonphosphorylated peptide sequence AAAR|LSLTDPLVAER, whereas slightly lower digestion efficiencies (∼80%) were yielded for peptides derived from the sequence LSLTDPLVAER|AGTDESR. Most likely, this effect can be attributed to the inhibitory effect of glutamic acid in proximity to the

Evaluating the Impact of Different Digestion Conditions

On the basis of our findings, we wondered whether changing standard conditions would increase the digestion efficiency for resistant phosphopeptide sequences. Though both applied data interpretation strategies (ηpsm, ηxic) yielded comparable results, we decided to use to the ηxic approach for the following experiments, as this strategy is less affected by instrument settings (such as dynamic exclusion) and chromatographic properties of the analytes. As previous studies have reported the benefit of organic solvents for tryptic digestion in global analyses,30,31 we performed digestions in the presence of 10% ACN or 5% 2765

dx.doi.org/10.1021/pr401181y | J. Proteome Res. 2014, 13, 2761−2770

Journal of Proteome Research

Article

digestion under standard conditions nor for ACN and TFE digestion, respectively. Both additives could not improve the digestion of the peptide sequence QQR|R|GpSLPEISNLR, which was still almost quantitatively cleaved into the missed-cleavage form R| GpSLPEISNLR (>99%), compared to the ∼1:1 ratio for the nonphosphorylated form. Besides the addition of organic compounds, we evaluated the effect of an increased trypsin concentration on the digestion efficiency of resistant phosphopeptide sequences. Using a trypsin to peptide ratio of 1:20 led to a nearly complete digestion of peptide sequences containing pY at P2′ (99% efficiency vs 90% for the standard digestion) and pT at P3′ (99% efficiency vs 49%) as well as a partial digestion of AAAR| LpSLTDPLVAER (31% vs 1%) and AAAR|LpTLTDPLVAER (49% vs 4%) to their fully tryptic forms (see Figure 6A). Notably not only did the relative share of the fully tryptic form increase, but also the absolute amount. Thus, the detected increase was not an artifact derived from unspecific cleavage,

Figure 4. Visualization of conformational changes occurring upon phosphorylation. Phosphorylation of serine at P2′ (mimicked by aspartate) induced a conformational change in the nonmodified sequence AAAR|LSLTDPLVAER (A), resulting in a close proximity of the positively charged arginine and the negatively charged aspartate residues (B). Similar changes were observed for pS at P2′ in QQR| R|GSLPEISNLR (C,D). Nitrogen (blue) and oxygen (red) atoms are labeled in the relevant amino acid residues.

TFE under otherwise standard digestion conditions. Both solvents increased the digestion efficiency of most resistant phosphopeptide sequences. However, using ACN, the overall digestion efficiency of the peptide sequences AAAR|LpSLTDPLVAER and AAAR|LpTLTDPLVAER was still below 11% (see Figure 5). In contrast, the addition of TFE improved the

Figure 5. Addition of organic solvents improved digestion efficiency. A final concentration of 10% ACN and 5% TFE, respectively, increased the digestion efficiency of most resistant phosphopeptides compared to a digestion without additives (trypsin to peptide ratio 1:50, duplicate analysis). Figure 6. An increase in trypsin concentration improved digestion efficiency. (A) Using a trypsin to peptide ratio of 1:20 instead of 1:50 was beneficial for the digestion of most resistant phosphopeptides, yielding a significant increase in digestion efficiency. (B) The addition of 5% TFE further increased the digestion efficiency of the highly resistant phosphopeptides containing pS and pT at P2′. Still, the digestion of the phosphorylated variants of QQR|R|GSLPEISNLR to their corresponding fully tryptic forms could not be improved significantly.

digestion efficiency of these peptides to 31% and 60% for the pS and the pT variants, respectively. Furthermore, the addition of TFE led to a nearly complete digestion (>99% efficiency) of the corresponding peptide sequence containing pT at P3′. To exclude the possibility that this increase in digestion efficiency results from unspecific cleavage, we searched unassigned spectra once more, but without enzyme constraints. No unspecific hydrolysis could be observed, neither for the 2766

dx.doi.org/10.1021/pr401181y | J. Proteome Res. 2014, 13, 2761−2770

Journal of Proteome Research

Article

Figure 7. Significantly improved overall performance with increased trypsin concentrations in large-scale analyses. (A) The highest number of unique phosphopeptides was identified using a trypsin to peptide ratio of 1:50, yet increasing trypsin concentrations led to the identification of more unique phosphorylation sites in the label-free study (phosphopeptide enrichment of 75 μg in triplicate). (B) These results are reflected by the number of phosphopeptide PSM with and without missed cleavage sites. The benefit of increased trypsin concentrations could be validated in an iTRAQ label-based approach (a total of 280 μg iTRAQ-labeled peptides) on the phosphopeptide PSM level (C) as well as on the phosphopeptide level, exemplarily presented for AAAR|LpSLTDPLVAER (D).

digestion efficiency, as no changes occurred after 4 h of incubation (data not shown). To gain more insights into the digestion kinetics of phosphopeptide sequences, shorter incubation times may be evaluated. Performing additional experiments, we could improve the digestion efficiency of resistant phosphopeptides (phosphorylated at P2′ or P3′) by the addition of 5% TFE as well as by increasing trypsin concentration from an enzyme to peptide ratio of 1:50 to a ratio of 1:20. Importantly, the accessibility of the tryptic cleavage site also depends on the overall peptide sequence, as none of the tested buffer conditions could improve the digestion of the phosphorylated variants of the artificial peptide sequence QQR|R|GSLPEISNLR to its fully tryptic form, indicating the need for further optimization of the digestion conditions, e.g., by evaluating other proteases. Initial results on the digestion of trypsin resistant phosphorylated variants of AAAR|LSLTDPLVAER using ArgC demonstrated an up to 10-fold increase in the digestion efficiency under standard conditions.

but rather due to specific proteolytic digestion, indicating a benefit for the quantification of low-abundant phosphopeptides by optimizing digestion conditions. However, also by using an increased trypsin concentration, no significant improvement could be obtained for the digestion of the nonmodified and phosphorylated forms of the peptide sequence QQR|R|GSLPEISNLR to the corresponding fully tryptic form. To further evaluate the optimization of digestion efficiency, we combined a trypsin to peptide ratio of 1:20 with organic additives (10% ACN and 5% TFE, respectively) for the digestion of highly resistant phosphopeptide sequences (see Figure 6B). The efficiency of digesting AAAR|LpSLTDPLVAER and AAAR|LpTLTDPLVAER to the corresponding fully tryptic forms could be improved considerably by the addition of TFE (42% vs 31% and 86% vs 49%), whereas ACN was only beneficial for the digestion of AAAR|LpTLTDPLVAER (69% vs 49%). Even though the digestion of the phosphorylated variants of QQR|R|GSLPEISNLR was only slightly influenced, combining relatively high trypsin to peptide ratios (1:20) and 5% TFE in the digestion buffer seems to be a straightforward way to improve the digestion efficiency for a variety of phosphopeptides. To evaluate the benefit of prolonged incubation times on digestion efficiency, we monitored the digestion of the phosphorylated variants (phosphorylated at P2′) and the nonmodified form of AAAR|LSLTDPLVAER after 4, 8, and 16 h of incubation (trypsin to peptide ratio 1:50, without additives). Here, prolonged incubation times could not improve

Impact of Digestion Conditions on Large-Scale Phosphoproteomic Studies

To validate our findings from the digestion of model peptides in a large-scale phosphoproteomic study, we analyzed the digestion efficiency of phosphopeptides in a human platelet sample using different trypsin to peptide ratios (1:10, 1:20, 1:50, and 1:100) for proteolytic digestion. In addition, we evaluated the impact of using 5% TFE in combination with a 2767

dx.doi.org/10.1021/pr401181y | J. Proteome Res. 2014, 13, 2761−2770

Journal of Proteome Research

Article

1:20 in combination with 5% TFE for proteolytic digestion and further employed trypsin to peptide ratios of 1:50 and 1:20 without additive as references. The data confirmed the beneficial effect of TFE for phosphopeptide quantification, as evident by the increased sum of reporter ion intensities of fully tryptic phosphopeptide PSM by more than 17% compared to a proteolytic digestion without additive (1:50 and 1:20 sample). Also, the amount of the fully tryptic peptide fragment LpSLTDPLVAER of the platelet glycoprotein 1b (P13224) slightly increased (1.2-fold) compared to a trypsin to peptide ratio of 1:20 without additive (data not shown). Interestingly, the intensities not only of the fully tryptic but also of the missed-cleavage phosphopeptide PSM and the amount of the individual missed-cleavage phosphopeptide AAAR|LpSLTDPLVAER increased by 1.5-fold in the presence of 5% TFE compared to the 1:20 sample without additive. This effect might be explained by the enhanced digestion of peptide sequences with multiple (n) missed-cleavage sites not only to the fully tryptic form but also to sequences with (n − 1) missed cleavages. Despite the positive effects of TFE in our experiments, we reproducibly observed an interference of TFE with the efficiency of the TiO2-based phosphopeptide enrichment, when compared to TFE-free samples (human platelet digests, data not shown). As this effect might lead to a loss of relevant peptides, we recommend the use of TFE only for targeted analyses and low-complexity samples, focusing on the quantification of single phosphoproteins without extensive phosphopeptide enrichment steps. Besides, we reproducibly observed a reduced digestion efficiency of the resistant model peptide AAAR|LpSLTDPLVAER in the presence of 50 mM TEAB compared to 50 mM ABC (20% lower). Indeed, in the presence of 5% TFE the digestion efficiency was even 50% lower (duplicate analysis, data not shown). As TEAB is often used in large-scale iTRAQ-based phosphoproteomic studies, its influence on the proteolytic digestion compared to ABC and other digestion buffers needs to be evaluated more in detail. In summary, increased trypsin concentrations yield a significant improvement for the identification and quantification of phosphorylation sites, as a consequence of slightly improved reproducibility and increased sensitivity, indicated not only by the higher number of fully tryptic PSM but also by increased amounts of the summed precursor ion areas and individual phosphopeptide intensities. Although increasing the trypsin to peptide ratio from 1:20 to 1:10 did not improve the overall performance in the label-free approach, a ratio of 1:10 is clearly beneficial for the digestion, and thereby for the quantification, of single resistant phosphopeptides such as AAAR|LpSLTDPLVAER in a high-complexity sample.

trypsin to peptide ratio of 1:20. Both approaches were followed by TiO2-based phosphopeptide enrichment. A semiquantitative label-free analysis of the enriched phosphopeptides obtained from 75 μg of platelet digest led to the identification of more than 1500 unique phosphopeptides (FDR on PSM level