variation and quantification among a target set of phosphopeptides in human plasma by multiple...
TRANSCRIPT
Electrophoresis 2014, 00, 1–11 1
Anna M. Zawadzka1
Birgit Schilling1
Jason M. Held2
Alexandria K. Sahu1
Michael P. Cusack1
Penelope M. Drake3
Susan J. Fisher3
Bradford W. Gibson1,4
1Buck Institute for Research onAging, Novato, CA, USA
2Division of Oncology andDepartment of Anesthesiology,Washington University Schoolof Medicine, St. Louis, MO, USA
3Department of Obstetrics,Gynecology and ReproductiveSciences, University ofCalifornia, San Francisco, CA,USA
4Department of PharmaceuticalChemistry, University ofCalifornia, San Francisco, CA,USA
Received March 29, 2014Revised April 26, 2014Accepted May 13, 2014
Research Article
Variation and quantification among a targetset of phosphopeptides in human plasmaby multiple reaction monitoring andSWATH-MS2 data-independent acquisition
Human plasma contains proteins that reflect overall health and represents a rich source ofproteins for identifying and understanding disease pathophysiology. However, few stud-ies have investigated changes in plasma phosphoproteins. In addition, little is knownabout the normal variations in these phosphoproteins, especially with respect to specificsites of modification. To address these questions, we evaluated variability in plasma pro-tein phosphorylation in healthy individuals using multiple reaction monitoring (MRM)and SWATH-MS2 data-independent acquisition. First, we developed a discovery work-flow for phosphopeptide enrichment from plasma and identified targets for MRM assays.Next, we analyzed plasma from healthy donors using an analytical workflow consistingof MRM and SWATH-MS2 that targeted phosphopeptides from 58 and 68 phosphopro-teins, respectively. These two methods produced similar results showing low variabilityin 13 phosphosites from 10 phosphoproteins (CVinter � 30%) and high interpersonalvariation of 16 phosphosites from 14 phosphoproteins (CVinter � 30%). Moreover, thesephosphopeptides originate from phosphoproteins involved in cellular processes govern-ing homeostasis, immune response, cell–extracellular matrix interactions, lipid and sugarmetabolism, and cell signaling. This limited assessment of technical and biological vari-ability in phosphopeptides generated from plasma phosphoproteins among healthy vol-unteers constitutes a reference for future studies that target protein phosphorylation asbiomarkers.
Keywords:
Biomarkers / Cancer / MS / Phosphorylation DOI 10.1002/elps.201400167
� Additional supporting information may be found in the online version of thisarticle at the publisher’s web-site
1 Introduction
Protein phosphorylation is a dynamic and reversible PTMknown to regulate cellular signaling pathways that controlmultiple biological processes. Significant changes in proteinphosphorylation occur in many disease states, including can-cer [1]. Some cancer-related phosphoproteins enter the blood-stream where they could potentially serve as biomarkers fordisease detection or in monitoring the efficacy of therapeutic
Correspondence: Dr. Bradford W. Gibson, Buck Institute for Re-search on Aging, 8001 Redwood Blvd., Novato, CA 94945, USAE-mail: [email protected]: +1-415-209-2231
Abbreviations: MARS, Multiple Affinity Removal System;MRM, multiple reaction monitoring; PE, plasma equivalents;SCX, strong cation exchange; TiO2, titanium dioxide
strategies [2, 3]. Similarly, phosphorylation status of nativeplasma proteins may reflect the disease status of the organ-ism during systemic responses such as inflammation [4]. Bi-ological fluids, including blood, are the preferred and mostaccessible samples for disease diagnosis and monitoring asthey reflect processes in distant parts of the body and are rel-atively easily available. Plasma phosphoproteome has been alargely overlooked source of potential biomarkers with onlylimited studies performed [5–8]. Our recent work has shownpotential for use of plasma phosphopeptides as biomarkers ofbreast cancer differentiating subtypes of this disease [3]. How-ever, despite growing numbers of plasma protein phospho-rylation sites discovered, nothing is known about biologicalvariability in these sites among healthy human population,let alone cancer patients. Moreover, most plasma and serumcollection protocols are not designed to stabilize phosphopro-teins against phosphatase activity and therefore we sought toimplement plasma collection protocol incorporating the
C© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
2 A. M. Zawadzka et al. Electrophoresis 2014, 00, 1–11
addition of phosphatase inhibitors early after blooddraw.
MS-based proteomics combined with selective phos-phopeptide enrichment strategies is a powerful techniquefor large-scale comprehensive characterization of the phos-phoproteome [3]. Additionally, quantitative MS approaches,namely multiple reaction monitoring (MRM), are increas-ingly used for biomarker verification and validation in clin-ical samples due to their reproducibility and sensitivity [9].However, MRM in its most quantitative version requiressynthetic stable isotope labeled peptides for normalizationand endogenous peptide verification. MRM assays also re-quire triple quadrupole instrumentation with significant up-front development efforts that are further constrained by thenumber of target peptides (typically �150) that can be mea-sured in a single analysis. Another recently developed tar-geted MS approach that is an alternative to MRM, SWATH-MS2, is based on a data-independent acquisition mass spec-trometric method [10]. In this approach, data are typicallyacquired on a fast high-resolution QTOF instrument by re-peated cycling through sequential isolation m/z windowsover the entire chromatographic separation. In contrast toMRM, SWATH-MS2 acquisitions record the fragment ionspectra of all analytes detectable in a sample, which are thenused for identification and quantification using a targeteddata extraction method. These data can be mined postacqui-sition and the method does not require extensive develop-ment. Overall, SWATH-MS2 has strengths of shotgun pro-teomics to detect large numbers of analytes and also pro-duce accurate quantification comparable with MRM repro-ducibility and sensitivity as recently shown for N-linked gly-coproteins in plasma [11] and other PTMs of biomarkers[12].
In this study, we developed first an optimized androbust discovery workflow to identify phosphopeptides inplasma, and second, an analytical workflow that used muchsmaller plasma volumes to assess variations among a targetset of phosphopeptides in healthy individuals using MRM-MS assays. Subsequently, an independent set of SWATH-MS2 acquisitions was obtained to assess the sensitivityand quantitative capabilities of this newer method in com-parison with the gold standard MRM-MS protocol. Thetwo methods produced similar, but complementary, resultswith overlap of over 40 phosphopeptides, which showed re-producible quantitation by both MRM and SWATH-MS2.Biological variability of phosphopeptides from numerous pro-teins that have known roles in cellular processes and sig-naling pathways altered in cancer was also assessed. Theseindependent analyses showed that while there was a rela-tively high level of biological variation among these targetedphosphopeptides in healthy individuals, a significant num-ber of phosphopeptides showed low biological interpersonalCVs (�30%). To our knowledge, this is the first effort toquantify relative differences in plasma protein phosphory-lation by MRM-MS and SWATH-MS2 in healthy humansubjects.
2 Materials and methods
2.1 Plasma preparation and immunodepletion
Human blood samples from healthy volunteers (18–50 years;four males, six females) were collected at the Univer-sity of California, San Francisco, according to the CPTACblood collection protocol (http://wikisites.mcgill.ca/djgroup/images/9/9f/CPTAC_plasma_protocol_20080110.pdf) afterinformed written consent was obtained. The protocol wasapproved by the UCSF Human Research Protection ProgramCommittee on Human Research (IRB #10–03275) and theBuck Institute (BUA B1022).
Within 30 min of collection, blood was centrifuged twiceat 4°C (1500 × g and 2000 × g for 15 min each), and phos-phatase and protease inhibitors (PhosSTOP and CompleteMini EDTA-free protease inhibitor cocktail tablets, Roche)were added to the plasma. The aliquots of plasma were storedat –80°C until further processing [3]. The 14 most abundantplasma proteins were immunodepleted according to the man-ufacturer’s instructions using a Multiple Affinity RemovalSystem (MARS) Human-14 column (10 × 100 mm) fromAgilent Technologies on a Waters 1525 HPLC system. Thedepleted proteins constituted about 92% of the total plasmaprotein, and the protein concentration of the depleted fractionwas between 4 and 5 mg/mL. The protein flow-through frac-tions were collected and readjusted to the original volume of200 �L per injection using 5K MWCO centrifugal concentra-tors (Agilent Technologies). Henceforth, an immunodepletedplasma volume refers to the original plasma volume (plasmaequivalents (PE)).
2.2 Trypsin digestion of plasma proteins
The MARS Hu-14 depleted plasma (1 mL PE) was denaturedwith 6 M urea, reduced with 20 mM DTT (30 min at 37°C),alkylated with 50 mM iodoacetamide (30 min at RT), anddigested overnight at 37°C with enzyme:substrate (1:50) ratiow/w of sequencing grade trypsin (Promega, Madison, WI) aspreviously described [13]. Following digestion, the sampleswere acidified with formic acid, and desalted using HLB OasisSPE cartridges (Waters, Milford, MA), and concentrated byvacuum centrifugation. Peptides were stored at –80°C untiluse.
2.3 Stable isotope labeled heavy phosphopeptides
as internal standards
Pure, stable isotope labeled synthetic peptides, labeled at theC-terminus with 13C6,15N4-Arg (R) of phosphorylated pep-tides, were synthesized by Thermo Scientific as AQUA �97%purity peptides (10 nmol): AAIpSGENAGLVR from ITIH1,ISApSAEELR from APOA4, LPTDpSELAPR from SEPP1.
C© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
Electrophoresis 2014, 00, 1–11 Proteomics and 2DE 3
2.4 Strong cation exchange (SCX) and HILIC
fractionation, and TiO2 and IMAC enrichment
of phosphopeptides
SCX and HILIC peptide fractionation was performed on aWaters 1525 HPLC system equipped with a 9.4 × 250 mmPolysulfoethyl A 5 �m column (PolyLC) and 4.6 × 250 mmTSKgel Amide-80 HR 5 �m particle column (Tosoh Bio-science), respectively [14, 15]. For SCX, the plasma samples(1000 �L PE) were loaded in 100% solvent A (7 mM KH2PO4,pH 2.7, 30% ACN) and eluted with the following gradient: 0%B for 2 min followed by 0% B to 25% B in 33 min, and then100% B in 1 min for 15 min at 3 mL/min. Solvent B consistedof 7 mM KH2PO4, pH 2.7, 350 mM KCl, and 30% ACN. ForHILIC, the plasma samples (500 �L PE) were loaded in 80%solvent B (98% ACN, 0.1% TFA) and eluted with the follow-ing gradient: 80% B for 5 min followed by 80% B to 60% B in40 min, and then 0% B in 15 min at 0.5 mL/min. Solvent Aconsisted of 98% HPLC-grade water (Honeywell) and 0.1%TFA. Ten SCX fractions were collected and desalted usingSep-Pak 500 mg tC18 cartridges (Waters). Twelve HILICfractions were collected and each enriched for phosphopep-tides after reducing their volume to 50 �L using a SpeedVacconcentrator (Savant, Thermo Scientific). Phosphopeptideswere enriched using titanium dioxide (TiO2) chromatog-raphy according to the manufacturer’s instructions(Titanosphere Phos-TiO kit, 200 �L columns, GL Sciences).For immobilized-metal affinity chromatography (IMAC)enrichment, 5 mg of Fe-NTA (Sigma-Aldrich) was used andthe samples were incubated in 250 mM CH3COOH/30%ACN. The same buffer was used to wash the gel three times,and phosphopeptides were eluted sequentially with 50 mMK2HPO4/NH4OH, pH 10, and 150 mM NH4OH/25% ACN(100 �L each) and acidified immediately with 20% TFA. Thesamples were desalted using Oasis HLB �Elution 96-wellplate (Waters). After removal of organic solvent using aSpeedVac concentrator, the phosphopeptide samples weresuspended in 0.1% formic acid and subsequently analyzed byLC-MS/MS. For discovery experiments, fractionated plasmasamples were pooled from 15 individual donors (Fig. 1).For MRM and SWATH-MS2 analyses, individual donorplasma (500 and 200 �L PE, respectively) was processed andanalyzed in duplicate. HILIC fractions from 29 to 60 minwere combined, as they contained �95% of all phosphopep-tides, yielding a total of one fraction. Heavy phosphopeptidestandards were spiked-in at 150 fmol each, and the sampleswere enriched using TiO2 (Fig. 3).
2.4.1 Nano-LC-ESI-MS/MS analyses, data-dependent
acquisitions
The peptide mixtures obtained after tryptic digestion, SCXor HILIC fractionation, TiO2 enrichment, and desalting wereanalyzed by RP nano-HPLC-ESI-MS/MS using an Eksigentnano-LC 2D HPLC system (Eksigent, Dublin, CA), which was
Figure 1. The experimental workflow developed for preparationof phosphopeptides from plasma for discovery experiments.
directly connected to a QTOF QSTAR Elite mass spectrome-ter (AB SCIEX, Concord, CA) in data-dependent mode.
2.4.2 Mass spectrometric database searches
Mass spectrometric data were analyzed using two sep-arate bioinformatics database search engine systems,ProteinPilotTM (AB SCIEX) version 4.0.8085 (revision 148085)[16] using the Paragon Algorithm version 4.0.0.0, (revision148083) [17] and an in-house MASCOT 2.3 server (MatrixScience) [18]. All data were searched using a publicly avail-able human Swiss-Prot UniProt release 2011_05 databaseof 20 239 protein sequences. For ProteinPilot searches,the following parameters were used: trypsin enzyme speci-ficity, carbamidomethyl (Cys) as a fixed modification, specialfactors including phosphorylation emphasis and urea denat-uration, and thorough search effort setting allowing for bio-logical modifications [17]. For database searches, a Protein-Pilot “peptide confidence” cut-off value of 95 was chosen,yielding a peptide-level local false discovery rate (FDR) of�3%. For MASCOT searches, the following parameters wereused: trypsin enzyme specificity, carbamidomethyl (Cys) as afixed modification, and the following variable modifications:phosphorylation at Ser, Thr, and Tyr, deamidation of as-paragine and glutamine residues, oxidization of methionine,
C© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
4 A. M. Zawadzka et al. Electrophoresis 2014, 00, 1–11
acetylation at the protein N-terminus, cyclization of N-terminal glutamine, and a maximum of three missed trypticcleavages. For QSTAR Elite data, a mass tolerance of 100 ppm(MS1) and 0.4 Da (MS2) was set for the precursor and productions, respectively. MASCOT peptide-spectral matches withsignificance threshold p � 0.05 were accepted. FDR analysiswas performed using the MASCOT automatic decoy search.In all cases, the peptide false-positive identification rate was�3%. Due to the phosphopeptide-centric approach, proteinidentifications were made based only on the identified phos-phopeptides and thus single-phosphopeptide identificationswere allowed.
2.4.3 MRM transition selection
MS/MS data from the pooled plasma samples obtained dur-ing discovery experiments were used to build spectral li-braries in Skyline [19]. Initially, two MRM-MS precursor-to-product ion transitions per phosphopeptide were designed inSkyline for all phosphopeptides in the library. The prelim-inary MRM-MS analysis was performed using the QTRAP5500 hybrid triple quadrupole/linear IT mass spectrometer(AB SCIEX, Foster City, CA) that is capable of high-sensitivityand multiplexed MRM acquisitions, and the data were up-loaded and analyzed in Skyline. Phosphopeptides with bothtransitions present and free of coeluting ions were selectedfor further assay development. The total of three to four tran-sitions per each target phosphopeptide were designed andthe list was split into three separate methods each containing�100 transitions for final sample analysis.
2.4.4 LC-MRM/MS
For selected reaction monitoring (MRM), samples were ana-lyzed by nano-LC-MRM-MS on a QTRAP 5500 mass spec-trometer. Chromatography was performed on a NanoLC-Ultra 2D LC system (Eksigent) with buffer A (0.1% v/v formicacid) and buffer B (90% ACN in 0.1% formic acid). Digestswere separated on a 75 �m id Integrafrit analytical column(New Objective, Woburn, MA) packed in-house with 10–12 cm of ReproSil-Pur C18-AQ 3 �m RP resin (Dr. Maisch,Germany) at a flow rate of 300 nL/min. Gradient was 3% Bfrom 0 to 5 min, increased to 7% B over 3 min, increased to25% B over the next 27 min, and increased to 40% B overthe next 7 min. Peptides were ionized using a PicoTip emit-ter (20 �m, 10 �m tip, New Objective). Data acquisition wasperformed using Analyst 1.5.1 (AB SCIEX) with an ion sprayvoltage of 2300 V, curtain gas of 20 psi, nebulizer gas of 15 psi,and interface heater temperature of 150°C.
The transitions, dwell times, and collision energy arelisted in Supporting Information Table 2. Four transitionswere assayed per peptide. Values of 100 and 40 were used asthe declustering potential and collision cell exit potential, re-spectively, for all transitions. MRM transitions were acquiredat unit resolution both in the first and third quadrupoles (Q1
and Q3). Samples were processed and analyzed in duplicatewith 150 fmol of each heavy phosphopeptide spiked-in beforeTiO2 enrichment and 125 �L PE was injected on the column.Standard curves were performed in duplicate by spiking in thestable isotope labeled phosphopeptides to determine the lin-ear range in a background matrix of 250 ng of a predigestedsix-protein mix (Michrom). Reproducibility of MRM mea-surements was tested using four replicates of pooled plasmasamples and was consistent with previous experience wherethe QTRAP 5500 was a part of recent multisite study thatassessed system suitability and reproducibility for MRM-MSanalysis [20].
2.4.5 Quantitative MRM data analysis
Skyline postacquisition software was used to process allMRM-MS data [19]. Each transition was individually inte-grated to generate peak areas and the peak area of the mostintense transition ion was used for analysis. All native phos-phopeptide data were normalized to the average of the mostintense transition ions of three heavy phosphopeptide stan-dards (y9–98 from AAIpSGENAGLVR, y8–98 from ISAp-SAEELR, and y9 from LPTDpSELAPR). If multiple precursorions were detected for the same phosphopeptide, only the onewith the most intense signal was included.
2.4.6 SWATH-MS2
Data acquisitions of nine plasma samples were performedby RP nano-HPLC-ESI-MS/MS using an Ultra Plus NanoLC2D HPLC system (Eksigent), which was directly connectedto a new-generation QTOF TripleTOF 5600 mass spectrom-eter (AB SCIEX, Concord, Canada) that became availablenear the end of this study and was capable of SWATH-MS2 acquisitions. Samples were initially reanalyzed in data-dependent mode to obtain MS/MS spectra for the 30 mostabundant parent ions following each survey MS1 scan tobuild spectral libraries. Additional data sets were recorded indata-independent mode using SWATH-MS2 acquisitions. Inthe SWATH-MS2 acquisition, instead of the Q1 quadrupoletransmitting a narrow mass range through to the collisioncell, a wider window of �25 Da is passed in incremental stepsover the full mass range from 400 to 1000. To increase over-all efficiencies, two injection replicate SWATH-MS2 experi-ments were performed per sample. The amount of sampleinjected on the column equaled to 55 �L PE. Additional de-tails describing mass spectrometric instrument parametersand settings, and all chromatographic setups and gradientconditions are found in Supporting Information Methods 1.Lastly, it should be noted that the high reproducibility of theTripleTOF 5600 was assessed in previous studies using newalgorithms that examined instrument stability with statisticalmetrics [21] and where the MS1 signal intensities were exam-ined for repeatability and reproducibility over extended timeperiod [22, 23].
C© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
Electrophoresis 2014, 00, 1–11 Proteomics and 2DE 5
2.4.7 Quantitative SWATH-MS2 data analysis
Data sets from SWATH-MS2 acquisitions were processedusing the full-scan MS/MS filtering module for data-independent acquisition within Skyline. The top eight frag-ment ions were extracted from SWATH-MS2 acquisitionswithin Skyline using a fragment ion resolution setting of10 000. The peak area of the most intense fragment ion wasused for quantitative analysis. All native phosphopeptide datawere normalized to the average of the most intense transitionions of two spiked heavy phosphopeptide standards (y8–98from ISApSAEELR and y9 from LPTDpSELAPR).
2.5 Statistical analysis
To assess the sample processing and instrument repro-ducibility, the CV between two process replicates per sam-ple (MRM-MS) or two process replicates with two technicalreplicates each per sample (SWATH-MS2) were determinedfor each fragment ion. Phosphopeptides were considered asreproducibly quantifiable only when process replicate CV was�30% for at least four of six samples analyzed by MRM in du-plicate and at least six of nine samples analyzed by SWATH-MS2. Next, CVinter for all eight MRM-MS and nine SWATH-MS2 sample acquisitions was calculated to assess biologicalvariation of phosphopeptides within the human population.Interpersonal variation was considered low when CVinter �
30% and high when CVinter � 30%.
2.6 Data accession
All raw data associated with this manuscript maybe downloaded from the Buck Institute ftp site atftp://sftp.buckinstitute.org:251. All confidently identifiedphosphorylated peptides were transferred to the data-sharingPanorama server [24], allowing for interactive web-based spec-tral viewing of all PTM-containing peptides in this study (at95% confidence score). The spectral viewer can be accessedat http://proteome.gs.washington.edu/software/panorama/PlasmaPhosphoproteome.html.
3 Results and discussion
3.1 Workflow development and discovery analysis
of phosphopeptides from plasma
In preparation for MRM-MS assay development, a discoveryworkflow for phosphopeptide enrichment and subsequentdata-dependent analysis from plasma were developed (Fig. 1).Pooled plasma samples were subjected to this workflow us-ing larger volumes than we would subsequently employ forour analytical workflow, allowing for a more in-depth analy-sis of the plasma phopshoproteome while building spectrallibraries that would be needed later. In this discovery work-flow, a previous reported CPTAC blood collection protocol
Figure 2. Numbers of unique phosphosites (left) and phospho-proteins (right) identified in healthy human plasma (A) after frac-tionation and phosphopeptide enrichment of 1 mL of plasma us-ing SCX-TiO2, SCX-IMAC, and HILIC-TiO2. (B) Comparison of thenumbers of phosphosite (left) and phosphoprotein (right) identi-fications obtained during discovery experiments with the resultsobtained for 200 �L of plasma processed with HILIC-TiO2 and an-alyzed by data-dependent acquisitions on a TripleTOF 5600 (en-circled).
was modified so that phosphatase and protease inhibitorscould be added at a very early stage after initial blood draw ofthe platelet-depleted plasma to preserve the phosphorylationstatus of the plasma proteome and limit contamination fromplatelets and other cell types [3]. Since the analysis and quan-tification of plasma and serum proteins are challenging dueto the complexity and large dynamic range of the plasma pro-teome, a combination of analytical approaches was employed.After immunodepletion of the 14 most abundant proteins andtrypsin digestion, the peptides were fractionated by off-lineSCX or HILIC chromatography and enriched for phospho-proteins by TiO2 and IMAC. The LC-MS/MS analysis on theQSTAR Elite of 1 mL PE allowed for identification of over 250unique phosphorylation sites with 95% confidence interval on134 phosphoproteins (Fig. 2A and Supporting InformationTable 1). Because MARS Hu-14 immunodepletion removes�97% of the 14 highest abundance proteins, phosphopep-tides originating from residual amounts of these proteinswere not considered in our data set (APOA1, APOA2, CO3,FIBA). Interestingly, only 78 of over 250 unique phosphositeswe identified had been previously reported [5, 6]. However, amore recent study by Jaros et al. [7] reported �500 phospho-proteins in human plasma, of which 58 are common with ourdiscovery results. The analysis of plasma samples preparedusing three combinations of fractionation and enrichmentmethods gave complementary sets of phosphosites with themost identified phosphosites following SCX fractionation.However, the highest number of phosphoproteins was iden-tified from plasma prepared by the HILIC-TiO2 workflow.Given that HILIC chromatography, unlike SCX fractionation,does not require subsequent desalting, we choose HILIC forour fractionation method in our analytical quantitative assays
C© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
6 A. M. Zawadzka et al. Electrophoresis 2014, 00, 1–11
Figure 3. The experimental workflow developed for preparationof phosphopeptides from plasma for quantitative mass spectro-metric measurements.
workflow for MRM-MS and SWATH-MS2 as described below(Fig. 3).
3.2 MRM-MS analysis
All phosphopeptides identified in our discovery workflowwere then used to build spectral libraries in Skyline [19]for subsequent MRM-MS assay development. A set of pre-liminary experiments was employed to develop an analyticalworkflow that was better optimized for sample preparationusing smaller plasma volumes that would be more typicallyavailable for biomarker studies. The final sample prepara-tion protocol included HILIC fractionation of 500 �L PEfollowed by TiO2 enrichment (Fig. 3). Heavy phosphopep-tide standards were spiked-in at 150 fmol per injection priorto TiO2 enrichment to partially control for technical varia-tion in the phosphopeptide-specific enrichment steps. MRMtransitions were refined only for phosphopeptides that weredetectable and final methods included three to four MRM-MS transitions per peptide, with a total of �500 transitions.MRM data were imported into Skyline and the most in-
tense transitions were used for quantification. Peak areaswere normalized to the mean of the three heavy peptidemost intense transitions. To test sample processing and in-strument reproducibility, a set of four individually preparedreplicate pooled plasma samples was analyzed and com-pared in Skyline. The CV for all spiked-in heavy peptidetransitions was �20% indicating very good sample process-ing and instrument reproducibility (Supporting InformationFig. 1).
Using this analytical workflow, six plasma sampleswere then processed in duplicate and two with one repli-cate each, for a total of eight samples. To select repro-ducible data for native phosphopeptides, CVs for each sam-ple process replicate were calculated and only phosphopep-tides with CVs less than 30% were considered for furtherstudy. The MRM analysis yielded quantitative data for 98unique phosphopeptide sequences corresponding to 90 phos-phosites from 58 phosphoproteins (Supporting InformationTable 3).
Next, the biological variation in these phosphopeptideslevels among the eight individuals was assessed based on cal-culating the CVinter. About half of the phosphosites showedrelatively low interpersonal variation below CVinter of 30%(Fig. 4A). The most abundant phosphopeptides examinedwere derived from alpha-2-HS-glycoprotein (FETUA), kinino-gen 1 (KNG1), antithrombin-III (ANT3), and secreted phos-phoprotein 1/osteopontin (OSTP). Phosphopetides from 7of 11 phosphoprotein members of the complement and co-agulation cascades quantified by MRM showed low biolog-ical variation. The other four, including a phosphopeptidefrom coagulation factor FA5, for which its concentration isknown to increase with age [25] had a relatively high biolog-ical variation. Phosphopeptides from proteins with functionin focal adhesion and extracellular matrix (ECM)–receptor in-teractions, including integrin 5 (ITA5), osteopontin (OSTP),tenaxin (TENXA), vitronectin (VTNC), showed low biologi-cal variation, whereas pS1454 from filamin A (FLNA) hadhigher variation. Phosphopeptides from protease inhibitorinteralpha globulin inhibitor H4 (ITHI4), which is involvedin acute inflammatory response, showed low variation, whilephosphopeptides from ITHI1 and ITHI2 were consider-ably higher. Other phosphosites with high biological vari-ation were derived from lipid carrier proteins apolipopro-teins APOL1 and APOA4 and IGF-binding proteins 3 and5 (IBP3 and IBP5) (Supporting Information Table 3 andFig. 5A and C).
As little is known about absolute concentrations of nativephosphopeptides, we used a stable isotope dilution MRM-MSapproach to target two phosphopeptides derived from APOA4and ITIH1 for which heavy standard peptides were available.Average concentration of ISApSAEELR phosphopeptide was226 nmol/L of plasma and 203 nmol/L for AAIpSGENAGLVR(Supporting Information Fig. 2). The reported concentrationof these proteins in plasma is 3–6 and 2–4 �mol/L, respec-tively [26]. Therefore, an estimated level of phosphorylationat these two phosphosites (S259 and S129) ranged from 4 to7%.
C© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
Electrophoresis 2014, 00, 1–11 Proteomics and 2DE 7Ta
ble
1.Q
uan
tita
tive
resu
lts
of
MR
Man
dS
WA
TH
-MS
2as
says
for
ph
osp
ho
pep
tid
esin
hu
man
pla
sma
Prot
ein
Acce
ssio
nM
odifi
edPh
osph
osite
sAs
core
Tran
sitio
nPr
ecur
sor
Prod
uct
Prec
urso
rPr
oduc
tM
ean
SDCV
inte
rM
ean
SDCV
inte
r
nam
enu
mbe
rse
quen
ce(%
)m
/zm
/zch
arge
char
geAL
Lal
l(%
)al
lal
l(%
)
ANT3
P010
08AT
EDEG
pSEQ
KIPE
ATN
RRS6
810
0y 1
570
4.32
856.
423
25
685
867
124
163
721
.896
919
2888
329
.8AN
T3P0
1008
KATE
DEGp
SEQK
IPEA
TNR
S68
100
y 669
4.98
687.
343
123
877
574
392
942
316
.51
097
017
227
005
20.7
CERU
P004
50N
NEG
TYYS
PNYN
PQpS
RS4
9910
0y 4
661.
9346
9.25
31
626
291
144
126
23.0
1840
449
3526
.8CF
AIP0
5156
VTYT
pSQE
DLVE
KS2
410
0y 9
746.
3310
30.5
12
122
365
1965
8.8
7299
1161
15.9
FETU
AP0
2765
CDSS
PDpS
AEDV
RS1
3810
0y 8
709.
2587
0.40
21
9031
374
621
888
971
24.2
489
547
192
423
018
.9FE
TUA
P027
65HT
FoxM
GVVS
LGpS
PSGE
VSHP
RS3
28y 1
072
6.33
1132
.48
31
954
205
02
193
977
23.0
120
799
926
353
221
.8IT
IH4
Q146
24N
VHpS
GSTF
FKS6
20y 8
602.
2649
5.71
22
245
101
7053
128
.810
867
2290
21.1
KNG1
P010
42DI
PTN
pSPE
LEET
LTHT
ITK
S275
100
y 17
740.
3599
5.97
32
142
060
931
168
021
.912
328
114
637
11.9
KNG1
P010
42ET
TCSK
EpSN
EELT
EpSC
ETKK
S332
,S33
9y 8
850.
6696
4.44
31
7512
720
199
26.9
1091
718
5117
.0LE
AP2
Q969
E1RC
pSLS
VAQE
S71
100
b 756
5.24
854.
362
181
320
2422
829
.832
159
8795
27.3
OSTP
P104
51IS
HELD
pSAS
SEVN
/ISHE
LDSA
pSSE
VNS3
0898
b 873
4.30
835.
392
151
070
1338
226
.256
2013
9324
.8PL
MN
P007
47KQ
LGAG
pSIE
ECAA
KS4
510
0y 5
514.
5757
8.26
31
213
933
3072
014
.459
995
1082
818
.0SE
PP1
P499
08Do
xMPA
pSED
LQDL
QKS2
6610
0Y 1
179
3.33
662.
302
26
670
889
154
670
223
.288
127
120
536
523
.3AL
DOA
P040
75GI
LAAD
ESTG
pSIA
KS3
910
0y 9
706.
8498
7.40
21
4160
828
035
67.4
5895
7210
122.
3AN
GL3
Q9Y5
C1ID
QDN
SpSF
DSLS
PEPK
S25
99y 4
929.
8947
0.26
21
4308
917
460
40.5
5642
3692
65.4
APOA
4P0
6727
ENAD
pSLQ
ASLR
PHAD
ELK
S174
100
y 15
691.
9983
1.43
32
294
168
135
256
46.0
8755
5105
58.3
APOA
4P0
6727
ISAp
SAEE
LRS2
5910
0y 8
528.
2484
4.42
21
219
702
112
030
51.0
2984
119
085
64.0
APOE
P026
49GE
VQAM
LGQp
STEE
LRS1
4710
0y 8
864.
3999
9.41
21
907
851
272
929
30.1
121
904
8612
670
.7CO
9P0
2748
AEQC
CEET
ASpS
ISLH
GKS2
61y 8
662.
9481
0.45
31
4200
223
074
54.9
3978
1470
37.0
CYTC
P010
34LV
GGPo
xMDA
pSVE
EEGV
RRS4
310
0y 1
563
2.96
793.
873
280
008
2656
133
.211
423
4639
40.6
FA5
P122
59LL
SLGA
GEFK
pSQE
HAK
S859
100
y 14
598.
9673
5.87
32
115
821
453
259
146
.024
453
1370
356
.0FB
LN1
P231
42pS
QETG
DLDV
GGLQ
ETDK
S147
100
y 893
6.40
847.
422
111
638
247
351
40.7
1531
663
6541
.6HA
BP2
Q145
20LI
ANTL
CNpS
RS4
7910
0y 8
621.
2910
15.4
02
153
525
2324
943
.447
1856
6412
0.1
IBP3
P179
36VD
YESQ
STDT
QNFp
SSES
KRS2
0197
y 15
763.
3284
2.38
32
261
577
9327
535
.727
365
2055
375
.1IB
P3P1
7936
YKVD
YESQ
STDT
QNFS
pSES
KRS2
02y 7
860.
3782
2.41
31
245
031
101
355
41.4
1658
852
7731
.8IT
IH1
P198
27AA
IpSG
ENAG
LVR
S129
100
y 961
9.30
982.
442
113
181
763
980
48.5
1812
916
50.6
ITIH
4Q1
4624
pSPE
QQET
VLDG
NLI
IRS2
2510
0y 7
631.
3180
0.46
31
550
666
173
907
31.6
6904
125
343
36.7
PLM
NP0
0747
QLGA
GpSI
EECA
AKCE
EDEE
FTCR
S45
100
y 19
923.
7011
51.4
63
289
3748
8654
.749
2525
4551
.7TE
NXA
Q164
73LG
PLSA
EGTT
GLAP
AGQT
SEEp
SRPR
S142
y 18
854.
7494
7.94
32
334
359
101
094
30.2
2665
689
0333
.4CB
PB2
Q96I
Y4SK
DHEE
LSLV
ApSE
AVR
S338
100
y 661
7.29
614.
333
193
083
126
594
628
.685
372
3192
537
.4CE
RUP0
0450
QSED
pSTF
YLGE
RS7
2599
y 475
6.31
474.
272
141
992
586
006
20.5
1341
241
2130
.7CE
RUP0
0450
RQpS
EDST
FYLG
ER/R
QSED
pSTF
YLGE
RS7
2210
0y 4
556.
5747
4.27
31
194
139
6683
134
.428
447
8274
29.1
CLUS
P109
09VT
TVAS
HTSD
pSDV
PSGV
TEVV
VKS3
9610
0y 1
079
8.72
507.
803
22
425
944
617
629
25.5
155
873
5297
034
.0F1
3BP0
5160
SGYL
LHGp
SNEI
TCN
RS3
7310
0y 4
600.
9355
0.24
31
114
296
828
579
025
.010
012
636
162
36.1
FETU
BQ9
UGM
5GS
VQYL
PDLD
DKN
pSQE
KS3
1510
0y 5
672.
6468
5.26
31
383
067
155
362
40.6
2608
143
4016
.6HE
P2P0
5546
GGET
AQpS
ADPQ
WEQ
LNN
KS3
797
y 768
4.96
931.
463
161
694
631
536
751
.177
401
1122
914
.5IB
P3P1
7936
YKVD
YESQ
STDT
QNFp
SSES
KR/Y
KVDY
ESQS
TDTQ
NFS
SEpS
KRS2
0197
y 19
860.
3710
95.4
83
216
568
465
781
39.7
3428
874
0821
.6
IBP5
P245
93IE
RDpS
REHE
EPTT
SEM
AEET
YSPK
/IERD
SRE
HEEP
TTpS
EMAE
ETYS
PKS1
1698
y 773
3.56
853.
394
150
736
419
598
138
.667
677
1890
627
.9
KNG1
P010
42Ep
SNEE
LTES
CETK
S332
100
y 881
8.31
967.
442
11
971
125
506
039
25.7
215
318
9305
543
.2KN
G1P0
1042
EpSN
EELT
ESCE
TKK
S332
100
y 658
8.57
752.
363
145
847
899
579
21.7
142
726
4564
032
.0PC
SK9
Q8N
BP7
HLAQ
ApSQ
ELQ
S688
100
b 960
2.77
960.
492
11
886
731
440
623
23.4
307
729
161
316
52.4
SEPP
1P4
9908
LPTD
pSEL
APR
S293
100
y 958
9.78
533.
232
223
347
6506
27.9
9394
2870
30.6
ZPI
Q9UK
55VV
QAPK
EEEE
DEQE
ASEE
KApS
EEEK
/VVQ
APK
EEEE
DEQE
ApSE
EKAS
EEEK
S61
100
y 23
739.
8292
0.04
43
165
424
4204
825
.462
784
3586
257
.1
C© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
8 A. M. Zawadzka et al. Electrophoresis 2014, 00, 1–11
Figure 4. The dependence ofbiological variability (CVinter)on the abundance of totalpeak areas of transition ionsof all plasma donor samplesin (A) MRM and (B) SWATH-MS2 measurements for 86phosphopeptides in MRM and187 in SWATH-MS2 analysis.Vertical dotted line representsCVinter = 30%.
3.3 Data-dependent analysis prior to SWATH-MS2
The initial data-dependent analysis using the TripleTOF 5600mass spectrometer that was used to build the spectral li-braries consumed less than half of the plasma volume thanwhat was used for MRM-MS analysis, yet still allowed forconfident identification of 193 phosphosites from 84 phos-phoproteins (Supporting Information Table 4). When com-paring discovery data obtained from the QSTAR Elite massspectrometer, these numbers are higher than obtained usingfractionated samples (Fig. 2B). Significantly, an additional 92phosphosites from 25 phosphoproteins were identified thatwere not detected in original QSTAR Elite discovery experi-ments (Fig. 2B). This dramatic improvement can be attributedto increased sensitivity and scanning efficiency in this next-generation TripleTOF 5600 instrument.
3.4 SWATH-MS2 analysis
SWATH-MS2 analysis was performed in attempt to minimizesample volume and assay development time and maximizenumbers of quantifiable phosphopeptides. Data-independentSWATH-MS2 acquisitions were obtained in duplicate fornine plasma samples prepared as two process replicates each.SWATH-MS2 data were analyzed in Skyline and the most in-tense transition ions were used for quantification. Peak areaswere normalized to the mean of the most intense transitionsof two spiked-in heavy peptides. To select reproducible data,CVs for native phosphopeptides of each sample process repli-cate were calculated and only phosphopeptides with processreplicate CVs of less than 30% were considered. The SWATH-MS2 analysis yielded 178 unique phosphopeptide sequencescorresponding to 139 phosphosites from 68 phosphoproteins(Supporting Information Table 4).
Figure 5. Mean peak area ofthe selected product ions ofphosphopeptides in plasmafrom healthy donors analyzedby (A) MRM and (B) SWATH-MS2 with low biological vari-ation (CVinter < 30%), and by(C) MRM and (D) SWATH-MS2with high biological variation(CVinter > 30%) consistent be-tween the two methods. Thelines represent averages ofall individual plasma sam-ples. Corresponding phospho-peptide sequences and tran-sition ions used for quantita-tion are described in detail inTable 1.
C© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
Electrophoresis 2014, 00, 1–11 Proteomics and 2DE 9
Interpersonal variability was evaluated based on calcu-lated CVinter for all nine plasma donor samples. Similar to theMRM-MS assay results, about half of the phosphopeptideshad low biological variation with CVinter � 30% (Fig. 4B). Themost abundant phosphopeptides were from alpha-2-HS-glycoprotein (FETUA), kininogen 1 (KNG1),antithrombin-III (ANT3), and fibronectin (FINC). Mostof the phosphopeptides from 14 phosphoprotein compo-nents of the complement and coagulation pathways that werequantified by SWATH-MS2 had low variability, while phos-phopeptides from 5 phosphoproteins (CBPB2, CFAH, CO9,F13B, and FA5) showed high biological variation (SupportingInformation Table 4). Also most of the phosphopeptidesrepresenting proteins involved in focal adhesion and ECM–receptor interactions had low biological variability with theexception for CD44 and collagen alpha 1(IV), and a doublyphosphorylated phosphopeptide FRIpSHELDSASpSEVNfrom osteopontin, not included in MRM-MS measurements.
Phosphopeptides from additional 29 phosphoproteinsnot included in the MRM assays were quantified by SWATH-MS2. Some interesting examples of phosphopeptides withhigh biological variation that may be linked to various dis-ease states included IGF-1, known to be connected withdiabetes-associated cancers [27], CD44 receptor for extracel-lular proteins involved in cell migration and tumor growthand progression [28], vitamin D binding protein (VTDB)known to affect inflammation and cell proliferation in can-cer and cardiovascular disease [29], adenyl cyclase associatedprotein 1 (CAP1) implicated in cell motility and tumor in-vasiveness [30], and SPARK-like protein 1 (SPRL1) with tu-mor suppressor function [31]. Interestingly, in addition to thehighly variable pS92 on SPRL1, another phosphosite pS295on SPRL1 had lower biological variability (Supporting Infor-mation Table 4).
3.5 Comparison between MRM and SWATH-MS2
quantification
Phosphopeptides quantified with high reproducibility by bothMRM-MS and SWATH-MS2 were compared (Table 1). Thir-teen phosphosites from 10 proteins showed low interpersonalvariation (CVinter � 30%) (Fig. 5A and B) and 16 phosphositesfrom 14 proteins revealed high biological variation (CVinter �
30%) (Fig. 5C and D) in both methods. However, 14 phospho-sites from 12 proteins had noticeably different CVinter valuesas measured by MRM-MS and SWATH-MS2. These largerdiscrepancies were primarily linked to phosphopeptides thatcontained multiple phosphorylation sites that were prone tomissed cleavage by trypsin, including phosphopeptides fromCERU, IBP3, IBP5, KNG1, and ZPI (Table 1). In some cases,phosphosites represented by several phosphopeptide forms,for example pS45 from plasminogen (PLMN), varied in their%CVinter, which appeared to be reflective of trypsin cleavageefficiency. Considering variability of the most abundant phos-phopeptide form, this phosphosite can be regarded as hav-ing low biological variation (Fig. 5). Lastly, phosphopeptides
from 19 proteins were quantified only by MRM-MS and from29 proteins only by SWATH-MS2 (Supporting InformationTables 3 and 4).
3.6 Relevance of phosphorylated proteins to disease
Many phosphoproteins for which phosphopeptides werequantified are known to be involved in processes and path-ways associated with human diseases, including cancer, andsome of them have been proposed as biomarkers [32]. Forexample, several phosphopeptides from phosphoproteins in-volved in the plasminogen activator/plasmin system werequantified, which is one of the major protease systems in-volved in tumor metastasis. In addition to plasminogen it-self, phosphopeptides from proteins involved in downregu-lation of the plasminogen activator/plasmin system and thussuppression of cell motility (alpha-2-antiplasmin, A2AP) [33]and from proteins known to stimulate cell migration andtumor invasiveness, such as Src/Rac-stimulating vitronectin(VTNC) and integrin alpha 5 (ITA5) [34], were quantifiedand shown to have low biological variability in the healthypopulation sampled here. A phosphopeptide from collagen-interacting tenascin-X was among ones with higher biologicalvariation and has been previously associated with breast can-cer proteome [35].
Several phosphopeptides identified in our recent studyexamining phosphoproteins in conditioned media of breastcancer cell lines [3] were also targeted here. Among thesewere phosphopeptides from CYTC and IBP5 specific forless-aggressive luminal-type tumors that showed high andlow interpersonal variability, respectively, by both MRM andSWATH-MS2 (Table 1). In addition, phosphopeptides thatwere characterized as basal breast cancer tumor specific fromOSTP, CD44, and IBP3 were quantified. While the phospho-peptide from OSTP with either phosphorylation at S308 orS310 showed low biological variability, phosphopeptides fromCD44 and IBP3 had higher interpersonal variation withinhealthy population.
4 Concluding remarks
Both MRM-MS and SWATH-MS2 methods for phospho-peptide quantification in human plasma were employedin this current study to evaluate phosphopeptide variationamong healthy individuals. The two methods produced re-producible relative phosphopeptide quantification data forpartially overlapping sets of phosphopeptides, with two-thirdsshowing biological variability of abundance consistent be-tween MRM and SWATH-MS2. Differences in interpersonalphosphorylation variability were shown for phosphopeptidesderived from proteins involved in maintaining hemostasis,immune response, cell surface interactions, diabetes path-ways, metabolism of lipids, and cell signaling pathways. Ourresults provide data that will be critical to first establishingthe base levels of protein phosphorylation variation withinhealthy human population. Such data will be critical for
C© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
10 A. M. Zawadzka et al. Electrophoresis 2014, 00, 1–11
future studies that might compare plasma samples amongcancer patients. For example, changes in phosphorylation ofproteins involved in signaling networks and cell-to-cell in-teractions may enable disease diagnosis and monitoring pro-gression of treatment. The developed phosphopeptide targetscan also complement already developed MRM-MS assays fornonphosphorylated plasma peptides [36, 37] and constitutea reference for selection of biomarker candidates. Indeed,the total number of phosphoproteins and defined phospho-sites that have been identified in human plasma to date hasincreased significantly just in the last year, including novelphosphosites identified in this current study [3, 7].
Our results also suggest that SWATH-MS2 analysis com-bined with phosphopeptide enrichment can provide repro-ducible and sensitive quantification of plasma phosphosites.Almost twice as many phosphopeptides were quantified fromhalf of the plasma volume compared to MRM-MS analysis.The SWATH-MS2 analysis did not require extensive methoddevelopment and phosphopeptide targets were not limitedto preselected ones like in the MRM-MS method. Additionalincorporation of heavy phosphopeptide standards, potentialsample preparation automation, or introducing a multiplexedStable Isotope Standard and Capture by Affinity-PeptideAntibodies (SISCAPA) approach would be expected to in-crease method sensitivity and technical reproducibility evenfurther.
We thank Lorri Reinders from the Buck Institute for helpwith preparation of plasma samples for immunodepletion and Dr.Michael McMaster from UCSF for help with obtaining plasmasamples. This work was supported by grants from the NationalCancer Institute, U24 CA126477 (S.J.F.) and a U24 Subcon-tract (B.W.G.) that are part of the NCI Clinical ProteomicTechnologies for Cancer initiative (http://proteomics.cancer.gov).We also acknowledge the support of several instruments fromthe NCRR shared instrumentation program, including the ABSCIEX QTRAP 5500 (S10 RR027953; B.W.G.), the QSTARElite (S10 RR024615; B.W.G.), and the TripleTOF 5600 (1S10OD016281; B.W.G.).
The authors have declared no conflict of interest.
5 References
[1] Blume-Jensen, P., Hunter, T., Nature 2001, 411, 355–365.
[2] Anderson, N. L., Anderson, N. G., Mol. Cell Proteomics2002, 1, 845–867.
[3] Zawadzka, A. M., Schilling, B., Cusack, M. P., Sahu, A. K.,Drake, P., Fisher, S. J., Benz, C. C., Gibson, B. W., Mol.Cell Proteomics 2014, 13, 1034–1049.
[4] Kelly-Spratt, K. S., Pitteri, S. J., Gurley, K. E., Liggitt, D.,Chin, A., Kennedy, J., Wong, C. H., Zhang, Q., Buson, T.B., Wang, H., Hanash, S. M., Kemp, C. J., PLoS One 2011,6, e19721.
[5] Carrascal, M., Gay, M., Ovelleiro, D., Casas, V., Gelpi, E.,Abian, J., J. Proteome. Res. 2010, 9, 876–884.
[6] Zhou, W., Ross, M. M., Tessitore, A., Ornstein, D., Van-meter, A., Liotta, L. A., Petricoin, E. F., 3rd, J. Proteome.Res. 2009, 8, 5523–5531.
[7] Jaros, J. A., Guest, P. C., Ramoune, H., Rothermundt,M., Leweke, F. M., Martins-de-Souza, D., Bahn, S., J. Pro-teomics 2012, 76, 36–42.
[8] Jaros, J. A., Martins-de-Souza, D., Rahmoune, H., Rother-mundt, M., Leweke, F. M., Guest, P. C., Bahn, S., J. Pro-teomics 2012, 76, 43–55.
[9] Picotti, P., Aebersold, R., Nat. Methods 2012, 9, 555–566.
[10] Gillet, L. C., Navarro, P., Tate, S., Rost, H., Selevsek, N.,Reiter, L., Bonner, R., Aebersold, R., Mol. Cell Proteomics2012, 11, O111 016717.
[11] Liu, Y., Huttenhain, R., Surinova, S., Gillet, L. C., Mourit-sen, J., Brunner, R., Navarro, P., Aebersold, R., Pro-teomics 2013, 13, 1247–1256.
[12] Held, J. M., Schilling, B., D’Souza, A. K., Srinivasan, T.,Behring, J. B., Sorensen, D. J., Benz, C. C., Gibson, B. W.,Int. J. Proteomics 2013, 2013, 791985.
[13] Keshishian, H., Addona, T., Burgess, M., Kuhn, E., Carr,S. A., Mol. Cell Proteomics 2007, 6, 2212–2229.
[14] McNulty, D. E., Annan, R. S., Mol. Cell Proteomics 2008,7, 971–980.
[15] Villen, J., Gygi, S. P., Nat. Protoc. 2008, 3, 1630–1638.
[16] Pappin, D. J., Hojrup, P., Bleasby, A. J., Curr. Biol. 1993,3, 327–332.
[17] Shilov, I. V., Seymour, S. L., Patel, A. A., Loboda, A.,Tang, W. H., Keating, S. P., Hunter, C. L., Nuwaysir, L. M.,Schaeffer, D. A., Mol. Cell Proteomics 2007, 6, 1638–1655.
[18] Perkins, D. N., Pappin, D. J., Creasy, D. M., Cottrell, J. S.,Electrophoresis 1999, 20, 3551–3567.
[19] MacLean, B., Tomazela, D. M., Shulman, N., Chambers,M., Finney, G. L., Frewen, B., Kern, R., Tabb, D. L., Liebler,D. C., MacCoss, M. J., Bioinformatics 2010, 26, 966–968.
[20] Abbatiello, S. E., Mani, D. R., Schilling, B., Maclean, B.,Zimmerman, L. J., Feng, X., Cusack, M. P., Sedransk, N.,Hall, S. C., Addona, T., Allen, S., Dodder, N. G., Ghosh,M., Held, J. M., Hedrick, V., Inerowicz, H. D., Jackson, A.,Keshishian, H., Kim, J. W., Lyssand, J. S., Riley, C. P., Rud-nick, P., Sadowski, P., Shaddox, K., Smith, D., Tomazela,D., Wahlander, A., Waldemarson, S., Whitwell, C. A.,You, J., Zhang, S., Kinsinger, C. R., Mesri, M., Rodriguez,H., Borchers, C. H., Buck, C., Fisher, S. J., Gibson, B. W.,Liebler, D., Maccoss, M., Neubert, T. A., Paulovich, A.,Regnier, F., Skates, S. J., Tempst, P., Wang, M., Carr, S.A., Mol. Cell Proteomics 2013, 12, 2623–2639.
[21] Ma, Z. Q., Polzin, K. O., Dasari, S., Chambers, M. C.,Schilling, B., Gibson, B. W., Tran, B. Q., Vega-Montoto,L., Liebler, D. C., Tabb, D. L., Anal. Chem. 2012, 84,5845–5850.
[22] Tabb, D. L., Vega-Montoto, L., Rudnick, P. A., Variyath,A. M., Ham, A. J., Bunk, D. M., Kilpatrick, L. E., Bill-heimer, D. D., Blackman, R. K., Cardasis, H. L., Carr, S.A., Clauser, K. R., Jaffe, J. D., Kowalski, K. A., Neubert,T. A., Regnier, F. E., Schilling, B., Tegeler, T. J., Wang,M., Wang, P., Whiteaker, J. R., Zimmerman, L. J., Fisher,S. J., Gibson, B. W., Kinsinger, C. R., Mesri, M., Ro-driguez, H., Stein, S. E., Tempst, P., Paulovich, A. G.,Liebler, D. C., Spiegelman, C., J. Proteome Res. 2010, 9,761–776.
C© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com
Electrophoresis 2014, 00, 1–11 Proteomics and 2DE 11
[23] Schilling, B., Rardin, M. J., MacLean, B. X., Zawadzka, A.M., Frewen, B. E., Cusack, M. P., Sorensen, D. J., Bere-man, M. S., Jing, E., Wu, C. C., Verdin, E., Kahn, C. R.,Maccoss, M. J., Gibson, B. W., Mol. Cell Proteomics 2012,11, 202–214.
[24] Sharma V., Eckels, J., Stergachis, A. B., Jaffe, J. D., Mac-Coss, M. J., 60th Annual ASMS Conference on MassSpectrometry & Allied Topics, Vancouver, Canada, 20–24 May 2012.
[25] Silliman, C. C., Dzieciatkowska, M., Moore, E. E., Kelher,M. R., Banerjee, A., Liang, X., Land, K. J., Hansen, K. C.,Transfusion 2012, 52, 417–424.
[26] Hortin, G. L., Sviridov, D., Anderson, N. L., Clin. Chem.2008, 54, 1608–1616.
[27] De Pergola, G., Silvestris, F., J. Obes. 2013, 2013, 291546.
[28] Louderbough, J. M., Schroeder, J. A., Mol. Cancer Res.2011, 9, 1573–1586.
[29] Malik, S., Fu, L., Juras, D. J., Karmali, M., Wong, B. Y.,Gozdzik, A., Cole, D. E., Crit. Rev. Clin. Lab. Sci. 2013, 50,1–22.
[30] Zhou, G. L., Zhang, H., Field, J., Cell Adh. Migr. 2014, 8,55–59.
[31] Sullivan, M. M., Sage, E. H., Int. J. Biochem. Cell Biol.2004, 36, 991–996.
[32] Polanski, M., Anderson, N. Leigh, Biomark. Insights2006, 2, 1–48.
[33] Hayashido, Y., Hamana, T., Ishida, Y., Shintani, T.,Koizumi, K., Okamoto, T., Oncol. Rep. 2007, 17,417–423.
[34] Noh, H., Hong, S., Huang, S., Theranostics 2013, 3,487–495.
[35] Zeng, Z., Hincapie, M., Pitteri, S. J., Hanash, S., Schalk-wijk, J., Hogan, J. M., Wang, H., Hancock, W. S., Anal.Chem. 2011, 83, 4845–4854.
[36] Anderson, L., Hunter, C. L., Mol. Cell Proteomics 2006,5, 573–588.
[37] Kuzyk, M. A., Smith, D., Yang, J., Cross, T. J., Jackson, A.M., Hardie, D. B., Anderson, N. L., Borchers, C. H., Mol.Cell Proteomics 2009, 8, 1860–1877.
C© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com