An Approach for Triplex-Isobaric Peptide Termini Labeling (Triplex

Jan 14, 2013 - Global cPILOT analysis of the APP/PS-1 mouse liver proteome ... Quantitative characterization of histone post-translational ... A novel...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/ac

An Approach for Triplex-Isobaric Peptide Termini Labeling (TriplexIPTL) Christian J. Koehler,† Magnus Ø. Arntzen,† Gustavo Antonio de Souza,‡ and Bernd Thiede*,† †

The Biotechnology Centre of Oslo, University of Oslo, P.O. Box 1125 Blindern, 0317 Oslo, Norway Department of Immunology, Oslo University Hospital Rikshospitalet and University of Oslo, 0424 Oslo, Norway

‡

S Supporting Information *

ABSTRACT: Isobaric peptide termini labeling (IPTL) is based on labeling of both peptide termini with complementary isotopic labels resulting in isobaric peptides. MS/MS analysis after IPTL derivatization produces peptide-specific fragment ions which are distributed throughout the MS/MS spectrum. Thus, several quantification points can be obtained per peptide. In this report, we present triplex-IPTL, a chemical labeling strategy for IPTL allowing the simultaneous quantification of three states within one MS run. For this purpose, dimethylation of the N-terminal amino group followed by dimethylation of lysines was used with different stable isotopes of formaldehyde and cyanoborohydride. Upon LC-MS/MS analysis, the combined samples revealed three corresponding isotopic fragment ion series reflecting quantitatively the peptide ratios. To support this multiplexing labeling strategy, we have further developed the data analysis tool IsobariQ and included multidimensional VSN normalization, statistical inference, and graphical visualization of triplex-IPTL data and clustering of protein profiling patterns. The power of the triplex-IPTL approach in combination with IsobariQ was demonstrated through temporal profiling of HeLa cells incubated with the kinesin Eg5 inhibitor S-Trityl-L-cysteine (STLC). As a result, clusters of quantified proteins were found by their ratio profiles which corresponded well to their gene ontology association in mitotic arrest and cell death, respectively.

S

identification and quantification of the peptides takes place on the MS/MS level after their isolation and fragmentation. Notably, at the MS (precursor ion) level, peptides originating from different samples colocalize in one peak. Thus, the complexity of the sample remains the same, independent of the number of different samples compared. Isobaric tagging for relative and absolute quantification (iTRAQ)6 and tandem mass tags (TMT)7 are reporter ion based isobaric methods which are composed of a reactive group to primary amines, a reporter and a balancer. During MS/MS fragmentation, the reporter ions with their unique masses are detected and can be assigned and used for relative quantification of the corresponding peptides. Reporter ion based isobaric quantification has disadvantages that contribute to reduced precision and accuracy: an unpredictable and variable level of signal compression is caused by the unavoidable coisolation of peptides during fragmentation,8 rare fragment ions of the same mass as reporter ions (‘phantom reporter ions’) will distort the quantification for that particular ion9 and isotopic impurities in the derivatization reagents need to be corrected for.8,10 In 2009, we addressed these issues with the introduction of isobaric peptide termini labeling (IPTL) as a novel approach

table isotope labeling of peptides and proteins has matured as a powerful tool for mass spectrometry-based quantitative proteomics. During the past decade several new techniques have been developed, and they can be classified into isotopic and isobaric labeling techniques.1 Using isotopic labeling techniques such as stable isotope labeling of amino acids in cell culture (SILAC),2 isotope-coded protein label (ICPL),3 or dimethylation,4 labeled peptides are altered in their precursor mass depending on the used isotopes. As a result, the same peptides from different samples appear as peaks with a mass shift in the MS spectrum. While the identification is typically achieved by the MS/MS spectrum, quantification is obtained by the comparison of the precursor peaks in the MS spectrum. A few technical drawbacks exist for isotopic quantification techniques. The number of peptides eluting from the column is at least doubled which increases with the number of simultaneous comparisons to be made and results in increased complexity of the MS spectra. As a consequence, interference of peaks is more likely and can complicate the quantification process. In addition, the ion intensities of the peptides are divided into different masses depending on their mass label and thus are less intense. Furthermore, the same peptides with different labels are often picked for MS/MS acquisition reducing the time to acquire other peptides of other proteins. Some of the disadvantages of isotopic labeling techniques can be overcome by isobaric labeling techniques.5 Here, the © 2013 American Chemical Society

Received: December 7, 2012 Accepted: January 14, 2013 Published: January 14, 2013 2478

dx.doi.org/10.1021/ac3035508 | Anal. Chem. 2013, 85, 2478−2485

Analytical Chemistry

■

for isobaric peptide quantification.11 IPTL is based on the use of two labels per peptide which are incorporated at the Nterminus and the C-terminus. The labels are chosen in isotopic variations so that the resulting net mass of all labels per peptide is equal, but the individual label on each peptide terminus is different. This strategy allows peptide identification and quantification within the same MS/MS spectrum. Fragment ions of the coeluting isobaric peptides result in peak pairs distributed throughout the whole mass range in the MS/MS spectrum which allows the method to be used with any mass analyzer, including Paul ion traps. The quantification points are directly coupled to the sequence which almost completely eliminates the interference of coisolated and cofragmented contaminating precursor. At the same time, several quantification points per peptide with reciprocal ratios for N-terminal and C-terminal fragments are obtained hence enabling statistical analysis of the data for each peptide and introducing a robustness of the method that contrasts to the single quantification points in reporter ion based isobaric tagging strategies. IPTL has been further developed since its introduction, and several variations of the chemistry have been published. The original IPTL labeling strategy was performed with endoproteinase Lys-C digests by chemical labeling of first the Cterminal lysines followed by the N-termini to achieve duplexIPTL. N-terminal succinylation followed by dimethylation of lysines enabled IPTL as a one pot reaction.12 The most costeffective labeling chemistry for IPTL was obtained by exploiting low pH-dependent site specific N-terminal dimethylation based as the first reaction.13 Other labeling strategies for IPTL allowed for the analysis of tryptic peptides using either oxygen18 labeling, guanidination, and dimethylation 14 or by combining SILAC and nonisobaric mTRAQ reagents.15 Moreover, index-ion triggered MS/MS ion quantification was developed for targeted quantification of peptides as a very interesting variation of the IPTL theme based on the combination of mTRAQ reagents and metabolic labeling.15 For IPTL data analysis, the software package IsobariQ was developed which permits the quantitative analysis of IPTL, iTRAQ, and TMT data.16 The performance of IPTL with the different MS/MS fragmentation techniques CID, ETD, and HCD was investigated, and optimal processing parameters were evaluated to increase the yield and the quality of quantitative information.17 Only a few methods for multiplex quantitative proteomics were reported. More than two samples can be compared using SILAC (typically up to 3),18 ICPL (4),3and dimethylation (3)19 for isotopic labeling and iTRAQ (8)20 and TMT (8)21 for reporter ion based isobaric labeling. In this manuscript, we present for the first time a labeling strategy for the comparison of three states using IPTL. Different combinations of isotopically labeled formaldehyde and cyanoborohydride were used for the site-specific dimethylation of the N-terminal amino group and subsequently for the dimethylation of the C-terminal lysine. This approach was validated using high resolution HCDMS/MS to identify and quantify proteins of HeLa cells exposed to the antimitotic inhibitor S-Trityl-L-cysteine (STLC) for 0 h (control), 16 h (mitotic arrest), and 40 h (apoptosis). The data analysis tool IsobariQ was further developed to allow for the quantitative analysis of triplex-IPTL data.

Article

EXPERIMENTAL PROCEDURES

Materials. Acetonitrile (MS grade) was purchased from Burdick Jackson, Seelze, Germany. Acetic acid, EDTA, endoproteinase Lys-C (sequencing grade), hydrochloric acid, and water (HPLC grade) were purchased from VWR, Oslo, Norway. Ammonium hydroxide, formaldehyde, formaldehyded2, formaldehyde-13C, formaldehyde-13Cd2, formic acid, iodoacetamide, methanol, sodium cyanoborodeuteride, sodium cyanoborohydride, transferrin human, triethylammonium hydrogen carbonate buffer, S-Trityl-L-cysteine, and trizma hydrochloride were bought from Sigma-Aldrich, Oslo, Norway. 1,4Dithiotreitol (DTT) and urea were purchased from Bio-Rad, Munich, Germany and dimethylsulfoxide from Roth, Karlsruhe, Germany. RPMI-1640 and fetal calf serum were obtained from Life Technologies, Oslo, Norway. Cell Culture and Induction of Mitotic Arrest and Apoptosis. HeLa cells were grown as a monolayer in RPMI1640 supplemented with 10% fetal calf serum and maintained in a humid incubator at 37 °C in a 5% CO2 environment to reach a density of 1 × 105 cells/mL. Five × 106 cells (50 mL) were treated with 5 μM STLC from a 5 mg/mL stock in DMSO for 0 h, 16 h, and 40 h. Cells were trypsinized, harvested, resuspended in 1 mL PBS, and centrifuged at 200 g. Cell pellets were frozen and stored in liquid nitrogen. Cell Lysis, Protein Precipitation, and in-Solution Lys-C Digestion. Pellets of cells treated with STLC for 0 h, 16 h, and 40 h were thawed on ice, and 800 μL SILAC Phosphoprotein lysis buffer B (Invitrogen, Oslo, Norway) was added. The cell slurry was homogenized with a pestle (20x) for mechanical breakage of the cells followed by sonication using an Ultrasonic processor (UP400s, Dr. Hielscher). Samples were centrifuged at 16,000 g for 20 min at 4 °C in a Heraeus Biofuge pico (Kendro, Hanau, Germany), and the supernatant was aliquoted in 40 μL aliquots. One aliquot of each time point was further used. To each aliquot, 400 μL of ice cold acetone/methanol (1:1) acidified with 0.1% hydrochloric acid (v/v) was added, vortexed, and precipitated at −20 °C overnight. Samples were centrifuged at 16,000 g for 20 min at 4 °C (Heraeus Biofuge pico), and the supernatant was discarded. Proteins were redissolved in 100 μL of 8 M urea in Lys-C buffer (25 mM Tris pH 8.5, and 1 mM EDTA). For reduction and alkylation of the cysteines, 5 μL of 200 mM DTT was added, and the samples were incubated at 37 °C for 1 h followed by the addition of 20 μL of 200 mM iodoacetamide for 1 h at room temperature in the dark. The alkylation reaction was quenched by adding 20 μL of 200 mM DTT. The proteins were digested with Lys-C (6.6 μg) in a final volume of 225 μL for 16 h at 37 °C. The digestion was stopped by adding 100 μL of 1% formic acid, and the generated peptides were purified using a Strata C18-E SPE column (Phenomenex, Værløse, Denmark) and dried using a Speed Vac concentrator (Savant, Holbrook, NY, USA). Triplex-IPTL Labeling. For N-terminal labeling, 100 μL of 1% acetic acid, pH 2.8 was directly added to the dried peptides and mixed. Subsequently, 4 μL of 4% formaldehyde, formaldehyde-d2, or formaldehyde-13Cd2 in water was added and thoroughly mixed followed by 4 μL of 600 mM sodium cyanoborohydride or sodium cyanoborodeuteride and incubated for 5 min using a thermomixer (Eppendorf, Hamburg, Germany). Subsequently, 16 μL of 1% ammonium hydroxide was added, vortexed, and incubated for one minute before 2479

dx.doi.org/10.1021/ac3035508 | Anal. Chem. 2013, 85, 2478−2485

Analytical Chemistry

Article

Table 1. Overview of the Satellite Neutral Loss Definitions in Mascota ″H″ heavy

″M1″ medium

CD2O NaBD3CN 36.075670 Da 13C(2) 2H(6) H(−2) 13C(2) 2H(6) C(−2) H(−6) 13C(2) 2H(2) C(−2) H(−2) 2H(4) H(−4) -

CD2O NaBH3CN 32.056407 Da 2H(4) C(2) 2H(4) H(−4) 13C(−2) 2H(−2) C(2) H(2)

modification formaldehyde cyanoborohydride monoisotopic mass composition neutral loss satellite neutral loss satellite neutral loss satellite neutral loss satellite

13

″L″ ″M1″ ″M2″ ″H″

″M2″ medium

″L″ light

CH2O NaBD3CN 32.050563 Da 13C(2) 2H(2) H(2) 13C(2) 2H(2) C(−2) H(−2) 2H(−4) H(4)

CH2O NaBH3CN 28.031300 Da C(2) H(4) 2H(−4) H(4) 13C(−2) 2H(−2) C(2) H(2) 13C(−2) 2H(−6) C(2) H(6)

13

a The heavy “H” and light “L” label require three neutral loss satellite ions, whereas neutral losses to match medium “M1” and “M2” peaks only require definitions for the N-terminal and lysine fragment ions, respectively.

adding 8 μL of 5% formic acid. Finally, the samples were purified using a C18-E SPE column (Phenomenex, Værløse, Denmark) and evaporated to dryness. Subsequently, the same procedure as described above was performed again for Cterminal labeling except using 100 μL of 200 mM triethylammonium bicarbonate instead of 100 μL of 1% acetic acid, pH 2.8. For evaluation of reaction specificity by MALDIMS, the same procedure was applied except that a quarter of all volumes was used to perform the labeling and μC18-ZipTip (Eppendorf, Hamburg, Germany) replaced the C18-E SPE column for sample purification. OFFGEL Fractionation of Triplex-IPTL Labeled Peptides. Combined triplex-IPTL labeled peptides of the complex HeLa samples were separated according to their isoelectric point using a 13 cm IGP DryStrip, pH 3−10 (GE Healthcare, Uppsala, Sweden) on a 3100 OFFGEL fractionator (Agilent, Santa Clara, CA, USA). Peptides were separated according to the manufacturer’s manual except that the concentration of glycerol was halved to 6% (v/v). For peptide separation, the predefined method OG12PE01 was used (voltage: 4.5 kV, current: 50 μA, power: 200 mW, time: 100 h). Subsequently, the peptides were purified using μ-C18 ZipTips (Millipore, Billerica, MA, USA), evaporated to dryness, and reconstituted in 1% formic acid, 2% acetonitrile in water for MS analysis. Mass Spectrometry. The experiments for the analysis of the HeLa proteomes were performed using a Dionex Ultimate 3000 nano-UHPLC system (Sunnyvale, CA, USA) connected to a quadrupole − Orbitrap (QExactive) mass spectrometer (ThermoElectron, Bremen, Germany) equipped with a nano electrospray ion source (Proxeon/Thermo). For liquid chromatography separation, an Acclaim PepMap 100 column (C18, 2 μm beads, 100 Å, 75 μm inner diameter) (Dionex, Sunnyvale CA, USA) capillary of 15 cm bed length was used. The flow rate was 0.3 μL/min, with a solvent gradient of 7% B to 32% B in 60 min. Solvent A was aqueous 2% acetonitrile in 0.1% formic acid, whereas solvent B was aqueous 90% acetonitrile in 0.1% formic acid. The mass spectrometer was operated in the data-dependent mode to automatically switch between MS and MS/MS acquisition. Survey full scan MS spectra (from m/z 300 to 1.750) were acquired in the Orbitrap with resolution R = 70,000 at m/z 200 (after accumulation to a target of 1,000,000 ions in the C-trap). The method used permitted sequential isolation of the most intense multiply charged ions, up to ten, depending on signal intensity, for fragmentation on the HCD cell using high-energy collision dissociation at a target value of 1,000,000 charges or maximum injection time of 128 ms (fixed injection time method).23 MS/MS scans were collected at 35,000 resolution at the Orbitrap cell. Target ions already

selected for MS/MS were dynamically excluded for 60 s. General mass spectrometry conditions were as follows: electrospray voltage, 2.4 kV; no sheath and auxiliary gas flow, heated capillary temperature of 250 °C, normalized HCD collision energy 30%. Ion selection threshold was set to 10,000 counts, and isolation width of 2.0 Da was used. To validate the specificity of the N-terminal dimethylation reaction and the subsequent lysine dimethylation, a few recombinant proteins digested with endoproteinase Lys-C were labeled by the three triplex-IPTL reactions. Each sample was analyzed during every step of the labeling procedure using an Ultraflex II (Bruker Daltonics, Bremen, Germany) MALDITOF/TOF mass spectrometer after external calibration as previously described.22 Furthermore, human transferrin was analyzed on an nanoLC-MS system using a Dionex Ultimate 3000 nano-UHPLC system (Sunnyvale, CA, USA) connected to a LTQ Orbitrap XL mass spectrometer (ThermoElectron, Bremen, Germany) to investigate the coelution of the three labeling reactions of the triplex-IPTL approach. Therefore, precursor masses of peptides of Lys-C digested human transferrin were selected for constant MS/MS fragmentation over a 15 min gradient, and the TIC of the selected precursors were compared against the XICs of IPTL peaks from selected IPTL-triplexes. The chromatograms were smoothed using boxcar smoothing with 3 points in the Xcalibur Qual Browser 2.2 SP1.48 (ThermoElectron, Bremen, Germany). Raw File Conversion and Mascot Search Parameters. Raw files were processed to generate peak lists in Mascot generic format (*.mgf) using ProteoWizard release version 3.0.331 (ProteoWizard MSData 3.0.3827; Proteowizard Analysis 3.0.3703) with the following filter parameters: --mgf --filter ″peakPicking true 2″, --filter MS2Deisotope --filter ″titleMaker < RunId>;; ; ;″. The resulting peaklists files from the 12 OFFGEL fractions were merged into one combined peaklist file using an in-house Python script. Database searches were performed using Mascot in-house version 2.3.02 to search from Swiss-Prot (11.2011, human, 20,252 sequences). Lys-C was selected as enzyme with zero missed cleavage site, and a tolerance of 10 ppm for the precursor ion and 0.05 Da for HCD-MS/MS fragments was applied. Carbamidomethylation of cysteines was set as fixed modification, while N-terminal protein acetylation was allowed as variable modification. Furthermore, four chemical labels of dimethylation with the composition C(2)H(4) for light “L”, 2H(4)C(2) for medium “M1″, 13C(2)2H(2)H(2) for medium “M2”, and 13C(2)2H(6)H(−2) for heavy “H” were used as variable modifications. The light and heavy labels were permitted as N-terminal and lysine specific, while the medium 2480

dx.doi.org/10.1021/ac3035508 | Anal. Chem. 2013, 85, 2478−2485

Analytical Chemistry

Article

Figure 1. The triplex-IPTL approach. A) Different variations of isotopically labeled formaldehyde and cyanoborohydride for dimethylation result in labels of different masses. The combination of label “L” and “H” and “M1” and “M2″, respectively, results in peptides of the same mass and containing the same number of deuterium isotopes. B) Triplex-IPTL labeling of three samples can be achieved by combining the labels (N-/Cterminus) L/H, M1/M2, and H/L. C) The MS/MS fragmentation pattern and correct peak assignments for b- and y-ions is depicted.

label “M1” was N-terminal specific only and the medium label “M2” lysine-specific only. The four dimethyl labels were extended with possible satellite neutral loss definitions to match peaks from a different label state (Table 1). Automatic decoy database searches were performed in Mascot to reveal a false discovery rate (FDR) for identifications. Quantitative Data Analysis. Quantitative data analysis was performed using IsobariQ version 2.0a. The import wizard of IsobariQ was used to load the data set into the main module with protein significance p < 0.05 from Mascot. No sequence suggestion below an ion score of 15 was loaded, and the minimal number of ratios for force-find hits was set to four.12 After import, the data were quantified with the following options: All labeling schema were required for peptide quantification, protein scoring was set to MudPit, and the MS/MS tolerance was set to 0.05 Da. Bold peptides (Mascot option) were required for the quantification, and unique and razor peptides were used17 with minimum two quantifiable peptides per proteins. As a note, quantifiable means that an MS/MS could be quantified but was not necessarily chosen to be used. The “normalize raw files independently” checkbox was not checked, Grubbs outlier detection was stringent, and only the top 12 most-intense triplex fragments in one MS/MS spectrum were used. The data were normalized using the multidimensional VSN method, and the quantified proteins were clustered first based on ratio significances and the significant clusters were further clustered based on protein profiles into five distinct groups using k-means. Gene ontology (GO) version 1.3499 and GO annotations (GOA) version 11/ 12/2012, both downloaded from http://www.geneontology. org, were utilized to calculate the number of proteins associated

with cell death (GO:0008219) and mitotic cell cycle (GO:0000278) by an in-house python script. Hypergeometric tests were applied to assess the statistical significance of the enrichment using the GO-annotated human genome in UniProtKB/Swiss-Prot as background (18,438 proteins).

■

RESULTS The Triplex-IPTL Labeling Strategy. IPTL is based on differential isotopic labeling of both peptide termini which results in isobaric intact peptide masses and quantifiable peptide fragment ions. A prerequisite for the development of the triplex-IPTL approach was to keep the terminal modifications of three states different in mass but at the same time balance the total mass of the labeled peptides. In addition, the number of deuterium and carbon-13 isotopes for all three states must be equal as coelution of peptides in the LC-system is strictly required for isobaric quantification as an integration of peak intensities over the elution profile is not applicable for the single MS/MS events. These requirements can be achieved via N-terminal specific dimethylation with subsequent dimethylation of the epsilon amino group of C-terminal lysines using different combinations of four stable isotopes of formaldehyde and two stable isotopes of cyanoborohydride (Figure 1). An outline of the triplex-IPTL approach with the detailed chemical combinations of the used isotopic labeling reagents is displayed in Figure 1A. The light label ″L″ is achieved through the dimethylation of an amino group utilizing formaldehyde and cyanoborohydride and results in a mass label of 28.0313 Da. The counter label to the light label is the heavy dimethylation indicated with an ″H″ using formaldehyde-13Cd2 and cyanoborodeuteride resulting in a mass label of 36.07567 Da. 2481

dx.doi.org/10.1021/ac3035508 | Anal. Chem. 2013, 85, 2478−2485

Analytical Chemistry

Article

Figure 2. HCD-MS/MS spectrum of a triplex-IPTL labeled peptide. The MS/MS spectrum of the peptide SQEQLAAELAEYTAK from human ezrin is displayed. The triplex-fragments are visible throughout the whole b- and y-ion series resulting in more than 20 quantification points for this peptide. The zoomed area between m/z 850 and m/z 1,050 shows the 4 Da mass intervals between the labels L, M1, and H for b-ions and L, M2, and H for y-ions.

according to their individual N-terminal label after the first reaction and to isobaric masses after the completion of the second labeling step (Supplementary Figure 1). The specificity of the two reactions was validated by MS/MS fragmentation as shown for the peptide SASDLTWDNLK of human transferrin (Supplementary Figure 2). In addition, the same peptide was monitored by LC-MS/MS using a 15 min gradient, and the extracted ion chromatograms of the y6-ions were compared over the elution profile of the peptide revealing that the individual labeled peptides coeluted during reversed-phase C18 liquid chromatography (Supplementary Figure 3). One unique feature of triplex-IPTL is the appearance of multiple peaks per peptide fragment ion where each individual peak represents one labeling state of the sample. During identification, Mascot allocates peptide sequences to the MS/ MS spectrum, where the highest score is ranked to the most probable. In a triplex-IPTL MS/MS spectrum, three matches can be assigned to the corresponding peptide differing in masses according to the variability of N-terminal and Cterminal modifications. In each of these peptide matches, only one set of peaks from the MS/MS spectrum can be matched, leaving the remaining peaks unidentified which results in a scoring penalty compared to a similar unlabeled peptide. To avoid this scoring penalty, satellite neutral loss ion must be defined (Table 1).17 Implementation of Multiplex Data Analysis in IsobariQ. IsobariQ was developed for the data analysis of duplex-IPTL, iTRAQ, and TMT data.16 To accommodate the triplex-IPTL labeling strategy, IsobariQ needed fundamental changes. With the incorporation of multiplexing in IPTL, every peptide spectrum will have three computed ratios, each with separate statistics, normalization, and significances. In the initial version of IsobariQ the multiplexing functionality was included for the reporter ion based techniques, but every ratio was treated independently. However, this is not optimal for normalization, as the different ratios may be dependent on

Together, both labels (L and H) result in a total mass label of 64.10697 Da introducing two carbon-13 and six deuterium isotopes into the labeled peptide. The second combination of labels is achieved through two medium dimethylation reactions indicated with ″M1″ and ″M2″. While the dimethylation ″M1″ uses formaldehyde-d2 in combination with cyanoborohydride resulting in a mass label of 32.056407 Da contributing four deuterium isotopes to the labeled peptide, the second medium dimethylation ″M2″ is achieved through the combination of formaldehyde-13C and cyanoborodeuteride with a mass of 32.050562 Da introducing another two deuterium and two carbon-13 isotopes. In sum, the two medium dimethylation reactions balance the mass and deuterium isotopes to the same stoichiometry as the first combination of the heavy and light label to 64.10697 Da with a total of two carbon-13 and six deuterium isotopes (Figure 1A). The combination of this labeling strategy results in triplex-IPTL labeling as displayed in Figure 1B. For simplicity, triplex-IPTL labeled peptides (light, medium, and heavy, respectively) were defined according to the N-terminal label. The light labeled sample 1 is modified using a light label “L” at the N-terminus and the heavy label ″H″ at the C-terminus, whereas these labels were swapped for the heavy labeled sample 3. For the medium labeled sample 2, ″M1″ at the N-terminus and ″M2″ at the C-terminus were used (Figure 1B). The sample assignment to peaks from N-terminal fragment ions represented by b-ions and C-terminal fragment ions represented by y-ions are displayed in Figure 1C. MS/MS fragment ions appear as triplexes with a mass distance of 4 Da. The succession from low to high m/z within these triplexes of the corresponding samples is opposite for b- and y-ions (Figure 1C). As an example for an HCD-MS/MS spectrum, a peptide corresponding to ezrin derived from HeLa cells exposed to STLC for different time points is shown in Figure 2. The specificity of the two labeling reactions was validated by applying of triplex-IPTL to Lys-C digested standard proteins and subsequent MALDI-MS analysis. All peptide masses shifted 2482

dx.doi.org/10.1021/ac3035508 | Anal. Chem. 2013, 85, 2478−2485

Analytical Chemistry

Article

Figure 3. IsobariQ data visualization and analysis. A) A ternary graph displays the distribution of all quantified proteins according to their H/M/L abundances. B) Five distinct clusters (I−V) were found by clustering of the quantified proteins based on their ratio profiles. When a protein cluster is selected, the grouped proteins are highlighted in the ternary graph (inset) to visualize the cluster in relation to the rest of the population (inset, red). The green line denotes the cluster median. C) An overview of GO term analysis is shown of the proteins found in the protein clusters I−V for “cell death” and “mitotic cell cycle”. The size of the circle reflects the relative representation of the proteins associated with the respective GO term within each cluster. Solid filling indicates statistical significance of overrepresentation (p < 0.05).

quantified, and clustered, and five distinct clusters were observed (Figure 3B). The five clusters comprehended proteins with profiles at 16 h/40 h with up−up (I), down−down (II), up−down (III), down−up (IV), and down-unchanged (V) (Figure 3B). For each of these clusters, GO functional association and overrepresentation analysis was performed for the two groups ‘cell death’ and ‘mitotic cell cycle’ by an inhouse python script (Figure 3C). This figure indicates that proteins associated with ‘cell death’ are present in all clusters, but dominating in cluster I and proteins associated with ‘mitotic cell cycle’ are also present in all clusters, but dominating in clusters III and IV. A complete list of the proteins in each cluster, their respective ratios, and corresponding GO association to “cell death” and/or “mitotic cell cycle” are given in Supplementary Table 1. Furthermore, this table indicates the presence of each protein previously reported to be involved in apoptosis according to the ApoptoProteomics database.25

each other. Heavy/light and medium/light both use the light as denominator and if heavy/light is normalized this would affect the medium/light ratios as well. Variance stabilizing normalization (VSN) as incorporated in the vsn-package in R24 includes this multidimensional normalization and is now reimplemented into IsobariQ for both IPTL and reporter ion based techniques. After all protein computations are complete, IsobariQ now has the ability to cluster proteins into groups based on the protein significance, protein profiles (protein ratio trend), or user selections by utilizing intercommunication with R and the k-means Hartigan and Wong algorithm. For each protein or protein cluster selected, IsobariQ displays for triplex-IPTL data the information in two new graphs, a protein profile graph and a ternary graph. The ternary graph displays data with three dimensions in one 2D figure (Figure 3A). When a protein cluster is selected, the grouped proteins are highlighted in the ternary graph (Figure 3B) to visualize the cluster in relation to the rest of the population. This kind of presentation particularly makes sense in combination with protein profile clustering as shown in Figure 3B where the distinct profiles display different locations on the ternary graph. The latest version of IsobariQ can be downloaded free of charge from http://www.biotek.uio. no/english/research/groups/thiede-group/software. Temporal Proteome Profiling of HeLa Cells Incubated with STLC Using Triplex-IPTL. HeLa cells exposed to the antimitotic inhibitor STLC for 0 h (control), 16 h (mitotic arrest), and 40 h (apoptosis) were used as a model to validate the feasibility of the triplex-IPTL approach applied to the analysis of a complex proteome. Triplex-IPTL labeling was performed using light labeling on the 0 h sample, medium labeling on the 16 h sample, and heavy labeling on the 40 h sample. The three samples were mixed and further separated by OFFGEL fractionation between pI 3−10. Twelve fractions were generated, each analyzed using an UHPLC-quadrupole Orbitrap LC-MS system and raw files processed and searched using Mascot. In IsobariQ, the proteins were normalized,

■

DISCUSSION Isobaric labeling methods for peptide quantification have some advantages in comparison to isotopic labeling techniques. Due to the isobaric masses of differentially labeled peptides, the complexity of the proteomic sample does not increase by the factor of comparisons as it does for isotopic labeling strategies. Consequently, the intensities of peaks derived from the same peptide of parallel analyzed samples are summed, and low abundancy peptides can exceed the limit of detection. MS/MS spectra typically contain less noise than MS spectra as fragmentation peaks derive from isolated peaks of a small mass window. Isobaric labeling approaches in mass spectrometry-based quantification include either reporter ion based labeling such as iTRAQ, and TMT or counterbalanced peptide terminal labels using IPTL. Multiplexing in quantitative proteomics is favorable as more information can be acquired in parallel which reduces resources, cost, and time. At the same 2483

dx.doi.org/10.1021/ac3035508 | Anal. Chem. 2013, 85, 2478−2485

Analytical Chemistry

Article

peaks which happen to match the exact mass of a ratiodetermining peak.9 For IPTL, the presence of multiple quantification points per MS/MS spectrum allows the recognition and removal of aberrant ratios through statistical outlier testing. Furthermore, quantification points are distributed over the whole mass range using IPTL. This distribution permits the use of any kind of mass analyzer and overcomes a limitation of reporter ion based isobaric approaches where ion traps are not feasible due to the 1/3 low mass cutoff. However, the protein C-terminal peptide if it is not a lysine, peptides containing internal lysine residues, and modified N-termini and lysines are not accessible for isobaric quantification by IPTL. STLC has been shown to be an allosteric inhibitor of the mitotic kinesin Eg5 with potential as an antimitotic chemotherapeutic agent.27 The primary function of Eg5 is to form the bipolar spindle during early prometaphase, and failure in separating the duplicated centrosomes leads to mitotic arrest by activation of the spindle checkpoint and ultimately cell death.28 Eg5 is essential for cell replication and thus ubiquitously expressed in proliferating tumor cells which make it a desired target for cancer therapy. Temporal protein profiling can be used to classify proteins into functional categories based on their relative abundances at different time points as described by Bull et al.29 In brief, if a maximum or minimum abundance is allocated to a specific time point by temporal protein profiling, the protein may be assigned to the process occurring at that time point. By applying this classification strategy to our data, proteins with involvement in mitotic arrest should be in clusters III and IV, while proteins involved in apoptosis should be in clusters I and II (Figure 3B). To assess this question we extracted the associated GO terms for every protein in each cluster and controlled if any of them were children of the two terms ‘cell death’ and ‘mitotic cell cycle’. By this means, we observed that the proportion of cell death associated proteins was largest in cluster I (25%), but they were also present in all other clusters (Figure 3C). Furthermore, many cell death associated proteins were found among the proteins with no significant regulation (data not shown). During cell death by apoptosis, several hundreds of proteins were reported to undergo cleavage by caspases.30 Actually, proteomics studies applied to study apoptosis revealed caspase substrates as the most frequently reported regulated proteins.25 In this experiment we have not performed any separation by molecular weight such as SDS-PAGE, thus differentiation between a full length protein and its cleaved version was not possible. Therefore, the proteins that show up- or down-regulation at the time of cell death (clusters I and II) should be unaffected by caspase cleavage. For proteins assigned to the mitotic cell cycle, we found that the dominating clusters are III and IV, corroborating the classification strategy described by Bull et al. Only very few of the detected proteins assigned to this category did not change abundance during the course of our experiment (data not shown). In conclusion, we have developed a chemical labeling strategy to perform triplex-IPTL experiments as shown for the temporal proteome profiling of HeLa cells exposed to STLC. The necessary features required for triplex-IPTL data analysis were implemented into the analysis software IsobariQ including cluster analysis which can be applied to visualize and analyze proteomic profiles.

time, it allows for higher confidence in the results by parallel processing of replicates or temporal profiling of proteins. While multiplexing is common for reporter ion based isobaric labeling techniques, it has not been reported with IPTL. Prerequisites for IPTL are the ability to selectively label the peptide termini, a minimum mass difference of 4 Da to avoid interference with the isotopic clusters of the peptide, and the number of used isotopes must be equal for all samples to exclude LC retention time shifts. Amino groups are highly reactive and best suited for the modification of peptides and can be generated on the C-terminus of peptides using Lys-C as endoproteinase. Using amino groups at both termini, at least one chemical modification reaction has to be highly selective. Qin et al. showed that N-terminal selective dimethylation of peptides is possible using 1% acetic acid (pH 2.8) and applied this approach to perform duplex-IPTL.13 In this report, we confirmed the specificity and extended this approach and presented triplex-IPTL labeling which can be achieved by combinations of different stable isotopes of formaldehyde and sodium cyanoborohydride. While the second label for the counter-terminus does not need the same selectivity as the first reaction, it still must be available in a stable isotopic variant with the same mass difference to the counter label. Consequently, we selected dimethylation reactions to create 4 Da mass differences between the light, medium, and heavy label (Figure 1A). To achieve isobaric labels with the same chemical composition and the same number of isotopes of the summed labels, the two medium labels must differ in their individual chemical composition (Figure 1A). Furthermore, we confirmed that triplex-IPTL labeled peptides coelute during chromatography because this is a prerequisite for quantification using IPTL. To avoid scoring penalties in Mascot, it was necessary that each modification was defined with satellite neutral losses to match the peaks of all three labels. We differentiated between the two medium labels as the mass difference of 0.0058 Da is in the range of the mass accuracy of a labeled peptide (5 ppm at 1,000 m/z) although the data searches were carried out with a higher tolerance. However, IPTL labeling provides a 2- to 3fold increase in sequence-describing peaks in MS/MS spectra, a fact that is currently not reflected in the process of calculating the probability of a correct match. Furthermore, the presence and identification of second peptides in MS/MS spectra has been described for the Andromeda search engine26 and could be a possible opportunity for improving the outcome of identification and quantification of IPTL data. Inaccuracies of mass spectrometry-based peptide quantifications can be caused by interferences between the peptide and coeluting peaks within the same m/z window at a given elution time. On the one hand, this interference alters the intensity of the peak in the MS trace of isotopic labeling techniques. On the other hand, coisolation and cofragmentation of such peaks distorts the intensities in reporter ion based isobaric labeling techniques. This phenomenon is referred to as peak compression or as the dampening effect and has been extensively studied with iTRAQ and TMT.8,10 This effect does not exist for IPTL as the quantitatively coded fragments are directly coupled to the peptides sequence with their terminal modifications. Co-isolated and fragmented contamination peaks would result in different peaks where it is very unlikely that several fragments are congruent. Another contributor to inaccuracy and imprecision in reporter ion based isobaric labeling strategies is the occurrence of phantom 2484

dx.doi.org/10.1021/ac3035508 | Anal. Chem. 2013, 85, 2478−2485

Analytical Chemistry

■

Article

Zou, X.; Biggs, C. A.; Wright, P. C. Anal. Bioanal. Chem. 2012, 404, 1011−1027. (11) Koehler, C. J.; Strozynski, M.; Kozielski, F.; Treumann, A.; Thiede, B. J. Proteome Res. 2009, 8, 4333−4341. (12) Koehler, C. J.; Arntzen, M. O.; Strozynski, M.; Treumann, A.; Thiede, B. Anal. Chem. 2011, 83, 4775−4781. (13) Qin, H.; Wang, F.; Zhang, Y.; Hu, Z.; Song, C.; Wu, R.; Ye, M.; Zou, H. Chem. Commun. 2012, 48, 6265−6267. (14) Yang, S. J.; Nie, A. Y.; Zhang, L.; Yan, G. Q.; Yao, J.; Xie, L. Q.; Lu, H. J.; Yang, P. Y. J. Proteomics 2012, 75, 5797−5806. (15) Yan, W.; Luo, J.; Robinson, M.; Eng, J.; Aebersold, R.; Ranish, J. Mol. Cell. Proteomics 2011, 10, M110.005611. (16) Arntzen, M. O.; Koehler, C. J.; Barsnes, H.; Berven, F. S.; Treumann, A.; Thiede, B. J. Proteome Res. 2011, 10, 913−920. (17) Koehler, C. J.; Arntzen, M. O.; Treumann, A.; Thiede, B. Anal. Bioanal. Chem. 2012, 404, 1103−1114. (18) Blagoev, B.; Ong, S. E.; Kratchmarova, I.; Mann, M. Nat. Biotechnol. 2004, 22, 1139−1145. (19) Boersema, P. J.; Aye, T. T.; van Veen, T. A.; Heck, A. J.; Mohammed, S. Proteomics 2008, 8, 4624−4632. (20) Choe, L.; D’Ascenzo, M.; Relkin, N. R.; Pappin, D.; Ross, P.; Williamson, B.; Guertin, S.; Pribil, P.; Lee, K. H. Proteomics 2007, 7, 3651−3660. (21) Werner, T.; Becher, I.; Sweetman, G.; Doce, C.; Savitski, M. M.; Bantscheff, M. Anal. Chem. 2012, 84, 7188−7194. (22) Schmidt, F.; Fiege, T.; Hustoft, H. K.; Kneist, S.; Thiede, B. Proteomics 2009, 9, 1994−2003. (23) Kelstrup, C. D.; Young, C.; Lavallee, R.; Nielsen, M. L.; Olsen, J. V. J. Proteome Res. 2012, 11, 3487−3497. (24) Huber, W.; von Heydebreck, A.; Sultmann, H.; Poustka, A.; Vingron, M. Bioinformatics 2002, 18 (Suppl 1), S96−104. (25) Arntzen, M. O.; Thiede, B. Mol. Cell. Proteomics 2012, 11, M111.010447. (26) Cox, J.; Neuhauser, N.; Michalski, A.; Scheltema, R. A.; Olsen, J. V.; Mann, M. J. Proteome Res. 2011, 10, 1794−1805. (27) Abualhasan, M. N.; Good, J. A.; Wittayanarakul, K.; Anthony, N. G.; Berretta, G.; Rath, O.; Kozielski, F.; Sutcliffe, O. B.; Mackay, S. P. Eur. J. Med. Chem. 2012, 54, 483−498. (28) Wang, F.; Good, J. A.; Rath, O.; Kaan, H. Y.; Sutcliffe, O. B.; Mackay, S. P.; Kozielski, F. J. Med. Chem. 2012, 55, 1511−1525. (29) Bull, V. H.; Fargestad, E. M.; Strozynski, M.; Thiede, B. Electrophoresis 2010, 31, 1873−1885. (30) Rotilio, D.; Della Corte, A.; D’Imperio, M.; Coletta, W.; Marcone, S.; Silvestri, C.; Giordano, L.; Di Michele, M.; Donati, M. B. Thromb. Res. 2012, 129, 257−262.

ASSOCIATED CONTENT

S Supporting Information *

Supplementary Figures 1−3 and Table 1. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +47-22840533. Fax +47-22840501. E-mail: bernd. [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the National Program for Research in Functional Genomics in Norway (FUGE, project no. 183418/S10) of the Norwegian Research Council, FUGEØst, and Inven2 to BT. M.A. would like to thank Wolfgang Huber (EMBL, Genome Biology Unit, Heidelberg, Germany) for discussions on VSN implementation. We thank Achim Treumann, NUPPA, University of Newcastle for his suggestions to improve this manuscript.

■

ABBREVIATIONS CID collision induced dissociation ETD electron transfer dissociation HCD higher-energy collisional dissociation FDR false discovery rate ICPL isotope-coded protein label IPTL isobaric peptide termini labeling iTRAQ isobaric tagging for relative and absolute quantification SILAC stable isotope labeling with amino acids in cell culture STLC S-Trityl-L-cysteine TIC total ion chromatogram TMT tandem mass tagging UHPLC ultra high performance liquid chromatography VSN variance stabilizing normalization XIC extracted ion chromatogram

■

REFERENCES

(1) Bantscheff, M.; Lemeer, S.; Savitski, M. M.; Kuster, B. Anal. Bioanal. Chem. 2012, 404, 939−965. (2) Ong, S. E.; Blagoev, B.; Kratchmarova, I.; Kristensen, D. B.; Steen, H.; Pandey, A.; Mann, M. Mol. Cell. Proteomics 2002, 1, 376− 386. (3) Schmidt, A.; Kellermann, J.; Lottspeich, F. Proteomics 2005, 5, 4− 15. (4) Hsu, J. L.; Huang, S. Y.; Chow, N. H.; Chen, S. H. Anal. Chem. 2003, 75, 6843−6852. (5) Treumann, A.; Thiede, B. Expert Rev. Proteomics 2010, 7, 647− 653. (6) Ross, P. L.; Huang, Y. N.; Marchese, J. N.; Williamson, B.; Parker, K.; Hattan, S.; Khainovski, N.; Pillai, S.; Dey, S.; Daniels, S.; Purkayastha, S.; Juhasz, P.; Martin, S.; Bartlet-Jones, M.; He, F.; Jacobson, A.; Pappin, D. J. Mol. Cell. Proteomics 2004, 3, 1154−1169. (7) Thompson, A.; Schafer, J.; Kuhn, K.; Kienle, S.; Schwarz, J.; Schmidt, G.; Neumann, T.; Johnstone, R.; Mohammed, A. K.; Hamon, C. Anal. Chem. 2003, 75, 1895−1904. (8) Ow, S. Y.; Salim, M.; Noirel, J.; Evans, C.; Rehman, I.; Wright, P. C. J. Proteome Res. 2009, 8, 5347−5355. (9) Christoforou, A. L.; Lilley, K. S. Anal. Bioanal. Chem. 2012, 404, 1029−1037. (10) Evans, C.; Noirel, J.; Ow, S. Y.; Salim, M.; Pereira-Medrano, A. G.; Couto, N.; Pandhal, J.; Smith, D.; Pham, T. K.; Karunakaran, E.; 2485

dx.doi.org/10.1021/ac3035508 | Anal. Chem. 2013, 85, 2478−2485