Kinase activity ranking using phosphoproteomics data (KARP) quantifies the contribution of protein kinases to the regulation of cell viability
2017; Elsevier BV; Volume: 16; Issue: 9 Linguagem: Inglês
10.1074/mcp.o116.064360
ISSN1535-9484
AutoresEdmund H. Wilkes, Pedro Casado, Vinothini Rajeeve, Pedro R. Cutillas,
Tópico(s)Bioinformatics and Genomic Networks
ResumoCell survival is regulated by a signaling network driven by the activity of protein kinases; however, determining the contribution that each kinase in the network makes to such regulation remains challenging. Here, we report a computational approach that uses mass spectrometry-based phosphoproteomics data to rank protein kinases based on their contribution to cell regulation. We found that the scores returned by this algorithm, which we have termed kinase activity ranking using phosphoproteomics data (KARP), were a quantitative measure of the contribution that individual kinases make to the signaling output. Application of KARP to the analysis of eight hematological cell lines revealed that cyclin-dependent kinase (CDK) 1/2, casein kinase (CK) 2, extracellular signal-related kinase (ERK), and p21-activated kinase (PAK) were the most frequently highly ranked kinases in these cell models. The patterns of kinase activation were cell-line specific yet showed a significant association with cell viability as a function of kinase inhibitor treatment. Thus, our study exemplifies KARP as an untargeted approach to empirically and systematically identify regulatory kinases within signaling networks. Cell survival is regulated by a signaling network driven by the activity of protein kinases; however, determining the contribution that each kinase in the network makes to such regulation remains challenging. Here, we report a computational approach that uses mass spectrometry-based phosphoproteomics data to rank protein kinases based on their contribution to cell regulation. We found that the scores returned by this algorithm, which we have termed kinase activity ranking using phosphoproteomics data (KARP), were a quantitative measure of the contribution that individual kinases make to the signaling output. Application of KARP to the analysis of eight hematological cell lines revealed that cyclin-dependent kinase (CDK) 1/2, casein kinase (CK) 2, extracellular signal-related kinase (ERK), and p21-activated kinase (PAK) were the most frequently highly ranked kinases in these cell models. The patterns of kinase activation were cell-line specific yet showed a significant association with cell viability as a function of kinase inhibitor treatment. Thus, our study exemplifies KARP as an untargeted approach to empirically and systematically identify regulatory kinases within signaling networks. Cells respond to their environment through a network of biochemical reactions driven by protein kinases (1Cohen P. The role of protein phosphorylation in neural and hormonal control of cellular activity.Nature. 1982; 296: 613-620Crossref PubMed Scopus (764) Google Scholar). Such networks regulate essential biological functions in normal cells and are often dysregulated in different diseases, including cancer (2Davies H. Bignell G.R. Cox C. Stephens P. Edkins S. Clegg S. Teague J. Woffendin H. Garnett M.J. Bottomley W. Davis N. Dicks E. Ewing R. Floyd Y. Gray K. Hall S. Hawes R. Hughes J. Kosmidou V. Menzies A. Mould C. Parker A. Stevens C. Watt S. Hooper S. Wilson R. Jayatilake H. Gusterson B.A. Cooper C. Shipley J. Hargrave D. Pritchard-Jones K. Maitland N. Chenevix-Trench G. Riggins G.J. Bigner D.D. Palmieri G. Cossu A. Flanagan A. Nicholson A. Ho J.W. Leung S.Y. Yuen S.T. Weber B.L. Seigler H.F. Darrow T.L. Paterson H. Marais R. Marshall C.J. Wooster R. Stratton M.R. Futreal P.A. Mutations of the BRAF gene in human cancer.Nature. 2002; 417: 949-954Crossref PubMed Scopus (8303) Google Scholar, 3Vivanco I. Sawyers C.L. The phosphatidylinositol 3-kinase AKT pathway in human cancer.Nature Rev. Cancer. 2002; 2: 489-501Crossref PubMed Scopus (5137) Google Scholar). The human genome codes for 518 protein kinase genes (4Manning G. Whyte D.B. Martinez R. Hunter T. Sudarsanam S. The protein kinase complement of the human genome.Science. 2002; 298: 1912-1934Crossref PubMed Scopus (6235) Google Scholar) and approximately half of these are expressed in a given cell type (5Wilhelm M. Schlegl J. Hahne H. Gholami A.M. Lieberenz M. Savitski M.M. Ziegler E. Butzmann L. Gessulat S. Marx H. Mathieson T. Lemeer S. Schnatbaum K. Reimer U. Wenschuh H. Mollenhauer M. Slotta-Huspenina J. Boese J.H. Bantscheff M. Gerstmair A. Faerber F. Kuster B. Mass-spectrometry-based draft of the human proteome.Nature. 2014; 509: 582-587Crossref PubMed Scopus (1318) Google Scholar, 6Kim M.S. Pinto S.M. Getnet D. Nirujogi R.S. Manda S.S. Chaerkady R. Madugundu A.K. Kelkar D.S. Isserlin R. Jain S. Thomas J.K. Muthusamy B. Leal-Rojas P. Kumar P. Sahasrabuddhe N.A. Balakrishnan L. Advani J. George B. Renuse S. Selvan L.D. Patil A.H. Nanjappa V. Radhakrishnan A. Prasad S. Subbannayya T. Raju R. Kumar M. Sreenivasamurthy S.K. Marimuthu A. Sathe G.J. Chavan S. Datta K.K. Subbannayya Y. Sahu A. Yelamanchi S.D. Jayaram S. Rajagopalan P. Sharma J. Murthy K.R. Syed N. Goel R. Khan A.A. Ahmad S. Dey G. Mudgal K. Chatterjee A. Huang T.C. Zhong J. Wu X. Shaw P.G. Freed D. Zahari M.S. Mukherjee K.K. Shankar S. Mahadevan A. Lam H. Mitchell C.J. Shankar S.K. Satishchandra P. Schroeder J.T. Sirdeshmukh R. Maitra A. Leach S.D. Drake C.G. Halushka M.K. Prasad T.S. Hruban R.H. Kerr C.L. Bader G.D. Iacobuzio-Donahue C.A. Gowda H. Pandey A. A draft map of the human proteome.Nature. 2014; 509: 575-581Crossref PubMed Scopus (1505) Google Scholar). Determining the contributions that different kinases make to cellular regulation is challenging, however, as each kinase's contribution is dependent on contextual information provided by the biochemical composition of the extracellular milieu (7Cowley S. Paterson H. Kemp P. Marshall C.J. Activation of MAP kinase is necessary and sufficient for PC12 differentiation and for transformation of NIH 3T3 cells.Cell. 1994; 77: 841-852Abstract Full Text PDF PubMed Scopus (1853) Google Scholar, 8Bakal C. Linding R. Llense F. Heffern E. Martin-Blanco E. Pawson T. Perrimon N. Phosphorylation networks regulating JNK activity in diverse genetic backgrounds.Science. 2008; 322: 453-456Crossref PubMed Scopus (104) Google Scholar, 9Creixell P. Palmeri A. Miller C.J. Lou H.J. Santini C.C. Nielsen M. Turk B.E. Linding R. Unmasking determinants of specificity in the human kinome.Cell. 2015; 163: 187-201Abstract Full Text Full Text PDF PubMed Scopus (52) Google Scholar) and changes as a result of development or disease. Given that each phosphorylation site is the result of protein kinase activity (opposed by a phosphatase), the phosphoproteome (the collection of all phosphorylation sites in cells) contains information regarding the net activity of all kinases expressed in the system being measured. Therefore, there is a strong community interest in the development of computational methods to extract kinase regulatory information from phosphoproteomics data (10Beltrao P. Bork P. Krogan N.J. van Noort V. Evolution and functional cross-talk of protein post-translational modifications.Mol. Syst. Biol. 2013; 9: 714Crossref PubMed Scopus (225) Google Scholar, 11Swaney D.L. Beltrao P. Starita L. Guo A. Rush J. Fields S. Krogan N.J. Villén J. Global analysis of phosphorylation and ubiquitylation cross-talk in protein degradation.Nat. Methods. 2013; 10: 676-682Crossref PubMed Scopus (401) Google Scholar, 12Tan C.S. Bodenmiller B. Pasculescu A. Jovanovic M. Hengartner M.O. Jørgensen C. Bader G.D. Aebersold R. Pawson T. Linding R. Comparative analysis reveals conserved protein phosphorylation networks implicated in multiple diseases.Science Signal. 2009; 2: ra39Crossref PubMed Scopus (160) Google Scholar). The phosphoproteome can now be efficiently explored by mass spectrometry (MS) (13Olsen J.V. Blagoev B. Gnad F. Macek B. Kumar C. Mortensen P. Mann M. Global, in vivo, and site-specific phosphorylation dynamics in signaling networks.Cell. 2006; 127: 635-648Abstract Full Text Full Text PDF PubMed Scopus (2822) Google Scholar, 14de Graaf E.L. Giansanti P. Altelaar A.F. Heck A.J. Single-step enrichment by Ti4+-IMAC and label-free quantitation enables in-depth monitoring of phosphorylation dynamics with high reproducibility and temporal resolution.Mol. Cell. Proteomics. 2014; 13: 2426-2434Abstract Full Text Full Text PDF PubMed Scopus (81) Google Scholar). Efforts have been made to compare relative kinase activation using MS-based phosphoproteomics data with methods such as kinase substrate enrichment analysis (KSEA), which allow the inference of differences in kinase activities between experimental conditions (15Casado P. Rodriguez-Prados J.C. Cosulich S.C. Guichard S. Vanhaesebroeck B. Joel S. Cutillas P.R. Kinase-substrate enrichment analysis provides insights into the heterogeneity of signaling pathway activation in leukemia cells.Science Signal. 2013; 6: rs6Crossref PubMed Scopus (201) Google Scholar, 16Mischnik M. Sacco F. Cox J. Schneider H.C. Schafer M. Hendlich M. Crowther D. Mann M. Klabunde T. IKAP: A heuristic framework for inference of kinase activities from phosphoproteomics data.Bioinformatics. 2015; 32: 424-431Crossref PubMed Scopus (42) Google Scholar, 17Drake J.M. Paull E.O. Graham N.A. Lee J.K. Smith B.A. Titz B. Stoyanova T. Faltermeier C.M. Uzunangelov V. Carlin D.E. Fleming D.T. Wong C.K. Newton Y. Sudha S. Vashisht A.A. Huang J. Wohlschlegel J.A. Graeber T.G. Witte O.N. Stuart J.M. Phosphoproteome integration reveals patient-specific networks in prostate cancer.Cell. 2016; 166: 1041-1054Abstract Full Text Full Text PDF PubMed Scopus (140) Google Scholar, 18Ochoa D. Jonikas M. Lawrence R.T. El Debs B. Selkrig J. Typas A. Villén J. Santos S.D. Beltrao P. An atlas of human kinase regulation.Mol. Syst. Biol. 2016; 12: 888Crossref PubMed Scopus (50) Google Scholar). Phosphoproteomics data obtained from drug perturbation experiments may also be used to infer signaling circuitry (19Saez-Rodriguez J. Alexopoulos L.G. Epperlein J. Samaga R. Lauffenburger D.A. Klamt S. Sorger P.K. Discrete logic modelling as a means to link protein signalling networks with functional analysis of mammalian signal transduction.Mol. Syst. Biol. 2009; 5: 331Crossref PubMed Scopus (271) Google Scholar, 20Terfve C.D. Wilkes E.H. Casado P. Cutillas P.R. Saez-Rodriguez J. Large-scale models of signal propagation in human cells derived from discovery phosphoproteomic data.Nat Commun. 2015; 6: 8033Crossref PubMed Scopus (56) Google Scholar, 21Wilkes E.H. Terfve C. Gribben J.G. Saez-Rodriguez J. Cutillas P.R. Empirical inference of circuitry and plasticity in a kinase signaling network.Proc. Natl. Acad. Sci. U.S.A. 2015; 112: 7719-7724Crossref PubMed Scopus (53) Google Scholar, 22Linding R. Jensen L.J. Ostheimer G.J. van Vugt M.A. T.M. Jørgensen C. Miron I.M. Diella F. Colwill K. Taylor L. Elder K. Metalnikov P. Nguyen V. Pasculescu A. Jin J. Park J.G. Samson L.D. Woodgett J.R. Russell Robert B. Bork P. Yaffe M.B. Pawson T. Systematic discovery of in vivo phosphorylation networks.Cell. 2007; 129: 1415-1426Abstract Full Text Full Text PDF PubMed Scopus (611) Google Scholar). These approaches, although useful for the comparison of signaling networks across conditions, do not provide information as to which kinases contribute more to signaling regulation relative to each other within an individual cell population. Here, we present the theoretical and experimental basis of a methodology, termed kinase activity ranking using phosphoproteomics data (KARP), which aims to infer the contributions made by individual kinases to signaling regulation in an untargeted manner. We first determined whether kinase activities could be inferred and ranked relative to one another from phosphoproteomics data obtained from a single cell population. We found that KARP accurately predicted the impact that targeting given kinases would have on cell viability, thus allowing an understanding, in quantitative terms, of the contribution of individual kinases to cell regulation. B-cell lymphoma and leukemia cell lines were routinely maintained in RPMI 1640 medium supplemented with 10% FBS and 100 U·ml−1 penicillin/streptomycin. Cells were maintained at a confluency of 0.5–2.0 × 106 cells·ml−1. Murine stromal cells were grown in IMDM 1The abbreviations used are: IMDM, Iscove's modified Dulbecco's medium; CDK, cyclin-dependent kinase; CK, casein kinase; DMSO, dimethyl sulfoxide; EGF, epidermal growth factor; ERK, extracellular signal-related kinase; FBS, fetal bovine serum; FWHM, full-width half-maximum; HEPES, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; HLB, hydrophilic-lipophilic balance; IGF-1, insulin-like growth factor 1; KARP, kinase activity ranking using phosphoproteomics data; KIS, kinase interacting with stathmin; LC, liquid chromatography; mTOR, mammalian target of rapamycin; MCF7, Michigan Cancer Foundation-7; MS-5, murine stromal-5; MS/MS, tandem mass spectrometry; P38D, p38 mitogen-activated kinase delta; PAK, p21-activated kinase; PKC, protein kinase C; RPMI, Roswell Park Memorial Institute; UPLC, ultra performance liquid chromatography; VBA, Visual Basic for Applications. medium (supplemented with 10% FBS and 100 U·ml−1 penicillin/streptomycin) and maintained at a confluency of 2.0–30.0 × 106 cells in 175 cm2 flasks. Murine stromal (MS)-5-cell-conditioned IMDM medium was generated by growing stromal cells in IMDM for 3 days. All cells were kept at 37 °C in a humidified atmosphere at 5% CO2. A total of 10 × 106 cells were seeded at 0.5 × 106 cells·ml−1 and left overnight. For each acute myeloid leukemia primary sample, 10 × 106 cells were seeded at 1 × 106 cells·ml−1 and left in the incubator for 2 h. Cells were subsequently harvested by centrifugation, washed twice with ice-cold phosphate-buffered saline—supplemented with 1 mm Na3VO4 and 1 mm NaF—and lysed in 0.2 ml of ice-cold urea lysis buffer (8 m urea in 20 mm HEPES (pH 8.0), supplemented with 1 mm Na3VO4, 1 mm NaF, 1 mm Na2H2P2O7, and 1 mm β-glycerol phosphate). Lysates were further homogenized by sonication, and any insoluble material removed by centrifugation. Protein concentration was estimated via the bicinchoninic acid assay. After normalizing each condition to a common protein concentration (0.5 μg·μl−1), each sample was reduced and alkylated by sequential incubation with 10 mm dithiothreitol and 16.6 mm iodoacetamide for 30 min at room temperature in the dark. For protein digestion, the urea concentration was reduced to 2 m through the addition of 20 mm HEPES (pH 8.0). Immobilized tosyl-lysine chloromethyl ketone-trypsin was then added, and samples incubated overnight at 37 °C. Trypsin beads were removed by centrifugation and the resultant peptide solutions were desalted using Oasis hydrophilic-lipophilic balance (HLB) 1cc solid-phase extraction cartridges as described previously (23Montoya A. Beltran L. Casado P. Rodríguez-Prados J.C. Cutillas P.R. Characterization of a TiO(2) enrichment method for label-free quantitative phosphoproteomics.Methods. 2011; 54: 370-378Crossref PubMed Scopus (94) Google Scholar). Phosphorylated peptides were enriched using TiO2 (GL Sciences) as previously described (21Wilkes E.H. Terfve C. Gribben J.G. Saez-Rodriguez J. Cutillas P.R. Empirical inference of circuitry and plasticity in a kinase signaling network.Proc. Natl. Acad. Sci. U.S.A. 2015; 112: 7719-7724Crossref PubMed Scopus (53) Google Scholar). The resulting phosphopeptide solutions were snap-frozen, dried with a SpeedVac, and stored at −80 °C until further use. For cell-line samples, each biological replicate was analyzed twice by LC-MS/MS as follows: Phosphopeptide pellets were resuspended in 14 μl of 0.1% TFA, and 4.0 μl per technical replicate was injected into a Waters NanoACQUITY UPLC system (Waters, Manchester, U.K.) coupled online to an LTQ-Orbitrap-XL mass spectrometer (Thermo Fisher Scientific). The samples were separated on a 100 min linear gradient between 5 and 35% acetonitrile (ACN) on an ACQUITY BEH130 C18 UPLC column (15 cm × 75 μm, 1.7 μm, 130 Å) at a flow rate of 300 nl·min−1. The top five most intense multiply charged ions in each MS1 scan were selected for collision-induced dissociation fragmentation (with multistage activation enabled). The resolution of the MS1 was set to 30,000 full-width half-maximum (FWHM). For each cell line, four independent biological replicates were performed, and these were analyzed in technical triplicate. Peptide identification was performed by matching deisotoped, MS/MS data to the Uniprot Swissprot human protein databases (September 2014 release, containing 20,233 entries), utilizing the Mascot server version 2.4. Mascot Distiller (version 2.5.1.0) was used to generate peak lists in the mascot generic format. Mass tolerances were set to 10 ppm and 600 mmu for the precursor and fragment ions, respectively. For the phosphoproteomics experiments, the allowed fixed modification was carbamidomethyl (Cys) and variable modifications were: phospho-Ser, phospho-Thr, phospho-Tyr, pyro-Glu (N-terminal), and oxidation-Met and were restricted the searches to tryptic peptides with one missed cleavage allowed. The identified phosphopeptides from each of the samples were collated and curated using in-house scripts. Unique phosphopeptides ions with expectancy < 0.05 were then included in the subsequent analyses. Mascot decoy database searches showed that with these settings produce a false discovery rate of ∼1%. Peptide quantification was performed as described before by our group (23Montoya A. Beltran L. Casado P. Rodríguez-Prados J.C. Cutillas P.R. Characterization of a TiO(2) enrichment method for label-free quantitative phosphoproteomics.Methods. 2011; 54: 370-378Crossref PubMed Scopus (94) Google Scholar, 24Casado P. Cutillas P.R. A self-validating quantitative mass spectrometry method for assessing the accuracy of high-content phosphoproteomic experiments.Mol. Cell. Proteomics. 2011; 10 (M110 003079)Abstract Full Text Full Text PDF PubMed Scopus (45) Google Scholar, 25Cutillas P.R. Vanhaesebroeck B. Quantitative profile of five murine core proteomes using label-free functional proteomics.Mol. Cell. Proteomics. 2007; 6: 1560-1573Abstract Full Text Full Text PDF PubMed Scopus (100) Google Scholar) and others (26Tsou C.C. Tsai C.F. Tsui Y.H. Sudhir P.R. Wang Y.T. Chen Y.J. Chen J.Y. Sung T.Y. Hsu W.L. IDEAL-Q, an automated tool for label-free quantitation analysis using an efficient peptide alignment approach and spectral data validation.Mol. Cell. Proteomics. 2010; 9: 131-144Abstract Full Text Full Text PDF PubMed Scopus (112) Google Scholar, 27Mann B. Madera M. Sheng Q. Tang H. Mechref Y. Novotny M.V. ProteinQuant Suite: A bundle of automated software tools for label-free quantitative proteomics.Rapid Comm. Mass Spectrom. 2008; 22: 3823-3834Crossref PubMed Scopus (50) Google Scholar). Briefly, Pescal software (written in Python v2.7) was then used to obtain peak areas of extracted ion chromatograms of each of the phosphopeptide ions in the database across all of the samples being compared. The retention times of each phosphopeptide ion, in each sample, were predicted through alignment of common phosphopeptides' retention times using an in-house linear modeling algorithm. Chromatographic peaks obtained from extracted ion chromatograms for each phosphopeptide in each sample were then integrated and the peak areas recorded. The mass-to-charge (m/z) and retention time (tR) tolerances were set to 7 ppm and 1.5 min, respectively. Processed data are shown in Supplemental Datasets 1 and 2. Phosphorylation sites were reported when phosphopeptides showed a delta score >10, otherwise the precise modification site was deemed ambiguous (28Savitski M.M. Lemeer S. Boesche M. Lang M. Mathieson T. Bantscheff M. Kuster B. Confident phosphorylation site localization using the Mascot Delta Score.Mol. Cell. Proteomics. 2011; 10 (M110 003830)Abstract Full Text Full Text PDF Scopus (222) Google Scholar). MS raw data have been deposited in PRIDE (project accession PXD005073, http://www.ebi.ac.uk/pride/archive/projects/PXD005073). Technical replicates were averaged, and peak areas for each phosphopeptide ion were then normalized to the sum of peptide intensities for each sample. Kinase–substrate matching was performed on these data as in reported before (15Casado P. Rodriguez-Prados J.C. Cosulich S.C. Guichard S. Vanhaesebroeck B. Joel S. Cutillas P.R. Kinase-substrate enrichment analysis provides insights into the heterogeneity of signaling pathway activation in leukemia cells.Science Signal. 2013; 6: rs6Crossref PubMed Scopus (201) Google Scholar) using a visual basic for applications (VBA) script against the PhosphoSitePlus database (downloaded in July 2014). The K-score was calculated for each kinase substrate group using the equation below (where m = the number of phosphorylation sites in the dataset matched to kinase K; α = the normalized intensity of the phosphorylation site i; n = the total number of phosphorylation sites in the dataset regardless of any kinase–substrate association; β = the normalized intensity of the phosphorylation site j; and t = the total number of known target phosphorylation sites in the PhosphoSitePlus database for kinase K). Data were visualized either using Microsoft Excel 2007/2010 or within the R statistical computing environment (v3.0.0) using a combination of the reshape2 and ggplot2 packages. Statistical significance of Pearson's correlation coefficients (r) was assessed through calculation of the corresponding t-statistics as below. An α = 0.05 was used, with P-values below this threshold considered statistically significant. FORMULA: t = r/√(1 − r^2/n − 2). K=∑n=1mαn∑j=1lβj·(mt)1/2·106 P31/Fuj cells were exposed to 1 mm sodium pervanadate or left untreated during 30 min (sodium pervanadate was prepared by mixing 30% H2O2 and 100 mm Na3VO4 pH 8.0 at 1:100 ratio during 15 min). Cells were then harvested and lysed as outlined above. After homogenization and protein quantification, treated and untreated cell lysates were mixed to a final protein concentration of 1.0 μg·μl−1. The proportions used were 0%, 25%, 50%, 75%, and 100% of pervanadate treated extracts with 100%, 75%, 50%, 25%, and 0% of untreated extracts. Protein mixtures were subsequently subjected to trypsin digestion and phosphopeptide enrichment as described above. K-score results were obtained from a meta-analysis of Supplementary Dataset 2 in Ref. (21Wilkes E.H. Terfve C. Gribben J.G. Saez-Rodriguez J. Cutillas P.R. Empirical inference of circuitry and plasticity in a kinase signaling network.Proc. Natl. Acad. Sci. U.S.A. 2015; 112: 7719-7724Crossref PubMed Scopus (53) Google Scholar). Briefly, MCF-7 cells were starved for 24 h and subsequently treated with 100 ng·ml−1 EGF or IGF-1 for 0, 5, 10, 30, or 60 min and processed for MS analysis as described in Ref. (21Wilkes E.H. Terfve C. Gribben J.G. Saez-Rodriguez J. Cutillas P.R. Empirical inference of circuitry and plasticity in a kinase signaling network.Proc. Natl. Acad. Sci. U.S.A. 2015; 112: 7719-7724Crossref PubMed Scopus (53) Google Scholar). K-scores were calculated as described above. Cell lines were seeded in 96-well plates (10,000 cells·well−1), left overnight and treated with vehicle (DMSO), or 1 to 1000 nm of AZD-5438 (CDK2i), GF-109203X (PKCαi; Tocris), PF-3758309 (PAKi; Calbiochem), Trametinib (MEKi; Selleckchem), MK-2206 (AKTi; Selleckchem), KU-0063794 (mammalian target of rapamycin (mTOR); Chemdea); or TAK 715 (P38αi). Cells were also treated with 0.01 to 10 μm of PKC-412 (PKC/Flt3i; Tocris) or 0.1 to 10 μm of TBB (CK2i). After 72 h, cells were stained with Guava ViaCount reagent (Millipore) as indicated by the manufacturer and cell number, and viability was measured using a Guava EasyCyte Plus instrument. All drugs were solubilized in DMSO, and all measurements were performed in triplicate. Flow cytometry data were analyzed using CytoSoft (v2.5.7). IC50 values were calculated using Graphpad PRISM (v5.03). The sensitivity coefficient (SC) was calculated using the equation below (where PCi = reduction in proliferation at Ci, IC50i = in vitro IC50 against primary target and Ci = inhibitor concentration at which proliferation is measured). Data were visualized using Microsoft Excel 2007/2010 or within the R statistical computing environment (v3.0.0), using a combination of the reshape2 and ggplot2 packages. SC=−log2(PCiCi·IC50i) We reasoned that the net activity of a kinase within a particular biological system would be proportional to its contribution to total phosphorylation, which can be calculated as the sum of the intensities of its known substrates relative to the sum of the intensities of all phosphorylation sites present in the dataset. We call the output of such analysis the K-score. To test the relevance of this concept, we analyzed the phosphoproteome of the P31/Fuji cell line (derived from acute myeloid leukemia) under basal conditions by label-free MS. We then computed the K-scores for each of the kinases represented in this dataset, as determined by known substrates defined in the PhosphoSitePlus database (29Hornbeck P.V. Zhang B. Murray B. Kornhauser J.M. Latham V. Skrzypek E. PhosphoSitePlus, 2014: Mutations, PTMs and recalibrations.Nucleic Acids. Res. 2015; 43: D512-D520Crossref PubMed Scopus (1680) Google Scholar) (Fig. 1A). This analysis revealed that CK2A1, CDK2, CDK1, ERK1, and CK1A were the five kinases that contributed most to cellular phosphorylation within this cell line. Kinases within the PhosphoSitePlus database are not equally represented, however, as the total number of substrates known for each kinase (t) was heterogeneous (Fig. 1B) perhaps reflecting how well-studied a given kinase is. In addition, the number of these substrates identified within the phosphoproteomics data (m) was also unequal (Figs. 1B–1D), probably because of a lack of expression of certain substrates in this cell line or due to the nature of the detectable phosphoproteome. We therefore decided to test the impact of normalization of the K-scores to account for this imbalance and thus undertook a systematic comparison of different normalization coefficients that incorporated the values of m and t. In order to test the performance of the different methods of normalization, we sought to compare the KARP outputs (K-scores) to an independent measure of cell regulation, as we reasoned that kinases with greater contribution to total phosphorylation may also contribute more to the signaling output. This hypothesis is supported by the known role of protein kinases in the regulation cell viability downstream of growth factor receptors, as the contribution of individual kinases in these networks to cell proliferation is thought to be proportional to their degree of activation (30Solit D.B. Garraway L.A. Pratilas C.A. Sawai A. Getz G. Basso A. Ye Q. Lobo J.M. She Y. Osman I. Golub T.R. Sebolt-Leopold J. Sellers W.R. Rosen N. BRAF mutation predicts sensitivity to MEK inhibition.Nature. 2006; 439: 358-362Crossref PubMed Scopus (1150) Google Scholar, 31Paez J.G. Jänne P.A. Lee J.C. Tracy S. Greulich H. Gabriel S. Herman P. Kaye F.J. Lindeman N. Boggon T.J. Naoki K. Sasaki H. Fujii Y. Eck M.J. Sellers W.R. Johnson B.E. Meyerson M. EGFR mutations in lung cancer: Correlation with clinical response to gefitinib therapy.Science. 2004; 304: 1497-1500Crossref PubMed Scopus (8482) Google Scholar, 32Heinrich M.C. Corless C.L. Duensing A. McGreevey L. Chen C.J. Joseph N. Singer S. Griffith D.J. Haley A. Town A. Demetri G.D. Fletcher C.D. Fletcher J.A. PDGFRA activating mutations in gastrointestinal stromal tumors.Science. 2003; 299: 708-710Crossref PubMed Scopus (2000) Google Scholar, 33Pao W. Girard N. New driver mutations in non-small-cell lung cancer.Lancet Oncol. 2011; 12: 175-180Abstract Full Text Full Text PDF PubMed Scopus (955) Google Scholar). Therefore, a relevant signaling output to consider in our cell models was the regulation of cell viability as a function of kinase activity ranking. As such, we treated P31/Fuji cells with a number of different small-molecule kinase inhibitors and measured the reduction in cell viability as a function of treatment. The effect of small molecule inhibitors on cell behavior is dependent on both the contribution of the target to the pathway flux as well as the affinity of the compound to the target (which is reflected by the in vitro IC50). To account for this, we normalized the drug-induced inhibition of cell viability by the reported in vitro IC50 of the compound against its known targets. We termed this value the sensitivity coefficient (SC) (Figs. 2A and 2B). By doing this, differences in inhibitor potencies were normalized, and kinase inhibitors with disparate in vitro IC50 values could be ranked against one another on the basis of the contribution of their targets to cell viability. As shown in Fig. 2C, K-scores were not associated with the IC50 of inhibitor effects on cell viability. We found, however, a statistically significant association between K-scores and SCs for given kinases. Normalization by the square root of the ratio between m and t for the specific kinase showed the strongest linear relationship between the K-scores and SC values for each individual kinase (r = 0.927, p < 0.01). Based on this observation, the final method for the calculation of K-scores was taken as K = Σα/Σβ · (m/t)1/2. The association between K-scores and reduction in cell viability as a function of kinase inhibitor treatment suggests that KARP provided a quantitative measure of the contribution that individual kinases made to the regulation of this cell population's viability. To investigate the impact that kinase popularity (i.e. the number of substrates per kinase present in the database) could potentially have on K-score values, we plotted the normalization factor in the final KARP formula (i.e. [m/t]1/2) as a function of t (number of substrates present in the PhosphoSite database) for each kinase identified (Fig. 3A). We found that the value of the normaliz
Referência(s)