Proteomics 2014, 14, 1845–1856
1845
DOI 10.1002/pmic.201400008
RESEARCH ARTICLE
Extracellular vesicles shed from gefitinib-resistant nonsmall cell lung cancer regulate the tumor microenvironment Do-Young Choi1∗ , Sungyong You2∗ , Jae Hun Jung1 , Jae Cheol Lee3 , Jin Kyung Rho4 , Kye Young Lee5∗∗ , Michael R. Freeman2,6,7 , Kwang Pyo Kim1∗∗ and Jayoung Kim2,6,7 1
Department of Applied Chemistry, College of Applied Science, Kyung Hee University, Yongin, Republic of Korea Departments of Surgery and Biomedical Sciences, Division of Cancer Biology and Therapeutics, Samuel Oschin Comprehensive Cancer Institute, Cedars−Sinai Medical Center, Los Angeles, CA, USA 3 Department of Oncology, Asan Medical Center, College of Medicine, University of Ulsan, Seoul, Korea 4 Asan Institute for Life Sciences, Asan Medical Center, College of Medicine, University of Ulsan, Seoul, Korea 5 Department of Internal Medicine, Konkuk University School of Medicine, Seoul, Korea 6 Boston Children’s Hospital, Harvard Medical School, Boston, MA, USA 7 Departments of Surgery and Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA, USA 2
Epidermal growth factor receptor (EGFR)-tyrosine kinase inhibitors (TKIs), including gefitinib, are the first-line treatment of choice for nonsmall cell lung cancer patients who harbor activating EGFR mutations, however, acquired resistance to EGFR-TKIs is inevitable. The main objective of this study was to identify informative protein signatures of extracellular vesicles (EV) derived from gefitinib-resistant nonsmall cell lung cancer cells using proteomics analysis. Nano-LC– MS/MS analysis identified with high confidence (false discovery rate < 0.05, fold change 2) 664 EV proteins enriched in PC9R cells, which are resistant to gefitinib due to EGFR T790M mutation. Computational analyses suggested components of several signal transduction mechanisms including the AKT (also PKB, protein kinase B)/mTOR (mechanistic target of rapamycin) pathway are overrepresented in EV from PC9R cells. Treatment of recipient cells with EV harvested from PC9R cells increased phosphorylation of signaling molecules, and enhanced proliferation, invasion, and drug resistance to gefitinib-induced apoptosis. Dose- and time-dependent pharmaceutical inhibition of AKT/mTOR pathway overcame drug resistance of PC9R cells and those of H1975 exhibiting EGFR T790M mutation. Our findings provide new insight into an oncogenic EV protein signature regulating tumor microenvironment, and will aid in the development of novel diagnostic strategies for prediction and assessment of gefitinib resistance.
Received: January 9, 2014 Revised: May 23, 2014 Accepted: June 11, 2014
Keywords: AKT / Cell biology / Extracellular vesicles / mTOR signaling pathway / Nonsmall cell lung cancer
Additional supporting information may be found in the online version of this article at the publisher’s web-site
Correspondence: Professor Jayoung Kim, Division of Cancer Biology and Therapeutics, Departments of Surgery and Biomedical Sciences, Samuel Oschin Comprehensive Cancer Institute, Cedars-Sinai Medical Center, 8700 Beverly Blvd., Davis Research Building 5071, Los Angeles, CA 90048, USA E-mail: [email protected] Abbreviations: ABC, ammonium bicarbonate; DEPs, differentially enriched EV proteins; EGFR, epidermal growth factor receptor;
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
EV, extracellular vesicles; KEA, kinase enrichment analysis; PE, pleural effusion; TKIs, tyrosine kinase inhibitors; NSCLC, nonsmall cell lung cancer ∗ These authors contributed equally to this work. ∗∗ Additional corresponding authors: Professor
Kye Young Lee, E-mail: [email protected]; Dr. Kwang Pyo Kim, E-mail: kimkp@ khu.ac.kr Colour Online: See the article online to view Figs. 1–6 in colour.
www.proteomics-journal.com
1846
1
D.-Y. Choi et al.
Introduction
The secretion of extracellular vesicles (EV) of several types from tumor cells has been identified as an important means of conditioning and alteration of the tumor microenvironment by malignant cells [1, 2]. Although EV may vary in size, biological function, and components, their clinical importance in cancer progression and the potential use for specific EV molecular species to serve as novel cancer biomarkers have become increasingly evident [3]. EV may alter the local tumor microenvironment via transfer of genetic and biochemical information, including coding and noncoding RNAs, miRNAs, lipids, metabolites, DNA fragments, and pathogenic or oncogenic proteins [4–12]. They may also play roles in establishing a favorable premetastatic niche and in epigenetic reprograming of collaborating cells involved in cancer progression [11, 13, 14]. EV have been found in biological fluids (e.g. urine, serum, pleural effusion (PE), saliva) [4, 6, 15, 16] as oncogenic cargo that are functionally relevant to mechanisms of tumor progression. These discoveries suggest that detection and analysis of these particles may be employed for noninvasive methods of monitoring patient response to therapy as well as for identification of novel therapeutic targets in attempts to overcome treatment resistance [17–19]. Lung cancer is the most common cause of death from cancer worldwide, and is diagnosed in approximately 174 000 Americans each year. Nonsmall cell lung cancer (NSCLC) represents about 85–90% of lung cancer diagnoses, and overexpression of the epidermal growth factor receptor (EGFR) has been detected in 80% of NSCLC patients [20]. However, clinical applications using EGFR-tyrosine kinase inhibitors (TKIs; e.g. gefitinib and erlotinib) are very active in EGFR mutations-positive NSCLC patients [21]. Development of acquired resistance is inevitable and unavoidable. Secondary EGFR mutations (e.g. T790M) is the most important mechanism which is detected in more than 50% of cases [21, 22]. Studies to identify novel prognostic biomarkers predicting sensitivity or resistance to EGFR inhibitors and to overcome such resistance are currently an active focus of research by a number of laboratories [23–25]. Discrimination of NSCLC cancer patients at higher risk (e.g. drug resist) could reduce mortality rates. Thus, it is urgent to develop new noninvasive prognostic methods such as molecular biomarkers present in biofluids for assessment of response to lung cancer chemotherapies [26–28]. In this study, we present the proteome profile of EV shed from gefitinib-resistant NSCLC cells compared to nonresistant isogenic cells. These findings provide important insights into the molecular components of EV from cells resistant to EGFR inhibition, and suggest possible clinical application within the subgroup of NSCLC patients who acquire resistance to gefitinib-based chemotherapy.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Proteomics 2014, 14, 1845–1856
2
Materials and methods
2.1 Reagents Antibodies against CD9, CD81, and CD63 are from Santa Cruz Biotechnology. Antibodies against AKT1, phosphorAKT1, GSK3, EIF4E, mTOR, phospho-mTOR, RELA, RAS, cleaved PARP, EGFR, and phosphor-EGFR were from Cell Signaling Technology. Gefitinib and BEZ235 were purchased from Active Biochem. Western blot apparatus and ECL detection kit were from BioRad and New England Nuclear.
2.2 Cell culture The isogenic and gefitinib-resistant subline, PC9R, was established as a gefitinib-resistant clone from parent PC9 cells by chronic incubation and selection of PC9 cells in medium containing gefitinib in a stepwise manner (from 0.01 to 2 M) for several months. For cell maintenance, PC9, PC9R, or H1975 NSCLC cells were cultured under a humidified atmosphere of 5% CO2 at 37⬚C in RPMI-1640 medium (Sigma, St. Louis, MO, USA) supplemented with 10% FBS, 100 g streptomycin, and 100 units/mL penicillin (Invitrogen, Carlsbad, CA, USA). The cells were then harvested by trypsinization and subcultured at a density of 6 × 103 cells/cm2 . The media were changed after 1 day of subculture during culture, and cell cultures were passaged again at 70−80% confluence.
2.3 Isolation of EV from conditioned medium To isolate EV, culture media from PC9 or PC9R cells were harvested and cell debris were removed by centrifugation at 500 × g for 15 min, followed by another centrifugation at 800 × g for 15 min. The supernatants were concentrated using Stirred Ultrafiltration Cells 8400 for molecular weightbased cutoff, and then was applied to 0.8 and 2.7 M sucrose cushions in 20 mM HEPES/150 mM NaCl buffer (pH 7.2). After ultracentrifugation at 100 000 × g for 2 h at 4⬚C, additional ultracentrifugation on a sucrose cushion and OptiPrep (50% iodixanol, Axis-Shield PoC AS, Oslos, Norway) density gradient was performed [12]. The protein concentrations of the EV fractions were measured by Bicinchoninic acid assay.
2.4 1D SDS-PAGE and enzymatic in-gel digestion Isolated EV were run on SDS-PAGE gel (4–12% gradient R Bis-Tris gel, Invitrogen), followed by staining with Novex GelCode Blue Stain Reagent (Pierce, Rockford, IL, USA). SDS-PAGE gel was sliced into ten pieces for in-gel digestion, as previously described [16]. Briefly, the excised gels were
www.proteomics-journal.com
1847
Proteomics 2014, 14, 1845–1856
destained with 50% ACN in 50 mM NH4 CO3 (ammonium bicarbonate, ABC). The destained gel pieces were dehydrated in 100% ACN and reduction steps were performed by DTT to give final concentration of 5 mM in 50mM ABC. After incubating the samples for 40 min at 37⬚C, reduced cysteine residues were alkylated by 50 mM idoacetamide (IAA) in darkness for 30 min at 25 ⬚C. The alkylated gel pieces were dehydrated in 100 ACN and digested with sequencing grade trypsin (Promega, Madison, WI, USA) in 50 mM ABC for 12 h at 37⬚C. Digested peptides were extracted from gel with 5% formic acid in 50% ACN. The extracts were dried and purified using C-18 Spin Columns (Thermo Scientific, Rockford, IL, USA) for MS analysis.
2.5 Nano-LC-ESI-MS/MS analysis and database search The tryptic peptides produced by the in-gel tryptic digestion were analyzed by Q Exactive mass spectrometry (Thermo Fisher Scientific, Bremen, Germany) coupled with an EASYnLC 1000 (Thermo Fisher Scientific). Briefly, the tryptic peptides were loaded onto trap column (100 m × 1.5 cm) packed with C18 -AQ 5-m-sized resin which loaded peptides were eluted with a linear gradient from 3 to 60% solvent B (0.1% formic acid in ACN) for 70 min at a flow rate 300 nL/min. The eluted peptides, separated by the analytical column (75 m × 12 cm), were sprayed into nano-ESI source with an electrospray voltage of 2.0 kV. The Q Exactive mass spectrometer was operated in top 8 data-dependent method, which dynamically choosing the most abundant precursor ions form survey scan (m/z; 300–2000) for higher-energy collisional dissociation (HCD) fragmentation. The normalized collision energy used was 27%. Dynamic exclusion duration was 30 s and isolation window of precursor was performed with 2. Survey scans were acquired at a resolution of 10 000 at m/z 200. The resolution for HCD spectra was set to 17 500 at m/z 200. MS/MS spectra were searched against IPI human protein (version 3.87, containing 91 491 entries), the sequencereversed decoy databases using SEQUEST of BioworksBrowser rev. 3.1. Full tryptic specificity was required with up to two missed cleavage sites allowed; mass tolerances for precursor ions and fragment ions were set to 10 ppm and 0.8 Da, respectively; fixed modification for carbamidomethylcysteine and variable modifications for methionine oxidation, N-terminal acetylation, and N/Q deamination were used. The SEQUEST search results were further evaluated using the PeptideProphet [29] and ProteinProphet [30] program in Trans Proteomic Pipeline v4.5.33, 34. We selected peptides with cutoff of FDR = 0.01. We then identified the proteins that have ProteinProphet probability 0.95. Among these proteins, we selected the high confidence proteins that were detected in more than two replicates of the three profiles in EV PC9 or PC9R samples and have more than two sibling peptides. C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
The relative quantitation of EV proteins was performed using the absolute protein expression (APEX) Quantitative Proteomics Tool [31]. We first generated a training dataset as 50 most abundant proteins with protein probabilities 0.95 and normalization factor = 1, then computed the predicted peptide count (Oi) of the EV proteins from the training set, and finally calculated APEX abundances of proteins using the protXML file and the urine protein Oi values [31]. The relative APEX abundances in EV from PC9 and PC9R samples were normalized using quantile normalization method [32]. Before identifying differentially enriched EV proteins (DEPs), quantified profiles were examined their distribution and degree of variation between samples for quality assessment (see Supporting Information Fig. 1) using SuperHirn [33]. Using the normalized APEX abundances, we identified DEPs using an integrative statistical method reported previously [34]. Briefly, for each protein, we first applied a t-test and a log2-median-ratio test to its abundances in three replicates. By permuting the replicates, we estimated empirical null distributions and for each protein we computed adjusted p values for the two tests. We then combined the p values from the two tests to compute the overall p value using Stouffer’s method [34]. Finally, we computed FDRs for the overall p values using Storey’s method [35]. We selected DEPs that have the combined FDR < 0.05 and fold change 2.
2.6 Pathway analysis To explore cellular processes, pathways, and subcellular localization represented by DEPs, we identified GO biological process, Kyoto Encyclopedia of Genes and Genomes KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway, and cellular component with p values computed from DAVID software less than 0.05 [36]. The kinase enrichment analysis (KEA) was performed by using Enrichr software (http://amp. pharm.mssm.edu/Enrichr/index.html) to compute kinase enrichment probability based on the distribution of kinase– substrate proportions in the background kinase–substrate database compared with kinases found to be associated with an input list of proteins [37]. To select the top ten kinases, the enrichment score was calculated by combining the p value computed using the Fisher exact test with the z-score of the deviation from the expected rank by multiplying these two numbers.
2.7 Network modeling To reconstruct a subnetwork describing enriched cellular processes in EV from PC9R, we selected 160 proteins that are involved in ten enriched cellular processes from the 664 PC9Rdriven EV-dominant proteins. We then built a Resistancedriven EV protein network model using the interaction information of the 160 proteins obtained from public databases www.proteomics-journal.com
1848
D.-Y. Choi et al.
including HPRD [38], BioGRID [39], STRING [40], and KEGG [41]. The nodes in the network were arranged such that the nodes with the same GOBPs [42] and KEGG pathways were grouped into the same network modules, resulting in the eight modules. 2.8 Cell proliferation assay PC9 cells (recipient) were seeded onto six-well culture plates at a density of 1 × 103 cells in growth medium. When cells grew to 80% confluence, they were washed with PBS twice and serum starved for additional 24 h. Same amount (50 g/mL) of isolated EV from PC9 (PC9 EV) or PC9R (PC9R EV; donor) were added to serum starved PC9 cells (recipient). Cells were incubated for additional 3 days with daily treatment with PC9 EV or PC9R EV. Cell proliferation was quantified by crystal violet staining. Briefly, cells were stained with crystal violet solution and quantified by dissolving stained cells in 10% acetic acid solution. Colorimetric reading was done by absorbance measurement at 570 nm.
Proteomics 2014, 14, 1845–1856
2.10 Apoptosis (or drug sensitivity) assay To assess the effect of PC9 EV or PC9R EV on gefitinibinduced apoptosis, PC9 cells were pretreated with PC9 EV or PC9R EV for 1 h, and followed by treatment of gefinitib (10 M) for additional 16 h. To determine whether dual inhibition of AKT and mTOR re-sensitizes PC9R cells, BEZ235 (100 nM) was treated in presence of gefitinib (0–10 M) for 3 days. Cell mass was quantified by crystal violet staining.
2.11 Statistical analysis The mean of 3 or more replicates was used as average. Data are presented as mean ± SD. The p values were calculated using a standard unpaired Student’s t-test for simple comparisons. Statistical significance is displayed as p < 0.05 (*) or p < 0.005 (**).
3
Results
2.9 Invasion assay
3.1 EV secreted from getinitib-resistant PC9R cells
Thick matrigel-coated inserts were prepared 1 day before experiment. PC9 cells (recipient) were serum starved for 4 h, and each cell suspension (1 × 105 cells/mL, 300 L total) was seeded on the upper surface of each insert. Exogenous PC9 EV or PC9R EV was added into culture medium, and recipient cells were allowed to degrade gelatin coating for 16 h at 37⬚C, 5% CO2 . Noninvasive cells were removed, and bottom side of the inserts was stained. After washing twice to remove background staining, 300 L of extraction solution [43] containing 10% acetic acid was used for extraction. Absorbance was read at 560 nm using a FLUOstar plate reader for quantification. At least triplicates were performed to obtain statistical signification.
In order to understand the biological features of NSCLC cells with acquired resistance to gefitinib, we used PC9R cells, which were developed by prolonged exposure to gefitinib in vitro [44, 45]. PC9R cells have been used in published studies as a model of gefitinib resistance due to acquired EGFR mutation (T790M) [46,47]. While parent PC9 cells were sensitive to gefitinib treatment (0–10 M), resulting in a significant decrease of cell number due to induced apoptosis, PC9R cells were not responsive to gefitinib even at the highest concentration, 10 M (Fig. 1A). To isolate the secreted EV from PC9 or PC9R, cells were first incubated in serum-free medium for 24 h. EV were isolated from the culture medium by serial centrifugations (to remove cell debris), sucrose cushion
Figure 1. Characterization of gefitinib-resistant PC9R, compared to PC9 control cells. (A) The two cell lines (2 × 102 per well) were seeded on 96-well plates in triplicates. Next day, cells were incubated in growth medium in the absence or presence of the varying doses of gefitinib (1, 0.01, 0.1, 1, or 10 M). Cell number was measured by crystal violet assay after 24-h incubation. (B) Representative electron microscopic images of EV shed from PC9 and PC9R cells. Size of bars, 200 nm. (C) Western blot analysis for EV markers (CD81, CD9, and CD63) validating components of isolated EV sample (N 3 independent trials).
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.proteomics-journal.com
1849
Proteomics 2014, 14, 1845–1856
Figure 2. A systems approach for the identification of EV proteins for gefitinib resistance. (A) Scheme of the proteomics approach to EV for gefitinib resistance. (B) KEGG pathways in which the identified proteins are involved. The number beside the term indicates the number of proteins that belong to the KEGG pathway annotation. (C) The subcellular localization of the identified proteins according to GOCCs (Gene Ontology Cellular Components).
ultracentrifugation, and OptiPrep density gradient ultracentrifugation, as described in Section 2. Isolated EV were checked by electron microscopy for size and intact structure (Fig. 1B), and characteristic EV markers (CD81, CD9, and CD63) [16] were verified by Western blot analysis (Fig. 1C).
3.2 MS proteomics identifies EV proteins characteristic of PC9R Stable alterations in specific pathways or networks might contribute to acquired resistance of NSCLC to gefitinib. In order to test this hypothesis and to understand the characteristic molecular features of drug-resistant PC9R cells, we attempted to identify protein components of EV associated with gefitinib resistance by proteomic analysis on EV isolated from PC9R and PC9 cells. An overview of the procedures used for quantitative EV-driven proteomic comparisons on EV of PC9R (PC9R EV) versus those of PC9 cells (PC9 EV) is illustrated in Fig. 2A. PC9R EV and PC9 EV were prepared in triplicates as described in Section 2. EV proteins were first visualized by 1D SDS-PAGE, followed by nanoLC-MS/MS analyses. LC-MS/MS was performed in triplicate generating a total of 30 proteome profiles for PC9R EV (n = 3) or PC9 EV (n = 3) samples. For protein identification, the resulting MS/MS spectra were searched against IPI human version 3.87 and a sequence-reversed decoy database using C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
SEQUEST (see Section 2). A total of 3121 EV proteins (2339 and 2584 proteins from PC9 EV and PC9R EV, respectively) were identified with an FDR for peptide identification
1845
DOI 10.1002/pmic.201400008
RESEARCH ARTICLE
Extracellular vesicles shed from gefitinib-resistant nonsmall cell lung cancer regulate the tumor microenvironment Do-Young Choi1∗ , Sungyong You2∗ , Jae Hun Jung1 , Jae Cheol Lee3 , Jin Kyung Rho4 , Kye Young Lee5∗∗ , Michael R. Freeman2,6,7 , Kwang Pyo Kim1∗∗ and Jayoung Kim2,6,7 1
Department of Applied Chemistry, College of Applied Science, Kyung Hee University, Yongin, Republic of Korea Departments of Surgery and Biomedical Sciences, Division of Cancer Biology and Therapeutics, Samuel Oschin Comprehensive Cancer Institute, Cedars−Sinai Medical Center, Los Angeles, CA, USA 3 Department of Oncology, Asan Medical Center, College of Medicine, University of Ulsan, Seoul, Korea 4 Asan Institute for Life Sciences, Asan Medical Center, College of Medicine, University of Ulsan, Seoul, Korea 5 Department of Internal Medicine, Konkuk University School of Medicine, Seoul, Korea 6 Boston Children’s Hospital, Harvard Medical School, Boston, MA, USA 7 Departments of Surgery and Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA, USA 2
Epidermal growth factor receptor (EGFR)-tyrosine kinase inhibitors (TKIs), including gefitinib, are the first-line treatment of choice for nonsmall cell lung cancer patients who harbor activating EGFR mutations, however, acquired resistance to EGFR-TKIs is inevitable. The main objective of this study was to identify informative protein signatures of extracellular vesicles (EV) derived from gefitinib-resistant nonsmall cell lung cancer cells using proteomics analysis. Nano-LC– MS/MS analysis identified with high confidence (false discovery rate < 0.05, fold change 2) 664 EV proteins enriched in PC9R cells, which are resistant to gefitinib due to EGFR T790M mutation. Computational analyses suggested components of several signal transduction mechanisms including the AKT (also PKB, protein kinase B)/mTOR (mechanistic target of rapamycin) pathway are overrepresented in EV from PC9R cells. Treatment of recipient cells with EV harvested from PC9R cells increased phosphorylation of signaling molecules, and enhanced proliferation, invasion, and drug resistance to gefitinib-induced apoptosis. Dose- and time-dependent pharmaceutical inhibition of AKT/mTOR pathway overcame drug resistance of PC9R cells and those of H1975 exhibiting EGFR T790M mutation. Our findings provide new insight into an oncogenic EV protein signature regulating tumor microenvironment, and will aid in the development of novel diagnostic strategies for prediction and assessment of gefitinib resistance.
Received: January 9, 2014 Revised: May 23, 2014 Accepted: June 11, 2014
Keywords: AKT / Cell biology / Extracellular vesicles / mTOR signaling pathway / Nonsmall cell lung cancer
Additional supporting information may be found in the online version of this article at the publisher’s web-site
Correspondence: Professor Jayoung Kim, Division of Cancer Biology and Therapeutics, Departments of Surgery and Biomedical Sciences, Samuel Oschin Comprehensive Cancer Institute, Cedars-Sinai Medical Center, 8700 Beverly Blvd., Davis Research Building 5071, Los Angeles, CA 90048, USA E-mail: [email protected] Abbreviations: ABC, ammonium bicarbonate; DEPs, differentially enriched EV proteins; EGFR, epidermal growth factor receptor;
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
EV, extracellular vesicles; KEA, kinase enrichment analysis; PE, pleural effusion; TKIs, tyrosine kinase inhibitors; NSCLC, nonsmall cell lung cancer ∗ These authors contributed equally to this work. ∗∗ Additional corresponding authors: Professor
Kye Young Lee, E-mail: [email protected]; Dr. Kwang Pyo Kim, E-mail: kimkp@ khu.ac.kr Colour Online: See the article online to view Figs. 1–6 in colour.
www.proteomics-journal.com
1846
1
D.-Y. Choi et al.
Introduction
The secretion of extracellular vesicles (EV) of several types from tumor cells has been identified as an important means of conditioning and alteration of the tumor microenvironment by malignant cells [1, 2]. Although EV may vary in size, biological function, and components, their clinical importance in cancer progression and the potential use for specific EV molecular species to serve as novel cancer biomarkers have become increasingly evident [3]. EV may alter the local tumor microenvironment via transfer of genetic and biochemical information, including coding and noncoding RNAs, miRNAs, lipids, metabolites, DNA fragments, and pathogenic or oncogenic proteins [4–12]. They may also play roles in establishing a favorable premetastatic niche and in epigenetic reprograming of collaborating cells involved in cancer progression [11, 13, 14]. EV have been found in biological fluids (e.g. urine, serum, pleural effusion (PE), saliva) [4, 6, 15, 16] as oncogenic cargo that are functionally relevant to mechanisms of tumor progression. These discoveries suggest that detection and analysis of these particles may be employed for noninvasive methods of monitoring patient response to therapy as well as for identification of novel therapeutic targets in attempts to overcome treatment resistance [17–19]. Lung cancer is the most common cause of death from cancer worldwide, and is diagnosed in approximately 174 000 Americans each year. Nonsmall cell lung cancer (NSCLC) represents about 85–90% of lung cancer diagnoses, and overexpression of the epidermal growth factor receptor (EGFR) has been detected in 80% of NSCLC patients [20]. However, clinical applications using EGFR-tyrosine kinase inhibitors (TKIs; e.g. gefitinib and erlotinib) are very active in EGFR mutations-positive NSCLC patients [21]. Development of acquired resistance is inevitable and unavoidable. Secondary EGFR mutations (e.g. T790M) is the most important mechanism which is detected in more than 50% of cases [21, 22]. Studies to identify novel prognostic biomarkers predicting sensitivity or resistance to EGFR inhibitors and to overcome such resistance are currently an active focus of research by a number of laboratories [23–25]. Discrimination of NSCLC cancer patients at higher risk (e.g. drug resist) could reduce mortality rates. Thus, it is urgent to develop new noninvasive prognostic methods such as molecular biomarkers present in biofluids for assessment of response to lung cancer chemotherapies [26–28]. In this study, we present the proteome profile of EV shed from gefitinib-resistant NSCLC cells compared to nonresistant isogenic cells. These findings provide important insights into the molecular components of EV from cells resistant to EGFR inhibition, and suggest possible clinical application within the subgroup of NSCLC patients who acquire resistance to gefitinib-based chemotherapy.
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Proteomics 2014, 14, 1845–1856
2
Materials and methods
2.1 Reagents Antibodies against CD9, CD81, and CD63 are from Santa Cruz Biotechnology. Antibodies against AKT1, phosphorAKT1, GSK3, EIF4E, mTOR, phospho-mTOR, RELA, RAS, cleaved PARP, EGFR, and phosphor-EGFR were from Cell Signaling Technology. Gefitinib and BEZ235 were purchased from Active Biochem. Western blot apparatus and ECL detection kit were from BioRad and New England Nuclear.
2.2 Cell culture The isogenic and gefitinib-resistant subline, PC9R, was established as a gefitinib-resistant clone from parent PC9 cells by chronic incubation and selection of PC9 cells in medium containing gefitinib in a stepwise manner (from 0.01 to 2 M) for several months. For cell maintenance, PC9, PC9R, or H1975 NSCLC cells were cultured under a humidified atmosphere of 5% CO2 at 37⬚C in RPMI-1640 medium (Sigma, St. Louis, MO, USA) supplemented with 10% FBS, 100 g streptomycin, and 100 units/mL penicillin (Invitrogen, Carlsbad, CA, USA). The cells were then harvested by trypsinization and subcultured at a density of 6 × 103 cells/cm2 . The media were changed after 1 day of subculture during culture, and cell cultures were passaged again at 70−80% confluence.
2.3 Isolation of EV from conditioned medium To isolate EV, culture media from PC9 or PC9R cells were harvested and cell debris were removed by centrifugation at 500 × g for 15 min, followed by another centrifugation at 800 × g for 15 min. The supernatants were concentrated using Stirred Ultrafiltration Cells 8400 for molecular weightbased cutoff, and then was applied to 0.8 and 2.7 M sucrose cushions in 20 mM HEPES/150 mM NaCl buffer (pH 7.2). After ultracentrifugation at 100 000 × g for 2 h at 4⬚C, additional ultracentrifugation on a sucrose cushion and OptiPrep (50% iodixanol, Axis-Shield PoC AS, Oslos, Norway) density gradient was performed [12]. The protein concentrations of the EV fractions were measured by Bicinchoninic acid assay.
2.4 1D SDS-PAGE and enzymatic in-gel digestion Isolated EV were run on SDS-PAGE gel (4–12% gradient R Bis-Tris gel, Invitrogen), followed by staining with Novex GelCode Blue Stain Reagent (Pierce, Rockford, IL, USA). SDS-PAGE gel was sliced into ten pieces for in-gel digestion, as previously described [16]. Briefly, the excised gels were
www.proteomics-journal.com
1847
Proteomics 2014, 14, 1845–1856
destained with 50% ACN in 50 mM NH4 CO3 (ammonium bicarbonate, ABC). The destained gel pieces were dehydrated in 100% ACN and reduction steps were performed by DTT to give final concentration of 5 mM in 50mM ABC. After incubating the samples for 40 min at 37⬚C, reduced cysteine residues were alkylated by 50 mM idoacetamide (IAA) in darkness for 30 min at 25 ⬚C. The alkylated gel pieces were dehydrated in 100 ACN and digested with sequencing grade trypsin (Promega, Madison, WI, USA) in 50 mM ABC for 12 h at 37⬚C. Digested peptides were extracted from gel with 5% formic acid in 50% ACN. The extracts were dried and purified using C-18 Spin Columns (Thermo Scientific, Rockford, IL, USA) for MS analysis.
2.5 Nano-LC-ESI-MS/MS analysis and database search The tryptic peptides produced by the in-gel tryptic digestion were analyzed by Q Exactive mass spectrometry (Thermo Fisher Scientific, Bremen, Germany) coupled with an EASYnLC 1000 (Thermo Fisher Scientific). Briefly, the tryptic peptides were loaded onto trap column (100 m × 1.5 cm) packed with C18 -AQ 5-m-sized resin which loaded peptides were eluted with a linear gradient from 3 to 60% solvent B (0.1% formic acid in ACN) for 70 min at a flow rate 300 nL/min. The eluted peptides, separated by the analytical column (75 m × 12 cm), were sprayed into nano-ESI source with an electrospray voltage of 2.0 kV. The Q Exactive mass spectrometer was operated in top 8 data-dependent method, which dynamically choosing the most abundant precursor ions form survey scan (m/z; 300–2000) for higher-energy collisional dissociation (HCD) fragmentation. The normalized collision energy used was 27%. Dynamic exclusion duration was 30 s and isolation window of precursor was performed with 2. Survey scans were acquired at a resolution of 10 000 at m/z 200. The resolution for HCD spectra was set to 17 500 at m/z 200. MS/MS spectra were searched against IPI human protein (version 3.87, containing 91 491 entries), the sequencereversed decoy databases using SEQUEST of BioworksBrowser rev. 3.1. Full tryptic specificity was required with up to two missed cleavage sites allowed; mass tolerances for precursor ions and fragment ions were set to 10 ppm and 0.8 Da, respectively; fixed modification for carbamidomethylcysteine and variable modifications for methionine oxidation, N-terminal acetylation, and N/Q deamination were used. The SEQUEST search results were further evaluated using the PeptideProphet [29] and ProteinProphet [30] program in Trans Proteomic Pipeline v4.5.33, 34. We selected peptides with cutoff of FDR = 0.01. We then identified the proteins that have ProteinProphet probability 0.95. Among these proteins, we selected the high confidence proteins that were detected in more than two replicates of the three profiles in EV PC9 or PC9R samples and have more than two sibling peptides. C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
The relative quantitation of EV proteins was performed using the absolute protein expression (APEX) Quantitative Proteomics Tool [31]. We first generated a training dataset as 50 most abundant proteins with protein probabilities 0.95 and normalization factor = 1, then computed the predicted peptide count (Oi) of the EV proteins from the training set, and finally calculated APEX abundances of proteins using the protXML file and the urine protein Oi values [31]. The relative APEX abundances in EV from PC9 and PC9R samples were normalized using quantile normalization method [32]. Before identifying differentially enriched EV proteins (DEPs), quantified profiles were examined their distribution and degree of variation between samples for quality assessment (see Supporting Information Fig. 1) using SuperHirn [33]. Using the normalized APEX abundances, we identified DEPs using an integrative statistical method reported previously [34]. Briefly, for each protein, we first applied a t-test and a log2-median-ratio test to its abundances in three replicates. By permuting the replicates, we estimated empirical null distributions and for each protein we computed adjusted p values for the two tests. We then combined the p values from the two tests to compute the overall p value using Stouffer’s method [34]. Finally, we computed FDRs for the overall p values using Storey’s method [35]. We selected DEPs that have the combined FDR < 0.05 and fold change 2.
2.6 Pathway analysis To explore cellular processes, pathways, and subcellular localization represented by DEPs, we identified GO biological process, Kyoto Encyclopedia of Genes and Genomes KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway, and cellular component with p values computed from DAVID software less than 0.05 [36]. The kinase enrichment analysis (KEA) was performed by using Enrichr software (http://amp. pharm.mssm.edu/Enrichr/index.html) to compute kinase enrichment probability based on the distribution of kinase– substrate proportions in the background kinase–substrate database compared with kinases found to be associated with an input list of proteins [37]. To select the top ten kinases, the enrichment score was calculated by combining the p value computed using the Fisher exact test with the z-score of the deviation from the expected rank by multiplying these two numbers.
2.7 Network modeling To reconstruct a subnetwork describing enriched cellular processes in EV from PC9R, we selected 160 proteins that are involved in ten enriched cellular processes from the 664 PC9Rdriven EV-dominant proteins. We then built a Resistancedriven EV protein network model using the interaction information of the 160 proteins obtained from public databases www.proteomics-journal.com
1848
D.-Y. Choi et al.
including HPRD [38], BioGRID [39], STRING [40], and KEGG [41]. The nodes in the network were arranged such that the nodes with the same GOBPs [42] and KEGG pathways were grouped into the same network modules, resulting in the eight modules. 2.8 Cell proliferation assay PC9 cells (recipient) were seeded onto six-well culture plates at a density of 1 × 103 cells in growth medium. When cells grew to 80% confluence, they were washed with PBS twice and serum starved for additional 24 h. Same amount (50 g/mL) of isolated EV from PC9 (PC9 EV) or PC9R (PC9R EV; donor) were added to serum starved PC9 cells (recipient). Cells were incubated for additional 3 days with daily treatment with PC9 EV or PC9R EV. Cell proliferation was quantified by crystal violet staining. Briefly, cells were stained with crystal violet solution and quantified by dissolving stained cells in 10% acetic acid solution. Colorimetric reading was done by absorbance measurement at 570 nm.
Proteomics 2014, 14, 1845–1856
2.10 Apoptosis (or drug sensitivity) assay To assess the effect of PC9 EV or PC9R EV on gefitinibinduced apoptosis, PC9 cells were pretreated with PC9 EV or PC9R EV for 1 h, and followed by treatment of gefinitib (10 M) for additional 16 h. To determine whether dual inhibition of AKT and mTOR re-sensitizes PC9R cells, BEZ235 (100 nM) was treated in presence of gefitinib (0–10 M) for 3 days. Cell mass was quantified by crystal violet staining.
2.11 Statistical analysis The mean of 3 or more replicates was used as average. Data are presented as mean ± SD. The p values were calculated using a standard unpaired Student’s t-test for simple comparisons. Statistical significance is displayed as p < 0.05 (*) or p < 0.005 (**).
3
Results
2.9 Invasion assay
3.1 EV secreted from getinitib-resistant PC9R cells
Thick matrigel-coated inserts were prepared 1 day before experiment. PC9 cells (recipient) were serum starved for 4 h, and each cell suspension (1 × 105 cells/mL, 300 L total) was seeded on the upper surface of each insert. Exogenous PC9 EV or PC9R EV was added into culture medium, and recipient cells were allowed to degrade gelatin coating for 16 h at 37⬚C, 5% CO2 . Noninvasive cells were removed, and bottom side of the inserts was stained. After washing twice to remove background staining, 300 L of extraction solution [43] containing 10% acetic acid was used for extraction. Absorbance was read at 560 nm using a FLUOstar plate reader for quantification. At least triplicates were performed to obtain statistical signification.
In order to understand the biological features of NSCLC cells with acquired resistance to gefitinib, we used PC9R cells, which were developed by prolonged exposure to gefitinib in vitro [44, 45]. PC9R cells have been used in published studies as a model of gefitinib resistance due to acquired EGFR mutation (T790M) [46,47]. While parent PC9 cells were sensitive to gefitinib treatment (0–10 M), resulting in a significant decrease of cell number due to induced apoptosis, PC9R cells were not responsive to gefitinib even at the highest concentration, 10 M (Fig. 1A). To isolate the secreted EV from PC9 or PC9R, cells were first incubated in serum-free medium for 24 h. EV were isolated from the culture medium by serial centrifugations (to remove cell debris), sucrose cushion
Figure 1. Characterization of gefitinib-resistant PC9R, compared to PC9 control cells. (A) The two cell lines (2 × 102 per well) were seeded on 96-well plates in triplicates. Next day, cells were incubated in growth medium in the absence or presence of the varying doses of gefitinib (1, 0.01, 0.1, 1, or 10 M). Cell number was measured by crystal violet assay after 24-h incubation. (B) Representative electron microscopic images of EV shed from PC9 and PC9R cells. Size of bars, 200 nm. (C) Western blot analysis for EV markers (CD81, CD9, and CD63) validating components of isolated EV sample (N 3 independent trials).
C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.proteomics-journal.com
1849
Proteomics 2014, 14, 1845–1856
Figure 2. A systems approach for the identification of EV proteins for gefitinib resistance. (A) Scheme of the proteomics approach to EV for gefitinib resistance. (B) KEGG pathways in which the identified proteins are involved. The number beside the term indicates the number of proteins that belong to the KEGG pathway annotation. (C) The subcellular localization of the identified proteins according to GOCCs (Gene Ontology Cellular Components).
ultracentrifugation, and OptiPrep density gradient ultracentrifugation, as described in Section 2. Isolated EV were checked by electron microscopy for size and intact structure (Fig. 1B), and characteristic EV markers (CD81, CD9, and CD63) [16] were verified by Western blot analysis (Fig. 1C).
3.2 MS proteomics identifies EV proteins characteristic of PC9R Stable alterations in specific pathways or networks might contribute to acquired resistance of NSCLC to gefitinib. In order to test this hypothesis and to understand the characteristic molecular features of drug-resistant PC9R cells, we attempted to identify protein components of EV associated with gefitinib resistance by proteomic analysis on EV isolated from PC9R and PC9 cells. An overview of the procedures used for quantitative EV-driven proteomic comparisons on EV of PC9R (PC9R EV) versus those of PC9 cells (PC9 EV) is illustrated in Fig. 2A. PC9R EV and PC9 EV were prepared in triplicates as described in Section 2. EV proteins were first visualized by 1D SDS-PAGE, followed by nanoLC-MS/MS analyses. LC-MS/MS was performed in triplicate generating a total of 30 proteome profiles for PC9R EV (n = 3) or PC9 EV (n = 3) samples. For protein identification, the resulting MS/MS spectra were searched against IPI human version 3.87 and a sequence-reversed decoy database using C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
SEQUEST (see Section 2). A total of 3121 EV proteins (2339 and 2584 proteins from PC9 EV and PC9R EV, respectively) were identified with an FDR for peptide identification
