Mining the Gastric Cancer Secretome: Identification of GRN as a...
0 downloads
52 Views
5MB Size
Article pubs.acs.org/jpr
Mining the Gastric Cancer Secretome: Identification of GRN as a Potential Diagnostic Marker for Early Gastric Cancer Hendrick Loei,† Hwee Tong Tan,† Teck Kwang Lim,‡ Kiat Hon Lim,§ Jimmy Bok-Yan So,∥ Khay Guan Yeoh,⊥ and Maxey C. M. Chung*,†,‡ †
Department of Biochemistry, Yong Loo Lin School of Medicine, National University of Singapore, 8 Medical Drive, Singapore 117597 ‡ Department of Biological Sciences, Faculty of Science, National University of Singapore, 14 Science Drive 4, Singapore 117543 § Department of Pathology, Singapore General Hospital, Pathology Building, Outram Road, Singapore 169608 ∥ Department of Surgery, Yong Loo Lin School of Medicine, National University of Singapore, 1E Kent Ridge Road, Singapore 119228 ⊥ Department of Medicine, Yong Loo Lin School of Medicine, National University of Singapore, 1E Kent Ridge Road, Singapore 119228 S Supporting Information *
ABSTRACT: Gastric cancer is the second leading cause of cancer deaths worldwide, and currently, there are no clinically relevant biomarkers for gastric cancer diagnosis or prognosis. In this study, we applied a 2D-LC-MS/MS based approach, in combination with iTRAQ labeling, to study the secretomes of the gastric cancer cell lines AGS and MKN7. By performing a comparative analysis between the conditioned media and the whole cell lysates, our workflow allowed us to differentiate the bona f ide secreted proteins from the intracellular contaminants within the conditioned media. Ninety proteins were found to have higher abundance in the conditioned media as compared to the whole cell lysates of AGS and MKN7 cells. Using a signal peptide and nonclassical secretion prediction tool and an online exosome database, we demonstrated that up to 92.2% of these 90 proteins can be exported out of the cells by classical or nonclassical secretory pathways. We then performed quantitative comparisons of the secretomes between AGS and MKN7, identifying 43 differentially expressed secreted proteins. Among them, GRN was found to be frequently expressed in gastric tumor tissues, but not in normal gastric epithelia by immunohistochemistry. Sandwich ELISA assay also showed elevation of serum GRN levels in gastric cancer patients, particularly those with early gastric cancer. Receiver operating characteristic (ROC) curves analysis confirmed that serum GRN can provide diagnostic discriminations for gastric cancer patients KEYWORDS: proteomics, gastric cancer, secretome, iTRAQ , GRN, granulin, biomarker
■
INTRODUCTION Gastric cancer (GC) is the fourth most common cancer in the world but is the second leading cause of cancer-related deaths, claiming almost 800 000 lives annually.1 In Singapore, it is the fifth and seventh most common cancer in men and women, respectively, and it is the fourth leading cause of cancer mortalities.2 Metastasis and peritoneal dissemination are the main causes of gastric cancer deaths. Also, despite advancement in therapeutic modalities for gastric cancer, many gastric cancer patients who have undergone curative surgery suffered recurrence, of which metastatic recurrence and peritoneal carcinomatosis are the most frequent patterns of recurrence.3,4 Presently, there is no clinically relevant biomarker for GC diagnosis and prognosis. Current gastrointestinal cancer biomarkers, such as carcinoembryonic antigen (CEA) and CA 19.9, have low specificity and sensitivity for GC diagnosis or © 2011 American Chemical Society
prognosis. Hence, there is a need for more sensitive and specific biomarkers for gastric cancer. To date, the specific molecular mechanisms associated with gastric cancer metastasis remain poorly defined. Cancer metastasis is a complex, multistep process where the cancer cells are required to acquire specific traits that enable them to overcome the barrier at each step of the metastasis cascade. Proteins secreted by tumors have a profound impact on tumor progression and metastasis, especially at the tumor invasion front. Different classes of secreted proteins, such as proteases, growth factors, cytokines, and angiogenic factors, are involved in various steps of cancer progression and metastasis, such as extracellular matrix (ECM) degradation, cell proliferation, Received: October 11, 2011 Published: December 29, 2011 1759
dx.doi.org/10.1021/pr201014h | J. Proteome Res. 2012, 11, 1759−1772
Journal of Proteome Research
Article
This would help avoid selecting these “red herrings” as targets for downstream validation works.
motility, survival, evasion of host immune responses, angiogenesis, etc.5 Hence, studying the proteins secreted by the tumor may yield a new range of potential biomarkers or therapeutic targets. Biomarker discovery by direct sampling of human blood serum has yielded limited success due to the presence of high abundance serum proteins, such as albumin, immonoglobulins, transferrin, fibrinogen, and haptoglobin. These high abundant proteins inevitably mask the low abundance proteins, which are usually the potential biomarkers. Hence, a number of research groups have turned to the alternative method of analyzing the cancer cell secretome for cancer biomarker discovery. The term “secretome” was initially coined by Tjalsma et al. in a genome-wide survey of the secreted proteins in Bacillus subtilis.6 It encompasses the total proteins that are secreted by a cell through the signal peptide-dependent classical secretory pathway or other nonclassical secretion mechanism. Proteins secreted by tumors into their microenvironment would subsequently enter the bloodstream and/or other body fluids depending on the type of cancer. Hence, studying the cancer secretome would be an attractive approach in identifying new candidate proteins as potential biomarkers for cancer diagnosis or prognosis. Several groups have adopted the secretome-based approach to identify novel biomarkers for various cancer types, such as lung cancer,7−11 pancreatic cancer,12 nasopharyngeal cancer,13 prostrate cancer,14 breast cancer,15,16 and colorectal cancer.17,18 In general, cell lines were incubated in serum-free media, and the conditioned media collected was concentrated for proteomic analysis. By studying the secretomes of their disease of interest, they were able to identify potential candidate biomarkers which were validated in clinical samples by Western blot, immunohistochemistry, or ELISA. This demonstrates the feasibility of a secretome-based strategy to identify novel biomarkers for cancer diagnosis or prognosis. In general, the study of cell line secretomes requires one to incubate the cells in serum-free culture media for a certain period of time, typically 24 h, before harvesting the conditioned media for sample preparation and proteomics analysis. This is to avoid masking of low abundance secreted proteins by the fetal bovine serum proteins. However, cells are likely to undergo spontaneous autolysis due to stress under serum-free conditions, releasing intracellular proteins into the conditioned media. As such, secretome studies are inevitably plagued by constant contamination from intracellular proteins. On the other hand, certain supposedly intracellular proteins without the N-terminal signal peptides can be secreted via nonclassical secretion pathways.19 Hence, not all the intracellular proteins present in the conditioned media are released into the conditioned media, solely due to cell autolysis. A filtering criterion has to be established to select bona f ide secreted proteins while avoiding the contaminants for downstream validation works. In this study, we adopted a proteomics approach using iTRAQ 8-plex coupled with 2-D LC MALDI-TOF/TOF MS to uncover and compare the secretomes of GC cell lines AGS and MKN7. AGS was derived from a primary gastric tumor while MKN7 was derived from a gastric cancer lymph node metastasis. On the basis of the notion that the relative abundance of bona f ide secreted proteins in the CM should be higher than that in the cell lysate, we compared the protein expression levels in the CM and the whole cell lysate to screen for potential contaminants.
■
MATERIALS AND METHODS
Cell Culture
The gastric cancer cell lines AGS, NCI-N87, and KATO-III were purchased from the American Type Culture Collection. MKN7, MKN74, and FU97 were purchased from the Japan Health Science Research Bank. Cells were cultured in RPMI1640 (Sigma Aldrich, St Louis, MO) supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA), 25 mM HEPES (Sigma Aldrich), 100 units/mL penicillin, 100 μg/mL streptomycin (Hyclone, Logan, UT) at 37 °C in 5% CO2. Conditioned Media and Sample Collection
Cells were grown to ∼70% confluency in 175 cm2 culture flasks, and the cell monolayer was washed three times in 1X phosphate buffered saline (PBS) and three times in serum- and phenol red-free RPMI-1640 media. The cells were subsequently incubated in serum- and phenol red-free media for 24 h. The conditioned media were collected and centrifuged at 1000g for 5 min to pellet floating cells. The supernatant was then centrifuged again at 100 000g for 1 h to pellet smaller vesicles and cell debris. The final supernatant was concentrated on a 5 kDa molecular weight cutoff Amicon Ultra-15 Centrifugal Filter Device (Millipore, Bedford, MA) at 3220g at 15 °C. The concentrated samples were then subjected to buffer exchange on the same centrifugal filter device by washing three times with 0.5 M TEAB pH 8.0 (Sigma Aldrich). Adherent cells were trypsinized, washed with PBS, and lysed in 0.5 M TEAB with 1% SDS at 100 °C for 10 min. The whole cell lysate was centrifuged at 20 000g for 1 h at 15 °C, and the final supernatant was collected. Protein concentrations were determined using the Pierce 660 nm Protein Assay with the Ionic Detergent Compatibility Reagent (Pierce Biotechnology, Rockford, IL). Absorbances were read at 660 nm in a Tecan Infinite M200 spectrophotometer (Tecan, Männedorf, Switzerland). iTRAQ 8-plex Labeling
The iTRAQ 8-plex reagents were purchased from Applied Biosystems Inc. (Forster City, CA), and the labeling was carried out following the protocol provided by the manufacturer with minor modifications. A set of AGS whole cell lysate and conditioned media samples and MKN7 whole cell lysate and conditioned media samples was labeled with iTRAQ reagent 113, 114, 115, and 116, respectively. A second biological replicate from a different cell culture passage was prepared and labeled in the same order with iTRAQ reagents 117, 118, 119, and 121. Briefly, 50 μg of each sample was reduced with 50 mM TCEP at 60 °C for 1 h, and cysteinyl residues were blocked with 200 mM methyl methanethiosulfonate (MMTS) at room temperature for 10 min. The SDS concentrations in all the samples were adjusted to 0.05% (w/v) prior to trypsinization. The samples were trypsinized at 37 °C for 16 h. The protein digests were subsequently dried by vacuum centrifugation and resuspended in 50 μL of 0.5 M TEAB. The iTRAQ reagent was added to each tryptic digest and incubated at room temperature for 2 h. Finally, the labeled peptides were then mixed and subjected to cation-exchange chromatography using a handheld cation-exchange cartridge system (Applied Biosystems) to remove interfering substances and excess labels in the sample. The eluate was further desalted using a SEP-PAK column 1760
dx.doi.org/10.1021/pr201014h | J. Proteome Res. 2012, 11, 1759−1772
Journal of Proteome Research
Article
(Millipore), lyophilized, and reconstituted in 50 μL of SCX buffer A (5 mM KH2PO4, pH 3, 5% ACN).
grouped by the ProGroup algorithm in the software to minimize redundancy. A decoy database search strategy was adopted to estimate the false discovery rate (FDR) for peptide identification. A corresponding randomized database was generated using the Proteomics System Performance Evaluation Pipeline (PSPEP) feature in the ProteinPilot Software 2.0.1. The FDR was calculated to be 3.64%, indicating a high reliability in the proteins identified. The results were then exported into Microsoft Excel for manual data interpretation. A 1.5-fold change cutoff for all iTRAQ ratios (1.5) was adopted to identify differentially expressed proteins between AGS and MKN7 cells.
2D Liquid Chromatography
Peptide separation was performed on an Ultimate 3000 liquid chromatography system (Dionex-LC Packings, Sunnyvale, CA) equipped with a Probot MALDI spotting device. Twelve microliters of sample was injected by microliter pickup injection into a 0.5 mm × 23.5 mm strong cation-exchange (SCX) NanoEase trap column (WATERS Corp., Milford, MA) in the first dimensional separation step. The SCX buffers A and B were 5 mM KH2PO4 buffer, pH 3, 5% ACN and 5 mM KH2PO4 buffer, pH 3, 5% ACN and 100 mM KCl, respectively. The flow rate was 6 μL/min. Nine fractions were obtained using step gradients of SCX buffer B: unbound, 0−2, 2−10, 10−20, 20−30, 30−40, 40−50, 50−75, and 75−100%. The eluting fractions were captured alternatingly onto two 0.18 mm × 23.5 mm Symmetry 300 C18 NanoEase trap columns (WATERS Corp.) and washed with 2% ACN, 0.05% TFA followed by gradient elution in a 300 μm × 150 mm Symmetry C18 NanoEase reversed phase column (C18 SYMMETRY 300 TM) (WATERS Corp.). Mobile phases A and B used for second-dimensional separation were 100% water with 0.05% TFA and 100% ACN with 0.04% TFA, respectively. LC fractions eluted from the RP column were then mixed directly with MALDI matrix solution (7 mg/mL α-cyano-4hydroxycinnamic acid (CHCA) and 130 μg/mL ammonium citrate in 75% ACN) at a flow rate of 5.4 μL/min via a 25 nL mixing tee before they were spotted in 28 × 44 spot arrays on 123 mm × 81 mm Opti-TOF LC/MALDI inserts (Applied Biosystems) using a Probot Micro Precision Fraction collector (Dionex-LC Packings), at a frequency of one spot per 5 s.
Bioinformatics and Annotations
Cross-referencing of IPI accession numbers to Uniprot accession numbers was performed with an in-house program based on version 3.68 of the IPI human database. The potential secretion pathways of proteins were predicted with the SecretomeP 2.0 server20 (http://www.cbs.dtu.dk/services/ SecretomeP/) for classical and nonclassical secretion. Protein sequences were retrieved from the Uniprot database and uploaded onto the SecretomeP 2.0 server for ab initio prediction of protein secretion. Potential exosomal release of the proteins was studied by manual annotation on the ExoCarta exosome database21 (http://exocarta.ludwig.edu.au/). Gene Ontology (GO) annotation of proteins was performed using the Software Tool for Researching Annotations of Proteins (STRAP).22 Western Blot Analysis
Ten micrograms of proteins were separated by SDS-PAGE on either 10 or 12.5% polyacrylamide gels and transferred onto PVDF membranes (Biorad, Hercules, CA). The membranes were blocked in either 5% milk or 2% BSA in Tris-buffered saline with 0.1% Tween-20 before being incubated with the following primary antibodies: anti-acrogranin (sc-11342; 1:1000), anti-Cathepsin C (sc-74590; 1:1000), anti-Cystatin C (sc-73878, 1:2000), anti-OPG (sc-71747; 1:1000), and antiPAI-1 (sc-5297; 1:2000) antibodies (Santa Cruz Biotechnology, Santa Cruz, CA); anti-FAM3C (14247-1-AP; 1:500), antiINHBA (60015-1-Ig; 1:1000), and anti-SPARC (15274-1-AP; 1:1000) antibodies (Proteintech Group, Chicago, IL); antiLOXL2 (MAB2639; 3 μg/mL) antibody (R&D Systems); and anti-TSP-1 (catalog number ab1823; 1:1000) antibody (Abcam, Cambridge, U.K.). The secondary antibodies used were sheep anti-mouse IgG HRP-conjugated (catalog number NXA931; 1:5000) from GE Healthcare (Uppsala, Sweden), goat anti-rabbit IgG HRP-conjugated (catalog number 1858415; 1:5000) from Pierce Biotechnology, and donkey anti-goat HRP-conjugated (catalog number sc-2020; 1:5000) from Santa Cruz Biotechnology.
MALDI-MS/MS Analysis
MS and MS/MS analyses were performed on a 4800 MALDITOF/TOF analyzer (Applied Biosystems/MDS Sciex) operating in MS-positive reflector mode. Instrument calibration and optimization was performed using calibration mixture 1 from the 4700 Proteomics Analyzer Mass Standards Kit (Applied Biosystems). Laser intensity was set to 4000 for MS and 4300 for MS/MS acquisition. Typically 1000 shots were accumulated in each spot and MS spectra were acquired between m/z 920 and 3900. The seven precursor ions with the highest peak intensity of each spot with S/N of at least 50 were automatically selected for MS/MS acquisition. MS/MS was performed using air as the collision gas at a collision energy of 1 kV and a collision gas pressure of ∼1 × 10−6 Torr, with an accumulation of 5000 shots for each spectrum. Peptide and Protein Identification and iTRAQ Quantitation
Protein identification and relative iTRAQ quantification were performed with the ProteinPilot Software 2.0.1 (Applied Biosystems/MDS Sciex) using the Paragon algorithm as the database search engine. Each MS/MS spectrum was searched against version 3.68 of the International Protein Index (IPI) human database (87 061 total proteins) released in December 2009. The search parameters allowed for cysteine modification by methyl methanethiosulfonate, and biological modifications specified within the algorithm. The detected protein threshold [Unused ProtScore (Conf)] in the software was set at 1.30 to achieve a 95% confidence. The identified proteins were
Tissue Samples and Microarrays
Three GC tissue microarrays were purchased from US Biomax, Inc. (Rockville, MD). They included a GC tissue with a matched adjacent normal tissue array (40 cases/80 cores) (catalog number ST801), a GC progression array containing normal, inflammation, hyperplasia, and adenocarcinoma tissues (80 cases/80 cores) (catalog number ST805), and a GC tissue with a matched lymph node metastasis tissue array (40 cases/ 80 cores) (catalog number ST810). 1761
dx.doi.org/10.1021/pr201014h | J. Proteome Res. 2012, 11, 1759−1772
Journal of Proteome Research
Article
Immunohistochemistry (IHC)
The study was approved by the relevant Institutional Ethics Review Boards.
Paraffin-embedded tissue sections (4 μm thickness) were stained with polyclonal rabbit anti-GRN antibody (18410-1AP) from Proteintech Group. The sections were deparaffinized in xylene and dehydrated in ethanol before being subjected to antigen retrieval by boiling in 1 mM Tris-EDTA (pH 9). Endogenous peroxidase activities were blocked with the DAKO REAL Peroxidase-Blocking Solution (DAKO Corp., Glostrup, Denmark) before incubation with the anti-GRN primary antibody (dilution factor 1:200) for 1 h at room temperature. The sections were then washed and developed with the Dako REAL EnVision Detection System (DAKO Corp.). The sections were incubated in the provided HRP-conjugated goat antirabbit/mouse secondary antibodies for 30 min, following which it was treated with the DAB+ chromogen-substrate buffer for 15 min. The sections were counterstained with hematoxylin for 30 s, washed, dehydrated in ethanol and xylene, and mounted. IHC staining intensity and percentage were evaluated by a skilled pathologist without prior knowledge of clinicopathological information. Expression of GRN was graded according to the staining intensity: 0 (no staining), 1+ (weak expression), 2+ (moderate expression), and 3+ (strong expression). Percentage of tumor cells stained was scored as follows: 0 (50% positive tumor cells). A staining index was calculated as the product of the staining intensity score and the score associated with the percentage of stained tumor cells.
Colorimetric ELISA for Quantitative Determination of Granulin (GRN) in Human Serum
The GRN concentrations in human serum of healthy individuals and gastric cancer patients were measured using a commercially available human Progranulin ELISA kit (Adipogen Inc., Seoul, Korea). Serum samples were diluted 1:200 using the 1X diluent provided in the kit, and the assay was performed according to the manufacturer’s instructions in duplicate. Data was analyzed with the statistical software OriginPro 8.5 (OriginLabs, Northampton, MA) and Microsoft Excel.
■
RESULTS
Secretomes of AGS and MKN7
The iTRAQ-labeled conditioned media protein samples of each cell lines were analyzed together with their respective whole cell lysates as shown in Figure 1. We identified 703 proteins with at least 95% confidence (Supporting Information Table 1). To identify proteins that are truly secreted by AGS and MKN7 cells, we compared the relative abundance of all the proteins in the conditioned media of each cell line with its respective whole cell lysate. We reasoned that truly secreted proteins of a cell should have a higher abundance in the conditioned media than in the cell lysate, i.e. CM/Lysate ratio >1. For a higher stringency, we used a higher cutoff of >1.5. Also, the ratio has to be reproducible in both biological replicates. Fifty-five and seventy-four proteins were found to have CM/Lysate ratio >1.5 in AGS and MKN7 cells, respectively, with an overlap of 39 proteins in between the cell lines (Figure 2A). The identities of these 90 proteins, together with their respective CM/Lysate ratios, are presented in Supporting Information Table 2.
Patient Population and Sample Collection
Preoperative serum samples were collected from gastric cancer patients admitted into the National University Hospital and Tan Tock Seng Hospital, Singapore, from 2006 onward. These patients were enrolled, with written informed consent, into the Gastric Cancer Biomarker Discovery II (GASCAD II) study. Blood was drawn from the patient prior to surgery, and the serum was obtained by centrifugation at 20 °C for 10 min in serum separator tubes. The serum was stored at −80 °C until use. Serum samples from noncancer individuals were obtained from a previous clinical studythe Gastric Cancer Epidemiology and Molecular Genetics Programme (GCEP). Subjects recruited under this program were systematically screened for early GC by endoscopy, and the subjects whose serum samples were used for the current study were determined to be free of GC at the point of blood collection.
Mode of Secretion of the Proteins in the CM
To investigate if the above-mentioned 90 proteins were secreted by the classical or nonclassical secretory pathways, we used the SecretomeP 2.0 prediction server for secretion pathway prediction. Seventy proteins (77.8%) were predicted to carry the N-terminal signal peptide that would target these proteins to the classical secretory pathway, 7 (7.8%) were predicted to be secreted nonclassically while the remaining 13 (14.4%) were unconfirmed (Figure 2B). Exosomal release is also a form of nonclassical secretion mechanism.19
Figure 1. Overview of the iTRAQ labeling strategy employed in this study for secretome and proteome comparison by 2D-LC MS/MS. Samples were trypsinized prior to iTRAQ labeling, and the labeled peptides were then mixed, separated by 2D-LC, and identified by MALDI-TOF/TOF MS/MS. Database search and iTRAQ analysis were performed using the ProteinPilot 2.0 software from AB SCIEX. 1762
dx.doi.org/10.1021/pr201014h | J. Proteome Res. 2012, 11, 1759−1772
Journal of Proteome Research
Article
To determine if the proteins without the signal peptide are released via exosomes, we consulted the Exocarta exosome database. This database is a collection of exosomal contents reported in multiple organisms from various publications.21 Thirteen out of the twenty proteins without the signal peptides were reported in Exocarta. In summary, up to 92.2% (83 out of 90) of proteins in the above list have the potential to be targeted to the extracellular space by various mechanisms. On the basis of these results, we are confident that our approach of comparing protein abundance in the CM with the cell lysate is a sound strategy to screen for potential contaminants in the CM for secretome studies.
Figure 2. Secretome of AGS and MKN7 cells. (A) Venn diagram showing the number of proteins with CM/Lysate ratio >1.5 that were identified in AGS and MKN7 cells. The intersection indicates the number of commonly secreted proteins between the two cell lines. (B) SecretomeP prediction of the above-mentioned 90 proteins with CM/ Lysate ratio >1.5. Amino acid sequences of these proteins were retrieved from Uniprot and imported into the SecretomeP 2.0 server for ab initio prediction for nonclassical secretion. 70 proteins were predicted to carry an N-terminal signal peptide, 7 were predicted to be secreted nonclassically, and 13 were unconfirmed.
Comparative Analysis between AGS and MKN7 Secretomes
AGS was derived from a primary gastric tumor while MKN7 was derived from a gastric cancer lymph node metastasis. Differences in the secretome and proteome profiles between these two cell lines would be likely to contribute to the metastatic
Table 1. Secreted Proteins with Higher Abundance in the Conditioned Media of MKN7 Compared to AGS IPI acc no.a
unusedb
totalc
peptd
% seq cove
IPI00926949.2
8.01
8.01
4
17.1
IPI00294839.6 IPI00329573.9 IPI00911063.1
4.5 7.26 3.69
4.5 7.35 3.69
2 4 2
11.5 9.3 18.2
IPI00339225.1 IPI00298362.3
33.57 7.28
33.57 7.4
21 3
24.6 14.5
IPI00827658.1 IPI00296099.6 IPI00028670.1 IPI00011229.1 IPI00902471.1
8.3 64.2 10.25 9.3 2.18
8.3 64.2 10.25 9.3 2.29
4 52 5 5 1
19.4 40.9 38 18.9 20.2
IPI00014572.1 IPI00783987.2 IPI00296713.4 IPI00926250.1 IPI00017704.3 IPI00871467.2 IPI00000494.6 IPI00924935.1
5.05 16.6 5.05 2.27 13.92 10.25 6 1.32
5.05 16.6 5.05 2.27 13.92 10.25 6 1.32
2 8 2 1 8 5 3 1
6.9 11.1 11.3 19.2 45.1 13.7 22.6 9.6
IPI00015117.2 IPI00917285.1
5.22 15.33
5.22 15.33
2 8
5.5 59.3
IPI00910781.1 IPI00944977.1 IPI00783665.4 IPI00301395.4
23.42 12.33 5.46 4
23.44 12.33 5.6 4.07
16 6 3 2
24.3 30.9 9.8 8.8
name SERPINE1 plasminogen activator inhibitor-1 isoform 2 precursor LOXL2;ENTPD4 lysyl oxidase homologue 2 COL12A1 isoform 1 of collagen alpha-1(XII) chain PLAT cDNA FLJ59355, highly similar to tissue-type plasminogen activator FN1 isoform 5 of fibronectin TNFRSF11B tumor necrosis factor receptor superfamily member 11B CD44 isoform 7 of CD44 antigen THBS1 thrombospondin-1 INHBA inhibin β A chain CTSD cathepsin D cDNA FLJ35128 fis, clone PLACE6008768, moderately similar to insulin-like growth factor-binding protein 3 SPARC SPARC C3 complement C3 (fragment) GRN isoform 1 of granulins FAM3C putative uncharacterized protein FAM3C COTL1 coactosin-like protein L1CAM isoform 1 of neural cell adhesion molecule L1 RPL5 60S ribosomal protein L5 TFRC cDNA FLJ57106, highly similar to transferrin receptor protein 1 LAMC2 isoform long of laminin subunit gamma-2 TIMP2 cDNA FLJ57920, highly similar to metalloproteinase inhibitor 2 GPI glucose-6-phosphate isomerase ALCAM cDNA, FLJ79012, highly similar to CD166 antigen LAMA5 laminin subunit alpha-5 CPVL probable serine carboxypeptidase CPVL
ave ratiof
sd
secreg
23.42
4.35
SP
10.95 10.57 10.08
4.10 1.60 1.34
SP SP SP
9.71 9.19
2.74 1.83
SP SP
8.29 7.34 7.22 7.13 7.04
1.02 1.79 2.49 0.67 1.17
SP SP SP SP SP
6.39 6.10 5.87 4.82 3.81 3.22 3.15 2.70
0.63 0.40 1.90 1.99 0.65 0.81 0.86 0.28
SP SP SP SP NC SP U NC
2.69 2.69
0.83 0.16
SP SP
2.67 2.37 2.09 1.85
0.43 0.32 0.56 0.21
U SP SP SP
ExoCartah
present absent present
present
a
IPI accession numbers of the proteins from the IPI human database v3.68. bUnused ProteinPilot score. Only proteins with ProteinPilot score >1.3 (>95% confidence) were considered. cTotal ProteinPilot score. Only proteins with ProteinPilot score >1.3 (>95% confidence) were considered. d The number of distinct peptides having at least 95% confidence. ePercentage sequence coverage of identified peptides. fAverage of four iTRAQ MKN7/AGS ratios, i.e. 116/114, 121/114, 116/118, and 121/118. The ratios were followed by the corresponding standard deviation (s.d.). g SecretomeP prediction for nonclassical secretion (NC). Classically secreted proteins would contain the N-terminal signal peptide (SP) while proteins predicted to be neither classically nor nonclassically secreted were labeled unconfirmed (U). hPrediction of exosomal release via manual annotation against the Exocarta database. Proteins that were not classically secreted and were present in the database were labeled “present”, while those that were not were labeled “absent”. Peptide information is available in Supporting Information Table 3. 1763
dx.doi.org/10.1021/pr201014h | J. Proteome Res. 2012, 11, 1759−1772
Journal of Proteome Research
Article
Table 2. Secreted Proteins with Higher Abundance in the Conditioned Media of AGS Compared to MKN7 IPI acc. no.a
unusedb
totalc
peptd
% seq cove
name
ave ratiof
sd
secreg
IPI00010333.1
2.02
2.02
1
9.1
MDK midkine
0.05
0.06
SP
IPI00019038.1
11.84
11.84
12
60.8
LYZ lysozyme C
0.07
0.02
SP
IPI00305477.5
38.79
38.79
55
75.2
CST1 cystatin-SN
0.09
0.07
SP
IPI00556442.1
3.52
3.52
2
12.7
IGFBP2 insulin-like growth factor binding protein 2 variant (fragment)
0.19
0.04
SP
IPI00642739.1
6.92
6.92
4
48.3
TIMP1 TIMP metallopeptidase inhibitor 1
0.24
0.06
SP
IPI00009950.1
4.92
4.92
3
8.4
LMAN2 vesicular integral-membrane protein VIP36
0.24
0.12
SP
2
18.2
PROCR endothelial protein C receptor precursor
0.27
0.07
SP
IPI00032293.1
16.04
16.04
14
72.6
CST3 cystatin-C
0.31
0.03
SP
IPI00002147.4
1.42
1.42
1
9.9
CHI3L1 chitinase-3-like protein 1
0.39
0.14
SP
IPI00009276.2
4
4
IPI00295542.5
9.68
9.69
4
33.8
NUCB1 nucleobindin-1
0.40
0.11
SP
IPI00024284.5
35.33
35.33
18
14.8
HSPG2 basement membrane-specific heparan sulfate proteoglycan core protein
0.41
0.12
SP
IPI00022810.1
5.23
5.23
2
17.7
CTSC dipeptidyl-peptidase 1
0.41
0.16
SP
IPI00023728.1
8.11
8.11
4
20.1
GGH γ-glutamyl hydrolase
0.41
0.08
SP
IPI00952639.1
14.34
14.34
8
16
APP cDNA FLJ50531, highly similar to amyloid β A4 protein
0.43
0.09
U
IPI00004656.3
6.52
6.52
8
26.1
B2M β-2-microglobulin
0.45
0.13
SP
IPI00908558.1
6.21
6.21
4
21.9
LSR cDNA FLJ55699, highly similar to Homo sapiens liverspecific bHLH-Zip transcription factor (LISCH7), transcript variant 3, mRNA
0.46
0.07
U
IPI00000013.1
6.84
6.89
4
23.7
CTSL2 cathepsin L2
0.55
0.02
SP
IPI00016353.1
11.03
11.03
5
27.1
DKK1 dickkopf-related protein 1
0.58
0.07
SP
ExoCartah
present
present
a
IPI accession numbers of the proteins from the IPI human database v3.68. bUnused ProteinPilot score. Only proteins with ProteinPilot score >1.3 (>95% confidence) were considered. cTotal ProteinPilot score. Only proteins with ProteinPilot score >1.3 (>95% confidence) were considered. dThe number of distinct peptides having at least 95% confidence. ePercentage sequence coverage of identified peptides. fAverage of 4 iTRAQ MKN7/AGS ratios, i.e. 116/114, 121/114, 116/118, and 121/118. The ratios were followed by the corresponding standard deviation (sd). gSecretomeP prediction for nonclassical secretion (NC). Classically secreted proteins would contain the N-terminal signal peptide (SP) while proteins predicted to be neither classically nor nonclassically secreted were labeled unconfirmed (U). hPrediction of exosomal release via manual annotation against the Exocarta database. Proteins that were not classically secreted and were present in the database were labeled “present”, while those that were not were labeled “absent”. Peptide information is available in Supporting Information Table 3.
potential in MKN7 cells. We investigated the expression levels of the above-mentioned 90 secreted proteins in AGS and MKN7 CM fractions (i.e., 116/114, 116/118, 121/114, and 121/118) and found 43 to be differentially expressed when a cutoff of >1.5 and
