Comparative and Targeted Proteomic Analyses of ... - ACS Publications

Article pubs.acs.org/jpr

Comparative and Targeted Proteomic Analyses of Urinary Microparticles from Bladder Cancer and Hernia Patients Chien-Lun Chen,†,# Yue-Fan Lai,‡,# Petrus Tang,§,# Kun-Yi Chien,‡,∥ Jau-Song Yu,‡,∥ Cheng-Han Tsai,‡ Hsiao-Wei Chen,∥ Chih-Ching Wu,⊥ Ting Chung,∥ Chia-Wei Hsu,‡ Chi-De Chen,‡ Yu-Sun Chang,‡,∥ Phei-Lang Chang,† and Yi-Ting Chen*,∥ †

Chang Gung Bioinformatics Center, Department of Urology, Chang Gung Memorial Hospital, Taoyuan 333, Taiwan Graduate Institute of Biomedical Sciences, College of Medicine, Chang Gung University, Taoyuan 333, Taiwan § Chang Gung Bioinformatics Center, Chang Gung University, Taoyuan 333, Taiwan ∥ Molecular Medicine Research Center, College of Medicine, Chang Gung University, Taoyuan 333, Taiwan ⊥ Graduate Institute of Medical Biotechnology and Department of Medical Biotechnology and Laboratory Science, Chang Gung University, Taoyuan 333, Taiwan ‡

S Supporting Information *

ABSTRACT: Bladder cancer is a common urologic cancer whose incidence continues to rise annually. Urinary microparticles are an attractive material for noninvasive bladder cancer biomarker discovery. In this study, we applied isotopic dimethylation labeling coupled with liquid chromatography-tandem mass spectrometry (LC−MS/MS) to discover bladder cancer biomarkers in urinary microparticles isolated from hernia (control) and bladder cancer patients. This approach identified 2964 proteins based on more than two distinct peptides, of which 2058 had not previously been reported as constituents of human urine exosomes/ microparticles. A total of 107 differentially expressed proteins were identified as candidate biomarkers. Differences in the concentrations of 29 proteins (41 signature peptides) were precisely quantified by LC− MRM/MS in 48 urine samples of bladder cancer, hernia, and urinary tract infection/hematuria. Concentrations of 24 proteins changed significantly (p < 0.05) between bladder cancer (n = 28) and hernia (n = 12), with area-under-the-curve values ranging from 0.702 to 0.896. Finally, we quantified tumor-associated calcium-signal transducer 2 (TACSTD2) in raw urine specimens (n = 221) using a commercial ELISA and confirmed its potential value for diagnosis of bladder cancer. Our study reveals a strong association of TACSTD2 with bladder cancer and highlights the potential of human urinary microparticles in the noninvasive diagnosis of bladder cancer. KEYWORDS: urine, microparticle, bladder cancer, biomarker discovery, biomarker verification, multiple reaction monitoring, multiplexed quantitation

■

INTRODUCTION

Urine exosomes/microparticles are stable under appropriate collection, storage, and processing conditions, highlighting their potential role in biomarker studies.4 However, most previous studies on the urine proteome have focused on differential proteomic profiling of soluble urine proteins between patients and controls.5−8 Information about the qualitative and quantitative composition of the exosomal/microparticle component of the clinical urine proteome is limited.9 In this study, we established a protein-extraction workflow for urinary microparticles for use in dimethylation labeling, and evaluated its reproducibility. To gain greater insight into quantitative differences in the profiles of urinary microparticles between hernia and bladder cancer patients, we used two types

Bladder transitional cell carcinoma is a major epidemiological problem whose incidence continues to rise each year.1,2 Urine cytology and cystoscopy are the major clinical tools for diagnosing bladder cancer. However, urine cytology shows poor sensitivity in the detection of low-grade bladder tumors, and additionally depends on the level of expertise of the pathologist for accurate interpretation.3 Conventional cystoscopy is an invasive procedure that, when used as a diagnostic tool, has a limited ability to detect flat lesions such as carcinoma in situ, which can lead to incomplete resection and higher recurrence rates. Therefore, a noninvasive, inexpensive, and highly sensitive/specific bladder cancer marker would be helpful in improving diagnosis, decreasing patient morbidity, and/or lowering costs associated with surveillance cystoscopy. © 2012 American Chemical Society

Received: November 22, 2011 Published: October 19, 2012 5611

dx.doi.org/10.1021/pr3008732 | J. Proteome Res. 2012, 11, 5611−5629

Journal of Proteome Research

Article

Hitachi). The supernatant was then removed and the pellet (microparticles) was resuspended in 50 μL of PBS. After vacuum drying the pellet, 5 μL of lysis buffer (10 mM Tris-HCl, 1 mM EDTA, 1 mM EGTA, 50 mM NaCl, 50 mM NaF, 20 mM Na4P2O7, 1 mM Na3VO4, and 1% Triton X-100) was added and the tube was incubated on ice for 15 min prior to addition of 45 μL of PBS. The amount of protein in each urinary microparticle sample was measured using a DC Protein Assay Kit (Bio-Rad, Hercules, CA), after which samples were stored at −20 °C for subsequent processing. The amount of microparticle protein extracted from individual urine samples ranged from 0.01 to 3 μg/mL.

of sample preparationspooled microparticle samples and individual samples without poolingfor discovery-phase experiments in this work. After integrating data sets from two discovery experiments, 29 proteins were further precisely quantified in additional individual urine microparticle samples by LC−MRM/MS. One of the target proteins, TACSTD2, was further verified by ELISA in a larger number of individual urine specimens. We herein demonstrate that differential isotope labeling via dimethylation, coupling with LC−CID−MS/MS and MRM−MS, can substantially facilitate an MS-based workflow for biomarker discovery/verification in urinary microparticles.

■

Electron Microscopy Analysis

MATERIALS AND METHODS

The quality of enriched microparticles was examined by electron microscopy (EM) imaging of exosomes. Briefly, the microparticle pellet was resuspended in 20−30 μL of 2% paraformaldehyde (PFA). A 5-μL drop of microparticle suspension was placed on clean Parafilm, after which a carbon-coated EM grid was placed on top of the drop and allowed to stand for 20 min to adsorb the fluid. The grid with adherent microparticles was washed with 100 μL of PBS, fixed with 50 μL of 1% glutaraldehyde for 5 min, and then washed with 10 100-μL drops of distilled water. The grid was transferred to a 50-μL drop of uranyl-oxalate solution and allowed to stand for 5 min. The grid was then incubated with a 50-μL drop of methyl cellulose-uranyl acetate on ice for 10 min. Excess fluid was removed and the grid was dried for 10 min at room temperature before examination with a JEM-2000EXII electron microscope (JEOL, Tokyo, Japan).

Clinical Specimens

In this study, age-matched hernia patients were chosen as the control subgroup because of no pathological problems in their urological system. The possible changes of urine proteome caused by age difference could be also excluded. The urine specimens were collected in the morning of the surgery dates of hernia and bladder cancer patients. Clinical specimens were collected as previously described.6,10 Briefly, the first morning urine samples were collected in the presence of a protease inhibitor cocktail tablet (one tablet/50 mL of urine; Roche, Mannheim, Germany) and sodium azide (1 mM) from nontumor volunteers and bladder cancer patients. The collected samples were centrifuged at 5000g for 30 min at 4 °C within 5 h to remove cells and debris, and the clarified supernatants were stored at −80 °C for subsequent processing. All urine samples were collected at Chang Gung Memorial Hospital, Taoyuan, Taiwan. The study protocol was approved by the Medical Ethics and Human Clinical Trial Committee at Chang Gung Memorial Hospital. In the discovery phase of this study, hernia patients (n = 9, 60.1 ± 8.2 years old) were defined as the control group. The urological disease group (n = 9, 65.3 ± 12.0 years old) included low-grade/early stage bladder cancer (LgEs; n = 2), high-grade/early stage bladder cancer (HgEs; n = 5), and high-grade/advanced-stage bladder cancer (HgAs; n = 2). In verification-phase experiments, additional individual samples, including 12 hernia patients (61.0 ± 8.8 years old) and 28 bladder cancer patients (67.8 ± 13.7 years old), were used for biomarker verification by LC−MRM/MS; these included LgEs (n = 7), HgEs (n = 17), and HgAs (n = 4) bladder cancers. An additional eight samples from patients (53.1 ± 9.5 years old) with hematuria (HU; n = 5) or urinary tract infection (UTI; n = 3) were also included as a disease control group. In total, 221 individual samples were used for mensurements of urinary TACSTD2 by ELISA. Supplemental Tables 1A−1C list the diagnosis, sex, and other clinical characteristics of all patient samples.

Western Blot Analysis

Soluble urinary proteins and microparticle proteins (5 μg) were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred electrophoretically onto polyvinylidene fluoride (PVDF) membranes (Bio-Rad, Hercules, CA) for characterization of microparticle proteins. The membranes were blocked by incubating for 1 h at room temperature with 5% nonfat dried milk in Tris/Tween-buffered saline (TBST; pH 7.4) containing 0.05% Tween 20 (Sigma, St Louis, MO). Membranes were probed with primary antibodies against TSG101 (anti-TSG101, 1:200; Santa Cruz Biotechnology, Santa Cruz, CA) and CD9 (anti-CD9 1:200; Santa Cruz Biotechnology) followed by incubation with horseradish peroxidase-conjugated secondary antibodies. Immunoreactive proteins were visualized using enhanced chemiluminescence detection, according to the manufacturer’s instructions (Western Lighting plus-ECL; Perkin-Elmer, The Netherlands). Total urine proteins (100 μg) from individual samples were resolved on SDS-PAGE gels and transferred electrophoretically onto PVDF membranes (Millipore, Billerica, MA) for biomarker verification. The membranes were blocked for 1 h at room temperature with 5% nonfat dried milk in Tris-buffered saline (J. T. Baker, Phillipsburg, NJ) with 0.1% Tween-20 (Sigma, St. Louis, MO) TBST. Afterward, the membranes were probed using anti-TACSTD2 antibody (BAF650, R&D) at 1:500 overnight at 4 °C. The membranes were probed with primary antibody followed by streptavidin−alkaline horseradish peroxidase-conjugated secondary antibody, and developed using enhanced chemiluminescence detection according to the manufacturer’s instructions (Millipore, Billerica, MA). The relative signal intensity of TACSTD2 protein detected in the blots was quantified using a computing densitometer (Molecular Dynamics, Sunnyvale, CA).

Enrichment of Urinary Microparticles by Ultracentrifugation

Urinary microparticle proteins were purified by ultracentrifugation as previously described.9,11 Briefly, 12.5-mL urine samples were thawed at 4 °C and centrifuged at 17 000g for 30 min (at 4 °C) to remove large cells and other debris. Then, the supernatant was centrifuged at 100 000g for 70 min at 4 °C (L8-80M; Beckman) to pellet the small vesicles that correspond to microparticles. The resulting pellet was transferred to a microcentrifuge tube, washed with 5 mL of phosphate-buffered saline (PBS) to eliminate contaminating proteins, and centrifuged at 100 000g for 70 min at 4 °C (CS150 GXL; 5612



Article

Flow Cytometric Analyses of Microparticle-Coated Beads

Comparative Analysis between a Pooled Bladder Cancer Sample and a Pooled Hernia Sample Using Online Strong Cationic Exchange (SCX) and Reverse-Phase LC/MS/MS

One microgram of microparticles was incubated with 1 μL of latex beads (surfactant-free, aldehyde sulfate 3.9 μm beads, Interfacial Dynamics) that had been washed twice in 1× PBS buffer (10 mM Na2HPO4, 0.9% NaCl, pH 7.4). The microparticle-coated beads were incubated in 100 μL of PBS buffer at room temperature for 30 min and incubated in a final volume of 200 μL PBS followed by rolling overnight at 4 °C. The microparticle-coated beads were then washed by PBS buffer containing 0.5% BSA (the PBS-BSA buffer) and centrifuged at 4000 rpm for 3 min (4 °C). Pellet of microparticle-coated beads was resuspended in 1 mL of PBSBSA buffer and centrifuged at 4000 rpm for 3 min twice. The microparticle-coated beads were then incubated with primary monoclonal antibodies of CD9 (anti-human CD9 conjugated biotin, 1:100, 13-0098; eBioscience, San Diego, CA) in 100 μL of PBS-BSA buffer for 30 min at 4 °C, washed twice by PBSBSA buffer and incubated with Streptavidin−Phycoerythrin (1:200, 016-110-084; Jackson ImmunoResearch Laboratories) in 100 μL PBS-BSA buffer for 30 min at 4 °C. After washing by PBS-BSA buffer, beads were analyzed by flow cytometry using a BD FACSCalibur flow cytometer instrument configured with BD CellQuest Pro 4.0.2 software (BD Bioscience).

Nine bladder cancer samples were pooled as a disease sample and nine hernia samples were pooled as an age-matched control sample. The pooled urinary microparticle samples from control and bladder cancer patients were labeled with light-formaldehyde and heavy-formaldehyde, respectively. The pooled hernia and bladder cancer tryptic peptides were mixed at a 1:1 ratio based on the total protein amount. A sample of the microparticle peptide mixture containing 50 μg of protein was then subjected to online two-dimensional (2D) strong cation exchange-reverse phase (SCX-RP) LC/MS/MS technology for identification of differentially expressed proteins. For online 2D SCX-RP chromatography using Dionex UltiMate 3000 LC systems (Germany), 50 μg of isotopic dimethyl-labeled peptide mixture was rehydrated with 50 μL of 0.1% (v/v) formic acid to produce a 1 μg/μL concentration, loaded onto a trap column (ZORBAX 300SB-C18, 5 μm, 0.3 × 5 mm; Agilent Technologies, Wilmington, DE), and washed using buffer systems of buffer A (0.1% formic acid in 30% acetonitrile) and buffer B (0.1 M NH4Cl and 0.1% formic acid in 30% acetonitrile) at a flow rate of 1.5 μL/min with a linear gradient from 13% buffer B to 16.5% buffer B for 120 min. The peptides were first fractionated on the SCX column (5 μm, 0.75 × 300 mm; Phenomenex) into 13 fractions at a flow rate of 1− 1.5 μL/min across the SCX column with a salt gradient from 16.5% buffer B to 100% buffer B for a total of 1560 min (120 min for each fraction). The peptides were further fractioned with a salt gradient using buffer systems of buffer B and buffer C (0.4 M NH4Cl and 0.1% formic acid in 30% acetonitrile). The peptides were separated into nine fractions at a flow rate of 1−1.5 μL/min across the same SCX column with a salt gradient from 100% buffer B to 100% buffer C for a total of 1080 min (120 min for each fraction). The eluted peptides in the total 22 fractions were then further separated sequentially on the second LC step using a resolving C18 nano reverse-phase LC column (1.7 μm, 120 mm; Waters) at a flow rate of 0.25 μL/ min using buffer systems of buffer D (0.05% formic acid in H2O) and buffer E (0.05% formic acid in 100% acetonitrile). The peptides were fractionated across the C18 column with a linear gradient of 7−13% buffer E (0.05% formic acid in 100% acetonitrile) for 3 min, 13−30% buffer E for 67 min, and 30− 45% buffer E for 20 min, and then washed and equilibrated for 30 min. The LC apparatus was coupled online to a linear ion traporbitrap (LTQ-Orbitrap; Thermo Fisher, San Jose, CA) operated using Xcalibur 2.0 software (Thermo Fisher). Intact peptides were detected in the Orbitrap at a resolution of 30 000. Internal calibration was performed using the ion signal of [Si(CH3)2O]6H+ at m/z 445.120025 as a lock mass. We used a data-dependent procedure that alternated between one MS scan and six MS/MS scans for the six most abundant precursor ions in the MS survey scan. The m/z values selected for MS/ MS were dynamically excluded for 180 s. The electrospray voltage applied was 1.8 kV. Both MS and MS/MS spectra were acquired using a single microscan with a maximum fill times of 1,000 and 100 ms for MS and MS/MS analyses, respectively. Automatic gain control was used to prevent overfilling of the ion trap, and 5 × 104 ions were accumulated in the ion trap for the generation of MS/MS spectra. The m/z scan range for MS scans was 350 Da to 2000 Da.

Tryptic Digestion of Urinary Microparticle Proteins

Six different procedures (Supplemental Figure 1) for trypsin digestion were evaluated by SDS-PAGE. In the optimized tryptic digestion procedure (procedure E), which was used for this study, 25 μg of urinary microparticle protein was reduced by dithiothreitol and incubation at 56 °C for 30 min. The sample was then alkylated by iodoacetamide at room temperature in the dark for 30 min. Sequencing-grade modified trypsin (Promega, Madison, WI) was added to microparticle proteins at a 1:25 (enzyme/substrate) ratio. The samples were incubated with trypsin at 37 °C for 16 h to achieve complete enzyme digestion. The tryptic peptides were stored at −20 °C for subsequent processing. Differential Dimethyl Labeling of N-Termini and Lysine Residues of Tryptic Peptides

The tryptic peptide sample (25 μg in 32 μL buffer at pH 8) was mixed with 7.5 μL of freshly prepared sodium cyanoborohydride (2 M). After the addition of 1.5 μL of 4% (w/w) formaldehyde-12CH2 (light labeling) or formaldehyde-13CD2 (heavy labeling) solution, the mixtures were allowed to react for 1 h at 37 °C. The reactions were quenched by adding 1.5 μL of 1 M ammonium bicarbonate. This procedure converts all primary amines (N-terminus and the side chain of lysine residues) in a tryptic peptide mixture to dimethylamines with isotopic dimethyl tags. The mass shifts per labeling site were +28 Da and +34 Da for light peptides and heavy peptides, respectively. The labeled samples were mixed in a 1:1 ratio (based on the total protein amount), and the pH of the mixture was adjusted to pH 3 by adding 2% trifluoroacetic acid in 20% acetonitrile (TFA solution/sample =1:3 [v/v]), followed by desalting and concentration using a C18 spin column (Pierce), according to the manufacturer’s protocol. The eluted samples were frozen and lyophilized to dryness and then analyzed by LC−MS/MS; the mass difference (6 Da) of the isotopic dimethyl labels was used to compare peptide abundance in different samples. 5613



Article

org/index.html# in May, 2011),16 1160 proteins from the HUExoDataBase (downloaded from the Web site http://dir. nhlbi.nih.gov/papers/lkem/exosome/ in May, 2011),17 8336 plasma proteins from the Sys-Bodyfluid database by Li et al,18 and 9504 plasma proteins from the HUPO Plasma Proteome Project.19 Molecular Function, Cellular Component, and Biological Process distributions among the identified urinary exosome proteins and differentially expressed proteins in the urine microparticle proteome were also analyzed using ProteinCenter software (version 3.7; Proxeon Bioinformatics).

Comparative Analysis of 18 Individual Urine Samples with a Pooled Internal Standard Sample Using One-Dimensional RP−LC−MS/MS Analysis

To compensate for the possible loss of information about individual variation due to sample pooling, we separately mixed the 18 individual microparticle samples used in the pooled strategy, described above, with a pooled microparticle sample (as a global internal standard). The internal standard was prepared by mixing equal amounts of microparticle peptides from the 18 individual samples. Peptides from each individual sample were separately labeled with formaldehyde-12CH2 (light labeling), and the internal standard was labeled with formaldehyde-13CD2 (heavy labeling). The light and heavy peptides were mixed at a 1:1 ratio, and 2 μg of the peptide mixture was subjected to one-dimensional nano reverse-phase LC/MS/MS analysis by LTQ-Orbitrap. The mixed peptides were separated on a resolving 10-cm analytical BioBasic C18 PicoFrit column (inner diameter, 75 μm) with a 15-μm tip (New Objective, Woburn, MA). Peptides were eluted at a flow rate of 0.25 μL/min across the analytical column with a linear gradient of 2−35% buffer E (0.05% formic acid in 100% acetonitrile) for 58 min, 35−90% buffer E for 11 min, 90% buffer E for 5 min, and then were equilibrated in 2% buffer E for 15 min. The parameters of LC-Orbitrap analysis were the same as those used for analysis of the pooled sample, described above. The peptide concentration ratio was determined by measuring peak-area ratios of 12CH2-dimethylated/13CD2-dimethylated peptides. The protein concentration ratios of individual samples/internal standard were determined using MaxQuant software (version 1.1.1.25).

LC−MRM/MS Analysis and Data Acquisition

A nanoACQUITY UPLC System was used for the injection of dimethylated peptides. LC−MRM/MS analysis of each sample took 70 min. One-microliter samples (representing 0.5 μg of heavy and light peptide mixtures) were injected onto a trap column (nanoACQUITY UPLC C18, 180 μm × 20 mm, 5-μm particle size; Waters) at a flow rate of 15 μL/min in buffer F (0.1% formic acid in H2O), and separated on a resolving analytical column (ACQUITY UPLC BEH130 C18, 75 μm × 150 mm, 1.7-μm particle size; Waters). Samples were then separated using a flow rate of 300 nL/min with a 0.5-min linear gradient from 3 to 10% solvent G (0.1% formic acid in acetonitrile), then a 31.5-min linear gradient from 10 to 30% solvent G, followed by a 10-min linear gradient from 30 to 40% solvent G, and finally a 1-min linear gradient from 40 to 97% solvent G. The analytical column was then reconditioned by holding solvent G at 97% for 10 min prior to ramping back down to 3% solvent G over 1 min and re-equilibrating for 16 min with 3% solvent G. A blank solvent injection (70-min analysis at 300 nL/min with the same LC gradient used for clinical samples) was run between all samples to prevent sample carryover on the HPLC column. An AB/MDS Sciex 5500 QTRAP with a nanoelectrospray ionization source controlled by Analyst 1.5.1 software (AB Sciex) was used for all LC-MRM/MS analyses. All acquisition methods used the following parameters: ion spray voltage, 1900−2000 V; curtain gas setting, 20 psi (UHP nitrogen); interface heater temperature, 150 °C; and MS operating pressure, 3.5 × 10−5 Torr; Q1 and Q3 were set to unit resolution (0.6−0.8 Da full width at half height). MRM acquisition methods were constructed using three MRM ion pairs per peptide with fragment-ion-specific tuned collision energy (CE) voltages and retention time constraints. A default collision cell exit potential of 35 V was used for all MRM ion pairs, and the scheduled MRM option was used for all data acquisition, with a target cycle time of 2 s and a 4-min MRM detection window. Transitions of 82 peptides (41 light peptides and 41 heavy peptides) corresponding to 29 target proteins were quantified in a LC−MRM/MS run (Supplemental Table 2).

Mass Spectrometric Data Processing and Bioinformatic Analysis

The raw data acquired by LTQ-Orbitrap were processed with the MaxQuant software version 1.1.1.25 according to the standard workflow.12,13 Database searches were performed in MaxQuant with the Andromeda search engine14 against the Swiss-Prot database (2010 version containing 517 100 sequences and 20 306 proteins in the forward database, including common contaminants), with an initial precursor mass tolerance of 7 ppm and fragment mass deviation of 0.5 Da. The identification of light- and heavy-labeled, dimethylated peptides included carbamidomethylation (cysteine) as a fixed modification and oxidation (methionine) as a variable modification. Two missed cleavages were allowed for trypsin digestion. The “identify” module in MaxQuant was used to filter identifications at a 1% false discovery rate (FDR) at the peptide and protein level using a reverse database, in which lysines and arginines were swapped with the preceding amino acid.15 Only peptides with a minimum length of six amino acids were considered for identification. Protein identification and quantitation results obtained from the 22 fractions generated by online SCX-RPLC-MS/MS were integrated using MaxQuant. For dimethyl-labeled peptide analysis, two ratio counts were set as a minimum requirement for quantification. The lists of identified proteins were filtered to eliminate reverse hits and 248 known contaminants. The identified proteins were further analyzed using the proteomics data mining and management software, ProteinCenter (version 3.7; Proxeon Bioinformatics, Odense, Denmark) to compare the identified urinary microparticle proteome with public-domain data sets, including 1139 genes of the urinary exosome from the Exocarta Database (version 2.1; downloaded from the Web site http://exocarta.

MRM−MS Data Analysis

Signature dimethylated peptides were determined using MRMPilot software (version 2.1; AB Sciex) according to the following principles: (1) +2 or +3 charge states; (2) peptide length within 7−20 ammonium acids; (3) no Met or Cys in the peptide sequence; (4) no missed cleavage sites; (5) dimethylated (K) and dimethylated (N-term) for light labeling; dimethylated-D413C2(K) and dimethylated-D413C2(K) (Nterm) for heavy labeling, and (6) both Q3 and Q1 ions were smaller than 1000 Da. The Q1/Q3 transition was further confirmed by performing protein database searches using MS/ MS spectra triggered by qualified Q1/Q3 transitions. The 5614



Article

Figure 1. Typical characteristics of microparticles enriched from human urine samples. (A) EM observation of microparticles purified from a urine sample. Scale bar = 100 nm. (B) Size distribution of urinary microparticles. A total of 42 microparticles observed in EM micrographs were analyzed. (C) Western blot analysis of exosomal markers, TSG101 and CG9, in microparticle and soluble urinary protein fractions (5 μg of protein) from a healthy volunteer and a bladder cancer patient. (D) Western blot analysis of TSG101 and CD9 in the exosomal fraction (5 μg of protein) of urine samples from a healthy volunteer, a hernia patient, and a bladder cancer patient. (E) Extracted microparticles coupled to latex beads were analyzed by flow cytometry. 95.6% of microparticles showed positive expression of the exosome marker, CD9,, which revealed the highly enrichment of urinary exosomes.

sequences of MS/MS spectra and retention times were confirmed using the MASCOT engine (version 2.2.04; Matrix Science, London, U.K.). For data acquisition, scheduled MRM was used to reduce cycle times and generate more points per peak. Tryptic peptides of all individual samples were labeled with a light dimethylated tag. The sample pooled from individual clinical samples was labeled with a heavy dimethylated tag and then spiked into all individual samples as an internal standard for comparison in the verification experiment by MRM-MS. All MRM data were processed using MultiQuant 1.1 (Applied Biosystems) using the MQL algorithm for peak integration. Automatic peak detection was used with a 5-min retention time window, with “report largest peak” enabled, and a 2-point Savitsky-Golay smooth with a peak-splitting factor of 4. The default MultiQuant values for noise percentage (40%) and baseline subtraction window (2 min) were used. Retention

times were determined empirically by analyzing a sample with longer MRM cycle times, which gives lower quality peaks, but allows retention time determination. The peak area of an endogenous light-labeled peptide signal was divided by the peak area of the corresponding heavy-labeled peptide signal to generate the peak-area ratio for statistical analysis. One LC− MRM/MS run was performed for each clinical sample. All integrated peaks were inspected manually to ensure correct peak detection and accurate integration. Three transitions per peptide were monitored by LC−MRM/MS. Reported calculated peak-area ratios are derived from the quantifier MRM transition, with two other qualifier transitions acting to verify retention times and reveal any signal interference. Qualified ion pairs were used to detect the presence of interference by ensuring that the relative signal intensity of ion pairs was consistent between the heavy and endogenous forms of all peptides. For specimens without detectable endogenous 5615



Article

Figure 2. Six procedures (A−F) used for optimization of tryptic digestion of urinary microparticle proteins. (A) Lanes 1 and 2 show microparticle protein profiles from a healthy volunteer prior to digestion. The tryptic products were also monitored by SDS-PAGE after digestion in lysis buffer (lane 3) and without buffer (lane 4). (B and C) Among the six procedures tested, procedure E generally showed the fewest bands in the high-mass range when tested using several clinical samples. Lane MP shows microparticle protein profiles prior to digestion.

peptide peak areas, the concentration values were expressed as “ND”. To perform statistical analyses, ND specimens were assigned values of zero. The determined concentration differences of target proteins/peptides are expressed in fold differences, assuming complete tryptic digestion and 100% peptide recovery.

diluted 3000-fold in PBS containing 1% BSA) was added and incubated for 40 min at room temperature. The substrate 4methylumbelliferyl phosphate (Molecular Probes, Eugene, OR) was diluted to 100 μM with an mixture alkaline phosphatase buffer (alkaline phosphatase buffer/PBS = 1:2), and 100 μL was added to each well. The fluorescence was measured with a SpectraMax M5 microplate reader (Molecular Devices, Sunnyvale, CA) with excitation and emission wavelength set at 355 and 460 nm, respectively.

Sandwich ELISA of TACSTD2

White polystyrene 96-well microtiter plates (Corning Corp., Corning, NY) were coated with goat anti-TROP2 (AF650, R&D) antibodies by incubation at 4000 ng/mL in PBS (50 μL in each well) for 2 h at room temperature. After washing, the plates were blocked by the addition of 200 μL per well of 1% BSA (Sigma)/PBS and incubated overnight at 4 °C. Fifty microliters of raw urine protein diluted 1:2 in blocking buffer from 81 hernia patients, 40 LgEs patients, 63 HgEs patients, and 37 HgAs patients was added and incubated for 1 h at room temperature. Recombinant TACSTD2 protein (650-T2, R&D) was used as a standard. Subsequently, the biotinylated antihuamn TACSTD2 (BAF650, R&D) antibody (1:50 dilution in PBS containing 1% BSA) was then applied and incubated for an additional 1 h at room temperature. The streptavidin−alkaline phosphatase (RPN1234, Amersham Bioscience, U.K.) (50 μL,

Statistical Analysis for Verification of Individual Biomarkers

The statistical package SPSS 13.0 (SPSS, Inc., Chicago, IL) was used for all analyses. Differences in concentrations of targeted urinary proteins (measured by LC-orbitrap-MS/MS or LC− MRM/MS) according to different clinical groups were analyzed using the nonparametric Mann−Whitney test. Receiver Operator Characteristic (ROC) curve and Area-Under-theCurve (AUC) analyses were applied to determine the optimal cutoff point that yielded the highest total accuracy with respect to discriminating different clinical classifications. The optimal cutoff point was determined using Youden’s index (J), calculated as J = 1 − (false positive rate + false negative rate) 5616



Article

= 1 − [(1 − sensitivity) + (1 − specificity)] = sensitivity + specificity − 1.20

■

using clinical samples, we evaluated the reproducibility of the sample preparation method used to obtain the urinary microparticle proteome using a model urine sample from a healthy individual. The urine sample was separated into three parts and processed as shown in Figure 3 to evaluate variations

RESULTS

Enrichment and Characterization of Urinary Microparticles by Ultracentrifugation

To confirm effective enrichment of the microparticle by centrifugation at 100 000g, we examined the pellet obtained from a urine sample by ultracentrifugation using EM, which showed numerous membrane vesicles 30−100 nm in diameter, a typical characteristic of urinary exosomes (Figure 1A,B). Proteins extracted from microparticle pellets were further analyzed by Western blotting to establish the presence of the known exosomal markers, TSG101 and CD9.11,16,21,22 As shown in Figure 1C, the exosomal markers TSG101 and CD9 were strongly enriched in the microparticle pellets compared with the supernatant fractions (soluble urine proteins), which did not show detectable levels of TSG101 or CD9. The results indicated that exosomal proteins were enriched in the extracted microparticles. TSG101 and CD9 levels showed no significant quantitative correlation with each other in urinary microparticles among different individuals (Figure 1D). The extracted microparticles coupled to latex beads were analyzed by flow cytometry, and more than 95% microparticles showed positive expression of exosomal marker, CD9, which revealed the high enrichment rate of urinary exosomes (Figure 1E). Taken together, the data of Figure 1 indicate that urinary exosomes were highly enriched in the extracted urinary microparticles by ultracentrifugation. Optimization of Trypsin Digestion Protocols for Urinary Exosome Proteins

Microparticles are membrane vesicles of endocytic origin secreted by different cell types. To test if the addition of lysis buffer could be helpful in improving trypsin digestion efficiency, we used SDS-PAGE to compare the digested products of microparticle proteins with and without addition of lysis buffer (Figure 2A, lanes 3 and 4). Fewer protein bands were observed in samples in which lysis buffer was used during trypsin digestion, indicating that the lysis buffer was helpful in breaking the microparticle membrane and improving the digestion efficiency of trypsin. The optimal digestion protocol has been shown to be analyte-dependent; thus, a single digestion protocol may not achieve optimal digestion of all urinary proteins.23 We then further optimized the tryptic digestion of urine microparticle proteins using six digestion procedures (A− F), summarized in Supplemental Figure 1, and evaluated each protocol’s efficacy by SDS-PAGE (Figure 2B,C). Digestions were performed without the addition of lysis buffer in procedures A-D and with lysis buffer in procedures E and F. Procedure E (one digestion with an enzyme/substrate ratio of 1/25 in the presence of lysis buffer) was generally found to produce the fewest protein bands in the whole-mass range. Therefore, procedure E was selected as the digestion protocol for all urine specimens used for biomarker discovery and verification by LC−MRM/MS analyses in this study.

Figure 3. Schematic representation of the experimental design used to evaluate the reproducibility of the urinary microparticle extraction and quantitative proteome analysis platform. Variations at the digestion/ labeling and whole-workflow levels were evaluated using peak-area ratios of heavy-labeled peptides to light-labeled peptides in mixture 1 and mixture 2, respectively.

at the digestion/labeling level (by measuring the heavy/light ratio in mixture 1) and at the whole-workflow level (by measuring the heavy/light ratio in mixture 2). Table 1 summarizes protein quantification based on heavy/ light peptide ratios in mixtures 1 and 2. Details of protein/ peptide identification and quantitation for this model sample are presented in Supplemental Tables 3A and 3B; a total of 242 and 247 proteins were quantified based on two distinct peptides in mixture 1 and mixture 2, respectively. The quantified data showed that the average ratios for the processed replicates were 0.99 ± 0.11 (coefficient of variation [CV] = 11%) and 0.97 ± 0.15 (CV = 15%) at digestion/labeling and whole-workflow levels, respectively. The variability in the quantitation platform used in this study were thus considered

Assessment of the Reproducibility of Sample Preparation for Urinary Microparticle Proteomics

The proteomic platforms used in this study included differential dimethylation labeling coupled with LC-based fractionation. Enrichment of urinary microparticle proteins is a multistep process. Before performing comparative proteomic analyses 5617



Article

workflow provides a solid basis from which to pursue the goal of urinary microparticle protein quantification by isotopic dimethylated labeling coupled with LC/MS/MS.

Table 1. Summary of Protein Quantitation and Experimental Variation at Digestion/Labeling Level (Mixture 1) and Whole-Workflow Level (Mixture 2)

Digestion/labeling level (mixture 1) Whole workflow level (mixture 2)

no. of proteins quantified

average ± SD

CV (%)

Quantitative Analysis of the Urinary Microparticle Proteome for Bladder Cancer Biomarker Discovery

242

0.99 ± 0.11

11%

247

0.97 ± 0.15

15%

It has been estimated that exosomes/microparticles contain only ∼3% of total urinary proteins,4 creating difficulties in obtaining the amounts of microparticle proteins needed for multidimensional fractionation from limited volumes of clinical samples. To gain greater insight into the quantitative differences in microparticle protein profiles between hernia and bladder cancer patients, we used two types of sample preparationpooled microparticle samples and individual samples without poolingfor discovery-phase experiments in

within reasonable ranges (both CVs were ≤15%). These results clearly indicate the good performance and reproducibility of the entire workflow, including the process of urinary microparticle enrichment, proteolytic digestion, dimethylation labeling, LC/ MS/MS analysis, and quantitative data analysis. The developed

Figure 4. Schematic representation of the discovery phase experimental design. (A) Preparation of a global internal standard from 18 individual samples. (B) Two types of sample preparations (pooled microparticle samples and individual samples without pooling) were used in the biomarker discovery pipeline described here. The detection of minor proteins in urinary microparticle was improved by preparing two pooled samples from nine hernia and nine bladder cancer patients and generating a 1:1 mixture of the two containing a total of 50 μg of isotopic dimethylated peptides for online 2D-SCX-RP-nanoLC−CID−MS/MS analysis. Person-to-person variations in the urinary microparticle proteome were assessed by employing a strategy that measures the concentration differences of urinary microparticle proteins in 18 individual samples relative to a pooled sample, used as an internal standard. 5618



Article

Figure 5. Summary of protein identification and quantitation results in the discovery phase and verification phase. A total of 107 proteins were selected as biomarker candidates during the discovery phase; 29 were detectable by LC−MRM/MS and 22 were verified as differentially expressed in urine microparticles of bladder cancer patients. Finally, urinary levels of protein TACSTD2 were measured by ELISA in 221 individual samples.

were quantified with at least two dimethylated peptides. Details of the protein/peptide identification and quantitation results for the pooled samples are shown in Supplemental Table 4. We first used ProteinCenter software to combine two published plasma data setsthe Sys-Bodyfluid database containing 8336 plasma proteins,18 and the HUPO Web site containing 9504 plasma proteins19yielding a reference plasma proteome containing a total of 15 873 plasma proteins. To analyze our microparticle proteome appropriately for biomarker discovery, we first divided the 2756 quantified microparticle proteins into two groups by comparing them with the 15 873 previously identified plasma proteins. In the first group were 1071 microparticle proteins that had been identified in previous studies of plasma proteomes; these were classified as plasma-associated microparticle proteins. The second part, comprising 1684 exosome proteins that had not been identified in plasma proteomes, were categorized as plasma-independent microparticle proteins. We then calculated the concentration ratios of microparticle proteins between the pooled bladder

the biomarker discovery pipeline of this work (Figure 4A,B). Protein identification and quantitation results are summarized in Figure 5. Pooled Comparison: Comparative Proteomics of a Pooled Bladder Cancer Sample and a Pooled Hernia Sample

To improve the detection of minor proteins in urinary microparticle, we used two pooled samples from nine hernia and nine bladder cancer patients to generate a 1:1 mixture containing a total of 50 μg of isotopic dimethylated peptides for online 2D-SCX-RP nanoLC−CID−MS/MS analysis. A total of 3760 nonredundant proteins with at least one distinct tryptic peptide were identified (FDR < 0.01 at protein and peptide levels) by combining the raw data from 22 fractions. To ensure the correctness of protein identification, we used 2945 identified proteins with at least two distinct tryptic peptides for subsequent biomarker discovery efforts on the premise that most assignments of proteins using more than a single peptide are likely correct.24 Among the 2945 identified proteins, 2756 5619



Article

that 624 proteins (97% of the individual data set) were identified in both data sets, and an additional 2321 proteins were identified only by pooled comparisons (Figure 6). Thus, a total of 2964 urinary microparticle proteins were identified (based on two or more peptides) in this study.

cancer sample and the hernia sample. To select candidate biomarkers, we set cutoff values of ratios (log values) at means ± two standard deviations, calculated separately for the plasmaassociated microparticle proteome and plasma-independent microparticle proteome. Proteins with a fold-change more than two standard deviations greater than the mean were defined as increased proteins, and proteins with fold-changes less than two standard deviations below the mean were defined as decreased proteins. The detailed distributions of the concentration ratios (bladder cancer to hernia) of microparticle proteins quantified in the two microparticle proteome subgroups are summarized in Supplemental Figure 2. In addition, several considerations were also included to filter out increased or decreased proteins as appropriate candidate biomarkers. First, histone proteins, which are considered markers of apoptotic blebs, were excluded.25 It has been reported that apoptotic blebs are similar to exosomes in size and flotation density, making it difficult to separate them using differential centrifugation.25 The source of urinary histone proteins in microparticle fractions is not clear. Second, three types of proteins, namely immunoglobulin, histone and complement proteins, were excluded from biomarker selection owing to the highly similar protein sequences of multiple isoforms, which might cause problems in clinical assays. After filtering, 50 increased proteins and eight decreased proteins (Supplemental Table 5A) were selected as candidate biomarkers from among the 1684 plasma-independent microparticle proteins, and 44 increased proteins and one decreased protein (Supplemental Table 5B) were considered as candidate biomarkers from among the 1071 plasma-associated microparticle proteins. Thus, a total of 103 proteins were selected as candidate biomarkers for further verification.

Figure 6. Comparison of MS-identified proteins in the two pooled samples (pooled comparison) and 18 individual samples (individual comparison). Combining the two data sets yielded a total of 2964 identified proteins.

Including immunoglobulin, histone, and complement proteins, a total of 168 proteins (Supplemental Table 5D) showed significant differences in concentration between bladder cancer and hernia urinary microparticle samples. To simplify subsequent biomarker-verification efforts, we excluded immunoglobulin, histone, and complement proteins as candidate biomarkers. After combining the 103 candidate proteins from pooled comparisons and 11 candidate proteins from individual comparison and removing redundancies, a total of 107 proteins (Supplemental Table 5E) were selected for the following biomarker validation by LC−MRM/MS.

Individual Comparison: Comparative Proteomics of 18 Individual Clinical Samples with a Pooled Internal Standard

Functional-Prediction Analysis of the Total Urinary Microparticle Proteome

Supplemental Figure 3 shows the Molecular Function, Cellular Component, and Biological Process distributions of total identified urinary microparticle proteins and differentially expressed proteins in the urinary microparticle proteome of hernia and bladder cancer. The top three Molecular Functions in both total microparticle proteome and differential proteome data sets were protein binding, catalytic activity, and metal ion binding. The top three Cellular Components among the 2964 total proteins were cytoplasm, membrane, and nucleus, whereas the top three Cellular Components among the 168 differentially expressed proteins (Supplemental Table 5D) were extracellular, membrane, and cytoplasm. The top three Biological Processes among the 2964 total proteins were metabolic process, regulation of biological process, and response to stimulus process, whereas the top three Biological Processes among the 168 differentially expressed proteins were metabolic process, response to stimulus process, and regulation of biological process. This analysis revealed no obvious differences in the distributions of global Molecular Functions between total identified proteins and differentially expressed proteins. However, it is interesting to note that the percentages of the Cellular Components, extracellular and membrane, increased significantly among the 168 differentially expressed proteins. Additionally, the percentages of the Biological Process, stimulus process, also increased significantly among the 168 differentially expressed proteins. The biological implications of the changes in Cellular Component and Biological Process distributions in bladder cancer-associated

To assess person-to-person variation in the urinary microparticle proteome, we applied a strategy to quantitatively analyze the urinary microparticle proteomes of 18 individual samples by comparing them to an internal standard (Figure. 4B). As illustrated in Figure 4B, the internal standard was prepared by pooling equal amounts of proteins from each of the 18 individual samples. The heavy-labeled peptides present in internal standard were used as a basis for global comparisons among all clinical samples. Using internal standard peptides as normalization factors, we could determine differences in the concentration of each peptide among individual samples. Using this strategy, we identified a total of 643 nonredundant proteins that were based on at least two distinct tryptic peptides (FDR < 0.01 at both protein and peptide levels). Of these, 616 proteins were quantified with at least two dimethylated peptides. Details of the protein quantitation results for the 18 individual samples are shown in Supplemental Table 6. Eleven proteinsnine that increased and two that decreasedshowed significant differences (p < 0.05, n = 18) between bladder cancer and hernia groups by nonparametric Mann−Whitney tests (Supplemental Table 5C), and were included as candidate biomarkers for further verification. Comparison of Pooled and Individual Data Sets of Urinary Microparticle Proteome

After the discovery phase, the two data sets from pooled comparisons and individual comparisons were compared and integrated using ProteinCenter software. This analysis showed 5620



Article

proteins in urinary microparticles are not clear and warrant further investigation. Biomarker Verification Using LC−MRM/MS

According to the results obtained from our workflow, the average concentration of urinary microparticle proteins was sample-dependent and ranged from 0.01 to 3 μg/mL. The small amounts (0.1−150 μg) of extractable microparticle proteins from limited volumes (10−50 mL) of clinical urine samples made verification of multiple biomarkers by Western blot analysis or other antibody-based immunoassays problematic. By contrast, only 1−2 μg of protein is required to quantify tens of candidate proteins by LC−MRM/MS analysis.26,27 The precision and reproducibility of MRM-based protein quantification in 1−2 μg human plasma proteins using stable isotope labeled synthetic peptides as internal standards have been demonstrated.37,38 Therefore, multiplexed MRM−MS assays offer an attractive alternative for the development of a useful panel of biomarkers. However, the costs associated with stable isotope-labeled synthetic peptides may limit the number of candidate proteins that can be verified. Therefore, we explored the possibility of using LC−MRM/MS to detect isotopiclabeled dimethylated peptides for multiple biomarker verification in urine samples of 28 bladder cancer, 12 hernia, and eight UTI/HU patients. Similar to the strategy used in individual comparisons, described above, we spiked each sample (containing light dimethyl-labeled peptides) with an internal standard sample pooled from these 48 samples (containing heavy dimethyl-labeled peptides). Collision energy was optimized to attain the high-intensity MRM signals necessary to guarantee adequate sensitivity. The mixtures were then quantified by LC−MRM/MS. Among the 107 candidate biomarker targets, 41 peptide sequences belonging to 29 proteins were detectable in a 70-min LC−MRM/MS run using a mere 0.25 μg of both microparticle proteins and internal standard peptides. Figure 7A shows the extracted ion chromatograms of MRM ion pairs in a LC− MRM/MS run of 41 urinary peptides, demonstrating effective quantitation of these proteins in a urine sample from a bladder cancer patient. Supplemental Table 2 lists the detailed sequences, Q1/Q3 transitions, and collision energies used in this study. Signals of 29 proteins in each clinical sample were then measured by LC−MRM/MS. The distribution of the average concentrations of 29 urine proteins (41 peptides) in 48 specimens from different disease groups is summarized in Supplemental Figure 4. The technical reproducibility of MRM−MS for relative quantitation of 41 urinary microparticle peptides was evaluated by analyzing three repeated injections of a selected bladder cancer sample in the Qtrap5500 mass spectrometer (Supplemental Figure 5); for 31 peptides (76%), CVs were less than 20%, and 37 peptides (90%) showed CVs less than 30%. This indicates that MRM−MS is capable of providing a reproducible platform for the relative quantitation of isotopic dimethyl-labeled peptides. The selection of signature peptides is a critical issue for protein quantitation by MRM-MS. Proteins were quantified accurately using unique peptides, that is, peptides with sequences specific for the target protein. However, for some proteins in this study without any unique, MRM−MSdetectable peptides, nonunique detectable peptides were used for quantitation as reference data. If the results of protein quantification by LC−MRM/MS are meaningful using nonunique peptides, the data should be further confirmed

Figure 7. Biomarker verification using LC−MRM/MS. (A) The extracted ion chromatograms of MRM ion pairs in a 70-min LC− MRM/MS run of 41 signature peptides corresponding to 29 proteins demonstrating quantitation of these proteins in a urine sample from a bladder cancer patient. A total of 0.5 μg microparticle protein was injected into the column. The arrows indicate the light and heavy peaks of three candidate biomarkers, HBA, HBB, and TACSTD2. (B− D) Verification of HBA, HBB, and TACSTD2 in 48 individual urine samples (12 hernia, 28 bladder cancer, and eight UTI/HU controls) by LC−MRM/MS analysis. The average concentration ratios and pvalues are indicated above each dot plot. The top p-value was calculated by comparing BC and UTI/HU subgroups; the left p-value was calculated by comparing BC and hernia subgroups, and the right p-value was calculated by comparing hernia and UTI/HU subgroups. All three proteins showed elevated expression levels in bladder cancer 5621



Article

0.001, r = 0.813−0.988), indicating a high confidence in the quantitative trends of two peptides of a protein. Although the correlation was high for all peptides examined, the concentration ratios of two peptides from one target protein were not exactly identical, possibly reflecting differences in digestion and recovery rates for the two peptides in urine microparticle samples. Moreover, if one peptide is not unique for the target protein, tryptic peptides from multiple source proteins may cause errors in protein quantification. In this situation, the quantitation results from the unique peptide should be adopted preferentially.

Figure 7. continued urinary microparticles (p < 0.05) compared to both hernia and UTI/ HU (n = 48). Analysis of the ROC curves for HBA, HBB, and TACSTD2 proteins, determined from the dot plots, yielded AUC values of 0.795, 0.75, and 0.735, respectively, for differentiating the bladder cancer group (n = 28) from the control group (hernia, n = 12).

with other approaches, such as ELISA. In this study, 12 proteins (A2M, APOA1, CA1, FGA, FGB, FGG, HBA1, HBB, SERPIND1, HP, RAB27B and TACSTD2) were quantified using two proteotypic peptides, and seven out of the 12 proteins (APOA1, CA1, FGB, HBB, SERPIND1, HP and TACSTD2) were quantified using two unique peptides. The quantitative results obtained with the two proteotypic peptides are summarized in Supplemental Table 7. The results obtained using two dimethylated peptides to quantitate 12 proteins in 48 urinary microparticle samples were highly correlated (p

Comparative and Targeted Proteomic Analyses of ... - ACS Publications

Recommend Documents