Article pubs.acs.org/jpr
Glycoproteomics Approach for Identifying Glycobiomarker Candidate Molecules for Tissue Type Classification of Non-small Cell Lung Carcinoma Yoshitoshi Hirao,† Hideki Matsuzaki,† Jun Iwaki,† Atsushi Kuno,† Hiroyuki Kaji,† Takashi Ohkura,† Akira Togayachi,† Minako Abe,† Masaharu Nomura,‡ Masayuki Noguchi,§ Yuzuru Ikehara,† and Hisashi Narimatsu*,† †
Research Center for Medical Glycoscience (RCMG), National Institute of Advanced Industrial Science and Technology (AIST), Central 2, 1-1-1 Umezono, Tsukuba, Ibaraki 305-8568, Japan ‡ Department of Surgery I, Tokyo Medical University, 6-7-1 Nishi-shinjuku, Shinjuku, Tokyo 160-0023, Japan § Department of Pathology, Institute of Basic Medical Science, Graduated School of Comprehensive Human Science, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8575, Japan S Supporting Information *
ABSTRACT: Histopathological classification of lung cancer has important implications in the application of clinical practice guidelines and the prediction of patient prognosis. Thus, we focused on discovering glycobiomarker candidates to classify the types of lung cancer tissue. First, we performed lectin microarray analysis of lung cancer tissue specimens and cell lines and identified Aleuria aurantia lectin (AAL), Hippeastrum hybrid lectin (HHL), and Concanavalia ensiformis agglutinin (ConA) as lectin probes specific to nonsmall cell lung carcinoma (NSCLC). LC−MS-based analysis was performed for the comprehensive identification of glycoproteins and N-linked glycosylation sites using lectin affinity capture of NSCLC-specific glycoforms of glycoproteins. This analysis identified 1092 AAL-bound glycoproteins (316 gene symbols) and 948 HHL/ConA-bound glycoproteins (279 gene symbols). The lectin microarray-assisted verification using 15 lung cancer cell lines revealed the NSCLC-specific expression of fibronectin. The glycosylation profiling of fibronectin indicated that the peanut agglutinin (PNA) signal appeared to differentiate two NSCLC types, adenocarcinoma and large cell carcinoma, whereas the protein expression level was similar between these types. Our glycoproteomics approach together with the concurrent use of an antibody and lectin is applicable to the quantitative and qualitative monitoring of variations in glycosylation of fibronectin specific to certain types of lung cancer tissue. KEYWORDS: Biomarker, glycoproteomics, glycosylation, lectin microarray, lung cancer, tissue
■
INTRODUCTION Lung cancer is one of the leading causes of cancer death, and its prevalence keeps increasing worldwide.1 Currently, lung cancer is diagnosed on the basis of histopathological findings and is classified into small cell lung carcinoma (SCLC) and non-small cell lung carcinoma (NSCLC). This classification is an important determinant for clinical management and prognosis of the disease because SCLC generally responds well to chemotherapy and radiation therapy, whereas NSCLC is treated with surgery. Therefore, an ideal diagnostic test should discriminate unerringly between SCLC and NSCLC.2,3 However, biological features of diseases, such as progression, metastatic susceptibility, sensitivity to therapeutics and radiation therapy, and prognosis, cannot be predicted fully from the histopathological observations obtained from cytological examination or pleural biopsy. To support or compensate for the morphological changes associated with © XXXX American Chemical Society
disease pathology, it is required that molecular diagnostic tools using concentrated liquid specimens of lung cancer such as pleural fluids, expectorations, and cellular tissues obtained by bronchoscopy be developed. At present, the neuroendocrine marker neural cell adhesion molecule (NCAM) has been used to reliably distinguish SCLC immunohistochemically. On the other hand, sialyl Lewis X is a good tumor marker for NSCLC.4,5 The expression of sialyl Lewis X is attributed to structural changes of the glycan epitopes on carrier proteins, whereas NCAM shows quantitative changes as a biomarker index. Special Issue: Proteomics of Human Diseases: Pathogenesis, Diagnosis, Prognosis, and Treatment Received: June 30, 2014
A
dx.doi.org/10.1021/pr5006668 | J. Proteome Res. XXXX, XXX, XXX−XXX
Journal of Proteome Research
Article
Lu134A, Lu135, and Lu139) and human NSCLC cell lines (lung adenocarcinoma: A549, HAL8, HAL24, NCI-H1355, NCI-H1975, NCI-H358, and LX1; large cell lung carcinoma: NCI-H1155, NCI-H1299, NCI-H2126, NCI-H460, and NCIH661) were used in this study. The cell lines were obtained from RIKEN Bioresource Center (Ibaraki, Japan; A549, Lu134A, Lu135, Lu139, HAL8, HAL24, and LX1) and American Type Culture Collection (Rockville, MD; NCIH524, NCI-H526, NCI-H2171, NCI-H1355, NCI-H1975, NCI-H358, NCI-H460, NCI-H661, NCI-H1155, NCI-H1299, and NCI-H2126). The cells were cultured in a growth medium comprising 10% fetal bovine serum (FBS) in Dulbecco’s modified Eagle’s medium supplemented with 4 mM Lglutamine, 4.5 g/L glucose, 1000 U/mL penicillin, and 1 mg/ mL streptomycin (Gibco, Grand Island, NY) and were incubated at 37 °C in a humidified atmosphere of 5% CO2 and 95% air. After the cells grew to confluence, the cells were incubated in FBS- and antibiotics-free medium for 24 h under the same conditions. The FBS-free conditioned cultured medium was collected as a protein source, concentrated to 10 times, and stored at −30 °C until use.
Over the past few years, new approaches adopting omicsbased techniques have been developed for biomarker identification to distinguish clinical classification of lung cancer.6−9 In particular, the comprehensive analysis of glycosylated proteins has been adopted instead of genomics or proteomics for biomarker discovery. Glycomics and glycoproteomics were introduced to cancer research10−13 because cancer cells are known to express aberrant glycosylation patterns such as increased branching, sialylation or fucosylation of N-linked glycans, and increased sialylation or structural truncation of O-linked glycans.11,14−17 Because glycan structures vary depending on the tissue type and disease progression, the analysis of aberrant glycoproteins derived from lung cancer cells should be a promising approach for discovering clinically useful biomarkers. For instance, a glycoproteomics approach using lectin (Lens culinaris lectin, LCA) capture, differential stable isotope-labeling of Trpcontaining peptides, and mass spectrometry (MS)-based relative quantitation and identification revealed that fucosylated haptoglobin exhibits a specific glycoform during lung cancer progression.6,18 The application of lectin affinity chromatography and MS-based glycoproteomics approaches has also been suggested for use in identifying increased fucosylation in NSCLC.19 Considering these concepts, we applied unique technologies that we developed previously to analyze glyco-alterations and then used these in a multilateral strategy designed for highthroughput discovery of cancer glycobiomarkers.20−23 In this study, we developed an integrated glycoproteomics approach using the combination of a lectin microarray and lectin capture followed by isotope-coded glycosylation site-specific tagging (IGOT)-liquid chromatography (LC)−MS. Using this strategy, we identified glycoproteins that show specific glyco-epitopes in NSCLC cell lines with lectin microarray, and we further characterized the lectin-binding profiles specific to two NSCLC tissue types: adenocarcinoma and large cell carcinoma. Although the histopathological classification of lung cancer has important implications in clinical practice, there is at present no reliable and convenient diagnostic method for the classification of NSCLC tissue type. The aim of this study was to prove the concept that the analysis of glycosylation patterns of a glycoprotein biomarker could distinguish NSCLC from SCLC. Here, we applied our glycoproteomics-based strategy for the discovery of feasible glycobiomarkers of NSCLC. The differential glycan profiling identified a lung adenocarcinomaspecific glycan structure of a glycobiomarker candidate that could be differentiated using a specific lectin. The combination of a rationally selected marker molecule and specific lectin probe can detect alterations in the glycan structure that reflect the progression of certain types of cancer.
■
Differential Glycan Analysis of Tissue Specimens Using Lectin Microarray
To select the lectins specific to lung cancer tissue types, crude glycoprotein solutions from tissue specimens were profiled using the lectin microarray as described previously.24,25 Each of the formalin-fixed paraffin-embedded tissue samples was deparaffinized and scratched off from the glass slide using a 21 G needle under a microscope. The scratched tissues were collected into a 1.5 mL microtube containing 200 μL of 10 mM citrate buffer (pH 6.0) and incubated for 1 h at 95 °C for antigen retrieval. The tissue pellets were washed with phosphate-buffered saline (PBS) by centrifugation, solubilized with 20 μL of PBS containing 0.5% Nonidet P40, and sonicated gently. The tissue suspension was incubated on ice for 1 h and centrifuged at 20 000g for 5 min at 4 °C. The obtained supernatants were used as detergent-solubilized glycoprotein extracts. Each 20 μL aliquot of the above glycoprotein solutions was incubated with 10 μg of Cy3-succinimidyl ester (Amersham Bioscience, Tokyo, Japan) for 1 h at room temperature in the dark. The obtained reaction product was loaded onto a Sephadex-G25 spin column to remove the excess fluorescent reagent. After centrifugation, the collected solution was adjusted to 200 μL with a probing buffer (TBS containing 1% Triton X-100 (TBSTx), and 500 mM glycine) and incubated for 2 h at room temperature to inactivate the residual fluorescent reagent completely. To each well on the glass slide, 60 μL of the Cy3-labeled glycoprotein solution was applied, and the section was incubated overnight at 20 °C in a humid chamber. After the binding reaction, fluorescent images of the array were acquired using a GlycoStation Reader 1200 evanescence scanner (GlycoTechnica Ltd., Sapporo, Japan). All data were analyzed with the Array Pro analyzer, version 4.5 (Media Cybernetics, Bethesda, MD). The relative intensities of positive lectins were calculated from their ratio to the fluorescent intensity of an internal-standard lectin, Urtica dioica agglutinin (UDA).
EXPERIMENTAL SECTION
Human-Derived Tissue Samples and Lung Cancer Cell Lines
The clinical specimens of lung cancer tissues and normal regions were collected upon resection of lung adenocarcinoma from 34 patients at the Tokyo Medical University. The institutional ethics committees of Tokyo Medical University and National Institute of Advanced Industrial Science and Technology approved this study, and informed consent for the use of clinical specimens was obtained from all participants. Human SCLC cell lines (NCI-H2171, NCI-H524, NCI-H526,
Differential Glycan Analysis of Cell Culture Supernatants Using Lectin Microarray
An aliquot of the cell culture supernatants was filtered through a Millex-HV 0.45 μm filter device (Millipore, Bedford, MA) and then concentrated using an Amicon Ultra-15 filter unit (10 K B
dx.doi.org/10.1021/pr5006668 | J. Proteome Res. XXXX, XXX, XXX−XXX
Journal of Proteome Research
Article
by hydrophilic interaction chromatography (HILIC) on Sepharose CL-4B (GE Healthcare, Waukesha, WI).
cutoff; Millipore) at 2000g for 10 min at room temperature. The protein concentration was measured with a Pierce BCA protein assay kit (Pierce, Rockford, IL), and a portion of the concentrate (200 ng of total protein) was labeled with Cy3 and then subjected to lectin microarray as described above. The relative signal intensities were calculated by mean normalization, with the mean set to 1.0.
Enzymatic Stable Isotope Labeling of Glycopeptides
Purified glycopeptides were treated with PNGase (2.5 mU, 37 °C overnight) in stable isotope 18O-labeled water to remove the glycan moiety and concomitantly label the glycosylated Asn of glycopeptides with the isotope, as described previously.23 This step is called IGOT.
Lectin Affinity Capture of Glycoproteins with Specific Glyco-Epitopes
Identification of the Labeled Peptides by Nanoflow LC−MS Analysis
To verify the lectins selected by the lectin microarray analysis, cell culture supernatants of lung cancer cell lines were subjected to lectin affinity capturing with Aleuria aurantia lectin (AAL), Hippeastrum hybrid lectin (HHL), and Concanavalia ensiformis agglutinin (ConA). Biotinylated AAL, HHL, and ConA were purchased from J-Oil Mills, Inc. (Tokyo, Japan) or Vector Laboratories (Burlingame, CA). Streptavidin-immobilized magnetic beads (Invitrogen, Carlsbad, CA) were washed three times in PBSTx, and the biotinylated lectins were mixed with the magnetic beads and incubated with gentle mixing at 1400 rpm for 2 h at 4 °C. Subsequently, the magnetic beads were recovered using a magnetic stand and washed three times in PBSTx. A portion of the cell culture supernatant (1 μg of total protein) was added to the magnetic beads and incubated with gentle mixing for 16 h at 4 °C. After incubation, the magnetic beads were eluted by PBS containing 0.2% sodium dodecyl sulfate (SDS) for 10 min at 98 °C. The samples captured by the three lectins were then subjected to SDS-polyacrylamide gel electrophoresis (PAGE) on a 10% polyacrylamide gel, and the gel was silver-stained (ATTO Corporation, Tokyo, Japan).
Stable isotope-labeled peptides were analyzed by the LC−MS method, as described previously.26 Briefly, the peptide mixture was injected into a C18 trap column (0.5 mm × 1 mm). After washing, the column was connected to a nanoflow LC system (flow rate: 100 nL/min), and the peptides were separated on a reverse-phase (C18) tip column (150 μm × 70 mm) by a linear gradient of MeCN (0−35% in 0.1% formic acid) for 70 min. The eluted peptides were sprayed directly into a quadrupole time-of-flight hybrid mass spectrometer (Q-TOF Ultima; Waters, Milford, MA). The spectra were obtained in the data-dependent MS/MS mode and were processed using MassLynx software (version 4.0; Waters) to create peak list files after smoothing by the Savitzky−Golay method (window channels, ± 3). The files were processed by the MASCOT algorithm (version 2.4.1; Matrix Science, Boston, MA) to assign peptides using the RefSeq protein sequence database (71 826 entries, downloaded in July 2014, http://www.ncbi.nlm.nih. gov/refseq/). The database search was performed with the parameters as described previously.26,27 Briefly, the parameters used are follows: enzyme, trypsin; maximum missed cleavage, 2; fixed modification, carbamidomethylation (Cys); variable modifications, deamination (pyroGlu, peptide N-terminal Gln), oxidation (Met), and IGOT (deamidation incorporating 18 O, +3 Da, Asn); peptide mass tolerance, 200 ppm; fragment mass tolerance, 0.5 Da. All results of the peptide search were exported in CSV files and processed by Microsoft Excel. We first selected the peptides with a rank of 1 and an expectation value
