Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Note Jinn-Shiun Chen, Kuei-Tien Chen and Chung-Wei Fan contributed equally to this article
E.-C. Chan, Department of Medical Biotechnology and Laboratory Science, Chang Gung University, 259 Wen-Hua 1st Road, Kweishan, Taoyuan, Taiwan, China Fax: +886 3 2118741 Tel: +886 3 2118800 (ext. 5220) E-mail: firstname.lastname@example.org
The aim of this study was to uncover the membrane protein profile differences between colorectal carcinoma and neighboring normal mucosa from colorectal cancer patients. Information from cellular membrane proteomes can be used not only to study the roles of membrane proteins in fundamental biological processes, but also to discover novel targets for improving the management of colorectal cancer patients. We used solvent extraction and a gel-assisted digestion method, together with isobaric tags with related and absolute quantitation (iTRAQ) reagents to label tumoral and adjacent normal tissues in a pairwise manner (n = 8). For high-throughput quantification, these digested labeled peptides were combined and simultaneously analyzed using LC-MS/MS. Using the shotgun approach, we identified a total of 438 distinct proteins from membrane fractions of all eight patients. After comparing protein expression between cancerous and corresponding normal tissue, we identified 34 upregulated and eight downregulated proteins with expression changes greater than twofold (Student’s t-test, P < 0.05). Among these, the overexpression of well-established biomarkers such as carcinoembryonic antigens (CEACAM5, CEACAM6), as well as claudin-3, HLA class I histocompatibility antigen A-1, tapasin and mitochondrial solute carrier family 25A4 were confirmed by western blotting. We conclude that gel-assisted digestion and iTRAQ labeling MS is a potential approach for uncovering and comparing membrane protein profiles of tissue samples that has the potential to identify novel biomarkers.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
isobaric tags with related and absolute quantitation
mitochondrial solute carrier family 25A4
Colorectal cancer (CRC) remains one of the most prevalent cancers in the western world and the third highest cause of cancer mortality in Taiwan . CRC is thought to evolve into invasive cancer from adenomatous polyps by acquired mutations in various genes . Development from adenoma into carcinoma takes 5–15 years, and there is therefore plenty of opportunity for early intervention. Approximately half of patients diagnosed with colorectal cancer die within 5 years of diagnosis, although an early diagnosis significantly improves patients’ outcomes. Unfortunately, few biomarkers are available for CRC analyses and none is sufficiently sensitive for screening purposes . Therefore, it is of great interest to identify proteins whose levels are consistently altered in CRC, both to improve the diagnosis and monitoring of CRC patients and because their function may reveal insight into critical events in tumorigenesis.
Various proteomic technologies have been used to search for new biomarkers in colorectal cancer [4–10]. There is increasing interest in sample prefractionation to reduce proteome complexity and gain deeper insight into the proteome. This strategy is particularly useful for low-abundance proteins such as membrane proteins. Membrane proteins account for ∼ 30% of the proteome and play critical roles in many biological functions such as cell signaling, cell–cell interactions, communication, transport mechanisms and energy . Information from membrane proteomes will help us understand the role of these proteins in fundamental biological processes, and it may also help us discover novel targets for biomedical therapeutics to improve patient management during pathogenesis . Thus, global analysis of membrane proteins in CRC may provide an important source of diagnostic or prognostic markers such as carcinoembryonic antigen-related cell adhesion molecule 5 (CEA).
Although high-throughput proteomic technologies can provide comprehensive analyses of soluble proteins, the analysis of membrane proteins has lagged behind because of their low concentration and high hydrophobicity. New tools and strategies are needed so that membrane fractions from cancer cells can be screened for candidate biomarkers. In this study, we utilized a technology combining gel-assisted digestion, isobaric tags with related and absolute quantitation (iTRAQ) labeling and LC-MS/MS for quantitative analysis of the membrane proteome of colorectal tissue. In brief, membrane proteins were solubilized with various types of detergents at high concentrations and subsequently incorporated into polyacrylamide gels without electrophoresis. Excess detergent was removed prior to protein digestion so that it would not interfere with the LC-MS/MS analysis. In addition, we also utilized a recently developed and widely used multiplexed quantitation strategy based on iTRAQ isobaric reagents [13–15]. The iTRAQ labeling strategy offers enhanced identification confidence and quantitation accuracy for proteomic research, especially for low-abundance proteins [16,17].
We used iTRAQ labeling together with gel-assisted digestion and mass spectrometry to detect differences in the protein expression profiles of membrane fractions from tumoral and adjacent normal mucosa from colorectal cancer patients. Differentially expressed proteins were identified by mass spectrometry and verified by western blotting. Initial validation studies confirmed the expression of claudin-3 (CLDN3) as a tumor-associated antigen in colorectal cancer. We also uncovered some candidates, such as HLA class I histocompatibility antigen A-1 (HLA-A1), tapasin (TAPBP) and mitochondrial solute carrier family 25A4 (SLC25A4), as potential biomarkers for monitoring CRC.
Quantitative analysis of membrane proteins from paired tumoral and adjacent normal tissue of CRC patients
A total of eight tumor tissues and eight matched normal tissues were collected from eight CRC patients (Table S1) and protein expression was compared between each tumor and adjacent normal tissue using LC-MS/MS analysis (Fig. 1). In our previous study using the same proteomic platform, quantitation of four independently purified membrane fractions from HeLa cells gave high accuracy (< 8% error) and precision (< 12% relative SD), demonstrating a high degree of consistency and reproducibility of this quantitation platform . We used the same quantitative strategy to enhance identification confidence and quantitation accuracy for proteomic research. A total of 438 proteins from both the tumor and normal tissue of eight patients was identified (false discovery rate = 2.25%). Figure 1 illustrates the flowchart for quantitative analysis of membrane proteins of the CRC samples and reveals 215, 299, 191 and 208 proteins from four 4-plex iTRAQ LC-MS/MS experiments, respectively. Statistical analysis of the expression level from eight CRC patients revealed changes in the expression of 42 proteins by more than twofold within 95% confidence levels (Student’s t-test; P < 0.05) of individual variation. Among the 42 identified proteins, 34 were upregulated and eight were downregulated (Table S2).
Differential protein expression analysis in CRC with hierarchical clustering
Cluster analysis was performed on our identified proteins to evaluate the relation between deregulated proteins and colorectal tissue samples and to identify interesting protein expression clusters. We initially uncovered 438 proteins from eight CRC patients and estimated their expression by comparing tumor tissues with adjacent normal tissues. By using a hierarchical clustering analysis, a clear distinction of expression patterns enabled the clustering of these proteins into several characteristic profiles, which split the 438 proteins into two main clusters: either upregulated (in red) or downregulated (in green) (Fig. 2). In cluster group 1, six proteins were notably downregulated in tumor tissues, including collagen I alpha-1 chain (−3.3-fold, P < 0.001), collagen I alpha-2 chain (−2.5-fold, P < 0.001), biglycan (−1.7-fold, P = 0.12), mimecan (−2.1-fold, P < 0.05), actin of aortic smooth muscle (−2.0-fold, P < 0.05) and myosin-11 (−1.7-fold, P = 0.11). In cluster group 2, 46 proteins were notably upregulated in tumor tissues, including isoform 1 of surfeit locus protein 4 (2.8-fold, P < 0.05), ITGB2, VDAC1, ADP/ATP translocase 1 (SLC25A4; 2-fold, P < 0.05), HLA-A1, VDAC2 and VDAC3, among others. Using cluster analysis with hierarchical partitioning of the expression profiles of identified proteins, the results from cluster groups 1 and 2 confirmed ∼ 73.8% (31/42) of the previously selected differentially expressed proteins (more than twofold within 95% confidence levels nof individual variation; Table S2) and added other interesting candidates, such as cytochrome c oxidase subunit 7C, NADH-ubiquinone oxidoreductase chain 4 or microsomal glutathione S-transferase 3 as possible CRC markers. For many of these proteins, there was a remarkable homogeneity of upregulated or downregulated expression across the eight pairs of CRC samples. Moreover, there were different cluster groups of proteins with less uniform patterns across the eight patients.
Functional classification of proteins identified in CRC
Proteins identified by mass spectrometry were classified by subcellular location and molecular function (Fig. 3). To better understand the probable roles of the membrane proteomes in terms of their biological functions, the subcellular localization and molecular functions of the 438 identified proteins were classified using the gene ontology (GO) consortium. The subcellular locations of these proteins are shown in Fig. 3A. We analyzed a total of 438 proteins, and ∼ 51% were found to be membrane bound or membrane associated. Among these, 27% were shown to be in the plasma membrane, including CEACAM5, CEACAM6, VDAC1, VDAC3, isoform 1 of tapasin (TAPBP), SLC25A4, HLA-A1, CLDN3, ITGB2, Galectin-3 and keratin type II cytoskeletal 8, and 24% were shown to be in organelle membranes (mitochondria or membrane-bound vesicles), including SEC11C, VDAC2 and cytochrome c oxidase subunit I. It is unclear whether the differentially identified mitochondrial proteins are related to the disease or whether they are sampling artifacts. Another 17% were shown to be in the extracellular space, including biglycan, collagen III alpha-1 chain, S100A8 and S100A9. Figure 3B shows the molecular function categorization of the proteins identified in CRC patients. Regarding major molecular functions, the proteins were mostly associated with binding functions (29.9%; S100A8, Galectin-3, keratin type II cytoskeletal 8), transporter activity (17.1%; VDAC1, VDAC2, VDAC3, TAPBP, SLC25A4) and catalytic activity (12.8%; cathepsin G, mitochondrial cytochrome c1 heme protein, component of pyruvate dehydrogenase complex mitochondrial precursor). A small number of proteins were also found associated with structural molecule activity (collagen I alpha-1 chain, collagen I alpha-2 chain, tubulin beta chain), molecular transducer activity (ITGB2, interferon-induced transmembrane protein 1, integrin alpha-6 and integrin alpha-M), signal transducer activity (S100A9, HLA-A1, CLDN3) and motor activity. For a few proteins (19.4%), no molecular function has yet been annotated.
Validation of differentially expressed proteins in CRC patients by western blotting
To further validate the results obtained from the relative comparative expression studies with LC-MS/MS, we examined the expression status of several of the identified proteins using western blotting. These representative proteins were selected based on changes of more than twofold in in their expression within the 95% confidence level (Student’s t-test; P < 0.05) of individual variation. In cases where the antibodies were suitable for western blotting, we tested their reactivity with CRC samples as a means of verification. Protein extracts from normal and tumoral tissues from another 16 patients were resolved by SDS/PAGE and blotted onto poly(vinylidene difluoride) membranes (Table S2). Figure 4 shows a representative compilation of immunoblotting for these proteins. These representative proteins included CLDN3, HLA-A1, SLC25A4 and TAPBP. The results of the western blot analysis in the tumoral and normal tissues confirmed the LC-MS/MS results. The expression levels of CLDN3, HLA-A1 and SLC25A4 were significantly higher in tumor compartment from CRC patients (P < 0.05). The protein expression of TAPBP was still differential, although less pronounced. TAPBP was upregulated in 12 of 16 CRC samples, but downregulated or not obviously changed between tumoral and matched normal samples in another four tissue pairs. Upregulation of CEA was not analyzed by immunoblot analysis. However, it has been unequivocally demonstrated in several earlier studies, using immunohistochemistry and immunoassays, that CEA expression is significantly elevated in neoplastic epithelium when compared with matched normal mucosa, and this was confirmed by our iTRAQ labeling MS analysis. These results demonstrate that some of the proteins identified by LC-MS/MS could serve as potential markers in future studies of CRC.
This study was aimed at identifying membrane proteins differentially expressed between colorectal cancer and normal tissue. We utilized iTRAQ labeling RPLC-MS/MS to explore the membrane protein profiles in paired CRC tissue samples. A commonly used strategy is multidimensional chromatography, where a first dimension, usually the strong cation-exchange chromatography, is combined with the second dimension RP-HPLC. However, the limited amount of membrane proteins extracted (5 μg·sample−1, a total of 20 μg for an iTRAQ analysis) from precious colorectal tissues restricted the use of fractionation prior to MS analysis. In our method, we decided to analyze the sample directly by RPLC-MS/MS three times to obtain a confident protein identification result. Using the iTRAQ labeling mass spectrometry, a total of 438 proteins were identified by our proteomic platform.
To better understand the roles of these identified proteins, they were grouped and analyzed according to their possible pathogenic roles. The clustering and molecular functions of the identified proteins can provide clues about their roles in the pathogenesis of CRC. In general, factors that contribute to the pathogenesis of CRC include the accumulation of mutations and the deregulation of gene expression. Of particular interest is the fact that a significant number of the proteins identified as differentially upregulated in tumor tissues may be functionally involved in the CRC tumorigenesis. Several clinically well-known biomarkers, such as CEACAM 5 and 6, were overexpressed in tumor tissues, compared with the matched normal colorectal tissues in our study. Although CEA is not an adequate screening tool for colorectal cancer patients, the assessment of CEA levels for prognosis has been shown to be an important variable in predicting postoperative outcomes. Data from studies on postoperative colorectal cancer patients have demonstrated that measurement of CEA every 3 months for at least 3 years is a valuable and cost-effective component of follow-up .
Our findings are in line with the results of several proteomics analyses. Alfonso et al. used a 2D-DIGE based approach to detect differentially expressed membrane proteins of colorectal cancer tissues. An important implication of the study is the conclusion that annexin A2, annexin A4 and VDAC appear as potential markers of interest for colorectal cancer diagnosis . A recent report detecting the changes of protein profiles associated with the process of colorectal tumorigenesis to identify specific protein markers for early colorectal cancer detection and diagnosis or as potential therapeutic targets. VDAC1, annexin A2 and Keratin 8 variant have been identified . Madoz-Gurpide et al. tested seven potential markers (ANXA3, BMP4, LCN2, SPARC, SPP1, MMP7 and MMP11) for antibody production and/or validation. ANXA3 was confirmed to be overexpressed in colorectal tumoral tissues . Kim et al.  analyzed CRC tissues using 2D difference in-gel electrophoresis on a narrow-range IPG strip and suggested S100A8 and S100A9 as candidates for serological biomarkers in combination with other serum markers that aid CRC diagnosis. Using our strategy combining gel-assisted digestion, iTRAQ labeling and LC-MS/MS analysis, identical or similar proteins were identified, including VDAC1, VDAC2, VDAC3, ANXA4 (2.5-fold, P < 0.05), ANXA5 (6.5-fold, P < 0.05), S100A8 (9.5-fold, P < 0.05) and S100A9 (8.5-fold, P < 0.05).
In addition to the well-known biomarkers and colorectal cancer-associated proteins such as CEA, CEACAM 6, VDAC and ANXA4, we identified several other proteins that may be potential novel markers for monitoring CRC but have not been unequivocally associated with colorectal carcinoma. Overexpression of CLDN3, HLA-A1, TAPBP and SLC25A4 in colorectal cancer has not been prominently reported, and there is interest in developing these proteins as diagnostic and prognostic markers for this disease. In western blotting analysis, CLDN3, HLA-A1 and SLC25A4 showed the best discriminatory power between tumoral and normal tissue. Our data provide important clues for the identification of differentially expressed membrane-associated proteins in CRC, and uncover several avenues for study of their roles in CRC carcinogenesis. Some of their functional roles and implications in CRC are discussed below.
CLDN3 was highly expressed in cancer tissues when tested by LC-MS/MS and western blotting. CLDN3 belongs to the claudin (CLDN) family, which consists of ∼ 23 proteins that are essential for the formation of tight junctions in epithelial and endothelial cells . Specifically, CLDN1, -3, -4, -5, -7, -10 and -16 have been found to be altered in various cancers . Overexpression of these proteins in cancer is unexpected, but recent work suggests that claudins may be involved in the survival of and invasion by cancer cells [24,25]. In addition, because claudins are surface proteins, they may represent useful targets for various therapeutic strategies. Interestingly, Clostridium perfringens enterotoxin is a ligand for CLDN3 and CLDN4 proteins, and binding of the toxin to these claudins leads to rapid cytolysis of cells . Preclinical studies have suggested that Clostridium perfringens enterotoxin may be effective against CLDN3- and CLDN4-expressing malignancies [27,28]. In our study, we found that overexpression of CLDN3 is significantly associated with CRC. In a previous study, CLDN3 expression was analyzed in 12 adenocarcinoma tissues and their paired normal mucosa, and was shown to be upregulated 1.5-fold in CRC . It would be worthwhile to further elucidate the value of this protein as a diagnostic and/or prognostic marker for CRC and to further understand its role in the survival and/or invasion in CRC cancer cells.
SLC25A4 was also significantly increased in CRC tissues compared with matched normal tissues. The solute carrier family 25 (SLC25) consists of proteins that are functionally and structurally related and that construct the transporters of a large variety of molecules . Following LC-MS/MS and western blotting analyses, SLC25A4 showed differential expression between tumor and normal tissues. This protein could be a valuable diagnostic marker or a target for monitoring patients’ conditions.
HLA-A1 was highly expressed in cancer tissues when tested by LC-MS/MS and western blot methods. Expression of human leukocyte antigen (HLA) class I presenting tumor-associated antigens on the tumor cell surface is considered to be a prerequisite for effective T-lymphocyte activation . As a consequence, HLA class I antigens can be downregulated or lost on malignant cells, and these variations may be associated with a poor prognosis [32,33].
In our study, expression of HLA-A1, determined by LC-MS/MS and western blotting, was upregulated in colorectal cancer in comparison with normal tissues. Although these results may appear controversial, only a few studies have reported the clinical impact of HLA class I expression in colorectal cancer, with contrasting results. Some studies have shown no significant correlation between staining intensity of HLA class I expression and survival [34,35], whereas others found that HLA class I expression correlated with the prognosis of CRC patients [36,37].
TAPBP may upregulate the expression of HLA class I molecules, and it was found to be upregulated in cancer tissues in this study using LC-MS/MS and western blotting. TAPBP plays multiple roles in the peptide-loading complex; it stabilizes the complex, aids in the appropriate selection of peptides, maintains appropriate HLA class I redox status and enhances TAP and HLA class I levels [38,39].
In summary, the strategy combining gel-assisted digestion and iTRAQ labeling LC-MS/MS has proven to be a potential means of identifying proteins in the membrane fraction from CRC tumoral samples. Some of the representative candidates, such as CLDN3, HLA-A1 and SLC25A4, appear to be promising markers for the detection of colorectal cancer.
Materials and methods
Monomeric acrylamide/bisacrylamide solution (40%, 29 : 1) was purchased from Bio-Rad (Hercules, CA, USA). Trypsin (modified, sequencing grade) was obtained from Promega (Madison, WI, USA). The BCA and Bradford protein assay reagent kits were obtained from Pierce (Rockford, IL, USA). SDS was purchased from GE Healthcare (Central Plaza, Singapore). Ammonium persulfate and N,N,N′,N′-tetramethylenediamine were purchased from Amersham Pharmacia (Piscataway, NJ, USA). EDTA was purchased from Merck (Darmstadt, Germany). Tris(2-carboxyethyl)-phosphine hydrochloride, triethylammonium bicarbonate, Na2CO3, NaCl, sucrose, magnesium chloride hexahydrate (MgCl2), Hepes, methyl methanethiosulfonate, trifluoroacetic acid and HPLC-grade acetonitrile were purchased from Sigma-Aldrich (St Louis, MO, USA). Formic acid was purchased from Riedel de Haen (Seelze, Germany). Water was obtained from Milli-Q® Ultrapure Water Purification Systems (Millipore, Billerica, MA, USA).
Patients and tumors
Clinical tissue samples from 56 patients with colorectal cancer were taken from freshly isolated surgical resections in the operating room at the Chang Gung Memorial Hospital, Tao Yuan, Taiwan. Malignant tissue (determined by pathological assessment) and adjacent normal tissue were prepared from the same resection. All formalin-fixed paraffin-embedded tumor blocks from equivalent specimens from the same tumor tissue were inspected for quality and tumor content, and a single representative tumor block from each case, containing at least 70% neoplastic cells, was selected for the study. Normal tissue was obtained from the distal edge of the resection at least 10 cm from the tumor. Written informed consent from all respective patients was obtained before surgery in accordance with medical ethics and approval by Human Clinical Trial Committee at Chang Gung Memorial Hospital. A total of eight tissue pairs containing tumoral and adjacent normal tissue were collected and analyzed by gel-assisted digestion and iTRAQ labeling MS. Other tissue pairs were utilized to verify potential targets from the above-mentioned LC-MS/MS analysis. Patients who had received any chemo- and/or radiotherapeutic treatment before surgery were excluded from this study.
Isolation of membrane proteins from tumoral and adjacent normal tissues
After surgery, paired tumoral and adjacent normal tissues were obtained from the same CRC patient and stored at −80 °C. Frozen tissues were unfrozen rapidly in a 37 °C water bath, washed with 0.9% (w/v) NaCl solution to remove blood, resuspended in STM solution (5 g·mL; 0.25 m sucrose, 10 mm Tris/HCl, 1 mm MgCl2) with protease inhibitors (protein : protein inhibitor = 100 : 1, v/v) and homogenized with a homogenizer (Polytron System PT 1200 E, Luzernerstrasse, Switzerland). The nuclei were removed by centrifugation at 260 g for 5 min at 4 °C, and the postnucleus supernatant was centrifuged at 1500 g for 10 min at 4 °C. The pellet was mixed with two-thirds the original homogenate volume of a 0.25 m STM solution containing protease inhibitors and resuspended in a homogenizer with three strokes of the loose-fitting pestle followed by one stroke of the tight-fitting pestle (Kimble/Kontes, Vineland). The resulting solution was centrifuged at 12 000 g for 1 h at 4 °C to pellet the membrane proteins. The pellet was washed twice with 1 mL of ice-cold 0.1 m Na2CO3 (pH 11.5), dissolved in 50 μL of 90% (v/v) formic acid to determine the membrane protein concentration by Bradford assay, and then vacuum dried to obtain a membrane pellet for subsequent proteolysis reactions.
Digestion of membrane proteins
Purified membrane proteins were subjected to gel-assisted digestion . In detail, the membrane protein pellet was resuspended in 50 μL of 6 m urea, 5 mm EDTA and 2% (w/v) SDS in 0.1 m triethylammonium bicarbonate and incubated at 37 °C for 30 min until completely dissolved. Proteins were chemically reduced by adding 1.28 μL of 200 mM Tris(2-carboxyethyl)-phosphine and alkylated by adding 0.52 μL of 200 mm methyl methanethiosulfonate at room temperature for 30 min. To incorporate proteins into a gel directly in an Eppendorf vial, 18.5 μL of acrylamide/bisacrylamide solution (40%, v/v, 29 : 1), 2.5 μL of 10% (w/v) ammonium persulfate, and 1 μL of 100%N,N,N′,N′-tetramethylenediamine was applied to the membrane protein solution. The gel was cut into small pieces and washed several times with 1 mL of triethylammonium bicarbonate containing 50% (v/v) acetonitrile. The gel samples were further dehydrated with 100% acetonitrile and then completely dried by SpeedVac. Proteolytic digestion was then performed with trypsin (protein/trypsin = 10 : 1, g/g) in 25 mm triethylammonium bicarbonate with incubation overnight at 37 °C. Peptides were extracted from the gel using sequential extraction with 200 μL of 25 mm triethylammonium bicarbonate, 200 μL of 0.1% (v/v) trifluoroacetic acid in water, 200 μL of 0.1% (v/v) trifluoroacetic acid in acetonitrile and 200 μL of 100% acetonitrile. The solutions were combined and concentrated in a SpeedVac.
iTRAQ labeling and LC-ESI MS/MS analysis
To label peptides with the iTRAQ reagent (Applied Biosystems, Foster City, CA, USA), one unit of label (defined as the amount of reagent required to label 100 μg of protein) was thawed and reconstituted in ethanol (70 μL) by vortexing for 1 min. The resulting peptides from the normal tissue of one patient were labeled with iTRAQ114 and peptides from tumor tissue of the same patient were labeled with iTRAQ115. The resulting peptides from normal tissue of another patient were labeled with iTRAQ116 and peptides from tumor tissue were labeled with iTRAQ117 and incubated at room temperature for 1 h. The same procedures were performed in the peptides from nontumor and tumor tissues of the remaining patients. Labeled peptides (5 μg each) were then pooled, vacuum dried and resuspended in 0.1% (v/v) trifluoroacetic acid (40 μL) for further desalting and concentration using Oasis® HLB uElution (Waters Corporation, Milford, MA, USA).
All MS/MS experiments for peptide identification were performed using a Waters nanoACQUITY UPLC pump system and a Waters Q-Tof premier mass spectrometer (Waters Corp.) equipped with a nano-ESI source. The nanoUPLC system used an aqueous mobile phase (buffer A) containing 0.1% formic acid in water and an organic mobile phase (buffer B) containing 0.1% (v/v) formic acid in acetonitrile. Desalting of the samples was performed for 1.5 min with 99% buffer A using a C18 trapping column (5 μm, 20 mm × 180 μm id; Waters Corp.). Samples were separated using a Waters ACQUITY™ BEH C18 Column (1.7 μm, 250 mm × 75 μm; Waters Corp.) at 300 nL·min−1 using a 120 min gradient.
During each LC injection, the mass spectrometer was operated in ESI positive V mode with a resolving power of 10 000. The voltage applied to produce an electrospray was 2.85 kV and the cone voltage was 35 eV. Argon was introduced as a collision gas and the collision flow rate was 0.35 mL·min−1. Data acquisition was operated in the data directed analysis mode. This mode included a full MS scan (m/z 400–1600, 0.6 s) and an MS/MS scan (m/z 100–1990, 1.2 s each scan) sequentially on the three most intense ions present in the full scan mass spectrum. Mass accuracy was calibrated with a synthetic human [Glu1]-Fibrinopeptide B solution (500 fmol·μL−1) due to the use of a NanoLockSpray source and sampled every 30 s. The collision energies were used to fragment each peptide ion on the basis of its mass-to-charge (m/z) values.
Data processing and analysis
For protein identification, data files from LC-MS/MS were searched against the non-redundant International Protein Index human sequence database v3.29  (68 161 sequences) from the European Bioinformatics Institute using the mascot algorithm (v2.2.1, Matrix Science, London, UK). Peak lists were generated and processed using mascot distiller v22.214.171.124 (Matrix Science). Search parameters for peptide and MS/MS mass tolerance were ± 0.1 Da and ± 0.1 Da, respectively, with allowance for two missed cleavages made from the trypsin digest and variable modifications of deamidation (Asn, Gln), oxidation (Met), iTRAQ (N-terminal), iTRAQ (Lys) and methyl methanethiosulfonate (Cys). Only proteins with a protein identification confidence interval of > 95% were confidently assigned. When unique peptides were identified to multiple members of a protein family, proteins with the highest sequence coverage were selected from the mascot search output. To evaluate the false discovery rate, we repeated the searches against a random database using identical search parameters and validation criteria.
For protein quantitation, we used multi-q  software to analyze the iTRAQ data. Raw data files from the Waters Q-Tof premier mass spectrometer were converted into files of mzXML format using masswolf (Institute for Systems Biology, Seattle, WA, USA), and the search results in mascot were exported in the xml data format. After the data conversions, multi-q selected unique iTRAQ-labeled peptides with confident MS/MS identification (mascot score ≥ 40), detected signature ions (m/z = 114, 115, 116, 117), and performed automated quantitation of peptide abundance. For the detector dynamic range filter, signature peaks with ion counts < 30 were filtered out by multi-q. To calculate protein ratios, the ratios of quantified unique iTRAQ peptides were weighted according to their peak intensities to minimize the standard deviation. The final protein quantitation results were exported to an output file in csv data format.
A total of 438 identified proteins were clustered based on normal Euclidean distance between them and average linkage. The treeview program was used to observe the hierarchical partitioning of expression profiles of identified proteins.
For subcellular localization and molecular function annotations, all the proteins identified in this study were analyzed using the Ingenuity Pathway Analysis Knowledge Base (http://www.ingenuity.com/) and gene ontology (GO) consortium .
Western blot and statistical analysis
Immunoblots of selected proteins were performed using tissue lysates from both tumoral and adjacent normal samples to confirm the LC-MS/MS findings. In total, tissue lysates from another patients with CRC were examined by immunoblotting. Briefly, each tissue sample was mixed with electrophoresis sample buffer containing 2% SDS and 5% 2-mercaptoethanol and boiled for 5 min. Proteins were separated by electrophoresis on 12% denaturing polyacrylamide gels and transferred to poly(vinylidene difluoride) membranes. These blots were blocked with 5% skim milk and then probed with the appropriate primary antibody (claudin-3 antibody; Abcam, Cambridge, MA, USA; SLC25A4 mAb, Abnova, Taipei, Taiwan; HLA Class 1 A1 antibody, Abcam; Tapasin antibody, Abcam) at a dilution of 1 : 1000 for 2 h, followed by incubation for 1 h with peroxidase-conjugated secondary antibody at room temperature. The blots were visualized by ECL and then exposed to Kodak biomax light films. The immunoblot images were acquired by Imagemaster (Amersham Pharmacia Biotech, NJ, USA). The protein level of each band was quantified by densitometry and analyzed with multi gauge version 2.0 software (Fuji PhotoFilm, Tokyo, Japan). Data were analyzed by an unpaired t-test using the statistical software spss/windows 11.0 statistical package (SPSS Inc, Chicago, IL, USA). P values of < 0.05 were considered statistically significant.
This work was supported by grants (CMRPD160097 and CMRPG371431) from Chang Gung University and Memorial Hospital, Taiwan.