Lectin Microarray-Based Sero-Biomarker Verification Targeting

(3) A pathogenic association between liver fluke infection and CC has been suggested ... from Thailand, which has the highest incidence of CC worldwid...
3 downloads 0 Views 2MB Size
Article pubs.acs.org/ac

Lectin Microarray-Based Sero-Biomarker Verification Targeting Aberrant O‑Linked Glycosylation on Mucin 1 Atsushi Matsuda,† Atsushi Kuno,† Tomomi Nakagawa,† Yuzuru Ikehara,† Tatsuro Irimura,‡ Masakazu Yamamoto,§ Yasuni Nakanuma,∥ Eiji Miyoshi,⊥ Shoji Nakamori,# Hayao Nakanishi,¶ Chutiwan Viwatthanasittiphong,● Petcharin Srivatanakul,▲ Masanao Miwa,∇ Junichi Shoda,⬟ and Hisashi Narimatsu*,† †

Research Center for Medical Glycoscience (RCMG), National Institute of Advanced Industrial Science and Technology (AIST), AIST Tsukuba Central 2, 1-1-1, Umezono, Tsukuba, Ibaraki 305-8568, Japan ‡ Juntendo University School of Medicine, 2-1-1, Hongo, Bunkyo-ku, Tokyo, 113-8421 Japan § Department of Surgery, Institute of Gastroenterology, Tokyo Women’s Medical University, 8-1, Kawada-cho, Shinjuku-ku, Tokyo, 162-8666 Japan ∥ Department of Human Pathology, Graduate School of Medical Sciences, Kanazawa University, 13-1 Takara-Machi, Kanazawa, Ishikawa 920-8641 Japan ⊥ Department of Molecular Biochemistry and Clinical Investigation, Osaka University Graduate School of Medicine, 2-2 Yamadaoka, Suita, Osaka 565-0871 Japan # National Hospital Organization Osaka National Hospital, 2-1-14 Hoenzaka, Chuo-ku, Osaka 540-0006, Japan ¶ Division of Oncological Pathology, Aichi Cancer Center Research Institute, 1-1 Kanokoden, Chikusa-ku, Nagoya, Aichi 464-8681, Japan ● Ubon Cancer Center of Thailand, 405 Khung a-wuth Road, Ubon Ratchathani 34000, Thailand ▲ National Cancer Institute of Thailand, 268/1 Rama VI, Ratchathewi, Bangkok 10400, Thailand ∇ Faculty of Bioscience, Nagahama Institute of Bio-Science and Technology, 1266 Tamura-Cho, Nagahama, Shiga 526-0829 Japan ⬟ Field of Basic Sports Medicine, Graduate School of Comprehensive Human Sciences, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8574, Japan S Supporting Information *

ABSTRACT: Glycoform of mucin 1 (MUC1) in cancerous cells changes markedly with cell differentiation, and thus, qualitative detection and verification of the MUC1 glycosylation changes have potential diagnostic value. We have developed an ultrasensitive method to detect the changes in cholangiocarcinoma (CC), which produces MUC1, and applied it in the diagnostics development. The focused glycan analysis using 43-lectin-immobilized microarray could obtain the glycan profiles of sialylated MUC1 in 5 μL of sera. The high-throughput analysis detected disease-specific alterations of glycosylation, and the statistical analysis confirmed that use of Wisteria floribunda agglutinin (WFA) alone produced a diagnostic score sufficient for discriminating 33 CC cases from 40 hepatolithiasis patients and 48 normal controls (p < 0.0001). The CC-related glycosylation change was verified by the lectin−antibody sandwich ELISA with WFA in two cohorts: (1) 78 Opisthorchis viverrini infected patients without CC and 78 with CC, (2) 33 CC patients and 40 hepatolithiasis patients (the same cohort used for the above lectin microarray). The WFA positivity distinguished patients with CC (opisthorchiasis: p < 0.0001, odds ratio = 1.047; hepatolithiasis: p = 0.0002, odds ratio = 1.018). Sensitive detection of qualitative alterations of sialylated MUC1 glycosylation is indispensable for the development of our glycodiagnostic test for CC.

C

holangiocarcinoma (CC) is the most common primary malignancy of the biliary tract and has a poor survival rate. Surgical management at a relatively early stage is the only potentially curative treatment. However, the late appearance of its clinical presentation makes CC difficult to detect at an early © 2015 American Chemical Society

Received: April 9, 2015 Accepted: June 19, 2015 Published: June 19, 2015 7274

DOI: 10.1021/acs.analchem.5b01329 Anal. Chem. 2015, 87, 7274−7281

Article

Analytical Chemistry

The detection system can be simplified to detect selected lectins of clinical relevance and may be most suitable for identifying CC-specific glycosylation of MUC1 after the use of a conventional lectin-antibody sandwich ELISA.

stage, and thus, CC is fatal in most cases by the time it becomes clinically evident.1,2 The epidemiological risk factors for CC are associated with chronic inflammation of the epithelium in bile duct diseases such as primary sclerosing cholangitis (PSC), liver fluke infection of Opisthorchis viverrini (opisthorchiasis), and intrahepatic gallstone (hepatolithiasis).2 PSC is a chronic biliary destructive disorder of unknown etiology and is the most common predisposing factor for CC in Western countries, accounting for 5−15% of cases of CC.3 A pathogenic association between liver fluke infection and CC has been suggested in many experimental and epidemiological studies.4−6 Most epidemiological data have been reported from Thailand, which has the highest incidence of CC worldwide (87 per 100 000 population).7,8 Hepatolithiasis is associated particularly with peripheral intrahepatic CC and is relatively common in parts of Asia where up to 10% of patients with hepatolithiasis develop CC.9 Therefore, the key to improving the survival rate of CC is being able to make a precise differential diagnosis between CC and these other diseases at an early stage. Carbohydrate antigen 19-9 (CA19-9) is used widely as a serological marker of CC. However, the diagnostic sensitivity is limited because this antigen is not synthesized in Lewisnegative individuals. 10 Its diagnostic specificity is also controversial because CA19-9 is found in normal embryonic tissue and is overexpressed in certain types of epithelial cancer and under inflammatory conditions.11,12 Moreover, it is not clear whether all CA19-9 antibodies are sufficiently specific to the CA19-9 epitope, sialyl-Lewis A, and thus, the diagnostic performance can be influenced greatly by the antibody clone used.13 Recent studies have confirmed in a limited number of patients that serum cytokeratin 19 fragment (CYFRA21-1) and carbohydrate antigen 242 (CA242) are more specific CC markers compared with CA19-9.14,15 Development of a novel and high-performance serodiagnostic marker may improve CC diagnosis. Mucins are large extracellular proteins with abundant Olinked glycosylation. Changes in mucin expression and glycosylation are associated with cancer progression. Among the mucin family, MUC1 is used widely as a histochemical or serological diagnostic marker of various cancers.16 However, the level of MUC1 expression is similar in both normal and tumor cells in some cases, meaning that qualitative changes in the glycosylation of MUC1 may have clinical relevance. There is remarkable diversity in glycan moieties between normal and cancer cells.17,18 Reports have shown that aberrant Oglycosylation of MUC1 reflects tumor progression, and the detection of the unique glycosylation pattern shown by antibody−lectin sandwich microarray is regarded as a useful biomarker of various diseases.19,20 Haab and colleagues demonstrated that the glycan structure of MUC1 is altered markedly at multiple steps of tumor development and progression including premalignancy under inflammatory conditions.21 Considering the main risk factors for CC, detection and monitoring of chronic inflammation in the bile duct at the premalignant step should be important for the early detection of CC and improvement in CC prognosis. Detection of changes in glycosylation of MUC1 during cancer progressions is important for the development of a clinically useful marker of CC. In this study, we report for the first time the detection of CC-specific glycan changes in serum sialylated MUC1 using an ultrasensitive lectin microarray and its diagnostic performance as a serological glycomarker of CC.



EXPERIMENTAL SECTION Serum and Plasma Specimens. Plasma specimens were collected from age- and sex-matched O. viverrini infected patients without CC (n = 78) and with CC (n = 78) in Thailand as Cohort 1 for the sandwich ELISA. As Cohort 2, serum specimens were collected from normal control (NC, n = 48), patients with hepatolithiasis (n = 46), and patients with CC (n = 33). NC specimens were prepared from total blood samples obtained from healthy volunteers without any history of CC, other cancer, or benign bile duct disease. Six hepatolithiasis patients were excluded because they did not have sufficient clinical information, and the remaining 40 hepatolithiasis patients comprised the hepatolithiasis patient group in the study. This cohort was used for the selection of the useful lectin by the differential glycan profiling with the antibody-overlay lectin microarray and the verification with sandwich ELISA. All specimens were collected from the 277 participants using Venoject II tubes (Terumo, Tokyo, Japan) according to the supplier’s manual. Each sample was divided into 2.0 mL aliquots in sterile cryotubes and immediately frozen at −80 °C until later use. The stored samples were thawed and centrifuged at 1500g for 10 min at 4 °C, and the supernatants were recovered to use in the enrichment process. All samples underwent only one freeze/thaw cycle. Age and sex for all subjects were recorded and the total bilirubin, α-fetoprotein, aspartate transaminase, alanine transaminase, and albumin were measured in accordance with the manual of each institute. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki. All patients provided their written informed consent for the analysis of their serum or plasma. The protocol for this study was approved by the ethical committees of Kanazawa University, Tokyo Women’s Medical University, Osaka University, Osaka National Hospital, Aichi Cancer Center Research Institute, Nagahama Institute of Bio-Science and Technology, University of Tsukuba, Ubon Cancer Center of Thailand, National Cancer Institute of Thailand, and the National Institute of Advanced Industrial Science and Technology. The experiments in this study were carried out in accordance with the approved guidelines. Culture Supernatants of CC Cell Lines. Two biliary tract cancer cell lines (KMC-1 and TGBC1TKB) used in this study were provided by the RIKEN BioResource Center through the National BioResource Project of the Ministry of Education, Culture, Sports, Science and Technology, Japan. KMC-1 was established from an intrahepatic CC and mucin-producing cell. 22 TGBC1TKB was established from metastasized gallbladder carcinoma of the lymph node and mucin-producing cell.23 RPMI 1640 medium (Life Technologies, Carlsbad, CA) supplemented with 5% fetal bovine serum, 100 units/mL penicillin, and 100 μg/mL streptomycin was used for cell culture. The cells were grown to confluence in a 150 cm2 flask and then washed five times with serum-free RPMI 1640 medium. The cells were then grown in serum-free RPMI 1640 medium for 2 days, and the culture supernatants were collected and used as the serum-free cell culture supernatants for the lectin microarray and ELISA. Immunoprecipitation of Sialylated MUC1. Sialylated MUC1 was immunoprecipitated using biotinylated antibody7275

DOI: 10.1021/acs.analchem.5b01329 Anal. Chem. 2015, 87, 7274−7281

Article

Analytical Chemistry

were performed in triplicate, and the mean value was used as the final value for each sample. Measurement of Serum CA19-9, CYFRA21-1, and CA242 Concentrations. The concentrations of serum CA19-9, CYFRA21-1, and CA242 were measured using commercial CA19-9, CYFRA21-1, and CA242 ELISA kits (DRC International, Marburg, Germany), respectively, following the manufacturer’s protocols. Statistics. The values are presented as the median ± standard deviation (SD). The Mann−Whitney U test was used to compare the clinicopathological data between patients with CC, other diseases, or NCs. All calculations were performed using either GraphPad Prism version 5.00 for Windows (GraphPad software, San Diego, CA) or Dr. SPSS II software for Windows (SPSS Inc., Chicago, IL). Receiver-operating characteristic (ROC) curve analysis was used to evaluate the differences between samples from patients with CC and other diseases based on the sensitivity and specificity at various cutoff levels. ROC analysis and calculation of an area under the ROC curve (AUC) were performed using GraphPad Prism. To evaluate the results, we used the cutoff values obtained from Youden’s index.28,29 For combinational analysis of WFApositive sialylated MUC1 (WFA-sialylated MUC1) and CA19-9, the following formula was used: Combination = (10 × [WFA-sialylated MUC]) + [CA19-9] (Figure 5).

conjugated streptavidin magnetic beads (Dynabeads, Life Technologies). The antisialylated MUC1 monoclonal antibody MY.1E12 was biotinylated using Biotin Labeling Kit-NH2 (Dojindo Laboratories, Kumamoto, Japan).24 Biotinylated MY.1E12 (500 ng) was reacted with streptavidin-coated magnetic beads in 20 μL of PBS containing 1% Triton X-100 (PBSTx) for 1 h at 4 °C. The beads conjugated with MY.1E12 were washed with PBSTx, and 20 μL of the samples (diluted in PBSTx) was added to the beads. The reaction between the antigen and MY.1E12 was continued by overnight incubation at 4 °C. The beads were washed with PBSTx, 10 μL of PBS containing 0.2% SDS was added to the beads, and the bound material was eluted after incubation for 10 min at 95 °C to remove sialylated MUC1 from immobilized-MY.1E12. For the lectin microarray analysis, the contaminant biotinylated MY.1E12 was depleted completely by addition of 40 μL of the streptavidin-coated magnetic beads, and the supernatants were collected as the immunoprecipitated samples. Antibody-Overlay Lectin Microarray. The antibodyoverlay lectin microarray was performed essentially as described previously.25 In brief, purified sialylated MUC1 from serum and culture supernatants as described above was diluted with PBSTx and then applied to the lectin array slide containing triplicate spots of 43 lectins (Supplementary Table S1). After incubation for 12 h at 20 °C, human serum polyclonal IgG (20 μg) was added, and the mixture was incubated for 30 min at 20 °C to reduce the background noise. The slide was washed three times with PBSTx, 60 μL of the biotinylated MY.1E12 solution (100 ng) in PBSTx was applied to the array, and the array was incubated for 1 h at 20 °C. The slide was washed three times with PBSTx, 100 ng of a Cy3-labeled streptavidin (GE Healthcare U.K. Ltd., Little Chalfont, U.K.) solution in PBSTx was added to the slide, and the slide was incubated for 25 min at 20 °C. The slide was washed with PBSTx and scanned with an evanescent-field fluorescence scanner (GlycoStation Reader 1200; GlycoTechnica Ltd., Yokohama, Japan). All data were analyzed with the Array-Pro Analyzer, version 4.5 (Media Cybernetics, Inc., Bethesda, MD). The net intensity of each spot was calculated by subtracting the background value from the signal intensity of three spots. The lectin signals of triplicate spots were averaged and normalized to the mean value of 43 lectins immobilized on the array, as described in a previous study.26 WFA-MY.1E12 Sandwich ELISA. The Wisteria floribunda agglutinin (WFA)-immobilized MY.1E12 sandwich ELISA was performed essentially as described previously.27 All specimens for analysis were diluted 1:10 with PBS containing 0.2% SDS and then heated for 5 min at 95 °C before the ELISA assay. The resulting solution (10 μL) was applied to the WFA-coated plates, and the plates were incubated for 30 min at room temperature. After washing with PBS containing 0.1% Tween 20 (PBST), the plate was incubated with 100 ng/well of MY.1E12 in PBST containing 0.1% BSA for 30 min at room temperature. The plate was washed extensively and then incubated with 50 μL of 1:2000-diluted solution of horseradish peroxidase-conjugated antimouse IgG (Jackson ImmunoResearch Laboratories Inc., Philadelphia, PA) in PBST for 30 min at room temperature. The substrate 3,3′,5,5′-tetramethylbenzidine (100 μL, Thermo Fisher Scientific, Fremont, CA) solution was added to each well and reacted for 30 min at 37 °C. The reaction was stopped by addition of 100 μL of 1 M sulfuric acid, and the OD value at 450 nm was measured. All experiments



RESULTS AND DISCUSSION Construction of a Highly Sensitive Glycan-Profiling Method Targeting Serum Sialylated MUC1. The diagnostic performance and specificity of MUC1 as a serological marker tend to be limited because of the presence of MUC1 in the serum from NCs or patients with benign disease.30 Thus, qualitative analysis of the glycosylation profile is an important key to developing a new diagnostic marker for CC and to improve the prognosis in CC patients. It remains difficult to analyze the glycoforms qualitatively, because biological methods such as mass spectrometry have insufficient throughput or sensitivity to allow comparisons between multiple samples including NCs for the differential analysis of various glycan structures of MUC1.31 Better methods based on microarray techniques have been reported recently for the differential analysis of MUC1 glycosylation, although these have yielded less detailed structural information.32,33 The concept of antibody-overlay lectin microarray can be used to obtain information about the glycosylation of target glycoproteins in a small amount of clinical samples.25,34 This methodology has facilitated the discovery of glycobiomarkers based on disease-specific glycosylation of target glycoproteins. However, most studies have focused on N-glycosylated proteins.35,36 The alternative microarray-based method (i.e., lectin-overlay antibody microarray) adopts various antimucin antibodies into the analysis of O-glycosylated proteins.20 This technique is more suitable for validating a vast number of target proteins using a small number of lectin probes but not for the simultaneous detection of multiple lectin bindings to a target protein. For instance, it requires at least 20 lectins as probes for differential glycan profiling of MUC1.37 Therefore, for the development of a new glycodiagnostic marker, we report here the glycan profiling of MUC1 in clinical specimens using lectin microarray (Figure 1A) to complement the previous report.25 First, the glycan profiling of MUC1 in cell culture supernatants of the sialylated MUC1-positive CC cell line (KMC-1) was obtained by the antibody-overlay lectin microarray with 7276

DOI: 10.1021/acs.analchem.5b01329 Anal. Chem. 2015, 87, 7274−7281

Article

Analytical Chemistry

calculated the coefficient of variation (CV) values for all lectin signals to determine the internal control lectin as reported previously.35,41 All lectin signals were normalized to the selected internal control Maclura pomifera agglutinin (MPA), which showed the lowest CV value (18%). In the comparative analysis of the lectin profiles (Supplementary Table S4 and Supplementary Figure S2), six lectins were identified as the lectin probes showing significant differences (p < 0.01) between the CC and hepatolithiasis patients: Ulex europaeus agglutinin-I (UEA-I, p = 0.0038), Aleuria aurantia lectin (AAL, p < 0.0001), Bauhinia purpurea alba lectin (BPL, p = 0.0014), Solanum tuberosum lectin (STL, p = 0.0067), WFA (p < 0.0001), and wheat germ agglutinin (WGA, p = 0.0006) (Figure 2A).

Figure 1. Antibody-overlay lectin microarray with MY.1E12. (A) Schematic diagram of an antibody-overlay lectin microarray with MY.1E12. (B) Scanned image of an antibody-overlay lectin microarray targeting sialylated MUC1 from KMC-1 culture supernatants and that after sialidase treatment. (C) Bar graph of 12 lectins binding to sialylated MUC1 from KMC-1. Neu, N-acetylneuramic acid; β-Gal, βgalactose; α-GalNAc, α-N-acetylgalactosamine; Sup, supernatant.

MY.1E12. Sialylated MUC1 from KMC-1 cells showed some significant lectin signals, and these signals were removed by digestion of sialic acid with sialidase treatment because MY.1E12 cannot bind to asialo-MUC1 carrying no sialic acid (Figure 1B,C). This result suggests that the MY.1E12-overlay lectin microarray is effective for glycan profiling of sialylated MUC1. Differential Glycan Profiling of Serum Sialylated MUC1. For the differential glycan profiling of sialylated MUC1 in multiple serum samples, we determined preliminarily the suitable sample volume for reliable analysis using pooled NC serum. The obtained dilution curves of sialylated MUC1 confirmed the detection of sialylated MUC1 in the linear response range in less than 5 μL of a serum (Supplementary Figure S1). Although the lectin microarray technique is similar to other microarray techniques, it differs in terms of the use of lectin-immobilized microarray chips. In contrast to the interaction between an antibody and antigen, affinity of lectins for their ligands is low in general (Kd > 10−6 M).38 However, it is known that the affinity of lectins binding to multivalent glycans or glycoproteins is often improved by placement of multiple binding sites or subunits.39,40 The lectin microarray with lectin immobilization detects multiple binding subunits resulting in strong affinity. Thus, the glycan profiling of MUC1 was achieved more effectively compared with the other techniques described above (Supplementary Table S2). We have demonstrated, for the first time, the highly sensitive glycan profiling of MUC1 in a tiny amount of serum using this lectin microarray. This methodology was applied to obtain an overview of the lectins that show the greatest differences of signal patterns between the CC patients and the other groups (hepatolithiasis and NCs). We performed differential glycan profiling of serum sialylated MUC1 using a set of 121 serum samples in cohort 2 (48 NCs, 40 hepatolithiasis patients, and 33 CC patients) and obtained 43 lectin signal intensities normalized relative to the mean signal intensity (Supplementary Table S3). Subsequent statistical analysis was performed to determine the most feasible lectin to correlate with the glycan change in sialylated MUC1 in CC patients. Specifically, we first

Figure 2. Dot plots and ROC curves of six lectins showing significant differences between patients with CC and those with hepatolithiasis in an antibody-overlay lectin microarray with MY.1E12 in cohort 2. (A) Dot plots show the results of antibody-overlay lectin microarray analysis with MY.1E12 of serum specimens from 48 NCs, 40 hepatolithiasis patients, and 33 CC patients. The whiskers indicate the highest and lowest values, and the box represents the interquartile range. The medians of the specimens are shown as the line across the box. Relative intensity represents the value normalized to the MPA signal. (B) Comparative analysis of ROC curve among six lectins. CC, cholangiocarcinoma; NC, normal control.

In particular, the patterns of AAL and WFA signals were better at distinguishing CC from hepatolithiasis patients compared with the other four lectins (p < 0.0001). WFA discriminated CC patients from NCs (p < 0.0001), but AAL did not (p = 0.0652). As further study, we performed the ROC curve analysis of the six lectins (Figure 2B), which showed clearly that the AUC of WFA (0.770) was the best among them. These results indicate that WFA is the most feasible lectin probe to distinguish CC from hepatolithiasis patients and NCs, and that the detection of WFA-sialylated MUC1 showed sufficient potential as a serological marker for CC diagnosis. Furthermore, the ultimate advantage of our glycan profiling is the applicability of multiple lectins, as described previously.35 The results of the glycan profiling showed that the combination of WFA and WGA produced a better statistical score for distinguishing CC samples from other samples compared with WFA or WGA alone (Supplementary Figure S3). We have reported previously that WFA was the best probe lectin based on the results of CC tissue glycome analysis, and that WFAsialylated MUC1 was a feasible CC diagnostic marker in bile specimens.27 To differentiate the target diseases, however, it is 7277

DOI: 10.1021/acs.analchem.5b01329 Anal. Chem. 2015, 87, 7274−7281

Article

Analytical Chemistry

high compared with that for total bilirubin (0.789). The levels of WFA-sialylated MUC1 and CA19-9 were significantly higher in O. viverrini-positive CC patients compared with those in opisthorchiasis patients (p < 0.0001, Figure 3). WFA-sialylated MUC1 and CA19-9 were suggested to feasible as CC markers. Comparison of WFA-Sialylated MUC1 and Other Markers in Serum from CC and Hepatolithiasis Patients. Because the feasibility of WFA-sialylated MUC1 as a CC marker was indicated in plasma (Figure 3), it was further confirmed using serum specimens from another cohort of 33 CC and 40 hepatolithiasis patients in cohort 2. The serum concentrations of WFA-sialylated MUC1 and conventional markers CA19-9, CYFRA21-1, and CA242 were measured and assessed for the association with the presence of CC by univariate and multivariate analyses (Figure 4 and Supple-

common that the selected best probe lectins are different between the glycome profile of the crude samples and the glycan profile of a single target glycoprotein.35,36,42,43 Therefore, it is essential to search for the optimal lectin for differential glycan profiling focused on a single target glycoprotein.44 Furthermore, selection of new lectin probe that is most appropriate for serum analysis was an important key to discriminate between CC and other controls in this study because it is quite difficult to specifically detect the serum glycoproteins secreted from CC tissues compared with bile specimens. We thus tried to identify the targeted glycan profiling of MUC1 for application in serodiagnosis, which resulted in the “reselection” of WFA as the best lectin probe to distinguish CC from the NC and other patient groups. This result agrees with our previous glycome analysis of serum, in which WFA was categorized as a minor lectin.36 In addition, successful assay and analysis of serum WFA-sialylated MUC1 are reported for the first time. Comparison of WFA-Sialylated MUC1 and Other Markers in Plasma from O. Viverrini-Positive CC and Opisthorchiasis Patients. The diagnostic value of WFAsialylated MUC1 was verified in comparison with other variables. The concentration of WFA-sialylated MUC1 in the plasma specimens was measured in 78 age- and sex-matched O. viverrini-positive CC and 78 opisthorchiasis patients without CC in cohort 1 by the WFA-immobilized MY.1E12 sandwich ELISA and assessed for the association with the presence of O. viverrini-positive CC by univariate and multivariate analysis (Figure 3 and Supplementary Table S5). Univariate analysis

Figure 4. Dot plots and ROC curves of serum WFA-sialylated MUC1, CA19-9, CYFRA21-1, and CA242 levels to distinguish CC from hepatolithiasis in cohort 2. (A) The levels of serum WFA-sialylated MUC1, CA19-9, CYFRA21-1, and CA242 are shown in dot plots. The values that differentiated CC from hepatolithiasis patients were identified by the Mann−Whitney U test. The whiskers indicate the highest and lowest values, and the box represents the interquartile range. The medians of the specimens are shown as the line across the box. (B) Comparative analysis of ROC curve among each marker. CC, cholangiocarcinoma.

mentary Table S6). The univariate analysis showed significant differences between CC and hepatolithiasis in some variables. However, the multivariate analysis showed that the levels of WFA-sialylated MUC1 (odds ratio: 1.018, 95% CI: 1.005− 1.031) and CA19-9 (odds ratio: 1.003, 95% CI: 1.001−1.006) were independently associated with the presence of CC. The levels of WFA-sialylated MUC1 and CA19-9 were significantly higher in CC patients compared with those in hepatolithiasis patients (p = 0.0002) (Figure 4A, upper panel). By contrast, although the CYFRA21-1 and CA242 levels differed significantly, their p-values (0.0022, 0.0040) were smaller than those of WFA-sialylted MUC1 and CA19-9 (Figure 4A, lower panel). In the ROC curve analysis, the AUC values for WFA-sialylated MUC1, CA19-9, CYFRA21-1, and CA242 were 0.738, 0.759, 0.720, and 0.696, respectively (Figure 4B). The AUC value of WFA-sialylated MUC1 was 0.859 for the WFA-MY.1E12 ELISA that included 48 NCs (Supplementary Figure S5). These results suggest that WFA-sialylated MUC1 is a feasible serological marker to discriminate CC from hepatolithiasis. For clinical application, the developed markers must be detectable by simpler methods using a single lectin such as the lectin−

Figure 3. Dot plots of plasma WFA-sialylated MUC1 and CA19-9 levels to distinguish CC from opisthorchiasis patients in cohort 1. The levels of plasma WFA-sialylated MUC1 and CA19-9 are shown in the dot plots. The values that differentiated CC from opisthorchiasis patients were identified by the Mann−Whitney U test. The whiskers indicate the highest and lowest values, and the box represents the interquartile range. The medians of the specimens are shown as the line across the box. WFAMY, WFA-MY.1E12 ELISA. CC, cholangiocarcinoma.

showed significant differences between samples from O. viverrini-positive CC and opisthorchiasis patients in some variables; however, multivariate analysis showed that the levels of WFA-sialylated MUC1 (odds ratio: 1.047, 95% confidence interval [CI]: 1.022−1.072), CA19-9 (odds ratio: 1.003, 95%, CI: 1.001−1.005), and total bilirubin (odds ratio: 1.022, 95% CI: 1.006−1.038) were independently associated with the presence of O. viverrini-positive CC. The ROC curve analysis was performed to evaluate the significance of the three markers (Supplementary Figure S4). The AUC values for WFAsialylated MUC1 (0.841) and CA19-9 (0.849) were remarkably 7278

DOI: 10.1021/acs.analchem.5b01329 Anal. Chem. 2015, 87, 7274−7281

Article

Analytical Chemistry

combination of WFA-sialylated MUC1 and CA19-9 levels seems to be a feasible method for the reliable diagnosis of CC. Although CA19-9 is used as a conventional serological marker to detect CC, it has some potential problems, as discussed in the Introduction. Improvements in CC prognosis largely rely on the identification of new, more precise CC markers that are superior to CA19-9 or that can address the problems of CA199. There was a low correlation between WFA-sialylated MUC1 and CA19-9 levels (Supplementary Figure S7), indicating that our marker has different characteristics from those of CA19-9. Although 5 of the 33 CC patients in Japan and 11 of the 78 CC patients in Thailand were completely Lewis negative (i.e., serum CA19-9 was not detectable), the serum or plasma levels of WFA-sialylated MUC1 in 11 of these 16 patients were higher than the cutoff value (Supplementary Table S7 and S8). Thus, the combined use of WFA-sialylated MUC1 with CA19-9 would yield superior statistical values for the serological CC diagnostic test. Recently, the combined diagnosis using several biomarkers or methods has been reported and is now used routinely.45−48 For instance, the combination of transient elastography with a serological marker-based algorithm is recommended for distinguishing severe fibrosis−cirrhosis (F3− F4) from the other stages (F0−F2) of hepatitis.45−47 Another recent study attempted to improve CC diagnosis using serum CA19-9 level in combination with other indices.11,15 However, none of the indices satisfactorily compensated for the disadvantage of CA19-9 discussed above. We previously studied WFA-sialylated MUC1 in combination with WFA-positive L1CAM (WFA-L1CAM) in bile. However, we could not analyze this combination in serum, because the serum concentration of WFA-L1CAM is very low compared to that in bile, and not detectable by currently available technology.48 As far as we know, this combinatorial test of WFA-sialylated MUC1 and CA19-9 is the most sensitive for identifying CC based on serum markers. In addition, we demonstrated that the combinatorial test was useful for diagnosis of CC arising from hepatolithiasis as well as that arising from opisthorchiasis. Although verification of the test to identify CC arising from PSC is needed, this combinatorial test has a potential for CC diagnosis worldwide. After the establishment of the assay system, further validation with a much larger cohort including other cancers will be performed in the near future.

antibody sandwich ELISA. In this study, serum WFA-sialylated MUC1 was detected successfully with the optimized WFAMY.1E12 sandwich ELISA, which was modified in a previous study for detecting WFA-sialylated MUC1 in bile.27 Our ELISA enabled highly sensitive detection of only 1 μL of multiple serum samples even in NCs because of the high-density lectincoated plates for improvement of the lectin−glycan interactions. In addition, each set of samples was run independently using several different batches of ELISA plates, and the repeat analyses by independent investigators showed that the expression patterns were highly reproducible (Supplementary Figure S6). Comparative Analysis of Combined Markers in CC Diagnosis. Serum and plasma WFA-sialylated MUC1 and CA19-9 showed greater ability to discriminate CC patients from hepatolithiasis and opisthorchiasis patients compared with the other biological indicators tested (Figures 3 and 4). A comparison of diagnostic performance with combinational analysis was performed using these markers in cohort 2 (Table 1 and Figure 5). In the ROC curve analysis, the AUC value of Table 1. Comparative Analysis of Combined Markers in CC WFAMY + CA19-9 WFAMY + CYFRA21-1 WFAMY + CA242 CA19-9 + CYFRA21-1 CA19-9 + CA242 CYFRA21-1 + CA242

AUC

odds ratio (95% CI)

0.842 0.789 0.798 0.751 0.811 0.751

1.005 (1.001−1.009) NS NS NS NS NS

AUC, area under the ROC curve; CI, confidence interval; WFAMY, WFA-MY.1E12 ELISA; CA19-9, carbohydrate antigen 19-9; CYFRA21-1, cytokeratin 19 fragments; CA242, carbohydrate antigen 242; NS, not significant.



CONCLUSION We established a highly sensitive, high-throughput method for glycanprofiling targeting to a slight amount of MUC1 from serum and identified WFA-sialylated MUC1 as a highperformance serodiagnostic marker of CC. Our marker would improve serological diagnostic performance CC in combination with CA19-9. Combinational use of several complementary markers is essential to the precise detection of CC.

Figure 5. Combinational analysis with WFA-sialylated MUC1 and CA19-9 in cohort 2. (A) Dot plot of the combinational analysis with two markers among NC, hepatolithiasis, and CC patients. The whiskers indicate the highest and lowest values, and the box represents the interquartile range. The medians of the specimens are shown as the line across the box. (B) ROC curve analysis of the combinational analysis with two markers between hepatolithiasis and CC patients. CC, cholangiocarcinoma.



ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.5b01329.

the combination of WFA-sialylated MUC1 and CA19-9 levels (0.842) was the best of all combinations and superior to the single-marker detection. The multivariate analysis showed that the combination of WFA-sialylated MUC1 and CA19-9 levels was independently associated with the presence of CC (odds ratio: 1.005, 95% CI: 1.001−1.009). The clinical value enabled us to identify 28 of 33 (85%) CC patients. Thus, the



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. 7279

DOI: 10.1021/acs.analchem.5b01329 Anal. Chem. 2015, 87, 7274−7281

Article

Analytical Chemistry Author Contributions

(22) Iemura, A.; Maruiwa, M.; Yano, H.; Kojiro, M. J. Hepatol. 1992, 15, 288−298. (23) Koike, N.; Todoroki, T.; Kawamoto, T.; Yoshida, S.; Kashiwagi, H.; Fukao, K.; Ohno, T.; Watanabe, T. Int. J. Oncol. 1998, 13, 1269− 1274. (24) Takeuchi, H.; Kato, K.; Denda-Nagai, K.; Hanisch, F. G.; Clausen, H.; Irimura, T. J. Immunol. Methods 2002, 270, 199−209. (25) Kuno, A.; Kato, Y.; Matsuda, A.; Kaneko, M. K.; Ito, H.; Amano, K.; Chiba, Y.; Narimatsu, H.; Hirabayashi, J. Mol. Cell. Proteomics 2009, 8, 99−108. (26) Tateno, H.; Toyota, M.; Saito, S.; Onuma, Y.; Ito, Y.; Hiemori, K.; Fukumura, M.; Matsushima, A.; Nakanishi, M.; Ohnuma, K.; Akutsu, H.; Umezawa, A.; Horimoto, K.; Hirabayashi, J.; Asashima, M. J. Biol. Chem. 2011, 286, 20345−20353. (27) Matsuda, A.; Kuno, A.; Kawamoto, T.; Matsuzaki, H.; Irimura, T.; Ikehara, Y.; Zen, Y.; Nakanuma, Y.; Yamamoto, M.; Ohkohchi, N.; Shoda, J.; Hirabayashi, J.; Narimatsu, H. Hepatology 2010, 52, 174− 182. (28) Akobeng, A. K. Acta Paediatr. 2007, 96, 644−647. (29) Zweig, M. H.; Campbell, G. Clin Chem. 1993, 39, 561−577. (30) Shigemura, M.; Konno, S.; Nasuhara, Y.; Shijubo, N.; Shimizu, C.; Nishimura, M. Clin. Chim. Acta 2013, 424, 148−152. (31) Storr, S. J.; Royle, L.; Chapman, C. J.; Hamid, U. M.; Robertson, J. F.; Murray, A.; Dwek, R. A.; Rudd, P. M. Glycobiology 2008, 18, 456−462. (32) Chen, S.; LaRoche, T.; Hamelinck, D.; Bergsma, D.; Brenner, D.; Simeone, D.; Brand, R. E.; Haab, B. B. Nat. Methods 2007, 4, 437− 444. (33) Chen, K.; Gentry-Maharaj, A.; Burnell, M.; Steentoft, C.; Marcos-Silva, L.; Mandel, U.; Jacobs, I.; Dawnay, A.; Menon, U.; Blixt, O. J. Proteome Res. 2013, 12, 1408−1418. (34) Meany, D. L.; Hackler, L.; Zhang, H.; Chan, D. W. J. Proteome Res. 2011, 10, 1425−1431. (35) Kuno, A.; Ikehara, Y.; Tanaka, Y.; Angata, T.; Unno, S.; Sogabe, M.; Ozaki, H.; Ito, K.; Hirabayashi, J.; Mizokami, M.; Narimatsu, H. Clin. Chem. 2011, 57, 48−56. (36) Kuno, A.; Ikehara, Y.; Tanaka, Y.; Ito, K.; Matsuda, A.; Sekiya, S.; Hige, S.; Sakamoto, M.; Kage, M.; Mizokami, M.; Narimatsu, H. Sci. Rep. 2013, 3, 1065. (37) Hirabayashi, J.; Yamada, M.; Kuno, A.; Tateno, H. Chem. Soc. Rev. 2013, 42, 4443−4458. (38) Blixt, O.; Head, S.; Mondala, T.; Scanlan, C.; Huflejt, M. E.; Alvarez, R.; Bryan, M. C.; Fazio, F.; Calarese, D.; Stevens, J.; Razi, N.; Stevens, D. J.; Skehel, J. J.; van Die, I.; Burton, D. R.; Wilson, I. A.; Cummings, R.; Bovin, N.; Wong, C. H.; Paulson, J. C. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 17033−17038. (39) Jégouzo, S. A.; Quintero-Martínez, A.; Ouyang, X.; dos Santos, Á .; Taylor, M. E.; Drickamer, K. Glycobiology 2013, 23, 853−864. (40) Coombs, P. J.; Harrison, R.; Pemberton, S.; Quintero-Martinez, A.; Parry, S.; Haslam, S. M.; Dell, A.; Taylor, M. E.; Drickamer, K. J. Mol. Biol. 2010, 396, 685−696. (41) Kuno, A.; Ikehara, Y.; Tanaka, Y.; Saito, K.; Ito, K.; Tsuruno, C.; Nagai, S.; Takahama, Y.; Mizokami, M.; Hirabayashi, J.; Narimatsu, H. Clin. Chim. Acta 2011, 412, 1767−1772. (42) Ocho, M.; Togayachi, A.; Iio, E.; Kaji, H.; Kuno, A.; Sogabe, M.; Korenaga, M.; Gotoh, M.; Tanaka, Y.; Ikehara, Y.; Mizokami, M.; Narimatsu, H. J. Proteome Res. 2014, 13, 1428−1437. (43) Sogabe, M.; Nozaki, H.; Tanaka, N.; Kubota, T.; Kaji, H.; Kuno, A.; Togayachi, A.; Gotoh, M.; Nakanishi, H.; Nakanishi, T.; Mikami, M.; Suzuki, N.; Kiguchi, K.; Ikehara, Y.; Narimatsu, H. J. Proteome Res. 2014, 13, 1624−1635. (44) Narimatsu, H.; Sawaki, H.; Kuno, A.; Kaji, H.; Ito, H.; Ikehara, Y. FEBS J. 2010, 277, 95−105. (45) Wandall, H. H.; Blixt, O.; Tarp, M. A.; Pedersen, J. W.; Bennett, E. P.; Mandel, U.; Ragupathi, G.; Livingston, P. O.; Hollingsworth, M. A.; Taylor-Papadimitriou, J.; Burchell, J.; Clausen, H. Cancer Res. 2010, 70, 1306−1313.

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank Drs. Yuki Matsuno and Akihiko Kameyama for helpful discussion and SMME analysis, Dr. Masahiro Noshima for helpful discussion and valuable statistical comments, Dr. Hiroaki Tateno and Ms. Jinko Murakami for preparing the lectin array, Dr. Hideki Matsuzaki for helpful discussion, Dr. Makoto Ocho for helpful statistical analysis, Dr. Thiravud Khuhaprema for preparation of the plasma specimens in Thailand, and Ms. Azumi Takahashi for assistance in preparing this manuscript. This work was supported in part by a grant from the “Medical Glycomics Project” from New Energy and Industrial Technology Development Organization (NEDO).



REFERENCES

(1) Khan, S. A.; Thomas, H. C.; Davidson, B. R.; Taylor-Robinson, S. D. Lancet 2005, 366, 1303−1314. (2) Blechacz, B.; Komuta, M.; Roskams, T.; Gores, G. J. Nat. Rev. Gastroenterol. Hepatol. 2011, 8, 512−522. (3) Burak, K.; Angulo, P.; Pasha, T. M.; Egan, K.; Petz, J.; Lindor, K. D. Am. J. Gastroenterol. 2004, 99, 523−526. (4) Thamavit, W.; Bhamarapravati, N.; Sahaphong, S.; Vajrasthira, S.; Angsubhakorn, S. Cancer Res. 1978, 38, 4634−4639. (5) Haswell-Elkins, M. R.; Mairiang, E.; Mairiang, P.; Chaiyakum, J.; Chamadol, N.; Loapaiboon, V.; Sithithaworn, P.; Elkins, D. B. Int. J. Cancer 1994, 59, 505−509. (6) Pinlaor, S.; Ma, N.; Hiraku, Y.; Yongvanit, P.; Semba, R.; Oikawa, S.; Murata, M.; Sripa, B.; Sithithaworn, P.; Kawanishi, S. Carcinogenesis 2004, 25, 1535−1542. (7) Watanapa, P. Br. J. Surg. 1996, 83, 1062−1064. (8) Watanapa, P.; Watanapa, W. B. Br. J. Surg. 2002, 89, 962−970. (9) Kubo, S.; Kinoshita, H.; Hirohashi, K.; Hamda, H. World J. Surg. 1995, 19, 637−641. (10) Narimatsu, H.; Iwasaki, H.; Nakayama, F.; Ikehara, Y.; Kudo, T.; Nishihara, S.; Sugano, K.; Okura, H.; Fujita, S.; Hirohashi, S. Cancer Res. 1998, 58, 512−518. (11) Ker, C. G.; Chen, J. S.; Lee, K. T.; Sheen, P. C.; Wu, C. C. J. Gastroenterol. Hepatol. 1991, 6, 505−508. (12) Goonetilleke, K. S.; Siriwardena, A. K. Eur. J. Surg Oncol 2007, 33, 266−270. (13) Partyka, K.; Maupin, K. A.; Brand, R. E.; Haab, B. B. Proteomics 2012, 12, 2212−2220. (14) Chapman, M. H.; Sandanayake, N. S.; Andreola, F.; Dhar, D. K.; Webster, G. J.; Dooley, J. S.; Pereira, S. P. J. Clin. Exp. Hepatol. 2011, 1, 6−12. (15) Ni, X. G.; Bai, X. F.; Mao, Y. L.; Shao, Y. F.; Wu, J. X.; Shan, Y.; Wang, C. F.; Wang, J.; Tian, Y. T.; Liu, Q.; Xu, D. K.; Zhao, P. Eur. J. Surg Oncol 2005, 31, 164−169. (16) Hollingsworth, M. A.; Swanson, B. J. Nat. Rev. Cancer 2004, 4, 45−60. (17) Julian, J.; Carson, D. D. Biochem. Biophys. Res. Commun. 2002, 293, 1183−1190. (18) Hakomori, S. Curr. Opin. Immunol. 1991, 3, 646−653. (19) Remmers, N.; Anderson, J. M.; Linde, E. M.; DiMaio, D. J.; Lazenby, A. J.; Wandall, H. H.; Mandel, U.; Clausen, H.; Yu, F.; Hollingsworth, M. A. Clin. Cancer Res. 2013, 19, 1981−1993. (20) Yue, T.; Goldstein, I. J.; Hollingsworth, M. A.; Kaul, K.; Brand, R. E.; Haab, B. B. Mol. Cell. Proteomics 2009, 8, 1697−1707. (21) Wu, Y. M.; Nowack, D. D.; Omenn, G. S.; Haab, B. B. J. Proteome Res. 2009, 8, 1876−1886. 7280

DOI: 10.1021/acs.analchem.5b01329 Anal. Chem. 2015, 87, 7274−7281

Article

Analytical Chemistry (46) Castéra, L.; Vergniol, J.; Foucher, J.; Le Bail, B.; Chanteloup, E.; Haaser, M.; Darriet, M.; Couzigou, P.; De Lédinghen, V. Gastroenterology 2005, 128, 343−350. (47) Ziol, M.; Handra-Luca, A.; Kettaneh, A.; Christidis, C.; Mal, F.; Kazemi, F.; de Lédinghen, V.; Marcellin, P.; Dhumeaux, D.; Trinchet, J. C.; Beaugrand, M. Hepatology 2005, 41, 48−54. (48) Matsuda, A.; Kuno, A.; Matsuzaki, H.; Kawamoto, T.; Shikanai, T.; Nakanuma, Y.; Yamamoto, M.; Ohkohchi, N.; Ikehara, Y.; Shoda, J.; Hirabayashi, J.; Narimatsu, H. J. Proteomics 2013, 85, 1−11.

7281

DOI: 10.1021/acs.analchem.5b01329 Anal. Chem. 2015, 87, 7274−7281