Serum Nardilysin, a Surrogate Marker for Epithelial–Mesenchymal Transition, Predicts Prognosis of Intrahepatic Cholangiocarcinoma after Surgical Resection

This article is featured in Highlights of This Issue, p. 451

Intrahepatic cholangiocarcinoma (ICC) is the second most common primary liver cancer following hepatocellular carcinoma (HCC), accounting for 5% to 15% of all primary liver cancers (1–3). There are marked geographic variations in the incidence of ICC, with a higher incidence in East Asia, whereas the number of patients with ICC has been reported to be increasing worldwide (3, 4). The survival rate of patients with ICC is poor because of the late presentation of the disease and limited therapies. Although surgical resection is the only curative treatment, 30% to 40% of patients with ICC have surgical indications (3). Moreover, the recurrence rate after surgical resection is 50% to 60% and the 5-year overall survival (OS) rate after surgical resection is only 25% to 31% (3, 5, 6), highlighting the need to optimize adjuvant strategies. Recent evidence has suggested that adjuvant chemotherapy is associated with prolonged survival, particularly in some advanced cases (7, 8). However, there are no established methods to define patient subgroups that need adjuvant strategies. The preoperative measurement of serum tumor markers may identify high-risk patients; however, few studies have investigated biomarkers in patients with ICC possibly due to the difficulties associated with collecting large numbers of serum samples from patients with ICC (9–11).

Nardilysin (N-arginine dibasic convertase, NRDC) is a zinc peptidase of the M16 family that selectively cleaves dibasic sites (12, 13). NRDC exhibits widespread expression throughout the body, and regulates multiple biological processes, such as myelination (14), body temperature homeostasis (15), and insulin secretion (16). Although NRDC is a soluble cytosolic protein without an obvious signal peptide or nuclear localization signal, it shuttles between the cytoplasm and nucleus and is secreted via an as yet unknown mechanism (17). We identified NRDC as a specific binding partner of heparin-binding epidermal growth factor-like growth factor (HB-EGF). Our subsequent studies demonstrated that NRDC enhanced the ectodomain shedding of HB-EGF and other membrane proteins, such as TNFα, through the activation of disintegrin and metalloproteinase (ADAM) proteases (18, 19). In addition to its extracellular functions, we recently clarified the nuclear functions of NRDC as a transcriptional coregulator, which modulates the transcriptional activity of histone deacetylase 3 (HDAC3; ref. 20), peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC-1α; ref. 15), and islet-1 (16). Furthermore, NRDC is strongly expressed in several cancer types, promotes tumor growth (19, 21–23), and is associated with a poor prognosis (19, 22, 23), suggesting important roles for NRDC in tumor biology. Our recent findings also indicated the clinical usefulness of serum NRDC in specific clinical settings. For example, serum NRDC levels were associated with the survival outcomes of postoperative patients with HCC with hepatitis C (23).

In this study, we retrospectively investigated serum expression levels of NRDC in patients with ICC who underwent surgical resection to clarify whether serum NRDC has potential as a postoperative prognostic indicator. We also examined NRDC mRNA expression in surgical specimens resected from patients and the pathophysiologic role of NRDC in ICC using an RNA interference method in ICC cell lines.

This study was designed to investigate whether serum NRDC has the ability to predict the outcomes of patients with ICC after surgical resection. We analyzed serum NRDC levels in two independent cohorts: a primary cohort at Kyoto University and an external validation cohort at Hyogo College of Medicine. In the primary cohort, we analyzed 79 consecutive patients with ICC who underwent surgical resection at Kyoto University Hospital (Kyoto, Japan) between January 2006 and July 2013. In the validation cohort, we analyzed 19 consecutive patients with ICC who underwent surgical resection at Hyogo College of Medicine (Nishinomiya, Japan) between January 2009 and January 2015. The final diagnosis of ICC was histologically confirmed. The follow-up data of the primary and validation cohorts were updated in April 2017 and February 2018, respectively. The surgical procedure in the primary cohort was reported previously (6, 7, 9). Serum samples from patients with ICC (n = 98) were obtained preoperatively at the time of admission. Fresh frozen cancer tissues were obtained from 43 of 79 patients with ICC in the primary cohort, from which mRNA was isolated. Surgical specimens from 20 of the 43 patients with ICC were also assessed by IHC.

Clinicopathologic and survival data were extracted from a prospectively maintained institutional database. Clinicopathologic data, including gender, age, hepatitis virus markers, the Child–Pugh classification, primary tumor characteristics, and treatment-related variables, were collected. Tumor characteristics and resection margins were ascertained based on a final pathological assessment. The tumor stage was assessed by the seventh edition of the American Joint Committee on Cancer (AJCC) classification (24). Post-surgical adjuvant chemotherapy was administered using gemcitabine (from 2006) and/or tegafur-gimeracil-oteracil potassium (S-1; from 2007) for tumors classified as stage II to IV according to the AJCC classification (from 2006). Recurrence was diagnosed based on imaging studies and tumor markers.

Written informed consent for the use of serum and resected tissue samples was obtained from all patients in accordance with the Declaration of Helsinki, and this study was approved by the institutional review committee of the Graduate School of Medicine, Kyoto University (approval code: R803-1) and Hyogo College of Medicine (approval code: 201807-010).

Serum was isolated from whole blood and stored at −80°C until analyzed. To quantify serum NRDC levels, an enzyme-linked immunosorbent assay (ELISA) was performed according to a previously described method (25). Briefly, to establish a sandwich ELISA system, all combinations of the 7 monoclonal antibodies for NRDC were tested, and the optimum combination of clone #231 for coating and #304 for detection was selected. An automated analyzer for the chemiluminescent enzyme immunoassay, SphereLight 180 (Olympus, Tokyo, Japan), was utilized to measure serum NRDC levels according to the manufacturer's protocol.

The human ICC cell lines HuCCT-1 and HuH28 were provided by the Japanese Collection of Research Bioresources Cell Bank (Osaka, Japan) and SSP-25 by the RIKEN Bioresource Center (Tsukuba, Japan). ICC cells were grown in RPMI1640 media supplemented with 10% FBS and 1% penicillin and streptomycin. The human colorectal cancer cell line HCT116 was provided by the ATCC and 293 T cells by the RIKEN Bioresource Center. These cells were growth in DMEM medium supplemented with 10% FBS and antibiotics. Cells were cultured at 37°C under 5% CO₂ and 95% relative humidity. These cells were incubated in serum-free medium for 24 hours before the initiation of experiments. To prepare condition medium (CM), confluent monolayers of ICC cells were incubated in RPMI1640 medium supplemented with 0.1% BSA for 48 hours. Cultured media were collected and centrifuged at 3,000 rpm at 4°C for 10 minutes, and the supernatant was harvested.

The gene knockdown (KD) of NRDC in ICC cells was performed by the transfection of lentiviral vectors expressing miRNA as previously reported (19, 23). The sequences targeting the NRDC gene were as follows: NRDC-KD1: 5′-CTGATGCAAACAGAAAGGAAA-3′; NRDC-KD2: 5′-GAGAAATGGTTTGGAACTCAA-3′. We used a control vector that contains a nontargeting sequence for any vertebrate gene as a negative control (NC). The efficiency of the gene KD of NRDC was evaluated by Western blotting and qRT-PCR analyses.

Cell proliferation was examined by a tetrazolium salt-based proliferation assay (WST-8 assay) using Cell Counting Kit-8 (CCK8; Dojindo Laboratories, Tokyo, Japan). Briefly, 96-well plates were seeded with cells at a density of 5,000 to 10,000 cells per well (HuCCT-1: 1.0 × 10⁴ cells, SSP-25: 5 × 10³ cell per well) and cultured for 72 hours. Ten microliters of CCK8 solution was added to each well and incubated at 37°C for 2 hours. The cell viability ratio was defined as the ratio of absorbance at 450 nm of KD and NC cells. All assays were performed in quadruplicate and repeated at least three times.

The migration of ICC cells was examined using a wound healing assay and chemotaxis assay as previously reported (26). In the wound healing assay, confluent ICC cells in 24-well plates were wounded using a sterile 200-μL pipette tip. Cells were then grown for an additional 12 and 24 hours with serum-free media. Wound closure was observed with an inverted microscope (Keyence) at 40× magnification. The cell migration distance was measured using ImageJ software (NIH) and compared with baseline measurements. All assays were repeated at least three times.

A chemotaxis assay was performed in 8-μm pore Transwell chambers (Corning Costar). Briefly, ICC cells (HuCCT-1: 5.0 × 10⁴ cells, SSP-25: 2.0 × 10⁴ cells) in serum-free media were placed in the upper chamber. The lower chamber was filled with 750 mL of RPMI1640 supplemented with 10% FBS as a chemoattractant. After the incubation at 37°C for 24 hours (HuCCT-1) and 12 hours (SSP-25), cells were fixed with 4% PFA at 20 minutes and stained with hematoxylin and eosin. Cells that migrated through the pores to the lower surface of the filter were counted under a microscope. A total of three random fields were counted in duplicate assays.

NC and NRDC-KD ICC cells (HuCCT-1: 1.0 × 10⁴ cells, SSP-25: 5.0 × 10³ cells per well) were seeded on 96-well plates in normal growth media. Twelve hours after seeding, gemcitabine was added at the indicated concentrations. After a 72-hour treatment, cell viability was examined using the WST-8 assay. The cell viability ratio was defined as absorbance at 450 nm of the sample divided by the absorbance of the control for each cell. All assays were performed in quadruplicate and repeated at least three times.

Seven- to nine-week-old male nude mice were used as recipients for xenotransplantation. A total of 1.0 × 10⁶ of HuCCT-1-NC, -KD1, or –KD2 cells were suspended in 100 μL PBS and subcutaneously injected into the right (NC) and left (KD1 or KD2) flank of nude mice (n = 7, of each). Tumor sizes were measured every week after the inoculation. Tumor volumes were calculated using the formula: length × (width)² × 0.52. Animal experiments were performed in accordance with the protocols approved by the Institutional Animal Care and Use Committee of Kyoto University. Animal experiments were performed in accordance with the protocols approved by the Institutional Animal Care and Use Committee of Kyoto University.

Four-micrometer-thick sections were incubated with an anti-human NRDC mouse monoclonal antibody (#102, established in our laboratory) at 4°C overnight. Envision polymer (DAKO), which is a horseradish peroxidase-labeled polymer conjugated with an anti-mouse IgG antibody, was used as a secondary antibody according to the manufacturer's protocol. Color was developed with diaminobenzidine solution (DAKO), followed by counterstaining with hematoxylin.

Total RNA was isolated from HuCCT-1 and SSP-25 cells and 43 fresh frozen cancer tissues using TRIzol reagent (Thermo Fisher Scientific) and cleaned using the DNase set and RNeasy Mini Kit (Qiagen). cDNA generated by reverse transcription with the Omniscript RT Kit (Qiagen) was subjected to a RT-PCR assay using the Step One Plus Real-Time PCR System (Applied Biosystems) and Fast SYBR Green Master Mix (Applied Biosystems) according to the manufacturer's protocols. The average expression levels of the target genes were normalized against GAPDH using the 2^−ΔΔCt method. The ICC cell line (SSP-25) was used as a reference sample (i.e., 2^−ΔΔCt value of SSP-25 cells = 1) when analyzing human samples. The primers used in this experiment are listed in Supplementary Table S1.

Twenty micrograms of cell lysates were electrophoresed on a 10% SDS-PAGE (SDS from Wako; acrylamide from Bio-Rad) and transferred to a nitrocellulose membrane (GE Healthcare). The primary antibodies are listed in Supplementary Table S2. The blot was observed with EZ-Capture II (ATTO) visualized by ECL Prime (GE Healthcare). β-Actin was used as the loading control. When analyzing protein levels in CM, Ponceau 3R Stain Solution (Wako) was used as the loading control. In figures, representative images were selected from at least three independent experiments, in which similar results were obtained.

Data from human clinical samples were expressed as median values (range). Regarding continuous variables, data were expressed as a median (range), and compared using the Mann–Whitney U test. Categorical variables were expressed as a number (%) and compared using the χ² test or Fisher exact test where appropriate. A ROC analysis was performed to evaluate the discriminatory power of predictors. The area under the ROC curve (AUC) was calculated. DeLong test was used to compare the AUC. The cut-off values, sensitivity, and specificity of serum NRDC variables were assessed by the Youden index. Other cut-off values were evaluated by clinically relevant values (5, 6). Relationships between continuous variables were evaluated by the Spearman correlation test (the value was expressed as ρ). To label the strength of the relationship, for absolute values of ρ, 0 to 0.19 is regarded as very weak, 0.2 to 0.39 as weak, 0.40 to 0.59 as moderate, 0.6 to 0.79 as strong, and 0.8 to 1 as very strong (27). OS was calculated from the day of surgical resection to the date of death or end of the follow-up period, whereas disease-free survival (DFS) was calculated using the date of death or recurrence as the time of the terminal event according to the Kaplan–Meier method. Survival was compared using a generalized Wilcoxon test. A multivariate analysis was performed by Cox's regression (Step-wise backward model) for variables identified as significant in the univariate analysis. When collinearity was encountered, a choice was made based on the P value and clinical reasoning. All analyses were two-sided, and differences were considered significant when P < 0.05. Statistical analyses were performed using JMP ver. 12.1 software.

Data from in vitro and in vivo experiments were analyzed using the Student t test and expressed as means ± SD. All statistical analyses were performed using JMP ver. 12.1 software (SAS Institute).

In the primary cohort, serum NRDC levels were measured in 79 preoperative patients with ICC who consecutively underwent surgical resection at Kyoto University Hospital between 2006 and 2013. In this study population, most patients had advanced stage disease [AJCC stage III/IV, n = 52 (65.8%); ref. 24]. R0 resection was performed on 63 patients (79.7%) (Supplementary Table S3). Serum NRDC levels were significantly higher in patients with ICC than in healthy controls (HC) [median; 1627.4 pg/mL (range: 351.9–11318.7) vs. 539.8 pg/mL (range: 5.9–1184.2), P < 0.001; Fig. 1A].

We then examined whether serum NRDC levels had prognostic value in patients with ICC after surgical resection. The median follow-up period was 41.6 months (range: 0.1–127.5 months). The median OS time was 47.6 months, with 3- and 5-year OS rates of 57.0% and 42.3%, respectively. The cut-off value for serum NRDC was selected as 1627.4 pg/mL based on the highest accuracy in relation to an outcome (death) using an ROC analysis (AUC values of 0.688). Patients with ICC were divided into two groups according to this cut-off value: high serum NRDC (n = 40) and low serum NRDC groups (n = 39). High serum NRDC levels correlated with the presence of multiple tumors (P = 0.006; Table 1). OS was significantly shorter in patients with high serum NRDC levels than in those with low serum NRDC levels (P = 0.002, Fig. 1B). The median OS and 3- and 5-year survival rates of high and low serum NRDC groups were 31.0 months versus 85.6 months, 42.5% versus 71.8%, and 22.3% versus 62.6%, respectively. Moreover, DFS was significantly shorter in patients with high serum NRDC levels than in those with low serum NRDC levels (P = 0.002, Fig. 1C). The median DFS and 1- and 3-year DFS rates of the high and low serum NRDC groups were 9.4 months versus 23.0 months, 42.5% versus 71.8%, and 15.0% versus 43.6%, respectively.

To confirm the prognostic relevance of serum NRDC as a biomarker, we performed univariate and multivariate analyses by Cox's hazard model using six potential confounders (Table 2). Among known prognostic factors (4–6), lymph node (LN) metastasis (P = 0.001), vascular invasion (P = 0.033), and multiple tumors (P < 0.001) were poor prognostic factors for OS. LN metastasis (P = 0.042), multiple tumors (P < 0.001), and poor differentiation (P = 0.026) were poor prognostic factors for DFS. After the multivariate analysis, serum NRDC levels were maintained as independent prognostic factors for OS (P = 0.019) and DFS (P = 0.009).

To validate the prognostic impact of NRDC in patients with ICC, we analyzed serum NRDC levels in an external independent cohort of 19 patients with ICC who underwent surgical resection at Hyogo College of Medicine between 2009 and 2015 (Supplementary Table S4). The median follow-up period was 25.4 months (range: 3.8–84.3 months) and the median OS was 27.9 months, with 3- and 5-year OS rates of 35.1% and 23.4%, respectively. Serum NRDC levels were significantly higher in the validation cohort (median; 1330, range: 528–14597 pg/mL) than in HC, but were not significantly different from those in the primary cohort (Fig. 1A). OS and DFS were significantly stratified according to cut-off values (1295 pg/mL: AUC values of 0.729), which were selected independently in the validation cohort by an ROC analysis. The clinical backgrounds of the high and low serum NRDC groups and Kaplan–Meier curves for OS and DFS are shown in accordance with this independent cut-off value (Supplementary Table S5; Fig. 1D and E). These results reinforced the predictive value of serum NRDC in postoperative patients with ICC.

Carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9) are serum tumor markers that are commonly measured in patients with ICC (3, 5, 6). A correlation was not observed between serum NRDC and CEA or CA19-9 levels (Supplementary Fig. S1). A Kaplan–Meier curve analysis revealed that the elevated serum CEA (≥5 ng/mL) and CA19-9 (≥37 IU/ml) levels in patients with ICC correlated with a poor prognosis in the primary cohort (Supplementary Fig. S2). An ROC analysis in relation to the outcome (death) showed that the prognostic ability of preoperative serum NRDC levels (AUC: 0.688) was equivalent to that of serum CEA (0.569) and CA19-9 (0.671) levels (Supplementary Fig. S3). We additionally analyzed the prognostic value of the combination of 3 markers (Supplementary Table S6) and found that the combination of NRDC and CA19-9 provided the highest AUC value (0.756), which had a significantly stronger prognostic value than CA19-9 alone (Supplementary Fig. S4).

The upregulated expression of NRDC in cancer tissue has been reported to have a poor prognostic impact in several cancer types (19, 22, 23). In addition to serum NRDC levels, we examined NRDC expression in surgical specimens resected from patients with ICC. An IHC analysis showed the membranous, cytosolic, and nuclear expression of NRDC in the cancer epithelium (Fig. 2A), which was consistent with previous findings (12–20). We quantified the mRNA expression levels of NRDC in cancerous tissues from 43 patients with ICC and analyzed the relationship with matched serum NRDC levels. As expected, a correlation was observed between mRNA and serum NRDC levels (ρ = 0.413, P = 0.006, Fig. 2B), suggesting that ICC tumors are a potential source of serum NRDC.

We also assessed the prognostic value of NRDC mRNA levels in tumors. The cut-off values of mRNA levels were selected based on the highest accuracy in relation to the outcome of death (cut-off 2^−ΔΔCt value: 1.708, AUC value of 0.634) and were used to divide patients into two groups: low NRDC mRNA (n = 11) and high NRDC mRNA (n = 32) groups. Using these cut-off values (patient characteristics are shown in Supplementary Table S7), OS and time to recurrence were significantly shorter in patients with high NRDC mRNA levels than in those with low NRDC mRNA levels (Fig. 2C and D).

We examined NRDC expression in three ICC cell lines and found that all cell lines strongly expressed NRDC (Supplementary Fig. S5). NRDC protein levels in ICC cells were similar to those in colon cancer HCT116 cells (28) and higher than those in 293T cells (13). Therefore, to examine the pathophysiologic roles of NRDC in ICC cells, we performed gene KD experiments using two ICC cell lines with high malignant potential: HuCCT-1 (obtained from metastatic ascites; ref. 29) and SSP-25 (spindle cell-type with mesenchymal features; ref. 30). The sufficient silencing of NRDC expression in both cells was confirmed by Western blotting and qRT-PCR analyses (Fig. 3A). Secreted NRDC in CM was also clearly decreased by the gene KD (Fig. 3B). We initially evaluated cell proliferation using the tetrazolium salt assay and found that the proliferation of HuCCT-1 and SSP-25 cells was significantly decreased by the gene KD of NRDC (Fig. 3C). To further assess the impact of NRDC on cell growth, control and NRDC-knocked down HuCCT-1 cells were used in tumor xenograft experiments. Tumor growth after subcutaneous implantation was markedly less in cells with the gene KD of NRDC than in control cells (Fig. 3D). Because we confirmed the similar effects of two different siRNAs on in vitro and in vivo cell proliferation, one (KD2) was selected for further experiments. In two different assays (wound healing assay and chemotaxis assay), HuCCT-1 cells in which NRDC was knocked down (NRDC-KD2) showed significantly less migratory potential than control cells (NC; Fig. 3E and F). The similar inhibitory effect of NRDC KD on cell migration was also confirmed in SSP-25 cells (Supplementary Fig. S6). Furthermore, the influence of NRDC levels on the chemosensitivity of HuCCT-1 and SSP-25 cells to gemcitabine was investigated. In both cell lines, NRDC-KD2 cells exhibited greater sensitivity to gemcitabine than NC cells (Fig. 3G; Supplementary Fig. S6).

To gain mechanical insights into the tumor-promoting and chemoresistant potential of NRDC in ICC, we investigated gene profiles in ICC cell lines. An analysis by qRT-PCR revealed that several epithelial–mesenchymal transition (EMT)- and cancer stem cell (CSC)-related genes were downregulated by the gene KD of NRDC in HuCCT-1 (Fig. 4A) and SSP-25 cells (Fig. 4B). For example, the mRNA levels of vimentin (VIM), a marker of mesenchymal cells, zinc finger E-box-binding homeobox 1 (ZEB1), an EMT-inducing transcription factor (EMT-TF), sex-determining region Y-box 2 (SOX2), a marker of CSC, and hypoxia-inducible factor-1α (HIF1A), a trigger of the EMT pathway, were downregulated by the gene KD of NRDC. Other EMT-TF, SNAI1 and TWIST1, were also significantly decreased by NRDC KD in SSP-25 cells (Fig. 4B). The Western blot analysis revealed that the protein expression levels of these genes were markedly reduced (Fig. 4A and B), whereas E-cadherin (CDH1), a marker of epithelial cells, was increased by NRDC KD in HuCCT-1 cells (Fig. 4A).

We then investigated whether relationships existed between the mRNA levels of NRDC and EMT-related genes in tumor tissues from patients with ICC. Notably, the expression levels of NRDC positively and strongly correlated with two EMT-TFs, ZEB1 (ρ = 0.679, P < 0.001) and SNAI1 (ρ = 0.647, P < 0.001), and HIF1A expression (ρ = 0.721, P < 0.001; Fig. 4C). NRDC mRNA levels correlated with SOX2 expression, but not with TWIST1 expression or the VIM/CDH1 ratio (Supplementary Fig. S7). We then examined the relationship between serum NRDC and mRNA levels of EMT-related genes in tumor tissues. Serum NRDC positively correlated with SNAI1 (ρ = 0.327, P = 0.032) and HIF1A (ρ = 0.301, P = 0.050) mRNA levels in surgical specimens. ZEB1 mRNA levels were also positively associated with serum NRDC levels (ρ = 0.287, P = 0.062; Fig. 4D). Therefore, serum NRDC has potential as a surrogate marker for EMT in the tumors of patients with ICC.

This study highlights the clinical implications of preoperative serum NRDC measurements in patients with ICC, which may contribute to the identification of patients with a poor prognosis. OS and DFS were significantly stratified by serum NRDC levels in the primary (development) and validation cohorts. The expression levels of NRDC in sera and cancerous tissues were significantly linked; therefore, an evaluation of pathophysiologic features in resected tissues provided an insight into the clinical significance of serum NRDC. NRDC mRNA levels correlated with EMT-related genes in primary tumors. Moreover, direct correlations were observed between serum NRDC and mRNA levels of major EMT-TF, suggesting that serum NRDC has potential as a surrogate marker for EMT features in primary tumors. This hypothesis was supported by a cell analysis because the gene silencing of NRDC in ICC cell lines reduced EMT-related gene expression. Functionally, the gene KD of NRDC was accompanied by attenuated proliferation/migration and increased chemosensitivity to gemcitabine. Importantly, a correlation was not noted between serum NRDC and CA19-9 or CEA, currently available prognostic markers for ICC. Furthermore, the combination of NRDC and CA19-9 had stronger prognostic value than either marker analyzed individually, suggesting that serum NRDC is a unique prognostic marker for patients with ICC after surgical resection.

Based on the prognostic impact of serum NRDC levels in patients with ICC after surgical resection, the source of NRDC in serum needs to be identified. According to the following findings of clinical studies: (i) serum NRDC levels were significantly higher in patients with ICC than in HC, (ii) a correlation between serum NRDC and NRDC expression levels in resected cancer tissue, we speculate that a potential source of serum NRDC may be the ICC tumor itself. Experiments using ICC cell lines also revealed that NRDC is secreted into CM, the amount of which correlated with the intracellular expression level of NRDC. Our previous analysis of patients with HCC also suggested that serum NRDC reflected the amount of NRDC in cancer tissues (23). Another possible source of serum NRDC is inflammatory cells adjacent to tumors. We recently reported that NRDC levels were increased in the synovial fluid of patients with rheumatoid arthritis (31). NRDC in macrophages regulates arthritis via the control of TNFα secretion because the macrophage-specific deletion of NRDC markedly ameliorated arthritis (31). We also demonstrated that inflammatory cells infiltrating the infarcted myocardium strongly express NRDC (25). However, clinical data from this study do not strongly support this hypothesis because preoperative C-reactive protein levels in ICC patients did not correlate with serum NRDC levels (data not shown). In any case, serial measurements of serum NRDC levels after surgical resection are needed to identify the real source, which will also further clarify the relationship between serum NRDC levels and tumor recurrence.

The significant link between patient prognosis and NRDC mRNA expression levels in cancerous tissues prompted us to examine the pathophysiologic functions of NRDC in ICC cells. The gene KD of NRDC in two different ICC cell lines attenuated cell proliferation and migration, indicating important roles for NRDC in ICC progression. Moreover, the gene KD of NRDC in HuCCT-1 cells resulted in increased sensitivity to gemcitabine. Because the poor prognosis of patients with ICC is mainly attributed to its highly metastatic characteristics, EMT in the pathogenesis of ICC has been attracting increasing attention from researchers (32). Accumulated evidence also suggests a relationship between chemoresistance and the acquisition of the EMT phenotype and/or existence of CSC within the tumor (33). Among EMT-related genes, the significance of EMT-TF, such as ZEB1, SNAI1, and TWIST1, has been emphasized (32–36). An analysis of surgical specimens also demonstrated that the strong expression of EMT-TF in resected tissue is associated with the poor prognosis of patients with ICC after surgical resection (32). We herein showed that the gene KD of NRDC was accompanied by marked reductions in several EMT-related genes. Moreover, strong correlations between NRDC and EMT-related genes (ZEB1, SNAI1, and HIF1A) were recapitulated in resected cancerous tissue from patients with ICC. Park and colleagues very recently demonstrated that NRDC regulated EMT-related genes, including SNAI1, in colon cancer cells (37); NRDC was responsible for the insulin-like growth factor-1 (IGF-1)-induced regulation of EMT-related genes. Together with our in vitro and in vivo data from ICC cells, NRDC appears to play important and general roles in the regulation of EMT.

Several signaling pathways triggered, for example, by TGFβ and EGF as well as hypoxia may induce EMT (35, 36). These signals lead to the activation of EMT-TF, such as ZEB1, SNAI1, and TWIST1, which directly or indirectly control key EMT-related genes. Because of the multiple functions of NRDC, there are several possibilities by which NRDC is involved in EMT. Extracellular NRDC may activate EGF or TNF receptor signaling by enhancing the ectodomain shedding of EGF receptor ligands or TNFα, respectively (12, 13, 18). Furthermore, nuclear NRDC may regulate the transcription of EMT-related genes, including EMT-TF (38, 39). Although the underlying mechanisms have not yet been elucidated in detail, the present results suggest that elevated NRDC levels in cancer tissue are associated with EMT programs. Based on the positive correlation between serum and tumor NRDC levels, it may be possible to assess the level of EMT features in primary tumors by measuring serum NRDCs.

In conclusion, this is the first study to demonstrate that serum NRDC has potential as a novel prognostic biomarker for ICC, which may reflect EMT features in primary tumor regions. We propose that the preoperative evaluation of serum NRDC is a potential clinical tool for predicting tumor recurrence in and the overall prognosis of patients after curative-intent surgery.

No potential conflicts of interest were disclosed.

Conception and design: T. Yoh, E. Hatano, S. Seo, E. Nishi

Development of methodology: T. Yoh, E. Hatano, Y. Kasai, R. Yamaguchi, M. Kurokawa, E. Nishi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Yoh, E. Hatano, H. Fuji, H. Sueoka, K. Iwaisako, K. Taura, R. Yamaguchi, J. Fujimoto

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T. Yoh, E. Hatano, Y. Kasai, M. Ohno, K. Iwaisako, K. Taura, E. Nishi

Writing, review, and/or revision of the manuscript: T. Yoh, E. Hatano, Y. Kasai, K. Taura, E. Nishi

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E. Hatano, K. Nishi, M. Ohno, J. Fujimoto, T. Kimura

Study supervision: E. Hatano, K. Toriguchi, K. Taura, T. Kimura, S. Uemoto

The authors gratefully acknowledge the patients who consented to blood collection for this study. The detection of serum NRDC was supported by Yoshiyuki Amano (Sanyo Chemical Industries). The authors thank Takayuki Kawai, Takahiro Nishio, Masayuki Okuno, Seidai Wada, Asahi Sato, Yoshinobu Ikeno, Yusuke Morita, Shintaro Matsuda, and Hiromi Iwai (Kyoto University) for their excellent help. This study was financially supported by Grants-in-Aid KAKENHI (17H04048, 17K09575, 17K16147, and 18H04694) and AMED (JP17cm0106608). It was also supported by the Takeda Science Foundation, SENSHIN Medical Research Foundation, and Suzuken Memorial Foundation.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.