Experiments using human materials were performed with the approval of the Ethics Committee of Kanazawa University Graduate School of Medicine (Permit no. 1985-3). Protocols for animal studies were approved by the Committee of the Institute for Experimental Animals, Kanazawa University Advanced Science Research Center (Permit no. AP-173905).

The human cholangiocarcinoma cell lines, CCKS-1, TFK-1 and HUCCT-1, were used. CCKS-1 was established in our laboratory from abdominal metastasis of cholangiocarcinoma (moderately differentiated adenocarcinoma) (24). TFK1 and HUCCT-1 were provided by Cell Resource Center for Biochemical Research, Tohoku University, Sendai, Japan. CCKS-1 was grown in Dulbecco's modified Eagle's medium/F-12 (Gibco; Thermo Fisher Scientific, Inc.), and TFK-1 and HUCCT-1 were grown in RPMI-1640 medium (Gibco; Thermo Fisher Scientific, Inc.) with 10% fetal bovine serum and 1% antibiotic-antimycotic solution (Gibco; Thermo Fisher Scientific, Inc.) at 37°C in an atmosphere of 5% CO₂. To block CDK signaling, the multi-CDK inhibitor roscovitine (Cell Signaling Technology, Inc.) was used.

Cell proliferative activity was examined using the WST-1 assay (Roche Diagnostics GmbH) according to the manufacturer's instructions. Cells were seeded on a 96-well plate at a density of ~2×10³−1×10⁴ cells/well. Cells were treated with roscovitine at the concentrations of 2, 10 and 20 µM. After 24, 48 and 72 h following the treatment, the WST-1 reagent was added and incubated for 1 h before reading the plate. Absorbance at 450 nm was measured using a microplate reader. Each assay was conducted in eight sets.

Cells were treated with 20 µM roscovitine for 24 h and processed for cell cycle analysis using the BrdU Flow Kit (Nippon Becton Dickinson Company, Ltd.) according to the manufacturer's instructions. Samples were analyzed using BD FACSCanto™ II Flow Cytometry (BD Biosciences). The percentages of cells in the G1, S, and G2/M phases of the cell cycle were assessed using BD FACSDiva^™ Software (BD Biosciences).

Cells were treated with 20 µM roscovitine for 72 h, and proteins were extracted using the T-PER Protein Extraction Reagent (Gibco; Thermo Fisher Scientific, Inc.). Protein concentrations were determined using the Bradfold method. Equal amounts of protein (50 µg) were loaded in each lane of an SDS-polyacrylamide gel (5–20%). Proteins were separated by electrophoresis and transferred to a nitrocellulose membrane. After blocking, the membrane was incubated at 4°C overnight with primary antibodies against p-CDK1 (Thr161; 1:500 dilution; cat. no. bs-3481R; rabbit polyclonal; Bioss Antibodies, Inc.), CDK1 (1:1,000 dilution; mouse monoclonal; cat. no. ab18; Abcam), and β-actin (1:1,000 dilution; cat. no. 4967). rabbit monoclonal; Cell Signaling Technology, Inc.). Following an incubation with the secondary antibody conjugated with peroxidase using Histofine Simple Stain MAX PO (Nichirei Corp.; cat. no. 424142) at room temperature for 1 h, the protein was visualized using 3,3′-diaminobenzidine tetrahydrochloride (DAB). Image analysis was performed using the NIH ImageJ software (National Institutes of Health). Each assay was conducted in five sets.

The protein levels of p-CDK1 (Thr161) and total CDK1 in cholangiocarcinoma cell lines were determined using an enzyme-linked immunosorbent assay (ELISA). The same samples used for the western blotting were analyzed, and the protein levels were determined using a RayBio^® Human phospho-CDK1 (Thr161) and Total CDK1 ELISA kit (RayBiotech) according to the manufacturer's instructions. Optical density was measured using a microplate reader at 450 nm, and the ratio of p-CDK1/CDK1 was calculated for each sample. Each assay was conducted in six sets.

DNA fragmentation attributable to apoptosis was assessed by the terminal deoxynucleotide transferase-mediated dUTP nick end-labeling (TUNEL) method. Cells were treated with 20 µM roscovitine for 72 h and processed in a paraffin-embedded cell block. After deparaffinization, sections were stained using a commercial kit (in situ Apoptosis Detection kit; Takara Bio, Inc.) according to the manufacturer's instructions. Color development was performed using DAB, and sections were counterstained with hematoxylin. Five high-power fields were randomly selected under a light microscope, and the percentage of apoptotic cells was calculated.

Carcinoma cell invasion was examined using the Corning^® BioCoat^™ Matrigel^® Invasion Chamber (Corning, Inc.) according to the manufacturer's instructions. Cells (~2-15×10⁴ cells/well) in a 0.5-ml suspension with or without 20 µM roscovitine were seeded in each well. Cells were incubated at 37°C for 72 h and then fixed at room temperature for 12 h with 10% neutral formalin. Cells were stained with Diff-Quick (Sysmex Corp.) at room temperature for 15 sec. Migrated cells were visualized under a light microscope, and the number of cells was counted in five randomly selected high-power fields.

Five-week-old BALB/cAnNCrj-nu/nu nude mice (male) were obtained from Charles River Laboratories, Inc. CCKS-1 was implanted subcutaneously into the right flanks of mice. On day 20, tumor-bearing mice were randomly divided into two groups (control and treatment groups), and five mice were included in each group. Roscovitine (ChemScene) was administered to mice by an intraperitoneal injection at a dose of 50 mg/kg for two 5-day series with a 2-day break in between. Control mice were treated with carrier solution alone. Tumor sizes were measured every day, and the tumor volumes were calculated using the following formula: (1/2) LxW², where L and W represented the longest tumor axis (mm) and the shortest tumor axis (mm), respectively. The fold difference from the initial tumor volume (measured at day 20) was defined as the tumor growth rate (25).

On day 33 (13 days after the first injection of roscovitine), the mice were sacrificed by inhalation of CO₂ with a 20% volume/min gas displacement flow rate until euthanasia was confirmed by the arrest of heartbeat and breathing. Formalin-fixed paraffin-embedded sections of dissected tumors were used for the apoptosis assay and the immunostaining of Ki-67.

Fifty-four cases of extrahepatic cholangiocarcinoma collected from Kanazawa University Hospital and its affiliated hospitals were examined. Sample collection was conducted during the periods from August 1994 to July 2016. All cases surgically resected were diagnosed as conventional cholangiocarcinoma. The median age of the patients was 69 years ranging from 35 to 84 years, and the male:female ratio was 33:21. The tumor-node-metastasis (TNM) classification was used according to the guidelines of the International Union Against Cancer (26). Cholangiocarcinoma cell lines (CCKS-1, TFK-1, and HUCCT-1) and tumor tissues obtained from a murine xenograft model were also used for the immunohistochemical analysis. Samples were fixed at room temperature for 48 h in 10% neutral formalin, embedded in paraffin, and 4-µm-thick paraffin-embedded sections were prepared.

After deparaffinization, antigen retrieval was performed by microwaving in 10 mmol/l citrate buffer (pH 6.0) for the immunostaining of p-CDK1, cyclin B1, and cyclin E1. Regarding the immunostaining of Ki-67, antigen retrieval was performed by heating in Tris-ethylenediaminetetraacetic acid buffer (pH 9.0) with a pressure cooker. Sections were immersed in 0.3% hydrogen peroxidase in methanol for 20 min to block endogenous peroxidase activity and then incubated in protein block solution (Dako; Agilent Technologies, Inc.). Sections were incubated overnight at 4°C with primary antibodies against p-CDK1 (Thr161; 1:100 dilution; cat. no. bs-3481R), rabbit polyclonal; Bioss Antibodies, Inc.), cyclin B1 (1:100 dilution; cat. no. ab32053; rabbit; monoclonal; Abcam), cyclin E1 (1:100 dilution; cat. no. ab33911; rabbit monoclonal; Abcam), or Ki-67 (prediluted; cat. no. 418071; rabbit monoclonal; Nichirei Corp.). Sections were then incubated with the secondary antibody conjugated with peroxidase using Histofine Simple Stain MAX PO [Nichirei Corp. (cat. no. 424142)] at room temperature for 1 h. Color development was performed using DAB, and sections were counterstained with hematoxylin.

The immunostaining of p-CDK2 was performed using the Novolink^™ Polymer Detection System (Leica Biosystems) according to the manufacturer's protocol. Briefly, after blocking endogenous peroxidase using Peroxidase Block for 5 min and a pretreatment with Protein Block for 5 min, sections were incubated at 4°C overnight with the primary antibody against p-CDK2 (Thr160; 1:100 dilution; rabbit polyclonal; Abcam cat. no. ab194868). Sections were then treated with Post Primary Block followed by Novolink Polymer at room temperature for 30 min. Immunoreactivity was visualized using DAB Working Solution, and sections were counterstained with hematoxylin.

A semiquantitative analysis of the immunohistochemical staining of p-CDK1, p-CDK2, cyclin B1, and cyclin E1 was performed for surgically resected specimens. Staining in the nuclei or cytoplasm of cholangiocarcinoma cells was individually evaluated, and staining in each section was graded as follows: (−), negative or <5% of carcinoma cells were positive; (+), >5% of carcinoma cells were positive.

The expression of Ki-67 was assessed by counting at least 1,000 cholangiocarcinoma cells in an area with the greatest staining density that was selected in an ×200 magnification field under a light microscope. The percentage of cells that were positive for Ki-67 was expressed as the Ki-67-labeling index.

Data were expressed as the mean ± standard deviation (SD). The significance of differences was analyzed using the Student's t-test and chi-squared test. In the univariate analysis, postoperative survival probability was calculated by the Kaplan-Meier method, and survival curves were compared by the Log-rank test using BellCurve^® for Excel software (Social Survey Research Information Co., Ltd.). A P-value <0.05 was considered to indicate a statistically significant difference.

The immunohistochemical expression of p-CDK1, p-CDK2, cyclin B1, and cyclin E1 was absent or weak in non-neoplastic biliary epithelial cells. In cholangiocarcinoma, the upregulation of protein expression was observed in a number of cases (Fig. 1). Positive immunohistochemical signals for CDKs and cyclins were observed in both the nuclei and cytoplasm of carcinoma cells, and the extent of positive staining varied from patchy to diffuse. Among the 54 cholangiocarcinoma cases examined, positive nuclear expression for p-CDK1, p-CDK2, cyclin B1, and cyclin E1 was observed in 29 cases (54%), 18 cases (33%), 32 cases (59%), and 34 cases (63%), respectively (Table I). No correlations were noted between the positivity of p-CDK1 and cyclin B1 or the positivity of p-CDK2 and cyclin E1 (data not shown).

The relationship between clinicopathological factors and the immunohistochemical expression of CDKs and cyclins was examined. The nuclear expression of p-CDK1 and cyclin B1 was positively associated with the presence of lymph node metastasis and the clinical stage (Table I). The nuclear expression of p-CDK2 and cyclin E1 was not associated with these clinicopathological factors. The expression of p-CDK1, p-CDK2, cyclin B1, and cyclin E1 in the cytoplasm of carcinoma cells was not associated with any clinicopathological factors (data not shown).

Postoperative follow-up data were analyzed with a median follow-up period of 21.8 months for the overall survival ratio and 18.0 months for the disease-free survival ratio. Patients with the positive nuclear expression of p-CDK1 had a significantly poor prognosis (Fig. 2). In contrast, no correlation was observed between patient prognosis and the nuclear expression of p-CDK2, cyclin B1, and cyclin E1 (Fig. 2). The cytoplasmic expression of p-CDK1, p-CDK2, cyclin B1, and cyclin E1 was not associated with patient prognosis (data not shown).

A group of disease-free surviving individuals was observed within the p-CDK1-positive cohort of patients. Disease-free surviving patients were also present in the p-CDK2-, cyclin B1-, and cyclin E1-negative groups. These surviving patients appeared to have cholangiocarcinoma with a well-differentiated histology.

The present results indicated that the activation of the CDK pathway involving CDK1 rather than CDK2 defined the tumor aggressiveness of cholangiocarcinoma. To further clarify their involvement in cholangiocarcinoma, in vitro experiments were performed.

The human cholangiocarcinoma cell lines, CCKS-1, TFK-1, and HUCCT-1 were used. Immunostaining revealed that p-CDK1 was expressed in the nuclei of all three cell lines, and its expression was stronger in CCKS-1 and HUCCT-1 than in TFK-1 (Fig. 3A). When cells were treated with the multi-CDK inhibitor roscovitine, cell proliferative activity was inhibited in the three cell lines in time- and dose-dependent manners, with 20 µM roscovitine exerting the strongest effects (Fig. 3B). In subsequent in vitro experiments, roscovitine was used at a dose of 20 µM.

The western blot analysis indicated that the treatment with roscovitine reduced the expression of p-CDK1 in the cells (Fig. 3C), and the analysis using ELISA further confirmed that the ratio of p-CDK1/CDK1 was significantly reduced by the treatment (Fig. 3D). Following treatment with roscovitine, the cell cycle was arrested at the G1 or G2/M phase in CCKS-1, TFK-1, and HUCCT-1 (Fig. 4), and this was accompanied by an increase in the percentage of apoptotic cells (Fig. 5A and B). Furthermore, the invasion of cholangiocarcinoma cells was significantly inhibited by the treatment with roscovitine (Fig. 5C and D).

The subcutaneous implantation of CCKS-1 was used in the analysis of the murine xenograft model. An intraperitoneal injection of roscovitine inhibited tumor growth in this model, and a significant difference was noted in the tumor growth rate between the untreated and treated groups between 10 and 13 days after roscovitine treatment (Fig. 6A and B). Tumors in the untreated group grew up to a volume of 847 mm³ (6.7-fold growth against the initial size), whereas tumors in the treated group had a maximum volume of 600 mm³ (3.0-fold growth against the initial size).

Similar to the results of the in vitro experiments, the apoptotic response of cholangiocarcinoma cells was significantly induced by the treatment (Fig. 6C and D). The Ki-67-labeling index of cholangiocarcinoma cells was significantly reduced by the treatment (Fig. 6C and E). Thus, roscovitine was an effective agent to treat cholangiocarcinoma in vivo.