Abstract
Background/Aim: Invasive papillary cholangio-carcinoma (IPC) is a minor subtype of extrahepatic cholangiocarcinoma. However, its etiology and characteristics remain unknown because of the unavailability of in vitro and in vivo models. We aimed to establish a novel preclinical model for translational research of IPC. Materials and Methods: A patient-derived xenograft (PDX) was engrafted in NOG mice and the cell line National Cancer Center human IPC (NCChIPC) was subsequently established from the PDX tumors. Immunohistochemistry and RNA-sequencing were used to determine the retention of original characteristics of patient tissues. Results: PDX tumors showed successful amplification, and the NCChIPC-derived xenograft largely retained the histopathological features of the original tumor with CK19, MUC1 and MUC5AC expression. Transcriptome analysis showed a high correlation between patient and preclinical models. Additionally, anticancer drugs response was analyzed in the NCChIPC PDX. Conclusion: These novel preclinical models here will help elucidate IPC etiology and facilitate translational research.
Invasive papillary cholangiocarcinoma (IPC) is a rare subtype of cholangiocarcinoma (CCA) that accounts for approximately 4% of all malignant epithelial tumors of the extrahepatic duct (1, 2). IPC is characterized by papillary proliferation of dysplastic biliary epithelium with delicate fibrovascular stalks and mucin production, which are designated as papillary CCA of intraductal growth (2-3).
Surgery is the only curative treatment for biliary tract cancers, including IPC. However, a limited number of patients are suitable for surgery, and most patients only report for treatment after the occurrence of locoregional and distant metastasis; thus, radical surgery or chemotherapy during advanced stages is rarely successful (4-9). No standard chemotherapeutic regimen has been established for inoperable cases or cases with recurrence after surgical resection (10, 11). The lack of a proper CCA experimental model is a major limitation despite the fact that the therapeutic outcome of CCA differs significantly depending on the degree of invasion and the choice of anticancer agent according to tumor subtyping. Currently, 15 different CCA cell lines are commercially available and several kinds of in vivo animal models have been reported (12, 13). However, cell lines are unable to retain the high heterogeneity of the CCA tumor, and animal models are difficult to approach considering cost and labor. Moreover, cell line establishment success rates are very low when cells are isolated directly from patient tissue due to stromal cell interference.
While a better prognosis for papillary CCA has been observed, treatment-related experience with this tumor is limited because it accounts for a minority of extrahepatic CCA (eCCA) cases (1, 14, 15). Considering the overall situation of CCA, a clear understanding of IPC etiology is lacking mainly due to unavailability of reliable models. The paucity of in vitro and in vivo models of IPC has hampered the development of new therapeutic modalities (15-17). Currently, there are two types of cell lines for papillary neoplasms of the bile duct: KBDC-11 for intraductal papillary neoplasm of bile duct and KKU-213 for mixed papillary CCA (18-19), that are used as a meaningful resource in CCA research. The etiology and characteristics of IPC is yet to be elucidated because of the absence of in vitro and in vivo models for IPC. Significant advances can be made by developing preclinical models that can replicate the native tumor environment and tumor heterogeneity. The establishment of a new cancer model is needed to provide precise insights into IPC-related molecular evolution patterns to determine its pathophysiology and design appropriate clinical treatment.
We aimed to establish useful in vivo and in vitro models that retain the intrinsic characteristics of the tumor and increase the success rate of model establishment. Patient-derived xenograft (PDX) models are an important missing component in the development of IPC therapeutics and would enable the examination of tumor tissue without affecting the heterogeneity, genomics, and architecture of IPC (17). Furthermore, the strategy was to establish a cell line from PDX tissue to increase the success rate of cell line establishment (19).
To the best of our knowledge, this is the first study to report the successful establishment of PDX and PDX-derived cell line for IPC using tissue obtained from a patient. Our results will help develop new suitable models for the translational and preclinical studies of IPC.
Materials and Methods
Ethics statement. The study was approved by the institutional review board (IRB) of the National Cancer Institute (NCC), and the patient provided written informed consent. All processes adhered to the tenets of the Declaration of Helsinki (IRB approval No: NCC-2015-0245). All animal studies were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) of the National Cancer Centre Research Institute (NCCRI) (NCC-16-313, NCC-21-313G). The NCCRI is a facility accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC) and abides by the guidelines of the Institute of Laboratory Animal Resources (ILAR) (Accredited unit-NCCRI: unit number 1392).
Patient details. Our study participant was a 71-year-old woman diagnosed with intraductal papillary neoplasm with an associated invasive carcinoma by pathological examination and imaging. The patient was admitted to the hospital for surgical resection in November 2016 with ECOG 1 and underwent surgery of the liver, extended right hemihepatectomy with bile duct resection, at the NCC. There was metastasis in one lymph nodes (1/10), the pathological stage was pT3N1, and the tumor was moderately differentiated (Table I).
Characteristics of the patient with invasive papillary cholangiocarcinoma included in this study.
Establishment of PDX from patient tissue. Female immunodeficient NOG mice aged 5-8 weeks (Harlan Laboratories, Inc., Indianapolis, IN, USA) were housed in a specific pathogen-free environment under controlled light and humidity conditions and were allowed food and water. The NOG mouse was anesthetized using 2% isoflurane in 100% oxygen. The patient tissue (F0) was removed from the medium and cut into approximately 3 mm3 pieces in a sterile petri dish with fresh medium and maintained on ice. To establish an F1 generation of PDX, a 5 mm horizontal incision was made on the flank of the mouse to create a subcutaneous pocket. a tumor piece mixed with growth medium and Matrigel® was inserted in the incision and sealed with a black silk suture. Povidone iodine was applied at the incision site. After completion of the procedure, the mouse was returned to the storage box and observed for complications. Tumor growth was monitored twice per week by measuring the tumor size using a caliper (Mitutoyo, Japan) and calculated using the formula: (Width2 × Length)/2. The mouse was euthanized with CO2 once the tumor size reached approximately 1,500 mm3. The tumor was then removed and triaged for cryopreservation and expansion in the secondary recipient mouse for establishing the F2 and F3 generations of PDX and for performing histological and molecular analyses.
H&E and immunohistochemical analysis. The tumor was fixed in formalin, embedded in paraffin, sectioned, and stained for histopathological assessment using hematoxylin and eosin (H&E). Immunohistochemical detection of CK19, MUC1, MUC2 and MUC5AC was performed using the rabbit anti-human MUC1 mAb clone: EPR 1025,1:3000 dilution, rabbit anti-human MUC2 mAb clone EPR 6145 1:5,000 dilution, rabbit anti-human MUC5AC mAb clone EPR 16904 1:500 dilution, and rabbit amt-human CK19 mAb clone Ep1580Y 1:400 dilution. Incubation with the respective antibodies was performed overnight at 4°C. Immunodetection was performed using an Envision Plus system (Dako, Carpinteria, CA, USA) with 3,3-diaminobenzidine (DAB/H2O2) chromogen. The immunostained sections were then counterstained with hematoxylin and coverslipped for microscopic assessment. All images were captured by Vectra® Polaris™ imaging system (PerkinElmer, Waltham, MA, USA).
Establishment of the human IPC cell line, NCChIPC. The cell line was established from the PDX F2. The tumor mass from F2 was minced and dissociated mechanically and chemically by GentleMACS™ dissociator (Milteny Biotec, Bergisch Gladbach, Germany) for 45 min. The cell suspension was washed by centrifugation and resuspended in Dulbecco’s modified Eagle’s medium (DMEM)/F-12 supplemented with epidermal growth factor (EGF) and the antibiotics zellshields™ (Minerva Biolabs, Berlin, Germany). The cell suspension was filtered through a strainer with a mesh size of 70 μm, suspended in a growth medium and incubated at 37°C and 5% CO2. The cells were seeded and cultured until the doubling time could be estimated. The cells were then trypsinized and counted. Primary cultured cells were observed periodically, and contamination with fibroblasts was aseptically removed by trypsinization until they were free of fibroblasts. The cultured cells were initially subcultured every week until they grew at a stable rate. When they reached 70% confluence using trypsin-EDTA (Invitrogen), the cells were cultured in growth media after a few passages of primary culture. Contamination with mycoplasma was monitored periodically using an e-myco™ Mycoplasma PCR detection kit [ver.2.0]. The culture remained free of mycoplasma during the experiments.
Cell proliferation assay. NCChIPC cells were seeded at a density of 5,000 cells/well (100 μl of suspended cells in enriched DMEM) in a 96-well plate. Cells suspended in growth medium were incubated in Incucyte Zoom (Sartorius, Essen Bioscience, Ann Arbor, MI, USA) at 37°C and 5% CO2. The cells were then monitored for 6 days and assayed for proliferation.
NCChIPC cell line-derived xenograft. To detect in vivo tumorigenicity, a mixture of NCChIPC cells (5.0×106) and 1:1 ratio of growth factor reduced Matrigel® was injected subcutaneously in an NOG mouse subcutaneously. The tumor growth was monitored twice per week. When the tumor size reached 1,500 mm3, the mouse was sacrificed, and the tissue samples were obtained and fixed in 10% phosphate-buffered formalin overnight. The tissue sample was then embedded in paraffin for histopathological evaluation.
Cytotoxicity assays for anticancer drugs. Gemcit® (Gemcitabine-HCl) and cisplatin were obtained from Dong-A ST CO.LTD, Seoul, Korea; albumin-bound paclitaxel (Abraxane®) was obtained from Celgene Corporation, NJ, USA; and Onivyde® (Irinotecan liposome injection) was obtained from Servier, France. 5-FU, oxaliplatin, erlotinib, epirubicin, carboplatin, and devimistat were purchased from MedChemExpress. For analyzing the drug responses, NCChIPC cells (4×103 cells/well) were seeded in 384-well plates and stabilized for 24 h, followed by incubation with the following drugs for 72 h: 0.001 μM to 10 μM of Gemcit®, albumin-binding paclitaxel (Abraxane®), and 0.01 μM to 100 μM of 5-FU, Onivyde®, oxaliplatin, cisplatin, erlotinib, epirubicin, caroboplatin, and devimistat. Cell cytotoxicity was measured using the Cell Titer-Glo® Viability assay kit (Promega Corporation, Madison, WI, USA). Plates were read using luminescence infinite 200 Pro (Life Sciences, Boston, MA, USA). Dose-dependent response was determined using the GraphPad Prism5 software.
Short tandem repeats. Short tandem repeat (STR) analysis was performed at 10 loci on different chromosomes to verify that the PDX, PDX-derived cells, and xenograft tissue F1, F2, and F3 samples were derived from the patient (F0). STR loci (TH01, D21S11, D5S818, D13S317, D7S820, D16S539, CSF1PO, AMEL, vWA, TPOX) amplification was performed using a Gene Print R10 system kit (Promega) according to the manufacturer’s instructions. Samples were run on an ABI 3730 DNA Analyzer (Thermo Fisher Scientific, Waltham, MA, USA) and analyzed using Gene Mapper v4.0.
RNA sequencing and data analysis. For library construction, total RNA from the frozen tissue samples or cell pellets was extracted using Trizol® (ThermoFisher Scientific), per the manufacturer’s protocol. RNA samples were pooled for RNA sequencing using TruSeq Stranded mRNA LT Sample Prep Kit (Illumina, San Diego, CA, USA) and sequenced using NovaSeq 6000 S4 Reagent Kit (Illumina). After performing a quality check using FastQC v0.11.7, spliced reads were mapped to the hg19 assembly using Bowtie 2 aligner 2.3.4.1. The fragments per kilobase of transcript per million mapped reads (FPKM) values were subsequently transformed to log2 values. The read count data were processed based on the TMM+CPM normalization method using EdgeR within R (R Development Core Team, Vienna, Austria) using Bioconductor (20). Data mining, data visualization, and statistical analysis were performed in Python version 3.9, Jupyter lab version 3.1.12 and Prism 5. Pearson’s correlation coefficients between tissues and NCChIPC were used to evaluate the correlation matrices.
Results
Establishment of PDX model for IPC. The overall strategy of the establishment of a preclinical model for IPC is summarized in Figure 1. The features of primary IPC patient tumors are shown in Figure 2A and B. To generate the PDX model, small pieces of tumor were engrafted on the left flank of a NOG mouse and tumor growth was monitored by palpating the site of the implant (Figure 2C). When the engrafted tumor reached 1,500 mm3, the mouse was sacrificed, and the tumor was removed by sharp dissection. Tissue from the first mouse was considered the first generation of PDX or F1 tissue. A second NOG mouse was then implanted with tissue from F1 to form the next generation in serial order. Tissue-derived xenograft morphology was analyzed using H&E staining. The retention of histopathological characteristics of xenografts derived from F1, F2, and F3 was determined by immunohistochemistry for IPC markers such as CK19, MUC1, 2, 5AC (Figure 2D). CK19 is well expressed in all xenograft tissues, similar to the parent tumor. MUC1 and MUC5AC are also expressed in all tissues; in contrast, MUC2 was not expressed in IPC, which was expected. Additionally, STR analysis was performed to confirm the paternity authentication for the presence of chromosomal aberrations in the PDX model compared to that in the original tumor. STR analysis at 10 loci demonstrated that PDX-derived models were unique and matched with the original patient tumor (Table II). This PDX model is a promising first report of IPC PDX.
Schematic diagram of the establishment of a preclinical model of invasive papillary cholangiocarcinoma (IPC) both in vitro and in vivo. The first step is the generation of a patient-derived xenograft (PDX) from the surgical tissue of IPC. Next, the primary patient tumor sample is engrafted in immunocompromised (NOG) mice. Then, the xenograft tumor was expanded in successive mice to develop the F2 and F3 passages. Thereafter, PDX tumor tissue-derived cells can be isolated (temporarily termed the NCChIPC cell line) and form tumor nodules with IPC characteristics by engraftment in immunodeficient mice, which can then be used as an in vivo model of IPC.
Establishment of IPC PDX model and retention of histopathological features of primary tumors by PDX tumors. (A) The patient abdominal CT image showed a bile duct defect as indicated by yellow arrowhead and (B) papillary tumor region of bile duct in patient surgical resected specimen is indicated by white circle. (C) The patient’s surgical tumor sample was subcutaneously engrafted in the flank of an immunocompromised NOG mouse for the generation of PDX F1. The tumor nodule is indicated by the yellow arrow. (D) Morphological and histological features between the patient (F0) and PDXs (F1-F3) are validated based on H&E staining and immunohistochemical analysis of CK19 and mucin subtypes. Histological and morphological features resembled the parent tumor. CK19 was expressed in all xenograft tissues similar to the parent tumor. MUC1 and MUC5AC were highly expressed in all the tissues. MUC2 were not expressed in IPC tissues. Scale bars=100 μm.
Short tandem repeat analysis performed at 10 loci on different chromosomes to verify that the PDX F1, F2, F3 and cell-derived xenografts were derived from the primary patient sample F0. PDX were consistent with F0 (patient).
Establishment of a new cell line from IPC. Cancer cells were isolated from PDX F2 tissue and showed an adherent epithelial-like morphology (Figure 3A). We termed these cell lines NCChIPC, human IPC at NCC. Proliferation monitored by analyzing the occupied cell area using the Incucyte™ machine revealed that the population doubling time was approximately 28-36 h (Figure 3B). We inoculated different anticancer drugs available for CCA and observed that Abraxane® was the most toxic, followed by epirubicin. Other drugs including Gemcit®, erlotinib, carboplatin, and 5-FU were less toxic (Figure 3C). These results suggest that the NCChIPC cell line can be used as an in vitro model for evaluating the drug response in IPC treatment
Establishment of the NCChIPC cell line for IPC. NCChIPC cells were isolated from the PDX tumor tissue. (A) The NCChIPC cell morphology of passage 10 in monolayer culture. (B) Proliferation curve of NCChIPC cells in culture medium through Incucyte. (C) Dose-response curve in NCChIPC according to different drugs. NCChIPC cells were seeded in a 384-well plate and incubated overnight. The cells were then exposed to drugs according to the indicated concentrations, there were different responses (toxicities) to different cancer medicines, with stronger effects of epirubicin, abraxane, gemcitabine and weaker effects with erlotinib. Error bars represent the standard deviation of three independent experiments.
Recapitulation of parental molecular characteristics in PDXs and NCChIPC cell line. To evaluate whether the established patient-derived preclinical models have retained the characteristics of the original tumor, we performed RNA sequencing to determine the similarity between the gene expression of the three generations of PDX (F1, F2, and F3) and NCChIPC cell lines with those of the F0 tissue (Figure 4A). Overall, the correlation of PDXs with F0 tissues showed very high similarity (94%-98%). Moreover, the NCChIPC cell line reflected about 81% of the patient’s transcriptomics characteristics. Next, we determined the RNA levels of 94 genes expressed in our library among 100 target genes corresponding to the proliferative or metabolic subclasses of eCCA, which was reported by Montal et al. (21) (Figure 4B). While only a few genes related to the metabolic class were overexpressed, most proliferation class-related genes were highly expressed and the level was retained not only in several passages of PDX generations but also in NCChIPC cells. These results suggest that our preclinical models for IPC reflect the genomic characteristics of the patient.
High correlation between patient (F0) and PDX (F1-F3) & NCChIPC cell line based on transcriptome evaluation. (A) A correlation matrix showing the Pearson correlation coefficient calculated from the expression of 27,686 genes (FPKM). (B) A heat-map showing the expression of genes related to two extrahepatic cholangiocarcinoma (eCCA) subclasses (proliferative and metabolic classes). The molecular subtype of patients with IPC was successfully engrafted as PDX and NCChIPC. Gene expression was analyzed by calculating the Z-score by using the normalized data.
Tumor formation by xenograft of NCChIPC cells and histological evaluation. Subcutaneous xenografts of NCChIPC cells successfully formed tumor nodules and exhibited a prominent papillary growing pattern of IPC morphologically by H&E staining despite of cell line implantation model (Figure 5). Immunohistochemical analysis of the tissues showed that they exhibited positive expression of CK19 and MUC5AC, which are the typical markers of IPC (Figure 5). These results strongly suggest that NCChIPC cell-derived in vivo models can be easily used in IPC research.
Tumor formation in vivo from subcutaneous implantation of the NCChIPC cell line in a NOG mouse. (A-B) The morphology of the NCChIPC cell-derived tumor xenograft (B) was similar to the IPC patient tumor tissue (A) as seen from H&E staining, revealing the papillary shape. (C-D) CK19 and MUC5AC were expressed in NCChIPC cell-derived xenograft tumor tissues. Scale bars=100 μm.
Discussion
IPC is a minor phenotype of CCA, for which preclinical models are difficult to establish. Nevertheless, it has been shown that preclinical models are key to basic and translational research, with the potential for invaluable assessment of human tumor biology, identification of therapeutic targets, and preclinical screening and evaluation of drugs for various cancers (22). Previously, cell line xenografts were used as standards in preclinical research; however, generally cell lines do not accurately reflect the true behavior of the host tumor and easily to adapt to in vitro growth, losing the original properties of the host tumor (23). PDXs are platforms that can represent the complexity and diversity of cancer and are known to preserve key biological properties of tumors from which they were derived and remain stable across passages (24, 25). These models can predict clinical outcomes and are useful for precision medicine. PDX models are established by engrafting patient tumor tissues in immunocompromised NOG mice and subsequently observing the passage of tumor cells from human tissues to the animal (23, 26). Xenografts derived directly from patients’ surgical tissue or biopsy samples with minimal in vitro manipulation retain the morphological and molecular markers of the source tumors despite serial passaging across several generations of mice.
In the present study, we established a PDX and cell line for IPC with an invasive phenotype using patient-derived tumor tissues. The PDX was established by engrafting surgically resected tumor tissues in NOG mice. The PDX expanded successfully to generate subsequent tumors at F1, F2, and F3 generations, with the retention of the histological and molecular features of the original tumor. Histology of original papillary carcinoma patient (F0) showed typical phenotypes of fibrovascular core and mitotic figures. This papillary growth pattern was retained throughout PDX passages. Meanwhile, at F3, a slightly complex papillary with a distinct shape was observed. However, overall, the invasive papillary-type was well-maintained. CK19 and mucin expression in F1, F2, and F3 tumors were consistent with that in the patient’s tumor tissue. CK19 as a positive marker of CCA showed a strong positive expression in all PDX and original tumor. In addition, we observed the expression of MUC1 and MUC5AC and no expression of MUC2 in all the tissues (Figure 2). MUC1 and MUC5AC were significantly expressed in both PDX and patient tissues. These ectopically expressing mucins produced by papillary CCA showed a significant correlation with a poor prognosis (27). MUC1 overexpression on cancer cells plays a role in metastasis and associated with aggressive disease. MUC5AC is also a bad prognostic factor when expressed in CCA (27). Therefore, it suggests that the expression of MUC1 and MUC5AC in both xenografts and patient tissue is associated with an aggrieve tumor behavior. These characters are inherited to NCChIPC cells showing papillary phenotype and MUC5AC expression (Figure 5). The NCChIPC cell line established from the PDX F2 generation showed a typical epithelial monolayer with polygonal shapes and regular dimensions, and grew exponentially as a monolayer in the growth medium (Figure 3).
Transcriptome analysis revealed that the gene expression in the NCChIPC cell line was very similar to that in the patient’s tissue and the PDX. The similarity in correlation coefficient values is higher when it is closer to 1; it was observed to be almost 0.8 or higher (Figure 4A), thus reflecting the high reliability of our cell line. Moreover, in the heat-map, we observed that the patient’s tumor belonged to the proliferative subclass (Figure 4B), and hence, anticancer drugs that target cell cycle or proliferation can be administered to this patient. Our cell line showed a very high response to Abraxane® which is known to have inhibitory effect of cell-cycle progression. Therefore, these results support the notion that that our model can be used to predict drug response. The tumorigenic ability of the NCChIPC cells was also very good. In addition, a distinctly shaped, IPC was formed despite changes in the stromal components that occur during engraftment, whereas a homogenous tumor is formed with a cell line in general (Figure 5).
In addition, STR profiling with 10 loci performed for validation of the PDX along with the tumor demonstrating that all PDX-derived models were unique and matched the original patient tissue, as observed in the DNA fingerprinting results (Table II). These results further proved that all xenografts were derived from the primary tumor of the IPC. STR validation using the corresponding results suggests that our model is reliable. This further supports that PDX, and the established cell line are reliable tools for pre-clinical use.
Despite the many advantages of PDX for cancer modeling, there are limitations such as loss of the tumor microenvironment, immune response (28, 29), selection of clonal subpopulations different from the original tumor (29, 30) and cost effectiveness (31). However, despite these challenges, the value and use of PDX in oncology for addressing preclinical models are improving, as they continually reflect the complexity, heterogeneity, and diversity of clinical tumors. Herein, we also used PDX for cell line establishment retaining the original heterogeneity and enhancing the success rate. NCChIPC and PDX-derived cell lines showed in vitro and in vivo heterogeneity to some extent with the characteristics of tissue architecture.
To the best of our knowledge, a PDX and corresponding cell line for IPC have not been previously reported. Therefore, this study is the first to establish PDX and cell lines as tools for detecting and understanding IPC.
In conclusion, our study generated a PDX, the NCChIPC cell line, and NCChIPC-derived xenograft as preclinical models. These novel preclinical models can help improve our understanding of the etiology of IPC and potentially facilitate translational research.
Funding
This research was supported by Grants from the National Cancer Center, Republic of south Korea (NCC-1810864, and NCC-1910193), and basic science research program through the National Research foundation of Korea (NRF) Funded by the ministry of science, CT & future planning (NRF-2019R1A2C1010919).
Acknowledgements
The Authors would like to express appreciation to the Pathology Department, the Genomics Core Facility and Animal Core Facility for the good support to the project. The Authors thank Ebiogen Inc. (Korea) for guide of data normalization and mining. The Authors would like to express deepest gratitude to Ms. Young Hwa Kang for excellent support in project administration.
Footnotes
Authors’ Contributions
SMW, S-JP and Y-HK: conceptualization. BM, SIC, A-RJ and MRL: establishment of models. Y-SL, MRL, JEI, JKK, S-JP and Y-HK: data curation and investigation. BM, MRL, Y-SL and JEI: validation. NYH and EKH; pathology decipher: BM and Y-HK: writing -original manuscript. BM, MRL, SMW and Y-HK writing-review and editing.
↵* These Authors contributed equally to this study.
Conflicts of Interest
The Authors have no conflicts of interest to declare regarding this study.
- Received September 28, 2021.
- Revision received October 22, 2021.
- Accepted November 8, 2021.
- Copyright © 2022 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.