1. Introduction
Cholangiocarcinoma (CCA) is the second most common primary liver malignancy. Cholangiocytes, the epithelial cells that line the biliary tree, undergo neoplastic transformation resulting in the formation of CCA [
1]. CCA can be categorized as intrahepatic, perihilar, or distal subtypes and accounts for 10–20% of all hepatobiliary malignancies [
2]. While potentially curable with surgery if diagnosed at an early stage, the overall clinical outcome of CCA continues to be poor due to its propensity for early local invasion, distant metastasis and high recurrence [
3,
4]. Furthermore, the majority of patients with CCA are diagnosed at advanced stages, and the administration of chemotherapy has shown limited efficacy [
5]. Given the aggressive disease course and lack of targeted treatment options for CCA patients, there is a pressing need for new and efficacious treatment modalities for this malignancy.
Immunotherapy, especially immune checkpoint inhibitors (ICIs), has proven to be an effective therapeutic modality in several chemoresistant malignant diseases including melanoma, renal and non-small cell lung cancers [
6]. Immune checkpoints (ICs) maintain self-tolerance and, during an immune response, ICs protect normal tissue from damage [
7]. However, tumor cells frequently exploit these ICs, serving as a prominent tumor immune evasion mechanism causing T-cell inactivation and downregulation of T-cell responses [
7,
8]. Immune checkpoint blockade strategies have been effective in reanimating the T-cell antigen-specific response and associated antitumor effects [
8]. Within the tumor microenvironment, ICs may serve as therapeutic targets for the treatment of primary liver malignancies including hepatocellular carcinoma (HCC) and CCA. ICIs hold great promise for HCC and CCA as the deregulation of the immune system contributes to the pathogenesis of these liver malignancies [
9,
10].
In the past few years, ICIs as monotherapy have elicited a durable and robust antitumor response in only a proportion of cancer patients, with variability both between types of cancers and between patients who share a histological type [
11,
12]. A trial of monotherapy with anti-PD1 antibody pembrolizumab in phase I/II in CCA patients showed a low response rate (10% to 20%), and little is known of the underlying mechanism of resistance [
11,
13]. Another phase I trial for CCA patients revealed that the combined treatment with anti-PD-1 antibody nivolumab and cisplatin plus gemcitabine was more effective than nivolumab as monotherapy [
14]. Similarly, combining anti-PD-L1 antibody durvalumab with the anti-CTLA4 antibody tremelimumab was more effective than monotherapies in CCA patients [
15].
Analysis of immune checkpoint marker testing such as PD-L1 alone for patient selection and predicting response to ICI therapy has proven insufficient in many cancers [
16]. Therefore, identifying and characterizing additional predictive biomarkers are of the utmost importance for the selection of a subset of CCA patients who are more likely to respond to ICI therapies. A better understanding of alternative checkpoint pathways may be required to increase the clinical benefits of ICIs in CCA patients. These alternative checkpoints may provide additional targets for rational combinatorial therapies that may enhance the effects of immunotherapy in CCA.
Combining the expression of immune checkpoints with additional biomarkers such as those identifying epithelial-to-mesenchymal transition (EMT) and cancer stem cells (CSCs) are also being considered in the management of some cancers [
10,
17,
18]. EMT is defined by the loss of epithelial properties and the concomitant gain of mesenchymal properties. EMT contributes to the invasion and metastasis of tumor cells and also to immunosuppression [
19,
20]. A correlation between the EMT phenotype with multiple ICs in many patient tumors has been reported [
10,
21]. In extrahepatic CCA, a close relationship between EMT and PD-L1 expression has been reported [
22]. However, little is known regarding the association of other ICs with an EMT phenotype in CCA. CSCs are also referred to as tumor-initiating or tumor-propagating cells. CSCs represent a small subpopulation of cells within the tumor that contributes to tumor initiation, metastasis and recurrence. CSCs are endowed with self-renewal, pluripotent properties and enhanced resistance to chemotherapy compared to the tumor bulk [
23]. The ability of PD-L1 to inhibit cancer stemness in CCA has been demonstrated [
24]. Other studies have reported the expression of PD-L1 and PD-L2 in CSCs derived from colon and breast cancers [
25]. Little is known regarding the association between expressions of other immune modulators and CCA-related CSC phenotype.
In this pilot study, we sought to identify prognostic immune-modulatory molecules in CCA patients. To this end, we analyzed CCA patient databases from The Cancer Genome Atlas (TCGA) and SurvExpress [
26,
27]. We correlated the expression of immune-related molecules with patient prognosis. Given that EMT and CSCs have a substantial role in CCA initiation and progression, we assessed the association of immune checkpoint molecule expression in aggressive CCA cell subpopulations such as CSCs and cells undergoing transforming growth factor (TGF)-β1- and tumor necrosis factor (TNF)-α-mediated EMT.
2. Materials and Methods
cBioPortal OncoPrint evaluation of immune checkpoint molecules: cBioPortal OncoPrint (
http://cbioportal.org, accessed on 10 May 2021) was used to generate a graphical summary of gene expression changes in immune checkpoint molecules across CCA patient samples. Within cBioPortal, we utilized the Cholangiocarcinoma (TCGA, Firehose Legacy) case set of 51 patients to evaluate gene changes in immune checkpoint genes. CCA patients are represented as columns and immune-modulatory genes are represented as rows. Genomic alterations, including copy number aberrations, changes in gene or protein expressions and mutations, are represented by glyphs and color codes [
27].
CCA patient databases: SurvExpress utilizes a gene expression database of different cancers to generate survival analyses of CCA patients (
http://bioinformatica.mty.itesm.mx:8080/Biomatec/SurvivaX.jsp, accessed on 10 May 2021). SurvExpress provided a CCA database of 35 patient samples (CHOL-TCGA Cholangiocarcinoma).
Evaluation of immune checkpoint molecules as prognostic biomarkers in CCA patients: SurvExpres was applied to evaluate the relationship between the expressions of 19 immune modulators with the survival of CCA patients based on a Cox regression analysis. The overall survival for CCA patients was estimated by Kaplan–Meier curves. The average intensity of quantile-normalized array data was used for genes with multiple probe sets. Survival and progression-free survival analyses were also performed using a CCA dataset of 36 patients in cBioPortal.
Cell culture and reagents: Prof. Mark Gorrell, Centenary Institute, Australia kindly gifted human CCA cell lines HuCCT-1 and CCLP-1. Human CCA cell line EGI-1 was sourced from Prof. John Mariadason, Olivia Newton-John Cancer Research Institute, Heidelberg, VIC, Australia. MycoAlert tests (ABM, Richmond, BC, Canada) confirmed the mycoplasma-free status of these cell lines. HuCCT-1 was cultured in Roswell Park Memorial Institute (RPMI); 1640 medium (Thermo Fisher Scientific Australia, Scoresby, VIC, Australia) supplemented with 10% fetal bovine serum (FBS) (Gibco, Life Technologies Australia Pty Ltd, Mulgrave, VIC, Australia) and 0.05% Gentamicin (Thermo Fisher Scientific Australia, Scoresby, VIC, Australia). Dulbecco’s modified Eagle’s medium (DMEM) (Thermo Fisher Scientific Australia, Scoresby, VIC, Australia) supplemented with 10% FBS (Gibco, Life Technologies Australia Pty Ltd, Mulgrave, VIC, Australia) and 0.05% Gentamicin (Thermo Fisher Scientific Australia, Scoresby, VIC, Australia) was used to culture CCLP-1. EGI-1 was cultured in DMEM (Thermo Fisher Scientific Australia, Scoresby, VIC, Australia) with 10% FBS (Gibco, Life Technologies Australia Pty Ltd, Mulgrave, VIC, Australia) and 1% penicillin/streptomycin (P/S) (Thermo Fisher Scientific Australia, Scoresby, VIC, Australia). Cells were cultured under a humidified atmosphere with 5% CO2 in the air at 37 °C. The cytokines TGF-β1 and TNF-α were procured from PeproTech, Cranbury, NJ, USA.
3-dimensional sphere enrichment assay: Trypsin-EDTA was used to detach cells grown as monolayers. Cells were suspended in serum-free stem cell medium following the removal of serum with 1 × PBS washes [
28]. The serum-free stem cell medium was prepared with DMEM/F12 medium (Thermo Fisher Scientific Australia, Scoresby, VIC, Australia) supplemented with 20 ng/mL recombinant human epidermal growth factor (rhEGF) (PeproTech, Cranbury, NJ, USA), 10 ng/mL recombinant human fibroblast growth factor (rhFGF) (PeproTech, Cranbury, NJ, USA), 2% B27 supplement without vitamin A (Invitrogen, Scoresby, VIC, Australia) and 1% N2 supplement (Invitrogen, Scoresby, VIC, Australia). Then, 5000 cells were plated per well in ultra-low attachment, 6-well plates (Corning, Melbourne, VIC, Australia). Cells were cultured in a humidified atmosphere of 5% CO
2 in air at 37 °C for 7 days. The spheres were collected by gentle centrifugation.
RNA extraction and cDNA synthesis: ISOLATE II RNA Mini Kit (Bioline, Eveleigh, NSW, Australia) was used for the purification of RNA [
29]. A NanoDrop 2000c spectrophotometer (Thermo Fisher Scientific Australia, Scoresby, VIC, Australia) was used to confirm RNA quantity and purity. Then, 1 µg RNA was reverse transcribed to cDNA with a Bioline SensiFAST cDNA Synthesis Kit (Bioline, Eveleigh, NSW, Australia).
Quantitative reverse transcription-PCR (qRT-PCR): Applied Biosystems ViiA 7 Real-Time PCR System was used for performing qRT-PCR with Lo-ROX SYBR Green (Bioline, Eveleigh, NSW, Australia) [
28]. Briefly, a 3-step cycle of the following conditions, 95 °C for 5 s, 63 °C for 20 s and 75 °C for 20 s was repeated for 40 cycles.
Beta-Actin (
ActB) was used as the housekeeping gene.
Table 1 lists the primers used in this study. The 2ΔΔCt method was used for data analysis. In this 2ΔΔCt method, candidate gene expression was normalized to
ActB expression, and copies of target gene per 10,000 copies of
ActB was used to present data.
Western blot analysis: Western blot analyses were performed as previously described [
29]. Briefly, cells were cultured and treated in 6-well plates. RIPA buffer (Thermo Fisher Scientific Australia, Scoresby, VIC, Australia) with Complete (Roche, Australia) and PhosSTOP (Roche, Sydney, NSW, Australia) protease and phosphatase inhibitors were used to lyse cells at 4 °C. Pierce BCA Protein Assay Kit (Thermo Fisher Scientific Australia, Scoresby, VIC, Australia) was utilized to measure total protein concentration. Then, 10 µg of protein was separated by electrophoresis (SDS-PAGE) in a polyacrylamide gel containing sodium dodecyl sulphate (SDS) and transferred to a polyvinylidene difluoride film (PVDF) membrane. Additionally, 5% skim milk in Tris-buffered saline containing 0.1% Tween 20 (TBS-T) was used to block the membranes. Next, the membranes were exposed to primary antibodies at 4 °C overnight. SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific Australia, Scoresby, VIC, Australia) detected the proteins on the membranes after exposure to HRP-conjugated secondary antibodies. β-Actin was the housekeeping control. Image Quant LAS 500 was used for image capture. Image Studio™ Lite v5.2 software was used for quantification.
Table 2 lists the antibodies used in this study.
Statistical analysis: Kaplan–Meier analysis was used to determine the relationship between immune modulator or EMT or CSC expression and CCA patient survival. A log-rank test was performed, and the
p-value for survival analysis was generated [
30]. In vitro experiments were repeated at least thrice, and representative results are presented. Using the Kolmogorov–Smirnov Test of Normality, we determined the normal distribution of the in vitro data sets (data not shown). Comparisons of in vitro data were performed with Student’s two-tailed
t-test with GraphPad Prism software version 8.00 (GraphPad Software Inc., San Diego, CA, USA). Statistical significance was set at *
p < 0.05, **
p < 0.01, ***
p < 0.005 and ****
p < 0.001.
4. Discussion
In the present pilot study, the expression of immune modulators IDO1, NT5E and FASLG was related to poor prognosis in CCA patients. Furthermore, a combination of various ICs with putative immune modulators PD-1, PD-L1 and CTLA-4 was associated with poor CCA patient prognosis. Moreover, EMT and CSC were closely associated with the modulation of immune checkpoint molecules in CCA. PD-L1 and NT5E expression was closely associated with EMT, while coordinate expression of NT5E and LSGAL9 with CSC marker ALDH1A1 was linked with poor overall survival in CCA patients.
Indoleamine 2,3-dioxygenase 1 (IDO1) is an intracellular heme-containing enzyme that contributes to the immune escape of tumors [
34]. Our observation of the association of high IDO1 with poor overall survival in CCA patients is consistent with observations in colorectal, non-small-cell lung and prostate cancers [
35]. In contrast, high IDO1 expression levels in HCC patients have been correlated with better survival outcomes, indicating that IDO1 may not have immunosuppressive functions in this cancer [
36,
37]. Elevated IDO1 expression in cancers has been correlated with single-agent ICI therapy resistance [
38]. Our findings suggest that combining IDO1 inhibitors with other ICIs may represent a promising strategy to expand CCA patient populations for immunotherapies. FASLG, a transmembrane protein of the tumor necrosis factor superfamily triggers apoptosis of T-cells [
39]. Ecto-5′-nucleotidase (CD73 or NT5E) is a glycophosphatidylinositol-anchored receptor enzyme that blocks activation of T-cell when adenosine binds to its receptor [
40]. A study found that CCA cell lines that expressed FASLG, induced cell death when cocultured with T-cells, indicative of the immune evasive function of FASLG/FAS axis in CCA [
39]. We and others have previously found that NT5E expression in cancers was associated with poor prognosis [
10,
41]. We found that IDO1 expression was associated with clinical parameters, such as a presence of risk factors for HCC and tumor stage II and III, while the expression of FASLG in CCA patients was associated with the lymph node stage. IDO1, FASLG and NT5E showed no association with clinical parameters including vascular invasion, gender and tumor stage T0–T3.
This pilot study revealed that IDO1, FASLG, CD80, HAVCR2, CD73, CTLA-4, LGALS9, VTCN1 and TNFRSF14 in combination with PD-L1 is linked with poor outcome in CCA patients. The Cluster of differentiation 80 (CD80) regulates T-cell activation by binding to CTLA4. A study on biliary tract cancers including CCA reported that strong CD80 expression in tumor tissue was closely associated with resistance to adjuvant chemotherapy [
42]. The T-cell immunoglobulin and mucin-domain 3 (TIM3 or HAVCR2) receptor limits T-cell responses by interacting with its ligand Galectin-9 (LGALS9 or Gal-9) [
43]. In animal models, combining anti-HAVCR2 and anti-PD-1 has shown to suppress tumor growth [
44].
V-set domain-containing T-cell activation inhibitor 1, VTCN1, (also named B7-H4, B7S1 or B7x) belongs to the B7 family and regulates T-cell-mediated antitumor responses. Studies in CCA patients showed high levels of VTCN1 expression were significantly related to poor prognosis [
45,
46]. LGALS9 is a tandem-repeat-type galectin that promotes antitumor immune responses by exerting antiproliferative effects on CAA cells [
47]. T-cell immunoglobulin and ITIM domain (TIGIT) is an inhibitory immune checkpoint of the poliovirus receptor (PVR)/desmin family [
48]. Inhibition of TIGIT alone or with PD-1 has shown to restore tumor-suppressive effects [
49]. The tumor stroma of CCA patients showed infiltration of lymphocytes expressing ICs, including PD1 and TIGIT [
50]. The function of tumor necrosis factor receptor superfamily member 14 (TNFRSF14) is not known in CCA.
We and others have noted PD-L1 expression was closely linked with EMT status [
22]. This is the first study to examine the relationship between TGF-β1- and TNF-α-induced EMT and upregulation of PD-L1 and NT5E expression. Furthermore, we and others have reported that PD-L1 expression negatively impacts CCA patient prognosis [
51]. In contrast, other studies show that CCA patients with low PD-L1 expression had a poorer prognosis [
24,
52].
A study showed that CCA-derived, CD133-positive CSCs displayed high levels of TGF-β1 and activation of the TGF-β1–pSmad2–EMT axis [
53]. Similarly, we found CCA-derived CSCs showed a high expression of TGF-β1 along with TNF-α. A study reported that the PD-L1 low cell fraction isolated from HuCCT-1 cells was enriched with CSC-related characteristics compared with the PD-L1 high cell fraction [
24]. This observation is consistent with our results demonstrating the downregulation of PD-L1 in CSCs enriched by HuCCT-1 cells. This study is the first to examine the relationship between CSC phenotype and other novel ICs including NT5E, LGALS9, TNFRSF14, FASLG and VTCN1. Our data suggest that CCA patient tumors with mesenchymal and CSC phenotypes might be targeted using immune checkpoint blockades.
This study is limited by a lack of CCA patient samples who have undergone treatment with immune checkpoint therapies. This pilot study, comprising a small number of CCA patients, may not provide adequate information and conclusions. Thus, further validation of the potential of immune checkpoint regulators as prognostic markers in larger cohorts of CCA patients will be more informative. Another limitation of this study is the lack of available clinical data for the SurvExpress CCA patients. Furthermore, the prognosis of CCA may be affected by the location of the tumor. In this study, we were unable to evaluate the independent association between distal, perihilar or intrahepatic CCA with the expression of immune checkpoints and patient prognosis. We were unable to assess the association between immune checkpoint expression and cancer-specific mortality. Further studies are needed to evaluate the correlation between immune checkpoint expression with parameters such as tumor location and cancer-specific mortality. While this study focused on the expression ICs in CCA tumor cells, comprehensive investigation is needed to validate the role of each individual IC in in vitro and in vivo CCA models. Studies particularly focusing on molecular mechanisms, functional assays, including motility and drug testing, and evaluation of ICs in animal models will be required to gain a better understanding of their function in CCA tissues. Further studies are needed to assess the expression of these molecules on tumor-infiltrating T-cells, which will also be helpful in predicting ICI responses. Additionally, studies on the co-expression of these immune checkpoints in CCA, as well as on how they influence or act with each other, will need to be assessed to enable effective and durable treatment.