Background
Cholangiocarcinoma (CCA) is a relatively rare malignancy in developed countries with 11,000 estimated new cases and 3,700 estimated deaths in the United States in 2016 [
1]. CCA represents a heterogeneous group of tumors arising from biliary epithelial cells (or cholangiocytes) and associated with the intra- or extra-hepatic biliary tract. The disease has most often progressed to an advanced stage at the time of clinical presentation and recurrence rate is high despite surgery. Single-agent molecular targeted or systemic chemotherapy is the conventional approach for treatment of CCA but is not very effective. Therapeutic resistance can arise from inter-individual variation in sensitivity, and specificity of the target as well as the agent. The use of multimodality approaches that include combination therapies is a rational approach to improve therapeutic outcomes if based on knowledge of cellular mechanisms associated with CCA growth and drug response. Recent studies have identified alterations in chromatin modulators as the most frequent genetic alterations in intrahepatic cholangiocarcinoma (iCCA) [
2,
3]. Amongst these, mutations in BRCA-1 associated protein-1 (BAP1) are the most common, occurring in 22-24% of cases; however, the clinical impact of mutation or loss of BAP1 in iCCA is yet unknown. The gene is located on chromosome 3p21 and encodes BAP1, a member of the ubiquitin C-terminal hydrolase superfamily of deubiquitinating enzymes that plays a critical role in chromatin remodeling together with ARID1A and PBRM1. Chromatin remodeling by BAP1 can involve interactions with several methylation and deacetylase components and result in modulation of gene expression. We have recently reported the involvement of long non-coding RNA (lncRNA) in therapeutic sensitivity in hepatocellular cancers, but their role in iCCA is unknown [
4,
5]. Moreover, the mechanisms by which lncRNA expression is regulated in cancer cells remain very poorly understood. Thus, we sought to evaluate the effect of BAP1 on the modulation of lncRNA involved in therapeutic sensitivity in iCCA. Our studies identified a novel regulatory niche between BAP1 and lncRNA nuclear paraspeckle assembly transcript 1 (NEAT-1) in modulating CCA phenotype and chemotherapeutic responses to gemcitabine individually and in combination with target-specific agents such as EZH2 or PARP inhibitor, GSK126 or olaparib, respectively. Recent evidence suggests a critical role of NEAT-1 in tissue development and as a transcription regulator promoting oncogenic transcriptome and cell regulatory pathways [
6‐
10]. BAP1 and NEAT-1 dictated exclusive drug sensitivity was observed across CCA cell lines suggesting a key mechanistic role in enhancing therapeutic effectiveness. These findings provide new insights into the role of tumor suppressor gene - BAP1 in regulating the expression of NEAT-1, a finding that could guide rationale combinatorial approaches to enhance therapeutic sensitivity in iCCA.
Methods
Cell lines, culture, and reagents
Human cholangiocarcinoma cell lines KMBC, HuCCT1, Mz-ChA-1, CCLP1, human non-malignant intrahepatic biliary cells H69, and human haploid cells with and without BAP1 knockout, HAP1 WT and HAP1 BAP1 KO cells (Horizon Genomics, Cambridge, UK) were obtained and cultured as follows: KMBC, HuCCT1, and CCLP1 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) high-glucose medium (HyClone, Logan, UT), containing 10% fetal bovine serum (FBS) and 1% antibiotic–antimycotic (Life Technologies, Grand Island, NY). Mz-ChA-1 cells were cultured in CMRL 1066 media with 10% FBS, 1% L-glutamine and 1% antibiotic–antimycotic. H69 cells were cultured in hormonally supplemented DMEM/nutrient mixture F-12 Ham (GIBCO BRL, Gaithersburg, MD) (3:1) containing adenine, insulin, triiodothyronine-transferrin, hydrocortisone, epinephrine, epidermal growth factor, penicillin/streptomycin and 10% FBS. Haploid cells were cultured in IMDM media with 10% FBS and 1% antibiotic–antimycotic. All cell lines were subjected to STR analysis. Gemcitabine was obtained from Selleckchem (Houston,TX), cisplatin (Sigma-Aldrich, St. Louis, MO), olaparib and GSK126 were obtained from Selleckchem (Houston, TX) and Cayman chemical (Ann Arbor, MI) respectively; trichostatin A (Santa Cruz Biotechnology, Santa Cruz, CA), and b-AP15 (VLX1500) were provided by Dr. Asher Chanan-Khan. Compounds were dissolved in 100% cell culture-grade DMSO and diluted with culture medium to the desired concentration with a final DMSO concentration of 0.1-0.25%. DMSO 0.1-0.25% (v/v) was used as a solvent control.
Spheroid cell growth assays
Multicellular tumor spheroids were generated by culturing cells in the AlgiMatrix 3D cell culture system (Grand Island, NY). 1 × 106 KMBC cells were seeded in a 6-well plate in the medium described above using AlgiMatrix scaffold and 10% firming buffer and allowed to form spheroids at 37 °C in a 5% CO2 incubator for 7 days. After a 7 day incubation period, the scaffold was dissolved using 5 ml dissolving buffer and spheroids were harvested. For therapeutic response studies, spheroids were seeded (50 per well) in an ultra-low attachment 96-well plate, and viable spheroids counted at selected time points using Cell Titer Glo3D assay (Promega, Madison, WI).
Transient transfections
siRNA against BAP1, NEAT-1 or non-targeting (NT) control sequences were obtained from Qiagen FlexiTube siRNA (Valencia, CA) (Additional file
1). Cells were transfected with either 100 nmol/L siRNA to BAP1 or 75 nmol/L siRNA to NEAT-1 or NT control using Lipofectamine 2000 (Life Technologies, Grand Island, NY) in Opti-MEM medium. Two different siRNA to each target gene were evaluated (sequences in Additional file
1), and the most efficacious one selected for further study. After 6 h, the medium was replaced with the respective complete culture medium containing 10% serum and the cells were incubated for 48 h before further study. BAP1 human cDNA clone, pCMV6-AC-BAP1 plasmid was purchased from Origene Technologies, Rockville, MD. KMBC cells were transfected using Lipofectamine 2000 (Life Technologies, Grand Island, NY) for 24 h with 2 μg BAP1 or pcDNA3.1 control vector before further study. BAP1 CRISPR/Cas 9 knockout plasmid and control CRISPR/Cas9 plasmid were obtained from Santa Cruz Biotechnology, Inc (Santa Cruz, CA). The plasmid was designed for maximum knockdown efficiency. Using Lipofectamine 2000 and plasmid transfection medium H69 cells were transfected with 1 μg of respective plasmid. After 48 h incubation the transfection efficiency was determined by fluorescent microscopy and cells were sorted by FACS analysis.
Assessment of therapeutic drug effect on cell growth
Cells were counted using the Vi-CellXR Cell Viability Analyzer (Beckman Coulter, Brea, CA), seeded at 1 × 103 cells/12 μl/well density in 384-well plate (Corning, NY) and incubated for at least 12 h prior to drug treatment. 200 μl, 2X of the highest drug concentration per drug was dispensed in quadruplets in a flat bottom 96-well plate (Corning, NY) followed by 100 μl of complete medium dispensed across rest of the plate. A BioTek Precision XS robot (Winooski, VT) was programmed to prepare serial dilutions across the plate. Cell viability was assessed after 72 h post-treatment using Cell Titer GloR 2.0 assay (Promega, Madison, WI) and a BioTek synergy HT- Plate Reader (Winooski, VT). Data was analyzed using GraphPad Prism to derive the EC50 or IC50 for each drug. For studies of drug interactions between gemcitabine and other drugs, cell viability was assessed after incubation with fixed ratio combinations based on individual drug IC50 values. Interactions were evaluated using a median effects analysis to derive a combination index (CI), with a CI < 1 indicating a synergistic interaction and CI > 1 suggesting an antagonistic effect.
DNA sequencing
Genomic DNA was isolated using a spin-column based protocol (Qiagen, Valencia, CA). Sequencing primers for BAP1 gene were designed using GeneRunner (Additional file
2: Table S1). The extracted DNA was amplified by PCR using GoTaq Hot Start mix (Promega, Madison, WI) and purified using Exo-SAP (Affymetrix, Santa Clara, CA) to remove excess primers and dNTPs as per the manufacturer’s instructions. 1.6pM of forward or reverse primers and 5 μl of the purified PCR product was used for DNA sequencing. Electropherograms obtained were analyzed using SeqScape v2.5 (ABI, Applied Biosystems, Foster City, CA).
RNA isolation and real-time PCR analysis
Total RNA was extracted from cells using TRIzol (Life Technologies) and treated with RNase-free DNase I (Qiagen, Valencia, CA). RNA concentration was measured using NanoDrop ND-2000 (Nano-Drop Technologies, Wilmington, DE). One microgram of RNA was reverse-transcribed to cDNA using iScript cDNA Synthesis Kit (BIO-RAD Laboratories, Inc., Hercules, CA), and real-time quantitative RT-PCR was performed using a LightCycler 96 System (Roche Diagnostics, Mannheim, Germany) to detect BAP1, NEAT-1, NDM29, MER11C, SNHG4, β-actin, and U6 using SYBR green I (SYBR® Advantage® qPCR Premix, Clontech, Mountain View, CA). RT-PCR primer sequences are listed in Additional file
2: Table S2. PCR based expression profiling of lncRNA was performed using the LncProfiler qPCR Array Kit (System Biosciences, Mountain View, CA), according to the manufacturer’s instructions. RNA from BAP1 knock-down/out cells were treated with DNase I and 2 μg of DNase-treated RNA was reverse transcribed. Real-time PCR was performed (2X Maxima SYBR Green with Rox) and the cycle number at which the reaction crossed a threshold (CT) was determined for each gene. Raw CT values were normalized using a median CT value (ΔCT = CT
lncRNA- CT
median). For each lncRNA, the relative amount of each lncRNA between 2 sample sets A and B was described using the equation 2
-ΔΔCT, where ΔΔCT = ΔCT
A -ΔCT
B.
Western blot analysis
Protein lysates were obtained from cells grown in 100-mm dishes. Nuclear and cytoplasmic fractions were obtained using the NE-PER extraction kit (Pierce, Rockford, IL), Protease Inhibitor Cocktail Tablet and PhosSTOP Phosphatase inhibitor (Roche, Indianapolis, IN) according to the manufacturer’s instructions. Protein concentrations in fractions were determined using the BCA Protein Assay (Pierce, Rockford, IL). Equivalent amounts of protein were mixed with 4X LDS sample buffer, incubated for 10 mins at 70 °C and resolved by electrophoresis in a 4–12% Bis-Tris gel (Life-Tech) followed by dry transfer to nitrocellulose membranes using iBlot Dry Blotting system (Life-Tech). After blocking (Odyssey Blocking buffer), membranes were incubated with respective primary antibodies [BAP1 (1:250) or PCNA (1:1000) (both from Santa Cruz Biotechnology, Santa Cruz, CA) diluted in Odyssey blocking buffer solution and incubated at 4 °C overnight, followed by infrared dye-labeled secondary antibodies (1:10000). The protein of interest was detected using the LI-COR Odyssey infrared imaging system (LI-COR Bioscience, Lincoln, NE). Relative expression was determined by probing against loading control.
Cell cycle analysis
Cells were plated at ~1.25 × 105 cells/well in a 6-well plate and incubated at 37 °C for 12 h followed by 6 h incubation in serum-free medium. The medium was replaced with complete medium containing either IC50 concentration of GSK126 or gemcitabine or a combination or DMSO control. After 24 h, cells were pelleted by centrifugation at 300 g for 5 mins, fixed using cold 70% ethanol (KMBC cells) or 4% paraformaldehyde (HuCCT1 cells) for 15–30 mins and washed twice with PBS. The cells were re-suspended in PBS and incubated with 4 mg/ml RNase for 15 mins, and then re-suspended in PBS and incubated with 10 mg/ml Propidium Iodide (PI) for up to 30 mins.100 μl of cell solution was then transferred to a 96-well plate, analyzed using an Acea Novocyte flow cytometer, and cell cycle analysis was performed using the integrated software.
Invasion assay
5 × 104 cells were suspended in 200 μl serum-free medium and loaded onto the upper compartment of Transwell (Corning, Lowell, MA) 24-well plates with a pore size of 8.0 μm. Serum-free medium (500 μl) was added to the bottom. After 24 h, cells that had migrated through the membrane were fixed and stained using Diff-Quik (Astral Diagnostics, West Deptford, NJ). Migrated cells were identified and quantitated using a microscope and average counts from 5 or more fields of cells were obtained for each group.
Anchorage independent growth assay
Cells transfected with siRNA to NEAT-1 or to respective nontarget control were seeded in 24-well plate in complete medium supplemented with 20% serum. Cells were grown in soft agar as described previously [
11]. The final concentration of the bottom and top feeder layers of the agar system was 1.2% and the cell suspension layer was 0.8%. Cells were incubated for 7 days in a humidified incubator at 37 °C. The total number of colonies was quantified as a direct proportion of fluorescence. Alamar Blue (Biosource International, Camarillo, CA) was added to the wells, and fluorescence was measured using a BioTek synergy HT- Plate Reader (Winooski, VT) (excitation 530/25 nm; emission 580/50 nm).
Analysis of lncRNA in human CCA
Raw sequences of 36 TCGA CCA RNAseq samples were obtained from TCGA website [
12]. These samples were analyzed using a Mayo Clinic custom developed bioinformatics analysis pipeline which aligned the raw sequences to GRCh37 using TopHat 2.0 [
13], counted the reads for known mRNAs and lncRNAs defined in ENSEMBL GTF file using featureCounts [
14]. One outlier sample was detected by principle component analysis and removed from further analysis. Genes having zero read counts in all remaining samples were removed and the remaining genes were normalized by CQN method [
15]. EdgeR R packages [
16] was applied to compare 10 samples with highest
BAP1 gene expression to 10 samples with lowest
BAP1 gene expression, and differentially expressed genes were identified.
Statistical analysis
Data were expressed as the mean ± standard deviation from at least three replicates, unless indicated otherwise. Comparisons between groups were performed using the two-tailed Student’s t test, one- or two-way ANOVA. Results were considered to be statistically significant when P < 0.05.
Discussion
Alterations in chromatin modulators like the nuclear deubiquitinating enzyme BAP1 are the most frequently observed genetic alterations reported in intrahepatic CCA, yet the molecular mechanisms by which they modulate cancer cell behavior are unknown. BAP1 can act as a tumor suppressor and can regulate several cellular processes through its interaction with other protein partners such as host cell factor 1, O-linked N-acetylglucosamine transferase, transcription factor Ying Yang1, ASXL1/2, and FoxK1/K2 and DNA repair proteins like BRCA1/BARD1 heterodimer and RAD51. In the present study, we identify alterations in long non-coding RNA gene expression as a contributor to tumor cell phenotype and differential therapeutic sensitivity of CCA cells that is related to BAP1 expression.
Alterations in BAP1 expression in other cancers such as renal cell carcinoma, breast carcinoma, small cell and non-small cell lung cancers, malignant mesothelioma, metastasizing uveal melanoma, and hepatic cancers can arise from chromosomal deletions [
19,
20]. Germline BAP1 mutations have been noted in several of these and other cancers such as meningioma, paraganglioma, basal cell carcinoma, ovarian cancers, and neuroendocrine tumors [
21,
22]. Genetic mutations in BAP1 could modulate BAP1 protein expression or could alter BAP1 function in the absence of alterations in protein expression. Indeed, genetic mutations in BAP1 are variably associated with loss of protein expression. BAP1 protein expression has been assessed as a prognostic biomarker in iCCA. Loss of BAP1 expression by immunohistochemistry was noted in ~26% of a cohort of 211 patients with iCCA, and was associated with higher histological grade, absence of lymphatic invasion and a trend towards improved prognosis [
3]. The greater sensitivity to gemcitabine and cisplatin in this setting may reflect favorable tumor biology and could contribute to these observations.
As a chromatin modulator, BAP1 is positioned to epigenetically modulate long non-coding RNAs (lncRNAs) and other genes at the intersection of cellular responses to an adverse microenvironment such as enhancement of survival signaling, modulation of cell proliferation and protein synthesis [
23]. Our observations implicate a role for NEAT-1 as a downstream-target of BAP1 that is involved in responses to therapy as well as in maintenance of phenotypic characteristics such as the proliferative, migratory and invasive capabilities of CCA cells. The mechanisms by which NEAT-1 can modulate BAP1 effects are unknown. A role for NEAT-1 has been described in several types of cancers such as breast, prostate, and lung cancer, with effects on promoting tumor growth through genetic or epigenetic mechanisms [
9,
10]. In pancreatic cancers, NEAT-1 can promote growth by modulating levels of miR-335-5p [
8]. NEAT-1 can regulate gene expression through its participation in paraspeckle nuclear body formation, and further studies to explore how these effects may be related to modulation of therapeutic response are warranted. In addition to NEAT-1, several other lncRNAs have been implicated as modulators of drug sensitivity [
24]. Examples include reports of lncRNA-HOTAIR in ER-induced tamoxifen resistance in breast cancer [
25], lncRNA-UCA1 in cisplatin-based resistance in ovarian cancer cells [
26], LncRNA-AK022798 in cisplatin-based drug resistance in gastric cancer cells [
27] and H19 and lncRNA-ROR in hepatocellular carcinoma cells [
4]. Consistent with the wide diversity of actions of lncRNA, several different mechanistic pathways have been implicated in these such as alterations in epithelial-mesenchymal transition, Wnt signaling, Notch 1, mitochondrial apoptosis, cell cycle regulation [
28‐
30]. In another study, lncRNA ENST00000563280 and NR-036444 were found to interact with critical cancer genes such as ABCB1, HIF1A, and FOXC2 to modulate doxorubicin-resistance in osteosarcoma patients [
31].
The dismal results observed with conventional therapies for CCA such as gemcitabine and cisplatin highlights the need for more effective drug regimens. Enhanced sensitivity to PARP inhibition has been reported in BAP1
−/− cells compared with BAP1
+/+ and BAP1
+/− cells, as well as in cancer cells harboring inactivating mutations in BRCA1 or-2 [
32,
33]. Our observations of increased sensitivity to PARP inhibition in the setting of reduced BAP1 expression are consistent with these reports and collectively support a role for BAP1 in DNA damage signaling and repair pathways that could promote the survival of cells with damaged DNA, drive neoplastic progression, as well as impact on response to drugs that induce DNA damage. Reduced BAP1 expression was also associated with enhanced sensitivity to bAP-15. An enhanced sensitivity to EZH2 inhibition was reported in mesothelioma cells lacking BAP1 [
18], and related to enhanced expression of SETD8, a methyl transferase that decreased EZH2 expression and further restricted cell proliferation in these cells. In contrast to observations in other cancers, sensitivity to pharmacologic inhibition of EZH2 by GSK126 was decreased in the setting of reduced BAP1 expression. In CCA patients, EZH2 overexpression is associated with tumor stage, lymph node positivity, and poor prognosis [
34]. Knockdown of EZH2 is reported to alter cell cycle regulation and induce G1 arrest leading to apoptosis in CCA cell lines [
35]. We observed an increase in EZH2 protein and reduced SETD8 with siRNA knockdown of BAP1 confirming that EZH2 represents a potential therapeutic target that could be modulated by BAP1. Combination therapy of an EZH2 inhibitor, 3-deazaneplanocin A, and gemcitabine has had potent synergistic effects in CCA cells by causing cell cycle arrest and increased apoptosis [
36].
Conclusions
Knowledge of BAP1 expression could guide an appropriate selection of candidate drugs for combination approaches. Interestingly, both olaparib and GSK126 had synergistic interactions with gemcitabine in CCA cells, with reduced cell viability in low-BAP1 KMBC cells in contrast to the effect observed in high-BAP1 expressing HuCCT1 cells. The effect of NEAT-1 on modulation of EZH2 expression further supports an effect of this lncRNA on tumor growth. We speculate that low BAP1 expressing cells may have a higher expression of both lncRNA NEAT-1 and EZH2, with a lower sensitivity to EZH2 inhibition.
Overall, there are limited treatment options for CCA, a cancer that has been noted to be increasing in incidence in the US and many other countries. The heterogeneity of these cancers is being recognized. The evaluation of BAP1 expression, or even EZH2 expression, should be considered in future therapeutic studies as they may guide appropriate selection of synergistic combination approaches that could improve outcomes through improved therapeutic responses.
Acknowledgements
Not applicable.