De Novo Lipid Synthesis Facilitates Gemcitabine Resistance through Endoplasmic Reticulum Stress in Pancreatic Cancer

Abstract

Pancreatic adenocarcinoma is moderately responsive to gemcitabine-based chemotherapy, the most widely used single-agent therapy for pancreatic cancer. Although the prognosis in pancreatic cancer remains grim in part due to poor response to therapy, previous attempts at identifying and targeting the resistance mechanisms have not been very successful. By leveraging The Cancer Genome Atlas dataset, we identified lipid metabolism as the metabolic pathway that most significantly correlated with poor gemcitabine response in pancreatic cancer patients. Furthermore, we investigated the relationship between alterations in lipogenesis pathway and gemcitabine resistance by utilizing tissues from the genetically engineered mouse model and human pancreatic cancer patients. We observed a significant increase in fatty acid synthase (FASN) expression with increasing disease progression in spontaneous pancreatic cancer mouse model, and a correlation of high FASN expression with poor survival in patients and poor gemcitabine responsiveness in cell lines. We observed a synergistic effect of FASN inhibitors with gemcitabine in pancreatic cancer cells in culture and orthotopic implantation models. Combination of gemcitabine and the FASN inhibitor orlistat significantly diminished stemness, in part due to induction of endoplasmic reticulum (ER) stress that resulted in apoptosis. Moreover, direct induction of ER stress with thapsigargin caused a similar decrease in stemness and showed synergistic activity with gemcitabine. Our in vivo studies with orthotopic implantation models demonstrated a robust increase in gemcitabine responsiveness upon inhibition of fatty acid biosynthesis with orlistat. Altogether, we demonstrate that fatty acid biosynthesis pathway manipulation can help overcome the gemcitabine resistance in pancreatic cancer by regulating ER stress and stemness. Cancer Res; 77(20); 5503-17. ©2017 AACR.

Figure 1. FASN expression correlates with disease progression and poor response to therapy in pancreatic cancer

(A) Pathway enrichment score comparisons for lipid metabolism pathway in gemcitabine-treated all stage or stage II pancreatic ductal adenocarcinoma patients with complete response or clinical progressive disease. (B) Pancreas tissues from KPC spontaneous progression model of pancreatic cancer and littermate controls were harvested at the indicated time points and the expression of FASN was determined by real-time PCR. Values normalized to Actb (beta-actin) mRNA levels are presented relative to normalized mRNA levels in control mice pancreas at 60 days post-birth. (C) Representative sections indicating the expression of FASN by IHC in human pancreatic tumors and uninvolved pancreas. (D) Kaplan-Meier plot showing overall survival of pancreatic cancer patients with low and high pancreatic tumor FASN expression (p = 0.0285, log rank test). (E) Correlation between FASN mRNA expression and gemcitabine IC50 in a panel of 17 pancreatic cancer cell lines. Correlation was evaluated with Pearson’s correlation; rho (r) value indicated in the figure. * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001

Figure 2. FASN inhibition synergistically enhances the anti-proliferative effect of gemcitabine in human pancreatic cancer cells

(A) Relative cell viability in PANC-1 and AsPC-1 cells upon treatment with a combination of gemcitabine and orlistat in three different schedules: simultaneously, sequentially and reverse sequentially for 72 hours. Cell viability was assessed using MTT assay. (B) Isobolograms showing the combination indices of gemcitabine and orlistat drug combinations in three different schedules: simultaneously: sequentially and reverse sequentially, for PANC-1, AsPC-1 gemcitabine-resistant cell lines and Capan-1 and HPAF-II gemcitabine-sensitive cell lines at 90% inhibition level. Combination index was calculated from MTT data using Compusyn software. (C) Synergistic effect of orlistat to gemcitabine is shown through cell counting. For the drug combination, gemcitabine and orlistat were used in a sequential schedule. (D) and (E) Clonogenic assay: 500 cells, which were previously treated with different gemcitabine concentrations for 72 hours, were seeded and orlistat was added subsequently. Colonies were stained with crystal violet 0.4% after 30 days and counted using Quantity One- 4.5.0 software. Error bars represent standard deviation. Asterisks indicate statistical significance: * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001.

Figure 3. Gemcitabine and orlistat combination induces cell cycle arrest, increased ROS levels, and apoptosis

(A) & (B) Cell cycle analysis after treatment with gemcitabine, orlistat, sequential drug combination for 48 hours. The number of cells in each phase of cell cycle upon indicated treatments was compared to the respective control by one-way ANOVA with Dunnett’s post hoc test. (C) & (D) ROS contents after treatment with gemcitabine, orlistat and drug combination for 24 hours analyzed by staining cells with CDFDA. Relative ROS levels upon indicated treatments were compared to the control by one-way ANOVA with Tukey’s post hoc test. (E) & (F) Cellular caspase 3/7 activity was determined by Caspase-Glo assay kit (Promega) and expressed relative to the control. Cells were treated for 72 hours. Relative caspase activity upon indicated treatments was compared to the control by one-way ANOVA with Tukey’s post hoc test. Error bars represent standard deviation. * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001.

Figure 4. Effect of Orlistat on gemcitabine uptake and metabolism

Real-time PCR analysis of genes involved in gemcitabine uptake (A), metabolism (B), and export (C) upon treatment with gemcitabine, orlistat, and combination of gemcitabine and orlistat for 48 hours, relative to ACTB control gene expression in PANC-1 cells. (D) Relative concentrations of gemcitabine (dFdC) and its active metabolite dFdCMP in gemcitabine-only, and gemcitabine-orlistat combination treated (24 hours) PANC-1 cell extracts, as determined by LC-MS/MS analysis. Error bars represent standard deviation. * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001.

Figure 5. Orlistat decreases the stemness of pancreatic cancer cell lines

(A) Relative mRNA levels of stemness associated genes in PANC-1 pancreatic cancer cell line upon 48 hours treatment with 100 µM orlistat for 48 hours, determined by real-time PCR analysis. The values are normalized to ACTB control gene expression levels. (B) Relative percentage of PANC-1 and CFPAC-1 cells positive for CD44/CD24, c-MET/CD133, CXCR4/CD133, and CD133/c-MET/CXCR4 stem cell markers upon treatment with control, gemcitabine 200 nM, orlistat 100 µM, and sequential drug combination for 48 hours. (C) Self-renewal capacity of the PANC-1 cells upon treatment with gemcitabine, orlistat and combination. Cells were grown in stem cell growth media in 6-well low attachment plates for 8 days. The single cells from primary spheres were cultured in similar conditions for another 8 days and secondary spheres (micrographs) were assessed. Spheres greater than 50 µM were counted. ns P ≥ 0.05, * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001.

Figure 6. Orlistat-induced ER stress causes synergism with gemcitabine

(A) & (B) Western blot analyses of ER stress-related protein levels in PANC-1 cells with indicated treatments at 48 hours post-treatment. (C) Isobologram showing the combination index of gemcitabine and thapsigargin in three different schedules: simultaneous, sequential, and reverse sequential in PANC-1. Combination index was calculated from MTT data using Compusyn software. (D) Relative percentage of PANC-1 cells dual positive for CD24/CD44 stem cell markers upon treatment with control, and thapsigargin for 48 hours. Gem: Gemcitabine, Thap: Thapsigargin. ** P ≤ 0.01, compared to the control by two sample Student’s t-test. (E) Kaplan-Meier survival comparisons between top and bottom tertiles for ER stress marker gene enrichment in TCGA patient tumor specimens. Comparisons are made for all stages (n = 20) and stage II only (n = 18) pancreatic ductal adenocarcinoma patients treated with gemcitabine only by utilizing Gehan-Breslow-Wilcoxon Test.

Figure 7. Enhanced inhibition of tumor growth in an orthotopic implantation model of pancreatic cancer by sequential combination of orlistat and gemcitabine

(A) Longitudinal bioluminescence imaging for tumor growth in athymic nude mice othotopically implanted with PANC-1 cells and treated with control, gemcitabine (Gem), Orlistat (Orli), and gemcitabine-orlistat combination (Gem+Orli) for indicated time points. (B) Average tumor volume as measured by calipers starting at day 30 when the tumor could be palpated. The tumor volume was calculated as (length × width²)/2. (C) Average excised tumor weights upon necropsy in different treatment groups. (D) The average mouse weight throughout the experiment, starting at the first week after tumor implantation. (E) Tumor tissue sections stained for cleaved caspase 3, CHOP, and ATF4 by immunohistochemistry from the four treatment groups at 200X Magnification. Error bars represent standard deviation. Asterisks indicate statistical significance: * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001.