Monensin inhibits cell proliferation and tumor growth of chemo-resistant pancreatic cancer cells by targeting the EGFR signaling pathway

Pancreatic ductal adenocarcinoma (PDAC) is one of the most deadly malignancies with <5% five-year survival rate due to late diagnosis, limited treatment options and chemoresistance. There is thus an urgent unmet clinical need to develop effective anticancer drugs to treat pancreatic cancer. Here, we study the potential of repurposing monensin as an anticancer drug for chemo-resistant pancreatic cancer. Using the two commonly-used chemo-resistant pancreatic cancer cell lines PANC-1 and MiaPaCa-2, we show that monensin suppresses cell proliferation and migration, and cell cycle progression, while solicits apoptosis in pancreatic cancer lines at a low micromole range. Moreover, monensin functions synergistically with gemcitabine or EGFR inhibitor erlotinib in suppressing cell growth and inducing cell death of pancreatic cancer cells. Mechanistically, monensin suppresses numerous cancer-associated pathways, such as E2F/DP1, STAT1/2, NFkB, AP-1, Elk-1/SRF, and represses EGFR expression in pancreatic cancer lines. Furthermore, the in vivo study shows that monensin blunts PDAC xenograft tumor growth by suppressing cell proliferation via targeting EGFR pathway. Therefore, our findings demonstrate that monensin can be repurposed as an effective anti-pancreatic cancer drug even though more investigations are needed to validate its safety and anticancer efficacy in pre-clinical and clinical models.

WST-1 assay. Cell proliferation was measured by using Premixed WST-1 Reagent (Takara Bio USA, Mountain View, CA) as previously described [28][29][30][31] . Briefly, subconfluent Panc-1 and MiaPaCa-2 cells were plated in 96-well plates and treated with different concentrations of drugs (gemcitabine, monensin, erlotinib or drugs combinations) for 48 h. The Premixed WST-1 substrate was added to the wells, and incubated at 37 °C for 30 min, followed byreading at 440 nm using a microplate reader. Each assay condition was performed in triplicate.
Cell wounding/migration assay. Cell wounding/migration experiments were carried out as previously described [32][33][34][35][36] . Specifically, cells were plated in 6-well culture plates to reach ~90% confluence. The monolayer cells were then scratched with pipette tips. At the indicated time points, wound healing status at the same locations was recorded. Each assay was set up in triplicate.
Transwell cell migration analysis. Transwell assay was carried out as previously described [37][38][39] . Briefly, resuspended Panc-1 or MiaPaCa-2 cells were place in the upper chamber containing a layer of the 8 µm pore Corning transwell membrane (Millipore-Sigma) and treated with 4 µM monensin or DMSO control, while the lower chamber was filled with culture medium. At 12 h post treatment, the cells that migrated through the membrane were fixed, stained, and counted (e.g., 10 high-power fields were counted to determine the average migrated cells).
Apoptosis analysis. The apoptosis analysis was determined by using the Annexin V staining assay as Chou-Talalay drug combination index determination. The drug combination effects between gemcitabine or EGFR inhibitor (erlotinib) and monensin was analyzed by using Chou-Talalay method 28,29,48,49 . Dose-dependent effects of each drug alone and in combinations on cell proliferation were first determined by WST-1 assay. The acquired data were calculated by using CompuSyn software (ComboSyn, Inc.). The obtained combination index (CI) from Chou-Talalay method provides a quantitative definition for additive effect (CI = 1), synergism (CI < 1), and antagonism (CI > 1) in different drug combinations as previously described 48,49 . Transfection and Gaussia luciferase assay. Gaussia luciferase (GLuc) reporter analysis was conducted as described [50][51][52][53][54] . A penal of cancer-associated signaling pathway reporters, such as E2F/DP1, Elk1/SRF, AP-1, NFκB, and STAT1/2 reporters, were homemade as previously described 29,33,52 . A constitutively active reporter pBG2Luc served as a control 29,33 . Briefly, Panc-1 cells were plated in 25 cm 2 flasks at subconfluence and transfected with 3 µg/flask of different reporter plasmids using Lipofectamine according to the manufacturer's instructions (Invitrogen). At 16 h post transfection, the transfected cells were reseeded into 24-well culture plates and treated with indicated concentrations of drug or vehicle control. At 24 h and 48 h post treatment, 50 µl of culture medium were taken for Gaussia luciferase assay using BioLux Gaussia Luciferase Assay Kit (New England Biolabs, NEB, Ipswich, MA). Each assay was carried out in triplicate.
Total RNA purification and Touchdown-quantitative real-time PCR (TqPCR). Subconfluent pancreatic cancer cells were treated with different concentrations of monensin for 48 h. RNA was isolated by using TRIZOL Reagents (Invitrogen) for reverse transcription with hexamer and M-MuLV reverse transcriptase (NEB). The cDNA products were used qPCR with the primers of the genes of interest designed with Primer3 program (Supplementary Table S1) 55 . TqPCR was done by using SYBR Green-based qPCR on a CFX-Connect unit (Bio-Rad Laboratories, Hercules, CA) as previously described 50,[56][57][58] . Each qPCR condition was done in triplicate. GAPDH was used to normalize gene expression levels.
Immunofluorescence staining. The immunofluorescence staining was performed as previously reported [59][60][61] . Briefly, Panc-1 and MiaPaCa-2 were exposed to various concentrations of drug or vehicle control. At 36 h, the cells were fixed and immunofluorescence stained with an anti-EGFR antibody (Santa Cruz Biotechnology, Santa Cruz, CA). Negative control was set up by incubating the cells with control IgG.

Statistical analysis.
All quantitative experiments were either repeated three times independently and/or performed in triplicate. The student's t test and one-way analysis of variance were used to calculate the statistical significance, which was defined as p < 0.05.

Results
Monensin suppresses cell proliferation and migration of gemcitabine-resistant pancreatic cancer cells. We first tested the effect of gemcitabine on two commonly-used human pancreatic cancer cell lines Panc-1 and MiaPaCa-2. Gemcitabine is one of the first-line chemotherapy agents for pancreatic cancer. When subconfluent Panc-1 and MiaPaCa-2 cells were exposed to escalating concentrations of gemcitabine, crystal violet staining demonstrated that Panc-1 cells were resistant to gemcitabine and survived well at an even high concentration of 40 µM, although MiaPaCa-2 cells were slightly more sensitive to gemcitabine (Fig. 1A, panel a). However, under the same growth condition monensin was shown to effectively suppress cell proliferation of both lines at a concentration as low as 1 µM (Fig. 1A, panel b), suggesting that monensin may act as a potent anticancer agent for pancreatic cancer cells.
We further carried out the WST-1 cell proliferation assays and confirmed the above findings. Specifically, Panc-1 cells were inhibited ineffectively by gemcitabine and showed about only 20% inhibition even at 80 µM of gemcitabine, while MiaPaCa-2 cells were inhibited by 80% at 20 µM of gemcitabine (Fig. 1B, panel a). On the other hand, the WST-1 assay results showed that the cell proliferation was drastically suppressed by monensin at a concentration of as low as 0.5 µM monensin for both Panc-1 (p < 0.01) and MiaPaCa-2 (p < 0.01) (Fig. 1B,  panel b). Taken together, the above results indicate that monensin can effectively suppress pancreatic cancer cell proliferation and overcome gemcitabine resistance in pancreatic cancer cells, particularly in Panc-1 cells.

Monensin suppresses cell wound healing and migration of pancreatic cancer cells.
We next tested whether monensin treatment impacts wound healing and cell migration of pancreatic cancer cells. When confluent Panc-1 and MiaPaCa-2 monolayer cells were scratched and added with 0 or 4 uM monensin, we found that the rate of wound closure was significantly lower in monensin-treated Panc-1 (Fig. 1C, panel a) and MiaPaCa-2 ( Fig. 1C, panel b) than that of the control group (or 0 µM monensin) at all examined the time points. For example, at 48 h, the wound gap was only ~20% and ~5% in Panc-1 and MiaPaCa-2 cells, respectively, compare to the starting time point (or 0 h) in the control groups. However, in the presence of 4 µM monensin the rate of wound closure in Panc-1 and MiaPaCa-2 was significantly reduced, and approximately 90% of the wound remained open in both cell lines (Fig. 1C, panels a vs. b).
We conducted the transwell cell migration assay to assess the effect of monensin on cell migration capability of pancreatic cells. While both Panc-1 and MiaPaCa-2 cells were shown to migrate through the transwell membrane rather effectively (Fig. 1D, panels a,b), the presence of monensin (at 4 µM) significantly reduced the average cell numbers migrated through the transwell membrane in both cell lines (p < 0.01) (Fig. 1D, panel c). Collectively, the above findings strongly suggest that, in addition to its ability to inhibit cell proliferation, monensin may significantly diminish the cell wound healing and migration capabilities of pancreatic cancer cells.

Monensin suppresses cell cycle progression and induces apoptosis in human pancreatic cancer cells.
To uncover possible mechanism through which monensin inhibits pancreatic cancer cell proliferation, we analyzed the cell cycle profile in monensin-treated pancreatic cancer cells and found a drastic increase in the sub-G1 phase and a remarkable decrease in S phase of Panc-1 and MiaPaCa-2 cells, compared with that of the control groups ( Fig. 2A, panels a,b), consistent with monensin's ability to inhibit cell proliferation of pancreatic cancer cells. Notably, the G1 phase in both lines also significantly decreased, which may be caused by the marked increase in sub-G1 populations.
Furthermore, Annexin-V based apoptosis analysis demonstrated that monensin was able to induce both early apoptosis and late apoptosis, compared with that of the control groups (Fig. 2B, panels a,b). For example, in Panc-1 cells, the percentage of early apoptotic cells and late apoptotic cells in the 4 µM monensin group were 12% and 20%, respectively, compared with 5.85% and 10.5% in the control group (p < 0.01) (Fig. 2B, panel a). Similarly, in MiaPaCa-2 cells the percentage of early apoptotic cells and late apoptotic cells in the 4 µM monensin group were 23% and 40.1%, respectively, compared with 1.95% and 5.77% in control group (p < 0.01) (Fig. 2B,  panel b). We also assessed the nucleus morphologic evidence of monensin-induced apoptosis in pancreatic cancer cells through Hoechst 33258 staining and found that monensin (at 4 µM for 48 h) induced significantly higher % of nuclear condensation and DNA fragmentation in Panc-1 and MiaPaCa-2 cells, compared with that of the control groups (data not shown). Therefore, the above findings suggest monensin may suppress pancreatic cancer cell proliferation at least in part by inhibiting cell cycle progression and inducing apoptosis.
Monensin acts synergistically with gemcitabine or EGFR inhibitor erlotinib on suppressing cell growth and inducing cell death of human pancreatic cancer lines. We tested if monensin would synergize with and/or sensitize pancreatic cancer cells to the currently used first-line chemotherapeutic drugs, such as gemcitabine and EGFR targeted inhibitor erlotinib, to inhibit pancreatic cancer cell proliferation. Qualitative crystal violet staining assay showed that, when subconfluent Panc-1 cells were co-treated with gemcitabine or erlotinib and different concentrations of monensin, menonsin was shown to enhance the inhibitory effects of gemcitabine (Fig. 3A, panel a) or erlotinib (Fig. 3A, panel b) in a dose-dependent fashion. Similar results were obtained in MiaPaCa-2 cells (Fig. 3B, panels a,b). The above findings indicate that monensin can potentiate the inhibitory effects that gemcitabine or erlotinib exerts on pancreatic cancer cells.  We also carried out the quantitative WST-1 assays on drug combinations (Fig. 3C, panels a,c). Using the data acquired from the WST-1 assays, we analyzed the combination index (or CI) with the well-established Chou-Talalay method 48,49 . We found the CI values for both monensin/gemcitabine and monensin/erlotinib combinations in Panc-1 cells were <1, indicating that these combinations may exhibit synergistic effects (Fig. 3C,  panel b). Similarly, in MiaPaCa-2 cells the CI values for both monensin/gemcitabine and monensin/erlotinib combinations were also <1, indicating that these combinations may exhibit synergistic effects (Fig. 3C, panel d).
Furthermore, we analyzed the effect of combinations of monensin and gemcitabine on inducing apoptosis in both Panc-1 and MiaPaCa-2 cells, and found that monensin significantly enhanced gemcitabine-induced early and late phases of apoptosis (Fig. 4A, panels a,b). Similarly, monensin significantly augmented erlotinib-induced early and late phases of apoptosis (Fig. 4B, panels a,b). Collectively, these findings strongly demonstrate that monensin can synergize with gemcitabine or erlotinib on inhibiting the cell growth and inducing cell death of pancreatic cancer cells.

Monensin suppressed several cancer-associated pathways and effectively inhibits the expression of EGFR in pancreatic cancer cells.
To explore the potential mechanisms through which monensin exerts anticancer effects on pancreatic cancer cells, we surveyed the effects of monensin on a panel of five well-characterized cancer-associated signal pathway reporters as we previously described 28,29,33,45,52,71,72 . When the Gaussia luciferase reporters were introduced into Panc-1 cells and treated with 0, 1 µM or 4 µM monensin for 24 h and 48 h, the Gaussia luciferase activities for the E2F/DP1, STAT1/2, NFκB, AP-1 and Elk-1/SRF reporters were significantly inhibited (p < 0.01) (Fig. 5A, panels a,b). It is noteworthy that we also analyzed the reporter activities for NFAT, HIF1A, RBP-JK, MYC/MAX, TCF/LEF, CREB, and TGFB/SMAD pathways and found their activities were not significantly affected by monensin.
Based on the results from the reporter assays and the previous reports from ours and other labs 19,29 , we speculated that monensin might target EGFR pathway. Thus, we analyzed the expression levels of the EGFR and EGFR-regulated downstream genes following monensin treatment in Panc-1 and MiaPaCa-2 cells. Using TqPCR 56 , we found that the expression of EGFR and RAF1 was significantly repressed in both cell lines (Fig. 5B,  panels a,b). Interestingly, KRAS and NRAS, and to a lesser extent MEK1, were shown to be significantly up-regulated by monensin, especially at 4 µM level (Fig. 5B, panels a,b). While we do not have any satisfactory explanations about such up-regulations, it is conceivable the up-regulation may be caused by negative feedback inhibitions upon monensin treatment. We also examined the effect of monensin on EGFR expression at protein level. Subconfluent Panc-1 and MiaPaCa-2 were treated with varied concentrations of monensin. EGFR expression was examined by immunofluorescence analysis. We found that while the membrane-bound and/or whole cell expression of EGFR was readily detected in the control or vehicle treated cells, EGFR expression, especially at cell membrane, was significantly suppressed by monensin, although there was significant nuclear stainging upon monensin treatment (Fig. 5C,  panel a,b). Nonetheless, the above results further indicate monensin can target EGFR signaling in pancreatic cancer.
We further analyzed EGFR expression in the clinical samples of pancreatic cancer. While EGFR expression in normal pancreatic samples was not apparently detectable, seven of the examined eight cases of pancreatic cancer samples exhibited strong EGFR staining in the cancerous ductal cells (Fig. 6A vs. B). The remaining samples had weaker but detectable EGFR expression in cancerous regions (data not shown). Taken together, these results demonstrate that EGFR may be targeted by monensin, which may at least in part explain how monensin exerts its effective anticancer activity against pancreatic cancer cells.

Monensin effectively blunts the tumor growth and inhibits cell proliferation and EGFR expression in the xenograft model of human pancreatic cancer in vivo.
Lastly, we examined the in vivo anticancer activity of monensin in the xenograft tumor model of human pancreatic cancer. The firefly luciferase-tagged Panc-1 cells were first injected into the flanks of athymic mice. After 3 days, the mice were randomly divided into two groups and treated with monensin (10 mg/kg body weight) or vehicle control. Tumor progression was monitored through whole body Xenogen bioluminescence imaging (Fig. 7A panel a). Xenogen imaging data analysis indicates that monensin effectively suppressed tumor growth at as early as 11 days after treatment, compared with the control group (Fig. 7A, panel b). At 4 weeks after treatment, the tumor masses recovered from the control group are significantly larger, either in individual tumors or bulk tumor volumes, than that of the monensin treatment group (Fig. 7B, panels a-c). These findings further confirm that monensin can effectively suppress the tumor growth of gemcitabine-resistant pancreatic cancer cells in vivo.
Nonetheless, histologic evaluation of the tumor masses retrieved from both groups did not show significant differences (Fig. 7C, panel a). However, IHC analysis indicated that the expression of the cell proliferation marker PCNA dramatically decreased in the tumor samples retrieved from the monensin treatment group, compared with that of the control group (Fig. 7C, panel b). Similarly, immunohistochemical staining indicated that the EGFR expression was significantly diminished in the tumor samples retrieved from the monensin treatment group, compared with that of the control group (Fig. 7C, panel c), consistent with the possibility that EGFR may be targeted by monensin in pancreatic cancer cells. Collectively, the in vitro and in vivo findings strongly demonstrate that monensin exerts a potent inhibitory effect on cell proliferation and tumor growth in drug-resistant pancreatic cancer cells, possibly through targeting the EGFR signaling pathway.

Discussion
Pancreatic cancer ranks fourth among cancer related deaths, and the disappointing five-year survival rate of below 5% results from drug resistance to all known therapies, as well as from disease presentation at a late stage when PDAC is already metastatic 5,9 . Most PDAC patients suffer from recurrence within 24 months and die of the progressively worsening treatment-resistant cancer. Recently, major treatment breakthroughs in many difficult-to-treat cancers, such as melanoma, have been facilitated by the identification of actionable mutant oncogenic driver genes 73 . While there has been an increasing understanding of the underlying biology of pancreatic cancer, unfortunately, until now, there are no actionable therapeutic targets for PDAC. Meanwhile, only modest improvement in effective systemic chemotherapy has been attained in pancreatic cancer. Therefore, pancreatic cancer is still one of the most lethal cancers with a dismal 5-year survival less than 5% 5 . Thus, there is an unmet clinical need to develop more effective and safe treatment for the clinical management of PDAC patients. The repurposing of existing non-cancer drugs represents a cost-effective alternative to develop new treatment options for cancer patients with high unmet medical needs. In this study, we demonstrate that monensin may be repurposed to treat chemo-resistant pancreatic cancer. Our results suggest the monensin may act synergistically with gemcitabine or erlotinib in combination chemotherapy for the treatment of drug-resistant pancreatic cancer.
Monensin was discovered as a polyether innophore antibiotic over half century ago 18,19 . It has a rather favorable biosafety profile as monensin is widely used in cattle and poultry feed 18,19,29 . It was reported that malignant cells were approximately 20 times more sensitive to monensin than normal cells 74 . We and others showed that monensin exhibits anti-proliferative effects on several other types of cancer, including renal cancer, lung cancer, colon cancer, myeloma, prostate cancer, and ovarian cancer cells 19,29,[74][75][76][77][78][79][80] . In this study, we further demonstrate that monensin exerts potent anticancer activity in chemo-resistant pancreatic cancer cells by inhibiting its proliferation, cell cycle progression, and cell migration and by inducing apoptosis. Moreover, monensin acts synergistically with gemcitabine or erlotinib to inhibit cell proliferation and induce apoptosis in chemo-resistant pancreatic cancer cells. Our in vivo study in xenograft tumor model of PDAC cells further validates the biosafety and anticancer efficacy of monensin as a repurposed anti-PDAC agent.
Mechanistically, monensin may accomplish its anticancer effect by targeting multiple signaling pathways, particularly the EGFR signaling pathway. Given the fact that monensin exerts its anti-proliferation effect in chemo-resistant pancreatic cancer cells at very low micromole concentrations when compared with gemcitabine or erlotinib. While more mechanism-based studies are needed, it was reported that monensin can reduce the expression of cyclin A, CDK6, and cyclin D1while inducing programmed cell death-related genes, such as caspase-3, caspase-8, Bax, and mitochondria transmembrane potential in some types of human cancer lines [76][77][78][79][80] . It has been recently shown that monensin can suppress Wnt signaling in colorectal cancer cells 81 , and EGFR signaling in ovarian cancer cells 29 .
In this study, we also investigated the effect of monensin on multiple cancer-related pathways and found that monensin can inhibit E2F/DP1, STAT1/2, NFκB, AP-1 and Elk-1/SRF pathways. Furthermore, the expression of EGFR and its downstream genes, such as RAF1 and BRAF, is effectively suppressed by monensin. The above findings suggest that monensin may exert its potent proliferation suppression effect through the inhibition of multiple growth factor-induced signal pathways, especially EGFR, which is found overexpressed in the clinical samples of pancreatic cancer. Interestingly, it was reported that monensin may impact the endocytic recycling pathway of EGFR 19,82 . We further demonstrate that the EGFR expression level in the PDAC xenograft tumors was significantly inhibited by monensin. Therefore, our findings strongly demonstrate that monensin can exert its potent anticancer activity in chemo-resistant PDAC cells at least in part by targeting the EGFR signaling pathway. It is conceivable that, even though both target EGFR, the mode of action for monensin should be distinct from that of erlotinib's since monensin can synergize with erlotinib in PDAC cells. Collectively, our findings suggest monensin can be repurposed to treat pancreatic cancer, although its safety and anticancer efficacy need to be further validated in preclinical and clinical studies.