This article is mentioned in:

Introduction

Pancreatic cancer is the fourth or fifth most common cause of cancer-related death in most western industrialized countries. The disease has a poor prognosis with a 5-year survival rate of <5%. In unresectable cases, chemo- or radiotherapy is used for palliation of symptoms and improvement of survival (1,2). Gemcitabine is currently the standard systemic treatment for unresectable and advanced pancreatic cancer, which can extend patient survival by 5–6 months (3). To date, a more satisfactory outcome has not been reported when gemcitabine is used in combination with other cytotoxic drugs (3–5).

The molecular basis for the resistance of pancreatic cancers to chemotherapy is not fully understood. It appears to result from various mechanisms that can each influence tumor resistance to gemcitabine. One form of drug resistance, termed multidrug resistance (MDR), is linked to the expression of proteins that act as drug efflux pumps. MDR is characterized by the development of broad cross-resistance to functionally and structurally unrelated drugs. This enhanced efflux of drugs is mediated by a super family of membrane glycoproteins, the ATP-binding cassette (ABC) transporters. The ABCB superfamily includes a 170-kDa transporter protein (ABCB1), also termed P-glycoprotein (P-gp). P-gp is comprised of the protein product of the human MDR1 gene in complex with the multidrug resistance-associated protein (MRP) (6). The expression of P-gp correlates with both the decreased accumulation of drugs, and the degree of chemoresistance in many different human cancer cells (7–9). Generally, 40–50% of patient tumors exhibit overexpression of P-gp (10). Although gemcitabine is a substrate for the ATP-dependent efflux pump (11), it is predominantly transported into the cell via facilitated diffusion mediated by the equilibrative nucleoside transporter (ENT), and sodium-dependent transporters (concentrative nucleoside transporter, CNT) (12,13). Pancreatic tumor cells can express high levels of ENT1, whereas members of the CNT family are present at only negligible levels (14).

Emerging evidence suggests that cancer stem cells (CSCs), a small subset of undifferentiated cells found within tumors, play a key role in tumorigenicity and malignancy (15). SP cells represent a small subpopulation of tumor cells that have properties associated with CSC in that they can rapidly efflux lipophilic fluorescent dyes producing a characteristic profile in fluorescence-activated flow cytometric analysis (16). SP cells are generally defined by their ability to efflux the fluorescent DNA binding dye Hoechst 33342 (H33342) and the differential emission spectra of this dye when it binds chromatin (16). The pumps responsible for dye efflux are attributed to the ABC superfamily transporters, including the MDR1/P-glycoprotein, ABCG2, MRP1 and ABCA2 (17–20). These pumps also transport chemotoxic drugs out of the cell, including vinblastine, doxorubicin, daunorubicin and paclitaxel (21). SP cells exhibit increased chemoresistance following in vitro exposure to gemcitabine (22,23). The FACS-based assay used to detect the presence of side populations is also currently under evaluation as a general method to identify and isolate CSCs subpopulations within tumor samples.

Verapamil is a calcium channel blocker that is utilized clinically to treat cardiac arrhythmias (24). It is also a first generation inhibitor of P-gp (25). When combined with chemotherapeutic agents, verapamil can help to promote intra-cellular drug accumulation (26). This has been demonstrated in non-small cell lung cancer, colorectal carcinoma, leukemia, and neuroblastoma cell lines (27–30). Based on this ability of verapamil to inhibit P-gp transport activity, it can also be used as an ‘SP’ blocker in the Hoechst 33342 assay as it will substantially reduce SP cells as visualized by flow cytometry analysis. Based on these observations, we hypothesized that verapamil treatment may directly exert anti-SP effects and therefore enhanced gemcitabine sensitivity in pancreatic cancer.

In this study, the biological characteristics of CSCs in pancreatic cancer SP cells including their self-renewal ability, resistance to gemcitabine, and general tumorigenicity were investigated in the context of verapamil treatment.

Materials and methods

Human pancreatic cancer cells and culture conditions

Human pancreatic adenocarcinoma cell lines L3.6pl (31) and AsPC-1 (American Tissue Culture Collection) were maintained in Dulbecco's minimal essential medium (D-MEM; Invitrogen GmbH, Karlsruhe, Germany), supplemented with 10% fetal bovine serum (Biochrom AG, Berlin, Germany), 2% MEM vitamin mixture (PAN Biotech GmbH, Aidenbach, Germany), 2% MEM NEAA (PAN Biotech GmbH), 1% penicillin streptomycin (PAN Biotech GmbH, Aidenbach, Germany) and 2% glutamax (Invitrogen GmbH). Cells were incubated in a humidified incubator (37°C, 5% CO2), grown in cell culture flasks, and passaged on reaching 70–80% confluence.

A gemcitabine-resistant pancreatic cancer cell line, termed L3.6plGres, was developed from the parental L3.6pl cell line by gradually increasing the concentration of gemcitabine (Gemzar; Lilly Deutschland GmbH, Giessen, Germany) in the cultured cells. Gemcitabine was first added at a concentration of 0.5 ng/ml (based on the IC50 value of L3.6pl). When the cells reached exponential growth, they were subcultured for two additional passages with 0.5 ng/ml gemcitabine or until the cells grew stably. The concentration of gemcitabine was then increased to 100 ng/ml and the cells were passaged until a stable gemcitabine-resistant pancreatic cancer cell line (L3.6plGres) was established.

Isolation of SP- and non-SP-cell fractions from L3.6plGres and AsPC-1 cell lines

SP- and non-SP-cell fractions were identified and isolated using a modified protocol described by Goodell et al (16). Briefly, 1×106/ml cells were re-suspended in D-MEM containing 2% fetal bovine serum and labeled with H33342 (Sigma-Aldrich GmbH, Steinheim, Germany) at a concentration of 2.5 μg/ml for 60 min in 37°C water bath, either alone or with 225 μM verapamil hydrochloride (Sigma-Aldrich GmbH). After 60 min the cells were centrifuged (300 g, 4°C) for 5 min, and then resuspended in ice-cold PBS containing 2% fetal bovine serum. The cells were passed through a 40-μm mesh filter and maintained at 4°C in the dark until flow cytometry analysis or sorting. Cells were counter-stained with 10 μg/ml propidium iodide to label dead cells, and the entire preparation was then analyzed using a BD-LSRII flow cytometer (BD Biosciences, Heidelberg, Germany) and FlowJo software (Treestar Inc., Ashland, OR, USA), or sorted using a MoFlo cell sorter with the Summit 4.3 software (Beckmann Coulter GmbH, Krefeld, Germany). Hoechst dye was excited at 355 nm (32), and fluorescence was measured at two wavelengths using a 450/50-nm (blue) band-pass filter and a 670/30-nm (33) long-pass edge filter. Following isolation the SP and non-SP cell fractions were used for in vitro and in vivo assays.

Cell viability and proliferation assay

Trypan blue (Sigma-Aldrich) staining was used to test for cell viability. The dye stains dead cells, and livings are distinguished by their ability to exclude the dye performing phase contrast microscopy. Cell viability was calculated using the following formula: Cell viability = unstained cells/unstained + trypan blue stained cells × 100%. Cell proliferation was measured using the Cell Counting Kit-8 (Dojindo Laboratories, Kumamoto, Japan) according to the manufacturer's instructions. In this assay 5,000–8,000 cells/well were plated in a 96-well plate and grown overnight, and then treated for 24 h with gemcitabine or verapamil. Cell proliferation was then determined using a VersaMax tunable microplate reader and Softmaxpro 5.2 software for data analysis (Molecular Devices, Sunnyvale, CA, USA).

Apoptosis assay

Cell apoptosis was analyzed using an Annexin V-FITC assay (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany) according to the manufacturer's instructions. After determination of cell numbers, 1×106 cells were washed with binding buffer and centrifuged. The cell pellet was resuspended in binding buffer, and 10 μl of Annexin V-FITC per 106 cells was added, mixed, and the preparation was incubated for 15 min in the dark at room temperature. The cells were then washed, and the cell pellet resuspended in binding buffer. The PI solution was added immediately prior to analysis using a BD-LSRII flow cytometer (BD Biosciences, Heidelberg, Germany) and FlowJo software version 7.6 (Treestar Inc.). Experiments were repeated at least three times.

Colony formation assay

To determine the ability of cells to form colonies, 500 cells in 2 ml of D-MEM medium were seeded into each well of a 6-well plate. The medium was changed two times per week and the assay was stopped when colonies were clearly visible without using a microscope (i.e., each single colony comprised approximately ≥50 cells). The resulting colonies were stained with 0.1% crystal violet and counted.

Flow cytometry analysis

For this staining procedure, the cells were incubated in the dark and on ice. Firstly they were blocked with PBS containing 0.5% albumin bovine (BSA) and 0.02% NaN3 with FCR blocking reagent (Miltenyi Biotec) for 15 min. Cells were then stained with the P-glycoprotein mouse mAb primary antibody (clone C219, Calbiochem, Darmstadt, Germany) for 45 min on ice. Afterwards they were incubated with the fluorescence-conjugated secondary antibody, goat anti-mouse IgG FITC (Abcam, Cambridge, UK) for 45 min. FITC isotype mouse immunoglobulins were used as negative controls. FITC was excited using a 488-nm laser and detected by a 530/30 band-pass filters. Dead cells were excluded by gating on forward and side scatter and eliminating PI-positive population cells. All the data were analyzed using FlowJo software version 7.6 (Treestar Inc.).

Western blot assay

The proteins ENT1 and P-gp were assayed by immunoblotting. Cells were resuspended in ice-cold RIPA buffer supplemented with a cocktail of protease/phosphatase inhibitors (Roche, Mannheim, Germany). Cells were further incubated on ice for 10 min and centrifuged at 14,000 g at 4°C for 10 min. After determination of the protein concentration using the BCA protein assay (Thermo Fisher Scientific, Rockford, IL, USA), an equal amount of protein was run on polyacrylamide gels and transferred semidry to polyvinylidene difluoride membranes (Amersham, Braunschweig, Germany). The corresponding primary antibody was incubated at 4°C overnight. After blocking for 2 h, the secondary antibody was added and incubated for 2 h at room temperature, following by washing. An enhanced chemiluminescense system(Amersham) was used for detection. Afterwards the membranes were reused for β-actin staining (Sigma-Aldrich GmbH) to ensure equal protein levels. The polyclonal goat anti-mouse/rabbit immunoglobulin HRP (Dako, Glostrup, Denmark) was used according to the manufacturer's instructions. For detection of the ENT1 protein, the anti-ENT1 antibody (Abcam) was used at 1:500 in 5% BSA-TBST buffer. For P-gp protein detection, the monoclonal mouse C219 antibody (Calbiochem) was used at 1:100 dilutions in 7.5% skim dried milk-PBST buffer. This antibody recognizes the two P-gp isoforms (34).

Orthotopic pancreatic cancer mouse model

Approval by the animal rights commission of the state of Bavaria, Germany was obtained for all animal experiments. Male athymic Balb/c nu/nu mice were purchased from Charles River, Inc. (Sulzfeld, Germany). Mice aged 6–8 weeks with an average weight of 20 g were maintained under a 12:12-h light-dark cycle and were used for the orthotopic pancreatic cancer mouse model. They were anesthetized using ketamine [100 mg/kg body weight (BW), xylazine (5 mg/kg BW] and atropine. The operation was carried out in a sterile manner. A 1-cm left abdominal flank incision was made, the spleen was exteriorized and 1×105 isolated SP- and non-SP L3.6plGres cells were injected into the subcapsular region of the pancreas. Before injection, cell viability was assessed by trypan blue staining and only cell preparations that showed ≥95% viability were used for injection. Mice were divided into an SP control group, non-SP control group and different drug treatment SP-groups. Four weeks after orthotopic implantation, therapy was initiated. One treatment group obtained daily (every workday) intraperitoneal injection of low concentration of verapamil (200 μM, 0.5 mg/Kg BW), the second treatment group received a higher concentration of verapamil (10 mM, 25 mg/Kg BW). Nine weeks after the injection of tumor cells, all the mice were sacrificed and examined for orthotopic tumor growth and development of metastases. The pancreatic tumors as well as other organs were isolated, weighed, and then used for H&E and immunohistochemical staining.

Immunohistochemistry

Formaldehyde-fixed and paraffin-embedded tissues were serially sectioned at 3 μm and allowed to dry overnight. Sections were deparaffinized in xylene followed by a graded series of ethanol (100, 95 and 80%) and rehydrated in phosphate-buffered solution, pH 7.5. Paraffin-embedded tissues were used for the Ki67 proliferation index assay, microvascular density analysis, and the TUNEL assay. TUNEL assay was carried out using an in situ cell death detection kit, Fluorescein (Roche Diagnostics GmbH, Mannheim, Germany), according to the manufacturer's instructions. The antibodies used for paraffin-embedded tissues were monoclonal rabbit anti-Ki67 antibody (ab 16667, Abcam) and polyclonal rabbit anti-CD31 (ab 28364, Abcam). The primary antibodies were diluted in PBS containing 3% bovine serum albumin (BSA). The sample slides were treated for 20 min with blocking solution (8% goat serum or rabbit serum in PBS with 3% BSA) before the primary antibody was applied. Endogenous peroxidase was blocked by incubation with 3% hydrogen peroxide (H2O2). Endogenous avidin and biotin were blocked using the Avidin/Biotin Blocking kit (Vector, USA).

Overnight incubation with the primary antibodies was followed by incubation with the respective biotinylated secondary antibodies (goat anti-rabbit, BA-1000, Vector), followed by the ABC reagent for signal amplification (Vectastain ABC-Peroxidase kits, PK-4000, Vector). Between the incubation steps, the slides were washed in TBS. 3,3′-diaminobenzidine (DAB, Dako, USA) was used to develop the color. Slides were counter-stained with hematoxylin, mounted in Kaisers Glycerinegelatine (Merck, Germany), and covered.

For quantification of the staining intensity, each index (Ki67, microvascular density CD31, and TUNEL) was evaluated in a blinded manner. Slides were observed under a high/low magnification scope (×200/×100), each one was evaluated within 3 fields, and the data were analyzed as mean positive signal (Ki67-positive cells, amount of microvessels, cells with strong FITC-fluorescence) of 3 fields.

Statistical analysis

Statistical evaluation was performed using the paired Student's t-test or ANOVA test (Microcal Origin) with p<0.05 considered to be statistically significant [p<0.05 (&); p<0.01 (#) p<0.001 (*); p<0.0001 (**)]. GraphPad Prism® 5.0 or Microsoft excel 2010 software were used to generate graphs and tables.

Results

Side population cells are enriched in L3.6plGres cells and show higher colony formation

The human pancreatic adenocarcinoma cell line L3.6pl was continuously cultured with gemcitabine, starting at 0.5 ng/ml, which was then gradually increased to 100 ng/ml over multiple cell passages to eventually develop a gemcitabine-resistant version of the cell line (L3.6plGres). The 24 h IC50 increased from 6.1±0.9 ng/ml in the parental L3.6pl cells, to 498.8±3.2 ng/ml in the L3.6plGres cells (L3.6pl vs. L3.6plGres p<1E−9). The general morphology of L3.6plGres cells was also changed with gemcitabine resistance into large fibroblastoid-like tumor cells (Fig. 1), while the parental L3.6pl cell line maintained its original round shape. By contrast, the AsPC-1 line, which was largely resistant to gemcitabine and presented with a spindle-like shape, did not show a change in morphology under gemcitabine selection (data not shown).

Figure 1 Microscopic images of L3.6pl and L3.6plGres following gemcitabine treatment. (A) L3.6pl cells: small and round cells with some irregular relatively larger cells. (B) L3.6plGres cells after treatment with gemcitabine over 30 days: showing enlarged cell shape. (C) L3.6plGres cells after treatment with gemcitabine over 60 days: the majority of L3.6plGres cells have changed their morphology into a fibroblastoid type with loss of polarity, increased intercellular separation and pseudopodia, with a minority of small round cells interspersed.

The L3.6pl and AsPC-1 cell lines were assayed for the presence of SP-cells. Verapamil hydrochloride, which blocks transporters of the ABC family and abrogates the ability of cells to efflux the dye, served as an SP cell control in the H33342 assay (SP cells will disappear or decrease in the presence of ABCG2 inhibitors owing to inhibition of Hoechst efflux from the cells). Both cell lines were found to contain a distinct proportion of SP cells. The percentage of SP cells increased from 0.9±0.22% in L3.6pl, to 5.38±0.99% in the L3.6plGres cells in response to continuous gemcitabine treatment. This proportion of SP cells could be diminished by treatment with verapamil hydrochloride. In the AsPC-1 cell line, 21.35±3.48% SP cells were identified, which was found to diminish to 3.56±0.87% in response to verapamil blockade (p<5E−10) (Fig. 2).

Figure 2 Side population detection. FACS analysis showing proportion of SP cells in L3.6pl (0.9±0.22%) as compared to L3.6plGres cells (5.38±0.99%). Using verapamil, the percentage of SP-cells was significantly reduced to 0.16±0.11% (L3.6pl versus L3.6pl + verapamil, p<5E−10). More SP cells (21.35±3.48%) were identified in the AsPC-1 cell line. Verapamil induced a significant reduction of SP cells to 3.56±0.87% (AsPC-1 versus AsPC-1 + verapamil, p<5E−10). Data are presented as mean ± standard deviation.

We then compared the colony formation ability of L3.6plGres, L3.6pl, as well as L3.6plGres-SP and NSP cells. The assay showed that L3.6plGres cells and L3.6plGres-SP cells have significantly higher colony formation ability, and thus enhanced CSC-like characteristics when compared to the L3.6pl or L3.6plGres-NSP cells (Fig. 3).

Figure 3 Chemo-resistant SP cells exhibit stronger colony formation ability. SP-enriched L3.6plGres cells are able to generate more colonies than L3.6pl. (A) The upper three wells were seeded with L3.6pl cells while the lower three wells were seeded with L3.6plGres cells. There were many more visible colonies in the lower three wells than the upper part. (B) L3.6plGres can form more colonies per 500 single cells than L3.6pl (#p<0.01). SP cells in L3.6plGres show enhanced colony formation ability relative to NSP cells. (C) The upper three wells were seeded with L3.6plGres-NSP cells, while the lower three wells were seeded with L3.6plGres-SP cells. There were many more visible colonies in L3.6plGres-SP than the L3.6plGres-NSP cells. (D) L3.6plGres-SP can form more colonies per 500 single cells than L3.6plGres-NSP (**p<0.0001).

The gemcitabine resistance of L3.6plGres cells was further validated by cell viability assays. The results show that L3.6plGres-SP cells have enhanced resistance to gemcitabine relative to respective NSP cells (Fig. 4).

Figure 4 L3.6plGres-SP cells are more resistant to gemcitabine than L3.6plGres-NSP cells. Immediately after FACS sorting, SP as well as NSP subpopulation cells from L3.6plGres were seeded. Different concentrations of gemcitabine were applied over 24 h. L3.6plGres SP cells maintained their gemcitabine resistance compared to NSP cells (#p<0.01, **p<0.0001).

Verapamil can effectively inhibit L3.6plGres and AsPC-1 SP cell proliferation

Since the percentage of SP cells in the parental L3.6pl cell line was relatively low, it was difficult to harvest sufficient cell numbers by FACS sorting for cell proliferation assays. Therefore, the proliferation capacity of the L3.6plGres cell line (enriched SP cells) was compared to control L3.6pl cells. In parallel, the SP and Non-SP cells from the AsPC-1 cell line were directly compared.

To investigate whether verapamil alone is effective against SP cells, we analyzed its effect on L3.6pl- and L3.6plGres-, AsPC-1-SP and -NSP cell proliferation. All of the cell lines tested showed a concentration-dependent reduction in cell proliferation. In response to 50, 100, 150 and 200 μM of verapamil, the L3.6plGres- and AsPC-1-SP cells were more sensitive to verapamil effects than L3.6pl- and AsPC-1-NSP cells (Fig. 5).

Figure 5 Verapamil alone can effectively inhibit the proliferation of L3.6plGres and AsPC-1 SP cells in vitro. (A) Treatment of L3.6pl and L3.6plGres cells for 24 h with increasing concentrations of verapamil led to a dose-dependent significant reduction of cell proliferation in L3.6plGres cells, which was much more sensitive to verapamil than L3.6pl cells (*p<0.001, **p<0.0001). (B) FACS sorted SP and NSP AsPC-1 cells were treated with increasing concentrations of verapamil. Cell proliferation was significantly reduced in AsPC-1-SP cells than AsPC-1-NSP cells (**p<0.0001).

Pro-apoptotic effect of verapamil in combination with gemcitabine in L3.6pl and L3.6plGres cells

The potential additive effect of verapamil in combination with gemcitabine was then investigated. L3.6plGres-SP cells were treated separately with verapamil (50 μM), gemcitabine (10 ng/ml), and in combination. Using FACS analysis, the level of apoptotic cells after combined treatment (at 24 h) was found to increase relative to treatment with verapamil or gemcitabine alone (Fig. 6A). In dose response experiments, the apoptosis of L3.6pl and L3.6plGres cells in response to increasing concentrations of gemcitabine was determined, while maintaining the concentration of verapamil at 50 μM. A dose-dependent increase in apoptosis was observed in L3.6pl cells, but not in the L3.6plGres cells (Fig. 6B and C).

Figure 6 Verapamil alone or combined with gemcitabine on SP apoptosis. (A) Sorted L3.6plGres-SP cells were treated with: 1, verapamil (50 μM); 2, gemcitabine (10 ng/ml); or 3, verapamil (50 μM) combined with gemcitabine (10 ng/ml) for 24 h. Compared to verapamil and gemcitabine treatment alone the proportion of apoptotic cells was substantially higher following combined treatment. (B) L3.6pl and L3.6plGres cells were treated over 24 h with different concentrations of gemcitabine (0, 10, 40 and 100 ng/ml) alone or in combination with verapamil (50 μM). Under combination therapy, the proportion of apoptotic L3.6pl cells increased in a dose-dependent manner (&p<0.05, #p<0.01, *p<0.001). (C) The percentage of apoptotic L3.6plGres was only slightly increasing following combination therapy.

The expression level of drug transporter proteins in L3.6pl and L3.6plGres

The expression of the drug transporter proteins P-gp and ENT1 were then analyzed by western blotting. The results revealed that the expression of P-gp was increased in L3.6plGres cells as compared to L3.6pl cells, while expression of ENT1 in L3.6plGres cells was found to be substantially lower than in L3.6pl cells. The expression of P-gp in both cell lines was found to decrease after 24-hour treatment with verapamil (50 μM), and in particular, showed a marked reduction in the L3.6plGres cells. The expression of ENT1 remained unchanged after verapamil treatment (Fig. 7).

Figure 7 Expression of drug transporter proteins in L3.6pl and L3.6plGres. (A) After treatment of L3.6pl and L3.6plGres cells with verapamil (50 μM) for 0, 6 and 24 h, all cells were analyzed by FACS. The proportion of P-glycoprotein positive cells decreased after treatment in both cell lines, however, significantly more in L3.6plGres (p<0.01). (B) The protein lysates from: 1, L3.6pl; 2, L3.6plGres; 3, L3.6pl treated with verapamil (50 μM) for 6 h; 4, L3.6pl treated with verapamil (50 μM) for 24 h; 5, L3.6plGres treated with verapamil (50 μM) for 6 h; 6, L3.6plGres treated with verapamil (50 μM) for 24 h. The expression of P-glycoprotein was significantly higher in L3.6plGres, than L3.6pl cells. Under treatment of verapamil for 24 h, the expression level of P-glycoprotein decreased in both cell lines, and showed a marked reduction in the L3.6plGres cells. (C) The expression level of ENT1 before/after verapamil treatment in L3.6pl and L3.6plGres cells. The protein lysates from: 1, L3.6pl; 2, L3.6plGres; 3, L3.6pl treated with verapamil (50 μM) for 24 h; 4, L3.6pl treated with verapamil (100 μM) for 24 h; 5, L3.6pl treated with verapamil (200 μM) for 24 h; 6, L3.6plGres treated with verapamil (50 μM) for 24 h; 7, L3.6plGres treated with verapamil (100 μM) for 24 h; 8, L3.6plGres treated with verapamil (200 μM) for 24 h. The results indicate that the expression of ENT1 was lower in L3.6plGres cells than in L3.6pl cells, and were unchanged under different concentrations (50, 100 and 200 μM) of verapamil.

Verapamil can effectively inhibit L3.6plGres -SP tumor growth in vivo

L3.6plGres-SP and NSP cells were injected into the pancreas of athymic Balb/c nu/nu mice. The following experimental groups were compared: group 1, L3.6plGres -SP control group (5 mice); group 2, L3.6plGres-NSP cell control group (3 mice); group 3, L3.6plGres-SP treatment group with high concentration of verapamil (25 mg/kg BW, 10 mM, 3 mice); and group 4, L3.6plGres-SP treatment group with low concentration of verapamil (0.5 mg/kg BW, 200 μM, 4 mice). In all groups, verapamil treatment was initiated four weeks after orthotopic tumor cell injection. Nine weeks later all the mice were sacrificed. Tumors and other organs were then harvested for analysis.

Macroscopically, the L3.6plGres-SP tumors showed a more aggressive phenotype compared to the L3.6plGres-NSP tumors. Verapamil treatment (low/high concentrations) substantially inhibited tumor growth and metastasis of the L3.6plGres-SP cells (Fig. 8). No changes in body weight, or general behavior were observed in the verapamil treatment groups (groups 3 and 4).

Figure 8 Verapamil therapy suppresses tumor growth induced by L3.6plGres-SP cells in an orthotopic human pancreatic cancer nude mouse model. Verapamil treatment significantly inhibited tumor growth in a dose-independent manner. (A) Macroscopic analysis of tumors from L3.6plGres-SP cells, L3.6plGres-NSP cells and L3.6plGres-SP cells with verapamil treatment (53). Verapamil resulted in a significant reduction of primary pancreatic tumor growth in a nude mouse model. (B) Tumor growth curves (depicted as longest diameter of a two-dimensional transcutaneous measurement) of L3.6plGres-SP cell implanted mice and mice under verapamil treatment. (C) Primary pancreatic tumor weight (including normal pancreatic tissue) upon sacrifice depicted as mean ± SD (**p<0.0001).

Immunohistochemical staining was used to establish a proliferation index, microvascular density, and apoptosis level in the experimental pancreatic tumors derived from the L3.6plGres-SP control and verapamil treatment groups (Fig. 9). The median percentage of Ki67+ cells within the tumors of the verapamil treatment group differed slightly from that seen in the L3.6plGres-SP group. The apoptotic signal was increased in the verapamil treatment group as compared to the L3.6plGres-SP control group. The microvascular density measured by CD31 staining revealed significantly reduced angiogenesis in tumors following verapamil treatment, as compared to the L3.6plGres-SP tumors in the control group. H&E staining showed large, pleomorphic cells with hyperchromatic nuclei in the L3.6plGres-SP control group.

Figure 9 Immunohistochemical analysis of L3.6plGres-SP induced tumors following verapamil treatment. H&E staining showed large, pleomorphic cells with hyperchromatic nuclei in the L3.6plGres-SP control group, while the normal, round, light nuclei composed most of the tumor tissue in the verapamil treatment group. Ki67+ signals are manifested by dark brown stained cells (53). The scatter plot of Ki67 index shows that the median percentage of Ki67+ proliferating cells within tumors of the verapamil treatment group decreased from that of the L3.6plGres-SP control group (#p=0.0027). TUNEL assay was used for apoptosis. Apoptotic cells are stained with fluorescent green. The scatter plot of the TUNEL index shows that the number of apoptotic cells in the verapamil treatment group is significantly higher than in L3.6plGres-SP control group (**p<0.0001). Immunohistochemical staining for CD31 was applied for detection of microvessel density. The red stained portions in the images are endothelial cells lining the vessels following anti-CD31 labeling. The scatter plot of microvessel density shows that the average microvessel density in the verapamil treatment group is significantly reduced compared to the L3.6plGres-SP control group (**p<0.0001).

Discussion

The general concept of cancer stem cells (CSCs) is a topic of increasing clinical interest. CSCs are proposed to show stem-like self-renewal and tumor initiation characteristics that are thought to help foster tumor recurrence and resistance to chemotherapy. CSCs can be enriched by several methods including by growth in serum-free defined media to induce sphere formation. They can also be directly isolated based on their expression of specific surface marker combinations, i.e., CD44+/CD24−/lin- for human breast cancers (35), EpCAMhigh/CD44+/CD166+ for colorectal cancer (36), CD34+/CD38− for acute myeloid leukemia (37), and Stro1+/CD105+/CD44+, a surface marker for bone sarcoma (38). CSC subpopulations can also be isolated by FACS sorting of side population (SP) cells based on their ability to efflux fluorescent dyes or chemotherapy reagents (39). SP cells have been widely linked to CSCs in various tumors and cancer cell lines, e.g., acute myeloid leukemia (40), neuroblastoma (19), melanoma (41), ovarian cancer (42), glioma cell lines (43), various human gastrointestinal cancer cell lines (44), and pancreatic cancer cell lines, including; MIAPaCa2, PANC-1, Capan-2, KP-1 NL and SW1990 (23,45–47).

In this study, we isolated and characterized SP cells from a highly metastatic pancreatic adenocarcinoma cell line L3.6pl, a chemotherapy-resistant variant of this line L3.6plGres, and from the AsPC-1 human pancreatic cell line (Fig. 2).

The SP-enriched L3.6plGres cells, which were shown in a previous study to have significantly increased ABCG2+, and CD24+ cells, showed higher colony formation ability in vitro than did the parental L3.6pl cells. L3.6plGres-SP cells also showed increased colony formation ability relative to L3.6plGres-NSP cells (Fig. 3) and exhibited increased gemcitabine chemotherapy resistance relative to the L3.6plGres-NSP cells suggesting enhanced stem cell-like properties (Fig. 4). SP and non-SP of AsPC-1 did not work as a good model for stem cell like SP study by our previous data (data not shown). SP proportion is highly abound in AsPC-1 and did not show statistic significance in tumorigenicity in vivo. Therefore, in this study, we focused on SP cells from L3.6pl and its corresponding resistant variant.

These observations were then validated in vivo. Following orthotopic implantation, significantly larger primary pancreatic tumors, and a higher incidence of liver metastases, were found in the L3.6plGres-SP cells, as compared to L3.6plGres-NSP cells, supporting the hypothesis that they show enhanced stem cell-like characteristics (Fig. 9).

Morphologic changes were associated with the establishment of gemcitabine-resistant L3.6pl cells. A fibroblastoid phenotype with loss of polarity, increased intercellular separation and pseudopodia was detected in the L3.6plGres cells (Fig. 1). This observation is consistent with the phenomenon of epithelial-to-mesenchymal transition (EMT). Molecular and phenotypic associations have been previously linked to the establishment of chemoresistance including the acquisition of an EMT-like cancer cell phenotype (48). SP cells have been previously associated with chemotherapy resistance and enhanced EMT (49,50).

Goodell et al (16) have shown that the molecular pumps responsible for H33342 efflux (used to differentiate SP from NSP cells) are generally members of the ABC superfamily including MDR1/P-gp, ABCG2, and MRP (7–9,51). Verapamil, the first generation P-gp inhibitor, has been reported to block dye efflux activity and potentially reverse the drug resistance caused by expression of P-gp (26). We show that verapamil effectively reduces Hoechst 33342 staining of FACS-associated SP cells in L3.6pl, L3.6plGres, and AsPC-1 cell lines (Fig. 2). Verapamil appears to target important characteristics associated with CRC/SP cells and may therefore show therapeutic efficacy for treating pancreatic cancer stem cells. We found that verapamil could also directly inhibit cell proliferation in a dose-dependent manner in vitro, and was able to inhibit SP cells more efficiently than NSP cells in both pancreatic cell lines tested (L3.6plGres and AsPC-1) (Fig. 5). Verapamil was further found to effectively prevent L3.6plGres-SP tumor growth in orthotopic in vivo tumor models (Fig. 8 and Table I). For a potential side effect, we checked the mice by bodyweight regular detection and animal behaviour, and speculated that no serious side effects existed. It is a limitation in this study that we did not have more biochemical parameters for a better evaluation of side effect.

Table I The incidence of primary pancreatic tumors as well as metastatic spread in L3.6plGres-SP, L3.6plGres-NSP and the therapy groups is indicated.a

Table I

The incidence of primary pancreatic tumors as well as metastatic spread in L3.6plGres-SP, L3.6plGres-NSP and the therapy groups is indicated.a

		L3.6plGres
			SP + verapamil
Group	SP	NSP	Low dose	High dose
Primary tumor	4/5	0/4	1/3	0/3
Liver metastasis	2/5	2/4	0/3	0/3

a High dose of verapamil (25 mg/kg) treatment to abolish detectable primary tumor and metastatic lesions.

The direct effects of verapamil on pancreatic cancer cells seen are supported by earlier reports that showed verapamil inhibition of proliferation and induced differentiation of human promyelocytic HL-60 cells (52), a growth inhibitory effect on human colonic tumor cells (54), and antiproliferative effects on brain tumor cells in vitro (55). A reversible, antiproliferative action of verapamil on human medulloblastoma, pinealoblastoma, glioma, and neuroblastoma tumor lines has also been reported (55).

SP cells isolated from L3.6plGres and AsPC-1 cell lines were more sensitive to verapamil than NSP cells. SP-enriched L3.6plGres cells expressed higher levels of P-gp than NSP L3.6pl cells. Verapamil treatment of L3.6plGres cells led to a significant reduction in P-gp expression as compared to L3.6pl cells (Fig. 7), which may explain in part the targeted effect of verapamil detected in SP cells.

Pro-apoptotic effects of verapamil were demonstrated by in vitro apoptosis assays (Fig. 6) and by ex vivo immunohistochemistry analysis of the tumor samples (Fig. 9). This enhanced apoptosis by verapamil may be explained by its actions as a calcium channel antagonist (52,54). Calcium ions (Ca2+) are cellular messengers that control cell and tissue physiology, and potentially ‘death’ signals (55,56). Calcium antagonists disrupt intracellular and extracellular Ca2+ equilibrium (57). Ca2+ is toxic at high concentrations, thus verapamil may help to foster apoptosis though disruption of the Ca2+ balance.

Other ABC superfamily members are also thought to contribute to SP chemotherapy resistance. Recently it was shown that verapamil and its derivative-NMeOHI(2) can trigger apoptosis though glutathione (GSH) extrusion mediated by MRP1 (58). The loss of GSH from cells can result in increased oxidative stress, a well-recognized trigger for apoptosis (59).

Although verapamil can enhance cytotoxicity when used in combination with chemotherapy agents, such as anthracyclines (28), paclitaxel (60), epipodophyllotoxins (61) and melphalan (62), little is known about the potential effects of verapamil when used alone or in combination with gemcitabine, the standard chemotherapy reagent for pancreatic cancer. We identified combined pro-apoptotic effects of verapamil and gemcitabine on L3.6pl but not on L3.6plGres cells. This may be explained in part by the expression level of ENT1, a major transporter for gemcitabine that was substantially lower in L3.6plGres cells compared to L3.6pl cells. The absence of ENT1 is also associated with a reduced survival in patients with gemcitabine treated pancreatic adenocarcinoma (63).

In conclusion, this study revealed that verapamil may act on pancreatic cancer cells in vitro by inhibiting P-gp transporters and inducing apoptosis of stem-like SP cells in L3.6pl, L3.6plGres and AsPC-1 cells. Verapamil can significantly inhibit pancreatic cancer tumor growth in vivo potentially by targeting stem-like side population cells.

Acknowledgements

The authors thank Anneli Tischmacher for her technical assistance. This study was supported by the German Research Society (DFG) grant BR 1614/7-1 and the DKTK/DKFZ 2013 (‘Stem cells in Oncology’) to C.J.B. L.Z., Y.Z. and Q.B. were supported by LMU-CSC (The China Scholarship Council) scholarship.

Abbreviations:

References