Molecular Mechanisms Involved in the Synergistic Interaction of the EZH2 Inhibitor 3-Deazaneplanocin A with Gemcitabine in Pancreatic Cancer Cells Free

This article is featured in Highlights of This Issue, p. 1619

With a 5-year survival rate of less than 5%, pancreatic ductal adenocarcinoma (PDAC) is the most lethal among the major solid tumors (1). Despite extensive clinical and scientific efforts, the grim prognosis of this disease has not improved over the past decade (2). The main reasons for the lack of efficient therapeutic strategies include invasive behavior and intrinsic resistance to most chemo/radio- and immunotherapy regimens (3).

Recently, PDAC emerged as a cancer stem cell (CSC)-driven disease. Pancreatic CSCs are highly tumorigenic and have the abilities to self-renew and produce differentiated progeny. Pancreatic CSCs possess the ability to undergo epithelial–mesenchymal transition (EMT; ref. 4) and form spheroids in serum-free medium containing well-defined growth factors. The PDAC spheroids showed increased proliferation, invasiveness, and metastasis (5). CSCs have also been associated with chemoresistance to gemcitabine (6–8). Against this background, studies on key determinants in CSCs can provide both biomarkers of PDAC aggressiveness and optimal novel targets to overcome chemoresistance.

Enhancer of Zeste Homolog-2 (EZH2) is a histone methyltransferase essential for self-renewal of CSCs (9). This protein is the catalytic subunit of Polycomb Repressive Complex 2 (PRC2), 1 of the 2 multimeric repressive complexes in the organization of the Polycomb group (PcG). PcG proteins act as important epigenetic mediators that can repress gene expression by forming multiple complexes leading to histone methylation, resulting in epigenetic control of gene expression (10, 11). In particular, EZH2 can silence several tumor suppressor genes by trimethylation at lysine 27 of histone H3 (H3K27me3; ref. 10), playing an important role in tumor development (12).

EZH2 is overexpressed in many cancer types, including PDACs. Recently, EZH2 expression has been associated with decreased E-cadherin expression and poor prognosis in patients with PDACs (13). Furthermore, Ougolkov and colleagues showed that EZH2 is an important factor in PDAC cell chemoresistance (14). In particular, EZH2 depletion by RNA interference sensitized PDAC cells to gemcitabine (14), which is used in the first-line treatment of PDACs. Because gemcitabine has very limited efficacy (5, 15), novel therapeutic strategies combining gemcitabine with targeted agents against EZH2 are warranted.

Recently, the cyclopentenyl analog of 3-deazaneplanocin A (DZNeP) was identified as a compound capable of reducing the levels of EZH2. This compound is a global histone methylation inhibitor suppressing the activity of S-adenosyl-l-homocysteine (AdoHcy) hydrolase, the enzyme responsible for the reversible hydrolysis of AdoHcy to adenosine and homocysteine (16). This results in intracellular accumulation of AdoHcy, which leads to inhibition of the S-adenosyl-l-methionine–dependent lysine methyltransferase activity. S-adenosyl-methionine metabolism represents a key cellular mechanism for methyl group donation for a variety of methylation-dependent metabolic processes that are disrupted by DZNeP, including production of methyltransferases such as EZH2 (17).

In the present study, we evaluated the EZH2 expression in PDAC tissues and cells and the growth inhibition by DZNeP in combination with gemcitabine in monolayer cell cultures and cells growing as spheroids in serum-free CSC medium. Furthermore, we characterized several factors, including cell-cycle perturbation, apoptosis induction, and inhibition of cell migration, as well as modulation of the expression of several genes involved in the DZNeP/gemcitabine interaction.

DZNeP was provided by Dr. Victor E. Marquez (NCI, NIH, Frederick, MD), whereas gemcitabine was a gift from Eli Lilly Corporation. The drugs were dissolved in sterile water and diluted in culture medium before use. RPMI medium, FBS, penicillin (50 IU/mL), and streptomycin (50 μg/mL) were from Gibco. All other chemicals were purchased from Sigma-Aldrich.

Eight PDAC cell lines (PANC-1, MIA-PaCa-2, BxPc3, Capan-1, PL45, HPAC, HPAF-II, and CFPAC-1), the human pancreatic duct epithelial-like cell line hTERT-HPNE, and skin fibroblasts Hs27 were obtained from the American Type Culture Collection, whereas 7 primary PDAC cultures (LPC006, LPC028, LPC033, LPC067, LPC111, LPC167, and PP437) were isolated from patients at the University Hospital of Pisa (Pisa, Italy), as described previously (18). Cells were cultured in RPMI-1640, supplemented with 10% heat-inactivated FBS and 1% streptomycin/penicillin at 37°C, and harvested with trypsin-EDTA in their exponentially growing phase. The cell lines were tested for their authentication by PCR profiling.

Total RNA was extracted using the TRIREAGENT-LS (Invitrogen), and its yields and purity were checked at 260—to 280 nm with NanoDrop 1000 Detector (NanoDrop Technologies). To prevent RNA degradation, the cells were harvested quickly on ice. One microgram of RNA was reverse-transcribed using the DyNAmo cDNA Synthesis Kit (ThermoScientific), according to the manufacturers' instruction. To evaluate whether the expression of EZH2 was similar in the primary cells and their originator tumors, we also extracted RNA from these 7 tumors, after laser microdissection with a Leica LMD6000 instrument (Leica), using the QIAamp RNA Micro Kit (Qiagen), as described previously (18).

Primers and probes to specifically amplify EZH2, hENT1, CD133, and E-cadherin (Hs01016789_m1, Hs00191940_m1, Hs01009250_m1, and Hs01023894_m1, respectively) were obtained from Applied Biosystems.

The quantitative real-time PCR reactions were carried out in the ABIPRISM 7500 Sequence Detection System instrument (Applied Biosystems).

We conducted a preliminary analysis of 3 housekeeping genes [β-actin, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and β-2-microglobulin] in all our PDAC cells. Because the values of β-actin were the closest to the geometric mean values of these housekeeping genes, we used this housekeeping gene for the normalization of all the following analyses. Preliminary experiments were carried out to show that the efficiencies of amplification of target and reference genes are approximately equal (18).

To evaluate modulation of EZH2 and H3K27me3 protein expression, the PANC-1, MIA-PaCa-2, and LPC006 cells were treated with 5 μmol/L DZNeP for 72 hours, as reported (19). Blotting procedures were conducted as described previously (18). Membranes were incubated overnight at 4°C with purified mouse anti-EZH2 mAb (BD Biosciences) at 1:1,000 dilution in blocking solution (Rockland in PBS-T), rabbit anti-H3K27me3 (1:1,000; Upstate Biotechnology), rabbit anti-hENT1 and rabbit anti-hCNT1 (1:1,000, kindly provided by M. Pastor-Anglada), and mouse anti-β-actin (1:50,000; Sigma-Aldrich). The membrane was then probed for 1 hour with the goat anti-mouse InfraRedDye (1:10,000; Westburg) or goat anti-rabbit InfraRedDye (1:10,000; Westburg) secondary antibodies. Fluorescent proteins were detected by an Odyssey Infrared Imager (LI-COR Biosciences), at 84-μm resolution, 0-mm offset, using high-quality settings. Then, the intensities of protein bands were quantified using the Odyssey v.3.0 Software (LI-COR Biosciences).

The LPC006 cells were grown in Chamber Slides System (Lab-Tek) in a humidified incubator. After 48 hours, the cells were fixed with 70% ethanol for 10 minutes. Immunocytochemistry was conducted using a monoclonal mouse anti-human E-cadherin antibody (Cell Signaling; EuroClone; 4°C overnight incubation and 1:30 dilution in PBS). Cells were then stained with avidin-biotin-peroxidase complex (UltraMarque HRP Detection). Negative controls were obtained replacing the primary antibody with PBS. The sections were reviewed and scored blindly by comparing the staining of treated cells versus untreated cells (positive control, basal expression), using a system based on staining intensity and on the number of positively stained cells, as described (18).

Cell growth-inhibitory effects of the DZNeP, gemcitabine, and their combination were studied using the sulforhodamine B assay. Cells were seeded in triplicate at 5 × 10³ cells per well and kept at 37°C for 24 hours. Then, the cells were treated for 72 hours with DZNeP (0.001–20 μmol/L), gemcitabine (0.001–500 nmol/L), and DZNeP at fixed concentration of 5 μmol/L simultaneously with gemcitabine (0.001–500 nmol/L). After 72 hours, plates were processed for the sulforhodamine B assay as described earlier (20). To determine whether the drug could kill cells, we also measured optical density on the day of drug addition because cell kill could lead to a decrease in the optical density on day 0. Therefore, for the cell growth inhibition curves, the optical density after 72 hours was corrected for the mean optical density observed for the control wells at the day of drug addition (day 0 value). The 50% growth-inhibitory concentration (IC₅₀) was calculated by nonlinear least-squares curve fitting (GraphPad Prism, Intuitive Software for Science).

The drug interaction of DZNeP and gemcitabine was evaluated by the median drug–effect analysis method originally described by Chou and Talalay (21). Cell growth inhibition of the combination was compared with the cell growth inhibition of each drug alone using the combination index (CI), where CI < 0.9, CI = 0.9–1.1, and CI > 1.1 indicated synergistic, additive, and antagonistic effects, respectively. Data analysis was conducted using CalcuSyn software (Biosoft). Because we considered growth inhibition lower than 50% as not relevant, CI values at fraction affected (FA) of 0.5, 0.75, and 0.9 were averaged for each experiment, and this value was used to calculate the mean between experiments.

Cell-cycle modulation and cell death induced by treatments with 5 μmol/L DZNeP, gemcitabine at IC₅₀ values, and their combination was investigated using 10⁵ cells, after 24 and 72 hours, by propidium iodide (PI) staining using a FACSCalibur flow cytometer (Becton Dickinson). Data analysis was conducted with CellQuest (Becton Dickinson), whereas cell-cycle distribution was determined using ModFit (Verity Software), as described (22).

DZNeP, gemcitabine, and their combination were also characterized for their ability to induce cell death by evaluating the sub-G₁ region of the fluorescence-activated cell-sorting analysis (23) and by fluorescence microscopy analysis with bisbenzimide staining (22).

Migration was evaluated using the Leica DMI300B (Leica) migration station integrated with the Scratch Assay 6.1 software (Digital Cell Imaging Labs). Briefly, cells were plated at a density of 3 × 10⁴ cells per well onto 96-well plates, and, after 24 hours, artificial wound tracks were created by scraping with a specific scratcher within the confluent monolayers. After removal of the detached cells by gently washing with PBS, the cells were fed with fresh medium and exposed to 5 μmol/L DZNeP, gemcitabine at IC₅₀, or their combination. The ability of the cells to migrate into the wound area was assessed by comparing the pixels in the images taken at the beginning of the exposure (time 0), with those taken after 4, 6, 8, 24, and 48 hours.

PANC-1, MIA-PaCa-2, and LPC006 spheroids were established by seeding 10,000 cells/mL in Dulbecco's Modified Eagles' Media (DMEM):F12 + GlutaMAX-I (1:1) with insulin-transferrin-selenium (1:1,000; Gibco, Invitrogen), in 24-well ultra-low attachment plates (Corning), according to manufacturers' protocol, as described earlier (24). Spheroids were generated for 10 to 14 days and then harvested for growth inhibition studies in 96-well plates, as well as for RNA isolation.

After checking their growth rate and stability, the spheroids were treated with 5 μmol/L of DZNeP, IC₅₀ values of gemcitabine, and their combination for 72 hours. The cytotoxic effects were evaluated by measuring the size and number of spheroids with the microscope Leica DMI300B (Leica), taking 9 pictures for each well. Spheroid volume (V) was calculated from the geometric mean of the perpendicular diameters D = (D_max + D_min)/2, (V = 4/3π(D/2)3).

Analysis by liquid chromatography/tandem mass spectrometry (LC/MS-MS) was used to determine total cytosolic adenosine as well as total phosphorylated deoxynucleosides. The latter was calculated from the difference before and after alkaline phosphatase treatment, as described previously (25). Approximately, 2 × 10⁶ cells were seeded into 6-well plates and exposed to either 5 μmol/L DZNeP, gemcitabine at IC₅₀ values, or DZNeP/gemcitabine combination for 24 hours, before being snap-frozen as a pellet. Cell pellets were resuspended in a known aliquot of phosphate buffer and precipitated with excess isopropyl alcohol. The supernatant was removed and evaporated to dryness via freeze drying. The dry samples were reconstituted in 200 μL of water, and 20 μL aliquots were used for analysis. The remaining samples were treated quantitatively with alkaline phosphatase (4 units) at 37°C overnight. Chromatography was conducted using a Dionex Ultimate 3000 micro HPLC coupled via a Turbo spray ionization source to a SCIEX API 3000 mass spectrometer (Applied Biosystems). Data analysis was conducted with v.1.52 Analyst software (AB Sciex) controlled by Dionex Mass Spectrometry Link combined with Chromeleon management software modules (ThermoScientific).

All experiments were carried out in triplicate and repeated at least twice. Data were expressed as mean values ± SE and analyzed by the Student t test or ANOVA followed by the Tukey multiple comparison test. The level of significance was P < 0.05.

The mRNA expression of EZH2 was detectable in all PDAC cells, as well as in the originator tissues of the primary tumor cell cultures. This expression differed among cells, ranging from 0.100 arbitrary units (a.u.) in LPC006 cells to 0.644 a.u. in PL45 cells (Fig. 1A). The mean expression in the tumor cells (0.372 ± 0.174 a.u.) was similar to the median (0.360 a.u.) and significantly higher (P < 0.01) than the expression detected in hTERT-HPNE cells (0.037 a.u.) and in fibroblasts (0.027 a.u.).

EZH2 gene expression in the 7 primary tumor cells and their originator tumors showed a similar pattern and were highly correlated with Spearman analysis (R² = 0.89, P = 0.01).

PANC-1, MIA-PaCa-2, and LPC006 cells were selected for further studies because of previous studies on expression of the CD133 CSC marker (24, 26) and their differential levels of EZH2.

The expression of EZH2 was also studied at protein level, both in untreated cells and in cells treated with 5 μmol/L DZNeP for 24 or 72 hours. As shown in Fig. 1B, DZNeP reduced the expression of EZH2, especially after 72 hours (e.g., 48%, 32%, and 36% reduction of EZH2 in PANC-1, MIA-PaCa-2, and LPC006 cells, respectively). In addition, we investigated the expression of the H3K27me3 protein, which was also reduced after 72-hour exposure.

Finally, as previous studies reported that DZNeP is an S-adenosyl-homocysteine hydrolase inhibitor, we verified this inhibition and measured the intracellular concentration of its product adenosine in PANC-1 cells by a specific LC/MS-MS method (16). Adenosine was significantly reduced after 72-hour exposure to DZNeP (Fig. 1C), indicating that S-adenosyl-homocysteine hydrolase was significantly inhibited.

Treatment with DZNeP showed minimal growth inhibition in PANC-1 cells. More than 50% of these cells were still growing after exposure at the highest concentration (20 μmol/L). MIA-PaCa-2 and LPC006 cells were much more sensitive, with IC₅₀ values of 1.0 ± 0.3 and 0.10 ± 0.03 μmol/L, respectively (Fig. 2A–C). On the contrary, gemcitabine was highly cytotoxic, with IC₅₀ values of 17.9 ± 1.3 nmol/L (PANC-1), 5.9 ± 0.8 nmol/L (MIA-PaCa-2), and 7.2 ± 1.3 nmol/L (LPC006).

On the basis of these results, as well as the modulation of EZH2 protein by 5 μmol/L DZNeP, combination studies were conducted using a fixed concentration of DZNeP at 5 μmol/L. DZNeP enhanced the antiproliferative activity of gemcitabine, reducing the IC₅₀ values of gemcitabine to 5.02 ± 1.31, 0.12 ± 0.04, and 0.03 ± 0.01 nmol/L in PANC-1, MIA-PaCa-2, and LPC006. The mean CI showed slight-to-moderate synergism in LPC006 cells, and strong synergism in the PANC-1 and MIA-PaCa-2 cells (Fig. 2D).

To evaluate the mechanisms underlying this synergistic interaction, several biochemical analyses were conducted with the simultaneous combination, as detailed below.

DZNeP, gemcitabine, and their combination affected the cell cycle of PDAC cells (Table 1). In particular, DZNeP significantly reduced the percentage of PANC-1 cells in the G₂–M phase from 27% to 19%, after 72 hours, whereas gemcitabine increased cells in G₂–M phase to 36% (P < 0.05). The drug combination significantly reduced the percentage of cells in the G₂–M phase. Hence, DZNeP blocked PANC-1 cells in the G₁–S boundary. Conversely, gemcitabine reduced the cells in this phase, and no modulation was detected after drug combination in the PANC-1 cells. Similar perturbations of the cell cycle were observed in MIA-PaCa-2 cells. However, in LPC006, the percentage of cells in the G₂–M phase was significantly reduced both after DZNeP alone and DZNeP/gemcitabine combination. Moreover, the DZNeP/gemcitabine combination significantly increased cells in the S-phase while reducing cells in the G₀–G₁ phase.

Analysis of the sub-G₁ region showed that drug treatments significantly enhanced cell death compared with control (Table 1). In particular, MIA-PaCa-2 cells treated with the combination exhibited the largest sub-G₁ signal (e.g., 34%).

Further analysis with fluorescence microscopy showed that cells exposed to DZNeP, gemcitabine, and their combination presented a typical apoptotic morphology with cell shrinkage, nuclear condensation and fragmentation, and rupture of cells into debris, after 72-hour exposure. In all cell lines, 5% to 9% of apoptotic cells were observed after gemcitabine treatment, whereas DZNeP exposure was associated with a higher percentage (6%–15%) of apoptotic cells; drug combination significantly increased the apoptotic index with respect to both control cells and gemcitabine-treated cells.

To investigate the effects of DZNeP, gemcitabine, and their combination on migratory behavior, a scratch assay was conducted in PANC-1, MIA-PaCa-2, and LPC006 cells. After exposure of PANC-1 cells to gemcitabine at IC₅₀, 5 μmol/L of DZNeP and their combination, a significant reduction of migration was observed after 48 hours (Fig. 3A; Supplementary Fig. S1). In particular, the percentages of cellular migration in PANC-1 were approximately 70%, 60%, 53%, and 38%, in untreated, gemcitabine-, DZNeP-, and their combination-treated cells, respectively.

Inhibition of migration of MIA-PaCa-2 and LPC006 with DZNeP or DZNeP/gemcitabine combination was much more effective than in PANC-1 cells (Fig. 3B and C; Supplementary Fig. S1). DZNeP significantly reduced cells migration with respect to controls after 8 hours, with inhibition of about 20% in the reduction of scratch area in LPC006 cells. In addition, the combination was also significantly more effective than DZNeP alone after 48 hours in both MIA-PaCa-2 and LPC006 cells.

Because previous studies suggested that EZH2 repressed E-cadherin expression (27), we investigated whether DZNeP could affect the levels of this target at both mRNA and protein level. DZNeP and its combination with gemcitabine significantly enhanced E-cadherin mRNA expression (Fig. 3D). Similarly, immunocytochemical analysis in LPc006 cells revealed a significant increase of E-cadherin protein staining after exposure to both DZNeP and DZNeP/gemcitabine combination (Fig. 3E; Supplementary Fig. S2).

Earlier studies illustrated that results of sensitivity to anticancer drugs, including gemcitabine, in 2-dimensional monolayer cell culture models were different from 3-dimensional (3D) culture models (28). Moreover, the use of serum-free CSC medium should select a population harboring CSCs characteristics, which should be selectively targeted by inhibitors of DZNeP. Thus, to determine whether DZNeP would enhance the efficacy of gemcitabine in 3D systems, we tested these drugs in spheroids of PANC-1, MIA-PaCa-2, and LPC006 cells. After 10 days of culture, we transferred in each well of 96-well plates about 10 spheroids that were approximately 500 μm in diameter (Fig. 4A and B). These growing spheroids were exposed to DZNeP, gemcitabine, and their combination for 72 hours. The growth of these spheroids was only slightly inhibited by gemcitabine (Fig. 4C), whereas DZNeP significantly reduced their volume. However, the DZNeP/gemcitabine combination remarkably increased the disintegration of these spheroids, which were significantly reduced in size compared with the spheroids exposed to gemcitabine-alone in all our 3 PDAC models. Spheroids from each treatment group were collected and used for PCR evaluation of CD133, which was significantly higher than in adherent cells (data not shown). This CSC marker was significantly increased after gemcitabine exposure, whereas DZNeP reduced its mRNA levels. Moreover, the DZNeP/gemcitabine combination significantly reduced CD133 expression in both PANC-1 and LPC006 cells, as shown in Fig. 4D.

Treatment of PANC-1 cells with gemcitabine, DZNeP, and their combination decreased the cellular concentrations of all deoxynucleotides from −25% (dAΣP after DZNeP) to −65% (dAΣP after gemcitabine). Of note, the levels of dTΣP were depleted by all treatments to not detectable levels (Fig. 5A). These changes in the pools of deoxynucleotides were associated with a significant increase in the mRNA expression of hENT1, after exposure to DZNeP and its combination with gemcitabine (Fig. 5B). An increased expression of both hENT1 and hCNT1 transporters was also observed at the protein level (Fig. 5C).

The present study shows that the combination of the EZH2 inhibitor DZNeP and the cytotoxic compound gemcitabine was strongly synergistic in a panel of PDAC cells characterized by different molecular properties.

The highly lethal nature of PDAC makes multiple areas of research a priority, including assessment of novel targets that might prevent or suppress the proliferative, invasive, and chemoresistant behavior of PDAC cells.

EZH2 has a master regulatory role in the fate of native embryogenic cells (29), as well as in cancer development via methylation-mediated repression of transcription of several genes (9, 30). Overexpression of EZH2 is a marker of advanced and metastatic disease in many solid tumors, including PDACs (13, 31, 32), and EZH2 nuclear accumulation is strongly associated with poor differentiation and prognosis of PDACs (13, 14).

Previous studies on PANC-1 and SW1990 showed that suppression of EZH2 expression by RNA interference with lentiviral shEZH2 markedly inhibited cellular proliferation in vitro and drastically diminished both tumorigenicity and liver metastasis in vivo (33). Furthermore, the transfection of shEZH2 construct cells sensitized MIA-PaCa-2 and Pac04.02 to doxorubicin and gemcitabine (14), suggesting that combination of EZH2 inhibitors with gemcitabine might overcome the intrinsic chemoresistance of PDACs.

To our knowledge, this is the first study evaluating the pharmacologic interaction of the small-molecule EZH2 inhibitor DZNeP with gemcitabine in PDAC cells (Fig. 6).

The expression of EZH2 was detectable in all our PDAC cells, including 7 primary tumor cell cultures, in their first passages, where the levels of EZH2 mRNA were comparable with their originator tumors, suggesting that these cells represent optimal preclinical models for our pharmacologic studies. Conversely, EZH2 levels were significantly lower in both fibroblasts and in the normal pancreatic ductal cells HPNE, in agreement with earlier data on normal pancreatic tissue and specimens from patients affected by pancreatitis (13).

Because DZNeP inhibits S-adenosyl-homocysteine hydrolase, a component of the methionine cycle, resulting in accumulation of the inhibitory S-adenosyl-homocysteine, its effects on histone methylation are global rather than EZH2-specific (31, 34), and we evaluated both the modulation of H3K27me3 expression and the perturbation of intracellular adenosine.

In our PDAC cells, using concentration and exposure time (5 μmol/L, 72 hours) similar to those used in other tumor cells (35), we observed a significant reduction of both EZH2 and H3K27me3 expression, as well as a dramatic decrease of intracellular adenosine content. Although DZNeP alone did not significantly affect proliferation of PANC-1 and MIA-PaCa-2 cells, these data suggested that DZNeP effectively reached its targets.

A recent phase III trial showed that the oxaliplatin/irinotecan/fluorouracil/leucovorin (FOLFIRINOX) regimen is an option for the treatment of metastatic patients with good performance status but was associated with increased toxicity (36). Thus, gemcitabine is still the standard first-line agent (37), and several studies are evaluating novel strategies to improve its activity against PDACs.

In the present study, we showed that DZNeP/gemcitabine combination was synergistic in 2 representative PDAC cell lines, PANC-1 and MIA-PaCa-2, and in the primary cell culture LPC006.

This synergistic interaction against cell proliferation was associated with a significant increase in apoptosis induction. This effect may be related to cell-cycle modulation, which was also important for the efficacy of the combination of the histone deacetylase inhibitor trichostatin A with gemcitabine (38). Cellular damage induced by chemotherapeutic drugs such as gemcitabine can convert some targets of EZH2 into critical survival factors. In this context, the blockade of EZH2 after the exposure to cytotoxic drugs could prevent cell damage repair, leading to apoptosis. In particular, previous studies in breast cancer cells resistant or sensitive to DZNeP led to the identification of a set of PRC2 target genes including TGFBI, IGFBP3, and PPPIR15A, which are involved in apoptosis (16), whereas TGFB signaling pathway is frequently deregulated in PDACs (39).

However, our findings show that the synergistic interaction of DZNeP with gemcitabine is also mediated by other mechanisms, which reduced PDAC aggressiveness and enhanced sensitivity to gemcitabine.

Because one of the major hallmarks and problems in the therapy of PDACs is its early local and systemic dissemination, we evaluated whether DZNeP might affect cell migration. In agreement with previous studies, showing that inhibition of EZH2 by DZNeP attenuated glioblastoma and mesothelioma cell migration/invasion (36, 40), we observed that inhibition of EZH2 by DZNeP and its combination with gemcitabine significantly reduced cell migration, as detected with wound-healing assay.

Several classes of proteins are participating to invasive PDAC phenotype, including cell–cell adhesion molecules like members of immunoglobulin and calcium-dependent cadherin families and integrins. One widely observed alteration in cell-to-environment interaction in PDAC involves E-cadherin, which couples adjacent cells by E-cadherin bridges, and a recent study showed that recruitment of histone deacetylases 1/2 by the transcriptional repressor ZEB1 downregulates E-cadherin expression in PDACs (41).

Keeping with previous evidence on inverse relationship between EZH2 and E-cadherin expression (13), our data also show that DZNeP-induced EZH2 inhibition resulted in an increase in both mRNA and protein expression of E-cadherin.

Recently, PDACs also emerged as a CSC-driven disease (42). This might at least partially explain its chemoresistant nature (8) and compounds targeting critical developmental genes keeping self-renewal in CSCs, including Sonic hedgehog, BMI-1, and EZH2, seem promising anticancer agents. For example the curcumin analog difluorinated-curcumin inhibited formation of pancreatospheres as well as PDAC growth by switching on several suppressor microRNAs and attenuating EZH2 expression (43).

DZNeP significantly reduced the volume of PDAC spheroids growing in serum-free stem cell medium. Gemcitabine only slightly reduced the volume of these spheroids, possibly by affecting some remaining bulk tumor cells, but it increased the expression of the CSC marker CD133, as observed previously (44), suggesting that exposure to gemcitabine might select a population of more aggressive cells. Conversely, DZNeP was able to effectively deplete the most aggressive subpopulation of PDAC cells, as suggested by the significant reduction of both spheroids and CD133 expression.

In addition to the effects of DZNeP on migration and spheroids, the present study also showed that it interfered with pivotal determinants for the activity of gemcitabine. In particular, different thymidylate synthase inhibitors upregulated hENT1 and increased gemcitabine sensitivity by depleting intracellular nucleotide pools (45, 46). Therefore, we analyzed the cellular deoxynucleotide pools and modulation of the expression of key nucleoside transporters (47). Gemcitabine, DZNeP, and their combination significantly depleted all the endogenous deoxynucleotides. The results achieved after exposure to gemcitabine might be explained by gemcitabine-induced inhibition of ribonucleotide reductase, as reported previously (48). DZNeP is not phosphorylated and does not get incorporated into DNA (49), but it markedly reduced endogenous deoxynucleotides. This might, at least in part, explain the significant upregulation of both hENT1 and hCNT1, potentially facilitating gemcitabine cytotoxicity.

In conclusion, inhibitors of EZH2, such as DZNeP, seem very promising anticancer agents by attacking key mechanisms involved in the proliferation, cell-cycle control, apoptosis, and migration of PDAC cells. Moreover, the favorable modulation of hENT1/hCNT1 transporter makes DZNeP an optimal candidate for combination with gemcitabine. The synergistic results observed in the present study may have critical implications for the rational development of innovative regimens including DZNeP and gemcitabine to improve the efficiency of the actual treatment of PDACs.

No potential conflicts of interest were disclosed.

Conception and design: A. Avan, F. Crea, E. Paolicchi, G.J. Peters, E. Giovannetti

Development of methodology: A. Avan, E. Paolicchi, N. Funel, E. Galvani, R.J. Honeywell, G.J. Peters, E. Giovannetti

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Avan, N. Funel, E. Galvani, G.J. Peters, E. Giovannetti

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Avan, F. Crea, E. Paolicchi, N. Funel, R.J. Honeywell, E. Giovannetti

Writing, review, and/or revision of the manuscript: A. Avan, F. Crea, E. Paolicchi, E. Galvani, V.E. Marquez, R. Danesi, G.J. Peters, E. Giovannetti

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F. Crea, E. Paolicchi, E. Giovannetti

Study supervision: F. Crea, G.J. Peters, E. Giovannetti

Provided the synthetic material (DZNep): V.E. Marquez

The authors thank M. Smits and Dr. T. Wurdinger (Department of Neuro-Oncology, VU University Medical Center, Amsterdam, Netherlands) for their advice on EZH2; Dr. C. Fedrigo (Department of Radiobiology, VU University Medical Center) for his help on experiments with spheroids; and A. Griffioen (Angiogenesis-group, VU University Medical Center) for the migration station to conduct wound-healing assays.

This work was supported by grants from the Netherlands-Organization for Scientific Research (Veni grant #91611046 to E. Giovannetti), AIRC-Marie Curie (International Fellowship to E. Giovannetti), Iran's National-Elites-Foundation (A. Avan), and CCA-VICI foundation (grant #2012-5-07 to A. Avan, G.J. Peters, and E. Giovannetti); and by the Intramural Research Program of the NIH, NCI, Center for Cancer Research (to V.E. Marquez).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.