Pomalidomide enhanced gemcitabine and nab-paclitaxel on pancreatic cancer both in vitro and in vivo

Background Chemotherapy with gemcitabine and nab-paclitaxel (gemcitabine/nab-paclitaxel) is recommended for unresectable pancreatic cancer. However, the therapeutic efficacy is attenuated by the antitumor agent-induced activation of nuclear factor-κB (NF-κB). Thalidomide inhibits NF-κB activation, therefore, we hypothesized that pomalidomide, a third-generation IMiD, would also inhibit NF-κB activation and enhance the antitumor effects of gemcitabine/nab-paclitaxel. Methods In vitro, we assessed NF-κB activity and apoptosis in response to pomalidomide alone, gemcitabine/nab-paclitaxel, or combination of pomalidomide and gemcitabine/nab-paclitaxel in human pancreatic cancer cell lines (PANC-1 and MIA PaCa-2). In vivo, we established orthotopic model and the animals were treated with oral pomalidomide and injection of gemcitabine/nab-paclitaxel. Results In pomalidomide and gemcitabine/nab-paclitaxel group, gemcitabine/nab-paclitaxel-induced NF-κB activation was inhibited and apoptosis was enhanced in comparison with those in the other groups both in vitro and in vivo. Especially, this study revealed for the first time that pomalidomide enhances p53 on pancreatic cancer cells. The tumor growth in the pomalidomide and gemcitabine/nab-paclitaxel group was significantly slower than that in the gemcitabine/nab-paclitaxel group. Moreover, pomalidomide induced G0/G1 cell cycle arrest and suppressed angiogenesis. Conclusions Pomalidomide enhanced the antitumor effect of gemcitabine/nab-paclitaxel by inhibition of NF-κB activation. This combination regimen would be a novel strategy for treating pancreatic cancer.


INTRODUCTION
Pancreatic cancer is one of the most lethal and aggressive of all malignancies and the fourth leading cause of death in the developed countries [1]. Currently, chemotherapy with gemcitabine and nab-paclitaxel (GN) is recommended for locally advanced unresectable and metastatic pancreatic cancer. However, the median overall survival remains unsatisfactory with 8.5 months [2,3].
Immunomodulatory drugs (IMiDs) have been considered as effective agents for treatment of colorectal [4,5], lung [6], and prostate cancers [7]. The detailed mechanism by which IMiDs induce apoptosis in solid tumors remains unclear. Pomalidomide (PMD) is a www.oncotarget.com third-generation IMiD derived from thalidomide and has antiproliferative effects on hematological malignancies [8]. Similar to thalidomide, PMD has direct antiproliferative, pro-apoptotic, and antiangiogenetic effects on multiple myeloma cells, as well as modulatory effects on the immune system [8][9][10].
IMiDs modulate serum levels of inflammatory cytokines and growth factors, such as IL-6, IL-12, IL-8 and granulocyte-macrophage colony-stimulation factor (GM-CSF), which contribute to cancer progression and antitumor immunity [11]. IMiDs enhance the cancer suppressing effect of natural killer (NK) and CD8+ cells by increasing IL-2 levels [12] and by suppression of T regulatory (T-reg) cell [13]. Moreover, IMiDs, including PMD, have been reported to have antitumor effect through induction of apoptosis [8], cell cycle arrest [14] and suppression of angiogenesis [15] in hematological malignancies. PMD has an immunosuppressive effect 15,000 to 50,000 times more potent than thalidomide and 10-fold more potent than lenalidomide on inhibition of tumor necrosis factor (TNF-α) [8,16]. Therefore, PMD is expected to have more potent antitumor activity as compared to thalidomide or lenalidomide in digestive cancer. In the past clinical trial, the combination of gemcitabine with lenalidomide showed no significant difference as compared to gemcitabine alone [17] on pancreatic cancer. However, considering the strength of the effects of pomalidomide, we believe that GNP therapy is expected to have a stronger effect than the combination of gemcitabine with lenalidomide.
The purpose of this study is to investigate the antitumor effect and to clarify the mechanisms of PMD in pancreatic cancer.

PMD suppressed the activation of nuclear NF-κB in vitro
To evaluate the activation of NF-κB in pancreatic cancer cell lines, the concentration of p65 in the nuclear extracts were measured using ELISA, and IκBα levels were evaluated using Western blot analysis. The concentrations of NF-κB p65 in the nuclear extracts of PANC-1 and MIA PaCa-2 in the GN group (GN group means cells treated with GN) were significantly greater than those in the C group (p < 0.01) (C group means cells treated with vehicle), whereas those in the GNP group (GNP group means cells treated with GNP) were significantly lower than those in the GN group (p < 0.01). In MIA PaCa-2 cells, the concentrations of NF-κB p65 in the nuclear extracts in the P group were significantly lower than that in the C group (p < 0.01, Figure 1A) (P group means cells treated with PMD alone). Similar to nuclear NF-κB concentrations, IκBα levels of PANC-1 and MIA PaCa-2 in the P group were greater than those in the C group, whereas phosphorylated IκBα levels of PANC-1 and MIA PaCa-2 in the GNP group were lower than those in the GN group ( Figure 1B).

In vivo analysis
The tumor growth in the GNP group was significantly slower than that in the GN group ( Figure  3A). Both tumor weights and volume of the excised tumor in the GNP group were significantly lower than those in GN group (weight, 385.3 ± 88.2 mg vs. 538.6 ± 36.2 mg, p < 0.05; volume, 0.36 ± 0.02 cm 3 vs. 0.76 ± 0.04 cm 3 , p < 0.01, respectively, n = 4 for each, Figure  3B). In contrast, there was no difference in the change in body weight among the treatment groups ( Figure  3C). In the GNP group, the number of Ki-67-positive cells in the excised tumor was significantly lower than that in the GN group (40.7 ± 3.2 vs. 63.3 ± 7.7 %, p < 0.01). Additionally, the number of Ki-67-positive cells in the P group was significantly lower than that in the C group (67.1 ± 2.4% vs. 74.8 ± 3.8%, p < 0.05, Figure   The activation of NF-κB was measured using ELISA at 2 hours after each treatment. The concentrations of NF-κB p65 in the nuclear extracts of PANC-1 and MIA PaCa-2 in GN group were significantly greater than those in C group (PANC-1, p < 0.01; MIA PaCa-2, p < 0.01), whereas those in GNP group were significantly lower than those in GN group (PANC-1, p < 0.01; MIA PaCa-2, p < 0.01). (B) Similar to nuclear NF-κB concentrations, IκBα levels of PANC-1 and MIA PaCa-2 in P group were greater than those in C group, whereas phosphorylated IκBα levels of PANC-1 and MIA PaCa-2 in GNP group were lower than those in GN group at 2 hours after treatment. (MIA PaCa-2) after treatment. In the cell proliferation assay, the viability in GNP group of PANC-1 and MIA PaCa-2 was significantly lower than that in GN group (PANC-1, p < 0.01; MIA PaCa-2, p < 0.01). (B) The proportion of early and late apoptotic cells was evaluated by Annexin/FITC analysis at 72 hours (PANC-1) and 48 hours (MIA PaCa-2) after treatment. The apoptotic cells in GNP group of PANC-1 and MIA PaCa-2 were significantly greater than those in GN group (PANC-1, p < 0.01; MIA PaCa-2, p < 0.05). (C) Western blot analysis demonstrated the expression of proteins related to apoptosis. After 2 hours treatment with PMD, gemcitabine and nab-paclitaxel were administered. PANC-1 and MIA PaCa-2 cells were harvested at 72 hours and 48 hours after treatment, respectively. The levels of p53, cleaved caspase-8 and -3 as well as cleaved PARP in GNP group were greater than those in GN group. www.oncotarget.com 3D). The TUNEL assay of the excised tumor showed that the number of apoptotic cells in the GNP group was significantly greater than that in the GN group (107 ± 2.6 vs. 63.0 ± 1.0, p < 0.01, Figure 3E).
NF-κB activation in the excised tumor was assessed using ELISA, and IκBα levels were evaluated using Western blot analysis. The concentrations of NF-κB p65 in the excised tumor in the GN group were significantly greater than those in the C group (p < 0.05), whereas those in the GNP group were significantly lower than those in the GN group (p < 0.05, Figure 4A). Similar to nuclear NF-κB concentrations, IκBα levels in the P group were greater than those in the C group, whereas phosphorylated IκBα levels in the GNP group were lower than those in the GN group, as observed in the in vitro experiments ( Figure 4B). Similar to the in vitro experiments, the level of p53, cleaved caspase-8 and 3 as well as cleaved PARP in the GNP group were greater than those in the other groups ( Figure 4C).

G0/G1 cell cycle arrest
To evaluate the effect of PMD on cell cycle progression in pancreatic cancer cells, we performed cell cycle analysis using flow cytometry. The proportion of cells in the G0/G1 phase in the P group was greater than that in the C group (PANC-1, 74.0 ± 1.0% vs. 66.3 ± 1.0%, p < 0.01; MIA PaCa-2, 54.2 ± 1.4% vs. 48.6 ± 2.2%, p < 0.01, Figure 5A). Additionally, the expression levels of phosphorylated Rb, which plays an important role in cell cycle progression [28,29] was examined. The level of phosphorylated Rb (Ser 780) protein in the P group was lower than that in the C group for PANC-1 and MIA PaCa- in GNP group was slower than that in GN group (p < 0.05). (B) At five weeks after treatment, both tumor weight and volume of excised tumor in GNP group were significantly lower than those in GN group (weight, p < 0.05; volume, p < 0.01) (n = 4 each). (C) Body weight changes during the treatment. There was no difference in the change in body weight among the experimental groups. (D) In GNP group, the number of Ki-67-positive cells in the excised tumor was significantly lower than that in GN group (p < 0.01). The number of Ki-67-positive cells in P group was significantly lower than that in C group (p < 0.05) (×400). (E) The TUNEL assay of the excised tumor showed that the number of apoptotic cells in GNP group was significantly greater than that in GN group (p < 0.01) (×400). www.oncotarget.com 2 ( Figure 5B). In this experiment, it was not appropriate to correctly evaluate the influence of PMD on GN with flow cytometry because GN also influence on cell cycle. Therefore, we simply examined the effect of PMD on pancreatic cancer cells.

PMD inhibited angiogenesis in pancreatic cancer cells
To investigate detailed molecular mechanism of antitumor effect by PMD, the expression of VEGF, which plays an important role in angiogenesis and is a target molecule of NF-κB was measured. The levels of VEGF protein in the GNP group were lower than those in the GN group, both in vitro and in vivo ( Figure 6A). Immunohistochemical staining of the excised tumors showed that the number of VEGF-positive cells in the GNP group was fewer than that in the GN group ( Figure 6B). The angiogenesis assay was performed to evaluate the effect of PMD on endothelial cell tube formation in PANC-1 and MIA PaCa-2. The number of connected cells in the P group was significantly fewer than that in the C group (PANC-1, 7.1 ± 1.3 vs. 12.5 ± 1.3, p < 0.01; MIA PaCa-2, 8.4 ± 0.6 vs. 11.8 ± 1.6, p < 0.05, respectively, Figure 6C). This experiment was not suitable for comparing GN and GNP group because all the cells in both GN group and GNP group were separated with each other. Therefore, we compared C group and P group.
In the current study, GN decreased IκBα and enhanced NF-κB activation, whereas PMD suppressed phosphorylation of IκBα and inhibited the NF-κB activation. Moreover, PMD had dual apoptotic effect of enhancing p53 and cleaved caspase-8-dependent pathway on pancreatic cancer. To the best of our knowledge, we firstly demonstrated enhancement of p53, which directly activates the transcription of apoptotic genes [37,38] by PMD on pancreatic cancer.
PMD is also known to induce G0/G1 phase cell cycle arrest [14]. The pRb regulates the local chromatin structure by recruiting histone deacetylase 1 (HDAC1) to modulate the balance of histone acetylation levels. The pRb blocks cell cycle progression by repressing E2F-target gene transcription through the recruitment of transcriptional corepressors and chromatin remodeling protein factors, such as HDACs, Sin3, CtBP and SWI/SNF, at promoter regions [28]. Hypophosphorylated pRb binds to DNA-

Figure 4: (A)
The concentrations of NF-κB p65 in the excised tumor in GN group were significantly greater than those in C group (p < 0.05), whereas those in GNP group were significantly lower than those in GN group (p < 0.05). (B) Similar to nuclear NF-κB concentrations, IκBα levels in P group were greater than those in C group, whereas phosphorylated IκBα levels in GNP group were lower than those in GN group, as observed in the in vitro experiments. (C) Similar to the in vitro experiments, the level of p53, cleaved caspase-8 and 3 as well as cleaved PARP in GNP group were greater than those in the other groups. www.oncotarget.com bound E2F transcriptional regulators and suppresses their target gene expression through the recruitment of HDACs, co-repressors and chromatin remodeling enzymes. When it is in a hyperphosphorylated state, the complex is dissociated, and the cell cycle progresses [39,40]. In the current study, we firstly demonstrated that PMD inhibited the Ser 780 phosphorylation of pRb and caused G0/G1 cell cycle arrest. These result suggested that PMD contributed to G0/G1 cell cycle arrest via the pRb pathway in pancreatic cancer cells.
The importance of angiogenesis in cancer progression has been demonstrated in various cancers [41]. VEGF is downstream of NF-κB signal transduction and is known to play a key role in neovascularization [42,43]. In the current study, we have shown that the level of VEGF was decreased and neovascularization was suppressed by PMD. Regarding solid tumor, previous report has also described that IMiDs inhibit angiogenesis in gastric cancer [44].
PMD has a manageable toxicity profile in patients who relapsed after multiple types of therapy for myeloma [45]. The most common grade 3 or 4 hematologic side effect is neutropenia, while non-hematologic grade 3 or 4 adverse effects are fatigue and pneumonia [8]. Grades 3 and 4 adverse events were reported in more than 5% of patients in the phase II clinical trials for PMD/dexamethasone for multiple myeloma [8]. Our in vivo study showed no significant body weight differences or treatment-related deaths. This indicated that current regimen does not have nutrition-related disadvantages.Nevertheless, another in vivo study using PMD shown thrombocytosis in combination with gemcitabine and S-1, therefore we must keep in mind checking platelet count in clinical use. In summary, PMD enhanced the antitumor effects of GN by inhibiting NF-κB activation. PMD induced caspase-8-mediated apoptosis, enhancement of p53, cell cycle arrest, and antiangiogenesis. Therefore, PMD with GN therapy may be a novel treatment with strong antitumor effect on pancreatic cancer.

Reagents
PMD was purchased from Tokyo Chemical Industry (Tokyo, Japan) and was dissolved in 1% dimethyl sulfoxide (DMSO) and stored at −80° C until use. Gemcitabine was purchased from Eli Lilly Japan (Kobe, Japan). Nab-paclitaxel was purchased from Taiho Pharmaceutical Co. Ltd (Tokyo, Japan).

Antibodies
Monoclonal antibodies to inhibitor of kappaBα (IκBα), cleaved caspase-3 and -8, and cleaved poly-ADP-ribose polymerase (PARP) were purchased from Cell Signaling Technology (Beverly, MA). Anti-bactin antibody was purchased from Sigma-Aldrich The in vitro angiogenesis assay revealed the number of connected cells in P group was significantly fewer than that in C group (PANC-1, p < 0.01; MIA PaCa-2, p < 0.05) (×200). www.oncotarget.com Conduct of Animal Experiments of the Science Council of Japan (2006). All the guidelines above are based on the Declaration of Helsinki.

In vitro experimental protocol
PANC-1 and MIA PaCa-2 cells were treated in 4 conditions: pomalidomide (100 μM; P group), gemcitabine (1000 nM and 494 nM, respectively) and nab-paclitaxel (500 nM and 683 nM, respectively) (GN group), gemcitabine and nab-paclitaxel with pomalidomide (GNP group), or vehicle only (1% DMSO; C group) for the appropriate time in each analysis. The concentrations of gemcitabine and nab-paclitaxel were determined according to the previous studies [30,32]. The appropriate doses were determined with reference to the IC50 and different concentrations of experiments we performed. In the GNP group, the cells were treated with PMD for 2 hours before the administration of gemcitabine and nab-paclitaxel.

In vivo experimental animal model and protocol
The orthotopic pancreatic cancer mice model was used as previously described [32,46]. The mice were continuously anesthetized with 2% isoflurane (Pfizer Japan Inc., Tokyo, Japan), and a 5-mm incision was made in the left flank on the splenic silhouette. The distal pancreas was gently delivered from the peritoneal cavity with spleen thorough the incision. 5.0 × 10 6 PANC-1 cells suspended in 40 µl of phosphate buffered saline (PBS) were injected into the tail of the pancreas. After injection, the pancreas was placed in the abdomen and the abdominal wound site was closed in two layers. At 4 weeks after the injection, the animals were randomly divided into four treatment groups, which were the same as the in vitro experimental groups (n = 4). Pomalidomide (0.5 mg/kg) was administered orally five times a week, and gemcitabine (50 mg/kg) and nab-paclitaxel (0.5 mg/ kg) were injected intraperitoneally once a week. The doses of antitumor agents were determined according to the previous studies [32,47]. In the C group, vehicle of pomalidomide (PBS) was administered orally five times a week, which vehicles of gemcitabine and nab-paclitaxel (distilled water and isotonic sodium, respectively) were injected intraperitoneally once a week. The tumor volume was evaluated sequentially once a week by magnetic resonance imaging (MRI). At five weeks after treatment, the animals were sacrificed and orthotopic pancreatic tumors were excised for the assessment.

Quantitative analysis of NF-κB activity
NF-κB is typically a heterodimer consisting of the p65 and p50 proteins, and it normally resides in the cytoplasm as an inactivate form as a complex with an IκBα. NF-κB activation signals lead to IκBα phosphorylation, and NF-κB is released from IκBα, and then translocated into the nucleus. Nuclear extracts from both in vitro and in vivo experiments were prepared using a nuclear extract kit (Active Motif, Carlsbad, CA) according to the manufacturer's instructions. The nuclear extracts were assayed using an enzyme-linked immunosorbent assay (ELISA) kit (Trans-AM NF-κB p65 Activation Assay; Active Motif) to detect and quantify the NF-κB p65 activity according to the manufacturer's instructions. The levels of NF-κB p65 were measured at 2 hours after treatment.

Western blot analysis
Whole protein extracts from both in vitro and in vivo experiments were prepared according to the protocol described previously [48]. The lysate protein was extracted by 2% sodium dodecyl sulfate, electrophoresed on 4% to 20% acrylamide gradient gels in Tris-glycine buffer, and transferred onto a nitrocellulose membrane. After incubating the blots in each primary antibody (1:1000 dilution) overnight, the membranes were then incubated with secondary antibody (1:10,000 dilution, Histofine; Nichirei, Tokyo, Japan). A luminol enhancer and peroxide solution, Clarity Max (Bio-Rad Laboratories, Inc., Hercules, CA), was used to detect the protein bands. Protein bands were detected using a Chemi Doc XRS+ system and evaluated using Image Lab software (Bio-Rad).

Cell proliferation assay
The cells were seeded into 96-well plates (5 × 10 3 cells/well) and treated for 72 hours (PANC-1) or 48 hours (MIA PaCa-2). The cell proliferation was measured using a CellTiter-Blue Cell Viability Assay Kit (Promega, Madison, WI) according to the manufacturer's instructions. The stained cells were analyzed with a PerkinElmer EnSpire plate reader (PerkinElmer Inc., Waltham, MA).

Assessment of apoptosis by flow cytometry
The cells (1 × 10 6 cells) were seeded and treated with each regimen for 72 hours (PANC-1) or 48 hours (MIA PaCa-2). These cells were washed with 1× binding buffer and incubated with 10 µl of Annexin V/ FITC (Annexin V/FITC Kit; Miltenyi Biotec GmbH, Cologne, Germany) for 15 minutes in the dark at room temperature. The stained cells were analyzed with an Attune NxT Flow Cytometer (Thermo Fisher Scientific, Waltham, MA). www.oncotarget.com

Cell cycle analysis
The cells were harvested after treatment for 72 hours (PANC-1) or 48 hours (MIA PaCa-2) and were fixed in 70% ethanol for 30 minutes at −20° C. After centrifugation, the cell pellets were washed with PBS and incubated at room temperature for 1 minute, followed by incubation with RNase (1 µg/ml) for 4 minutes. Finally, these cells were incubated with propidium iodide (1 mg/ml) (Sigma-Aldrich, St Louis, MO) for 30 minutes. DNA content was determined with an Attune NxT Flow Cytometer (Thermo Fisher Scientific).

In vitro angiogenesis assay
To evaluate the antiangiogenic effect of PMD in pancreatic cancer, an in vitro angiogenesis assay kit (Trevigen Inc., Gaithersburg, MD) was used. Pancreatic cell lines were seeded into 96-well plates (5 × 10 3 cells/well) on top of an aliquot of 50 µl of basement membrane. These cells were incubated with PMD for 24 hours and fixed with 3.7% methanol. These cells were continuously stained with cell staining solution according to the manufacturer's instructions. This assay evaluates the ability of the cells to connect. The number of intersections was counted in three random fields per well for three wells. This experiment was conducted with reference to [49].

Immunohistochemical staining
Paraffin sections of tumor tissues were stained immunohistochemically using anti-Ki-67 and VEGF antibodies as the primary antibodies and the DAKO Envision Kit/HRP as a secondary antibody (DAKO, Carpenteria, CA). A TdT-mediated dUTP nick end labeling (TUNEL) assay was performed using an In Situ Cell Death Detection Kit with fluorescein (Roche Diagnostics, Basel, Switzerland) to evaluate the induction of apoptosis. These measurements were performed according to the manufacturer's instructions. The number of Ki-67-positive cells and TUNEL-positive cells were counted in three random fields (×400 and ×200, respectively) in three tumors.

Statistical analysis
Data are expressed as the mean ± SD. Non-paired t-test (2-tailed) and repeated measurements analysis (ANOVA) of variance were used for statistical analysis with SPSS 23.0 (IBM Japan, Tokyo, Japan). All p values were considered statistically significant when the associated probability was < 0.05.