Targeting notch pathway enhances rapamycin antitumor activity in pancreas cancers through PTEN phosphorylation

The prevalence in expression of a potential oncogene helps determine the significance of its role in cancer. To better understand the role of Notch pathway in pancreas cancer, we developed a pancreas tissue microarray with associated clinical data from 86 patients (Table 1). We also examined the expression of Notch1-4 and their ligands, Jagged1 and DLL4. Notch3 was most prevalent with greater expression in 84% of resected cancers, followed by Notch4 at 31% (Table 2 Figure 1). Interestingly, none of the tumor cells expressed Notch1, and only one of the 86 tumors surveyed expressed Notch2. Notch1 and DLL4 were expressed predominantly in endothelial cells, suggesting that, while not significantly expressed in tumor cells, they are important in tumor angiogenesis. We also tested the dataset for correlation between different Notch family members and clinical characteristics, such as overall survival, stage and tumor grade. No association between Notch receptors and clinical characteristics was observed. However, we noted that Notch3 expression correlated with Jagged1, but not for Delta-like 4, suggesting that Jagged1 is the ligand for Notch3 (Figure 1C, Table 3) [11]. Of note, eighty-five percent of the tumors surveyed with IHC exhibited high expression of EGFR (Figure 1B). Notch3 also correlates with EGFR expression (Figure 1D, Table 4), consistent with our previous finding in lung cancer that Notch3 and EGFR pathways cooperate in maintaining the oncogenic phenotype [12]. Notch receptors are activated by proteolytic cleavages after ligand binding, resulting in the release of the cytoplasmic domain (NICD). We were able to demonstrate that several human pancreas cancer cell lines expressed the activated forms or NICD of Notch receptors (Figure 2A). In addition, pancreas cancer cell lines developed from overexpressing K-rasG12D and TGF-β knockout mice showed Notch1 ICD and Notch3 ICD expression (Figure 2B), further supporting the role of Notch pathway in pancreas cancers [13]. Similar to our previous observation, Jagged1 is also highly expressed in nearly all of cell lines tested [14]. We found no difference in Notch expression between cell lines with K-ras mutation alone (K162, K512, K518) and those with both K-rasG12D and TGF-β knockout (K375, K389, K399). When K162 and K399 were treated with MRK003, γ-secretase inhibitor, dose-dependent down regulation of activated Notch3 was observed (Figure 2C). Interestingly, while we observed suppression of the activated form of Notch, we observed a rise in HES1 and HEY1 transcripts, suggesting that Notch modulates cancer phenotype in pancreas through non-canonical pathways (Figure 2D, E).

To determine whether inhibiting Notch activation reduces tumor phenotype, we utilized both dominant-negative Notch3 receptor and a γ-secretase inhibitor (GSI). When BxPc3 was transfected with dominant-negative Notch3 or treated with 25 μM of MRK003, colonies were significantly reduced in number, as compared to vector controls (VC) or DMSO control (C) (Figure 3A). A significant body of literature has supported a role for Notch signaling in apoptosis. Similar to our previous observation in lung cancer, inhibiting Notch in serum-free condition resulted in enhanced cancer cell death measured with PI staining (Figure 3B). The Bcl-2 family plays an important role in apoptosis through the activation of the mitochrondria-dependent caspase pathway. Using Notch3 siRNA, we showed that Notch regulates Bcl-xL expression and Bcl-2 (Figure 3C). When MRK003 was used, a similar effect on Bcl-xL could be found, accompanied by an increase in cleaved PARP, a marker of caspases activation (Figure 3D). To determine whether γ-secretase inhibitors possess activity in vivo, we inoculated xenografts with K162 and K399 cell lines developed from a mouse model of pancreas cancer. The γ-secretase inhibitors DAPT and MRK003 suppressed tumor growth by 25% to 50%, suggesting that the Notch pathway plays a role in the survival of cancer cells in both in vitro and in vivo models (Figure 3E).

The Notch pathway is known to crosstalk with other oncogenic pathways such as the EGFR and the Akt pathway [12, 15]. Interestingly, unlike observations in lung cancer, inhibition of the Notch pathway in pancreas cancer had no appreciable effect on ERK activation (Figure 4A). On the other hand, Akt phosphorylation was inhibited by MRK003 in pancreas cancer cell line K399. PTEN (phosphatase and tensin homolog) is a well-known negative regulator of Akt. In hypoxia, Notch1 has been shown to suppress PTEN transcription, leading to Akt activation [15]. However, while Notch is known to regulate Akt through the transcriptional regulation of PTEN, we did not detect a difference in total PTEN levels. Rather the phosphorylation of PTEN at Ser380 was altered, when GSI was used (Figure 4A). While not much is known about the phosphorylation of PTEN, recent evidence suggests that it regulates protein stability [16]. While some findings indicate that phosphorylation of PTEN improves stability but reduces PTEN function, others have shown that the loss of phospho-PTEN in migrating cells leads to the activation of Akt [17]. Cdc42, a member of the Rho GTPase family, is important in Akt-mediated cell survival and motility, and its activation is inhibited by PTEN [18, 19]. We noted a decrease in Cdc42 when treated with GSI, suggesting that Notch regulates Akt-dependent cell survival through PTEN and Cdc42.

How PTEN is regulated through phosphorylation is intensely investigated. In a recent model of chemotaxis proposed by Li et al., Rock1, a member of the Rho-associated, coiled-coil containing protein kinases, is activated by Rho-GEF (guanine nucleotide-exchange factor) and RhoA, another Rho GTPase family member. Activated Rock1 then binds and phosphorylates PTEN [17, 20]. Rho proteins and Rock proteins are important regulators of cell migration, proliferation and apoptosis [21]. To examine the role of the Rho GTPase pathway in Notch-induced PTEN phosphorylation in pancreas cancer, we examined the effect of GSI on Rock1 and RhoA. Interestingly, we noted an increase in the expression of RhoA with increasing dose of GSI, whereas the expression of Rock1 remained essentially unchanged (Figure 4B). The effect of Notch signaling on RhoA appears to be transcriptionally mediated (4C). To determine whether Notch modulation of PTEN phosphorylation is dependent on RhoA/Rock1, we examined the effect of GSI in the presence of Rock1 inhibitor Y27632 (4D). Whether the observations in the chemotaxis model can be translated into a cancer model requires further validation. The loss of PTEN phosphorylation by GSI in the presence of Y27632 suggests, however, that the Notch effect on PTEN depends on the RhoA/Rock1 pathway.

The observed redundancy in oncogenic pathways may require that multiple pathways are inhibited in order to enhance tumor cytotoxicity. The PI3K/Akt/mTOR pathway is activated in the majority of pancreas cancers. Because of the crosstalk between Notch and Akt, we examined whether the combination of the mTOR inhibitor Rapamycin and MRK003 will result in improved tumor cytotoxicity. While some studies suggest that Rapamycin induces Akt activation, we noted that in K399 rapamycin inhibits Akt phosphorylation, and that this inhibition was enhanced, when Rapamycin was combined with MRK003 [22] (Figure 5A). Again, we observed a change in phospho-PTEN, but not total PTEN, when Notch pathway is inhibited. Furthermore, the level of phospho-PTEN was increased when MRK003 was combined with rapamycin. Foxo3a is a member of the forkhead family which acts as tumor suppressor by promoting cell cycle arrest and apoptosis. It is inactivated by Akt. The combination of Rapamycin and MRK003 led to a slight increase in the tumor suppressor Foxo3a and pro-apoptotic Bim, a member of the BH-3 only Bcl-2 family. Moreover, we noted an increased expression of RhoA, when cancer cells were treated with MRK003, and the change was enhanced when Rapamycin was added (Figure 5B). No change in Rock1 level was detected. Taken together, these observations support the hypothesis that Notch and mTOR cooperate in regulating Akt through PTEN phosphorylation and RhoA.

While results from preclinical studies using mTOR inhibitors in pancreas cancers have been promising, their low efficacy in early clinical studies indicate that these agents possess minimal clinical activity when administered as single agents [23]. Redundancy in the biological system and results from clinical trials suggest that targeting multiple targets will result in augmented tumor suppression. Because we observed Akt suppression when GSI was added to Rapamycin, we tested whether inhibiting the Notch pathway will enhance tumor suppression with mTOR inhibitor in vitro. In both human and murine pancreas cell lines, K399 and Panc-1, respectively, the combination of MRK003 and rapamycin inhibited proliferation to a greater degree than Rapamycin or MRK003 alone (Figure 5C, D). These findings suggest that Notch can enhance Rapamycin in inhibiting pancreas cancer growth through the modulation of Akt.

Human pancreas cancer cell lines Panc-1, HRAF-II and BxPC3 were obtained from American Type Culture Collection (ATCC). Murine pancreas cancer cell lines K399, K389, K375, K162, K152, and K518 were developed ex vivo from tumors of mice overexpressing K-rasG12D and TGF-β knockout, and were obtained from Dr. H. Moses [13]. The formulation and the in vivo dosing schedule of γ-secretase inhibitor MRK003 were provided by Merck Co., Inc, and were described previously [28]. The mTOR inhibitor rapamycin and the Rock1 inhibitor Y27632 were obtained from Sigma-Aldrich and CalBiochem, respectively. The γ-secretase inhibitor DAPT (N-[N-(3,5-Difluorophenacetyl)-L-alanyl]-S-phenylglycine t-butyl ester), was also obtained from Sigma-Aldrich. The dominant-negative Notch3 (DN) and VC (control) constructs were transfected into BxPC3 and selected with G418, as previously described [12]. Notch3 siRNA3 sequences were also described previously [14].

De-identified tumor and adjacent normal tissues were obtained under an IRB-approved protocol at Vanderbilt University Medical Center. Before constructing a TMA block, serial 5-μm sections were cut from each donor block. One of these sections was stained with H&E for marking morphologically representative areas of the tumor. Using a Beecher Instruments Tissue Arrayer (Silver Springs, MD), tissue cylinders with a diameter of 0.6 mm were punched from the four targeted areas in each donor block and deposited into a 9 × 14 (~126 cores) TMA block, which contained 76 cores of adenoma tissue and 50 cores of adjacent, non-malignant tissue as controls. The TMA blocks were warmed to 36°C for 30 minutes, and multiple serial 5 μm sections were cut and placed on charged slides.

The Notch3 antibody 1E4 (a gift from Dr. Anne Joutel) was used for immunohistochemistry, and the method was described previously [29]. Jagged1 (C-20) and Notch4 (H-225) were purchased from Santa Cruz, whereas Notch1 (3608), DLL4 (HPA023392) and Notch2 (C651.6D6HN) antibodies were obtained from Cell Signaling Technology, Sigma-Aldrich, and the Developmental Studies Hybridoma Bank (DSHB), respectively. Human EGFR antibody (31G7) was purchased from Zymed. The IHC staining was scored on a composite scale of 0 to 3 by two independent observers, including one pathologist. In case of disagreement, the decision was deferred to the pathologist. The tumors that scored 2 or better were considered positive. For immunoblotting, Notch1 (3447), Notch3 (2889), phospho-Akt, total Akt, PTEN, pPTEN, RhoA, Rock1, cdc42, Bcl-xL, Bcl-2 and PARP were obtained from Cell Signaling Technology. For specific use in murine cell lines, Jagged1 (H114), Notch1 (C-20), and Notch3 (M-134) were obtained from Santa Cruz, and Notch2 (C651.6D6HN) and Notch4 were purchased from DSHB and Orbigen, respectively.

Total RNA was isolated from K399 cells using Sure Prep RNA Purification Kit (Fisher Scientific). cDNA synthesis was carried out using iScript cDNA Synthesis Kit, according to manufacturer's recommendation (Bio Rad). Primers for murine GAPDH were AATGGGGTGAGGCCGGTG (sense) and CAGAAGGGGCGGAGATGATG (antisense). Murine RhoA primers were CCATGTACCCAAAAGCGCC (sense) and CAAATGTGCCCATCGTCCTG (antisense). Experiments were performed at annealing temperature of 55°C for 39 cycles.

Cells were plated into 96-well microtitre plates at 10% FCS and at 50% confluency in 200 μl DMEM. After 48 hours of treatment with inhibitors, 50 μl of MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) stock solution (2 mg/ml) was added to each well, and the plates were incubated for 4 hours. MTT formazan crystals were then resolubilized by adding 150 μl 100% dimethylsulfoxide (DMSO) to each well. Plates were agitated on a plate shaker for 5 min, and the absorbance at 540 nm was determined using a scanning multi-well spectrophotometer (BioRad). For soft agar assays, transfected cells were plated at a density of 5000 cells/plate using 35 mm Petri dishes and suspended in 0.4% agar containing 10% FCS RPMI and 50 μg/ml of G418 selective antibiotic over 0.8% base agar. The plates were incubated at 37°C and 5% CO2 in a humidified chamber for 14 days. Cell death was determined as follows: Cells were stably transfected with Notch3 DN or treated with MRK003 for 24 hours and were maintained in 10% FCS-RPMI or serum-free medium. Then, they were stained with propidium iodide (Calbiochem, La Jolla, CA). The percentage of dead cells was determined with a Beckman Coulter FACS Calibur Flow Cytometer.

Animal experiments were performed according to the protocol approved by Vanderbilt University IACUC. Athymic 4-to 6-week-old female nude mice (nu+/nu+) were used for the tumor xenograft models. Panc1 or K399 (1 × 106 cells in the volume of 200 ml of PBS) was inoculated subcutaneously (s.c.) into the right posterior legs of nude mice. Treatment was initiated when tumors were palpable. MRK-003 (150 mg/kg) was administered orally for three consecutive days per week for 2 weeks. MRK-003 was diluted in 0.5% methylcellulose. The tumors were measured every 2 days with a caliper. Tumor Volume (TV) was calculated with the formula: TV = (Length) × (Width)2/2. Percentage tumor volume (% TV) on day X was calculated as: %TV = (tumor volume on day ×/tumor volume on day 1) × 100.

The size of implanted tumors at precise time points after treatment was compared with that of control groups. Unless specifically stated, statistical inference in all comparative experiments both in vivo and in vitro was obtained using unpaired, two-sided Student's t-tests. For TMA, protein expression was correlated using Pearson's correlation coefficients. For all determinations, differences were considered significant at P < 0.05.