N-cadherin haploinsufficiency increases survival in a mouse model of pancreatic cancer

Abstract

Pancreatic ductal adenocarcinoma (PDA) is often detected at a late stage, hence the identification of new therapies that have potential to block tumor progression is critical for this lethal disease. N-cadherin upregulation has been observed in many cancers including PDA, however a causal role for this cell adhesion receptor in disease progression has yet to be defined. The concomitant expression of oncogenic Kras^G12D and mutant p53 (Trp53^R172H) in the murine pancreas results in metastatic PDA that recapitulates the cognate features of human pancreatic cancer providing an excellent animal model to identify genes required for tumor progression. Here we determine the consequences of genetically manipulating N-cadherin expression in a mouse model of PDA. Remarkably, mice with reduced N-cadherin expression (i.e. N-cad −/+) survived 25% longer (177 vs. 142 days, p <0.05) than animals expressing two wild-type N-cadherin (Cdh2) alleles. The survival benefit is likely due to a cumulative effect of N-cadherin’s role in different aspects of tumorigenesis including tumor cell survival, growth, migration and invasion. Interestingly, reduced Hedgehog signaling may contribute to the better prognosis for the N-cad −/+ mice. Moreover, the matrix metalloproteinase MMP-7, associated with poor prognosis in PDA, was reduced in N-cad −/+ tumors. Finally, N-cad −/+ tumor cells exhibited decreased FGF-stimulated ERK1/2 activation consistent with N-cadherin’s ability to promote FGFR signaling. These data support a critical role for N-cadherin in PDA and its potential prognostic value. Additionally, this study provides in vivo genetic evidence that the cell surface protein N-cadherin represents a promising therapeutic target for the treatment of pancreatic cancer.

Introduction

Pancreatic ductal adenocarcinoma (PDA) is the fourth leading cause of cancer death with a median survival of <6 months and a dismal 5-year survival rate of <5%. The mortality rate is so high in part because pancreatic cancer usually does not produce symptoms until after it has metastasized, thus rendering the malignancy inoperable (1). Therefore, treatment strategies that specifically target and prevent metastases have the potential to significantly improve the prognosis of this devastating disease.

Cadherin switching (i.e. E-cadherin to N-cadherin) associated with epithelial-mesenchymal transition (EMT) is implicated in the transition from benign tumors to invasive, malignant cancer and the subsequent metastatic dissemination of tumor cells (2). Ectopic expression of N-cadherin increases tumor cell motility, implicating cadherin switching in the regulation of cell behavior (3, 4). Activating mutations in the human K-ras protooncogene are found in over 90% of invasive PDA and are sufficient to initiate the disease in mice (5). In >50% of PDA cases, missense mutations are found in the p53 tumor suppressor gene leading to chromosomal instability and malignant progression (1). Interestingly, oncogenic K-ras is sufficient to upregulate N-cadherin expression in pancreatic ductal cells (6–8) suggesting that K-ras may play an active role in cadherin switching in pancreatic cancer. In contrast, mutant p53 was shown to decrease N-cadherin expression in the presence of mutant K-ras (8) suggesting a complex regulation of this cadherin in premalignant verses malignant disease. Importantly, knockdown of N-cadherin in BxPC-3 pancreatic tumor cells led to decreased tumor size and metastases in an orthotopic animal model (9). Taken together these studies suggest that interfering with N-cadherin function may prove beneficial in pancreatic cancer.

In this study, we investigated the in vivo role of N-cadherin in the development and progression of PDA. We demonstrate for the first time that interfering with N-cadherin expression is sufficient to prolong survival in a genetically defined murine model of pancreatic cancer. Moreover, these studies indicate that N-cadherin influences various signaling pathways previously shown to be critical for pancreatic cancer.

Results and Discussion

To determine the consequences of interfering with N-cadherin function in pancreatic cancer, we genetically manipulated N-cadherin expression in a mouse model of PDA. In this model, Pdx1/Cre activates oncogenic K-ras and dominant negative p53 in the developing pancreas leading to metastatic PDA with a median survival of 5 months (10). The LSL-K-ras^G12D; LSL-Trp53^R172H; Pdx1/Cre (KPC) mice were bred with two independent N-cadherin mutant strains containing either a floxed (Ncad^fl) allele or a germline null (Ncad^lacZ) allele (11, 12). At the embryonic and adult stage, Pdx1/Cre; Ncad^fl/fl pancreata appeared normal, indicating N-cadherin was not required for pancreas development (13). To confirm N-cadherin was upregulated in murine PDA, we performed immunohistochemistry on KPC Ncad^+/+ tumors. Heterogeneous N-cadherin expression was observed primarily in less differentiated areas of the tumor whereas E-cadherin was associated with ductal structures (Fig. 1A). To determine the requirement for N-cadherin in PDA, we examined the survival of the KPC mice with altered N-cadherin gene dosage. Remarkably, KPC Ncad^−/+ (Ncad^fl/+, n=18, 173 days; Ncad^lacZ/+, n=17, 177 days) survived about 25% longer than KPC Ncad^+/+ (n=16, 142 days, p <0.05) mice (Fig. 1B). In contrast, there was no significant change in survival between KPC Ncad^−/− (Ncad^fl/fl, n=14, 140 days; Ncad^lacZ/fl, n=17, 137 days) compared to KPC Ncad^+/+ mice. Hence, the present study will focus on understanding the survival benefit in the KPC Ncad^−/+ mice and N-cadherin-null tumors will serve as an additional control. To examine N-cadherin expression levels, Western analysis was performed on primary tumor cell lines derived from the KPC mice. As predicted by the genotypes, N-cadherin was reduced in the KPC Ncad^−/+ and absent in the KPC Ncad^−/− tumor cells (Fig. 1C). E-cadherin levels were not changed in KPC tumor cells. Moreover, β-catenin and p120-catenin displayed normal expression pattern in KPC tumor cells (Suppl. Fig. 1). Initially, we focused on primary tumors and metastases of terminally ill animals. At the terminal stage, we observed no apparent difference between the tumor pathology of KPC mice regardless of N-cadherin genotype (data not shown). All KPC mice exhibited similar characteristics of end-stage PDA including ascites, regional disease, and distant metastasis. The incidence of lymph node, liver, and lung metastases was similar between the different N-cadherin genotypes (Suppl. Table I). Next, we examined pancreas pathology at an earlier age (Fig. 1D). Tumor development was delayed in the KPC Ncad^−/+ compared to KPC Ncad^+/+ and Ncad^−/− mice at 3 – 4 months of age consistent with abundant interspersed acini remaining in the KPC N-cad^−/+ pancreas. At this early age, six of eleven KPC Ncad^+/+ animals developed PDA whereas only three of ten KPC Ncad^−/+ animals had PDA consistent with a delay in tumor development.

Figure 1.

Prolonged survival of pancreatic cancer mice expressing reduced levels of N-cadherin. (a) Immunohistochemical analysis of E-cadherin (antibody from Santa Cruz) and N-cadherin (antibody from Invitrogen) in KPC Ncad^+/+ tumors. (b) Kaplan-Meier survival analysis of control, KPC Ncad^+/+, KPC Ncad^−/+, and KPC Ncad^−/− mice. Animals heterozygous for N-cadherin (n=17) survived 25% longer (177 vs. 142 days, p <0.05) than N-cadherin wild-type mice (n=16). KPC Ncad^−/− mice had a similar lifespan as KPC Ncad^+/+ (140 vs. 142 days). Transgenic mice were in a mixed genetic background. All mouse experiments were performed under the approval of the Thomas Jefferson University IACUC. (c) Immunoblot analysis of N-cadherin and E-cadherin expression in tumor cell lines derived from the KPC mice. (d) Representative tumor pathology observed in 4 month old KPC mice. An experienced pancreatic pathologist (A.K.W.) reviewed tissue specimens in a blinded fashion.

N-cadherin has been implicated in various aspects of tumorigenesis including tumor cell survival, migration, and invasion (4, 9, 14, 15). To investigate possible explanations for the increased survival of the KPC Ncad^−/+ animals, we first examined tumor cell survival using the apoptotic marker cleaved caspase-3. Interestingly, large foci of dying cells were observed in the KPC Ncad^−/+ tumors compared to small clusters of apoptotic cells in the KPC Ncad^+/+ and Ncad^−/− tumors (Fig. 2A). The increased caspase-3 activity was also observed by Western blot analysis (Fig. 2B). The KPC Ncad^−/+ tumor cells exhibited increased cell death in vitro demonstrating the phenotype was not dependent on the tumor environment (Suppl. Fig. 2).

Figure 2.

N-cadherin regulates tumor cell survival and growth. (a) Immunohistochemical analysis of the apoptotic marker cleaved caspase-3 in KPC tumors. Note the large foci of apoptotic cells in the Ncad −/+ compared to Ncad +/+ and Ncad −/− tumor. (b) Immunoblot analysis of total caspase-3 and cleaved caspase-3 in KPC tumors. Primary antibodies used for immunohistochemistry and Western blot: Caspase-3 and Cleaved Caspase-3 (Cell signaling). (c) Decreased anchorage-independent growth of N-cadherin heterozygous tumor cells. All in vitro assays were performed on low passage number cell lines (<P8). Crystal violet-stained colonies of KPC cells grown for 3 weeks in soft agar. Colonies larger than 100μm were quantified from 30 fields (10X magnification) for each cell line. Each experiment was done in triplicate and repeated twice. Results are expressed mean ± SEM of triplicates. *, p <0.05. (d) qPCR of Hedgehog target genes. Total RNA was isolated from primary cancer cell lines derived from the KPC mice. The data showed is the representative result from three independent experiments. *, p <0.05; **, p <0.01. Expression of the target gene was compared with the expression level of GAPDH. Primer sequences used were Gli-1 forward: 5′ GGAAGTCCTATTCACGCCTTG A 3′, reverse: 5′ CAACCTTCTTGCTCACACATGTAAG 3′; Ptc forward: 5′ CCCTAACAAAAATTCAACCAAACCT 3′, reverse 5′ GCATATACTTCCTGGATAAACCTTGAC 3′; GATA4 forward: 5′-CTGGAGGCGAGATGG-3′, reverse 5′-GGTGGTGGTAGTCTGG-3′; MMP9 forward: 5′-GATCCCCAGAGCGTCATTC-3′, reverse: 5′-CCACCTTGTTCACCTCATTTTG-3; GAPDH forward: 5′-CCACTCTTCCACCTTCGATG-3′, reverse: 5-TCCACCACCCTGTTGCTGTA-3′. (e) For proliferation assay, BrdU Labeling and Detection Kit (Roche) was used on the KPC tumor cells. *, p <0.05. Results shown in all graphs were analyzed by one way ANOVA analysis.

To investigate the cellular and molecular changes associated with the survival advantage in the KPC Ncad^−/+ animals, tumor cell lines were derived from the KPC mice as previously described (16). To determine the relative tumorigenicity of the different KPC tumor cell lines, anchorage-independent cell growth was observed for up to 3 weeks in soft agar. As expected all three genotypes formed colonies in soft agar, however there was a reduced number of large KPC Ncad^−/+ colonies compared to KPC Ncad^+/+ and Ncad^−/− colonies (Fig. 2C) consistent with a delay in tumor development. Moreover, KPC Ncad^−/+ exhibited decreased cell proliferation as determined by 5-bromo-2′-deoxyuridine (BrdU) incorporation (Fig. 2E).

Activation of sonic hedgehog (Shh) signaling occurs in the majority of PDA and is implicated in multiple stages of tumorigenesis (17–19), therefore we performed RT-PCR analysis on Shh target genes. Initially, we examined expression of Gli-1, a transcription factor downstream of Shh signaling. Gli-1 expression was decreased along with the Shh receptor Patched (Ptc) (Fig. 2D) consistent with decreased Shh signaling in the KPC Ncad^−/+ tumor cells. The transcription factor, GATA-4, upregulated in human PDA (20) was markedly decreased in the KPC Ncad^−/+ tumor cells (Fig. 2D). Furthermore, N-cadherin was shown to regulate MMP-9 in breast cancer cells (3, 15), and it was also decreased in the KPC Ncad^−/+ tumor cells (Fig. 2D).

Next we examined the matrix metalloproteinase MMP-7 important for acinar-ductal cell metaplasia and metastasis in animal models and its expression is associated with poor survival in PDA patients (21, 22). Importantly, MMP-7 was decreased in the KPC Ncad^−/+ compared with KPC Ncad^+/+ and Ncad^−/− tumors (Fig. 3A) consistent with increased survival of the KPC Ncad^−/+ animals. Moreover, MMP-7 was also decreased in the KPC Ncad^−/+ tumor cell line (Fig. 3B).

Figure 3.

Reduced MMP-7 expression in N-cadherin heterozygous tumors. (a) Immunohistochemical analysis of MMP-7 (antibody from R&D) in KPC tumors. (b) Immunoblot analysis of MMP-7 expression in primary tumor cell lines derived from KPC mice.

N-cadherin has been shown to interact with FGFR thus facilitating FGF signaling possibly by blocking its ligand-induced internalization thus increasing FGFR stability on the cell surface (15). FGFR2 has been implicated in migration and invasion of pancreatic cancer cells (23), and importantly N-cadherin and FGFR2 co-immunoprecipitate in murine pancreatic tumor cells (8). Therefore, we examined FGFR2 expression in the KPC tumors. Similar to human PDA (23), FGFR2 was not only detected at the cell surface but also in the cytoplasm in the KPC tumor cells with no apparent difference in FGFR2 expression between the N-cadherin genotypes (Fig. 4A). FGF10 produced by pancreatic stromal cells stimulates FGFR2 signaling in human pancreatic cancer cells resulting in increased migration and invasion (23). Importantly, N-cadherin was shown to enhance FGF2/FGFR1 signaling in breast cancer cells leading to sustained activation of the MAPK-ERK pathway (15). To examine the effect of N-cadherin reduction on FGFR signaling, serum-starved KPC tumor cells were stimulated with FGF10 and MAPK-ERK activation was examined over time (Fig. 4B). Interestingly, KPC Ncad^−/+ tumor cells did not exhibit sustained ERK1/2 phosphorylation at 2 hrs compared to KPC Ncad^+/+ and Ncad^−/− tumor cells consistent with reduced FGF10/FGFR2 signaling. As a control, heparin alone did not stimulate the KPC Ncad^+/+ or Ncad^−/+ cells, however we did observe activation of ERK in the Ncad^−/− cells (data not shown) suggesting that Ncad^−/− tumor cells may utilize a different mechanism to activate the ERK pathway. The FGF10/FGFR2 pathway was shown to be important for migration and invasion of human pancreatic tumor cells (23), therefore we performed well-established migration and invasion assays. Consistent with reduced FGF signaling, KPC Ncad^−/+ tumor cells exhibited decreased migration (Fig. 4C) and invasion (Fig. 4D) in the Boyden chamber assays compared to KPC Ncad^+/+ and Ncad^−/− tumor cells.

Figure 4.

Compromised FGF signaling in N-cadherin heterozygous tumor cells. (a) Immunohistochemical analysis of FGFR2 expression (antibody from Santa Cruz) in KPC tumors. (b) For FGF assays, 1.7×10⁵ KPC cells were resuspended in DMEM/F12 complete medium and plated in 6 cm dishes pre-coated with 15 μg/ml rat tail collagen type I (BD). Cells were grown for 24 h, culture medium were removed, dishes were washed once with DPBS and then added with DMEM/F12 plain medium. After 24 h, cells were harvested (0 time point) or stimulated with 1 μg/ml heparin (Sigma) alone (control), or with 50 ng/ml recombinant human FGF10 (R&D) plus 1μg/ml heparin for 10 m, 2 h and 24 h. KPC cells were then harvested, lysed and immunoblotted for phospho-ERK and ERK (Cell Signaling). The graph depicts the fold change of phospho-ERK/total ERK compared to 0 time point of corresponding KPC cell line. Western analysis was performed using different individual KPC cell lines and repeated at least twice. Note the nonsustained activation of ERK in the KPC Ncad^−/+ cells. (c) Tumor cells were seeded on noncoated Boyden chamber membranes for migration assays. (d) Tumor cells were seeded on BD BioCoat^tm Matrigel^tm chamber membranes for invasion assays. After 48 h of incubation, cells that migrated or invaded through the membranes were fixed, stained and counted under light microscope. The data showed is one representative result from three independent experiments. At least two independent cell lines from each genotype were used for the assays. Results shown in the graph were analyzed by one way ANOVA analysis. **, p <0.01. (e) Immunohistochemical analysis of NCAM (antibody from Millipore) in KPC tumors. (f) Immunoblot analysis of NCAM expression in primary tumor cell lines derived from KPC mice.

The neural cell adhesion molecule (NCAM), belonging to the Ig superfamily, like N-cadherin interacts with FGFR and it is important in tumor progression (24, 25). Remarkably, NCAM was decreased in the KPC Ncad^−/+ tumors (Fig. 4E, F) consistent with the idea that these cells are less responsive to FGF stimulation. Moreover, NCAM was increased in the KPC Ncad^−/− tumors suggesting that enhanced NCAM/FGFR signaling may be compensating for loss of N-cadherin.

Perspective

In this study, we demonstrate for the first time that interfering with N-cadherin expression can prolong survival in a spontaneous highly metastatic pancreatic cancer model (10). To our knowledge, this is the first demonstration of a specific genetic manipulation leading to increased survival in the KPC mouse model. It was reported that pharmacological inhibition of hedgehog signaling plus gemcitabine could extend survival of the KPC animals (26). Interestingly, it was recently reported that genetic loss of MMP7 (also decreased in KPC Ncad^−/+) was not sufficient to increase survival, although metastasis was decreased in the absence of MMP7 (22). The overall increase in survival of the KPC Ncad^−/+ mice is likely due to N-cadherin’s role in multiple pathways critical for tumor progression. We show that altering N-cadherin expression in the context of mutant K-ras and p53 impinges on several key steps of tumor progression including tumor cell survival, growth, migration, and invasion. The decreased response of KPC Ncad^−/+ tumor cells to FGF10 stimulation is consistent with N-cadherin’s ability to regulate FGFR signaling in breast cancer cells (15, 27). Hence, we hypothesize that N-cadherin levels fall below a critical threshold required for effective FGF10/FGFR2 signaling in KPC Ncad^−/+ tumors. However, it is unlikely that decreased FGF signaling alone is sufficient to prolong survival in the KPC model. Importantly, other molecules known to be involved in PDA were also affected by reduced N-cadherin expression in the KPC model. It is intriguing to speculate that N-cadherin may influence Shh signaling via its ability to bind the cell surface protein Cdo (28), which can bind directly to Shh (29).

Paradoxically, generating tumors devoid of N-cadherin did not result in a survival advantage whereas reducing N-cadherin levels prolonged survival of the KPC Ncad^−/+ mice. Since the N-cadherin gene is presumably deleted before tumor initiation, the KPC Ncad^−/− tumors must rely on an alternative mechanism for the development and progression of PDA. It is remarkable that tumor cell survival, growth, migration and invasion of the Ncad^−/− tumor cells were similar to the Ncad^+/+. We hypothesize that the lack of a survival benefit in the KPC Ncad^−/− mice is due to compensation by an alternative cell adhesion receptor. NCAM is a potential candidate as it interacts with FGFR and it is upregulated in response to loss of E-cadherin (25) similar to N-cadherin. In future studies, it will be interesting to determine the importance of NCAM upregulation in Ncad^−/− tumors.

The increased survival of the KPC Ncad^−/+ mice is especially exciting as the N-cadherin antagonist ADH-1 was recently shown to block tumor growth, invasion and metastasis in an orthotopic pancreatic cancer model (14). ADH-1 is a cyclic pentapeptide that contains the cell adhesion recognition site, His-Ala-Val, which is found in the first extracellular (EC1) domain of N-cadherin. Furthermore, ADH-1 enhanced the cytotoxic effects of chemotherapy in a melanoma xenograft model resulting in reduced tumor growth (30). Interestingly, FGFR1 expression decreased in mice treated with the chemotherapeutic agent melphalan plus ADH-1, but not when mice were treated with either compound alone. Of note, ADH-1 recently received orphan drug designation from the FDA for its use in conjunction with melphalan for the treatment of Stage IIB/C, III, and IV malignant melanoma. Moreover, monoclonal antibodies against the ectodomain of N-cadherin slowed growth, invasion and metastasis of prostate cancer cells in a xenograft model (31). Taken together, either interfering with N-cadherin function on the cell surface with an inhibitor or genetically suppressing N-cadherin levels (this study) is sufficient to impede tumor cell growth, survival, invasion, and metastasis thus supporting N-cadherin as a bona fide therapeutic target for different cancers. In future studies it will be important to determine whether interfering with N-cadherin function with either ADH-1 or an anti-N-cadherin monoclonal antibody can mimic the N-cadherin haploinsufficiency survival phenotype in the KPC model.