The Notch Signaling Pathway Is Related to Neurovascular Progression of Pancreatic Cancer

Abstract

Objective:

To analyze the potential role of the Notch signaling pathway in pancreatic cancer angiogenesis and invasion.

Background:

Angiogenesis, pain, and early neuroinvasion are clinical features of pancreatic cancer. Blood vessels and nerves develop together and use common routes through the organism. The Notch pathway (Notch-1/4, Jagged-1/2, Delta-1) appears crucial in this process. The current study analyzed the Notch pathway in pancreatic cancer and characterized its angiogenic and invasive effects.

Methods:

Five PaCa cell lines were cultured for the in vitro experiments. Real-time quantitative RT-PCR was done to quantify mRNA expression in 31 human PaCa specimens, and immunohistochemistry was used to localize protein expression within tumor specimens. Activation of the Notch signaling was done by transfection of PaCa cells with a constitutive active Notch-1 mutant (Notch-IC). Overexpression of Jagged and Delta was achieved by transfection of full-length cDNA. Spheroid assays were used to study angiogenesis and ELISAs to measure VEGF, bFGF, and angiogenin expression. Matrigel invasion assays were used to analyze tumor cell invasion.

Results:

Notch-3 and Notch-4 mRNA were significantly (P < 0.001) overexpressed in PaCa. Immunohistochemistry revealed protein accumulation of Notch-1 as well. All ligands were significantly up-regulated. A positive immunosignal of ligands was seen in nerves, blood vessels, and ductal tumor cells. Transfection of PaCa cells with the constitutive active Notch-IC mutant and with Jagged-1 revealed increased levels for VEGF. Concomitantly, recombinant Jagged-1 increased sprouting of endothelial cells in the spheroid assay.

Conclusion:

The Notch pathway most likely regulates neurovascular development in pancreatic cancer. Activation of this signaling pathway by constitutive Notch-1 mutants and by Jagged-1 causes an angiogenic and invasive tumor phenotype. Specific blockade of Notch signaling may therefore be beneficial for patients with pancreatic cancer.

The Notch signaling pathway is well known as a neuronal pathway. This study shows that the Notch pathway plays a pivotal role in angiogenesis in pancreatic cancer. Jagged-1, a Notch ligand, and constitutive activation of Notch-1 regulate VEGF gene expression and angiogenesis.

Pancreatic ductal adenocarcinoma represents the fifth leading cause of cancer death in Western countries.¹ Its prognosis has not improved over decades.¹ The recent identification of pancreatic intraepithelial neoplasia (PanIN) as precursors of pancreatic cancer has enabled analysis of premalignant lesions.^2,3 Recently, it has been shown that the Notch signaling pathway, a evolutionary conserved pathway in neurogenesis, also tightly regulates pancreatic development and possibly differentiation of PanIN lesions.^4–7 Members of the Notch gene family encode transmembrane receptors that are involved in cell interaction mechanisms and cell fate decisions during development and postnatal life.^4,8 Mammals have 4 known Notch genes and at least 2 families of Notch ligands, designated “Delta” and “Jagged.”⁹ Notch signals affect cell differentiation, proliferation, and apoptosis.⁴ The most notable feature of the Notch pathway is its induction of lateral inhibition, whereby a single cell is programmed to differentiate through activation of Notch signaling while other neighboring cells retain their undifferentiated state dependent on their ligands and microenvironment.⁵

Upon receptor–ligand interaction, Notch proteins are cleaved within the transmembrane domain. Notch cleavage releases the Notch intracellular domain (Notch-IC), which, dependent upon presenilin-1, translocates to the cell nucleus.^10,11 This “active” form of Notch (Notch-IC) likely participates in neoplastic cell transformation.^12–15 Nevertheless, experimental evidence suggested different roles of Notch signaling in cancer growth, since in some cancer entities it acts as a tumor suppressor gene, whereas in others it possesses oncogenic activity.4,6,16,17

Notch signaling is also important in pancreatic development. Suppression of Notch activity leads to differentiation of pancreatic progenitor cells into endocrine cells, paralleled by a depletion of exocrine progenitor cells.^7,18,19 In pancreatic carcinogenesis, Notch signaling was shown to mediate the tumor-initiating effects of transforming growth factor (TGF)-α by expanding a population of undifferentiated precursor cells.⁶ Notch was also activated as a direct consequence of EGF receptor activation and was required for TGF-α-induced changes in epithelial differentiation.⁶ The role of Notch signaling in angiogenesis is less clear.

Angiogenesis is a process by which new blood vessels are formed from the existing vasculature. This process is relatively rare in healthy human adults and might represent an effective anticancer strategy. Two recent developments in this field have been the delineation of the mechanism by which hypoxia acts as a proangiogenic stimulus via the oxygen-sensing hypoxia-inducible factor-α, and the identification of several novel extracellular angiogenic signaling pathways. The latter include Notch/Delta, the ephrin/Eph receptor, hedgehog, sprouty, and slit/roundabout families. Gene inactivation studies revealed that Notch-1 null mice die due to abnormalities in blood vessels.²⁰ Similarly, inactivation of Jagged-1 results in embryonic lethality due to vascular defects.²¹ The vascular abnormalities of Notch-1 and Jagged-1 mutants are similar, supporting the hypothesis that Jagged-mediated activation of the Notch pathway promotes angiogenesis and contributes to remodeling of the vascular system. Localization studies have suggested a role for the Notch pathway in arterial/venous specification.^20,22

In this study, we analyzed the expression of members of the Notch gene family and their ligands Delta and Jagged in various human pancreatic cell lines and pancreatic disorders. We further analyzed their potential as regulators of tumor neoangiogenesis and tumor cell invasion in pancreatic cancer. This study demonstrates differential expression of the Notch gene and identifies Jagged-1 as a new mediator of tumor angiogenesis and tumor cell invasion in pancreatic cancer.

MATERIALS AND METHODS

Human Pancreatic Cancer Specimens

Human pancreatic cancer tissues used for RNA extraction were obtained from 31 patients undergoing resection for pancreatic cancer at the University Hospital of Bern, Switzerland, and the University of Heidelberg, Germany. Normal human pancreatic tissue specimens (n = 22) were obtained through an organ donor program. Tissue samples for RNA extraction were frozen in liquid nitrogen in the operating room and stored at −80°C. The study protocol was approved by the Ethics Committees at the Universities of Bern and Heidelberg.

For immunohistochemistry, a series of 46 pancreatic cancer tissue specimens were analyzed (27 men and 19 women). None of the patients was treated prior to resection. According to the TNM classifications of UICC, there were 6 patients with stage I, 13 with stage II, 22 with stage III, and 5 with stage IV disease. Twelve tumors were well differentiated (grade 1), 28 tumors were moderately differentiated (grade 2), and 6 tumors were poorly differentiated (grade 3).

Immunohistochemistry

Three-micron sections of formalin-fixed tissues were deparaffinized and rehydrated as described previously.^23–25 Immunostaining was performed using the DAKO Envision System (DAKO, Carpinteria, CA) according to the manufacturer's instructions. Briefly, sections were heated to 56°C for 3 hours, dewaxed in xylene, and rehydrated. Endogenous peroxidase activity was blocked and slides were incubated in normal goat or rabbit serum for 20 minutes, and the primary antibodies were added and incubated at 4°C overnight. The following antibodies were used for immunohistochemistry: the mouse polyclonal antibody against Notch-1 (Ab-1; Neomarkers, Fermont, CA) was used in a 1:100 dilution; the rabbit polyclonal antibody against Notch-2 (25–255; Santa Cruz Biotechnology, Heidelberg, Germany) was used in a 1:500 dilution; the rabbit polyclonal antibodies against Notch-3 (M-134; Santa Cruz Biotechnology) and Notch-4 (H-225; Santa Cruz Biotechnology) were used in 1:100 dilutions. The goat polyclonal antibody against Jagged-1 (C-20; Santa Cruz Biotechnology) was used in a 1:500 dilution and the goat polyclonal antibody against Delta (C-20, Santa Cruz Biotechnology) was used in a 1:100 dilution.

After thorough rinsing in TBS-Tween, incubation of the secondary antibody labeled with biotin was followed by incubation with streptavidin peroxidase and color development by DAB (3,3′-diaminobenzidine tetrahydrochloride). To ensure antibody specificity, control slides were incubated either in the absence of primary antibody or with a nonspecific IgG antibody. All slides were analyzed by 2 independent observers.

Cell Culture

Human pancreatic cancer cell lines (AsPc-1, BxPC-3; Capan-1, MIA PaCa-2, and PANC-1) and the rat pancreatic cancer cell line DSL-6A/C1 were purchased from the American Tissue Type Culture Collection (ATCC, Rockville, MD) and were cultured as previously described.²⁵ These cell lines were chosen because of their different degree of cellular differentiation.²⁴ For angiogenesis assays, the human umbilical vein endothelial cell (HUVEC) was cultured in endothelial cell growth medium containing an endothelial cell growth supplement (Promocell, Heidelberg, Germany).

Transfection of Pancreatic Cancer Cells With Notch-IC, Jagged-1, and Delta-1

A retroviral vector containing the full-length cDNAs encoding either the human Delta-1 or Jagged-1 along with an LZRS-linker-IRES-enhanced green fluorescent protein (eGFP) element was kindly provided by Leonor Parreira (Instituto de Histologia e Embriologia, Faculdade de Medicina de Lisboa, Lisboa, Portugal). These vectors permitted the coexpression of cDNAs and the marker gene eGFP. The CMV-Notch-1 IC vector was a generous gift from Dr. F. Radtke (Ludwig Institute for Cancer Research, Lausanne, Switzerland). For transfection experiments, MIA PaCa-2 and AsPC-1 cells were transfected with a full-length Jagged-1 and Delta-1 cDNA expression plasmids using the Lipofectamine Plus Reagent according to the manufacturer's instructions (Invitrogen, Karlsruhe, Germany). The Notch-1 IC cDNA expression plasmid was cotransfected with an empty pEGFP vector. Forty-eight hours after transfection, cells were analyzed for GFP expression by fluorescence microscopy. Analysis was only continued if > 90% of the cells expressed GFP. In control experiments, an empty retroviral GFP expression plasmid was transfected.

In Vitro Invasion Assays

The function of Jagged-1 in tumor cell invasion was tested with Matrigel invasion assays (BD Biosciences, Heidelberg, Germany). The basement membrane of Costar Transwell was reconstituted and 5 × 10³ DSL-6A/C1 cells were added. Recombinant Jagged protein was added (10, 50, 100, 250, 500 ng/mL) and incubated for 24 hours. The noninvading cells were removed from the upper surface of the membrane. Cells adhered to the lower surface were fixed in 75% methanol mixed with 25% acetone and then stained with 1% Toluidine blue. To calculate the total number of invading cells, the membranes were scanned and cell number in every microscopic cutout of the mosaic image of the membrane was counted using the software Zeiss KS300 (Carl Zeiss AG, Oberkochen, Germany). Assays were done in duplicates and repeated twice.

Determination of Angiogenic Cytokines Upon Recombinant Overexpression of Jagged-1 and Delta-1

Cells were transfected as described with Jagged-1, Delta-1, and Notch-1 IC and grown for 48 hours in DMEM medium supplemented with 10% FBS. They were then washed 3 times with PBS and changed to 1 mL of DMEM medium supplemented with 1% FBS. After the 48-hour incubation period, the supernatant and the cells were harvested. The amount of protein in the supernatant of cells was determined with the Quantikine human VEGF, basic FGF, and angiogenin enzyme-linked immunosorbent assay kits (R&D Systems, Wiesbaden, Germany), according to the manufacturer's instructions. Protein levels were calculated as pg/mL as described.^26,27

Real-Time Quantitative RT-PCR

A total of 10⁶ cells were collected in 300 μL lysis buffer of the MagnaPure mRNA Isolation Kit I (ROCHE Diagnostics, Mannheim, Germany) and mRNA was isolated with the MagnaPure-LC device using the mRNA-I standard protocol. The elution volume was set to 50 μL. Tissue samples were disrupted by one run with the RiboLyser (ThermoHYBAID, Heidelberg, Germany) in lysing matrix “D” tubes (Q-BIOgen, Heidelberg, Germany) containing 400 μL lysis buffer from the MagnaPure mRNA Isolation Kit II (ROCHE Diagnostics). The RiboLyser tubes were centrifuged at 4°C for 1 minute at 13,000 rpm; 300 μL of the lysate was collected and mixed with 600 μL capture buffer containing oligo-dT. After centrifugation at 13,000 rpm for 5 minutes, 880 μL of this mix was transferred into a MagnaPure sample cartridge and mRNA was isolated with the MagnaPure-LC device using the mRNA-II standard protocol. The elution volume was set to 50 μL.

An aliquot of 8.2 μL mRNA was reverse transcribed using AMV-RT and oligo-(dT) as primer (First Strand cDNA synthesis kit, Roche) according to the manufacturer's protocol in a thermocycler. After termination of the cDNA synthesis, the reaction mix was diluted to a final volume of 500 μL and stored at −20°C until PCR analysis.

Primer sets were as follows: Notch-1 (NM_017617) CAATGTGGATGCCGCAGT TGTG and CAGCACCTTGGCGGTCTCGTA; Notch-2 (NM_024408) AAAAATGGGGCCAA CCGAGAC and TTCATCCAGAAGGCGCACAA; Notch-3 (NM_000435) AGATTCTCATCCG AAACCGCTCTA and GGGGTCTCCTCCTTGCTATCCTG; Notch-4 (NM_004557) GCGGAG GCAGGGTCTCAACGGATG and AGGAGGCGGGATCGGAATGT; Jagged-1 (NM_000214) CGGGATTTGGTTAATGGTTATC and ATAGTCACTGGCACGGTTGTAGCAC; Jagged-2(NM_002226) ACCAGGTGGACGGCTTTG and CCGCGACAGTCGTTGA; Delta-1 (NM_005618) CCTACTGCACAGAGCCGATCT and ACAGCCTGGATAGCGGATACAC. The PCR was performed in a LightCycler machine using LightCycler FastStart DNA Sybr GreenI kit (RAS, Mannheim Germany). To control for specificity of the amplification products, a melting curve analysis was performed. No amplification of unspecific products was observed. To correct for differences in the content of total RNA, the calculated copy numbers were normalized according to the average expression the housekeeping gene Cyclophilin B (CPB) and HPRT.

In Vitro Angiogenesis Assays

To study potential pro- and antiangiogenic effects of Jagged-1, a 3-dimensional spheroid model of endothelial cell differentiation was used, as previously described.²⁸ Briefly, standardized endothelial cell spheroids were generated by seeding 750 HUVECs suspended in corresponding culture medium containing 0.25% (wt/vol) carboxymethylcellulose in nonadherent round-bottom 96-well plates (Greiner, Frickenhausen, Germany). Under these conditions, HUVECs form a single EC spheroid. These spheroids were harvested within 24 hours and used for the corresponding experiments as described.²⁹ In brief, HUVE cell spheroids were embedded into collagen gels. A collagen stock solution (2 mg/mL; pH 7.4) was prepared and mixed with endothelial cell growth medium basal medium (Promocell) with 40% FCS (Biochrom, Berlin, Germany) containing 0.5% (wt/vol) carboxymethylcellulose to prevent sedimentation of spheroids prior to polymerization of the collagen gel. Recombinant human VEGF and rat Jagged-1 were added in the indicated concentrations. The spheroid-containing gel was rapidly transferred into prewarmed 24-well plates and allowed to polymerize. The gels were incubated at 37°C in 5% CO₂ at 100% humidity. After 48 hours, in vitro angiogenesis was quantified by measuring the length of the sprouts that had grown out of each spheroid (ocular grid at 100× magnification). At least 10 spheroids were analyzed per experimental group and experiment. The following cytokines were used for the stimulation of HUVE cells: rhu VEGF (25 ng/mL), FGF-2 (25 ng/mL), and recombinant rat Jagged-1 (25, 50, 250 ng/mL). All recombinant proteins were purchased from R & D Systems. Each assay was performed at least 3 times.

Statistical Analysis

Median and mean values of the respective RT-PCR results were statistically analyzed using the SAS program (Statistical Analysis System, Version 6.11, SAS Institute, Cary, NC) and the SPSS program (version 10.0, SPSS, Munich, Germany). The t test procedure for unpaired samples was used to compare the overall expression in cancerous and normal pancreatic tissue samples. P values less than 0.05 were considered as significant.

RESULTS

Expression of Notch Receptors and Notch Ligands in Cell Lines

Expression of various members of this angiogenic pathway, consisting of Notch-1, -2, -3, and -4 as well as the Notch ligands Jagged-1 and -2 and Delta-1, was measured in 6 human pancreatic cancer cell lines (Fig. 1). A remarkably different expression pattern was observed in this set of cell lines. The Notch-2 gene was consistently expressed in all cell lines, however, at different levels. The highest levels of Notch-2 expression were detectable in the undifferentiated cell lines MIA PaCa-2 and PANC-1. The AsPC-1, BxPC-3, and Capan-1 cell lines were entirely devoid of Notch-3 and Notch-4 gene expression (Fig. 1A–C). The Jagged ligands were also expressed in all cell lines, although as for Notch-2, different expression levels were detectable. The highest expression of Jagged-1 was seen in Capan-1 and the T3M4 cell line (Fig. 1C, F). High levels of Jagged-2 expression were found in MIA PaCa-2 and T3M4 cells. The highest expression of Delta-1 was present in T3M4 cells. In contrast, AsPC-1 and Capan-1 cells did not express the Delta-1 gene at all.

FIGURE 1. Real-time quantitative RT-PCR analysis: Expression of Notch-1, Notch-2, Notch-3, Notch-4, Jagged-1, Jagged-2, and Delta was quantified in 6 human pancreatic cancer cell lines (A–F). Notch-2 was detectable in all pancreatic cancer cell lines, whereas other members of the Notch gene family were differentially expressed. Among the ligands, Jagged-1 was expressed in all cell lines, whereas Jagged-2 and Delta-1 mRNA expression was present in some. Normalization of expression levels was done using cyclophilin-B as a housekeeping gene. *P < 0.001. #P = 0.0593.

Expression of Notch Receptors in Human Specimens

Expression of the members of the Notch gene family was determined by real-time quantitative PCR. Included in this analysis were samples from 31 pancreatic cancer patients and 22 previously healthy organ donors. All 4 members of the Notch gene family were detectable in normal pancreatic tissues (Fig. 2). In cancer specimens, Notch-1 gene expression was found at constitutively low levels (Fig. 2A). Higher mRNA levels were detectable for Notch-2 both in normal and pancreatic cancer specimens (Fig. 2A). The Notch-2 gene was not significantly up-regulated in pancreatic cancer samples in comparison to normal pancreatic tissue (Fig. 2A) (P = 0.0593). A significant up-regulation was detectable for the Notch-3 and Notch-4 genes, however (Fig. 2A). A 2.9-fold increase in Notch-3 mRNA expression was seen in pancreatic cancer specimens when compared with the expression level in normal pancreatic tissue. Similarly Notch-4 mRNA expression was 1.8-fold increased in cancer specimens (Fig. 2A).

FIGURE 2. Real-time quantitative RT-PCR: human specimens. Expression of Notch-1, Notch-2, Notch-3, Notch-4, Jagged-1, Jagged-2, and Delta was quantified in 31 human pancreatic cancer specimens (black bar) and compared with expression of these mRNA moieties in normal pancreatic tissue specimens (clear bar) (n = 22). Notch-3 and Notch-4 were statistically significantly overexpressed (*P < 0.001), whereas differences in Notch-2 mRNA expression did not reach statistical significance (#P = 0.0593). The ligands Jagged-1 and -2 as well as Delta were significantly overexpressed in pancreatic cancer specimens (*P < 0.001).

Expression of Notch Ligands in Human Specimens

Expression of the ligands Jagged-1, Jagged-2, and Delta-1 was measured in the same specimens mentioned before. Jagged-1 mRNA expression was up-regulated (1.8-fold) in pancreatic cancer specimens but was also detectable in normal pancreatic tissues (Fig. 2B). In contrast, Jagged-2 mRNA was expressed in normal pancreatic tissue specimens at constitutively low levels but was sharply increased (6.7-fold) in pancreatic cancer specimens (Fig. 2B). Delta-1 expression was detected at low levels in normal pancreatic specimens; however, a 2-fold up-regulation was detectable in pancreatic cancer (Fig. 2B).

Localization of Notch Expression in Human Tissue Specimens

Localization of Notch gene products in human tissue specimens was done by immunohistochemistry. Notch-1 immunoreactivity was detectable in ductal pancreatic cancer cells, but more importantly an intense signal was seen in nerves within the tumor mass, particularly when tumor cells and nerves were in close proximity to each other (Fig. 3A–D). Strong Notch-2 immunoreactivity was present in ductal pancreatic cancer cells and in vascular smooth muscle cells of tumor blood vessels (Fig. 3E–G). Similarly, a strongly positive immunosignal for Notch-3 was seen in vascular smooth muscle cells of tumor blood vessels, whereas endothelial cells were devoid of Notch-3 immunoreactivity (Fig. 3H–J). Immunohistochemical analysis of Notch-3 revealed a heterogeneous staining pattern. Strongly positive areas within cancer specimens were located next to cancer ducts that did not stain positive for Notch-3. Notch-4 expression was seen in endothelial linings of blood vessels but not in vascular smooth muscle cells of tumor blood vessels (Fig. 3K, L). Furthermore, ductal cancer cells exhibited positive immunostaining for Notch-4 as well.

FIGURE 3. Immunohistochemistry of Notch family members in pancreatic cancer samples. Notch-1 immunoreactivity was present in ductal pancreatic cancer cells (A; black arrow) but also in intratumoral nerves (B; white arrow), especially when nerves were infiltrated by cancer cells (C, D). Notch-2 was detectable in tumor cells (E, F) and also in nerves. Please note the nuclear staining (activated Notch) in ductal cancer cells (G). Notch-3 immunoreactivity was marked by staining in ductal pancreatic cancer cells but also in vascular smooth muscle cells (I, J; gray arrow). Tumor cells stained positive for Notch-3 (H), but periendothelial vascular smooth muscle cells also exhibited Notch-3 immunoreactivity (I, J). Notch-4 was found in tumor cells but also in endothelial linings (K, L; red arrow).

Localization of Notch Ligands in Human Tissue Specimens

Localization of Jagged and Delta in human specimens revealed that Jagged-1 was strongly expressed in tumor cells (Fig. 4A–C). Moreover, a strong signal was detectable in the islets of Langerhans, where Notch-1 was also positive (Fig. 4F). Furthermore, a strong signal for Jagged-1 was present in regions of tumor invasion, eg, when tumor cells invaded into nerves, similar to what was observed with Notch-1 staining (Fig. 4C). The vascular staining pattern resembled that of an arterial preference of this protein (Fig. 4D). Vascular smooth muscle cells of arteries were found to be positive, whereas neighboring veins remained Jagged-1 negative (Fig. 4D). The most obvious observation was the relationship between Jagged-1 staining and the pancreatic islets, even when a ductal transformation was seen in the islets (Fig. 4E). Jagged-2 staining in human specimens did not reveal a specific positive signal, which could be due to antibody related issues. In contrast, Delta-1 staining resulted in highly specific immunostaining of cancer cells (Fig. 4G, H). Moreover, Delta-1 staining was primarily found in tumor arteries. Apart form a minority of endothelial cells, which stained weekly positive in vascular hot spots, positive immunosignals were found in vascular smooth muscle cells of arteries constantly (Fig. 4I).

FIGURE 4. Immunohistochemistry of Notch ligands in pancreatic cancer tissues. Strong immunoreactivity was present for Jagged-1 in ductal cancer cells (A; black arrow), in so-called “tubular complexes” (B), but also in intratumoral nerves (B; white arrow), especially when they were infiltrated by cancer cells (C, D). Furthermore, a strong signal was detectable in the islets of Langerhans (E, F). Cytoplasmic Delta-1 immunoreactivity was found in cancer cells (G, H) but also in vascular smooth muscle cells, whereas the endothelial linings remained negative (I, red arrow).

Overexpression of Jagged-1 and Delta-1

The MIA PaCa-2 and AsPC-1 cell lines were transfected with retroviral cDNA expression plasmids encoding the human Jagged-1 and Delta-1 gene. These 2 cell lines were chosen on the basis of differential Jagged-1 and -2 expression with both cell lines devoid of Delta expression. VEGF, basic FGF, and angiogenin protein concentrations were measured in the cell culture supernatant 48 hours after cell transfection. Successful transfection was confirmed by fluorescence microscopy in which 90% of the cells were required to be GFP positive for further protein analysis (data not shown). In the case of Jagged-1 transfection of pancreatic cancer cells, increased VEGF protein concentrations were detectable in cell culture supernatants (Fig. 5A). The amount of secreted basic FGF and angiogenin was not different upon Jagged-1 transfection (Fig. 5B, C). Transfection of Delta-1 did not affect VEGF protein secretion or basic FGF gene expression. However, anigogenin expression was increased upon Delta-1 overexpression in both cell lines, regardless of their constitutive expression levels (Fig. 5C).

FIGURE 5. Overexpression of Notch-1 IC, Jagged-1, and Delta-1 in cultured pancreatic cancer cells: Transfection studies in pancreatic cancer cells were performed as described in Materials and Methods. Angiogenin, VEGF, and basic FGF were measured by specific ELISAs. Overexpression of Jagged-1 resulted in increased VEGF levels in the supernatant of pancreatic cancer cells (A). Similarly, expression of a constitutive active Notch variant (Notch-1 IC) also increased VEGF levels in both tested cell lines. Delta-1 overexpression did not result in changes in the VEGF levels. Basic FGF secretion was higher in the case of Notch-1-IC expression in the MIA PaCa-2 cell line but not in the AsPC-1 cell line (B). Neither Jagged-1 nor Delta-1 transfection changed basic FGF protein expression (B). Low levels of angiogenin in MIA PaCa-2 cells were not changed by Jagged-1 or Delta-1 transfection. A high constitutive expression level of angiogenin was detectable in the AsPC-1 cell line. Upon transfection of this cell line with Delta-1, angiogenin expression increased further. No changes in the expression levels of angiogenin were seen upon transfection with Jagged-1 or Notch-1 IC. *P < 0.05.

Expression of a Constitutive Active Notch-1 Domain

When the Notch receptor binds to its ligand, which is located in neighboring cells, the intracellular domain of the Notch protein (Notch-IC) is released and translocates to the nucleus where it interacts with the transcription machinery. To study the effects of constitutively active Notch-1 signaling, transient overexpression studies were done by transfecting pancreatic cancer cells with vectors expressing Notch-1 IC domain (CMV-Notch-1 IC). Transient transfection was performed as described above. Transfection of Notch-1 IC resulted in increased cytokine levels for VEGF in both cell lines tested (Fig. 5A) but also resulted in a selective increase in basic FGF protein levels in MIA PaCa-2 but not in the AsPC-1 cell lines (Fig. 5B). Furthermore, angiogenin protein was increased in AsPC-1 but not in MIA PaCa-2 cells (Fig. 5C).

Angiogenic Potential of Jagged-1 in Endothelial Cell Spheroid-Based In Vitro Angiogenesis Assay

To explore the ability of Jagged-1 to induce angiogenesis in endothelial cell spheroids as focal starting points for in-gel-based 3-dimensional in vitro angiogenesis, spheroids of a defined cell number (750 cells/spheroid) were seeded in collagen gels and the outgrowth of capillary-like structures was assessed qualitatively and quantitatively. Endothelial cells originating from the embedded spheroids invade the gel to form complex networks of capillary-like structures (Fig. 6). In control experiments, HUVECs were grown without stimulating factors (Fig. 6A), basal sprouting, or in the presence of 25 ng/mL VEGF, which was used as a well-characterized positive control (Fig. 6B). Addition of Jagged-1 to HUVEC spheroids resulted in a dose-dependent increase in the sprouting rate and length of HUVEC spheroids, indicating that endothelial cells can be stimulated by Jagged-1 (Fig. 6C–E). Interestingly, supramaximal doses of 250 ng of Jagged-1 stimulated endothelial cell sprouting similar to VEGF, even if the doses were 10-fold higher (Fig. 6E, F). Lower doses of Jagged-1 were also able to potently induce angiogenic sprouting as well (Table 1).

FIGURE 6. Three-dimensional endothelial cell sprouting assay: Endothelial cell spheroids were seeded as described. Basal sprouting activity is shown in A. Addition of VEGF (25 ng) resulted in vigorous sprouting (B) of HUVECs into this matrix. D, E, Increasing doses (25, 50, and 250 ng/mL) of recombinant Jagged-1 protein were added to HUVEC spheroids. The highest tested dose of 250 ng exhibited an angiogenic stimulus similar to that of VEGF at a dose of 25 ng. The cumulative increase under various growth conditions is summarized in F. *P < 0.05.

TABLE 1. HUVEC Spheroid Cultures as In Vitro Angiogenesis Assay

Tumor Cell Invasion Assays

Matrigel invasion assays were performed to test whether Jagged-1 is a mediator of tumor cell invasion as suggested by the results of the immunostaining. A syngen rat model system was used, consisting of the rat pancreatic cancer cell line DSL-6A/C1 and recombinant rat Jagged protein. In each well, 5 × 10³ cells were seeded and stimulated with increasing doses of Jagged-1 protein (10, 50, 100, 250, and 500 ng/mL). After an incubation period of 24 hours, tumor cell invasion was quantified by counting the number of cells, which migrated through the 8-μm pores of the Transwell membrane. In comparison with untreated cells, addition of Jagged-1 resulted in a significant increase of tumor cell invasion (Fig. 7). Already at lowest concentration of Jagged-1 protein (10 ng/mL) led to a sharp increase in the invasive phenotype. Dose escalation up to 500 ng/mL did not accelerate tumor cell invasion further (Fig. 7).

FIGURE 7. Matrigel Invasion Assays: The rat pancreatic cancer cell line DSL-6A/C1 was used in the Matrigel invasion assays. In each well, 5 × 10³ cells were seeded. Recombinant rat Jagged protein was added in the indicated doses. The invasion assay was analyzed after 24 hours. Invaded cells adhered to the lower surface were counted as described in Materials and Methods. * P < 0.05.

DISCUSSION

It has recently been suggested that the Notch pathway may play a role in pancreatic carcinogenesis.⁶ This study highlighted that ectopic Notch activation resulted in accumulation and expansion of metaplastic ductal epithelium. Furthermore, it has been shown that Notch is activated by EGF receptor activation and is required for TGF-α-induced changes in epithelial differentiation, thus providing the first experimental evidence that Notch mediates the tumor-initiating effects of TGF-α.⁶

The angiogenic effect of the Notch pathway has not been studied in pancreatic cancer. In the present study, we found that Notch family members were differentially expressed in cultured pancreatic cancer cell lines. In human pancreatic cancer tissues, Notch-3 and Notch-4 were found to be expressed at higher levels when compared with normal pancreatic tissue. In the case of the Notch-4 expression pattern, cultured cancer cells did not express relevant levels, whereas whole tissue homogenates of not dissected cancer specimens expressed Notch-4 mRNA abundantly. It is therefore likely that nonepithelial components, eg, the vasculature contributed to increased Notch-4 expression. While Notch-2 was only slightly elevated, the highest relative increase in mRNA expression was found for Notch-3. It is well known from studies in zebra fish that Notch-3 maintains cell–cell interactions and cellular communication between vascular smooth muscle cells and endothelial cells. Furthermore, Notch-3 maintains arterial vessel homeostasis by promoting vascular smooth muscle cell survival. This hypothesis is well supported by our present immunohistochemical findings. Besides being expressed in the vasculature, Notch-3 was also expressed by ductal tumor cells, which also expressed Notch-1 and Notch-2. Notch-1 expression was detectable not only in pancreatic cancer cells but also abundantly in intratumoral nerves. In the case of neural/perineural invasion, a strong signal was detectable by immunohistochemistry. This is not surprising because the Notch pathway was long thought to be a “neuronal”-specific pathway and decisively regulates many pathways in the development of the nerve system. In conjunction with its newly discovered role in oncogenic signaling, this might provide some clues as to why pancreatic cancer possesses this high degree of neuroinvasion.

The Notch ligands were also differentially expressed by pancreatic cancer cells without a correlation to the individual cell differentiation status. In pancreatic cancer tissue specimens, all 3 ligands were significantly up-regulated, with the highest relative increase in Delta-1 expression, even so pancreatic cancer cell lines in culture hardly expressed Delta-1 indicating that Delta-1 expression in culture might be impaired. Spatial expression analysis in tissue samples indicated that Jagged-1 expression was mainly present in invasive tumor areas, a finding that was recently described for prostate cancer as well.³⁰ Furthermore, Jagged-1 was strikingly expressed in the islets of Langerhans, even in normal pancreatic tissue specimens. Delta expression was also present in ductal cancer cells and in vascular smooth muscle cells of tumor arteries, but not in the endothelial compartment, suggesting that Delta participates in oncogenic and angiogenic signaling.

Despite expression of some of the Notch receptors and Notch ligands appeared to be associated with cellular differentiation in vitro, there was no correlation with any of these factors with tumor grade or tumor stage in human specimens.

To further elaborate the angiogenic properties of the Notch pathway, transfection studies were done using cDNA of Jagged-1 and Delta-1 as well as a constitutive active Notch-1 IC construct. Recombinant overexpression of Jagged-1 and constitutive active Notch-1 IC but not Delta-1 overexpression resulted in an increase in VEGF protein secretion in both pancreatic cancer cell lines, indicating for the first time that activated Notch signaling up-regulates VEGF protein secretion. Previously, it had been shown that VEGF induced Notch-1 and Delta expression.³¹ Recently, however, Jagged-2 overexpression increased VEGF protein levels in myeloma cells.³² Our findings are further supported by the observation that Notch-1 activation up-regulated HIF-1α, one of the main regulators of VEGF gene expression in breast cancer cells.³³ Furthermore, Notch-1 IC induces ErbB2 expression, and activated ErbB2 in turn induces expression of VEGF by activation of the transcription factor Sp1.^34,35 Similar mechanisms may apply for the observation that Jagged-1 up-regulates VEGF mRNA expression.^36,37 Thus, it is not surprising that we identified Jagged-1 and activated Notch signaling as new regulators of VEGF gene expression. Alternatively, overexpression of Jagged-1 and Notch-1 IC may activate signaling pathways that are not Notch controlled under physiologic conditions. However, how and when Notch signaling cross talks with any of the above-mentioned pathways have to be further investigated.

The effect of activated Notch signaling on angiogenin secretion was more heterogeneous, with up-regulation of angiogenin in AsPC-1 and MIA PaCa-2 cells upon Delta-1 overexpression. Notch-1 IC overexpression, however, did not alter angiogenin expression. It is not surprising that Delta-1 overexpression increased angiogenin expression since angiogenin is known to act perivascularly where Delta expression was also seen by immunohistochemistry.

The observation that Jagged-1 is actively involved in angiogenesis was additionally tested in a 3-dimensional endothelial cell sprouting assay. In this system, Jagged-1 appeared as a potent stimulus for endothelial cell sprouting and endothelial cell invasion into the gel matrix; even so, the doses necessary for induction of sprouting were slightly higher than those for VEGF-induced angiogenesis. The same phenomenon was seen in the Matrigel invasion assays, where Jagged-1 increased the invasive phenotype of pancreatic cancer cells as suggested by the observation in the immunohistochemical studies.

Taken together, our present study defines, for the first time, a new role of the Notch signaling pathway in pancreatic cancer angiogenesis and identifies Jagged-1 as a new regulator of VEGF gene expression, angiogenesis, and invasion in this disease.

Discussions

Dr. O'Sullivan: Thank you very much, Dr. Büchler, for this fine presentation. I am very flattered to be invited to discuss it; were flattery a ligand, it would be very promiscuous because there would be no scarcity of receptors.

This paper comes from the genre of “transferable concepts of embryogenesis to carcinogenesis.” Cancer development in organogenesis is similar phenomena involving proliferation, angiogenesis, tissue remodeling, and cell migration. Recent studies have suggested, for a number of cancers, that signal transduction pathways, which have been silenced after organ development, are reactivated and become important in the carcinogenic process. A fine example of this, as presented here, is the Notch signal transduction pathway. Another is Hedgehog, which in the context of pancreatic cancer has similar implications.

The link between cancer development and embryogenesis is particularly important because it allows you to use developmental biology to gain insight into tumor biology.

You seem to suggest that the expression of the Notch pathway in the pancreatic cancer converts it to an accelerated invasive pattern, and is also pro-angiogenic. The finding of the Notch signal transduction pathway in some of the neurovascular elements beside the tumor is, I think, a new observation by you. I congratulate you for that.

Much of the work in this area has been done by Leach and his group at the Johns Hopkins, and they have shown that the Notch signaling pathway is intact in pancreatic cancer. In other words, the downstream genes are activated or activatable. They also found Notch activation to be critical for acinar to duct metaplasia and to be present in the earliest or precursor stages of cancer, such as metaplastic lesions and pancreatic intraepithelial neoplasia. My first question is: did you find overexpression of Notch in the neurovascular architecture of intraduct papillary neoplasia or in pancreatic intraepithelial neoplasia? Did you find these elements near the precursor forms of disease?

The progression of invasive pancreatic cancer is a multistep process that involves activation of kRas and mutation/deletion of several oncogenes and tumor suppressor genes. Two phenomena, the expression of the signal transduction notch pathway and mutation/deletion of critical oncogenes, seem to occur in parallel. In Leach's work, not all invasive cancers expressed Notch. This raises the question: is there uniform staining of the pancreatic cancer for Notch or are there prognostic and morphologic differences between those carcinomas of pancreas that express the Notch and those that do not, because it seems that the kRas activation and the mutation deletion of genes could confer a carcinoma phenotype in the absence of this signal transduction pathway?

You seem to propose 2 mechanisms for angiogenesis: the production of the growth factors VEGF and bFGF and also an angiogenic response to the Notch ligand, Jagged-1. Do you know if this is operating through the notch pathway in the endothelial cells? Umbilical cells are embryologic in origin: have you looked at responsiveness in adult-type endothelium?

Metastasis to regional lymph node occurs early in pancreatic cancer, and recently Alitalo and his group from Finland have shown, again in murine pancreatic model, that VEGF C is essential for lymph angiogenesis, and lymph angiogenesis is essential for metastasis. Have you looked at the VEGF isoforms that you produced in response to Notch activation?

The EGF receptor activation plays an important role in the Notch activation during tumorigenesis, and this receptor seems to be activated in the Hedgehog pathway. Do you have any information on coexpression of the 2 signal transduction systems in this tumor?

I think your work is very important because elucidation of these signal transduction pathways could identify therapeutic targets and perhaps control of a deadly cancer. It could also be a rich field for proteomic discovery of circulating biomarkers.

Thank you again for the privilege of discussing this paper.

Dr. Büchler: Thank you for your kind words. First of all, the Johns Hopkins study shows clearly that the Notch pathway is implicated in the oncogenic signaling pathway. We did not look at IPMTs. However, we found that Notch receptors and their ligands were present in PanIN lesions. However, we analyzed too few of these lesions to draw any conclusions with regard to statistical significance.

With regard to the third question, the uniform staining pattern, this is really a good question. To draw a conclusion from the fact that specimens did not stain positive in immunohistochemical analysis is difficult. This could be related, for example, to issues of tissue fixation. We also observed a heterogeneous staining pattern, but we could not establish a correlation with clinical parameters. For instance, we tried to correlate immunohistochemical data with tumor grading and with the tumor stage. With regard to lymph node metastasis, we had the problem that most pancreatic cancer patients have: of course, positive lymph nodes. We just had a few who did not have positive lymph node status. Altogether, we could not establish the link between expression of Notch family members and clinical parameters.

With regard to the angiogenesis mechanisms: as a matter of fact, it was previously not reported that Jagged-1 regulates VEGF gene expression. We tried to prove this by 2 principles: one was intracellular active signaling mutant and the other was Jagged-1 overexpression. Published data suggested that Notch-1 activation up-regulated HIF-1α, one of the main regulators of VEGF gene expression. Furthermore, Notch-1 IC induces ErbB2 expression, and activated ErbB2 in turn induces expression of VEGF by activation of the transcription factor Sp1. In summary, we did identify for the first time that VEGF expression is regulated by Notch signaling, but it has been suggested in the literature indirectly before.

Regarding the type of endothelial cells we used, for our assay we used HUVECs, which are umbilical vein cells. I assume that adult cells would display similar characteristics, but scientifically we have no evidence for cells other than HUVECs.

With regard to lymph angiogenesis and lymphatic metastasis, we have data from another study showing that VEGF-C was expressed in pancreatic cancer, but in this study we measured different VEGF isoforms. What we measured was VEGF-A, which is the predominant form for the vessel development with regard to blood vessels.

The hedgehog pathway and the Notch pathway are likely linked, but we did not analyze this in the current study.

Dr. Eggermont: I have a question about the perineural invasion, which is always an intriguing phenomenon in pancreatic cancer, and that is part of the reason why the local pain problems maybe so predominant in many cases. How would you be able to further formalize this perineural invasion in terms of taking it one step further than the pathologic observations that you have made thus far. Do you have animal models available or any other experimental setting where you can study this, or will you encounter species specificity problems that need to be overcome or that will force you to work with syngeneic systems. I personally do not know whether there is any species specificity involved in some of these ligands that are essential in the process. Would you be able to address this in a model to study this further?

Dr. Büchler: This is an excellent question since it proposes a model for nerve development. We actually chose this set of factors because they are so predominant in development of the bowel and in development and differentiation of neural tissue.

Honestly, studying nerve development experimentally is very difficult, and we do not have any assay available where we can easily study nerve development and neural invasion in vitro or in vivo. The only method I can think of is that we have a specific inhibitor 1 day, which we can then apply in the orthotopic murine model to study nerve invasion. For now, we are entirely dependent on observation in tissue specimens; and once an inhibitor is available, we can certainly address this further.

Dr. Tuynman: Thank you for your nice presentation. You have shown that Notch-3 and -4 are important in pancreatic cancer, and you hypothesized that the recombinant Jagged-1 is a specific ligand for the Notch receptor. But I am curious whether you know that the ligand is really specific. Does Jagged-1 not also activate the IGF receptor, for instance? This is essential if you want to translate these results to therapy, where you have to choose to block either Notch-3 or -4 or block the ligand activity.

Dr. Büchler: Again, a brilliant question addressing the binding specificity of Notch receptors and their ligands. We rely upon data from murine transgenic studies, and I agree that these are studies that are done in mice, and they could be different from the human situation. With these data in mind, we also made the observation that Notch 3, for instance, is expressing the muscle cells of a vessel, and Notch 4 is in the endothelial cells. This expression was also observed in murine experiments. With regard to the specificity, Jagged-1 is considered a specific ligand for Notch. However, I agree that it is very possible that other factors possibly bind or activate the Notch signal in pathways, especially if it is present in high doses as in cancer, where almost every single factor is overexpressed. There are reasons to believe that, for example, EGF could very well stimulate the Notch pathway; and as a matter of fact, this has been indicated by Dr. Leach's study.