Function of Nuclear Factor κB in Pancreatic Cancer Metastasis¹

NFκB³ is a family of pleiotropic transcription factors that regulates the transcription of a many genes that play key roles in embryonic development, lymphoid differentiation, apoptosis, and immune and inflammatory responses (1, 2, 3). Several reports suggest that the members of the NFκB and their inhibitor, the IκB families, are involved in the development of cancer (3, 4, 5). For instance, c-rel, a member of the NFκB family, was first identified as a cellular homologue of the v-rel oncogene, suggesting that other members of the NFκB family are also oncogenes (4, 5). In addition, the genes encoding c-rel, bcl-3, p105(p50), and p100(p52) are located at sites of recurrent genomic rearrangements in cancer (6, 7, 8, 9, 10). We first reported that RelA, the p65 subunit of NFκB, is constitutively activated in human pancreatic adenocarcinoma and human pancreatic cancer cell lines but not in normal pancreatic tissues or in immortalized/nontumorigenic pancreatic epithelial cells (11). Furthermore, we found that overexpression of downstream target genes of NFκB that are important in cancer metastasis, such as urokinase plasminogen activator, are inhibited by ectopical expression of an IκBαM in these pancreatic cancer cell lines (12). However, it was still not known whether inhibition of constitutive NFκB activity would suppress tumorigenic and metastatic phenotypes in pancreatic cancer cells.

It has been demonstrated recently that NFκB activation is obligatory for retinal angiogenesis; the administration of an NFκB inhibitor suppressed retinal neovascularization (13). Other observations indicate that inhibition of NFκB activity by expressing IκBαM resulted in decreased expression of VEGF and IL-8, angiogenesis, invasion, tumorigenesis, and metastasis in an ovarian cancer cell line, but did not inhibit tumorigenesis in a prostate cancer cell line (14, 15). It is now well established that angiogenesis is essential for carcinogenesis, and the growth of both primary and metastatic tumors. Growth beyond 1–2 mm³ requires tumors to develop an adequate blood supply (16, 17, 18). Angiogenesis, in turn, contributes to metastasis by facilitating the shedding of tumor cells into newly formed blood vessels (19). Tumor angiogenesis is, in part, regulated by angiogenic factors that are produced and secreted by tumor cells. Several positive regulators of endothelial cells have been identified: VEGF and IL-8 are both involved in the process of new vessel formation. VEGF subunits, particularly of the two smaller isoforms (VEGF121 and 165), have little or no heparin-binding activity and induce their angiogenic effects by binding to the specific transmembrane tyrosine kinase receptors KDR/flk-1 and flt-1, which are expressed selectively on vascular endothelial cells, thus stimulating the growth of endothelial cells (20). IL-8 was identified originally as potent activator and chemoattractant for neutrophils (21). Subsequent studies have revealed that IL-8 triggers angiogenesis in vivo via mechanisms that are mediated by direct stimulation of endothelial cell growth or by indirect leukocyte-dependent effects (22, 23). Indeed, human recombinant IL-8 can induce proliferation and migration of human umbilical vein endothelial cells, and it can also stimulate vascularization in a rat cornea assay (24).

Human pancreatic cancer has a very poor prognosis, even after curative resection, and is currently the fifth leading cause of cancer deaths in the United States (25). The overall 5-year survival rate continues to be dismal, at 1–3% (26). Most patients with pancreatic cancer have locally advanced, unresectable disease or metastasis at the time of diagnosis (27). Currently, chemotherapy, radiation therapy, and surgery are largely ineffective in treating this disease (27). To provide a better understanding of the function of constitutive NFκB activity in pancreatic cancer, we determined the role of NFκB in a metastatic pancreatic cancer cell line AsPc-1, using a mouse orthotopic pancreatic cancer model. We showed that constitutive NFκB activity plays a key role in the regulation of the proangiogenic factors, VEGF and IL-8, and in induction of metastasis. Our results suggest that blocking NFκB signaling cascades may provide a potential novel therapeutic strategy for pancreatic cancer.

The human pancreatic adenocarcinoma cell lines AsPc-1, Panc-1, MiaPaCa-2, CaPan-1, MDAPanc-28, and AsPc-1 were obtained from American Type Culture Collection (Rockville, MD). All of the human pancreatic adenocarcinoma cell lines and murine embryonic fibroblast cells were cultured in DMEM (Life Technologies, Inc., Gaithersburg, MD) containing 10% fetal bovine serum, 100 units/ml penicillin (Life Technologies, Inc.), and 10 mg/ml streptomycin (Life Technologies, Inc.) in a 37°C incubator with 5% CO₂.

Female athymic BALB/c nude mice were purchased from Charles River Laboratories, Inc. (Wilmington, MA). One million viable AsPc-1 and AsPc-1/IκBαM cells suspended in 50 μl of PBS were injected into the pancreatic parenchyma of nude mice, being careful to avoid possible leakage of tumor cells from the injected site. Mice were housed in cabinets with laminar flow under specific pathogen-free conditions. Animals were maintained according to institutional regulations in facilities approved by the Association for the Assessment and Accreditation of Laboratory Animal Care in accordance with current regulations and standards of the United States Department of Agriculture, United States Department of Health and Human Services, and NIH. The animal protocol was approved by the Institutional Animal Care and Use Committee at The University of Texas M. D. Anderson Cancer Center. Mice were used in experiments once they reached 8 weeks of age.

The CMV-Flag-IκBαM/puro^r retroviral vector was generated by replacing the XhoI-HindIII fragment of the pRetro-On/puro^r construct, which contains a TRE minCMV promoter and an rTTA sequence, with XhoI-BamHI fragment of CMV-Flag-IκBαM construct, which has mutations (S32, 36A) of the NH₂ terminus and a COOH-terminal PEST sequence, which specifically inhibits phosphorylation of IκBα and subsequent NFκB activation. The CMV-IκBαM construct was provided by Dr. Inder M. Verma (Salk Institute, La Jolla, CA). The CMV-Flag-IκBαM/puro^r and pRetro-On/puro^r control retroviruses were generated, and infections were performed as described previously (28). Pooled puromycin-resistant cells that express Flag-tagged IκBαM were used for subsequent analyses.

One microgram of HIV-κB or VEGF reporter gene construct containing Firefly Luciferase was cotransfected into tumor cells with an internal control, p-TK Renilla Luciferase, using lipotransfection method (FuGENE 6; Roche, Indianapolis, IN) in triplicate. The HIV-κB reporter gene construct contains two HIV-κB enhancers, and the VEGF reporter gene construct contains 1.5 kb of VEGF promoter. The activity of both Firefly and Renilla Luciferase were determined 48 h after transfection using the Dual-Luciferase Reporter Assay System (Promega, Madison, WI).

EMS were performed using nuclear extracts as described previously (29). Briefly, cells were stimulated with 10 ng/ml of TNF-α or 100 nm PS341 for various time intervals as indicated. Ten μg of nuclear extract in a 10-μl reaction volume containing 75 mm NaCl, 15 mm Tris-HCl (pH 7.5), 1.5 mm EDTA, 1.5 mm DTT, 25% glycerol, 20 mg/ml BSA, and 1 μg of poly(dI·dC) was incubated on ice for 40 min. Double-stranded oligodeoxyribonucleotide DNA probes (wild-type κB: 5′-AGTTGAGGGGACTTTCCCAGGC-3′, mutant κB: 5′-AGTTGAGGCGACTTTCCCAGGC-3′, and Oct-1: 5′-TGTCGAATGCAAATCACTAGAA-3′; Sigma, Woodlands, TX) were end-labeled and added to the reaction mixture, incubated at room temperature for 20 min, and applied to a 4% nondenatured polyacrylamide gel containing 0.25 × Tris-borate EDTA buffer [22.5 mm Tris, 22.5 mm borate, and 500 μm EDTA (pH 8.0)]. Equal loading of nuclear extracts was confirmed by determining Oct-1 DNA binding activity. For competition assays, a 100-fold molar excess of unlabeled oligonucleotides was added to the binding reaction. For supershift assay, 2 μl of polyclonal antibodies against p65 and p50 were added to the reaction mixture on ice 40 min before the probe was added. After electrophoresis, the gel was dried for 1 h at 80°C and exposed to Kodak X-ray film (Eastman Kodak Co., Rochester, NY) at −80°C.

For Northern blot analysis, total RNA was extracted using TRIZOL Reagent (Life Technologies, Inc.) according to the manufacturer’s protocol. Fifteen μg of RNA were electrophoresed on a 1% denaturing formaldehyde agarose gel, transferred to a nylon membrane using capillary blotting, and UV cross-linked. Reverse transcription-PCR was performed to obtain cDNA probe for Northern blot. In brief, 1 μg of total RNA extracted from MDAPanc-28 cells was incubated at 42°C for 1 h with 100 ng of oligo(dT)_12–18 primer (Life Technologies, Inc.), 250 μm of deoxynucleotide triphosphate (Promega), 1× incubation buffer of avian myeloblastosis virus reverse transcriptase (Roche), 20 units of RNase inhibitor (Roche), and 25 units of avian myeloblastosis virus reverse transcriptase (Roche) in a final volume of 20 μl. The samples were then heated to 90°C for 5 min to terminate reaction. One μl of the cDNA reaction was then subjected to 30 PCR cycles (denaturing at 94°C for 1 min, annealing at 56°C for 1 min, and polymerization at 72°C for 1 min) using 250 μU of TaqDNA polymerase (Roche), 1× PCR reaction buffer (Roche), 250 μm deoxynucleotide triphosphate (Promega), and 500 nm specific primers (IL-8: 5′-primer, GGACCCCAAGGAAAACTGGG; 3′-primer, GCTGGCAATGACAAGACTGGG; VEGF: 5′-primer, GGCCTCCGAAACCATGAA; 3′-primer, TGGTGAGAGATCTGGTTCCC) in a final reaction volume of 50 μl. The PCR products were extracted and subsequently cloned into a pCR 2.1-TOPO vector (Invitrogen, Carlsbad, CA) for sequence analysis. The sequences of all of the cDNA fragments agreed with those obtained from GenBank. The cDNA probes (EcoRI-EcoRI) were labeled with [α-³²P]dCTP (Amersham Pharmacia Biotech Inc., Piscataway, NJ) using a random primer labeling kit (Roche) and used for hybridization. Equal loading of mRNA samples was determined by rehybridizing the same membrane filter with a cDNA probe for GAPDH as described previously (29).

Western blot analysis of VEGF using whole cell protein extracts was performed as described previously (30). Briefly, to determine the levels of VEGF in conditioned medium, cell-culture supernatants were centrifuged and concentrated. Then 50 μg of protein sample (per lane) was mixed with an equal volume of 2× sample buffer (31), and samples were separated using 8% SDS-PAGE and then transferred onto a polyvinylidene difluoride membrane (Osmonics, Westborough, MA). The membrane was blocked with 5% nonfat milk in PBS containing 0.2% Tween 20 and incubated with a rabbit antibody against VEGF (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). Membranes were washed in PBS containing 0.2% Tween 20 and probed with horseradish peroxidase-coupled secondary goat antirabbit antibodies (Amersham, Arlington Heights, IL). Proteins were detected using the Lumi-Light Western blotting substrate (Roche) according to the manufacturer’s instructions.

Paraffin-embedded sections (for VEGF staining) and frozen sections (for CD-31 staining) were obtained from marginal regions of resected mouse pancreas, and the tumors derived from implanted AsPc-1 and AsPc-1/IκBαM human pancreatic cancer cell lines. The sections were processed for H&E and immunohistochemical stainings. A monoclonal anti-CD31 antibody (PharMingen, San Diego, CA; 1:50) and a rabbit polyclonal anti-VEGF antibody (Santa Cruz Biotechnology, Inc.; 1:500) were used in immunohistochemical studies. Briefly, paraffin-embedded sections were dewaxed in xylene, rehydrated using graded concentrations of alcohol, and treated with 0.5% hydrogen peroxide in methanol for 10 min to block endogenous peroxidase activity. Next, sections were washed three times with PBS for 5 min each time, and nonspecific reactions were blocked by incubating the sections in PBS containing 10% normal goat serum (for VEGF) or 10% normal rabbit serum (for CD-31) for 10 min at room temperature. Sections were then washed three times in PBS for 15 min each. Immune complexes were visualized using the LSAB2 system (Dako, Corp., Carpinteria, CA) and incubated with 0.5 mg/ml diaminobenzidine and 0.03% (v/v) H₂O₂ in PBS for 7 min. Sections were counterstained with hematoxylin and mounted. Mouse lung specimens known to have immunoreactivity for CD-31 were used as positive controls. Positive immunoreactivity for VEGF was determined by a reddish-brown precipitate in the cytoplasm. Blood vessel detected by antiCD-31 antibody were counted in the five areas of the tumor sections using a light microscope at low power (×40 and ×100). The average count in each area was defined as the IMD, and individual IMDs were determined for 200 fields (equivalent to 0.74 mm²). As Weidner et al. (32) have described, neither a vessel lumen nor the presence of RBCs was required to identify a microvessel (32).

Each experiment was performed independently at least twice with similar results each time; one representative result for each experiment was presented. The significance of the data was determined using the Student t test (two-tailed). A value of P < 0.05 indicates significance. All of the statistical analyses were performed using StatView 5.0 (Abacus Concepts Inc., Berkeley, CA).

To determine the role of constitutive NFκB activity in tumorigenesis and metastasis, and in the regulation of the expression of proangiogenic molecules, we constructed a retroviral vector expressing Flag-tagged phosphorylation defective mutant of IκBα (S32, 36A; IκBαM). The human pancreatic tumor cell line AsPc-1 was infected with Flag-tagged IκBαM and a control retrovirus, and these cells were selected for resistance to puromycin (800 ng/ml). Nuclear extracts from pooled puromycin-resistant control AsPc-1 cells and AsPc-1 expressing IκBαM (AsPc-1/IκBαM) were prepared for EMSA. Constitutive RelA/NFκB activity is found in nuclear extracts from AsPc-1 cell lines (Fig. 1,A, Lane 1), consistent with our previous report that NFκB is activated constitutively in human pancreatic adenocarcinoma cells (11). NFκB DNA binding activity is completely abolished in AsPc-1/IκBαM cells (Fig. 1,A). Furthermore, the expression of IκBαM effectively inhibited the TNF-α-mediated NFκB activation in both cell lines (Fig. 1,A, Lanes 6–8). Competition and supershift assays showed the presence of RelA and p50 in the NFκB binding activity in AsPc-1 cells (Fig. 1,B). Consistent with these results, κB-Luciferase reporter-gene assays showed that AsPc-1 cells have high level of constitutive NFκB reporter gene activity, and κB reporter gene assays also indicated an inhibitory effect of IκBαM on NFκB reporter gene activity (Fig. 1 C). These results suggest that the expression of IκBαM inhibits the constitutive and TNF-α-mediated NFκB DNA-binding activities in pancreatic cancer cell line AsPc-1.

To determine whether inhibition of NFκB can suppress tumorigenesis and metastasis, we performed an in vivo experiment using an orthotopic nude mouse model. Six weeks after injection of AsPc-1 and AsPc-1/IκBαM cells, all of the mice became sick and were sacrificed. Pathological examination was performed to determine the extent of tumor formation and metastasis. As shown in Fig. 2,A, 100% (10 of 10) of animals injected with AsPc-1 and AsPc-1/IκBαM cells presented pancreatic tumors. The mice injected with AsPc-1 cells displayed aggressive manifestation of cancer: 36% (4 of 11) had jaundice, 72% (8 of 11) had liver metastasis, 100% (11 of 11) had tumor formation in pancreas, 100% (11 of 11) had peritoneal metastasis, and 72% (8 of 11) had ascites accumulation (Fig. 2, B and C). However, the mice injected with AsPc-1/IκBαM cells revealed extinguished incidence of clinical manifestations: 0% (0 of 10) had jaundice, 0% (0 of 10) had liver metastasis, and 10% (1 of 10) had peritoneal metastasis and ascites formation (Fig. 2, B and C), although the weight of pancreatic tumors was not reduced in these mice as compared with control mice injected with AsPc-1 cells (Fig. 2 D). Taken together, these results suggest that inhibition of constitutive RelA/NFκB activity suppresses the liver metastasis in an orthotropic pancreatic cancer model.

To determine whether inhibition of constitutive NFκB activity down-regulated the expression of NFκB target genes, we performed immunohistochemical staining to analyze the expression of VEGF and the extent of IMD in the pancreas and livers obtained from the mice injected with AsPc-1 and AsPc-1/IκBαM cells (Fig. 3, A and B). Consistent with the phenotypes of AsPc-1-derived tumors, much higher levels of VEGF expression were found in the tumors derived from AsPc-1 cells, and significantly higher IMDs as indicated by CD-31 immunostainings were detected in the adjacent pancreas tissues as compared with those tumors from the mice injected with AsPc-1/IκBαM cells (Fig. 3, A and B). These results suggest that levels of VEGF or the extent of vascularity might be proportional to the aggressiveness of these tumors, because the process of angiogenesis is essential for the growth of metastatic lesions. Our results have shown that inhibition of NFκB-inducible VEGF expression remarkably suppresses metastasis.

Among a variety of proangiogenic genes, VEGF is one of the potential downstream target genes regulated by NFκB (33, 34). To determine levels of VEGF expression in the different pancreatic cancer cell lines, we carried out Western blot analysis using cell culture supernatant collected from Panc-1, MiaPaCa-2, CaPan-1, MDAPanc-28, and AsPc-1 human pancreatic cancer cell lines, and from murine embryonic fibroblast cells, which served as a positive control. As shown in Fig. 4,A, proteins with molecular weights of M_r 23,000, 18,000, and 12–15,000 were detected in these cells, corresponding to VEGF165, VEGF121, and products of processed VEGF, respectively (35). Interestingly, AsPc-1 cells produced a high level of VEGF, whereas Panc-1 cells express an undetectable level of VEGF. To determine the role of NFκB activity in the regulation of VEGF expression, we analyzed VEGF promoter activity using a plasmid containing a Luciferase gene regulated by the 1.5-kb VEGF promoter (VEGF-Luc). This reporter-gene construct was transiently transfected into AsPc-1 and AsPc-1/IκBαM cells. As shown in Fig. 4,B, the expression of IκBαM significantly reduced VEGF promoter-mediated transcriptional activity in both AsPc-1 cells. Moreover, consistent with the result obtained by reporter gene analysis, AsPc-1/IκBαM cells secreted significantly less VEGF protein into the culture supernatant than did control AsPc-1 cells (Fig. 4 C). Taken together, these results suggest that inhibition of constitutive RelA/NFκB activity by using IκBαM resulted in the down-regulation of VEGF expression in AsPc-1 cells. Furthermore, our results suggest that the level of constitutive RelA/NFκB activity is associated with the level of VEGF expression, suggesting that RelA/NFκB regulates VEGF expression.

TNF-α has been suggested to play an important role in tumor angiogenesis and cancer progression (35, 36, 37, 38, 39, 40); therefore, we investigated the role of NFκB activity in regulating TNF-α-mediated VEGF and IL-8 expression. AsPc-1 and AsPc-1/IκBαM cells were stimulated with TNF-α, and total RNA was extracted from the cells at various time points. Northern blot analysis for VEGF and IL-8 were performed. As shown in Fig. 5,A, AsPc-1 cells constitutively expressed a detectable level of VEGF mRNA (Fig. 5, Lane 1), which is in agreement with the high levels of VEGF protein secreted into culture medium (Fig. 4, A and C). In response to TNF-α stimulation, expression of VEGF was increased and peaked at 1 h in AsPc-1 cells (Fig. 5, Lanes 2 and 3). This constitutive and TNF-α-induced VEGF mRNA expression was effectively inhibited by overexpression of IκBαM in AsPc-1 cells (Fig. 5, Lanes 7–10), which were in accordance with results from previous EMSA analysis (Fig. 1,A). AsPc-1 cells expressed undetectable level of IL-8 mRNA; however, TNF-α remarkably induced expression of IL-8 (Fig. 5, Lanes 2–4). IκBαM effectively inhibited TNF-α-mediated IL-8 expression although a weak induction was still observed (Fig. 5, Lanes 7–9). Taken together, these results indicate that TNF-α-mediated expression of VEGF and IL-8 is NFκB-dependent. However, induction of TNF-α-mediated IL-8 mRNA expression is only partially inhibited by IκBαM in AsPc-1 cells, implying the involvement of an alternative signal activation in IL-8 gene regulation.

Our results indicated that specific NFκB inhibition by IκBαM remarkably attenuated VEGF expression. To support these observations, we used the pharmacological NFκB proteasome inhibitor PS-341 to block activation of NFκB, as proteasome-mediated degradation of phosphorylated IκBα is required for NFκB activation (1, 2, 3). AsPc-1 cells were pretreated with 100 nm of PS-341 for 2 h and then stimulated with 10 ng/ml TNF-α for 30 min. Nuclear extracts were prepared, and NFκB DNA binding activity was determined by EMSA. Both constitutive and TNF-α-inducible NFκB DNA binding activities were remarkably inhibited by PS-341 (Fig. 6,A, Lanes 2 and 4). Next, we investigated the effect of PS-341 on the expression of VEGF. Total RNA and whole cell extracts were prepared for Northern and Western blot analyses of VEGF, respectively. AsPc-1 and AsPc-1/IκBαM cells were pretreated with 100 nm of PS-341 followed by TNF-α (10 ng/ml) stimulation for various time intervals. Constitutive VEGF expression was down-regulated significantly by PS-341 (Fig. 6,B, Lanes 1 and 6). Moreover, TNF-α-inducible VEGF expression, observed in (Fig. 6,B, Lanes 2 and 3), was completely inhibited by PS-341. In PS-341-mediated down-regulation of VEGF expression, the level of VEGF protein was undetectable at 24 h, and this inhibitory effect lasted until 72 h (Fig. 6 C, Lanes 1–5). These results demonstrate that blocking of RelA/NFκB activity by PS-341 inhibits VEGF expression in AsPc-1 cells, suggesting that PS-341 is a potential antiangiogenesis agent for treating pancreatic cancer.

In this study, we investigated the effect of blocking NFκB signaling on the tumorigenesis, metastasis, and angiogenesis of AsPc-1 human pancreatic cancer cell line. We inhibited constitutive NFκB activity by stable expression of a phosphorylation-defective IκBα mutant (S32, 36A). This inhibition effectively suppressed liver metastasis from the metastatic human pancreatic tumor cell line AsPc-1 in an orthotopic nude mouse model. Furthermore, ectopic expression of IκBαM significantly inhibited constitutive RelA/NFκB activity and expression of the major proangiogenic molecules VEGF and IL-8, hence decreasing neoplastic angiogenesis. Inhibition of NFκB signaling by ectopic expression of IκBαM or by using pharmacologic NFκB inhibitor PS-341 also significantly reduced TNF-α-induced VEGF and IL-8 expression in the AsPc-1/IκBαM pancreatic cancer cells. These results suggest that the blocking of NFκB signaling reduces angiogenic potential, tumorigenesis, and metastasis of pancreatic cancer.

The suppression of tumor aggressiveness represented as jaundice and metastatic formation in AsPc-1 cells that we found are consistent with the previous report that blocking of constitutive NFκB activity by overexpression of IκBαM decreases tumorigenicity and malignant ascites in an ovarian cancer model (14). Expression of IκBαM also decreases tumor growth and regional lymph node metastasis in a prostate cancer model (15). However, constitutive NFκB activation has not been reported in these tumors. Therefore, the activation of NFκB may be specific to these two tumor cell lines. Interestingly, the tumorigenicity of Panc-1, a well-differentiated pancreatic cancer cell line, was inhibited by the overexpression of IkBαM,⁴ whereas the tumorigenicity in this poorly differentiated and metastatic AsPc-1 cell was not suppressed by overexpression of IκBαM. We showed previously that Panc-1 expressed constitutively activated NFκB and retained wild-type Smad4/DPC4 gene (11, 30). Collectively, these results suggest that additional genetic alterations in AsPc-1 cells may preclude IκBαM-mediated tumor suppression observed in Panc-1 cells.

Our results provided evidence that blocking NFκB activity, which has been shown to be constitutive in 70% of pancreatic cancers, clearly decreases the clinical manifestation, such as jaundice, hepatic and peritoneal metastasis, and ascites accumulation, which all resembles the advanced clinical stages in patients with pancreatic cancer. These findings may provide therapeutic significance for inhibition of NFκB for patients with pancreatic cancer, because most of these patients die from metastatic disease to the lymph nodes, liver, lungs, or peritoneum. For these reasons, our results may provide a molecular basis for novel therapeutic strategies targeting NFκB activity in patients with metastatic pancreatic cancer.

Our results showed that AsPc-1 cells produce rapidly growing and highly vascularized tumors, accumulated ascites, and extensive liver and peritoneal metastasis. The inhibition of constitutive NFκB activity by using IκBαM resulted in a significant reduction of metastatic potential and clinical manifestation in AsPc-1 cells. Our findings suggest that the overexpression of VEGF may correlate with tumor aggressiveness, indicated by clinical manifestations such as jaundice, ascites accumulation, tumorigenesis, and liver metastasis in a mouse orthotopic pancreatic cancer model. These results are consistent with previous reports that VEGF is commonly overexpressed in human pancreatic cancers, and that this factor may contribute to the angiogenesis and metastasis in this disease (41, 42). Recent studies suggest that VEGF is the best validated target for antiangiogenesis therapies. Overwhelming genetic, mechanistic, and animal efficacy data show that blocking VEGF accomplished by using antisense VEGF or an anti-VEGF neutralizing antibody results in an inhibition of angiogenesis and tumor growth in pancreatic cancer (43, 44).

In the current study we also used the proteasome inhibitor PS-341, a dipeptidyl boronic acid that specifically inhibits the 26S proteasome, and, hence, blocked IκB protein degradation and subsequent NFκB activation (45, 46). PS-341 inhibited constitutive and TNF-α-induced RelA/NFκB activity, and suppressed VEGF expression. These results strongly indicate that NFκB inhibition blocks VEGF expression, which may provide therapeutic strategies, directly linking targeting NFκB inhibition with antiangiogenesis therapy in patients with pancreatic cancer. TNF-α has been reported to function as an angiogenic factor, inducing angiogenesis in the cornea and chorioallantoic membrane in vivo (36, 37). TNF-α activates NFκB and enhances several angiogenesis-promoting factors, such as basic fibroblast growth factor, IL-6, VEGF, and IL-8 (35). Our results also revealed that TNF-α enhances NFκB DNA-binding activity with subsequent transactivation of major proangiogenic molecules, such as VEGF and IL-8, in pancreatic cancer cell lines. Furthermore, expression of IκBαM efficiently inhibited TNF-α-mediated NFκB activation and subsequent induction of VEGF and IL-8 expression. Yoshida et al. (35) demonstrated that TNF-α enhances production of IL-8 and VEGF in endothelial cells, and induces angiogenesis in rabbit corneas that are blocked by anti-IL-8 or anti-VEGF antibodies. Moreover, they have reported that NFκB antisense oligonucleotides block TNF-α-dependent induction of IL-8 and VEGF expression, and inhibit TNF-α-induced tubular morphogenesis in endothelial cells. Consistent with these reported data, our present study indicated that NFκB plays an important role in TNF-α-mediated transactivation of VEGF and IL-8 in pancreatic cancer cells. Our findings provide important implications for the therapeutic usefulness of NFκB inhibition to antiangiogenesis therapy in patients with pancreatic cancer.

We thank Dr. Inder M. Verma for generously providing the CMV-IκBαM and Dr. David McConkey for kindly providing PS-341. We also thank Kerry Wright and Pat Thomas for editorial assistance.