The Nuclear Factor-κB RelA Transcription Factor Is Constitutively Activated in Human Pancreatic Adenocarcinoma Cells¹

Pancreatic adenocarcinoma is the fifth leading cause of adult cancer mortality in the United States. (1). However, the epidemiology of pancreatic cancer provides few clues about its etiology and pathogenesis. Strategies for early detection of pancreatic cancer have not yet been developed, and most pancreatic adenocarcinomas present with metastatic or locally advanced disease at the time of diagnosis (2). Therapeutic options for patients with advanced disease are few, because chemotherapy and irradiation are largely ineffective, and metastatic disease frequently develops even after potentially curative surgery (2, 3, 4). Nonetheless, studies of the biology of pancreatic adenocarcinoma have revealed that activation of specific oncogenes and inactivation of specific tumor suppressor genes play critical roles (2, 5).

The c-rel member of Rel/NF³-κB pleiotropic transcription factors was first identified as a cellular homologue of the v-rel oncogene, indicating the possibility that other members of Rel/NF-κB are oncogenes (6). Rel/NF-κB controls the expression of numerous genes involved in the immune response, embryonic development, lymphoid differentiation, oncogenesis, and apoptosis (6, 7). The Rel/NF-κB family consists of p65, Rel (v-rel), RelB, p50 (p105), and p52 (p100), which can form heterodimers and homodimers among themselves (6, 7). In most cell types, Rel/NF-κB proteins are sequestered in the cytoplasm in an inactive form through their noncovalent association with the inhibitor IκB (8, 9). This association masks the nuclear localization signal of Rel/NF-κB, thereby preventing Rel/NF-κB nuclear translocation (8, 9). Activation of Rel/NF-κB is controlled by IκBα and does not require protein synthesis, thereby allowing rapid and efficient gene regulation (6, 7, 8, 9). Previously, we cloned IκBα cDNA and its promoter and described a feedback inhibition pathway to control IκBα gene transcription that down-regulates transient activation of Rel/NF-κB (10). We also showed that enhanced IκBα degradation was responsible for constitutive activation of Rel/NF-κB activity in mature murine B-cell lines (11). Stimulation of cells by various inducers that leads to phosphorylation of IκBα at serine residues 32 and 36 by the recently identified and cloned IκB kinases, IKK1 and IKK2, triggers the rapid degradation of the inhibitor (12, 13, 14, 15, 16, 17, 18, 19). Consequently, Rel/NF-κB proteins are released and translocated into the nucleus, where they activate the expression of target genes. MEKK1, a kinase involved in TNF-α-induced NF-κB activation, was identified as one of the tightly associated subunits in IκB kinase complex and activates the IκBα kinase complex by phosphorylation (15, 20, 21). These findings have provided functional analysis of components of an IκB kinase complex for a better understanding of the signal transduction cascade leading to activation of Rel/NF-κB.

Several reports suggest that members of the Rel/NF-κB and IκB families are involved in the development of leukemias and lymphomas (22). The genes encoding c-rel, bcl-3, nfkb1, and nfkb2 have been shown to be located at sites of recurrent genomic rearrangements in these lymphoid cancers (22, 23, 24, 25, 26). Carried by a highly oncogenic retrovirus, v-rel causes aggressive tumors in young birds and is able to transform avian lymphoid cells and fibroblasts (27, 28). The mutated c-rel oncogene also transforms cells (29). Furthermore, the expression of IκBα antisense resulted in constitutive activation of RelA and oncogenic transformation of NIH/3T3 cells (30), suggesting that relA is an oncogene. A number of reports have shown that activation of NF-κB may be critically involved in tumorigenesis. The Tax protein from the human T-cell leukemia virus (HTLV-1) is a potent activator of Rel/NF-κB, and the growth of Tax-induced tumors in mice was inhibited by antisense relA constructs (31). Ras and the MAP kinases are involved in the activation of Rel/NF-κB transcription factors (32, 33). However, activation of NF-κB RelA in nonlymphoid human cancers has not been identified previously. The signal cascades leading to RelA activation and RelA downstream target genes that are relevant to tumorigenesis remain unclear. An interesting possibility is that mutated Ras may involve activation of Rel/NF-κB (34). We therefore undertook a study to determine the activity of NF-κB RelA transcription factors in pancreatic adenocarcinoma cells in which oncogenic activation of the K-ras gene by mutation has been identified frequently (85%; Ref. 35).

Tissue samples from patients with adenocarcinoma of the pancreas were obtained at the time of surgery and were histologically confirmed to be adenocarcinoma. Histologically normal pancreas was also obtained from the region of the pancreatic neck, proximal to the site of tumor. All human pancreatic cell lines (MDAPanc-3, MDAPanc-28, MDAPanc-48, ASPC-1, BXPC-3, Capan-1, Capan-2, CFPAC-1, Hs766T, MiaCaPa-2, and Panc-1) were grown in the original cell culture medium specified by American Type Culture Collection (Rockville, MD) and DMEM containing 10% FCS. Human pancreatic tumor cell lines MDAPanc-3, MDAPanc-28, and MDAPanc-48 were established by M. Frazier and D. B. Evans (M. D. Anderson Cancer Center). SGH cell lines CK2, CK4, and Pan-1 and other immortalized/nontumorigenic cell lines were provided by P. M. Pour and T. Lawson (University of Nebraska, Omaha, NE). These cells were grown in DMEM/F12 containing 10% Nuserum IV, 1 μm dexamethasone, 2 μg/ml insulin, 10 ng/ml epidermal growth factor, 20 μg/ml bovine pituitary extract, 100 units/ml penicillin, and 100 μg/ml streptomycin.

Formalin-fixed, paraffin-embedded adenocarcinoma tissues were obtained at the time of pancreatectomy from patients at our institution. Immunohistochemistry for specific detection of activated RelA was performed using monoclonal anti-human RelA antibody that recognizes an epitope overlapping the nuclear localization signal and IκBα binding site of RelA (CDTDDRHRIEEKRKRKT; Boehringer Mannheim, Indianapolis, IN). A control peptide was synthesized (CDTDDRHRIEEKRK-RKT) for competition assays. For detecting RelA proteins, polyclonal anti-RelA antibody that recognizes an epitope corresponding to amino acids 3–19 in the NH₂-terminal domain of RelA (Santa Cruz Biotechnology, Santa Cruz, CA) and a control peptide (amino acids 3–19) were used. Immunohistochemical staining was performed as described previously (36).

The pancreatic cell lines were treated with TPA (50 ng/ml) or TNF-α (5 ng/ml) for 60 min. Some of the pancreatic cells were treated with N-tosyl-l-phenylalanine chloromethyl ketone (50 μm) for 1 h and curcumin (50 μm) for 6 h. The nuclear extracts were prepared according to the method of Andrews and Faller (37). The concentration of the extracts was ∼5 mg/ml. For EMSAs, 10 μg of nuclear extract were incubated with 1 μg poly(deoxyinosinic-deoxycytidylic acid) (Pharmacia) in a binding buffer [75 mm NaCl, 15 mm Tris (pH 7.5), 1.5 mm EDTA, 1.5 DTT, 25% glycerol, and 20 μg BSA] for 30 min at 4°C. ³²P-Labeled double-stranded oligonucleotides (5′-CTCAACAGAGGGGACTTTCCGAGAGGCCAT-3′) containing the κB site found in the HIV long terminal repeat were used as probes. The mutant κB site for HIV long terminal repeat (5′-CTCAACAGAGTTGACTTTTCGAGAGGCCAT-3′) was used for competition studies. The competition was performed with a 50-fold excess of unlabeled wild-type or mutant κB oligonucleotides. The supershift experiments were performed with anti-RelA antibody in the absence or presence of the control peptide (Santa Cruz Biotechnology, Santa Cruz, CA). The binding of the probe was performed for 20 min at room temperature in a total volume of 15 μl. The reactions were analyzed on 4% polyacrylamide gels containing 0.25× TBE (Tris/borate/EDTA) buffer.

The cells stimulated by either TPA or TNF-α for 1 h were harvested at the same time intervals as in EMSA, and RNA was isolated as described by Chomczynski and Sacchi (38). RNA (25 μg) was electrophoresed through a 1.2% agarose gel containing formaldehyde, transferred to a Hybond nylon filter (MSI), and UV cross-linked. The blots were hybridized with a ³²P-labeled 1.1-kb human IκBα (MAD3) cDNA (EcoRI-EcoRI) probe, exposed, stripped, and rehybridized with the cDNA probe for glyceraldehyde-3-phosphate dehydrogenase as described previously (10).

Histologically normal and adenocarcinoma cells of the pancreas that were obtained and frozen at the time of surgery were ground into fine powder in liquid nitrogen, lysed, and homogenized in 250 μl of lysing buffer. Twenty-five μg of the lysates were resolved by SDS-PAGE, transferred to nylon membranes (Immobilon-P; Millipore, Bedford, MA), and detected with IκBα antibody specific for the NH₂ terminus (amino acids 1–56) of the IκBα protein. The subsequent Western blot analysis was carried out with an ECL western blotting kit (Amersham) according to the manufacturer’s recommendations.

One μg of the HIV κB-CAT reporter plasmid, 2.5 μg of β-actin promoter LacZ expression plasmids, and 5 μg of IκBαM, RafDN, MEKK1DN, and CMVpBS (control) expression plasmids were used in each cotransfection. Forty-eight h after lipotransfection, the cells were collected. Relative transfection efficiency was determined by cotransfected LacZ expression plasmid, and subsequent β-galactosidase activities in cell extracts were used to normalize the transfection efficiencies. CAT assays were performed as described previously (10), and CAT activity (percentage of conversion to acetylated chloramphenicol) was determined by phosphoimage analysis.

To investigate whether Rel/NF-κB-DNA binding activities were altered in human pancreatic adenocarcinoma, we first carried out immunohistochemical analyses. The monoclonal antibodies (Boehringer Mannheim) used in this study detect only activated RelA proteins (36). They do this by recognizing an epitope overlapping the nuclear localization signal and IκBα binding site of RelA. Therefore, these monoclonal antibodies can selectively and specifically bind to the activated form of RelA (36) and are useful in differentiating between activated and inactivated forms of RelA. When the affinity-purified monoclonal antibodies, which detect only the activated RelA proteins, were used in the analysis, specific RelA staining was detected in 16 of 24 pancreatic adenocarcinomas but not in normal human pancreatic ductal epithelial cells or surrounding fibroblastic stroma (Fig. 1, A and B). The staining for RelA was competed away using the control peptide (Fig. 1 C). Similar levels of RelA proteins were detected by the polyclonal rabbit anti-RelA antibodies, which recognize all of the RelA proteins, in both normal human pancreatic ductal epithelial cells and pancreatic adenocarcinoma (data not shown). These results show that RelA is activated in human pancreatic adenocarcinoma cells but not in the normal pancreatic ductal cells.

To confirm the finding that RelA-DNA binding activity is activated in human pancreatic adenocarcinoma, 24 paired histologically normal and tumor tissue samples from patients with adenocarcinoma of the pancreas were obtained at the time of surgery, and the cytoplasmic and nuclear extracts were prepared as described previously (10). Using EMSA, constitutive RelA NF-κB activity was detected in 70% (14 of 20) of human pancreatic adenocarcinoma tissues but not in the paired normal tissues (Fig. 2,A). Using the unlabeled wild-type and mutant κB oligonucleotides in competition and specific anti-RelA antibody in supershift, EMSA indicated that the binding activities in the extracts were specific to κB sites and to RelA (Fig. 2,B). Furthermore, most if not all of the κB binding activity was shifted by anti-RelA antibody, pointing out that the constitutive κB DNA binding activity contained RelA protein, not c-Rel or RelB (Fig. 2 B). Our results show that RelA is frequently activated in human pancreatic adenocarcinoma tissues but not in normal pancreatic tissues.

To address the question of whether any known RelA target genes are induced in these tumor tissues, we analyzed the expression of IκBα, a previously identified RelA inducible gene (6, 10). As shown in Fig. 2,C, the level of IκBα protein in the cytoplasmic extracts was increased at least 10-fold in pancreatic tumor tissues compared with that in the adjacent normal pancreatic tissues. Taken together, these results (Figs. 1 and 2) demonstrated the constitutive activation of RelA in the majority of human pancreatic adenocarcinomas. This is the first demonstration that RelA is constitutively activated in a nonlymphoid human cancer. Interestingly, the activated RelA was detected in 65% (13 of 20) of the pancreatic adenocarcinoma tissues that were shown previously to carry a K-ras mutation at codon 12 (39, 40).

To determine whether RelA-DNA binding activity also was activated in human and SGH pancreatic adenocarcinoma cell lines, we performed EMSA using the nuclear extracts from control cells (Jurkat cells stimulated with 50 μg/ml TPA), human pancreatic adenocarcinoma cell lines, the tumorigenic SGH pancreatic cell line Pan-1, and nontumorigenic SGH pancreatic cell lines CK2 and CK4. Our results show that constitutive RelA activation was detected in the human pancreatic tumor cell lines MDAPanc-28, MDAPanc-48, Capan-1, Capan-2, Panc-1, BxPC-3, MiaCaPa-2, AsPC-1, CFPAC-1, and SGH pancreatic tumor cell line Pan-1 but not in human pancreatic tumor cell lines MDAPanc-3 and Hs766T or in nontumorigenic SGH cell lines CK2 and CK4. Fig. 3 shows examples of RelA activity in the pancreatic cell lines that we studied. Constitutive RelA DNA binding activity was detected in the human pancreatic tumor cell lines CFPAC-1, Capan-1, BxPC-3, AsPC-1, and SGH pancreatic tumor cell line Pan-1 but not in nontumorigenic SGH cell lines CK2 and CK4 (Fig. 3,A). The RelA-DNA binding activity was further characterized with or without phorbol myristate acetate and TNF-α stimulation in six human pancreatic tumor cell lines. Fig. 3,B shows that stimulation of MDAPanc-28, Capan-1, AsPC-1, CFPAC-1, and BxPC-3 cells with phorbol myristate acetate (50 μg/ml) or TNF-α (5 ng/ml) did not further induce RelA-DNA binding activity, indicating that the RelA was already activated. In HS766T cells, RelA activity was inducible by TPA or TNF-α (Fig. 3,B). When unlabeled wild-type and mutant κB oligonucleotides were used in competition and the anti-RelA specific antibody was used in EMSA supershifts, the results indicated that the binding activities in the extracts were specific to κB sites and RelA (Fig. 3 C). No differences in oct-1 and AP-1 DNA binding activities were detected in these pancreas cancer cell lines tested, in which RelA is constitutively activated and in MDAPanc-3 cells, one of the two human pancreatic cancer cell lines with inducible RelA activity (data not shown). The results obtained from further analyses were consistent with RelA activation being detected by multiple gel-shift analyses of independent samples at different passages in the original cell-culture medium specified by American Type Culture Collection and in DMEM containing 10% FCS (data not shown). Thus, we have concluded that RelA is constitutively activated in most pancreatic tumors but not in normal pancreatic tissues and nontumorigenic SGH pancreatic cells and that constitutive activation of RelA is a stable alteration in these pancreatic adenocarcinoma cells. These cell lines have provided a useful in vitro system for studying the signal transduction pathway leading to constitutive RelA activation and the role of constitutively activated RelA in pancreatic adenocarcinoma.

Because the inhibitor IκBα is one of the downstream target genes regulated by RelA (6, 10), Northern blot analysis for determining the level of IκBα mRNA was carried out to confirm RelA constitutive activities. As shown in Fig. 3,D, the levels of IκBα mRNA were induced by TPA (50 μg/ml) or by TNF-α (5 ng/ml) only in Hs766T cells, which is consistent with the inducible RelA activity (Fig. 3 B, Lanes 16–18), but the levels of IκBα mRNA were already up-regulated without any stimulation and were not further induced by TPA or TNF-α in MDAPanc-28, Capan-1, ASPC-1, CFPAC-1, or BXPC-3 cells. Therefore, these results provide further evidence that RelA is constitutively activated in most human pancreatic adenocarcinoma cells.

Curcumin, a potent antioxidant and cancer chemopreventive agent, has been shown to inhibit kinase activity and TNF-α-induced activation of NF-κB at a step before the phosphorylation of IκBα (41, 42). We therefore examined the effect of curcumin on constitutive RelA activity in the human pancreatic tumor cell line MDAPanc-28. Our data showed that constitutive RelA-DNA binding activity in pancreatic cancer cells was totally inhibited by curcumin (25 μm), as evidenced by the complete absence of such κB-specific DNA binding activity in extracts from cell lines after 6 h of treatment, whereas there was no reduction of RelA-DNA binding activity in control cells with or without DMSO treatment (Fig. 4,A). Moreover, the constitutive RelA activity was totally inhibited by N-tosyl-l-phenylalanine chloromethyl ketone treatment, as we reported previously (Ref. 10; Fig. 4 A). These results suggest that the constitutive RelA activity in pancreatic adenocarcinoma cells is induced by the activation of the upstream signal-transduction cascades leading to activation of RelA.

To determine whether constitutive activation of RelA can be inhibited by specific inhibitors, we performed a series of transfection experiments to analyze the inhibition of constitutive RelA activity by expression of dominant-negative mutants of IκBα, Raf, and MEKK1 (Fig. 4,B). The results show that dominant-negative IκBα, c-Raf, and MEKK1 inhibited RelA-induced transcriptional activation specifically through κB sites in a CAT reporter gene (Fig. 4,B) but did not inhibit a β-actin promoter/β-galactosidase reporter gene (Fig. 4,C). The inhibition of NF-κB activation by dominant-negative IκBα, Raf, and MEKK1 in pancreatic tumor cell lines is consistent with the earlier results obtained in other cell lines (43, 44). These results shown in Fig. 4 suggest that constitutive activation of RelA in pancreatic adenocarcinoma cells is induced by activation of upstream signal cascades involving Ras and MAP kinases.

To determine the effect of curcumin on the growth of pancreatic cancer cells, we treated MDAPanc-28 cells with or without Taxol, a chemotherapy agent used in treatment of pancreatic cancer and a known apoptotic inducer, in the presence and absence of curcumin. 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay was used to determine the percentage of viable cells at the end of each time point. The results shown in Fig. 4 D indicated that curcumin did not inhibit the growth of MDAPanc-28 cells, but curcumin-mediated inhibition of RelA activity may sensitize these cells to Taxol-induced apoptosis. Our finding is consistent with the previous reports that RelA plays a key role in regulation of apoptosis (43, 44, 49).

We have demonstrated that RelA-DNA binding activity is constitutively activated in the majority of human pancreatic adenocarcinomas and human and SGH pancreatic adenocarcinoma cell lines but not in normal pancreatic tissues or the nontumorigenic SGH pancreatic cell lines tested. RelA may be constitutively activated in pancreatic tumor cells by upstream signaling cascades, possibly involving Ras and MAP kinases.

In our immunohistochemical analyses for detecting Rel/NFκB-DNA binding activities in human pancreatic adenocarcinoma, the monoclonal antibodies detected the activated RelA proteins only by recognizing an epitope overlapping the nuclear localization signal and IκBα binding site of RelA and therefore selectively and specifically bound to the activated form of RelA (36). This selectivity and specificity of the monoclonal antibody were characterized previously, and it has been shown that these monoclonal antibodies cannot bind to RelA proteins when they are associated with the inhibitor IκBα (36). In our early studies of constitutive Rel/NFκB activity in B cells undergoing differentiation (6, 11), we found that only ∼10–20% of total Rel/NFκB proteins were detectable in the nucleus. It is unclear why the majority of Rel/NFκB proteins that are freed from IκBα remain in the cytoplasm (6). Because only ∼10–20% of RelA proteins can be localized in the nucleus and 80–90% of RelA still remain in the cytoplasm when RelA proteins are activated constitutively by long-term stimulation (11), this monoclonal anti-RelA antibody was useful in differentiating between activated and inactivated forms of RelA and facilitated the detection of the activated RelA proteins. Our data from immunohistochemical analyses showed that RelA was constitutively activated in human pancreatic adenocarcinoma. The conclusion is further supported by the detection of RelA-DNA binding activity in the nucleus of pancreatic adenocarcinoma, but not in normal pancreatic tissues, and by the detection of overexpressed IκBα protein in pancreatic adenocarcinoma but not in normal pancreatic tissues.

The subsequent analyses also show constitutive RelA activity in nine human pancreatic tumor cell lines and in the SGH pancreatic tumor cell line Pan-1 but not in the immortalized/nontumorigenic SGH cell lines CK2 and CK4, which were used to compensate for the lack of nontumorigenic human pancreatic ductal cell lines. The tumorigenic pancreatic ductal epithelial cell lines from SGH, with similarities in pathology, morphology, and molecular alterations to human pancreatic cancer, provide a biologically relevant in vivo model for analyzing the molecular alterations in signal transduction cascades (45, 46). These results raise the interesting possibility that RelA is an oncogene and that its constitutive activity plays a critical role in pancreatic tumorigenesis. Recently, others have shown that constitutive activation of RelA and oncogenic transformation have been achieved by the expression of IκBα antisense in NIH/3T3 cells (30).

Little is known about the involvement of Rel/NFκB transcription factors and their inhibitor IκBα in oncogenesis of nonlymphoid cancers. Thus far, only three reported cases of the alteration in expression of Rel/NF-κB have been associated with nonlymphoid cancers. p65, p52 (p100), and p50 (p105) proteins are highly expressed in some breast and lung cancers (22, 47). However, no alterations in κB DNA binding activity associated with overexpressed Rel/NF-κB proteins have been demonstrated previously. It has been shown that overexpression of RelA in transgenic mouse thymocytes specifically increased the level of inhibitor IκBα but not the overall NF-κB-binding activity in unstimulated cells when compared with those of control thymocytes (48). These results indicate that cytoplasmic retention of overexpressed RelA by IκBα is the major in vivo mechanism controlling the potential excess of NF-κB activity in long-term RelA-overexpressing cells and explain why overexpression of RelA, RelB, and c-Rel does not induce tumors in transgenic mice. The constitutive RelA activity induced by the antisense expression of inhibitor IκBα has been reported to transform NIH/3T3 cells, demonstrating that RelA is also a transforming gene besides mutated c-Rel and v-Rel (30). However, the fibroblast cell lines established from the IκBα knockout mice, which exhibited constitutive RelA activity and a postneonatal lethal phenotype, were not transformed (49), suggesting that constitutive activation of RelA alone was unable to transform murine fibroblast cells and may require a cooperating oncogene such as ras for tumorigenic transformation. A recent report showed that activation of RelA by oncogenic Ha-ras-induced signaling is required for cellular transformation (50). This report supports the possibility that the Ras and MAP kinase signal transduction pathway is involved in the constitutive activation of RelA in pancreatic tumor cells.

Possible explanations for the constitutive RelA activity in pancreatic cancer cells are: (a) IκBα is mutated and therefore cannot bind to RelA and mask the nuclear translocation signal in RelA; (b) mutations in RelA prohibit IκBα binding to RelA; or (c) the RelA upstream signal-transduction cascades are constitutively activated. Our data demonstrate that constitutive RelA activity in MDAPanc-28 cells can be inhibited by the nonspecific kinase inhibitor, curcumin. Curcumin inhibits TNF-α-induced activation of NFκB at a step before IκBα phosphorylation (41). Our data also show that expression of dominant-negative IκBα, Raf, and MEKK1 resulted in almost complete inhibition of constitutive RelA-activated transcription specifically through κB sites in MDAPanc-28 and Capan-1 cells (Fig. 4). These results suggest that constitutive RelA activity in pancreatic adenocarcinoma cells is induced through activation of upstream signal-transduction cascades for RelA. If constitutive RelA activity is caused by a mutation in either IκBα or RelA that prohibits their interaction, the constitutive RelA activity would not be inhibited by blocking the upstream signaling pathways using curcumin or expression of dominant-negative IκBα, Raf, and MEKK1. Additionally, these results suggest that MAP kinase signaling cascades are involved in the constitutive activation of RelA in pancreatic tumors, possibly involving mutated K-ras.

Ras initiates two divergent signaling cascades that activate distinct MAP kinases. Activation of ERKs by Ras is mediated via Raf-1 and MEK kinases, whereas JNK activation is mediated by another Ras-responsive protein kinase, MEKK (51, 52). Recent reports showed that Raf-1 is commonly used by multiple inducers that activate Rel/NF-κB (33) and that MEKK1 has been identified as one of the tightly associated subunits in IκB kinase complex and phosphorylates the IκBα kinase complex, which, in turn, phosphorylates IκBα proteins and activates Rel/NF-κB transcription factors (15, 16, 17, 18, 19, 20, 21). These results are consistent with the possibility that Ras and MAP kinases are involved in constitutive activation of RelA in pancreatic tumor cells. Our results suggest that point mutation in the ras oncogene might correlate with the constitutive RelA activity in pancreatic cancers. Activated RelA was detected in 65% (13 of 20) of the pancreatic adenocarcinoma tissues that were shown previously to carry a K-ras mutation at codon 12 (39, 40), and 89% (8 of 9) human pancreatic tumor cell lines that carry the same mutated K-ras gene have constitutive RelA activity. However, despite this correlation, of the 11 human pancreatic tumor cell lines that we studied, both BxPc3 and HS766T cells express a wild-type K-ras gene. For BxPc3 cells, RelA is constitutively activated (Fig. 3, A and B), and for HS766T cells, RelA activity is inducible (Fig. 3 B). MDAPanc-3 cells, which carry a mutated K-ras gene, also has inducible RelA activity (data not shown). It is possible that signal transducers upstream or downstream of K-ras are constitutively activated in BxPc3 cells, and additional genetic alterations might be needed to activate RelA in HS766T and MDAPanc-3 cells. The mechanisms for constitutive activation of RelA in pancreatic cancer remain unknown. Determination of the activities of IκBα kinases, Ras, and MAP kinases from these pancreatic tumor cell lines and expression of a transfected mutated K-ras gene, raf-1, MEKK1, and IKKs into nontumorigenic SGH cells will help to elucidate the signal-transduction pathways leading to the constitutive activation of RelA in pancreatic adenocarcinomas.

We are grateful to Drs, Inder M. Verma and Dan Van Antwerp at the Salk Institute for IκBαM expression plasmid; to Dr. Bin Su in the Department of Immunology at University of Texas M. D. Anderson Cancer Center for RafDN and MEKK1DN; to Dr. Marsha L. Frazier in the Department of Gastrointestinal Oncology at University of Texas M. D. Anderson Cancer Center for providing MDAPanc-3, MDAPanc-28, and MDAPanc-48 cell lines; and to Drs. Terry Lawson and Coral Kolar at Eppley Institute for Cancer Research, the University of Nebraska, for providing SGH pancreatic cell lines. We thank Di Shen for technical assistance, members of the Chiao laboratory for helpful discussions, and Nancy G. Arora and Pat Thomas for editorial assistance.