Keratinocyte Growth Factor/Fibroblast Growth Factor-7-regulated Cell Migration and Invasion through Activation of NF-κB Transcription Factors*

Keratinocyte growth factor (KGF)/fibroblast growth factor-7 (FGF-7) is a paracrine- and epithelium-specific growth factor produced by cells of mesenchymal origin. It acts exclusively through FGF-7 receptor (FGFR2/IIIb), which is expressed predominantly by epithelial cells, but not by fibroblasts, suggesting that it might function as a paracrine mediator of mesenchymal-epithelial interactions. KGF/FGF-7 plays an essential role in the growth of epithelial cells and is frequently overexpressed in cancers of epithelial origin such as pancreatic cancer, switching paracrine stimulation of KGF/FGF-7 to an autocrine loop. Less is known, however, about the signaling pathways by which KGF/FGF-7 regulates the response of epithelial cells. To delineate the signaling pathways activated by KGF/FGF-7 and examine cellular response to KGF/FGF-7 stimulation, we performed functional analysis of KGF/FGF-7 action. In this report, we show that KGF/FGF-7 activated nuclear factor κB (NF-κB), which in turn induced expression of VEGF, MMP-9, and urokinase-type plasminogen activator and increased migration and invasion of KGF/FGF-7-stimulated human pancreatic ductal epithelial cells. Expression of phosphorylation-defective IκBα (IκBαS32A,S36A), which blocked NF-κB activation, inhibited KGF/FGF-7-induced gene expression and cell migration and invasion. Our results demonstrate for the first time that KGF/FGF-7 induces NF-κB activation and that NF-κB plays an essential role in regulation of KGF/FGF-7-inducible gene expression and KGF/FGF-7-initiated cellular responses. Thus, these findings identify one signaling pathway for KGF/FGF-7-regulated cell migration and invasion and suggest that paracrine sources of KGF/FGF-7 are one of the malignancy-contributing factors from tumor stroma.

cially IL-1 are potent inducers of KGF expression, suggesting that infiltrating inflammatory cells such as monocytes and neutrophils produce inflammatory cytokines, including IL-1, which induce KGF/FGF-7 expression from local mesenchymal cells to promote epithelial proliferation (3,4). KGF/FGF-7 is one of the AP-1-regulated genes induced by the pro-inflammatory cytokines (14). However, KGF/FGF-7-mediated downstream signaling pathways are still unclear.
Nuclear factor B (NF-B) is a family of pleiotropic transcription factors that control the expression of numerous genes involved in growth, tumorigenesis, tumor metastasis, differentiation, embryonic development, apoptosis, and inflammation (15)(16)(17)(18). Interaction of c-Rel, RelA, and RelB with their inhibitors, the IBs, results in inactive complexes in the cytoplasm by masking the nuclear localization signal (19,20). In most cell types, NF-B proteins are sequestered in the cytoplasm by the inhibitor IB in an inactive form (16,19,20). On stimulation, IB is phosphorylated by IB kinase (IKK) and polyubiquitinated, which triggers its rapid degradation by proteasome (21)(22)(23). Consequently, NF-B proteins are released and translocated into the nucleus, where they activate the expression of target genes (15)(16)(17)24). One of the key target genes regulated by NF-B is its inhibitor IB␣. A feedback inhibition pathway for control of IB␣ gene transcription and down-regulation of transient activation of NF-B activity has been described previously (25)(26)(27).
Members of the NF-B family are involved in the development of cancer. Many tumors have acquired genetic alterations in the signaling pathways that regulate NF-B activation. In one study, for example, defective IB␣ led to constitutive nuclear NF-B activity, which in turn conferred a growth advantage to Hodgkin disease tumor cells (28). Elevated IKK activities also were found in some of the tumor cells, suggesting that IKK is activated by as yet unidentified aberrant upstream signaling cascades (28). We previously reported that RelA, the p65 subunit of the NF-B transcription factor, is constitutively activated in most pancreatic cancer tissues and human pancreatic cancer cell lines but not in normal pancreatic tissues or immortalized pancreatic ductal epithelial cells (29,30). Constitutive RelA activity plays a key role in pancreatic cancer metastasis and tumor progression (31,32). We recently showed that IL-1 autocrine stimulation is involved in constitutive NF-B activity in pancreatic cancer. Key features of pancreatic cancer are marked proliferation of stromal fibroblasts and deposition of extracellular matrix components such as matrix metalloproteinases (MMP) and collagens, suggesting that microenvironmental cellular interactions are important in the pathogenesis of this disease. The role of tumor-associated stromal fibroblasts in pancreatic cancer pathogenesis remains unclear.
In the study reported here, we show that NF-B activation and NF-B-regulated genes are induced in response to stimulation by epidermal growth factor (EGF) and KGF/FGF-7 in immortalized/nontumorigenic human pancreatic ductal epithelial (HPDE) cells (33) and that Ras, phosphatidylinositol 3-kinase, and Akt are involved in these pathways. Inhibition of NF-B activation by a phosphorylation-defective IB␣ (IB␣S32A,S36A) blocked KGF/FGF-7-induced gene expression and cell migration and invasion. Our findings suggest that NF-B plays an essential role in the regulation of KGF/FGF-7inducible gene expression and KGF/FGF-7-initiated cellular responses, and that paracrine stimulation of KGF/FGF-7 might be one of the factors contributing to growth of malignancies from tumor stroma.

EXPERIMENTAL PROCEDURES
Cell Lines and Reagents-Human papillomavirus type 16 early gene 6 and 7-immortalized/nontumorigenic HPDE cells have been described elsewhere (33). Mouse embryonic fibroblasts (MEFs) were grown in Dulbecco's modified Eagle's medium, which contained 4.5 g/liter glucose, glutamine, and nonessential amino acids, and was supplemented with 10% heat-inactivated fetal bovine serum. Human KGFR cDNA was obtained from Dr. T. Miki (NCI, National Institutes of Health). EGF and KGF were obtained from R&D Systems. HPDE cells were cultured in keratinocyte-SFM (serum-free medium, Invitrogen). Anti-phospho-IB␣ (Ser-32), anti-phospho-Akt, and anti-IB␣ antibodies were obtained from Cell Signaling Technology. Anti-Ras antibody was obtained from Calbiochem. Anti-Paxillin antibody was obtained from Neo-Markers, Inc. Anti-␤-actin antibody was obtained from Sigma. Anti-uPA antibody was obtained from American Diagnostica Inc. Anti-human MMP9 was obtained from Binding Site Inc., and anti-vascular endothelial growth factor (VEGF), anti-p65/NF-B, anti-Akt, and anti-MEKK3 antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA).
Growth Curve and Cell Cycle Analysis-HPDE cells were kept in keratinocyte-SFM without growth factors for 48 h and were stimulated or not stimulated with KGF/FGF-7 or EGF for various periods. Cells were counted every 24 h in triplicates, and cell cycle profiles of HPDE cells unstimulated or stimulated with KGF (100 ng/ml) or EGF (100 ng/ml) at different time intervals were analyzed by flow cytometry. Each experiment was performed independently at least three times with similar results each time. Results are expressed as the means Ϯ S.E. of three independent experiments. All of the statistical analyses were performed using StatView 5.0 (Abacus Concepts Inc., Berkeley, CA).
Nuclear Extract Preparation and Electrophoretic Mobility Shift Assay-HPDE cells were starved for 48 h, then stimulated with EGF (100 ng/ml) or KGF (100 ng/ml) for different time intervals or TNF␣ (10 ng/ml) for 30 min as control. Electrophoretic mobility shift assay (EMSA) was performed as previously described by Schmidt et al. (34). Nuclear extracts were prepared according to the method of Andrews and Faller (35). DNA binding assays for NF-B proteins were performed with nuclear extracts (10 g) as described by Wang et al. (29). 32 P-Labeled double-stranded oligonucleotides (5Ј-CTCAACAGA-GGGGACTTTCCGAGAGGCCAT-3Ј) containing the B site found in the HIV long terminal repeat were used as probes. The mutant B site for HIV long terminal repeat (5Ј-CTCAACAG-AGTTGACTTTTCGAGAGGCCAT-3Ј) was used for competition studies. The competition was performed with a 50-fold excess of unlabeled wild-type or mutant B oligonucleotides. The supershift experiments were performed with anti-RelA antibody (Santa Cruz Biotechnology). The reactions were ana-lyzed on 4% polyacrylamide gels containing 0.25ϫ Tris/borate/ EDTA buffer.
Western Blot Analysis-HPDE cells were starved for 48 h, then treated with EGF or KGF (100 ng/ml) for various time periods, or with TNF␣ (10 ng/ml) for 30 min as a positive control. Cell cytoplasmic extracts were used for detection of phosphor-IB␣, IB␣, Paxillin, and ␤-actin protein levels. Cell nuclear extracts were used to detect RelA/NF-B and Paxillin protein levels. For detection of uPA, MMP9, and VEGF levels in conditioned medium, the conditioned media from unstimulated and growth factor-stimulated HPDE cells were harvested at the same ratio of medium volume to cell number and subjected to dialysis for 24 h; dialyzed medium (6 ml) was then dried to a volume of 40 -80 l. The concentrated medium or 100 g of protein extracts was resolved by SDS-PAGE, transferred to nylon membranes (Immobilon-P, Millipore), and probed with antibodies. The subsequent Western blot analysis was carried out with Lumi-light Western blot substrate (Roche Applied Science).
Transfection and Luciferase Reporter Assays-HPDE cells and wild-type MEFs at 70% confluence in growth factor-free medium were transfected with either 1.0 g of B-Luc promoter-reporter construct alone or co-transfected with various expression vectors using FuGENE 6 transfection reagent (Roche Applied Science); pRL-CMV Renilla luciferase was cotransfected as a control reporter vector. After 24 -48 h, the cells were treated or not treated with EGF or KGF (100 ng/ml) for 1-4 h. At the end of treatment, reporter activity was determined using a dual-luciferase reporter gene assay according to the manufacturer's instructions (Promega, Madison, WI). The results are shown as the means Ϯ S.E. of three independent experiments.
Reverse Transcription-PCR-Total RNA was extracted using TRIzol reagent according to the protocol of the manufacturer (Invitrogen). The RNA was then subjected to reverse transcription into cDNA. The primers used for PCR amplification of the KGFR were 5Ј-CTCAAGCACTCGGGGATAAA-3Ј and 5Ј-CTGTTTTGGCAGGACAGTGA-3Ј; the 150-bp product corresponded to nucleotides 1352-1501 of human KGFR cDNA, which are located in the IgIIIb exon of human KGFR cDNA (29). The PCR conditions were as follows: 94°C for 5 min, then 30 cycles at 94°C for 30 s, 55°C for 30 s, and 72°C for 30 s, and finally extension at 72°C for 7 min. Glyceraldehyde-3-phosphate dehydrogenase was used as a positive control.
Cell Migration and Invasion Assay-Invasion assay was performed in 24-well plates by using a BD Biocoat growth factorreduced Matrigel invasion chamber (BD Biosciences) with an 8.0-m pore size PET membrane. Each membrane had a thin layer of GFR Matrigel Basement Membrane Matrix, which serves as a reconstituted basement membrane in vitro. The inserts were rehydrated by adding 0.5 ml of warm culture medium at 37°C to the inserts for 2 h. Invasion of KGF/FGF-7stimulated HPDE/IB␣M cells was determined, and HPDE/puromycin (HPDE/CTL) cells were used as the control. These HPDE cells, kept in SFM without growth factors for 24 h, were seeded (5 ϫ 10 4 cells in 0.5 ml of SFM) to the invasion chambers with or without KGF (100 ng/ml) and heparin (29.4 g/ml), and 750 l of complete keratinocyte-SFM was added to the lower well of the invasion plate. Cells were incubated at 37°C in a humidified atmosphere of 95% air and 5% CO 2 for 48 h. Noninvading cells from the interior of the inserts were removed by using cotton-tipped swabs. Invading cells on the bottom side of the membrane were fixed by 10% formaldehyde for 10 min, and then stained with 1% crystal violet, and at least three random fields per insert were counted under an Olympus microscope. A representative field of each experiment was photographed with 4ϫ lens and 2.5ϫ magnification. Results are expressed as the means Ϯ S.E. of three independent experiments.
Migration assay was performed in 24-well plates by using Falcon cell culture inserts, which have a PET membrane with 1 ϫ 10 5 8.0-m pores per cm 2 (BD Biosciences Labware). Before the assay was run, HPDE cells were starved for 24 h, and 700 l of complete keratinocyte-SFM was added to the lower well of the migration plate. The cells (5 ϫ 10 4 cells/0.3 ml) were seeded to the insert in keratinocyte-SFM with or without KGF (100 ng/ml) and heparin (29 g/ml). The incubation, staining, counting, and photographing procedures were the same as for the invasion assay. The representative fields of each experiment are shown. Results are expressed as the means Ϯ S.E. of three independent experiments.

KGF/FGF-7 Induces Proliferation, Invasion, and Migration in
Immortalized HPDE Cells-KGF/FGF-7 is a potent mitogenic factor for epithelial cells derived from a variety of tissues and plays an important role in tissue repair (8,9). To determine whether KGF/FGF-7 has any effects on immortalized/nontumorigenic HPDE cells, we stimulated HPDE cells with KGF/ FGF-7 using EGF-treated cells as a positive control. Our results show that KGF/FGF-7 induced the HPDE cells to proliferate as determined by cell counting and flow cytometry analysis for cell cycle profile (Fig. 1, A and B). Cell migration and invasion assays were performed as described in "Experimental Procedures." The representative fields of cell migration and invasion experiment are shown ( KGF/FGF-7 Induces NF-B Activation in Immortalized HPDE Cells-To test whether KGF/FGF-7-mediated pathways could activate NF-B, HPDE cells were kept in growth factorfree medium for 48 h, and then stimulated with KGF/FGF-7 (100 ng/ml) or EGF (100 ng/ml, as a positive control) at various time intervals. Our results show that both KGF/FGF-7 and EGF induced activation of NF-B, and a comparable amount of nuclear extracts was used as determined by Oct-1 EMSA for loading controls ( Fig. 2A). KGF/FGF-7-induced NF-B activation was time-dependent, and the activity peaked at 2 h ( Fig.  2A). As shown in Fig. 2B, to confirm the specificity of the KGF/ FGF-7-and EGF-induced B DNA binding activity, competition and supershift assays were performed by using the same nuclear extracts for NF-B EMSA in Fig. 2A. The unlabeled oligonucleotides containing wild-type B binding sites completely blocked NF-B DNA binding activity, whereas the unlabeled oligonucleotides with mutant B binding sites had no effect on NF-B DNA binding activity (Fig. 2B). In supershift experiments, the mobility of the NF-B binding activities was further retarded after incubation with p65 antibody as indi-cated by an arrow (Fig. 2B). These results suggest that the NF-B DNA binding activity was specific and that one of the subunits of the DNA binding complex was p65 (RelA). As shown in Fig. 2C, IB␣ phosphorylation and degradation were induced by KGF/FGF-7 and EGF in a timedependent manner as determined by Western blot analysis using the cytoplasmic extracts, isolated for the experiments described in Fig.  2A, with anti-phospho-IB␣ and anti-IB␣ antibodies. Because IB␣ phosphorylation by IKK, polyubiquitination, and proteasome-mediated degradation are the necessary steps for NF-B activation (16), these results further indicate that KGF/FGF-7 and EGF induced NF-B activation. Interestingly, the degradation patterns of IB␣ varied slightly between EGF and KGF/ FGF-7 stimulation (Fig. 2C, lanes 4  and 7). Because IB␣ is one of the genes transcriptionally regulated by NF-B, the expression levels of IB␣ may oscillate depending on the nature of the stimulation (25,26). To further demonstrate that KGF/FGF-7 induces NF-B activation as observed in NF-B EMSA and IB␣ immunoblots, the levels of nuclear RelA/NF-B were determined in the nuclear extracts from the HPDE cells stimulated with EGF and KGF/FGF-7 at various time intervals. The results show that the levels of RelA/NF-B were increased in the nuclear fractions in the EGF and KGF/FGF-7 time course stimulation (Fig. 2C). The relative protein loading was shown by the use of anti-␤-actin antibody, and the quality of the nuclear extracts was determined by probing both cytoplasmic and nuclear extracts with an antibody against Paxillin, a cytoplasmic protein. As shown in Fig. 2C, the findings show that NF-B activation was induced by KGF/FGF-7. To determine the expression of NF-B-regulated genes that are induced by KGF/ FGF-7, the NF-B reporter gene assays were performed in HPDE cells. The results showed that KGF/FGF-7 induced NF-B-dependent promoter activity similar to that induced by EGF and TNF␣, further indicating that
KGF/FGF-7 Induces Expression of NF-B Downstream Target Genes-To determine whether expression of NF-B downstream target genes was induced by KGF/FGF-7, levels of uPA, MMP9, and VEGF expression in HPDE cells were examined (Fig. 3). The HPDE cells were kept in growth factor-free medium for 48 h, and then stimulated with KGF/FGF-7 (50 ng/ml) for various time intervals (Fig. 3). The conditioned media from these cells were harvested at the same volume/cell number ratios at each time point and were subjected to dialysis and concentration prior to Western blot analysis. The loading control for the concentrated conditioned media was shown by the use of Coomassie Blue staining of the identical gels. The results show that KGF/FGF-7 induced the expression of NF-B downstream target genes uPA, MMP9, and VEGF and that these inductions were dose-and time-dependent (Fig. 3A). However, the patterns of uPA, MMP9, and VEGF expression induced by KGF/FGF-7 were different. At 24-and 48-h time points, only uPA expression was increased without KGF/FGF-7 stimulation and possibly by other stimuli in the media.
To further demonstrate the expression of uPA, MMP9 and VEGF were induced by KGF/FGF-7 through NF-B activation, time-and dose-dependent phosphorylation, and degradation of IB␣, and the levels of nuclear RelA/NF-B were determined in cytoplasmic and nuclear extracts by immunoblotting. As shown in Fig. 3B, IB␣ phosphorylation and degradation were induced by KGF/FGF-7 in a time-and dose-dependent manner as determined by Western blot analysis using anti-phospho-IB␣ and anti-IB␣ antibodies. At 24-and 48-h time points, both IB␣ phosphorylation and degradation appeared to be rather constant in the cells with and without KGF/FGF-7 stimulation, suggesting the IB␣ re-synthesis and phosphorylation/ degradation through the IB␣ autoregulation loop reached the steady state (25,26). The levels of nuclear RelA/NF-B were increased in the nuclear fractions in the KGF/FGF-7 time course stimulation (Fig. 3B). The quality of the nuclear extracts was determined by probing both cytoplasmic and nuclear extracts with an antibody against Paxillin, and relative protein loading was shown by the use of anti-␤-actin antibody as shown in Fig. 3B. Although the increased level of nuclear RelA/NF-B in the KGF/FGF-7-stimulated cells was consistent with IB␣ phosphorylation and degradation, it should be mentioned that FIGURE 2. EGF and KGF activate NF-B. A, HPDE cells were cultured in keratinocyte-SFM without growth factors for 48 h, and then were stimulated with 100 ng/ml KGF or EGF for the indicated time and fractionated into cytoplasmic and nuclear extracts. Nuclear extracts (15 g) were used in this analysis with an HIV B probe. An Oct-1 probe was used as a loading control for quality and quantity of cell nuclear extracts. B, competition and supershift assays were performed using 15 g of the same nuclear extracts from KGF-or EGFstimulated cells for determining the specificity of inducible RelA/p50 NF-B DNA binding activity. The supershifted NF-B band was indicated by an arrow. C, for Western blot analysis for IB␣ phosphorylation and degradation, cytoplasmic protein extracted as described for A was probed with anti-phosphorylated IB␣ (p-IB␣), anti-IB␣, and anti-Paxillin antibodies. For nuclear p65/NF-B immunoblots, the same nuclear extracts used in NF-B EMSA in A were probed with an anti-p65/NF-B and anti-Paxillin antibodies. Relative protein loading was shown by using anti-␤-actin antibody. D, the B reporter gene assay was performed with wild-type and mutant B-luciferase reporter gene constructs as well as TK Renilla for control. After transfection with the reporter gene constructs, HPDE cells were cultured in growth factorfree medium for 48 h and then stimulated with KGF, EGF, or TNF␣ for 2 h, and the dual luciferase assay was performed.
the nuclear RelA/NF-B detected in Western blot reflected the total RelA/NF-B proteins, which include those from an inactive complex with IB␣ unable to bind to B enhancer in the nucleus, as part of the feedback regulation mechanism, and those that are capable of binding to B enhancer as determined by NF-B EMSA.
KGF/FGF-7-induced NF-B Activation Is Inhibited by Mouse IB␣ (S32A,S36A) Phosphorylation Mutant IB␣M-To determine the role of NF-B in KGF/FGF-7-mediated signaling pathways, we used retroviral infection to generate stable clones (HPDE/IB␣M) that expressed IB␣M (IB␣S32A,S36A) with mutated PEST domain for increase of its stability (36). As shown in Fig. 4A, the expression of IB␣M appeared as a faster migration band, which was largely due to 3 amino acid residues shorter in length, and the lack of phosphorylation on PEST region in this mutated mouse IB␣ was confirmed in pooled puromycin-resistant HPDE cell clones by Western blot using anti-IB␣ antibody. In the presence of IB␣M in HPDE cells, the endogenous IB␣ protein was reduced to a minimum level (Fig. 4A), and the same finding in pancreatic cancer cell line was reported (31). The reduction of endogenous IB␣ protein suggest the inhibition of NF-B-mediated basal transcription of IB␣ by IB␣M and the faster turnover rate for endogenous IB␣. TNF␣-and KGF/FGF-7-induced NF-B activation was inhibited completely by IB␣M as determined by EMSA (Fig. 4,  B and C). To further demonstrate that TNF␣and KGF/FGF-7-induced NF-B activation is suppressed by IB␣M, inhibition of TNF␣and KGF/FGF-7-regulated IB␣ phosphorylation and degradation and level of nuclear RelA/NF-B were determined in HPDE/CTL (infected with a retroviral vector only expressing a puromycin-resistant gene) and HPDE/ IB␣M cells by Western blot analysis (Fig. 4, B and C). The endogenous IB␣ level was very low, and phosphorylation of endogenous IB␣ induced by KGF/FGF-7, a much weaker NF-B inducer as compared with TNF-␣, cannot be readily detected in HPDE/IB␣M cells, but endogenous IB␣ degradation was detected and IB␣M was not degraded (Fig. 4C). The quality of the nuclear extracts and relative protein loading were determined by Western blot analysis described above. Taken together these data demonstrate that the TNF␣and KGF/ FGF-7-induced IB␣ phosphorylation and degradation and increase of nuclear RelA/NF-B protein were inhibited by expression of IB␣M.

KGF/FGF-7-induced Expression of NF-B-regulated Genes Is Inhibited by Mouse IB␣ (S32A,S36A) Phosphorylation Mutant
IB␣M-To determine whether KGF/FGF-7-inducible expression of NF-B downstream target genes was inhibited by IB␣M, levels of uPA, MMP9, and VEGF in the conditioned media from HPDE/CTL and HPDE/IB␣M cells were examined as already described above. The loading control for the concentrated conditioned media was shown by the use of Coomassie Blue staining of the identical gels (Fig. 5A). The results of Western blot analysis show that KGF/FGF-7-induced expression of the NF-B downstream target genes uPA and MMP9 was inhibited by IB␣M, further suggesting that NF-B plays an essential role in regulation of KGF/FGF-7-inducible gene expression (Fig. 5A). The presence of a low level of uPA in HPDE/IB␣M cells at 12 h with KGF/FGF-7 stimulation might be due to the KGF/FGF-7-induced activation of NF-B from degradation of endogenous IB␣. The disappearance of uPA in HPDE/IB␣M cells at 24 and 48 h of KGF/FGF-7 stimulation suggest that the residual NF-B activity induced from endogenous IB␣ complex was suppressed by IB␣M expressed at higher level with greater stability. Interestingly, the KGF/FGF-7 induced expression of VEGF is not inhibited by IB␣M, suggesting VEGF may be regulated by other transcription factors mediated by KGF/FGF-7.

KGF/FGF-7 Induces NF-B Activation
To further demonstrate KGF/FGF-7-induced expression of uPA, MMP9 and VEGF were suppressed by IB␣M-mediated inhibition of NF-B activation, time-dependent phosphorylation, and degradation of IB␣, and increased nuclear RelA/ NF-B protein levels were determined in cytoplasmic and nuclear extracts by immunoblotting. As shown in Fig. 5B, IB␣ phosphorylation and degradation were induced by KGF/FGF-7 in a time-dependent manner as determined by Western blot

. Phosphorylation-defective IB␣ mutant (IB␣M) inhibited KGF-induced NF-B activation.
A, expression of IB␣M in IB␣M retrovirusinfected HPDE cells was determined by Western blot analysis using anti-IB␣ antibody. Relative protein loading was shown by using anti-␤-actin antibody. B and C, EMSA were performed to examine TNF␣-and KGF-induced NF-B DNA binding activity in HPDE/CTL and HPDE/IB␣M cells as described. The HPDE/CTL and HPDE/IB␣M cells were cultured in keratinocyte-SFM without growth factors for 48 h, then stimulated with TNF␣ for 30 min or with 100 ng/ml KGF for the indicated times; the cells were then fractionated into cytoplasmic and nuclear extracts. Nuclear extracts (15 g) were used in this analysis with an HIV B probe. The cytoplasmic extracts (50 g) were subjected to Western blot using anti-phosphorylated IB␣ (p-IB␣), anti-IB␣, and Paxillin antibodies, and the nuclear extracts were probed with an anti-p65/NF-B and anti-Paxillin antibodies. Relative protein loading is shown by using anti-␤actin antibody.  MARCH 2, 2007 • VOLUME 282 • NUMBER 9 analysis using anti-phospho-IB␣ and anti-IB␣ antibodies. In the KGF/FGF-7 time-course stimulation, the levels of nuclear RelA/NF-B were increased in the nuclear fractions of HPDE/ CTL cells but not in HPDE/IB␣M cells (Fig. 5B). The nuclear extracts showed very little contaminants with cytoplasmic proteins as determined by the level of Paxillin, and protein loadings in each lane were comparable as determined by the use of anti-␤-actin antibody (Fig. 5B). These results suggest that IB␣M inhibited the KGF/FGF-7-induced expression of the NF-Bregulated downstream target genes uPA and MMP9.

KGF/FGF-7-induced Cell Migration and Invasion Are Suppressed by IB␣M-mediated Inhibition of NF-B-To
demonstrate whether NF-B activity plays a key role in KGF/ FGF-7-induced cellular responses, cell migration and invasion assays were performed as described under "Experimental Procedures" using HPDE/IB␣M cells and HPDE/CTL cells as a control. The representative fields of the cell migration experiment are shown (Fig. 6A) as are the representative fields of the cell invasion experiment (Fig.  6C). Results are expressed as the means Ϯ S.E. of three independent experiments (Fig. 6, B and D). The results in Fig. 6 showed that KGF/ FGF-7 significantly enhanced HPDE/CTL cell migration and invasion, whereas its effects on migration and invasion of HPDE/IB␣M cells were suppressed by IB␣Mmediated inhibition of NF-B activity. Taken together, these results suggest that NF-B plays an essential role in regulation of KGF/FGF-7-inducible gene expression and KGF/FGF-7-mediated cell migration and invasion.
KGF Receptor-mediated Signaling Cascades Induce NF-B Activation-To determine whether KGFR, a splicing variant of FGFR-2, FGFR2-IIIb, relays KGF/FGF-7 signals for induction of NF-B activation, and to generate a specific cell line for identifying the signaling components in KGF/FGF-7-induced NF-B activation, an expression vector for a human KGFR was constructed. Wild-type MEFs cells, which do not express mouse KGFR, were transfected with this expression vector to obtain stable clones. As shown in Fig. 7A, human KGFR transcript was detected by reverse transcription-PCR in these G418resistant clones after transfection of KGFR expression plasmid, indicating that human KGFR was expressed. NF-B was activated by KGF/FGF-7 in the MEFs/KGFR but not in the parental MEFs (Fig. 7B, lanes 2 and 5). A low level of NF-B activity was detected in MEFs/KGFR without exogenous KGF/FGF-7, suggesting that expression of human KGFR in MEFs partially connected the KGF/FGF-7 autocrine stimulation loop (Fig. 7B, lane 4). The expression of KGF/FGF-7 in MEFs cells was detected by Western blot (data not shown). EGF-induced NF-B activation served as a positive control (Fig.  7B). NF-B reporter gene assays showed that KGF/FGF-7-induced NF-B activity was dependent on expression of KGFR (Fig. 7C). Furthermore, the high level of reporter gene activity in the cell transfected with KGFR without exogenous KGF/ FGF-7 stimulation is consistent with the results from the NF-B EMSA in Fig. 7B, suggesting KGF/FGF-7 that autocrine stimulation is at work. In Fig. 7 (D and E), both KGF/FGF-7and EGF-induced NF-B reporter gene activity was observed in MEFs/KGFR cells only transfected with NF-B-luciferase reporter gene. In the MEFs/KGFR cells co-transfected with NF-B-luciferase reporter gene and phosphorylation-defective
Activation of Akt Plays a Key Role in KGF/FGF-7-induced NF-B Activation-The reports by Romashkova et al. and Ozes et al. (37,38) show that NF-B has a role in growth factor signaling and define an anti-apoptotic Ras/phosphatidylinositol 3-kinase/Akt/IKK/NF-B pathway (37,38). To identify the signaling molecules involved in growth factor-mediated activation of IKK and NF-B, B-luciferase reporter gene assays were performed using MEFs/KGFR co-transfected with various Akt mutants. The results show that a myristoylated active form of Akt (myr-Akt) greatly enhanced KGF/FGF-7-and EGF-induced NF-B reporter gene activity, and constitutively activated Akt mutant (AktDD) induced NF-B reporter gene activity without exogenous growth factors, whereas activation-defective (AktAA) and kinase-dead (AktK179M) AKT mutants inhibited KGF/FGF-7-and EGF-induced NF-B reporter gene activity (Fig. 8A). The expression levels of various transfected Akt mutants are comparable as determined by anti-Akt immunoblot using the same extracts for the NF-B reporter gene assay (Fig. 8A). Although kinase-dead MEKK3 (MEKK3KM) only reduced KGF/FGF-7-and EGF-induced NF-B activation by ϳ3-fold, Ras (RsN17) mutant, Akt (Akt-K179M) mutants, and LY294002, a specific inhibitor of phosphatidylinositol 3-kinase, the upstream kinase of Akt, completely inhibited KGF/FGF-7-and EGF-induced NF-B reporter gene activity (Fig. 8B). The expression levels of transfected kinase-dead MEKK3KM, RsN17, and Akt-K179M mutants were very similar as determined by the immunoblot using the same extracts for the NF-B reporter gene assay (Fig.  8B). As shown in Fig. 8C, phosphorylation of Akt and KGF/ FGF-7-and EGF-induced NF-B DNA binding activity also were inhibited by LY294002 in HPDE cells. Taken together, these results show that the signaling pathways mediated by KGF/FGF-7 involve activation of Akt and NF-B.

DISCUSSION
We have demonstrated for the first time that KGF/FGF-7, a paracrine and epithelium-specific growth factor produced by cells of mesenchymal origin, is an inducer of NF-B activation and that NF-B plays an essential role in regulation of KGF/ FGF-7-inducible gene expression. In this report, we provide crucial data supporting this novel function of KGF/FGF-7, as follows: 1) KGF/FGF-7 stimulated growth, migration, and invasion of immortalized HPDE cells (Fig. 1); 2) KGF/FGF-7 induced NF-B activation as demonstrated by EMSA (Fig. 2); 3) KGF/FGF-7 induced expression of NF-B downstream target genes (Fig. 3)  A, expression of KGFR and glyceraldehyde-3-phosphate dehydrogenase (GAPDH, control) in MEFs was determined by reverse transcription-PCR after FGFR2/IIIb transfection. B, KGF-induced NF-B DNA binding activity in MEFs and MEFs/KGFR was determined by EMSA as described. MEFs and MEFs/KGFR were cultured in Dulbecco's modified Eagle's medium supplemented with 10% heatinactivated fetal calf serum for 48 h and then fractionated into cytoplasmic and nuclear extracts. Nuclear extracts (10 g) were used in this analysis with an HIV B probe. C-E, dual luciferase assays were performed to determine the activation of NF-B reporter gene. C, the B luciferase and TK Renilla reporter genes were co-transfected into MEFs with KGFR; the cells were cultured in serum-free medium for 48 h and then stimulated or not stimulated with KGF (100 ng/ml) for 2 h. D and E, the B luciferase and TK Renilla reporter genes were co-transfected into MEFs with phosphorylation-defective IB␣ mutant and kinase-dead IKK1/␣ and IKK2/␤. These transfectants were cultured in SFM for 48 h and then stimulated or not stimulated with EGF (100 ng/ml) or KGF (100 ng/ml) for 2 h, and protein extracts were isolated for dual luciferase assay. (Fig. 6); 6) KGF/FGF-7 receptor-mediated signaling cascades induced NF-B activation (Fig. 7); and 7) activation of Akt played a key role in KGF/FGF-7-induced NF-B activation (Fig.  8). These results suggest that NF-B plays an essential role in regulation of KGF/FGF-7 signaling cascades and KGF/FGF-7induced cellular responses.
The tumor microenvironment is composed of fibroblasts, various inflammatory cells, and vascular structures. Interaction between tumor cells and their microenvironment is a highly dynamic process, with presumed effects on tumor progression. The interplay is based on three basic processes: production of autocrine and paracrine factors, cell-matrix contact, and signaling by direct cell-cell interactions (39,40). To establish cell culture models for the study of the tumorigenesis of human pancreatic cancer in vitro, we have developed immortalized cultures of HPDE cells derived from normal epithelium (41). The aim of these studies was to investigate the immortalization-transformation sequence of these cells to understand better the implication of various mutations, including the role of paracrine factors such as KGF/FGF-7, a potent mitogenic paracrine mediator of epithelial cells, in the development of pancreatic cancer. We investigated the transforming potential of KGF on pancreatic epithelial cells, because one of the key features of pancreatic epithelial cancer is a marked proliferation of stromal fibroblasts. We tested exogenous effects of KGF/FGF-7 on human pancreatic epithelial cells by mimicking paracrine or autocrine KGF/FGF-7 stimulation. KGF/FGF-7 stimulated HPDE cells and enhanced cell growth, migration, and invasion, suggesting the transforming potential of KGF/FGF-7 on human pancreatic epithelial cells.
We previously showed that NF-B is constitutively activated in most human pancreatic cancer tissues and cell lines but not in normal pancreatic tissues and immortalized pancreatic ductal epithelial cells (29,42). A number of recent studies have shown that phosphorylation mutant IB␣ (S32A,S36A) (IB␣M)-mediated inhibition of constitutive NF-B activity in human pancreatic cancer cells suppresses tumorigenesis and liver metastasis in an orthotopic nude mouse model, suggesting that constitutive NF-B activation plays an important role in pancreatic tumor progression and metastasis (31,32). We previously found that an IL-1 autocrine mechanism accounts for the constitutive activation of NF-B in metastatic human pancreatic cancer cell lines (42). Our results also demonstrate that regulation of IL-1 expression is primarily dependent on growth FIGURE 8. Akt plays a key role in regulation of KGF/FGF-7-induced NF-B activation. A, the B luciferase reporter genes and TK Renilla were transfected into MEFs/KGFR with control expression vector or with various AKT mutants (Myrkt, AktDD, AktAA, and Akt-K179M) as indicated. These MEFs were cultured in SFM for 48 h and then stimulated with either 100 ng/ml EGF or 100 ng/ml KGF/FGF-7 for 2 h, and protein extracts were isolated from these cells for dual luciferase assays and Western blots to determine the expression of Akt mutants and ␤-actin as indicated. B, MEFs/KGFR cells were transfected with control expression vector, kinase-dead MEKK3 and Akt, or a mutant Ras (RsN17), cultured in Dulbecco's modified Eagle's medium without growth factors for 48 h and then either stimulated with EGF (100 ng/ml) or KGF (100 ng/ml) for 2 h or treated or not treated with LY294002 (30 M) for 2 h. Dual luciferase assay and Western blots for determining the expression of Akt, MEKK3, Ras mutants, and B-actin were performed as indicated. C, HPDE cells were starved for 48 h, and then treated with 30 M LY294002 for 30 min. Then the cells were treated with 100 ng/ml EGF or KGF/FGF-7 for 1 h. The nuclear and cytoplasmic extracts were isolated; nuclear extracts were used to determine NF-B activities by EMSA and cytoplasmic extracts to determine phosphorylation of Akt by Western blot analysis using an anti-phospho-Akt antibody.

KGF/FGF-7 Induces NF-B Activation
factor-regulated AP-1 activity (43). These findings suggest a possible mechanism by which constitutive activation of NF-B in metastatic human pancreatic cancer cells is initially induced by KGF/FGF-7 and further enhanced by IL-1 autocrine stimulation, because IL-1 is one of the downstream target genes regulated by NF-B. Interestingly, Chedid et al. (3) reported that the pro-inflammatory cytokine IL-1 strongly induces the expression of KGF/FGF-7 in fibroblasts from multiple sources (3). Furthermore, it has been shown that, in a cell co-culture system, expression of KGF/FGF-7 is strongly enhanced in fibroblasts, and expression of IL-1 is greater in co-cultured keratinocytes than in monocultures (44). Thus it was postulated that IL-1, which had no effect on keratinocyte proliferation, induces the expression of growth factors to stimulate keratinocyte proliferation, such as KGF/FGF-7 in fibroblasts. This is consistent with our finding that KGF/FGF-7 produced from fibroblasts induced NF-B activation and stimulated the expression of NF-B-regulated genes such as IL-1, which in turn induced expression of KGF/FGF-7 in fibroblasts, resulting in a double paracrine stimulation loop for a dynamic and reciprocal modulation of cytokine and growth factor production in epithelial mesenchymal interactions. The mutually induced signaling circuits for growth regulation may have in vivo functional significance, because marked proliferation of stromal fibroblasts is one of the key features of pancreatic cancer. Furthermore, the significance of this reciprocal modulation of the growth of epithelial cells is regulated by a double paracrine mechanism through release of pro-inflammatory cytokine IL-1 from epithelial cells, which elicit enhanced expression of growth factors, particularly KGF/FGF-7, in fibroblasts; thus IL-1, in addition to its pro-inflammatory function, may play an essential role in regulating fibroblast proliferation.