Induction of the Angiogenic Modulator Fibroblast Growth Factor-binding Protein by Epidermal Growth Factor Is Mediated through Both MEK/ERK and p38 Signal Transduction Pathways*

Fibroblast growth factor-binding protein (FGF-BP) is a secreted protein that binds and activates fibroblast growth factors (FGF-1 and FGF-2) and induces angio-genesis in some human cancers. FGF-BP is expressed at high levels in squamous cell carcinoma (SCC) cell lines and tumor samples and has been shown to be rate-lim-iting for the growth of SCC tumors in vivo . In this study, we examine the regulation of FGF-BP by epidermal growth factor (EGF) and the signal transduction mechanisms that mediate this effect. We found that EGF treatment of the ME-180 SCC cell line caused a rapid induction of FGF-BP gene expression. This induction was mediated transcriptionally through the AP-1 (c-Fos/ JunD) and CCAAT/enhancer-binding protein elements as well as through an E-box repressor site in the proximal regulatory region of the FGF-BP promoter. Pharmacological inhibition of protein kinase C and mitogen-activated protein kinase/extracellular signal-regulated kinase kinase 1/2 (MEK1/2) completely blocked EGF induction of FGF-BP mRNA, whereas inhibition of phos-phatidylinositol 3-kinase had no effect. Additionally, both EGF- and anisomycin-induced FGF-BP mRNA was abrogated by inhibition of p38 mitogen-activated protein kinase, demonstrating a role for p38 in the regulation of FGF-BP. Co-transfection

A pivotal process in a healing wound as well as in a growing tumor is the development of new blood vessels, or angiogenesis. Some of the most potent angiogenic stimulators are the fibroblast growth factors, including the classical angiogenic activators of this family, FGF-1 1 and FGF-2 (1). High concentrations of biologically active FGF-1 and FGF-2 are found in extracts of normal human tissues that are not necessarily undergoing active new blood vessel growth (1). This is due to the storage of FGF-1 and FGF-2 in the extracellular matrix, where they are found tightly bound to membrane-attached heparan sulfate proteoglycans (2), which quenches their biological activity. One mechanism by which FGF-1 and FGF-2 are released from the extracellular matrix is through the secretion of the carrier protein FGF-BP that binds to FGF-1 and FGF-2 in a noncovalent and reversible manner (3). FGF-BP is actively secreted from cells and, once bound, prevents degradation of FGF-1 and FGF-2 (3,4). The importance of FGF-BP secretion in promoting tumor growth was shown in studies using the nontumorigenic adrenal carcinoma cell line SW-13, which has high expression of FGF-2 but is negative for FGF-BP. Stable overexpression of FGF-BP in SW-13 cells led to a dramatic increase in FGF-2dependent colony formation and formation of highly vascularized tumors in nude mice (4). Additional studies were carried out to characterize the biological role of FGF-BP in highly tumorigenic cell lines such as ME-180 (human cervical squamous cell carcinoma) and LS174T (colon adenocarcinoma), which express high levels of endogenous FGF-BP (4,5). Reduction of FGF-BP mRNA levels in these cell lines using ribozyme targeting significantly inhibited tumor development and angiogenesis (6). These studies demonstrated that FGF-BP serves as a rate-limiting angiogenic modulator for some tumor types (6,7).
FGF-BP is highly expressed in squamous cell carcinoma (SCC) cell lines from lung, bladder, skin, and cervix and is positive in primary SCC tumor samples (4). Furthermore, its expression has been shown to be up-regulated during mouse embryonic development of the skin, lung, and intestine and is low in most adult tissues (8). A potential role for FGF-BP during skin carcinogenesis was described in studies showing dramatic FGF-BP up-regulation in human 2 and mouse skin treated topically with DMBA and TPA (8). These observations suggested several mechanisms that might be involved in the direct regulation of the FGF-BP gene. First, DMBA treatment has been shown to cause a specific point mutation in the ras oncogene (9), suggesting that the Ras signal transduction pathway might regulate FGF-BP expression. This is also indicated by the observation that FGF-BP is induced in ras-transformed keratinocytes (8). Second, the role of TPA as a direct regulator of FGF-BP gene expression was confirmed upon TPA treatment of several SCC cell lines, including ME-180, which caused rapid transcriptional induction of the FGF-BP gene (10). We further identified that Sp1, AP-1, and C/EBP sites within the proximal FGF-BP promoter are all required for TPA regulation of FGF-BP (10).
SCC cell lines and tumors, including the ME-180 cell line, typically express high levels of EGF receptors (EGFRs) (11,12), and overexpression of EGFR in SCC has been shown to confer greater tumorigenicity (13), which led us to investigate the possible role and mechanisms of EGF regulation of FGF-BP. EGF signaling occurs predominantly through binding to its receptor EGFR (HER1) and its dimerization partner ErbB2 (HER2/Neu). Autophosphorylation of activated EGF receptors stimulates a number of signal transduction pathways, including the classical Ras/Raf/MAP kinase kinase (MEK)/MAP kinase (ERK) pathway, which is known to phosphorylate and activate AP-1 (14). MEK/ERK activation occurs either through phosphorylated EGFR recruitment of the Shc-Grb2-SOS complex and subsequent Ras activation or through recruitment of phospholipase C␥ and subsequent PKC activation. PKC can in turn modulate Raf through both Ras-dependent and -independent mechanisms (15,16). Other signaling pathways induced by EGF include the stress-activated protein kinases such as JNK and p38 (17), the PI 3-kinase pathway (18), and the Janus kinase/signal transducers and activators of transcription pathway (19).
Here we show a possible link between EGF signaling and angiogenic activation through the regulation of the FGF-BP gene in SCC. We found that EGF treatment induces rapid up-regulation of FGF-BP transcription occurring through the AP-1 and C/EBP sites in the FGF-BP promoter. Furthermore, inhibition of either the MEK/ERK pathway or the p38 pathway abrogates induction by EGF, implicating dual activation of these MAP kinases as an important step in FGF-BP regulation.

EXPERIMENTAL PROCEDURES
Cell Culture and Reagents-The ME-180 cervical squamous cell carcinoma cell line was obtained from American Type Culture Collection (ATCC; Manassas, VA). Cells were cultured in improved minimum essential medium (IMEM) (Biofluids Inc., Rockville, MD) with 10% fetal bovine serum (Life Technologies, Inc.). Actinomycin D, calphostin C, and wortmannin were purchased from Sigma. Anisomycin and tyrphostin AG1478 were from Alexis Corp. SB203580, SB202190, and SB202474 were from Calbiochem, and U0126 was purchased from RBI (Natick, MA). All compounds were dissolved in Me 2 SO.
Northern Analysis-ME-180 cells were grown to 80% confluence in 10-cm dishes, washed twice in serum-free IMEM, and incubated for 16 h in serum-free IMEM prior to treatment. Cells were pretreated for 1 h with the indicated drug or with vehicle alone (Me 2 SO; final concentration 0.1%). EGF or anisomycin treatment was for 6 h unless otherwise indicated. Total RNA was isolated with RNA STAT-60 TM (Tel-Test Inc.), and Northern analysis was carried out as described previously (10). Hybridization probes were prepared by random-primed DNA labeling (Amersham Pharmacia Biotech) of purified insert fragments from human FGF-BP (4) and human GAPDH (CLONTECH). Quantitation of mRNA levels was performed using a PhosphorImager (Molecular Dynamics, Inc.).
Transient Transfections and Reporter Gene Assays-24 h before transfection, ME-180 cells were plated in six-well plates at a density of 750,000 cells/well. pRL-CMV Renilla luciferase reporter vector (Promega) was included as a control for transfection efficiency. For each transfection, 1.0 g of FGF-BP promoter-luciferase construct, 0.1 ng of pRL-CMV (transfection efficiency control), and 8 l of LipofectAMINE (Life Technologies, Inc.) were combined and added to cells for 3 h in serum-free conditions as described previously (10). For co-transfections, 1.0 g of Ϫ118/ϩ62Luc FGF-BP promoter construct, 500 ng of expression vector, 0.1 ng of pRL-CMV, and 8 l of LipofectAMINE were added to cells. Transfected cells were treated for 18 h with EGF (5 ng/ml) in serum-free IMEM before cell lysis in 150 l of passive lysis buffer (Promega). 20 l of extract was assayed for both firefly and Renilla luciferase activity using the Dual-Luciferase TM reporter assay system (Promega). Due to a small background induction (1.5-2 fold) of the pRL-CMV plasmid by EGF, all luciferase values were normalized for protein content. There were no significant differences, however, in the transfection efficiencies between plasmid constructs in untreated controls as determined by Renilla luciferase assay. Protein content of cell extracts was determined by Bradford assay (Bio-Rad).
Immunocomplex Akt Kinase Assay-ME-180 cells were grown in 150-mm dishes and pretreated as described above for Northern analysis. Cells were treated with EGF (5 ng/ml) for 5 min. Cell lysates were prepared by scraping the cells into immunoprecipitation buffer (50 mM Tris-HCl, pH 8, 150 mM NaCl, 0.5% deoxycholic acid, 1% Nonidet P-40, 10% glycerol, 1 mM sodium orthovanadate, 1 M okadaic acid, 50 mM sodium fluoride, 2 g/ml aprotinin, 1 g/ml pepstatin A) and incubating for 15 min at 4°C in a rotating rack. The lysates were cleared by centrifugation, and protein content was measured with the Bio-Rad protein assay kit. Lysates were precleared with protein G-Sepharose and incubated for 4 h at 4°C with 3 g of sheep anti-Akt1 antibody (Upstate Biotechnology, Inc., Lake Placid, NY). Immunocomplexes were captured with protein G-Sepharose at 4°C for 1 h. The beads were then washed with 50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 1 mM dithiothreitol. The Akt kinase assay was carried out as described previously (26), using the K9 peptide as a substrate.

EGF Treatment Increases FGF-BP mRNA in SCC Cells-We
have shown previously that phorbol ester (TPA) treatment of the human cervical SCC cell line ME-180 results in a time-and dose-dependent increase of FGF-BP mRNA, which is mediated via the PKC signal transduction pathway (10). Similarly, treatment of ME-180 cells with EGF resulted in a rapid increase in the steady-state levels of FGF-BP mRNA with no effect on GAPDH mRNA levels ( Fig. 1). FGF-BP mRNA induction was detectable after 1 h of treatment and was maximal after 6 h with an average increase of 4.5-fold. The rapid and transient nature of EGF induction of FGF-BP mRNA is identical to that seen after TPA treatment (10), suggesting that these agents might induce FGF-BP through a similar transcriptional mechanism.
To determine whether FGF-BP is up-regulated by EGF at the transcriptional level, we tested the effect of the transcription inhibitor actinomycin D on the EGF induction of FGF-BP mRNA. Pretreatment with actinomycin D completely blocked the induction of FGF-BP by EGF (see Fig. 4). Combined treatment with EGF and cycloheximide had no effect on induction of FGF-BP mRNA (data not shown), indicating that de novo protein synthesis is not required for the EGF response. Furthermore, transient transfection of the full-length FGF-BP promoter from Ϫ1060 to ϩ62 into ME-180 cells resulted in a 4.5-fold increase in luciferase activity upon EGF treatment (Fig. 2). These data demonstrate that EGF, like TPA, can directly increase the rate of FGF-BP gene transcription.

Identification of EGF Response Elements within the FGF-BP Promoter-Functional analysis of the FGF-BP promoter has
shown that TPA-induced transcription involves a combination of both positive and negative regulatory elements located within the first 118 base pairs of the proximal promoter (10). Full TPA induction was mediated by a C/EBP consensus site between Ϫ48 and Ϫ40, as well as through a juxtaposed Sp1/ AP-1 element between Ϫ76 and Ϫ59. The Sp1(b) site between Ϫ76 and Ϫ68 was described previously as an Ets element based on its homology to other Ets consensus binding sites (10). However, we have now determined through supershift analysis that this site is bound by the Sp1 transcription factor (Fig. 3D). To investigate whether these same regulatory elements play a role in EGF induction of the FGF-BP promoter, we transiently transfected a series of mutated promoter constructs into ME-180 cells and tested their ability to be induced after treatment with EGF. Deletion of promoter sequences from Ϫ1060 to Ϫ118 had no effect on the level of promoter activity and resulted in a similar 5-fold EGF induction (Fig. 2), demonstrating that the EGF regulatory region of the promoter is located within the first 118 base pairs of the promoter. Although the upstream Sp1 site (Sp1(a)) drives a significant portion of basal promoter activity (10), deletion of this site had no effect on EGF induction (data not shown). Deletion of the Sp1(b) site within the context of the Ϫ118/ϩ62 promoter fragment also had no effect on EGF induction (Fig. 2). Basal activity of the promoter, however, did drop significantly in the absence of the Sp1(b) element ( Fig. 2 and Ref. 10). In contrast, mutation of the AP-1 site resulted in a significant 40% decrease in EGF induction as compared with the wild-type Ϫ118/ϩ62 promoter. Similarly, complete deletion of the juxtaposed Sp1(b)/AP-1 site resulted in a 46% decrease in EGF response. Mutation of the AP-1 site had no effect on basal promoter activity, demonstrating that changes in EGF responsiveness can occur independently of changes in the basal rate of transcription. These results show that unlike TPA induction, which requires the entire Sp1(b)/AP-1 element, EGF induction is driven mainly through the AP-1 site in the promoter, reflecting possible differences in the mechanisms by which each of these agents regulate the transcription of FGF-BP.
To test the contribution of the C/EBP element to EGF induction, we introduced an internal deletion of this site within the context of the Ϫ118/ϩ62 promoter. Removal of the C/EBP site significantly reduced the EGF effect by 46% but had no influence on basal promoter activity (Fig. 2). Together, these data demonstrate the requirement for an intact C/EBP site and AP-1 site in the EGF induction of the FGF-BP promoter.
In addition to these positive regulatory elements, we recently showed that TPA regulation of the FGF-BP promoter involves a negative regulatory element, lying just downstream of the AP-1 site, which shows similarity to an E-box factor binding site (10). We wanted to determine whether this E-box repressor site also played a role in the regulation by EGF. Point mutation of this site at position Ϫ58, or deletion of this site from Ϫ57 to Ϫ47 resulted in a dramatic increase in EGF induction of the promoter from about 5-fold in the wild-type promoter to about 8 -10-fold in the repressor mutants (Fig. 2), indicating loss of repressor activity. Therefore, the repressor element between Ϫ58 and Ϫ47 also plays a regulatory role in EGF induction by limiting the transcriptional response to growth factor stimulation.
AP-1, C/EBP, and Sp1 Binding to the FGF-BP Promoter-Because the AP-1 site in the FGF-BP promoter appears to be important during TPA and growth factor regulation of FGF-BP, we determined which members of the Fos and Jun family might be binding to the FGF-BP AP-1 site. To identify transcription factor binding to the AP-1 site, we carried out gel shift analysis using labeled promoter sequence fragment from Ϫ70 to Ϫ51 (Fig. 3A), which was incubated in the presence of nuclear extracts from control or EGF-treated ME-180 cells. Protein binding in the uppermost complex, which has previously been shown to represent AP-1 (10), is highly induced by EGF treatment (Fig. 3B, lanes 1 and 2). To determine the composition of the AP-1 complex, we used Fos and Jun-specific antibodies for supershift analysis. As shown in Fig. 3B, the addition of a Fos antibody that recognizes all Fos family members (lane 3) or that specifically recognizes c-Fos (lane 4) resulted in a supershifted complex. Antibodies against FosB, Fra-1, or Fra-2 had no effect. Furthermore, incubation with a general Jun family antibody (Fig. 3B, lane 8) or with a JunD-specific antibody (lane 11) either blocked or supershifted the AP-1 complex, respectively. The c-Jun-or JunB-specific antibodies had no effect on AP-1 binding. These results demonstrate that c-Fos and JunD are the major components of AP-1 binding to the FGF-BP promoter.
The binding of C/EBP to the FGF-BP promoter was investi- gated using a promoter fragment spanning the C/EBP site from Ϫ55 to Ϫ30 (Fig. 3A). Gel shift analysis with this fragment in the presence of nuclear extracts from untreated or EGF-treated ME-180 cells showed no significant change in the binding of C/EBP (Fig. 3C, lanes 1 and 2). This finding is consistent with previously published observations that PKC activation phosphorylates and enhances C/EBP transcriptional activity with no effect on DNA-binding activity (27). The specificity of C/EBP binding in the upper complex of the gel shift (Fig. 3C, arrow) is demonstrated through competition with a molar excess of unlabeled Ϫ55/Ϫ30 fragment (lane 9) as well as with a C/EBP consensus element (lane 10). Incubation with an antibody against C/EBP␤ that also cross-reacts with other members of the C/EBP family resulted in one prominently supershifted band (Fig. 3C, lane 3). In addition, antibodies specific for C/EBP␤ (lane 4) and C/EBP␦ (lane 7) also supershifted the complex, whereas antibodies for C/EBP␣ (lane 5), C/EBP␥ (lane 6), and C/EBP⑀ (lane 8) had little effect on C/EBP binding. Because the C/EBP site is required for full EGF induction of the FGF-BP promoter (Fig. 2), these results suggest that C/EBP␤ and -␦ are involved in mediating the EGF effect. The mechanism by which these factors are activated in response to EGF, however, remains to be determined.
As mentioned previously, the promoter element between Ϫ80 and Ϫ63 shows some homology to an Ets factor binding site. Closer analysis of this site also revealed homology to a variant Sp1 site, which was previously shown to respond to EGF treatment (28). In fact, supershift analysis of the Ϫ80 to Ϫ63 element showed specific binding of Sp1 (Fig. 3D). This element has been denoted Sp1(b) in order to distinguish it from the Sp1 site further upstream (Sp1(a)). Interestingly, the binding of Sp1 to the Sp1(b) site increased in the presence of nuclear extracts from EGF-treated cells (Fig. 3D, lanes 1 and 2) despite the fact the removal of the Sp1(b) site alone has no impact on the overall EGF induction of the FGF-BP promoter (Fig. 2).

Role of PKC and MEK1/2 Signal Transduction Pathways in
FGF-BP Regulation-In order to differentiate between the possible signaling pathways involved in EGF induction of FGF-BP, we tested pharmacological inhibitors of signal transduction components for their effect on FGF-BP regulation. We found that treatment with the EGFR tyrosine kinase inhibitor tyrphostin AG1478 reduced EGF induction of FGF-BP to 30% (Fig. 4), which was not significantly different from the basal level of expression (without EGF or drug treatment) of approximately 25% (data not shown). As expected, EGFR tyrosine kinase activity is essential for the EGF effect on the FGF-BP gene. In addition, we have shown previously that TPA induction of FGF-BP transcription was mediated through a PKC-dependent pathway (10). To establish whether PKC activation was also required for the EGF effects on FGF-BP, we treated ME-180 cells with the specific PKC inhibitor calphostin C (29) and found that this completely blocked EGF induction of FGF-BP mRNA (Fig. 4). Therefore, PKC activation is central in mediating FGF-BP transcriptional activation upon either EGF or TPA stimulation.
Since the MEK/ERK pathway is known to be stimulated by EGF, we investigated the contribution of the MAP kinase kinases MEK1 or MEK2 to FGF-BP using pharmacological inhibitors of MEK1/2. Treatment with the drug U0126, which is a potent inhibitor of both MEK1 and MEK2, abrogated EGF induction of FGF-BP mRNA (Fig. 4). Consistent with the role of MEK in FGF-BP induction, treatment with a less potent MEK inhibitor PD98059 also blocked the EGF effect on FGF-BP, albeit at higher concentrations (data not shown). Therefore, activation of the MEK pathway is a necessary step in EGF regulation of FGF-BP.
EGF is also known to stimulate intracellular signaling via the PI 3-kinase pathway (18). In order to test the contribution of PI 3-kinase to FGF-BP regulation, we used the drug wortmannin to specifically inhibit PI 3-kinase activity. Pretreat-

FIG. 2. Effect of FGF-BP promoter mutations on the transcriptional induction by EGF.
The histogram on the left shows the impact of each promoter deletion on the basal (uninduced) luciferase activity of each construct. The basal activity of the Ϫ118/ϩ62 construct was set at 100%. The right histogram shows the transcriptional activity in the presence of EGF and is expressed as -fold induction of EGF-treated over untreated for each construct. ME-180 cells were transiently transfected with the indicated FGF-BP promoter luciferase constructs and were either untreated or treated with 5 ng/ml of EGF for 18 h. Promoter constructs are described under "Experimental Procedures" and in Ref. 10. Values represent the mean and S.E. from at least three separate experiments, each done in triplicate wells. Statistically significant differences relative to the Ϫ118/ϩ62 promoter construct are indicated (*, p Ͻ 0.05; **, p Ͻ 0.01, t test). ment with wortmannin at two different concentrations had no effect on EGF induction or FGF-BP mRNA (Fig. 5A). Endogenous PI 3-kinase activity in ME-180 cells, as measured by immunoprecipitated Akt kinase activity, was induced approximately 2-fold by EGF and was effectively blocked by the same concentrations of wortmannin and under the same experimental conditions used for the analysis of FGF-BP (Fig. 5B). Therefore, we conclude that activation of PI 3-kinase upon EGF treatment does not play an essential role in the regulation of FGF-BP expression. Furthermore, the lack of effect by wort- mannin served as a negative control and demonstrates specificity of the inhibitions observed in Fig. 4.
Induction of FGF-BP through p38 Kinase-Stimulation of the JNK/p38 MAP kinase pathway has also been shown to regulate AP-1 activity in response to mitogens and stress (17). We therefore tested whether JNK or p38 activation could induce FGF-BP gene expression by treating with the antibiotic anisomycin. Anisomycin treatment at concentrations below 200 nM is known to be an effective stimulator of both JNK and p38 kinase activity (30). Treatment of ME-180 cells with anisomycin alone resulted in a significant and dose-dependent increase of FGF-BP mRNA levels up to 2.3-fold (Fig. 6A), suggesting that activation of the JNK/p38 pathway might be involved in the regulation of FGF-BP.
To further investigate the involvement of p38 kinase in FGF-BP regulation, we tested the contribution of p38 to FGF-BP induction using the p38-specific inhibitors SB203580 and SB202190, which have no inhibitory activity for JNK or ERK1/2 (31)(32)(33). Treatment with SB203580 or SB202190 significantly reduced EGF induction of FGF-BP mRNA by 50% to 75% in a dose-dependent manner (Fig. 6B). In contrast, there was no reduction after treatment with SB202474, a drug with a similar structure as SB203580 and SB202190 but with no inhibitory activity for p38. Additionally, p38 inhibition completely blocked anisomycin induction of FGF-BP (Fig. 6B, inset), demonstrating that both anisomycin and EGF induction of FGF-BP mRNA require p38 activation.
To investigate a possible additive or synergistic interaction between ERK and p38 pathways in EGF-induced FGF-BP expression, we treated cells simultaneously with the MEK inhibitor (U0126) and the p38 inhibitor (SB202190) and examined the effect on FGF-BP mRNA. As shown in Fig. 7, treatment with suboptimal doses of U0126 alone or with SB202190 alone resulted in approximately 20% decrease in EGF-induced FGF-BP. Simultaneous treatment with both inhibitors resulted in a 40% reduction of FGF-BP mRNA levels, indicating that the contribution of each of these pathways is additive and not synergistic.
Contribution of MEK/ERK and p38 Pathways to FGF-BP Promoter Activity-Inhibition of MEK and p38 with pharmacological agents demonstrated the importance of these kinases in mediating the induction of FGF-BP by EGF. To confirm the role of each of these pathways in the regulation of FGF-BP transcription, we expressed dominant-negative mutants that specifically suppress the enzymatic activity of their endogenous counterparts. Co-transfection of a dominant-negative form of MEK2 with the FGF-BP promoter luciferase construct (Ϫ118/ ϩ62) into ME-180 cells resulted in a significant decrease in the amount of EGF induction (Fig. 8A). In addition, inhibition of ERK, the MAP kinase target of MEK, by co-transfection of a kinase-deficient dominant negative ERK2 mutant, also resulted in a significant decrease in EGF induction (Fig. 8A). Inhibition of the MEK/ERK signaling pathway with these dominant negative mutants led to nearly a 50% decrease in EGF induction of the FGF-BP promoter. These results support the hypothesis that activation of MEK and ERK by EGF are nec-

FIG. 4. Effect of signal transduction inhibitors on EGF induction of FGF-BP mRNA.
FGF-BP mRNA levels from ME-180 cells treated with 5 ng/ml EGF for 6 h. Cells were pretreated for 1 h with vehicle alone, 5 g/ml actinomycin D (transcription inhibitor), 100 M tyrphostin AG1478 (EGFR tyrosine kinase inhibitor), 100 nM calphostin C (PKC inhibitor), or 10 M U0126 (MEK1/2 inhibitor). Northern blot signal intensities of FGF-BP mRNA were quantitated, normalized to GAPDH, and expressed relative to mRNA levels after EGF treatment alone (without inhibitor), which was set to 100%. Basal FGF-BP level (without EGF or inhibitor) was approximately 25%. Values represent the mean and S.D. of at least two separate experiments.

FIG. 5. Effect of PI 3-kinase inhibition on EGF-induced FGF-BP mRNA.
A, FGF-BP mRNA levels after pretreatment with vehicle alone or with the indicated concentration of wortmannin (PI 3-kinase inhibitor) for 1 h, followed by 6 h of treatment with 5 ng/ml EGF. FGF-BP mRNA levels were analyzed by Northern blot, normalized for GAPDH, and expressed relative to untreated control. B, inhibition of PI 3-kinase activity by wortmannin. ME-180 cells were pretreated for 1 h with or without wortmannin, stimulated for 5 min with 5 ng/ml EGF, and assayed for Akt kinase activity (downstream target of PI 3-kinase) as described under "Experimental Procedures." essary for induction of the FGF-BP gene.
To test the contribution of p38 and JNK to EGF induction of FGF-BP, we co-transfected dominant negative mutants of JNK and p38 along with the FGF-BP promoter into ME-180 cells (Fig. 8A). Each dominant negative mutant contains an alanine and phenylalanine in place of the activating threonine or tyrosine phosphorylation sites (22,23). Expression of the dominant-negative JNK mutant had no significant effect on EGF induction compared with the empty vector. On the other hand, expression of the dominant-negative p38 mutant consistently reduced EGF induction of FGF-BP transcription. Although inhibition of p38 reduced promoter activity by only 23%, this difference was statistically significant from empty vectortransfected cells. Additionally, we tested whether overexpression of wild-type JNK or p38 constructs could affect EGF in-duction of the FGF-BP promoter. In agreement with the dominant-negative data, expression of JNK had no significant effect on promoter activity, whereas expression of p38 consistently resulted in a higher level of EGF induction of FGF-BP transcription, which was significantly higher than the empty vector control (Fig. 8B). Furthermore, expression of a constitutively active mutant of MKK6, MKK6(Glu), a MAP kinase kinase that specifically phosphorylates p38 (25), was also tested for its effect on FGF-BP induction. Overexpression of MKK6(Glu) caused a significant increase in both basal and EGF-induced FGF-BP promoter activity (Fig. 8B). Overall, these data implicate ERK and p38 as being necessary for the full induction of FGF-BP by EGF. The lack of effect on FGF-BP promoter activity by either JNK overexpression or by inhibition of JNK through expression of a dominant-negative JNK mutant or expression of JIP, which prevents nuclear translocation of activated JNK (24) (Fig. 8B), suggests that JNK activation does not play a role in the regulation of FGF-BP in these cells. DISCUSSION This study shows for the first time that transcription of FGF-BP, an important mediator of FGF activation in SCC, is directly induced by EGF. The EGF family of growth factors, which include EGF, transforming growth factor-␣, and other structurally related peptides, cause cellular signaling through the EGFR pathway, regulating proliferation and differentiation of many tissue types. Deregulation of the EGF-induced signaling network is known to play an important role in the tumorigenesis of several human cancers, including neoplasms of the brain, lung, breast, ovary, pancreas, prostate, and colon, as well as in SCC of the skin and cervix (11,12,34). Here, we have shown that EGF up-regulates FGF-BP gene expression, suggesting a link between activated EGF signaling in a cell and subsequent FGF activation, ultimately leading to activation of FGF-mediated processes, such as angiogenesis, during development and tumor growth.
The up-regulation of FGF-BP by EGF was found to show characteristics similar to TPA induction of this gene, including a rapid and transient increase in FGF-BP mRNA levels, and the requirement of transcriptional elements within the TPA FGF-BP mRNA levels from ME-180 cells treated with 5 ng/ml EGF for 6 h. Cells were pretreated for 1 h with vehicle alone or with suboptimal concentrations of U0126 (MEK1/2 inhibitor) and/or SB202190 (p38 inhibitor). Northern blot signal intensities of FGF-BP mRNA were quantitated, normalized to GAPDH, and expressed relative to mRNA levels after EGF treatment alone (without inhibitor), which was set to 100%. Basal FGF-BP level (without EGF or inhibitor) was approximately 25%. Values represent the mean and S.D. of at least two separate experiments. regulatory region (Ϫ118 to ϩ62) of the promoter. Transcriptional regulation of FGF-BP by EGF required the AP-1 site between Ϫ65 and Ϫ61, which is bound by c-Fos and JunD family members (Figs. 2 and 3B). The finding that c-Fos is associated with FGF-BP up-regulation is significant, since an important role for c-Fos during SCC formation has been observed. Studies with c-fos knockout mice demonstrated that c-fos is required for malignant transformation of skin papillomas into malignant carcinomas, since c-fos(Ϫ/Ϫ) papillomas became desiccated and hyperkeratinized, lacked vascularization, and remained benign (35). Since the FGF-BP gene was previously found to be highly induced during skin carcinogenesis (8), it seems possible that this gene could be a target of c-Fos activation during skin SCC tumor formation, providing at least one mechanism for the recruitment of new blood vessels to the tumor.
The other required EGF response element in the FGF-BP promoter is the C/EBP site between Ϫ47 and Ϫ33, which is predominantly bound by the C/EBP␤ and C/EBP␦ family members (Figs. 2 and 3C). C/EBP proteins are a family of leucine zipper transcription factors that play a central role in the acute phase response and in a number of cell differentiation pathways (reviewed in Ref. 36). Regulation of C/EBP activity occurs at multiple levels, including increased gene expression (37,38), nuclear localization (39), enhanced DNA binding (40), and posttranscriptional modification by protein kinases (27,41). Because we observed no increase in C/EBP binding to the FGF-BP promoter after EGF treatment, it seems likely that stimulation of C/EBP transcriptional activity occurs through a post-translational modification. This conclusion is consistent with other studies demonstrating that TPA stimulation of the PKC pathway in hepatoma cells results in increased phosphorylation of C/EBP␤, enhanced transcriptional efficacy, and no obvious changes in DNA binding (27,42).
C/EBP family members recognize similar DNA elements in their target genes, where they bind either as homodimers or heterodimers with other C/EBP family members or with other leucine zipper factors (43). Whereas C/EBP␣ is generally associated with growth arrest and cellular differentiation, C/EBP␤ and C/EBP␦ are often correlated with gene activation during cellular proliferation, inflammation, and tumorigenesis (44 -46). Analysis of different C/EBP target gene promoters has demonstrated a finely tuned regulation of gene expression through the interplay of different C/EBP factors. During the acute phase response, for example, the amount of C/EBP␣ homodimers or heterodimers is reduced and replaced by C/EBP␤ and C/EBP␦ complexes (47)(48)(49). Interestingly, study of cyclooxygenase-2 promoter regulation by C/EBP during skin carcinogenesis showed a change in C/EBP complexes from C/EBP␣⅐C/EBP␤ in normal skin to predominantly C/EBP␤⅐C/ EBP␦ complexes in skin SCC (50). The presence of C/EBP␤⅐C/ EBP␦ complexes on the FGF-BP promoter in ME-180 SCC cells raises the possibility that FGF-BP up-regulation during tumor formation may be in part due to an interplay between the different C/EBP family members.
In trying to delineate the signal transduction pathway regulating FGF-BP gene expression, we found that activation of PKC plays a central role, since inhibition of PKC with calphostin C blocks both the EGF (Fig. 4) and TPA (10) induction of this gene. PKC activation in response to growth factor stimulus can lead to stimulation of the classical MAP kinase pathway Raf/MEK/ERK and subsequent activation of AP-1. PKC-mediated signaling through this pathway can be achieved through either Ras-dependent (15) or Ras-independent (16) mechanisms. The involvement of Ras in FGF-BP regulation is suspected, since FGF-BP expression is increased after DMBA treatment and in ras-transformed keratinocytes (8). A direct role for Ras in the activation of FGF-BP expression, however, has yet to be determined. In addition to PKC, the EGF-activated MEK/ERK pathway plays a significant role in the regulation of FGF-BP gene expression, since pharmacological and dominant-negative inhibition of MEK or ERK abrogates EGF induction. Although the exact target of ERK activation was not examined in this study, ERK has been shown to phosphorylate and activate both AP-1 (14) and C/EBP (41) family members.
p38 is a JNK-related MAP kinase that is activated in response to a variety of stimuli including growth factors, phorbol esters, cytokines, and environmental stress (17). Using a number of different approaches, we show in this study that in FIG. 8. Effect of dominant negatives on EGF induction of FGF-BP promoter activity. A, ME-180 cells were transiently cotransfected with the Ϫ118/ϩ62 FGF-BP promoter construct along with either empty vector or with dominant negative mutant constructs for MEK2, ERK2, JNK, or p38. Transfected cells were either untreated or treated with 5 ng/ml EGF for 18 h. EGF induction is determined by the -fold increase in luciferase activity and expressed relative to the empty vector control, which is set at 100%. Basal (uninduced) promoter activity was approximately 18% of EGF-treated. Values represent the mean and S.E. from at least three separate experiments, each done in triplicate wells. Statistically significant differences relative to the empty vector control are indicated (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001, t test). B, ME-180 cells co-transfected with Ϫ118/ϩ62 FGF-BP promoter construct and either empty vector or expression vectors for JNK, p38, MKK6 (Glu), or JIP. Cells were treated as in A, and the FGF-BP promoter activity is shown relative to the untreated empty vector control. Statistical analysis was as described in A, and differences in the absence or presence of EGF are compared with the empty vector control in the absence or presence of EGF, respectively. addition to ERKs, p38 contributes to the EGF induction of FGF-BP. We demonstrated the inhibition of EGF-induced FGF-BP by pyridinyl imidazole compounds (SB202190 and SB203580), which selectively inhibit p38␣ and p38␤2 isoforms but have no effect on the activity on other p38 isoforms, JNK, or ERK (31)(32)(33). In addition, we have shown that FGF-BP promoter activity is inhibited by expression of a dominantnegative p38 mutant but is activated by overexpression of wild-type p38. p38 is phosphorylated and activated by the dual specificity protein kinases MKK3, MKK4, and MKK6 (23,25,(51)(52)(53)(54). Although overexpression of the p38-specific kinase MKK6 can stimulate FGF-BP transcription, the signal transduction cascades that connect EGFR activation to phosphorylation of p38 in ME-180 cells remain unknown. One possible pathway that is currently under investigation is the involvement of two members of the Rho family of GTPases, Rac and CDC42, which are known to regulate the activity of both JNK and p38 (55)(56)(57)(58). Furthermore, Rac and CDC42 can be activated downstream of Ras (55), thereby connecting p38 and JNK activation to growth factor effects on cell growth and proliferation. Transcription factor targets of p38 include CREB and ATF1 (59,60), ATF2 (25,61), MEF-2C (62), and the C/EBP family members CHOP (63) and C/EBP␤ (64,65), suggesting that C/EBP␤ and/or C/EBP␦ could be targets of p38 activation on the FGF-BP promoter.
In general, EGF regulation of FGF-BP gene expression seemed to be more dependent on the MEK/ERK pathway, since inhibition of MEK1/2 with U0126 completely abrogated induction, and expression of dominant negative MEK or ERK reduced induction by 50%. p38, on the other hand, appears to play a somewhat lesser role, since pharmacological inhibition of p38 caused a maximum 50% reduction of FGF-BP, and expression of dominant negative p38 reduced induction by only 23%. Although stimulation of the p38 pathway plays a less prominent role, regulation of FGF-BP by p38 may be independent of MEK/ERK activation, since they function in an additive rather than synergistic manner (Fig. 7). While the mechanism for these differences in activity of each pathway remains unclear, one explanation could be the differences in their transcription target specificities at the level of the FGF-BP promoter.
This study also examined the possible role of JNK in the regulation of the FGF-BP gene. Based on several lines of evidence, we have concluded that JNK is unlikely to be involved in FGF-BP regulation. Inhibition of JNK activity, either through expression of a dominant-negative JNK mutant or expression of JIP, had no effect on EGF induction of the FGF-BP promoter. Overexpression of wild-type JNK also had no effect on FGF-BP expression. In addition to this, we examined the activation of Elk1 by MEKK, a potent activator of JNK (66), and found no effect on this pathway by dominant negative JNK or JIP expression (data not shown). Furthermore, we found no activation of FGF-BP gene expression in the presence of UV light (data not shown), which is another known stimulator of the JNK pathway (17). Together, these findings indicated that there may be very little JNK activity in ME-180 cells and that JNK activation does not significantly contribute to FGF-BP gene expression in response to EGF.
In conclusion, this study demonstrates that the growth factor EGF induces FGF-BP gene transcription and characterizes the mechanisms by which this effect is accomplished. The EGFmediated pathways leading to FGF-BP transcription and subsequent angiogenic activation include the selective activation of MEK/ERK and p38 MAP kinase pathways. This study highlights several targets for potential anti-angiogenic therapy of human cancers, which utilize FGF-BP's angiogenic properties for tumor growth.