A Phosphatidylinositol 3-Kinase-regulated Akt-Independent Signaling Promotes Cigarette Smoke-induced FRA-1 Expression*

The FRA-1 proto-oncogene is overexpressed in a variety of human tumors and is known to up-regulate the expression of genes involved in tumor progression and invasion. The phosphatidylinositol 3-kinase (PI3K)-Akt pathway is also known to regulate these cellular processes. More importantly, respiratory toxicants and carcinogens activate both the PI3K-Akt pathway and FRA-1 expression in human bronchial epithelial (HBE) cells. In this study we investigated a potential link between the PI3K-Akt pathway and the cigarette smoke (CS)-stimulated epidermal growth factor receptor-mediated FRA-1 induction in non-oncogenic HBE cells. Treatment of cells with LY294002, an inhibitor of the PI3K-Akt pathway, completely blocked CS-induced FRA-1 expression. Surprisingly pharmacological inhibition of Akt had no significant effect on CS-induced FRA-1 expression. Likewise the inhibition of protein kinase C ζ, which is a known downstream effector of PI3K, did not alter FRA-1 expression. We found that the PI3K through p21-activated kinase 1 regulates FRA-1 proto-oncogene induction by CS and the subsequent activation of the Elk1 and cAMP-response element-binding protein transcription factors that are bound to the promoter in HBE cells.

then phosphorylate Akt at two critical threonine and serine residues. Phosphorylation of these residues is required for Akt activation (4). Activated Akt phosphorylates various effector molecules. In addition to Akt, PI3K-dependent activation of MAP kinases, such as ERK1/2, and of certain PKC isoenzymes, such as PKC and PKC␦, has also been documented in various cell types (3).
CS and its individual components, such as NNK (4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone) and nicotine, have been shown to activate PI3K-Akt pathway in vitro and in vivo (5,6). Moreover an elevated level of phosphospecific form of AKT (Ser 473 ) has been found in pulmonary metaplasias and dysplasias (7), NNK-induced lung tumors (5), and human lung tumors derived from smokers (5,8). These data suggest a potential role for PI3K-Akt pathway in CS-induced respiratory pathogenesis. Consistent with this observation, an enhanced activity of Akt, a major downstream effector of PI3K, has been found in various cancer cell lines/tumor tissues (3,9). However, the downstream mechanisms of CS-induced P13K-Akt pathway are unclear.
The activator protein-1 family of transcription factors (c-JUN, JUNB, JUND, c-FOS, FOSB, FRA-1, and FRA-2) regulate several cellular responses induced by various oxidants and toxicants (10). These proteins play key roles in the initiation and progression of tumors (11). However, the links between PI3K-Akt signaling and CS-induced activator protein-1 activation have not been defined in HBE cells. A high level of FRA-1 expression has been detected in various tumors and lung cancer cells (12)(13)(14)(15), and some known carcinogens, such as CS and asbestos, persistently activate FRA-1 expression in lung epithelial cells (16). Several observations suggest that FRA-1 may play a key role in carcinogen-induced lung epithelial cell proliferation and transformation. For example, FRA-1 is required for asbestos-induced mesothelial cell transformation (17). It up-regulates the expression of genes coding for airway squamous cell metaplastic marker (18), matrix metalloproteinase-9 (19), and CD44 and c-met proto-oncogene (20), which are implicated in toxin-induced respiratory pathogenesis. FRA-1 controls cell motility, invasion, and maintenance and progression of the transformed state in several cell types (for reviews, see Refs. 11 and 21). Recently several studies demonstrated a critical role for this transcription factor in controlling both cell motility and invasiveness of several tumorigenic cell lines (22)(23)(24).
Mutational induction of FRA-1 gene in tumor tissues or cell lines is not known to date. Thus, an abnormal signaling and subsequent transcription factor activation may contribute to high level expression of FRA-1. Recently we have shown that metalloproteinase (MMP)-epidermal growth factor receptor (EGFR)-mediated MAP kinase signaling plays an obligatory role in controlling CS-stimulated FRA-1 proto-oncogene induction in HBE cells (25). Although the activation of PI3K-AKt pathway by cigarette smoke and NNK has been reported earlier by others (5)(6)(7), none of these studies showed the upstream signals that activate P13K pathway by CS and the subsequent downstream effectors that control activator protein-1 proto-oncogene induction. Here we show, for the first time, that EGFR-dependent PI3K-regulated CS-stimulated FRA-1 expression occurs through a PAK1-c-Raf-MEK1/2-ERK1/2 pathway that culminates in the activation of transcription factors Elk1 and CREB independently of Akt.
Cell Culture and CS Exposure-The 1HAEo cell line, an SV40-transformed nonmalignant HBE cell line (28), was maintained in minimum essential medium containing 10% fetal calf serum. Cells were grown in 25-cm 2 flasks (BD Biosciences) to 90% confluence, serum-starved for 14 h, and then exposed to mainstream CS as described previously (25). In brief, CS was drawn into a syringe-driven device and then delivered via the tube into cell culture flasks that were inverted to expose cells directly to smoke for 2 min. After that the flasks were restored to their original orientation so that the cells were covered with culture medium, and the flasks were placed in the 37°C incubator. Throughout this study, we used CS at a dose of 2 cc/25-cm 2 flask. No cell cytotoxicity was observed under the conditions of CS exposure.
Kinase Immunoblot Analysis-Cells were exposed to CS for various time points, washed three times with chilled phosphate-buffered saline containing 1 mM Na 3 VO 4 and then lysed in MAP kinase lysis buffer (20 mM Tris, pH 7.5, with 150 mM NaCl, 0.2 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM Na 3 VO 4 , 5 mM ␤-glycerol phosphate, and 1 g/ml leupeptin). Cell lysates were separated on an SDS-PAGE gel and transferred to a polyvinylidene difluoride membrane. The membranes were blocked with 5% nonfat dry milk. Blots were then incubated overnight at 4°C with primary antibodies as indicated under "Results" and washed three times with Tris-buffered saline with Tween 20 before probing with horseradish peroxidase-conjugated secondary antibodies for 1 h at room temperature. Blots were then visualized with ECL reagent (Amersham Biosciences). Band intensities were quantified by densitometry. Kinase phosphorylation was normalized to that of total kinase protein levels and then expressed as a percentage of that untreated control.
Immunoprecipitation of PAK1 and c-Raf-Cells were exposed to CS for various time points and then harvested in ice-cold cell lysis buffer.
Lysates (ϳ200 g) were incubated with 5 l of anti-PAK1 or anti-c-Raf antibodies with gentle rocking overnight at 4°C. The immunocomplex was precipitated using protein A-agarose beads (20 l of 50% bead slurry) with gentle rocking for 3 h at 4°C. After centrifugation, the beads were washed with cell lysis buffer five times followed by addition of 20 l of 3ϫ SDS sample buffer. Beads were boiled for 5 min and then centrifuged to remove agarose beads prior to electrophoresis. The sample was analyzed by Western blotting with phosphospecific antibodies anti-PAK1 and anti-c-Raf.
Gene Expression Analysis-After CS exposure, cells were washed extensively with cold phosphate-buffered saline containing 1 mM Na 3 VO 4 and harvested into MAP kinase lysis buffer. Protein samples (ϳ40 g) were resolved by 10% SDS-PAGE and transferred to polyvinylidene difluoride membranes. Membranes were blocked overnight at 4°C in Tris buffer containing 0.1% Tween and 5% nonfat milk and incubated at room temperature for 1 h with FRA-1 antibody. Membranes were then incubated with horseradish peroxidase-conjugated secondary antibody for 1 h, and immunoreactive bands were detected using the ECL reagent (Amersham Biosciences). Blots were stripped and probed with ␤-actin antibodies to ensure a comparable loading of extracts. For Northern blot analysis, after exposing to CS, total RNA was isolated from cells using the TRIzol reagent. RNA (ϳ15 g) was separated on a 1.2% agarose gel, blotted onto a Nytran membrane, and sequentially hybridized with 32 P-labeled FRA-1 and ␤-actin cDNAs as probes as described previously (25). The total amount of FRA-1 mRNA and protein in each sample was quantified with a Gel Doc 2000 system (Bio-Rad) and normalized to that of ␤-actin band. For transient transfections, cells were transfected with 100 ng of the Ϫ379 to ϩ32 bp FRA-1 promoter reporter luciferase construct (hereafter abbreviated as 379-Luc, see Ref. 29 for details) along with 5 ng of the Renilla luciferase plasmid (pRL-TK, Promega Corp.). After overnight incubation, cells were serum-starved for 16 h and then exposed to CS for 5 h, and luciferase activity was analyzed as described previously (25).
Electrophoretic Mobility Shift Assay (EMSA)-Serum-starved cells were exposed to CS, and nuclear extracts were prepared as described previously (26). Briefly the EMSAs were performed using 2-3 g of nuclear extract in 20 l of binding buffer on ice for 10 min prior to the addition of the appropriate 32 P-labeled double-stranded oligo probe. After incubating for 30 min at room temperature, samples were resolved on a 4% polyacrylamide gel containing 2% glycerol. To demonstrate the presence of a specific protein in the complexes, nuclear extracts were mixed with 1-2 g of anti-CREB (catalog number 06-519, Upstate Cell Signaling) antibodies and incubated on ice for 2 h prior to adding the probe.
Chromatin Immunoprecipitation (ChIP) Assays-ChIP assays were carried out as described earlier (26). Briefly cells (ϳ1 ϫ 10 7 ) were exposed to filtered air or CS for 60 min, and ChIP was performed using a commercially available kit (Upstate Biotechnology Inc.). Chromatin was cross-linked by adding formaldehyde (1%) to the tissue culture medium for 10 min at 37°C. A fraction of the soluble chromatin (1%) was saved for measurement of total chromatin input. Precleared chromatin was incubated with specific antibodies for 18 h at 4°C. DNA recovered from the immunoprecipitated products was used as a template for PCR with FRA-1 promoter-specific primers (26).
Statistical Analysis-Data are expressed as the mean Ϯ S.E. Statistical significance was determined using t test and accepted at p Ͻ 0.05. All assays were performed in two or three (n ϭ 2-3) independent samples, and each experiment was repeated at least two times.

PI3K-mediated Signaling Is Required for CS-stimulated FRA-1
Expression-To determine whether the PI3K-mediated signaling is necessary for CS-induced FRA-1 expression, HBE cells were treated with LY294002 for 30 min and then exposed to CS for 3 h, and FRA-1 mRNA and protein expression was analyzed by Northern and Western blot analyses, respectively ( Fig. 1). Consistent with our previous results (25), CS significantly (ϳ3-fold) stimulated the expression of FRA-1 mRNA (Fig. 1A) and protein levels ( Fig. 1B) as compared with the cells exposed to filtered air. However, pretreatment of cells with PI3K inhibitor LY294002 completely suppressed CS-induced FRA-1 mRNA expression. Previously we have shown that CS stimulates FRA-1 induction mainly at the transcription level, and the Ϫ379 bp promoter contains sufficient information to mediate such induction (26). To verify that PI3K signaling regulates FRA-1 transcription, cells were transfected with the 379-Luc promoter reporter construct along with the pRK-TL reference plasmid. Promoter activity in response to CS was analyzed in the absence or presence of LY294002 (Fig. 1C). As expected, CS significantly elevated luciferase activity, which was abrogated in the presence of LY294002. No effect of CS on the basal promoter activity was observed. Collectively these observations strongly support a prominent role for PI3K-dependent signaling in mediating the CS-stimulated FRA-1 transcription in pulmonary epithelial cells.
CS Activates PI3K Signaling in HBE Cells-To determine whether CS activates PI3K signaling, we monitored the activation of PDK1 ( Fig. 2A) and its downstream effector Akt (Fig. 2B). Cells were exposed to CS for 0 -60 min, and equal amounts of total lysates from duplicate samples were separated by PAGE and immunoblotted using phosphospecific PDK1 (Ser 241 ) and Akt (Ser 473 ) antibodies. As shown in Fig and 60 min post-CS exposure, respectively. We also found that CS stimulates phosphorylation of Ser 473 in another HBE cell line and in the lung tissues of mice exposed to CS (data not shown). Collectively these data suggest that CS stimulates the PI3K-Akt signaling, consistent with results reported by others in the lung cell types (5,6).
CS Activates PI3K-Akt Pathway through EGFR-Recently we have shown that the activation of EGFR is essential for CS-stimulated FRA-1 induction (25). Given the observation that PI3K inhibitor completely blocks FRA-1 induction and EGFR-mediated signaling activates the PI3K-Akt pathway in response to growth factors in other cell types (30,31), we next examined whether the activation of PI3K-Akt pathway was mediated through EGFR. Cells were treated with AG1478 prior to the exposure, and CS-stimulated Akt phosphorylation was analyzed by immunoblot analysis using phospho-Ser 473 antibodies (Fig. 2C). AG1478 (lanes 7 and 8) com- Western blot (WB) analyses, respectively. Results shown are a representative blot of two independent experiments. C, cells were transfected with 100 ng of 379-Luc along with pRL-TK (5 ng). After overnight incubation, cells were serum-starved for 16 h, treated with Me 2 SO or PI3K inhibitor LY294002 for 30 min, and then exposed to filtered air or CS for 5 h, and luciferase activity was assayed using a dual luciferase kit. The percent increase in CS-stimulated luciferase activity over the filtered air-exposed samples in the presence of Me 2 SO was considered equal to 100 (first bar). Promoter activity in response to CS in the presence of LY294002 is minimal (second bar). Values represent mean Ϯ S.E. (n ϭ 3) of a representative experiment. The experiment was repeated twice. FIGURE 2. CS stimulates PI3K-Akt signaling, and EGFR and MMP activity is required for this activation in HBE cells. Cells were exposed to filtered air (Ϫ) or CS (ϩ) for 0 -60 min and harvested in MAP kinase lysis buffer. Lysates (ϳ40 g) were separated by SDS-PAGE and immunoblotted with phosphospecific PDK1 (A) and Akt (Ser 473 ) (B) antibodies. The membranes were subsequently probed with anti-PDK1 or anti-Akt antibodies, respectively. Bands were quantified and normalized with the respective total protein content (bottom panels). The level of kinase phosphorylation of filtered air-exposed ("0" min) control group was designated as one arbitrary unit (AU). Quantification of a representative blot of two independent experiments is shown (means Ϯ S.D.). *, p Ͻ 0.05 compared with filtered air-exposed samples. C, cells were treated with AG1478 (AG, 2 M) for 40 min and then exposed to CS for 30 min. Akt phosphorylation was analyzed using phosphospecific Akt (Ser 473 ) antibodies. *, p Ͻ 0.001 compared with filtered air-exposed (-CS) samples; $, p Ͻ 0.05 compared with Me 2 SO (DMSO)-treated CS-exposed samples; and #, p Ͻ 0.05 compared with Me 2 SO-treated filtered air-exposed samples. D, cells were treated with 25 M GM6001 for 30 min and then exposed to CS. The total lysates were immunoblotted using phosphospecific Akt (Ser 473 ) antibodies. Quantification of CS-stimulated Akt activation was performed. Quantification of a representative blot of two independent experiments is shown (means Ϯ S.D.). *, p Ͻ 0.001 compared with filtered air-exposed (ϪCS) samples; #, p Ͻ 0.05 compared with Me 2 SO (DMSO)-treated CS-exposed samples.
pletely inhibited CS-stimulated (lanes 3 and 4) Akt phosphorylation. Interestingly inhibition of EGFR activity with AG1478 significantly stimulated basal levels Akt phosphorylation (lanes 5 and 6) indicating a negative crosstalk between EGFR and Akt signaling in the unstimulated state. We have shown earlier that MMP activity is required for CS-stimulated EGFR mediated signaling in the lung. Consistent with this result, treatment of cells with MMP inhibitor GM6001 markedly blocked (Fig. 2D, lanes 7 and 8) CSinduced Akt phosphorylation (Fig. 2D, lanes 3 and 4). Collectively these results suggest that signaling pathways involving EGFR and MMP regulate the activation of PI3K-Akt pathway in response to CS.
Akt-mediated Signaling Is Not Required for CS-stimulated FRA-1 Induction-Enhanced Akt activity has been linked to tobacco-induced cellular responses (5, 6) and tumorigenesis (3). Because both Akt and FRA-1 regulate cellular processes such as cell motility and invasiveness, we sought to determine whether Akt-dependent signaling was required for PI3K-mediated CS-stimulated FRA-1 induction. To examine this aspect, we used two pharmacological inhibitors of Akt, Akt inhibitor II and Akt inhibitor IV. Akt inhibitor II specifically inhibits the activation of Akt without affecting other upstream or downstream kinases (32). Akt inhibitor IV inhibits phosphorylation/activation of Akt by targeting its immediate upstream kinase, PDK1, but not PI3K (33). We first examined the effects of these inhibitors on Akt-mediated signaling under our experimental conditions by monitoring the activation of glycogen synthase kinase, a major downstream target of Akt, in response to EGF. As shown in Fig. 3A, pretreatment of cells with either Akt inhibitor II (top panel, lanes 4 and 5) or Akt inhibitor IV (bottom panel, lanes 4 -6) markedly suppressed EGF-stimulated glycogen synthase kinase phosphorylation (lanes 2 and 3). To confirm the specificities of these inhibitors, these membranes were stripped and probed with phosphospecific ERK1/2 antibodies. As anticipated, both Akt inhibitor II and Akt inhibitor IV had no significant effect on EGF-induced ERK1/2 activation (Fig.  3A). To determine whether Akt-mediated signaling is required for FRA-1 induction, cells were exposed to CS in the presence or absence of either Akt inhibitor II or Akt inhibitor IV, RNA was isolated, and Northern blot analysis was performed. Neither Akt inhibitor II (top panel, lanes 4 -6) nor Akt inhibitor IV (bottom panel, lanes 4 -6) had an appreciable inhibitory effect on the CS-induced expression of FRA-1 mRNA (Fig. 3B, lanes 2 and 3). Thus, although CS-activates Akt, it is not required for regulating FRA-1 expression in response to CS. We next examined whether PKC, which is a known downstream targets of PI3K (3), is involved in this process using the PKC myristoylated pseudosubstrate inhibitor (Fig. 3C). This inhibitor had no effect on CS-stimulated of FRA-1 expression. To determine whether or not CSactivates PKC, we monitored the levels of phosphorylation of this kinase using phosphospecific antibodies. As shown in Fig. 3D, CS stimulated the phosphorylation of PKC in HBE cells as early as 5 min (lane 2) compared with unstimulated cells (lane 1); phosphorylation remained above the basal level through 60 min (lane 5). However, pretreatment of cells with PI3K inhibitor markedly suppressed both basal (Fig. 3E, compare lane 1 and lane 3) and CS-stimulated PKC phosphorylation (Fig. 3E, compare lane 2 and lane 4).
The PI3K Regulates CS-induced FRA-1 Expression via the MEK1/2-ERK1/2 Pathway-The above results indicate that PI3K regulates FRA-1 induction independently of Akt and PKC. Recently we have shown a critical role for ERK1/2-dependent signaling in controlling CSstimulated FRA-1 expression in HBE cells (25). Because PI3K regulates ERK1/2 kinases in response to growth factors in several cell types (3) and because EGFR activity is critical for CS-stimulated FRA-1 induction, we next examined whether PI3K induces FRA-1 through ERK1/2 signaling. were treated with the Akt-specific pharmacological inhibitors II (Akt-iII, 50 M) and IV (Akt-iIV, 10 M) for 30 min. Me 2 SO (DMSO) was used as a vehicle control. Cells were treated without (Ϫ) or with (ϩ) epidermal growth factor (EGF) (20 ng/ml) and immunoblotted with phosphospecific antibodies as indicated. B, cells were treated with Akt inhibitors II and IV for 30 min and exposed to filtered air (Ϫ) or CS (ϩ) for 3 h. Northern blot analysis was carried out under conditions identical to those in Fig. 1. C, the effect of PKC pseudosubstrate inhibitor (PKC -i) on CS-stimulated FRA-1 mRNA expression. D, analysis of PKC phosphorylation in response to CS exposure. Cells were treated with CS for 0 -60 min, and lysates were immunoblotted with phosphospecific PKC antibodies. E, the effect of PI3K inhibitor on CS-stimulated PKC phosphorylation. Cells were treated with Me 2 SO (lanes 1 and 2) or with 20 M LY294002 (LY, lanes 3 and 4) prior to CS (ϩ) or filtered air (Ϫ) exposure. A representative autoradiogram of two independent experiments is shown. p, phospho; GSK, glycogen synthase kinase. Cells were treated with LY294002 prior to CS exposure, and CS-induced ERK1/2 kinase activation was assayed by immunoblot analysis (Fig. 4A). PI3K inhibitor completely blocked CS-stimulated ERK1/2 activation. To further confirm that PI3K regulates ERK1/2 through the activation of MEK1/2 but not via a cross-talk between Akt and ERK, we analyzed the effect of PI3K inhibitor on CS-stimulated MEK1 activation using phosphospecific MEK1 antibodies (Fig. 4B). Indeed LY294002 robustly blocked CS-induced MEK1 phosphorylation (compare lanes 3 and 4 and lanes 7 and 8) indicating that PI3K regulates ERK signaling by activating MEK1, the known upstream activator of ERK1/2.

PAK1 via c-Raf Mediates CS-activated PI3K-MEK-ERK
Pathway-Both PAK1 and/or Raf kinases have been shown to mediate the cross-talk between the PI3K and ERK pathways (34). To examine whether PI3K activates MEK-ERK signaling via PAK1 or Raf, cells were exposed to CS for 0, 15, and 30 min, and CS-induced PAK1 and c-Raf kinase activation was assessed by immunoprecipitation and Western blot analysis using the respective total and phosphospecific antibodies (Fig. 5). As shown in Fig.  5A, CS stimulated both PAK1 (top panel) and c-Raf phosphorylation (bottom panel). Also pretreatment of cells with PI3K inhibitor LY 294002 markedly suppressed CS-stimulated PAK1 and c-Raf phosphorylation (Fig. 5B,  compare lanes 2 and 4). The basal level c-Raf phosphorylation was high in the presence of LY294002 (Fig. 5B, lane 3). To determine the role of PAK1 in CS-stimulated MEK-ERK1/2 pathway, cells were transfected with a GST-tagged dominant mutant PAK1 expression vector encoding amino acids 83-149 of PAK1 (hereafter referred to as GST-PAK1). When expressed mutant GST-PAK1 inhibits PAK1 kinase activity and blocks PAK1 translocation to the centrosomes during mitosis (27). Cells were then exposed to CS, and the activation of c-Raf was monitored by immunoprecipitation and Western blot analysis (Fig. 6A). Overexpression of GST-PAK1 (lanes 3 and 4) significantly inhibited CS-stimulated c-Raf phosphorylation (Fig. 6A) as compared with empty vector-transfected CS exposed cells (lanes 1 and 2). Likewise expression of GST-PAK1 markedly suppressed CS-stimulated MEK1/2 phosphorylation (Fig. 6B). To determine whether c-Raf activation is necessary for CS-stimulated MEK-ERK pathway, cells were treated with c-Raf -specific inhibitor GW5074 prior to CS exposure, and MEK-ERK activation was monitored (Fig. 6C). CS-stimulated MEK1/2 and ERK1/2 activation was significantly suppressed by c-Raf inhibitor (lane 4) as compared with vehicle-treated CS-exposed cells (lane 2). Collectively these results indicate an important role for PAK1 in mediating a cross-talk between the PI3K and Raf-MEK-ERK1/2 signaling pathways.

The Inhibition of PI3K or ERK Pathways Suppresses CS-induced Elk1
and CREB Phosphorylation-To determine other pathways activated by the PI3K, we examined the effect of LY294002 on Elk1 and CREB protein phosphorylation (Fig. 7A). These two transcription factors are downstream targets of ERK signaling and are known to regulate the induction of c-FOS (35) and FRA-1 (26) in response to tumor promoters and mitogens. We also used the MEK-ERK pathway inhibitor PD98059 (Fig. 7B) to compare with the results obtained with the PI3K inhibitor LY294002. As shown in Fig. 7A, CS stimulated the phosphorylation of Elk1 and CREB (lanes 3 and 4) compared with filtered air-exposed samples (lanes 1 and 2). Both LY294002 (Fig. 7A, lanes 7 and 8) and PD98059 (Fig. 7B, lanes 7 and 8) completely blocked such phosphorylation. Together these results show that PI3K-ERK pathway controls the activation of transcription factors Elk1 and CREB in response to CS. To further confirm the role of Elk1 and ATF/CREB proteins in the tran- FIGURE 5. Effect of PI3K inhibitor on CS-stimulated PAK1 and c-Raf activation. A, cells were exposed to filtered air (Ϫ) or CS (ϩ) for different time points as indicated. Cell lysates were immunoprecipitated (IP) with anti-PAK1 or anti-c-Raf antibodies overnight. The immunocomplex was precipitated using protein A-Sepharose beads. Western blot (WB) analysis was performed using the phosphospecific antibodies for PAK1 (pPAK1) and c-Raf (pc-Raf). Lysates (ϳ40 g of protein) were separated and immunoblotted with ␤-actin antibodies to ensure equal amounts of protein present in the immunoprecipitation. B, cells were treated with Me 2 SO (DMSO) or LY294002 (20 M) for 30 min prior to CS (ϩ) or to filtered air (Ϫ) exposure. After a 30-min postexposure, cell extracts were prepared and precipitated with anti-PAK1 or anti-c-Raf antibodies. Western blot analysis was performed using the phosphospecific antibodies for PAK1 and c-Raf. Lysates (40 g of protein) were separately immunoblotted with ␤-actin or c-Raf antibodies. Each experiment was repeated four times (n ϭ 4).

FIGURE 6. The PI3K regulates MEK-ERK signaling through PAK1/c-Raf activation.
Cells were transfected with the GST-PAK1 mutant construct for 48 h. An empty GST vector was used as control. Cells were exposed to CS for 30 min, and whole cell lysates were harvested. A, cell lysates were immunoprecipitated with anti-c-Raf antibody and then immunoblotted with anti-phosphospecific c-Raf antibody. *, p Ͻ 0.03 compared with filtered air-exposed (ϪCS) samples; #, p Ͻ 0.05 compared with Me 2 SO (DMSO)treated CS-exposed samples (n ϭ 3). B, cell lysates as in A were immunoblotted with phosphospecific MEK1 antibody. The membrane was subsequently probed with ERK2 antibody. C, cells were treated with Me 2 SO or GW5074 (10 M) for 40 min prior to CS (ϩ) or to filtered air (Ϫ) exposure. After a 15-min postexposure, cells were lysed, and MEK1/2 and ERK1/2 phosphorylation was analyzed. The membranes were stripped and probed with the respective total antibodies. Each experiment of B and C was repeated at least twice to obtain reproducible results. p, phospho.
scriptional up-regulation of FRA-1 by CS, cells were transfected with the wild-type 379-Luc promoter-reporter construct or with 379-Luc constructs bearing mutations in the SRE or in the ATF binding site (Fig.  7C). The SRF has been shown to be critically required for an efficient binding of Elk1 to the SRE (35). After overnight incubation, cells were exposed to CS for 5 h, and reporter expression was analyzed. Consistent with our previous results (25), CS significantly stimulated luciferase activity driven by the wild-type FRA-1 promoter. However, mutations either in the SRE or ATF site significantly reduced CS-stimulated FRA-1 promoter activity. This result implies that protein binding to these sites is required for the induction of FRA-1 transcription in HBE cells.
CS Exposure Does Not Alter Elk1 Protein Binding at the FRA-1 Promoter in Vivo-Recently we have reported that the Ϫ276 to Ϫ239 bp segment composed of SRE (containing a ternary complex factor site and the CArG box) and a putative ATF site plays a critical role in mitogeninduced FRA-1 transcription in lung epithelial cells (26). Furthermore we have shown that specific knock-down of Elk1 transcription factor significantly attenuates tumor promoter-stimulated FRA-1 transcription (26). To examine whether PI3K signaling altered the engagement of Elk1 with the endogenous FRA-1 promoter, we performed a ChIP assay (Fig. 8). Cells were treated with Me 2 SO (lanes 1-4), LY294002 (lanes 5-8), or PD98059 (lanes 9 -12) prior to CS exposure. After a 60-min incubation with CS, DNA-protein complexes were cross-linked by formaldehyde, soluble chromatin was immunoprecipitated using the anti-Elk1 antibodies and non-immune IgG, and DNA fragments were recovered from the immunoprecipitation products were used as templates for PCR with FRA-1 promoter-specific primers (Fig. 8A). As expected, ChIP assays with the non-immune IgG showed no amplification of the FRA-1 promoter. Elk1 was constitutively engaged with the promoter irrespective of CS exposure (lanes 1-4). Treatment of cells either with PI3K inhibitor (lanes [5][6][7][8] or ERK inhibitor (lanes 9 -12) did not significantly alter the binding of Elk1 to the promoter.
Inhibition of PI3K and ERK Signaling Does Not Affect CREB Binding to the FRA-1 Promoter-Unlike Elk1, we were unable to amplify the FRA-1 promoter to detectable levels in ChIP assays using CREB anti-bodies even at greater than 50 PCR cycles. We therefore performed EMSA assays to examine whether CS enhanced the recruitment of CREB to the ATF site located at position Ϫ248 of the FRA-1 promoter (Fig. 9A) and whether PI3K and ERK inhibitors affected such binding. Cells were exposed to CS for 60 min in the presence or absence of LY294002 or ERK inhibitor PD98059, nuclear extracts were prepared, and EMSA was performed using the ATF site of FRA-1 promoter as a probe. Incubation of both filtered air-exposed and CS-exposed nuclear extracts with the ATF probe yielded a single specific complex (Fig. 9B,  indicated with an arrow). There was no difference in the binding of CREB to ATF site. When incubated with the CREB antibodies (lanes 3    Fig. 1. CS-induced Elk1 and CREB activation was analyzed using their phosphospecific antibodies. The ␤-actin band was used for loading control. Results shown are from a representative experiment that was repeated at least twice. B, the effect of ERK inhibitor PD98059 (25 M) on CS-stimulated Elk1 and CREB phosphorylation. C, cells were transfected with the wild-type 379-Luc promoter-reporter construct or with mutant 379-Luc constructs bearing mutations in SRE or the ATF binding site. pRL-TK was used to monitor transfection efficiency. After overnight incubation, cells were exposed to CS for 5 h, and reporter expression was analyzed. Data are expressed as -fold over room air-exposed (Ctr) samples. The data represent the values of six independent samples. *, p Ͻ 0.01 compared with room air-exposed samples. $ and #, p Ͻ 0.05 and p Ͻ 0.04, respectively, as compared with Me 2 SO (DMSO)-treated CS-exposed samples. AU, arbitrary units; p, phospho; WT, wild type; mt, mutant. and 4) this complex was supershifted. CREB antibodies used in the present study do not cross-react with other ATFs. Collectively these results (Figs. 8 and 9) indicate that both Elk1 and CREB bind to the SRE and the ATF sites of the FRA-1 promoter in unstimulated cells, and CS stimulation has little effect on their steady-state DNA binding.

DISCUSSION
The present study, for the first time, demonstrated a potential link between the EGFR-activated PI3K-dependent ERK MAP kinase signaling and CS-induced FRA-1 proto-oncogene expression (Fig. 1). We showed that the inhibition of EGFR kinase activity markedly suppressed both CS-induced PI3K-Akt signaling (Fig. 2C) and FRA-1 induction (25). Consistent with this result, a general inhibitor of MMP activity, which is essential for CS-induced EGFR activation (36 -38), also had similar effects on CS-induced PI3K-Akt signaling (Fig. 2D) and FRA-1 expression (25). Thus, PI3K appears to be one of the downstream effectors of CS-induced MMP-EGFR signaling. The Akt pathway has been shown to regulate HBE cell survival and transformation promoted by NNK (5, 6), a potent carcinogen present in the CS. However, results obtained with Akt-specific pharmacological inhibitors in this study indicate that Akt has no significant effect on CS-induced transcriptional activation of FRA-1. Previous studies have shown that Akt both negatively and positively regulates expression of activator protein-1. For example, Akt inhibits the expression of c-FOS, a close relative of FRA-1, by down-regulating the ERK MAP kinase pathway, and a high level of Akt activity is inversely correlated with c-FOS expression in glioblastoma cells (39). In contrast, enhanced Akt activity is required for a high level expression of FRA-1 in prostate cancer cells (40). A different study showed Akt-dependent stimulation of c-FOS occurs through its SRE in HeLa cells (41). Thus, it appears that Akt distinctly regulates the expression of FOS family members in a cell type-and signal-specific manner.
Our data indicate that PI3K signals activate ERKs (Fig. 4A), which are essential for high level induction of FRA-1 by CS (25), tumor promoters, and mitogens in HBE cells (data not shown). Consistent with this, a prominent role for ERK signaling in mitogen-inducible FRA-1 mRNA expression has been shown in other cell types (17,(42)(43)(44). Although a requirement for PI3K pathway in mitogen-induced ERK activation has been suggested in other cells, the present study went further to demonstrate a link between CS-induced PI3K signaling and ERK activation for the first time in HBE cells (Fig. 4A). The activation of ERKs occurred via MEK1 in a PI3K-dependent manner. Consistent with our studies, the inhibition of PI3K pathway blocks platelet-derived growth factor-and epidermal growth factor-stimulated c-FOS expression via ERK signaling (45,46). The PI3K inhibitor blocked CS-induced ERK1/2 activation (Fig. 4A). Supporting this observation, several studies have shown that PI3K modulates MEK-ERK pathway at the level of PAK1 and/or Raf (47)(48)(49). PAK1 is a member of the serine/threonine kinase family that is known to control various cellular processes including cytoskeletal reorganization, cell migration, and MAP kinase signaling (50). PAK1 has been shown to facilitate the cross-talk between the PI3K and ERK signaling pathways (34). For example, PI3K may act through either Raf-1 or PAK1 to phosphorylate MEK1 (47)(48)(49), an upstream activator of ERK1/2 kinases. Furthermore the activation of PAK1 by benzo-(a)pyrene, a carcinogen found in CS, in 293T and HeLa cells has been reported (51). Our studies demonstrated that inhibition of PI3K pathway blocked the activation of PAK1 (Fig. 5), and the ectopic expression of a PAK1 mutant inhibited CS-activated c-Raf and MEK1 phosphorylation (Fig. 6). Our novel findings indicate that PAK1 mediates the crosstalk between the PI3K and ERK pathways by activating c-Raf and MEK1 kinases in response to CS in HBE cells.
Recently we have demonstrated that the Ϫ379 to ϩ32 bp genomic fragment, devoid of introns, contains sufficient information to drive FRA-1 transcription in response to tumor promoters and CS exposure in pulmonary epithelial cells (25,26,29). A Ϫ379/Ϫ237 bp bipartite enhancer comprises the Ϫ318 12-O-tetradecanoylphorbol-13-acetate response element, SRE (276/Ϫ237), and an ATF site, which are necessary for a high level induction of FRA-1 (26,29) in response to 12-Otetradecanoylphorbol-13-acetate. The recruitment of c-JUN to the Ϫ318 12-O-tetradecanoylphorbol-13-acetate response element and the binding of the Elk1, serum response factor, and CREB/ATF-1 proteins to the respective ternary complex factor, CArG, and ATF sites are necessary for FRA-1 expression (26,29). In the present study, we found that CS stimulates the phosphorylation of Elk1 and CREB transcription factors, whereas inhibitors of PI3K or ERK1/2 blocked such activation. The inhibition of P13K did not block CS-stimulated p38 phosphorylation (data not shown) suggesting that ERK1/2 alone may integrate CS-induced PI3K-dependent signals into Elk1.
Because Elk1 and CREB are also known to regulate the induction of c-FOS and FOSB (52,53), it is likely that Elk1 and CREB act as downstream targets of CS-induced ERK1/2 signaling and regulate FRA-1 induction. Consistent with this notion, CS failed to significantly stimulate transcriptional activation from the FRA-1 promoter bearing mutations in CArG box and ATF sites suggesting a critical requirement of Elk1 and ATF/CREB protein binding at these sites for the induction of FRA-1 by CS (Fig. 7C). However, ChIP (Fig. 8) and EMSA (Fig. 9) analyses showed that CS did not significantly alter the constitutive binding of both Elk1 and CREB proteins to the FRA-1 promoter. Although inhibitors of PI3K and ERK signaling did not inhibit DNA binding of Elk1 and CREB, inhibitors markedly attenuated their CS-inducible phosphorylation status (Fig. 7, A and B) and FRA-1 mRNA expression (Fig. 1). These results are consistent with our previous ChIP analyses demonstrating the constitutive binding of Elk1, serum response factor, and CREB proteins to the FRA-1 promoter (26). What then prevents the Elk1 and CREB functions in the presence of PI3K-inhibtiors? Phosphorylation of Elk1 at Ser 383 and Ser 389 has been shown to enhance the recruitment of ternary complexes at the SRE and for transcriptional activation (54). Similarly phosphorylation of CREB at Ser 133 has been shown to be required for its transcriptional activity (55). Thus, CSinduced phosphorylation of CREB and Elk1 may promote transcrip- tional activation via cofactor recruitment. Indeed the phosphorylation of Elk1 at Ser 383 by ERK has been shown to facilitate the recruitment of the transcriptional co-activator p300 (56).
In summary, based on the above observations, we propose the following model for the CS-induced expression of the FRA-1 (Fig. 10). In the unstimulated state, the Elk1 and CREB proteins are constitutively bound to the FRA-1 promoter. Upon CS stimulation, EGFR-mediated signaling activates PI3K pathway leading to both Akt and PAK1 activation (step 1). PAK1 then stimulates the phosphorylation of c-Raf/MEK/ ERK1/2 (step 2). ERKs translocate into the nucleus and phosphorylate Elk1 and CREB proteins bound to the promoter (step 3) resulting in enhanced FRA-1 transcription (step 4).