Evidence That Phosphatidylinositol 3-Kinase- and Mitogen-activated Protein Kinase Kinase-4/c-Jun NH2-terminal Kinase-dependent Pathways Cooperate to Maintain Lung Cancer Cell Survival*

Cancer cells in which the PTEN lipid phosphatase gene is deleted have constitutively activated phosphatidylinositol 3-kinase (PI3K)-dependent signaling and require activation of this pathway for survival. In non-small cell lung cancer (NSCLC) cells, PI3K-dependent signaling is typically activated through mechanisms other than PTEN gene loss. The role of PI3K in the survival of cancer cells that express wild-type PTEN has not been defined. Here we provide evidence that H1299 NSCLC cells, which express wild-type PTEN, underwent proliferative arrest following treatment with an inhibitor of all isoforms of class I PI3K catalytic activity (LY294002) or overexpression of the PTEN lipid phosphatase. In contrast, overexpression of a dominant-negative mutant of the p85α regulatory subunit of PI3K (Δp85) induced apoptosis. Whereas PTEN and Δ85 both inhibited activation of AKT/protein kinase B, only Δp85 inhibited c-Jun NH2-terminal kinase (JNK) activity. Cotransfection of the constitutively active mutant Rac-1 (Val12), an upstream activator of JNK, abrogated Δp85-induced lung cancer cell death, whereas constitutively active mutant mitogen-activated protein kinase kinase (MKK)-1 (R4F) did not. Furthermore, LY294002 induced apoptosis of MKK4-null but not wild-type mouse embryo fibroblasts. Therefore, we propose that, in the setting of wild-type PTEN, PI3K- and MKK4/JNK-dependent pathways cooperate to maintain cell survival.

Certain cancer cell types with PTEN gene loss have constitutively active PI3K and undergo apoptosis in response to pharmacologic or genetic inhibition of PI3K (24). Most nonsmall cell lung cancer (NSCLC) cell lines demonstrate hallmarks of PI3K pathway activation, such as phosphorylation of AKT and its downstream mediators, but have a wild-type PTEN gene (25)(26)(27)(28)(29)(30). Despite having wild-type PTEN, NSCLC cells undergo apoptosis in response to PI3K pathway inhibition (25). The apoptosis reported by Brognard et al. (25) may depend in part on the absence of serum, which rescues cells from apoptosis induced by PI3K inhibition (18,19,31,32). Thus, serum-induced activation of other peptide growth factor-induced signaling pathways can overcome the pro-apoptotic effect of PI3K inhibition. In this study, we investigated the signaling pathways that interact with PI3K to control NSCLC cell survival. Using pharmacologic and genetic approaches, we found that inhibition of PI3K-dependent signaling alone induced proliferative arrest, whereas inhibition of both PI3K and MKK4/JNK-dependent pathways induced apoptosis. These findings indicate that, in the setting of wild-type PTEN, PI3Kand MKK4/JNK-dependent pathways cooperate to maintain cell survival.
Apoptosis and cell cycle progression were measured by TUNEL with the APO-BRDU staining kit (Phoenix Flow Systems, San Diego, CA). Floating cells and attached cells were dispersed with trypsin-EDTA, pelleted, washed, and fixed in 1% paraformaldehyde for 15 min on ice and then fixed in 70% ethanol. The fixed cells were washed and incubated with DNA labeling solution containing terminal deoxynucleotidyltransferase reaction buffer, deoxynucleotidyltransferase enzyme, and bromodeoxyuridine triphosphate (BrdUrd-dUTP). The cells were rinsed before being resuspended with fluorescein-PRB-1 antibody solution and analyzed by flow cytometry in the presence of propidium iodide/RNase solution. Analyses of 3,000 to 10,000 events were done with a FACScan flow cytometer (BD Pharmingen) equipped with a 488-nm argon ion laser and two software packages: CellQuest 3.1 (BD Pharmingen) and ModFit LT 2.0 (Verity Software House, Topsham, ME). Live gating of the forward and orthogonal scatter channels was used to exclude debris and to selectively acquire cell events. A dual display of DNA area (linear red fluorescence) and BrdUrd-dUTP incorporation (FITC-PRB-1) was used to determine the percentage of propidium iodine-stained cells that were apoptotic.
Apoptosis was also determined by the detection of nucleosomal DNA fragmentation by using the TACS apoptotic DNA laddering kit (Trevigen, Inc., Gaithersburg, MD) according to the manufacturer's protocol. Briefly, DNA was isolated from cells after adenovirus transfection or LY294002 treatment by incubating them in lysis buffer. DNA samples were subjected to electrophoresis on a 1.5% agarose gel and visualized by ethidium bromide staining.
Immune Complex Kinase Assay-H1299 cells were incubated for 2 h with Ad5-CMV, Ad5-⌬p85, or Ad5-PTEN at 1 ϫ 10 3 , 5 ϫ 10 3 , or 1 ϫ 10 4 p/cell in serum-free conditions, changed to complete medium, and incubated for 48 h. Cells were then washed twice in 1ϫ phosphate-buffered saline, serum-starved for 24 h, treated with 50 ng/ml EGF for 15 min, and lysed in lysis buffer. Extracts were subjected to immunoprecipitation (100 g) with antibodies to JNK1, AKT1/2, or ERK1/2 by rotation at 4°C overnight. Protein A-G-agarose beads (20 l) were added, and the solution was incubated at 4°C for 1 h. The beads were washed three times with lysis buffer and once with kinase buffer (20 mM Hepes (pH 7.5), 20 mM ␤-glycerol phosphate, 10 mM MgCl 2 , 1 mM dithiothreitol, and 50 mM sodium orthovanadate). Kinase assays were performed by incubating the beads with 30 l of kinase buffer, to which 20 M cold ATP, 5 Ci of [␥-32 P]ATP (2,000 cpm/pmol), and 2 g of GST-c-Jun, GST-GSK3␤, or MBP as substrates were added. The kinase reaction was performed at 30°C for 20 min. The samples were then suspended in 1ϫ Laemmli buffer and boiled for 5 min, and the samples were analyzed by 12% SDS-PAGE. The gel was dried and autoradiographed.
Immune complex assays were also performed with COS-7 cells, which were transiently transfected for 6 h with 5 g of plasmids containing Rac-1 (Val 12 ), MKK1 (R4F), or empty vector using FuGENE. The cells were then transfected with Ad5-⌬p85 or Ad5-CMV (1 ϫ 10 3 or 5 ϫ 10 3 particles/cell) and incubated in complete medium for 24 h. The cells were then changed to serum-free medium for 24 h, treated with IGF-1 (50 ng/ml) for 15 min, and lysed. JNK and ERK were immunoprecipitated from 100 g of total cell lysates and subjected to kinase assays using GST-c-Jun and MBP, respectively, as substrates.
Rac-1 Activity Assays-Pull-down assays with GST-tagged p21 binding domain (PBD) of PAK-1 were performed as follows. COS-7 cells were co-transfected with 2 g of HA-tagged p85␣, HA-tagged p110␣, and ⌬p85 (2, 4, or 6 g) using LipofectAMINE (Invitrogen). Total amount of DNA transfected per plate was equalized with empty vector. After 6 h, transfectants were washed and changed to normal growth medium. After 24 h, transfectants were serum-starved for 16 h, treated with 50 ng/ml EGF or IGF-1 for 15 min, and lysed. PAK-1 PBD-agarose (5 g in a 50% slurry) was added to the lysates and the mixture was incubated for 1 h at 4°C. The bead pellet was collected by centrifugation (5 s at 14,000 ϫ g) and the supernatant was drained off. The beads were then washed and suspended in 20 l of 1ϫ Laemmli sample buffer. Proteins were separated by 12% SDS-PAGE, transferred to nitrocellulose membrane, and blotted against Rac-1 and CDC42 polyclonal antibodies.

PI3K-dependent Pathway Contributes to NSCLC Cell
Proliferation and Survival-We investigated the effects of PI3K inhibition on the proliferation and viability of H1299 NSCLC cells, which have a wild-type PTEN gene (27). H1299 cells were transfected with recombinant adenoviruses that express PTEN (Ad5-PTEN) or ⌬p85 (Ad5-⌬p85), a p85␣ dominant-negative mutant lacking the inter-SH2 residues required for binding to the p110 catalytic domain (35). Transfection of H1299 cells with Ad5-PTEN or Ad5-⌬p85 increased the expression of the adenoviral gene products and suppressed pAKT levels (Fig. 1), providing evidence that these adenoviral vectors effectively blocked PI3K-dependent signaling. When H1299 cells were incubated with Ad5-PTEN or Ad5-⌬p85, cell number decreased in a dose-dependent fashion (Fig. 2, A and B). H1299 cell number also decreased in a dose-dependent manner after treatment with LY294002, a competitive inhibitor of ATP binding to all isoforms of class I PI3K (Fig. 2C). Other NSCLC cell lines with wild-type PTEN (H358, Calu-6, H460, H661, H226B, H441, H1299, and H226Br) underwent a similar decrease in cell number following treatment with LY294002 or transfection with Ad5-⌬p85 or Ad5-PTEN (data not shown).
We next investigated whether PI3K inhibition induced proliferative arrest or apoptosis of NSCLC cells by performing flow cytometric analysis of H1299 cells transfected with Ad5-PTEN or Ad5-⌬p85 and then stained with propidium iodide (Fig. 3A). Ad5-PTEN transfection induced proliferative arrest in the G 0 /G 1 phase of the cell cycle, with minimal evidence of programmed cell death, as shown by the lack of a hypodiploid peak. Although Ad5-⌬p85 transfection also caused an accumulation of cells in G 1 , its most striking effect was apoptosis, as indicated by the appearance of a hypodiploid peak. We examined this finding further by using terminal deoxynucleotidetransferase nick-end labeling (TUNEL), a more sensitive assay for apoptosis, and found low levels of DNA fragmentation in cells transfected with Ad5-PTEN (Fig. 3B). In contrast, transfection with Ad5-⌬p85 produced much more DNA fragmentation, which is compatible with the induction of high levels of apoptosis (Fig. 3B).
We investigated the effect of Ad5-PTEN and Ad5-⌬p85 on signaling events known to contribute to apoptosis, proliferative arrest, or both (Fig. 4). Ad5-⌬p85 transfection reduced the levels of procaspase-9, procaspase-3, and poly(ADP-ribose) polymerase, demonstrating evidence of caspase activation and proteolysis of a caspase-3 substrate. In contrast, Ad5-PTEN transfection decreased CDK2 levels and increased p27 CDK inhibitor levels without evidence of caspase activation or poly-(ADP-ribose) polymerase cleavage. Together, these findings support a role for PI3K in the proliferation of NSCLC cells and demonstrate a pro-apoptotic effect of ⌬p85.
⌬p85 Inhibits the Activity of MAP Kinases-⌬p85 induced apoptosis of NSCLC cells whereas PTEN did not. Therefore, we hypothesized that inhibition of the PI3K/AKT pathway was required but not sufficient to induce apoptosis. We sought to identify additional survival signals typically activated by peptide growth factors that are inhibited by ⌬p85. Receptor tyrosine kinases maintain NSCLC cell survival, in part, by activating MAP kinases (37). We investigated the role of MAP kinases in ⌬p85-induced cell death. H1299 NSCLC cells were incubated with Ad5-PTEN or Ad5-⌬p85, treated with EGF, and subjected to in vitro kinase assays of JNK and ERK activity (Fig. 5). ERK activity increased in cells incubated with Ad5-CMV. Ad5-PTEN and Ad5-⌬p85 had similar, dose-dependent effects on ERK activity. Relative to the effect of Ad5CMV, ERK activity increased with low dose (10 3 particles/cell) and decreased with high dose (5 ϫ 10 3 or 10 4 particles/cell) Ad5-PTEN or Ad5-⌬p85. JNK activity decreased minimally after Ad5-PTEN and, to a much greater extent, after Ad5-⌬p85 incubation. Thus, ⌬p85 was unique in its ability to inhibit JNK activity.
We investigated the mechanism by which ⌬p85 inhibited JNK. p85␣ associates with Rac-1, an upstream activator of JNK, and activates Rac-1 through association with a multiprotein complex that binds to p85 SH2 domains (38). We investi- Ad5-CMV or treated with medium alone (Ϫ). The cells were then serumstarved for 24 h, treated with EGF (50 ng/ml) for 15 min, and lysed. JNK and ERK were immunoprecipitated from 100-g aliquots of the cell lysate and subjected to immune complex kinase assays (KA) using GST-c-Jun and MBP as substrates to examine the activities of JNK and ERK, respectively. As a control, JNK1 and ERK1/2 expression were examined by Western blot analysis (W) of whole cell lysates.

FIG. 6. Effect of ⌬p85 on Rac-1 activation by peptide growth factors.
A, COS-7 cells were treated for 15 min with 50 ng/ml IGF-1, EGF, or medium alone. B, COS-7 cells were co-transfected overnight with plasmids containing HA-p85 and increasing doses of ⌬p85, serumstarved for 16 h, and treated with IGF-1 for 15 min. Cells were lysed, and Rac-1 activity was analyzed by performing a pull-down assay on cell extracts using a GST-tagged PBD of PAK-1, followed by Western analysis (W) to quantitate PAK-1-associated Rac-1 or CDC42. To examine relative transfection efficiencies in each sample, Western analysis of HA expression was performed on total cell extracts. gated whether wild-type p85␣ and ⌬p85 differ in their ability to activate Rac-1. We quantitated Rac-1 activity in cell extracts using a pull-down assay with a GST-tagged PBD of PAK-1, which associates selectively with GTP-bound (activated) Rac-1 or CDC42. PBD-associated proteins are subjected to Western analysis to quantitate Rac-1 and CDC42. We performed this experiment in COS-7 cells, in which peptide growth factors activate Rac-1 through a PI3K-dependent mechanism (21,22). Using this assay we showed that Rac-1 is activated by treatment with EGF or IGF-1 (Fig. 6A). COS cells were co-transfected with wild-type p85 and increasing amounts of ⌬p85 and treated with EGF to activate Rac-1. Relative to the effect of wild-type p85, ⌬p85 inhibited peptide growth factor-induced activation of Rac-1 but not CDC42 (Fig. 6B). Thus, in contrast to the stimulatory effect of p85␣, ⌬p85 inhibited Rac-1.
p85␣ serves both to stabilize p85 protein and to inactivate PI3K lipid kinase activity (39). Therefore, we tested the hypothesis that Ad5-⌬p85 inhibits intracellular signaling activity by increasing intracellular p85 protein levels. We incubated H1299 NSCLC cells with various doses of Ad5-⌬p85 or an adenoviral vector expressing full-length p85␣ (Adex1CAp85␣-HA) and examined their relative effects on EGF-induced phosphorylation and activation of AKT, JNK, and ERK by Western blotting and in vitro kinase assays (Fig. 7). We measured the intensity of pAKT, GST-GSK3, total AKT, and p85 bands by densitometric scanning and corrected for differences in total p85 and AKT protein levels at each virus dose. pAKT levels in cells transfected with 10 3 , 5 ϫ 10 3 , and 10 4 particles/ cell of Ad5-⌬p85 were 20.6, 74.6, and 56.7%, respectively, of pAKT levels in cells transfected with the same doses of Adex1CAp85␣-HA (Fig. 7, top). In vitro kinase assays using GST-GSK3 as substrate demonstrated that Ad5-⌬p85 inhibited AKT kinase activity to a greater extent than Adex1CAp85␣-HA (Fig. 7, top). In contrast to Ad5-⌬p85, Adex1CAp85␣-HA increased the phosphorylation and activity of JNK and minimally increased ERK activity (Fig. 7, middle and bottom). Treatment with LY294002 did not block the effects of Adex1CAp85␣-HA on JNK and ERK (data not shown), providing evidence that PI3K activity was not required. These findings indicate that increasing the levels of wild-type p85␣ was not sufficient to recapitulate the effects of ⌬p85 on AKT, JNK, and ERK.
MAP Kinase Signaling Contributes to NSCLC Cell Survival-We investigated the importance of JNK and ERK inhibition in ⌬p85-induced cell death by examining whether co-transfection of upstream activators of these kinases would block ⌬p85-induced cell death. In COS-7 cells, a constitutively active mutant Rac-1 (Val 12 ) blocked ⌬p85-induced inhibition of JNK and ERK, whereas a constitutively active mutant MAPK/ERK kinase (MKK1) (R4F) blocked ⌬p85-induced inhibition of ERK but not JNK (Fig. 8A). In H1299 NSCLC cells, Ad5-⌬p85induced cell death was abrogated by Rac-1 (Val 12 ) but not by MKK1 (R4F) (Fig. 8, B and C). Together, these findings indicate that ⌬p85-induced cell death requires inhibition of JNK but not ERK.
On the basis of these findings, we hypothesized that Rac-1 and its downstream mediators (PAK-1/MKK4/JNK) cooperate with PI3K-dependent signaling to maintain cell survival. To test this hypothesis, we examined whether PI3K inhibition would be sufficient to induce apoptosis of MKK4-null MEF cells. These cells did not activate JNK in response to EGF (Fig.  9A). We treated MKK4-null and wild-type MEF cells with LY294002 and examined them for evidence of apoptosis. LY294002 treatment induced apoptosis in MKK4-null MEF cells, but wild-type MEF cells demonstrated minimal evidence of cell death (Fig. 9, B and C). Together, these findings support the hypothesis that PI3K-and MKK4-dependent pathways cooperate to maintain cell survival.

DISCUSSION
Cancer cells with PTEN gene loss require activation of the PI3K-dependent pathway for survival. However, PI3K-dependent signaling is activated in cancer cells through mechanisms other than PTEN gene loss, as demonstrated in NSCLC cells, which typically have evidence of PI3K pathway activation and express wild-type PTEN (25)(26)(27)(28)(29)(30). Here we investigated the role of PI3K in the survival of a NSCLC cell line that expresses wild-type PTEN. We found that LY294002 treatment and PTEN overexpression arrested cells in the G 0 /G 1 phase of the cell cycle with minimal evidence of apoptosis. Unexpectedly, we found that ⌬p85 induced marked apoptosis.
Several lines of evidence presented here support the hypothesis that ⌬p85 induced apoptosis through the combined inhibition of MAP kinase-and PI3K-dependent pathways. First, in addition to inhibiting AKT, ⌬p85 inhibited JNK activity. Second, constitutively active Rac-1 (Val 12 ) blocked ⌬p85-induced apoptosis of H1299 cells. Third, LY294002 treatment induced apoptosis in MKK4-null MEF cells but not wild-type MEF cells. The cooperative effect between PI3K and MAP kinase pathways was specific to JNK-dependent signaling, as introduction of constitutively active mutant MKK1 did not block apoptosis of H1299 cells induced by ⌬p85. This finding is consistent with previous reports that MAP kinase family members play distinct biological roles in tumor cells (40,41). Together, these findings indicate that, in the setting of wild-type PTEN, PI3Kand MKK4/JNK-dependent pathways cooperate to maintain cell survival.
Mechanisms by which ⌬p85 inhibited JNK activity have not been fully defined. Introduction of wild-type p85␣ did not recapitulate the effects of ⌬p85 on JNK and AKT, suggesting that FIG. 7. Effect of wild-type p85␣ and ⌬p85 on AKT, JNK, and ERK. H1299 cells were incubated with no virus (lanes designated Ϫ and EGF), empty vector (Ad5) at 10 4 particles/cell, Adex1CAp85␣-HA (Ad-WTp85), which expresses wild-type p85␣, or Ad5-⌬p85 at 10 3 , 5 ϫ 10 3 , or 10 4 particles/cell. After transfection, the cells were grown in complete medium at 37°C for 48 h. The cells were then subjected to serum-free conditions for 12 h (Ϫ), treated for 15 min with EGF (50 ng/ml), and lysed. Lysates were subjected to either Western blot analysis (20 g/sample) using antibodies to the indicated proteins or immunoprecipitation (100 g/sample) to isolate AKT, JNK, and ERK for in vitro kinase assays using GST-GSK3, GST-c-Jun, and MBP, respectively, as substrates.
⌬p85 functions through mechanisms other than increasing intracellular levels of p85␣, which inactivates PI3K lipid kinase activity through changes in the stoichiometry of p85:p110 (39). Alternatively, ⌬p85 may inhibit recruitment of p85-associated proteins required for activation of MAP kinase pathways by receptor tyrosine kinases. Supporting this possibility, we found that ⌬p85 inhibited Rac-1 activation by peptide growth factors. The inhibitory effect of ⌬p85 on peptide growth factor-induced Rac-1 activation in lung cancer cells is not consistent with previous reports that overexpression of ⌬p85 stimulates Rac-1 activity in T lymphocytes (42). Although we have yet to identify the mechanism by which ⌬p85 inhibits Rac-1, cell type-specific factors may be important. Recent findings indicate that p85 activates Rac-1 by associating with a multiprotein complex (including Eps8, Abi1, and SOS-1) that binds to p85 SH2 domains (38). We hypothesize that, in certain cell types, the p85 inter-SH2 domain is also required for Rac-1 activation. Recent findings have shown that this region of p85 contains several motifs that, in addition to binding to the p110 catalytic subunit, interact with other factors regulated by GTPase-and tyrosine kinase-dependent pathways (43), supporting the possibility that another multiprotein complex associates with this region. Additional studies will be needed to identify these proteins and to examine their role in Rac-1 activation.
Although adenoviral vectors expressing PTEN and ⌬p85 shared the ability to inhibit AKT activity, they differed in other downstream signaling events. Ad5-PTEN increased p27 levels and decreased CDK2 levels, which has been described previously in cells transfected with adenoviral vectors expressing PTEN (18,44) and is consistent with the G 0 /G 1 proliferative arrest we observed. In contrast to Ad5-PTEN, Ad5-⌬p85 did not increase p27 or decrease CDK2 levels. This finding was somewhat surprising, given the ability of ⌬p85 to inhibit PI3K-dependent signaling. Analysis of the dose-dependent effects of Ad5-PTEN and Ad5-⌬p85 demonstrated that Ad5-PTEN was a more potent inhibitor of pAKT levels than was Ad5-⌬p85, which could account for their differences in downstream signaling. The adenoviral dose-dependent changes in ERK activity we observed in NSCLC cells transfected with Ad5-PTEN differs from observations in glioblastoma and prostate cancer cells transfected with Ad-PTEN and in PTEN-null embryonic stem cells, in which ERK activity did not change (19,(45)(46)(47). This difference could be the result of cell type-specific factors or nonspecific effects of exogenous PTEN in H1299 cells.
Previous studies have shown that MKK4 expression and activity are altered in human tumor cells and that MKK4 can act as both a promoter and a suppressor of human tumorigenesis. The MKK4 gene is deleted or mutated in a subgroup of pancreatic, biliary, and breast carcinomas, and reintroduction of MKK4 inhibits the metastatic ability of certain tumor cells,

FIG. 8. Effects of constitutively active mutants of MKK1 (R4F) and
Rac-1 (Val 12 ) on Ad5-⌬p85-induced MAP kinase inhibition and apoptosis. A, COS-7 cells were transiently transfected for 6 h with 5-g plasmids containing Rac-1 (Val 12 ), MKK1 (R4F), or empty vector using FuGENE. The cells were then transfected with Ad5-⌬p85 or Ad5-CMV (1 ϫ 10 3 or 5 ϫ 10 3 particles/ cell) and incubated for 24 h in complete medium. The cells were then changed to serum-free medium for 24 h, treated with IGF-1 (50 ng/ml) for 15 min, and lysed. JNK and ERK were immunoprecipitated from 100-g aliquots of total cell lysates and subjected to kinase assays (KA) using GST-c-Jun and MBP, respectively, as substrates. As a control, JNK1 and ERK1/2 levels were examined by Western blot analysis (W). B, H1299 cells were transfected with 5 g of plasmids containing Rac-1 (Val 12 ), MKK1 (R4F), or empty vector (Ϫ). The next day, the cells were incubated with Ad5-CMV at 5 ϫ 10 3 particles/ cell (Ϫ) or Ad5-⌬p85 at 1 ϫ 10 3 or 5 ϫ 10 3 particles/cell. Two days later, floating and adherent cells were isolated, fixed in 1% paraformaldehyde and 70% ethanol, stained with propidium iodide and APO-BrdUrd, and subjected to flow cytometric analysis. The percentage of dead cells was determined by quantification of the pre-G 1 cell population and is indicated in the upper right corner of each flow diagram. C, H1299 cells were transfected with Rac-1 (Val 12 ) and then incubated with increasing amounts of Ad5-⌬p85, as described in panel b. Genomic DNA was isolated from floating and adherent cells and subjected to 1.5% gel electrophoresis for DNA fragmentation analysis.
demonstrating that MKK4 has tumor suppressor activity (48 -50). Potentially mediating this effect, Ras pathway activation increases the expression of p53 and p16 INK4a , which induces premature cellular senescence; conversely, inactivation of p53 or p16 prevents Ras-induced growth arrest (51). In contrast to these studies, MKK4 is known to be a downstream mediator of Rac-1, and Rac-1 activation contributes to Ras-induced cellular transformation (20), indicating that MKK4 plays a role in cellular transformation. Supporting the latter hypothesis, we found that Rac-1 activation rescued lung cancer cells from ⌬p85-induced apoptosis, and MKK4 cooperated with PI3K to maintain MEF cell survival. This finding supports in vitro studies of lung cancer cells demonstrating that MKK4-depend-ent pathways play a dominant role in mutant Ras-induced colony formation (52,53). Thus, MKK4 and its downstream mediators play apparently contradictory roles in the regulation of cellular growth and transformation that may depend on the presence of cell type-specific factors or the activity of tumor suppressor pathways that inhibit the mitogenic and transforming effects of MKK4.
Findings presented here have implications for the design of effective therapeutic approaches for lung cancer. Signal transduction inhibitors are being assessed in clinical trials as therapeutic agents for several types of cancer. The enthusiasm for these agents has been fueled by the efficacy of ABL kinase inhibitors in the treatment of chronic myelogenous leukemia, which arises from a reciprocal chromosomal translocation involving the Bcr and Abl genes (54). However, unlike chronic myelogenous leukemia, in which constitutively active Abl is sufficient to induce the disease (55), lung tumorigenesis is a multistep process leading to aberrant activity of a variety of oncogenic and tumor suppressive pathways. These pathways act in combination to induce malignant transformation of NHBE cells and to maintain the survival of lung cancer cells (56). Thus, in patients with lung cancer, inhibition of multiple pathways may be necessary to induce tumor regression. The findings presented here support the hypothesis that PI3K-and MKK4/JNK-dependent pathways cooperate in lung cancer cells to maintain their survival, and combination therapy targeting these pathways should be considered in future clinical trials.