Requirement of c-Jun NH2-Terminal Kinase for Ras-Initiated Tumor Formation

ABSTRACT The c-Jun NH2-terminal kinase (JNK) signal transduction pathway causes increased gene expression mediated, in part, by members of the activating transcription factor protein (AP1) group. JNK is therefore implicated in the regulation of cell growth and cancer. To test the role of JNK in Ras-induced tumor formation, we examined the effect of compound ablation of the ubiquitously expressed genes Jnk1 plus Jnk2. We report that JNK is required for Ras-induced transformation of p53-deficient primary cells in vitro. Moreover, JNK is required for lung tumor development caused by mutational activation of the endogenous KRas gene in vivo. Together, these data establish that JNK plays a key role in Ras-induced tumorigenesis.

Jnk2 Ϫ/Ϫ mice). In contrast, studies of glioblastoma, prostate cancer, and lung carcinoma cell lines have identified important roles for JNK2 (but not JNK1) (6,13,41,53). Together, these data confirm that JNK may play a role in cancer development, but the relative roles of the JNK1 and JNK2 isoforms are unclear.
It is established that the Jnk1 and Jnk2 genes exhibit partially redundant functions (14,15,30,46). Studies of compound mutants with ablation of both ubiquitously expressed Jnk genes are therefore required. These studies have been compromised by the early embryonic lethal phenotype of Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ mice (34). Nevertheless, Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ murine embryonic fibroblasts (MEF) have been prepared. These cells exhibit an early senescence phenotype that is caused by engagement of the p53 tumor suppressor pathway (14,46). Immortalized wildtype and JNK-deficient fibroblasts have been isolated using the 3T3 protocol that disrupts the p53 pathway (32). Retroviral transduction assays with ectopic expression of activated Ras demonstrated that both wild-type (WT) and JNK-deficient 3T3 cells can exhibit properties of transformed cells (32). JNK may therefore not be required for Ras-induced transformation, but this conclusion is subject to a number of important caveats. It is likely that the 3T3 immortalization protocol does not cause the same genetic alterations in both WT MEF and Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ MEF. Moreover, these studies of 3T3 cells expressing ectopic Ras may not be informative for understanding the mechanism of epithelial cell tumor formation by mutational activation of the endogenous KRas gene.
The purpose of this study was to test the requirement of JNK for Ras-induced tumorigenesis. First, we examined whether JNK is required for the Ras-induced transformation of primary MEF with a defined genetic background. Second, we examined the requirement of JNK for epithelial cell transformation using an established model of lung cancer in mice. We report that JNK is critically required for Ras-induced transformation of MEF in vitro and for Ras-induced lung tumor formation in vivo.

Mice.
We have described Jnk1 Ϫ/Ϫ mice (19), Jnk2 Ϫ/Ϫ mice (52), mice with conditional expression of Jnk1 (14), and mice with conditional expression of KRas G12D (29). Mice with Trp53 gene ablation (18) were provided by Stephen Jones, University of Massachusetts Medical School. Mice with conditional expression of Trp53 (36) were obtained from the Jackson Laboratory (strain B6.129P2-Trp53 tm1Brn /J [stock number 008462]). Mice expressing 4-hydroxy-tamoxifen-stimulated Cre (2) were obtained from the Jackson Laboratory [strain B6;129-Gt(ROSA)26Sor tm1(cre/ERT)Nat /J (stock number 004847)]. The mice used in this study were backcrossed to the C57BL/6J strain (Jackson Laboratories) and were housed in a facility accredited by the American Association for Laboratory Animal Care (AALAC). The Institutional Animal Care and Use Committee (IACUC) of the University of Massachusetts approved all studies using animals.
Genotype analysis. Genotype analysis was performed by PCR using genomic DNA as the template. The Jnk1 LoxP (1.1-kb) and Jnk1 ⌬ (0.4-kb) alleles were identified using the amplimers 5Ј-CCTCAGGAAGAAAGGGCTTATTTC-3Ј FIG. 1. Growth properties of Trp53 Ϫ/Ϫ Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ embryonic fibroblasts. (A) Loss of contact growth inhibition in Trp53 Ϫ/Ϫ MEF cultures is prevented by disruption of Jnk1 plus Jnk2. Wild-type (WT), Trp53 Ϫ/Ϫ , and Trp53 Ϫ/Ϫ Jnk Ϫ/Ϫ MEF were examined using phase-contrast microscopy. Representative images are shown of cells growing in sparse (top) or confluent (bottom) culture conditions. (B) Growth of Trp53 Ϫ/Ϫ cells in low serum concentrations is prevented by disruption of Jnk1 plus Jnk2. WT, Trp53 Ϫ/Ϫ , and Trp53 Ϫ/Ϫ Jnk Ϫ/Ϫ MEF were plated on day 0 (D ϭ 0) and cultured for 7 days (D ϭ 7) in medium supplemented with different concentrations of serum. Relative cell numbers were measured by staining with crystal violet. The data are presented as means Ϯ standard deviations (n ϭ 5) and are representative of results from three independent experiments. Statistically significant differences (P Ͻ 0.01) between WT MEF and mutant MEF are indicated with an asterisk. (C, D) Soft agar colony formation by Trp53 Ϫ/Ϫ MEF requires JNK. Cells were grown in soft agar supplemented with 10% serum. Colony formation was assessed after 20 days. (C) Representative images are shown of colonies formed by WT, Trp53 Ϫ/Ϫ , and Trp53 Ϫ/Ϫ Jnk Ϫ/Ϫ cells. The number of colonies (Ͼ10 cells) formed by each cell type was counted and is shown in panel D. Statistically significant differences (P Ͻ 0.01) between WT MEF and mutant MEF are indicated with an asterisk.
Cell culture. Primary mouse fibroblasts (embryonic day 11) were prepared and cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 1% penicillin-streptomycin, and 1% glutamine (Invitrogen). All studies were performed using low-passage primary MEF (Ͻ4 passages). Cre-mediated gene ablation was performed by treating MEF with 1 M 4-hydroxytamox- . Soft agar assays were performed using methods described previously (32). Cell number was measured using a hemocytometer and by staining with crystal violet (32). Retroviral transduction assays. The retroviral vectors pWZL-Blast-E-cadherin and pWZL-Blast-dn-cadherin (39) were obtained from Addgene (numbers 18800 and 18804). A human Nox4 cDNA (10) was cloned by blunt-end ligation into the EcoRI site of the retroviral vector pWZL-Blast. Recombinant retroviruses were prepared and employed for transduction assays in MEF (35) by using selection with 5 g/ml blasticidin (6 days).
ELISA. Multiplexed enzyme-linked immunosorbent assay (ELISA) was performed using a Luminex 200 machine (Millipore). Cell lysates were probed using the BioPlex phosphoprotein and total target assay kit to measure phosphorylated and total JNK, extracellular signal-regulated kinase (ERK), p38, and AKT (Bio-Rad).
Immunofluorescence analysis. Cells grown on coverslips were washed with PBS, fixed in 4% paraformaldehyde ( (Vector Labs). Images were examined using a Leica TCS SP2 confocal microscope. DNA fragmentation assays. Genomic DNA fragmentation was measured using the cell death detection ELISA PLUS kit (Roche) according to the manufacturer's recommendations (35).
Analysis of tissue sections. Histology was performed using tissue fixed in 10% formalin for 24 h, dehydrated, and embedded in paraffin. Sections (7 m) were cut and stained using hematoxylin and eosin (Biocare Medical). Immunofluorescence analysis was performed using deparaffinized sections treated with the endogenous biotin-blocking kit (Invitrogen), staining (4°C, 12 h) with biotinconjugated anti-PCNA, and incubation (25°C, 1 h) with Alexa Fluor 633-conjugated streptavidin. The sections on coverslips were washed and mounted on slides using VectaShield medium containing DAPI (Vector Labs.). Images were examined using a Leica TCS SP2 confocal microscope.
Microscopic examination of exponentially growing cultures of WT, Trp53 Ϫ/Ϫ , and Jnk Ϫ/Ϫ Trp53 Ϫ/Ϫ primary MEF showed no major differences in cell morphologies (Fig. 1A). However, these cultures were markedly different when grown to high density. The WT MEF formed a growth-arrested monolayer at confluence (Fig. 1A). In contrast, the Trp53 Ϫ/Ϫ MEF did not exhibit contact growth inhibition and formed densely packed cultures with multiple layers (Fig. 1A). The Trp53 Ϫ/Ϫ overgrowth phenotype was strongly suppressed in confluent cultures of Jnk Ϫ/Ϫ Trp53 Ϫ/Ϫ MEF (Fig. 1A). This conclusion was confirmed by measurement of cell numbers (Fig. 1B). Moreover, the effect of p53 deficiency to increase cell growth in low concentrations of serum was suppressed by JNK deficiency (Fig. 1B). To test the role of JNK in three-dimensional cultures of p53-deficient MEF, we examined the growth of cells in soft agar. WT cells did not grow in soft agar, but Trp53 Ϫ/Ϫ MEF formed small colonies ( Fig.  1C and D). JNK deficiency suppressed this effect of p53 deficiency to cause growth in soft agar ( Fig. 1C and D).
Together, these data demonstrate that the effect of p53 deficiency to cause loss of contact growth inhibition and proliferation in soft agar requires JNK. JNK is required for Ras-induced suppression of contact growth inhibition. We employed a conditional gene ablation strategy using primary MEF prepared from Rosa26-Cre ERTϩ/Ϫ FIG. 4. JNK suppresses stress-induced apoptosis in p53-deficient MEF. (A) MEF were exposed to the indicated doses of UV radiation. Lysates were prepared at 4 h postirradiation, and genomic DNA fragmentation was measured by ELISA. Statistically significant differences (P Ͻ 0.01) between WT MEF and mutant MEF are indicated with an asterisk. (B) MEF were exposed to UV radiation (60 J/m 2 ). Relative cell numbers were measured by staining with crystal violet. The data are presented as means Ϯ standard deviations (n ϭ 3) and are representative of results from three independent experiments. Statistically significant differences (P Ͻ 0.01) between WT MEF and mutant MEF are indicated with an asterisk. (C) MEF were exposed to UV (60 J/m 2 ). At 6 h postirradiation, the cells were replated at the indicated dilutions. Colony formation was assessed after 10 days in culture by staining with crystal violet. Representative images are shown. The mean number of colonies per dish (n ϭ 5) is indicated. (D) Total cell lysates were prepared from MEF at the indicated times after exposure to UV radiation (60 J/m 2 ). The amount of PARP and caspase 3 was assessed by immunoblot analysis using antibodies that detect both the cleaved (designated with an asterisk) and full-length forms of these proteins. The expression of ␣-tubulin was examined to confirm equal loading of each lane.  Fig. 2A). Together, these data demonstrate that JNK is required for Ras-induced suppression of contact growth inhibition.
Loss of E-cadherin has been associated with reduced cellcell adhesion and accelerated progression to carcinoma (40). Increased E-cadherin expression by JNK-deficient cells ( Fig.  2B and C) may therefore contribute to the increased contact growth inhibition of Ras-transformed MEF in vitro ( Fig. 2A). To test this hypothesis, we examined the effect of increased expression of E-cadherin in control MEF and the expression of dominant negative (dn) E-cadherin in JNK-deficient MEF (Fig. 3A). These changes in E-cadherin did not alter cell morphology at confluence (Fig. 3B) or proliferation (Fig. 3C). However, increased expression of E-cadherin did suppress the growth of control Ras-transformed MEF in soft agar (Fig. 3D). Ras-transformed JNK-deficient MEF grew poorly in soft agar, but expression of dn-E-cadherin did not increase soft agar growth (Fig. 3D). Together, these data indicate that E-cadherin expression may only partially contribute to the phenotype of JNK-deficient MEF transformed by Ras. The mechanism that accounts for the increased expression of E-cadherin by Ras-transformed JNK-deficient MEF is unclear. We detected increased expression of E-cadherin (Cdh1) mRNA by Ras-transformed JNK-deficient MEF compared with that by Ras-transformed control MEF (Fig. 2D). This observation indicates that JNK may function to repress Ecadherin gene expression. Previous studies have established that repressors, including Snail (Snai1) and Slug (Snai2), downregulate E-cadherin expression (4,8,24). Indeed, we found that the increased expression of E-cadherin by Rastransformed JNK-deficient MEF was associated with significantly decreased expression of Slug (Fig. 2D). It is therefore possible that JNK-regulated expression of Slug may contribute to increased E-cadherin expression by Ras-transformed JNKdeficient MEF.
Role of p53 in JNK-mediated apoptosis. It is established that the JNK signaling pathway contributes to the apoptotic response to some insults by activating the intrinsic (mitochondrial) pathway (46). The mechanism is mediated, in part, by phosphorylation of members of the Bcl2-related protein family, including Bim, Bmf, and Mcl-1 (26,27,38). This signaling pathway may influence tumor development (49).
A role for p53 in JNK-dependent apoptosis has been proposed (7,20,21). To test the requirement of p53 in JNKmediated apoptosis, we examined the effect of UV radiation on JNK-and p53-deficient MEF (Fig. 4). UV-stimulated genomic DNA fragmentation assays demonstrated that both p53 Ϫ/Ϫ MEF and Jnk Ϫ/Ϫ Trp53 Ϫ/Ϫ MEF exhibited resistance to UV radiation (Fig. 4A). However, time course analysis of cell survival demonstrated that Trp53 Ϫ/Ϫ MEF were partially resistant to UV radiation and that Jnk Ϫ/Ϫ Trp53 Ϫ/Ϫ MEF exhibited significantly increased resistance to UV radiation (Fig. 4B). This conclusion was confirmed by the analysis of colony formation following exposure to UV radiation (Fig. 4C). These data demonstrate that JNK can regulate apoptosis independently of p53. To obtain biochemical evidence to support this conclusion, we examined UV-stimulated activation of caspase 3 and cleavage of the caspase substrate PARP (Fig. 4D). We found that p53 deficiency did not prevent caspase 3 activation, but caspase 3 activation was strongly suppressed in Jnk Ϫ/Ϫ Trp53 Ϫ/Ϫ MEF. These data demonstrate that p53 is not required for JNK-dependent apoptosis. JNK deficiency may therefore contribute to tumor development in a p53-deficient genetic background by suppression of stress-induced apoptosis. JNK deficiency suppresses Ras-stimulated transformation in vitro. The Trp53 Ϫ/Ϫ and Jnk Ϫ/Ϫ Trp53 Ϫ/Ϫ MEF represent genetically defined primary cells that can be employed to test the requirement of JNK for transformation by Ras in vitro (Fig.  4). Control studies confirmed the presence of a functional JNK signaling pathway in Trp53 Ϫ/Ϫ MEF but not in Jnk Ϫ/Ϫ Trp53 Ϫ/Ϫ MEF (Fig. 5A). Microscopic examination demonstrated that activated Ras caused p53 Ϫ/Ϫ MEF to exhibit a spindle-shaped morphology, but this morphological change was reduced in Jnk Ϫ/Ϫ p53 Ϫ/Ϫ MEF (Fig. 5B). Measurement of growth in response to different concentrations of serum demonstrated that activated Ras increased the saturation density of both Trp53 Ϫ/Ϫ and Jnk Ϫ/Ϫ Trp53 Ϫ/Ϫ MEF, although the JNK-deficient MEF exhibited less growth (especially in low serum concentrations) than p53 Ϫ/Ϫ MEF (Fig.  5C). This conclusion was confirmed by direct measurement of cell proliferation in low (1%) and high (10%) serum concentrations ( Fig. 5D and E).
Since JNK deficiency did not prevent Ras-stimulated growth (Fig. 5), we questioned whether the loss of JNK might influence other signal transduction pathways in MEF with activated Ras. No significant differences in the activation of ERK, p38 MAP kinase, and AKT were detected when Trp53 Ϫ/Ϫ and Jnk Ϫ/Ϫ Trp53 Ϫ/Ϫ MEF were compared (Fig. 6A). However, we detected reduced levels of reactive oxygen species (ROS) in Jnk Ϫ/Ϫ Trp53 Ϫ/Ϫ MEF compared with those in Trp53 Ϫ/Ϫ MEF (Fig. 6B). The ROS in Trp53 Ϫ/Ϫ MEF with activated Ras was suppressed by diphenyleneiodonium chloride (DPI), an inhibitor of NADPH oxidases (Nox) enzymes, but not by the mitochondrial electron transport inhibitor rotenone (Fig. 6C). These data are consistent with Nox-dependent, rather than mitochondrion-dependent, ROS accumulation (31,51). Nox1 and Nox4 have been implicated in ROS production by MEF (17). JNK deficiency caused no change in the expression of Nox1 (data not shown), but JNK deficiency did cause reduced expression of Nox4 (Fig. 6D). Expression of the Nox enzyme subunit p22 phox was not altered by JNK deficiency (Fig. 6D).
ROS production has been implicated in tumorigenesis (44). The levels of ROS detected in Trp53 Ϫ/Ϫ MEF being greater than those in Jnk Ϫ/Ϫ Trp53 Ϫ/Ϫ MEF suggests that JNK deficiency may alter the tumorigenic potential of these MEF with activated Ras. To test this prediction, we examined the growth of these MEF in soft agar assays in vitro. We found that activated Ras caused growth of large colonies of Trp53 Ϫ/Ϫ MEF in soft agar, but activated Ras did not cause the growth of large Jnk Ϫ/Ϫ Trp53 Ϫ/Ϫ MEF colonies in soft agar (Fig. 7A and B). Retroviral transduction of Nox4 did not rescue the defect in soft agar growth or contact growth inhibition detected in cultures of Ras-transformed Jnk Ϫ/Ϫ Trp53 Ϫ/Ϫ MEF (data not shown). This observation demonstrates that reduced expres-sion of Nox4 is not sufficient to account for the phenotype of Ras-transformed JNK-deficient MEF.
To test the effect of JNK deficiency on tumor formation in vivo, we examined the subcutaneous growth of MEF with activated Ras in syngeneic mice. We found that Ras-transformed Trp53 Ϫ/Ϫ MEF formed tumors, but this growth was strongly suppressed by JNK deficiency (Fig. 7C and D). Together, these data demonstrate that JNK is required for Ras-induced transformation of p53-deficient MEF in vitro.

JNK deficiency suppresses Ras-induced lung tumor formation in vivo.
In vitro studies suggest that JNK is required for Ras-stimulated transformation (Fig. 7). To test whether JNK contributes to Ras-stimulated transformation in vivo, we examined the effect of activated Ras in an established model of lung tumorigenesis (29). Lung-specific expression of KRas G12D was studied using nasal instillation of adenovirus-Cre (29). We compared the effect of endogenous KRas activation in Jnk1 ϩ/ϩ Jnk2 ϩ/ϩ (control) mice and in mice with conditional compound JNK deficiency (Jnk1 LoxP/LoxP Jnk2 Ϫ/Ϫ ). Expression of KRas G12D caused lesions in the lungs, including atypical adenomatous hyperplasia, epithelial hyperplasia of bronchioles, and adenomas ( Fig. 8A and B). Comparison of control mice with JNK-deficient mice demonstrated that the number of hyperplastic lesions and adenomas was greatly reduced in the JNK-deficient mice ( Fig. 8C and D). Immunofluorescence analysis of lung sections demonstrated that JNK deficiency caused a marked decrease in staining for the proliferation marker PCNA (Fig. 8E). No major differences in apoptosis markers (caspase 3 activation and PARP cleavage) between control and JNK-deficient lungs were detected (Fig. 8F). These data demonstrate that JNK is required for the efficient formation of Ras-induced lesions in the lung. The major role of JNK is reflected by increased proliferation rather than altered apoptosis. The low level of lung tumorigenesis detected in JNK-deficient mice suggests that JNK may not be essential for Ras-induced lung cancer. These tumors may represent JNK-independent formation of lung cancer. Alternatively, these tumors may arise because of incomplete Cre-mediated ablation of Jnk genes. Indeed, the possible presence of incomplete gene ablation in this model of KRas-induced lung cancer is consistent with previous reports using Rac1 LoxP/LoxP mice (33). We tested whether incomplete Jnk gene ablation might contribute to the low level of tumorigenesis in JNK-deficient mice by genotype analysis of genomic DNA isolated from these tumors. This analysis demonstrated that all of the JNK-deficient lung tumors examined retained a Jnk1 LoxP allele (Fig. 9). Indeed, immunofluorescence analysis of the phosphorylation of the JNK substrate c-Jun on Ser 63 demonstrated that the JNKdeficient tumors retained a functional JNK signaling pathway (Fig. 9). We did not find in JNK-deficient mice any tumors that lacked expression of the Jnk1 LoxP allele or phosphorylation of the JNK substrate c-Jun. These data indicate that JNK may be essential for KRas-induced lung tumorigenesis.

DISCUSSION
This study demonstrates that JNK can play a major role in Ras-induced tumor formation. Specifically, JNK is essential for transformation of primary p53 Ϫ/Ϫ MEF by activated Ras in vitro. Moreover, JNK is required in a mouse model of Rasinduced lung tumor formation in vivo. Together, these data provide strong evidence that the JNK signaling pathway may contribute to the development of cancer.
Further studies are needed to identify the signaling pathway that leads to Ras-induced JNK signaling. A recent study established that Rac1 is required for Ras-induced lung tumor formation (33). Since Rac1 can cause JNK activation (3,12,37), it is possible that this role of Rac1 may reflect a requirement of Rac1 for JNK activation in this model of Ras-induced lung tumorigenesis. Targets of Rac1 that may mediate JNK pathway activation include members of the mixed-lineage protein kinase family (22).
Additional studies are required to identify JNK substrates that are relevant to lung tumorigenesis. It is possible that AP1 family transcription factors, which are activated by JNK, may contribute to Ras-induced lung tumor formation. However, many other targets of JNK signaling, in both the cytoplasm and the nucleus, may contribute to JNK-dependent tumorigenesis (48). Detailed analysis of these JNK substrates is required to identify those relevant to lung tumorigenesis.
JNK prevents contact growth inhibition. JNK may contribute to dysregulated cellular proliferation by interfering with contact growth inhibition. Trp53 Ϫ/Ϫ and Trp53 Ϫ/Ϫ KRas G12D primary MEF exhibit defects in contact growth inhibition and form multiple layers when grown to confluence ( Fig. 1A and  2A). In contrast, compound deficiency of Jnk1 plus Jnk2 prevents the overgrowth phenotype of both Trp53 Ϫ/Ϫ and Trp53 Ϫ/Ϫ KRas G12D primary MEF ( Fig. 1A and 2A). This effect of JNK correlates with altered expression of the cadherin family of adhesion proteins ( Fig. 2B and C). Strikingly, Rastransformed Trp53 Ϫ/Ϫ Jnk Ϫ/Ϫ primary MEF express high levels of E-cadherin ( Fig. 2B and C). This expression of E-cadherin may contribute to increased intercellular adhesion. Moreover, in the context of Ras-induced epithelial cell tumorigenesis, increased E-cadherin expression might decrease epithelialmesenchymal transition (EMT) and progression to carcinoma (5). Indeed, it is established that EMT of lung epithelium is suppressed by JNK deficiency (1). These actions of JNK to decrease the intercellular adhesion of cells expressing KRas G12D may contribute to tumor formation.
JNK and the tumor microenvironment. This study has focused on the role of JNK in tumor cells. However, JNK may also play roles within the tumor microenvironment, including expression of cytokines/growth factors (15) and the function of the innate and adaptive immune systems (15,23,45). Thus, the finding that Jnk1 Ϫ/Ϫ mice exhibit decreased (and Jnk2 Ϫ/Ϫ mice exhibit increased) carcinogen-induced skin cancer compared with wild-type mice (11,42) may be explained, in part, by the opposite effects of JNK1 and JNK2 deficiency on CD8 T cellmediated tumor immunosurveillance (23,45). Moreover, the effects of JNK1 deficiency to suppress carcinogen-induced hepatocellular carcinoma (28) may be accounted for, in part, by decreased expression of inflammatory cytokines (e.g., interleukin 6 [IL-6] and tumor necrosis factor [TNF]) by hepatic innate Amplimers employed for PCRs designed to distinguish between these genes are indicated. (C) Sections of lung tumors were stained using an antibody to phospho-Ser 63 -c-Jun and DAPI and examined by fluorescence microscopy (left). The fluorescent images were merged with images obtained using differential interference contrast microscopy (right). Ser 63 is a phosphorylation site of c-Jun by JNK. These data demonstrate that lung tumors in JNK-deficient mice retain a functional JNK signaling pathway.
immune cells (15). These considerations indicate that a full understanding of the role of JNK in cancer should take account of JNK functions in both tumor cells and the tumor microenvironment. JNK and tumor suppression. The results of this study demonstrate that Ras-induced transformation of primary MEF and Ras-induced lung tumor formation requires JNK. However, these data do not exclude the possibility that JNK may also contribute to tumor suppression. Indeed, published studies have implicated JNK in both tumor formation and tumor suppression (16,49). Precedent for this type of dual activity during tumor development is provided by the transforming growth factor ␤ (TGF-␤) pathway (43). Roles of the JNK pathway in tumor suppression most often correlate with late-stage tumor metastasis (49). Indeed, studies of human cancer genetics and biochemical analysis of human tumor-derived cell lines support the conclusion that the JNK signal transduction pathway may contribute to metastasis suppression (49). Moreover, Ras transformation of established cell lines demonstrates that loss of JNK can cause increased metastatic potential (32). Further studies will be required to formally test the hypothesis that JNK may act as a suppressor of late-stage tumor development in vivo. In contrast, this study demonstrates that JNK is required for early stage Ras-induced tumor formation in vitro and in vivo.