Transformation by Oncogenic RAS Sensitizes Human Colon Cells to TRAIL-induced Apoptosis by Up-regulating Death Receptor 4 and Death Receptor 5 through a MEK-dependent Pathway*

RAS oncogenes play a major role in cancer development by activating an array of signaling pathways, most notably mitogen-activated protein kinases, resulting in aberrant proliferation and inhibition of apoptotic signaling cascades, rendering transformed cells resistant to extrinsic death stimuli. However, tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) is able to kill specific tumor cells through the engagement of its receptors, death receptor 4 (DR4) and death receptor 5 (DR5), and the activation of apoptotic pathways, providing promising targets for anticancer therapies. In this study, we show that TRAIL induces cell death in human colon adenocarcinoma cells in a MEK-dependent manner. We also report a prolonged MEK-dependent activation of ERK1/2 and increased c-FOS expression induced by TRAIL in this system. Our study reveals that transformation of the colon cell line Caco-2 by Ki- and mainly by Ha-ras oncogenes sensitizes these cells to TRAIL-induced apoptosis by causing specific MEK-dependent up-regulation of DR4 and DR5. These observations taken together reveal that RAS-MEK-ERK1/2 signaling pathway can sensitize cells to TRAIL-induced apoptosis by up-regulating DR4 and DR5 and overall imply that TRAIL-based therapeutic strategies using TRAIL agonists could be used in cases of human colon cancers bearing RAS mutations.

The initiating mutagenic changes in the genome producing an early adenoma from normal colonic mucosa may provide a favorable environment (i.e. increased replication rate, increased survival signals, and alterations of the DNA repair mechanisms) for other mutations to occur. In this way, they traverse from early to intermediate to late adenoma and finally adopt a malignant phenotype, as characterized by the multistage carcinogenesis model (1). Avoidance of apoptosis is a basic event that occurs during carcinogenesis. Changes in the apoptotic program and/or activation of the antiapoptotic path-ways are key events allowing progression through various stages of carcinogenesis (2).
Apo2L/TRAIL 1 (Apo2 ligand/tumor necrosis factor-related apoptosis-inducing ligand) is a cytokine that can selectively induce apoptosis in tumor cells although leaving normal cells largely unaffected. The effects of TRAIL are mediated through the tumor necrosis factor receptor superfamily members, TRAIL-R1 (DR4) and TRAIL-R2 (DR5). TRAIL can also bind to the decoy receptors, the role of which remains to be identified (3,4). There are four distinct TRAIL receptors, all belonging to the tumor necrosis factor receptor superfamily: TRAIL-R1 (DR4) (5), TRAIL-R2 (DR5) (6), TRAIL-R3 (TRID/DcR1/lymphocyte inhibitor of TRAIL (LIT)), and TRAIL-R4 (DcR2) (7). The latter two are incapable of transducing the signal because they either do not contain a death domain or contain a truncated death domain. TRAIL ligation to its functional receptors leads to recruitment of the adapter protein Fas-associated death domain (FADD) to the cytoplasmic region of the receptor (8) followed by recruitment of procaspase-8 or procaspase-10 (9, 10) resulting in the formation of the death-inducing signaling complex. At this level, procaspases undergo autocatalytic processing to be activated, leading to the activation of the effector caspases (11), mitochondrial dysfunction, and ultimately apoptosis.
Even though TRAIL receptor activation is mainly linked to the induction of apoptosis, evidence implicates TRAIL signaling in the activation of c-Jun NH 2 -terminal kinase (12), NFB (13,14), and ERK (15,16), suggesting that multiple signaling pathways are activated through the TRAIL/TRAIL-R system. It has been shown that incubation with TRAIL can activate a positive feedback loop causing the up-regulation of TRAIL-R2 (17) that can potentially render the cells more sensitive to TRAIL in a time-dependent manner. Activation of AP-1 (18) can lead to TRAIL-R1 up-regulation. Similarly TRAIL-R2 has been shown to be up-regulated through NFB (17). These results taken together allow us to assume that molecules not directly implicated in the TRAIL pathway play an important role in the regulation of TRAIL receptor expression.
Considering the limitations of the current therapies in targeting cancer cells only, TRAIL provides an attractive target for cancer therapy (19). Currently preclinical studies using soluble recombinant TRAIL derivatives, as well as monoclonal antibodies mimicking TRAIL effects, induce apoptosis in a broad range of human cancer cells and tumor regression in mice without significant systemic toxicity (4). In addition, TRAIL can complement current strategies because it can render death sensitivity to tumors by its synergistic action with chemotherapeutic drugs or UV treatment (20,21).
RAS genes play a very important role in the development of human cancer. Activating mutations in the three best studied RAS genes, Ha-RAS, Ki-RAS, and N-RAS, have been detected in a large variety of human tumors (22). Studies in mice that are genetically modified to express ras oncogenes have very clearly shown that activated Ras promotes the initiation, progression, and maintenance of several types of tumors (23,24). Activating Ki-RAS mutations are found in about 50% of colon carcinomas. Activated RAS mediates its biological activity through interaction with various downstream effector targets, thus activating pathways like MEK, PI3K, and Rho GTPases. RAS regulates a RAF-MEK-ERK1/2 kinase cascade. The MEK pathway has been shown to be active in human colon adenocarcinoma cells (25) as well as in human colorectal tumors (26). The activated ERK1/2 translocate to the nucleus and regulate the expression of target genes like c-FOS and/or FRA-1 (27,28).
Sensitization to TRAIL-induced apoptosis very often is mediated by up-regulation of TRAIL receptors DR4 and DR5 (29). Interestingly, expression of DR4 and DR5 has been found elevated in human colonic carcinomas as compared with normal colonic mucosa (30); we aimed to pinpoint oncogenic pathways that make TRAIL more efficient against cells representing the later stages of carcinogenesis. We specifically examined the sensitivity of an intermediate colon adenocarcinoma cell line, Caco-2, to TRAIL-induced apoptosis subsequent to its transformation with two active RAS isoforms (KRASV12 and HRASV12) that are very potent activators of ERK1/2 kinases. Caco-2 cells have been reported previously to be insensitive to TRAIL-induced apoptosis. We found that in Caco-2 human colon cells, the two RAS oncogenes present tumor-promoting potential, the Ha-RAS clones giving more and larger tumors in nude mice and a different gene expression profile. 2 In this study we have provided strong evidence for the importance of ERK in the regulation of TRAIL-R1 and TRAIL-R2. Moreover, we denote the significance of RAS oncogenes in the regulation of TRAIL receptors and the subsequent sensitivity to TRAILmediated apoptosis, both via cellular transformation and through direct signaling.
Specifically, we have demonstrated the functional importance of the ERK-1/2 signal in modulating TRAIL-induced apoptosis in RAS-transformed cells, in which aberrant Ras activation up-regulates the expression of TRAIL receptors, thus modulating the subsequent sensitivity to TRAIL-mediated apoptosis by an ERK-dependent mechanism. We also have demonstrated the relevance of this mechanism to its potential exploitation in colorectal cancer therapy.

MATERIALS AND METHODS
Cell Cultures-The Caco-2, DLD-1, and HT29 cells were obtained from ATCC. The human colon cell lines Caco-2 and HT-29 and the RAS-overexpressing clones derived from Caco-2 were cultured in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen), antibiotics, and nonessential amino acids (Invitrogen). The human colon adenocarcinoma cell line DLD-1 and the derived clone that has the oncogenic Ki-RAS allele disrupted (DKO-4) (31) were cultured in RPMI 1640 (Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen) and antibiotics. The human colon cell lines AAC1 and RGC2, obtained from Professor C. Paraskeva (United Kingdom), were cultured in the same manner as the HT-29 cells with the addition of insulin, hydrocortisone, and 20% fetal bovine serum (Invitrogen). RGC2 and AAC1 are clonogenic, nontumorigenic human colonic adenoma-derived cell lines (32). The MEK inhibitors PD98059 and U0126 were purchased from Alexis Biochemicals. The PI3K inhibitor wortmannin was purchased from Sigma. The Caco-2 clones constitutively expressing active RAS proteins will be described in a forthcoming article. 2 In short, cDNA expressing either V12 Ki-RAS4B or Ha-RASV12 was ligated into pcDNA3 plasmid. The resulting plasmid was transfected into Caco-2 cells using the CaPO 4 precipitation technique and selected with Geneticin (Invitrogen). The colonies expressing between 2-and 3-fold exogenous protein were selected for further analysis. As a control, Caco-2 cells were also transfected with empty pcDNA3 expression cassette, and a number of colonies were isolated and expanded.
Cytotoxicity and Apoptosis Assays-For cell viability cells were plated on 6-well plates, fixed with absolute methanol, stained with 0.5% crystal violet for 10 min, and washed three times with phosphatebuffered saline; the remaining crystal violet was extracted using 30% acetic acid and absorbance was measured at 595 nm. The percentages of viable, necrotic, and apoptotic cells were assessed by exposure to the DNA binding dyes Hoechst no. 33342 (Sigma) and propidium iodide (Sigma); apoptotic measurements were carried out using a fluorometric method and counting cell numbers under a fluorescent inverted microscope (Nikon Eclipse).

FIG. 1. Response of human colon adenoma-carcinoma cell lines to TRAIL treatment and steady state protein levels of apoptotic factors.
A, the cells were incubated with 500 ng/ml TRAIL, and viability was measured after 24 h. B, HT29 cells were incubated for 12 h with or without 500 ng/ml TRAIL, the nuclei were stained with Hoechst and propidium iodide, and 500 cells from random fields were checked for apoptotic or necrotic nuclei. The arrows indicate apoptotic nuclei. AP, number of apoptotic nuclei; an insignificant number of necrotic nuclei was found and is not shown. At the same time, whole cell lysates were checked with WB for characteristic PARP cleavage. A representative Western blot is shown. C, the relative mRNA levels of DR4 and DR5 in Caco-2, DLD-1, and HT29 cells were analyzed by RT-PCR and normalized to GAPDH; a representative image of four experiments is shown. D, the relative protein levels of FADD, caspase-8, caspase-3, DR4, and DR5 were also analyzed by Western blot and normalized to total ERK2 levels.
ERK2 protein expression and/or Ponceau staining was used for protein loading control; ERK2 was also used for normalization. All of the experiments were repeated at least three times, and S.D. function was used for error bar generation. All controls in every experiment were treated with the same amount of buffer/dilutant as the treated cells.

TRAIL Induces Apoptosis Specifically in Colon Carcinoma Cell Lines, and This Effect Correlates with High DR4 and DR5
Expression Levels-TRAIL has been shown to exert enhanced apoptotic activity on tumor cells, whereas non-tumor cells have been reported to be resistant to TRAIL-induced death in many systems (4). To examine the apoptotic effects of TRAIL in an in vitro system of human colorectal carcinogenesis, we subjected human adenoma and carcinoma colon cells to different concentrations and time periods of TRAIL treatment. The three colon adenoma cell lines examined in the study, AAC1 (32), RGC2, and Caco-2, showed no difference in viability after a 24-h treatment with 500 ng/ml TRAIL (Fig. 1A). Even after 72 h of TRAIL treatment no evidence of cell death in adenoma cells was detected (data not shown). On the other hand, the carcinoma cell lines HT-29 and DLD-1 had reduced viability after TRAIL treatment ( and at the protein level by WB analysis (Fig. 1D), respectively. We detected increased levels of DR4 and DR5 mRNA and protein levels in HT29 and DLD-1 cells as compared with Caco-2 cells that were correlated with the sensitivity of the cell lines to TRAIL-induced cell death. Expression levels of other factors like FADD, caspase-8, and caspase-3 did not correlate with the sensitivity of the examined cell lines to TRAIL-induced apoptosis (Fig. 1D).
TRAIL Induces a Rapid and Sustained MEK-dependent Activation of ERK1/2 Followed by High mRNA Levels of c-fos-TRAIL receptor activation has been reported to regulate MAP kinase signaling pathways like c-Jun NH 2 -terminal kinase (12) and ERK (15), and these pathways play an important role in TRAIL-induced apoptosis in many systems. To examine the cascade of MAP kinase signaling events in the different colon cell lines after TRAIL treatment, we have followed the activation of MEK pathway from the cell membrane to the nucleus. Steady state basal phosphorylation levels of the ERK1/2 kinases were detected in colon cell lines ( Fig. 2A). Treatment with TRAIL increased the phosphorylated levels of MEK and p42/44 (ERK1/2) in HT29 cells ( Fig. 2A). There was a strong and rapid ERK1/2 activation peaking 15 min after treatment with 500 ng/ml TRAIL. ERK phosphorylation followed MEK activation that peaked after ϳ5 min of treatment with TRAIL. We then examined the ELK-1-responsive immediate early gene c-FOS in the HT29 cells and found an increase at the mRNA level of 1.9-fold at 30 min and 2.5-fold at 90 min after treatment with TRAIL, although this increase was abolished by co-treatment with the MEK inhibitor PD98059 (Fig. 2B). Remarkably, c-FOS induction by TRAIL was sustained for 90 min and did not show the characteristic transient induction observed in many systems after growth factor stimulation (Fig. 2B). The DLD-1 cell line was also found to have increased mRNA levels of the c-FOS gene after 25 min of incubation with TRAIL, an effect that was abolished by co-treatment with the MEK inhibitor (data not shown). importance of the regulation of TRAIL signaling at the receptor level. To explore the mechanism of TRAIL-induced MEK signaling and its effect on TRAIL receptor expression, we examined the potential feedback mechanism in the colon cell lines.

Treatment with TRAIL Increases the Levels of Its Receptors
Pretreatment of the HT29 cell line for 16 h with 100 M MEK inhibitor PD98059 prevented TRAIL from significantly reducing cell viability (Fig. 3A) and from inducing apoptosis. Specifically, HT29 cells treated for 24 h with TRAIL showed significant reduction in viability relative to the control. Pretreatment for 16 h with the MEK inhibitor of cells to be treated with TRAIL resulted in an increase of cell viability (Fig. 3A), inhibition of PARP cleavage (Fig. 3B), a reduced number of apoptotic nuclei (Fig. 3C), and reduced caspase-8 activation (data not shown). Specifically, 32 Ϯ 3.8% (mean Ϯ S.D.) of the nuclei counted in the cells treated with TRAIL had apoptotic characteristics, whereas in the case of cells pretreated with the MEK inhibitor PD98059 only 16 Ϯ 2.1% of the nuclei showed apoptotic characteristics after treatment with TRAIL (Fig. 3C). On the other hand, in the DLD-1 cell line MEK inhibition failed to produce any significant differences in TRAIL-induced cell death (data not shown).
Additional evidence for the regulatory role of the MEK pathway on the expression levels of TRAIL receptors was provided by the reduced steady state levels of the functional TRAIL receptors DR4 and DR5 after treatment with MEK inhibitor, as determined by RT-PCR in the DLD-1 and HT29 cells (Fig. 4A) and by Western blot analysis in the HT29 cells (Fig. 4B). This clearly showed that in both the HT29 and the DLD-1 cell lines the expression levels of these receptors were at least partially dependent on MEK activation. This effect was lower on the DLD-1 cells, indicating that in this cell line the role of MEK signaling in the expression of the death receptors is not very important. In the same experiment we have shown that the mRNA expression levels of both of the TRAIL receptors increase after treatment with TRAIL (Fig. 4A). To determine whether the actual expression levels of the DR4 and DR5 receptors were altered on the cell surface, we performed FACS analysis on the HT29 cells and confirmed the observations made by Western blot and RT-PCR. There was a slight decrease in the cell surface levels of DR4 and DR5 after a 1-h incubation with MEK inhibitor PD98059, whereas a 16-h incubation showed a ϳ42 Ϯ 4.2% and a 51.5 Ϯ 4.6% decrease in the levels of DR4 and DR5, respectively. The decoy and FAS receptors showed no significant changes after treatment with PD98059, indicating that the effect of the MEK inhibitor was not universal for death receptor levels in the cells under investigation (Fig. 5).

Human Colon Cell Transformation by Oncogenic Ki-and Ha-RAS Up-regulates DR4 and DR5 Followed by Increased
Sensitivity to TRAIL-To examine the ability of RAS oncogenes (which are very potent activators of ERK1/2 kinases) to sensitize nontransformed human cells to TRAIL-induced apoptosis, we examined whether their overexpression would sensitize the previously unresponsive intermediate colon adenoma cell line Caco-2 to TRAIL-induced apoptosis.
The Caco-2 cell line was stably transfected with the Ki-RASV12, the Ha-RASV12, and the empty vector ( (Fig. 6A, Neo) as a control, 2 and clones with low-to-moderate expression of RAS were chosen for analysis. The response of the various RAS-transformed cells to TRAIL was determined by cell viability assay and the presence of PARP cleavage. The Ha-RAStransformed cells (Fig. 6A, H2 and H13) showed responsiveness to TRAIL in reducing cell viability by ϳ35% after 24 h and by 73% after 62 h of treatment, whereas the Ki-RAS (Fig. 6A, K15) clones were more resistant; beginning to show significantly reduced viability after 62 h of treatment (Fig. 6A); these clones gave representative results of the various neo-transformed and Ki-and Ha-RAS-transformed cells tested. The reduced viability occurred by apoptosis as detected by Hoechst staining (data not shown) and by Western blot for the characteristic PARP cleavage (Fig. 6B). Moreover, we examined the steady state levels of phosphorylated Akt at Ser-473, a downstream effector of PI3K survival signaling, to see whether these levels correlate with the differential responsiveness of Ki-and Ha-RAS-transformed cells to TRAIL. Steady state levels of p-Akt were higher in both clones relative to neo, although Ha-RAS seemed to be more effective in activating PI3K signaling (Fig. 6C) as reported previously in other systems (33). Using a PI3K inhibitor (wortmannin) we were able to reduce the phosphorylation levels of Akt in all clones (Fig. 6C) with a corresponding reduction in viability after TRAIL treatment (Fig. 6D). These results show a uniform response of the cells to PI3K inhibition allowing us to assume that it is unlikely that PI3K signaling plays an important role in the differential responsiveness of the clones to TRAIL.
To determine the expression levels of DR4 and DR5 in the RAS-overexpressing clones, we performed RT-PCR and Western blot analysis. RT-PCR analysis of the steady state mRNA levels for DR4 and DR5 indicated their significant overexpression in the RAS-transformed clones in both cases as compared with the control (neo)-transfected cells (Fig. 7A). Western blot analysis of the expression levels of DR4 and DR5 showed a 1.7-fold up-regulation of these receptors on average, whereas caspase-3, caspase-8, and FADD protein levels did not show significant differences between the RAS clones; remarkably in the case of DR5 protein levels in the Ha-RAS clones there is an additional slower migrating band repeatedly giving a strong signal on Western blot at about 50 kDa (Fig. 7B). Overexpression of DR4 and DR5 in the Caco-2 RAS-transformed cells was confirmed on the cell surface by FACS analysis (Fig. 7C). In addition to DR4 and DR5 up-regulation, we noted that in the case of the Ha-RAS clones there is a strong up-regulation and cell surface localization of the Fas receptor as analyzed by FACS (Fig. 7C), whereas in both cases the decoy receptor levels were not altered as compared with the control Caco-2 neo clones. To further confirm the effect of oncogenic RAS on DR4 FIG. 6-continued and DR5 expression we checked the steady state mRNA expression of these receptors in DKO-4 cells, which were derived from DLD-1 cells and have their oncogenic Ki-RAS allele disrupted (31). RT-PCR analysis showed that both DR4 and DR5 expression is reduced in the DKO-4 clone relative to the DLD-1 cells transfected with the control cassette (Fig. 7D).

Inhibition of ERK1/2 in the Ki-and Ha-RAS Clones Partially Reverses the Increased Expression Levels of DR4 and DR5-We
investigated whether ras transformation of Caco-2 cells was followed by a corresponding increased ERK activity; WB analysis for activated ERK1/2 showed increased phosphorylation levels of ERK1/2 in both the Ki-and the Ha-RAS clones that could be decreased after MEK inhibition (Fig. 8A). The use of the specific inhibitors for MEK reduced the increased expression of DR4 and DR5 on the mRNA level (Fig. 8B) as well as on the protein level as detected by Western blot analysis (Fig. 8C). The decrease in protein and mRNA levels of DR4 and DR5 was accompanied by a respective decrease in sensitivity to TRAILinduced apoptosis when the H2 clone was treated with the MEK inhibitor PD98059 prior to treatment with TRAIL (Fig.  8D). We note that these results were reproducible when we used the UO126 compound for specific inhibition of MEK (data not shown). DISCUSSION The discovery of specific cellular targets for colorectal cancer therapy resulting in the generation of selective agents against aberrantly regulated gene products present in cancer cells will potentially result in less toxicity than that observed in tradi-tional therapies. As such, the most attractive therapeutic agent would target regulated cell properties such as survival, cell cycle, signal transduction, and metastasis. Studies with this approach are currently in progress to predict which targets and therapeutic candidates will be most promising in the treatment of colorectal cancer (34).
Constitutively activated Ras increases the tumorigenic potential of cells because it causes deregulation of important intracellular signaling pathways. Ras proteins transduce signals from receptor tyrosine kinases to a downstream cascade of protein kinases regulating the growth, survival, and cytoskeletal processes that are aberrant in malignant cells. Activating mutations of RAS occur in almost half of the incidences of colorectal cancer. Major RAS-regulated signaling pathways are the MEK and PI3K pathways, which play an important role in cell proliferation and survival.
The death receptor ligands tumor necrosis factor, FAS ligand, and TRAIL are all able to induce apoptosis by binding to their cell membrane receptors. Recombinant forms of these ligands can potentiate the antitumor effects of cytotoxic agents in both in vitro and in vivo models.

TRAIL-induced Cell Death in Colon Cells Is Correlated with DR4 and DR5 Overexpression in a MEK-dependent Manner-
We have used human colon cell lines to show that the apoptotic agent TRAIL activates the MEK pathway in a rapid and sustained manner and that this pathway is involved in regulating DR4 and DR5 expression levels. Specifically, by inhibiting MEK we found that the mRNA and protein levels of DR4 and DR5 were down-regulated, and this had a striking effect on the ability of TRAIL to induce cell death in the HT29 cells and the oncogenic ras-transformed Caco-2 clones. Moreover, we found that in cells nonresponsive to TRAIL-induced apoptosis ERK1/2 activation by overexpression of activated RASV12 isoforms consequently up-regulated expression of DR4 and DR5 and sensitized the cells to death, whereas disruption of oncogenic Ki-RAS reduced both the phosphorylation levels of ERK proteins and death receptor expression. Our findings are further supported by the evidence that expression of the human death receptor 4 is regulated by AP-1 (18). On the other hand transcriptional up-regulation of the death receptor 5 gene is dependent on NFB (17) and AP-1 (18). Therefore, TRAIL receptor expression as well as the sensitivity of cells to TRAIL is regulated by multiple factors besides MEK, including signaling pathways, transcription factors, and the general oncogenic transformation status of the cells due to co-operative activities of existing oncogenic mutations. We believe that in the case of the DLD-1 cells these factors have a high impact on their sensitivity. It has to be noted that DLD-1 cells bear a Ki-RAS as compared with a B-RAF mutation in HT-29 cells, as well as a different APC mutation than HT-29 cells, which results in differential signaling to the nucleus (35). These differences, taken together with the observation that these cells constitu-tively express much higher levels of TRAIL receptors and, therefore, after treatment with PD98059 still conserve high levels of these receptors, can explain the inability of MEK inhibition to reduce their sensitivity.
Our findings provide a mechanistic explanation of how MEK signals promoting cell growth in most cases can also under certain conditions mediate cell death, in this case by inducing elevated expression of TRAIL receptors and sensitizing resistant cells to TRAIL-induced cell death. The decision between life and death in a particular cell type must depend on the balance of signaling pathways; in the case of the MEK pathway the duration and the intensity of the signal is important, the most characteristic example being that of PC12 cells, in which the duration of ERK activity determines whether the cells proliferate or differentiate (36). Importantly, we have detected a rapid and sustained activation of ERK1/2 regulated by MEK for more than 3 h after TRAIL treatment and a subsequently sustained overexpression of c-FOS for more that 90 min. This is in contrast to the transient activation of MEK pathway and c-FOS expression after typical growth factor stimulation reported in many cell types. These results are consistent with previous studies in which it was found that TRAIL mediates FADD-dependent overexpression of the c-FOS transcription factor (37).

FIG. 7-continued
Transformation by Ras Renders Resistant Colon Cells Sensitive to TRAIL-To examine whether the ability of the MEK signaling pathway mediates the TRAIL effect in colon cells, we developed clones of colon cells overexpressing oncogenic ras forms because Ras activates the MEK pathway in many cell systems. Hence, we established stable Caco-2 cell lines constitutively expressing Ki-RASV12 (Caco-KV12) or Ha-RASV12 (Caco-HV12), and a control cell line stably transfected with the empty expression cassette (Caco-Neo). Caco-2 cells are from an intermediate adenoma colon-derived cell line that is often used in the study of enterocyte differentiation because upon reaching confluency the cells differentiate into a mature enterocyte phenotype. These cells bear no mutations in any of the three RAS loci, nor are they known to bear mutations in any of the genes that function as RAS effectors (e.g. B-RAF). Therefore, Caco-2 cells represent an ideal model system in which to introduce RAS mutations and examine their effects on MAP kinase signaling.
The isolation and characterization of RAS-overexpressing clones will be described in a forthcoming article. 2 Briefly, we have demonstrated that constitutively active RASV12 isoforms transform Caco-2 cells indicated by growth in soft agar and the formation of tumors in severe combined immunodeficient mice. Moreover, only Ha-RASV12 was capable of inducing an epithelial to mesenchymal transition when compared with Ki-RASV12, and these findings were further supported by gene expression profile analysis. 2 RAS overexpression converted the TRAIL-resistant Caco-2 cells to cells responsive to TRAIL tumoricidal activity and, especially in the case of Ha-RAS-overexpressing clones, in a more dramatic way. Tumor cells can become resistant to TRAIL by distinct mechanisms. We examined the reasons for induced sensitivity to cell death by RAS overexpression in these cell lines. We analyzed expression levels of TRAIL downstream components like FADD and caspases 3 and 8, but no changes related to TRAIL sensitivity were detected. Expression levels of DR4 and DR5 were increased in a MEK-dependent manner. Of special interest is the slower migrating form of DR5, which is predominantly expressed in the Ha-RAS-overexpressing clone, the most sensitive of all the cell lines tested to TRAIL-induced apoptosis. In addition, the Ha-RAS-transformed clones presented a dramatic overexpression of the Fas receptor, but this was not correlated with MEK activation. These findings argue that colon cells transformed by RAS oncogene can be sensitive to TRAIL-induced cell death in a MEKdependent manner. Overexpression of the Ha-RAS oncogene has a better apoptosis sensitization potential in general, potentially because it strongly up-regulates the FAS receptor as well.
Recently, RAS sensitization of human embryonic kidney and foreskin fibroblast cells to tumoricidal activity of TRAIL has been reported (38), although it has not been correlated with TRAIL receptor expression and MEK signaling pathways. We believe that the presence of oncogenic RAS acts upon the sensitivity of the cells to TRAIL through MEK signaling and through oncogenic transformation in cooperation with other oncogenes. The latter could be one of the reasons that the Ha-RAS-transformed clones have higher sensitivity to TRAIL because oncogenic Ha-RAS is known to have a greater transforming potential as compared with Ki-RAS. In our model system, which is relevant to human colorectal cancer, a mechanistic explanation for the observed effect of RAS oncogene is provided. Transformation with the c-MYC oncogene (39) pro-

FIG. 8. Consequences of MEK inhibition on protein and mRNA levels of the apoptosis-inducing TRAIL receptors in the RAStransformed clones.
A, steady state phosphorylated levels of ERK1/2 proteins and the effect of the PD98059 (PD) on these levels were determined by Western blot analysis using phosphor-specific antibodies against ERK-1/2. B, DR4 and DR5 mRNA levels were analyzed by RT-PCR, which was performed on RNA extracted from cells that were either incubated for 5 h with a 100 M concentration of the MEK inhibitor or not, and the values were normalized to GAPDH (shown below each lane); a selected representative image from four repeats is shown. C, DR4 and DR5 protein levels were examined by Western blot analysis with or without 16 h of incubation with a 100 M concentration of the MEK inhibitor PD98059. D, a viability assay was performed on the H2 clone with or without 16 h of pretreatment with PD98059. The graph represents the decrease in viability relative to respective control starting from the moment that TRAIL was added to the growth medium.
vides resistant cells with cell death sensitivity as well; thus the presence of specific activated oncogenes can provide an explanation of the sensitivity of tumor cells to TRAIL-induced apoptosis in general.
RAS as a Determining Factor for TRAIL Specificity in Killing Tumors-Careful selection of therapeutic strategy based on molecular phenotyping of the patient's tumor is evidently to become a major issue in the development of targeted anticancer therapeutics.
The soluble recombinant TRAIL, as well as monoclonal antibodies that mimic its effect, are of interest for cancer therapy and are already in clinical trials (40,41). Recent studies in the mouse suggest that TRAIL immunoselects tumors for TRAIL resistance (42). Data showing the impressive selective antitumor activity of soluble TRAIL in vitro have generated considerable excitement and have resulted in the development of TRAIL as a novel anticancer agent; only recently, moreover, a few key studies have addressed the natural role of TRAIL in immunity against cancer.
Importantly, not all cancer cells are sensitive to the cytotoxic effects of TRAIL. Moreover, abnormalities of various components of death receptor pathways have been identified in human cancer including loss of FAS expression, deletion or loss of TRAIL receptor DR4, mutation of TRAIL receptor DR5, overexpression of TRAIL decoy TRID, and overexpression of Fas decoy receptor, as well as overexpression of the activated caspase-8/10 inhibitor, FLIP. These changes can determine the resistance or sensitivity of a particular cancer cell type to TRAIL-induced apoptosis (43). In this study we have used an in vitro system based on human colorectal adenocarcinoma cells, and we have shown that the presence of an activated form of RAS oncogene can shift the balance of a resistant colon cell line toward sensitivity to TRAIL-induced apoptosis by up-regulating DR4 and DR5. In support of this hypothesis, cell lines originally bearing a mutant Ki-RAS allele that was disrupted (31) had reduced TRAIL receptor levels (Fig. 7D). Although in our cell system the presence of activated RAS can provide sensitivity to TRAIL, it is possible that in human tumors other genetic changes of TRAIL downstream components, such as those mentioned above, can also contribute to sensitivity, and this hypothesis will be tested in the future. We have also demonstrated that it is specific transformation by RAS that renders colon cells sensitive because the MEK pathway is activated in this system and its inhibition prevents TRAIL-induced death. It is likely that other oncogenes capable of activating the MEK pathway may potentially provide these same properties to tumor cells. On the other hand, the PI3K survival pathway does not play an important role in the sensitivity to TRAIL on the part of the Ki-and Ha-RAS clones.
The findings of our study provide a mechanistic basis for a pharmacogenomic approach and could be further exploited therapeutically. More than half of all human colon cancers bear oncogenic mutations on RAS genes (22). TRAIL and TRAIL agonists can potentially be used to destroy tumors that bear RAS or other oncogenic mutations that cause DR4 and DR5 receptor overexpression. Other approaches to inhibiting the components of RAS pathways are also currently in process and have already provided agents that have encouraging anticancer effects. Therefore, it is essential that careful therapeutic strategy selection should be made because the combination of molecules inhibiting RAS pathway components with TRAIL agonists may not give the desired synergistic effect. In recent years, efforts have been made toward the generation of "smart" anticancer drugs that will target specific molecules depending on the molecular phenotyping of the patient's tumor, and many of these potential or current drugs influence signaling path-ways. This raises the problem that an incorrect selection of multiple drugs could give conflicting results if the cross-talk of those pathways is not adequately taken into consideration.