Influence of K-ras Activation on the Survival Responses of Caco-2 Cells to the Chemopreventive Agents Sulindac and Difluoromethylornithine¹

Mutational activation of the K-ras oncogene is an important genetic alteration in colorectal neoplasia. K-rasmutations have been detected in ∼50% of sporadic human colorectal tumors (1, 2). K-ras mutations have been detected in aberrant crypt foci, as well as in adjacent regions of histologically normal mucosa (3). These findings suggest that the mutation of K-ras may be a relatively early event in the temporal development of colon cancer. K-ras is also mutated in chemically induced rodent tumors with a frequency similar to that of human cancers (4, 5). Although the role of K-ras in tumorigenesis is unclear, activation of this gene has been correlated with deficient apoptosis in human colorectal neoplasms (6).

The NSAIDs,⁴including piroxicam (7, 8), indomethacin (9),and sulindac (10, 11), effectively inhibit colon carcinogenesis in the AOM-treated rat model. NSAIDs also inhibit the development of tumors harboring an activated K-ras(12). Accumulating evidence suggests that NSAIDs inhibit carcinogenesis via the induction of apoptosis in tumor cells (10, 13, 14, 15). A number of studies suggest that the chemopreventive properties of the NSAIDs, including the induction of apoptosis, is a function of their ability to inhibit prostaglandin synthesis (14, 16, 17). Recent studies, however, indicate that NSAIDs may act through both prostaglandin-dependent and -independent mechanisms (18, 19, 20, 21). Sulindac sulfone, a metabolite of the NSAID sulindac, lacks COX-inhibitory activity yet induces apoptosis in tumor cells (10, 15), possibly by a mechanism involving inhibition of a cyclic guanosine 3′,5′-monophosphate phosphodiesterase (22), and inhibits tumor development in several rodent models of carcinogenesis (10, 15, 20).

The polyamine biosynthesis inhibitor also decreases K-rasmutations and colon carcinogenesis in the AOM-treated rat. The combination of DFMO and the NSAID piroxicam has been shown to have a synergistic chemopreventive effect in the azoxymethane-treated rat model of colon carcinogenesis (7), although DFMO exerted a greater suppressive effect than piroxicam on K-ras mutation and tumorigenesis when each agent was administered separately (7, 23, 24). In one study, administration of DFMO or piroxicam to AOM-treated rats reduced the number of tumors harboring K-ras mutations from 90 to 36 and 25%, respectively (8). Both agents also reduced the amount of biochemically active p21 ras in existing tumors. (23).

The importance of K-ras activation in NSAID-mediated chemoprevention has not yet been determined. NSAIDs induce apoptosis in both colon tumor cell lines and animal tissues and appear to reduce K-ras tumor development; however, K-rasactivation has not been investigated as a mechanism of NSAID-mediated cytotoxicity. It is also not known whether such cytotoxicity is dependent on the anti-inflammatory properties of the NSAIDs. The effect of polyamine depletion on NSAID-mediated chemoprevention has been studied only in piroxicam-treated rat models. The NSAID sulindac is metabolized to two different molecules that differ in their ability to inhibit COX, yet both are able to exert chemopreventive effects via the induction of apoptosis. Sulindac sulfone lacks COX-inhibitory activity and most likely facilitates the induction of apoptosis in a manner independent of blocking prostaglandin synthesis.

A number of clinical trials are under way that use a combination of DFMO and NSAIDs to potentiate chemoprevention. Although DFMO has been used successfully in combination with the NSAID piroxicam in animal models, the combination of DFMO and sulindac has not been investigated. Unlike piroxicam, sulindac is metabolized to two chemopreventive agents that most likely differ in their mechanism of action. Polyamines may play a role in the efficacy of one or both of these agents, because polyamines have been shown to mediate the induction of apoptosis in many rodent cell lines (25, 26, 27, 28). Several studies have shown that the proto-oncogene c-myc is a transcriptional activator of ODC (29, 30) and that ODC is a mediator of c-myc-induced apoptosis in murine myeloid cells (31, 32). Polyamine depletion in c-myc-overexpressing murine myeloid cells prevented apoptosis upon growth factor withdrawal (31).

Caco-2 cells were obtained from the American Type Culture Collection(Rockville, MD). Caco-2 cells were maintained in MEM (Life Technologies, Inc., Grand Island, NY) supplemented with 10% fetal bovine serum and a 1% penicillin (10,000 units/ml)-streptomycin(10,000 mg/ml) solution (Life Technologies, Inc.). Transfected Caco-2 cells resistant to neomycin were maintained in 400 μg/ml G418. Cells were maintained at 37°C in the presence of 5%CO₂ in air.

All cells were passaged into new medium every 2–3 days, just prior to confluence. Drug additions were performed at the time of cell seeding and in MEM for each cell line unless otherwise noted. DFMO was provided by Marion Merrel Dow Co. (Cincinnati, OH). The two metabolites of sulindac, sulindac sulfone and sulfide, were generous gifts from Dr. Gary Piazza (Cell Pathways, Inc., Horsham, PA).

Caco-2 cells were removed from the monolayer by treatment with trypsin(∼1500 units/ml; Calbiochem, San Diego, CA)-EDTA (0.7 mm)and counted using a hemocytometer. A sample of the cell suspension was combined in a 1:1 volume ratio with trypan blue dye (Life Technologies,Inc.), and at least two independently prepared suspensions were counted on a hemocytometer, two counts each. For all cell types, viability was determined by the percentage of cells able to exclude the trypan blue dye.

Apoptosis was quantitated by morphological examination of cells on prepared slides. Each sample was prepared by pooling trypsinized cells with the aspirated culture medium and an additional saline wash of the tissue culture plate. Slides for light microscopy were prepared by placing 20,000 or 40,000 cells into a cytospin cup and pelleting the cells onto slides via centrifugation using a cytospin (Shandon Lipshaw,Pittsburgh, PA) at 600 rpm for 2 min. Cells were affixed to the slides by immersion of the slide into 100% methanol for at least 1 min. Slides were stained using a 1:10 (for Chinese hamster ovary cells) or a 1:20 (for Caco-2 cells) dilution of Modified Giemsa stain (Sigma Chemical Co., St. Louis, MO). Apoptotic cells were identified by characteristic chromatin condensation, cytoplasmic vacuole formation,cell shrinkage, and formation of apoptotic “bodies” (33). Apoptotic bodies that were not enclosed by membranes were not included in the cell scoring. At least 500 cells were scored for each treatment for each day, and frequencies were expressed as a percentage of the total cells counted.

Caco-2 cell survival was determined by colony-forming efficiency. Caco-2 cells or the K-ras-transfected clones were plated at a cell density of 2 × 10⁵ cells per 60-mm dish in MEM, in the presence or absence of DFMO at varied concentrations. Cells were grown for 24 h, and then the DFMO/MEM media was removed. The plates were rinsed twice with saline, and MEM containing varied concentrations of sulindac sulfide or sulfone was added. Cells were grown for an additional 24 h and then replated at three serially diluted concentrations, with three plates seeded for each concentration. Approximately 21 days later, the plates were stained for colony formation.

Plates were removed from the incubator, and an equal volume of a 3:1 volume ratio of methanol and acetic acid was added to each plate for 5 min. The supernatant was aspirated, and the colonies were stained by the addition of ∼1 ml of crystal violet stain (5 mg/ml in 100%ethanol; Sigma). The plates were then rinsed in distilled water and air dried. Colonies were defined as consisting of a minimum of 50 cells,with each plate containing 20 or more colonies for inclusion in the data set. Standard deviations were prepared from data in which all three plates contained a sufficient number of colonies to be included.

A cDNA encoding an activated K-ras^Val12 was purchased from the American Type Culture Collection (Rockville, MD). This cDNA was ligated into the multiple cloning site of a pCDNA3 mammalian expression vector (Invitrogen Corp., Carlsbad, CA), and vectors containing the K-ras insert were isolated according to standard protocols (34). Large-scale amounts of plasmid(mg quantities) suitable for mammalian cell transfection were purified using the Nucleobond Plasmid kit (Clontech Laboratories, Inc., Palo Alto, CA).

The calcium phosphate transfection method was used to introduce the pCDNA3-K-ras plasmid into Caco-2 cells, according to established protocols (34).

One week after transfection, geneticin (Life Technologies, Inc. Germantown, MD) was added to the media at a concentration of 400μg/ml. Stable clones were isolated with the use of trypsin-soaked filter discs placed over isolated colonies on a tissue culture plate and then transferred to the wells of a 48-well plate. Clones were maintained in 400 μg/ml of geneticin and screened via a RFLP method to confirm presence of the mutant K-ras (4) and by Western blot for K-ras expression.

Whole-cell lysates of Caco-2 cells were prepared according to commercially available protocols (Santa Cruz Biotechnology, Santa Cruz,CA). Plates were kept on ice, and cells were scraped off the tissue culture plates in the presence of RIPA buffer (PBS, 1% NP40, 0.5%sodium deoxycholate, 0.1% SDS, 100 μg/ml phenylmethylsulfonyl fluoride, 30 μg/ml aprotinin, and 100 mm sodium orthovanadate). The cell suspension was then passed twice through a 22-gauge needle and was centrifuged at 10,000 × g for 20 min to clarify the lysate.

Characterization of K-ras expression in transfected Caco-2 clones was performed by Western blot, using a mouse monoclonal antibody directed against both normal and activated K-ras (Santa Cruz Biotechnology). Samples were electrophoresed onto a 12.5% acrylamide gel, and the proteins were then transferred onto a Hybond nitrocellulose membrane (Amersham Life Science, Inc., Arlington Heights, IL). Samples were run at 50 V for 16 h in a transfer apparatus (Bio-Rad Laboratories, Hercules, CA) in transblot buffer(30% methanol, 50 mm Tris, and 191 mm glycine). The nitrocellulose membrane was blocked for 1 h in Blotto (TBS, 5% nonfat dry milk, and 0.05%Tween 20; Sigma). A mouse monoclonal IgG directed against K-ras (Santa Cruz Biotechnology) was diluted 1:100 in Blotto and applied to the membrane for 2 h. Membranes were washed in Tween-TBS for three washes of 5 min each, and then a horse radish peroxidase-conjugated goat antimouse IgG secondary antibody was diluted in Blotto and applied to the membrane. The membrane was incubated in secondary antibody for 1 h, and then the membrane was washed again in Tween-TBS. Visualization was made with the ECL Detection system(Amersham Life Science, Inc., Arlington Heights, IL).

SDS-PAGE was performed according to the method of Laemmli (35), except that bisacrylamide was replaced with N,N′-diallyltartardiamide (Bio-Rad Laboratories). Samples were diluted in electrophoresis buffer (2% SDS, 5%β-mercaptoethanol, and 3% sucrose in 300 mmTris-HCl, pH 7.0) and heated at 100°C for 5 min before loading on a 12.5% acrylamide gel.

Protein measurements of the lysates used for Western blotting were performed with the DC protein assay kit (Bio-Rad Laboratories), using the modified method suitable for use with detergents. Protein concentrations for both assay types were estimated by comparing the absorbance of unknown samples to that of known BSA standard concentrations in the same assay.

Analysis of covariance model was used in the study, where the dose was the covariant variable. The data were normalized by dividing each value by the mean value of colony-forming efficiency at 0 dose level.

Caco-2 human colon adenocarcinoma cells were used to determine the effects of sulindac and DFMO on cells containing an activated K-ras. Caco-2 cells are one of the few colon tumor lines that contain a normal K-ras gene (36, 37). These cells possess many of the characteristics of mature enterocytes,including spontaneous differentiation and hydrolase secretion (38).

Caco-2 cells were stably transfected with a plasmid directing the overexpression of an activated K-ras, and individual clones were subsequently isolated. Ten clonal populations were characterized by Western blot using an antibody that reacts with both normal and mutated K-ras proteins (Fig. 1). K-ras protein was detectable in the parental cells and was present in increased amounts in most of the transfectants. One clone, 13, had levels of K-ras below that of the parental Caco-2 cells,whereas clones 14, 22, 59, 80, 74, and 87 were intermediate expressors,and contained levels of K-ras similar to the uncloned K-ras-transfected population. Clones 60, 66, and 96 showed the highest level of expression. Clones 60 and 66 were used for further study.

Growth of parental Caco-2 cells as well as clones 60 and 66 were measured in the presence of vehicle, 120 μm sulindac sulfide, or 600 μm sulindac sulfone (Fig. 2). In the presence of vehicle only, the expression of activated K-ras did not confer a growth advantage to either transfected cell line. The number of viable cells in all sulfide-treated cultures remained relatively unchanged throughout the 10-day growth period, whereas treatment of cultures with the sulfone derivative resulted in a pronounced cell loss beginning 6 days after treatment. K-ras activation did not confer a selective toxicity in the presence of either sulindac metabolite.

Sulindac has been shown to cause apoptosis in the AOM-treated rodent model. In this model, sulindac also reduces the number of tumors harboring an activated K-ras, which suggests that apoptosis may occur by a K-ras-dependent mechanism. To test this hypothesis, Caco-2 parental cells and clones 60 and 66 were treated with vehicle, 120 μm sulindac sulfide, or 600μ m sulindac sulfone, and apoptosis of each culture was quantitated by light microscopy (Fig. 3). The percentage of cells undergoing apoptosis in response to vehicle alone were minimal, with a peak apoptosis index of 4%. Caco-2 parental cells exposed to sulfide displayed a gradual increase in apoptosis that peaked at 8 days in culture. Clone 60 cells, and to a lesser extent,clone 66, displayed an earlier onset of apoptosis. Clone 60 reached peak apoptosis levels at day 2 compared with day 8 of the parental cells, whereas clone 6 reached a peak level of apoptosis by day 6. In both the Caco-2 parental and clone 60 cells, induction of apoptosis by sulindac sulfone was more pronounced than with the sulfide. Upon exposure to the sulfone, clone 60 cells reached a peak level of apoptosis by day 4, compared with day 8 in the parental cells. K-ras appears to accelerate the onset of apoptosis in response to both sulindac sulfone and sulfide.

In an effort to better characterize the apoptosis response to the sulindac metabolites in cells with an activated K-ras, the colony-forming efficiencies of Caco-2 and clone 60 cells were measured after exposure to increasing concentrations of sulindac sulfide or sulfone (Fig. 4). Cells were treated for 24 h and then replated at diluted concentrations in fresh media. By this method, a 50% decrease in cell number occurred at approximately 225 μmsulindac sulfide and 800 μm sulindac sulfone. The expression of an activated K-ras did not affect the final measure of viability caused by either sulindac metabolite,because the transfected cells displayed the same toxicity as the parental cells. Activation of K-ras was unable to confer a selective cytotoxicity in the presence of either sulindac metabolite.

Caco-2 cells and K-ras transfectants exhibit a growth cessation in response to polyamine depletion, which is not accompanied by a loss of viability in short-term experiments. Polyamine depletion in these experiments appears to have no effect on sulindac-induced cytotoxicity. To further characterize the effects of polyamine depletion, both alone and in combination with sulindac, colony-forming efficiencies of Caco-2 and clone 60 cells were analyzed after exposure to 5 mm DFMO and increasing concentrations of sulindac sulfide or sulfone (Fig. 5). Exposure of the K-ras-transfected cells to 5 mm DFMO had a much more pronounced effect on cell survival than that exhibited by the parental Caco-2 cells. Treatment of the transfected cells with 5 mm DFMO alone resulted in a 60% decrease in cell survival compared with untreated transfectants, whereas in the parental cells, DFMO treatment alone had little effect. For both Caco-2 and clone 60 cell lines, the combination of DFMO and increasing concentrations of either sulindac metabolite resulted in an even further decrease in cell survival, which suggests that DFMO did not inhibit sulindac-induced cytotoxicity.

When the survival of DFMO-treated clone 60 or parental Caco-2 cells was normalized to like DFMO-treated controls, the survival rates of the DFMO-treated cells mirrored those of the non-DFMO-treated cells in response to increasing concentrations of sulindac sulfide (Fig. 6). The interaction of DFMO with either sulindac sulfide or sulindac sulfone was evaluated using the analysis of covariance model, where the dose was the covariant variable. This analysis was unable to detect any interaction between either DFMO and sulindac sulfide or DFMO and sulindac sulfone (Table 1).

The 60% decrease in survival of clone 60 cells treated with 5 mm DFMO suggests that polyamine depletion was selectively toxic to cells expressing an activated K-ras. To further examine this response, Caco-2 cells and clones 60 and 66 were treated with increasing concentrations of DFMO for 24 h and then replated for colony formation (Fig. 7). Caco-2 cells were refractory to DFMO treatment and maintained nearly 100% survival at concentrations up to 5 mm DFMO. In contrast, both clones 60 and 66 exhibited increasing toxicity with increasing concentrations of DFMO, with 50% cell survival occurring at only 0.5 mm DFMO. At 5 mmDFMO, cell survival for both transfectants was decreased to 20–40% of controls. In these experiments, polyamine depletion was selectively toxic to cells with an activated K-ras at concentrations as low as 50 μm. Cultures treated with DFMO for varying times and concentrations were evaluated for induction of apoptosis, using assessment of morphological criteria as described in“Materials and Methods.” We found no evidence for induction of apoptosis in any DFMO-treated cultures. Thus, although DFMO killed cells expressing the activated ras gene, this process involved an apoptosis-independent mechanism.

In the present study, the polyamine synthesis inhibitor DFMO was selectively toxic to cells containing an activated K-ras. K-ras activation did not affect the quantitative level of cytotoxicity, as measured by colony formation, exerted by either metabolite of the NSAID sulindac. However, the rate of killing, as measured by apoptosis induction, induced by the sulindac metabolites was increased by K-ras activation. These results suggest that the cytotoxic effects of DFMO and the NSAIDs occur through unique K-ras-dependent mechanisms.

The expression of an activated K-ras did not appear to change the growth rate of transfected Caco-2 cells, even in clones that expressed high levels of the protein. Thus, the role of activated K-ras expression is cell specific. Activated K-ras expression is associated with an increase in growth rates in DLD-1 and HCT116 human colon cell lines (39). Others have also failed to observe an effect of activated H-ras expression on growth of Caco-2 cells (40). However, both de Vries et al.(40) and we found that ras activation increases tumorigenicity of Caco-2 cells. Just as ras plays pivotal roles in proliferation, ras also participates in several mechanisms leading to apoptosis. Activated K-ras increases the rate of ceramide-induced apoptosis in colon cancer cells (41). Our findings indicating enhanced rates of killing induced by sulindac derivatives may be related, because NSAIDs may induce apoptosis in part by a ceramide-dependent mechanism (42). The role of ras in regulating proliferation and apoptosis have been reviewed recently (43, 44).

A number of studies suggest that NSAID-mediated chemoprevention occurs through the induction of apoptosis (13, 15, 45, 46). However, the role of K-ras activation in this process has not been investigated extensively. One group has reported that K-ras activation in rat enterocytes confers resistance to apoptosis induced by sulindac sulfide, although not with sulfone (47). In the present study, we found that treatment of Caco-2 cells with sulindac sulfide and sulfone resulted in a K-ras-dependent induction of apoptosis. Apoptosis induction by both sulfide and sulfone occurred several days earlier in K-ras-transfected cells than in parental cells, although maximal levels of apoptosis in both transfectants and parental cells were similar. Our results are consistent with those of other groups who showed that ras activation induces the expression of the antiapoptotic protein COX-2 (48). Similarly, we found that K-ras activation in Caco-2 cells up-regulates COX-2 (49). The difference between our results and those of Arber et al.(47) is unknown but may relate to different patterns of gene expression induced by K-ras in rat enterocytes compared with human cells.

The K-ras-dependent acceleration of apoptosis seen with sulindac treatment appeared to be inconsequential with respect to overall cytotoxicity. Survival studies with increasing concentrations of sulindac sulfide or sulfone did not reveal any differences in colony-forming efficiency between K-ras transfectants and parental Caco-2 cells, although cytotoxicity in both cell types was correlated with increased drug concentrations. These data suggest that although K-ras activation may increase the rate at which cells die from apoptosis, the final level of cell survival influenced by sulindac metabolites is unaffected by an activating mutation in K-ras.

Treatment of Caco-2 or cells expressing either a normal or mutant K-ras with DFMO and either sulindac sulfide or sulfone resulted in additive effects on cell survival. Our results corroborate previous studies in animal models with DFMO and the NSAID piroxicam (7). In the present study, treatment of Caco-2 cells with DFMO did not increase the toxicity of the sulindac metabolites in long-term survival studies. Thus, DFMO and sulindac appear to use unique mechanisms to suppress cell viability. This possibility is further supported by the selective cytotoxicity of DFMO seen in cells with an activated K-ras. Treatment of both K-rasclones 60 and 66 with increasing concentrations of DFMO led to a dose-dependent decrease in colony-forming efficiency, whereas survival of parental cells was unaffected. Twenty-four h of treatment with 5 mm DFMO was sufficient to reduce colony-forming efficiency of the K-ras transfectants by 40%, although cells were subsequently incubated in normal media for an additional 24 h prior to replating. Concentrations as low as 50μ m DFMO were also moderately cytotoxic. In short-term growth experiments, where cells were exposed to 5 mm DFMO continuously over a 6-day time course,exclusion of trypan blue dye in either the K-ras cells or parentals was unaffected, and we failed to observe any apoptotic cells in DFMO-treated cultures (data not shown). The cells appeared to be metabolically active during this time period, yet the colony-forming efficiency data indicate they were unable to proliferate. The cytotoxicity of DFMO on cells with an activated K-ras is consistent with other studies showing potent chemopreventive effects of DFMO on AOM-induced tumorigenesis as well as the development of tumors with an activated K-ras (8, 24).

The mechanism of K-ras-dependent toxicity of DFMO is not yet established. In results to be presented elsewhere, we show that K-ras activation suppresses the expression of the spermidine spermine N′-acetyltransferase, the first enzyme in polyamine catabolism.⁵Treatment of cells expressing an activated K-ras with DFMO are thus unable to generate new polyamines either from synthesis or catabolism. This deficit may be sufficient to kill these cells.

Administration of both piroxicam and DFMO in the AOM-rat model decreased both the number of existing tumors as well as the number of tumors with an activated K-ras (7). The mechanism by which DFMO prevents K-ras-dependent tumorigenesis in the AOM-rat model is unknown. It has been proposed that NSAIDs and DFMO act through a common pathway involving the inhibition of prostaglandin synthesis (7), because ODC activity can be inhibited by agents that inhibit COXs (7, 50). The induction of apoptosis may occur through the inhibition of prostaglandin synthesis, or in the case of sulindac sulfone, a prostaglandin-independent mechanism that is also independent of K-ras activation. The activation of K-ras may lead to the down-regulation of genes that are not necessary for growth but are necessary for cell survival, such as spermidine spermine N′-acetyltransferase. This down-regulation may not cause toxicity unless the cell undergoes the additional stress of polyamine depletion, in which case the genes normally expressed during conditions of stress are not available to protect the cell from a cytotoxic response.