Nonsteroidal Anti-inflammatory Drugs Induce Colorectal Cancer Cell Apoptosis by Suppressing 14-3-3ε

Human and animal studies have shown that nonsteroidal anti-inflammatory drugs (NSAIDs) suppress colorectal tumorigenesis (1). Population-based epidemiologic studies have shown that use of aspirin and NSAID reduces risk of colorectal cancer (2, 3). Animal experiments provide evidence for control of colorectal cancer by NSAIDs such as sulindac and selective cyclooxygenase-2 (COX-2) inhibitors (4, 5). Results from clinical trials have documented the efficacy of sulindac and selective COX-2 inhibitors in reducing the size and numbers of colon adenomatous polyps in patients with familial adenomatous polyposis, a precancerous condition (6, 7). The action of NSAIDs on controlling colorectal cancer is attributed to the inhibition of COX-2 (8), which is overexpressed in a majority of patients with colorectal cancer (9, 10). NSAID could also exert its action on colorectal cancer in a COX-2–independent manner (11). It is well documented that NSAIDs, notably sulindac, indomethacin, and selective COX-2 inhibitors exert their actions by inducing colorectal cancer cell apoptosis (12, 13). The mechanism by which these agents induce apoptosis is not entirely clear. It was proposed that NSAID-induced apoptosis might be mediated by down-regulation of peroxisome proliferator–activated receptor-δ (PPARδ; ref. 14). Several studies have shown that PPARδ is overexpressed in colorectal cancer and PPARδ is activated by prostacyclin generated by colorectal cancer cells (15). Activation of PPARδ accelerates intestinal adenoma growth whereas genetic disruption of PPARδ decreases the tumorigenicity of human colorectal cancer cells (16, 17). It has been reported that PPARδ plays an important role in cellular resistance to apoptotic insults (18, 19). Activation of PPARδ was reported to be associated with increased expression of phosphoinositide-dependent kinase-1 and activation of Akt (20). Akt is recognized as a key signaling molecule for transmitting survival messages. We have recently reported that PPARδ activation by prostacyclin or L-165041, a synthetic PPARδ ligand, increases the expression of 14-3-3 proteins, especially the 14-3-3ε isoform (21). Promoter analysis identifies three PPAR response elements (PPRE) located between −1426 and −1477 of the 14-3-3ε promoter region. Deletion of these PPREs abrogates the protective effect of prostacyclin or L-165041 (21). Our previous findings suggest that PPARδ-mediated 14-3-3ε up-regulation plays a crucial role in endothelial cell resistance to apoptosis (21). It is unclear whether 14-3-3ε is involved in colorectal cancer cell survival. We postulated in this study that NSAIDs induce apoptosis by reducing 14-3-3ε via suppressing PPARδ. Reduction in 14-3-3ε protein levels results in mitochondrial Bad translocation and colorectal cancer cell apoptosis. The results show that sulindac sulfide, sulindac sulfone, indomethacin, and SC-236 suppressed HT-29 14-3-3ε protein and promoter activity, which correlated with the suppression of PPARδ protein expression. The reduction in 14-3-3ε levels was accompanied by enhanced mitochondrial Bad translocation and HT-29 apoptosis.

Cell culture and reagents. HT-29 and DLD-1 cells were obtained from American Type Culture Collection (Manassas, VA). HT-29 were maintained in DMEM and DLD-1 were maintained in RPMI 1604 supplemented with 10% fetal bovine serum and penicillin-streptomycin at 37°C in a humidified 5% CO₂ atmosphere. Cell culture media and antibiotics were purchased from BRL Life Technologies (Grand Island, NY). Sulindac sulfide and sulindac sulfone were obtained from ICN Biomedicals, Inc. (Solon, OH). Indomethacin, SC-236, MG-132, and cell-permeable caspase 3 inhibitor, ac-DVED aldehyde (DEVD-CHO), were from Calbiochem (San Diego, CA). We used 10 μmol/L of MG-132 and 50 μmol/L of DEVD-CHO, which were reported to block proteasome and caspase 3 activities, respectively (22, 23).

Recombinant adenoviral vectors. Replication-defective recombinant adenoviruses were generated by homologous recombination and amplified in 293 cells as described previously (24, 25). The recombinant viruses were purified by CsCl density gradient centrifugation and the virus titers were determined by a plaque assay as previously described (24). Ad-PPARδ was kindly provided by Drs. Kinzler and Vogelstein at Johns Hopkins University. HT-29 cells were transfected with Ad-GFP (green fluorescent protein) or Ad-PPARδ at a multiplicity of infection (moi) of 50 for 48 h before treatment with indomethacin or sulindac sulfide.

Preparation of mitochondrial fraction. HT-29 cells treated with indicated reagents (sulindac sulfide, 160 μmol/L; sulindac sulfone, 300 μmol/L; indomethacin, 800 μmol/L; or SC-236, 40 μmol/L) for 8 h were washed thrice with PBS and harvested by centrifugation. Mitochondrial fractions were prepared by a mitochondria isolation kit (Sigma, St. Louis, MO) as described previously (26). The mitochondrial and cytosolic fractions were purified by two-step gradient centrifugation and stored at −20°C. Mitochondrial Bad protein level was analyzed by Western blotting. Heat shock protein 60 (Hsp60) was used as a mitochondrial marker.

Immunoprecipitation. HT-29 cells treated with the indicated reagents were harvested and lysed in 1× radioimmunoprecipitation assay buffer (Upstate, Charlottesville, VA) with cocktail protease inhibitors (Sigma). Lysate proteins (200 μg) were immunoprecipitated with a human 14-3-3ε antibody. The immunoprecipitated complex was pulled down with protein A/G-agarose (Santa Cruz Biotechnology, Santa Cruz, CA). After washing five times, the proteins were analyzed by Western blotting using a Bad antibody.

Western blot analysis. Cells were lysed in radioimmunoprecipitation assay buffer containing protease inhibitors. Protein concentrations were determined by a Bio-Rad (Hercules, CA) protein assay kit. Cell lysate proteins (25 μg) were applied to each lane and analyzed by Western blots as previously described (25). Rabbit polyclonal antibodies against 14-3-3ε (diluted at 1:250), and goat polyclonal antibodies against Hsp60 (1:2,000) were purchased from Santa Cruz Biotechnology. Rabbit polyclonal antibody against PPARδ (1:500) was obtained from Cayman Chemical (Ann Arbor, MI). Rabbit polyclonal antibody against Bad (1:500) was purchased from Cell Signaling (Beverly, MA). Monoclonal antibody against actin (1:5,000) was obtained from Oncogene (Boston, MA). Donkey anti-rabbit or donkey anti-mouse IgG conjugated with horseradish peroxidase were purchased from Santa Cruz Biotechnology. Protein bands were visualized by an enhanced chemiluminescence system (Pierce, Rockford, IL).

Assay of apoptosis. Apoptotic cells were analyzed by Hoechst staining according to the instructions of the manufacturer (Invitrogen, Carlsbad, CA). Hoechst 33258 is a cell-permeable dye which is sensitive to nuclear DNA damage and is useful in detecting apoptosis of unfixed living cells (27). In brief, HT-29 cells transiently transfected with Ad-PPARδ, Ad-GFP, or stably transfected HT-29 cells were treated with 160 μmol/L of sulindac sulfide for 24 h. After washing, they were incubated with 1 μg/mL of Hoechst 33258 in the dark for 15 min, centrifuged and washed with PBS. The washed cells were examined under fluorescent microscopy (excitation at 352 nm and emission at 461 nm). Apoptotic cells exhibited strong nuclear staining with overt nuclear morphologic changes as previously described (27). The percentage of apoptotic cells was determined by counting at least 200 cells.

Plasmid constructs and luciferase reporter assay. A human 14-3-3ε expression construct was amplified by PCR and subcloned into pCDNA 3.1+ vector (Invitrogen) and a 14-3-3ε promoter subcloned into pGL3 vector were prepared as previously described (21). PPRE reporter (14) was kindly provided by Drs. Kinzler and Vogelstein at Johns Hopkins University. For the luciferase assay, cells were transfected with the reporter constructs by Effectene transfection kit (Qiagen, Valencia, CA) for 48 h. After treatment with the indicated reagents, cells were lysed and luciferase activity was measured using a kit from Promega (Madison, WI), and the emitted light was determined in a luminometer.

Establishment of 14-3-3ε stable cell line. Control vector (pCDNA3.1+) and human 14-3-3ε expression vectors which contain a neomycin-resistant cassette were transfected into HT-29 cells by using an Effectene transfection kit for 48 h. The transfected cells were screened by incubation with 500 μg/mL of G418 (Sigma) for 4 weeks. The cell lines which stably expressed 14-3-3ε were isolated from individual colonies and maintained in DMEM with 10% fetal bovine serum and 200 μg/mL of G418.

Statistical analysis. ANOVA was used to analyze statistical differences among multiple groups. P < 0.05 was considered to be statistically significant.

Suppression of 14-3-3ε by sulindac sulfide and COX-2 inhibitors. To evaluate the effect of sulindac on 14-3-3ε protein levels, we treated HT-29 with sulindac sulfide at 0 to 160 μmol/L for 4 to 24 h. 14-3-3ε proteins were analyzed by Western blotting. Sulindac sulfide did not have an effect at 4 h of treatment, suppressed 14-3-3ε at 8 h only at 160 μmol/L, and exerted a concentration-dependent inhibition of 14-3-3ε at 24 h of treatment (Fig. 1A). Sulindac sulfone had no effect at 300 μmol/L but suppressed 14-3-3ε expression at 600 μmol/L (Fig. 1B). Indomethacin, a nonselective COX inhibitor, or SC-236, a selective COX-2 inhibitor, also concentration-dependently inhibited 14-3-3ε protein levels (Fig. 1C). Sulindac sulfide suppressed 14-3-3ε protein levels in DLD-1 with a pattern similar to that of HT-29 (Fig. 1D).

Inhibition of 14-3-3ε promoter activity by sulindac sulfide. To determine whether sulindac sulfide inhibits 14-3-3ε at the transcription level, we transfected HT-29 with a human 14-3-3ε promoter construct (−1628/+24) and treated the transfected cells with various concentrations of sulindac sulfide for 24 h. Sulindac sulfide concentration-dependently inhibited 14-3-3ε promoter activity (Fig. 2A). Sulindac-induced 14-3-3ε protein suppression was not influenced by MG-132, a proteasome inhibitor, or DEVD-CHO, a caspase 3 inhibitor (Fig. 2B). Taken together, these results suggest that sulindac suppresses 14-3-3ε primarily at the transcriptional level.

Restoration of 14-3-3ε by adenoviral PPARδ transfer. We have recently discovered that 14-3-3ε transcription is mediated by PPARδ (21). We were, therefore, interested in learning whether sulindac-induced 14-3-3ε suppression involves PPARδ. HT-29 cells were treated with sulindac sulfide (160 μmol/L) or indomethacin (800 μmol/L) for 24 h and PPARδ proteins were analyzed by Western blots. PPARδ levels became undetectable following treatment with either agent (Fig. 3A). To ensure that PPARδ is active in HT-29, we transfected the cells with a PPRE construct and treated the cells with sulindac sulfide for 24 h. Sulindac reduced the PPRE activity in a concentration-dependent manner (Fig. 3B). To determine whether PPARδ overexpression rescues 14-3-3ε from sulindac or indomethacin suppression, we transduced HT-29 with Ad-PPARδ (50 moi) or Ad-GFP control (50 moi). PPARδ protein levels were increased by Ad-PPARδ transduction (Fig. 4A), which is accompanied by enhanced PPRE activity (Fig. 4B) and 14-3-3ε promoter activity (Fig. 4C). PPARδ proteins in Ad-PPARδ–transduced cells were not altered by sulindac or indomethacin (Fig. 4A), but the PPARδ transcriptional activity (Fig. 4B) and the 14-3-3ε promoter activity (Fig. 4C) were partially suppressed. 14-3-3ε proteins were increased by Ad-PPARδ overexpression which were suppressed by sulindac sulfide and indomethacin (Fig. 4D).

Mitochondrial Bad translocation and cell death by sulindac and COX-2 inhibitors. 14-3-3 proteins bind phosphorylated Bad and sequester it in cytosol. Apoptotic signals induce Bad to release from 14-3-3 and translocate to the mitochondria where it disrupts the protective action of Bcl-XL, thereby promoting apoptosis via the mitochondrial pathway (28, 29). It was reported that up-regulation of 14-3-3ε protein levels by PPARδ ligands enhances Bad binding and reduces Bad translocation to mitochondria (21). Because sulindac suppresses 14-3-3ε in HT-29, it is predicted that sulindac reduces cytosolic Bad sequestration and reciprocally enhances Bad localization to mitochondria. To confirm this, we did two types of experiments. First, we prepared mitochondrial and cytosolic fractions from HT-29 treated with or without sulindac sulfide or sulfone, indomethacin, or SC-236, and Bad in the mitochondrial fraction was analyzed by Western blotting. As a control, the mitochondria-resident protein Hsp60 was concurrently analyzed. Second, we immunoprecipitated cell lysates of HT-29 treated with or without sulindac sulfide or indomethacin by a 14-3-3ε antibody and analyzed Bad in the immunoprecipitate with a Bad antibody. The results show that in the absence of NSAIDs, most Bad were localized to the cytosol, whereas Hsp60 was predominantly detected in the mitochondrial fraction with a trace quantity detected in the cytosolic fraction (Fig. 5A). Bad was detected at a very low level in mitochondrial fraction (Fig. 5A). Sulindac and indomethacin enhanced mitochondrial Bad (Fig. 5A) and suppressed binding of Bad to 14-3-3ε (Fig. 5B). SC-236 enhanced mitochondrial Bad to an extent comparable to sulindac sulfide and indomethacin (Fig. 5A). These results suggest that sulindac, indomethacin and SC-236 induce Bad translocation to mitochondria as a consequence of severe 14-3-3ε suppression.

Rescue of NSAID-induced apoptosis by Ad-PPARδ and 14-3-3ε. Sulindac sulfide and indomethacin induced HT-29 apoptosis as evidenced by an increase in Hoechst-positive cells. Apoptosis was reduced by transduction of cells with Ad-PPARδ (Fig. 5C). To ascertain the role of 14-3-3ε in NSAID-induced HT-29 apoptosis, we prepared a HT-29 cell line overexpressing 14-3-3ε (HT-29^14-3-3ε) and a control cell line (HT-29^CTR; Fig. 6A). In HT-29 control cells, sulindac suppressed 14-3-3ε proteins (Fig. 6A) accompanied by a large increase in apoptosis (Fig. 6B). Overexpression of 14-3-3ε proteins in HT-29^14-3-3ε cells (Fig. 6A) prevented apoptosis by >50% (Fig. 6B).

Our findings show for the first time that sulindac and other NSAIDs drastically suppress 14-3-3ε expression in colorectal cancer cells, resulting in reduced Bad sequestration in cytosol, increased mitochondrial Bad, and enhanced apoptosis. The 14-3-3 family comprises at least seven members in mammalian cells (30). They share sequence homology and biochemical properties, interact with more than 100 cellular proteins, and have diverse biological activities (31). They bind phosphorylated Bad, thereby sequestering Bad in cytosol and suppressing Bad-induced mitochondrial injury and the consequential apoptosis via the intrinsic (mitochondrial) apoptotic pathway (32). It was reported in endothelial cells that among the 14-3-3 members, 14-3-3ε is expressed in abundance, and its cellular level is regulated and is crucial in protecting cells from oxidative damage (21). The results in this study show that 14-3-3ε is constitutively expressed in HT-29 and DLD-1 cells and is a target of NSAID-induced apoptosis. Sulindac, indomethacin, and SC-236 suppress 14-3-3ε expression in HT-29 accompanied by increased translocation of Bad to mitochondria and apoptosis. Stable overexpression of 14-3-3ε alleviates sulindac-induced apoptosis. Taken together, these results strongly suggest that the suppression of 14-3-3ε is a major mechanism by which NSAIDs induce colorectal cancer apoptosis.

14-3-3ε transcription requires PPARδ and is regulated by the level of PPARδ, as evidenced by increased 14-3-3ε proteins in Ad-PPARδ–transduced cells. Our results are consistent with a previous report that sulindac suppresses PPARδ transcription activity (14), and provide new data which attribute reduced PPARδ transcription activity to the suppression of PPARδ protein expression. The results show a good correlation between sulindac- and indomethacin-induced 14-3-3ε and PPARδ suppressions. Furthermore, PPARδ overexpression by Ad-PPARδ transduction rescues 14-3-3ε from suppressions by sulindac sulfide at 80 μmol/L or indomethacin at 400 μmol/L. These data indicate that sulindac and other NSAIDs suppress PPARδ expression, thereby blocking PPARδ-mediated 14-3-3ε transcription. Although Ad-PPARδ abrogates 14-3-3ε suppression by sulindac and indomethacin at 80 and 300 μmol/L, respectively, it only partially corrected 14-3-3ε levels suppressed by higher concentrations of sulindac (160 μmol/L) or indomethacin (800 μmol/L). Lack of complete correction by Ad-PPARδ may be due to the following reasons. First, activated PPARδ promotes transcription by forming a heterodimer with RXR nuclear receptor (14). Overexpression of PPARδ alone without concurrent RXR overexpression may, therefore, be inadequate to exert maximal transcriptional activation. Second, sulindac and other NSAIDs have diverse COX-2–dependent and COX-2–independent effects on signaling transduction and gene expression. Lack of a complete rescue of 14-3-3ε suppression and cell apoptosis may suggest that PPARδ is not the only pathway via which NSAIDs suppress 14-3-3ε. Furthermore, differential effects of Ad-PPARδ (50 moi) on rescuing 80 versus 160 μmol/L sulindac-induced 14-3-3ε suppression may reflect dose-dependent effects on diverse cellular functions. Lastly, lack of a correlation between the magnitude of PPARδ overexpression and 14-3-3ε rescue could also be attributed to the requirement of PPARδ posttranslational modification for its full transactivation activity. Because neither MG-132 nor DEVD-CHO blocks the 14-3-3ε–suppressing effect of sulindac, it is unlikely that the reduction in 14-3-3ε protein levels are due to degradation by ubiquitin-proteasome or caspase 3.

PPARδ expression is regulated by the β-catenin-Tcf (T cell factor) transcriptional program (14). The wild-type adenomatous polyposis coli tumor suppressor gene product binds and targets β-catenin for degradation, thereby suppressing PPARδ expression (14). Adenomatous polyposis coli mutation results in increased β-catenin, which translocates to the nucleus and complexes with Tcf to form an active transcription factor that binds to specific responsive elements on PPARδ and up-regulates PPARδ expression in familial adenomatous polyposis (14, 33). It is reported that sulindac and celecoxib inhibit β-catenin expression and/or the transcriptional activity of the β-catenin/Tcf complex (34, 35). It is, therefore, possible that sulindac, SC-236, and other NSAIDs suppress PPARδ expression through the β-catenin/Tcf transcription pathway.

Sulindac and other NSAIDs may induce colorectal cancer cell apoptosis by COX-2–dependent and COX-2–independent mechanisms (36–39). COX-2 is overexpressed in colorectal cancer and COX-2-derived prostaglandin I₂ (PGI₂) has been shown to activate PPARδ in colorectal cancer cells (15). PPARδ activation confers resistance to apoptosis in colorectal cancer as well as a number of non–cancer cells such as endothelial, renal interstitial, cardiomyocytes and keratinocytes (18, 19, 21). Sulindac sulfide and selective COX-2 inhibitors may induce apoptosis by blocking COX-2 derived PGI₂ production, thereby suppressing PPARδ activation and the antiapoptosis action of PPARδ. COX-2–derived prostaglandin E₂ (PGE₂) up-regulates β-catenin expression (40) and NSAIDs may further suppress β-catenin–mediated PPARδ expression by blocking PGE₂ production. However, a recent report has shown that sulindac, selective COX-2 inhibitors, and a host of classic NSAIDs inhibit β-catenin by a COX-2–independent mechanism (41). Our results reveal that sulindac sulfone, which does not inhibit COX-2, is capable of inhibiting 14-3-3ε expression, suggesting that NSAIDs suppress 14-3-3ε by a COX-2–independent mechanism. Regardless of the COX-2 involvement, an important finding is that reduction in 14-3-3ε levels by sulindac and other NSAIDs results in a decline in Bad sequestration and an increase in Bad translocation to the mitochondria where Bad binds and inactivates the antiapoptotic Bcl-2 family proteins, notably Bcl-XL and Bcl-2. It was reported that sulindac suppresses Bcl-XL expression, creating an imbalance between Bcl-XL and Bax, and subsequent mitochondria-mediated apoptosis (42). Moreover, sulindac inhibits Bcl-2 expression (43), further enhancing Bad- and Bax-mediated mitochondria damage, Smac/DIABLO release, and apoptosis (44). Taken together, our results and the published data suggest that sulindac and other NSAIDs induce colorectal cancer cell apoptosis by multiple mitochondrial apoptotic processes as proposed in Fig. 6C. First, NSAIDs suppress PPARδ, thereby reducing 14-3-3ε level and Bad sequestration. Second, NSAIDs inhibit extracellular signal–regulated kinase, thereby blocking Bad phosphorylation (45), and further reducing Bad sequestration leading to excessive Bad translocation to the mitochondria. Lastly, NSAIDs inhibit Bcl-XL and Bcl-2 expression resulting in a reduction in the ratio of antiapoptotic to proapoptotic Bcl-2 family proteins on the mitochondrial membrane with consequent loss of mitochondrial membrane potential, release of proapoptotic factors especially Smac/DIABLO resulting in caspase activation and apoptosis. These processes are interrelated and mutually reinforcing. As PPARδ or 14-3-3ε overexpression rescues apoptosis, the PPARδ/14-3-3ε transcriptional pathway represents a major pathway for colorectal cancer cell survival.

Grant support: NIH (National Heart, Lung, and Blood Institute, HL-50675).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Susan Mitterling and Nathalie Huang for excellent editorial assistance.