|Home | About | Journals | Submit | Contact Us | Français|
Insulin-like growth factor 2 mRNA-binding protein-1 (IMP-1) is an oncofetal protein that binds directly to and stabilizes oncogenic c-Myc and regulates in turn its post-transcriptional expression and translation. In contrast to normal adult tissue, IMP-1 is re-expressed and/or overexpressed in human cancers. We demonstrate that knock-down of c-Myc in human colon cancer cell lines increases the expression of mature let-7 miRNA family members and downregulates several of its mRNA targets: IMP-1, Cdc34, and K-Ras. We further demonstrate that loss of IMP-1 inhibits Cdc34, Lin-28B, and K-Ras, and suppresses SW-480 cell proliferation and anchorage-independent growth, and promotes caspase and lamin-mediated cell death. We also found that IMP-1 binds to the coding region and 3′UTR of K-Ras mRNA. RNA microarray profiling and validation by reverse transcription PCR reveals that the p53-inducible pro-apoptotic protein, CYFIP2, is upregulated in IMP-1 knock-down SW480 cells, a novel finding. We also show that overexpression of IMP-1 increases c-Myc and K-Ras expression, and LIM2405 cell proliferation. Furthermore, we show that loss of IMP-1 induces Caspase-3 and Parp–mediated apoptosis, and inhibits K-Ras expression in SW480 cells, which is rescued by CYFIP2 knock-down. Importantly, analysis of 228 patients with colon cancers reveals that IMP-1 is significantly upregulated in differentiated colon tumors (p ≤ 0.0001) and correlates with K-Ras expression (r=0.35, p ≤ 0.0001) relative to adjacent normal mucosa. These findings indicate that IMP-1, interrelated with c-myc, acts upstream of K-Ras to promote survival through a novel mechanism that may be important in colon cancer pathogenesis.
Colorectal cancer (CRC) is the second leading cause of cancer mortality in the United States. Deregulated expression of the oncogenic transcription factor c-Myc occurs in a broad range of human cancers and is often associated with poor prognosis, indicating a key role for this oncogene in tumor progression. Importantly, c-Myc expression is elevated in approximately 70% of colon tumors due to defective Wnt signaling (1, 2). Recent evidence indicates that c-Myc induction causes global transcriptional repression of human microRNA (miRNA) genes, including members of the let-7 family that are downregulated in human cancer cells and tumors (3, 4). The let-7 family of miRNAs act as tumor suppressors and inhibit the translational expression of oncogenic mRNAs including K-Ras, c-Myc, Hmga2, and Cdc34, and suppress cancer cell growth, proliferation, and tumor formation in vivo (5-12). K-Ras is frequently mutated in human tumors and plays key roles in regulating diverse cellular pathways important for cell growth, differentiation, and survival (13). Indeed, 40-50% of human colon cancers harbor activating mutations in the K-Ras proto-oncogene and is associated with progression from an adenoma to adenocarcinoma. Thus, the K-Ras signaling pathway represents an attractive target for cancer therapy (14-18).
The human c-Myc mRNA coding region determinant-binding protein (CRD-BP), also known as insulin-like growth factor2 (IGF2) mRNA-binding protein (IMP-1), is expressed during early embryonic mammalian development and functions in translational stability by binding and shielding several mRNAs that play critical roles in cell growth and proliferation from proteolytic degradation including c-Myc (19-24). Consistent with it's oncofetal function, loss of IMP-1 in mice causes perinatal lethality, dwarfism, and impaired intestinal morphogenesis (25). In striking contrast to normal adult tissues, IMP-1 re-expression has been reported in breast, ovarian, and colorectal tumors (26). Furthermore, IMP-1 is a positive predictor of poor clinical outcome in colon cancer patients (27). Recent work has revealed that the β-catenin/Tcf complex upregulates IMP-1 mRNA and protein expression, necessary for the stabilization and induction of c-Myc and β-TrCP1 mRNAs in CRCs, and maybe involved in the suppression of apoptosis (24, 28). Moreover, increased IMP-1 levels positively correlate with activation of β-catenin/Tcf signaling in primary colorectal tumors (24). Importantly, IMP-1 is a direct let-7 target and promotes cell cycle progression, growth, and migration (29). These studies suggest IMP-1 plays a role in regulating human cancer progression.
Herein, we report a molecular mechanism by which c-Myc positively modulates IMP-1 expression in colon cancers, in part by negative regulation of let-7 miRNAs. We also show that loss of IMP-1 downmodulates K-Ras expression downstream of β-catenin, and concomitantly inhibits colon cancer cell proliferation, anchorage-independent growth, and survival in monolayer and organotypic (3D) cell culture. Furthermore, we identify a novel pro-apoptotic gene target, CYFIP2, which is downregulated by IMP-1, and mediates the regulation of cell survival and K-Ras expression in colon cancer cells. In contrast to our knock-down studies, IMP-1 overexpression increases c-Myc and K-Ras expression, and colon cancer cell proliferation. For the first time we find that IMP-1 directly interacts with K-Ras mRNA and is highly elevated in colon cancer cells and tumors and positive correlates with K-Ras relative to normal mucosa, thus suggesting a novel interrelationship with K-Ras in vivo.
Caco-2, SW-480, and LIM2405 cells were purchased from the American Type Culture Collection (ATCC) and maintained in accordance with the manufacturer's recommendations. Cells were transfected with human β-catenin siRNA (gift from Dr. Jeffrey Drebin, University of Pennsylvania), SignalSilence c-Myc siRNAs I and II (Cell Signaling Technology), IMP-1 siRNA, CYFIP2 siRNA (Santa Cruz Biotechnology), or scrambled negative control siRNA (Ambion) using Lipofectamine 2000 reagent (Invitrogen), as per manufacturers instructions. The IMP-1/pMSCV-PIG retroviral vector was purchased from Addgene (plasmid 21659) and used to stably infect LIM2405 cells as described previously (30, 31).
Total RNA was isolated from tissues and cultured cells using the mirVana miRNA isolation kit (Ambion) and analysis of mRNA expression by reverse transcription-PCR (RT-PCR) as described previously (32). Normal human colon was obtained from the Cooperative Human Tissue Network at the Hospital of the University of Pennsylvania (with Institutional Review Board approval). Human primary let-7a3-b intron, CYFIP2, and β-actin PCR products were amplified using the following oligonucleotide primer pairs:
RT–PCR products were resolved by 1% TAE agarose gel electrophoresis.
Quantitative Real-time PCR (qRT-PCR) was performed on an Applied Biosystems 7900HT Real-Time PCR System. The reverse transcription was performed using the TaqMan® miRNA Transcription kit, followed by quantification of hsa-IMP-1 and mature hsa-let-7a and -7b, using predesigned TaqMan® Assays (Applied Biosystems), according to the manufacturer's recommendations. β-actin or U47 endogenous controls (Applied Biosystems) were used as an internal standard to normalize. PCR reactions were performed in triplicate. Data were analyzed using ABI PRISMs 7000 sequence detection system software (Applied Biosystems).
We purchased the following antibodies: IMP-1 (for IHC), c-Myc, β-catenin, cleaved Caspase-3 (Asp175)(5A1E) and Parp (Asp214), Lamin A/C, (Cell Signaling Technology), β-catenin (for IHC), Cdc34, Cyclin D1 (BD Transduction Laboratories), Ras clone 10 (Upstate), IMP-1, K-Ras (Santa Cruz), Caspase-8 (Enzo Life Sciences), Cyfip2 (Abcam), Lin28B (Abgent), and K-Ras (for IHC) (Spring Bioscience).
Immunoblotting was performed as described previously (33). The membranes were stripped using Blotfresh Western Stripping Reagent (SignaGen) and re-probed for anti-β-actin (Sigma-Aldrich) to confirm equal loading. Relative band intensities were quantified using Adobe Photoshop software, and normalized to the most intense band, for each antibody.
Twelve tissue microarrays (TMAs) were constructed with two cores of representative areas of each carcinoma and two cores of normal adjacent mucosa under IRB approval. A uniform cohort of 228 patients (133 males and 95 females) with colon cancer, 88 in stage 2, and 140 in stage 3, which were diagnosed between November 1993 and October 2006. Rectal tumors were excluded from the study. The pathology reports were available in all cases. H&E slides and paraffin blocks were retrieved from the Surgical Pathology files of the Hospital Clinic, Barcelona. H&E slides from the original surgery and subsequent tumors were examined. All patients were treated with surgery and 5-fluorouracil. Slides were imaged using a Nikon TE2000 Eclipse microscope with a QiCam (Q Imaging, Surrey, Canada) camera and IPLab imaging software (BD Bioscience Bioimaging, Rockville, MD). The H-Score, is the product of the intensity of the staining (0, 1, 2 or 3) by the percent of stained cells (0 to 100). The distributions of nuclear β-catenin, K-Ras, and IMP-1 intensity scores, percent stained cells, and H-scores were right-skewed and did not follow a normal distribution. Thus, Wilcoxon signed rank tests were used to compare normal vs. tumor tissue, paired within each patient. Spearman correlation coefficients, stratified by tumor type, were used to assess the strength of linear association between nuclear β-catenin, K-Ras, and IMP-1 measures. Spearman correlation coefficients > 0.50 indicate a strong association and any greater than 0.30 indicate a moderate association. All analyses were performed using SAS version 9.2 (SAS Institute, Cary, NC).
WST-1 cell proliferation and anchorage-independent growth were performed as described previously (31).
Gene expression profiling and bioinformatics analysis was performed as described previously (33), using total RNA isolated from IMP-1 and scramble control siRNA SW-480 cells.
SW480 cells were transiently transfected with either scramble control or IMP-1 siRNAs, seeded on Type I Collagen embedded with human fetal colonic fibroblasts (ATCC), grown in organotypic culture (OTC), harvested, and fixed as previously described (34).
The K-Ras coding cDNA and its 3′UTR subcloned into pGEM-T easy vector were linearized by Nco I and Sal I restriction enzymes respectively. GLI1 expression plasmid and the fragment encompassing nucleotides 2713-3600 of GLI1 cDNA (35) were used respectively as positive and negative controls for the binding assay of K-Ras coding region with IMP-1. The full-length GLI1 subcloned into pOTB7 (ATCC) was linearized immediately 3′ to the target DNA insert by Bgl II restriction enzyme. Fragments encompassing nucleotides 17-597 and 577-1182 of β-TrCP1 cDNA (24) were subcloned into pcDNA3.1, linearized by Xba I restriction enzyme, and used respectively as negative and positive controls for the binding assay of the 3′UTR region of K-Ras with IMP-1. In vitro transcription, transfection, whole cell extraction, and UV cross-linking were performed as described previously (35).
We first sought to determine the relative abundance level of endogenous c-Myc protein in eight human colon cancer cell lines by western blot analysis and selected Caco-2/15 cells for further c-Myc knock-down and biochemical studies based on it's lower abundance of c-Myc relative to the other cell lines surveyed (Fig.1A). Caco-2/15 cells were transiently transfected with two different c-Myc siRNAs either individually or together. We found that a combination of the two c-Myc siRNAs (1 and 2) more effectively reduced endogenous c-Myc protein expression by 82% compared to transfection with the individual c-Myc siRNA -1 (37%) and -2 (43%), or the scramble control cells (Fig.1B). Interestingly, we also observed a concomitant decrease in expression of the let-7 miRNA target IMP-1 by 73%, as well as Cdc34 and total Ras, by 30% and 16% respectively, compared to control cells (Fig1B), suggesting let-7 regulation by c-Myc. To test this, total RNA was isolated from Caco-2/15 cells transfected with either a scramble control or c-Myc siRNAs. By PCR analysis using let-7 intron-specific primers, we detected no change in primary transcript levels of the let-7a3-b bi-cistronic cluster in c-Myc knock-down cells relative to scramble controls (Fig.1C). However, quantitative real-time PCR (qRT-PCR) showed an increase in both mature let-7a and let-7b miRNA expression (Fig.1D), suggesting that c-Myc modulates expression of known let-7 targets, IMP-1, Cdc34, and Ras, in part by negatively regulating endogenous mature let-7 miRNA levels in human colon cancer cells.
Our novel finding that c-Myc siRNA inhibits IMP-1 expression suggests a potential positive feedback mechanism. To explore this possibility further, we first determined the relative mRNA and protein expression level of IMP-1 in several human colon cancer cell lines by qRT-PCR and immunoblot analysis, respectively. We discovered that IMP-1 mRNA was upregulated by 3-fold or greater in 8 of 14 human colon cancer cell lines surveyed, relative to normal colonic epithelium (NCE) (Fig.2A). IMP-1 mRNA levels were most notably increased in Caco-2 (511-fold) and SW-480 (38-fold) cells (Fig.2A). Consistent with this finding, IMP-1 protein was highly expressed in Caco-2, SW-480, and SW-620 cells, and abundant levels of β-catenin, a positive regulator of IMP-1 (Fig.2B). To examine the biochemical role of IMP-1 in c-Myc signaling, SW480 cells were transiently transfected with control, c-Myc, or IMP-1 siRNAs. By western blot, we found that loss of IMP-1 decreased endogenous levels of Cdc34 (68%), Lin-28B (53%), a let-7 repressor and mediator of c-Myc driven cellular proliferation, and K-Ras (60%), relative to scramble controls, but not to the same extent as c-Myc inhibition (36). (Fig.2C). Furthermore, we did not detect any significant changes in c-Myc or mature let-7a or −b miRNA expression levels (data not shown). To assess for effects of IMP-1 in β-catenin signaling, we transfected SW-480 cells with control scramble, c-Myc, IMP-1, and β-catenin siRNAs. Western blot analysis showed that loss of β-catenin reduced endogenous protein levels of oncogenic c-Myc (89%), IMP-1 (67%), and Cyclin-D1 (98%), relative to scramble control cells (Fig.2D). Interestingly, K-Ras was also reduced in these cells (90%) to a greater extent than by IMP-1 knock-down (42%), compared to controls (Fig.2D). Since IMP-1 has been shown to bind and stabilize several mRNAs including oncogenic c-Myc, we hypothesized that it might regulate K-Ras expression by binding to K-Ras mRNA (37). Indeed, protein-RNA binding assays revealed that IMP-1 interacts directly with both the coding region and 3′UTR of K-Ras mRNA, compared to negative controls (Fig.2E). Taken together, these findings suggests that IMP-1 functions downstream of β-catenin and may modulate K-Ras expression in human colon cancer cells via binding to K-Ras mRNA.
To determine the biological role of IMP-1 in human colon cancer cell lines, we transfected SW-480 cells with scramble control or IMP-1 siRNAs and examined the effects on regulating cell proliferation and anchorage-independent growth. We found that loss of IMP-1 expression reduced colon cancer cell proliferation in WST-1 assays by approximately 2-fold (p=0.0001) and significantly suppressed both colon cancer colony size formation and number in soft agar assays by 3-fold, compared to scramble controls (p=0.008) (Fig.3A and B). Intriguingly, transfection with IMP-1 siRNA also increased trypan blue uptake by 2.5-fold (p=0.013) compared to scramble control cells (Fig.3C). Loss of IMP-1 also induced cell rounding, membrane blebbing, and floating cells, suggesting the induction of cell death (Fig. 3C). To examine this further, adherent and floating SW-480 cells were collected and harvested 24 and 48 hours post-transfection with scramble or IMP-1 siRNAs. By immunoblot analysis we found that two markers of apoptosis, Caspase-3 and Lamin A/C cleavage products, were both increased in IMP-1 knock-down cells by 2-fold compared to control cells (Fig.3D). In addition, we also observed a 3.6-fold increase in cleaved Caspase-8 in IMP-1 siRNA transfected cells but no change in caspase-9 cleavage (data not shown) relative to scramble controls, suggesting that IMP-1 functions via induction of an extrinsic or mitochondrial-mediated cell death pathway (Fig.3E). Taken together, these data support the notion that IMP-1 is a positive modulator of color cancer cell proliferation, anchorage-independent growth, and survival.
To identify novel genes that are involved in mediating IMP-1 functions, expression profiling analysis was performed on total RNA isolated from scramble control and IMP-1 siRNA SW-480 cells. Gene expression analysis confirmed that IMP-1 was reduced by approximately 1.6-fold in knock-down cells compared to scramble controls (Supplementary Table I). We focused on the cytoplasmic FMR1-interacting protein 2 (CYFIP2), which was the most upregulated gene target (2.3-fold; (p=0.00001) by IMP-1 loss compared to control cells (Supplementary Table I). Importantly, CYFIP2 is a p53-inducible gene that is sufficient for inhibition of colon cancer proliferation, caspase activation, and the induction of apoptosis (38). Consistent with the gene microarray results, PCR analysis confirmed that CYFIP2 mRNA levels were increased by 4.5-fold in cells transfected with IMP-1 siRNA but was reduced or undetectable in scramble controls (Figs.4A and B). We next asked whether CYFIP2 inhibition by IMP-1 mediates cell survival and K-Ras expression. To test this, SW-480 cells were transiently transected with scramble control, IMP-1, CYFIP2, or a combination of IMP1 and CYFIP2 siRNAs. By western blot, we found that co-transfection of IMP-1 and CYFIP2 siRNAs reversed Caspase-3 and Parp cleavage and rescued K-Ras inhibition by IMP-1 loss, similar to levels observed in control cells (Fig.4C). Based upon these findings, we propose that IMP-1 promotes colon cancer cell survival and regulates K-Ras expression, in part by suppressing CYFIP2.
As a complementary approach to our knock-down studies and to further investigate the role of IMP-1, we chose to modify LIM2405 cells, which express low endogenous IMP-1 mRNA and protein levels (Figs. 2A and 2B), with a retrovirus expressing IMP-1 or the vector alone as a control. We found that IMP-1 upregulated endogenous levels of c-Myc and K-Ras by 3.8-fold and 2.6-fold respectively, relative to vector controls in LIM2405 cells (Fig.5A). IMP-1 also increased LIM2405 cell proliferation in WST-1 assays by 1.5-fold (p=0.00125) (Figs.5B and 5C). We next wanted to examine whether or not IMP-1 expression can rescue K-Ras inhibition by β-catenin knock-down. To test this, IMP-1 and vector alone LIM2405 cells were transfected with scramble or β-catenin siRNAs. Western blot analysis showed that K-Ras levels were 1.7-fold higher in IMP-1 cells transfected with β-catenin siRNA compared to vector alone cells (Fig.5D). These results support the notion that IMP-1 regulates K-Ras expression downstream of β-catenin signaling.
We next examined the effect of IMP-1 on colonic epithelium in organotypic culture (OTC). SW480 cells were transiently transfected with IMP-1 or scramble control siRNAs and seeded on Collagen Type I matrix embedded with colonic fetal fibroblasts. In contrast to scramble controls, IMP-1 siRNA induced the formation of apoptotic bodies indicative of cells undergoing death (Fig.6A). Consistent with this notion and our biochemical monolayer culture results, we also detected a 2.4 fold (p=0.02) increase in cleaved Caspase-3 positive cells relative to scramble control cells by IHC (Figs.6A and B). Together these findings suggest that IMP-1 plays a key role in modulating cell survival.
Based on our biochemical finding that IMP-1 expression is elevated in human colon cancer cell lines, we sought to determine whether or not there is a correlation between IMP-1, β-catenin, and K-Ras expression levels, in a cohort of 228 paired patient normal and adjacent colon cancer tissue specimens by IHC. In comparison to adjacent normal colonic mucosa, we observed a significant upregulation of nuclear β-catenin, K-Ras, and IMP-1 in paired colon cancers by 9.9-fold, 8.3-fold, and 33.8-fold (p ≤ 0.0001), respectively (Fig.7A). Importantly, IMP-1 predominantly localized in the cytoplasmic compartment of colon tumors and positively associated with K-Ras intensity (r=0.35, p ≤ 0.0001) and H-scores (r=0.32, p ≤ 0.0001) among the tumor tissues (Fig. 7B and Supplementary Fig.S1, compared to adjacent normal colonic mucosa (Fig.7B). Collectively, our findings suggest that IMP-1 expression is upregulated in human colon cancers and correlates with K-Ras levels.
We report that loss of IMP-1 inhibits expression of let-7 miRNA oncogenic protein targets, Cdc34 and K-Ras, and the let-7 repressor Lin-28B, and concomitantly suppresses colon cancer cell proliferation, anchorage-independent growth, and triggers caspase-mediated cell death in both monolayer and 3D culture. By gene profiling analysis and PCR analysis we have discovered a novel pro-apoptotic target, CYFIP2, which is upregulated by IMP-1 knock-down. Importantly, we have identified a new function of IMP-1 to bind to K-Ras mRNA and regulate cell survival and K-Ras expression via inhibition of CYFIP2. We further show that a gain of IMP-1 expression occurs in human colon cancer cells and correlates with K-Ras levels in patient colon tumors.
To date, little is known about the molecular mechanisms that are involved in the complex regulation of IMP-1 in mammalian cells (26). Constitutive activation of β-catenin/Tcf has been shown to upregulate IMP-1 levels in colon cancers and is necessary for the stabilization and induction of c-Myc (24, 29). In addition, a recent study demonstrated that overexpression of the anti-tumorigenic protein 15S-Lipoxygenase (15S-Lox-2) suppresses human prostate carcinoma cell growth and proliferation by downregulation of IMP-1 (39). Our novel finding that loss of c-Myc reduces IMP-1 expression and re-expression of IMP-1 increases c-Myc levels suggests a new potential feedback mechanism whereby c-Myc and IMP-1 may reciprocally modulate expression of each other in colon cancer cells. Oncogenic c-Myc induction has been shown to repress members of the let-7 family in human cancer cells and IMP-1 is a direct let-7 target (3, 4, 29). Consistent with this, inhibition of IMP-1 by targeted knock-down of c-Myc resulted in an increase in mature let-7 miRNA expression, suggesting that c-Myc regulation of IMP-1 is mediated in part by let-7 in colon cancer cells. Furthermore, loss of IMP-1 inhibited expression of let-7 oncogenic targets, Cdc34, K-Ras, and Lin-28B, but did not alter the endogenous levels of c-Myc, pri- or mature let-7 miRNAs (data not shown), suggesting that IMP-1 is regulated by additional genes involved in promoting the cancer cell phenotype.
We showed that loss of IMP-1 suppressed colon cancer cell proliferation and anchorage-independent growth, and increased trypan blue uptake, caspase and lamin-mediated cell apoptosis. By gene profiling microarray analysis, we also identified a new IMP-1 target, CYFIP2, a direct target of the tumor suppressor p53 that is sufficient for caspase activation and colon cancer apoptosis (Supplementary Table I)(38).
K-Ras mutations are relatively frequent and appear in colonic adenomatous polyps and colon cancers of mice and humans (40). Importantly, we observed that knock-down of IMP-1 or it's upstream regulator, β-catenin/Tcf, decreased levels of K-Ras protein levels in colon cancer cells. Conversely, IMP-1 expression rescues inhibition of K-Ras by β-catenin siRNA. In addition, we found that IMP-1 binds directly to the coding region and 3′UTR of K-Ras mRNA. Furthermore, we also showed that CYFIP2 knock-down reverts Caspase-3 and Parp-mediated apoptosis, and inhibition of endogenous K-Ras expression due to loss of IMP-1. These findings indicate that K-Ras may be a novel downstream target of IMP-1 signaling and is regulated via direct interaction with K-Ras mRNA and the suppression of CYFIP2. Interestingly, IMP-1 was upregulated in patient colon tumors and correlated with K-Ras expression, suggesting that gain of IMP-1 maybe a novel mechanism to increase endogenous K-Ras levels.
Overexpression of IMP-1 has been reported in a variety of cancers, suggesting an important role of IMP-1 in cancer development (41-44). Constitutively active β-catenin/Tcf transcription factor upregulates IMP-1 in colon cancer cells, and elevated levels of IMP-1 correlate with β-catenin/Tcf signaling in primary colorectal tumors (24, 28). We found that nuclear β-catenin levels were significantly increased in colon cancers compared to normal colonic mucosa, however, there was no correlation between IMP-1 and β-catenin expression in tumors, suggesting that IMP-1 dysregulation may occur independently of β-catenin in vivo.
In summary, we report that IMP-1 binds to K-Ras mRNA and plays an important role in regulating K-Ras expression potentially via repression of the pro-apoptotic protein, CYFIP2, downstream of β-catenin/Tcf, and promotes colon cancer cell proliferation, anchorage-independent growth, and survival (Supplementary Fig.S2). Furthermore, we show that human colon tumors express high levels of IMP-1 relative to normal colonic epithelium and present new evidence that IMP-1 positively correlates with K-Ras in colon cancers. Based on our novel findings, we propose that IMP-1 maybe a candidate for targeted therapeutic intervention in human cancers with dysregulation of K-Ras expression and signaling.
We thank D. Baldwin for the gene expression profiling and J. Tobias for the microarray statistical analysis (Penn Microarray Facility Bioinformatics Group), J. Drebin (Department of Surgery) for the β-catenin siRNA, X. Yang (Department of Cancer Biology) for the Caspase-8 antibody, G.P. Swain, D. Budo, and S. Campbell (Penn Morphology Facility) for IHC, and C. Brensinger for the TMA statistical analysis (Center for Clinical Epidemiology & Biostatistics). This work was supported in part by NIH DK056645, the National Colon Cancer Research Alliance, the Hansen Foundation (to AKR), and NCI grant CA121851 (to VSS).