Cancer chemoprevention uses natural, synthetic, or biological substances to reverse, suppress, or prevent either the initial phase of carcinogenesis or the progression of neoplastic cells to cancer1. It holds promise for overcoming problems associated with the treatment of late-stage cancers. However, the broad application of chemoprevention is compromised currently by limited effectiveness and potential toxicity. In an effort to overcome these challenges, we developed a new chemoprevention approach that specifically targets premalignant tumor cells for apoptosis. This procedure is based on our finding that a deficiency in the adenomatous polyposis coli (APC) gene and subsequent activation of β-catenin lead to the repression of cellular caspase-8 inhibitor cFLIP expression through activation of c-Myc and that 9-cis-retinyl acetate (RAc) independently upregulates tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) death receptors and suppresses decoy receptors. Thus, the combination of TRAIL and RAc induces apoptosis in APC-deficient premalignant cells without affecting normal cells in vitro. In addition, we show that short-term and non-continuous TRAIL and RAc treatment induces apoptosis specifically in intestinal polyps, strongly inhibits tumor growth, and prolongs survival in multiple intestinal neoplasms C57BL/6J-ApcMin/J (ApcMin) mice. With our approach, we further demonstrate that TRAIL and RAc induce significant cell death in human colon polyps, providing a potentially selective approach for colorectal cancer chemoprevention by targeting APC-deficient cells for apoptosis.
TRAIL, or Apo2L, is a membrane-bound TNF family ligand2,3. While TRAIL induces apoptosis in cancer cells, it does not harm normal cells4,5. To test the response of premalignant tumor cells to TRAIL, we isolated primary adenoma cells from intestinal polyps (benign adenomas) of ApcMin mice6-8. We also isolated intestinal epithelial cells from wild-type mice as controls (Supplementary Fig. 1a). TRAIL treatment did not induce any apoptosis in the isolated epithelial cells (Fig. 1a and Supplementary Fig. 1b). Previous reports indicate that certain retinoids synergize TRAIL-induced apoptosis in some cancer cell lines by inducing either TRAIL or its death receptors 9-11. We explored the effect of RAc on TRAIL sensitivity in isolated primary adenoma and normal epithelial cells. Although RAc alone had no effect on the survival of either normal cells from control mice or adenoma cells, it specifically sensitized the adenoma cells to TRAIL-induced cell death (Fig. 1a and Supplementary Fig. 1b). These results show that RAc and TRAIL synergistically induce apoptosis in intestinal adenoma cells derived from ApcMin mice.
The major difference between adenoma cells and normal cells is the loss of the wild-type APC allele in adenoma cells12. To test the role of APC in cell sensitivity to TRAIL and RAc, we knocked down APC expression in an immortalized normal human colon epithelial cell line, NCM356, using APC small interfering RNA (APC-siRNA) (Fig. 1b). We observed that knockdown of APC and treatment with TRAIL and RAc resulted in caspase 8 cleavage and an increase in caspase 3/7 activity (Fig. 1c and Supplementary Fig. 1c). Furthermore, caspase 8 inhibitor Z-IETD blocked caspase 3/7 activity (Supplementary Fig. 1c), suggesting that caspase 8 processing resulted in its activation. Similar results were obtained using an immortalized normal human bronchial epithelial cell line (BEAS-2B-BW1799 or BW1799) (Supplementary Fig. 2). These results demonstrate that APC deficiency is the key factor in sensitizing normal cells to apoptosis induced by the combination of TRAIL and RAc.
Inactivation of APC leads to stabilization and nuclear translocation of β-catenin and transcriptional activation of target genes, such as c-Myc and cyclin D113. To investigate the signaling events in the APC pathway required for sensitizing cells to TRAIL and RAc, we transfected either stabilized β-catenin14 or full-length c-Myc into NCM356 cells (Supplementary Fig. 3a and b). Expression of either β-catenin mutant or c-Myc led to TRAIL- and RAc-mediated caspase 8 cleavage and activation (Fig. 2a and Supplementary Fig. 3c). Transfection of cyclin D1 had no effect on TRAIL- and RAc-induced apoptosis in NCM356 and other cells (data not shown). We then determined the requirement of c-Myc for sensitizing cells to TRAIL and RAc. Knockdown of APC led to an increase in c-Myc expression, as expected. Co-transfection of c-Myc-siRNA blocked APC deficiency-mediated c-Myc induction and caspase 8 cleavage/activation in response to TRAIL and RAc (Fig. 2b and Supplementary Fig. 3c). We further confirmed these observations using independent siRNAs (Supplementary Fig. 4). These results indicate that the effect of APC on cell response to TRAIL and RAc is largely dependent on APC/β-catenin-mediated activation of c-Myc.
Other studies have shown that c-Myc directly suppresses expression of cFLIP and that this activity correlates with TRAIL sensitivity15. NCM356 cells expressed readily detectable cFLIP and transfection of c-Myc resulted in a significant decrease in cFLIP expression and caspase 8 cleavage (Fig. 2c). While direct suppression of cFLIP expression by cFLIP-siRNA led to caspase 8 cleavage and activation in NCM356 cells treated with TRAIL and RAc, rescue of cFLIP expression by transfection of a cFLIP expression plasmid prevented c-Myc-induced caspase 8 processing and activation in response to TRAIL and RAc (Fig. 2c and Supplementary Fig. 3c). Similar results were obtained in BW1799 cells (Supplementary Fig. 5). In addition to suppressing cFLIP expression, c-Myc could also act on mitochondria to facilitate TRAIL-induced apoptosis16. To test this possibility, we performed rescue experiments using Bcl-2 and Bcl-XL. Overexpression of either Bcl-2 or Bcl-XL did not inhibit c-Myc-mediated caspase 8 processing and activation in response to TRAIL and RAc (Fig. 2c and Supplementary Fig. 3c). These results indicate that APC deficiency sensitizes normal cells to TRAIL and RAc by suppressing cFLIP expression through the activation of β-catenin and c-Myc.
Previous reports have shown that certain retinoids sensitize some cancer cells to TRAIL-induced apoptosis by upregulating either TRAIL or TRAIL receptors DR4 and DR5 9,10. To test the effect of RAc on normal cells in our setting, we examined the expression of TRAIL receptors in NCM356, BW1779, and another immortalized normal human colon epithelial cell line CRL1831 cells after RAc treatment. Although RAc did not induce TRAIL expression (data not shown), it induced DR4 or DR5 expression in these cells (Fig. 2d and Supplementary Fig. 6a). Interestingly, RAc strongly suppressed DcR1 expression in these cells (Fig. 2d and Supplementary Fig. 6a). In addition, the DcR2 expression was inhibited by RAc in NCM356 cells (Fig. 2d). We further showed that RAc caused similar changes in mRNA levels in NCM356 cells (Supplementary Fig. 6b), suggesting that RAc modulates TRAIL receptor expression at the levels of mRNA.
Overexpression of DR5 sensitized APC-knockdown cells to TRAIL-induced caspase 8 cleavage and activation (Fig. 2e and Supplementary Fig. 6c). Pretreatment of APC-knockdown cells with a DcR1-neutralizing antibody to block TRAIL-DcR1 interaction also partially induced caspase 8 cleavage and activation after TRAIL treatment, and this effect was further enhanced in DR5-transfected cells (Fig. 2e and Supplementary Fig. 6c). Similar results were obtained in BW1799 cells (Supplementary Fig. 6d and e). These data suggest that both induction of DR5 and suppression of DcR1 contribute to the effect of RAc in sensitizing APC-deficient cells to apoptosis.
We next tested the effect of TRAIL and RAc in ApcMin mice in vivo. We first treated the 3-month-old ApcMin mice with TRAIL and low-dose RAc for 15 treatment cycles over six weeks. We observed that TRAIL and RAc significantly inhibited the growth of intestinal polyps as compared to controls (Fig. 3a). Histological analysis of tumors and adjacent normal tissues in treated mice showed that many cells in tumors, but not in normal tissues, contain condensed nuclei, suggesting that cells in the tumors are dying (Fig. 3b). This observation was confirmed by terminal deoxynucleotidyl transferase biotin-dUTP nick end labeling (TUNEL) analysis (Fig. 3b). When the total numbers and sizes of the polyps were compared 1 month after the six week treatment period, the treated mice displayed an average 4- to 5-fold decrease in the total number of polyps relative to the mice treated with PBS control (Fig. 3c). The reduction in large polyps (≥2 mm) was much more pronounced (to about 1/30 of control) (Fig. 3c). These data demonstrate that the combination of TRAIL and RAc induces apoptosis specifically in Apc-defective polyps and effectively suppresses the tumor growth.
Biochemically, accumulation of soluble β-catenin and induction of c-Myc expression were seen in the polyps of ApcMin mice but not in normal tissue (Supplementary Fig. 7a). cFLIP protein was readily detectable in normal tissue, while it was significantly reduced in polyps (Supplementary Fig. 7b). Consistent with our in vitro observations, RAc induced DR5 expression and repressed DcR1 in both the polyps in ApcMin mice.
Since TRAIL and RAc together target APC-deficient cells for apoptosis, this combination has the potential to achieve chemoprevention in short-term therapy. To test this, we applied two short consecutive TRAIL and RAc treatment cycles to ApcMin within a week and analyzed the intestinal tumors 2 weeks after treatment. After only two treatments, we observed a 69% reduction in the number of polyps when 3 mg/kg of TRAIL and 68 μg/kg of RAc were injected into ApcMin mice (Fig. 3d). The tumor numbers were further reduced to 10% of the control when 10-fold higher doses of TRAIL and RAc were used (Fig. 3d). These data indicate that short-term treatment with TRAIL and RAc is capable for inhibiting growth of premalignant cells in ApcMin mice. To test the effect of TRAIL and RAc treatment on the survival of ApcMin mice, we applied 5 non-consecutive TRAIL and RAc treatments to ApcMin mice within 4 months (average 1 treatment every 3 weeks). All mice in the control group treated with PBS died within 7 months; however, five of the seven treated mice survived beyond 8 months (P≤0.001) (Fig. 3e) demonstrating that non-continuous TRAIL and RAc treatment yields a long-term survival benefit for ApcMin mice. While it is likely that the two early deaths in the treatment group resulted from a late start in treatment, it cannot be ruled out that intrinsic resistance, heterogeneity, genetic or epi-genetic changes of premalignant cells can influence the effect of TRAIL treatment. Further study will provide a more in-depth assessment.
As a first step to demonstrate the efficacy and relevance of TRAIL and RAc treatment in humans, we treated biopsy samples of both normal and tumor regions from consenting familial adenomatous polyposis (FAP) patients under organ culture conditions as reported 17,18. A cross section of normal colon tissue showed a round and defined structure of individual villus consisting of a monolayer of colon epithelial cells and membrane β-catenin staining pattern, while dysplastic colon polyps showed much larger villi, bigger nuclei, an irregular structure consisting of multiple epithelial layers, and enhanced nuclear β-catenin staining (Fig. 4a). Treatment of normal tissue slices with TRAIL and RAc did not induce significant cell death, while the same treatment in the polyp samples resulted in significant cell death, as detected by TUNEL assay (Fig. 4a). Quantification of TUNEL-positive cells showed that TRAIL and RAc killed an average of 57% of the cells in the human colon polyps (Fig. 4b). These results verify that TRAIL and RAc are effective against human polyps from FAP patients in vitro and that our approach holds promise for the chemoprevention of colon cancer in humans.
A major concern in chemoprevention is potential toxicity associated with treatment. We have shown that TRAIL and RAc target the APC-β-catenin-c-Myc signaling pathway for apoptosis. Since activation of β-catenin is involved in stem cell self-renewal and maintenance in adult tissues19, TRAIL and RAc could possibly interact in a detrimental way with stem cells. In order to investigate this, we first tested the effect of TRAIL and RAc on intestinal stem cells (ISCs) in ApcMin mice. Using a putative ISC marker leucine-rich-repeat-containing G-protein-coupled receptor (Lgr5)20, we noted that most of the Lgr5 positive cells in the normal sections were located at the bottom of the crypt as reported20 (Fig. 4c). Staining of active caspase 3 revealed extensive staining in tumor sections similar to the findings shown in Figure 3b (Fig. 4c). More importantly, no staining was detected in serial sections from the normal region, including Lgr5+ cells in the crypt regions where the ISC resides (Fig. 4c). We also studied the effect of long-term TRAIL and RAc treatment on tissue-resident stem cells isolated from pararenal adipose tissue (mASCs) of ApcMin mice21 (Supplementary Fig. 8). We found no detectable effect on morphology (data not shown), neural differentiation (data not shown), proliferation and apoptosis, senescence, osteogenesis, and adipogenesis both in vitro and in vivo (Supplementary Fig 9 and 10). Taken together, these results suggest that TRAIL and RAc exert no negative effect on stem cells (ISCs and mASCs) in mice.
In summary, we discovered a synergistic interaction among TRAIL, RAc, and tumor suppressor APC, which results in the specific induction of apoptosis in APC-deficient cells by TRAIL and RAc (Supplementary Fig. 11). Induction of apoptosis represents a most potent cellular mechanism against cancer, and selectively eliminating premalignant tumor cells by TRAIL and RAc is an effective method of chemoprevention. More importantly, we demonstrate that the chemopreventive effect of TRAIL and RAc can be achieved by short-term intermittent and non-continuous treatment cycles. Thereby, the potential side effects and costs often associated with long-term treatment could be minimized and controlled.