C.M.M. and J.-H.K. contributed equally to this article
Cancer Cell Biology
Nonsteroidal anti-inflammatory drugs suppress cancer stem cells via inhibiting PTGS2 (cyclooxygenase 2) and NOTCH/HES1 and activating PPARG in colorectal cancer
Article first published online: 7 AUG 2013
Copyright © 2013 UICC
International Journal of Cancer
Volume 134, Issue 3, pages 519–529, 1 February 2014
How to Cite
Moon, C. M., Kwon, J.-H., Kim, J. S., Oh, S.-H., Jin Lee, K., Park, J. J., Pil Hong, S., Cheon, J. H., Kim, T. I. and Kim, W. H. (2014), Nonsteroidal anti-inflammatory drugs suppress cancer stem cells via inhibiting PTGS2 (cyclooxygenase 2) and NOTCH/HES1 and activating PPARG in colorectal cancer. Int. J. Cancer, 134: 519–529. doi: 10.1002/ijc.28381
- Issue published online: 14 NOV 2013
- Article first published online: 7 AUG 2013
- Accepted manuscript online: 13 JUL 2013 11:58AM EST
- Manuscript Accepted: 24 JUN 2013
- Manuscript Received: 10 DEC 2012
- Ministry of Education, Science and Technology (Basic Science Research Program through the National Research Foundation of Korea (NRF)). Grant Number: 2009–0072370
- Yonsei University College of Medicine for 2010. Grant Number: 6–2010-0036
- colonic cancer stem cell;
- nonsteroidal anti-inflammatory drugs;
- cyclooxygenase 2;
Cancer stem cells (CSCs) play a pivotal role in cancer relapse or metastasis. We investigated the CSC-suppressing effect of nonsteroidal anti-inflammatory drugs (NSAIDs) and the relevant mechanisms in colorectal cancer. We measured the effect of NSAIDs on CSC populations in Caco-2 or SW620 cells using colosphere formation and flow cytometric analysis of PROM1 (CD133)+CD44+ cells after indomethacin treatment with/without prostaglandin E2 (PGE2) or peroxisome proliferator-activated receptor γ (PPARG) antagonist, and examined the effect of indomethacin on transcriptional activity and protein expression of NOTCH/HES1 and PPARG. These effects of indomethacin were also evaluated in a xenograft mouse model. NSAIDs (indomethacin, sulindac and aspirin), celecoxib, γ-secretase inhibitor and PPARG agonist significantly decreased the number of colospheres formation compared to controls. In Caco-2 and SW620 cells, compared to controls, PROM1 (CD133)+CD44+ cells were significantly decreased by indomethacin treatment, and increased by 5-fluorouracil (5-FU) treatment. This 5-FU-induced increase of PROM1 (CD133)+CD44+ cells was significantly attenuated by combination with indomethacin. This CSC-inhibitory effect of indomethacin was reversed by addition of PGE2 and PPARG antagonist. Indomethacin significantly decreased CBFRE and increased PPRE transcriptional activity and their relative protein expressions. In xenograft mouse experiments using 5-FU-resistant SW620 cells, the 5-FU treatment combined with indomethacin significantly reduced tumor growth, compared to 5-FU alone. In addition, treatment of indomethacin alone or combination of 5-FU and indomethacin decreased the expressions of PROM1 (CD133), CD44, PTGS2 (cyclooxygenase 2) and HES1, and increased PPARG expression. NSAIDs could selectively reduce the colon CSCs and suppress 5-FU-induced increase of CSCs via inhibiting PTGS2 (cyclooxygenase 2) and NOTCH/HES1, and activating PPARG.
bisphenol A diaglycidyl ether
fibroblast growth factor
RBPJ (CBF1)-responsive element
cancer stem cell
IAP-binding mitochondrial protein
Dulbecco's modified Eagle's medium
epidermal growth factor
NOTCH 1 intracellular domain
nonsteroidal anti-inflammatory drugs
phosphate buffered saline
peroxisome proliferator-activated receptor γ
prostaglandin-endoperoxide synthase 1
prostaglandin-endoperoxide synthase 2
recombination signal binding protein for immunoglobulin kappa J region
signal transducer and activator of transcription
Colorectal cancer (CRC) remains the fourth leading cause of cancer-related deaths in the world despite improvements of chemotherapeutic agents and the emergence of novel anti-cancer therapies. Conventional therapies, including chemotherapy and radiotherapy, target rapidly dividing tumor cells and suppress tumor growth. However, these agents are limited by therapy-resistant cancer cells and thus fail to eradicate the disease. Recent evidence has suggested that a small subset of cells isolated on the basis of phenotypic and molecular characteristics, referred to as cancer stem cells (CSCs), have the capacity to self-renew and differentiate, and may play a pivotal role in tumor initiation, recurrence and resistance to chemotherapy.
The existence of CSCs in CRC has been demonstrated in experimental and clinical studies, showing that small sets of CRC cells, such as PROM1 (CD133)+, EpCAM high/CD44+ and PROM1 (CD133)+CD44+ cells, have tumor-initiating properties in vivo.[5, 6] These studies showed that a small number of cancer cells that express CSC markers can initiate and propagate CRC, unlike the majority of other cancer cells. Moreover, in a study evaluating the relationship between CSC and chemo-resistance in CRC, CSCs were enriched in residual tumors following classical chemotherapy and retained the ability to generate tumors. Similarly, in CRC cell lines, treatment with 5-fluorouracil (5-FU) or oxaliplatin increased the proportion of PROM1 (CD133)+CD44+ cells in vitro. Based on these results, CSCs are now recognized as a specific target for the complete elimination of CRC. Other researchers have attempted to identify selective agents for the inhibition of CSCs and the underlying signaling pathways.[10, 11] Nonsteroidal anti-inflammatory drugs (NSAIDs) are well-known chemopreventive drugs in CRC, and recent studies have shown that NSAIDs may reduce the recurrence and mortality of CRC.[12-14] Therefore, we hypothesized that the beneficial effects of NSAIDs on tumor recurrence and mortality could be related to suppression of CSCs, and we investigated the CSC-suppressing effects of NSAIDs, especially on chemotherapy-induced CSCs. Lastly, we sought to understand the underlying relevant mechanisms involving PTGS2 (cyclooxygenase 2, COX2)-dependent and -independent pathways.
Material and Methods
In vitro cell line study
Cells and cell culture
The human CRC cell line (SW620 and Caco-2) was purchased from the Korean Cell Line Bank (Seoul, Republic of Korea). Cells were cultured in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (Hyclone, Logan, UT), 100 units/ml penicillin, 100 µg/ml streptomycin (Invitrogen, Carlsbad, CA) and 2 mM l-glutamine (Life Technologies, Carlsbad, CA). All cells were maintained in 5% CO2 incubator at 37°C. We used a 5-FU-resistant cell line in all xenograft mouse experiments. This cell line was developed by escalating doses of 5-FU serially in SW620 cells based on modifications of protocols from prior studies. Briefly, cells were exposed to an initial dose of 10 µM 5-FU for 72 hr and cultured in a drug-free condition during the defined period. After cells recovered from the prior dose of 5-FU, the 5-FU concentration was increased serially until it reached 1.0 mM.
Drugs and antibodies
Indomethacin, sulindac, aspirin, 5-FU, celecoxib, prostaglandin E2 (PGE2) and N-[N-(3,5-difluorophenacetyl)-l-alanyl]-(S)-phenylglycine-t-butyl ester (DAPT) were purchased from Sigma (St. Louis, MO). 7-Amino-4-chloro-3-methoxyisocoumarin (JLK6), rosiglitazone and bisphenol A diaglycidyl ether (BADGE) were purchased from Tocris Bioscience (Minneapolis, MN), Cayman Chemical (Ann Arbor, MI) and Tocris Bioscience (Bristol, UK), respectively. Antibodies used for flow cytometry, Western blotting and immunohistochemical (IHC) analysis were as follows: anti-PROM1 (CD133) (Cell Signaling Technology, Danvers, MA), anti-CD44, anti-HES1, and anti-peroxisome proliferator-activated receptor γ (PPARG) (Santacruz, Delaware, CA), anti-COX-2 (Invitrogen), phycoerythrin (PE)-conjugated anti-PROM1 (CD133), PE-conjugated mouse-IgG1 (Miltenyi Biotec, Bergisch Gladbach, Germany) and fluorescein (FITC)-conjugated anti-CD44 antibody (BD Biosciences, Franklin Lakes, NJ).
Colosphere culture assay
The sphere-forming abilities of CSCs were evaluated by colosphere culture assays. To evaluate CSC-inhibitory effects, SW620 cells were plated with 2,000 cells/well in 24-well ultra-low adhesive plates (Corning Incorporated, NY) in a 1 ml serum-free medium in the presence of each agent, including NSAIDs (indomethacin 6.25, 12.5, 25.0 µM, sulindac 25, 50, 100 µM, aspirin 0.25, 0.5, 1.0 mM and celecoxib 5, 10, 20, 40 µM), γ-secretase inhibitor (DAPT 2.5, 5.0, 10.0 µM, JLK6 2.5, 5.0, 10.0 µM) and PPARG agonist (rosiglitazone 2.5, 5.0, 10.0 µM). The serum-free medium used was DMEM-F12 supplemented with B27 (Life Technologies), 20 ng/ml epidermal growth factor (EGF) (R&D Systems, Minneapolis, MN), 10 ng/ml fibroblast growth factor (bFGF) (R&D Systems), penicillin–streptomycin and 2 mM l-glutamine (Life Technologies). Cells were cultured in a 5% CO2 incubator at 37°C and 500 µl of a medium with each agent was changed every 72 hr. To measure the inhibitory effect on colosphere formation, the number of colospheres was counted under a microscope (Olympus Bx51 microscope) at day 14.
Flow cytometric analysis
To investigate CSC-suppressing effects of indomethacin, Caco-2 and SW620 cells were treated with control, indomethacin (Caco-2, 12.5 µM; SW620, 100 µM), 5-FU (2 µM) and a combination of 5-FU and indomethacin. After 72 hr, flow cytometric analysis of CSC markers (PROM1 [CD133] and CD44) was performed. To evaluate the effects of PGE2 or BADGE on indomethacin-induced inhibition of CSCs, changes in PROM1 (CD133)+CD44+ cells were measured in Caco-2 and SW620 cells after 72- or 96-hr treatment with control, 5-FU (2 µM) alone and combination of 5-FU and indomethacin (Caco-2, 12.5 µM; SW620, 100 µM) with/without PGE2 (Caco-2, 10, 20 µM; SW620, 5, 10 µM) or BADGE (6.25 µM). The prepared cells were detached by accutase (Millipore, Billerica, MA), washed with phosphate-buffered saline (PBS) and then resuspended in FACS buffer (1× PBS, 1% bovine serum albumin [BSA], 2 mM ethylene diamine tetraacetic acid). Primary antibodies (PE-conjugated anti-PROM1 [CD133] [Miltenyi Biotec], FITC-conjugated anti-CD44 [BD Biosciences]) were added and incubated for 10 min in dark (4°C or ice) according to the manufacturer's instructions. As controls for flow cytometric analysis, we used FITC-conjugated mouse IgG1, κ isotype control, (clone P3 [cat. no. 11–4714-81]; eBioscience, San Diego, CA) and PE-conjugated mouse IgG2b, κ isotype control (clone eBMG2b [cat.no. 12–4732-81]; eBioscience). Samples were then washed and analyzed using a BD LSRII (BD Biosciences) coupled to a computer with BD FACS Diva software for data analysis.
Luciferase reporter assay
293T and SW620 cells were transfected with reporter plasmid containing the firefly luciferase gene driven by a RBPJ (CBF1)-responsive element (CBFRE-Luc), or PPAR-responsive element 3 (PPRE3-Luc), respectively. The constructs of CBFRE-Luc and PPRE3-Luc that we used have been described elsewhere.[15, 16] CBFRE-Luc was generously donated by S. Hayward (The Johns Hopkins University School of Medicine, Baltimore, MD), and PPRE3-Luc was generously donated by JW Kim (Yonsei University College of Medicine, Seoul, Republic of Korea). Renilla vector pRL-TK was used as a transfection control. After 6-hr CBFRE-Luc transfection, the cells were in a fresh medium for 24 hr, then pretreated with control vehicle, indomethacin 25, 50, 75, 100 µM or DAPT 30 µM for 1 hr. For 24 hr, the cells were treated with 20 mM of LiCl (Sigma), an inhibitor of GSK3, to upregulate NOTCH signaling. As for the transfection of PPRE-Luc, the cells were in a fresh medium for 24 hr after 6-hr transfection and treated with control, indomethacin 6.25, 12.5, 25.0 or rosiglitazone 10 µM for 24 hr. After drug treatment, the cells were lysed with lysis buffer (Promega, Madison, WI) and incubated at room temperature for 15 min. Relative luciferase activities were measured using a Dual-Luciferase Reporter Assay System (Promega) and a Berthold luminometer according to the manufacturer's instructions.
Western blot analysis
For the analysis of HES1 and PPARG expression in SW620 cells after 96-hr treatment with control or indomethacin (25 or 50 µM), cells were washed twice with PBS and lysed in protein extraction solution (iNtRON Biotechnology, Gyeonggi, Republic of Korea). Protein quantification was performed using a protein bicinchoninic acid assay (Pierce, Rockford, IL). We fractionated protein extracts (10 µg) by 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred the proteins to polyvinylidene fluoride membranes (Bio-Rad, Hercules, CA). Membranes were incubated with primary antibodies overnight at 4°C, followed by incubation with secondary antibodies. The membrane was developed using an ECL Western blotting detection kit (Amersham Biosciences, Freiburg, Germany).
In vivo xenograft mouse experiments
Xenograft tumor models
Six-week-old male Balb/c athymic nude mice were purchased from the Japan SLC and acclimated for 1 week. All mouse experiments were approved by the Committee of Care and Use of Laboratory Animals of Yonsei University College of Medicine and performed in accordance with institutional guidelines and policies.
Equal numbers (5 × 105) of 5-FU-resistant SW620 cells were suspended in 150 µl matrigel diluted 1:1 in DMEM and injected subcutaneously into the left rear flank of each mouse. When tumors reached a palpable size, a total of 20 mice were allocated randomly to four treatment groups (control, indomethacin alone, 5-FU alone, 5-FU and indomethacin combination treatment).
5-FU (30 mg/kg body weight) was administered intraperitoneally at alternative days (three times per week). Indomethacin (1.0 mg/kg body weight) was dissolved in a 0.5% carboxy methylcellulose (CMC, Sigma) solution and given by oral gavage every day for 16 days, and the same volume of 0.5% CMC solution without indomethacin was given to control and 5-FU alone-treated mice group on the same day. Tumor masses were measured every other day using calipers, and tumor volume was calculated based on the following formula: volume = (length × width)/2. All mice were euthanized at 16 days after first drug treatment and the tumor masses were dissected. The excised tumors were measured using calipers and placed in 10% buffered formalin for IHC.
All IHC studies were performed on formalin-fixed, paraffin-embedded tissue sections using anti-PROM1 (CD133), anti-CD44, anti-PTGS2 (COX2), anti-HES1 and anti-PPARG antibody. Briefly, 5 µm-thick sections were deparaffinized in xylene and hydrated in gradually decreasing concentrations of alcohol. The antigen retrieval was performed in 10 mM sodium citrate buffer (pH 6.0), using a pressure cooker for 10 min. After incubation with 3% hydrogen peroxide to block endogenous peroxidase activity, a blocking reagent (1.5% normal serum [Vector Laboratories, Burlingame, CA] in 1× TBS [50 mM Tris-Cl, 50 mM NaCl, pH 7.5]) was added to the sections for 10 min. Slides were then consecutively incubated with 1:100 dilution of primary antibody (overnight at 4°C) and secondary antibody (30 min at room temperature). Slides were developed with a Vectastain ABC kit (Vector Laboratories) and counterstained with hematoxylin. The staining was independently interpreted by two researchers (CM Moon and JH Kwon). For blindness, another researcher (JS Kim) assigned new identification codes to each slides before IHC staining and randomly selected and assigned slides to two researchers. All IHC staining was evaluated by light microscopy and the immunoactivity was scored according to the proportion of immunostained tumor cells. We counted the PROM1 (CD133), CD44, PTGS2 (COX2), HES1 and PPARG-stained cells among 100 tumor cells in the five different fields (total counted cell number = 500) under 400× microscope, then totaled the cell counts and converted the counts into percentages.
All analyses were performed using SPSS 18.0 (PASW Statistic, SPSS, IBM, Chicago, IL). In this study, the Mann–Whitney U test for nonparametric data and two-tailed Student's t test for parametric data were used to determine statistical significance. p-Values less than 0.05 were considered significant.
NSAIDs inhibited colosphere-forming capacity
First, we investigated the efficacy of NSAIDs in inhibiting colosphere-forming capacity in SW620 cells. Indomethacin significantly decreased the number of colospheres compared to control in a dose-dependent manner (relative numbers of spheres; control, 100% vs. 6.25, 12.5 and 25.0 µM indomethacin, 60.8, 44.0 and 17.2%; p < 0.05). Sulindac (control, 100% vs. 50 and 100 µM sulindac, 56.0 and 28.5%; p < 0.05) and aspirin (control, 100% vs. 0.25, 0.5 and 1.0 mM aspirin, 77.3, 54.1 and 23.2%; p < 0.05) also showed the same phenomenon (Fig. 1), suggesting the CSC-suppressing effect of NSAIDs.
Indomethacin decreased PROM1 (CD133)+CD44+ cells and suppressed the 5-FU-induced increase of PROM1 (CD133)+CD44+ cells
To evaluate the inhibitory effects of indomethacin on CSCs and 5-FU-induced CSC population, a flow cytometric analysis of PROM1 (CD133) and CD44 in Caco-2 and SW620 cells was performed after treatment with indomethacin with/without 5-FU for 72 hr (Fig. 2). The treatment of indomethacin (Caco2, 12.5 µM; SW620, 100 µM) significantly decreased PROM1 (CD133)+CD44+ cells (control vs. indomethacin; Caco-2, 7.0 vs. 4.8%; p < 0.05 [Fig. 2c] and SW620, 14.0 vs. 10.6%; p < 0.05 [Fig. 2d]), treatment with 2.0 µM of 5-FU led to significant increases in PROM1 (CD133)+CD44+ cells (control vs. 5-FU; Caco-2, 7.0 vs. 13.2%; p < 0.05 [Fig. 2c] and SW620, 14.0 vs. 25.6%; p < 0.05 [Fig. 2d]), and 5-FU and indomethacin combination treatment for the same period of time significantly reduced the proportion of PROM1 (CD133)+CD44+ cells compared to treatment with 5-FU alone (5-FU vs. 5-FU + indomethacin; Caco-2, 13.2 vs. 7.9%; p < 0.05 [Fig. 2c] and SW620, 25.6 vs. 17.7%; p < 0.05 [Fig. 2d]). These results suggest that indomethacin can selectively suppress CSCs and the 5-FU-induced increase of CSCs in CRC cells.
NSAIDs showed CSC-inhibitory effect through the PTGS2 (COX2)-dependent pathway
We investigated whether the PTGS2 (COX2)-dependent pathway, the most well-known action mechanism of NSAIDs, is a relevant mechanism in CSC suppression by NSAIDs. In the colosphere culture assay, PTGS2 (COX2) selective inhibitor (celecoxib) significantly decreased the number of colospheres (relative number of spheres; control, 100% vs. 10, 20 and 40 µM celecoxib, 65.2, 19.5 and 4.2%; p < 0.05) (Fig. 3a), and the CSC-inhibitory effect of indomethacin was prevented by the addition of PGE2 (Figs. 3b and 3c). Compared to Caco-2 cells treated with 2 µM 5-FU alone, the proportion of PROM1 (CD133)+CD44+ cells significantly decreased after a 72-hr combined treatment with 5-FU and indomethacin (28.3 vs. 18.2%, p < 0.05). In this setting, additional treatments with PGE2 (10 or 20 µM) for the same period of time significantly increased the proportion of PROM1 (CD133)+CD44+ cells in a dose-dependent manner (10 µM, 24.9% and 20 µM, 33.6%) (Fig. 3b). Similarly, the treatment of SW620 cells with 5-FU and indomethacin with/without PGE2 for 96 hr showed the same phenomenon (Fig. 3c). These results suggest that NSAIDs suppress the 5-FU-induced increase of CSCs via a PTGS2 (COX2)-dependent pathway.
NSAIDs showed CSC-inhibitory effect through NOTCH/HES1 and PPARG pathway
Since indomethacin is also known as a γ-secretase inhibitor[18, 19] and PPARG agonist,[20, 21] colosphere culture and luciferase reporter assays using CBFRE-Luc or PPRE-Luc were performed after treatment with indomethacin, γ-secretase inhibitor and PPARG agonist. As shown in Figure 4a, treatment with γ-secretase inhibitor (DAPT and JLK6) resulted in a significant decrease in the number of colospheres in SW620 cells (relative number of spheres; control, 100% vs. 2.5, 5.0 and 10 µM DAPT, 37.1, 29.2 and 19.6%; p < 0.05 and control, 100% vs. 2.5, 5.0 and 10 µM JLK6, 71.4, 57.3 and 31.7%; p < 0.05). PPARG agonist (rosiglitazone) also significantly reduced spheres in this setting (control, 100% vs. 2.5, 5.0 and 10 µM rosiglitazone, 74.0, 47.9 and 32.1%; p < 0.05). In CBFRE-Luc reporter assays using LiCl, an inhibitor of GSK3, which is a negative regulator of NOTCH1/NICD, indomethacin treatment (50, 75 and 100 µM) as well as DAPT (30 µM), significantly attenuated the LiCl-induced increase of CBFRE transcriptional activity in a dose-dependent manner (p < 0.05) (Fig. 4b). In addition, treatment with 6.25, 12.5 and 25.0 µM of indomethacin as well as 10 µM of rosiglitazone significantly increased PPRE transcriptional activity (p < 0.05) (Fig. 4c). Indomethacin also significantly decreased HES1 expression and increased PPARG expression in Western blot assays (Fig. 4d). Moreover, as shown in Figures 4e and 4f, the inhibitory effect of indomethacin on the 5-FU-induced increase of PROM1 (CD133)+CD44+ cells was reversed with the addition of BADGE, which is a competitive inhibitor of PPARG agonist. Compared to those treated with 5-FU alone, Caco-2 cells treated with 5-FU and indomethacin showed the decreased proportion of PROM1 (CD133)+CD44+ cells (18.6%), which increased to 36.3% following the addition of 6.25 µM BADGE (p < 0.05) (Fig. 4e). Similarly, compared to SW620 cells treated with 5-FU and indomethacin, additional treatment with BADGE resulted in a significant increase in PROM1 (CD133)+CD44+ cells (22.3 vs. 34.8%, p < 0.05) (Fig. 4f), suggesting that the inhibition of PPARG activity by BADGE could suppress the CSC-inhibitory effect of indomethacin. These results suggest that indomethacin can down-regulate NOTCH/HES1 signaling and up-regulate the expression of PPARG, and these molecular pathways are related to the CSC inhibitory effects of indomethacin.
Indomethacin inhibited CSCs in an in vivo xenograft mouse model
In a xenograft mouse experiment, we used 5-FU-resistant SW620 cells. Based on the proportion of PROM1 (CD133)+CD44+ cells, 5-FU-resistant cells had a higher proportion of CSCs than the parent SW620 cells (47.1 vs. 15.4%) (Fig. 5a). 5-FU was administered during these experiments to maintain a high proportion of CSCs in tumor xenografts and simulate the conditions of patients undergoing chemotherapy. Xenograft tumor nodules were formed in all nude mice and were harvested on day 16. As shown in Figure 5b, the size of tumors was larger than others in the order given follow: control, 5-FU alone, indomethacin alone and 5-FU and indomethacin combination group. However, the difference of tumor size at day 16 among these groups could not reach statistical significance except the difference between 5-FU and 5-FU + indomethacin group. The combination of 5-FU and indomethacin decreased the tumor size by 41.3% compared to tumors in mice treated with 5-FU alone (p < 0.05, Figs. 5c and 5d).
IHC staining of xenograft tumors was performed for CSC markers (PROM1 [CD133] and CD44), PTGS2 (COX2), HES1 and PPARG (Fig. 6a). When the stained cells were analyzed quantitatively, indomethacin significantly decreased PROM1 (CD133)+ and CD44+ cells (control vs. indomethacin; PROM1 [CD133], 30.5 vs. 11.9%; p < 0.05, and CD44, 43.4 vs. 22.0%; p < 0.05), whereas 5-FU increased these cells compared to controls (control vs. 5-FU; PROM1 [CD133], 30.5 vs. 49.4%; p < 0.05, and CD44, 43.4 vs. 60.8%; p < 0.05). Furthermore, the combination treatment with 5-FU and indomethacin significantly decreased PROM1 (CD133) and CD44-positive cells compared to 5-FU alone (5-FU vs. 5-FU + indomethacin; PROM1 [CD133], 49.4 vs.27.9%; p < 0.05, and CD44, 60.8 vs. 30.5%; p < 0.05) (Fig. 6b). Indomethacin treatment also showed significant down-regulation of PTGS2 (COX2) (control vs. indomethacin; 32.6 vs. 18.3%; p < 0.05) and HES1 (40.3 vs. 23.9%; p < 0.05), and up-regulation of PPARG expression (36.5 vs. 55.7%; p < 0.05) in xenograft tumors compared to the control group. In addition, compared to mice treated with 5-FU alone, the 5-FU treatment combined with indomethacin reduced PTGS2 (COX2) (5-FU vs. 5-FU + indomethacin; 49.6 vs. 33.6%; p < 0.05) and HES1 expression (57.4 vs. 40.4%; p < 0.05) and increased PPARG expression (25.7 vs. 43.0%; p < 0.05) (Fig. 6c).
In this study, we evaluated whether NSAIDs suppress CSCs to increase the anti-cancer effects of current chemotherapy in human CRCs. We found that indomethacin significantly decreased colosphere formation and suppressed PROM1 (CD133)+CD44+ cells and the 5-FU-induced increase in PROM1 (CD133)+CD44+ cells. Similarly, the anti-CSC effects of indomethacin were also observed in a xenograft mouse model.
NSAIDs have been identified as chemopreventive agents in CRC, and experimental and clinical studies have consistently reported that NSAIDs may reduce the risk of colorectal adenoma or cancer.[23, 24] In animal models, either nonselective or PTGS2 (COX2)-selective NSAIDs have been shown to suppress CRC growth.[25, 26] Because CSCs are involved in tumor initiation and growth, our results suggest that the preventive and therapeutic effects of NSAIDs might be related to their CSC-suppressing ability in CRC. In particular, we found that treatment with indomethacin suppressed 5-FU-induced increases of CSCs. In the Cancer and Leukemia Group B 89803 trial, which was a postoperative adjuvant chemotherapy trial of patients with stage III colon cancer, aspirin intake was significantly associated with a lower risk of cancer recurrence or death. Based on these reports and our results, we would hypothesize that NSAIDs like indomethacin could prevent CRC recurrence through CSC-inhibiting effects. However, further clinical and experimental studies are needed to further explore this issue.
The CSC-inhibiting effects of NSAIDs and relevant mechanisms in CRC have not been elucidated. The anticarcinogenic activity of NSAIDs in CRC may depend mostly on the inhibition of PTGS2 (COX2) activity, because prostaglandins (PGs) play important roles in tumorigenesis in CRC.[28-30] PTGS2 (COX2) overexpression was documented in 85% of human CRC cases and about 50% of colorectal adenomas and was also identified in an animal model, showing that a PTGS2 (COX2) gene null mutation significantly reduced the number and size of polyps on ApcΔ716 mice. In our study, we found that the anti-CSC effects of NSAIDs were related to a PTGS2 (COX2)-dependent pathway. As for the CSC-related downstream of PTGS2 (COX2), earlier study reported that PTGS2 (COX2) overexpression leads to the production of PGE2, which ultimately stimulates the β-catenin-mediated transcription in colon cancer. The WNT/β-catenin pathway is thought to be involved in the regulation of CSCs and is one of the most interesting therapeutic targets in CSCs.
We also investigated the PTGS2 (COX2)-independent pathways of NSAIDs as a mechanism of anti-CSC activity. Previous studies reported that traditional NSAIDs have anti-cancer effects via PTGS2 (COX2)-independent mechanisms.[35, 36] Recently, second mitochondria-derived activator of caspase (DIABLO)-mediated apoptosis was identified as a mechanism for the removal of oncogenic intestinal stem cells by NSAID treatment. However, there is no direct evidence showing the CSC-inhibiting effect of NSAIDs through a PTGS2 (COX2)-independent pathway in CRC. In several previous reports that were not related to CSC studies, NSAIDs were shown to inhibit NOTCH/HES1 signaling pathway[18, 19] and activate the PPARG expression.[20, 21] NOTCH/HES1 signaling has been shown to be oncogenic in CRCs through inhibiting the terminal differentiation of epithelial cells, and the dysregulation of the NOTCH/HES1 signaling was implicated in the self-renewal and maintenance of CSCs in CRC. Meanwhile, PPARG activation resulted in growth arrest and induced differentiation of colon cancer cells. In addition, the CSC-inhibitory effect of PPARG agonist through the inhibition of JAK-STAT pathway was demonstrated in brain CSCs. Therefore, we sought direct evidence of effects of NSAIDs on CSCs by inhibiting NOTCH signaling and activating the PPARG pathway. Based on previous evidence and our results, the NOTCH pathway and PPARG may be related to CSCs in CRC and were down- and up-regulated by NSAIDs, respectively. Thus, we suggest that NSAIDs suppress colon CSCs via PTGS2 (COX2)-independent pathways, such as the NOTCH pathway and PPARG as well as a PTGS2 (COX2)-dependent pathway.
Meanwhile, in this study, the CSC-suppressing effect of indomethacin in SW620 cells was more distinct in in vivo mouse experiments compared to in vitro studies. For this phenomenon, we assumed that indomethacin could influence the tumor microenvironment as well as CSCs itself in a mouse model. It has been revealed that cancer cells constantly interact with the complex microenvironment during carcinogenesis. Similarly, the proliferation and survival of CSCs could be enhanced by the activated tumor microenvironment including various cellular components such as inflammatory cells and fibroblast, its cytokines and growth factors, and these additional effects in mouse model might be blocked by NSAIDs, showing more distinct effects in in vivo model.
Of note, this study showed different CSC-suppressing efficacies and doses of indomethacin between Caco-2 and SW620 cells, suggesting that CSC-maintaining mechanisms escaping the effects of NSAIDs might be involved differently in these two cell lines. As for possible causes of this phenomenon, we can consider several mechanisms including different drug efflux. Previous study has identified the close relationship between PTGS2 (COX2) and P-glycoprotein (P-gp) in colonic inflammatory conditions including inflammatory bowel diseases (IBDs) and the different expression levels of MDR1 were noted in the different cell lines although they were documented in leukemia cells. Therefore, the different activity of P-gp-medicated drug efflux would be one of possible causes for the different CSC-inhibitory efficacies of indomethacin between Caco-2 and SW620 cells. Furthermore, like different cell types, difference in the type of NSAIDs is also thought to influence the CSC-inhibiting effect. The effect on PTGS2 (COX2)-independent pathways (inhibition of NOTCH/HES1 and activation of PPARG) may be diverse depending on the types of NSAIDs, whereas the inhibitory effect on PTGS2 (COX2) activity may be similar in various NSAID agents except low dose aspirin. Especially, because low dose aspirin is not sufficient to inhibit PTGS2 (COX2), its chemopreventive benefit may mainly depend on other mechanisms.[14, 46] However, to confirm our results and verify these issues, further experiments including other cell lines and human samples were warranted.
To understand the effect of NSAIDs on CSCs, it would be also important whether the concentration of these agents determined in vitro is in line with plasma levels in humans. As for indomethacin, previous study has demonstrated that the plasma concentration 1.5 hr after single oral dose of 25 mg indomethacin was 3.3 ± 0.9 µM, which was about 4–30 times lower than those of our study (12.5 µM in Caco-2 and 100 µM in SW620). Similarly, the effective concentration for inhibiting colosphere formation by celecoxib ranged from 10 to 40 µM in our study, which is much higher than the range (1–3 µM) with clinical efficacy in arthritis. However, the relatively high concentration of indomethacin and celecoxib in this study is comparable to those in previous studies demonstrating the inhibitory effect of NSAIDs on tumor growth. The different action mechanisms of NSAIDs based on their different effect and its concentrations need to be elucidated in further experiments.
The development and discovery of NSAIDs derivatives, which can reveal the higher activity of these multitargeted mediators, may allow the development of more effective CSC-targeted therapies. The application of multiple existing agents, including metformin and statins, for other clinical purposes (termed repositioning) may lead to novel therapeutic options that selectively suppress CSCs and improve sensitivity to conventional chemotherapy.[11, 49] Low doses of metformin, a well-known diabetes drug, have inhibitory effects on CSCs, and the combination of metformin and doxorubicin effectively reduces tumor mass and prevented relapses in breast cancer. Salinomycin has also been identified as a selective inhibitor of breast CSCs by high-throughput screening of various compounds. Statins may also be chemosensitizing agents that reduce stemness and induce the differentiation of CRC cells.
In conclusion, our results demonstrate that NSAIDs have CSC-inhibitory effects in CRC, and suppress chemotherapy-induced increases of CSCs through inhibition of PTGS2 (COX2) and NOTCH/HES1, and activation of PPARG. NSAIDs may therefore be used as adjunctive treatments to improve the efficacy of conventional chemotherapy in patients with CRC.
- 27Influence of regular aspirin use on survival for patients with stage III colon cancer: findings from Intergroup trial CALGB 89803 [abstract 3530]. J Clin Oncol 2005;23:3530., , .
- 48Preliminary study of the safety and efficacy of SC-58635, a novel cyclooxygenase 2 inhibitor: efficacy and safety in two placebo-controlled trials in osteoarthritis and rheumatoid arthritis, and studies of gastrointestinal and platelet effects. Arthritis Rheum 1998;41:1591–602., , , et al.