Curcumin has promising potential in cancer prevention and therapy by interacting with proteins and modifying their expression and activity, which includes transcription factors, inflammatory cytokines and factors of cell survival, proliferation and angiogenesis. miR-21 is overexpressed in many tumours, promoting progression and metastasis. In the present study, we examined the potential of curcumin to regulate miR-21, tumour growth, invasion and in vivo metastasis in colorectal cancer. In Rko and HCT116 cells, we identified two new transcriptional start sites of the miR-21 gene and delineated its promoter region. PMA stimulation induced miR-21 expression via motifs bound with AP-1 (activator protein 1) transcription factors. Curcumin treatment reduced miR-21 promoter activity and expression in a dose-dependent manner by inhibiting AP-1 binding to the promoter, and induced the expression of the tumour suppressor Pdcd4 (programmed cell death protein 4), which is a target of miR-21. Curcumin-treated Rko and HCT116 cells were arrested in the G2/M phase with increasing concentrations. Furthermore, curcumin inhibited tumour growth, invasion and in vivo metastasis in the chicken-embryo-metastasis assay [CAM (chorionallantoic membrane) assay]. Additionally, curcumin significantly inhibited miR-21 expression in primary tumours generated in vivo in the CAM assay by Rko and HCT116 cells (P<0.00006 and P<0.035 respectively). Taken together, this is the first paper to show that curcumin inhibits the transcriptional regulation of miR-21 via AP-1, suppresses cell proliferation, tumour growth, invasion and in vivo metastasis, and stabilizes the expression of the tumour suppressor Pdcd4 in colorectal cancer.
Curcumin (diferuloylmethane), a yellow-coloured polyphenol, is an active component of the spice turmeric (Curcuma longa) and has been used for many centuries as a food additive and a traditional medicine in Asian countries. Curcumin has chemo-preventive and therapeutic properties against many tumours in in vitro and in vivo models and has been included in clinical trials . It suppresses inflammation and cell proliferation and induces apoptosis, and sensitizes tumour cells to cancer therapies [2–4]. It also suppresses invasion and angiogenesis in cancer [2,5,6]. Curcumin physically binds to different proteins such as thioredoxin reductase, COX2 (cyclo-oxygenase-2), PKC (protein kinase C) and 5-LOX (5-lipoxygenase) and modulates different molecular target molecules such as transcription factors [AP-1 (activator protein 1), NF-κB (nuclear factor κB), STAT-3 (signal transducer and activator of transcription 3) and β-catenin], growth factors and their receptors, cytokines and enzymes [5,7]. Curcumin is also known to inhibit Akt/mTOR (mammalian target of rapamycin) signalling in different cancer cells [7–9]. Curcumin has been administrated up to 10 g/day in humans without toxicity, and serum concentrations of 1.77±1.87 mol/l were observed in patients treated with 8 g/day . Many diseases, especially cancer, are the result of deregulation of more than 500 different genes . Since curcumin is known for being non-toxic at higher doses and for its role in inhibiting tumour progression by modulating a large number of molecules and mechanisms [2,5,6], it has been suggested that it should be used as a potential drug in cancer therapy.
miRNAs (microRNAs) are a class of ~22 nt endogenous, non-coding RNAs generated from a larger stem–loop structure, which can be expressed in a cell- and tissue-specific manner, influencing mRNA stability and translation. They control a wide range of biological functions such as cellular proliferation, differentiation and apoptosis [12–14]. Disruptions of miRNA-target gene regulation, including genomic instability and impaired miRNA processing, have been associated with a growing range of cancers [13,15]. Previous papers have provided strong evidence that miRNAs are key molecules involved in cancer initiation and progression [16,17].
miR-21 shows strong evolutionary conservation across a wide range of vertebrate species in mammalian, avian and fish clades, suggesting a conserved role for miR-21 in gene regulation . Gene expression studies across tumour samples and cancer cells showed that miR-21 is the only miRNA up-regulated in a large-scale profiling of 540 human samples derived from 363 specimens representing six types of solid tumours (breast, colon, lung, pancreas, prostate and stomach) and 177 respective normal control tissues . Other reports also showed that it is up-regulated in other cancer types such as glioblastoma, oesophageal cancer, ovarian carcinoma, cholangiocarcinoma, B-cell lymphoma, hepatocellular carcinomas, cervical cancer, uterine leiomyomas, head and neck cancer, chronic lymphocytic leukaemia, squamous cell carcinoma of the tongue and papillary thyroid carcinoma . This extensive miRNA profiling in different cancers compared with normal tissues/cells identified miR-21 as a key molecule associated with neoplastic transformation and as a biomarker. Functional studies of miR-21 with either sense or antisense transfection experiments in colorectal , breast [21,22], glioblastoma , hepatocellular  and neuroepithelial  cancer cells revealed its role in inducing cell proliferation, anti-apoptosis, migration, invasion and metastasis by differentially regulating molecules such as Pdcd4 (programmed cell death protein 4), PTEN (phosphatase and tensin homologue deleted on chromosome 10) and Bcl-2. These studies strongly suggest its role in oncogenesis/tumour progression.
Pri-miR-21 (primary transcript containing miR-21) is one of the first human miRNA genes for which the structure was determined as located on chromosome 17 residing within the tenth intron of the gene TMEM-49 (transmembrane protein-49) [26,27]. Pri-miR-21 and TMEM-49 are overlapping genes, and in the same direction of transcription, are independently transcribed and terminated with their respective poly(A) signal . In human promyelotic HL-60 cells, miR-21 expression is induced after PMA stimulation [27,28]. Two previous studies in cervical cancer and leukaemia cell lines suggest that the miR-21 gene has two different TSS (transcription start sites): one (CATT) is at intron 10 of TMEM-49  and the second one (ATAAC) is at 891 bp downstream of the TSS reported by Fujita et al. , which is in intron 11 of TMEM-40 . Pri-miR-21 promoter regulation is controlled by AP-1, PU.1, the SWI/SNF complex and δEF1 (zinc finger homeodomain enhancer-binding protein 1) [27,29,30]. BMP-6 (bone morphogenetic protein-6) down-regulates miR-21 expression by inhibiting the expression and binding of AP-1 and δEF1 to the miR-21 promoter . c-Jun and c-Fos are the major regulators of miR-21 among AP-1 family members after PMA stimulation .
Combining the promising data on curcumin's function as an antitumorigenic, and the role of miR-21 in tumorigenesis and progression, we performed the present study to determine the pri-miR-21 maximal promoter region and TSS sites in colorectal cancer, and especially to study the effect of curcumin on the transcriptional regulation and expression of miR-21, cell cycle regulation, tumour growth, invasion and in vivo metastasis in this cancer entity. As a result, we show that the traditional medicine compound curcumin inhibits miR-21 transcriptional regulation and expression, tumour growth, invasion and in vivo metastasis, and stabilizes the miR-21 target Pdcd4 in colorectal cancer. Furthermore, we identified multiple TSS for pri-miR-21, among which two new TSS are the major ones, one at 35 bp and the other at 28 bp upstream of the TSS reported by Fujita et al. . Moreover, we report on a new additional AP-1 site (−663/−659 bp) in the upstream region of pri-miR-21. Finally, we observed that the TSS previously reported by Cai et al. , which is located in intron 10 of TMEM-49, is spliced from the pri-miR-21 promoter in colorectal cancer cells.
MATERIALS AND METHODS
Cell culture, drug, treatments and materials
Cells were obtained from the A.T.C.C., and grown at 37°C with A.T.C.C.-recommended medium supplemented with 10% (v/v) fetal bovine serum. Curcumin was purchased from Sigma–Aldrich and dissolved in DMSO. Mock-treated control (DMSO) cells were handled in an identical manner to curcumintreated cells. Control miR and pre-miR21 were from Ambion, TaqMan primer probes for the quantification of miR-21 and RNU6B (RNA U6 small nuclear 2) were from Applied Biosystems and oligonucleotides were obtained from Metabion.
Preparation of protein extracts and immunoblotting
Cells were washed with PBS after curcumin treatment and lysed in extraction buffer (Biosource). Protein concentrations were determined by using a BCA (bicinchoninic acid) kit (Pierce, Rockford, IL, U.S.A.). For immunoblotting, samples (40 μg per lane) were boiled for 5 min, separated by SDS/PAGE and transferred to PVDF membranes. After transfer, the membranes were blocked with 5% (w/v) non-fat dried skimmed milk powder containing TTBS (Tris-buffered saline with 0.1% Tween 20) for 3 h at room temperature (21°C) and then probed with the indicated primary antibodies versus Pdcd4 (Abcam), p-c-Jun (no. sc 822x, Santa Cruz Biotechnology) and glyceraldehyde 3-phosphate dehydrogenase (no. sc 1616, Santa Cruz Biotechnology) for 2 h at room temperature. After three washes with TTBS, blots were incubated with appropriate secondary antibodies conjugated with horseradish peroxidase. After final washes with TTBS, the membranes were exposed to a film after ECL® analysis (Amersham Biosciences).
RNA isolation and RT–PCR (reverse transcription–PCR)
Total RNA was extracted from tissue culture cells and CAM (chorionallantoic membrane) tumours with TRIzol® (Invitrogen) according to the manufacturer's instructions. The RNA samples were quantified using a spectrophotometer and visualized on a Mops-formaldehyde gel for quality assurance. A 1 μg portion of total RNA was reverse transcribed by random hexamer primers using SuperScript II reverse transcriptase (Invitrogen), or 100 ng of total RNA was reverse transcribed by gene-specific primers (Applied Biosystems). Expression of pre-miR-21 was determined by SYBER green PCR [primer pairs: pre-miR-21 (sense: CATTGTGGGTTTTGAAAAGGTTA, antisense: CCACGACTAGAGGCTGACTTAGA) and R18 (RN18S1) normalizing control (sense: CGCCGCTAGAGGTGAAATTC, antisense: CATTCTTGGCAAATGCTTTCG)], and mature miRNA was determined by the TaqMan miRNA assay using RNU6B as the internal normalizing control. All TaqMan PCRs were performed in triplicate.
H-tetrazolium bromide] assay
Cells (3×103 per well) were seeded on to a 96-well plate in a total volume of 100 μl of a medium with 10% (v/v) FBS (fetal bovine serum). After 24 h, the cells were treated with either DMSO or curcumin at concentrations between 0 and 28 μM. After 48 h, the growth inhibitory effect was evaluated using 20 μl per well of CellTiter96® AqueousOneSolution (Promega) according to the manufacturer's instructions.
Cell cycle analysis
Rko and HCT116 cells were treated for 24 h with DMSO or 2.5 and 20 μM curcumin. After 24 h, the cells were washed twice with PBS and harvested by trypsinization. The cells were washed again with PBS and fixed in 70% cold ethanol for 1 h. The cells were washed with PBS once and then incubated with 4 μg of RNase A (Roche Applied Science) for 30 min at room temperature. Propidium iodide was added to the cell suspension at a final concentration of 20 μg/ml and incubated for 30 min. The cells were then analysed by flow cytometry using FACScan (Becton Dickinson). The results were quantified by using Cell Quest software (Becton Dickinson).
RACE (rapid amplification of cDNA ends) and generation of luciferase reporter constructs
RACE and luciferase reporter constructs were generated as described previously [31,32]. Briefly, the 5′-RACE-ready first-strand cDNA was synthesized using 1 μg of total RNA from Rko cells and human placenta. The 5′ flanking sequence of miR-21 was amplified using a universal primer (UMP, Clontech) and a specific primer: 5′-GTAGGCTCCTGTTGGTCAAAAGAGG-3′. The resulting PCR products were gel purified, cloned into the pGEM-T easy vector (Promega) and ten different positive clones of each were sequenced. RT–PCR was performed with a panel of colorectal, cervical (HeLa) and breast (MCF-7) cancer cells to confirm the pre-miR-21 TSS and splicing of intron 11 of TMEM-49 from the pre-miR-21 promoter, with sense 5′-GCCATACCAGAGTACAGTATCAGC-3′ and antisense 5′-CATCCAGGACAACCAGTTTTCTCC-3′ primers.
Approx. 2.0 kb of upstream sequence of the pre-miR-21 promoter was amplified by PCR from human genomic DNA and cloned into the pGL3-basic vector (Promega) using KpnI and Mlu1 restriction sites. The 1060 bp construct was generated from the 2 kb construct by PvuII and MluI digestion, backbone blunted using T4 DNA pol I and ligated. The 448 bp construct was made from the 1060 bp construct by KpnI and SacI digestion, backbone blunted using T4 DNA pol I and ligated. Mutant constructs were generated by a QuikChange® XL site-directed mutagenesis kit from Stratagene (La Jolla, CA, U.S.A.) using the 1060 bp maximum promoter activity-containing construct as a template. The sequences of oligonucleotides used for cloning and mutation are depicted in Supplementary Table S1 (http://www.bioscirep.org/bsr/031/bsr0310185add.htm). Successful incorporation of the mutations was confirmed by automated sequencing.
Transfection and luciferase assays
For luciferase assays, 0.5×106 Rko or HCT116 cells were plated on to 24-well plates and, after 24 h, transfected with Lipofectamine™ 2000 (Invitrogen). pRL-TK (25 ng, Renilla luciferase; Promega) was transfected along with luciferase reporter plasmids (1 μg) to normalize transfection efficiency. After 12 h of transfection, cells were either mock treated with DMSO or treated with increasing concentrations of curcumin (2.5–20 μM). After 24 h of treatment, cells were washed twice with PBS and lysed with 100 μl of passive lysis buffer (Promega) for 20 min. Luciferase activity of 20 μl of cell lysate was measured via the dual-luciferase reporter assay system (Promega) according to the manufacturer's instructions. Assays for all samples were performed in triplicate, and the results were averaged.
ChIP (chromatin immunoprecipitation) assay
ChIP assays were performed as described previously  using anti-p-c-Jun (sc-822 X), anti-c-Fos (sc-44 X) or anti-IgG (as the control, sc-2338 X) antibodies. Rko cells were either mock treated with DMSO or treated with curcumin (10 μM) 1 h before PMA stimulation. At 24 h post-stimulation, cells were processed as described in . p-c-Jun and c-Fos binding to the pri-miR-21 promoter was measured using SYBER green PCR with two sets of specific primers: set 1, sense 5′-GCCTCCCAAGTTTGCTAATG-3′ and antisense 5′-TGTACTCTGGTATGGCACAAAGA-3′; and set 2, sense 5′-GAGATCAGGCCATTGCACTC-3′ and antisense 5′-GCAACACTGCCTAATGCTTG-3′.
Cell migration and invasion assay
Rko and HCT116 cells were pretreated with curcumin (10 and 16 μM respectively). After 24 h, cells were trypsinized, and 3×105 cells were plated on to Boyden chambers either coated with 10 μg matrigel per well (for invasion assays) or uncoated (for in vitro migration assays) in a serum-free medium containing 0.1% BSA and respective curcumin concentrations. FBS medium (10%) in the lower chamber served as the chemoattractant. After 13 h, non-invading cells were removed with cotton swabs. Invaded cells were trypsinized and counted using the ATP-luminescence-based motility-invasion assay (Promega) as previously described .
The CAM assay was performed as described in [33,34] with a few modifications. Briefly, 2×106 cells were inoculated on the CAM of 10-day-old chicken embryos. On days 12 and 14, chicken embryos were treated intravenously with curcumin or DMSO respectively. Untreated embryos served as an additional control. On day 17, the embryos were killed. Metastasis was determined by harvesting lungs and liver, and processing the tissue for genomic human DNA by quantitative alu-PCR.
Statistical analyses were performed using SPSS statistical software (SPSS, Chicago, IL, U.S.A.). The Student's t test was performed for data with normal distribution and the Wilcoxon test for all other analyses. P≤0.05 was considered statistically significant.
RACE analysis and the transcription initiation site of
pri-miR-21 is located in intron 10 of the TMEM-49 gene, and both of these genes are transcribed in the same direction. TMEM-49 expression (see Supplementary Figure S1 at http://www.bioscirep.org/bsr/031/bsr0310185add.htm) did not correlate with the miR-21 expression levels reported by Asangani et al.  in the screened panel of cancer cell lines [20, 27], suggesting that both genes have different promoters. Cai et al.  and Fujita et al.  reported two different TSS for miR-21 and both are located in different introns of the TMEM-49 protein coding genes with a distance in between of 891 bp. To define the transcription initiation site(s) for pri-miR-21, 5′-RACE was performed using a TMEM-49-specific primer anchored within exon 12 (see Supplementary Figure 2a at http://www.bioscirep.org/bsr/031/bsr0310185add.htm), with total RNA from Rko cells and human placenta. 5′-RACE amplification from RNA of both samples generated two amplicons with approximate sizes of 1200 and 500 bp respectively (Figure 1a). Interestingly, no specific amplification of the 500 bp fragment was observed in the normal human placenta. This result was not surprising since miR-21 might not be expressed in human placenta. The 1200 bp fragment size corresponds to the TMEM-49 transcript. Cloning and sequencing of the 500 bp PCR products revealed multiple 5′-start sites, with most of the clones including a region consisting of 35 bp upstream of the TSS reported by Fujita et al.  (see Supplementary Figures S2b and S3a at http://www.bioscirep.org/bsr/031/bsr0310185add.htm). Interestingly, no specific band or sequence corresponding to the TSS reported by Cai et al.  was observed (results not shown). Furthermore, RT–PCR performed with the specific primers binding to TMEM-49 intron 10 (pri-miR-21 promoter) and exon 12 in a panel of cancer cells (Supplementary Figure S2a) resulted in a 477 bp fragment (Figure 1b). These observations suggest a splicing mechanism of intron 11 of TMEM-49 from the pri-miR-21 promoter, which is a common mechanism of classical genes (Supplementary Figure 3b), and two novel major TSS at 35 and 28 bp upstream of the TSS previously reported by Fujita et al.  for the pri-miR-21 promoter.
RACE analysis to determine the miR-21 TSS and promoter characterization
pri-miR-21 transcriptional activation and expression
To determine the impact of PMA stimulation on miR-21 expression, Rko and HCT116 cells were stimulated with 100 nM PMA in a time-dependent manner. miR-21 was up-regulated after 12 h and reached maximum up-regulation after 24 h of PMA stimulation (see Supplementary Figures S4a and S4b at http://www.bioscirep.org/bsr/031/bsr0310185add.htm). These results suggest that PMA induces miR-21 expression in colorectal cancer cells. Furthermore, to delineate the miR-21 promoter region required for constitutive and PMA-inducible activity, different luciferase constructs were made either with or without the TSS reported by Cai et al.  and our new TSS, and transfected into Rko and HCT116 cells (Figure 1c). Cells were stimulated with DMSO or 100 nM PMA 12 h post-transfection, and after 24 h of stimulation, cells were harvested and analysed. Deletion of 1000 bp downstream of the novel 35 bp TSS determined by us and the TSS reported by Fujita et al.  showed a significant increase in the promoter activity either with or without PMA stimulation, and deletion of 500 bp from the 5′ flanking region of these two TSS decreased the promoter activity in both conditions when compared with the 1 kb fragment (Figure 1c). These results suggest that the maximum miR-21 promoter is spanning 1000 bp upstream of the novel TSS determined by the present study as well as the TSS reported by Fujita et al. , which is 35bp downstream of our novel TSS.
pri-miR-21 promoter activity
Rko and HCT116 cells were treated with curcumin for 48 h to determine the IC50 values for cell proliferation (Figure 2a). The MTT assay determined the cell lines' IC50 as 10 μM (Rko) and 16 μM (HCT116). To determine the activity of curcumin to regulate pri-miR-21 promoter activity, the maximum promoter activity-giving construct was transfected into Rko and HCT116 cells. At 12 h post transfection, cells were treated for 24 h with curcumin in different concentrations. Curcumin significantly reduced the promoter activity in both cell lines (Figure 2b). To determine the functional relevance of an upstream new AP-1-binding motif (at −663/−659 bp: AP-1 IV) in the promoter, a mutational construct specifically for this site was generated and luciferase assays were performed. Either with or without PMA stimulation, the promoter activity was decreased. In particular, with PMA stimulation, the promoter activity of this AP-1 IV mutant construct was significantly reduced when compared with the PMA-stimulated wild-type construct. Additionally, combinatorial AP-1 mutation constructs (AP-1-binding motifs −58/−55 bp: AP-1 I; −165/−162 bp: AP-1 II; and −226/−223 bp: AP-1 III, proximal to the TSS reported by Fujita et al. ) were generated along with the new upstream AP-1-binding motif and transfected into Rko and HCT116. Mutation of all four AP-1 motifs reduced the basal promoter activity significantly when compared with the wild-type construct, the activity of this mutant being significantly reduced compared with other combinational constructs mutated for different AP-1 motifs. Moreover, PMA-induced promoter activity was most significantly reduced by the construct mutated for all four AP-1 motifs, with an activity comparable with basal promoter activity. The curcumin-inhibited promoter activity was not reduced further when compared with its DMSO control with the four AP-1 mutant construct. However, curcumin pretreatment inhibited PMA-induced promoter activity in both cell lines (Figure 2c). ChIP analysis was performed to determine the in vivo relevance of AP-1 transcription factor binding to the promoter for curcumin treatment, PMA stimulation and a combination of both. PMA stimulation increased the binding of p-c-Jun/AP-1 and c-Fos/AP-1 to the pri-miR-21 promoter (Figure 1d). Under similar conditions, curcumin decreased the binding of p-c-Jun and c-Fos. Additionally, 1 h pretreatment with curcumin followed by PMA stimulation could not increase the binding of p-c-Jun and c-Fos to the promoter (Figure 2d). Similar results were observed in a Western-blot analysis; curcumin inhibited c-Jun activation in terms of phosphorylation (Figures 2e and 2f; Supplementary Figure 4c). These results suggest that curcumin inhibits pri-miR-21 transcriptional regulation, and also its PMA induction, through AP-1 family members.
Curcumin inhibits AP-1 binding to miR-21 and
pri-miR-21 promoter activity
Curcumin represses miR-21 expression and induces Pdcd4 protein expression
To demonstrate the effect of curcumin on pre-miR-21 and miR-21 expression and function, quantitative PCR and Western-blot analyses were performed for pre-miR-21, miR-21 and Pdcd4, which has been shown to be an miR-21 post-transcriptional target , with 24 h of curcumin treatment. Rko and HCT116 cells showed a significant reduction in pre-miR and miR-21 expressions after curcumin treatment in a dose-dependent manner (Figures 3a and 3c). Moreover, curcumin inhibited the PMA-induced miR-21 expression, but not completely (Supplementary Figure 4d). In parallel, Pdcd4 protein amounts were increased significantly (P<0.05) in both cell lines (Figures 3b and 3d). These results suggest that curcumin is inhibiting miR-21 expression and function in colorectal cancer, as shown by a reduction of its target Pdcd4, an important tumour suppressor in colorectal cancer.
Curcumin inhibits miR-21 expression
Curcumin inhibits the cell cycle, migration and invasion
Curcumin arrests tumour cells in the G2/M phase of the cell cycle
Curcumin has been reported to inhibit cell growth and to induce apoptosis in cancer cells [2,5,35]. Dose-dependent treatment of curcumin inhibited the growth of Rko and HCT116 cells (Figure 4a and 4b). FACS was carried out to analyse the cell cycle profile of curcumin-treated cells. The proportion of cells with actively replicating DNA, which represents the S phase, did not change significantly with 24 h of curcumin treatment. However, the proportion of cells containing 2N DNA, which represents the G1 phase, decreased significantly in curcumin-treated Rko (74–36%) and HCT116 (67–22%) cells. Similarly, the fraction of cells with 4N DNA, which represents the G2/M phase, increased in Rko (13–47%) and HCT116 (17–67%) cells in a dose-dependent manner. These results suggest that treatment of Rko and HCT116 cells with curcumin results in the arrest of cells in the G2/M phase of the cell cycle.
Curcumin inhibits migration and invasion
To demonstrate the ability of curcumin to inhibit migration and invasion in colon cancer, Rko and HCT116 cells were pretreated with DMSO or curcumin (10 and 16 μM respectively). After 24 h, cells were either seeded in Boyden chambers for migration, or in 10 μg matrigel-coated chambers for invasion. Curcumin-treated cells showed significantly decreased migration (Rko: P=0.01; HCT116: P=0.04) and invasion (Rko: P=0.05; HCT116: P=0.02) in both cell lines when compared with DMSO-treated cells (Figures 4c and 4d). These results suggest that curcumin inhibits migration and invasion in colorectal cancer cells.
Curcumin inhibits tumour and metastasis formation
To study the ability of curcumin to regulate in vivo metastasis, we employed the CAM assay of the chicken embryo, which is able to differentiate primary tumour growth and metastasis, among other phenomena, in a highly sensitive and quantitative in vivo setting [33,34]. For investigating the regulation of primary tumour growth, Rko and HCT116 cells were inoculated on the upper CAM of 10-day-old chicken embryos either with, or without, 24 h pretreatment with DMSO or curcumin (20 μM). On day 12, the embryos with pretreated cells were treated on the upper CAM with DMSO or curcumin. The primary tumours on the upper CAM were removed from the experiments performed with the pretreated cells, weighed and quickly frozen in liquid nitrogen. For the metastasis assay, on days 12 and 14 the embryos with plain cells were treated intravenously with 20 μM curcumin or DMSO. On day 17, the intravenously treated embryos were killed and their lungs and livers were harvested for DNA. The number of cells metastasized into the liver and lungs was measured with a TaqMan-based PCR amplifying human alu-sequences on the chicken background [33,34]. We found that curcumin was able to significantly reduce primary tumour growth of Rko (P=0.02) and HCT116 (P=0.016) cells (Figures 5a and 5c), and also distant metastasis of Rko cells (liver, P=0.01: lungs, P=0.011). In HCT116 cells, we observed a trend for metastasis inhibition when compared with respective DMSO controls (Figures 5b and 5d). These results suggest that curcumin inhibits tumour growth and in vivo metastasis of colorectal cancer cell lines.
Curcumin reduces tumour growth and metastasis
Curcumin inhibits miR-21 expression in primary tumours
in vivo in the CAM assay
To support the in vitro results on curcumin down-regulation of miR-21, total RNA was harvested from the frozen resected tumours isolated on the upper CAM from both Rko and HCT116 cells. Quantitative PCR was performed for pre-miR-21 and miR-21 expression, with DMSO- or curcumin-treated samples from both cell lines. We observed that all samples/tumours (n = 5) treated with curcumin showed reduced pre-miR-21 and miR-21 expression when compared with DMSO-treated samples/tumours from both cell lines. Pre-miR-21 expression in individual samples is shown in Supplementary Figure S5 (http://www.bioscirep.org/bsr/031/bsr0310185add.htm). The mean expression of pre-miR-21 was significantly down-regulated in curcumin-treated samples in Rko (P=0.0002) and HCT116 (P=0.041). Likewise, mean miR-21 expression in Rko (P=0.00006) and HCT116 (P=0.0035) was significantly down-regulated when compared with the mean of DMSO-treated samples/tumours (Figures 6a and 6b). These results confirm the in vivo relevance of our in vitro findings. Taken together, these results show that curcumin significantly inhibits tumour growth and in vivo metastasis and that this is, in part, mediated by miR-21.
Curcumin inhibits miR-21 expression
in vivo (CAM tumours)
Cancer is a complex disease occurring as a result of a progressive accumulation of genetic aberrations and epigenetic changes that enable cells to escape from normal endogenous and environmental controls . Cancer arises as a localized disease, progresses to an invasive primary tumour and finally becomes metastatic [37,38]. Colon cancer is one of the major causes of cancer-related death worldwide, representing the third most common and second leading cause of cancer death. However, too little is known still about the mechanisms of cancer metastasis and especially strategies to inhibit metastasis. miRNAs have been shown to have an important role in tumour biology, including oncogenesis, progression, invasion, metastasis and angiogenesis. miRNAs are now recognized as key players of post-transcriptional control of gene expression. It has been shown that miRNAs are aberrantly expressed or mutated in cancer, suggesting that they may function as a novel class of oncogenes or tumour suppressor genes, and ongoing investigations have started to implicate specific miRs in metastasis [18,39].
miRNA genes were reported to be transcribed by RNA polymerase II to produce pri-miRNAs with a terminal modification of a 5′-7-methyl guanylate (m7G) cap and a 3′ poly (A) tail, like traditional protein coding genes. This process has also been reported to be regulated by known transcription factors. miR-21 is one of the first human miRNAs whose structure was determined [26,40], and gene expression studies on different human tumour samples reported that miR-21 is up-regulated, especially in lung and colorectal tumour tissues [18–20]. However, the transcriptional regulation of miR-21 is not completely understood and is also not clear in terms of TSS. Cai et al.  reported a TSS 2445 bp upstream of the pri-miR-21 gene, in the TMEM-49 intron 11, which is constitutively active, in HEK (human embryonic kidney)-293 cells. Loffler et al.  reported a different TSS 50 bp upstream of the TSS reported by Cai et al., which is regulated by conserved enhancer-binding motifs of STAT3 (signal transducer and activator of transcription 3), in melanoma cells. A third TSS was reported for pri-miR-21 by Fujita et al.  in cervical cancer (HeLa) and leukaemia (HL-60) cell lines , to be located in intron 10 of TMEM-49. However, the previously reported transcriptional start sites were not detected in the study by Fujita et al. . Moreover, the TSS reported by Fujita et al. minimally overlaps approx. 60 bases with the TSS of Cai et al., and both are functioning separately, suggesting that each is an independent promoter [26,27]. In the present study, with 5′-RACE analysis in colorectal cancer with the specific primer in exon 12 of TMEM-49, we determined multiple TSS for pri-miR-21, which are either identical with, or even as novel as, TSS located around the TSS reported by Fujita et al. . No specific bands were observed to support the TSS previously reported by Cai et al.  and Loffler et al.  in our data. Interestingly, intron 11 of TMEM-49 was observed to be spliced from the pri-miR-21 gene to support RNA polymerase II transcription, which is a common mechanism in the processing of normal protein coding genes [18,26], and correspondingly, the TSS reported by Cai et al. and Loffler et al. were located within TMEM-49 intron 11. In the present study, the construct that included the Cai-TSS was not able to induce promoter activity unlike the TSS reported by Fujita et al. and the novel TSS presented by us, suggesting that the TSS reported by Cai et al.  is not essential for the pri-miR-21 promoter in colorectal cancer cells. Our findings of the pri-miR-21 TSS within intron 10 support the findings of Fujita et al. in colorectal cancer. Certainly, this TSS is close to the core promoter, with a typical structure for most of the human coding genes, suggesting the general biological importance of this TSS. Still, we think that the most likely explanation for the discrepancies found for diverse TSS in the other studies is that many TSS might, in part, be cancer entity specific [27,42]. Ozsolak et al.  identified and assayed the transcriptional activity of four regions upstream of pri-miR-21. Two of them are very similar to the promoters reported by Cai et al. and Fujita et al. They observed a superior induction of promoter activity with the region identified by Fujita et al. and also found that the promoter reported by Cai et al. is more active in HeLa cells, but not active in melanoma cells. These observations support the notion that there may be cell lineage and/or tumour type specificity for the usage of the respective TSS and the transcriptional regulation of pri-miR-21.
Subsequently, our promoter analyses on pri-miR-21 revealed that PMA-induced AP-1 binds to the pri-miR-21 promoter and triggers the transcription of the pri-miR-21 gene. Additionally, we found a novel AP-1-binding motif (−663/−659 bp) in the upstream sequence of the pri-miR-21 promoter, which is also important for core promoter activity and PMA-stimulated promoter activity in colorectal cancer. Transcriptional regulation of pri-miR-21 has been shown to be controlled by AP-1, PU.1, the SWI/SNF complex, STAT3 and δEF1 [27,29,30,41]. Enhancement of endogenous activity of AP-1 can be induced through various signalling axes, and is associated with oncogenic transformation [44,45]. It has been shown that enhanced miR-21 expression in several cancers is due to transcriptional regulation by an enhanced activation of AP-1 [16,18,44,45]. PMA induces miR-21 expression through AP-1 family transcription factors, mainly c-Jun and c-Fos , as we have also shown here. Curcumin physically binds and modulates different target molecules such as AP-1 and NF-κB [2,5]. Correspondingly, we have now shown that curcumin reduces miR-21 expression by inhibiting the activation and binding of AP-1 to the pri-miR-21 promoter. Taken together, the results suggest that curcumin reduces miR-21 expression by inhibiting AP-1 function.
This is the first study to show that curcumin inhibits invasion and in vivo metastasis by reducing miR-21 expression. Pdcd4 is a tumour suppressor shown to inhibit phorbol ester-induced neoplastic transformation  and tumour promotion and progression  and also to suppress tumour progression in human colon carcinoma cells by down-regulating MAP4K1 (mitogen-activated kinase kinase kinase kinase 1) transcription and inhibiting c-Jun activation and AP-1-dependent transcription [48,49]. Recently, we showed that Pdcd4 is post-transcriptionally inhibited by miR-21 . Pdcd4 down-regulation in lung and colorectal cancers has been associated with poor patient prognosis [50,51]. Ras-induced AP-1 activity enhanced miR-21 expression and down-regulated Pdcd4 by an autoregulatory loop mechanism . Based on these results, we hypothesized that curcumin might inhibit miR-21 expression by down-regulating AP-1. Consistent with this, curcumin treatment increased Pdcd4 expression and inhibited AP-1 in the present study, leading to a hypothetical model of interactions as depicted in Figure 6(c). Our findings suggest that curcumin-induced tumour suppression, and inhibition of migration, invasion and in vivo metastasis, might in part be mediated by miR-21-triggered, Pdcd4-dependent mechanisms [20–22,24]. Additionally, we found that curcumin inhibited the cell cycle in G2/M phase, which is in line with earlier findings [9,52,53]. This again could in part be mediated by miR-21-targeted cell cycle regulators [18,23]. However, our present findings certainly cannot exclude other important mechanisms mediated by curcumin in the inhibition of miR-21-regulated and -induced cancer progression (Figure 6c) [2–11,35].
Taken together, our results suggest that curcumin is an inhibitor of miR-21 and down-regulates miR-21 expression by inhibiting AP-1 in colorectal cancer. Curcumin also inhibits in vivo tumour growth, invasion and distant metastasis in colorectal cancer. Additionally, curcumin treatment induces tumour suppressor Pdcd4 expression. Our results add to the understanding of the antitumorigenic activities of curcumin and the regulatory mechanisms of miR-21 that function in cancer progression. Our findings also explain a possible mechanism of Pdcd4-mediated tumour suppression induced by curcumin treatment.
activator protein 1
fetal bovine serum
nuclear factor κB
programmed cell death protein 4
primary transcript containing miR-21
rapid amplification of cDNA ends
signal transducer and activator of transcription 3
transcription start site
Tris-buffered saline with 0.1% Tween 20
Giridhar Mudduluru, Jonahunnatha George-William, Santoshi Muppala, Irfan Asangani and Kumarswamy Regalla carried out the experiments; Giridhar Mudduluru and Jonahunnatha George-William collected and/or assembled the data; Giridhar Mudduluru, Jonahunnatha George-William and Santoshi Muppala analysed and interpreted the data; Giridhar Mudduluru, Laura Nelson and Heike Allgayer prepared the manuscript; and Heike Allgayer obtained financial support. All authors approved the final version of the manuscript.
We thank Erika Hillerich and Paolo Ceppi for their excellent help and a critical appraisal of the manuscript prior to submission.
H.A. was supported by the Alfried Krupp von Bohlen und Halbach Foundation (Award for Young Full Professors), Essen; Hella-Bühler-Foundation, Heidelberg; Dr Ingrid zu Solms Foundation, Frankfurt/Main; the Hector Foundation, Weinheim, Germany; the FRONTIER Excellence Initiative of the University of Heidelberg; BMBF, Germany; and the Walter Schulz Foundation, Munich, Germany.
These authors contributed equally to this work.