Inhibition of cyclo-oxygenase 2 expression in colon cells by the chemopreventive agent curcumin involves inhibition of NF-κB activation via the NIK/IKK signalling complex


Colorectal cancer is a major cause of cancer deaths in Western countries, but epidemiological data suggest that dietary modification might reduce these by as much as 90%. Cyclo-oxygenase 2 (COX2), an inducible isoform of prostaglandin H synthase, which mediates prostaglandin synthesis during inflammation, and which is selectively overexpressed in colon tumours, is thought to play an important role in colon carcinogenesis. Curcumin, a constituent of turmeric, possesses potent anti-inflammatory activity and prevents colon cancer in animal models. However, its mechanism of action is not fully understood. We found that in human colon epithelial cells, curcumin inhibits COX2 induction by the colon tumour promoters, tumour necrosis factor α or fecapentaene-12. Induction of COX2 by inflammatory cytokines or hypoxia-induced oxidative stress can be mediated by nuclear factor kappa B (NF-κB). Since curcumin inhibits NF-κB activation, we examined whether its chemopreventive activity is related to modulation of the signalling pathway which regulates the stability of the NF-κB-sequestering protein, IκB. Recently components of this pathway, NF-κB-inducing kinase and IκB kinases, IKKα and β, which phosphorylate IκB to release NF-κB, have been characterised. Curcumin prevents phosphorylation of IκB by inhibiting the activity of the IKKs. This property, together with a long history of consumption without adverse health effects, makes curcumin an important candidate for consideration in colon cancer prevention.


Cyclo-oxygenase 2 (COX2), an inducible isoform of prostaglandin H synthase (PGHS) which mediates prostaglandin synthesis during inflammation, is selectively overexpressed in colon tumours and is thought to play an important role in colon carcinogenesis (Kargman et al., 1995). Genetic knock-out or pharmacological inhibition of COX2 has been shown to protect against development of colonic tumours in mice which harbour a germline knock-out mutation of the adenomatous polyposis coli (APC) tumour suppressor gene, or in rats exposed to the colon carcinogen azoxymethane (Oshima et al., 1996; Kawamori et al., 1998). Germline and somatic mutations in the APC gene are important early events in human colon carcinogenesis (Nishisho et al., 1991; Powell et al., 1992). Overexpression of COX2 in colonic epithelial cells may promote tumour development by causing resistance to apoptosis and facilitating alterations in cell adhesion properties (Tsujii and Dubois, 1995). Non-steroidal anti-inflammatory agents (NSAIDs), which directly inhibit COX2 activity (Vane and Botting, 1995), cause regression of adenomatous polyps in familial adenomatous polyposis (FAP) patients (Giardiello et al., 1995), and may reduce the risk for sporadic colon cancer (Rosenberg et al., 1991; Thun et al., 1991). It has therefore been suggested that COX2 is an important target for the chemopreventive effects of these agents (Dubois and Smalley, 1996; Gardiello et al., 1997). However, the chronic administration of NSAIDs causes serious side-effects, thought to be due to concomitant inhibition of COX1 (Eberhart and DuBois, 1995), a constitutively expressed isoform of PGHS, making the development of selective COX2 inhibitors highly desirable. Such agents could act either by direct inhibition of the cyclo-oxygenase or peroxidase component of COX2 and/or by inhibition of COX2 gene expression (Subbaramaiah et al., 1997).

Curcumin, like NSAIDS, is a potent anti-inflammatory agent due to inhibition of prostaglandin synthesis (Huang et al., 1992). However, it has been shown to be only a very weak direct inhibitor of cyclo-oxygenase enzyme activity (Srivastava and Srimal, 1985). It also has chemopreventive activity in animal models of colon cancer (Rao et al., 1993; Pereira et al., 1996), but its mechanism of action is not well understood. It has been shown to inhibit COX2 expression (Kelley et al., 1996), and in separate studies to be a potent inhibitor of NF-κB activation (Sanjaya and Aggarwal, 1995; Bierhaus et al., 1997; Kumar et al., 1998), but not of its binding to DNA (Bierhaus et al., 1997).

To investigate the mechanisms of colon cancer chemoprevention by curcumin we tested the hypothesis that it acts through inhibition of COX2 gene induction by the model colon tumour promotors, tumour necrosis factor α (TNFα) and fecapentaene-12. The lumenal concentration of TNFα is increased in patients with inflammatory bowel disease (Braegger et al., 1992), a condition which predisposes to colorectal cancer (Gyde et al., 1998). Fecapentaene-12, the most abundant form of a group of mutagenic chemicals found in the faeces of individuals who consume a Western diet high in fat and meat (Schiffman et al., 1989), has been shown to be a tumour promotor in a rat model of colorectal cancer (Zarcovic et al., 1993).

The overexpression of COX2 in colon tumour cells is thought to be due to alterations in transcriptional control (Kutchera et al., 1996). Several transcription factors have been implicated, but their precise roles have yet to be elucidated. The COX2 gene is induced by a wide variety of stimuli including oncogenic viruses, growth factors, cytokines, and tumour promoters (Hershman, 1994). Overexpression in response to the viral oncogene v-src is mediated in part by the ras-MAP kinase signalling pathway via an AP1 transcription factor which binds to a cyclic AMP response element (CRE) in the human gene promoter (Xie and Herschman, 1995). However, overexpression caused by hypoxia in human umbilical vein endothelial cells and by IL-1 in rheumatoid synoviocytes has been shown to be mediated by NF-κB (Crofford et al., 1997; Schmedtje et al., 1997). We investigated whether the chemopreventive activity of curcumin in colon cells is related to inhibition of COX2 expression via modulation of signalling pathways that regulate the stability of the NF-κB-sequestering protein IκB. Serine/threonine and tyrosine kinases that mediate activation of NF-κB by the TNF receptor 1 (TNFR1) via phosphorylation of the inhibitory protein IκB, have recently been cloned (Malinin et al., 1997; DiDonato et al., 1997). They form a complex which includes NF-κB inducing kinase (NIK) and two isoforms of the IκB kinase (IKKα and IKKβ) (Woronicz et al., 1997; Cohen et al., 1998). We present data showing that curcumin does indeed inhibit COX2 expression and NF-κB DNA binding induced by physiologically relevant concentrations of TNFα and fecapentaene-12 in human colon epithelial cells, and show that curcumin achieves this by inhibiting phosphorylation of IκB by the NIK/IKK signalling complex.


The ability of curcumin to inhibit tumour promotor-induced COX2 protein expression was first examined. Phorbol 12-myristate 13-acetate (PMA) (50 ng/ml) was the most powerful inducer of COX2 protein in human colon epithelial (HCEC) cells, causing a 25-fold induction after 6 h of treatment, which was completely inhibited by curcumin (20 μM) (Figure 1a, lanes 2 and 3). TNFα (10 ng/ml) and fecapentaene-12 (40 μM) caused smaller three and tenfold inductions, respectively, which were also inhibited by curcumin (Figure 1a, lanes 4 – 7).

Figure 1

Effects of curcumin (CUR) on COX2 expression by PMA, TNFα or fecapentaene-12 (FEC) in HCEC cells. (a) Western blot of COX2 in control (vehicle treated) HCEC cells or cells treated as indicated with 50 ng/ml PMA, 50 ng/ml PMA +20 μM curcumin, 10 ng/ml TNFα, 10 ng/ml TNFα +20 μM curcumin, 40 μM fecapentaene-12 +20 μM curcumin for 6 h. Western blot data are representative of three separate experiments. Arbitrary units represent the sum of the pixel values in the band minus the background. (b) Effect of TNFα on expression of COX2 mRNA in control (vehicle treated) HCEC cells and cells treated as indicated with TNFα (0.1 – 10 ng/ml) for 2 h. Levels of mRNA were determined by RT – PCR. *Significantly higher than untreated cells as determined by Student's t-test, p>0.05. (c) Effect of curcumin on induction of COX2 mRNA by TNFα in control (vehicle treated) HCEC cells and cells treated as indicated with curcumin (40 μM), TNFα (1 ng/ml) or TNFα+curcumin (10 – 40 μM). Results are means±s.d. of three measurements performed in three separate experiments. *,**Significantly lower than TNFα alone as determined by Student's t-test, P<0.05, P<0.001

We next examined the effects of TNFα (0.1 – 10 ng/ml) and fecapentaene-12 (1 – 40 μM) on COX2 mRNA levels by RT – PCR. TNFα caused a dose dependent induction (Figure 1b), which was maximal after 2 – 3 h, returning to baseline after 12 h (data not shown). Curcumin (10 – 40 μM) inhibited the induction of COX2 mRNA by TNFα, also in a dose dependent manner (Figure 1c, samples 4 – 6). Similar induction was evident after treatment with fecapentaene-12, which was inhibited by 75% in the presence of 40 μM curcumin (data not shown). The viability of cells after 4 h of treatment was greater than 90%, as determined by the MTT assay. Curcumin alone at concentrations of 25 μM and above had a marked (60 – 95%) inhibitory effect on HCEC cell growth, whereas 10 μM had little or no inhibitory effect (data not shown).

Since the promoter region of COX2 contains two NF-κB binding sites, and since TNFα is a potent activator of NF-κB in many cell types, and fecapentaene-12 has been shown to act as a cofactor for protein kinase C (Hoshina et al., 1991), which is known to activate NF-κB (Diaz-Meco et al., 1993), we used electrophoretic mobility shift assays (EMSA) to examine the ability of curcumin to block NF-κB activation by these agents. After treatment of cells with either TNFα (1 ng/ml) or fecapentaene-12 (20 μM), there was a marked induction of nuclear protein binding to an oligonucleotide containing the `consensus' NF-κB binding sequence (Figure 2a, lane 2; Figure 2b, lane 14). Pretreatment of cells with curcumin (10 – 40 μM) caused an inhibition of protein-DNA binding induced by either TNFα (Figure 2a, lanes 3 – 5) or fecapentaene-12 (Figure 2b, lane 15). To eliminate the possibility that the inhibitory effects of curcumin on NF-κB-DNA binding could be attributed to its antioxidant properties, we tested the ability of caffeic acid phenyl ester (CAPE), an antioxidant which is structurally related to curcumin, to inhibit binding. CAPE (10 – 40 μM) had no significant inhibitory effect on the induction of protein-DNA binding induced by TNFα (Figure 2a, lanes 6 – 8), nor did it affect the induction of COX2 mRNA levels by either tumour promoter (Figure 3a). Similar results were obtained when N-acetyl cysteine (5 mM), a classical antioxidant, was substituted for CAPE (Figure 3b). These results suggest that curcumin inhibited activation and translocation of NF-κB to the nucleus in a manner unrelated to its antioxidant capacity.

Figure 2

Effects of curcumin on NF-κB DNA binding to `consensus' NF-κB oligonucleotides. Nuclear proteins were extracted from HCEC cells and incubated for 40 min on ice with end-labelled oligonucleotides prior to electrophoresis. (a) EMSAs with NF-κB `consensus' oligonuleotide in HCEC cells treated as indicated with vehicle (control), 1 ng/ml TNFα, TNFα +curcumin (10 μM – 40 μM), TNFα+CAPE (10 – 40 μM), 40 μM curcumin alone or 40 μM CAPE alone; (b) vehicle (control), 20 μM fecapentaene-12, fecapentaene-12+20 μM curcumin, or 20 μM curcumin alone

Figure 3

Effect of CAPE or N-acetyl cysteine (NAC) on COX2 mRNA levels following exposure of HCEC cells to TNFα or fecapentaene-12 (FEC). (a) Lack of inhibition of COX-2 mRNA induction by CAPE in HCEC cells treated as indicated with vehicle (control), CAPE 20 μM, TNFα (1 ng/ml), TNFα+CAPE (20 μM), fecapentaene-12 (20 μM), fecapentaene-12 (20 μM)+CAPE (20 μM); (b) Lack of inhibition of COX-2 mRNA induction by N-acetylcysteine (NAC) in HCEC cells treated as indicated with vehicle (control), NAC (5 mM), TNFα (1 ng/ml), TNFα (1 ng/ml)+NAC (5 mM), fecapentaene-12 (20 μM), fecapentaene-12 (20 μM)+NAC (5 mM). Results are means (±range) of measurements performed in two different experiments. For TNFα exposures results are from a single experiment. Similar results were obtained with TNFα at 10 ng/ml (data not shown). *Significantly higher than control (P<0.05), **not significantly different from fecapentaene alone, as analysed by Student's t-test

To assess the ability of curcumin to inhibit NF-κB transactivation, we performed transient transfection assays with an NF-κB-luciferase reporter construct (p6NF-κB), containing 6 NF-κB consensus sequences. As HCEC cells displayed poor transfection efficiency, we used the colon carcinoma cell line SW480 for these studies. Exposure of these cells, transiently transfected with p6NF-κB and a β-galactosidase plasmid pCMVgal, to TNFα (10 ng/ml) caused a threefold induction of p6NF-κB luciferase activity (Figure 4, column 6). Curcumin (20 μM) blocked this induction of luciferase activity (column 8), indicating that it inhibits the transactivating potential of NF-κB.

Figure 4

Inhibition of TNFα-induced NF-κB dependent transactivation by curcumin. SW480 cells were transfected with the plasmid tk-36-LUC (`empty' cassette), columns 1 – 4, or p6NF-κB-tk-LUC, columns 5 – 8 and treated as indicated with vehicle (control), TNFα (10 ng/ml), curcumin alone (20 μM) or TNFα (10 ng/ml)+curcumin (20 μM) prior to measuring luciferase reporter gene activity. Values represent the means±s.d. of three separate transfections carried out in duplicate. *Significantly higher than untreated sample as analysed by Student's t-test, P<0.01

To determine whether curcumin would inhibit the phosphorylation and degradation of the NF-κB sequestering protein IκB, we performed Western blot analyses with IκB-specific antibody. Curcumin completely blocked the degradation of IκB caused by TNFα (10 ng/ml) in HCEC cells (Figure 5, lanes 3 – 5). Instability of IκB is thought to be controlled by phosphorylation via the NIK/IKK signalling complex (Woronicz et al., 1997). To test the possibility that curcumin acts through inhibition of this complex, we measured its ability to inhibit NF-κB-mediated alkaline phosphatase reporter gene activity (p4NF-κB), induced either by exposure to TNFα (10 ng/ml) or by overexpression of the NIK kinase, following transfection of a NIK expression construct (pcDNA3-NIK). These experiments were carried out in HEK293 human embryonic kidney cells, to circumvent the very poor transfection efficiency in HCEC cells. Curcumin (20 μM) inhibited TNFα-mediated induction of p4NF-κB alkaline phosphatase activity by 60% (Figure 6a) and a similar inhibition (40%) was observed following overexpression of NIK (Figure 6b), suggesting that curcumin acts via inhibiting NIK or kinases downstream of NIK, namely IKKα or β. Inhibition in this system was less pronounced than the inhibition of IκB degradation observed in Western blot experiments with pNF-κB6 (see above), because curcumin was added after transfection and overexpression of NIK in order to prevent it from interfering with the transfection per se. When we performed the same transfection experiments in SW480 colon carcinoma cells, which had been pretreated with curcumin for 1 h prior to exposure to TNFα, the induction of p6NF-κB reporter gene activity was inhibited by greater that 90% (Figure 4).

Figure 5

Effect of curcumin on TNFα-induced IκB degradation. HCEC cells were treated as indicated with 10 ng/ml TNFα for various times (5 min-1 h) in the absence or presence of 20 μM curcumin which was added to the media 1 h prior to TNFα. Cytosolic proteins were subjected to analysis by Western blotting using an IκBα specific antibody (Santa Cruz)

Figure 6

Effect of curcumin or salicylate on the activity of components of the NIK/IKK complex. HEK293 cells were transfected with NIK, IKKα or IKKβ expression constructs in combination with an NF-κB reporter construct or with the reporter construct alone. Following the dual transfections (b,c,d) cells were exposed to curcumin (20 μM), salicylate (10 mM) or vehicle control prior to measuring alkaline phosphatase reporter gene activity at 2, 4 and 6 h after the addition of inhibitor to the medium. Cells transfected with the reporter construct alone (a) were exposed to TNFα (10 ng/ml), after a 1 h pretreatment with curcumin (20 μM) and alkaline phosphatase reporter gene activity was measured at 2, 4 and 6 h after the addition of TNFα. Results represent the mean±s.e.m. and are representative of at least three separate experiments. *Significantly different from reporter gene activity measured in the absence of curcumin or salicylate as determined by balanced ANOVA analysis, P<0.05

In a similar transfection experiment using a construct, pFLAG-IKKα, to overexpress IKKα, or pFLAG-IKKβ to overexpress IKKβ, curcumin was again able to inhibit the reporter gene activity of the NF-κB co-transfected reporter constructs by 45 and 60% respectively (Figure 6c). Salicylate (10 mM) also inhibited the IKKα and IKKβ overexpression mediated activation of the NF-κB reporter by 45 and 70%, respectively (Figure 6d).


Our results demonstrate that the dietary constituent curcumin inhibits activation of NF-κB which in turn inhibits expression of the COX2 gene induced by the tumour promoters PMA, TNFα or fecapentaene-12 in human colon epithelial cells. The data suggest that curcumin blocks tumour promoter-mediated NF-κB transactivation by inhibiting the NIK/IKK signalling complex, probably at the level of IKKα/β. These data are consistent with the recent observations of Morteau et al. (1998) who have shown that NF-κB is critical for the induction of COX-2 gene expression by TNFα in human colon tumour cells. Overexpression of COX2 in colon epithelial cells, which occurs during colon carcinogenesis, causes resistance to apoptosis (Tsujii and Dubois, 1995), suggesting that treatment with curcumin might reinstate susceptibility to apoptosis. This interpretation is consistent with the observation that curcumin increases the percentage of apoptotic cells in colon tumours of rats exposed to azoxymethane, while decreasing tumour incidence by 50% (Samaha et al., 1997). NF-κB activation prevents TNFα-induced cell death by blocking apoptosis (Beg and Baltimore, 1996). It is therefore possible that COX2 is one of the downstream mediators of this effect as previously suggested (Zeng-gang et al., 1996).

The magnitude of inhibition of p4NF-κB reporter gene activity by curcumin was similar whether the reporter gene induction was driven by overexpression of NIK, IKKα or IKKβ or by addition of TNFα, suggesting that curcumin has little or no inhibitory effect at the level of the TNF receptor. It is unlikely that the inhibitory effects of curcumin on the NIK/IKK complex are due to non-specific antioxidant activity, since neither CAPE nor N-acetyl cysteine had any effect on NF-κB (p65)-DNA binding induced by the tumour promoters. Higher concentrations of CAPE than those used in this study have been shown to inhibit TNFα-mediated NF-κB-DNA binding in U937 cells, but probably through blocking DNA binding directly, since there was no inhibition of IκB degradation (Natarajan et al., 1996).

It is unlikely that fecapentaene-12 activates the NIK/IKK pathway at the level of the TNF receptor, but we have shown previously that it induces oxidative stress in cells (Plummer and Faux, 1994). As reactive oxygen intermediates are a common denominator in NF-κB activating signals, they may be involved with the effects of fecapentaene-12 on the NIK/IKK pathway. However, as already mentioned, fecapentaene-12 can act as a cofactor for protein kinase C (Hoshina et al., 1991), which is also known to activate NF-κB (Diaz-Meco et al., 1993). A similar mechanism may be operated by PMA (Ghosh and Baltimore, 1990). Both TNF and fecapentaene-12 activate AP1 (Zeng-gang et al., 1996; Holloway et al., 1998), which is important in the induction of COX2 transcription by v-Src (Xie and Herschman, 1995) and PMA (Subbaramaiah et al., 1998). Since curcumin inhibits AP1 dependent transactivation (Huang et al., 1991), some of the inhibitory effects of curcumin on COX2 gene induction by the tumour promoters may also be mediated by inhibition of AP-1 (Kelley et al., 1996).

The ability to inhibit the NIK/IKK signalling complex may be common to the action of other anti-inflammatory chemopreventive agents. Aspirin and salicylate, shown previously to inhibit IκB phosphorylation and degradation (Kopp and Ghosh, 1994), have recently been shown to inhibit the phosphorylation of IκB by specifically reducing ATP binding to IKKβ (Yin et al., 1998). Salicylate was also shown to inhibit COX2 induction by LPS in macrophages (Tordman et al., 1995) and endothelial cells (Du et al., 1998) and our unpublished immunofluorescence data is consistent with this observation. Yin et al. (1998) have shown that salicylate inhibits IKKβ selectively in in vitro kinase assays. In contrast our studies show that salicylate can inhibit NF-κB activation by IKKα overexpression in addition to its inhibition of IKKβ-mediated NF-κB activation, albeit to a lesser extent. The reason for this apparent discrepency may reflect the fact that in our overexpression system the ability of IKKα to activate NF-κB is in some way dependent on IKKβ and that the effects of both inhibitors are mediated through inhibition of this latter kinase.

Whether or not to use aspirin as a cancer chemopreventive agent in large numbers of individuals at risk of developing colon cancer is currently hotly debated (Vainio, 1997; Trujillo et al., 1994), in view of the possible toxicity associated with chronic aspirin administration. If curcumin can be shown to have chemopreventive activity against colon carcinogenesis in human clinical trials, comparable to that already demonstrated unequivocally in animal models of this disease, its long history of dietary consumption without adverse health effects might make it an important alternative to aspirin for chemoprevention of this disease. It is difficult to assess the potential biological relevance of the concentrations of curcumin used in this study since the bioavailability of curcumin in the colonic epithelium is not known, and since access to the colon does not necessarily require systemic absorption, a comparison with plasma concentrations may be misleading. Studies to assess the bioavailability of curcumin in humans are ongoing.

In this study we have shown that curcumin inhibits the induction of COX2 in human colonic cells by tumour promoters that occur in the human colon. Since overexpression of this enzyme is probably important in the pathogenesis of human colon cancer, our data also suggest that measurement of the effects of curcumin on COX2 gene expression may be a useful surrogate biomarker for the assessment of its biological activity in chemoprevention trials. Finally, our finding that curcumin acts in part through inhibition of the NIK/IKK signalling cascade provides a focus for the rational development of novel chemopreventive agents.

Materials and methods

Cell culture

Human colon epithelial cells (HCEC) and SW480 colon carcinoma cells were kindly provided by Dr Andrea Pfeifer (Nestec Ltd. Research Centre, Lausanne, Switzerland) and Professor Christos Paraskeva (Bristol University), respectively. Cells were grown in Dulbecco's minimal essential medium (DMEM) (Gibco – BRL Ltd.), supplemented with 10% foetal calf serum (FCS). For HCEC cells tissue culture vessels were precoated with medium containing 10 μl/ml Vitrogen 100 (Collagen Corp.), 2.5 μg/ml human fibronectin (Sigma) and 50 μg/ml BSA (Gibco – BRL) prior to plating. Fecapentaene-12 (SR1 International, Menlo Park, CA, USA), TNFα (Sigma), curcumin (Sigma) and sodium salicylate (Sigma) were diluted to concentrations of 10 – 40 μM, 0.1 – 10 ng/ml, 10 – 40 μM and 1 mM, respectively, in DMEM. In incubations containing curcumin or salicylate, cells were pretreated for 1 h with these agents prior to the addition of the tumour promoters. Curcumin stock solutions were made up in dimethylsulphoxide (DMSO) immediately before each experiment in light-impervious tubes. The handling and storage of fecapentaene-12 was carried out according to a previously described procedure (Plummer and Faux, 1994). The viability of cells was determined by the MTT assay (Scudiero et al., 1988). Protein assays were performed using the Bradford reagent (Biorad) or the Lowry method (Sigma).

Western blot analysis

Cells were washed with buffer (20 mM Tris base, 150 mM NaCl, 5 mM glucose, 2 μg/ml leupeptin and 20 μg/ml aprotinin, pH 7.4), lysed in homogenisation buffer (20 mM Tris HCl, 2 mM EDTA, 2 mM EGTA, 6 mM β-mercaptoethanol, 2 μg/ml leupeptin and 2 μg/ml aprotinin, pH 7.5), sonicated for 20 – 30 s and centrifuged at 12 000 g for 35 min at 4°C. After electrophoresis, proteins were electroblotted to nitrocellulose, the membrane was blocked with 10% Marvel and 0.1% Tween 20 for 2 h at room temperature, incubated with primary antibody in blocking buffer (10 mM Tris, 0.14 M NaCl, 5% milk powder (Marvel) pH 7.6) overnight at room temperature and then horseradish peroxidase secondary antibody diluted 1 : 1500 (v/v) in blocking buffer for 1.5 h at room temperature, before visualisation by the ECL method (Amersham). Rabbit polyclonal antibodies raised against COX2 and IκBα (Santa Cruz Biotechnology Inc.) were used at 1 : 1000 dilution (v/v) in blocking buffer. Visualisation and quantitation of bands was carried out by a scanning densitometer (Molecular Dynamics) using Imagequant software.


Semi-quantitative RT – PCR analysis of COX2 mRNA levels was performed according to the method of Hla and Maciag (1991). Briefly, 1 μg of RNA, extracted from cells using TRIzol reagent (Gibco – BRL), according to the manufacturer's instructions, was reverse transcribed in 10 μl RT buffer containing MMLV reverse transcriptase (10 u/μl), RNasin (Gibco – BRL) (1 u/μl), dATP, dGTP, dTTP and dCTP (1 mM), random hexamers (Boeringer Mannheim) (15 pmoles/μl), MgCl2 (5 mM) and dithiothreitol (DTT) (1 mM). The reaction product was diluted 1 in 10 with distilled water and 10 μl of this subjected to PCR by adding 40 μl PCR buffer containing COX2 or GAPDH primers (1.25 pmoles/μl), MgCl2 (2.5 mM), dATP, dGTP, dCTP, dTTP (1 mM) and Taq DNA polymerase (Gibco/BRL) (0.2 u/μl) and heating for 4 min at 94°C followed by 22 (GAPDH primers) or 25 (COX2 primers) cycles of 94°C 1 min, 58°C 1 min and 72°C 1 min. PCR reaction products were subjected to electrophoresis on 1.5% agarose gels, stained with ethidium bromide and photographed prior to densitometric measurements using a Molecular Dynamics computing densitometer. The amount of amplified product was confirmed by this method to be linear with respect to the input RNA for both COX2 and GAPDH primers. The densities of the COX2 bands were normalised with respect to the GAPDH bands in parallel PCR reactions.

Electrophoretic mobility shift assays (EMSA)

Nuclear extracts were made according to the method of Staal et al. (1990) from 5×106 HCEC cells. EMSA were performed using the NF-κB `consensus' oligonucleotide- 5′-AGTTGAGGGGACTTTCCCAGGC-3′. Nuclear protein extract (4 μg) was incubated with 0.25 pmoles 32P-end-labelled oligonucleotide in binding buffer containing 20 mM Hepes (pH 7.5), 4% ficol, 0.5 μg/ml poly DIDC, 0.1 mM MgCl2 and 0.1 mM DTT, on ice for 40 min. The DNA: protein complex formed was separated from free oligonucleotide on a 4% non-denaturing polyacrylamide gel. Following electrophoresis, the gel was dried and visualization and quantitation of radioactive bands performed by a PhosphorImager (Molecular Dynamics) using image-quantTM software. Specificity of binding was checked by incubating in the presence of an excess of `cold' NF-κB oligo or an unrelated oligo containing an AP1 `consensus' DNA binding sequence – 5′-GCTTGATGAGTCAGCCGGAA-3′ (Promega).

DNA transfections

SW480 cells (1.6×107) were transfected in serum free DMEM with 1.25 pmoles of the plasmid p6NF-κB-tk-LUC (p6NF-κB) (kindly provided by Dr Patrick Baeuerle, University of Freiburg) and 0.1 pmoles pCMVB (Promega) by electroporation. The `empty' cassette (tk-36-LUC) was used as a negative control. The plasmid pCMVB was cotransfected with the other constructs to enable normalisation of luciferase activity to β-galactosidase activity, thus controlling for differences in transfection efficiency. Following transfection, cells were resuspended in DMEM containing 10% FCS and allowed to recover for 5 h. Cells were then switched to DMEM containing 1% FCS for 24 h prior to exposure to TNFα or fecapentaene-12 at final concentrations of 10 ng/ml or 20 μM, respectively, for 2 h in the absence or presence of 20 μM curcumin. Curcumin was added in DMEM with 2% FCS 1 h prior to the TNF/fecapentaene-12 exposures. Luciferase and β-galactosidase enzyme activities were measured using Promega assay kits with a Wallach MicroBeta 1450 plate reader or Labsystems iEMS reader, respectively. Luciferase activity was expressed in relative units after normalisation to β-galactosidase

HEK 293 cells were transfected with NIK (pcDNA3-NIK), IKKα (pFlag-IKKα) or IKKβ (pFlag-IKKβ) expression constructs (IκB kinase constructs were kindly provided by David Goeddel) together with an NF-κB alkaline phosphatase reporter construct (p(NF-κB)4-tk-sPAP) or with p(NF-κB)4-tk-sPAP alone using Fugene (Boeringer) according to the manufacturer's instructions. After 18 h cells were incubated in the absence or presence of curcumin (20 μM). pNF-κB4-tk-sPAP transfectants were incubated with TNFα (10 ng/ml) 1 h prior to the addition of the curcumin-containing medium. All transfections included either the β-lactamase expression construct pRSV-lactamase or the β-galactosidase expression construct pRSVβ-Gal to assess transfection efficiency. Measurements of alkaline phosphatase reporter gene activity were made at various times after addition of curcumin to the medium.


RT – PCR data were analysed using the Student's t-test. Transfection data were analysed by Student's t-test or balanced ANOVA combined with Fisher's test.



nuclear factor kappa B




tumour necrosis factor alpha




phorbol 12-myristate 13-acetate






caffeic acid phenyl ester


N-acetyl cysteine


NF-κB inducing kinase


I kappa B kinase


electrophoretic mobility gel shift assay


analysis of variance


  1. Beg AA and Baltimore D. . 1996 Science 274: 783–784.

  2. Bierhaus A, Zhang Y, Quehenberger P, Luther T, Haase M, Muller M, Mackman N, Zeigler R and Nawroth PP. . 1997 Thromb. Haemo. 77: 772–782.

  3. Braegger CP, Nicholls S, Murch SH, Stephens S and MacDonald TT. . 1992 Lancet 339: 89–91.

  4. Cohen L, Henzel WJ and Baeuerle PA. . 1998 Nature 395: 292–296.

  5. Crofford LJ, Tan B, McCarthy CJ and Hla T. . 1997 Arthritis Rheum. 40: 226–236.

  6. Diaz-Meco MT, Berra E, Municio MM, Sanz I and Lozano J. . 1993 Mol. Cell. Biol. 13: 4770–4775.

  7. DiDonato JA, Hayakawa M, Rothwarf DM, Zandi E and Karin M. . (1997) Nature 388: 548–554.

  8. Du M, Xu XM, Sansores-Garcia L, Zu Z and Wu KK. . 1998 Blood 92: 2253.

  9. Dubois RN and Smalley WE. . 1996 J. Gastroenterol. 31: 898–906.

  10. Eberhart CE and DuBois RN. . 1995 Gastroenterol. 109: 285–301.

  11. Gardiello FM, Hamilton SR, Hylind LM, Yang VW, Tamez P and Casero RA. . 1997 Cancer Res. 57: 199–201.

  12. Ghosh S and Baltimore D. . 1990 Nature 344: 678–682.

  13. Giardiello FM, Offerhaus GJA and Dubois RN. . 1995 Eur. J. Cancer 31A: 1071–1076.

  14. Gyde SN, Prior P, Allan RN, Stevens A, Jewell DP, Truelove SC, Lofberg R, Brostrom O and Hellers G. . 1998 Gut 29: 206–217.

  15. Hershman HR. . 1994 Cancer Metast. Rev. 13: 241–256.

  16. Hla T and Maciag T. . 1991 J. Biol. Chem. 266: 24059–24063.

  17. Holloway K, Munks RJL, Peer M-A and Plummer SM. . 1998 Br. J. Cancer 78: 157.

  18. Hoshina S, Ueffing M, Morotomi M and Weinstein IB. . 1991 Biochem. Biophys. Res. Commun. 176: 505–510.

  19. Huang M-T, Lysz T, Ferraro T and Conney AH. . 1992 In: Cancer Chemoprevention, Wattenberg L, Lipkin M, Boone CW and Kelloff (eds). CRC Press Inc. pp.375–391.

  20. Huang TS, Lee SC and Lin JK. . 1991 Proc. Natl. Acad.Sci. USA 88: 5292–5296.

  21. Kargman SL, O'Neill GP, Vickers PJ, Evans JF, Mancini JA and Jothy S. . 1995 Cancer Res. 55: 2556–2559.

  22. Kawamori T, Rao CV, Seibert K and Reddy BS. . 1998 Cancer Res. 58: 409–412.

  23. Kelley DJ, Sacks PG, Ramonetti JT, Schantz SP and Dannenberg AJ. . 1996 Proc. Am. Assoc. Cancer Res. 37: 909.

  24. Kopp E and Ghosh S. . 1994 Science 265: 956–959.

  25. Kumar A, Dhawan S, Hardegen NJ and Aggarwal BB. . 1998 Biochem. Pharmacol. 55: 775–783.

  26. Kutchera W, Jones DA, Matsunami N, Groden J, McIntyre TM, Zimmerman GA, White RL and Prescott SM. . 1996 Proc. Natl. Acad. Sci. USA 93: 4816–4820.

  27. Malinin NL, Boldin MP, Kovalenko AV and Wallach D. . 1997 Nature 385: 540–544.

  28. Morteau JO, Han DS and Balfour Sartor R. . 1998 Immunology 95: 537–543.

  29. Natarajan K, Sanjaya S, Burke Jr TR, Grunberger D and Aggarwal BB. . 1996 Proc. Natl. Acad. Sci. USA 93: 9090–9095.

  30. Nishisho I, Nakamura Y, Miyoshi Y, Miki Y, Ando H, Horii A, Koyama K, Utsunomiya J, Baba S, Hedge P, Markham A, Krush AJ, Peterson G, Hamilton SR, Nilbert MC, Levy DB, Bryan TM, Preisinger AM, Smith KJ, Su L-K, Kinzler KW and Vogelstein B. . 1991 Science 253: 665–669.

  31. Oshima M, Dinchuk JE, Kargman SL, Oshima H, Hancock B, Kwong E, Traskos M, Evans JF and Taketo MM. . 1996 Cell 87: 803–809.

  32. Pereira MA, Grubbs CJ, Barnes LH, Li H, Olson GR, Eto I, Juliana M, Whitaker LM, Kelloff GJ, Steele VE and Lubet RA. . 1996 Carcinogenesis 17: 1305–1311.

  33. Plummer SM and Faux SP. . 1994 Carcinogenesis 15: 449–453.

  34. Powell SM, Zilz N, Beazer-Barclay Y, Bryan M, Hamilton SR, Thibodeau SN, Vogelstein B and Kinzler KW. . 1992 Nature 359: 235–237.

  35. Rao VV, Simi B and Reddy BS. . 1993 Carcinogenesis 14: 2219–2225.

  36. Rosenberg L, Palmer JR, Zauber AG, Warshauer ME, Stolley PD and Shapiro S. . 1991 J. Natl. Cancer. Inst. 83: 355–358.

  37. Samaha HS, Kellof GJ, Steele V, Rao CV and Reddy BS. . 1997 Cancer Res. 57: 1301–1305.

  38. Sanjaya S and Aggarwal BB. . 1995 J. Biol. Chem. 270: 24995–25000.

  39. Schiffman MH, Andrews AW, Daniels J, Robinson A, Smith L, Nair PP and Wilkins TD. . 1989 Mutat. Res. 222: 351–357.

  40. Schmedtje JF, Yan-Shan J, Lui W-L, Dubois RN and Runge MS. . 1997 J. Biol. Chem. 272: 601–608.

  41. Scudiero DA, Shoemaker RH, Paull KD, Monks A, Tierney S, Nofziger TH, Currens MJ, Seriff D and Boyd MR. . 1988 Cancer Res. 48: 4827–4833.

  42. Srivastava R and Srimal RC. . 1985 Indian J. Med. Res. 81: 215–223.

  43. Subbaramaiah K, Chung WJ, Michaluart P, Telang N, Tanabe T, Inoue H, Jang M, Pezzuto JM and Dannenberg AJ. . 1998 J. Biol. Chem. 273: 21875–21882.

  44. Subbaramaiah K, Zakim D, Weksler BB and Dannenberg AJ. . 1997 Proc. Soc. Exp. Biol. Med. 216: 201–210.

  45. Thun MJ, Nanboordiri MM and Heath CW. . 1991 New Eng. J. Med. 325: 1593–1596.

  46. Tordman C, Coge F, Andre N, Rique H, Spedding M and Bonnet J. . 1995 Biochem. Biophys. Acta 1256: 249–256.

  47. Trujillo MA, Garewal HS and Sampliner RE. . 1994 Digest. Diseases Sci. 39: 2260–2266.

  48. Tsujii M and Dubois RN. . 1995 Cell 83: 493–501.

  49. Vainio H. . 1997 Non-steroidal Anti-inflammatory Drugs, IARC Handbooks of Cancer Prevention Volume 1 Lyon: IARC press.

    Google Scholar 

  50. Vane JR and Botting RM. . 1995 Inflammation Res. 44: 1–10.

  51. Woronicz JD, Gao X, Cao Z, Rothe M and Goeddel DV. . 1997 Science 278: 866–869.

  52. Xie W and Herschman HR. . 1995 J. Biol. Chem. 270: 27622–27628.

  53. Yin MJ, Yamamoto Y and Gaynor RB. . 1998 Nature 396: 77–80

  54. Zarcovic M, Qin X, Nakatsuru Y, Oda H, Nakamura T, Shamsuddin AM and Ishikawa T. . 1993 Carcinogenesis 14: 1261–1264.

  55. Zeng-gang L, Hsu H, Goeddel DV and Karin M. . 1996 Cell 87: 565–576.

Download references


We thank Helen Ball for technical assistance and also Andreas Gescher and Ann Hudson for helpful discussions and comments on the manuscript. SM Plummer benefited from a BACR mid-career fellowship which supported part of this work.

Author information



Corresponding author

Correspondence to Simon M Plummer.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Plummer, S., Holloway, K., Manson, M. et al. Inhibition of cyclo-oxygenase 2 expression in colon cells by the chemopreventive agent curcumin involves inhibition of NF-κB activation via the NIK/IKK signalling complex. Oncogene 18, 6013–6020 (1999).

Download citation


  • colon cancer
  • cyclo-oxygenase 2
  • curcumin, NF-κB

Further reading