Nuclear factor-κB and IκB kinase are constitutively active in human pancreatic cells, and their down-regulation by curcumin (diferuloylmethane) is associated with the suppression of proliferation and the induction of apoptosis


  • Lan Li M.S.,

    1. Department of Experimental Therapeutics, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Bharat B. Aggarwal Ph.D.,

    1. Department of Experimental Therapeutics, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Shishir Shishodia Ph.D.,

    1. Department of Experimental Therapeutics, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • James Abbruzzese M.D.,

    1. Department of Gastrointestinal Medical Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Razelle Kurzrock M.D.

    Corresponding author
    1. Department of Experimental Therapeutics, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    2. Department of Gastrointestinal Medical Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    3. Phase 1 Program, Division of Cancer Medicine, The University of Texas M.D. Anderson Cancer Center, Houston, Texas
    • Phase 1 Program, Division of Cancer Medicine, P.O. Box 422, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030
    Search for more papers by this author
    • Fax: (713) 745-2374



Pancreatic carcinoma is a lethal malignancy, with the best available therapeutic option—gemcitabine—yielding response rates of < 10%. Because nuclear factor-κB (NF-κB) has been determined to play a role in cell survival/proliferation in human pancreatic carcinoma, this transcription factor is a potential therapeutic target.


The authors investigated the ability of curcumin (diferuloylmethane), an agent that is pharmacologically safe in humans, to modulate NF-κB activity.


NF-κB and IκB kinase (IKK) were constitutively active in all human pancreatic carcinoma cell lines examined, and curcumin consistently suppressed NF-κB binding (as assessed using an electrophoretic mobility gel-shift assay) and IKK activity. Curcumin decreased the expression of NF-κB–regulated gene products, including cyclooxygenase-2 (as assessed using immunoblot analysis), prostaglandin E2, and interleukin-8 (as assessed using an enzyme-linked immunoassay), all of which have been implicated in the growth and invasiveness of pancreatic carcinoma. These changes were associated with concentration- and time-dependent antiproliferative activity (as assessed using a 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide [MTT] assay) and proapoptotic effects (as assessed via annexin V/propidium iodide staining [fluorescence-activated cell sorting, as well as with the induction of polyadenosine-5′-diphosphate-ribose polymerase cleavage).


Curcumin down-regulated NF-κB and growth control molecules induced by NF-κB in human pancreatic cells. These effects were accompanied by marked growth inhibition and apoptosis. Through these findings, the authors provided a biologic rationale for the treatment of patients with pancreatic carcinoma using this nontoxic phytochemical. Cancer 2004. © 2004 American Cancer Society.

Adenocarcinoma of the pancreas is one of the most devastating neoplastic diseases. As the fifth leading cause of cancer-related mortality, it remains a major public health problem, accounting for an estimated 30,000 deaths per year in the United States.1 Therapeutic options for patients with unresectable, metastatic, or recurrent disease are extremely limited, and almost no patient survives for 5 years with this malignancy. Gemcitabine is the only chemotherapeutic agent licensed for use to treat patients with pancreatic carcinoma. However, gemcitabine results in an objective tumor response rate of < 10%,2 and its impact on survival is minor.2 The finding that it is considered to be the most active agent against this disease, despite such limited efficacy, underscores the need for new approaches.

One of the mechanisms that could drive resistance to apoptosis in pancreatic carcinoma is the activation of nuclear transcription factor-κB (NF-κB).3 During normal homeostasis, NF-κB is present in the cytoplasm as an inactive heterotrimer composed of p50, p65, and IκBα subunits.4 After activation, IκBα undergoes phosphorylation and ubiquitination-dependent degradation by the proteasome. As a result, nuclear localization signals on the p50-p65 heterodimer are exposed, leading to nuclear translocation and binding to a specific consensus sequence found in the promoters of diverse growth-regulatory genes. This binding activates gene transcription. The phosphorylation of IκBα is mediated via the activation of IκB kinase (IKK).4 The IKK complex consists of three proteins: IKKα, IKKβ, and IKKγ/NF-κB essential modulator, with IKKα and IKKβ being the kinases responsible for phosphorylating IκBα. A wealth of data now indicate that NF-κB regulates the expression of various genes that play crucial roles in apoptosis, proliferation, and transformation.5 Among the NF-κB–regulated genes are IκBα, cyclooxygenase-2 (COX-2), and interleukin-8 (IL-8).6, 7 Pancreatic carcinoma cells often express high levels of IL-8 and COX-2, and these molecules have been identified as being responsible for the proliferative and metastatic potential of this neoplasm.8, 9 Therefore, NF-κB may be an important target for pancreatic carcinoma treatment.

Epidemiologic and animal studies have shown that microchemicals present in the diet could be effective agents for the prevention and treatment of malignancy.10–33 Some of these compounds have significant antitumor activity in vitro and in preclinical models. Because of their proven pharmacologic safety, there is an increasing interest in plant-derived chemicals that exhibit such activity. In particular, curcumin (diferuloylmethane) is a nonnutritive food chemical present in tumeric (Curcuma longa). This compound is pharmacologically safe, as is indicated by its consumption as a dietary spice for centuries.13 Curcumin has antiproliferative and proapoptotic effects against diverse tumors in vitro.25, 29, 30 In vivo, it has been found to suppress carcinogenesis of the skin,16–19 the forestomach,33 the breast,20 the colon,21–23 and the liver24 in murine models.

In the current study, we investigated the in vitro antitumor activity of curcumin against human pancreatic carcinoma cell lines. Our results demonstrate that all five pancreatic carcinoma cell lines studied expressed constitutively active NF-κB, which was suppressed by curcumin through inhibition of IKK activity. This was associated with down-regulation of expression of gene products regulated by NF-κB, (e.g., COX-2, prostaglandin E2 [PGE-2] and IL-8) and with antiproliferative effects, as well as with the induction of apoptosis.


Cell Lines

Five human pancreatic carcinoma cell lines (i.e., BxPC-3, Capan-1, Capan-2, ASPC-1, and HS766-T) were purchased from American Type Cell Culture (Manassas, VA). ASPC-1, BxPC-3, and Capan-1 were grown in Roswell Park Memorial Institute (RPMI)-1640 media supplemented with 20%, 10%, and 15% fetal calf serum (FCS), respectively. Capan-2 was grown in McCoy 5A medium supplemented with 10% FCS and HS766-T was grown in Dulbecco modified eagle medium supplemented with 10% FCS. All media were purchased from Invitrogen (Carlsbad, CA).

Electrophoretic Mobility Shift Assay (EMSA) for Nuclear Factor-κB

EMSA was used to assess NF-κB binding before and after curcumin treatment. Cells seeded at 5.0 × 105 cells per 100 μm plate and incubated overnight were treated with 50% inhibitory (IC50) concentrations of curcumin (final concentration in dimethylsulfoxide [DMSO], 0.1%) and incubated for 72 hours. Cells were harvested by trypsinizing and washing twice in cold phosphate-buffered saline (PBS), stored at −80 °C, and used within 2 weeks.

Cells were thawed on ice and induced to swell by adding cold cytoplasmic extraction buffer (CEB: 10 μM 4-[2-hydroxyethyl]-1-piperazineethanesulfonic acid [HEPES], pH 7.9; 10 μM KCl; 0.1 μM ethylenediaminetetraacetic acid [EDTA]; 0.1 μM ethyleneglycotetraacetic acid [EGTA]; 1.0 μM dithiothreitol [DTT]; 0.5 μM phenylmethanesulfonyl fluoride [PMSF]; 2 μg/mL leupeptin; 2 μg/mL aprotinin; and 0.5 mg/mL benzamidine), and this step was followed by a 15-minute incubation on ice. The cytoplasmic cell fraction was lysed by adding 3.125 μL 10% Igepal (Sigma-Aldrich, St. Louis, MO) per 100 μL CEB and vortexing for 20 seconds. The cell suspension was centrifuged for 5 minutes. The cytosolic supernatant was discarded, and cold nuclear extraction buffer (20 μM HEPES, pH 7.9; 400 μM NaCl; 1.0 μM EDTA; 1.0 μM EGTA; 1.0 μM DTT; 0.5 μM PMSF; 2 μg/mL leupeptin; 2 μg/mL aprotinin; and 0.5 mg/mL benzamidine) was added to the pellet. Nuclear suspensions were incubated for 30 minutes on ice (vortexing every 10 minutes) and then centrifuged for 10 minutes. Nuclear extracts from the supernatant were collected and stored at −80 °C and used within 4 days.

Nuclear extracts were thawed on ice, and protein concentration was determined using the Bio-Rad DC protein assay (Bio-Rad Laboratories, Hercules, CA). The binding reaction was initiated by adding 8 μg nuclear extract to binding buffer (100 μM HEPES, pH 7.9; 50 μM EDTA; 100 μM DTT; and 10% glycerol), 2 μg poly(dI:dC) (Amersham Biosciences, Piscataway, NJ), 3.0 × 105 counts per minute 32P-labeled NF-κB double-stranded oligonucleotide, and 1% Igepal (total volume, 20 μL), and incubating the mixture for 15 minutes at 37 °C. The reaction was terminated by adding 4 μL 6X DNA loading dye and placing samples on ice. Samples were loaded on a prerun 5.5% polyacrylamide gel and electrophoresed. The gel was dried and placed on film for autoradiography.

For supershifts, 2 μg rabbit polyclonal anti-p50 or anti-p65 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) was added to the nuclear extract and incubated at room temperature for 5 or 15 minutes before being added to the binding reaction. Quantitation of autoradiographs was performed on the Fluorchem 8900 imaging system using AlphaEaseFC software (Alpha Innotech, San Leandro, CA).


The consensus sequence wild-type and mutant blunt-end double-stranded oligonucleotides used for the NF-κB EMSAs were purchased from Santa Cruz Biotechnology. The NF-κB wild-type sense strand sequence is 5′-AGTTGAGGGGACTTTCCCAGGC-3′, and the mutant sequence is 5′-GTTGAGGCGACTTTCCCAGGC-3′. Oligonucleotides were labeled on the 32P end with T4 polynucleotide kinase (New England Biolabs, Beverly, MA), purified with Quick Spin G-50 Sephadex columns (Roche Diagnostics, Indianapolis, IN), and stored at −20 °C.

IκB Kinase Assay

A kinase assay based on phosphorylation of substrate was used to examine IKK activity. The IKK complex was precipitated from whole-cell extracts with antibodies against IKKα and IKKβ and then treated with 20 μL protein A/G-Sepharose (Pierce, Rockford, IL). After 2 hours, the beads were washed with lysis buffer and then assayed in kinase assay mixture containing 50 μM HEPES (pH 7.4), 20 μM MgCl2, 2 μM DTT, 20 microcuries [γ-32P]adenosine triphosphate (ATP), 10 μM unlabeled ATP, and 2 μg substrate glutathione-S-transferase (GST)-IκBα (i.e., residues 1–54 of IκBα conjugated with GST). After incubation at 30 °C for 30 minutes, the reaction was terminated by boiling with 5 μL 5X sodium dodecyl sulfate (SDS) sample buffer for 5 minutes. Finally, the protein was resolved on a 10% polyacrylamide gel under reducing conditions, the gel was dried, and the radioactive bands were visualized using a PhosphorImager (Alpha Innotech Corp., San Leandro, CA). To determine the total amounts of IKKα and IKKβ in each sample, 30 μg of the whole-cell extract protein was resolved on a 7.5% acrylamide gel and then electrotransferred to a nitrocellulose membrane. The membrane was blocked with 5% nonfat milk protein for 1 hour and then incubated with either anti-IKKα or anti-IKKβ (1:1000 dilution) for 1 hour. The membrane was then washed and treated with horseradish peroxidase–conjugated secondary anti-mouse IγG, antibody, and the proteins were detected by chemiluminescence analysis (Amersham Biosciences).

Interleukin-8 Enzyme-Linked Immunoassay (ELISA)

Levels of IL-8 protein in conditioned media before and after curcumin exposure were determined by ELISA. Cells were seeded and treated with IC50 concentrations of curcumin for 72 hours under the conditions described for NF-μB EMSA. After treatment, media from cells exposed to curcumin or to DMSO only were harvested and centrifuged to remove cellular debris. Aliquots of the supernatant were stored at −80 °C. Samples were thawed, and the Quantikine human IL-8 immunoassay kit (R & D Systems, Minneapolis, MN) was used to determine the IL-8 concentration. A standard curve was generated using recombinant IL-8, and the assay was performed according to the manufacturer's instructions, with the exception that samples were read (on a Molecular Devices [Sunnyvale, CA] plate reader) at 490 nm with wavelength correction at 560 nm. Results were calculated by generating a four-parameter logistic curve fit using SOFTmax Pro Version 2.6 software (Molecular Devices). The lower limit of sensitivity for the assay was 10 pg/mL.


Immunoblotting was performed to determine steady-state levels of COX-2 as well as the IKKα and IKKβ proteins and to assess polyadenosine-5′-diphosphate-ribose polymerase (PARP) cleavage (which is indicative of apoptosis). In brief, cells were washed 3 times with ice-cold PBS and lysed in 1 mL of a solution containing 50 μM HEPES (pH 7.0),150 μM NaCl, 1.5 μM MgCl, 1% Triton, 10 μM sodium pyrophosphate, 1 μM PMSF (Sigma), 10 μg/mL leupeptin (Sigma), and 10 μg/mL aprotinin. Samples were centrifuged (speed, 14,000 rpm) at 4 °C for 5 minutes. The supernatant containing the whole-cell extract was retained for immunoblotting. Protein content was measured using the BCA protein assay kit (Pierce Endogen, Rockford, IL), and samples were run on a 6–10% SDS gel. Fractionated proteins then were transferred to a 0.45 μM nitrocellulose membrane (Bio-Rad Laboratories) in 39 μM glycine, 48 μM Tris, and 1.3 μM SDS at 100 volts for 1 hour. The blot was blocked in Tris-buffered saline plus 0.05% Tween 20 (TBST) (Cayman Chemical, Ann Arbor, MI) and 5% milk or TBST plus 5% bovine serum albumin (Sigma) overnight and then incubated with primary antibody (dilution, 1:2000–5000) for a minimum of 2 hours at room temperature or at 4 °C overnight. The signal was detected using a secondary antibody (horseradish peroxidase–conjugated anti-mouse immunoglobulin [Ig] or anti-rabbit Ig, as appropriate [dilution, 1:5000; Amersham Biosciences]) in conjunction with the ECL detection kit (Amersham Biosciences) and then autoradiographed.


Antibodies used for immunoblotting included the following: anti-tubulin monoclonal antibody (MoAb) (Sigma); anti-PARP rabbit polyclonal antibody (Cell Signaling, Beverly, MA); anti-COX-2 MoAb (Cayman Chemical, Ann Arbor, MI); and anti-IKKα and anti-IKKβ MoAbs (Imagenex, San Diego, CA). For EMSAs, rabbit polyclonal antibodies against p50 and p65 (Santa Cruz Biotechnology) were used.

PGE-2 Assay

PGE-2 levels in conditioned media were assessed according to the manufacturer's instructions using an enzyme immunoassay (Cayman Chemical). A standard curve was created using purified PGE-2 supplied by the manufacturer. The lower limit of sensitivity for the assay was 15 pg/mL.

3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) Cell Proliferation Assay

Proliferation/survival of cells after curcuin exposure was assessed using the MTT assay. Cells were seeded at a concentration of 5.0 × 103 cells per well in a 96-well plate and incubated overnight. Next, cells were treated in triplicate with various concentrations of curcumin (Sigma-Aldrich) dissolved in DMSO, with this solution being added to the appropriate growth media to yield a final DMSO concentration of 0.1%. Cells were incubated for 72 hours. The medium then was aspirated from the wells, cells were rinsed with PBS, and 200 μL MTT (Sigma-Aldrich) solution (final concentration, 0.5 mg/mL) was added. (The MTT solution was prepared by adding 1 part of a stock solution of MTT [5 mg/mL] dissolved in PBS to 9 parts serum-supplemented media.) Cells were incubated for 5 hours, after which the medium was again aspirated and the precipitated formazan dissolved by adding 200 μL DMSO and placing the resulting mixture on a shaker for 10–20 minutes. Samples were read at 560 nm (with subtraction of the reference absorbance at 650 nm) on a plate reader (Molecular Devices). The mean and standard error for each treatment were determined and then converted to percentages relative to the control absorbance (i.e., the absorbance of a sample containing 0.1% DMSO only). The concentration at which cell growth was inhibited by 50% (IC50) was determined by linear interpolation, using the formula IC50 = ([50% − low percentage]/[high percentagelow percentage]) × (high concentrationlow concentration) + low concentration. IC90 was determined in a similar fashion, with ‘90%’ substituted for ‘50%’ in the preceding formula.

Cell Recovery Assay

To determine whether cells treated with curcumin recovered their proliferative capacity after the removal of curcumin, the cell recovery assay was performed. Cells were seeded at a concentration of 5.0 × 105 cells per 100 μm plate and incubated overnight. These cells subsequently were treated with various concentrations of curcumin dissolved in DMSO (final DMSO concentration in growth media, 0.1%); incubation lasted for 72 hours, after which the cells were trypsinized and counted. Next, for each treatment, identical numbers of cells (3.0–5.0 × 103 cells per well, depending on the cell line) were seeded in triplicate on 96-well plates in curcumin-free medium and incubated for 72 hours. The degree of cell recovery was assessed using the MTT assay.

Annexin V-Propidium Iodide (PI) Staining for Apoptotic Cells

Cells were seeded at 2.5–5.0 × 105 cells per 100 μm plate, depending on the cell line, and incubated overnight. Next, cells were treated with various concentrations of curcumin in DMSO (final DMSO concentration in growth media, 0.1%) and incubated for 72 hours. These cells then were harvested by rapid trypsinization (i.e., lasting < 5 minutes) to minimize potentially high background annexin V levels in adherent cells. Cells subsequently were washed and stained with fluorescein isothiocyanate (FITC)–annexin V–PI according to the protocol outlined in the Annexin V–FLUOS staining kit (Roche Diagnostics). Stained cells were placed on ice and protected from light until they were evaluated on flow cytometry, which was performed using an Epics XL-MCL flow cytometer in conjunction with System II Version 3.0 software (hardware and software supplied by Beckman Coulter, Miami, FL). A laser excitation wavelength of 488 nm was used for flow cytometric analysis. The green signal from FITC–annexin V was measured at 525 nm, and the red signal from PI was measured at 620 nm. Viable cells are those that stain negatively for both annexin V and PI. Cells in early apoptosis are positive for annexin V and negative for PI, whereas those that are necrotic or in late apoptosis are positive for both annexin V and PI.


We used the human pancreatic cell lines ASPC-1, BxPC-3, Capan-1, Capan-2, and HS766-T to investigate the effect of curcumin on constitutive NF-κB and IKK activity and on NF-κB–regulated gene expression, cell proliferation, and apoptosis.

High Levels of Constitutive Nuclear Factor-κB Activation in Pancreatic Carcinoma Cell Lines

NF-κB has been implicated in the growth of diverse neoplasms, including pancreatic carcinoma.3 The NF-κB pathway regulates numerous downstream oncogenic and growth-related signals. EMSAs revealed constitutive binding of NF-κB transcription factor in all five pancreatic carcinoma cell lines tested (Fig. 1). Levels of binding were highest in ASPC-1, Capan-1, and Capan-2 cells.

Figure 1.

Constitutive nuclear factor-κB (NF-κB) binding as demonstrated by electrophoretic mobility shift assay (EMSA) in pancreatic carcinoma cell lines. NF-κB is constitutively activated in all five pancreatic cell lines. (A) EMSA with NF-κB probe. (B) Mutant probe. The same amount of protein (8 μg) was used in each lane.

Inhibition of Activation of the Nuclear Factor-κB Transcription Factor in Pancreatic Carcinoma Cell Lines by Curcumin

In all five cell lines tested, EMSA revealed that curcumin inhibited NF-κB activation in a dose-dependent manner (Fig. 2). The designation ‘NF-κB’ encompasses a family of proteins; thus, various combinations of Rel/NF-κB proteins can constitute an active NF-κB heterodimer that binds to a specific sequence in DNA. To show that the inhibited band visualized on EMSA in pancreatic carcinoma cells was indeed NF-κB, we incubated nuclear extracts from pancreatic carcinoma cells with antibodies against either the p50 (NF-κB1) or the p65 (RelA) subunit of NF-κB. In both cases, the band in question shifted to a higher molecular mass (Fig. 2), suggesting that the major NF-κB band in pancreatic carcinoma cells consisted of p50 and p65 subunits. A nonspecific minor band also was observed but was not supershifted by either antibody. Treatment with preimmune serum had no effect on the band, whereas treatment with unlabeled NF-κB in 100-fold excess caused it to disappear completely (data not shown). When the mutated NF-κB oligonucleotide was used, the expected NF-κB bands (p50/p65) were not observed (Fig. 1, right hand panel).

Figure 2.

Effect of curcumin on nuclear factor-κB (NF-κB) binding as assessed by electrophoretic mobility shift assay. Curcumin inhibited NF-κB binding in all five pancreatic cell lines. Cells were exposed to a 50% inhibitory concentration (IC50) or a 75% inhibitory concentration (IC75) of curcumin for 72 hours. The same amount of protein (8 μg) was used in each lane. Supershift assays using anti-p50 and anti-p65 antibodies confirmed that the band shown contained the p50 and p65 subunits of NF-κB.

Constitutive Activity of IκB Kinase in Pancreatic Carcinoma and Suppression of Activity by Curcumin

Phosphorylation is necessary for the degradation of IκBα and the subsequent release of NF-κB. Phosphorylation of IκBα is mediated through IKK. An in vitro kinase assay using immunoprecipitated IKK from untreated pancreatic carcinoma cells and GST-IκBα as a substrate revealed constitutive IKK activity in all pancreatic carcinoma cell lines investigated, whereas under similar conditions, an assay using immunoprecipitated IKK from curcumin-treated cells revealed decreased kinase activity (Fig. 3, top row). Immunoblot analysis of cell extracts from untreated and curcumin-treated cells revealed no significant changes in protein levels of the IKK subunits IKKα and IKKβ in treated cells (Fig. 3, middle and bottom rows).

Figure 3.

Effect of curcumin on IκB kinase (IKK) activity as assessed by the IKK activation assay. (Top row) IKK was immunoprecipitated and the kinase assay performed to assess IKK activity; IKK activity was suppressed in all cell lines after exposure to 50% inhibitory concentrations of curcumin. (Middle and bottom rows) Western blotting was performed to evaluate total IKKα and IKKβ levels; steady-state levels of IKKα and IKKβ remained unchanged after curcumin exposure. DMSO: dimethylsulfoxide; GST: glutathione-S-transferase.

Reduction of Steady-State Nuclear Factor-κB–Regulated Gene Product (Interleukin-8 and Cyclooxygenase-2) Levels by Curcumin

The expression of IL-8, which plays a role in the growth and metastatic potential of pancreatic carcinoma,9 is regulated by NFκB. In the current study, 72 hours of exposure to IC50 levels of curcumin substantially decreased IL-8 levels (29–79%) as assessed by enzyme-linked immunosorbent assays in all 5 cell lines tested (Fig. 4).

Figure 4.

Effect of curcumin on interleukin-8 (IL-8) levels as assessed by enzyme-linked immunosorbent assay (ELISA). Curcumin exposure substantially reduced IL-8 levels in conditional media for all five pancreatic carcinoma cell lines. Cells were exposed to 50% inhibitory concentrations of curcumin for 72 hours. The decrease in IL-8 levels ranged from 29% to 79%. Mean values ± standard errors (SE) from triplicate experiments are shown. (IL-8 levels in control samples: ASPC-1, 16,525 pg/mL; BxPC-3, 2666 pg/mL; Capan-1, 21,649 pg/mL; Capan-2, 7329 pg/mL; and HS766-T, 473 pg/mL.)

COX-2 plays a role in the growth of diverse tumors, including pancreatic carcinoma,8, 34, 35 and its expression is regulated by NF-κB.7 Steady-state levels of COX-2 protein were assessed via Western blotting of protein extracts (100 μg per lane) obtained from pancreatic carcinoma cell lines both before and after exposure to IC90 levels of curcumin for 72 hours. BxPC-3, Capan-1, and Capan-2 cells expressed COX-2, whereas ASPC-1 and HS766-T did not (Fig. 5A). In cell lines expressing COX-2, dose-dependent decreases in COX-2 expression were documented after exposure to curcumin (Fig. 5B).

Figure 5.

Effect of curcumin on cyclooxygenase-2 (COX-2) levels as assessed by Western blot. (A) Steady-state levels of COX-2 protein were compared in five pancreatic carcinoma cell lines. BxPC-3, Capan-1, and Capan-2 cells were found to express COX-2. (B) Steady-state levels were reduced after 72 hours of exposure to 90% inhibitory (IC90) levels of curcumin. Tubulin levels remained unchanged.

Reduction of PGE-2 Levels in Pancreatic Carcinoma Cell Lines by Curcumin

COX-2 synthesizes PGE-2, a molecule that plays a role in promoting the metastatic process. Therefore, the effect of curcumin-induced down-regulation of COX-2 on PGE-2 levels warrants further investigation. Control levels of PGE-2 produced by the cell lines studied were undetectable (< 15 pg/mL for ASPC-1 cells, 485 pg/mL for BxPC-3 cells, 1023 pg/mL for Capan-1 cells, 530 pg/mL for Capan-2 cells, and 40 pg/mL for HS766-T cells). Hence, levels were significantly higher in cell lines with high expression of COX-2 (BxPC-3, Capan-1, and Capan-2; Fig. 5A). In addition, exposure to IC50 levels of curcumin significantly decreased PGE-2 expression in these lines (Fig. 6) in a concentration-dependent manner (data not shown).

Figure 6.

Effect of curcumin on prostaglandin E2 (PGE-2) levels as assessed by enzyme immunoassay: Curcumin exposure significantly reduced PGE-2 levels. PGE-2 concentrations in conditioned media for COX-2–expressing cells were assessed after 72 hours of growth in the presence (at 50% inhibitory [IC50] levels) and absence of curcumin. (PGE-2 levels in control samples: BxPC-3, 485 pg/mL; Capan-1, 1023 pg/mL; Capan-2, 530 pg/mL.) Black bars: control samples; gray bars: dimethylsulfoxide (0.1%); white bars: curcumin at IC50 (72-hour incubation).

Inhibition of Proliferation/Survival of Pancreatic Carcinoma Cell Lines In Vitro by Curcumin

Exposure to curcumin (duration, 72 hours) inhibited pancreatic cell growth in all 5 lines tested in a concentration-dependent and time-dependent manner (Fig. 7). Control samples were exposed to 0.1% DMSO (because curcumin is dissolved in 0.1% DMSO in experimental samples). Proliferation and survival were assessed using the MTT assay. (MTT is a pale yellow substrate that is cleaved by living cells to yield a dark blue formazan product. This colorimetric change reflects active cell proliferation/survival.) IC50 values varied from approximately 5 μM for BxPC-3 and Capan-1 cells to 46 μM for Capan-2, and IC90 values ranged from 6.75 μM for Capan-1 cells to 94.5 μM for Capan-2 cells (Table 1).

Figure 7.

Effect of curcumin on proliferation/survival as assessed by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. Curcumin inhibited the growth of all five pancreatic carcinoma cell lines tested. Controls were exposed to 0.1% dimethylsulfoxide (DMSO), because curcumin was dissolved in 0.1% DMSO for the experimental samples. Mean values ± standard errors from triplicate experiments after 72 hours of exposure to curcumin are shown. (A) Dose-dependent inhibition. Closed circles: Capan-2; ×: BxPC-3; closed triangles: HS766-T; open circles: ASPC-1; open triangles: Capan-1. (B) Time-dependent inhibition of cells exposed to 90% inhibitory (IC90) levels of curcumin (as identified in [A]). Black bars: control samples; white bars: samples exposed to IC90 levels of curcumin for 24 hours; striped bars: samples exposed to IC90 levels of curcumin for 48 hours; speckled bars: samples exposed to IC90 levels of curcumin for 72 hours.

Table 1. Inhibitory Concentrations of Curcumina
 Cell line
  • IC50: concentration at which 50% growth inhibition was observed; IC90: concentration at which 90% growth inhibition was observed.

  • a

    Table entries represent mean values from quadruplicate assays.

IC50 (μM)115.45.4467.0
IC90 (μM)2713.56.7594.521.6

Irreversibility of the Growth-Inhibitory Effects of Curcumin

After exposure to curcumin for 72 hours, pancreatic carcinoma cells were replated in fresh media, and recovery of proliferation/survival was assessed (using the MTT assay) after an additional 72 hours. (Curcumin concentrations were approximately equal to the IC50 and IC90 values for each cell line.) There was a concentration-dependent loss of ability to recover in all five cell lines, suggesting that the cells had undergone irreversible changes (e.g., apoptosis) (Fig. 8).

Figure 8.

Failure to recover proliferation/survival (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide [MTT] assay) after curcumin exposure. (A) BxPC-3 cells exposed to 90% inhibitory (IC90) levels of curcumin. (B) Capan-1 cells exposed to IC90 levels of curcumin. (C) Capan-2 cells exposed to 50% inhibitory (IC50) and IC90 levels of curcumin. The antiproliferative effects of curcumin on pancreatic cells were irreversible. In this experiment, 5 × 103 cells were plated at Time 0. After 72 hours of growth in the presence or absence of curcumin, 5 × 103 cells were replated in fresh media. At 72 hours, the MTT assay was performed. Curcumin-treated cells failed to recover. Mean values ± standard errors from triplicate experiments are shown. Black bars: 72 hours of treatment; striped bars: 72 hours of treatment and 72 hours of recovery.

Induction of Apoptosis in Pancreatic Carcinoma Cell Lines by Curcumin

Apoptosis was first assessed by annexin V/PI staining (FACS analysis) after 72 hours of exposure to curcumin. Annexin V binds to cells that express phosphatidylserine on the outer layer of the cell membrane, a characteristic feature of cells entering apoptosis. This technique allows discrimination of live cells (unstained by either fluorochrome) from apoptotic cells (stained with annexin V). As shown in Figure 9A, there was a dose-related increase in apoptosis after curcumin exposure. At IC90, curcumin induced 73–95% apoptosis in all 5 pancreatic carcinoma cell lines after 72 hours. Apoptosis also was correlated with duration of exposure to curcumin, with early apoptosis appearing at 24 hours and late apoptosis/necrosis appearing by 72 hours (data not shown).

Figure 9.

Effect of curcumin on apoptosis. Curcumin induced apoptosis in pancreatic carcinoma cells. Programmed cell death was assessed by annexin V/propidium iodide staining (fluorescence-activated cell sorting analysis) after 72 hours of exposure to curcumin. (A) Mean percentages of apoptotic cells ± standard errors from triplicate experiments are shown. (B) Apoptosis also was assessed by Western blotting for polyadenosine-5′-diphosphate-ribose-polymerase cleavage after 24 or 72 hours of exposure to curcumin (50% inhibitory [IC50] or 90% inhibitory [IC90] levels). Blots of HS766-T and Capan-2 cells were washed and then exposed to anti-tubulin antibodies to verify that protein loading was equal across all lanes. kD: kilodalton; DMSO: dimethylsulfoxide.

A hallmark of apoptosis is cleavage of the native 116-kilodalton (kD) form of PARP to an 85 kD form of this molecule. PARP is an enzyme involved in DNA damage and repair mechanisms. It synthesizes the poly(adenosine diphosphate–ribose) polymer onto chromatin and other target proteins. During apoptosis, PARP is cleaved by the protease caspase-3, an important downstream apoptotic caspase. Western blotting using anti-PARP antibodies demonstrated progressive concentration-dependent and time-dependent PARP cleavage after 24–72 hours of incubation with IC50 to IC90 levels of curcumin (Fig. 9B)


There has been a recent surge in interest in phytochemicals as medicinal anticancer agents, due to the favorable efficacy and toxicity profiles of these agents. Curcumin (diferuloylmethane) is derived from the root of the plant Curcuma longa. For centuries, it has been consumed in the diet and used as an herbal medicine in several Far Eastern countries. Furthermore, a large body of data indicates that curcumin has potent antitumor effects against a variety of malignant cell lines in vitro and chemopreventive effects in murine carcinoma models.33, 36–38 Finally, this compound has demonstrated little in the way of toxicity in animals as well as in preliminary Phase I trials in humans.39

Because of the central role of NF-κB in cell survival and proliferation, we explored this transcription factor as a target for curcumin. Our results indicate that NFκB is constitutively active in all five human pancreatic cell lines examined (Fig. 1) and that curcumin inhibited NF-κB binding (Fig. 2). Suppression of NF-κB appeared to be mediated by curcumin-induced attenuation of IKK, which had high constitutive activity in all pancreatic carcinoma cell lines investigated (Fig. 3); IKK is known to be responsible for NF-κB activation. Our results are in agreement with previous reports of inhibition of IKK by curcumin in colon carcinoma cells and in multiple myeloma.7, 40

Several potential mechanisms could explain why NF-κB down-regulation by curcumin abrogates the survival of malignant cells.41–44 Indeed, the constitutive transcription of numerous angiogenic and tumorigenic chemokines modulated by NF-κB facilitate the proliferation and survival of malignant cells.44 In the current study, down-regulation of expression of the gene products COX-2 and IL-8, whose syntheses are known to be regulated by NF-κB, was observed (Figs. 4, 5). These observations are noteworthy for several reasons. In particular, high levels of COX-2 have been associated with a variety of malignancies, including pancreatic carcinoma,8, 34, 35 and ectopic COX-2 expression has been demonstrated to suppress apoptosis and promote tumor cell proliferation.41, 42 It is likely that the down-regulation of COX-2 by curcumin is mediated by the inhibition of NF-κB binding, which is required for COX-2 expression.7

PGE-2 expression also was suppressed by curcumin (Fig 6) and was correlated with the down-regulation of COX-2 (Fig. 5B). These results are consistent with the known metabolism of the arachidonic acid pathway, as COX-2 is the enzyme that directs PGE-2 synthesis. PGE-2 inhibits apoptosis by stimulating Bcl-2. It also enhances the proliferative, migratory, and invasive activity of malignant cells and transactivates the epidermal growth factor receptor (EGFR).45, 46 EGFR and Bcl-2 are involved in growth and escape from apoptosis, respectively, in pancreatic carcinoma.32, 47, 48 Hence, interference with the NFκB/COX-2/PGE-2 pathway may play a key role in the antiproliferative effects of curcumin.

Curcumin also substantially decreased IL-8 levels (Fig. 4). This finding is important, because IL-8 is a pleiotropic chemokine that previously has been shown to play a role in the promotion of malignant cell proliferation as well as in angiogenesis.9 Furthermore, serum IL-8 levels are high in patients with pancreatic carcinoma and are correlated with poor outcome.43 IL-8 expression is regulated in part by NF-κB,9 and therefore, suppression of NF-κB activation by curcumin may underlie the observed decreases in IL-8 levels.

The suppressive effects of curcumin on NF-κB and on NF-κB-related gene products were accompanied by significant antiproliferative activity in all 5 of the human pancreatic carcinoma cell lines that were examined, with 3 of these lines having 50% inhibitory concentrations (IC50) that were ≤ 10 μM (MTT assay; Fig. 7; Table 1). These antiproliferative effects were irreversible, as was demonstrated by the consistent failure of these cells to recover after removal of curcumin from the growth media (Fig. 8). Significant apoptosis (73–95% annexin V/PI staining) was evident after 72 hours of curcumin exposure in all cell lines (Fig. 9A). PARP cleavage also was observed and confirmed the occurrence of programmed cell death (Fig. 9B).

Pancreatic carcinoma is an aggressive and incurable malignancy that is almost uniformly resistant to chemotherapy. The mainstay of therapy for pancreatic carcinoma is gemcitabine, an agent with a response rate of < 10% and an associated median survival of approximately 6 months. Nonspecific toxicity, which outweighs or diminishes salutary effects, is a major problem in drug development. Numerous studies have shown that curcumin is pharmacologically safe. Furthermore, it was demonstrated recently (in Phase I clinical trials) that humans can tolerate ≤ 8 g daily when curcumin is administered orally.39 The results presented in the current study indicate that curcumin can suppress NF-κB and IKK activity; COX-2, PGE-2, and IL-8 expression; and cell proliferation in pancreatic malignancies, and all of these curcumin-induced changes are associated with potent proapoptotic effects. Taken together with the unfavorable outlook for patients with pancreatic carcinoma, our observations suggest that curcumin warrants investigation in the clinical setting.