Because a role for nuclear factor-κB (NF-κB) has been implicated in the pathogenesis of pancreatic carcinoma, this transcription factor is a potential target for the treatment of this devastating disease. Curcumin (diferuloylmethane) is a phytochemical with potent NF-κB-inhibitory activity. It is pharmacologically safe, but its bioavailability is poor after oral administration.
The authors encapsulated curcumin in a liposomal delivery system that would allow intravenous administration. They studied the in vitro and in vivo effects of this compound on proliferation, apoptosis, signaling, and angiogenesis using human pancreatic carcinoma cells. NF-κB was constitutively active in all human pancreatic carcinoma cell lines evaluated and liposomal curcumin consistently suppressed NF-κB binding (electrophoretic mobility gel shift assay) and decreased the expression of NF-κB-regulated gene products, including cyclooxygenase-2 (immunoblots) and interleukin-8 (enzyme-linked immunoassay), both of which have been implicated in tumor growth/invasiveness. These in vitro changes were associated with concentration and time-dependent antiproliferative activity (3-[4,5-dimethylthiazol-2-yl]2,5-diphenyltetrazolium bromide assay [MTT assay]) and proapoptotic effects (annexin V/propidium iodide staining [fluorescence-activated cell sorting] and polyadenosine-5′-diphosphate-ribose-polymerase cleavage).
The activity of liposomal curcumin was equal to or better than that of free curcumin at equimolar concentrations. In vivo, curcumin suppressed pancreatic carcinoma growth in murine xenograft models and inhibited tumor angiogenesis.
Pancreatic carcinoma is an aggressive, chemotherapy-resistant malignancy. As the 5th leading cause of cancer-related mortality, it remains a major public health problem, accounting for an estimated 30,000 deaths per year in the U.S.1 Surgical resection is the only curative therapy, but to the best of our knowledge virtually no patients survive 5 years. Gemcitabine, approved by the Food and Drug Administration in 1996, is the only chemotherapeutic agent licensed for use in pancreatic carcinoma. However, gemcitabine results in an objective tumor response in approximately 5% of patients and the impact on survival was minor.2 That it is considered the most active agent in this disease, despite such limited efficacy, highlights the desperate need for new therapeutic strategies.
Epidemiologic and animal studies have shown that the microchemicals present in the diet can be effective agents for the prevention of cancer.3–26 Some of these compounds have significant antitumor activity in vitro and in preclinical models. Because of their lack of toxicity, there is an increasing interest in plant-derived chemicals that exhibit such activity. The possibility of using phytochemicals with established chemopreventive activities and preclinical antitumor effects as a novel approach for management of established cancer merits exploration.
Curcumin (diferuloylmethane) is a food chemical present in tumeric (Curcuma longa). It has been found to be pharmacologically safe as indicated by its consumption as a dietary spice for centuries.6 This compound has potent antiproliferative and proapoptotic effects in vitro.18, 22, 23 In murine models, curcumin suppresses carcinogenesis of the skin,9–12 the forestomach,26 the breast,13 the colon,14–16 and the liver17 in mice.
Because curcumin has diverse effects on signaling molecules relevant to cancer, multiple mechanisms of action could account for its antitumor effect. Importantly, curcumin is a potent inhibitor of NF-κB, a transcriptor factor implicated in the pathogenesis of several malignancies including pancreatic carcinoma.27 NF-κB-regulated genes include IκBα, cyclooxygenase-2 (COX-2), and interleukin-8 (IL-8).28, 29 Pancreatic carcinoma cells often express high levels of IL-8 and COX-2, and these molecules appear to play a role in the proliferative and metastatic potential of this neoplasm.30, 31 Therefore, suppression of NF-κB and its downstream effectors may be an innovative strategy for treatment of pancreatic carcinoma.
In the current study, we investigated the in vitro and in vivo antitumor activity of liposomal curcumin against human pancreatic carcinoma cells. Our results demonstrate that all six pancreatic carcinoma cell lines studied expressed constitutively active NF-κB, which was suppressed by liposomal curcumin. NF-κB inhibition was associated with growth suppression, apoptosis, and down-regulation of expression of gene products (IL-8 and COX-2) regulated by NF-κB. In vivo, liposomal curcumin inhibited pancreatic carcinoma growth and demonstrated antiangiogenic effects.
MATERIALS AND METHODS
The human pancreatic carcinoma cell lines BxPC-3, Capan-1, Capan-2, ASPC-1, HS766-T, and MiaPaCa2 were purchased from American Type Culture Collection (Manassas, VA) and cultured using the manufacturer' instructions.
1,2-Dimyristoyl-sn-glycero-3-phosphocholine (DMPC) and (DMPC/DMPG) (DMPG = 1,2-dimyristoyl-sn-glycero-3-[phospho-rac- (1-glycerol)] [sodium salt]) were obtained as dry powder from Avanti Polar Lipids (Alabaster, AL). Curcumin, dimethylsulfoxide (DMSO), acetone, and tert-butanol were obtained from Sigma Chemical Company (St. Louis, MO).
Different total lipid curcumin ratios (weight/weight) ranging from 10:1 to 4:1 were tested before settling on a fixed ratio of 10:1. The lyophilization procedure involved several steps. First, curcumin was dissolved in 50 mg/mL DMSO. The lipid (e.g., DMPC) was dissolved in 20 mg/mL tert-butanol. The 2 solutions were mixed and filtered through a 0.22-μM filter for sterilization. Aliquots of this solution were placed in lyophilization vials. The vials were frozen in a dry ice acetone bath and lyophilized for 24 hours to remove all DMSO and tert-butanol. The vials were stored at −20 °C. The lipid formulation included DMPC or DMPC/DMPG in some experiments.
MTT Cell Proliferation Assay
Proliferation/survival of cells after liposomal curcumin exposure was assessed by the MTT (3-[4,5-dimethylthiazol-2-yl]2,5-diphenyltetrazolium bromide) assay using the manufacturer's instructions (Sigma-Aldrich Company, St. Louis, MO). Cells were incubated for 72 hours. The mean value and standard error for each treatment were determined, then converted to percent relative to control (empty liposomes). The concentration at which cell growth was inhibited by 50% (IC50) was determined by linear interpolation using the formula [(50% − low percentage)/(high percentage − low percentage)] × (high concentration − low concentration) + low concentration. IC90 was similarly determined, with 90% substituting for 50% in the above equation.
Cell Recovery Assay
To determine whether cells treated with liposomal curcumin recovered their proliferative capacity after removal of curcumin, the cell recovery assay was used. Cells were seeded at 5.0 × 105 cells per 100 μm plate and incubated overnight. Cells were treated with various concentrations of liposomal curcumin and incubated for 72 hours. Cells were trypsinized and counted, and the same number of cells for each treatment (3.0–5.0 × 103 cells per well, depending on the cell line) were then seeded in triplicate into 96-well plates in curcumin-free medium and incubated for 72 hours. The degree of cell recovery was determined via MTT assay.
Annexin V/Propidium Iodide 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. Cells were treated with various concentrations of liposomal curcumin or free curcumin in DMSO (the final DMSO concentration was 0.1%) and incubated for 72 hours. The cells were harvested by quick (< 5 minutes) trypsinization to minimize potentially high annexin V background levels in adherent cells. Cells were then washed and stained with fluorescein 5(6)-isothiocyanate (FITC)/annexin V/propidium iodide (PI) as directed by the annexin V-FLUOS staining kit (Roche Diagnostics, Indianapolis, IN). Stained cells were placed on ice and protected from light until read via flow cytometry. Cells were analyzed on an Epics XL-MCL flow cytometer using the System II version 3.0 software (both hardware and software are from Beckman Coulter, Miami, FL), with the laser excitation wavelength set at 488 nanometers (nm). The green signal from FITC/annexin V was measured at 525 nm and the red signal from PI was measured at 620 nm. Cells staining negative for both annexin V and PI are viable. Cells that are annexin V+/PI− are in early apoptosis, whereas cells that are necrotic or in late apoptosis are annexin V+/PI+.
Electrophoretic Mobility Shift Assay for Nuclear Factor-κB
Electrophoretic mobility shift assay (EMSA) was performed using standard procedures and was used to detect NF-κB binding before and after liposomal curcumin treatment. Cells seeded at 5.0 × 105 cells per 100 μm plate and incubated overnight were treated with IC50 concentrations of liposomal curcumin and incubated for 72 hours.
Nuclear extracts were thawed on ice, and protein concentration was determined by the Bio-Rad DC protein assay (Bio-Rad Laboratories, Hercules, CA). For supershifts, 2 μg of rabbit polyclonal anti-p50 or 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 its addition in the binding reaction.
Quantitation of autoradiographs was performed on the Fluorchem 8900 imaging system using AlphaEase FC 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 32P-end-labeled with T4 polynucleotide kinase (New England Biolabs, Beverly, MA), purified with Quick Spin G-50 Sephadex columns (Roche Diagnostics), and stored at −20 °C.
Interleukin-8 Enzyme-Linked Immunoadsorbent Assay
Levels of IL-8 protein in conditioned media before and after liposomal curcumin exposure were determined by enzyme-linked immunoadsorbent assay (ELISA) using the Quantikine human IL-8 immunoassay kit by R & D Systems (Minneapolis, MN). The lower limit of sensitivity of the assay is 10 pg/mL.
Immunoblotting using standard procedures was performed to assess steady-state levels of COX-2 protein and to determine polyadenosine-5′-diphosphate-ribose-polymerase (PARP) cleavage (the latter reflecting apoptosis). A protein assay to measure protein content was performed by using the bicihonimic acid (BCA) protein assay kit (Pierce Endogen, Rockford, IL). The samples were run on a 6–10% sodium dodecyl sulfate gel. The signal was detected by secondary antibody (horseradish peroxidase-conjugated antimouse immunoglobulin [Ig] or antirabbit Ig, as appropriate [1:5000]) (Amersham Pharmacia Biotech, Piscataway, NJ) and the enhanced chemiluminescence detection kit (Amersham Pharmacia Biotech), and then autoradiographed.
Antibodies used for immunoblotting included monoclonal antitubulin antibody (Sigma Chemical Company), anti-PARP rabbit polyclonal (Cell Signaling, Beverly, MA), and anti-COX-2 monoclonal antibody (MoAb) (Cayman Chemical, Ann Arbor, MI). For EMSAs, rabbit polyclonal anti-p50 and anti-p65 (Santa Cruz Biotechnology) were used. Anti-vascular endothelial growth factor (anti-VEGF) MoAb was obtained from Oncogene (Boston, MA) and rat antimouse CD31 from BD Pharmingen (San Diego, CA).
Female athymic nu/nu mice (3–5 weeks old) obtained from Harlan Sprague Dawley (Indianapolis, IN) were maintained 5 per cage in microisolator units. Animals were given a commercial diet and water. Mice were quarantined for ≥ 1 week before experimental manipulation. All animal experiments were performed at The University of Texas M. D. Anderson Cancer Center (MDACC) (Houston, TX) under protocol 10-02-13231 approved by the Institutional Animal Care and Use Committee of The University of Texas MDACC.
A total of 5 × 106 BxPC-3 or MiaPaCa2 cells collected in 100 μL RPMI media in log phase growth were injected subcutaneously on 1 side of the abdomen of 3–5 week-old female nude mice. Once tumor masses became established, animals were randomized to receive intravenous liposomal curcumin (intravenous tail vein; 40 mg/kg body weight, 3 time per week; this is the maximum volume that could be injected), empty liposomes, or saline. Tumor size and body weight were measured with calipers three times a week. The tumor volume was calculated using the following formula: volume = (length × width 2)/2, in which width was the shortest measurement in millimeters.
Immunohistochemistry for Angiogenesis
Immunohistochemistry studies were performed by using formalin-fixed, paraffin embedded sections (5 μm), heat-induced antigen retrieval (Dako Corporation, Carpinteria, CA), and 1:200 monoclonal anti-VEGF antibody (Oncogene) or a 1:50 dilution of rabbit polyclonal anti-IL-8 antibody (Biosource International, Camarillo, CA). The frozen sections were cut from snap-frozen tissue and embedded in OCT compound (Miles, Elkhart, IN). Separate sections were incubated with 1:200 rat antimouse CD31 antibody (BD Pharmingen). The detection system was the LSAB2 detection kit (Dako Corporation). The secondary antibody is a biotinylated antibody (Dako Corporation) and forms a complex with peroxidase-labeled streptavidin. Counterstaining was performed by using Gill's hematoxylin (Sigma Chemical Company). Negative controls were performed by replacing the primary antibody with anti-IgG1 (the same isotype with the primary antibody) (Oncogene).
We used the human pancreatic cell lines ASPC-1, BxPC-3, Capan-1, Capan-2, HS766-T, and MiaPaCa2 to investigate the effect of liposomal curcumin on cell proliferation and apoptosis and on constitutive NF-κB activity and NF-κB-regulated gene expression.
Liposomal Curcumin Inhibits Proliferation/Survival of Pancreatic Carcinoma Cell Lines In Vitro
Exposure to liposomal curcumin (for 72 hours) inhibited pancreatic cell growth in all 6 lines tested in a concentration and time-dependent manner (Fig. 1). Proliferation/survival was assessed by 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 concentrations varied from approximately 2.0 μM for Capan-1 1 to 37.8 μM for Capan-2. The IC50 for free curcumin ranged from 5.4 μM (BxPC-3 and Capan-1 cells) to 46 μM (Capan-2 cells) (Table 1). The antiproliferative effects of liposomal curcumin were equivalent or better than those free of free curcumin at equimolar concentrations in all cell lines.
Table 1. Inhibitory Concentration of Liposomal Curcumina
Pancreatic carcinoma cell line
IC50: concentration at which 50%; growth inhibition occurred; IC90: concentration at which 90% growth inhibition occurred.
Mean of triplicate experiments (3-[4,5-dimethylthiazol-2-yl]2,5-diphenyltetrazolium bromide [MTT] assay after 72 hours of exposure to curcumin).
Growth-Inhibitory Effects of Liposomal Curcumin Are at Least Partially Irreversible
After exposure to liposomal curcumin for 72 hours, pancreatic carcinoma cells were replated in fresh media, and recovery of proliferation/survival was assessed (by MTT assay) after an additional 72 hours. (The concentrations of liposomal curcumin were approximately the IC50 and IC90 for each cell line.) There was a concentration-dependent loss of ability to recover in all six cell lines, although this was not complete in all cell lines. These results suggest that the cells had at least partially undergone irreversible changes such as apoptosis (Fig. 2).
Liposomal Curcumin Induces Apoptosis in Pancreatic Carcinoma Cell Lines
Apoptosis was first assessed by annexin V/PI staining (FACS analysis) after 72 hours of exposure to liposomal 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 allows discrimination of live cells (unstained by either fluorochrome) from apoptotic cells (stained with annexin V). There was a dose-related increase in apoptosis after liposomal curcumin exposure and the effects of liposomal curcumin were equal to or better than those of free curcumin at equimolar concentrations (Fig. 3A).
A hallmark of apoptosis is cleavage of PARP from the native 116 kilodalton (kD) to 85 kD. PARP is an enzyme involved in DNA damage and repair mechanisms. It synthesizes the poly(ADP-ribose) polymer to chromatin and other target proteins. During apoptosis, PARP is cleaved by the protease, caspase-3, an important downstream apoptotic caspase. Western blot analysis using anti-PARP antibody demonstrated PARP cleavage after 72 hours of incubation with IC50 to IC90 levels of liposomal curcumin (Fig. 3B and data not shown).
Liposomal Curcumin Inhibits Activation of the NF-κB Transcription Factor in Pancreatic Carcinoma Cell Lines
NF-κB is a transcription factor that has been implicated in the growth of diverse neoplasms including pancreatic carcinoma.27 The NF-κB pathway regulates numerous downstream oncogenic and growth signals. We have previously demonstrated, with the use of EMSAs, constitutive binding of NF-κB transcription factor in all pancreatic carcinoma cell lines tested.32 Liposomal curcumin inhibited NF-κB activation in a dose-dependent manner in all of these lines (representative data on Fig. 4). At equimolar concentrations, liposomal curcumin was as effective or better than free curcumin at suppressing NF-κB activity (Fig. 4).
Because NF-κB is a family of proteins, various combinations of the Rel/NF-κB protein can constitute an active NF-κB heterodimer that binds to a specific sequence in DNA. To show that the retarded band visualized by EMSA in pancreatic carcinoma cells was indeed NF-κB, we incubated nuclear extracts from pancreatic carcinoma cells with antibody to either the p50 (NF-κB1) or the p65 (RelA) subunit of NF-κB. Both shifted the band to a higher molecular mass (Fig. 4), thus suggesting that the major NF-κB band in pancreatic carcinoma cells consisted of p50 and p65 subunits. A nonspecific minor band was also observed. This was not supershifted by the antibody. Preimmune serum had no effect and excess unlabeled NF-κB (100-fold) caused complete disappearance of the band (data not shown). Use of mutated NF-κB oligonucleotide (instead of the NF-κB oligonucleotide) failed to yield the NF-κB bands (data not shown).
Liposomal Curcumin Decreases the Steady-State Level of NF-κB-Regulated Gene Products (Interleukin-8 and Cyclooxygenase-2)
IL-8 has been implicated in the growth and metastatic potential of pancreatic carcinoma.31 IL-8 expression is regulated by NF-κB. In our current experiments, 72 hours of exposure to IC50 levels of curcumin substantially decreased IL-8 levels (35–90%) as assessed by ELISA in all 6 lines tested (Fig. 5).
COX-2 has been implicated in the growth of diverse tumors including pancreatic carcinoma8, 33–35 and its expression is regulated by NF-κB.29 Steady-state levels of COX-2 protein were assessed by Western blotting of protein extracts (100 μg per lane) derived 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 expressed COX-2, but ASPC-1 and HS766T did not.32 In COX-2–expressing lines, a dose-dependent decrease in COX-2 expression was seen after exposure to liposomal curcumin (Fig. 6). These changes were dose dependent (data not shown).
For both IL-8 and COX-2, the suppressive effect of liposomal curcumin was equal to or greater than that of free curcumin at equimolar concentrations.
Liposomal Curcumin Inhibits Pancreatic Tumor Xenograft Growth in Murine Models
Liposomal curcumin suppressed the growth of both BxPC-3 and MiaPaCa2 tumors in murine models (Fig. 7). Animals received 40 mg/kg of liposomal curcumin intravenously 3 times weekly. This dose was the maximum that could be delivered based on volume injected. The mice demonstrated no overt drug-related toxicity.
Liposomal Curcumin Inhibits Angiogenesis in Vivo
Treatment with liposomal curcumin resulted in reduced tumor size and visible blanching of tumors (Fig. 8A). In addition, expression of CD31 (endothelial cell marker), as well as VEGF and IL-8, was decreased in both BxPC-3 and MiaPaCa2 tumor xenografts (by immunohistochemistry analysis), consistent with an antiangiogenic effect (Fig. 8B and data not shown).
Plant-derived chemicals have been used successfully for a wide spectrum of medicinal purposes. Their salutary effects often derive from a favorable therapeutic to toxicity index. Curcumin (diferuloylmethane) is a component of the root of the plant Curcuma longa. A wealth of data indicates that this compound has potent antitumor effects against a variety of cancer cell lines in vitro, and chemopreventive effects in murine carcinoma models.26, 33–35 Furthermore, curcumin is virtually devoid of side effects in animals as well as in early Phase I trials in humans.36 Free curcumin is, however, highly hydrophobic and cannot be administered systemically. Liposome encapsulation of curcumin makes this agent amenable to intravenous dosing and circumvents the problem of poor oral availability that limits the utility of free curcumin.
Because of the central role of NF-κB in cell survival and proliferation, we explored this transcription factor as a target for liposomal curcumin. During normal homeostasis, NF-κB is present in the cytoplasm as an inactive heterotrimer composed of the p50, p65, and IκBα subunits.37 After activation, IκBα undergoes phosphorylation and ubiquitination-dependent degradation by the proteosome. 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. Considerable data now indicate that NF-κB regulates the expression of various genes that play critical roles in apoptosis, proliferation, and transformation.38 We have previously reported that NF-κB is constitutively active in all the human pancreatic cell lines evaluated.32 Liposomal curcumin decreased NF-κB binding and its effects were as potent as these of free curcumin (Fig. 4). NF-κB inhibition was accompanied by marked in vitro growth suppression and apoptosis (Figs. 1–3).
Several mechanisms could explain why NF-κB down-regulation by liposomal curcumin attenuates proliferation of cancer cells.39–42 First, the constitutive transcription of various angiogenic and tumorigenic chemokines modulated by NF-κB facilitates proliferation and survival of malignant cells.42 In our studies, down-regulation of expression of the gene products COX-2 and IL-8, whose synthesis is known to be regulated by NF-κB, was observed (Figs. 5 and 6). Importantly, in this regard, high levels of COX-2 have been implicated in the growth of several malignancies including pancreatic carcinoma30, 43, 44 and ectopic COX-2 expression suppresses apoptosis.39, 40 The down-regulation of COX-2 by liposomal curcumin is most likely mediated by the attenuation of NF-κB binding, which is needed for COX-2 expression.29 Liposomal curcumin also decreased IL-8 expression (Fig. 6). This is noteworthy because IL-8 is a pleiotropic cytokine that promotes tumor growth and vascularization.9 IL-8 transcription is controlled, at least in part, by NF-κB.31
The inhibitory effects of liposomal curcumin on the NF-κB apparatus were associated with substantial antiproliferative activity against all six of our human pancreatic carcinoma cell lines (Table 1). Consistent failure of the cells to recover after removal of liposomal curcumin from the media suggested a cytotoxic, rather than cytostatic, effect (Fig. 2). Marked programmed cell death was observed after 72 hours of treatment with liposomal curcumin (Fig. 3).
Liposomal curcumin was also active in in vivo models. We treated animals bearing human pancreatic carcinoma xenografts (BxPC-3 and MiaPaCa2) with systemically administered liposomal curcumin. Significant tumor growth inhibition without overt host toxicity was observed (Fig. 7). Tumors from animals treated with liposomal curcumin demonstrated an antiangiogenic effect, including obvious blanching of tumors on visual inspection and attenuation of CD31 (an endothelial marker) and VEGF and IL-8 expression (Fig. 8). These results are consistent with those of Arbiser et al.,45 who demonstrated that curcumin suppresses corneal neovascularization in mice.
Pancreatic carcinoma is a devastating illness, and virtually all patients die, usually within 1 year. This malignancy is highly resistant to chemotherapy. The best agent available, gemcitabine, improved survival by only 6 weeks in a pivotal, randomized trial.2 Centuries of use of curcumin in Far Eastern countries demonstrate that this phytochemical is pharmacologically safe. In addition, in Phase I clinical trials, humans can tolerate ≤ 8 g per day when free curcumin is ingested.36 The bioavailability of oral curcumin is, however, poor.36, 46 A liposome-encapsulated formation of curcumin circumvents this problem by permitting intravenous administration. The results presented in the current study demonstrate that liposomal and free curcumin are equipotent in their ability to suppress NF-κB activity, COX-2, and IL-8 expression, as well as cell proliferation/survival of pancreatic carcinoma cells. In vivo, liposomal curcumin inhibits pancreatic cell growth in murine xenograft models and these effects are accompanied by a potent antiangiogenic response. No overt host toxicity is noted when maximal volumes are administered to mice. Taken together with the dismal outlook for patients with pancreatic carcinoma, our observations suggest that liposomal curcumin should be investigated in the clinical setting.