Curcumin induces proapoptotic effects against human melanoma cells and modulates the cellular response to immunotherapeutic cytokines

Curcumin [1,7-bis(4-hydroxy-3methoxyphenyl)-1,6-heptadien-3,5-dione] is the primary bioactive component isolated from turmeric, a dietary spice made from the rhizome of Curcuma longa (1). Curcumin has chemopreventative and chemotherapeutic properties in numerous experimental models and has been shown to inhibit a variety of cellular targets that promote survival and metastasis, including nuclear factor κ-B (NF-κB) and signal transducer and activator of transcription-3 (STAT3), among others (1–3). These data suggest that curcumin may be an effective means to modulate pathways that contribute to chronic inflammation leading to cancer. Despite its broad target specificity, curcumin has a favorable safety profile in humans (no dose-limiting toxicity up to 12 g/d; refs. 4, 5), although dietary curcumin has poor bioavailability (6). Nonetheless, curcumin has shown antitumor effects in vivo when administered via dietary supplement, oral gavage, or injection, even at the low concentrations achieved in these physiologic systems (7–14). Although our understanding of how curcumin can elicit antitumor activity is incomplete, this agent is being investigated in human oncology trials and its use by cancer patients as a form of alternative medicine will likely continue or even increase as anecdotal reports of its anticancer potential become widespread.

One clinical area in which effective chemopreventative agents could make a tremendous impact is in the adjuvant setting of melanoma. Here patients have been rendered disease free following surgical excision of a high-risk lesion (>4 mm in depth, or with lymph node involvement). These patients are at a high risk of local recurrence or even metastatic disease. The only Food and Drug Administration–approved adjuvant therapy is the cytokine IFN-α. The use of this agent is somewhat controversial due to its toxicity profile, but a recent meta-analysis showed a significant benefit of IFN-α in relation to both relapse-free and overall survival in the adjuvant setting of melanoma (15, 16). Cytokine immunotherapy with single agent, high-dose IFN-α and interleukin-2 (IL-2) are also mainstays of treatment in the setting of metastatic disease where they can mediate complete clinical responses in approximately 10% of patients (17, 18). The mechanisms by which IFN-α or IL-2 mediates antitumor activity involve stimulation of host immune effector cells (18–21), although IFN-α can also exert direct antiproliferative, proapoptotic, and antiangiogenic effects on melanoma cells (22–26).

In addition to the clinical use of cytokines, endogenously produced type I interferons (IFN-α, IFN-β) and type II interferons (IFN-γ) are critical components of tumor immunosurveillance (27–29). In a similar manner, IL-2 has been shown to play an important role in survival, proliferation, and effector function of natural killer (NK) and T cells (30–32). The adjuvant setting of melanoma and the presence of micrometastatic disease represent scenarios in which effective tumor immunosurveillance could be of great clinical benefit. Based on these data, identifying chemopreventative agents that might complement the cellular actions of interferons and interleukins could have the greatest potential for clinical success.

In the present report, we show that curcumin can induce apoptosis and down-regulate proteins involved in the survival of human melanoma cells. However, pretreatment with curcumin inhibited IFN-α– and IFN-γ–induced signal transduction and downstream gene expression in human melanoma cell lines. Curcumin also significantly inhibited the response of peripheral blood mononuclear cells (PBMC) from healthy donors to key cytokines involved in antitumor immunity (IFN-α, IFN-γ, IL-2) at the level of signal transduction and gene expression. Finally, curcumin adversely modulated the function of NK cells to secrete IFN-γ in response to IL-12, and inhibited cytotoxicity of A375 human melanoma cells and NK-sensitive K562 target cells in vitro. Although the ability of curcumin to inhibit the response to cytokines is desirable for the prevention of cancers associated with chronic inflammation, the present data caution that it may negate the cellular response to immunotherapeutic cytokines that mediate cancer immunosurveillance.

A375 and Hs294T human metastatic melanoma cell lines and K562 chronic myelogenous leukemia cells were purchased from the American Type Culture Collection. The 1106 MEL and F01 human metastatic melanoma cell lines were a kind gift from Dr. Soldano Ferrone (University of Pittsburgh, Pittsburgh, PA) and were cultured as previously described (23). The radial growth phase WM 1552c and vertical growth phase WM 793b human melanoma cell lines were provided by Dr. M. Herlyn (Wistar Institute, Philadelphia, PA) and cultured as previously described (33). Recombinant human IFN-α2b was purchased from Schering-Plough, Inc. Recombinant human IFN-γ and IL-2 were purchased from R & D Systems, Inc. Recombinant human IL-12 was provided by Genetics Institute.

PBMCs were isolated from source leukocytes of healthy donors (American Red Cross) via density gradient centrifugation using Ficoll-Paque (Amersham, Pharmacia Biotech) as previously described (34). Where indicated, NK cells were enriched from source leukocytes by negative selection with Rosette Sep reagents (Stem Cell Technologies, Inc.).

Curcumin was prepared via condensation of pentane-2,4-dione and vanillin according to the procedure of Venkateswarlu (35). Recrystallization of the crude precipitate from ethanol provided curcumin as orange needles, mp: 181°C to 183°C (182-184°C; ref. 36). The spectral data for the synthetic curcumin (¹H NMR and ¹³C NMR) were also identical to that reported for the natural product (36). This approach routinely resulted in curcumin that was of >95% purity as determined by nuclear magnetic resonance spectroscopy, which is equal or superior to most commercially available samples (data not shown).

Lysates were prepared from melanoma cell lines or PBMCs in Laemelli buffer and assayed for protein expression by immunoblot analysis as previously described with antibodies to STAT1 (BD Biosciences), pSTAT1, STAT3, pSTAT3, poly ADP ribose polymerase (PARP), Bcl-2, cyclin D1, caspase-3 (Cell Signaling Technology), or β-actin (Sigma; ref. 23). Following incubation with the appropriate horseradish peroxidase–conjugated secondary antibody, immune complexes were detected using the SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific).

Phosphatidyl serine exposure was assessed in tumor cells by flow cytometry using allophycocyanin-labeled Annexin V (BD Pharmingen) as previously described (37). Propidium iodide exclusion was not used concurrently in these experiments due to the observation that curcumin produces considerable spectral overlap at wavelengths used to detect this viability exclusion dye. Each analysis was done utilizing at least 10,000 events.

Real time PCR was used to evaluate the expression of the IFN-α–stimulated genes (IFIT2, ISG-15), IFN-γ–stimulated gene (IRF1), and IL-2–stimulated gene [cytokine-induced SH2-containing protein (CIS)] in melanoma cell cultures or PBMCs as previously described (38, 39). Following cytokine treatment, total RNA was isolated using via TRIzol extraction (Invitrogen), quantitated, and converted to cDNA using random hexamers (Invitrogen) as primers for first-strand synthesis. Two microliters of the resulting cDNA were used as a template to measure mRNA transcripts by real time reverse transcription-PCR (RT-PCR) with predesigned primer/probe sets (Assays On Demand, Applied Biosystems) and 2X TaqMan Universal PCR Master Mix (Applied Biosystems) per the manufacturer's recommendations. Predesigned primer/probe sets for 18 s rRNA (Applied Biosystems) were used to normalize expression values (housekeeping gene) and data were acquired and analyzed using the ABI Prism 7900 HT Sequence Detection System (Applied Biosystems).

Culture supernatants from normal human NK cells were assayed for IFN-γ by ELISA using commercially available monoclonal antibody (mAb) pairs as previously described (40). Standard curves were prepared for each plate, and all plasma samples from a single donor were run in duplicate and in parallel to minimize interassay variability.

To measure IFN-γ and granzyme B (GrB) secretion, ELISPOT experiments were conducted using MultiScreen 96-well plates (Millipore) as described previously (41). The plates were coated overnight at 4°C with 50 μL/well of antihuman GrB (Mabtech) or antihuman IFN-γ capture antibody (Mabtech) diluted to 10 μg/mL in PBS. After incubation, coated plates were blocked with 1% fetal bovine serum in PBS for 1 h at room temperature, then washed four times with media. Freshly isolated NK cells (effectors) were incubated overnight in media with either 20 μmol/L curcumin or DMSO (vehicle). Effector cells were then coincubated in triplicate with A375 melanoma cells as targets, or K562 (positive control target; American Type Culture Collection) at an effector:target rato of 20:1 for 4 h. Targets and effectors cultured alone were used as negative controls. After incubation, the plates were washed four times with 100 μL/well of PBS/0.05% Tween-20, and 100 μL of biotinylated monoclonal antihuman GrB or IFN-γ–detecting antibody (Mabtech) diluted to 1 μg/mL were added. After a 3-h incubation at room temperature, the plates were washed four times with PBS-Tween, and 100 μL/well of strepavidin-alkaline phosphatase (Mabtech) diluted to 1:1,000 with PBS were added for 2 h at room temperature. After the incubation, the plates were washed again with PBS-Tween, and 150 μL of BCIP/NBT phosphatase substrate were added to each well. Plates were developed for 5 min at room temperature in the dark and the reaction was stopped by rinsing plates with distilled water. The membranes were air-dried and spots were visualized and counted using the ImmunoSpot Imaging Analyzer system (Cellular Technology Ltd.).

Mixed effects regression models were used to analyze the results of the PBMC and ELISPOT experiments, including a random effect for donor and treatment. For the PBMC experiments, fold changes in gene expression were log-transformed to meet the model assumptions of normality. For ELISPOT experiments, data were obtained in at least three independent experiments with different NK cell donors. P values ≤0.05 were considered to be statistically significant.

Treatment of multiple human metastatic melanoma cell lines (A375, FO1, Hs294T, 1106 MEL) with curcumin for 48 hours induced a dose-dependent increase in Annexin V–positive cells that was most prominent at doses >10 μmol/L (Fig. 1A). Immunoblot analysis was used to confirm curcumin-induced apoptosis in representative lines by measuring the processing of caspase-3 and cleavage of PARP (Fig. 1B). Consistent with the documented ability of curcumin to inhibit the STAT3 and NF-κB signal transduction pathways (1, 42, 43), a reduced level of basal phosphorylated STAT3 and Bcl-2 proteins was observed in melanoma cells following curcumin treatment (Fig. 1C). Individual melanoma cell lines displayed differential curcumin sensitivity at the lower 10 μmol/L dose, but apoptosis was relatively uniform across all cell lines (∼50%) after a 48-hour treatment with higher doses of curcumin (30 μmol/L; Fig. 1A). Curcumin also induced apoptosis at comparable concentrations in both the WM 1552c radial growth phase (41% apoptosis at 20 μmol/L curcumin) and WM 793b vertical growth phase (36% apoptosis at 20 μmol/L curcumin) human melanoma cell lines.

IFN-α and IFN-γ are clinically relevant cytokines that can induce direct proapoptotic effects against melanoma cells (26, 44). To determine whether curcumin can modulate the cellular response to these cytokines, representative human melanoma cell lines (A375 and FO1) were pretreated with curcumin (20 μmol/L) or vehicle (DMSO; negative control) for 6 hours (prior to the induction of apoptosis) and subsequently cultured for 15 minutes with PBS (vehicle), IFN-α (10⁴ U/mL), or IFN-γ (10 ng/mL) to stimulate signal transduction. These doses were chosen as they induce a maximal degree of STAT1 phosphorylation at this time point (23). Cell lysates were evaluated by immunoblot analysis. Reduced levels of IFN-α and IFN-γ-induced STAT1 phosphorylation, and IFN-α–induced STAT3 phosphorylation was observed in these melanoma cell lines following pretreatment with curcumin (Fig. 2A and B). These data indicated that curcumin could modulate signal transduction events involved in promoting tumor progression (e.g., STAT3-mediated) or apoptosis (e.g., STAT1-mediated; ref. 45).

The effect of curcumin on downstream IFN-induced gene expression was next evaluated by real time PCR. In these experiments, melanoma cell lines were pretreated for 1 hour with curcumin (20 μmol/L) or DMSO (vehicle) as a control and cultured in the presence of PBS, IFN-α (10⁴ U/mL), or IFN-γ (10 ng/mL) for an additional 4 hours. Curcumin pretreatment of the multiple human melanoma cell lines led to consistent reductions in expression of the proapoptotic IFN-regulatory factor-1 (IRF1) gene expression following IFN-γ stimulation as compared with DMSO-pretreated cells (Fig. 2C). This gene has been previously shown to be transcribed via a STAT1-STAT1 homodimer binding to a γ-activated sequence element and is well characterized as an IFN-γ–responsive gene (46). Similar reductions in transcription of IFN-α–stimulated genes regulated by the ISGF3 complex (e.g., ISG-15) were also noted following curcumin pretreatment in representative melanoma cell lines (Fig. 2D and not shown).

Stimulation of host immune effector cells is thought to contribute to the antitumor potential of interferons or interleukins (20, 21, 47). Similar to data obtained in melanoma cell lines, a 5-hour pretreatment of PBMCs isolated from healthy donors with curcumin (10 μmol/L or 20 μmol/L) also inhibited the ability of IFN-α and IFN-γ to phosphorylate STAT1 or STAT3 (Fig. 3A). Curcumin pretreatments also led to a striking inhibition of IL-2–induced STAT5 phosphorylation in PBMCs (Fig. 3B). These data suggested the inhibitory effects of curcumin were broad and not specific to the interferons. Subsequent real time PCR analysis in multiple donors further indicated that curcumin pretreatment significantly inhibited the IFN-γ–induced transcription of IRF1, and the IFN-α–induced transcription of ISG-15 in PBMC from all donors examined (Fig. 4A and B; P = 0.02 for IRF1; P = 0.05 for ISG-15). In some donors, the ability of curcumin to inhibit IFN responsiveness was dramatic. However, as shown in prior published studies, there was a high degree of donor variability in the individual responsiveness to IFN stimulation (38, 48). The curcumin-mediated inhibition of gene expression seemed to be dose-dependent, and was more pronounced for IFN-α–induced genes as compared with IFN-γ–induced genes (Fig. 4C and D). Similar data were obtained in PBMCs following curcumin pretreatment for the IL-2–responsive gene CIS (Fig. 4E), and were reproduced following a 4-hour costimulation of PBMCs with curcumin and these cytokines (not shown).

We hypothesized that curcumin-mediated inhibition of signal transduction might also lead to a similar decrease in the function of NK cells. This cell subset can directly eliminate malignant cells and is activated in response to stimulation with a variety of immunoregulatory cytokines (49). Cotreatment of NK cells from healthy donors with curcumin and IL-12 (10 or 50 ng/mL) led to reduced secretion of IFN-γ as compared with NK cells cotreated with DMSO (vehicle) and IL-12 for 24 hours (Fig. 5A). Importantly, overnight treatment of healthy donor NK cells with curcumin (20 μmol/L) also impaired GrB release from NK cells against the A375 melanoma cell line, and the NK-sensitive K562 cell line, serving as a positive control target (Fig. 5B). Targets or effector NK cells cultured alone showed no GrB release (data not shown). Although an inherent degree of interindividual variability was observed in the NK cell response, the inhibitory effect of curcumin on NK cell function was consistently observed in three of three individual donors. Curcumin-treated NK cells exposed to A375 targets exhibited on average an 89 GrB spots/well decrease (± 29) over vehicle-treated NK cells from the same donor (P = 0.03). A similar trend was seen against the K562 line where curcumin-treated NK cells yielded on average 65 less GrB spots/well (± 16) than vehicle-treated NK cells (P = 0.02).

Consistent with our ELISA data obtained following IL-12 stimulation (Fig. 5A), curcumin also led to a slightly reduced production of IFN-γ by NK cells in the presence of A375 melanoma and K562 target cell lines (Fig. 5C). Curcumin-treated NK cells exposed to A375 targets exhibited on average a 101 IFN-γ spots/well (± 68) decrease compared with vehicle-treated NK cells from the same donor (P = 0.12). A similar effect was seen against the K562 line where curcumin-treated NK cells produced on average 57 less IFN-γ spots/well (±18) than untreated NK cells (P = 0.03).

The use of medicinal natural products is widespread in patients with various types of cancer. Recent estimates indicate that one third of Americans use medicinal natural products daily, and that possibly >50% of cancer patients use them as well (1). Curcumin represents a compound used in dietary supplements, and its antitumor potential is currently being evaluated in human clinical trials for various cancers, including multiple myeloma, pancreatic cancer, and colon cancer (50). Although the precise mechanisms by which many medicinal natural products elicit antitumor activity are unknown, these compounds do possess bioactive components that inhibit growth and promote apoptosis of malignant cells. However, few studies have explored in detail the off-target effects of these compounds on human immune cells, or their ability to modulate the cellular response to immunotherapeutic cytokines (51–53).

Due to the ability of the immune system to regulate tumor outgrowth, the interaction between bioactive compounds and host immune function is an important area of research. Data from the present study showed that curcumin induced apoptosis of human melanoma cells in vitro. Curcumin also inhibited the IFN-responsiveness of melanoma cells and abrogated signal transduction, gene expression, and functional activity of PBMCs in response to numerous cytokines including IFN-α, IFN-γ, IL-2, and IL-12. These data indicate that despite its beneficial, direct antitumor actions, curcumin (and potentially other natural products) may adversely modulate the cellular response to clinically relevant cytokines.

Curcumin has shown cytotoxic activity against a variety of tumor targets, suggesting that it has broad potential as an effective chemopreventative or chemotherapeutic agent (1). The present data further confirm that single-agent curcumin is active against human melanoma cells. Previous reports have also shown that curcumin induces apoptosis against the B16F10 murine melanoma cell line, and that this cell death was mediated by endoplasmic reticulum stress and down-regulation of the antiapoptotic protein Mcl-1 (54). In human melanoma cells, curcumin has also been shown to induce apoptosis through a Fas receptor/caspase-8 pathway, to suppress NF-κB activity, and induce G₂-M cell cycle arrest (55–57). Consistent with these observations, our data showed that curcumin induced apoptosis in radial growth phase, vertical growth phase, and metastatic human melanoma cell lines, and down-regulated basal pSTAT3 and Bcl-2. Because curcumin can inhibit multiple prosurvival pathways, there is considerable interest in studying whether this compound can augment existing therapies for melanoma or other malignancies (42, 43).

The ability of curcumin to inhibit various aspects of innate immune function and Jak-STAT signal transduction can be considered controversial with respect to its clear antitumor activity (52, 53, 58–60). Indeed, these inhibitory properties of curcumin are likely beneficial for preventing cancers that arise from chronic inflammation and for reducing certain mediators that contribute to immune suppression. For example, curcumin suppressed STAT1 activation in murine dendritic cells and inhibited indoleamine 2,3-dioxygenase–mediated suppression of T-cell responses (53). Oral administration of curcumin to mice bearing ascites carcinoma also inhibited tumor-induced T-cell death (51). Conversely, curcumin has been shown to inhibit basic functional properties of immune effector cells that could facilitate immune-mediated recognition of tumors, including IL-2–induced proliferation of cytotoxic T lymphocytes and expression of the CD80 and CD86 costimulatory molecules (60). Data from the present report suggest that curcumin may adversely affect the responsiveness of immune cells to clinically relevant cytokines that are often administered to patients with advanced melanoma. Of note, previous studies have shown that STAT1 signal transduction in immune cells was required for the antitumor actions of exogenously administered IFN-α in murine models of melanoma (61). Similarly, the STAT5 transcription factor plays a crucial role in normal immune function by promoting IL-2–induced cell cycle progression in T cells (62–64) and regulates NK cell–mediated proliferation and cytolytic activity (65–67). These data caution that despite its anti-inflammatory, chemopreventative properties, curcumin may potentially negate the antitumor properties of therapeutically administered cytokines.

It has been difficult to truly characterize the immunomodulatory effects of curcumin because many prior studies done in murine models have not documented plasma or tissue concentrations of curcumin. This information would have allowed for a more accurate interpretation as to whether the concentrations achieved in vivo were relevant for inducing tumor cell apoptosis. Efforts to improve the in vivo bioavailability and absorbtion of curcumin are being rigorously investigated and have produced promising initial results. For example, simply heating curcumin has been shown to increase its water solubility by 12-fold (68). In a similar manner, coadministration of curcumin together with piperine, an extract from black pepper, has been shown to increase the bioavailability of curcumin by 1,000% in normal human donors (5). This effect is primarily due to the ability of piperine to inhibit hepatic and intestinal glucuronidation, a major mechanism by which curcumin is rapidly metabolized (5). The present data suggest careful attention should be given to the immunomodulatory effects of curcumin once the hurdle of achieving systemic levels that induce apoptosis is overcome.

Although our understanding of melanoma development is in its infancy, the presence of activated lymphocytes in regressing tumors, the clinical activity of exogenous cytokines such as IL-2 and IFN-α, and the high incidence of autoimmune vitiligo observed in patients who respond clinically to IFN-α suggest that the immune system can influence the development and metastatic progression of melanoma (20, 23, 69). Consequently, melanoma represents a form of cancer that may be amenable to immune-based therapy. In particular, melanoma patients rendered disease free following surgery who are at high risk for recurrence represent a population in which chemopreventative agents could have a positive impact. Ideally, these agents could complement the actions of exogenous or endogenous immunostimulatory cytokines to provide a potentially greater clinical benefit.

In conclusion, data from the present study raise concern that utilizing medicinal natural products could negate the effects of immunotherapy or of the ability of the immune system to recognize tumors. Curcumin represents a single example of the numerous alternative medicine or nutritional supplements available over-the-counter that are used regularly in the absence of strong scientific data. In line with these concerns are recent reports that green tea polyphenols block the antitumor properties of bortezomib and other boron-based proteasome inhibitors against multiple myeloma cells (70). These data suggest that the ability of bioactive natural products to modulate immune function and cytokine responsiveness deserves further investigation in a well-controlled manner before recommending their widespread use.

No potential conflicts of interest were disclosed.

We thank Dr. William E. Carson, III, for critical review of the manuscript, and the OSU CCC Analytical Cytometry, Nucleic Acid and Biostatistics Shared Resources.