Curcumin Induces a p53-Dependent Apoptosis in Human BasalCell Carcinoma Cells

Shiou-Hwa Jee, Shing-Chuan Shen,* Chung-Ren Tseng, Hsien-Ching Chiu, and Min-Liang Kuo*Department of Dermatology, College of Medicine, National Taiwan University, Taipei, Taiwan; *Laboratory of Molecular and Cellular Toxicology, Institute ofToxicology, College of Medicine, National Taiwan University, Taipei, Taiwan

Curcumin, a potent antioxidant and chemopreventiveagent, has recently been found to be capable of inducingapoptosis in human hepatoma and leukemia cells by wayof an elusive mechanism. Here, we demonstrate thatcurcumin also induces apoptosis in human basal cellcarcinoma cells in a dose- and time-dependent manner,as evidenced by internucleosomal DNA fragmentationand morphologic change. In our study, consistent withthe occurrence of DNA fragmentation, nuclear p53 pro-tein initially increased at 12 h and peaked at 48 hafter curcumin treatment. Prior treatment of cells withcycloheximide or actinomycin D abolished the p53increase and apoptosis induced by curcumin, suggestingthat either de novo p53 protein synthesis or some proteinssynthesis for stabilization of p53 is required for apoptosis.In electrophoretic mobility gel-shift assays, nuclearextracts of cells treated with curcumin displayed distinctpatterns of binding between p53 and its consensus bindingsite. Supportive of these findings, p53 downstream tar-

Curcumin (diferuloylmethane) is a major active com-ponent of the food flavoring turmeric (Curcuma longa).Topical application of curcumin on the skin of miceresults in inhibiting benzo (a) pyrene- and DMBA-induced skin carcinogenesis (Huang et al, 1992). The

same blocking effect is also observed in various carcinogenic modelsthrough dietary administration of curcumin (Azuine and Bhide, 1992;Huang et al, 1993). Aside from its anti-carcinogenic effects, curcuminexhibits remarkable anti-inflammatory and antioxidant properties in vivo(Sharma, 1976; Mukhopadhyay et al, 1982; Rao et al, 1982; Toda et al,1985). The pharmacologic safety of curcumin is demonstrated by thefact that it has been consumed for centuries at levels of up to 100 mgper d by people in certain countries (Ammon and Wahl, 1991);however, the mechanism underlying the diverse effects of curcumin isnot fully understood. One possible molecular mechanism that hasbeen suggested is that curcumin suppresses the phorbol ester-inducedtranscriptional factor c-jun/AP-1 (Huang et al, 1991). Recently, Korutlaand Kumar (1994) showed that curcumin is capable of inhibiting the

Manuscript received January 13, 1998; revised May 11, 1998; accepted forpublication June 10, 1998.

Reprint requests to: Dr. Min-Liang Kuo, Laboratory of Molecular andCellular Toxicology, Institute of Toxicology, College of Medicine, NationalTaiwan University, No 1, Sec 1, Jen-Ai Road, Taipei, Taiwan.

Abbreviations: BCC, basal cell carcinoma; CIP, cyclin-dependent kinaseinteracting protein; Gadd45, growth arrest and DNA damage-inducible protein.

0022-202X/98/$10.50 · Copyright © 1998 by The Society for Investigative Dermatology, Inc.

656

gets, including p21CIP1/WAF1 and Gadd45, could beinduced to localize on the nucleus by curcumin withsimilar p53 kinetics. Moreover, we immunoprecipitatedextracts from basal cell carcinoma cells with differentanti-p53 antibodies, which are known to be specific forwild-type or mutant p53 protein. The results reveal thatbasal cell carcinoma cells contain exclusively wild-typep53; however, curcumin treatment did not interfere withcell cycling. Similarly, the apoptosis suppressor Bcl-2 andpromoter Bax were not changed with the curcumintreatment. Finally, treatment of cells with p53 anti-sense oligonucleotide could effectively prevent curcumin-induced intracellular p53 protein increase and apoptosis,but sense p53 oligonucleotide could not. Thus, ourdata suggest that the p53-associated signaling pathway iscritically involved in curcumin-mediated apoptotic celldeath. This evidence also suggests that curcumin may bea potent agent for skin cancer prevention or therapy. Keyword: DNA strand breaks. J Invest Dermatol 111:656–661, 1998

intrinsic kinase activity of epidermal growth factor receptors, leadingto growth inhibition of A431 cells. In addition, curcumin is a potentprotein kinase C inhibitor in NIH 3T3 cells (Jiang et al, 1996b). Thesefindings indicate that curcumin may well be a potent anti-tumor agentfor some cancer cells by way of a mechanism that modulates thesignaling transduction pathway.

Apoptosis, a mode of cell death, plays a crucial role in embryonicdevelopment, metamorphosis, hormone-dependent atrophy, and tumorgrowth as a physiologic event that regulates cell number and eliminatesdamaged cells (Tata, 1996; Ishizuya-Oka et al, 1997). Several studieshave demonstrated that apoptosis may be involved in cell deathinduced by chemotherapeutic agents, including cisplatin, camptothecin,amsacrine, etoposide, and teniposide (Eischen et al, 1997; Holm et al,1994). There is accumulating evidence that the efficiency of anti-tumor agents is related to the intrinsic propensity of the target tumorcells to respond to these agents by apoptosis (Villunger et al, 1997).Recent evidence also clearly shows that suppression of apoptosis bytumor-promoting agents in preneoplastic cells is thought to be animportant mechanism in tumor promotion (Shibata et al, 1996).Supportive of this notion, a considerable incidence of apoptosisassociated with remodeling has been observed in nodular lesions duringtheir disappearance induced by administration of the chemopreventiveagent S-adenosyl-L-methionine (Garcea et al, 1989; Pascale et al, 1992).In this context, it is noteworthy that apoptosis-inducing ability seemsto have become a primary factor in considering the efficiency ofchemopreventive agents.

P53 is well known for suppressing cellular proliferation through two

VOL. 111, NO. 4 OCTOBER 1998 CURCUMIN INDUCES APOPTOSIS IN BCC CELLS 657

mechanisms, each operating in a distinct context. In normal fibroblasts,p53 induces G1 arrest in response to DNA-damaging agents, presumablyto allow the cells to perform critical repair functions before progressingthrough the cell cycle (Linke et al, 1997; Loignon et al, 1997). Inother contexts, such as in abnormally proliferating cells or irradiatedthymocytes, induction of p53 leads to programmed cell death orapoptosis (Midgley et al, 1995). Moreover, wild-type p53 protein isincreased during apoptosis induced by DNA-damaging agents (Zhanet al, 1996), and by male hormone depletion (Colombel et al, 1992).Therefore, the regulation and expression of the p53 gene may be ofcentral importance to the induction of apoptosis in preneoplastic andtumor cells, and intrinsic to the efficacy of a range of chemopreventiveand chemotherapeutic agents.

In this study, we first report that curcumin induced apoptosis inhuman basal cell carcinoma cells accompanied by the upregulation ofp53 protein and its downstream targets p21CIP1/WAF1 and Gadd45;however, curcumin treatment did not interfere with cell cycling, orchange other apoptosis-related proteins, i.e., Bcl-2 or Bax. Treatmentof basal cell carcinoma (BCC) cells with a p53-specific anti-senseoligonucleotide resulted in attenuation of curcumin-induced apoptosis.Our data suggest that p53-related signaling, but not the Bcl-2 family,plays an important role in apoptotic cell death induced by curcumin.

MATERIALS AND METHODS

Cell culture and chemicals Curcumin, cycloheximide, actinomycin D,and staurosporine were obtained from Sigma (Santa Ana, CA). Human BCCwere established as described previously (Yen et al, 1996). The cell line wascultured in an RPMI 1640 medium supplemented with 10% fetal calf serum.The p53-specific anti-sense (59-CGGCTCCTCCATGGCAGT-39) and sense(59-ACTGCCATGGAGGAGCCG-39) phosphorothioates (S-oligos) were syn-thesized and purified by high-performance liquid chromatography by Genset.

Electron microscopy Curcumin-treated and -untreated cells (0.1%dimethylsulfoxide) were fixed with fixative containing 2% glutaraldehyde and2% paraformaldehyde in phosphate-buffered saline (PBS) for 15 min. Followingseveral rinses in PBS, cells were post-fixed in 1% osmium tetroxide, dehydratedin alcohol, and embedded in an Epon-Araldite mixture. The sections were cutand doubly stained with uranyl acetate and lead citrate before being examinedwith a Joel 2000EXII electron microscope at 100 kV.

DNA extraction and electrophoresis Cells (1 3 106) were treated withdifferent concentrations of curcumin for 16 h or a fixed concentration forvarying exposures of time, then harvested and washed twice in ice-cold PBS,resuspended in 500 µl TE (10 mM Tris-HCl pH 7.6, 1 mM ethylenediaminetetraacetic acid pH 8.0), and lyzed in 500 µl lysis buffer (3% sodium dodecylsulfate, 50 mM Tris, pH 12.6) at room temperature for 10 min. DNA wasextracted with phenol and chloroform before precipitation with 95% ethanol.DNA pellets were finally solubilized in TE buffer and treated with RNase Afor 40 min prior to 1.2% agarose gel electrophoresis.

Flow cytometry analysis At an indicated time (0, 6, 12, 24, 48, or 72 h),curcumin-treated or -untreated cells were harvested and fixed in 75% ethanolat –20°C for at least 1 h. After centrifugation at 800 rpm for 5 min at 4°C,cell pellets were resuspended in 0.5 ml of a buffer (0.5% Triton X-100/PBSand 0.05% RNase A) and incubated for 30 min. Finally, 0.5 ml of propidiumiodide solution (50 µg per ml) was added, and cells were allowed to stand onice for 15–30 min. Fluorescence emitted from the propidium iodide-DNAcomplex was quantitated after laser excitation of the fluorescent dye by aFACSsor flow cytometer (Becton Dickinson, San Jose, CA). The effect ofvarious modulators on curcumin-triggered apoptosis was also examined andquantitated by flow cytometric analysis.

Immunoblotting and immunoprecipitation Cellular lysates were pre-pared as described previously (Kuo et al, 1996). A 50 µg sample of each lysatewas subjected to electrophoresis on 10% sodium dodecyl sulfate-polyacrylamidegels for detection of p53, p21CIP1/WAF1, Gadd45, Bcl-2, and Bax. The sampleswere then electroblotted on nitrocellulose paper. After blocking, blots wereincubated with anti-p53 (Oncogene Science, Manhasset, NY), anti-p21CIP1/WAF1 (Transduction, Lexington, KY), anti-Gadd45, anti-Bcl-2, andanti-Bax (Santa Cruz Technology, Santa Cruz, CA) antibodies in 10 mM TrispH 7.5, 100 mM NaCl, 0.1% Tween 20 (PBST) for 1 h followed by twowashes (15 min each) in PBST, and then incubated with horseradish peroxidase-conjugated goat anti-mouse IgG (Amersham, Arlington Heights, IL) for 30 min.After washing, blots were incubated for 1 min with the western blotting reagent

ECL (Amersham) and chemilluminescence was detected by exposure of thefilters to Kodak-X-Omat films for 30 s to 30 min.

Cell lysates were prepared as described previously (Kuo et al, 1997). Equalamounts of cell lysates (400 µM) were preabsorbed with preimmune mouseserum and protein A-sepharose for 1 h. Subsequently, immunoprecipitationwas performed with anti-p53 monoclonal antibody (Oncogene Science) at 4°Cfor 15 h, and immunocomplexes were then absorbed with protein A-sepharose.After washing four times with RPIPA buffer, the immunoprecipitates wereseparated on 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis,and electroblotted on PVDF filters or autoradiographed with Kodak X-OMAT film.

Immunohistochemistry Cells were fixed in 80% cold-methanol, and furtherblocked with 3% bovine serum albumin in phosphate-buffered saline containinganti-p53, anti-p21CIP1/WAF1, or anti-Gadd45 antibodies. After being washedwith PBS, cells were incubated with fluorescent isothiocyanate-conjugatedsecondary antibodies, and detected by fluorescence microscope.

Nuclear extract preparation and gel shift Nuclear extracts of BCC cellstreated with curcumin or solvent control were prepared as described previously(Kuo et al, 1997). To detect p53-specific DNA-binding activity, unlabeledwild-type and mutant p53 consensus DNA oligonucleotides were used in theexperiment. Briefly, the 25 µl of DNA-binding reaction mixture containedunlabeled p53 consensus fragments (Santa Cruz Technology), 3000–6000 cpmof 32P end-labeled DNA fragments, 20 µg nuclear extract, and a buffercomposed of 20 mM HEPES pH 7.6, 20% glycerol, 10 mM sodium chloride,0.2 mM ethylenediamine tetraacetic acid, 0.5% NP-40, 1 mM dithiothreitol,and 1 mM phenylmethylsulfonyl fluoride. Binding reactions and subsequentanalysis on native 4% polyacrylamide gels was performed at room temperatureas previously described (Zhang et al, 1994). Gels were then dried and exposedto X-ray film.

RESULTS

Curcumin induces apoptosis in BCC cells Previous studies haveshown that the LD50 of curcumin on cells was 20 µM for humanleukemia cells and 50 µM for human hepatoblastoma cells (Jiang et al,1996a; Kuo et al, 1996). In this study, the dose-dependent curve wasmeasured by trypen blue exclusion and the 24 h LD50 of curcuminon BCC cells was 50 µM (data not shown). To determine the modeof cell death induced by curcumin, BCC cells were treated with50 µM curcumin, and morphologic alterations were subsequentlyexamined via electron microscopy. Curcumin-treated BCC cellsshowed morphologic changes characteristic of apoptosis, includingdisappearance of microvilli, cell shrinkage, chromatin condensation,and appearance of membrane blebbing (Fig 1B) when compared withuntreated cells (Fig 1A). Agarose gel electrophoresis of curcumin-treated chromosomal DNA showed a ladder-like pattern of DNAfragments consisting of multiples of µ180–200 base pairs. An earlyDNA fragmentation was seen at 12 h after exposure to 50 µMcurcumin (Fig 2A). To further determine the degree of apoptosis, weemployed flow cytometry to quantitate the sub-G1 peak (for apoptoticcells). Figure 2(B) shows that when BCC cells were exposed to 50 µMcurcumin for 24, 48, and 72 h, the result was 30%, 50%, and 63%apoptosis, respectively.

Curcumin induces an increase in p53 protein and its DNAbinding activity, but does not cause cell cycle arrest P53 hasbeen found to be importantly involved in apoptosis induced by a broadrange of agents (Yonish-Rouach et al, 1996; Canman and Kastan,1997). We thus examined whether p53 protein could be induced inBCC cells treated with curcumin as determined by using westernblotting. Nuclear extracts from BCC cells treated with 50 µM curcuminfor the indicated durations of time showed that p53 protein began toincrease at 12 h, and gradually increased to a maximum level by 48 hafter treatment (Fig 3A). The kinetics of the increase in p53 proteincorrelated well with the onset of apoptosis induced by curcumin(compare Figs 2B and 3A); however, when BCC cells were givenprior treatment with cycloheximide (a protein synthesis inhibitor)or actinomycin D (an RNA synthesis inhibitor), curcumin-inducedupregulation of p53 protein and apoptosis were effectively abolished(Fig 3A, B). Because curcumin has been reported to be a potentprotein kinase C inhibitor (Liu et al, 1993), we selected staurosporine,a well-known protein kinase C inhibitor, to examine whether stauros-

658 JEE ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 1. Apoptosis occurring in human basal cell carcinoma cells aftertreatment with curcumin. BCC cells were treated with solvent control (0.1%dimethylsulfoxide) (A) or curcumin (50 µM) for 12 h (B). After removal ofthe curcumin, the cells were fixed and examined as described in Materials andMethods. Original magnification, 6003; scale bar: 6 µm.

porine could induce upregulation of p53 protein and apoptosis in thecell system. Interestingly, staurosporine induced an increase in p53protein that was consistent with the occurrence of apoptosis (Fig 3A, B).

Next, we examined the relative activity of p53 from nuclear extractsof curcumin-treated BCC cells to bind to its known cognate sequence,by using an electrophoretic mobility gel shift assay. In nuclear extractsof curcumin-treated BCC cells, a band corresponding to the p53-DNA-binding complex was easily detected without curcumin treatment(Fig 3C, lane 1). The band intensity began to increase at 12 h andreached a maximum at 24 h after curcumin treatment (Fig 3C, lanes2–6). The specificity of this binding was confirmed by competitionanalysis with an excess of unlabeled homologous or mutant oligonucleo-tides (Fig 3C, lanes 7 and 8).

These findings led us to examine the state of the p53 protein in theBCC cell line. To resolve this issue, we performed immunoprecipitationassay by using different anti-p53 antibodies, which have been knownto specifically recognize wild-type or mutant p53 proteins. As Fig 4depicts, an abundance of p53 proteins could be immunoprecipitatedfrom extracts of BCC cells by using pAb1801 antibodies (specific toboth wild-type and mutant p53 proteins) and pAb246 antibodies(specific to wild-type p53 protein), but that was not detected usingpAb240 antibody (specific to mutant p53 protein). pAb240 antibody,however, could immunoprecipitate a substantial amount of p53 proteinfrom the cell lysate of human colon SW620 cells, which are knownto harbor mutant p53. These results strongly indicate that BCC cellscontain wild-type p53 protein.

It has been reported that induction of wild-type p53 protein bysome cytotoxic agents often leads to growth arrest at the G1 stage,with cells subsequently undergoing apoptosis. To investigate whethercurcumin-treated BCC cells also show growth arrest prior to apoptosis,the DNA contents of BCC cells treated with curcumin for variousperiods were analyzed by flow cytometry (Fig 5). The DNA content

Figure 2. Internucleosomal DNA fragmentation in BCC cells treatedwith curcumin. (A) BCC cells were exposed to 50 µM curcumin for 0–48 h.DNA from cells exposed to different time points was extracted, electrophoresedthrough 2% agarose gels, and visualized by staining with ethidium bromide.(B) The apoptotic cells in the sub-G1 region were quantitated by flowcytometry. The data presented here are the mean value of three independentexperiments. Bars indicate standard errors.

profile was not altered by the first 6 h of curcumin treatment. A sub2NDNA peak was not observed until 12 h of exposure, the timing ofwhich is consistent with the appearance of the DNA ladder. Throughoutthe course of apoptosis, there was no obvious alteration in thedistribution of each cell cycle stage.

Curcumin induces an increase in p53 downstream targetsp21CIP1/WAF1 and Gadd45 p21CIP1/WAF1 and Gadd45 are down-stream targets of p53 that can be used as intracellular markers forexamining the biologic effect of p53. Western blot analysis showedthat both p21CIP1/WAF1 and Gadd45 initially increased at 12 h aftercurcumin treatment, and increased steadily thereafter (Fig 6A). Appar-ently, the induction kinetics of both proteins are closely consistentwith the upregulation of p53 protein (Fig 3A). Furthermore, theimmunofluorence analysis showed that p53, p21CIP1/WAF1, and Gadd45were almost exclusively located in the nucleus of curcumin-treatedBCC cells (Fig 6B). In contrast, there was no apparent increase inthese proteins in either the cytoplasm or the nucleus of untreated cells.These results suggest that curcumin cannot only upregulate p53protein and its downstream targets, but also promotes their subsequenttranslocation to the nucleus.

Attenuation of curcumin-induced apoptosis by anti-sense p53oligonucleotides To demonstrate that p53 protein is specificallyinvolved in curcumin-induced apoptosis, BCC cells were treated with20 µM p53-specific anti-sense oligonucleotides for 16 h prior to theaddition of 50 µM curcumin for another 24 h. Under this condition,

VOL. 111, NO. 4 OCTOBER 1998 CURCUMIN INDUCES APOPTOSIS IN BCC CELLS 659

Figure 3. Increase of p53 protein was closely associated with curcumin-induced apoptosis. (A) BCC cells were treated with 50 µM curcumin,staurosporine (50 ng per ml; STP), or pretreated with cycloheximide (10 µM;CHX), actinomycin D (2 ng per ml; Act D) for 3 h followed by 50 µMcurcumin treatment. The expression of p53 protein was detected by westernblotting. (B) The percentage of hypodiploid cell in BCC cells under differenttreatment as described in (A) was measured by flow cytometric analysis. (C)Induction of p53 binding activity by curcumin was detected by gel retardationassay in BCC cells. wt, 25-fold molar excess of the nonradioactive p53CONoligonucleotide was added prior to the addition of [32P] p53CON; mut, 25-fold molar excess of the mutated p53CON was added prior to the addition of[32P] p53CON.

the curcumin-induced apoptosis (sub-G1 cells) could be attenuated bythe p53-specific anti-sense oligonucleotide, but not by its senseoligonucleotide (Fig 7A). To further examine whether the intracellularp53 protein would be reduced by anti-sense oligonucleotide, wedetermined the change of p53 protein level by using flow cytometry.Indeed, the curcumin-induced increment of p53 protein was effectivelyinhibited by treatment with p53-specific anti-sense oligonucleotide,but that was not affected by sense oligonucleotide (Fig 7B). Theseresults indicate that p53 protein indeed plays a critical role in thecurcumin-induced apoptotic cell death process.

Finally, we examined whether the other apoptosis-related targets,Bcl-2 and Bax, were also involved in apoptosis induced by curcumin.Western blot analysis showed that neither the Bcl-2 nor the Baxprotein levels were changed during the course of apoptosis in curcumin-treated cells when compared with the level of α-tubulin (data notshown).

Figure 4. Identification of the state of p53 proteins in BCC cells. Cellextracts from BCC cells were immunoprecipitated by pAb1801 (specific toboth wt p53 and mut p53 protein), pAb 240 (specific to mut p53), or pAb 246(specific to wt p53), respectively, and followed by western blotting to detectp53 protein using pAb 1801 as prob. mut, control of mutated p53 protein fromSW620 cell extracts, which were immunoprecipitated by pAb 240 antibodyand detected by western blotting described in Materials and Methods.

Figure 5. Flow cytometric analysis of DNA content in BCC cells treatedwith curcumin. One million BCC cells were treated with 50 µM curcuminfor the indicated time and then permeabilized and subjected to flow cytometricanalysis as described in Materials and Methods.

DISCUSSION

In this study, we demonstrated that curcumin is able to induce apoptosisin BCC cells, accompanied by an increase in p53, p21CIP1/WAF1, andGadd45 protein levels. Although our and other previous studies haveshown that curcumin-induced apoptosis occurs in human leukemiaand hepatoma cells (Jiang et al, 1996a; Kuo et al, 1996), the mechanismof action of curcumin is not fully understood. Here, we report thatthe process of apoptosis induced by curcumin is strictly correlatedwith the induction of the p53-associated signaling pathway. Moreimportantly, this is the first demonstration that the p53-associatedpathway is indispensable for curcumin-mediated apoptosis, as evidencedby the p53 anti-sense oligonucleotide experiment. Cycloheximide andactinomycin D could abolish an curcumin-induced increase in p53protein and apoptosis in BCC cells, suggesting that either de novosynthesis of p53 protein or protein synthesis for stabilization of p53protein is required for the cell death process. The results presentedhere, however, are different from other studies that have shown thatDNA-damaging agents induce an increase in p53 protein via a post-transcriptional mechanism (Kren et al, 1996; Chang et al, 1997).Anyhow, further investigation is required to clarify the inductivemechanism of p53 by curcumin.

Curcumin does not cause DNA damage but affects many kinds ofcellular protein kinases, including protein kinase C and epidermalgrowth factor receptor kinase (Korutla et al, 1995; Jiang et al, 1996b).Therefore, the difference in the inductive mechanism of p53 between

660 JEE ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 6. Induction of CIP1 and Gadd45 proteins by curcumin in BCCcells. (A) Immunoblot analysis of CIP1 and Gadd45 proteins in BCC cells.Cell extracts were prepared and immunoblotting performed as described inMaterials and Methods. (B) Immunostaining of intracellular p53, Gadd45, CIP 1protein by specific antibody. Cells were treated with curcumin (50 µM) for12 h and fixed in methanol at –20°C followed by staining with anti-p53(Oncogene Science), anti-CIP1 (Transduction), or anti-Gadd45 (Santa CruzTechnology) antibody, and observed with fluorescence microscope. Control,cells cultured with 0.1% dimethylsulfoxide for 12 h; 12 h, cells treated with50 µM curcumin for 12 h.

curcumin and DNA-damaging agents is likely due to the different celltypes, or the different cellular targets that are attacked by them. Somewell-known protein kinase inhibitors, such as staurosporine (a proteinkinase C inhibitor) and H7 (a nonspecific serine-threonine kinaseinhibitor), have also been found to induce an increment of p53 proteinin MCF-7 cells (Jeoung et al, 1995a, b).

In this study, we also showed that staurosporine treatment couldcause the expression of p53 protein in BCC cells, which could beabolished by prior treatment with cycloheximide or actinomycin D(data not shown). Taken together, these findings suggest that curcumininduces the upregulation of p53 protein in BCC cells, possibly throughmodulation of protein kinase-related signaling pathways.

In many cases, the induction of wild-type p53 protein followingexposure to DNA-damaging agents has been shown to arrest cells atthe G1 phase of the cell cycle (Shao et al, 1995; Pellegata et al, 1996).Our results show that although curcumin can upregulate the expressionof p53, it does not interfere with cell cycling. Similar results have beenreported for myeloid leukemia S6 cells and hepatoma HepG2 cells,where the induction of wild-type p53 failed to arrest cell growth

Figure 7. Attenuation of curcumin induced the increase of p53 proteinand apoptosis by anti-sense p53 oligonucleotides. (A) BCC cells werepretreated with 20 µM sense (S) or anti-sense (AS) oligonucleotides for 16 hfollowed by 50 µM curcumin (Cur) treatment for 24 h. The percentage ofapoptotic cell was measured by flow cytometric analysis. C, control group. (B)Flow cytometric analysis of p53 protein levels in BCC cells in the presence ofanti-sense or sense p53 S-oligos. Cell samples were trypsinized and fixed in75% ethanol. Then 0.5% Triton X-100-premeabilized samples were stainedwith fluorescent isothiocyanate-conjugated anti-p53 monoclonal antibody todetermine the p53 protein.

during apoptosis (Yonish-Rouach et al, 1993; Jiang et al, 1996a). Theactual mechanism for these two distinct phenotypes remains unclear.We suggest that growth arrest and apoptosis may be only two of manydifferent independent events controlled by the p53 gene. In some casesthese events are synchronously regulated; however, in other cases theresponses are uncoupled.

Because of the increment in p53-DNA binding activity as well asthe expression of two p53 downstream targets, p21CIP1/WAF1 andGadd45, in BCC cells after curcumin treatment, we suggest that thep53-associated function in the cell line is normal. Supportive ofthis hypothesis, we found that BCC cells harbor wild-type p53 asdemonstrated by immunoprecipitation assay using different anti-p53antibodies.

The experiment that exposed BCC cells to p53-specific anti-senseoligonucleotide resulted in only a 50% reduction of apoptosis inducedby curcumin. This indicates that mechanism(s) other than p53 may beinvolved in curcumin-induced apoptosis. Another possibility is thatthe efficiency of anti-sense oligonucleotide is limited by its toxicity toBCC cells; however, flowcytometric analysis clearly showed that anti-sense p53 oligonucleotides effectively reduced curcumin-induced p53protein increase in BCC cells. Based on these findings, we think thatthe former possibility is appropriate for explaining the partial inhibitionof anti-sense p53 on curcumin-mediated apoptosis. Again, the inductionof p21CIP by curcumin did not lead to G1 arrest that is in contrast toother studies showing the p21CIP is a critical regulator for G1 arrest(Waldman et al, 1995).

The members of the Bcl-2 gene family are major regulators ofapoptosis, but their importance in apoptosis induced by curcumin in

VOL. 111, NO. 4 OCTOBER 1998 CURCUMIN INDUCES APOPTOSIS IN BCC CELLS 661

BCC cells remains unclear. Although the expression of Bcl-2 and Baxin M1 murine leukemia cells is highly inducible by activating the ts-mutant of the p53 gene (Miyashita et al, 1994; Guillouf et al, 1995),our data showed no correlation between p53 induction, onset ofapoptosis, and the expression of Bcl-2 and Bax. This result is also incontrast to previous studies showing that curcumin treatment causes adecrease of Bcl-2 protein during the course of apoptosis in humanleukemia HL-60 cells (Samaha et al, 1997; Rao et al, 1995b). Thediscrepancy between these two studies may be primarily due to differenttissue origin.

The chemoprevention of cancer using curcumin has been intensivelyinvestigated (Kuo et al, 1996). Among them, curcumin has been foundto effectively block carcinogen-induced skin carcinogenesis in animals(Rao et al, 1995a; Chen et al, 1996). Recently, it has become clearthat the action mechanism of chemopreventive agents is closely relatedto their ultimate ability to induce apoptosis (Samaha et al, 1997).Previous studies (Jiang et al, 1996a) have demonstrated that curcuminwas able to induce apoptosis in erbB2-transformed NIH 3T3 cells,mouse sarcoma S180, human colon HT-29, human kidney cancer 293,and hepatoma HepG2 cells, but failed to induce apoptosis in humanforeskin fibroblast, rat primary hepatocytes, and mouse embryofibroblasts. In our unpublished data, we found that curcumin failed totrigger a typical apoptosis in primary keratinocytes even increased thedose up to 100 µM. Thus, based on these observations, cancer cellsseem to be more sensitive to curcumin. Understanding the modes ofaction of curcumin should provide useful information for its possibleapplication in cancer prevention, and perhaps also in cancer therapy.

This study was supported by National Science Council of Taiwan, the Republic ofChina, NSC 84-2622-B002-007 and NSC 85-2622-B002-011.

REFERENCES

Amman HP, Safayhi H, Mack T, Sabieraj J: Mechanism of anti-inflammatory actions ofchurchmen and boswellic acids. J Ethnopharmacol 38:113–119, 1993

Ammon HP, Wahl MA: Pharmacology of Curcuma longa. Planta Med 57:1–7, 1991Azuine MA, Bhide SV: Chemopreventive effect of tumeric against stomach and skin

tumors induced by chemical carcinogens in Swiss mice. Nutr Cancer 17:77–83, 1992Canman CE, Kastan MB: Role of p53 in apoptosis. Adv Pharmacol 41:429–460, 1997Chang TC, Tsai LC, Hung MW, Chu LL, Chu JT, Chen YC: Effects of transcription and

translation inhibitors on a human gastric carcinoma cell line. Potential role of Bcl-X (S) in apoptosis triggered by these inhibitors. Biochem Pharmacol 53:969–977, 1997

Chen YC, Kuo TC, Lin-Shiau SY, Lin JK: Induction of HSF70 gene expression bymodulation of Ca (12) ion and cellular p53 protein by curcumin in colorectalcarcinoma cells. Mol Carc 17:224–234, 1996

Colombel M, Olsson CA, Ng PY, Buttyan R: Hormone-regulated apoptosis results fromreentry of differentiated prostate cells onto a defective cell cycle. Cancer Res 52:4313–4319, 1992

Eischen CM, Kottke TJ, Marrtins LM, et al: Comparison of apoptosis in wild-type andFas-resistant cells: Chemotherapy-induced apoptosis is not dependent on Fas/Fasligand interactions. Blood 90:935–943, 1997

Garcea R, Diano L, Pascale R, et al: Inhibition of promotion and persistent nodule growthby s-adenosyl-L-methionine in rat liver carcinogenesis: role of remodeling andapoptosis. Cancer Res 49:1850–1856, 1989

Guillouf C, Grana X, De Selvakumaran M, Luca A, Giordano A, Hoffman B, LiebermannDA: Dissection of the genetic programs of p53-mediated G1 growth arrest andapoptosis: blocking p53-induced apoptosis unmasks G1 arrest. Blood 85:2691–2698, 1995

Holm B, Jensen PB, Sehested M, Hansen HH: In vivo inhibition of etoposide-mediatedapoptosis, toxicity, and antitumor effect by the topoisomerase II-uncouplinganthracycline aclarubicin. Cancer Chemothe Pharmacol 34:503–508, 1994

Huang TS, Lee SC, Lin JK: Suppression of c-jun/AP-1 activation by an inhibitor of tumorpromotion in mouse fibroblast cells. Proc Natl Acad Sci USA 88:5292–5296, 1991

Huang M-T, Lysz T, Ferraro T, Abidi TF, Laskin JD, Conney AH: Inhibitory effects ofcurcumin on tumor initiation by benzo [a]pyrene and 7,12-dimethyl-benz[a]anthracene. Carcinogenesis (Lond) 13:2183–2186, 1992

Huang MT, Ma W, Lou Y-R, Ferraro T, Reuhl K, Newmark H, Conney AH: Inhibitoryeffect of dietary curcumin on gastrointestinal tumorigenesis in mice. Proc Am AssocCancer Res 34:555, 1993

Ishizuya-Oka A, Stolow MA, Ueda S, Shi YB: Temporal and spatial expression of an

intestinal Na1/PO4 3– cotransporter correlates with epithelial transformation duringthyroid hormone-dependent frog metamorphosis. Dev Genet 20:53–66, 1997

Jeoung DI, Tang B, Sonenberg M: Effects of tumor necrosis factor-alpha on antimitogenicityand cell cycle-related proteins in MCF-7 cells. J Biol Chem 270:18367–18373, 1995a

Jeoung DI, Tang B, Sonenberg M: Induction of tumor suppressor p21 protein by kinaseinhibitors in MCF-7 cells. Biochem Biophys Res Commun 214:361–366, 1995b

Jiang MC, Yang-Yen HF, Lin JK, Yen JJ: Differential regulation of p53, c-myc, Bcl-2 andBax protein expression during apoptosis induced by widely divergent stimuli inhuman hepatoblastoma cells. Oncogene 13:609–616, 1996a

Jiang MC, Yang-Yen HF, Yen JJ, Lin JK: Curcumin induces apoptosis in immortalizedNIH 3T3 and malignant cancer cell lines. Nut Cancer 26:111–120, 1996b

Korutla L, Kumar R: Inhibitory effect of curcumin on epidermal growth factor receptorkinase activity in A431 cells. Biochim Biophys Acta 1224:597–600, 1994

Korutla L, Cheung JY, Mendelsohn J, Kumar R: Inhibition of ligand-induced activationof epidermal growth factor receptor tyrosine phosphorylation by curcumin.Carcinogenesis 16:1741–1745, 1995

Kren BT, Trembley JH, Steer CJ: Alteration in mRNA stability during rat liver. Am JPhysiol 270:G763–G777, 1996

Kuo ML, Chou YW, Chau YP, Huang TS. Resistance to apoptosis induced by alkylatingagents in v-Ha-ras-transformed cells due to defect in p53 function. Mol Carcinogenesis18:221–231, Year?

Kuo ML, Huang TS, Lin JK: Curcumin, an antioxidant and anti-tumor promoter, inducesapoptosis in human leukemia cells. Biochim Biophys Acta 1317:95–100, 1996

Linke SP, Clarkin KC, Wahl GM: p53 mediates permanent arrest over multiple cell cyclesin response to gamma-irradiation. Cancer Res 57:1171–1179, 1997

Liu JY, Lin SJ, Lin JK: Inhibitory effects of curcumin on protein kinase C activity inducedby 12-o-tetradecanoyl-phorbol-13-acetate in NIH 3T3 cells. Carcinogenesis 14:857–861, 1993

Loignon M, Fetni R, Gordon AJ, Drobetsky EA: A p53-independent pathway for inductionof p/21WAF1CIP1 and concomitant G1 arrest in UV-irradiated human skin fibroblasts.Cancer Res 57:3390–3394, 1997

Midgley CA, Owens B, Briscoe CV, Thomas DB, Lane DP, Hall PA: Coupling betweengamma irradiation, p53 induction and the apoptotic response depends upon celltype in vivo. J Cell Sci 108:1843–1848, 1995

Miyashita T, Krajewski S, Krajewska M, et al: tumor suppressor p53 is a regulator of bcl-2 and bax gene expression in vitro and in vivo. Oncogene 9:1799–1805, 1994

Samaha HS, Kelloff GJ, Steele V, Rao CV, Reddy BS: Modulation of apoptosis bysulindac, curcumin, phenylethyl-3-methylcaffeate, and 6-phenylhexyl isothiocyanate:apoptosis index as a biomaker in colon cancer chemoprevention and promotion.Cancer Res 57:1301–1305, 1997

Mukhopadhyay A, Basu N, Ghatak N, Gujral PK: Anti-inflammatory and irritant activitiesof curcumin analogues in rats. Agents Actions 12:508–515, 1982

Pascale RM, Marras V, Simile MM, et al: Chemoprevention of rat liver carcinogenesis bys-adenosyl-L-methionine: a long-term study. Cancer Res 52:4979–4986, 1992

Pellegata NA, Antoniono RJ, Redpath JL, Stanbridge EJ: DNA damage and p53-mediatedcell cycle arrest: a reevaluation. Proc Natl Acad Sci USA 93:15209–15214, 1996

Rao TS, Basu N, Siddiqui HH: Anti-inflammatory activity of curcumin analogues. IndianJ Med Res 75:574–578, 1982

Rao CV, Rivenson A, Simi B, Reddy BS: Chemoprevention of colon cancer by dietarycurcumin. Ann New York Acade Sci 786:201–204, 1995a

Rao CV, Rivenson A, Simi B, Reddy BS: Chemoprevention of colon carcinogenesis bydietary curcumin, a naturally occurring plant phenolic compound. Cancer Res55:259–266, 1995b

Shao ZM, Dawson MI, Li XS, et al: p53 independent G0/G1 arrest and apoptosis inducedby a novel retinoid in human breast cancer cells. Oncogene 11:493–504, 1995

Sharma OP: Antioxidant activity of curcumin and related compounds. Biochem. Pharmcol25:1811–1812, 1976

Shibata MA, Maroulakou IG, Jorcyk CL, Gold LG, Ward JM, Green JE: p53-independentapoptosis during mammary tumor progression in C3 (1) /SV40 large T antigentransgenic mice: suppression of apoptosis during the transition from preneoplasia tocarcinoma. Cancer Res 56:2998–3003, 1996

Tata JR: Metamorphosis: an exquisite model for hormonal regulation of post-embryonicdevelopment. Biochem Soc Symposia 62:123–136, 1996

Toda S, Miyase T, Arichi H, Tanizawa H, Takino Y: Natural antioxidant III. Antioxidativecomponents isolated from rhizome of Curcuma Longa L. Chem Pharmaceut Bullet(Tokyo) 33:1725–1728, 1985

Villunger A, Egle A, Kos M, Hartmann BL, Geley S, Kofler R, Greil R: Drug-inducedapoptosis is associated with enhanced Fas (Apo-1/CD95) signaling in human T-acute Lymphatic leukemia cells. Cancer Res 57:3331–3334, 1997

Waldman T, Kinzler KW, Vogel S: p21 is nessary for the p53-medated G1 arrest in humancancer cells. Cancer Res 55:5187–5190, 1995

Yen HT, Chiang LC, Wen KH, Tsai CC, Yu HS: The expression of cytokines by anestablished basal cell carcinoma cell line (BCC-1/KMC) compared with culturednormal keratinocytes. Arch Dermatol Res 288:157–161, 1996

Yonish-Rouach E, Grunwald D, Wilder S, et al: p53-mediated cell death: relationship tocell cycle control. Mol Cell Biol 13:1415–1423, 1993

Yonish-Rouach E, Choisy C, Deguin V, Breugnot C, May E: the role of p53 as atranscription factor in the induction of apoptosis. Beharing Inst Mitt 97:60–71, 1996

Zhan Q, Alamo I, Yu K, et al: The apoptosis-associated gumma-ray response of BCL-X(L) depends on normal p53 function. Oncogene 13:2287–2293, 1996

Zhang W, Guo XY, Hu Gy Liu WB, Shay JW, Deisseroth AB: A temperature-sensitivemutant of human p53. Embo J 13:2353–2344, 1994