chemoprevention of chemically induced skin …...therapeutic discovery chemoprevention of chemically...

Therapeutic Discovery

Chemoprevention of Chemically Induced Skin Tumorigenesisby Ligand Activation of Peroxisome Proliferator–ActivatedReceptor-b/d and Inhibition of Cyclooxygenase 2

Bokai Zhu1,2, Robert Bai1, Mary J. Kennett1, Boo-Hyon Kang3, Frank J. Gonzalez4,and Jeffrey M. Peters1,2

AbstractLigand activation of peroxisome proliferator–activated receptor-b/d (PPARb/d) and inhibition of

cyclooxygenase-2 (COX2) activity by nonsteroidal anti-inflammatory drugs (NSAID) can both attenuate

skin tumorigenesis. The present study examined the hypothesis that combining ligand activation of

PPARb/d with inhibition of COX2 activity will increase the efficacy of chemoprevention of chemically

induced skin tumorigenesis over that observed with either approach alone. To test this hypothesis, wild-

type and Pparb/d-null mice were initiated with 7,12-dimethylbenz[a]anthracene (DMBA), topically treated

with 12-O-tetradecanoylphorbol-13-acetate to promote tumorigenesis, and then immediately treated with

topical application of the PPARb/d ligand GW0742, dietary administration of the COX2 inhibitor

nimesulide, or both GW0742 and nimesulide. Ligand activation of PPARb/d with GW0742 caused a

PPARb/d-dependent delay in the onset of tumor formation. Nimesulide also delayed the onset of tumor

formation and caused inhibition of tumor multiplicity (46%) in wild-type mice but not in Pparb/d-null mice.

Combining ligand activation of PPARb/d with dietary nimesulide resulted in a further decrease of tumor

multiplicity (58%) in wild-type mice but not in Pparb/d-null mice. Biochemical and molecular analysis of

skin and tumor samples show that these effects were due to the modulation of terminal differentiation,

attenuation of inflammatory signaling, and induction of apoptosis through both PPARb/d-dependent andPPARb/d-independent mechanisms. Increased levels and activity of PPARb/d by nimesulide were also

observed. These studies support the hypothesis that combining ligand activation of PPARb/d with

inhibition of COX2 activity increases the efficacy of preventing chemically induced skin tumorigenesis

as compared with either approach alone. Mol Cancer Ther; 9(12); 3267–77. �2010 AACR.

Introduction

Cyclooxygenase (COX) signaling pathways haveimportant roles in modulating skin carcinogenesis.COX is the central enzyme in prostanoid biosynthesisthat catalyzes the conversion of arachidonic acid to pros-taglandin H2, which is then converted to biologicallyactive lipids such as thromboxane, prostaglandin E2

(PGE2), and prostacyclin by different enzymes (1). Thereare 2 isoforms of COX, COX1 and COX2. While COX1 isconstitutively expressed, COX2 is induced by tumorpromoters, growth factors, and cytokines (2). Resultsfrom experimental animal models have established acausal relationship between COX2 and skin carcinogen-esis. For example, genetic disruption of both COX1 andCOX2 can prevent skin tumorigenesis (3) and nonster-oidal anti-inflammatory drugs (NSAID) that inhibit COXactivity as well as inhibit UV-induced and chemicallyinduced skin carcinogenesis (4–7). The proliferativeeffects of COX2 are due primarily to increased synthesisof prostaglandins, which directly influence cell growthafter binding to specific cell surface receptors, includingthe prostaglandin E, prostaglandin F, and prostaglandin Iclass of receptors (8, 9). For example, protumorigeniceffect of PGE2 can be mediated by the EP2 receptor(10). Although prostaglandins can mediate their biologi-cal effects through specific prostaglandin receptors suchas prostaglandin E, prostaglandin F, and prostaglandin I,they might also modulate the activities of peroxisomeproliferator–activated receptors (PPAR).

Authors' Affiliations: 1Department of Veterinary and Biomedical Sciencesand The Center for Molecular Toxicology and Carcinogenesis, and 2Inte-grative Biosciences Graduate Program, Huck Institutes for Life Sciences,The Pennsylvania State University, University Park, Pennsylvania; 3Pre-clinical Research Center, Chemon, Jeil-Ri, Yangji-Myeon, Cheoin-Gu,Yongin-Si, Gyeonggi-Do, Korea; and 4Laboratory of Metabolism, NationalCancer Institute, Bethesda, Maryland

Note: Supplementary material for this article is available at MolecularCancer Therapeutics Online (http://mct.aacrjournals.org/).

Corresponding Address: Jeffrey M. Peters, Department of VeterinaryScience and The Center for Molecular Toxicology and Carcinogenesis,The Pennsylvania State University, University Park, PA 16802. Phone:814-863-1387; Fax: 1-814-863-1696. E-mail: [email protected]

doi: 10.1158/1535-7163.MCT-10-0820

�2010 American Association for Cancer Research.

MolecularCancer

Therapeutics

www.aacrjournals.org 3267

on June 12, 2020. © 2010 American Association for Cancer Research. mct.aacrjournals.org Downloaded from

http://mct.aacrjournals.org/

Three distinct isoforms, PPARa, PPARb (alsoreferred to as PPARd or PPARb/d), and PPARg , exist,with essential roles in the regulation of adipogenesis,lipid metabolism, cell proliferation/apoptosis, celldifferentiation, inflammatory responses, and carcino-genesis (11–16). PPARs regulate these pathways bymodulation of gene expression through direct andindirect mechanisms. PPARb/d is found at very highlevels in the nucleus of epithelium including intestineand in keratinocytes (17). In the absence of ligands,nuclear PPARb/d can also be coimmunoprecipitatedwith its heterodimerization partner RXRa, suggestingthat PPARb/d has an important constitutive role inthe epithelium (17). Thus, it is not surprising thatimportant roles for PPARb/d have been observed inskin. For example, Pparb/d-null mice exhibit enhancedepidermal hyperplasia in response to phorbol estertreatment (18, 19) and exacerbated chemicallyinduced skin tumorigenesis in a 2-stage carcinogenbioassay as compared with wild-type mice (20), sug-gesting that PPARb/d inhibits epidermal cell prolif-eration in response to stimuli. Consistent with thisidea, PPARb/d-dependent inhibition of skin tumor-igenesis is found after topical application of thePPARb/d ligand GW0742 (21). The chemopreventiveeffects of ligand activation of PPARb/d are mediated inpart by the induction of unidentified target genes ornontranscriptional events that modulate terminal differ-entiation and inhibit cell proliferation and/or inhibition ofproinflammatory signaling (reviewed in refs. 11, 14,15).

Some reports suggest that NSAIDs attenuates carci-nogenesis by inhibiting PPARb/d expression and/oractivities, although this view has yet to be experimen-tally confirmed, and there are many inconsistencieswith this hypothesis in the literature (reviewed in refs.14, 15). For example, the hypothesis that NSAIDsinhibit cancer by decreasing PPARb/d expression/function is inconsistent with the observation thatPPARb/d expression following exposure to NSAIDsis either unchanged or increased in human cancer celllines (22). Furthermore, inhibition of chemicallyinduced skin tumorigenesis is found in both wild-typeand Pparb/d-null mice following treatment with theCOX1/COX2 inhibitor sulindac, suggesting thatNSAIDs mediate chemoprevention of chemicallyinduced skin tumorigenesis through PPARb/d-inde-pendent mechanisms (6). This is consistent with arecent report showing that combining COX2 inhibitionwith ligand activation of PPARb/d resulted inincreased efficacy in the inhibition of preexisting skintumor multiplicity (7). Collectively, these observationssuggest that combining these 2 therapeutic approacheswill increase the efficacy of chemoprevention as com-pared with either agent alone. Thus, the effect ofcombining COX2 inhibition and ligand activation ofPPARb/d on chemoprevention of skin carcinogenesiswas examined.

Materials and Methods

Two-stage chemical carcinogenesis bioassayFemale wild-type and Pparb/d-null mice on a C57BL/6

genetic background (19), 6 to 8 weeks of age, wereinitiated with 50 mg of 7,12-dimethylbenz[a]anthracene(DMBA; Sigma-Aldrich). One week after initiation, micewere treated topically with 5 mg of 12-O-tetradecanoyl-phorbol-13-acetate (TPA; NCI Chemical CarcinogenReference Standard Repository), 3 days per week for41 weeks. Mice from both genotypes were randomlydivided into 1 of the following 4 groups: a) control dietand topical application of acetone; b) control diet andtopical application of GW0742 (5 mmol/L); c) nimesulidediet (400 mg/kg) and topical application of acetone; or d)nimesulide diet (400 mg/kg) and topical application ofGW0742 (5 mmol/L). Because C57BL/6 mice weighing 20to 30 g typically consume approximately 4 g of food perday (23), the estimated dose of nimesulide ranged from 50to 80 mg/kg of body weight per day. The concentrationsof topical GW0742 and nimesulide in the diet were basedon previous work showing inhibition of chemicallyinduced skin tumorigenesis by GW0742 or nimesulidein related models (7, 21). After 42 weeks, mice wereeuthanized by overexposure to carbon dioxide. Tumorsamples were either fixed or snap-frozen in liquid nitro-gen for future analysis. Fixed tumor samples wereembedded in paraffin, sectioned and stained with hema-toxylin and eosin, and scored for benign or malignantpathology by 2 independent pathologists.

Short-term bioassayFemale wild-type and Pparb/d-null mice were accli-

mated to either a control or nimesulide diet (400 mg/kg)for 1 week and then treated topically with acetone orTPA dissolved in acetone (5 mg) followed 1 hour laterby topical application of either acetone or GW0742(5 mmol/L) every other day for a total of 3 applications.Micewere fed either the control or nimesulide diet duringthis periodof topicalGW0742 treatment.Micewere eutha-nized 6 hours after the last acetone or GW0742 treatment,and skin samples were obtained for RNA and proteinisolation.

Keratinocyte culturePrimary mouse keratinocytes were isolated from 2-day

postnatal wild-type and Pparb/d-null mice as describedpreviously (24). Keratinocytes were cultured in low-calcium (0.05 mmol/L) Eagle’s minimal essential med-ium with 8% chelexed fetal bovine serum at 37�C and 5%carbon dioxide.

Caspase 3/7 activity assaySkin samples were ground to a fine powder in liquid

nitrogen and then homogenized in buffer containing10 mmol/L of Tris (pH ¼ 7.5), 100 mmol/L of NaCl,1 mmol/L of EDTA, and 0.01% Triton-X100. For in vitroanalysis of caspase 3/7 activity, primary keratinocytes

Zhu et al.

Mol Cancer Ther; 9(12) December 2010 Molecular Cancer Therapeutics3268



were cultured as described earlier for 2 days beforetreatment with either DMSO, 1 mmol/L of GW0742,and 500 mmol/L of nimesulide or the combination of1 mmol/L of GW0742 and 500 mmol/L of nimesulidefor 24 hours. Cells were then trypsinized and lysed inthe Tris buffer as described earlier for 30 minutes on ice.Homogenates were centrifuged at 16,000 � g, and thesupernatant was used for analysis. Caspase 3/7 activitywas measured using a luminescent assay (Promega).

Western blot analysisPrimary keratinocytes were cultured as described ear-

lier for 2 days before treatment with either DMSO,1 mmol/L of GW0742, and 500 mmol/L of nimesulide orthe combination of 1 mmol/L of GW0742 and 500 mmol/Lof nimesulide for 24 hours. Cells were then trypsinizedand then lysed in buffer containing protease inhibitors.Samples were sonicated to facilitate cell lysis before cen-trifugation at 16,000 � g at 4�C for 30 minutes, and thesupernatant was used for Western blot analysis. Proteinfrom skin samples was isolated similarly with the samebuffer. Separation of proteins by electrophoresis, transferto membranes, and blocking was done as previouslydescribed (25). After overnight incubation at 4�C withthe primary antibody, membranes were incubated withbiotinylated secondary antibody (Jackson ImmunoRe-search Laboratories) for 1 hour at room temperaturefollowed by incubation with 125I-labeled streptavidin.Membranes were exposed to plates and the level of radio-activity was quantified with filmless autoradiographicanalysis. Hybridization signals for specific proteins werenormalized to the signal for the loading control lactatedehydrogenase (LDH) or actin. The following primaryantibodies were used: anti-PARP (Cell Signaling Technol-ogy), anti-K1 (Covance), anti-K10 (Covance), anti-PPARb/d (17), anti-actin (Rockland), and anti-LDH(Rockland). The ratio of cleavedPARP to uncleavedPARPwas calculated using Optiquant software.

RNA isolation and quantitative real-timePCR analysisTotal RNA was isolated from skin and tumor samples,

using TRIZOL reagent (Invitrogen). Reverse transcriptionand quantitative real-time PCR (qPCR) were carried outas previously described (25). Primers for keratin 1 (K1),keratin 10 (K10), angiopoetin-like protein 4 (Angptl4),interleukin 6 (IL6; Il6), and tumor necrosis factor-a(TNFa; Tnfa) have been previously described (7, 21,26, 27). The relative level of mRNA was normalized tothat of glyceraldehyde 3-phosphate dehydrogenase(Gapdh) or 18s RNA levels.

Statistical analysisThe significance of tumor incidence between each

treatment and genotype was determined by chi-squaretest for trend analysis (Prism 5.0, GraphPad Software,Inc.). Fisher’s exact test was used to determine the sig-nificance of the incidence of mice with keratoacanthomas

and/or squamous cell carcinomas (SCC). For all otheranalysis, a 1-tailed student t test was used.

Results

Ligand activation of PPARb/d and inhibition ofCOX2 enhances chemoprevention of chemicallyinduced skin tumorigenesis

Combining ligand activation of PPARb/d with COX2inhibition results in a modest decrease in multiplicity ofpreexisting tumors in a chemotherapeutic model (7).Because later stage tumors can be resistant to therapiesdesigned to regress tumor growth, the effect of combin-ing ligand activation of PPARb/d with COX2 inhibitionwas examined in a chemoprevention model. Markedchanges were observed in both genotypes (Fig. 1; Sup-plementary Fig. 2). The onset of papilloma formation wassooner and the incidence of papilloma was greater incontrol Pparb/d-null mice than in control wild-type miceprior toweek 16 of the 2-stage bioassay (P� 0.05; Fig. 1A),consistent with previous studies (7, 20, 21). Topical

A

B

C

Figure 1. Chemoprevention of chemically induced skin tumorigenesis bycombining ligand activation of PPARb/d and inhibition of COX2. Wild-type(þ/þ) and Pparb/d-null (�/�) mice were treated with topical GW0742(5 mmol/L), dietary nimesulide (400 mg/kg), or the combination of GW0742and nimesulide during a 42-week, 2-stage bioassay (initiation with DMBAand promotion with TPA) as described in Materials and Methods. A,incidence and onset of skin tumor formation. B, skin tumor multiplicity.C, average tumor size per mouse. Values represent the mean.

Skin Cancer Chemoprevention by Nimesulide and GW0742

www.aacrjournals.org Mol Cancer Ther; 9(12) December 2010 3269



application of the PPARb/d ligand GW0742, or dietarynimesulide, caused a delay in the onset of tumor forma-tion (P � 0.05; Fig. 1A). These effects were not found inPparb/d-null mice. Compared with control, combiningligand activation of PPARb/d with inhibition of COX2activity caused a delay in the onset of tumor formationin wild-type mice, and this effect was not found inPparb/d-null mice (Fig. 1A). In response to eitherGW0742 or nimesulide, the percentage of wild-type micewith skin tumors from week 11 to week 16 was lower butnot statistically different than control wild-type mice(Fig. 1A). However, in response to both GW0742 andnimesulide, the percentage of wild-type mice with skintumors from week 11 to week 16 was decreased ascompared with control wild-type mice (P � 0.05;Fig. 1A). These effects of GW0742, nimesulide, or thecombination of GW0742 and nimesulide were not foundin Pparb/d-null mice (Fig. 1A). Skin tumor multiplicitywas significantly greater (29%–30%) in control Pparb/d-null mice than in control wild-type mice from week 20until week 42 of the 2-stage bioassay (P � 0.05; Fig. 1B).Ligand activation of PPARb/d with GW0742 resulted indecreased (20%–40%) tumor multiplicity in wild-typemice during week 37 to week 42 of the bioassay, andthis effect was not found in Pparb/d-null mice (P � 0.05;Fig. 1B). Interestingly, skin tumor multiplicity was lower(24%–27%) in Pparb/d-null mice in response to topicalGW0742 from week 20 to week 30 of the bioassay than incontrol Pparb/d-null mice (P � 0.05; Fig. 1B). Dietarynimesulide caused a decrease (30%–46%) in tumor multi-plicity inwild-typemice during week 24 to week 42 of thebioassay, and this effect was not found in Pparb/d-nullmice (P � 0.05; Fig. 1B). The combination of topicalapplication of GW0742 and dietary nimesulide resultedin a marked decrease (57%–69%) of tumor multiplicityfrom week 22 onward in wild-type mice, and the effectwas greater than either GW0742 or nimesulide treatmentalone from week 21 to week 40 (P � 0.05; Fig. 1B).In Pparb/d-null mice, the combination of GW0742 withnimesulide caused a decrease (27%–42%) in tumor multi-plicity from week 20 to week 30 (P � 0.05; Fig. 1B).

Average tumor size was greater in the Pparb/d-nullmice than in wild-type mice, but this difference wasnot statistically significant (Fig. 1C). Topical GW0742or the combined treatment of topical GW0742 and dietarynimesulide did not cause a significant decrease in averagetumor size in either genotype (Fig. 1C). Dietary nimesu-lide caused a decrease in average tumor size in wild-typemice, and this effect was not observed in Pparb/d-nullmice (P � 0.05; Fig. 1C). Closer examination of the dis-tribution of the tumor size also revealed some strikingdifferences (Fig. 2). The percentage of control Pparb/d-nullmice with tumors in the 2- to 3-mm size rangewas greaterthan control wild-type mice (Fig. 2A). In addition, theaverage percentage of total tumors per mouse in the 1- to2-mm range was greater in control wild-typemice than incontrol Pparb/d-null mice, and this difference was con-sistent with a greater percentage of total tumors per

mouse in the 2- to 3-mm size range and greater than 5-mm size ranges in control Pparb/d-null mice as comparedwith control wild-type mice (Fig. 2B). In wild-type micefed nimesulide, the percentage of mice with tumors in the3- to 5-mm size range and the percentage of mice withtumors greater than 5-mm in size were significantly lessas compared with control wild-type mice (Fig. 2A). Simi-larly, the average percentage of total tumors per mousegreater than 5 mm was lower in wild-type mice fednimesulide as compared with control wild-type mice(Fig. 2B). The average percentage of total tumors permouse in the 1- to 2-, 2- to 3, and 3- to 5-mm rangewas similar in wild-type mice fed nimesulide as com-pared with control wild-type mice (Fig. 2B). Dietarynimesulide had no effect on the distribution of tumorswith different sizes in Pparb/d-null mice as comparedwith control Pparb/d-null mice (Figs. 2A and B). However,the average percentage of total tumors permouse in the 1-to 2-mm size range was lower in Pparb/d-null mice fednimesulide (Fig. 2B) than in wild-type mice fed nimesu-lide. This difference was due to the increase in theaverage percentage of total tumors per mouse in the 2-to 3-mm and greater than 5-mm size ranges in Pparb/d-null mice fed nimesulide as compared with similarlytreated wild-type mice (Fig. 2B). In wild-type mice trea-ted with GW0742, the percentage of mice with tumors inthe 2 to 3-mm size range was greater, whereas thepercentage of mice with tumors in the 3- to 5-mm andgreater than 5-mm size ranges was less than that ofcontrol wild-type mice (Fig. 2A). This effect was notfound in GW0742-treated Pparb/d-null mice (Fig. 2A).GW0742 had no effect on the average size distributionof total tumors per mouse in either genotype (Fig. 2B).The percentage of wild-type mice treated with bothtopical GW0742 and dietary nimesulide with tumors inall size ranges was markedly lower than that of controlwild-type mice, and this effect was not found in similarlytreated Pparb/d-null mice (Fig. 2A). The average percen-tage of total tumors per mouse in the 2- to 3-mm rangeand 3- to 5-mm ranges was lower in wild-type micetreated with both topical GW0742 and dietary nimesulidethan in control wild-type mice; these effects were notfound in similarly treated Pparb/d-null mice (Fig. 2B).

Themajority of representative skin lesions examined inall groups were squamous cell papillomas (data notshown). Skin lesions macroscopically suspected of beingSCC were examined for histopathology. Skin lesionsmacroscopically suspected of being SCC were notobserved in wild-type mice treated with nimesulide.For control, nimesulide-treated, GW0742-treated, andnimesulide þ GW0742-treated wild-type mice, 2 of 8, 0of 7, 3 of 10, and 2 of 10 mice, respectively, had lesionsmacroscopically suspected of being SCC. For control,nimesulide-treated, GW0742-treated, and nimesulide þGW0742-treated Pparb/d-null mice, 5 of 8, 3 of 10, 4 of10, and 5 of 10 mice, respectively, had lesions macro-scopically suspected of being SCC. Histopathologic ana-lysis revealed that these lesions were typically either

Zhu et al.




keratoacanthomas or SCC. A higher incidence of kera-toacanthomas was observed in control Pparb/d-null mice(3/8) than in control wild-typemice (1/8; SupplementaryFig. 3A). No keratoacanthomas were found in wild-typemice fed dietary nimesulide, but GW0742, nimesulide, ortheir combined treatment did not cause any statisticallysignificant changes in the incidence of keratoacanthomain either genotype (Supplementary Fig. 3A). The averagenumber of keratoacanthomas per mouse was comparablebetween both genotypes, although no keratoacanthomaswere noted in wild-type mice fed nimesulide (Supple-mentary Fig. 3B). Although 25% of control wild-typemice(2/8) had SCC, no SCCs were found in wild-type micetreated with dietary nimesulide or topical GW0742 andonly 10% of wild-type mice treated with both dietarynimesulide and topical GW0742 (1/10) had SCC

(Supplementary Fig. 3C). SCCs were found in 25% ofcontrol Pparb/d-null mice (2/8), 20% of nimesulide-trea-ted Pparb/d-null mice (2/10), none of GW0742-treated,and 40% of nimesulide þ GW0742-treated Pparb/d-nullmice (4/10; Supplementary Fig. 3C). None of these dif-ferences achieved statistical significance. The averagenumber of SCCs per mouse was comparable betweenboth genotypes, although no SCCs were observed innimesulide-treated or GW0742-treated wild-type miceor GW0742-treated Pparb/d-null mice (SupplementaryFig. 3D). One hemangioma was observed in 1 GW0742-treated Pparb/d-null mouse, and 1 malignant basal celltumor was found in 1 nimesulide þ GW0742-treatedPparb/d-null mouse (data not shown). Interestingly, poly-morphonuclear neutrophil infiltrates were more com-monly observed in Pparb/d-null mouse skin lesions

Figure 2. Skin tumor size followingligand activation of PPARb/d andinhibition of COX2. Wild-type(þ/þ) and Pparb/d-null (�/�) micewere treated with topical GW0742(5 mmol/L), dietary nimesulide(400 mg/kg), or the combination ofGW0742 and nimesulide during a42-week, 2-stage bioassay(initiation with DMBA andpromotion with TPA) as describedin Materials and Methods. A,incidence of mice with differenttumor sizes. These valuesrepresent the percentage of micewithin a given group that exhibitedskin tumors with the indicated sizerange. B, distribution of averagetumor size for each treatmentgroup. Mice within each treatmentwere used to calculate thepercentage of tumors of thatparticular size range for eachtreatment group. *, significantlydifferent from control wild-type,P � 0.05.

A

B





than in wild-type mouse lesions (Supplementary Fig. 4),consistent with past results (19). In addition, polymor-phonuclear neutrophil infiltrates were less common inskin lesions fromwild-typemice treatedwith nimesulide,GW0742, or the combined treatment but were morecommonly found in similarly treated Pparb/d-null mice.

Effect of GW0742 and nimesulide on terminaldifferentiation markers

Ligand activation of PPARb/d or inhibition of COXactivity can both induce terminal differentiation in pri-mary keratinocytes and skin (3, 30, 31). To determinewhether the enhanced efficacy of inhibiting chemicallyinduced skin tumorigenesis by combining GW0742 withnimesulide was due, in part, to modulation of terminaldifferentiation, expression of differentiation markers wasexamined. Dietary nimesulide, topical GW0742, and topi-cal GW0742 in combination with dietary nimesulideincreased expression of keratin 1 (K1) protein in wild-type mouse skin as compared with control, and this effectwas not found in Pparb/d-null mouse skin (Fig. 3). Dietarynimesulide or topical GW0742 did not alter expression ofkeratin 10 (K10) protein in mouse skin from either geno-type (Fig. 3). However, topical GW0742 in combinationwith dietary nimesulide increased expression of K10protein in wild-type mouse skin as compared with con-trol, and this effect was not found in Pparb/d-null mice(Fig. 3)

Effect of GW0742 and nimesulide on theinflammatory response

Inflammation can influence different stages of tumor-igenesis. Secretion of proinflammatory signaling mole-cules by immune and somatic cells such as TNFa andIL6 can act on cancer cells and promote tumor growthand malignant conversion (reviewed in ref. 32). TheNSAID nimesulide is known to attenuate inflammationby inhibiting COX2 activity and the subsequent produc-tion of arachidonic acid metabolites. In addition, ligand

activation of PPARb/d is also known to have anti-inflammatory activities in rodent and human models(reviewed in refs. 14, 15, 33). To determine whetherattenuation of inflammation could, in part, underlie theobserved inhibition of chemically induced skin tumor-igenesis, expression of the mRNA encoding 2 importantproinflammatory cytokines, TNFa and IL6, was exam-ined in both the tumor samples and mouse skin. Tumorsfrom Pparb/d-null mice from all treatment groups had ahigher level of Il6 mRNA (2- to 20-fold) and TnfamRNA(2- to 3-fold) than that of similarly treated wild-typemice (Fig. 4A and B). Dietary nimesulide caused asignificant decrease of both Il6 mRNA (53% lower)and Tnfa mRNA (79% lower) in tumors from wild-typemice but not in tumors from Pparb/d-null mice. Tumorsfrom wild-type mice treated only with topical GW0742or the combination of topical GW0742 and dietarynimesulide exhibited a decrease in mRNA encodingIl6 and Tnfa, but this change was only statisticallysignificant for Tnfa mRNA (73%–77% lower; Fig. 4B).No change in expression of Il6 or Tnfa mRNA wasfound in tumors from Pparb/d-null mice treated withtopical GW0742 or the combination of topical GW0742and dietary nimesulide (Fig. 4A and B). These data areconsistent with the presence of polymorphonuclearneutrophil infiltrates found more commonly in Pparb/d-null mice than in wild-type mice (SupplementaryFig. 4).

A short-term bioassay was also carried out usingwild-type and Pparb/d-null mice that were acclimatedto either a control or nimesulide diet for 1 week and thentreated with or without TPA followed 1 hour later witheither acetone (vehicle control) or GW0742. The ratio-nale for this approach is that TPA is known to increaseinflammatory signaling that could influence tumor pro-motion. Expression of Il6 mRNA was similar in bothcontrol wild-type and control Pparb/d-null mouse skin(Fig. 4C). Expression of Il6 mRNAwas increased in bothwild-type and Pparb/d-null mouse skin in response to

Figure 3. Expression of differentiation markers in skin following ligand activation of PPARb/d and inhibition of COX2. Protein was isolated fromwild-type (Pparb/dþ/þ) and Pparb/d-null (Pparb/d�/�) mouse skin following treatment with GW0742, nimesulide, or both GW0742 and nimesulide.Western blotswere carried out as described in Materials and Methods. Hybridization signals for keratin 1 (K1) and keratin 10 (K10) were normalized to actin.Values represent the mean � SEM. *, significantly different from control, P � 0.05.

Zhu et al.




TPA treatment but was markedly higher (8-fold vs. 106-fold, respectively) in Pparb/d-null mouse skin than inwild-type mouse skin (Fig. 4C). Expression of Il6 mRNAwas not influenced by GW0742, nimesulide, or GW0742þ nimesulide treatment in either control or TPA-treatedwild-type mouse skin or control Pparb/d-null mouseskin (Fig. 4C). Dietary nimesulide in Pparb/d-null miceresulted in lower Il6 mRNA (63% lower) followingtopical TPA, and a similar effect was also found inPparb/d-null mice that were treated with GW0742 andnimesulide (84% lower) following topical TPA treat-ment (Fig. 4C). Expression of Tnfa mRNA was similarin both control wild-type and control Pparb/d-nullmouse skin (Fig. 4D). Expression of Tnfa mRNA wasincreased in both wild-type and Pparb/d-null mouseskin following TPA treatment, and this effect wasgreater in Pparb/d-null mouse skin than in wild-typemouse skin (Fig. 4D). Expression of Tnfa mRNA wasnot influenced by GW0742, nimesulide, or combinedGW0742 and nimesulide treatment in either control orTPA-treated mouse skin from either genotype (Fig. 4D).Whereas the relative change in expression of TnfamRNA in response to TPA was approximately 2-foldin all groups of both genotypes, the relative fold changein expression of TnfamRNA was less (1.1-fold) in TPA-treated wild-type mouse skin treated with dietary nime-sulide and topical GW0742 than in similarly treated

control wild-type mouse skin, and this effect was notobserved in Pparb/d-null mice (Fig. 4D).

Effect of GW0742 and nimesulide on apoptosisBecause combining ligand activation of PPARb/d by

GW0742 with inhibition of COX2 activity by dietarynimesulide caused the most marked effect on tumormultiplicity, the effect of these treatments on apoptosiswas examined. There is compelling evidence that1mechanism bywhich nimesulide inhibits tumorigenesisis through the induction of apoptosis (reviewed inref. 34). In contrast, the effect of ligand activation ofPPARb/d on apoptotic signaling remains uncertain. Thisis due to conflicting studies suggesting that ligand acti-vation of PPARb/d causes proapoptotic signaling, anti-apoptotic signaling, or has no effect on apoptosis(reviewed in refs. 14, 15). Caspase 3/7 activity andpoly(ADP-ribose) polymerase (PARP) cleavage weremeasured to determine whether ligand activation ofPPARb/d with GW0742 and/or inhibition of COX2 bynimesulide modulate apoptosis in mouse skin and ker-atinocytes. In control wild-type and Pparb/d-null mouseskin, dietary nimesulide or topical GW0742 did not mod-ulate caspase 3/7 activity compared with control in eithergenotype (Fig. 5A). However, the combined treatment ofGW0742 with nimesulide caused an increase in caspase3/7 activity in wild-type mouse skin, and this effect was

A

C D

B

Figure 4. Effect of ligand activation of PPARb/d and/or inhibition of COX2 on Il6 and Tnfa mRNA in skin tumors and skin. Total RNA was isolated fromtumor samples or skin samples from wild-type (Pparb/dþ/þ) and Pparb/d-null (Pparb/d�/�) mice following treatment with GW0742, nimesulide, or bothGW0742 and nimesulide as described in Materials and Methods. The mRNA encoding Il6 (A and C) or Tnfa (B and D) was quantified from skin tumors (A and B)or mouse skin following acute TPA treatment (C and D) using qPCR and normalized to 18s mRNA. Values represent the mean � SEM. *, significantlydifferent from control, P � 0.05.





not seen in Pparb/d-null mouse skin. Because different celltypes can influence apoptosis, the effect of GW0742 andnimesulide was examined in primary keratinocytes fromwild-type and Pparb/d-null mice. Consistent with resultsobtained from the analysis of whole skin, GW0742 did notalter caspase 3/7 activity or PARP cleavage in eitherwild-type or Pparb/d-null primary keratinocyte (Fig. 5B–D). Incontrast, culturing primary keratinocytes with nimesu-lide increased apoptotic signaling in primary keratino-cytes in both genotypes, as evidenced by an increase incaspase 3/7 activity and PARP cleavage (Fig. 5B–D).Cotreatment of wild-type primary keratinocytes withnimesulide and GW0742 led to enhanced caspase 3/7activity and PARP cleavage as compared with thatobserved with either compound alone, but this increasewas not found in Pparb/d-null keratinocytes (Fig. 5B–D).

Effect of GW0742 and nimesulide on expression andfunction of PPARb/d

Nimesulide is not known to activate PPARb/d byacting as an agonist. One mechanism that may explainsome of the modest PPARb/d-dependent changes result-ing from nimesulide treatment is increased expressionand function of PPARb/d. Indeed, increased expressionof PPARb/d has been observed following exposure toNSAIDs including nimesulide (22). Thus, expression andfunction of PPARb/d were examined. Interestingly, diet-ary nimesulide, topical GW0742, and the combined treat-ment of GW0742 and nimesulide all increased expressionof Pparb/dmRNA in wild-type mouse skin (Fig. 6A). This

increase in expression was also found at the protein level,but the changes were not statistically significant (Fig. 6B).However, examination of expression of the PPARb/dtarget gene Angptl4 showed that dietary nimesulide,topical GW0742, and the combined treatment ofGW0742 and nimesulide all increased expression ofAngptl4 mRNA in wild-type mouse skin, and this effectwas not found in similarly treated Pparb/d-null mouseskin (Fig. 6C).

Discussion

Consistent with past studies (20, 21), chemicallyinduced skin tumorigenesis was exacerbated in Pparb/d-null mice as compared with wild-type mice, asassessed by differences in the onset tumor formation,the incidence of keratoacanthomas, and tumor multipli-city. Furthermore, ligand activation of PPARb/d inhib-ited chemically induced skin tumorigenesis throughPPARb/d-dependent mechanisms similar to results frompast studies (7, 21). Dietary nimesulide was also effectivefor chemoprevention as shown by decreased tumor mul-tiplicity and a decrease in tumor size distribution. Ascompared with dietary nimesulide or topical GW0742,the combination of dietary nimesulide and topicalGW0742 enhanced the chemopreventive activity of eitheragent alone, most notably by the prolonged markeddecrease in tumor multiplicity. Interestingly, the effectof GW0742, nimesulide, and the combined treatment ofnimesulide and GW0742 seems to be due, in part, to the

A

C D

B Figure 5. Effect of ligandactivation of PPARb/d and/orinhibition of COX2 on apoptoticsignaling. Total cytosolic proteinwas isolated from wild-type(Pparb/dþ/þ) and Pparb/d-null(Pparb/d�/�) mouse skin orprimary keratinocytes followingtreatment with GW0742,nimesulide, or both GW0742 andnimesulide as described inMaterials and Methods. Caspase3/7 activity in mouse skin (A) orprimary keratinocytes (B) wasnormalized to total proteinconcentration. C, representativeWestern blot showing the full-length and cleaved PARP proteinin response to control (DMSO-D),GW0742 (G), nimesulide (N), or thecombination of both GW0742 andnimesulide (G þ N). D,quantification of the ratio ofcleaved PARP to full-length PARPfrom Western blots. ND, notdetected. Values represent themean � SEM. *, significantlydifferent from control, P � 0.05.

Zhu et al.




modulation of PPARb/d-dependent and PPARb/d-inde-pendent mechanisms that influence differentiation,inflammation, and apoptosis.PPARb/d-dependent chemoprevention of chemically

induced skin tumorigenesis by GW0742 is likely due, inpart, to enhanced terminal differentiation, as observed inthe present study and previous reports (7, 21, 28, 29, 31).However, reduced expression of Tnfa mRNA was alsoobserved in skin tumors from GW0742-treated wild-typemice but not in similarly treated Pparb/d-null mice.Because activating PPARb/d is known to inhibit inflam-matory signaling (reviewed in ref. 33), it is possible thatinhibition of inflammatory signaling by PPARb/d alsocontributes to the mechanisms underlying the chemopre-ventive effects of GW0742 in this model. This is consistentwith the reduced accumulation of infiltrating poly-morphic neutrophils in GW0742-treated skin tumors.The mechanism underlying GW0742-dependent inhibi-tion of skin tumor multiplicity in both wild-type andPparb/d-null mice is uncertain, but this has been found

previously (21). One possible mechanism is that GW0742inhibits myeloperoxidase activity through direct enzymeinhibition (28). Because myeloperoxidase is found inneutrophils that accumulate during tumor promotionwith TPA, it is possible that GW0742 inhibits the activityof infiltrating neutrophils. Additional studies are neededto determine how GW0742 inhibits chemically inducedskin tumorigenesis through PPARb/d-independentmechanisms.

It is of interest to note that the chemopreventive effectof nimesulide was also dependent on PPARb/d. Indeed,delayed onset of skin tumorigenesis, reduced tumormultiplicity, and larger proportion of smaller versuslarger tumors were all observed in wild-type mice fedthe nimesulide diet, and these effects were diminished insimilarly treated Pparb/d-null mice. One possiblemechan-ism that may underlie this effect is the observed increasein PPARb/d function resulting from nimesulide treat-ment. Similar increases in PPARb/d expression and func-tion have also been observed in colon cancer cell lines,and these changes were also associated with the inhibi-tion of cell growth by nimesulide (22). The increase inPPARb/d expression by nimesulide could lead toenhanced terminal differentiation or anti-inflammatoryactivities. This is consistent with the observed increase inK1 expression and the inhibition of Il6 and Tnfa mRNAin skin tumors found in wild-type mice treated withdietary nimesulide but not in similarly treatedPparb/d-null mice. Although dietary nimesulide at theconcentration used in the present study is known toinhibit COX2 activity in mouse skin (7), expression ofCOX2 is also known to be higher in phorbol ester–treatedPparb/d-null mouse skin than in control (35). Further-more, inhibition of COX2 activity is found in wild-typemouse skin and this effect is diminished in Pparb/d-nullmouse skin (7). Thus, the observed PPARb/d-dependentchemoprevention by nimesulide could be due to differ-ences in stoichiometry between nimesulide and COX2.Further studies are needed to examine this possibility.

The efficacy of chemoprevention of chemically inducedskin tumorigenesis was greatest when nimesulide wascombined with GW0742. This was most evident by theprolonged inhibition of tumor multiplicity. Interestingly,this effect was due to both PPARb/d-dependent andPPARb/d-independent mechanisms. Inhibition of tumormultiplicity was observed in both wild-type andPparb/d-null mice from week 20 to week 32, after whichthis was found in wild-type but not in Pparb/d-null mice.This is of interest because dietary nimesulide was onlyeffective for inhibiting tumor multiplicity in wild-typemice but not in Pparb/d-null mice, whereas GW0742 waseffective in both genotypes during this time frame. Com-bining inhibition of COX2 activity with inhibition ofmyeloperoxidase activity could result in synergistic oradditive effects that contribute to the observed enhancedchemoprevention by both nimesulide and GW0742.However, results from the present study also show thatnimesulide effectively increases apoptotic signaling in

A

B

C

Rel

ativ

e P

parβ

/δm

RN

A

Figure 6. Effect of ligand activation of PPARb/d and/or inhibition of COX2on expression and activity of PPARb/d. Protein or mRNAwas isolated fromwild-type (Pparb/dþ/þ) or Pparb/d-null (Pparb/d�/�) mouse skin followingtreatment with GW0742, nimesulide, or both GW0742 and nimesulideas described in Materials and Methods. Expression of Pparb/d mRNA(A) or PPARb/d protein (B) was quantified by qPCR or Westernblotting, respectively. C, expression of the PPARb/d target Angptl4 mRNA.*, significantly different from control, P � 0.05.





mouse keratinocytes in both genotypes. This suggeststhat the observed PPARb/d-independent inhibition oftumor multiplicity resulting from the combination ofnimesulide and GW0742 could be influenced, in part,by increased apoptotic signaling. Why the observed che-moprevention becomes dependent on PPARb/d duringthe later stages of the bioassay is uncertain but could bedue to the combined effects on differentiation and anti-inflammatory activities that become more dominant dur-ing this period. Because of the striking enhanced chemo-prevention of chemically induced skin tumorigenesis bycombining ligand activation of PPARb/d with the inhibi-tion of COX2 activity, as compared with either agentalone, it will be of great interest to determine whetherthis approach can be used for UV-induced skin tumor-igenesis, a more predominant etiologic risk factor for skincancer in humans. Alternatively, whether inhibiting EPreceptor activity and activating PPARb/d will provide asafer approach, due to known issues associated with

COX2 inhibitors, should be of interest based on theseoriginal studies. Combining inhibition of COX2 signalingwith ligand activation of PPARb/d could provide a newapproach for chemoprevention of skin tumorigenesis.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

The authors thank Drs. Andrew Billin and Timothy Willson forproviding the GW0742 used for these studies.

Grant Support

This work supported in part by National Institutes of Health (CA124533,CA126826, CA141029, and CA140369 to J.M.Peters)

Received 08/31/2010; revised 10/01/2010; accepted 10/08/2010;published 12/14/2010.

References1. Smith WL. The eicosanoids and their biochemical mechanisms of

action. Biochem J 1989;259:315–24.2. Hla T, Ristimaki A, Appleby S, Barriocanal JG. Cyclooxygenase gene

expression in inflammation and angiogenesis. Ann NY Acad Sci1993;696:197–204.

3. Tiano HF, Loftin CD, Akunda J, Lee CA , Spalding J , Sessoms A , et al.Deficiency of either cyclooxygenase (COX)-1 or COX-2 alters epider-mal differentiation and reduces mouse skin tumorigenesis. CancerRes 2002;62:3395–401.

4. Fischer SM, Conti CJ, Viner J, Aldaz CM, Lubet RA. Celecoxib anddifluoromethylornithine in combination have strong therapeutic activ-ity against UV-induced skin tumors in mice. Carcinogenesis 2003;24:945–52.

5. Pentland AP, Schoggins JW, Scott GA, Khan KN, Han R. Reduction ofUV-induced skin tumors in hairlessmice by selective COX-2 inhibition.Carcinogenesis 1999;20:1939–44.

6. Kim DJ, Prabhu KS, Gonzalez FJ, Peters JM. Inhibition of chemically-induced skin carcinogenicity by sulindac is independent of peroxi-some proliferator-activated receptor-b/d (PPARb/d). Carcinogenesis2006;27:1105–12.

7. Bility MT, Zhu B, Kang BH, Gonzalez FJ, Peters JM. Ligand activationof peroxisome proliferator-activated receptor-b/d and inhibition ofcyclooxygenase-2 enhances inhibition of skin tumorigenesis. ToxicolSci 2010;113:27–36.

8. Hata AN, Breyer RM. Pharmacology and signaling of prostaglandinreceptors: multiple roles in inflammation and immune modulation.Pharmacol Ther 2004;103:147–66.

9. Tober KL, Thomas-Ahner JM, Kusewitt DF, Oberyszyn TM. Effects ofUVB on E prostanoid receptor expression in murine skin. J InvestDermatol 2007;127:214–21.

10. Sung YM, He G, Fischer SM. Lack of expression of the EP2 but notEP3 receptor for prostaglandin E2 results in suppression of skin tumordevelopment. Cancer Res 2005;65:9304–11.

11. Burdick AD, Kim DJ, Peraza MA, Gonzalez FJ, Peters JM. The role ofperoxisome proliferator-activated receptor-b/d in epithelial cell growthand differentiation. Cell Signal 2006;18:9–20.

12. Lee CH, Olson P, Evans RM. Minireview: lipid metabolism, metabolicdiseases, and peroxisome proliferator-activated receptors. Endocri-nology 2003;144:2201–7.

13. Peraza MA, Burdick AD, Marin HE, Gonzalez FJ, Peters JM. Thetoxicology of ligands for peroxisome proliferator-activated receptors(PPAR). Toxicol Sci 2006;90:269–95.

14. Peters JM, Gonzalez FJ. Sorting out the functional role(s) ofperoxisome proliferator-activated receptor-b/d (PPARb/d) in cell pro-liferation and cancer. Biochim Biophys Acta 2009;1796:230–41.

15. Peters JM, Hollingshead HE, Gonzalez FJ. Role of peroxisome-pro-liferator-activated receptor b/d (PPARb/d) in gastrointestinal tractfunction and disease. Clin Sci (Lond) 2008;115:107–27.

16. Lalloyer F, Staels B. Fibrates, glitazones, and peroxisome prolif-erator-activated receptors. Arterioscler Thromb Vasc Biol 2010;30:894–9.

17. Girroir EE, Hollingshead HE, He P, Zhu B, Perdew GH, Peters JM.Quantitative expression patterns of peroxisome proliferator-activatedreceptor-b/d (PPARb/d) protein in mice. Biochem Biophys Res Com-mun 2008;371:456–61.

18. Michalik L, Desvergne B, Tan NS, Basu-Modak S, Escher P,Rieusset J, et al. Impaired skin wound healing in peroxisomeproliferator-activated receptor (PPAR)a and PPARb mutant mice. JCell Biol 2001;154:799–814.

19. Peters JM, Lee SST, Li W, Ward JM, Gavrilova O, Everett C, et al.Growth, adipose, brain and skin alterations resulting from targeteddisruption of the mouse peroxisome proliferator-activated receptorb/d. Mol Cell Biol 2000;20:5119–28.

20. KimDJ, Akiyama TE, Harman FS, Burns AM , ShanW ,Ward JM , et al.Peroxisome proliferator-activated receptor b (d)-dependent regulationof ubiquitin C expression contributes to attenuation of skin carcino-genesis. J Biol Chem 2004;279:23719–27.

21. Bility MT, Devlin-Durante MK, Blazanin N, Glick AB, Ward JM, KangBH, et al. Ligand activation of peroxisome proliferator-activatedreceptor-b/d (PPARb/d) inhibits chemically-induced skin tumorigen-esis. Carcinogenesis 2008;29:2406–14.

22. Foreman JE, Sorg JM, McGinnis KS, Rigas B , Williams JL, ClapperML, et al. Regulation of peroxisome proliferator-activated receptor-b/d by the APC/b-CATENIN pathway and nonsteroidal antiinflamma-tory drugs. Mol Carcinog 2009;48:942–52.

23. Bachmanov AA, Reed DR, Beauchamp GK, Tordoff MG. Food intake,water intake, and drinking spout side preference of 28 mouse strains.Behav Genet 2002;32:435–43.

24. Dlugosz AA, Glick AB, Tennenbaum T, Weinberg WC, Yuspa SH.Isolation and utilization of epidermal keratinocytes for oncogeneresearch. Methods Enzymol 1995;254:3–20.

25. Palkar PS, Borland MG, Naruhn S, Ferry CH, Lee C, Sk UH, et al.Cellular and pharmacological selectivity of the PPARb/d antagonistGSK3787. Mol Pharmacol 2010;78:419–30.

Zhu et al.




26. Murray IA, Krishnegowda G, Dinatale BC, Flaveny C, Chiaro C, Lin JM,et al. Development of a selective modulator of aryl hydrocarbon (Ah)receptor activity that exhibit anti-inflammatory properties. Chem ResToxicol 2010;23:955–66.

27. Shan W, Palkar PS, Murray IA, McDevitt EI, Kennett MJ, Kang BH,et al. Ligand activation of peroxisome proliferator-activated receptorb/d (PPARb/d) attenuates carbon tetrachloride hepatotoxicity bydownregulating proinflammatory gene expression. Toxicol Sci2008;105:418–28.

28. Kim DJ, Bility MT, Billin AN, Willson TM, Gonzalez FJ, Peters JM.PPARb/d selectively induces differentiation and inhibits cell prolifera-tion. Cell Death Differ 2006;13:53–60.

29. Schmuth M, Haqq CM, Cairns WJ, Holder JC, Dorsam S, Chang S,et al. Peroxisome proliferator-activated receptor (PPAR)-b/d stimu-lates differentiation and lipid accumulation in keratinocytes. J InvestDermatol 2004;122:971–83.

30. Akunda JK, Lao HC, Lee CA, Sessoms AR, Slade RM, Langenbach R.Genetic deficiency or pharmacological inhibition of cyclooxygenase-1

or -2 induces mouse keratinocyte differentiation in vitro and in vivo.FASEB J 2004;18:185–7.

31. Westergaard M, Henningsen J, Svendsen ML, Johansen C, JensenUB, Schrder HD, et al. Modulation of keratinocyte gene expressionand differentiation by PPAR-selective ligands and tetradecylthioaceticacid. J Invest Dermatol 2001;116:702–12.

32. Grivennikov SI, Greten FR, Karin M. Immunity, inflammation, andcancer. Cell 2010;140:883–99.

33. Kilgore KS, Billin AN. PPARb/d ligands as modulators of theinflammatory response. Curr Opin Investig Drugs 2008;9:463–9.

34. Renard JF, Julemont F, de Leval X, Pirotte B. The use of nimesulideand its analogues in cancer chemoprevention. Anticancer AgentsMedChem 2006;6:233–7.

35. Kim DJ, Murray IA, Burns AM, Gonzalez FJ, Perdew GH, Peters JM.Peroxisome proliferator-activated receptor-b/d (PPARb/d) inhibits epi-dermal cell proliferation by down-regulation of kinase activity. J BiolChem 2005;280:9519–27.





2010;9:3267-3277. Mol Cancer Ther Bokai Zhu, Robert Bai, Mary J. Kennett, et al. and Inhibition of Cyclooxygenase 2δ

/βActivated Receptor-−Ligand Activation of Peroxisome Proliferator Chemoprevention of Chemically Induced Skin Tumorigenesis by

Updated version

http://mct.aacrjournals.org/content/9/12/3267

Access the most recent version of this article at:

Cited articles

http://mct.aacrjournals.org/content/9/12/3267.full#ref-list-1

This article cites 35 articles, 9 of which you can access for free at:

Citing articles

http://mct.aacrjournals.org/content/9/12/3267.full#related-urls

This article has been cited by 2 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://mct.aacrjournals.org/content/9/12/3267To request permission to re-use all or part of this article, use this link



http://mct.aacrjournals.org/content/9/12/3267.full#ref-list-1

http://mct.aacrjournals.org/content/9/12/3267.full#related-urls

http://mct.aacrjournals.org/cgi/alerts

mailto:[email protected]



chemoprevention of chemically induced skin …...therapeutic discovery chemoprevention of chemically...

Documents