Oxidative Stress and AP1 Activity in Tamoxifen Resistant Breast Tumors In Vivo

Oxidative Stress and AP-1Activity in Tamoxifen-Resistant Breast TumorsIn Vivo

Rachel Schiff, Praveen Reddy,Markku Ahotupa, EsterCoronado-Heinsohn, Matt Grim,Susan G. Hilsenbeck, RichardLawrence, Susan Deneke, RafaelHerrera, Gary C. Chamness,Suzanne A. W. Fuqua, Powel H.Brown, C. Kent Osborne

Background: Most breast cancers, eventhose that are initially responsive totamoxifen, ultimately become resistant.The molecular basis for this resistance,which in some patients is thought to in-volve stimulation of tumor growth bytamoxifen, is unclear. Tamoxifen in-duces cellular oxidative stress, and be-cause changes in cell redox state canactivate signaling pathways leading tothe activation of activating protein-1(AP-1), we investigated whether tam-oxifen-resistant growth in vivo is asso-ciated with oxidative stress and/or ac-tivation of AP-1 in a xenograft modelsystem where resistance is caused bytamoxifen-stimulated growth. Methods:Control estrogen-treated, tamoxifen-sensitive, and tamoxifen-resistantMCF-7 xenograft tumors were assessedfor oxidative stress by measuring levelsof antioxidant enzyme (e.g., superoxidedismutase [SOD], glutathione S-trans-ferase [GST], and hexose monophos-phate shunt [HMS]) activity, glutathi-one, and lipid peroxidation. AP-1 pro-tein levels, phosphorylated c-jun levels,and phosphorylated Jun NH2-terminalkinase (JNK) levels were examined bywestern blot analyses, and AP-1 DNA-binding and transcriptional activitieswere assessed by electrophoretic mobil-ity shift assays and a reporter gene sys-tem. All statistical tests are two-sided.Results: Compared with control estro-gen-treated tumors, tamoxifen resistanttumors had statistically significantly in-creased SOD (more than threefold; P =.004) and GST (twofold; P = .004) ac-tivity and statistically significantly re-duced glutathione levels (greater thantwofold; P<.001) and HMS activity (10-fold; P<.001). Lipid peroxides were not

significantly different between controland tamoxifen-resistant tumors. Weobserved no differences in AP-1 proteincomponents or DNA-binding activity.However, AP-1-dependent transcrip-tion (P = .04) and phosphorylated c-Junand JNK levels (P<.001) were statisti-cally significantly increased in thetamoxifen-resistant tumors. Conclu-sion: Our results suggest that the con-version of breast tumors to a tamoxi-fen-resistant phenotype is associatedwith oxidative stress and the subse-quent antioxidant response and withincreased phosphorylated JNK andc-Jun levels and AP-1 activity, whichtogether could contribute to tumorgrowth. [J Natl Cancer Inst 2000;92:1926–34]

Tamoxifen is the most prescribed drugfor the prevention and treatment of breastcancer (1,2). However, in breast cancerpatients, the disease eventually progresseswith the emergence of tamoxifen-resistanttumor cells. Tamoxifen is thought to actprimarily by competitive blockade of theestrogen receptor (ER) (3,4). Experimen-tal and clinical evidence suggests that animportant form of tamoxifen resistance isthe acquired ability of the tumor cells tobe stimulated, rather than inhibited, by thedrug after prolonged treatment (5–9).

We have developed an in vivo experi-mental model for tamoxifen resistance us-ing ER-positive MCF-7 human breastcancer cells grown in athymic nude mice(5). Tamoxifen treatment suppresses tu-mor growth for several months, butgrowth eventually resumes as the tumorsbecome stimulated by tamoxifen (5). Themechanisms underlying the conversionfrom growth suppression to growth stimu-lation are still unclear. However, severalstudies using the MCF-7 in vivo modelhave already discarded a number of po-tential mechanisms for growth stimula-tion by tamoxifen, including alteredtamoxifen uptake or metabolism (6,10)and lost or altered ER (11).

Another possible mechanism forgrowth stimulation by tamoxifen is an al-tered intracellular redox status leading toactivation of downstream signaling path-ways. Cellular redox status is a balancebetween the rate of pro-oxidant genera-tion, either exogenous or endogenous, andthe cellular enzymatic and nonenzymaticantioxidant capacities. A number of stud-ies (12–16) have shown that, dependingon the cellular microenvironment, tamoxi-

fen can affect the intracellular redox sta-tus as either a pro-oxidant or an antioxi-dant. For example, tamoxifen has the abil-ity to protect lipids, proteins, and DNAagainst oxidative damage (13) and can it-self be activated into reactive electro-philic metabolites (14). Moreover, evi-dence suggests that tamoxifen can inducephase I and phase II metabolizing en-zymes (15), which may contribute to itsbeneficial antioxidant activity but mayalso be responsible for its own activation.It is also known that changes in the intra-cellular redox status can lead to the acti-vation of important transcription factors,including activating protein-1 (AP-1)(17,18).

AP-1 is a heterodimeric transcriptionfactor that is composed of various mem-bers of the Jun and Fos families (19) andbinds to DNA at specific AP-1 bindingsites. AP-1 activity is determined in partby phosphorylation of these complexcomponents. Importantly, the transcrip-tional activity of c-Jun is increased byphosphorylation by the Jun NH2-terminalkinases (JNKs)/stress-activated proteinkinases (SAPKs), which are preferentiallyactivated by a variety of environmentaland cellular stresses (20), including oxi-dative stress (21). AP-1 activity can alsobe coregulated by protein–protein interac-tions between AP-1 and the ER (22). Fur-thermore, tamoxifen can function as anagonist in coactivating ER/AP-1 on pro-moters regulated by AP-1 sites (22–26).

The observation that AP-1 is importantin several mitogenic signaling pathways(27,28) led us to hypothesize that an in-crease in cellular AP-1 activity, perhapsresulting from tamoxifen-induced oxida-tive stress and the changes in intracellularredox status, may contribute to the devel-

Affiliations of authors: R. Schiff, P. Reddy, S. G.Hilsenbeck, R. Herrera, G. C. Chamness, S. A. W.Fuqua, P. H. Brown, C. K. Osborne, The BreastCenter and the Departments of Molecular and Cel-lular Biology and Medicine at Baylor College ofMedicine, Houston, TX; M. Ahotupa, MCA Re-search Laboratory, Department of Physiology, Uni-versity of Turku, Finland; E. Coronado-Heinsohn,M. Grim, S. Deneke (Department of Medicine), R.Lawrence (Institute for Drug Development), TheUniversity of Texas Health Science Center, San An-tonio.

Correspondence to: C. Kent Osborne, M.D., TheBreast Center at Baylor College of Medicine, 1 Bay-lor Plaza, MS: 600, Houston, TX 77030 (e-mail:[email protected]).

See “Notes” following References.”

© Oxford University Press

1926 REPORTS Journal of the National Cancer Institute, Vol. 92, No. 23, December 6, 2000

by guest on April 2, 2014

http://jnci.oxfordjournals.org/D

ownloaded from

http://jnci.oxfordjournals.org/


opment of tamoxifen-resistant tumorgrowth. In this study, we looked for evi-dence of oxidative stress and changes inAP-1 activity in the MCF-7 in vivo nudemouse model of tamoxifen resistance.

MATERIALS AND METHODS

Breast Cancer Cells

ER-positive MCF-7 human breast cancer cells(originally obtained from Dr. H. Degani at the Weiz-mann Institute of Science, Rehovoth, Israel) wereused for all experiments, unless otherwise stated.Tissue culture methods have been described previ-ously (29). Exponentially growing cultures weretreated with 12-O-tetradecanoylphorbol 13-acetate(TPA) (50 ng/mL) in the presence of serum-containing medium for the indicated times. A doxo-rubicin-resistant subclone of MCF-7, MCF-7 Adria(obtained from Dr. K. Cowan, National Cancer In-stitute, Bethesda, MD), known to express high levelsof c-Jun, was used as an internal standard in theelectrophoretic mobility shift assay. ER-negativehuman MDA-MB-435 breast cancer cells were cul-tured as described previously (30) and were used toobtain ER-negative xenograft tumors.

Athymic Nude Mouse Model ofTamoxifen-Stimulated Growth

Animal care was in accordance with institutionalguidelines. Four- to 6-week-old female ovariecto-mized BALB/c athymic nude mice (Harlan Sprague-Dawley Inc., Madison, WI) were given a subcuta-neous injection in the mammary fat pad of 5 × 106

MCF-7 cells or their transfectant derivatives (seebelow) and hormonally treated as described previ-ously (5,7). Estradiol pellets (0.25 mg; InnovativeResearch, Rockville, MD) were placed subcutane-ously in the interscapular region of the mice tostimulate tumor growth. When tumors reached a di-ameter of 8–12 mm (2–4 weeks), each mouse wasrandomly allocated to one of the following fourgroups: 1) control estrogen-treated, 2) removal ofthe estrogen pellet (i.e., estrogen withdrawal only),3) estrogen withdrawal and treatment with 500 �gof tamoxifen citrate (AstraZeneca, Macclesfield,U.K.) in peanut oil (subcutaneously injected dailyMonday through Friday), or 4) estrogen withdrawaland treatment with 5 mg of ICI 182,780 (AstraZen-eca) in castor oil (subcutaneously injected weekly).Tumor growth was assessed and tumor volumeswere measured as described previously (29).

Tumors were removed during estrogen treatment(control estrogen-treated tumors) and at varioustimes after the treatment with the antiestrogen drugstamoxifen and ICI 182,780. Antiestrogen-sensitivetumors are defined as those harvested during the first3 months of treatment when tumor growth is inhib-ited by tamoxifen. Thus, these tumors are defined astamoxifen sensitive (tamoxifenS) and ICI 182,780sensitive (ICIS), respectively. Tumors were usuallyharvested 2 weeks after the initiation of treatmentunless otherwise stated. For the DNA-binding stud-ies, tumors were also harvested at 1, 2, and 3 monthsafter tamoxifen treatment began. After 3–5 monthsof continuous treatment, growth resumes and tumor

progression is evident as an increase in tumor vol-ume. Tumors that first undergo growth inhibitionand then resume growth after prolonged antiestrogentreatment are defined as tamoxifen resistant (tamoxi-fenR) or ICI 182,780 resistant (ICIR). We haveshown previously that tamoxifen resistance in thismodel is due to tamoxifen-stimulated tumor growth(5,7).

In the estrogen-withdrawal group of mice, tumorswere removed after 2 weeks (estrogen withdrawalS)or several months later after tumor growth resumed(estrogen withdrawalR). Each tumor analyzed wasfrom a different mouse; tumor tissues were removedfrom each mouse and kept at −190 °C for lateranalyses.

Antioxidant Enzyme Assays

Antioxidant enzyme activities were assessed inthe control (estrogen-treated), tamoxifenS, andtamoxifenR groups (five tumors per group). Homog-enates of frozen tumors (20 [wt/vol]) were preparedin a 0.25 M sucrose solution (0 °C) with a Potter–Elvehjem glass–Teflon homogenizer driven by anelectric drill at 500 rpm with pulse homogenizationsix times at 20-second intervals. Homogenates werecentrifuged for 10 minutes at 10 000g at 4 °C toremove nuclei, mitochondria, and lysosomes, andthe supernatants were collected. The activities ofsuperoxide dismutase (SOD) (Cu/Zn form) and cata-lase were measured in the tumor homogenates, andthe activities of glutathione S-transferase (GST) andthe hexose monophosphate shunt (HMS) were mea-sured in the 10 000g supernatants. Enzyme activitieswere assayed with optimal incubation times and pro-tein concentrations to ensure the linearity of the re-action velocity. SOD activity (�g/mg protein) wasmeasured by luminometric detection of the superox-ide anion produced in the xanthine–xanthine oxidasesystem (31). Catalase activity (�g/mg protein) wasdetermined spectrophotometrically by measuring therate of disappearance of H2O2 (32). GST activity(nmol � min−1 � mg−1 protein) was measured spectro-photometrically with 1-chloro-2,4-dinitrobenzene asthe substrate (33). HMS activity (nmol � min−1 �

mg−1 protein) was assessed spectrophotometricallyby the production of reduced nicotinamide adeninedinucleotide phosphate (NADPH) with the use ofglucose 6-phosphate as the substrate (34); HMS ac-tivity represents the sum of glucose 6-phosphate de-hydrogenase and 6-phosphogluconate dehydroge-nase activities.

Lipid Peroxidation andGlutathione Levels

Lipid peroxidation was assessed in tumors fromthe control estrogen-treated, tamoxifenS, andtamoxifenR groups (five tumors per group) by thequantitation of the appearance of conjugated dienedouble bonds in lipid extracts (35). Briefly, lipidswere extracted from 10 000g supernatants with chlo-roform–methanol, dried under a nitrogen atmo-sphere, redissolved in cyclohexane, and analyzedspectrophotometrically at 233 nm to quantify dieneconjugation as detected by peak absorption. Lipidperoxidation is expressed as �Abs/mg protein,where �Abs is the difference in absorbance betweenthe sample and the cyclohexane solvent.

Reduced glutathione (GSH) and oxidized gluta-

thione (GSSG) levels (�mol/g wet wt) were deter-mined from the control estrogen-, tamoxifen-, estro-gen-withdrawal-, and ICI 182,780-treated tumors(four tumors per group). The glutathione content oftumor cytosol supernatants was estimated by high-performance liquid chromatography (36). Proteincontent was measured by the Bradford method (Bio-Rad Laboratories, Richmond, CA).

Protein Extraction and Western BlotAnalysis of Phosphorylated c-Jun andJNKs/SAPKs Forms

Phosphorylated forms of c-Jun and JNKs/SAPKswere analyzed with the use of PhosphoPlus antibodykits (Cell Signaling, Inc., Beverly, MA) according tothe manufacturer’s directions. Briefly, pellets of invitro cultured cells or ground, frozen tumor powdersfrom the control estrogen-treated, tamoxifenS, andtamoxifenR groups (at least eight tumors per group)were manually homogenized in lysis buffer (CellSignaling, Inc.). After microcentrifugation at 14 000gfor 30 minutes at 4 °C, supernatants were collected,and the protein concentration was determined. Ali-quots (25 �g) of protein from each sample wereseparated under denaturing conditions by electro-phoresis with 10% polyacrylamide gel containingsodium dodecyl sulfate and transferred by electrob-lotting onto nitrocellulose membranes (Schleicher &Schuell, Inc., Keene, NH). The blots were stainedwith StainAll Dye (Alpha Diagnostic International,Inc., San Antonio, TX) to confirm uniform proteintransfer. Separate membranes were then reacted witheither c-Jun- or JNK/SAPK-specific PhosphoPlusantibodies that specifically recognize the phosphor-ylated forms. The membranes were stripped and re-blotted for total c-Jun and JNK/SAPK with antibod-ies that recognize the respective proteinsindependently of their phosphorylation status. Blotswere developed by chemiluminescence (Cell Signal-ing, Inc.). Each sample was analyzed twice on sepa-rate immunoblots. Bands were quantified by densi-tometric scanning of developed films with the use ofthe Image 1.61/ppc software program of the Na-tional Institutes of Health, Bethesda, MD.

Electrophoretic Mobility Shift Assay

Nuclear extracts from cells or individual tumors(five tumors per group) were prepared as describedpreviously (28) with minor changes. Briefly, cells orground-up, frozen tumor powders were disrupted inlysis buffer (i.e., 10 mM HEPES, 1 mM EDTA, 60mM KCl, 0.5 mM dithiothreitol [DTT], 0.5% Noni-det P-40 [NP-40], and protease inhibitors [1 mMphenylmethyl sulfonyl fluoride, 0.4 �M aprotinin,and 10 �M leupeptin]), and nuclei were isolated bymicrocentrifugation at 2500g at 4 °C for 10 minutes.The isolated nuclei were lysed by three cycles offreezing/thawing in nuclear suspension buffer (250mM Tris [pH 7.8], 400 mM KCl, 0.5 mM DTT, 20%glycerol, and protease inhibitors) and microcentri-fuged at 16 000g at 4 °C for 10 minutes, and thesupernatants (nuclear protein extracts) were col-lected. Protein concentrations were determined withthe use of the Bradford method. Electrophoretic mo-bility shift assays were performed with 10 �g ofnuclear protein extract in a 20-�L reaction mixturecontaining 20 mM HEPES (pH 7.9), 40 mM KCl, 1

Journal of the National Cancer Institute, Vol. 92, No. 23, December 6, 2000 REPORTS 1927



ownloaded from



mM EGTA [i.e., ethylene glycol-bis(�-aminoethylether)-N,N�-tetraacetic acid], 1 mM phenylmethylsulfonyl fluoride, 0.5 mM DTT, 1% glycerol, 2 �gof poly (dI-dC), and 100 pg of [�-32P]adenosine tri-phosphate end-labeled, double-stranded oligo-nucleotide probe containing an AP-1 binding site (5�

CTAGTGATGAGTCAGCCGGATC 3�; Strata-gene, La Jolla, CA; the AP-1 binding site is under-lined). Reaction mixtures were incubated for30 minutes at room temperature. For oligonucleotidecompetition experiments, the reaction mixtures werepreincubated with a 100-fold excess of unlabeled,cold oligonucleotide probes containing an AP-1 orSp-1 binding site (Stratagene) for 20 minutes beforethe addition of the radioactive probes. The reactionmixtures were separated on 5% nondenaturing poly-acrylamide gels at 4 °C. After electrophoresis anddrying, gels were autoradiographed, and shiftedbands were quantified on a PhosphoImager (Mo-lecular Dynamics, Sunnyvale, CA). The AP-1 DNA-binding activity of individual tumors was normal-ized between gels with the use of an internalstandard extract of MCF-7 Adria cells. Electropho-retic mobility shift assays of the same extracts wererepeated a minimum of two times.

AP-1/CAT Reporter Constructs,Stable Transfection, andChloramphenicol AcetyltransferaseAssay

To study AP-1-dependent gene transcription, weused a chloramphenicol acetyltransferase (CAT) re-porter system as described previously (37,38). Thereporter construct, Col-TREx5/TKCAT (TREx5),contains five copies of a consensus AP-1 bindingsite (5� ATGAGTCAG 3�) upstream of the herpessimplex virus-thymidine kinase (HSV-tk) minimalpromoter. The same site is also a synthetic consen-sus TPA responsive element (TRE). We also used acontrol construct, TRE�-72/TKCAT (TRE�-72),upstream to position −109 of the HSV-tk promoter,that contains a point mutation in the AP-1 site (5�

TGGAGTCAG 3�) that eliminates both basal andinducible AP-1 activities (37,38).

To generate stable transfection clones, we platedMCF-7 cells at a density of 8 × 105/100 mm2. After24 hours, the cells were cotransfected with 10 �g ofthe TREx5 or TRE�-72 AP-1/CAT reporter con-structs and with 1 �g of the pSV2neo selection plas-mid (Clontech Laboratories, Inc., Palo Alto, CA)containing the neomycin resistance gene with theuse of the LipofectAMINE reagent (Life Technolo-gies, Inc. [GIBCO BRL], Rockville, MD), accordingto the manufacturers directions. G418-resistant (600�g/mL) individual clones were screened for TPA-inducible AP-1 transcriptional activity (50 ng/mLTPA for 8 hours), and CAT assays were performedas described below.

TPA-inducible TREx5 clones and noninducibleTRE�-72 clones were grown in nude mice andtreated with tamoxifen as described above. Tumorswere harvested from the control estrogen-treated,tamoxifenS, and tamoxifenR groups (four to eightmice per group). Tumor samples were homogenized,and extracts were made as described previously (39).Protein concentrations were determined, and thesame amount of protein from each clone was ana-lyzed for CAT activity by thin-layer chromatogra-phy and quantitated on a PhosphoImager (Ambis,

San Diego, CA). We calculated the relative CATactivity by dividing the CAT activity of the tamoxi-fen-treated tumors by the CAT activity of the controlestrogen-treated tumors of the clone. All CAT as-says were repeated two times with each sample.

Statistical Methods

Differences in the mean values of tumor antioxi-dant enzyme activities (SOD, HMS, and GST), glu-tathione levels, DNA-binding activities, and AP-1-dependent CAT activities among the treatmentgroups were analyzed by Student’s t test as pairwisecomparisons with respect to the control estrogen-treated group or to the tamoxifenS group, as speci-fied. Bonferroni’s correction was used to adjust formultiple comparisons. The mean values of westernblot band densities of phosphorylated c-Jun andJNKs/SAPKs were compared by two-way analysisof variance. For purposes of statistical analyses, datawere transformed by taking logarithms to equalizevariances. All P values are two-sided.

RESULTS

Antioxidant Enzyme Activity

We have developed and studied anexperimental in vivo model that mimicsthe clinical scenario of acquired resis-tance of breast cancer to tamoxifen orother endocrine therapies, such as estro-gen-withdrawal or ICI 182,780 treatment(7). In the nude mouse model, ER-positive MCF-7 xenograft tumors are es-tablished in the presence of estrogen (con-trol estrogen-treated tumors), and theestrogen is withdrawn before the start ofany antiestrogen treatment. Tumor growthis initially inhibited, but it eventually re-sumes after continued treatment. Tumorsinhibited by tamoxifen are defined astamoxifen-sensitive (tamoxifenS) tumors,and tumors that resumed growth are de-fined as tamoxifen-resistant (tamoxifenR)tumors. We have shown previously thattamoxifen stimulates tamoxifenR tumorgrowth (5,7). In parallel, tumors inhibitedby estrogen withdrawal or ICI 182,780are defined as estrogen-withdrawalS orICIS tumors, respectively, and tumors thatresumed growth are defined as estro-gen-withdrawalR or ICIR tumors, respec-tively.

To investigate the relationship betweentamoxifen resistance and altered cellularredox status, we first measured the activ-ity levels of different antioxidant enzymesin control estrogen-treated, tamoxifenS,and tamoxifenR tumors grown in nudemice (Table 1). SOD activity was notablyincreased by tamoxifen, with a greaterthan fourfold increase in tamoxifenS tu-mors (P<.001) and a greater than three-fold increase in tamoxifenR tumors (P �

.004). Catalase activity was not statisti-cally significantly different among the tu-mor groups. GST activity was statisticallysignificantly increased in the tamoxifenR

tumors (twofold; P � .004) relative to thecontrol estrogen-treated or the tamoxifenS

tumors. TamoxifenR tumors also had in-creased protein levels of GST-Pi, a mem-ber of the GST enzyme complex, as mea-sured by western blot analysis (data notshown).

The most striking effect of tamoxifenwas a profound inhibition of the produc-tion of NADPH by the HMS (Table 1). IntamoxifenS tumors, the HMS activity wasstatistically significantly reduced to lessthan half the activity detected in the con-trol estrogen-treated tumors (P<.001). IntamoxifenR tumors, the HMS activity wasfurther statistically significantly reducedby another fourfold to 10-fold in total(P<.001). Thus, tamoxifen treatment andthe development of tamoxifenR byMCF-7 breast tumors in vivo are associ-ated with changes in antioxidant activi-ties, suggesting that the tumors are expe-riencing oxidative stress.

Oxidative Stress

Lipid peroxidation is a process gener-ated by the effect of reactive oxygen spe-cies and occurs when the antioxidantdefense mechanisms are being over-whelmed (40). Glutathione, via its redoxcycling, is a potent antioxidant that pro-vides cells with a substantial degree ofprotection against oxidative stress (41).Because decreased HMS activity andNADPH levels would be expected togreatly reduce glutathione levels, we nextmeasured lipid peroxidation and glutathi-one levels (Table 1). Lipid peroxidationwas statistically significantly higher in thetamoxifenS tumors (P � .016) but thenreturned to baseline levels after resistanceemerged (Table 1). In contrast, levels ofboth GSH and GSSG were markedly re-duced in the tamoxifenR tumors as com-pared with the control estrogen-treated tu-mors (greater than twofold; P<.001 and P� .02, respectively). GSH levels were de-creased only slightly in tamoxifenS tu-mors compared with those in control es-trogen-treated tumors, and GSSG levelswere reduced by 1.7-fold in tamoxifenS

tumors compared with those in control es-trogen-treated tumors. Importantly, therewere marked and statistically significantdifferences in the GSH levels between thetamoxifenS and tamoxifenR tumors(P<.001). These results suggest that the




ownloaded from



development of tamoxifen resistance inMCF-7 breast tumors in vivo is associatedwith increased susceptibility to oxidativestress and depletion of glutathione levelsas the tumors attempt to respond to theoxidative stress.

To confirm that the marked decrease intotal glutathione levels in resistant tumorswas specific for tamoxifen, we measuredglutathione levels in tumors from micetreated either with estrogen withdrawal orwith the estrogen antagonist ICI 182,780.Although, compared with the levels incontrol estrogen-treated tumors, glutathi-one levels decreased slightly in tumorsthat had acquired resistance to estrogenwithdrawal and in ICIS or ICIR tumors,glutathione levels in the tamoxifenR tu-mors were still substantially lower thanthose in all other tumors.

To learn whether tamoxifen-inducedoxidative stress is mediated though theER, we measured total glutathione levelsin xenograft tumors of ER-negativeMDA-MB-435 human breast cancer cells.Tamoxifen does not inhibit in vivo growthof this cell line, and long-term experi-ments are not possible because the tumorsbecome too large. Tamoxifen treatmentfor 29 days resulted in no reduction inglutathione levels in these tumors (datanot shown), suggesting that the effectof tamoxifen might be mediated throughthe ER.

Tamoxifen Resistance and AP-1DNA-Binding Activity

Because oxidative stress has beenshown to activate the AP-1 transcription

factor, which increases cell proliferation(27,28), we explored the effect of tamoxi-fen on AP-1 composition and activity inMCF-7 tumors in vivo. AP-1 is a heterodi-meric transcription factor complex that iscomposed of proteins from the Jun andFos families. Comparison of control es-trogen-treated, tamoxifenS, and tamoxif-enR tumors revealed that there were noapparent changes in messenger RNA orprotein levels of the Jun and Fos familymembers c-Jun, JunD, JunB, Fra-1, andc-Fos (data not shown).

Using electrophoretic mobility shift as-says, we next compared AP-1 DNA-binding activity in nuclear extracts fromcontrol estrogen-treated, tamoxifenS, andtamoxifenR tumors (Fig. 1, A). Tamoxif-enS nuclear extracts were made from es-tablished tumors 2 weeks, 1 month, 2months, and 3 months after tamoxifentreatment began. Although there was amodest reduction in AP-1 DNA-bindingactivity during the first 2 months oftamoxifen treatment, there were no statis-tically significant differences in AP-1DNA-binding activity between control es-trogen-treated tumors and tamoxifen-treated tumors at any time (Fig. 1, A).DNA-binding assays performed withAP-1 oligonucleotides containing one orfive copies of the AP-1 site producedcomparable results (data not shown). Us-ing specific antibodies to various AP-1family members (c-Jun, JunD, Fra-1, andc-Fos), we determined the composition ofthe AP-1 DNA-binding complexes. Wefound no appreciable differences in thecomposition of the DNA-binding com-

plexes among any of the control ortamoxifen-treated groups (data notshown).

Although tamoxifen treatment did notappear to change AP-1 DNA-binding ac-tivity, we detected a fourfold decreasein AP-1 DNA-binding activity in estro-gen-withdrawalS tumors (P<.001) andin estrogen-withdrawalR tumors (P<.001) (Fig. 1, B). We observed a 3.5-fold decrease in AP-1 DNA-bindingactivity in ICIR tumors (P<.001). Oneaberrant tumor in the ICIS treatmentgroup did not show a decrease in AP-1DNA-binding activity. Thus, althoughtamoxifen, estrogen withdrawal, andICI 182,780 all inhibited tumor growthin vivo, only tamoxifen maintainedAP-1 DNA-binding activity at initial lev-els.

Association of Increased AP-1Transcriptional TransactivatingActivity With Development ofTamoxifenR Growth

AP-1 DNA-binding activity does notnecessarily reflect the transcriptional ac-tivity of this transcription factor complex(42). To determine whether AP-1 DNA-binding activity is a direct reflection of itsability to promote transcription in thissystem, we developed stable transfectantsof MCF-7 cells expressing CAT reportergene constructs containing either fivecopies of a synthetic AP-1 DNA-bindingsite (TREx5) or a control mutated AP-1DNA-binding site lacking basal and in-ducible AP-1 activities (TRE�-72) up-

Table 1. Antioxidant enzyme activity, lipid peroxidation, and glutathione levels in hormonally treated MCF-7 tumors*

Tumorgroup

Mean (95% confidence interval)

SOD,†�g/mg protein

Catalase,†�g/mg protein

GST,† nmol/minper mg protein

HMS,† nmol/minper mg protein

LPO,†�Abs/mg protein

GSH,‡�mol/g protein

GSSG,‡�mol/g protein

E2 1.25 1.74 0.93 12.10 717 2.725 0.434(0.29–2.21) (1.33–2.15) (0.60–1.26) (10.34–13.86) (488–946) (2.692–2.758) (0.279–0.589)

TamS 5.01 2.38 0.97 5.42 2217 2.391 0.248(3.76–6.26) (1.69–3.07) (0.77–1.17) (4.66–6.18) (1243–3191) (2.322–2.460) (0.179–0.317)

TamR 4.34 2.61 1.83 1.30 874 1.297 0.156(3.54–5.15) (1.77–3.45) (1.50–2.16) (1.03–1.57) (504–1244) (1.281–1.313) (0.103–0.209)

−E2R ND ND ND ND ND 2.162 0.563

(2.101–2.223) (0.357–0.769)ICIS ND ND ND ND ND 2.002 0.301

(1.975–2.029) (0.242–0.360)ICIR ND ND ND ND ND 2.139 0.250

(2.053–2.225) (0.179–0.321)

*SOD � superoxide dismutase (Cu/Zn form); GST � glutathione S-transferase; HMS � hexose monophosphate shunt; LPO � lipid peroxidation; GSH �

reduced glutathione; GSSG � oxidized glutathione; ND � not done; E2 � control estrogen tumors; TamS � tamoxifen-sensitive tumors; TamR � tamoxifen-resistant tumors; −E2

R � estrogen-withdrawal resistant tumors; ICIS � ICI 182,780-sensitive tumors; ICIR � ICI 182,780-resistant tumors.†Tumors from five mice per treatment group. �Abs � difference in absorbance between the sample and the cyclohexane solvent.‡Tumors from four mice per treatment group.




ownloaded from



stream to a minimal promoter (37). TPAinduced CAT activity in vitro in thetwo TREx5 clones tested but not in theTRE�-72 clone (Fig. 2, A). The stabletransfectants were then grown as tumorsin the nude mice and treated with tamoxi-fen. In tumors grown from the controlTRE�-72 clone, the low basal CAT ac-tivity gradually declined during tamoxi-fen treatment (Fig. 2, B and C). In con-

trast, a statistically significant increasein CAT activity was detected in tamoxif-enR tumors in both TREx5 clones (two-fold to threefold; P � .04 and P � .006for TREx5 clones 1 and 2, respectively).The finding that AP-1 transcriptional ac-tivity increased at the time of tamoxifenR

growth was reproducible in two differentclones and in two independent in vivo ex-periments.

Hyperphosphorylation of c-Jun inTamoxifenR Tumors

AP-1 transcriptional activity is in-creased by phosphorylation at two spe-cific serine residues in the c-Jun compo-nent of AP-1 (42). These residues, Ser 63and Ser 73, are phosphorylated by theJNKs (43,44). Furthermore, JNK activitycan be increased by various stresses (20),including oxidative stress (21) and/or glu-tathione depletion (45), conditions ob-served in our model with tamoxifen treat-ment, especially during tamoxifenR growth.

Because we observed an induction ofoxidative stress in the tumors during theemergence of tamoxifen resistance, wedetermined the phosphorylation status ofc-Jun from control estrogen-treated,tamoxifenS, and tamoxifenR tumors (Fig.3, A). No statistically significant changeswere detected between the control estro-gen-treated tumors and tamoxifenS tu-mors. However, the tamoxifenR tumors(of both untransfected MCF-7 tumors andstable TREx5 transfectants) had statisti-cally significantly higher levels of phos-phorylated c-Jun (greater than twofold;P<.001), even after correcting for minorchanges in total c-Jun levels.

Because the JNKs are the major ki-nases known to phosphorylate c-Jun atSer 63 and Ser 73, we next measuredphosphorylated (i.e., active) JNK familymembers in the control estrogen-treated,tamoxifenS, and tamoxifenR tumors (Fig.3, B). We saw that the tamoxifenR tumorscontained statistically significantly higherlevels of both the 46-kd and 54-kd phos-pho-JNK forms compared with thetamoxifenS tumors (>1.8-fold and 1.5-fold; P<.001 and P � .008, respectively,for 46-kd and 54-kd phospho-JNK forms,after correcting for minor changes in thelevel of total JNK). Thus, both increasedphospho-c-Jun levels and increased JNKactivity accompanied the increase in AP-1-dependent transcription in the tamoxif-enR tumors.

DISCUSSION

The emergence of tamoxifen resistanceis a major problem in the treatment ofbreast cancer, and understanding themechanisms by which resistance arisescould have major clinical implications forpreventing or circumventing it. Our re-sults show that the development of ac-quired tamoxifen resistance of xenograftMCF-7 tumors in vivo is associated withboth increased susceptibility to oxidative

Fig. 1. Activating protein-1 (AP-1) DNA-binding activity of control estrogen-, tamoxifen-, estrogen-withdrawal-, and ICI 182,780-treated MCF-7 breast cancer xenograft tumors. Nuclear extracts from tumorgroups (five mice per group) were analyzed by electrophoretic mobility shift assay with the use of an AP-1oligonucleotide probe. A) Tumors from control estrogen-treated (E2), tamoxifen-sensitive (the groups in-cluded 2 weeks [2w], 1 month [1m], 2 months [2m], and 3 months [3m] tamoxifen treatment during thegrowth-inhibited [TamS] phase), and tamoxifen-resistant (during the growth-stimulated [TamR] phase) xe-nograft mice were harvested and analyzed for AP-1 DNA-binding activity. B) Tumors from control estrogen-treated (E2,), estrogen-withdrawal-treated (−E2), or estrogen-withdrawal-treated/ICI 182,780-treated xeno-graft mice at the sensitive, growth-inhibited phase (2 weeks of treatment, −E2

S or ICIS) and at the resistant,growth-stimulated phase (−E2

R or ICIR) were harvested and analyzed for AP-1 DNA-binding activity. Toppanels: representative electrophoretic mobility shift assay gels of the tumor nuclear extracts. For TamS, only2 weeks of treatment is shown. An internal standard extract of MCF-7 Adria cells (MCF-7/Adr) was includedin each panel for normalization of the AP-1 binding reactions between gels. An excess of unlabeled AP-1oligonucleotide (+ AP-1) was used to competitively inhibit specific AP-1 binding, and a nonspecific Sp-1oligonucleotide (+ Sp-1) was added as a further negative control for this specificity. Arrows point to thespecific AP-1 complexes, and the nonspecific band is designated “ns.” Bottom panels: relative AP-1DNA-binding activity in the treated tumors. Specific AP-1 complexes were quantitated on a PhosphoImager,and the DNA-binding activity of individual tumors was normalized between gels. AP-1 DNA-bindingactivity in the tumor groups was calculated relative to the control estrogen-treated group, and calculatedmeans (±95% confidence intervals) were analyzed statistically by Student’s t test as pairwise comparisonsversus the control estrogen-treated group. * � P<.001. Electrophoretic mobility shift assays with the samesamples were repeated a minimum of two times.




ownloaded from



stress and increased AP-1 activity.Tamoxifen affects the intracellular redoxstatus in breast tumors, increases lipidperoxidation, and induces the activity of a

number of antioxidant enzymes. Pro-longed tamoxifen treatment resulted in tu-mors that were tamoxifen resistant andgrowth stimulated and had a reduced an-

tioxidant cellular capacity, as evidencedby a striking decrease in HMS activityand a marked depletion of glutathione.These oxidative changes after prolonged

Fig. 2. Activating protein-1 (AP-1) transcriptional activity in tamoxifen-treatedtumors. MCF-7 breast cancer cells were stably transfected with chloramphenicolacetyltransferase (CAT) constructs containing either five copies of a consensusAP-1 site (TREx5 clones) or a control mutated AP-1 site (TRE�-72 construct).A) 12-O-Tetradecanoylphorbol 13-acetate (TPA) induction of TRE�-72 andTREx5 clones in vitro. Cells of two TREx5 clones (clone TREx5/1 and cloneTREx5/2) and one TRE�-72 clone were left untreated (−) or were treated withTPA (50 ng/mL) for 8 hours, and extracts were analyzed by the CAT assay, asdescribed in the “Materials and Methods” section. In B and C, AP-1-dependentCAT activity is indicated in tamoxifen-treated tumors. The transfectant cloneswere injected into nude mice and treated with estrogen and tamoxifen. Tumors(four to eight mice per group) were harvested for CAT analysis after establish-ment in the presence of estrogen (E2), 2 weeks after tamoxifen treatment began(during the tamoxifen-inhibited growth phase, TamS), and at the appearance oftamoxifen-resistant growth (TamR). All tumors were from individual mice. CATassays were performed with the use of the same amount of protein extract forindividual tumors of each clone. B) CAT assay of clones TRE�-72 and TREx5.The TRE�-72 clone has low basal CAT activity, so its autoradiographs wereexposed longer than those of the TREx5 clones. All of the tumors analyzed fromthe stable transfectants are shown. CAT assays of the same samples were re-peated twice. C) Quantitation of relative CAT activity of the TRE/CAT clonesshown in B. The CAT assays were quantitated on a PhosphoImager, and therelative CAT activity in the tumor groups was calculated relative to that of thecontrol estrogen-treated group of each clone. TamoxifenR calculated means(±95% confidence intervals) were analyzed statistically by Student’s t test com-pared with the tamoxifenS group. CP � chloramphenicol.

Fig. 3. Phosphorylation of c-Jun and Jun NH2-terminal kinase (JNK) in thetamoxifen-treated tumors. Protein extracts (25 �g) of control estrogen-treated(E2), tamoxifen-sensitive (TamS), and tamoxifen-resistant (TamR) MCF-7 breastcancer xenograft tumors were analyzed by western blot analysis with antibodiesthat recognize the phosphorylated forms of the proteins. A) Blots probed withanti-phospho c-Jun antiserum specific for Ser 63 (top panel) and anti-total c-Junantiserum, which recognizes c-Jun independently of its phosphorylation status(bottom panel). Controls were MCF-7 cells either untreated (−) or treated invitro (+) with 12-O-tetradecanoylphorbol 13-acetate (TPA) at a dose of 50ng/mL for 1 hour. B) Blots probed with anti-phospho JNK (54 kd and 46 kd)antiserum recognizes Thr 183 and Tyr 185 phosphorylation. Arrows point to the54-kd and 46-kd JNK family members. A representative gel is shown of two orthree tumors per group. Mean values of western blot band densities of phos-phorylated c-Jun and JNKs/stress-activated protein kinases (SAPKs) were com-pared by two-way analysis of variance. Ab � antibody; M.W. � molecularweight.




ownloaded from



treatment appear to be specific to tamoxi-fen because this marked reduction in glu-tathione levels and, consequently, in-creased susceptibility to oxidative stresswere not found in tumors from micetreated with estrogen, with estrogen with-drawal, or with the pure antiestrogen ICI182,780.

A remaining open question is to whatextent all of the oxidative changes are ERdependent. It is known that tamoxifen’santiangiogenic effects and its potential toinduce oxidative stress are, at least in part,ER independent (46–48), although therole of the two ER subtypes is still unclear(49). A recent study (50) demonstratingthat the oncostatic action of melatonin onbreast cancer cells is mediated by in-creased glutathione levels and is restrictedto ER-positive cells suggests a link be-tween cellular redox state and ER. In ad-dition, the lack of oxidative stress in ER-negative MDA-MB-435 tumors fromtamoxifen-treated mice and preliminaryin vitro data also suggest that tamoxifen-induced oxidative stress in MCF-7 tumorsmay be ER mediated.

Because changes in the cellular redoxstatus can lead to the induction of AP-1and because AP-1 is important in a vari-ety of mitogenic signaling pathways(27,28), we studied this transcription fac-tor complex in our tamoxifenR in vivomodel. We found that the AP-1 transcrip-tional activity was increased as tumorsprogressed to a tamoxifenR phenotype. Itis possible that the observed increase inthe antioxidant enzyme GST, whose ex-pression is regulated by AP-1 (51), re-flects a parallel increase in AP-1-dependent transcription. Other reportshave shown that prolonged tamoxifentreatment dramatically affects the expres-sion of a number of phase II enzymes,such as GST (52–54), probably throughthe antioxidant response element con-tained in the promoter of these genes (54)and that these genes may also be regu-lated by AP-1 (51). In agreement with ourfinding, Astruc et al. (25) also have re-ported that prolonged tamoxifen treat-ment markedly increases the cellular re-sponse to inducers of AP-1.

Although we found increased AP-1transcriptional activity in the tamoxifenR

tumors, neither expression of Jun and Fosfamily members nor AP-1 DNA-bindingactivity was altered. These findings aresimilar to those in other reports (27,28).In contrast, Dumont et al. (55) reportedthat the progression of MCF-7 tumor cells

(MCF-WES cells) to a tamoxifen-stimulated/tamoxifen-resistant phenotypeis associated with increased AP-1 DNA-binding activity. However, MCF-WEStamoxifenR tumors have a markedly de-creased ER content, and although the tu-mors are still estrogen sensitive, they areglobally resistant to all antiestrogens. Incontrast, the tamoxifenR tumors in thisstudy, which have increased AP-1 tran-scriptional activity, express high levels ofER, remain estrogen dependent (5), andare growth inhibited by pure steroidal an-tiestrogens (6,7). Thus, although bothtypes of tamoxifenR tumors share a com-mon regulatory pathway, namely AP-1,they may differ in how the pathway isactivated. Whether these differences inAP-1 activation account for the diver-gence in their cellular phenotype remainsto be investigated.

Our data also demonstrate that in-creases in JNK activity and c-Jun phos-phorylation are associated with thetamoxifenR phenotype. Increased c-Junphosphorylation by tamoxifen may poten-tiate c-Jun transcriptional activity (42,56)and could also enhance the agonistic ef-fects of tamoxifen at AP-1 sites, as pro-posed by Webb et al. (22). JNK can beactivated by diverse stress stimuli (20,42),including oxidative stress (21). Oxidativestress can alter the level of intracellularglutathione, which can be a key regulatorfor the induction of JNKs (45). Becauseboth oxidized and reduced glutathionelevels are markedly decreased in ourtamoxifenR tumors but not in our tamoxi-fenS tumors, our cumulative data suggestthe possibility that chronic tamoxifen ad-ministration leads to oxidative stress anda reduction in glutathione levels followedby activation of JNK and increased AP-1activity. The increased AP-1 activitycould then provide the tumor cells with asufficient growth stimulus to offset anygrowth-inhibitory effects of the antiestro-gen mediated via the ER pathway. Recentdata demonstrating a substantial increasein JNK activity in tamoxifen-resistant hu-man breast tumors (57) further supportthis hypothesis.

Although our data show that prolongedtamoxifen treatment of MCF-7 tumors innude mice results in oxidative stress andincreased AP-1 activity, to conclude thatthese pathways mediate tamoxifenR

growth, it will be necessary to determinewhether increasing or inhibiting AP-1 ac-tivity or cellular sensitivity to oxidativestress will affect the emergence of the

tamoxifenR phenotype. In this context, itis interesting that overexpression of c-Junin MCF-7 cells results in a tamoxifenR

phenotype both in vitro and in vivo (58).Confirmation of the role of oxidativestress and AP-1 in the development oftamoxifenR could provide new strategiesto delay or even to prevent this importantclinical problem.

REFERENCES

(1) Early Breast Cancer Trialists’ CollaborativeGroup. Systemic treatment of early breast can-cer by hormonal, cytotoxic, or immunetherapy. 133 randomised trials involving31,000 recurrences and 24,000 deaths among75,000 women. Lancet 1992;339:71–85.

(2) Fisher B, Costantino JP, Wickerham DL, Red-mond CK, Kavanah M, Cronin WM, et al.Tamoxifen for prevention of breast cancer: re-port of the National Surgical Adjuvant Breastand Bowel Project P-1 Study. J Natl CancerInst 1998;90:1371–88.

(3) Katzenellenbogen BS, Fang H, Ince BA, Pak-del F, Reese JC, Wooge CH, et al. William L.McGuire Memorial Symposium. Estrogen re-ceptors: ligand discrimination and antiestrogenaction. Breast Cancer Res Treat 1993;27:17–26.

(4) Katzenellenbogen BS, Montano MM, Le GoffP, Schodin DJ, Kraus WL, Bhardwaj B, et al.Antiestrogens: mechanisms and actions in tar-get cells. J Steroid Biochem Mol Biol 1995;53:387–93.

(5) Osborne CK, Coronado E, Allred DC, WiebeV, DeGregorio M. Acquired tamoxifen resis-tance: correlation with reduced breast tumorlevels of tamoxifen and isomerization of trans-4-hydroxytamoxifen. J Natl Cancer Inst 1991;83:1477–82.

(6) Osborne CK, Jarman M, McCague R, Coro-nado EB, Hilsenbeck SG, Wakeling AE. Theimportance of tamoxifen metabolism intamoxifen-stimulated breast tumor growth.Cancer Chemother Pharmacol 1994;34:89–95.

(7) Osborne CK, Coronado-Heinsohn EB, Hilsen-beck SG, McCue BL, Wakeling AE, McClel-land RA, et al. Comparison of the effects of apure steroidal antiestrogen with those oftamoxifen in a model of human breast cancer.J Natl Cancer Inst 1995;87:746–50.

(8) Gottardis MM, Jordan VC. Development oftamoxifen-stimulated growth of MCF-7 tu-mors in athymic mice after long-term anties-trogen administration. Cancer Res 1988;48:5183–7.

(9) Legault-Poisson S, Jolivet J, Poisson R, Ber-etta-Piccoli M, Band PR. Tamoxifen-inducedtumor stimulation and withdrawal response.Cancer Treat Rep 1979;63:1839–41.

(10) Wolf DM, Langan-Fahey SM, Parker CJ, Mc-Cague R, Jordan VC. Investigation of themechanism of tamoxifen-stimulated breast tu-mor growth with nonisomerizable analogues oftamoxifen and metabolites. J Natl Cancer Inst1993;85:806–12.

(11) Osborne CK, Fuqua SA. Mechanisms of




ownloaded from



tamoxifen resistance. Breast Cancer Res Treat1994;32:49–55.

(12) Day BW, Tyurin VA, Tyurina YY, Liu M,Facey JA, Carta G, et al. Peroxidase-catalyzedpro- versus antioxidant effects of 4-hy-droxytamoxifen: enzyme specificity and bio-chemical sequelae. Chem Res Toxicol 1999;12:28–37.

(13) Wei H, Cai Q, Tian L, Lebwohl M. Tamoxifenreduces endogenous and UV light-induced oxi-dative damage to DNA, lipid and protein invitro and in vivo. Carcinogenesis 1998;19:1013–8.

(14) Dehal SS, Kupfer D. Evidence that the cat-echol 3,4-Dihydroxytamoxifen is a proximateintermediate to the reactive species binding co-valently to proteins. Cancer Res 1996;56:1283–90.

(15) Nuwaysir EF, Dragan YP, McCague R, MartinP, Mann J, Jordan VC, et al. Structure–activityrelationships for triphenylethylene antiestro-gens on hepatic phase-I and phase-II enzymeexpression. Biochem Pharmacol 1998;56:321–7.

(16) Ahotupa M, Hirsimaki P, Parssinen R, MantylaE. Alterations of drug metabolizing and anti-oxidant enzyme activities during tamoxifen-induced hepatocarcinogenesis in the rat. Carci-nogenesis 1994;15:863–8.

(17) Meyer M, Schreck R, Baeuerle PA. H2O2 andantioxidants have opposite effects on activationof NF-kappa B and AP-1 in intact cells: AP-1as secondary antioxidant-responsive factor.EMBO J 1993;12:2005–15.

(18) Sen CK, Packer L. Antioxidant and redoxregulation of gene transcription. FASEBJ 1996;10:709–20.

(19) Angel P, Karin M. The role of Jun, Fos and theAP-1 complex in cell-proliferation and trans-formation. Biochim Biophys Acta 1991;1072:129–57.

(20) Kyriakis JM, Banerjee P, Nikolakaki E, Dai T,Rubie EA, Ahmad MF, et al. The stress-activated protein kinase subfamily of c-Jun ki-nases. Nature 1994;369:156–60.

(21) Guyton KZ, Liu Y, Gorospe M, Xu Q, Hol-brook NJ. Activation of mitogen-activated pro-tein kinase by H2O2. Role in cell survival fol-lowing oxidant injury. J Biol Chem 1996;271:4138–42.

(22) Webb P, Lopez GN, Uht RM, Kushner PJ.Tamoxifen activation of the estrogen receptor/AP-1 pathway: potential origin for the cell-specific estrogen-like effects of antiestrogens.Mol Endocrinol 1995;9:443–56.

(23) Uht RM, Anderson CM, Webb P, Kushner PJ.Transcriptional activities of estrogen and glu-cocorticoid receptors are functionally inte-grated at the AP-1 response element. Endocri-nology 1997;138:2900–8.

(24) Paech K, Webb P, Kuiper GG, Nilsson S, Gus-tafsson J, Kushner PJ, et al. Differential ligandactivation of estrogen receptors ERalpha andERbeta at AP1 sites. Science 1997;277:1508–10.

(25) Astruc ME, Chabret C, Bali P, Gagne D, PonsM. Prolonged treatment of breast cancer cellswith antiestrogens increases the activating pro-tein-1-mediated response: involvement of theestrogen receptor. Endocrinology 1995;136:824–32.

(26) Webb P, Nguyen P, Valentine C, Lopez GN,Kwok GR, McInerney E, et al. The estrogenreceptor enhances AP-1 activity by two distinctmechanisms with different requirements for re-ceptor transactivation functions. Mol Endocri-nol 1999;13:1672–85.

(27) Philips A, Chalbos D, Rochefort H. Estradiolincreases and anti-estrogens antagonize thegrowth factor-induced activator protein-1 ac-tivity in MCF7 breast cancer cells without af-fecting c-fos and c-jun synthesis. J Biol Chem1993;268:14103–8.

(28) Chen TK, Smith LM, Gebhardt DK, Birrer MJ,Brown PH. Activation and inhibition of theAP-1 complex in human breast cancer cells.Mol Carcinog 1996;15:215–26.

(29) Osborne CK, Hobbs K, Clark GM. Effect ofestrogens and antiestrogens on growth of hu-man breast cancer cells in athymic nude mice.Cancer Res 1985;45:584–90.

(30) Lemieux P, Oesterreich S, Lawrence JA, SteegPS, Hilsenbeck SG, Harvey JM, et al. Thesmall heat shock protein hsp27 increasesinvasiveness but decreases motility of breastcancer cells. Invasion Metastasis 1997;17:113–23.

(31) Laihia JK, Jansen CT, Ahotupa M. Lucigeninand linoleate enhanced chemiluminescent as-say for superoxide dismutase activity. FreeRadic Biol Med 1993;14:457–61.

(32) Beers RF, Sizer IW. A spectrophotometricmethod of measuring the breakdown of hydro-gen peroxide by catalase. J Biol Chem 1952;195:133–40.

(33) Habig WH, Pabst MJ, Jacoby WB. GlutathioneS-transferase. The first enzymatic step in mer-capturic acid formation. J Biol Chem 1974;249:7130–9.

(34) Glock GE, McLean P. Further studies on theproperties and assay of glucose-6-phosphatedehydrogenase and 6-phosphogluconate dehy-drogenase of rat liver. Biochem J 1953;55:400–8.

(35) Corongiu FP, Lai M, Milia A. Carbon tetra-chloride, bromotrichloromethane and ethanolacute intoxication. New chemical evidence forlipid peroxidation in rat tissue microsomes.Biochem J 1983;212:625–31.

(36) Livesey JC, Reed DJ. Measurement of gluta-thione–protein mixed disulfides. Int J RadiatOncol Biol Phys 1984;10:1507–10.

(37) Angel P, Imagawa M, Chiu R, Stein B, ImbraRJ, Rahmsdorf HJ, et al. Phorbol ester-inducible genes contain a common cis elementrecognized by a TPA-modulated trans-actingfactor. Cell 1987;49:729–39.

(38) Szabo E, Preis LH, Brown PH, Birrer MJ. Therole of jun and fos gene family members in12-O-tetradecanoylphorbol-13-acetate inducedhemopoietic differentiation. Cell Growth Dif-fer 1991;2:475–82.

(39) Adrian GS, Rivera EV, Adrian EK, Lu Y,Buchanan J, Herbert DC, et al. Lead suppresseschimeric human transferrin gene expression intransgenic mouse liver. Neurotoxicology 1993;14:273–82.

(40) Mylonas C, Kouretas D. Lipid peroxidationand tissue damage. In Vivo 1999;13:295–309.

(41) Meister A, Anderson ME. Glutathione. AnnuRev Biochem 1983;52:711–60.

(42) Karin M. The regulation of AP-1 activity bymitogen-activated protein kinases. J Biol Chem1995;270:16483–6.

(43) Derijard B, Hibi M, Wu IH, Barrett T, Su B,Deng T, et al. JNK1: a protein kinase stimu-lated by UV light and Ha-Ras that binds andphosphorylates the c-Jun activation domain.Cell 1994;76:1025–37.

(44) Kallunki T, Su B, Tsigelny I, Sluss HK, Deri-jard B, Moore G, et al. JNK2 contains a speci-ficity-determining region responsible for effi-cient c-Jun binding and phosphorylation.Genes Dev 1994;8:2996–3007.

(45) Wilhelm D, Bender K, Knebel A, Angel P. Thelevel of intracellular glutathione is a key regu-lator for the induction of stress-activated signaltransduction pathways including Jun N-terminal protein kinases and p38 kinase by al-kylating agents. Mol Cell Biol 1997;17:4792–800.

(46) Gagliardi AR, Hennig B, Collins DC. Anties-trogens inhibit endothelial cell growth stimu-lated by angiogenic growth factors. AnticancerRes 1996;16:1101–6.

(47) Gundimeda U, Chen ZH, Gopalakrishna R.Tamoxifen modulates protein kinase C via oxi-dative stress in estrogen receptor-negativebreast cancer cells. J Biol Chem 1996;271:13504–14.

(48) Ferlini C, Scambia G, Marone M, Distefano M,Gaggini C, Ferrandina G, et al. Tamoxifen in-duces oxidative stress and apoptosis in oestro-gen receptor-negative human cancer cell lines.Br J Cancer 1999;79:257–63.

(49) Barton M. Vascular effects of estrogens: rapidactions, novel mechanisms, and potential thera-peutic implications. Chung Kuo Yao Li HsuehPao 1999;20:682–90.

(50) Blask DE, Wilson ST, Zalatan F. Physiologicalmelatonin inhibition of human breast cancercell growth in vitro: evidence for a glutathione-mediated pathway. Cancer Res 1997;57:1909–14.

(51) Bergelson S, Pinkus R, Daniel V. Induction ofAP-1 (Fos/Jun) by chemical agents mediatesactivation of glutathione S-transferase and qui-none reductase gene expression. Oncogene1994;9:565–71.

(52) Nuwaysir EF, Daggett DA, Jordan VC, PitotHC. Phase II enzyme expression in rat liver inresponse to the antiestrogen tamoxifen. CancerRes 1996;56:3704–10.

(53) Fei P, Matwyshyn GA, Rushmore TH, KongAN. Transcription regulation of rat glutathioneS-transferase Ya subunit gene expression bychemopreventive agents. Pharm Res 1996;13:1043–8.

(54) Montano MM, Katzenellenbogen BS. The qui-none reductase gene: a unique estrogen recep-tor-regulated gene that is activated by anties-trogens. Proc Natl Acad Sci U S A 1997;94:2581–6.

(55) Dumont JA, Bitonti AJ, Wallace CD, Bau-mann RJ, Cashman EA, Cross-DoersenDE. Progression of MCF-7 breast cancer cellsto antiestrogen-resistant phenotype is accom-panied by elevated levels of AP-1 DNA-binding activity. Cell Growth Differ 1996;7:351–9.

(56) Bannister AJ, Oehler T, Wilhelm D, Angel P,




ownloaded from



Kouzarides T. Stimulation of c-Jun activity byCBP: c-Jun residues Ser63/73 are required forCBP induced stimulation in vivo and CBPbinding in vitro. Oncogene 1995;11:2509–14.

(57) Johnston SR, Lu B, Scott GK, Kushner PJ,Smith IE, Dowsett M, et al. Increased activatorprotein-1 DNA binding and c-Jun NH2-terminal kinase activity in human breast tumorswith acquired tamoxifen resistance. Clin Can-cer Res 1999;5:251–6.

(58) Smith LM, Wise SC, Hendricks DT, SabichiAL, Bos T, Reddy P, et al. cJun overexpressionin MCF-7 breast cancer cells produces a tumor-igenic, invasive and hormone resistant pheno-type. Oncogene 1999;18:6063–70.

NOTES

Supported in part by Public Health Service grantP50CA58183, a Breast Cancer SPORE grant fromthe National Cancer Institute, National Institutes ofHealth, Department of Health and Human Services.

Manuscript received February 2, 2000; revisedSeptember 11, 2000; accepted September 20, 2000.




ownloaded from



Oxidative Stress and AP1 Activity in Tamoxifen Resistant Breast Tumors In Vivo

Documents

Transcript of Oxidative Stress and AP1 Activity in Tamoxifen Resistant Breast Tumors In Vivo