Breast Cancer Research and Treatment 79: 399–407, 2003.© 2003 Kluwer Academic Publishers. Printed in the Netherlands.

Report

Role of CRE-binding protein (CREB) in aromatase expressionin breast adipose

Mariam Sofi, Morag J. Young, Theodora Papamakarios, Evan R. Simpson, and Colin D. ClynePrince Henry’s Institute of Medical Research, Clayton, VC., Australia

Key words: adipose, aromatase, breast cancer, cyclic AMP, CREB, CYP19

Summary

Estrogen biosynthesis from C19 steroids is catalyzed by aromatase cytochrome P450. Aromatase is expressedin breast adipose tissue through the use of a distal, cytokine-responsive promoter (promoter I.4). Breast tumors,however, secrete soluble factors that over-stimulate aromatase expression through an alternative proximal cAMP-responsive promoter, promoter II. We have mapped the cAMP-responsive regions of promoter II by transienttransfection of 3T3-L1 preadipocytes with aromatase promoter II reporter genes. 5′ deletion and mutation anal-yses identified two cAMP response element (CRE)-like sequences (CRE1 and CRE2) that were essential forcAMP-induced promoter II activity. Electrophoretic mobility shift analysis demonstrated that CRE binding protein(CREB) bound to each element, and that this interaction was enhanced in the presence of cAMP. Quantificationof CREB mRNA expression in adipose tissue from normal and tumor bearing breast adipose tissue revealed thatCREB expression is approximately five times higher in tumor bearing than in normal breast adipose tissue. Thus,the over expression of aromatase in adipose tissue surrounding breast tumors could arise through increases in bothCREB expression and CREB transcriptional activity. Pharmacological inhibition of CREB activity, previouslyshown to have anti-proliferative effects on cancer cells, might therefore have additional benefits through inhibitionof aromatase expression and thus estrogen production in breast adipose.

Introduction

Estrogen is the primary hormonal factor that stimu-lates growth of hormone-dependent breast cancers [1].Approximately 70% of all breast cancers are hormone-dependent, and anti-estrogen treatment is generally thefirst line adjuvant therapy for patients with metastaticestrogen receptor (ER)-positive tumors. Estrogen syn-thesis from C19 steroids is catalyzed by the enzymearomatase cytochrome P450 (P450arom, the productof the CYP19 gene) [2]. In humans, aromatase isexpressed in the ovary and placenta, but also in vari-ous extra-glandular sites including brain, bone, andadipose tissue [3]. Regulation of the CYP19 genethat encodes aromatase is complex: the 90 kbp ge-nomic region immediately upstream of the first codingexon (exon II) contains at least six untranslated firstexons that are spliced onto CYP19 transcripts in atissue-specific manner [4–6]. Each untranslated exonis associated with its own unique promoter region,

which is in turn regulated by discreet cohorts of hor-mones, second messengers and transcription factors.In the ovary, for example, CYP19 expression is regu-lated by FSH, which induces the proximal promoterII through cAMP-dependent mechanisms [7, 8]. Inplacenta, promoter I.I regulates CYP19 expression inresponse to retinoids [9], whereas in bone and adiposetissue, a distal promoter (promoter I.4) drives CYP19expression under the control of glucocorticoids, class1 cytokines or TNFα [10–13].

In postmenopausal women, aromatase activity inadipose tissue is the major source of circulating estro-gens [9, 14]. In normal breast adipose tissue CYP19 isexpressed at low levels from promoter I.4 in responseto locally produced cytokines and systemic glucocor-ticoids. In the presence of a tumor, however, aromataseactivity and CYP19 expression in the surroundingadipose tissue are inappropriately elevated [15–18].This induction occurs in response to factors derivedfrom the tumor itself and/or infiltrating immune cells.

400 M Sofi et al.

One such factor is prostaglandin E2 (PGE2), one ofthe most potent inducers of CYP19 expression thusfar described. Importantly, PGE2 induces the gonadal-type promoter, promoter II, by activating EP1 and EP2receptors linked to protein kinase A and C pathways,respectively [19]. Breast tumors therefore trigger pro-moter switching such that the increase in local CYP19expression arises through aberrant transcription frompromoter II [20–22].

The molecular mechanisms regulating promoter IIin ovary have been well characterized. Basal tran-scription from promoter II requires activity of anorphan member or the nuclear receptor superfamily,steroidogenic factor-1 (SF-1) [8]. FSH-induced tran-scription occurs through activation of the CRE bindingprotein (CREB), which binds to a cAMP response ele-ment (CRE) approximately 60 bp upstream of the SF-1binding site [23]. The actions of SF-1 and CREB areboth necessary and sufficient for transcription frompromoter II in ovary.

Comparatively less is known regarding the regula-tion of promoter II in breast adipose tissue, althoughthe mechanisms appear to differ from those in ovary.Adipose tissue does not express SF-1; however a re-lated nuclear receptor, LRH-1, appears to substitutefor SF-1 as a basal transcription factor in this tissue[24]. The mechanisms of hormone-induced transcrip-tion from promoter II are less well understood. In thepresent study we have mapped the cAMP-regulatedregions of promoter II in adipose fibroblasts. Wefind that two distinct CRE-like sequences are requiredfor cAMP-induced transcription, and that cAMP in-duces binding of CREB to each site. Importantly, theexpression of CREB in human adipose tissue is ele-vated in tissue proximal to a breast tumor, comparedwith normal breast adipose tissue. Induction of CREBexpression may therefore contribute to the overexpres-sion of aromatase and increased estrogen productionobserved in breast cancer.

Methods

Plasmids

PII-914/+17 is a CYP19 promoter II/luciferase con-struct containing −914/−17 bp of human CYP19promoter II in the vector pGL3 (Promega Corp.,Sydney, NSW). 5′-deleted promoter constructs weregenerated by PCR using pII-914/+17 as template,generating appropriate restriction sites (PstI, SalI). Pu-tative cis-elements were disrupted by PCR-directed

mutagenesis. PCR fidelity and integrity of the res-ultant plasmids were confirmed by direct sequencing.The pCMV-β vector (Promega) encodes full-length β-galactosidase and was used to correct for transfectionefficiency.

Cell culture, transfection and reporter gene assays

Human adipose stromal cells were isolated fromsubcutaneous adipose tissue obtained from womenundergoing reduction mammoplasty (Mr Alan Kalus,Windsor, Australia), and cultured as described pre-viously [25]. All procedures were approved by theHuman Ethics Committee, Monash Medical Centre.3T3-L1 cells were cultured in DMEM supplemen-ted with 10% FBS at a density of 40,000 cells/ml.Cells were transfected for 22 h with 1.2 µg of totalDNA comprising 1.0 µg luciferase reporter and 0.2 µgpCMV-β, using Fugene6 reagent (Roche Diagnostics,Castle Hill, NSW). Cells were serum-starved for 24 hprior to experimental treatment, after which luciferaseand β-galactosidase activities of soluble cell extractswere measured using the Luciferase Assay System(Promega) and Galacto-light system (Applied Biosys-tems, Melbourne, Vic.), respectively.

Electrophoretic mobility shift assay

Nuclear extracts were prepared from confluent 3T3-L1cells and human adipose stromal cells by the methodof Schreiber et al. [26]. Five µg of nuclear extractwere incubated with 20,000 cpm 32P-labeled probe in20 µl binding buffer (20 mM HEPES pH 8.0, 1 mMEDTA, 10% glycerol, 50 mM KCl, 50 µg/ml polydI.dC/dI.dC, 1 mg/ml BSA, 10 mM dithiothreitol) for15 min at room temperature before electrophoresisusing a 5.4% polyacrylamide gel and 0.5 × TBE(final concentrations 44.5 mM Tris, 44.5 mM boricacid, 1 mM EDTA, pH 8.0) as running buffer for 3 hat 200 V. Gels were dried and radioactive complexesvisualized by phosphorimaging on a Molecular Dy-namics STORM phosphorimager. Where antibodieswere included in the reaction, protein extract and an-tibody were pre-incubated on ice for 10 min beforeaddition of probe.

Reverse transcription-PCR

Total RNA was prepared from confluent 3T3-L1 cellsand human adipose stromal cells using the QiaAMPRNA Blood Mini kit (Qiagen, Clifton Hill, Vic.). Total

CREB regulates aromatase expression 401

Figure 1. 5′-Deletion mapping of the cAMP-responsive regions of CYP19 promoter II expressed in 3T3-L1 preadipocytes. 3T3-L1 preadipo-cytes were transfected with the CYP19 promoter II reporter constructs depicted above, along with a constitutively active β-galactosidaseexpression construct, and incubated in the presence or absence of forskolin (25 µM) for 8 h. Cell were then lysed and assayed for luciferase andβ-galactosidase activities. Data are normalized to a percentage of the basal activity of pII-516, are expressed as mean ± SEM, and representdata combined from three to eight independent experiments.

RNA was prepared from whole adipose tissue (ob-tained from reduction mammoplasty or mastectomyspecimens) by guanidine isothiocyanate extraction fol-lowed by cesium chloride density ultracentrifugation.RNA integrity was verified by agarose gel electro-phoresis. First strand cDNA synthesis from 250 ngtotal RNA was performed using AMV reverse tran-scriptase (Roche) primed by random hexamers.

Real time PCR amplification of CREB and β2-microglobulin was performed on the LightCycler(Roche) using SYBR Green reaction mix (Roche)and the primers described above. cDNA sampleswere diluted 1:20 in water immediately before use.Experimental samples were quantified by compar-ison with standards of known concentration (0.1 –1000 fg/µl). PCR reactions were carried out usingthe following primer sets (all 5′–3′): CREB-sense:GAA GCT GAA AAC CAA CAA ATG ACA; CREB-antisense: CAA TAG TGC TAG TGG GTG CTG TG;β2microglobulin-sense: TGA ATT GCT ATG TGTCTG GGT; β2microglobulin-antisense: CCT CCATGA TGC TGC TTA CAT; GAPDH-sense: CGGAGT CAA CGG ATT TGG TCG TAT; GAPDH-antisense: AGC CTT CTC CAT GGT GGT GAAGAC. Statistical significance was assessed by Stu-dent’s t-test.

Results

To begin to identify cis-regulatory elements that me-diate cAMP-induced transcriptional activation of pro-moter II in preadipocytes, a series of 5′-deletedCYP19 promoter II luciferase reporter constructs were

Table 1. Comparison of CRE-like sequences withinCYP19 promoter II

Consensus CRE: T G A C G T C A

CRE1: −292 T G A A G T C A−285

CRE2: −211 T GCA C G T C A−203

AP1: −497 T G A – G T C A−491

Figure 2. Mutational analysis of CRE-like sequences within CYP19promoter II. 3T3-L1 preadipocytes were transfected with the CYP19promoter II reporter constructs depicted above, harboring mutationsin either the AP1, CRE1 or CRE2 sites, and incubated in the pres-ence or absence of forskolin (25 µM) for 8 h. Cell were then lysedand assayed for luciferase and β-galactosidase activities. Data arenormalized to a percentage of the basal activity of pII-516, are ex-pressed as mean ± SEM, and represent data combined from threeindependent experiments.

prepared (pII-914/+17 to pII-100/+17). These con-structs were transiently transfected into 3T3-L1 cellsalong with a constitutively active β-galactosidase ex-pression vector to control for transfection activity.Figure 1 shows normalized luciferase activity of these

402 M Sofi et al.

Figure 3. EMSA of 3T3-L1 nuclear proteins binding to CRE1 and CRE2. (A) 3T3-L1 preadipocyte nuclear extracts were incubated with a32P-labelled oligonucleotide duplex spanning CRE1 (left) or CRE2 (right), and analyzed by gel electrophoresis. Incubations were preformedin the absence (lanes 1 and 5) or presence (lanes 2–4, 6–8) or a 200-fold molar excess of the non-radiolabeled competitor probes indicated.(B) Effect of cAMP on nuclear protein binding to CRE1 and CRE2. 3T3-L1 preadipocytes were incubated with forskolin (25 µM) for thetimes indicated, nuclear extracts prepared and EMSA performed as above. (C) CREB is a component of the protein/DNA complexes formed onboth CRE1 and CRE2. 3T3-L1 nuclear extracts were incubated with antibodies directed against CREB or c-fos for 10 min on ice before beingsubjected to EMSA. The position of supershifted bands is indicated (ss). (D) CREB binding is increased in the presence of forskolin. 3T3-L1cells were incubated in the presence or absence of forskolin (25 µM) for 3 h and nuclear extracts prepared. Extracts were incubated on ice for10 min and EMSA conducted as above.

constructs under basal conditions, and after treat-ment with forskolin for 8 h. Basal activities of allconstructs were not significantly different. Forskolintreatment increased activity of constructs contain-ing between −914/+17 and −347/+17 bp by 3.5- to5-fold. Shorter constructs (containing −277+17 bpor less) were not induced by forskolin, indicatingthe presence of cAMP-responsive sequences between−347 and −277.

Sequence analysis of the promoter region between−417 and +17 revealed the presence of an octameric

element (CRE2, −292/−285) sharing 7 out of8 bp homology with the consensus palindromic CRE(Table 1). Downstream of this element exist twosequences previously shown to contribute to cAMP-induction of promoter II in ovary (CRE1, −211/−203;SF1, −132/−124). An additional CRE-like sequence(AP1, −497/−491) exists upstream of these sites. Toexamine the contribution of each of these putative cis-elements to cAMP-induced transcriptional activity, thesites were mutated individually in the context of alonger promoter fragment (pII-516/+17, Figure 2).

CREB regulates aromatase expression 403

Figure 4. EMSA of human breast preadipocyte nuclear protein binding to CRE1 and CRE2. (A) Human preadipocyte nuclear extracts wereincubated with a 32P-labelled oligonucleotide duplex spanning CRE1 (left) or CRE2 (right), and analyzed by gel electrophoresis. Incubationswere preformed in the absence (lanes 1 and 4) or presence (lanes 2, 3, 5 and 6) or a 200-fold molar excess of the non-radiolabeled competitorprobes indicated. (B) Effect of cAMP on nuclear protein binding to CRE1 and CRE2. Human preadipocytes were incubated with forskolin(25 mM) for the times indicated, nuclear extracts prepared and EMSA performed as above. (C) CREB is a component of the protein/DNAcomplexes formed on both CRE1 and CRE2. Human preadipocyte nuclear extracts were incubated with an antibody directed against CREB for10 min on ice before being subjected to EMSA.

Mutation of either the upstream AP1 site or the prox-imal SF-1 site did not significantly alter either basalor forskolin-induced promoter activity. Mutation ofCRE1, but not CRE2, decreased basal promoter by ap-proximately 80%. Mutation of either CRE2 or CRE1,however, completely abolished the ability of the pro-moter to respond to forskolin. Thus, cAMP-inducedtranscription from promoter II in 3T3-L1 preadipo-cytes requires two distinct CRE-like sequences, CRE1and CRE2.

To identify transcription factors that bind these ele-ments in preadipocytes, EMSA was performed usingeach binding site as probe and nuclear extracts from3T3-L1 cells (Figure 3(A)). Nuclear extracts formedat least four distinct complexes (c1–c4) with CRE1(lane 1). Formation of each complex was completelyabolished in the presence of a 200-fold molar excessof non-radiolabeled CRE1 (lane 2). A probe con-taining a mutation in the core CRE motif (mCRE1)failed to compete with CRE1 in formation of c1, c2and c3, but competed for formation of c4 (lane 4),suggesting that c1–c3, but not c4, represent sequence-specific protein/DNA interactions. Non-radiolabeledCRE2 competed with CRE1 for formation of c1, but

not c2–c4 (lane 3). Using CRE2 as probe, three spe-cific protein/DNA complexes were observed (c5–7,lane 5). Formation of each complex was inhibited bynon-radiolabeled CRE2 or CRE1 (lanes 6 and 7), butnot by mCRE (lane 8). Thus, each CRE-like sequenceforms specific protein/complexes with 3T3-L1 nuclearextracts. Further, the lowest mobility complexes (c1and c5) appear to be common to each probe.

To determine if any of the protein/DNA com-plexes are formed in a cAMP-responsive manner,3T3-L1 cells were treated with forskolin for varioustimes and nuclear extracts prepared (Figure 3(B)). Us-ing CRE1 as probe, binding intensities of c1, andto a lesser extent c2, increased in response to for-skolin, reaching a maximum following 3 h of treat-ment. Formation of the other complexes was notaltered by forskolin treatment. In contrast, forma-tion of each complex formed with CRE2 increased inresponse to forskolin, with maximal binding again ob-served following 3 h of treatment (Figure 3(B), rightpanel). For both CRE1 and CRE2, an antibody dir-ected against CREB specifically displaced the lowestmobility complexes (c1 and c5), producing supershift-ed complexes (Figure 3(C)). Moreover, the intensity

404 M Sofi et al.

Figure 5. Expression of CREB in normal breast adipose tissue andadipose tissue surrounding breast tumors. Total RNA was isolatedfrom adipose tissue obtained from either reduction mammoplastyprocedures (normal) or mastectomy samples containing a palpabletumor (adjacent to tumor). RNA was reverse transcribed and cDNAused as template for PCR for both CREB and GAPDH. (A) PCRproducts obtained following 30 cycles for CREB and 25 cycles forβ2-microglobulin (previously determined to correspond to the lin-ear phase of the amplification curve). (B) Southern blotting usingan internal 32P-labelled probe to quantitate the PCR products ob-tained in (A). (C) Quantitative real-time PCR analysis of CREB andβ2-microglobulin expression in normal and tumor-bearing adiposetissue. Graphical data are expressed as box plots where the box ex-tends from the 25th to the 75th percentile, with a line at the median.The error bars extend to the highest and lowest values. Note thattumor sample number 7 was excluded from the analyses since it liesout with two standard deviations of the mean of the other values.Statistical significance is indicated at ∗p < 0.05, ∗∗p < 0.01.

of this supershifted complex was increased follow-ing treatment with forskolin (Figure 3(D)). Thus,CREB derived from 3T3-L1 preadipocyte nuclear

extracts binds to both CRE1 and CRE2 in response tocAMP.

Because important species differences exist withrespect to hormonal regulation of aromatase expres-sion [27], the EMSAs were repeated using humanadipose stromal cell nuclear extracts as the source ofprotein (Figure 4). Although a large amount of non-specific binding was observed (NS), these experimentsrevealed that for both CRE1 and CRE2, the lowestmobility complex formed with human adipose stromalcell nuclear extract is sequence specific (Figure 4(A)),increases in response to forskolin (Figure 4(B)), andis recognized by an antibody directed against CREB(Figure 4(C)). Note that this antibody appears to dis-place binding of human CREB rather than form asupershifted complex, as it does with mouse CREB(Figure 3(C)). These data provide strong evidencethat cAMP-induced transcription from promoter II inpreadipocytes is mediated, at least in part, by CREBbinding to two distinct CRE-like elements.

Since aromatase expression from promoter II inpreadipocytes is increased in adipose tissue surround-ing a breast tumor, we next asked if increased ex-pression of CREB might contribute to this aberrantexpression. Breast adipose tissue specimens were col-lected from healthy individuals and from mastectomyspecimens containing a palpable tumor. RNA wasextracted and semi-quantitative RT-PCR performed(Figure 5(A)). CREB expression was detected in onlyone out of five normal adipose tissue specimens. How-ever, eight out of eight mastectomy adipose specimensexpressed CREB, at varying levels. This differencewas statistically significant (p > 0.01) when quan-titated by southern blotting and expressed relativeto expression of GAPDH (Figure 5(B)). To confirmthis, a quantitative real-time PCR method for CREBwas developed. Using this method, CREB expres-sion was detected in two of eight normal adiposetissue specimens, and five of eight mastectomy spe-cimens. The mean expression levels (relative to thehousekeeping gene β2-microglobulin) in normal andtumor-containing adipose tissue were 0.02 and 0.11,respectively (p < 0.05).

Discussion

Recent clinical trials have demonstrated that aro-matase inhibitors are superior to traditional ERantagonists for breast cancer treatment in the neo-adjuvant setting [28]. The source of estrogen thatdrives growth of ER-positive tumors in postmeno-

CREB regulates aromatase expression 405

pausal women is local synthesis by adipose fibroblastssurrounding the tumor, as well as tumor stromal cellsand epithelium [19]. In these tissues, aromatase ex-pression is inappropriately elevated through activationof the proximal promoter II, rather than the nor-mal distal promoter I.4 [20–22]. Elucidation of themechanisms by which transcription from promoter IIis induced in breast cancer is critical for a clearerunderstanding of this disease, as well as for the devel-opment of more efficacious, tissue-specific aromataseinhibitors.

In the present study we show that promoter II isregulated by the transcription factor CREB, whichbinds to two distinct CRE-like sequences within pro-moter II in a cAMP-dependent manner. Thus, a num-ber of differences exist between the regulation ofpromoter II in ovary, and in adipose. For example,transcription from promoter II in ovary requires theproximal CRE (CRE1) and SF-1 sites. These sitesare both necessary and sufficient for cAMP induc-tion of promoter II in ovary. In contrast, a min-imal promoter II containing both elements (pII-211,Figure 1) does not respond to cAMP in preadipo-cytes. Responsiveness to cAMP requires an additionalCRE-like sequence (CRE2), located approximately80 bp upstream of CRE1 (−292/−285). Althougheach CRE bound a number of proteins from preadipo-cytes nuclear extract in EMSA, antibody supershiftexperiments revealed that CREB is a component ofthe protein/DNA complexes at both sites. This sug-gests that subtle differences exist between differentcell types in the cAMP regulation of promoter II.Although CRE2 has previously been shown to con-tribute to cAMP induction of aromatase in WS3TFbreast tumor fibroblasts [29], these cells do not expressCREB, and the predominant CREB in tumor fibro-blasts appears to be a 60 kDa species [29]. By contrast,adipose fibroblasts express CREB, which likely me-diates cAMP induced aromatase expression in thesecells.

The identity of the other CREBs derived frompreadipocytes nuclear extract is unknown at present,although Chen et al. recently reported that the distalCRE1 overlaps a binding site for the transcriptionalrepressor SnaH [30]. Thus the activity of promoter IIis likely determined, at least in part, by the balance ofpositive and negative transcription factors binding tothese CREs. It has been proposed that in normal breastepithelial and stromal cells, SnaH is expressed andbinds promoter II, preventing other positive factors,such as CREB, from binding. Cancerous breast tissue,

by contrast, expresses lower levels of SnaH, thusallowing access of transcriptional activators to thepromoter II CREs [30].

Additional mechanisms of tumor-induced aro-matase activity exist. MCF-7 conditioned mediumstrongly increased aromatase expression in humanadipose fibroblasts in a cAMP-independent mannerthat involved binding of CEBPβ to sites upstream ofthe CREs described here [31]. Therefore, differentconditions likely up-regulate aromatase expression viadifferent mechanisms.

The DNA binding activity of CREB is stronglyactivated by phosphorylation by PKA in response tocAMP. Thus, tumor derived factors that activate ad-enylyl cyclase, such as PGE2, would be predictedto increase CREB phosphorylation and consequentlyactivate transcription from promoter II. The currentresults suggest that in addition to CREB activity, ex-pression of CREB itself may be increased in adiposetissue surrounding breast tumors. It is unclear whetherthis reflects up-regulation of CREB by tumor-derivedfactors or a linkage between CREB expression levelsin breast adipose and breast cancer risk. Althoughno studies have yet shown any direct link betweenCREB expression and breast cancer, CREB may beinvolved in tumor growth and metastasis. Thus, In-hibition of CRE-mediated transcription using CRE-transcription factor decoy oligonucleotides inhibitsthe growth of ER-positive MCF-7 breast cancer cells[32, 33] as well as the growth of cultured ovariancancer cells [34]. The growth inhibitory effect ofthese oligonucleotides in ovarian cancer cells was as-sociated with down regulation of protein kinase Aactivity and export of CREB from the nucleus tothe cytosol [34]. In addition, CREB and its associ-ated proteins appear to contribute to the acquisitionof metastatic phenotype of human melanoma cells[35]. Observations such as these have stimulated in-terest in CREB and other CREBs as molecular targetsfor cancer therapy [36]. The pivotal role of CREBin regulating aromatase expression and estrogen syn-thesis in breast adipose tissue, as shown here, sug-gests that inhibition of aromatase expression wouldbe an additional important benefit of such anti-CREBtherapy.

Acknowledgement

This work was supported by the Victorian BreastCancer Research Consortium.

406 M Sofi et al.

References

1. Davidson NE, Lippman ME: The role of estrogens in growthregulation of breast cancer. Crit Rev Oncogenesis 1: 89–111,1989

2. Simpson ER, Mahendroo MS, Means GD, Kilgore MW,Hinshelwood MM, Graham-Lorence S, Amarneh B, Ito Y,Fisher CR, Michael MD: Aromatase cytochrome P450, theenzyme responsible for estrogen biosynthesis. Endocr Rev 15:342–355, 1994

3. Bulun SE, Simpson ER: Regulation of aromatase expressionin human tissues. Breast Cancer Res Tr 30: 19–29, 1994

4. Means GD, Kilgore MW, Mahendroo MS, Mendelson CR,Simpson ER: Tissue-specific promoters regulate aromatasecytochrome P450 gene expression in human ovary and fetaltissues. Mol Endocrinol 5: 2005–2013, 1991

5. Mahendroo MS, Means GD, Mendelson CR, Simpson ER:Tissue-specific expression of human P-450AROM. The pro-moter responsible for expression in adipose tissue is differentfrom that utilized in placenta. J Biol Chem 266: 11276–11281,1991

6. Toda K, Shizuta Y: Molecular cloning of a cDNA showingalternative splicing of the 5′- untranslated sequence of mRNAfor human aromatase P-450. Eur J Biochem 213: 383–389,1993

7. Steinkampf MP, Mendelson CR, Simpson ER: Regulation byfollicle-stimulating hormone of the synthesis of aromatasecytochrome P-450 in human granulosa cells. Mol Endocrinol1: 465–471, 1987

8. Michael MD, Kilgore MW, Morohashi K, Simpson ER:Ad4BP/SF-1 regulates cyclic AMP-induced transcriptionfrom the proximal promoter (PII) of the human aromataseP450 (CYP19) gene in the ovary. J Biol Chem 270:13561–13566, 1995

9. Simpson ER, Zhao Y, Agarwal VR, Michael MD, Bulun SE,Hinshelwood MM, Graham-Lorence S, Sun T, Fisher CR,Qin K, Mendelson CR: Aromatase expression in health anddisease. Recent Prog Horm Res 52: 185–213, 1997

10. Shozu M, Simpson ER: Aromatase expression of humanosteoblast-like cells. Mol Cell Endocrinol 139: 117–129, 1998

11. Zhao Y, Nichols JE, Bulun SE, Mendelson CR, Simpson ER:Aromatase P450 gene expression in human adipose tissue.Role of a Jak/STAT pathway in regulation of the adipose-specific promoter. J Biol Chem 270: 16449–16457, 1995

12. Zhao Y, Mendelson CR, Simpson ER: Characterization of thesequences of the human CYP19 (aromatase) gene that mediateregulation by glucocorticoids in adipose stromal cells and fetalhepatocytes. Mol Endocrinol 9: 340–349, 1995

13. Zhao Y, Nichols JE, Valdez R, Mendelson CR, Simpson ER:Tumor necrosis factor-alpha stimulates aromatase gene ex-pression in human adipose stromal cells through use of anactivating protein-1 binding site upstream of promoter 1.4.Mol Endocrinol 10: 1350–1357, 1996

14. Siiteri PK, Macdonald PC: Role of extraglandular estrogenin human endocrinology. In: Green RO, Astwoon EB (eds)Handbook of Physiology. American Physiological Society,1973, pp 619–629

15. Thorsen T, Tangen M, Stoa KF: Concentration of endogenousoestradiol as related to oestradiol receptor sites in breast tumorcytosol. Eur J Cancer Clin On 18: 333–337, 1982

16. van Landeghem AA, Poortman J, Nabuurs M, Thijssen JH:Endogenous concentration and subcellular distribution of es-trogens in normal and malignant human breast tissue. CancerRes 45: 2900–2906, 1985

17. Bulun SE, Price TM, Aitken J, Mahendroo MS, Simpson ER:A link between breast cancer and local estrogen biosynthesissuggested by quantification of breast adipose tissue aromatasecytochrome P450 transcripts using competitive polymerasechain reaction after reverse transcription. J Clin Endocr Metab77: 1622–1628, 1993

18. Harada N: Aberrant expression of aromatase in breast cancertissues. J Steroid Biochem 61: 175–184, 1997

19. Zhao Y, Agarwal VR, Mendelson CR, Simpson ER: Estro-gen biosynthesis proximal to a breast tumor is stimulated byPGE2 via cyclic AMP, leading to activation of promoter II ofthe CYP19 (aromatase) gene. Endocrinology 137: 5739–5742,1996

20. Zhou D, Zhou C, Chen S: Gene regulation studies of aro-matase expression in breast cancer and adipose stromal cells.J Steroid Biochem 61: 273–280, 1997

21. Agarwal VR, Bulun SE, Leitch M, Rohrich R, SimpsonER: Use of alternative promoters to express the aromatasecytochrome P450 (CYP19) gene in breast adipose tissues ofcancer-free and breast cancer patients. J Clin Endocr Metab81: 3843–3849, 1996

22. Harada N, Utsumi T, Takagi Y: Tissue-specific expressionof the human aromatase cytochrome P-450 gene by altern-ative use of multiple exons 1 and promoters, and switchingof tissue-specific exons 1 in carcinogenesis. P Natl Acad SciUSA 90: 11312–11316, 1993

23. Michael MD, Michael LF, Simpson ER: A CRE-like sequencethat binds CREB and contributes to cAMP-dependent reg-ulation of the proximal promoter of the human aromataseP450 (CYP19) gene. Mol Cell Endocrinol 134: 147–156,1997

24. Clyne CD, Speed CJ, Zhou J, Simpson ER: Liver receptorhomologue-1 (LRH-1) regulates expression of aromatase inpreadipocytes. J Biol Chem 277: 20591–20597, 2002

25. Ackerman GE, Smith ME, Mendelson CR, MacDonald PC,Simpson ER: Aromatization of androstenedione by humanadipose tissue stromal cells in monolayer culture. J ClinEndocr Metab 53: 412–417, 1981

26. Schreiber E, Matthias P, Muller MM, Schaffner W: Rapiddetection of octamer binding proteins with ‘mini-extracts’,prepared from a small number of cells. Nucleic Acids Res 17:6419, 1989

27. Hinshelwood MM, Dodson MM, Sun T, Simpson ER: Reg-ulation of aromatase expression in the ovary and placenta:a comparison between two species. J Steroid Biochem 61:399–405, 1997

28. Munster PN, Horton J: Tamoxifen v.s. the aromataseinhibitors: news from San Antonio, 2001. Cancer Control 8:478–479, 2001

29. Zhou D, Chen S: Identification and characterization of acAMP-responsive element in the region upstream from pro-moter 1.3 of the human aromatase gene. Arch BiochemBiophys 371: 179–190, 1999

30. Okubo T, Truong TK, Yu B, Itoh T, Zhao J, Grube B, ZhouD, Chen S: Down-regulation of promoter 1.3 activity of thehuman aromatase gene in breast tissue by zinc-finger protein,snail (SnaH). Cancer Res 61: 1338–1346, 2001

31. Zhou J, Gurates B, Yang S, Sebastian S, Bulun SE: Malignantbreast epithelial cells stimulate aromatase expression via pro-moter II in human adipose fibroblasts: an epithelial–stromalinteraction in breast tumors mediated by CCAAT/enhancerbinding protein beta. Cancer Res 61: 2328–2334, 2001

32. Park YG, Nesterova M, Agrawal S, Cho-Chung YS: Dualblockade of cyclic AMP response element (CRE) and AP-

CREB regulates aromatase expression 407

1-directed transcription by CRE-transcription factor decoyoligonucleotide. Gene-specific inhibition of tumor growth. JBiol Chem 274: 1573–1580, 1999

33. Lee YN, Park YG, Choi YH, Cho YS, Cho-Chung YS: CRE-transcription factor decoy oligonucleotide inhibition of MCF-7 breast cancer cells: cross-talk with p53 signaling pathway.Biochemistry 39: 4863–4868, 2000

34. Alper O, Bergmann-Leitner ES, Abrams S, Cho-ChungYS: Apoptosis, growth arrest and suppression of invasive-ness by CRE-decoy oligonucleotide in ovarian cancer cells:protein kinase A downregulation and cytoplasmic exportof CRE-binding proteins. Mol Cell Biochem 218: 55–63,2001

35. Xie S, Price JE, Luca M, Jean D, Ronai Z, Bar-Eli M:Dominant-negative CREB inhibits tumor growth and metas-tasis of human melanoma cells. Oncogene 15: 2069–2075,1997

36. Cho-Chung YS, Park YG, Nesterova M, Lee YN, Cho YS:CRE-decoy oligonucleotide-inhibition of gene expression andtumor growth. Mol Cell Biochem 212: 29–34, 2000

Address for offprints and correspondence: Colin D. Clyne, PrinceHenry’s Institute of Medical Research, PO Box 5152, Clayton Vic.3168, Australia; Tel.: +61-3-9594-4372; Fax: +61-3-9594-6125;E-mail: colin.clyne@med.monash.edu.au