(CANCER RESEARCH 58. 5466-5472. December I. 1998]

Expression of AP-2 Transcription Factors in Human Breast Cancer Correlates withthe Regulation of Multiple Growth Factor Signalling Pathways1

Bruce C. Turner,2 Jian Zhang, Andrew A. Gumbs, Mary G. Mäher,Leni Kaplan, Darryl Carter, Peter M. Glazer,Helen C. Hurst, Bruce G. Haffty, and Trevor Williams3

Departments of Therapeutic Radiology \B. C. T.. A. A. C., M. G. M.. P. M. G., B. G. H.¡and Pathology ¡L.K.. D. CJ, Yale University School of Medicine and Department ofMolecular. Cellular and Developmental Biology ¡J.Z.. T. WJ, Yale University, New Haven, Connecticut 06511. and Gene Transcription Laboratory. Imperial Cancer Researcht'unti Oncology Unit, Hammersmith Hospital. London WI2 ONN, United Kingdom ¡H.C H.I

ABSTRACT

The AP-2 transcription factors are required for normal growth and

morphogenesis during mammalian development. Previous in vitro studieshave also indicated that the AP-2 family of proteins may be involved in theetiology of human breast cancer. The AP-2 genes are expressed in manyhuman breast cancer cell lines, and critical AP-2-binding sites are presentin both the ERBB-2 (HER2/neu) and estrogen receptor promoters. Wehave now characterized immunological reagents that enable specific AP-2family members, including AP-2a and AP-2y, to be detected in human

breast cancer epithelium. Data obtained with these reagents demonstratethat whereas AP-2t»and Al'-2y are both present in benign breast epithe-

lia, there is a significant up-regulation of \P-2y expression in breastcancer specimens (/" = 0.01). There was also a significant correlation

between the presence of the AP-2a protein and estrogen receptor expression i/' = O.U18) and between specimens containing both AP-2o/AP-2yproteins and ERBB-2 expression i/' = 0.003). Furthermore, we detectedan association i/' = 0.04) between the expression of AP-2y and the

presence of an additional signal transduction molecule implicated inbreast cancer, the insulin-like growth factor I receptor. Analysis of theproximal promoter of the insulin-like growth factor I receptor revealed anovel AP-2-binding site. Thus, AP-2 proteins may directly regulate the

transcription of this growth factor receptor. Taken together, these datastrongly support a role for the AP-2 gene family in the control of cell

growth and differentiation in breast cancer.

INTRODUCTION

Recent advances in molecular genetics have provided new insightsinto the etiology of human breast cancer. Whereas mutations in theBRCAI and BRCA2 genes have been identified in some patients withfamilial breast cancer, other molecular markers may help to classifysporadic breast tumors, which account for approximately 90% ofnewly diagnosed breast cancers (1-5). In particular, the expressionstatus of two critical signal transduction molecules, ERBB-2 (HER2/neu) and the ER,4 has important implications for patient prognosis,

appropriate adjuvant treatment, and treatment response (1, 2, 6-8).

Insight into the regulatory networks controlling the expression of

Received 6/29/98; accepted 10/5/98.The costs of publication of this article were defrayed in part hy the payment of page

charges. This article must therefore be hereby marked tuh-ertixenient in accordance with

18 U.S.C. Section 1734 solely to indicate this fact.1Supported by the Yale University School of Medicine Department of Therapeutic

Radiology. ACS-IRG Award 1N31-38 (to T. W.). American Cancer Society Grant RPG-98-096-OI-MGO (to T. W.), and NIH Gram ROI CA 77833 (to T. W.). B. C. T. is anAACR-AFLAC Scholar. J. Z. is the recipient of a breast cancer research fellowship fromthe Department of the Army (DAMD17-96-1-6094).

2 Present address: Department of Radiation Oncology, Hahncmann University. Mail-

stop 200. Broad and Vine Streets. Philadelphia. PA 19102 and Department of Pathology.University of Pennsylvania School of Medicine. John Morgan Building. Room 252, 36thand Hamilton Walk, Philadelphia. PA 19104.

' To whom requests for reprints should be addressed, at Department of Molecular.

Cellular and Developmental Biology. Yale University. KBT 1034. 266 Whitney Avenue,New Haven. CT 06511. Phone: (203) 432-3230; Fax: (203) 432-5690; E-mail:twill ^biomed.med.yale.edu.

4 The abbreviations used are: ER. estrogen receptor: DDFS, distant disease-free

survival; OS. overall survival; PR. progesterone receptor; IGF-IR, insulin-like growthfactor 1 receptor; EMSA. electfophorelic mobility shift assay; BBE. benign breast epithelium; IDC. invasive ductal carcinoma; NS. not significant.

these important signal transduction pathways in breast cancer mayyield new gene targets for prognosis and therapeutic intervention.

In normal mammary gland development, ER molecules are important for signaling ductal proliferation during puberty and pregnancy.The ER gene is also expressed at elevated levels in approximately50% of all breast carcinomas (6, 8, 9). Patients with ER-positive

breast carcinomas have an improved survival rate and a longer disease-free interval than patients whose tumors lack ER expressionbecause the ER-positive tumors are usually hormone sensitive, andtheir growth can be controlled with hormonal therapy. ERBB-2 over-

expression is also a common molecular alteration observed in humanbreast cancer, in which it occurs in approximately 15% of breasttumors (2). Moreover, high levels of ERBB-2 have been shown to

correlate with poor prognosis and to predict a worse response tochemotherapy, mastectomies, and breast-conserving therapy (2, 7).The oncogenic potential of ERBB-2 has also been demonstrated in

both cultured cells and transgenic mouse models, strongly suggestingthat this protein has a role in the pathogenesis of human breast cancer(2).

The AP-2 gene family of transcription factors has been implicatedin regulating both ER and ERBB-2 expression during the progressionfrom normal breast epithelium to breast cancer (10-12). All threeknown members of this gene family, AP-2a, AP-2ß,and AP-2y, are

expressed in human breast cancer cell lines (11). During the normalcourse of vertebrate embryogenesis, these genes are transcribed intissues undergoing complex morphogenetic changes. Major sites ofAP-2 expression include the developing neural crest, kidney, epidermis, and face and limb buds (13-15). Thus far, two of these geneshave been disrupted by homologous recombination. AP-2a knockout

mice have multiple defects in the development of the neural tube,face, eyes, limbs, and body wall (16, 17). Mice lacking the AP-2/3

gene have fewer gross phenotypic defects but nevertheless die shortlyafter birth due to a failure of collecting duct formation in the kidneys(18). Together, these findings establish the importance of the AP-2

gene family for normal growth and morphogenesis during embryonicdevelopment.

The three related AP-2 proteins all bind as homo- or heterodimersto a GC-rich element typified by the DNA sequence GCCNNNGGC(11, 19). One related version of this sequence, 5'-GCCCTGCGGGG-3', occurs in the 5' untranslated region of the ER gene and augments

ER promoter activity in vitro (12, 20). In the context of the ERBB-2

promoter, our previous studies have shown that the sequence element5'-GCTGCAGGC-3', which is required for high-level expression of

this gene in human breast cancer cell lines, also acts as an AP-2-binding site (10). Moreover, we have demonstrated that the AP-2proteins can act through this binding site to regulate ERBB-2 promoteractivity (11). Further support for the role of the AP-2 proteins inERBB-2 overexpression came from an examination of breast cancercell lines. In these in vitro experiments, high levels of ERBB-2expression correlated with the presence of one or more AP-2 proteins,especially that of AP-2a and AP-2y (10, 11). No ERBB-2 overexpression was found in the absence of AP-2, although AP-2 expression

5466

Research. on January 23, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from

AP-2a AND AP-2-y EXPRESSION IN HUMAN BREAST CANCER

could be detected in some cell lines that lacked ERBB-2 transcription

(21).The presence of AP-2 proteins in several breast cancer cell lines

raised the possibility that an in vivo analysis of this gene family couldprovide mechanistic insight into the etiology of mammary tumors.Therefore, we have now examined the incidence of AP-2 protein

expression in vivo, both in BBE and breast cancer specimens, todetermine whether expression of the AP-2 proteins was up-regulated

during the progression toward breast cancer. We have also extendedthese studies to determine whether the presence of the AP-2a andAP-27 transcription factors in human breast cancer specimens corre

lated with the expression of candidate target genes, especially that ofgrowth factor receptor molecules. As outlined below, our findingsstrongly support an important role for AP-2 in the etiology of human

breast cancer.

MATERIALS AND METHODS

Patient Population and Clinical Characteristics. The breast cancer database has 2400 patients treated between 1973 and 1993 at Yale UniversitySchool of Medicine with a median follow-up of 13 years and a minimumfollow-up of 4 years. From this group of patients, we identified 81 patients whohad experienced early-stage breast cancer and were treated with lumpectomyand radiation therapy whose archival paraffin-embedded tumor blocks were

available. All patients were treated by lumpectomy. with or without axillarydissection, followed by full-course radiation therapy to the intact breast to a

median dose of 64 Gy. The age, tumor histology, tumor size, stage distribution,axillary lymph node status, date of radiation therapy treatment, menopausa!status, type of adjuvant systemic treatment, length of follow-up, DDFS, and

OS were available on all patients in the study (22). All patients were seen infollow-up every 6 months for the first 3 years and yearly thereafter. The 81early-stage breast cancer patients presented at a mean age of 53 years and hada median follow-up of 10 years. The 10-year DDFS and OS for the patient

population in this study are 60 and 73%, respectively. The majority of thepatients had a histológica! diagnosis of IDC (73 of 81 patients) or invasivelobular carcinoma (5 of 8! patients), and some patients had invasive medullarycarcinoma (3 of 81 patients). All tumors specimens were derived from patientswith T, (tumor < 2 cm) or T2 (tumor = 2-5 cm) breast tumors, and there was

no bias toward either small or large breast tumors in this study. Note thatalthough 81 total samples were used in the study, there was not always enoughtissue available to use for all probes. In particular, 80 samples were processedfor AP-2a expression, of which 74 were also analyzed for the presence of theAP-2-y protein. One sample was analyzed for AP-2-y but not AP-2a expression.

With respect to ERBB-2, 76 samples were analyzed; 75 were assessed forAP-2« status, 70 were assessed for AP-27 status, and 69 were assessed forboth AP-2a and AP-27 status. We were also able to obtain the paraffin-

embedded blocks of 12 women who underwent reduction mammoplasties inwhich a pathological analysis revealed no evidence of malignancy. A protocolfor the study was approved by the Human Investigations Committee at theYale University School of Medicine.

Immunohistochemistry. After the identification of both benign und invasive breast cancer cases, the individual 10% formalin-fixed and paraffin-

embedded blocks were evaluated for either BBE or invasive carcinoma byH&E staining and processed for immunohistochemical staining as describedpreviously (22). The 5-/j.m-thick sections were deparaffini zed in xylene and

washed in ethanol before treatment with hydrogen peroxide to quench endogenous peroxidase. After rehydration, slides used for AP-2a immunohistochem-istry were pressure-cooked for 5 min in 9 mM sodium citrate (pH 6.0). Slidesused for AP-2-y staining or staining with the SC-184 polyclonal antiserum

(Santa Cruz Biotechnology, Santa Cruz. CA) did not require antigen retrieval.Subsequently, all sections were washed in 1% PBS and blocked with goatsuppressor serum for 30 min. Next, immunological reagents capable of distinguishing between the three AP-2 gene products were added overnight at4°C.AP-2a status was determined by using the AP-2a-specific mouse mono

clonal antibodies 3B5 or 5E4 (11. 17). either undiluted from tissue culturesupernatant or as a 1:1000 dilution of ascites fluid. The 3B5 antibody recognizes an epitope in the DNA-binding domain of AP-2a, and an AP-2«peptide.

GINIPDQTVIKKGPVSLSKSNSNAVSAIPINK (23), can act as a specificblocking reagent. AP-2-y status was analyzed using a 1:2(K) dilution of theepitope-selected rabbit polyclonal antiserum -y96. The peptide immunogen for

this antiserum. SYMNPGDQSPADSNKTLEKC. is located near the COOHterminus of the human AP-2-y protein (11). This peptide was coupled to

keyhole limpet hemocyanin before immunization and was also used for theepitope selection and peptide blocking experiments related to the AP-2-y

antiserum. Western blot analysis indicated that immunological reagents 3B5and y96 are specific for AP-2a and AP-2-y, respectively (data not shown). The

batch of SC-184 polyclonal antiserum used in these experiments (#EO94)reacted equally well with AP-2a and AP-2ßand was also capable of recognizing AP-27. Dut with less avidity (data not shown).

After primary antibody incubation, the slides were treated for 30 min witha 1:500 dilution of biotinylated antimouse (a) or antirabbit (7; SC-184)antisera followed by peroxidase-conjugated streptavidin. The immune com

plexes were then visualized with the chromogenic substrate diaminoben/.idinetetrahydrochloride. Subsequently, the slides were counterstained with hema-toxylin. Positive AP-2 protein staining was determined by nuclear staining

only when it was present within either the BBE or the invasive breast cancercomponent. The same tumor specimens were also analyzed using a monoclonalantibody to the ERBB-2 protein (Oncogene Science, Uniondale. NY), a polyclonal antibody to the ßsubunit of the IGF-IR (Santa Cruz Biotechnology),

and the previously described estrogen receptor immunocytochemistry assaymethod for the ER and PR receptors (Abbott Laboratories. Chicago. IL; Refs.7. 22. and 24). Control samples were processed and stained as described above,except thai the primary antibody was omitted, or the primary antibody waspreincubated with the appropriate blocking peptide for I h before applicationto the slide.

The pathologist who processed, stained, and graded the specimens wasblinded to the clinical histories of the patients. Each slide was rated on thefollowing point scale: (ai O, no stain; (h) 1+. light stain; (c) 2 + , moderatestain; (d) 3 + . heavy stain: and (e) 4 + , intense staining. The percentagedistribution of staining within BBE or the invasive carcinoma component wasscored. A value of s2+ in intensity was considered positive, and an H-score(defined as the product of intensity and distribution) of >50 was used as acutoff for AP-2 and ERBB-2, and an H-score of &75 was used as a cutoff forIGF-IR, ER, and PR expression as described previously (7, 22).

Electrophoretic Mobility Shift Assays. A double-stranded oligonucleo-tide, 5'-GGGCCGGGCCCTCGGCGCGCCG-3'. was synthesized correspond

ing to nucleotides -97 to -75 in the proximal human IGF-IR promoter (25).

The oligonucleotide corresponding to the well-characterized AP-2-binding site

in the human metallothionein IIA promoter has been reported previously ( 19).Both oligonucleotides were labeled with polynucleotide kinase and used asprobes for EMSAs or used as cold competitor DNA molecules. EMSA wasperformed using in v/fw-translated AP-2 protein as described previously (19).

Statistical Analysis. All patient data including demographics, age. stage.ER/PR status, lymph node status, histology, radiation therapy technique, andadjuvant systemic chemotherapy as well as outcomes including local, regional,and distant metastasis were entered into a blinded computerized database, andall calculations were performed using the PRODAS database package (Conceptual Software Inc., Houston. TX). The actuarial survival curves werecalculated using the life table method, the difference among the curves wasdetermined by the generalized Wilcoxon test, and the differences between thecategorical variables were tested by standard )f analysis. The factors affectingsurvival were determined by a univariate regression analysis according to theCox proportional hazards regression model. P £ 0.05 was considered toindicate statistical significance.

RESULTS

Established human breast cancer cell lines often express high levelsof the AP-2a and AP-2-y proteins (10, 11). We determined whether

this phenomenon also occurred in primary breast cancer tissue. Forthis purpose, we developed specific immunological reagents that werecapable of recognizing these two proteins in formalin-fixed paraffin-

embedded tissue. We had previously generated mouse monoclonalantibodies 3B5 and 5E4 that are specific for the AP-2a protein (11,

17). We therefore optimized staining procedures so that these two5467

Research. on January 23, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from

AP-2a AND AP-2-y EXPRESSION IN HUMAN BREAST CANCER

.-"•'&&

B.ï-

-HÕAP2 alpha -HOAP2 alpha

>;.

. C1,^

+ct AP2 alpha+peptide

-a AP2 alpha

+a AP2 gamma -a AP2 gamma

. .. * ^.«••.1•»•«»....

•;>>?V. * -••-•' -,

•*.-.•.v;-- •-*•v •.•" •

+a AP2 gamma+peptide

Fig. \. Detection of AP-2«and AP-2y protein expression in IDC. A and B. the presenceof AP-2a in the nucleus ibnnvn) was detected with monoclonal antibody 3B5. C. nuclearstaining was lost when the 3B5 antibody was preincubated with a specific blockingpeptide. /}. breast cancer specimen stained only with antimouse secondary antibody. £.section stained with rabbit polyclonal anliserum y96 for AP-2y expression. F. breastcancer specimen stained only with antirabbit secondary antibody. C. sample in which the7% antiserum was preincubated with a specific blocking peptide.

antibodies could be used for the immunohistochemical analysis ofhuman tissue specimens (see "Materials and Methods"). A represent

ative example of the staining pattern obtained for AP-2a in a human

breast cancer biopsy using the 3B5 monoclonal antibody is shown inFig. 1, A and B. The AP-2a staining pattern was strictly nuclear,

which is consistent with the known subcellular localization of thistranscription factor. The specific nuclear staining was completelyblocked when the antibody was preincubated with its correspondingpeptide epitope before immunostaining, demonstrating the specificityof the immune reaction (Fig. 1C). In addition, the nuclear stainingpattern was not visualized when breast cancer sections were stainedwith the secondary antibody alone (Fig. ID). We also determined thatthe 5E4 monoclonal antibody, which recognizes a different AP-2a

epitope. also generated an equivalent pattern of staining when used onadjacent tissue sections (data not shown). These data indicate theefficacy of these AP-2a immunological reagents.

To study AP-27 expression, we generated and characterized a novelepitope-selected rabbit polyclonal antiserum, 796, which does not

cross-react with AP-2or or AP-2ß(data not shown). The immunore-activity of breast sections stained with the -y96 antiserum indicated

that AP-27 protein was also localized to the nucleus (Fig. IE). The

staining was not visualized when the tissue sections were stained withthe secondary antibody alone (Fig. IF). Similarly, the specific nuclearstaining was blocked when the antibody was preincubated with therelevant peptide before incubation with breast tissue sections (Fig.1C). Taken together, these studies demonstrate the efficacy of ourimmunological reagents and indicate that the AP-2a and AP-2-y

proteins can be detected in breast cancer biopsies.AP-2 Expression in BBE. The presence of the AP-2a and AP-2-y

proteins in a breast cancer specimen might indicate that the expressionof these genes had been activated in the progression toward breastcancer. Alternatively, the presence of the AP-2 proteins might indicate

that this gene family is normally expressed during the growth anddevelopment of BBE. We therefore examined the expression ofAP-2a and AP-2-y in 12 BBE specimens from patients undergoing

reduction mammoplasty. Both of these proteins could be detected inthe ductal epithelium of BBE samples, where they were again predominantly nuclear in localization (data not shown). We found that 3of 12 (25%) BBE samples stained for AP-2a protein, and 4 of 12(33%) BBE cases expressed the AP-27 protein (Table 1). Variouscombinations of AP-2 protein expression were observed in BBE: (a)AP-2a alone; (b) AP-27 alone; (c) both AP-2o and AP-27; or (<f)

neither protein detected. We also noted that in comparison to thetumor tissue, both the intensity and distribution of AP-27 staining

were weaker in the BBE than those seen in the invasive cancerspecimens (see below).

AP-2 Expression in Human Breast Cancer. We next used thespecific AP-2 reagents to study the expression of AP-2a and AP-27 ina well-characterized population of patients for which breast cancerspecimens were available. AP-2a protein expression was detected at

significant levels in 28 of 80 (35%) breast tumor specimens analyzed(Table 1). Similar levels of AP-2a protein were observed in both the

breast cancer specimens and the BBE tissue samples discussed above(P = NS). In contrast, we found that 63 of 75 (84%) breast cancerscontained elevated levels of AP-27 protein, whereas only 33% of BBEcases had AP-27 immunoreactivity (P = 0.01). This finding indicatesthat there is a clear up-regulation of AP-27 expression during the

transition from BBE to invasive breast cancer.We were also able to analyze the data for the coexpression of

AP-2a and AP-27 m 74 of the breast tumor specimens. A total of 65

tumors (88%) contained appreciable levels of at least one of the twoAP-2 proteins, and concomitant expression of both proteins was

detected in 25 of 74 (34%) tumor specimens (Table 1). Only 9 of 74(12%) tumors lacked expression of both AP-2a and AP-27, which

Table 1 AP-2 prolt'in expression in BBE und 1DC

The expression of AP-2«protein in 80 breast cancer tumor specimens and AP-2y in

75 breast tumors was determined by imniunohislochemistry techniques, and nuclearstaining was scored independently by a single pathologist using the same histológica!criteria. An H-score ^ 50 was used as a cutoff for AP-2 protein expression. Seventy-foursamples were analyzed in terms of both AP-2a and AP-2-x expression. Differences in theexpression of AP-2 protein between BBE and IDC were determined using ^2 analysis.

AP-2a+AP-2a-A.P-2y+

AP-2f-AP-2a+/AP-2-y+

AP-2a-/AP-2-y+AP-2a+ /AP-2-y-AP-2a-/AP-2y-BBE3/12(25%)

9/12(75%)4/12(33%)

8/12(67%)2/12(17%)

2/12(17%)1/12(8%)7/12(58%)IDC28/80

(35%)52/80(65%)63/75

(84%)12/75(16%)25/74

(34%)37/74 (50%)

3/74 (4%)9/74(12%)PNS0.01NS

5468

Research. on January 23, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from

AP-2a AND AP-2y EXPRESSION IN HUMAN BREAST CANCER

Table 2 AP-2 protein expression and the association with DDFS and OS

The expression of AP-2 proleins in breast cancer paraffin-embedded specimens was

determined by immunohistochemistry techniques. DDFS and OS curves were calculatedby the life table method, with differences between the curves tested by Ine Mantel-Haenzel

statistics test.

ProteinexpressionAP-2a+

AP-2a-AP-2-X+

AP-2-X-1

0-yearDDFS74%

62%60%

69%10-year

OS65%

63%65%

51%PNSNS

Table 3 Correlation of AP-2y expression with ER tmtl PR levels

ER and PR status was determined by ¡mmunohistochemistry (24) with anH-score a 75 required for hormone receptor positivity. Differences in ER and PR statusbetween the population groups were compared using standard x~ analysis.

ER + ER-

AP-2-X+AP-2y-

24/75 (32%)4/75 (5%)

39/75 (52%)8/75(11%)

NS

PR + PR-

AP-2y+AP-2-y-

22/75 (29%)3/75 (4%)

41/75(55%)9/75(12%)

NS

stands in marked contrast to our findings in BBE, in which 7 of 12(58%) specimens expressed neither of these two proteins (Table 1).

The AP-2 family of transcription factors also contains the relatedAP-2ßgene. The expression of this gene was previously detected ata lesser frequency than AP-2a and AP-2y in established human breastcancer cell lines (11). Similarly, using an AP-2ß-specific antisera

(11), we found that the expression of this protein also occurred at alower frequency in human breast cancer specimens (data not shown),but accurate quantification of AP-2ßprotein is not possible at thistime with the available reagents. The SC-184 antiserum (see "Materials and Methods") is capable of recognizing AP-2a, AP-2ß,and, to

a lesser extent, AP-2y. By using this antiserum to examine theexpression status of AP-2ßin the nine tumors that were negative forboth AP-2a and AP-2y, we were able to determine that these samplesalso lacked AP-2ßexpression (data not shown). Taken together, our

data demonstrate that the majority of breast cancers surveyed expressone or more of the AP-2 proteins, whereas only a small fraction areAP-2 negative. These findings indicate that AP-2 expression may be

an important marker for breast tumor progression in vivo.Survival Characteristics of AP-2-expressing Breast Tumors.

The high incidence of AP-2a and AP-2y expression in mammary

tumors prompted us to examine whether these genes could act asmolecular markers for predicting prognosis. In particular, we wereinterested in determining whether breast tumors with altered levels ofAP-2 were associated with changes in survival rates. The 10-yearDDFS rate for breast cancer patients with high levels of AP-2a was74% compared with 62% for those with low levels of AP-2a(P = NS; Table 2). The 10-year OS for breast cancer patients withhigh and low levels of AP-2a was 65 and 63%, respectively(P = NS). The 10-year DDFS for patients with high levels of AP-2y

was 60%, and it was 69% for breast cancer patients with low levels ofAP-2y (P = NS; Table 2). Similarly, there was no difference in the10-year OS between patients with breast tumors containing highlevels of AP-2-y protein and those with breast tumors containing lowlevels of AP-2y protein. We also found that the local breast relapse-

free survival for breast cancer patients treated with lumpectomy andradiation therapy was not dependent on the level of either AP-2a orAP-2-y, suggesting that AP-2 is not critical in mediating effects related

to ionizing radiation (data not shown). Finally, the combined presence

of AP-2a and AP-2-y in a tumor sample did not influence the OS or

DDFS.Hormone Receptor Status and AP-2 Protein Levels. It has been

postulated that the AP-2-y protein may be one of the transcription

factors controlling ER expression in vitro (12). Therefore, the dataobtained for AP-2 expression status were analyzed in terms of the ER

and PR positivity of the breast tumors. We found no correlationbetween AP-2y expression and ER/PR positivity (Table 3). Similarly,there was no correlation between AP-2o status and PR expression.However, there was a significant correlation between AP-2a statusand ER positivity (Table 4). Whereas 15 of 28 (54%) AP-2a-positive

tumors were ER positive, only 14 of 52 (27%) tumors that did notexpress AP-2a were ER positive (P = 0.018). The breast tumors

could be further classified into those specimens that coexpressed bothAP-2a and AP-2y (double positives) and those that expressed only

one of these two proteins (single positives). ER expression wasdetected in 13 of 25 (52%) AP-2a/AP-2y double positive breastcancer specimens and 13 of 40 (33%) AP-2 single positive breastcancer specimens (P = 0.07). In the nine tumors that did not expressany AP-2 protein, two were still ER positive, whereas three were PRpositive. The latter finding indicated that the presence of AP-2 pro

teins was not essential for the transcription of these two hormonereceptors.

Coexpression of AP-2 and the Oncoprotein ERBB-2. The AP-2proteins can regulate ERBB-2 promoter activity in vitro, and theexpression of ERBB-2 correlates with the presence of one or moreAP-2 proteins in human breast cancer cell lines (10, 11). The presenceof ERBB-2 protein had previously been detected (7) at elevated levels

in 8 of 75 (11%) breast cancer specimens used in the current analysisof AP-2 expression. We found that only 2 of 75 (3%) breast tumorshad high levels of ERBB-2 protein when there were only low levels ofAP-2a (P = 0.015; Table 5). None of the breast tumors had ERBB-2immunoreactivity when there were low levels of AP-2-y. However, the

most striking correlation concerned the expression of ERBB-2 intumors that contained both AP-2a and AP-2y protein. Of the 38tumors that expressed either AP-2a or AP-2y alone, only 1 had highlevels of ERBB-2. In marked contrast, 6 of 24 (25%) AP-2a/AP-2y

Table 4 Correlation of AP-2a expression with ER tinti PR levels

ER and PR status was determined by immunohistochemistry (24) with anH-score ^ 75 required for hormone receptor posilivily. Differences in ER and PR statusbetween the population groups were compared using standard x~ analysis.

A. ER + ER-

AP-2a +AP-2«-

15/80(19%)14/80(17.5%)

13/80(16%)38/80 (47.5%)

(MMX

B. PR + PR-

AP-2a +AP-2a-

9/80(11%)17/80(21%)

19/80(24%)35/80(44%)

NS

Table 5 Coexpression of AP-2 tint! ERBB-2

The expression of AP-2 and ERBB-2 proteins in breast cancer specimens was determined by immunohistochemical techniques.

ProteinexpressionAP-2a-

AP-2a+AP-2y-

AP-2y+AP-2a-/AP-2-y-

AP-2a+ or AP-2-V+AP-2a+/AP-2y+ERBB-2

+2/75

(3%)6/75(8%)0/70

(0%)7/70(10%)0/69

(0%)1/69(1%)6/69 (9%)ERBB-2-46/75

(61%)21/75(28%)11/70(16%)

52/70(74%)8/69(12%)

36/69(52%)18/69(26%)P0.0150.220.003

5469

Research. on January 23, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from

AP-20 AND AP-2> EXPRESSION IN HUMAN BREAST CANCER

double-positive tumors expressed ERBB-2 (P = 0.003). Our findingsstrongly support previous in vitro studies that indicate that AP-2 maybe an important transcriptional regulator of ERBB-2.

Coexpression of AP-2 and the IGF-IR. The correlations foundbetween AP-2, ERBB-2, and ER expression prompted us to examine

another signaling molecule that has been implicated in breast cancer,namely, the IGF-IR. Overexpression of the IGF-IR may predict for

poor survival characteristics in breast cancer patients, and the receptorhas been demonstrated to mediate an antiapoptotic phenotype (26).The presence of the IGF-IR protein had previously been detected (22)

in 17 of 75 (23%) breast cancer specimens used in the current analysisof AP-2 expression. Significantly, we found that elevated levels ofIGF-IR expression did not occur in the absence of AP-2y (P = 0.04;

Table 6). Similarly, in the nine samples that were scored as negativefor all AP-2 proteins, we did not find sections with high expression ofthe IGF-IR (P = 0.08). In contrast, there was no association betweenthe expression of AP-2a and the levels of the IGF-IR or betweenAP-2of and AP-2y double positivity and IGF-IR expression. Together,these data indicate that the AP-2 gene family, in particular, AP-2y,may be important for the regulation of IGF-IR expression in human

breast cancer.Identification of a Proximal AP-2-binding Site in the IGF-IR

Promoter. Previous EMSA studies of the rat IGF-IR promoter indicated that a weak AP-2-binding site might be present between nucle-otides —331and —135(27). Our further examination of the human

and rat IGF-IR promoter sequence revealed that a more proximalAP-2 site might also exist between -90 and -80. This sequence,5'-GCCCTCGGC-3', was conserved between both species (25) andmatched the previously identified consensus AP-2-binding site, 5'-GCCNNNGGC-3'. When a labeled oligonucleotide corresponding to

this IGF-IR sequence was used as a probe for EMSA, a slower-migrating band was produced upon the addition of AP-2a protein.This protein-DNA complex was competed away by an excess of anunlabeled oligonucleotide corresponding to the standard AP-2-bind-

ing site from the human metallothionein Ila gene (Fig. 2). Additionalcompetition studies confirmed that the proximal IGF-IR sequenceacted as an AP-2-binding site, although it was approximately 10-foldweaker in its ability to bind AP-2 than the standard metallothionein

site (data not shown).

DISCUSSION

Most breast tumors arise spontaneously in women with no overtfamily history of breast cancer and do not appear to involve deleterious mutations in BRCA1 and BRCA2 (4). Therefore, it is critical toidentify and characterize other genes that may regulate the progression toward mammary carcinoma. One successful approach to identifynew genes involved in breast cancer has been to determine how theexpression of known prognostic indicators, such as the ER andERBB-2 genes, are up-regulated in breast cancer cell lines. In vitro

Table 6 Cotxpftahm nfAP-2 ami IGF-IRThe expression of AP-2 and IGF-IR in breast cancer lumpectomy specimens was

determined by ¡mmunohislochemisiry techniques. AP-2—indicates that none of the threeAP-2 proteins was detected in a tumor. AP-2+ indicates lhat at least one of Ihe three AP-2

proteins was present.

ProteinexpressionAP-2a-

AP-2a+AP-2-y-AP-2-X+AP-2-

AP-2 +IGF-IR

+14/80(17%)

5/80(6%)0/75

(0%)17/75(23%)0/74

(0%)17/74(23%)1GF-IR-38/80

(48%)23/80(29%)12/75(16%)

46/75(61%)9/74(12%)

48/74 (65%)PNS0.040.08

probe

competitor

IGF-I-R

hMtlla

1234

IGF-I-R

hMtlla

consensus

GGGCCGGGCCCTCGGCGCGCCG

TGACCGCCCGCGGCCCGTG

GCCNNNGGCFig. 2. Identification of an AP-2-binding site in the proximal IGF-IR promoter by

EMSA. The sequences of the IGF-IR and human metallothionein Ha (hMtlla) probes andcompetitor oligonucleotides are given, and the homology to the AP-2 consensus is shownin bold. All lanes contain the IGF-IR probe and rabbit reticulocyte lysate programmedwith AP-2cr mRNA. Lane I. no competitor. Lanes 2-4 contain increasing amounts (1. 10,

and 50 ng) of the human metallothionein Ila competitor oligonucleotide.

studies examining the c/i-regulatory sequences responsible for the

activation of these two genes have implicated the same transcriptionfactor family: AP-2 (10-12). Our current in vivo examination of AP-2

gene expression in breast cancer specimens also strongly supports arole for these transcription factors in the etiology of human breastcancer. The majority of tumor biopsies examined (88%) express atleast one of the AP-2 proteins. This finding largely reflects a significant increase in AP-2y expression in the progression from BBE toIDC. The presence of AP-2a and AP-2y in breast cancer could simply

reflect the normal expression of these genes in the ductal epithelia ofthe healthy mammary gland. Under these circumstances, the AP-2

proteins may represent benign markers of ductal epithelia that play norole in the growth and proliferation of this tissue during the progression to mammary cancer. However, our evidence strongly supportsthe alternative hypothesis, namely, that the AP-2 genes are intimately

involved in controlling growth factor expression in human breastcancer.

We had previously shown that the AP-2 proteins bind to a criticalregion of the ERBB-2 promoter and are capable of activating ERBB-2expression in vitro (10, 11). In addition, we had found that ERBB-2

expression in human breast cancer cell lines only occurred in thepresence of the AP-2 proteins (10, 11). The data obtained in vivoreinforce the connection between AP-2 and ERBB-2 expression, be-

5470

Research. on January 23, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from

AP-2a AND AP-2-y EXPRESSION IN HUMAN BREAST CANCER

cause no significant ERBB-2 protein was detected in tumors lackingAP-2. Indeed, the majority of ERBB-2 expression was limited totumors containing both the AP-2a and AP-2-y proteins. Moreover, a

reexamination of our previous in vitro data also revealed a highincidence of AP-2a/AP-2y coexpression in breast cancer cell linesthat are ERBB-2 positive (11). One possible explanation for thesecorrelations comes from the finding that the AP-2a and AP-2y genes

can form functional heterodimers in vitro (11). Therefore, the correlation between ERBB-2 expression and AP-2o/AP-2y double positiv-

ity in vivo may indicate that there is a preference for heterodimersbetween these two transcription factors in the activation of ERBB-2transcription. It is also noteworthy that ERBB-2 expression was notdetected in all mammary tumors that contained both AP-2a andAP-2y proteins. This finding indicates that the AP-2 transcription

factors represent only one component of the regulatory hierarchyrequired for ERBB-2 promoter activity (28). Although AP-2 expression may not be sufficient for ERBB-2 transcription, our present

findings strongly support previous in vitro studies that suggest that theAP-2 proteins are critical transcriptional regulators of ERBB-2 geneexpression in breast cancer. A recent study of childhood medulloblas-toma also indicated that the AP-2 gene family may be involved in the

regulation of epidermal growth factor receptor family gene expressionin this class of tumor. In medulloblastomas, the presence of the AP-2

proteins (the various family members were not distinguished) showedsignificant correlations with the expression of ERBB-2, ERBB-3, andERBB-4 (29). Therefore, it seems likely that AP-2 transcription fac

tors may play a major role in the regulation of the epidermal growthfactor receptor gene family in many different classes of tumor.

Previous studies in breast cancer cell lines had also indicated thatAP-2 may regulate ER expression. The AP-2y protein was isolatedfrom the MCF-7 breast carcinoma cell line based on its ability to

interact with an important regulatory element present within the ERgene (12). It has yet to be demonstrated that AP-2 can directly activate

ER transcription in vitro, but our in vivo findings support a role forAP-2 in the regulation of ER expression in breast carcinoma. Inparticular, we detected a significant correlation between AP-2a expression and ER positivity. This finding reveals that AP-2a may be amore important regulator of ER expression than AP-2-y. Finally, with

respect to the ER, several tumors were found that expressed this genein the absence of any AP-2 protein. Therefore, it is apparent that AP-2

is not absolutely required for ER transcription. Nevertheless, thecorrelations that we observed again link AP-2 with a growth factor

signaling pathway that is critical for the outcome of human breastcancer.

Our data also implicate the AP-2 gene family in the regulation ofIGF-IR expression in mammary cancer. The IGF-IR is highly ex

pressed in several types of cancer, including breast cancer, and itsregulation may be important for the development of malignant transformation (26). The expression of the IGF-IR is hormonally regulated,

and the transcription of this gene is also dependent upon transcriptionfactor Spl (30). Our data suggest that AP-2 may also be an importantregulator of IGF-IR gene expression. In support of this hypothesis,previous studies have shown that AP-2 can interact with the distalregion of the IGF-IR promoter (27), and we have now identified amore proximal AP-2-binding site located between nucleotides -80and —90.Additional studies will be required to determine whether theAP-2 proteins can directly regulate IGF-IR expression, although inthis regard, cotransfection experiments have implicated the AP-2proteins as transcriptional regulators of another insulin-like growthfactor regulatory protein, IGFBP-5 (31 ). Our analysis of breast cancerbiopsies indicated that AP-2y displayed the most significant correlation with IGF-IR gene expression. Taken together with the AP-2a/ERcorrelation, these data indicate that the closely related AP-2a and

AP-2y transcription factors may have different downstream targets or

may differentially regulate the same downstream targets.The data obtained in our analysis strongly support an important role

for the AP-2 proteins in the control of gene expression in breastcancer. The finding that both AP-2a and AP-2y are expressed to alimited extent in BBE raises the possibility that the AP-2 genes are

also required for normal mammary gland development. However,because AP-2a knockout mice die perinatally and lack the ventral

body wall, the role of this transcription factor in the normal development and maturation of the mammary gland remains to be determined(16, 17). Similarly, no information is currently available on the role ofthe AP-2y gene in mammary gland development. Nevertheless, giventhe requirement of the AP-2 family for embryonic growth and morphogenesis (16-18), it is striking that the AP-2a and AP-2y genes are

expressed in the ductal epithelia of the normal breast, a tissue thatundergoes profound growth and remodeling. We therefore postulatethat these genes are involved in normal breast development, and thattheir function is subverted during the progression toward breast cancer. Further support for the role of AP-2 in mammary gland biology

comes from the study of gene expression. In addition to our findings,which demonstrate a connection between AP-2 and the in vivo ex

pression of three growth factor receptors in the breast, importantAP-2-binding sites are also associated with the regulatory regions ofcollagenase, p21wall/Lir>l, integrins, cytokeratins, E-cadherin, and

mouse mammary tumor virus (32-39). Additional studies are clearlyneeded to determine how the AP-2 genes are regulated during mam

mary gland development and to ascertain how these proteins influencethe growth and maturation of this organ.

Evidence obtained in vitro concerning the role of AP-2 in

cellular transformation suggests two alternative hypotheses for thepresence of increased expression of AP-2 in breast cancer. Onepossibility is that the AP-2 proteins are proliterative signals that

activate a variety of growth factor pathways. The inappropriateactivation of the AP-2 proteins would thus drive ductal epithelia

cells continually through the cell cycle. This theory would besupported by experiments in which overexpression of the AP-2agene increased the oncogenic potential of PA-1 teratocarcinomacells (40). An alternative hypothesis is that the AP-2 proteins aremore akin to tumor suppressors, and the activation of the AP-2

genes would represent a failed attempt to halt cell proliferation. Inthis instance, the up-regulation of AP-2 target genes would occuras a secondary consequence of increased AP-2 levels. Several lines

of evidence also support this theory. First, transfection of anAP-2a expression construct into HepG2 hepatocarcinoma cells orSW480 adenocarcinoma cells can inhibit their growth and tumor-

igenicity (38). The observed decrease in cell growth correlated withan increase in the expression of the AP-2 target gene, p2lwa"/c'P'.

Chromosomal mapping of AP-2a also supports a role for this gene asa potential tumor suppressor. The human AP-2a gene, TFAP2A, islocated at chromosome 6p24, a region associated with loss of het-

erozygosity in several types of cancer, including follicular center celllymphoma, ovarian carcinoma, and head and neck squamous cellcarcinoma (41-44). It is also noteworthy that studies in cell-freesystems have demonstrated that the AP-2a protein can directly interact with SV40 large T antigen, human T-cell lymphotrophic virus type

I tax, and adenovirus EIA in vitro (36, 37, 45). The ability of theseproteins to interact with AP-2 may explain in part how these viral

transactivators corrupt normal cellular function. Taken together, thesein vitro studies establish a link between the AP-2 proteins, cell cycle

control, and cellular transformation. Clearly, additional experimentswill be required to understand the in vi'vo role of the AP-2 gene family

in cell fate determination. Knowledge of how the AP-2 proteins

5471

Research. on January 23, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from

AP-2o AND A.P-2y EXPRESSION IN HUMAN BREAST CANCER

regulate growth factor signaling in vivo may lead to important applications in the treatment and diagnosis of human breast cancer.

ACKNOWLEDGMENTS

We thank S. E. Turner. M. DiGiovanna. D. Stern, and the Section of CriticalTechnologies of the Yale University School of Medicine for assistance. T. W.thanks the Pew Scholars Program in the Biomedicai Sciences for their pastsupport.

REFERENCES1. Ben/. C. C'.. Brand!. B. H.. and Zänker.K. S. Gene diagnostics provide new insights

into breast cancer prognosis and therapy. Gene (Amst.). 159: 3-7. 1995.2. Hynes. N. E.. and Stem. D. F. The biology of erbB-2/neu/HER-2 and its role in

cancer. Biochim. Biophys. Acta. 1198: 165-184, 1994.

3. Steeg. P. S.. and Abrams. J. S. Cancer prognostics: past, present and p27. Nat. Med..3: 152-154, 1997.

4. Stratlon. M. R., and Wooster. R. Hereditary predisposition to breast cancer. Curr.Opin. Genet. Dev.. 6: 93-97, 1996.

5. Walker. R. A., and Varley. J. M. The molecular pathology of human breast cancer. In:N. R. Lemoine and N. A. Wright (eds.). The Molecular Pathology of Cancer, pp.31-57. Cold Spring Harbor. NY: Cold Spring Harbor Laboratory Press. 1993.

6. Chevallier. B„Heint/.mann. F.. Mosseri. V.. Dauce. J. P.. Bastit. P.. Graie. Y..Brunelle. P.. Basuyau. J. P.. Comoz. M.. and Asselain, B. Prognostic value of estrogenand progesterone receptors in operable breast cancer. Results of a univariate andmullivariate analysis. Cancer (Phila.). 62: 2517-2524, 1988.

7. Haffly. B. G., Brown. F.. Carter. D., and Flynn, S. Evaluation of HER-2/neu

oncoprotein expression as a prognostic indicator of local recurrence in conservativelytreated breast cancer: a case-control study. Int. J. Radial. Oncol. Biol. Phys., 35:751-757. 1996.

8 McGuire. W. L. Hormone receptors: their role in predicting prognosis and responseto endocrine therapy. Semin. Oncol.. 5: 428-433, 1978.

9. King. R. J. B. A discussion of the roles of oestrogen and progestin in humanmammary carcinogencsis. J. Steroid Biochern. Mol. Biol., 39: 811-818. 1991.

10. Bosher. J. M.. Williams. T. and Hurst. H. C. The developmentally regulated transcription factor AP-2 is involved in c-erbB-2 overexpression in human mammarycarcinoma. Proc. Nati. Acad. Sci. USA, 92: 744-747, 1995.

11. Bosher, J. M.. Totty, N. F., Hsuan, J. J., Williams. T.. and Hurst. H. C. A family ofAP-2 proteins regulates c-erbB-2 expression in mammary carcinoma. Oncogene. 13:1701-1707. 1996.

12. McPherson. L. A.. Baichwal. V. R.. and Weigel, R. J. Identification of ERF-1 as amember of the AP-2 transcription factor family. Proc. Nati. Acad. Sci. USA, 94:4342-4347. 1997.

13. Chazaud, C.. Oulad-Abdelghani. M.. Bouillel. P.. Décimo.D.. Chambón. P.. andDollé.P. AP-2.2. a novel gene related to AP-2. is expressed in the forebrain. limbsand lace during mouse embryogenesis. Mech. Dev.. 54: 83-94. 1996.

14. Mitchell. P. J.. Timmons, P. M., Hébert,J. M.. Rigby. P. W. J.. and Tjian, R.Transcription factor AP-2 is expressed in neural crest cell lineages during mouseembryogenesis. Genes Dev., 5: 105-119. 1991.

15. Moscr. M.. Imhof. A.. Pscherer, A.. Bauer. R.. Amselgruber. W.. Sinowaiz. F..Hofstädler,F., Schule. R., and Buettner. R. Cloning and characterization of a secondAP-2 transcription factor: AP-20. Development (Camb.), 121: 2779-2788. 1995.

16. Schorle. H.. Meier. P.. Buchen, M.. Jaenisch. R., and Mitchell. P. J. Transcriptionfactor AP-2 essential for cranial closure and craniofacial development. Nature(Lond.), 381: 235-238, 19%.

17. Zhang. J.. Hagopian-Donaldson. S.. Serbedzija. G.. Elsemore. J.. Plehn-Dujowich. D..McMahon. A. P.. Flavell, R. A., and Williams, T. Neural tube, skeletal and body walldefects in mice lacking transcription factor AP-2. Nature (Lond.). 381: 238-241,

1996.18. Moser. M., Pscherer, A.. Roth, C., Becker. J., Mucher, G., Zerres, K.. Dixkens, C.,

Weis. J.. Guay-Woodford. L.. Buettner. R.. and Fassler. R. Enhanced apoptolic celldeath of renal epithelial cells in mice lacking transcription factor AP-2ß.Genes Dev..//: 1938-1948. 1997.

19. Williams. T.. and Tjian. R. Analysis of the DNA binding and activation properties ofthe human transcription factor AP-2. Genes Dev.. 5: 670-682. 1991.

20. deConinck, E. D.. McPherson. L. A., and Weigel. R. J. Transcriptional regulation ofestrogen receptor in breast carcinomas. Mol. Cell. Biol.. 15: 2191-2196. 1995.

21. Creaser, P.. D'Argenio. D. A., and Williams. T. Comparative and functional analyses

of the AP-2 promoter indicate that conserved octamer and initiator elements arecritical for activity. Nucleic Acids Res.. 13: 2597-2605. 1996.

22. Turner. B. C., Haffty, B. G.. Narayanan. L.. Yuan. J.. Havre. P. A.. Gumbs, A. A..Kaplan. L.. Burgaud. J-L.. Carter. D.. Baserga. R.. and Glazer. P. M. Insulin-like

growth factor 1 receptor overexpression mediates cellular radioresistance and localbreast cancer recurrence after lumpectomy and radiation. Cancer Res., 57: 3079-

3083, 1997.23. Williams. T., Admon. A.. Lüscher.B., and Tjian. R. Cloning and expression of AP-2.

a cell-type-specific transcription factor that activates inducible enhancer elementsGenes Dev., 2: 1557-1569. 1988.

24. King, W. J.. DeSombre, E. R., Jensen, E. V., and Greene. G. L. Comparison ofimmunocytochemical and steroid-binding assays for estrogen receptor in humanbreast tumors. Cancer Res., 45: 293-296. 1985.

25. Cooke. D. W., Banken, L. A.. Roberts, C. T. J.. LeRoith. D.. and Casella. S. J.Analysis of the human type I insulin-like growth factor receptor promoter region.Biochem. Biophys. Res. Commun., 177: 1113-1120. 1991.

26. Baserga, R. The insulin-like growth factor I receptor: a key to tumor growth? CancerRes., 55: 249-252, 1995.

27. Werner. H., Bach, M. A.. Stannard. B., Roberts, C. T.. and LeRoith, D. Structural andfunctional analysis of the insulin-like growth factor I receptor gene promoter. Mol.Endocrinol.. 6: 1545-1558. 1992.

28. Bales. N. P.. and Hurst. H. C. Transcriptional regulation of type 1 receptor tyrosinekinases in the mammary gland. J. Mammary Gland Biol. Neoplasia. 2: 153-163.

1997.29. Gilbertson, R. J.. Perry, R. H., Kelly, P. J., Pearson, A. D. J.. and Lunec. J. Prognostic

significance of HER2 and HER4 coexpression in childhood medullohlastoma. CancerRes., 57: 3272-3280. 1997.

30. Werner. H.. Hernandez-Sanchez, C.. Karnieli. E., and LeRoith, D. The regulation ofIGF-I receptor gene expression. Int. J. Biochem. Cell Biol.. 27: 987-994. 1995.

31. Duan, C.. and Clemmons, D. R. Transcription factor AP-2 regulates human insulin-like growth factor binding protein-5 gene expression. J. Biol. Chem., 270: 24844-

24851. 1995.32. Byrne. C. Regulation of gene expression in developing epidermal epithelia. Bioes-

says, 19: 691-698. 1997.

33. Fini, M. E.. Bartlell. J. D.. Matsuhara. M., Rinehart, W. B., Mody, M. K.. Girard.M. T., and Rainvillc. M. The rabbit gene for 92-kDa matrix mctalloprotcinase. Roleof AP-1 and AP-2 in cell type-specific transcription. J. Biol. Chem.. 269: 28620-

28628. 1994.34. Hennig, G.. Lowrick. O.. Birchmeier. W.. and Behrens, J. Mechanisms identified in

the transcriptional control of epithelial gene expression. J. Biol. Chem., 277: 595-

602, 1996.35. Mellentin-Michelotti, J.. John. S., Pennie, W. D.. Williams, T.. and Hager. G. L. The

5' enhancer of the mouse mammary tumor virus long terminal repeat contains a

functional AP-2 element. J. Biol. Chem., 269: 31983-31990. 1994.

36. Mitchell. P. J.. Wang, C.. and Tjian. R. Positive and negative regulation of transcription in rilro: enhancer-binding protein AP-2 is inhibited by SV40 T antigen. Cell. 50:847-861, 1987.

37. Somasundaram. K.. Jayaraman, G.. Williams. T., Moran, E., Frisch. S., andThimmapaya. B. Repression of a matrix metalloprotea.se gene by EIA correlates withits ability to bind to cell type-specific transcription factor AP-2. Proc. Nati. Acad.Sci. USA. 93: 3088-3093. 1996.

38. Zeng. Y-X., Somasundaram. K.. and EI-Deiry. W. F. AP-2 inhibits cancer cell growthand activates p21w»'"':ln|expression. Nat. Genet., 15: 78-82, 1997.

39. Zulter, M. M.. Sanloro, S. A.. Painter. A. S.. Tsung, Y. L., and Gafford. A. The humana2 integrin gene promoter. Identification of positive and negative regulatory elementsimportant for cell-type and developmentally restricted gene expression. J. Biol.Chem.. 269: 463-469. 1994.

40. Kannan. P.. Buetlner. R.. Chiao, P. J.. Yim, S. O.. Sarkiss. M.. and Tainsky. M. A.N-raj oncogene causes AP-2 transcriptional self-interference, which leads to transformation. Genes Dev.. 8: 1258-1269, 1994.

41. Califano, J.. van der Riet. P.. Westra, W., Nawroz. H.. dayman. G.. Pianladosi. S..Corio. R.. Lee. D.. Greenberg. B.. Koch. W.. and Sidransky. D. Genetic progressionmodel for head and neck cancer: implications for field cancerization. Cancer Res., 56:2488-2492, 1996.

42. Davies, A. F.. Stephens. R. J., Olavesen, M. G.. Heather. L.. Dixon. M. J.. Magee. A..Flinter. F.. and Ragoussis. J. Evidence of a locus for orofacial clcfling on humanchromosome 6p24 and STS content map of the region. Hum. Mol. Genet., 4:121-128, 1995.

43. Foulkes, W. D., Ragoussis. J.. Stamp, G. W. H.. Allan. G. J., and Trowsdale, J.Frequent loss of heterozygosity on chromosome 6 in human ovarian carcinoma. Br. J.Cancer, 67: 551-559. 1993.

44. Randerson, J., Cawkwell. L.. Jack, A.. Lewis, F., Johnson. P.. Evans, P., Barrans. S..and Morgan. G. J. Fluorescent polymerase chain reaction of a panel of CA repeats onchromosome 6 in the indolent phase of follicular centre cell lymphoma. Br. J. Cancer.74: 942-946. 1996.

45. Muchardt. C.. Seeler, J-S.. Nirula. A.. Gong, S.. and Gaynor. R. Transcription factorAP-2 activates gene expression of HTLV-I. EMBO J., //: 2573-2581. 1992.

5472

Research. on January 23, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from

1998;58:5466-5472. Cancer Res   Bruce C. Turner, Jian Zhang, Andrew A. Gumbs, et al.   Signalling Pathways

FactorCancer Correlates with the Regulation of Multiple Growth Expression of AP-2 Transcription Factors in Human Breast

  Updated version

  http://cancerres.aacrjournals.org/content/58/23/5466

Access the most recent version of this article at:

   

   

   

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

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

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

  Permissions

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

.http://cancerres.aacrjournals.org/content/58/23/5466To request permission to re-use all or part of this article, use this link

Research. on January 23, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from