Activator Protein-2 Overexpression Accounts for Increased Insulin Receptor Expression in Human Breast Cancer

The peptide hormone insulin exerts its biological effects by interacting with the insulin receptor (IR), a specific glycoprotein in the plasma membrane of insulin target cells, which generates intracellular signals via a tyrosine kinase cascade (1). The IR belongs to the tyrosine kinase growth factor receptor family and consists of two identical extracellular α-subunits that contain the insulin-binding domain and two transmembrane β-subunits that have ligand-activated tyrosine kinase activity (1–3). When insulin binds to the IR, the receptor is first activated by tyrosine autophosphorylation, and then the IR tyrosine kinase phosphorylates various effector molecules, such as IRS-1, leading to hormone action (1–3). Overexpression of receptor molecules with intrinsic tyrosine kinase activity in cells has been involved in the induction of a transformed phenotype (4, 5), and the mediation of a proliferative response through the IR has been widely shown in various transformed cells (6–8). In epithelial cells, the IR is usually expressed at low levels. However, it has been shown experimentally that overexpression of IRs in these cells can increase responses to insulin and induce a ligand-mediated neoplastic transformation. In this regard, it has been shown that overexpression of functional IRs can occur in human breast cancer and other epithelial tumors, including ovarian and colon cancer, in which the IR may exert its oncogenic potential by directly affecting cell metabolism and/or by synergizing with other oncogenes in influencing growth and differentiation (6–10). However, very little is known about the molecular mechanisms that account for the increased IR expression in human breast cancer, and understanding these mechanisms can be of valuable help from both biological and clinical aspects.

As part of an investigation into the molecular basis of regulation of the human IR gene, we previously showed that IR gene expression in eukaryotic cells is positively regulated by the architectural factor high-mobility group A1 (HMGA1; refs. 11–13), a small basic protein that binds to AT-rich regions of certain gene promoters and functions mainly as a specific cofactor for gene activation (14–17). HMGA1 by itself has no intrinsic transcriptional activity; rather, it has been shown to trans-activate promoters through mechanisms that facilitate the assembly and stability of stereospecific DNA-protein complexes that drive gene transcription (14–17). HMGA1 performs this task by modifying DNA conformation and by recruiting transcription factors to the transcription start site, facilitating DNA-protein and protein-protein interactions (14–17). We showed that transcriptional activation of the human IR promoter requires the assembly of a transcriptionally active multiprotein-DNA complex, which includes, in addition to HMGA1, the ubiquitously expressed transcription factor Sp1 and the CCAAT-enhancer binding protein β (C/EBPβ; ref. 12). By physically interacting with Sp1 and C/EBPβ, HMGA1 facilitates the binding of both factors to the IR promoter in vitro and supports transcriptional synergism between these factors in vivo (12).

The transcription factor activator protein-2 (AP-2) activates transcription via GC-rich DNA sequences (18, 19), and its expression is required for normal growth and morphogenesis during mammalian development (20). An involvement of AP-2 family of proteins in the etiology of human breast cancer has been shown in previous studies, in which a significant up-regulation of AP-2α in breast cancer specimens has been observed (21). In addition, AP-2 genes are expressed in many human breast cancer cell lines, and AP-2 consensus binding sites can be detected in both c-erbB-2 and estrogen receptor promoters, whose activity is increased in breast cancers (22, 23). In our study, overexpression of the IR in both human breast cancer cell lines and in whole breast tumor tissue correlated closely with AP-2α mRNA and protein expression levels. Therefore, to explore the biochemical mechanisms involved in IR overexpression in breast cancer, we have examined the functional significance of AP-2α for IR gene transcription using the human AP-2-deficient HepG2 cell line. There are no AP-2 DNA-binding sites on the promoter region of the IR gene; however, overexpression of AP-2 in HepG2 cells greatly enhanced IR gene transcription, supporting a role for this transcription factor in the regulation of IR gene expression. In this article, we show for the first time that AP-2 physically interacts with HMGA1 and Sp1 in vitro and in vivo, and physical interaction between these factors contributes to efficient functional cooperation in the transactivation of the IR gene by AP-2. Our findings show that AP-2α may play significant molecular roles in IR overexpression in malignant breast cells and provide mechanistic insight into the etiology of human breast cancer.

Cells and protein extracts. HepG2 human hepatoma cells, human cervix epithelioid carcinoma cells (HeLa), and human normal and breast cancer cells (HBL-100, MCF-7, MDA-MB 231, MDA-MB 468, and MDA-MB 453; American Type Culture Collection, Manassas, VA) were maintained in DMEM (Life Technologies, Inc., Carlsbad, CA) supplemented with 10% fetal bovine serum. Specimens of tissue from patients with breast cancer were collected intraoperatively and frozen in liquid nitrogen. Tumor tissue with increased IR expression was selected and analyzed for each patient together with the healthy surrounding tissue. Nuclear and cytoplasmic extracts were prepared from cultured cells and breast tissue as described previously (11). For each extract, an equal number of nuclei were homogenized, and the final protein concentration in the extracts was determined by the modified Bradford method (Bio-Rad, Hercules, CA).

Reverse transcription-PCR, immunoprecipitation, and Western blot analysis. Total cellular RNA was extracted from cells and tissues using the RNAqueous-4PCR kit and subjected to DNase treatment (Ambion, Austin, TX). cDNA was synthesized from total RNA with the RETROscript first strand synthesis kit (Ambion) and used for PCR amplification. PCR products were electrophoretically resolved on 2% agarose gel, transferred to GeneScreen-plus nylon membranes (DuPont, Fayetteville, NC), and hybridized with ³²P-labeled PCR-generated probes for the IR, AP-2α, C/EBPβ, and tubulin. For quantitation of specific mRNA, hybridized membranes were exposed to X-ray film as needed to provide adequate intensity of bands, and the band intensities of the autoradiograms were quantified by densitometry. The values obtained were then normalized to those of the tubulin and C/EBPβ genes. Western blot analyses of IR, AP-2, and C/EBPβ were done on total cellular lysates or nuclear extracts from cells and tissues as previously described (11–13). For immunoprecipitation studies with HMGA1 or Sp1 antibody, aliquots of HeLa or HepG2 nuclear extract were incubated for 12 hours with rotation at 4°C with 10 μL of antibody-coupled protein A beads. Beads were recovered by gentle centrifugation and washed thrice with 500 μL of NETN wash buffer [0.1% NP40, 150 mmol/L NaCl, 1 mmol/L EDTA, 50 mmol/L Tris-HCl (pH 8)] for 5 minutes. Protein was removed from the beads by boiling in sample buffer for 5 minutes and analyzed by SDS-PAGE and immunoblotting. Antibodies used for these studies were as follows: anti-HMGA1 (11), anti-Sp1 (PEP 2; 1:500), anti-AP-2α (H-79; 1:1,000), and anti-IRβ (C-19; 1:1,000; Santa Cruz Biotechnology, Santa Cruz, CA). For covalent coupling of antibodies to protein A-Sepharose (Amersham, Piscataway, NJ), anti-HMGA1 or anti-Sp1 polyclonal antibody was mixed with beads and bound for 1 hour with rotation at room temperature. After extensive washing with 200 mmol/L sodium borate (pH 8), solid dimethyl pimelimidate (Sigma, St. Louis, MO) was added to a final concentration of 20 mmol/L, and the components were mixed on roller for 30 minutes at room temperature. To stop the reaction, the beads were washed twice in 200 mmol/L ethanolamine (pH 8) and incubated on roller for 2 hours at room temperature in 200 mmol/L ethanolamine. Antibody-coupled protein A beads were washed twice in PBS solution and used in immunoprecipitation studies.

Glutathione S-transferase pull-down assay. ³⁵S-labeled AP-2 and its deletion mutants were synthesized in vitro by using the TNT-T7 quick-coupled transcription/translation system (Promega, Madison, WI). Glutathione S-transferase (GST) fusion protein expression vectors for HMGI and derivatives (kindly provided by Dimitris Thanos), Sp1 (a kind gift from Hans Rotheneder and Erhard Wintersberger), and the pGEX-2TK control vector (Amersham) were transformed into the BL21 strain of Escherichia coli (Stratagene, La Jolla, CA), expanded in suspension culture, and induced for 2 hours with 0.5 mmol/L isopropyl-d-thiogalactopyranoside (Sigma). Bacteria were pelleted, sonicated in ice-cold PBS lysis buffer [1% NP40; 10% glycerol; 1 mmol/L DTT; 1 mmol/L phenylmethylsulfonyl fluoride (PMSF); 10 μg of pepstatin A, leupeptin, and aprotinin/mL; and 0.4 mg of lysozyme/mL], and centrifuged. The resultant supernatant was then added to 300 μL of glutathione-agarose beads, mixed on a rotating wheel at 4°C for 2 hours, and centrifuged. Bound GST-fused proteins in the pellet were washed five times with lysis buffer and resuspended in 300 μL of binding buffer [50 mmol/L NaCl, 20 mmol/L Tris-HCl (pH 8), 0.05% NP40, 0.25% bovine serum albumin, 1 mmol/L PMSF, 1 mmol/L DTT]. Bound protein was quantitated with the Coomassie protein assay reagent (Pierce Co., Rockford, IL), and 0.5 μg of each GST-protein bound to glutathione-agarose beads was incubated with 7 μL of in vitro translated ³⁵S-labeled protein in 150 μL of binding buffer at 4°C for 2 hours. Reactions were terminated by centrifugation; the precipitate was washed thrice with protein binding buffer and subjected to a 10% acrylamide SDS-PAGE; and proteins were visualized by autoradiography.

Oligonucleotides and gel shift analysis. Wild-type and mutant C2 (300 bp) sequence of the human IR promoter gene was generated by PCR amplification using the previously described recombinant plasmid pCAT-C2 (24) and its mutagenized clones as template (see below) and the following primers: C2 forward, 5′-TCGAGTCACCAAAATAAACAT-3′ (for wild-type and mutated oligonucleotides); C2 reverse, 5′-TGCAGGGGAGGGAGGTGCCGC-3′ (for wild-type oligonucleotides); pCAT-C2 reverse, 5′-ATTGGGGATATATCAACGGTGGTATATCC-3′ (for mutated oligonucleotides). ³²P-labeled C2 was used in gel shift assays as previously described (11). Double-stranded 26-mer oligonucleotides containing wild-type or mutated binding sites for AP-2α (Santa Cruz Biotechnology) were used in competition studies and decoy experiments.

Chromatin immunoprecipitation. Chromatin immunoprecipitation assay was done as described previously (13), using HepG2 cells (1 × 10⁷) transfected with the pC-hu/AP-2α expression vector. Formaldehyde-fixed DNA-protein complex was immunoprecipitated with anti-HMGA1, anti-Sp-1, or anti-AP-2α antibody. Primers for the human IR promoter sequence (ref. 24; forward, 5′-AACCACCTCGAGTCACCAAAA-3′ and reverse, 5′-AGAGGGAGGGAAAGCTTGCAG-3′) were used for PCR amplification of immunoprecipitated DNA (30 cycles), using PCR ready-to-go beads (Amersham Biosciences, Piscataway, NJ). PCR products were electrophoretically resolved on 1.5% agarose gel and visualized by ethidium bromide staining.

Plasmids and mutagenesis. Eukaryotic expression plasmids used in this study were as follows: pcDNA1-HMGA1 (12), pEVR2/Sp1 (12), pC-hu/AP-2α (a kind gift from Ko Fujimori), full-length and various truncated forms of AP-2α (25). Site-directed mutagenesis of DNA binding sites for HMGA1 and/or Sp1 in pCAT-C2 was done by the overlap extension method (26), using wild-type C2 sequence as initial template in PCRs and the following primers (Life Technologies): C2-HMGA1 forward, 5′-GCCCACTATGAACCCAATAGCAACCTGGTAGAGAAAGG-3′ and C2-HMGA1 reverse, 5′-CCTTTCTCTACCAGGTTGCTATTGGGTTCATAGTGGGC-3′; C2-Sp1 forward, 5′-CCCGGCACAGGGAGGCTTGGAGACGTGCGGGGCG-3′ and C2-Sp1 reverse, 5′-CGCCCCGCACGTCTCCAAGCCTCCCTGTGCCGGG-3′ (first round); C2-Sp1 forward, 5′-GACGTGCGGGGCTTGGCGGGACCGAGCTGCACCTCCCTCC-3′ and C2-Sp1 reverse, 5′-GGAGGGAGGTGCAGCTCGGTCCCGCCAAGCCCCGCACGTC-3′ (second round). Mutagenized bases are in lowercase letters. Mutations were confirmed by sequence analysis.

Transcription studies. Recombinant plasmid pCAT-C2 containing wild-type or mutant DNA binding sites for HMGA1 and/or Sp1 and effector vectors for AP-2, HMGA1, or Sp1 were transiently transfected into HepG2 or MCF-7 cells by the calcium phosphate precipitation method, and CAT activity was assayed 48 hours later as previously described (24). As an internal control of transfection efficiency, β-galactosidase activity was measured in addition to protein expression levels (11). For decoy experiments, MCF-7 cells were cotransfected with pCAT-C2 reporter vector plus 20 μg of double-stranded oligonucleotides containing wild-type or mutated AP-2 binding site sequences and chloramphenicol acetyltransferase (CAT) activity was measured as described above. Small interfering RNA (siRNA) targeted to HMGA1 (ACCACCACAACTCCAGGAA), Sp1 (AATGAGAACAGCAACAACTCC), and AP-2α (siGENOME SMART pool reagent), plus nonspecific siRNA controls with a similar GC content (NS-IX and NS-X) were obtained from Dharmacon (Lafayette, CO); 100-200 pmol siRNA duplex was transfected into cells at 50% to 60% confluency using LipofectAMINE 2000 reagent (Invitrogen, Carlsbad, CA), and cells were analyzed 48 to 96 hours later. HeLa nuclear extracts were employed in in vitro transcription studies using the HeLa cell extract in vitro transcription kit (Promega) plus the linearized pCAT-IR plasmid (24) as the DNA template. Hybridization of RNA transcripts with a ³²P-labeled CAT primer and reverse transcription were carried out as reported previously (27). For competition experiments, double-stranded oligonucleotides containing wild-type or mutated HMGA1 and/or Sp1 or AP-2 binding sites were added to untreated HeLa extracts, and the extracts were incubated for 15 minutes at room temperature before exposure to the IR-CAT template. (Note: Sequences for non-indicated oligonucleotide primers are available upon request.)

Expression of IR and AP-2 in breast cancer cells and tissues. Increased IR protein content in cultured human breast cancer cells and whole breast tumor tissues has been reported earlier (7, 28). In agreement with these previous observations, here, we show that IR mRNA levels were higher in certain human breast cancer cell lines (MCF-7, MDA-MB 231, and MDA-MB 468) compared with cells derived from normal human breast (HBL-100; Fig. 1). In addition, IR mRNA levels in breast cancer specimens from six unrelated patients were significantly higher than those observed in the corresponding surrounding normal tissues (Fig. 1). IR mRNA abundance in cultured breast cancer cells and tissues paralleled IR protein expression levels (Fig. 1). As measured by reverse transcription-PCR, the abundance of IR mRNA in breast cancer cell lines and tumoral breast tissues correlated closely with AP-2 mRNA. The increase in AP-2 mRNA levels in both cells and tissues paralleled the increase in AP-2 protein expression levels. These findings indicate that an up-regulation of AP-2 may occur in human breast cancer, and this overexpression correlates closely with IR mRNA and protein expression levels.

AP-2 interacts with HMGA1 in the absence of DNA. We previously showed that the architectural transcription factor HMGA1 plays significant molecular roles in the context of the human IR gene, whose transcriptional activation is dependent on the formation of a multiprotein-DNA complex, in which protein-protein and DNA-protein interactions are needed for transcriptional synergism between the various transcription factors in vivo (11–13). As stated above, there are no AP-2 DNA-binding sites on the promoter region of the IR gene. Therefore, in an attempt to identify the biochemical mechanisms underlying the stimulatory role of AP-2 in IR gene expression, we did experiments designed to investigate whether HMGA1 was necessary for recruitment and binding of AP-2 to this nucleoprotein complex controlling IR gene transcription. Direct physical association between HMGA1 and multiple transcription factors has been reported (29, 30). To analyze the ability of HMGA1 and AP-2 to interact with each other in vitro, in the absence of DNA, we first did a GST pull-down assay, in which in vitro translated ³⁵S-labeled AP-2 was analyzed for its ability to be specifically retained by a GST-HMGI affinity resin. As shown in Fig. 2A, AP-2 was retained by GST-HMGI but not GST alone, suggesting that AP-2 interacts physically with HMGA1 in vitro. A bona fide interaction between HMGI and AP-2 was observed in the presence of high concentrations of ethidium bromide (not shown), which has been shown to disrupt DNA-dependent protein-protein contact (31). Interaction between HMGA1 and AP-2 was further investigated in coimmunoprecipitation studies with cell nuclear extract and an antibody against HMGA1 immobilized on protein A beads. As shown in Fig. 2B, immunoprecipitation of HMGA1 from HeLa nuclear extracts followed by Western blot analysis for AP-2 revealed a major specific band, which migrated in a position corresponding to the size of AP-2. When the same transfer was reprobed with the anti-HMGA1 antibody, a unique specific band, which migrated in a position corresponding to the size of HMGA1, was detected. Taken together, these data unequivocally indicate that HMGA1 and AP-2 physically interact either in vitro or in vivo in the context of the intact cell and suggest that physical and functional cooperation between HMGA1 and AP-2 on the IR promoter might occur through direct contact as well.

To elucidate which regions of HMGA1 and AP-2 were required for physical interactions, we first carried out GST pull-down assays using GST-linked deletion mutant versions of HMGI. As shown in Fig. 2C, HMGI mutant 1-74, which lacks the COOH-terminal tail plus the third basic repeat, displayed an evident binding activity for AP-2 similar to that of the wild-type (1-107) protein. No binding activity was detected with mutants 1-54 and 65-107 containing the first and third basic repeats, respectively. Conversely, binding activity was detected within the HMGI mutant 54-74, containing the second basic repeat, whereas HMGI mutants lacking any basic repeat (clones 31-54 and 65-74) were unable to bind AP-2. Taken together, these data indicate that, in the context of HMGA1, the minimal region required for specific interaction with AP-2 corresponds to the middle basic repeat flanked by the spacer region between the middle and the third basic repeats (amino acids 54-74), revealing, in this respect, a behavior similar to that shown by the interactions of HMGA1 with nuclear factor-κB (27), Sp1, and C/EBPb (12). As a second step to characterize the AP-2 protein domains involved in HMGA1 binding, similar experiments were done with GST-HMGI wild-type and in vitro synthesized ³⁵S-labeled AP-2 mutants. As shown in Fig. 2C, full-length AP-2 and its mutant ΔN165 were both specifically retained by the agarose beads coupled with GST-HMGI but not by GST alone. In addition, AP-2 mutants ΔN227 and ΔC390 were not bound by GST-HMGI. These data show that binding of AP-2 to HMGI does not need the NH₂-terminal region upstream amino acid 165. Instead, the simultaneous presence of both regions 165-227 and the COOH-terminal domain downstream amino acid 390 of AP-2 are required for AP-2-HMGA1 interaction. Thus, these findings indicate that AP-2 interacts with HMGA1 through few multiple functional domains that can function as potent activators of transcription in the presence of HMGA1.

AP-2 interacts with Sp1 in the absence of DNA. A functional interplay among AP-2 and Sp1 family factors has been shown in various pathways during activation of gene transcription in eukaryotes (32, 33). On the other hand, functional cooperation between Sp1 and HMGA1 for the transcriptional activation of the IR promoter has been reported previously by our group (12). To gain support for the assumption that AP-2 and Sp1 interact directly at the IR promoter, we first tried to determine whether the two factors bind to each other in solution. To this end, a pull-down assay using GST-fused Sp1 and in vitro synthesized ³⁵S-labeled AP-2 was done. As shown in Fig. 3A, AP-2 was efficiently retained by wild-type GST-Sp1 but not GST alone. To verify that the interaction observed in vitro between AP-2 and Sp1 resembles the interaction in vivo, we did coimmunoprecipitation studies using nuclear extracts from HepG2 cells cotransfected with expression vectors for AP-2 and Sp1, in the presence of an antibody against Sp1 immobilized on protein A beads. As shown in Fig. 3B, immunoprecipitation of Sp1 from nuclear extracts of transfected HepG2 cells followed by Western blot analysis for AP-2 revealed a unique specific band, which migrated in a position corresponding to the size of AP-2. When the same transfer was reprobed with an anti-Sp1 antibody, a major band that migrated in a position corresponding to the size of Sp1 was observed. Thus, these findings indicate that these two factors physically interact in vivo. Further evidence of the interaction between the transcription factors investigated in this study was obtained by performing coimmunoprecipitation experiments with nuclear extracts from transfected HepG2 cells and the antibody against HMGA1 immobilized on protein A beads. As expected, immunoprecipitation of HMGA1 followed by immunoblot analyses with specific antibodies allowed the identification of three proteins, including, in addition to HMGA1, Sp1 and AP-2 (Fig. 4). Therefore, we conclude that a physical complex consisting of HMGA1, AP-2, and Sp1 exists in vivo in the context of the intact cell.

AP-2 interacts with HMGA1 and Sp1 in the presence of DNA sequences for the IR promoter. Physical interaction among AP-2, HMGA1, and Sp1 has been characterized further by gel shift assay, using the previously described radiolabeled fragment C2 of the human IR promoter gene (24). Whereas binding sites for Sp1, flanked by AT-rich HMGA1 binding sites, have been identified within this DNA region, no putative binding sites for AP-2 have been detected within this element of the IR promoter. As shown in Fig. 5A, the binding of either pure recombinant HMGA1 (11) or pure Sp1 (a kind gift from Robert Tjian) to this probe produced a protein-DNA complex that was recognized and supershifted to a slower-migrating form by the anti-HMGA1 or anti-Sp1 antibody, respectively. Nuclear extracts from HepG2 cells (which do not express AP-2) transfected with AP-2 expression plasmid were used as a source of AP-2 nuclear protein. The incubation of nuclear extracts from transfected HepG2 cells with C2 probe yielded a higher protein-DNA complex that was in part recognized and supershifted by the anti-AP-2 antibody. The two additional complexes, other than that including AP-2, in HepG2 nuclear extracts, included Sp1 and HMGA1. The existence of AP-2 in a cocomplex with HMGA1 and Sp1 was confirmed in competition studies using nuclear extracts from HeLa cells, a cell line naturally expressing either HMGA1, Sp1, or AP-2 nuclear protein. In HeLa nuclear extracts, the formation of a large complex involving the simultaneous presence of all three proteins on the same probe C2 was prevented by using relatively small amounts of nuclear extracts in the presence of increasing concentrations of competitor DNA for AP-2 (Fig. 5A). Thus, these results show that AP-2 did not bind C2 directly but rather forms a multiprotein-DNA complex with Sp1 and HMGA1 independently of AP-2 binding to DNA. These data were substantiated by chromatin immunoprecipitation assay, showing that HMGA1, Sp1, and AP-2 indeed bind to the endogenous IR locus in vivo in AP-2α-overexpressing HepG2 cells (Fig. 5B).

Induction of the IR promoter by AP-2α requires HMGA1 and Sp1. In the light of the above results indicating that AP-2 physically interacts with HMGA1 and Sp1, it was important to see whether these interactions had a functional role in the transcriptional activity of the IR gene promoter. The effect of AP-2α on IR transcription was first evaluated by transiently transfecting the AP-2-null cell line HepG2 (naturally expressing both HMGA1 and Sp1) with the C2 IR promoter element (pCAT-C2) in the presence of an expression vector encoding the wild-type AP-2α. As shown in Fig. 6A, overexpression of AP-2 in HepG2 cells induced a concentration-dependent stimulation of IR gene transcription. Because AP-2 did not bind the IR promoter element directly, this result supports a model in which AP-2 activates IR gene transcription independently of AP-2 binding to DNA. To examine the possibility that transactivation of the IR promoter by AP-2 could involve HMGA1 and Sp1, we generated CAT constructs with the C2 regulatory element mutated at either or both of the HMGA1 and Sp1 binding sites. Mutation of the HMGA1 or Sp1 binding site in the C2 promoter element (pCAT-C2/HMGA1m or pCAT-C2/Sp1m, respectively) reduced CAT activity in AP-2-transfected HepG2 cells to 15% to 20% of that with the native (pCAT-C2) promoter. Mutation of both binding sites together (pCAT-C2/HMGA1m/Sp1m) almost completely abolished CAT activity. Likewise, IR-CAT activity was impaired in HepG2 cells transfected with AP-2α mutant effector plasimds (AP-2α ΔC390, AP-2α ΔN165, and AP-2α ΔN227), providing further evidence of the existence of important functional interactions among HMGA1, Sp1, and AP-2 in the context of the IR gene (Fig. 6A). Consistent with these observations, endogenous IR protein and mRNA (data not shown) levels were very low in AP-2α-overexpressing HepG2 cells subjected to siRNA-mediated knockdown of HMGA1 and/or Sp1 (Fig. 6B).

Finally, we showed that the interplay among AP-2, HMGA1, and Sp1 had functional sequelae also in MCF-7 cells, a cell line ideally suited for studying the effects of these proteins on transcription because it does not express appreciable levels of either HMGA1 or Sp1 (34, 35), whereas AP-2 is constitutively expressed (21). As shown in Fig. 6C, simultaneous overexpression of HMGA1 and Sp1 in these cells led to a significant increment in CAT activity that exceeded that seen with either factor alone and required intact binding sites for HMGA1 and Sp1, as shown by diminished activity of constructs mutated at either or both of these sites. Induction of CAT activity was at least in part dependent on endogenous AP-2, as revealed by AP-2 decoy (36) and siRNA strategies. Cotransfection of a cis element decoy against the AP-2 binding site resulted in a 50% reduction in CAT activity, whereas transfection of the mutated AP-2 decoy element, which fails to bind the AP-2 protein in vitro, was ineffective. siRNA-mediated knockdown of endogenous AP-2α in MCF-7 cells showed similar results (Fig. 6C). In concert with these findings, endogenous IR expression was reduced in MCF-7 cells subjected to siRNA-mediated knockdown of AP-2α (Fig. 6D). These results were supported further by in vitro transcription studies using HeLa nuclear extracts in the presence of synthetic oligonucleotides with either the HMGA1, Sp1, or AP-2 binding sequence (Fig. 7). Together, these data indicate that AP-2α plays a positive role in IR gene expression and suggest that AP-2α may regulate IR promoter activity through direct association with HMGA1 and Sp1 rather than binding to DNA.

Breast cancer is an important public health problem because it affects about one in 10 women at some time during their lives and represents a leading cause of cancer deaths among women. A wide variety of studies have shown that the IR is frequently overexpressed by most human breast cancers, suggesting a role for this tyrosine kinase receptor protein in breast cancer cell biology. Furthermore, ligand-dependent malignant transformation and increased cell growth occur in cultured breast cells overexpressing IRs. The IR can exert its oncogenic potential in malignant breast cells via abnormal stimulation of multiple cellular signaling cascades, enhancing growth factor–dependent proliferation and/or by directly affecting cell metabolism. However, little is known about the molecular mechanisms that account for the increased IR expression in human breast cancer.

In target cells, the IR has been shown to be under the regulation of hormones, metabolites, and differentiation (24). The transcription of the IR gene is also dependent upon transcription factor Sp1 (12). Our data implicate AP-2α in the up-regulation of IR expression in breast cancer, and this up-regulation may be important for the development of malignant transformation. As shown by immunoblot and gene expression analyses, AP-2α was significantly increased in nuclear extracts from cultured breast cancer cells and human breast cancer specimens, and this increase correlated closely with IR overexpression in both breast cancer cell lines and breast cancer tissues. Activation of gene expression by AP-2 has been reported for many different genes, in which AP-2 DNA binding sites have been identified in the promoter region of the gene (37). In our study, we provide compelling evidence that increased IR expression in breast cancer is dependent on the assembly of a transcriptionally active multiprotein-DNA complex, which includes, in addition to AP-2, the architectural factor HMGA1 and the ubiquitously expressed transcription factor Sp1. The observation that AP-2 forms a protein-protein complex with HMGA1 and Sp1 in the activation of IR promoter, in the absence of AP-2 DNA-binding sites, is consistent with previous studies showing that AP-2 is capable of regulating gene transcription indirectly, through the formation of heteromeric protein-protein interactions between AP-2 and c-Myc (38) and AP-2 and E1A (39), in the absence of AP-2 protein-DNA interaction. In addition, our observation agrees closely with the results reported in previous works, indicating that transcriptional activation by AP-2 may require combinatorial mechanisms, as the DNA sequences implicated in transactivation by AP-2 do not always resemble an AP-2 binding site, and the proline-rich activation domain of AP-2 is a poor transactivator when positioned in a distal position (33, 38, 40).

High levels of AP-2 protein expression in human breast cancer have been reported in previous reports, in which expression of AP-2 correlated with the regulation of multiple growth factor signaling pathways (22, 41). Herein, we show for the first time that AP-2α induces transcriptional activation of the IR gene, and this activation occur independently of AP-2 binding to the IR promoter. As shown in transient transfection studies using the AP-2-null cell line HepG2, transactivation of the IR promoter by AP-2 required intact DNA binding sites for HMGA1 and Sp1 and needed direct physical association of AP-2 with HMGA1 and Sp1, which represents a fundamental prerequisite for AP-2α to properly stimulate IR gene transcription in the absence of specific AP-2 consensus binding sites. This result closely resembles a previous one for transcriptional activation of the CYP11A1 gene, in which a functional interplay between AP-2 and Sp1 has been shown in the context of the CYP11A1 gene promoter, in the absence of an AP-2 binding site (33).

Stabilization of transcription factor-DNA interactions by HMGA1 plays a critical role in gene regulation. In particular gene contexts, HMGA1 acts to stabilize transcription factor-DNA and protein-protein interactions through a combination of its own binding to a proximal minor groove AT-rich sequence and through direct interaction with an adjacent transcription factor through one of its additional AT-hook motifs or through direct effects on the conformation of multisubunit proteins (14–17, 42). Our data suggest that induction of IR gene transcription by AP-2α in breast cancer can occur independently of AP-2 binding to DNA and support a model in which HMGA1 may facilitate and stabilize interactions between AP-2 and Sp1 in the preinitiation complex at the IR promoter. To our knowledge, this is the first report describing a quantitative abnormality in a trans-acting factor, which affects the levels of expression of the IR gene in human breast cancer. We believe that AP-2α overexpression leading to IR overexpression might be a critical event in breast carcinogenesis, and that AP-2α is a potential target for therapeutic intervention in human breast cancer.

Grant support: Ministero dell’Università e della Ricerca, Italy protocol 2004062059-002 and Telethon, Italy grant GGP04245 (A. Brunetti).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Ko Fujimori (Osaka Bioscience Institute), Tom Maniatis (Harvard University, Cambridge, MA), Hans Rotheneder (University of Vienna), Guntram Suske (Philipps University Marburg, Marburg, Germany), Dimitris Thanos (Columbia University), Robert Tjian (University of California, Berkeley), and Erhard Wintersberger (University of Vienna) for providing reagents and Anna Malta for secretarial help.