PEA3 transactivates the Muc4/sialomucin complex promoter in mammary epithelial and tumor cells.

Sialomucin complex (SMC, rat Muc4) is a heterodimeric glycoprotein composed of two subunits, the mucin component ascites sialoglycoprotein ASGP-1 and the transmembrane subunit ASGP-2, which is aberrantly expressed on the surfaces of a variety of tumor cells. Up-regulation of the Muc4/SMC gene in the 13762 sublines of the rat mammary adenocarcinoma correlates with the overexpression of transcription factor PEA3 and the receptor tyrosine kinase ErbB2. Here we report that PEA3 is capable of transactivating the Muc4/SMC promoter in a dose-dependent manner via direct attachment to a PEA3 binding site. ERM and ER81, the other two members of the PEA3 subfamily of transcription factors, could not transactivate the Muc4/SMC promoter. Transcriptional activation of Muc4/SMC by PEA3 is potentiated by Ras and MEKK1 kinases. These data suggest that expression of PEA3 in mammary tumors leads to up-regulation of Muc4/SMC transcription, the gene product of which may contribute to the metastatic potential of mammary tumors.

Muc4/SMC 1 is a membrane mucin isolated from the highly metastatic 13762 ascites sublines of a rat mammary adenocarcinoma (1). Muc4/SMC consists of a peripheral, O-glycosylated glycoprotein subunit (2), ascites sialoglycoprotein-1 (ASGP-1), which is tightly, but non-covalently bound to an N-glycosylated integral membrane glycoprotein ASGP-2 (3). The latter has two epidermal growth factor-like domains and can act as an intramembrane ligand and modulator of the receptor tyrosine kinase ErbB2 (4). Muc4/SMC is transcribed from a single gene as a 9-kilobase transcript (5) and translated into a single Ϸ300-kDa polypeptide, which is proteolytically cleaved into the two subunits early in its transit to the cell surface (6). Recently, it was demonstrated that the N-terminal amino acid sequence of ASGP-1 and the complete sequence of ASGP-2 from rat SMC are 60 -70% identical to the sequence of human MUC4 (7). Rat SMC differs from human MUC4 by the absence of a 16-amino acid tandem repeat, which was the original identifying characteristic of MUC4 (8). A number of functions have been attributed to Muc4/SMC. In tumor cells ASGP-1 confers anti-recognition and anti-adhesive properties (9), which were demonstrated by the transfection of Muc4/SMC cDNA into A375 melanoma cells. Expression of Muc4/SMC in these cells was found to abolish cell-matrix adhesion and cell-cell interactions (10). In addition, the expression of Muc4/SMC in A375 cells reduces their killing by natural killer cells (11), which may be important in tumor progression of neoplastic mammary cells. Further, the presence of Muc4/SMC on exposed epithelia such as cornea, vagina, and cervix suggests a protective role for Muc4/SMC (12,13). Muc4/SMC is also proposed to modulate cellular signaling through the EGF family of receptors via its interaction with ErbB2 (4), the critical receptor for formation of active heterodimers of the class I (ErbB) receptor tyrosine kinases. This interaction may play a role in the constitutive phosphorylation of ErbB2 in the 13762 ascites cells (14), and the rapid growth of these cells in vivo.
The expression patterns of both human MUC4 and rat Muc4/ SMC have been studied in detail. MUC4 is expressed in a broad range of secretory epithelial cells. By Northern blotting and/or in situ hybridization, human MUC4 mRNA has been detected in numerous normal tissues such as trachea, colon, stomach, cervix, uterus, prostate, and lung (15,16), but not in normal pancreas, gall bladder, liver or biliary epithelial cells, and breast tissues (17,18). Western blot and immunohistochemistry reveal that rat Muc4/SMC protein is expressed in a number of normal rat tissues including colon, small intestine, trachea, uterus, cornea, and mammary gland (19). Muc4/SMC is overexpressed on the surface of the 13762 rat mammary adenocarcinoma (Ͼ10 6 copies/cell) at a level approximately 100-fold higher than in lactating mammary gland or approximately 10,000-fold higher than in normal virgin gland (1,2,20,21). Interestingly, human MUC4 has also been shown to be aberrantly expressed in several epithelial cancers such as lung, pancreas, and gall bladder carcinoma (17,18,22,23), as well as in various cancer cell lines (24 -26). Thus, Muc4/SMC must be tightly regulated, because aberrant expression of this protein could have deleterious consequences.
We have cloned and characterized the 5Ј-flanking region of the rat Muc4/SMC gene (27). When compared with the recently published promoter sequence of the human MUC4 (28), the rat sequence shows ϳ70% homology over the first 464 nucleotides upstream of the first ATG, and then both sequences diverge. Functional studies of the 5Ј-flanking region of MUC4 in pancreatic cells have shown that two regulatory regions control MUC4 transcription: Ϫ219/Ϫ145 and Ϫ2781/Ϫ2572 (28). Analysis of the rat and human MUC4 promoter sequences showed the presence of several Ets regulatory elements, which may bind PEA3 transcription factor (27,28). In addition, analysis of the 5Ј-flanking regulatory of another two mucins MUC2 and MUC5AC shows the presence of several putative binding sites for the transcription factor PEA3 (29,30). The ets genes, which currently comprise nearly 30 paralogs in mammals, encode transcription factors bearing conserved sequence-related DNA binding domains (the ETS domain) (31). Ets proteins are capable of regulating transcription by binding to sites in the promoters of their cognate target genes. DNA binding is achieved by interaction between the ETS domain and a ϳ10-bp sequence element termed the Ets binding site, comprising a highly conserved core sequence, 5Ј-GGA(A/T)-3Ј. PEA3 (32) is the founding member of a subfamily of Ets proteins, which also include ER81 (33) and ERM (31,34). Analyses of the transcriptional properties of individual PEA3 subfamily members reveal that they commonly activate transcription (32)(33)(34)(35). A few PEA3 target genes have been identified; transient transfection studies suggest that PEA3 is capable of regulating transcription of genes whose products facilitate cell motility and invasion (reviewed in Ref. 36).
PEA3 is overexpressed in the vast majority of human breast tumors and in nearly all of those of the ErbB2-positive subclass (37,38). PEA3 is similarly overexpressed in transgenic mouse models of this malignancy (38,39). Expression of dominantnegative PEA3 in the mouse mammary gland of MMTV-ErbB2 transgenic mice dramatically reduces the incidence of mammary tumors (39). These findings suggest the possibility that PEA3 plays a role in mammary oncogenesis or tumor progression. To assess the role of the PEA3 gene in Muc4/SMC expression, we measured the ability of PEA3 to affect the transcriptional activity of Muc4/SMC. The transactivation capacity of PEA3 subfamily ets genes has been shown to be dramatically increased by Ras, Raf-1, MEK, and MAPKs ERK-1 and ERK-2 (40 -42). The JNK/SAPK pathway is also involved in PEA3 group member activation (42). To learn specifically how the Muc4/SMC activity is regulated by PEA3, we inquired whether constitutively activated Ha-Ras (Ha-RasV12) and Ras downstream effectors are capable of enhancing the transcriptional activity of Muc4/SMC by PEA3 in HC11 mouse mammary epithelial cells. The mouse mammary epithelial cell line HC11 has been shown to be a good model for studying mammary cell differentiation and gene regulatory mechanisms taking place in mammary cells (43). Our findings suggest that ERK and JNK/SAPK pathways independently stimulated PEA3-dependent Muc4/SMC gene expression in HC11 cells. Transcriptionally activated PEA3 stimulates Muc4/SMC expression, thus accounting in part for the increased Muc4/SMC mRNA and protein levels found in 13762 ascites cells compared with normal mammary epithelial cells.
Cells-MAT-B1 and MAT-C1 cells are ascites sublines of the 13762 rat mammary tumor. Both sublines were passaged by weekly transfers into 2-3-month-old female Fischer rats and also maintained as frozen stocks (51). The sublines differ in morphology and in the O-glycosylation of ASGP-1. Cells used in these studies were washed three times with cold phosphate-buffered saline after recovery from the peritoneal cavity. The MAT BIII cultured subline of the 13762 rat mammary adenocarcinoma was obtained from the ATCC (Manassas, VA). The normal mouse mammary epithelial cell line HC11 was a gift from Dr. Jeffrey Rosen (Baylor College of Medicine, Houston, TX). Cell Culture-Primary mammary epithelial cell cultures were established using previously described protocols (52,53). Briefly, mammary glands excised from virgin, pregnant, or lactating female Fischer 344 rats were minced and resuspended in digestion media comprising 1 mg/ml collagenase type II (Worthington Biochemical Corp., Freehold, NJ), and 100 units/ml penicillin, 100 g/ml streptomycin in Ham's F-12 medium and incubated at 37°C with shaking for 45 min. Fully and partially digested epithelial cell clusters were pelleted and incubated a second time in digestion buffer at 37°C with shaking for 45 min. Digested epithelial cell clusters were pelleted, resuspended in PBS, and passed through a 520-m cell sieve to remove undigested material. Mammary epithelial cell clusters in the resulting filtrate were captured on a 70-m nylon membrane. Cell clusters were collected by rinsing the membrane with PBS and were subsequently washed three times in PBS prior to plating. Incubating freshly isolated cells on a plastic tissue culture plate for 1 h permitted attachment and removal of fibroblasts. The isolated cell preparations contained Ͼ85% epithelial cells (54), as judged by staining for cytokeratins (55) and Muc4/SMC (21). The HC11 mouse mammary cell line was cultured in RPMI medium supplemented with 10% FBS, 5 g/ml insulin, and 10 ng/ml EGF. The MAT-B1 cells were cultured in Dulbecco's modified Eagle's medium (high glucose) supplemented with 10% FBS. The 13762 MAT BIII cell line was cultured in McCoy's 5A medium supplemented with 10% FBS.
Whole Cell Extracts-Mammary tissue from virgin, pregnant, and lactating rats as well as from ascites and cultured cells, were lysed with PBS plus 0.5% SDS containing protease inhibitors and sonicated. The lysates were centrifuged for 20 min at 12,000 ϫ g and pellet discarded. Lysates were stored at Ϫ80°C until used for immunoblot analyses.
Northern Blots-Total RNA was isolated using TRI Reagent TM (Molecular Research Center, Inc., Cincinnati, OH), and 25 g were electrophoresed on 1% formaldehyde-agarose gels. Resolved RNAs were transferred to ZetaProbe positively charged nylon membranes (Bio-Rad) and cross-linked with a Stratalinker (Stratagene, La Jolla, CA). The membranes were prehybridized for at least 2 h at 42°C in prehybridization solution (50% formamide, 5ϫ SSC, 5ϫ Denhardt's reagent, 0.1% SDS, and 0.5 mg/ml salmon sperm DNA). The probe, A2G2-#9 (56), a 1.7-kb DNA fragment that spans the 5Ј unique region and four tandem repeats of Muc4/SMC cDNA, was random primer-labeled with [ 32 P]dCTP using a random primer labeling kit (Roche Molecular Biochemicals). The membranes were hybridized overnight at 42°C in prehybridization solution containing 0.1 g/ml dextran sulfate and the labeled probe. Following hybridization, membranes were washed once at room temperature in 2ϫ SSC with 0.1% SDS for 15 min, twice at 50°C in 2ϫ SSC with 0.1% SDS for 20 min each, and once at 50°C in 0.1ϫ SSC with 0.1% SDS for 15 min. Signals were detected by exposure with Kodak XAR-5 x-ray film.
Immunoblot Analyses-Whole cell and nuclear extracts (1-15 g of protein) were analyzed in 7% polyacrylamide gels with the Mini-PRO-TEAN II system (Bio-Rad). Protein concentration was determined using BCA protein assay reagent and bovine serum albumin as the standard. Resolved proteins were transferred to nitrocellulose membranes and blocked with 5% nonfat milk (Bio-Rad) in Tris-buffered saline with 0.1% Tween 20. The membranes were then probed with primary and secondary antibodies. Proteins were detected by the Renaissance TM enhanced chemiluminescence kit (PerkinElmer Life Sciences).
Luciferase Reporter Constructs-For PEA3 transactivation studies, a fragment containing 335 bases of the Muc4/SMC promoter region was generated by PCR of rat uterine genomic DNA. To 100 ng of genomic DNA, 10 pmol of oligonucleotide primers, Taq polymerase, 1 l of 10 mM dNTPs, and 3 l of 25 mM MgCl 2 were added. The primers used were: Left (2003-2024), 5Ј-GCTCTAGACCTGGCTGCCTTCTACAGTA-3Ј; and Right (2394 -2413), 5Ј-CGGGATTCCTCAGGAATCTCCACAGCA-3Ј. The nucleotide positions are given according to the Muc4 promoter sequence (GenBank TM accession no. AF240632). Reactions were temperature-cycled in a Gene Cycler (Bio-Rad) under the following conditions: 94°C for 2 min (1 cycle); 94°C for 45 s, 55°C for 30 s, and 72°C for 1 min (35 cycles); and 72°C for 10 min (1 cycle). Oligonucleotide primers were designed to introduce an XbaI site at the 5Ј end of the 335-bp fragment and a BamHI at the 3Ј end. This allowed the successful ligation of the fragment into the pGL3-Basic Vector (Promega). After gel purification, the fragment was digested with XbaI and BamHI. This fragment was then ligated to the pGL3-Basic vector digested with NheI and BglII. The ligation mixture was transformed into XL-1 Blue competent cells (Stratagene), plated, propagated in LB broth supplemented with ampicillin (100 g/ml), and purified using a Maxiprep Kit (Qiagen). Accurate cloning was confirmed by DNA sequencing at the DNA Core Laboratory (University of Miami, Miami, FL). The construct was designated WT-PEA3.
Site-directed Mutagenesis-Mutation of the PEA3 binding site was performed on WT-PEA3 using the QuikChange site-directed mutagenesis kit (Stratagene). 10 ng of WT-PEA3 was combined with complementary mutant oligonucleotide primers (125 ng), 1 l of 10 mM dNTPs, and 2.5 units of PfuTurbo DNA polymerase. The oligonucleotide primers used were as follows: sense (2167-2102; the nucleotide position is defined according to the sequence of the Muc4 promoter (GenBank TM accession no. AF240632), 5Ј-GGGGCCGTAGGCAGGTTGGTGCCCTC-TCTCACTTCC-3Ј; and antisense, 5Ј-GGAAGTGAGAGAGGGCACCAA-CCTGCCTACGGCCCC-3Ј (the underlined letters represent the bases along the PEA3 binding sequence that were mutated). The reaction mixture was temperature-cycled in a Gene Cycler (Bio-Rad). Cycling conditions were 95°C for 30 s (1 cycle), then 95°C for 30 s, 55°C for 1 min, and 68°C for 10 min (12 cycles). This series of cycles serves to amplify the MT-PEA3 construct.
Cell Transfections and Luciferase Assays-Transfections were performed in triplicate using LipofectAMINE (Invitrogen), according to the instructions from the manufacturer. Briefly, cells were seeded in 6-well plates at 5 ϫ 10 5 cells/well 24 h before transfection. Each well was transfected for 3 h using a mixture of 1 ml of serum-free Opti-MEM medium, 1-2 g of plasmid DNA (including 10 ng of pRL-CMV), and 15 l of LipofectAMINE. Where necessary "empty" vectors were included to maintain a constant amount of DNA. The media were then changed to the appropriate medium optimum for growth of each cell type. Lysates were prepared 24 -48 h after transfection, and firefly and Renilla luciferase activities were measured using a Dual Luciferase Reagent kit (Promega) and a LB 9507 luminometer (Berthold GmbH). Firefly activity was normalized to Renilla activity, and results were expressed as relative light units. For experiments with kinase inhibitor U0126, the HC11 cells were cultured in serum-free RPMI medium for 24 h before transfection followed by 30-min treatment of U0126 at 80 M and then the transfection was done as described above.
Plasmid Constructs-The following plasmids were used for expressing PEA3 derivatives as N-terminal 6-histidine-tagged and C-terminal Myc/6-histidine-tagged proteins in mammalian cell transfections. The pHis-N152PEA3 (encoding mPEA3 amino acids 152-480) was con-structed by inserting the EcoRI/XbaI-digested PCR-derived fragment into the same sites in pcDNA3.1/His A (Invitrogen). The primers used for the PCR cloning were: Left (801-818), 5Ј-GGAATTCCAGAACAG-CAGCAGAGCCTCC; and Right (1762-1781), 5Ј-GCTCTAGAATCCAC-CTCTGTGGCCGAAGG. The pmyc-His C440 PEA3 (encoding mPEA3 amino acids 1-440) was constructed by inserting the NheI-EcoRI-digested PCR-derived fragment into the same sites in pcDNA3.1/myc-His (Ϫ) A (Invitrogen). The primers used for this PCR reaction were: Left (346 -367), 5Ј-GCGGCTAGCCGGATGGAGCGGAGGATGAA; and Right (1645-1665), 5Ј-GGAATTCCAGCCTTCAGAGCTGGACGTT. Nucleotide position is defined according to the sequence of mPEA3 (Gen-Bank TM accession no. X63190). The PCR reactions were conducted under the following conditions: initial annealing at 98°C for 2 min, followed by 25 cycles of denaturation at 98°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 1.5 min. The sequences of all plasmids constructed from PCR-derived products were verified by automated dideoxy sequencing at the DNA Core Laboratory (University of Miami, FL).
Expression of GST-PEA3 Fusion Protein in Escherichia coli-GST fusion protein encoding a fragment of the mouse PEA3 protein was constructed in the vector, pGEX-4T2 (Amersham Biosciences), using PCR and/or restriction endonuclease digestion. The following oligonucleotides were used to amplify the ETS domain (amino acids 334 -417, 255 bp) of the mouse PEA3 cDNA: Left primer (1350 -1369), 5Ј-CGG-GATCCCGGGGTGCCTTACAACTGTG; and Right primer (1591-1609), 5Ј-CGGAATTCCGGCTCGCACACAAACTTGT. Nucleotide position is defined according to the sequence of mPEA3 (GenBank accession no. X63190). PCR was conducted under the following conditions: initial annealing at 98°C for 2 min, followed by 25 cycles of denaturation at 98°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 30 s. The PCR product was digested with BamHI and EcoRI (sites incorporated into 5Ј end and 3Ј end primers, respectively) ligated between the same sites of pGEX-4T2 vector and transformed into the E. coli strain BL-21 (Stratagene). Protein induction was carried out as described in the GST handbook from Amersham Biosciences. Briefly, E. coli were grown overnight in 2ϫ YT medium containing 100 g/ml ampicillin in a 37°C shaker. Cultures were diluted 1:10 into 2ϫ YT, and protein induction was carried out by the addition of 1 mM isopropyl-1-thio-␤-D-galactopyranoside (Sigma) when the cells reached an absorbance of 0.6 -0.8 measured at 600 nm. Two hours following the induction, whole cell lysates were prepared by sonication in a phosphate-buffered saline. These lysates were subjected to protein purification using the affinity matrix Glutathione-Sepharose TM 4B (Amersham Biosciences). Eluted fractions were stored at Ϫ80°C until use.
Preparation of Nuclear Extracts and Electrophoretic Mobility Shift Assays-Transfections were carried out as described above, except that the HC11 cells were initially plated in 60-mm (diameter) dishes at a density of ϳ3 ϫ 10 5 cells/dish. The cells were transfected with a total of 6 g of DNA comprising 3 g of expression plasmid DNA and 3 g of pBluescript II SK (Stratagene) as a carrier. Nuclear extracts were prepared 48 h after transfection using NE-PER ® nuclear and cytoplasmic extraction reagents (Pierce). Binding reactions were performed in a final volume of 20 l containing 0.2 ng of radiolabeled DNA, 1 g of poly(dI-dC) (Amersham Biosciences), 1-10 g of protein (using either nuclear cell extracts from transfected HC11 cells or purified GST-fusion proteins) in 50 mM Tris-HCl (pH 7.5), 50 mM NaCl, 2 mM dithiothreitol, 2 mM EDTA, and 2 mM spermidine. The reactions were carried out for 20 min at room temperature. To stop the reaction, 5 l of 10% glycerol ϩ 0.01% bromphenol blue was added, and then samples were run on 5% polyacrylamide gels (40:1 ratio acrylamide/bisacrylamide) in 0.5ϫ TBE (1ϫ TBE ϭ 89 mM Tris borate (pH 8.0) and 2 mM EDTA). When employed, competitor DNAs were added at 100 M excesses before the addition of the radiolabeled DNA. A radiolabeled double-stranded oligonucleotide, corresponding to positions 2073-2092 within the Muc4 promoter (GenBank TM accession no. AF240632) (5Ј-GTAGGCAGGAAG-GTGCCCTC-3Ј and the complementary sequence 5ЈGAGGGCACCTTC-CTGCCTAC-3Ј) was used as a probe in these assays. Prior to their use, the DNA strands were annealed and labeled by T4 polynucleotide kinase using [␥-32 P]ATP. The double-stranded oligonucleotide was purified by G-25 spun columns.
Figure Generation-Original autoradiographs were scanned in HP PrecisionScan Pro, and the final figures were produced using Adobe Photoshop 6.0. In several of the figures, lanes on the same autoradiogram were juxtaposed to facilitate representation of data. When lanes from the same autoradiogram were spaced together, this was achieved in such a way as to reveal a thin line between adjacent lanes. Otherwise, the final figures are faithful representations of the original autoradiograms.

MUC4/SMC, ErbB2, and PEA3 Are Up-regulated in 13762
Tumor Cells Compared with Normal Mammary Epithelial Cells-Muc4/SMC was originally isolated from highly proliferative and metastatic ascites sublines of the 13762 rat mammary adenocarcinoma (12). In addition, as a ligand for ErbB2, Muc4/SMC can potentiate phosphorylation of ErbB2 (4). Because PEA3 is a key regulator of ErbB2 overexpression in human and mouse breast tumors (37,38), we decided to explore the link among Muc4/SMC, ErbB2, and PEA3 in 13762 rat mammary adenocarcinoma sublines and normal mammary epithelial cells. PEA3 is a transcription factor known to enhance transcription of various genes, including metalloproteinases (57), vimentin (58), and urokinase-type plasminogen activator (59). To compare the level of Muc4/SMC, ErbB2, and PEA3 expression in the rat tumor cells to that in normal mammary epithelial cells, cells from three different sublines of the 13762 rat mammary adenocarcinoma and normal mammary tissue were solubilized in 1% SDS, and total protein was quantified by BCA assay. MAT-B1 and MAT-C1 cells are ascites sublines of the 13762 rat mammary tumor. The MAT-C1 subline exhibits highly branched microvilli and restricted concanavalin A receptor mobility and is xenotransplantable. The MAT-B1 cells have a more normal surface architecture, with unbranched, curved microvilli, ruffles, and ridges. Their concanavalin A receptors are highly mobile, and the cells are not xenotransplantable (60). The MAT BIII is a cultured subline established from a transplantable rat ascites tumor derived from the 13762 mammary adenocarcinoma. We decided to include in this study the mouse mammary epithelial cell line HC11 for two reasons. First, analysis of the cytokeratins expressed in HC11 cells suggests they have characteristics of a mammary-specific stem cell (61). Second, by growing HC11 cells in EGF-containing medium, and upon confluence withdrawing EGF, these cells become competent in responding to lactogenic hormone treatment and expressing milk proteins (62). This makes HC11 cells useful for studying molecular mechanisms regulating expression of Muc4/SMC in mammary gland. 0.1 g of MAT-B1 and MAT-C1, and 10.0 g of MAT-BIII, HC11, virgin, and lactating mammary total protein were loaded for immunoblot analysis with anti-ASGP-2 mAb 4F12. The Muc4/SMC bands from the 0.1 g of MAT-B1 and 10 g of lactating mammary tissues are of similar intensity (Fig. 1). Lactating mammary tissue contains ϳ85% epithelial cells (54), whereas the MAT-B1 cells are a pure cell preparation. Thus, if a pure preparation of mammary epithelial cells were loaded for this analysis, the signal would be ϳ15% more intense. These data indicate that there are ϳ85-100-fold more Muc4/SMC protein expressed in the MAT-B1 tumor cells than in the normal lactating mammary tissue. Interestingly, there is ϳ100-fold more Muc4/SMC protein present in lactating mammary tissue than in virgin mammary tissue (ϳ16-fold more per cell). Thus, the MAT-B1 cells express ϳ10,000-fold more Muc4/SMC than normal virgin mammary tissue. Corrected for cell number, MAT-B1 cells express roughly 1500-fold more Muc4/SMC per cell than virgin mammary epithelial cells. The MAT-C1 cells also have a very high level of Muc4/SMC expression compared with normal mammary epithelial cells from HC11 cell line and virgin rat mammary gland. The cultured subline MAT-BIII has a lower level of Muc4/SMC expression compared with rat epithelial cells from lactating mammary gland. Interestingly, when we put the 13762 ascites cells in culture, we observed a downregulation of Muc4/SMC (63). The transmembrane subunit of Muc4/SMC acts as an intramembrane ligand for the receptor tyrosine kinase ErbB2 and potentiates its phosphorylation (64). Protein expression studies by Western blotting show similarities compared with mammary epithelial cells between Muc4/SMC and ErbB2 expression in MAT-B1 and MAT-C1 ascites cells (Fig. 1). ErbB2 also plays a role in human malignancies. It is amplified and/or overexpressed in ϳ25% of human breast carcinomas and in many other malignancies (65,66). Because it has been reported that PEA3 is overexpressed in ErbB2-induced breast tumors and their metastases (37,38) and PEA3 can also stimulate transcription of ErbB2, the receptor for Muc4/SMC (67), we decided to examine the level of PEA3 expression in rat tumor cells versus normal mammary epithelial cells. Western blotting analysis showed elevated PEA3 levels in MAT-B1, MAT-C1, and MAT-BIII 13762 sublines when compared with normal mammary epithelial cells from HC11 cell line and virgin or lactating rat mammary gland (Fig. 1). This correlation suggests that PEA3 may regulate the transcription of Muc4/SMC gene. Indeed, scanning the Muc4/ SMC promoter has shown an authentic PEA3 binding site and several core elements (5Ј-GGA(A/T)-3Ј) for ETS factors (27). However, it is not known whether the PEA3 site is functional.

Muc4/SMC Transcript Level Is Amplified in 13762 Ascites
Tumor Cells-Mutations in the Muc4/SMC promoter or other genes (transcriptional activators or repressors) may contribute to overexpression of the Muc4/SMC gene in the tumor cells. To compare the steady state level of Muc4/SMC transcript in MAT-B1, MAT-C1, and normal mammary tissue, Northern blot analysis was performed. Total RNA was isolated from MAT-B1, MAT-C1, or mid-and late-pregnant rat mammary tissue, and 25 g of total RNA were subjected to Northern blot analysis with the probe A2G2-#9. A single 9.2-kb band was stained in the MAT-C1 and normal mammary tissue samples, whereas two bands were stained in the MAT-B1 sample, one of 9.2 kb and another of ϳ7.5 kb ( Fig. 2A). This smaller band may represent a transcript produced from a form of Muc4/SMC truncated in the region of the tandem repeats in the MAT-B1 cells similar to the spliced variant forms of human pancreatic tumor-associated mucin MUC4 (68). The stained bands were quantified by densitometry. The MAT-B1 and MAT-C1 cells contain 6-and 9-fold more Muc4/SMC transcript at steady state than normal rat mammary epithelial cells, respectively (Fig. 2B). Thus, there is some defect in the tumor cells that allows for the presence of more Muc4/SMC transcript in these cells.
PEA3, but Not ERM or ER81, Can Transactivate the Muc4/ SMC Promoter-Transient transfections were performed to determine whether the Muc4/SMC promoter was susceptible to regulation by the PEA3 subfamily of ets genes. The mouse mammary epithelial cell line HC11 was selected for these experiments because it has been shown to be a good model system for studying mammary cell differentiation and gene regulatory mechanisms taking place in mammary cells (43). HC11 cells were co-transfected with PEA3, ERM, and ER81 expression vectors and with WT-PEA3, a luciferase reporter construct containing 335 bp of the Muc4/SMC promoter. We observed a dramatic rise in promoter activity in a dose-dependent manner after transfection of PEA3 cDNA, but not with ERM or ER81 expression vectors (Fig. 3). The same result was obtained using a luciferase reporter linked to ϳ500 bp of Muc4/SMC promot-er, 2 which contains two perfect consensus binding sites for PEA3. When the PEA3 binding site was mutated, the transactivation activity of PEA3 was reduced by 50% (Fig. 3A). As expected, the pGL3-Basic vector produced the lowest level of luciferase activity. Taken together, these studies demonstrated that PEA3 could activate transcription by interacting with a binding site on the Muc4/SMC promoter.
Autoinhibition of PEA3 DNA Binding within the Muc4/SMC Promoter-The PEA3 DNA-binding site (containing the central 5Ј-GGA(A/T)-3Ј motif) within the Muc4/SMC promoter represents a high affinity binding site for many ETS-domain transcription factors. DNA binding by mPEA3 was therefore investigated by gel retardation analysis using this site (Fig. 4). Using the mouse PEA3 sequence (GenBank TM accession no. X63190) to search in the human and rat genome data bases (www.ncbi.nih.gov/Genomes/) showed over 90% amino acid identity between the three proteins and ϳ99% amino acid sequence conservation within the ETS domains (32,69). 2 For this reason we consider it appropriate to use the expression vector for mPEA3 in binding experiments within the Muc4/ SMC promoter. Full-length mouse PEA3 expressed in HC11 cells bound the rPEA3 consensus site very weakly (Fig. 4C,  lanes 2-8). Moreover, there were no differences in the compe-2 A. Perez and K. L. Carraway, unpublished data. tition for complex formation between PEA3-related and nonrelated sequences (Fig. 4C, lanes 3-6), and no supershifted complex was observed when PEA3 antibody was included in the binding reaction (Fig. 4C, lanes 2 and 7). As several ETS domain transcription factors exhibit autoinhibition of their DNA binding activity (70,71), a series of truncated PEA3 proteins were constructed (Fig. 4) and transiently transfected into HC11 cells. Nuclear extracts from the truncated proteins were tested for binding to a PEA3 site within the Muc4/SMC promoter in comparison to the full-length protein (Fig. 4C). Deletion of either N-terminal (151 amino acids) or C-terminal (40 amino acids) stimulated the DNA binding (Fig. 4C, compare   lanes 2, 9, and 15), but again the specificity of the binding could not be shown. The retarded complexes containing the truncated proteins were competed with non-related as well as with PEA3related sequences (Fig. 4C, compare lanes 9 and 15 to lanes  10 -12 and 16 -18). Interestingly, cold competition with SP1 oligonucleotide clearly reduces the binding of full-length and truncated forms of PEA3 to its binding site within the Muc4/ SMC promoter (Fig. 4C, lanes 10, 16, and 18). This results shows that SP1 may be part of this retarded complex. Moreover, two SP1 binding sites are flanking the PEA3 consensus element within the Muc4/SMC promoter (27). However, the competition with mutant PEA3 oligonucleotide induces a de- FIG. 4. DNA binding of PEA3 within the Muc4/SMC promoter is regulated by intramolecular inhibition. A, schematic representation of full-length mPEA3 and two truncated proteins. The numbers of the amino acids truncated are indicated. B, full-length and truncated proteins were expressed by transient transfections in HC11 cells. Nuclear extracts were obtained 48 h after transfection, as described under "Experimental Procedures," and equimolar amounts of each protein was loaded onto 12% SDS-PAGE followed by immunoblotting with a monoclonal anti-PEA3 antibody. Nuclear extract from untransfected HC11 cells was included as a negative control for PEA3 expression. C, the DNA-binding activity of mPEA3 and the truncated derivatives were analyzed by gel retardation analysis. Equimolar amounts of each protein (indicated above each lane) were incubated with a 32 P-labeled double-stranded oligonucleotide containing the PEA3 motif within the Muc4/SMC promoter. Controls for the specificity of the binding include: competition with unlabelled PEA3-related oligonucleotides (rPEA3 and mPEA3) and non-related oligonucleotides (mMPEA3, SP1). Inclusion of antibodies against PEA3 and actin was done to show the presence of the transcription factor PEA3 in DNA-protein complex. *-Corresponds to a partially degraded protein complex. crease of shifted band stronger than the rat and mouse PEA3 oligonucleotide (Fig. 4C, lanes 3-5). Therefore, the data indicate that SP1 competition may not be specific and could be an experimental artifact. The same results were seen when nuclear extracts from MAT-B1 cells were used. 2 Taken together, these results demonstrate that PEA3 DNA binding is inhibited by an intramolecular mechanism involving residues from domains located N-and C-terminal of the ETS domain. However, DNA binding activity could be uncovered by incubation with PEA3-specific antibodies (against the N-and C-terminal regions of PEA3) or by deleting the two regulatory regions that flank the ETS domain (71).
PEA3 Exerts Its Transactivating Function through the PEA3 Binding Site on the Muc4/SMC Promoter-To show that PEA3 exerts its ability to activate the Muc4/SMC promoter via binding to a consensus site on the Muc4/SMC promoter, we constructed a GST fusion protein encoding the Ets domain of mouse PEA3 (GST-EtsmPEA3) (Fig. 5A). We decided to use only the Ets domain of PEA3 in the binding experiments because the N-terminal activating domain and the C-terminal Ets domain are flanked by regions that negatively regulate DNA binding (71). GST and GST-EtsmPEA3 were expressed in E. coli, partially purified using affinity matrix Glutathione-Sepharose TM 4B and subjected to immunoblot analysis using a polyclonal anti-GST antibody (Fig. 5B). Electrophoretic mobility shift assay was performed to confirm the DNA binding potential of PEA3 to Muc4/SMC promoter (Fig. 5C). Using a 32 P-labeled oligonucleotide that includes a PEA3 binding site within Muc4/SMC promoter (27), GST-EtsmPEA3 binding was apparent in lane 3, but not in lane 2 with GST alone (Fig. 5C). The specificity of the PEA3 binding was demonstrated by competition with a 100 M excess of unlabeled rat and mouse consensus PEA3 sequence (Fig. 5C, lanes 4 and 5) and the absence of a competition effect when using a mutant PEA3 or SP1 oligonucleotides (Fig. 5C, lanes 6 and 7). It is important to notice that competition with the mouse PEA3 consensus sequence is more robust than competition with the rat PEA3 consensus sequence (5Ј-AGGAAGG-3Ј) (Fig. 5C, lanes 4 and 5). Mutation of the G and/or T at the 3Ј extremity of the motif (5Ј-AGGAAGT-3Ј) interfered with the capacity of the oligonucleotide to act as a competitor in DNA binding assays (30). Direct confirmation of binding of GST-EtsmPEA3 to Muc4/ SMC promoter was indicated by a supershift assay using a GST antibody (lane 8). In contrast, no supershifted DNA-protein complex was evident when an anti-actin antibody was used in place of GST antibody (lane 9).
Effect of Endogenous and Transfected PEA3 on Muc4/SMC Transcriptional Activation in Normal Epithelial (HC11) and Ascites (MAT-B1) Cells-To determine the effect of endogenous PEA3 on Muc4/SMC transcriptional activation, HC11 and MAT-B1 cells were transfected with the wild-type or the mutated PEA3 site of the Muc4/SMC promoter luciferase reporter (WT-PEA3 and MT-PEA3; Fig. 6A). A significant decrease in luciferase activity was observed in MAT-B1 cells transfected with the mutant PEA3 reporter compared with the wild type. In contrast, transfection of both reporters in HC11 cells showed FIG. 5. Analysis of the specific DNA binding activity of recombinant GST-EtsmPEA3 to the PEA3 element within the Muc4/SMC promoter. A, schematic representation of the GST fusion protein chimera. Both proteins GST and GST-EtsmPEA3 were expressed in E. coli and partially purified by affinity chromatography as described under "Experimental Procedures." 1 g of either GST or GST-EtsmPEA3 were loaded onto 12% SDS-PAGE followed by immunoblotting with a polyclonal anti-GST antibody. B, equimolar amounts of GST (lane 1) or GST-EtsmPEA3 (lanes 2-9) were incubated with a 32 P-labeled doubledstranded oligonucleotide containing the PEA3 binding site, corresponding to positions 2073-2092 within the Muc4/SMC promoter (GenBank Accession number AF240632, 5Ј-GTAGGCAGGAAGGTGC-CCTC-3Ј). The reaction mixture was electrophoresed in a 5% non-denaturing polyacrylamide gel and exposed to autoradiography for 2 h at Ϫ70°C. Lane 8 shows the supershifted PEA3 DNA-protein complex. As specific competitors we used the rPEA3 (lane 4) and the PEA3 consensus sequence (32), mPEA3 (lane 5) 5Ј-GATCCAGGAAGTGACTAACG-3Ј oligonucleotides. Lanes 6 and 7 show competition with non-related PEA3 sequences, the mutant PEA3 (5Ј-GATCTAAAAAGA-CTAACG-3Ј) and SP1 consensus sequence (5Ј-ATTCGATCGGGGCGGGGC-GAGC-3Ј). significantly less difference in luciferase activity, suggesting that overexpression of PEA3 in MAT-B1 cells is responsible for this effect. Next, we decided to elucidate the signaling mechanism implicated on transcriptional activation of Muc4/SMC in MAT-B1 cells by PEA3. Cotransfection of the WT-PEA3 reporter with the PEA3 expression vector in HC11 and MAT-B1 cells showed a greater induction of luciferase activity in tumor cells compared with the reporter alone (Fig. 6B). Because the expression of transfected PEA3 is the same in both cell types (data not shown) and the MAPK cascade stimulates the transcriptional activity of PEA3 (42), the 4-fold induction of luciferase activity in MAT-B1 cells compared with HC11 cells could be the result of an increase in ERK activity. To determine whether the ERK pathway is involved in up-regulation of PEA3 in 13762 ascites cells, we determined the expression levels of PEA3 (Fig. 1B) and ERK in mammary epithelial and tumor cells (Fig. 6C). MAT-B1, MAT-C1, and MAT-BIII cells exhibited higher levels of PEA3 expression (Fig. 1) than normal mammary epithelial cells. Normal mammary epithelial and tumor cells express similar levels of ERK-1 and ERK-2; however, ERK-1 and ERK-2 had a higher degree of activation (phosphorylation) in tumor cells (Fig. 6C). Moreover, the ERK phosphorylation levels paralleled the Muc4/SMC expression levels in these cells. These data suggest that the ERK pathway could be involved in the induction of Muc4/SMC by PEA3.

PEA3-dependent Muc4/SMC Transcriptional Activation Involves Both the MAPK/ERK and SAPK/JNK Cascades in HC11
Cells-Ras has multiple downstream effectors that activate divergent signaling pathways, such as the GTP-binding protein Rac1 (and Cdc42) and the serine threonine kinase Raf-1 (72). The pathway downstream of Raf is well defined (73); Raf phosphorylates and activates MEK1, which in turn phosphorylates and activates ERK1 and ERK2. Signaling events downstream of Rac1 and Cdc42 are less well understood. More recently, Raf-independent Ras-activated MAPK pathways have been identified. For example, the Ras effector MAPK kinase kinase 1 (MEKK1) activates SEK, which in turn activate members of the SAPK/JNK subfamily of MAPKs (74 -77). Thus, Ras-dependent kinase cascades can activate both the ERKs and SAPK/JNKs.
To determine whether the expression of Muc4/SMC is regulated by Ras, we inquired whether RasV12, a constitutively active form of Ras, was capable of enhancing PEA3-dependent Muc4/SMC transcriptional activity. To this end we cotransfected HC11 cells with a WT-PEA3 Muc4/SMC reporter, PEA3, and RasV12 expression vectors. Transfection of the reporter plasmid DNA with the vector encoding RasV12 alone increased the amount of luciferase activity recovered from the lysates ϳ2-fold, suggesting the presence of one or more Ras-stimulatable transcription factors endogenous to HC11 cells (Fig. 7A,  lane 3). Co-expression of PEA3 and RasV12 markedly increased PEA3-dependent luciferase expression (compare lanes 4 and 5). To determine whether Ras functioned through a MAPK/ERK pathway to stimulate PEA3-dependent luciferase activity, we determined the ability of MEK-1 inhibitor U0126 to suppress the activation of Muc4 reporter activity elicited by RasV12. The inhibitory effect of U0126 was partial (compare lanes 5 and 7), suggesting that ERK pathway is not the only mechanism involved in the transcriptional activation of Muc4/ SMC by PEA3. Transfected cells were also analyzed by immunoblotting with anti-ERK and anti-phospho-ERK antibodies. When RasV12 was overexpressed, ERK activation and luciferase expression were higher than in U0126-treated cells.
To learn whether the SAPK/JNK pathway may also potentiate the transcriptional activation of Muc4/SMC by PEA3 in HC11 cells, we transfected the Muc4/SMC reporter (WT-PEA3) FIG. 6. Effect of mutating the PEA3 site in the Muc4/SMC promoter on reporter assays and ERK activation on HC11 and MAT-B1 cells. A, HC11 and MAT-B1 cells were transfected with 1 g of WT-PEA3 (wild type) or MT-PEA3 (mutated) Muc4-luciferase reporter. After 24 h, cell lysates were prepared and used for luciferase activity. The effect of endogenous PEA3 from HC11 and MAT-B1 cells on mutated PEA3 site is shown as a fold decrease of mutated reporter versus wild type in each cell type. B, HC11 and MAT-B1 cells were co-transfected with 1 g of wild type PEA3 reporter and 0.5 g of the PEA3 expression vector (pCANmycPEA3). Fold activation compares the activity in HC11 and MAT-B1 cells. Luciferase activity represents data that have been normalized to Renilla activity. Columns, means; bars, S.D.; n ϭ 3. C, PEA3, ERK1&2, pERK1&2 and actin protein levels were determined by Western blotting. Ten micrograms of protein from each cell type was loaded onto 10% denaturing polyacrylamide gel and electrophoresed for 3 h, followed by transfer onto a nitrocellulose membrane. The proteins were detected using the respective antibodies, i.e. anti-PEA3, anti-ERK (1&2), anti-phospho-ERK (1&2) or anti-actin. with PEA3 and MEKK1 expression vectors. To ensure that we could measure an additional effect of MEK kinase 1 on PEA3stimulated luciferase gene expression, we used a concentration of the PEA3 expression vector that yielded a 2-fold increase in luciferase activity under conditions of serum starvation (Fig.  7B, compare lanes 2 and 3). Co-expression of PEA3 and MEKK1 led to nearly a 4-fold increase in PEA3-dependent luciferase expression (compare lanes 3 and 5). The endogenous activity was increased 2-fold following the expression of MEKK1 (compare lanes 2 and 4). The ability of constitutively activated MEK1 (data not shown) and MEKK1 to directly stimulate PEA3-dependent Muc4/SMC transcriptional activity supports the notion that the activity of Muc4/SMC is independently regulated by two different MAPK cascades.

DISCUSSION
Muc4/SMC is overexpressed on the surface of the 13762 ascites tumor cells (Ͼ10 6 copies/cell) (2). There are many mechanisms by which regulation of Muc4/SMC expression can be disrupted in tumors. At the genomic level, gene amplification can cause overexpression when each extra copy of the gene expresses even at normal levels. For example, MUC1 is overexpressed in a number of breast cancers by amplification of small chromosomal regions containing the MUC1 gene (78). In the 13762 ascites tumor cells, comparing genomic Southern blots of normal rat liver and 13762 ascites tumor cell indicated that the Muc4/SMC gene is amplified by ϳ3-fold (56). An ϳ3fold amplification alone does not explain the level of Muc4/SMC overexpression in the 13762 tumor cells. Gross genetic rearrangement or deletion could cause overexpression by altering the type or presence of regulatory elements present in the gene. Genomic Southern blots using probes from the 5Ј or 3Ј end of the cDNA have not shown any gross rearrangements or deletions in the tumor Muc4/SMC gene (56). Thus, there must be other mechanisms for overexpression of Muc4/SMC in these cells. Muc4/SMC may be overexpressed in the 13762 tumor cells because of changes in the rate of transcription or message half-life, or changes in the rate of Muc4/SMC translation or Muc4/SMC protein half-life. It is likely that more than one of these mechanisms is responsible for Muc4/SMC overexpression in the tumor cells. Availability of the Muc4/SMC transcriptional regulatory region helped to understand how it is regulated in normal tissues and how its misregulation may lead to aberrant expression in tumor cells. The high relative promoter activity present within the first 300 bp downstream of Muc4/ SMC transcription start site drew attention to several transcription factor-binding sites (27). One of these, the PEA3 binding site, served as the focus for this study. Our main goal was to determine whether PEA3 could activate transcription through this binding site.
It was shown that PEA3 has a substantial impact on Muc4/ SMC transcriptional activation. PEA3 was capable of transactivating Muc4/SMC promoter in a dose-dependent manner via direct attachment to its binding site. PEA3 appears to be a specific transactivator for Muc4/SMC gene because neither ERM nor ER81 could transactivate Muc4/SMC promoter. Mutation of the PEA3 binding site reduced the transcriptional activity of PEA3 by 50% (Fig. 4A). It is important to mention that besides the perfect PEA3 binding site there are several potential Ets-binding sites (5Ј-GGA(A/T)-3Ј) present in the minimal PEA3-responsive fragment of the Muc4/SMC promoter (27), although none corresponds exactly to a consensus PEA3 site. Thus, it seems likely that overexpression of PEA3 and its binding to any of these core Ets sites is responsible for the remaining 50% transactivation by PEA3 in the mutated Muc4/SMC reporter. Another possibility may be that PEA3 may increase the gene expression of other proteins that then activate the promoter. Thus, it may have an indirect effect. Our studies used only 335 bp of the Muc4/SMC promoter to demonstrate the effect of PEA3 on the Ets binding site on this promoter. If the whole length of the Muc4/SMC promoter were used, it might mask the activity of PEA3 as numerous positive and negative cis-elements on this promoter come into play with their respective transcription factors (27). Muc4/SMC gene is not the only mucin gene target of PEA3 transcription factor. PEA3 strongly transactivates MUC4 promoter in transient FIG. 7. Ras and MEKK1 kinases potentiate transactivation of Muc4/SMC promoter by PEA3. A, HC11 cells were serum-starved for 24 h followed by 30 min incubation with 80 M of U0126 before transfection with 1 g of either pGL3-Basic or WT-PEA3 with or without 0.5 g of PEA3 and/or RasV12 expression vectors. Twenty-four hours after transfection cells were extracted and subjected to luciferase activity assay and to immunoblot analysis with anti-ERK, anti-phospho-ERK or anti-actin as indicated at the right of each panel. B, additive effect of constitutively-activated MEKK1 and PEA3 on the expression of luciferase reporter in transient transfection on HC11 cells. HC11 cells were serum-starved for 24 h before transfection with 0.5 g of WT-PEA3 luciferase reporter and the expression vectors for PEA3 and MEKK1. Luciferase activity was measured 24 h after transfection. Luciferase activity represents data that have been normalized to Renilla activity. The values shown are the average of three experiments plus and minus the standard deviation. transfection assays in the pancreatic cancer cell line Capan-1. 3 The elevated PEA3 levels in 13762 tumor cells that also overexpressed Muc4/SMC and ErbB2 led us to investigate the potential role of this transcription factor in the Muc4/SMC regulation pathway in cancer (Fig. 1). Therefore, it is highly likely that the overexpression of PEA3 seen in various tumors contributes to the amplification of Muc4/SMC transcript. How does PEA3 overexpression initially take place? Although this question has not been definitively answered, PEA3 can be activated through two distinct MAPK cascades (42). Similarly, ErbB2 (the receptor for Muc4/SMC) regulates PEA3 activity through two Ras-dependent pathways: the ERK and the SAPK/JNK cascades (79). In addition, we have recently demonstrated that in mammary epithelial cells, Muc4/SMC is partially regulated at the transcript level by an ERK-dependent pathway (80). Our present findings have shown that the same two MAPK cascades, the ERK and the SAPK/JNK pathways, independently stimulate the PEA3-dependent Muc4/SMC transcriptional activation. Finally, note that transcriptional activation of MUC2 and MUC5AC by factors of EGF family also involves the Ras/ Raf/ERK signaling cascade (81).
The phosphorylation of PEA3 through the ERK pathway carries multiple consequences. Activated PEA3 may enhance transcription of its target genes, which likely include the ErbB2 and PEA3 genes (67). Up-regulation of PEA3 activity and abundance in tumor cells may also enhance the invasive and metastatic potential of these cells. PEA3 is overexpressed in metastatic mammary tumors (38), and ectopic overexpression of PEA3 in human mammary MCF-7 cells, which express very low levels of ErbB2 and PEA3, confers an invasive phenotype to these cells (82). This may result in part from PEA3-mediated transcriptional activation of genes encoding matrix-degrading proteinases (82). Furthermore, our results suggested that Muc4/SMC is a target of PEA3 transcriptional activation and overexpression of PEA3 in rat mammary tumor cells contributes to their metastatic potential by up-regulating Muc4/SMC expression.