Downregulation of Muc1 in MMTV-C-Neu Tumors

Downregulation of Muc1 in MMTV-c-Neu tumors

Melissa C Adriance1,3 and Sandra J Gendler*,1,2

1Tumor Biology Program, Mayo Medical/Graduate School, Mayo Clinic Scottsdale, Scottsdale, AZ 85259, USA; 2Department of Biochemistry and Molecular Biology, Mayo Medical/Graduate School, Mayo Clinic Scottsdale, Scottsdale, AZ 85259, USA

MUC1 is a tumor antigen, overexpressed in approxi- MUC1 is found to be overexpressed in approximately mately 90% of human breast cancers. In normal glandular 90% of human breast cancers, with a clinical signifi- epithelia, MUC1 is expressed at the apical surface; cance of poor patient prognosis (Zotter et al., 1988; however, in carcinomas an aberrantly glycosylated form McGuckin et al., 1995; Croce et al., 2003). Further, of MUC1 is upregulated and expressed around the entire increases in MUC1 expression at least as high as ten fold surface of the cell. Previously, we have shown that a lack have been detected (Hilkens et al., 1984); however, the of Muc1 significantly delays tumor progression and/or processes responsible for the stimulation of MUC1 onset in MMTV-PyV-mT and MMTV-Wnt-1 transgenic expression during cancer are currently unclear. The mice. Here we show that, unlike the models mentioned complexity of MUC1 expression is suggested by the fact above, a loss of Muc1 in MMTV-c-Neu mice (MMTV-c- that hormones, cytokines, and gene modifications are Neu/Muc1À/À) altered neither mammary tumor onset nor implicated in its regulation. progression. Moreover, characterization of MMTV-c- A role for MUC1 during tumor progression has been Neu/Muc1 þ / þ tumors revealed that Muc1 expression was demonstrated in studies utilizing transgenic models of repressed at the level of transcription. In contrast, normal breast cancer. MUC1 is the human designation and mammary gland tissue adjacent to tumor tissue expressed Muc1 is the mouse designation. Crossing the MMTV- Muc1 and pregnant mammary glands from c-Neu PyV-mT (MTag) and MMTV-Wnt-1 models onto a transgenic animals expressed high levels of Muc1. We Muc1 null background affects tumor formation. In the found that transient transfection of activated ErbB2 into MMTV-PyV-mT model, MMTV-PyV-mT/Muc1À/À human embryonic kidney 293/MUC1 cells resulted in the mice display a significantly slower tumor progression repression of MUC1 expression. Further, transient and a trend toward decreased metastasis (Spicer et al., transfection of activated ErbB2 resulted in the inhibition 1995). Moreover, MMTV-Wnt-1 mice on a Muc1 null of Muc1 transcriptional activation in luciferase reporter background display a delay in tumor onset (almost assays. These data suggest that the activation of ErbB2, double the time to tumor onset in MMTV-Wnt-1/ which only occurs in c-Neu tumors, selectively inhibits Muc1 þ / þ mice) (Schroeder et al., 2003). Further, in Muc1 expression. MMTV-Wnt-1 tumors Muc1 associates with b-catenin, Oncogene (2004) 23, 697–705. doi:10.1038/sj.onc.1207165 suggesting that the cytoplasmic domain of MUC1 may allow MUC1 to facilitate molecular interactions that Keywords: transmembrane mucin; erbB receptors; promote tumorigenesis and metastasis. ErbB2; MUC1 Previously, our laboratory demonstrated that MUC1/ ErbB2 complexes are detected in mammary glands from lactating MUC1 transgenic mice. In this experimental model, MUC1 potentiated ErbB signaling, as seen by strong induction of Erk1/2 phosphorylation. ErbB2, as Introduction well as MUC1, is overexpressed in human breast cancers, and overexpression is associated with poor During the progression to carcinoma, a change in the prognosis. ErbB2 overexpression occurs in 80% of cellular profile of gene expression occurs. One gene ductal carcinoma in situ (DCIS) and in 25% of invasive highly induced during the progression to carcinoma is cancers (Slamon et al., 1987; Revillion et al., 1998; MUC1. MUC1 is a transmembrane mucin with a Witton et al., 2003). To examine the functional heterodimeric structure consisting of a large negatively significance of Muc1 during ErbB2-mediated tumorigenesis, we crossed MMTV-c-Neu mice onto a Muc1- charged extracellular domain and a smaller subunit À/À consisting of transmembrane and cytoplasmic domains. deficient background (MMTV-c-Neu/Muc1 )ora Muc1 wild-type background (MMTV-c-Neu/Muc1 þ / þ ). We found that the loss of Muc1, unlike previous *Correspondence:SJ Gendler, Mayo Clinic Scottsdale, 13400 E. Shea findings, did not alter tumor progression in the Blvd., Scottsdale, AZ 85259, USA; E-mail:[email protected] 3 MMTV-c-Neu mice. Surprisingly, MMTV-c-Neu/ Present address:Lawrence Berkeley National Lab, Cell and Mole- þ / þ cular Biology, 1 Cyclotron Road, MS 83-101, Berkeley, CA 94720, Muc1 tumors displayed no to very low Muc1 USA expression, while adjacent normal tissues in the mam- Received 27 June 2003; revised 26 August 2003; accepted 27 August 2003 mary gland expressed expected levels of Muc1, which Activated ErbB2 represses MUC1 expression MC Adriance and SJ Gendler 698 may explain the lack of an effect of deleting the Muc1 gene on tumor development in this model. The decrease in Muc1 in the MMTV-c-Neu tumors was at the level of mRNA, and it was a tumor-specific event. Siegel et al. previously demonstrated that activating mutations occur with the c-Neu transgene in a tumor-specific manner (Siegel et al., 1994, 1999; Siegel and Muller, 1996). Thus, we examined if the activation of ErbB2 was responsible for mediating the inhibition of Muc1. We found that transient transfection of human embryonic kidney (HEK) 293/MUC1 cells with activated ErbB2 resulted in repression of MUC1 and that this repression was due to the inhibition of transcriptional activation. These findings suggest that during neoplastic progression, Neu negatively regulates Muc1 expression in the c- Neu tumors.

Figure 1 Loss of Muc1 does not impact tumorigenesis in MMTV- c-Neu transgenic mice. MMTV-c-Neu transgenic mice were crossed Results onto a Muc1 null background. Tumor onset was monitored in MMTV-c-Neu/Muc1 þ / þ and MMTV-c-Neu/Muc1À/À mice. Tu- mor incidence of 50% occurred at 39 weeks for MMTV-c-Neu/ Loss of Muc1 has no effect on tumor onset in Muc1À/À and 46 weeks for MMTV-c-Neu/Muc1 þ / þ (P ¼ 0.13 log- MMTV-c-Neu transgenic mice rank test) In order to understand the role of Muc1 in tumor progression and metastasis, we crossed MMTV-c-Neu transgenic mice onto a Muc1 null background. MMTV- c-Neu transgenics develop unifocal mammary adeno- carcinomas surrounded by a hyperplastic mammary epithelium after a long latency period of approximately 7–12 months (Guy et al., 1992). The c-Neu transgenic model mimics the human breast cancer situation in that metastases occur in the lung (Guy et al., 1992). We found that there was no signiﬁcant difference in the time to tumor onset between MMTV-c-Neu/Muc1À/À mice and MMTV-c-Neu/ Muc1 þ / þ mice (50% tumor incidence at 39 weeks for MMTV-c-Neu/Muc1À/À mice and 46 weeks for MMTV-c-Neu/ Muc1 þ / þ mice; P ¼ 0.13 log-rank test) (Figure 1). Moreover, the rate of metastasis in MMTV-c-Neu mice on Muc1 wild-type or null backgrounds was not signiﬁcantly different.

Muc1 is not expressed in MMTV-c-Neu/Muc1 þ / þ mammary tumors As loss of Muc1 had no effect on tumor onset or progression in MMTV-c-Neu tumors, we examined the level of expression of Muc1 in MMTV-c-Neu/Muc1 þ / þ tumors. Mammary carcinomas in the mouse induced by mouse mammary tumor virus show high levels of expression of Muc1 both at the RNA and protein levels (SJ Gendler, unpublished data). Moreover, Spicer et al. Figure 2 MMTV-c-Neu/Muc1 þ / þ transgenic mice express Muc1 showed that MTag mammary carcinomas express high in normal mammary glands, although Muc1 is largely absent in levels of the tumor antigen Muc1 (Spicer et al., 1995). tumors. Immunohistochemical detection of Muc1 was performed using anti-Muc1 CT2 (a–c, and e) and of c-Neu using anti-Neu (Sc- Unlike the tumor models mentioned before, Muc1 284) (d and f). (a, c–f) are c-Neu-induced tumors, and (b) represents protein was absent (or expressed at very low levels) normal mammary gland in the MMTV-c-Neu/Muc1 þ / þ mammary tumors (Figure 2a); however, normal appearing mammary expression, 21 tumors had very low Muc1 expression tissue adjacent to the c-Neu tumors displayed strong (Figure 2c), and 10 tumors were completely negative for apical expression of Muc1 (Figure 2b). We stained a Muc1 (Figure 2e). However, the tumors had strong total of 33 MMTV-c-Neu/Muc1 þ / þ mammary tumors expression of the Neu transgene by immunostaining and found that two tumors were positive for Muc1 (Figure 2d and f).

Oncogene Activated ErbB2 represses MUC1 expression MC Adriance and SJ Gendler 699 To quantitate the level of Muc1 protein expressed, Muc1 mRNA was reduced in MMTV-c-Neu tumors immunoblot analysis was performed and it was demon- To examine the mechanism by which Muc1 is down- strated that Muc1 was present at low levels, but high modulated in the c-Neu-induced tumors, Northern blot levels of c-Neu were easily detected (Figure 3a). This analysis of total cellular RNA was performed and result contrasted with the results in the nontransgenic showed low levels of Muc1 mRNA expression lactating mammary gland and MTag mammary tumor, (Figure 3b). The expected level of Muc1 expression in which strong Muc1 expression was observed, but low would be that seen in the positive controls (a mammary levels of c-Neu were observed (Figure 3a). gland of a lactating mouse and ET mouse pancreas carcinoma); the observed expression was much lower than expected, suggesting that Muc1 is downregulated at the level of transcription. This was compared with the high levels of keratin 18 and Neu mRNA detected (Figure 3b). The c-Neu tumors were highly epithelial as shown by their strong expression of keratin 18 mRNA. The decreased Muc1 mRNA paralleled the reduction of Muc1 protein detected in these c-Neu tumors, suggesting that the reductions in steady-state levels of Muc1 RNA lead to decreases in Muc1 protein.

Muc1 expression increased in c-Neu transgenic mice during pregnancy To determine if overexpression of c-Neu is sufficient to induce the downregulation of Muc1 expression, mammary glands from pregnant MMTV-c-Neu/Muc1 þ / þ Tg mice were analysed. There is an increase in Muc1 expression at day 10 of pregnancy and maximal levels are reached by day 14 of pregnancy. The c-Neu transgene is under the control of the MMTV promoter, suggesting that there should be an increase in the expression of the transgene during pregnancy and lactation, which was what we saw (Figure 4d). Transgenic MMTV-c-NeuMuc1 þ / þ mammary glands taken at day 15 of pregnancy showed the expected high level of Muc1 expression (Figure 4a and c). The amount of Muc1 expressed at day 15 of pregnancy was equivalent to the amount of expression in the MTag tumor (Figure 4c). The Muc1 staining pattern of the transgenic pregnant mammary gland paralleled the pattern found in the nontransgenic pregnant mammary gland, where Muc1 stained predominantly on the apical side of the lumenal cells (Figure 4a and b). These data suggested that overexpression of Neu was not sufficient to downregulate Muc1 and that a tumor-specific event was responsible for the decreased Muc1 expression. This is supported by the fact that it has been previously shown that the c-Neu tumors contain extracellular deletions in the juxtamembrane domain that oncogeni- cally activate the c-Neu transgene (Siegel et al., 1994, 1999; Siegel and Muller, 1996).

Figure 3 Muc1 is not highly expressed in MMTV-c-Neu/Muc1 þ / þ tumors. (a) Lysate (200 mg) was prepared from mammary glands Activated ErbB2 induces downregulation of MUC1 during lactation, MTag mammary carcinomas (positive controls), expression and MMTV-c-Neu tumors, separated by SDS–PAGE, transferred onto PVDF membrane, and probed with anti-Muc1 (CT2) and To determine whether overexpression or activation of anti-Neu (Sc-284) antibodies. (b) Total RNA (20 mg) was isolated ErbB2 was necessary for ErbB2-mediated inhibition of from mammary glands during lactation, ET mouse pancreas tumor MUC1, we transiently transfected HEK 293/MUC1 (positive controls), and MMTV-c-Neu tumors, run on 1.2% cells with either an empty vector (pcDNA 3.1), wild-type formaldehyde–agarose gels and transferred onto nylon membranes. The membrane was stained with methylene blue for the 28S ErbB2 (ErbB2 WT), or activated ErbB2 (ErbB2 VE). ribosomal band for loading control. RNA was hybridized with The activated ErbB2 contains a point mutation in the either a Muc1 probe (pMuc10), ErbB2 probe, or a keratin 18 probe transmembrane domain at residue 659 that results in an

Oncogene Activated ErbB2 represses MUC1 expression MC Adriance and SJ Gendler 700

Figure 5 Transient transfection of activated ErbB2 represses MUC1 expression. (a) Protein lysates (250 mg) from HEK 293/ MUC1 cells transiently transfected with 1.0 mg of either an empty vector (empty), wild-type ErbB2 (ErbB2 WT) or activated ErbB2 (ErbB2 VE) were immunoprecipitated (IP) with anti-MUC1 (CT2) and immunoblotted with anti-MUC1 (CT2) (top panel). Protein lysates (5 mg) were immunoblotted with anti-MUC1 (CT2) (middle Figure 4 Muc1 expression is increased in c-Neu transgenic panel) or anti-Neu (Sc-284) to detect ErbB2 (bottom panel). mammary glands during pregnancy, suggesting that high levels of Immunoblot chemiluminescence was performed with West Dura c-Neu are not sufficient to downregulate Muc1. Immunohisto- extended duration chemiluminescent substrate. (b) HEK 293 cells chemical detection of Muc1 using anti-Muc1 (CT2) on c-Neu were transfected with 500 ng of Muc1-pGL3 in the presence of transgenic mammary glands during pregnancy (a) and nontrans- either empty vector control (control), wild-type ErbB2, or activated genic mammary glands during pregnancy (b) demonstrated that the ErbB2 (mg amount listed in the figure). Cells were cotransfected anticipated high levels of Muc1 were present in c-Neu transgenic with 10 ng of pCMV-b-gal to normalize for transfection efficiency. mammary glands during pregnancy. (c) Lysates (200 mg) were Results represent percent of repression versus control values. The prepared from MTag mammary tumors (positive control) and c- results are an average of three independent experiments performed Neu transgenic mammary glands during pregnancy, separated on in triplicate SDS–PAGE, transferred onto PVDF membrane, and probed with anti-Muc1 (CT2) antibody. (d) Lysates (100 mg) were prepared from virgin and pregnant c-Neu transgenic mammary glands, drive expression of the reporter gene (data not shown). separated on SDS–PAGE, transferred onto PVDF membrane, and To determine if overexpression or activation of ErbB2 probed with anti-Neu (Sc-284) antibody was responsible for the inhibition of Muc1, we performed reporter assays in HEK 293 cells cotrans- amino acid change from valine to glutamic acid, and fected with either an empty vector (control), wild-type results in ligand-independent signaling and increased ErbB2, or activated ErbB2 (ErbB2 VE). Reporter assays kinase activity of ErbB2. This mutation is homologous demonstrated that ErbB2 signaling inhibited Muc1 to the point mutation originally identified in the rat transcriptional activation in a dose-dependent manner. oncogene. Transfection with the empty vector or wild- We found that mutationally activated ErbB2 had the type ErbB2 did not result in the downmodulation of greatest inhibition (Figure 5b). Transient transfection of MUC1 expression (Figure 5a); however, we found that 0.5 mg of activated ErbB2 resulted in 80% repression, transfection with activated ErbB2 resulted in a dramatic while transfection of 0.5 mg of the wild-type gene decrease in the expression of MUC1 (Figure 5a). resulted in a 40% repression. Addition of more ErbB2 (2.0 mg) was able to induce an 80% repression. These ErbB2 signaling inhibits Muc1 transcriptional activation findings are consistent with an earlier report, which showed that transient transfection of ErbB2 could To investigate the effect of ErbB2 signaling on the inhibit Muc1 transcriptional activation. Moreover, these transcriptional activity of Muc1, we performed reporter data suggest that decreases detected in Muc1 mRNA in assays in HEK 293 cells with mouse Muc1 promoter MMTV-c-Neu tumors were due to decreases in Muc1 deletion/reporter constructs. The mouse and human transcription. promoters are well conserved within 500 bp of the transcription start site and have 74% identity (Spicer Location of the Muc1 promoter region responsible for et al., 1991). We utilized serial deletions of mouse Muc1 ErbB2 repression of Muc1 promoter constructs (containing various amounts of 50 flanking sequence) in the pGL3 basic vector. The full- To identify the region necessary for the activated ErbB2- length promoter is denoted À1836/ þ 33 and we mediated inhibition, we performed reporter assays with arbitrarily took this construct to be 100% for our Muc1 promoter deletion constructs (Figure 6a). These reporter assays in comparison with the various deletion assays suggested that the location of the region constructs. Promoter analysis demonstrated that all the necessary for ErbB2-mediated inhibition is located deletion constructs had basal activity and were able to between À247 and À108 (Figure 6b). This region

Oncogene Activated ErbB2 represses MUC1 expression MC Adriance and SJ Gendler 701

Figure 7 MUC1 is highly expressed in ErbB2-negative breast tumors, although areas of discordant expression of ErbB2 and MUC1 are detected in breast tumors overexpressing ErbB2. Immunohistochemical detection with anti-MUC1 (BC2) (left panels) and anti-ErbB2 (DAKO HercepTest) (right panels). (a and b) HER2 negative tumor, (c and d) HER2 positive tumor

human breast tumors that overexpress HER2; however, areas of discordant expression between the two proteins were observed. In some areas of the sections ErbB2 Figure 6 ErbB2-mediated inhibition of Muc1 requires region À247/À108 within Muc1 promoter. (a) Diagram of Muc1 promoter staining was very strong (Figure 7d), while these same serial 50 deletion constructs, which contain the native Muc1 areas in adjacent sections showed low to no positive promoter plus 33 transcribed base pairs of Muc1 inserted upstream expression for MUC1 (Figure 7c). Thus, even though all of firefly luciferase in the pGL3-Basic vector. (b) HEK cells were the sections were MUC1 positive, areas of reciprocal transfected with 500 ng of Muc1 50 deletion promoter constructs in the presence of either empty vector pcDNA 3.1 (control) or expression between ErbB2 and MUC1 were detected. activated ErbB2. Cells were cotransfected with CMV-b-gal to normalize for transfection efficiency. These results are an average of three independent experiments performed in triplicate Discussion contains a previously identified DNAse I hypersensitive site that was found to correlate with Muc1 expression. In the MMTV-c-Neu transgenic mouse model we found Additionally, within this region is an AP-2 site that crossing the MMTV-c-Neu model onto a Muc1 null conserved in both the mouse and human genes. Thus background elicited no effect on either tumor onset or these data suggest that an element within the À247/ tumor progression. This result differed from our À108 region is necessary for activated ErbB2-mediated previously published results, where we showed that a inhibition of Muc1. lack of Muc1 significantly delayed tumor progression and/or onset in MMTV-PyV-mT and MMTV-Wnt-1 transgenic mice (Spicer et al., 1995; Schroeder et al., ErbB2 and MUC1 expression in human breast tumors 2003). Further characterization of the MMTV-c-Neu/ To examine the effect that overexpression of ErbB2 has Muc1 þ / þ tumors revealed that the expression of Muc1 on MUC1 expression in human breast cancer, we was dramatically decreased, and that this decrease was compared the staining pattern of MUC1 and ErbB2 at the level of mRNA. Muc1 was expressed in the on adjacent sections in five ErbB2-negative (HER2-0) normal mammary gland and high levels of Muc1 were versus five ErbB2-positive (HER2-3 þ ) human breast detected in pregnant mammary glands from MMTV-c- tumors. The breast cancer sections were examined by a Neu transgenic mice, suggesting that the inhibition of pathologist at Mayo Clinic Scottsdale and were Muc1 was a tumor-specific event. To address the classified according to modified Bloom–Richardson consequence of overexpression of MUC1 during criteria. All cases classified as ErbB2 negative did not ErbB2-mediated tumorigenesis, we are currently cross- display ErbB2 staining (Figure 7b), while MUC1 ing MMTV-c-Neu transgenic mice with mice that staining was observed throughout the tumor sections overexpress full-length human MUC1. The MUC1 with strong staining in both membrane and cytoplasmic transgenic animals express a genomic clone of MUC1 areas of the tumor cells (Figure 7a). Moreover, the that contains the endogenous promoter and all the invasive areas of the breast tumors highly expressed regulatory sequences necessary to drive expression MUC1. Additionally, MUC1 was highly expressed in (Rowse et al., 1998a). Interestingly, in one bitransgenic

Oncogene Activated ErbB2 represses MUC1 expression MC Adriance and SJ Gendler 702 animal we found that MUC1 is highly expressed in the ErbB2 in human breast cancer (Turner et al., 1998). normal mammary gland; however, MUC1 is largely Thus, our findings demonstrate that mutationally absent in the mammary tumor, with staining present in activated ErbB2 can regulate MUC1 expression at the focal areas. level of transcription. Activating mutations occur with high frequency in the ErbB2 has been shown to induce the loss of epithelial c-Neu transgene in a tumor-specific manner (Siegel and properties in favor of the acquisition of a more Muller, 1996; Siegel et al., 1999). These mutations result mesenchymal phenotype, thus promoting invasion. In in activation of the gene and promote receptor vitro, ErbB2 has been shown to regulate negatively dimerization and increased activity. We tested whether epithelial adhesive proteins such as E-cadherin and a2 overexpression or activation of ErbB2 led to the integrin at the level of transcription in the process of inhibition of MUC1 expression. Transient transfection stimulating a cellular program favoring a more me- of activated ErbB2 into HEK 293/MUC1 cells resulted senchymal phenotype. During tumor development in in a dramatic decrease in MUC1 expression, while MMTV-c-Neu mice, somatic mutations in the Neu transfection of an empty vector or wild-type ErbB2 had transgene resulting in the formation of an activated no effect on MUC1 expression. Consistent with our form of ErbB2 may promote an epithelial to mesench- data, Engelman et al. (1998) also found that only ymal transition (EMT), and thus downregulate Muc1. activated ErbB2 inhibited caveolin-1 expression in cell Interestingly, work in MDCK cells, a polarized epithe- culture; moreover, caveolin-1 expression was down- lial cell line, showed that the expression of activated regulated in MMTV-c-Neu tumors, presumably due to ErbB2 favored a fibroblastic phenotype, while wild-type somatic mutations in the transgene. In contrast, Scibetta ErbB2 had no effect on cell morphology (Khoury et al., et al. (2001) found that exogenous overexpression of 2001). Moreover, snail, a zinc-finger transcriptional wild-type ErbB2 in immortalized mammary epithelial repressor activated during EMT, has been shown to cells was sufficient to repress MUC1 expression. inhibit E-cadherin and MUC1 expression (Guaita et al., Interestingly, they found that ErbB2-mediated inhibi- 2002). It is somewhat of a paradox then, that despite the tion of MUC1 was less effective in breast cancer cell fact that MUC1 is an epithelial antigen, it is highly lines. The HEK 293/MUC1 cells we tested highly expressed in invasive breast cancers. expressed MUC1 at levels comparable to the MDA- Owing to our findings in the MMTV-c-Neu tumors MB-468 breast cancer cell line. This may explain the and HEK 293/MUC1 expressing cells, we were inter- discrepancy between our findings in HEK 293/MUC1 ested in examining the effect that overexpression of cells and their findings in the mammary epithelial cells. ErbB2 had, if any, on MUC1 expression in human Utilizing Muc1 promoter-luciferase constructs in breast cancers. Both MUC1 and ErbB2 are over- transient transfection assays, we demonstrated that this expressed in human breast cancers with a clinical inhibition was at the transcriptional level. Moreover, we significance. It is interesting to note that MUC1 is provided evidence that a region located between À247/ highly expressed in invasive disease, while the over- À108 in the Muc1 promoter is required for this expression of ErbB2 predominates earlier in tumor ErbB2-mediated transcriptional regulation. This À247/ development. MUC1 is overexpressed in approximately À108 region contains a previously identified DNase I 90% of human breast cancers, while ErbB2 over- hypersensitive site that has been shown to correlate with expression occurs in 80% of DCIS and 25% of invasive MUC1 expression (Shiraga et al., 2002). Further, this cancers (Slamon et al., 1987; Zotter et al., 1988; hypersensitive site was also present in transgenic mice Revillion et al., 1998; Witton et al., 2003). In human expressing the human gene, suggesting that mice express cancers, oncogenic activation of ErbB2 is generally the trans-acting factors that potentially bind to this site thought to occur mostly due to the overexpression of (Shiraga et al., 2002). In the human gene, this ErbB2, and not activation. However, an alternative hypersensitive site contains a pyrimidine mirror repeat form of ErbB2 with increased oncogenic activity has (PMR) that is associated with the formation of H-DNA been detected in human breast cancer cell lines and some (Nelson et al., 1996). Regions of DNA that contain human breast tumors (Kwong and Hung, 1998; Siegel polypyrimidine or polypurine tracts involving a mirror et al., 1999). Interestingly, the deletions found in the repeat have the capability of forming a triplex structure human gene overlap the somatic mutations that occur in known as H-DNA (Mirkin et al., 1987). The formation the c-Neu transgene during tumor development in of H-DNA within promoter regions is thought to be MMTV-c-Neu mice. To determine if ErbB2-mediated involved in the regulation of gene expression (Larsen inhibition of MUC1 occurs in human breast cancers, we and Weintraub, 1982). In the mouse gene, the PMR is determined MUC1 expression in tumors that over- not conserved within the hypersensitive site; however, expressed HER2/Neu and tumors that were negative immediately 30 to this region the human and mouse for HER2/Neu. Examinations of our sample subset of genes are highly homologous, suggesting that this human breast cancers suggested that overexpression conservation in sequence may reflect an area important of ErbB2 does not result in a global downregulation of for the regulation of expression (Shiraga et al., 2002). MUC1. However, we did find areas with reciprocal Also within this region an AP-2 site is conserved in the expression of ErbB2 and MUC1. Previously, another human and mouse genes. AP-2 has both stimulatory and group studying breast cancer found that the patterns of repressive functions; interestingly, the expression of AP- ErbB2 and MUC1 expression were overlapping, but 2 has been shown to correlate with the overexpression of discordant (Akewanlop et al., 2001). Thus, because the

Oncogene Activated ErbB2 represses MUC1 expression MC Adriance and SJ Gendler 703 expression of MUC1 and ErbB2 is not completely and contain the native Muc1 promoter inserted up- concordant, the targeting of both MUC1 and ErbB2 in stream of luciferase in the pGL3-Basic vector (Promega) therapeutic protocols may result in affecting a wider (J Nicholas et al., manuscript submitted). range of breast tumors. The disparity between the findings in the mouse and Tumor generation system human breast tumors may be due to a lower frequency of ErbB2-activating mutations in human tumors. FVB mice transgenic for the unactivated rat c-Neu Additionally, unlike human tumors, c-Neu tumors do protooncogene under the control of the mouse mam- not contain a significant stromal component. Thus, the mary tumor virus promoter (line #202) (a generous gift differing environments may result in a different gene from Dr W Muller, McMaster University, London, expression pattern. In vitro, Scibetta et al. showed that Ontario, Canada) were utilized to examine Muc1 activating ras mutations had a negative effect on MUC1 expression during tumor development. The resulting transcriptional activation; however, it is important to offspring were screened for the presence of the c-Neu note that in human breast cancer, activating mutations transgene by PCR amplification as previously described in ras are a rare event (Rochlitz et al., 1989; Scibetta (Rowse et al., 1998b). Muc1 null mice were also screened et al., 2001). We found that in c-Neu transgenics, Muc1 by PCR as previously described (Spicer et al., 1995). was still expressed at high levels during pregnancy, Positive c-Neu males were bred back onto FVB mice suggesting that the machinery driving the increase in homozygous for either Muc1 null or wild-type alleles. Muc1 expression during pregnancy over-rides any ErbB2- Starting at 12 weeks mice were palpated once a week for mediated inhibition. These activating signals may be the presence of mammary tumors. Palpable tumors were turned on during progression to carcinoma in the human measured by calipers, and tumor weight was calculated breast and explain why overexpression of ErbB2 does not according to the formula:grams ¼ [(length) Â (width)2]/ result in an inhibition of MUC1 expression. 2, where length and width are measured in centimeters. At the end point of E2 g tumor weight, tumors were dissected and a portion was fixed in methacarn for immunohistochemical analysis. The lungs were dis- Experimental procedures sected, fixed in methacarn, and scored for metastasis Cell culture and DNA constructs using a dissecting microscope.

HEK 293 cells (purchased from ATCC) and HEK 293/ Northern analysis MUC1 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 5% fetal Total RNA was isolated from tissues dissected from calf serum, 1% glutamax, 1% pen/strep, 1 mM sodium MMTV-c-Neu tumors, lactating mammary gland, ET pyruvate (Life Technologies), and 1 Â nonessential pancreas tumor, and homogenized in Trizol (Life amino acids (Sigma). The medium for HEK 293/ Technologies). RNA (20 mg) was fractionated on 1.2% MUC1 cells and HEK 293/vector control cells was formaldehyde–agarose gel and transferred onto a nylon supplemented with zeocin 125 mg/ml (Life Technolo- membrane (Zeta probe, Biorad) in 10 Â SSC. The gies). HEK 293/MUC1 cells were generated by cotrans- membrane was baked at 801C for 1 h and was cross- fecting HEK 293 cells with a Sac II 10 kb clone of linked by UV irradiation. The membrane was stained human MUC1 (containing the endogenous promoter with methylene blue solution to detect 18S and 28S and all the regulatory sequences necessary to drive ribosomal RNA prior to hybridization. Blots were expression; Peat et al., 1992 and pcDNA3.1zeo vector hybridized overnight with a mouse Muc1 probe, (Invitrogen, Life Technologies) using lipofectamine pMuc10, a 30 cDNA probe to unique sequence (Spicer (Invitrogen, Life Technologies)). Selection was begun et al., 1991), a Neu cDNA probe (a gift from Dr D Lee, 72 h post-transfection by supplementing the media with University of North Carolina, Chapel Hill, NC, USA), zeocin 125 mg/ml. HEK 293 cells expressing MUC1 were and a keratin 18 cDNA probe (a gift from Dr R Oshima, selected by fluorescent-activated cell sorting on a B-D Burnham Institute, La Jolla, CA, USA). Vantage (Becton Dickson, San Jose, CA, USA) with B27.29, an antibody that recognizes the MUC1 tandem Immunoblot analysis repeat and used to ensure surface expression. Human ErbB2 cDNA in the pcDNA3 expression Mammary glands and tumors were homogenized in vector was a kind gift from Dr Yosef Yarden 20 mM HEPES, pH 8.0, 150 mM NaCl, 1% Triton X- (Weizmann Institute, Israel). Human constitutively 100, 2 mM EDTA, 2 mM sodium ortho vanadate, 50 mM active ErbB2 cDNA (containing the point mutation ammonium molybdate, 10 mM sodium fluoride, and a valine to glutamic acid at residue 659) was kindly complete inhibitor cocktail (Sigma Chemical Company). provided by Dr Nancy Hynes (Friedrich Miescher Cellular debris was removed by centrifugation at Institute, Switzerland) and cloned into the pcDNA3.1 15 000 g for 3 min. The protein concentration was expression vector at the HindIII site. Reporter plasmids determined by BCA assay (Pierce) and samples were containing the Muc1 promoter full-length and 50 frozen at À801C. Immunoblots were performed using deletion constructs were kindly provided by Dr Yaacov 100–200 mg of protein extract per sample, separated by Barak (Jackson Laboratory, Bar Harbor, ME, USA) electrophoresis on SDS–PAGE gels. Proteins were

Oncogene Activated ErbB2 represses MUC1 expression MC Adriance and SJ Gendler 704 transferred onto PVDF membranes (Immobilon). Blots gen, Life Technologies) were analysed 48 h post- were probed with anti-Neu (Sc-284; Santa Cruz Biotech- transfection for MUC1 expression. MUC1 levels in nology) and an Armenian hamster monoclonal antibody transfected cells were determined by immunoprecipita- to Muc1 cytoplasmic tail, CT2 (Schroeder et al., 2001). tion and immunoblotting as described above with the following exceptions. Immunoprecipitations were per- Immunohistochemistry formed from 250 mg of protein lysate, using Protein A/G agarose conjugate (Invitrogen, Life Technologies). Im- Mouse tissues were fixed in methacarn, paraffin munoprecipitations were performed with anti-MUC1 embedded (Mayo Clinic Scottsdale Histology Core), (CT2), followed by subsequent immunoblot analysis and 5 mM sections were cut. Slides were deparaffinized in with CT2. Immunoblot analysis for ErbB2 was xylene, rehydrated, blocked in normal goat serum, and performed with anti-Neu, Sc-284 (Santa Cruz Biotech- incubated with primary antibodies anti-Muc1 mono- nology). clonal CT2 and anti-Neu rabbit polyclonal (Santa Cruz Biotechnology) overnight at 41C. Slides were washed in PBS, incubated with HRP-conjugated secondary anti- Transient transfections and reporter assays 0 0 body, washed in PBS, and developed with 3 3 -diami- The effect of ErbB2 signaling on Muc1 transcriptional nobenzidine (Vector Laboratory) and counterstained activity was measured by cotransfection of empty vector with Meyers hematoxylin (Sigma Diagnostics, St Louis, (pcDNA3.1), wild-type ErbB2 (ErbB2 WT), or activated MO, USA). ErbB2 (ErbB2 VE) with reporter plasmid containing the Formalin-fixed, 5 mm paraffin-embedded human Muc1 promoter linked to firefly luciferase (pMuc1- breast cancer sections were stained for MUC1 and LUC) constructs into HEK 293 cells using lipofectamine ErbB2 as follows. ErbB2 staining was performed with the according to the manufacturer’s instructions (Invitro- HerceptTest (DAKO). For MUC1 staining, slides were gen, Life Technologies). Cotransfections included a deparaffinized in xylene, rehydrated, blocked in normal CMV-b galactosidase plasmid (Clontech, Palo Alto, goat serum, and incubated with primary antibody anti- CA, USA) as a normalization control for transfection MUC1 BC2, a mouse monoclonal antibody that efficiency. At 48 h following transfection, firefly lucifer- recognizes the epitope APDTR in the tandem repeat in ase activity was measured using the Luciferase Reporter MUC1; BC2 was a kind gift from Dr M McGuckin Assay System according to the manufacturer’s instruc- (University of Melbourne, Australia). Anti-MUC1 BC2 tions (Promega, Madison, WI, USA) on a lumicount was diluted 1 :20 000 in PBS with 1.5% normal goat microplate luminometer (Parkard, Meridan, CT, USA). serum (Xing et al., 1989). Sections were incubated with b-galactosidase activity was measured with the Gal- 1 anti-MUC1 antibody overnight at 4 C, followed by Screen Reporter Gene Assay System according to the 2 Â 1 min PBS washes, 1 h incubation in HRP-conjugated manufacturer’s instructions (Tropix, Bedford, MA, anti-mouse (Pierce) diluted 1 :500 in PBS with 1.5% USA) on a lumicount microplate luminometer (Par- normal goat serum, followed by 2 Â 1 min PBS washes. 0 0 kard). The total micromolar amount of DNA trans- Slides were developed with 3 3 -diaminobenzidine (Vec- fected per sample was standardized with the addition of tor Laboratory, Burlingame, CA, USA) and counter- an empty vector. The experiments were repeated at least stained with Meyers hematoxylin (Sigma) and followed three times. by 10 min of water washes. Sections were dehydrated and cover slipped in the Mayo Clinic Scottsdale Histology Core Facility. Approval for this study was obtained from Acknowledgements the Mayo Institutional Review Board. We thank Christopher J Sterner for excellent technical assistance in the breeding and palpation of the c-Neu/ Muc1 þ / þ and c-Neu/Muc1À/À mice. We also thank Dr Yosef Pathological examination Yarden and Dr Nancy Hynes for ErbB2 constructs, Dr Pathological examination of the breast tumor sections was Yaacov Barak for Muc1 deletion constructs, Dr William performed by Dr T Lidner at Mayo Clinic Scottsdale. Muller for the c-Neu transgenic mice, Dr David Lee for the c- Neu probe, Dr Robert Oshima for the keratin 18 cDNA probe, Sections were graded according to a Modified Bloom– Jim Tarara and Linda Murphy for technical support, and Dr Richardson criteria (Bloom and Richardson, 1957). Patrick Roche for providing human breast tumor sections. We also thank Suresh Savarirayan, and animal care attendants for Analysis of MUC1 expression animal care, Marvin Ruona for computer graphics, and Carol Williams for manuscript preparation and submission. This HEK 293/MUC1 cells transfected with an empty vector work was supported by NIH CA064389 (SJG), NIH pre- (pcDNA3.1) or wild-type ErbB2 (wild-type), or acti- doctoral fellowship CA090205 (MCA), and the Mayo Foun- vated ErbB2 (ErbB2 VE) using lipofectamine (Invitro- dation.

References

Akewanlop C, Watanabe M, Singh B, Walker M, Kufe DW Croce MV, Isla-Larrain MT, Rua CE, Rabassa ME, Gendler and Hayes DF. (2001). Cancer Res., 61, 4061–4065. SJ and Segal-Eiras A. (2003). J. Histochem. Cytochem., 51, Bloom H and Richardson W. (1957). Br. J. Cancer, 11, 359–377. 781–788.

Oncogene Activated ErbB2 represses MUC1 expression MC Adriance and SJ Gendler 705 Engelman JA, Lee RJ, Karnezis A, Bearss DJ, Webster M, Rowse GJ, Tempero RM, VanLith ML, Hollingsworth MA Siegel P, Muller WJ, Windle JJ, Pestell RG and Lisanti MP. and Gendler SJ. (1998a). Cancer Res., 58, 315–321. (1998). J. Biol. Chem., 273, 20448–20455. Schroeder JA, Adriance MC, Thompson MC, Camenisch TD Guaita S, Puig I, Franci C, Garrido M, Dominguez D, Batlle and Gendler SJ. (2003). Oncogene, 22, 1324–1332. E, Sancho E, Dedhar S, De Herreros AG and Baulida J. Schroeder JA, Thompson MC, Gardner MM and Gendler SJ. (2002). J. Biol. Chem., 277, 39209–39216. (2001). J. Biol. Chem., 276, 13057–13064. Guy CT, Webster MA, Schaller M, Parsons TJ, Cardiff RD Scibetta AG, Albanese I, Morris J, Cooper L, Downward J, and Muller WJ. (1992). Proc. Natl. Acad. Sci. USA, 89, Rowe P and Taylor-Papadimitriou J. (2001). DNA Cell 10578–10582. Biol., 20, 265–274. Hilkens J, Buijs F, Hilgers J, Hageman P, Calafat J, Shiraga T, Smith D, Nuthall HN, Hollingsworth MA and Sonnenberg A and van der Valk M. (1984). Int. J. Cancer, Harris A. (2002). Mol. Med., 8, 33–41. 34, 197–206. Siegel PM, Dankort DL, Hardy WR and Muller WJ. (1994). Khoury H, Dankort DL, Sadekova S, Naujokas MA, Muller Mol. Cell. Biol., 14, 7068–7077. WJ and Park M. (2001). Oncogene, 20, 788–799. Siegel PM and Muller WJ. (1996). Proc. Natl. Acad. Sci. USA, Kwong KY and Hung MC. (1998). Mol. Carcinog., 23, 93, 8878–8883. 62–68. Siegel PM, Ryan ED, Cardiff RD and Muller WJ. (1999). Larsen A and Weintraub H. (1982). Cell, 29, 609–622. EMBO J., 18, 2149–2164. McGuckin MA, Walsh MD, Hohn BG, Ward BG and Wright Slamon DJ, Clark GM, Wong SG, Levin WJ, Ullrich A and RG. (1995). Hum. Pathol., 26, 432–439. McGuire WL. (1987). Science, 235, 177–182. Mirkin SM, Lyamichev VI, Drushlyak KN, Dobrynin VN, Spicer AP, Parry G, Patton S and Gendler SJ. (1991). J. Biol. Filippov SA and Frank-Kamenetskii MD. (1987). Nature, Chem., 266, 15099–15109. 330, 495–497. Spicer AP, Rowse GJ, Lidner TK and Gendler SJ. (1995). Nelson KL, Becker NA, Pahwa GS, Hollingsworth MA and J. Biol. Chem., 270, 30093–30101. Maher III LJ. (1996). J. Biol. Chem., 271, 18061–18067. Turner BC, Zhang J, Gumbs AA, Maher MG, Kaplan L, Peat N, Gendler SJ, Lalani N, Duhig T and Carter D, Glazer PM, Hurst HC, Haffty BG and Williams Taylor-Papadimitriou J. (1992). Cancer Res., 52, T. (1998). Cancer Res., 58, 5466–5472. 1954–1960. Witton CJ, Reeves JR, Going JJ, Cooke TG and Bartlett JM. Revillion F, Bonneterre J and Peyrat JP. (1998). Eur. J. (2003). J. Pathol., 200, 290–297. Cancer, 34, 791–808. Xing PX, Tjandra JJ, Stacker SA, Teh JG, Thompson CH, Rochlitz CF, Scott GK, Dodson JM, Liu E, Dollbaum C, McLaughlin PJ and McKenzie IF. (1989). Immunol. Cell Smith HS and Benz CC. (1989). Cancer Res., 49, 357–360. Biol., 67 (Part 3), 183–195. Rowse GJ, Ritland SR and Gendler SJ. (1998b). Cancer Res., Zotter S, Hageman PC, Lossnitzer A, Mooi WJ and Hilgers J. 58, 2675–2679. (1988). Cancer Rev., 11–12, 55–101.

Oncogene