Ectopic Runx2 Expression in Mammary Epithelial Cells Disrupts Formation of Normal Acini Structure: Implications for Breast Cancer Progression

The transcription factor Runx2 is highly expressed in breast cancer cells compared with mammary epithelial cells and contributes to metastasis. Here we directly show that Runx2 expression promotes a tumor cell phenotype of mammary acini in three-dimensional culture. Human mammary epithelial cells (MCF-10A) form polarized, growth-arrested, acinilike structures with glandular architecture. The ectopic expression of Runx2 disrupts acini formation, and electron microscopic ultrastructural analysis revealed the absence of lumens. Characterization of the disrupted acini structures showed increased cell proliferation (Ki-67 positive cells), decreased apoptosis (Bcl-2 induction), and loss of basement membrane formation (absence of B4 integrin expression). In complementary experiments, inhibition of Runx2 function in metastatic MDA-MB-231 breast cancer cells by stable expression of either short hairpin RNA-Runx2 or a mutant Runx2 deficient in subnuclear targeting resulted in reversion of acini to more normal structures and reduced tumor growth in vivo. These novel findings provide direct mechanistic evidence for the biological activity of Runx2, dependent on its subnuclear localization, in promoting early events of breast cancer progression and suggest a molecular therapeutic target. [Cancer Res 2009;69(17):6807–14]


Introduction
Loss of cell polarization and luminal filling of mammary glands are crucial structural alterations in breast cancer (1,2). The expression of cancer-related genes in breast epithelial cells can lead to disorganization of mammary gland structure in in vitro models and cause a pathology similar to epithelial tumors, indicating a critical requirement for the fidelity of tissue and cellular organization of acini (2)(3)(4). Molecular events leading to disruption of the glandular structure include expression of antiapoptotic genes, reduced expression of proapoptotic factors, inactivation of pRB, and ectopic expression of ErbB2 and tissue inhibitor of metalloproteinase (1,4,5). However, the role of transcription factors in regulating the expression of genes implicated in the formation of the luminal space and in cell polarization during mammary epithelial gland formation is not well defined.
The Runx transcription factors (Runx1, Runx2, and Runx3) are essential for organogenesis, and mutations in these genes have been linked to several types of cancer (6,7). Runx1 and Runx3 mutations promote leukemia and gastric cancers, respectively (8,9). Runx2 is a key factor for bone formation (10,11) and deregulation of Runx2 is associated with osteosarcoma (12,13). In normal mammary epithelial cells, Runx2 is expressed at low levels, but it is expressed at high levels in metastatic cancer cells (14)(15)(16) and promotes bone metastatic properties of breast and prostate cancer cells (17)(18)(19)(20)(21). The consequences of aberrant Runx2 expression in breast cancer cells have been well studied (20). Runx2 directly increases expression of metastatic genes [IHH, matrix metalloproteinase (MMP)-9, MMP-13, transforming growth factor-h receptor, vascular endothelial growth factor, etc.] and increases activity of the osteolytic cycle in metastatic bone disease (21). However, the consequences of induced expression of Runx2 in normal breast epithelial cells in contributing to a cancer cell phenotype are unknown and constitute the focus of this study. Understanding the relationship between deregulation of transcription factors and associated molecular and architectural abnormalities that result in histologic phenotypes observed in tumors is critical for developing novel therapeutic targets for early stages of cancer progression.
Normal MCF-10A mammary epithelial cells cultured on a basement membrane form polarized, growth-arrested, acini-like spheroids, recapitulating several aspects of glandular architecture in vivo (1,22). In these studies, we took advantage of the threedimensional culture model to examine the contribution of Runx2 to the tumorigenic phenotype of mammary acini. Ectopic expression of Runx2 in MCF-10A cells results in disrupted acini structures that resemble cancer phenotype. Importantly, complementary experiments in metastatic breast cancer MDA-MB-231 cells, which do not form an organized structure in basement membrane, show that knockdown of endogenous Runx2 or expression of mutant Runx2 proteins results in reversion to more normal acini-like structures. Thus, our results directly show that induced expression of Runx2 in mammary epithelial cells to levels found in metastatic breast cancer cell lines contributes to early events in development of tumor cell properties and may play a critical role in the loss of cellular organization of mammary acini tissue. three-dimension in a-MEM containing 10% fetal bovine serum (Invitrogen). The construction and transduction procedure for adenovirus expressing Runx2 and control are reported previously (23). Monolayer and threedimensional cultures of MCF-10A cells were set up according to Debnath and colleagues (1,24); briefly, single-cell suspensions in 2% Matrigel (BD Biosciences) were overlaid on a thin layer of Matrigel and cultures were fed every second day with 2% Matrigel in DMEM/F12. Generation of stable MDA-MB-231-short hairpin RNA (shRNA) control or Runx2 cells by lentiviral gene delivery is reported previously (21).
Western blot analysis. Runx2 protein in normal and breast cancer cells was detected by Western blot analysis as described previously (18). Wholecell lysates or nuclear lysates were mixed with direct lysis buffer and separated in a 10% SDS-PAGE. Proteins were transferred to polyvinylidene difluoride membrane and incubated with mouse monoclonal Runx2 antibody (R&D Systems) followed by incubation with horseradish peroxidase-conjugated secondary antibodies (Santa Cruz Biotechnology). Immunoreactive proteins were detected using an enhanced chemiluminescence kit (Pierce).
Electrophoretic mobility shift analysis. Nuclear extracts were prepared from normal mammary epithelial MCF-10A or breast cancer MDA-MB-231 cells as described previously (18). Complementary oligonucleotides representing the consensus Runx binding element were synthesized: 5 ¶-ACCCTGAGTTCTGTGGTTGTTTCCTGTGGGTCT-3 ¶. The plus strand (10 pmol) was labeled with [g-32 P]ATP for 1 h at 37jC with T4 polynucleotide kinase (New England Biolabs). For immunoshift analysis, 200 ng Runx2 antibody was incubated with nuclear extract at 22jC for 30 min before the addition of probe. The samples were electrophoresed at 200 V for 3 h at 4jC. Gels were dried and subjected to autoradiography at 70jC.
Electron microscopy. For electron microscopy of MCF-10A acini, tissue structures were harvested by suspending the cultures in ice-cold serum-free DMEM/F-12 and centrifuging at 600 Â g for 5 min followed by three washes in cold DMEM/F-12 to remove the Matrigel. Acini were fixed in 2.5% glutaraldehyde in 0.1 mol/L sodium cacodylate (pH 7.4) at 4jC for 1 to 2 h and then washed in 0.1 mol/L sodium cacodylate. Samples were postfixed in 1% osmium in 0.1 mol/L sodium cacodylate at 4jC for 30 min, washed again, and dehydrated in graded ethanol. Samples were embedded in Epon and the transitional solvent was propylene oxide. Sections were stained with 1.4% uranyl acetate in 40% ethanol and then with lead citrate. Sections were imaged with a Philips CM10. Negatives were scanned and processed digitally. Size measurements were made from the digitized images.
Immunofluorescence studies. For confocal microscopy, cells were cultured in reduced growth factor Matrigel on four-well chamber slides (Thermo Scientific Nalgene and Nunc) before fixation and staining as described (24) and then imaged with a Leica SP1 confocal microscope. DNA was stained with DRAQ5 after the second antibody wash.
Mammary fat pad injections and bioluminescence imaging. Anesthetized mice were inoculated with 1 Â 10 6 MDA-MB-231 cells with Matrigel under mammary fat pads. Bioluminescence images were acquired by using the IVIS Imaging System (Xenogen) 2 to 15 min after injection 150 mg/kg D-luciferin (Gold BioTech) in PBS.

Results
Ectopic Runx2 expression disrupts normal acini structure.
To investigate the consequences of Runx2 expression in normal epithelial cells, we first characterized Runx2 levels in the MCF-10A cells, which form normal acini structures in three-dimensional cultures. Analysis of total RNA and nuclear extracts from MCF-10A cells shows the presence of low levels of Runx2 compared with metastatic MDA-MB-231 cells by real-time PCR analysis and Western blotting (Fig. 1A, left and right). The presence of Runx2 DNA-binding activity in MCF-10A cells was further confirmed by antibody supershift studies (see Supplementary Fig. S1). Examination of subnuclear organization using in situ immunofluorescence microscopy shows that Runx2 is localized in nuclear foci in MCF-10A cells albeit at lower levels compared with MDA-MB-231 cells (Fig. 1B). The observed low levels of Runx2 in MCF-10A cells make this cell line a suitable model to study the effect of Runx2 expression on mammary acini formation and the potential induction of a phenotype related to cancer progression.
MCF-10A cells were transduced with adenovirus expressing either Runx2 or h-galactosidase as control. Increased levels of Runx2 were confirmed by Western blot analysis ( Supplementary  Fig. S2) and cells were grown in three-dimensional culture for 20 days after transduction. We find that ectopic expression of Runx2 causes a significant disruption of normal acini structure as examined by light microscopy at days 2 and 20 (Fig. 1C). Runx2treated cells form disorganized cell masses compared with the organized acini structures formed by control cells (infected with AdLacZ). Runx2-expressing cells exhibited structures with a 2-fold increased diameter compared with controls (AdRunx2 176.6 F 29.5 Amol/L versus AdLacZ control 91.43 F 16.32 Amol/L). We examined the ultrastructure of acini by electron microscopy, which revealed that the disorganized structures from Runx2-expressing cells lack a hollow lumen and apoptotic cells. Control cells formed polarized acini with hollow lumens (Fig. 1D). These results show that ectopic expression of Runx2 in normal mammary epithelial cells results in disorganization of mammary acini architecture.
Runx2 expression disrupts acinus formation by increasing proliferation and decreasing basement membrane formation. Formation of acini structure is a result of stringent control of cell proliferation, polarization, apoptosis, and formation of basement membrane (1). Our findings (Fig. 1) suggest that ectopic Runx2 expression results in disruption of normal acini structure in threedimensional culture of MCF-10A cells due to either continuous cell proliferation or inability of the MCF-10A cells to undergo apoptosis. We find a significant increase in proliferation of cells transduced with Runx2 adenovirus compared with control cells at day 20 of differentiation as shown by Ki-67 staining (Fig. 2A). The low levels of Ki-67 staining in control cells (AdLacZ) indicate that MCF-10A cells are primarily nonproliferative. These results suggest that continued proliferation of Runx2-expressing cells may contribute to disorganization of acini structure.
Acini formation is a highly regulated process in which cells communicate with each other to form a polarized structure. These structures also show the presence of basement membrane around well-formed acini. We examined the MCF-10A three-dimensional cultures treated with Runx2 or control adenovirus for h 4 integrin as a marker for basement membrane. MCF-10A cells infected with control virus exhibit well-formed basement membrane as shown by h 4 integrin staining after day 20 of culture (Fig. 2B), whereas ectopic expression of Runx2 prevents basement membrane formation in culture. These findings show that Runx2 overexpression in MCF-10A cells causes loss of polarization as these cells continue to proliferate in disorganized acini. Moreover, Runx2-expressing cellular aggregates accumulate characteristics similar to architectural features of breast cancer.
Runx2 activates cancer-related Bcl-2 and IL-8 genes in MCF-10A cells. The organizational changes in MCF-10A acini with Runx2 ectopic expression might be due to alteration of Runx2 target gene expression. To gain further insight into molecular alterations leading to disorganization of acini architecture due to ectopic expression of Runx2, we performed a focused cDNA expression profile of cancer-related genes. Total RNA from MCF-10A cells transduced with either control LacZ or Runx2 adenovirus was labeled and hybridized with membranes containing probes for 96 cancer-related genes ( Supplementary  Fig. S3). Our results show a significant increase in expression of Bcl-2 and IL-8 genes in cells treated with Runx2 adenovirus compared with controls. We validated the increased expression of these genes by quantitative reverse transcription-PCR analysis (Fig. 2C, left). We further confirmed the in vivo recruitment of Runx2 on Bcl-2 and IL-8 gene promoters by chromatin immunoprecipitation assay in MCF-10A cells (Fig. 2C, right). These studies identified Bcl-2 and IL-8 as novel Runx2-responsive genes. Increased Bcl-2 expression in response to Runx2 overexpression might contribute to luminal filling by reducing apoptosis. These results suggest that aberrant Runx2 expression in mammary epithelial cells leads to activation of downstream cancer-related genes and disrupts the balance between cell growth and apoptosis.
Loss of functions associated with the Runx2 COOH terminus prevents cell growth and tumor cell-like phenotype. To understand the molecular mechanism of Runx2-mediated regulation of cellular processes required for maintaining highly organized epithelial glandular structures, we used mutants of Runx2. A point mutant (RY: Y428A and R398A) and one deletion mutant (DC: Bioluminescence images were acquired by using the IVIS Imaging System (Xenogen) 2 to 10 min after 12 d of cell injections. Images were set at the indicated pseudocolor scale to show relative bioluminescent. D, quantitation of luciferase activity shows inhibition of tumor growth under mammary fat pad in shRNA-Runx2 animals compared with control group. Analysis was done with Living Image software (Xenogen) by measuring photon flux (measured in photons/s/cm 2 /steradian) using a region of interest drawn around the bioluminescence signal to be measured (n = 6). Data were normalized to the signal obtained right after xenografting (day 0). SD. *, P < 0.005; **, P < 0.001, compared with empty vector controls (Student's t test).
Runx2 1-361) of Runx2 protein are known to abrogate subnuclear targeting to the nuclear matrix and to disrupt organization of Runx2 regulatory complexes with coregulatory proteins involved in the transforming growth factor-h and c-Src signaling pathways (7,25,26). Both these pathways are linked to tumor progression (27). Furthermore, the mutants failed to promote in vivo bone formation (11) and prevented tumor growth when expressed in MDA-MB-231 cells (15,17).
MDA-MB-231 cells stably expressing the Runx2 mutants were generated by lentiviral mediated gene delivery, and protein levels were confirmed by Western blot analysis (Fig. 3A). We examined the functional activity of these Runx2 mutants on endogenous levels of cancer-related markers and found that both mutations reduce activation of Runx2 target genes (MMP-2, MMP-9, vascular endothelial growth factor, and osteopontin) compared with wildtype Runx2 (Supplementary Fig. S4). The morphology of threedimensional cultures of MDA-MB-231 cells expressing the RY mutant Runx2 protein showed significantly smaller acini compared with empty vector control, and this difference was more striking on days 2 and 4 (Fig. 3B). Cells expressing wild-type Runx2 form large and disorganized structures compared with empty vector control cells, whereas functionally deficient DC-Runx2-expressing and Runx2-RY-expressing cells show reversion to acini structures compared with the empty vector control. Taken together, our results show that Runx2 interactions with the nuclear matrix and its coregulatory proteins via the COOH terminus are critical for mediating disorganization of mammary acini.
Depletion of Runx2 in metastatic cancer cells partially restores organized acini structure and inhibits tumor growth in vivo. The MDA-MB-231 metastatic breast cells do not form acini-like structures but instead make highly branched masses when grown in three-dimensional Matrigel culture (Fig. 4A, top). We tested the hypothesis that abnormal levels of expressed Runx2 (see Fig. 1) are contributing to the unregulated cell growth that prevents acini formation in the three-dimensional culture model. The experimental design examined whether depleting Runx2 in MDA-MB-231 cells would rescue normal acini structure. We generated MDA-MB-231 cells stably expressing a shRNA for Runx2 and achieved knockdown of >80% of endogenous Runx2 as shown by Western blot analysis ( Supplementary Fig. S5). The threedimensional culture of shRNA-Runx2-expressing cells for 12 days showed significant reversion in that the structures that formed more closely resembled MCF-10A acini than the structures formed by the parental cells or empty vector controls (Fig. 4A, bottom). The morphology and diameter of acini formed by shRNA-Runx2 cells (110 Am at day 12 of culture) were comparable with normal mammary epithelial cell acini. Microscopic characterization of the reverted structures revealed reduced proliferation compared with controls, as examined by Ki-67 immunostaining (Fig. 4B, top), and significantly higher staining of h 4 integrin around these structures, indicating the formation of basement membrane (Fig. 4B, bottom).
We also performed mammary fat pad injections of firefly luciferase-labeled MDA-MB-231 cells stably expressing either shRNA-Runx2 or control shRNA in the severe combined immunodeficient mouse model. Tumor growth was visualized by luciferase bioluminescence imaging 12 days after cell injections (Fig. 4C). Animals injected (n = 6) with MDA-MB-231 cells expressing shRNA-Runx2 showed a significant reduction in tumor growth compared with a control group (expressing empty vector) as shown by quantitation of luciferase activity (Fig. 4D). Examining tumor growth up to 4 weeks revealed a 41% decrease in tumor volume in shRNA-Runx2 animals (1.8 cm 3 ; n = 6) compared with controls (3.1 cm 3 ; n = 6). Thus, our results from in vitro and in vivo studies show that depleting Runx2 levels in metastatic breast cancer cells restores mammary gland organization by regulating cell proliferation and polarization and reduces tumor growth in mammary fat pads.

Discussion
Our results provide evidence that elevated Runx2 expression in the three-dimensional culture model of normal mammary epithelial cells activates several key cancer-related genes and disorganizes acinar architecture to resemble a tumorigenic phenotype. A striking feature of the loss of glandular architecture and formation of proliferating cell aggregates by Runx2 overexpression is the absence of h 4 integrin, an essential component for organization of glandular epithelium (28). We show that depletion of endogenous Runx2 in highly aggressive metastatic breast cancer cells limits tumor formation in the mammary gland in vivo and reverts cancer cell aggregates into more normal acini-like structures in vitro. Combined with our previous reports (17,18,21) and based on present findings, we propose that Runx2 not only promotes metastatic properties of cancer cells but also can initiate tumorigenic properties in normal mammary epithelial cells.
We identified mechanisms by which ectopic Runx2 expression contributes to the altered phenotype of MCF-10A cells, including increased proliferation and decreased apoptosis. Runx factors have been linked to apoptosis (29,30). Our data show that Runx2 induces Bcl-2 expression in MCF-10A cells, which likely contributes to luminal filling via antiapoptotic events and loss of glandular structure. Also, coexpression of Bcl-2 with cyclin D1 in MCF-10A cells is known to result in luminal filling (31). Runx2 stimulates expression of the cell cycle regulator cyclin D1, an essential factor for proliferation and an oncogene identified in metastatic breast cancer cells (21). Apoptosis may also be decreased due to the synergy between Runx2 and c-Myc, which causes a low rate of apoptosis and proliferative advantage of lymphoma cells (32). The ineffectiveness of the subnuclear targeting deficient Runx2 mutant in causing acini disruption may likely be due to the loss of interacting coregulatory factors in organized nuclear foci with Runx2. Runx2 facilitates transforming growth factor-h-mediated growth effects on breast cancer cells by interaction with Smads (20). Our results are consistent with other findings that Runx2 functions as an oncogene when abnormally expressed in tumor cells (20,32). This property of Runx2 is in contrast to its tumor suppressor function in normal cells (33).
Our results show that expression of Runx2 causes loss of basement membrane and cell polarity leading to disorganized acini. The insulin-like growth factor-I signaling pathway has been implicated in disorganization of acinar morphogenesis in threedimensional cultures (1), and this pathway stimulates Runx2 activity (34). Activation of the insulin-like growth factor-I receptor increases proliferation, survival signaling, and loss of apicobasal polarity in mammary epithelial cells and promotes early breast cancer lesions (35). In breast cancer cells, inhibition of phosphoinositide 3-kinase or mitogen-activated protein kinase results in nearly complete reversion to acini structure (36). We show that Runx2 depletion in metastatic breast cancer cells restores more normal-appearing, acini-like structures in three-dimensional cultures. Studies have shown that loss of the Runx2 COOH terminus leads to elimination of domains that contain phosphoinositide 3-kinase/mitogen-activated protein kinase phosphorylation sites (34,37). Interestingly, Runx2 synergizes with phosphoinositide 3kinase/Akt signaling to enhance cell migration (38). Thus, Runx2 may be responding in vivo to insulin-like growth factor and/or phosphoinositide 3-kinase signaling pathways that lead to mammary gland disorganization during early tumor progression.
Profiling cDNAs in breast cancer cells has revealed a set of genes highly expressed in response to mitogenic signals that enhance tumorigenic properties (39). Runx2 activates MMP-2, MMP-9, and osteopontin involved in migration and invasion of MCF-10A cells through p38 mitogen-activated protein kinase signaling in response to transforming growth factor-h (40,41). Recent studies identified Runx2 as an essential component of transforming growth factor-hmediated activation of the vicious cycle operative in the bone microenvironment during breast cancer metastasis (18,21). Therefore, we propose that increased expression of Runx2 and its target genes related to tumor growth and metastasis in the MCF-10A cells are promoting the tumor phenotype in the threedimensional culture model. This conclusion is supported by the suppression of the tumor cell phenotype of the MDA-MB-231 cell in Matrigel by Runx2 shRNA and Runx2 subnuclear targeting mutants. In this study, we further identified Bcl-2 and IL-8 as novel Runx2 target genes. Bcl-2 was shown to be induced in hematopoietic cells by the leukemic factor AML1(Runx1)-ETO (42). IL-8 enhances cell proliferation, survival, and MMP expression in CXCR1-and CXCR2-expressing endothelial cells and regulates angiogenesis (43). Our chromatin immunoprecipitation studies show that Runx2 associates with promoters of these genes in regions with consensus sequences for Runx regulatory element. Therefore, these studies indicate that depletion of Runx2 and the resulting down-regulation of Runx2 target genes can promote differentiation of breast cancer cells into more normal epithelial cells as we have observed in the three-dimensional culture model.
Taken together, our studies have established that the induction of Runx2 levels in normal mammary epithelial cells enhances tumorigenic properties and promotes disorganization of acini-like structures formed by these cells. We propose that Runx2 activates cancer-related genes in response to deregulated cell signaling pathways during early stages of breast cancer. As such, Runx2 is a novel therapeutic target that takes advantage of unique subnuclear targeting properties for a broad spectrum of biological activities in tumor cells.

Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.