Homeobox D10 Induces Phenotypic Reversion of Breast Tumor Cells in a Three-Dimensional Culture Model

Research Article

Meritxell Carrio,1 Gemma Arderiu,1 Connie Myers,2 and Nancy J. Boudreau1

1Department of Surgery, University of California San Francisco, San Francisco, California and 2Department of Cell and Molecular Biology, Life Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California

Abstract considering the role for Hox genes in regulating breast tissue Homeobox (Hox) genes are master regulatory genes that direct morphogenesis and cell-extracellular matrix interactions in normal organogenesis and maintain differentiated tissue function. We tissues, it is likely that loss of Hox gene expression might underlie previously reported that HoxD10 helps to maintain a changes in cell-extracellular matrix interactions, which give rise to quiescent, differentiated phenotype in endothelial cells by the aberrant morphology and behavior of breast tumors. Indeed, suppressing expression of genes involved in remodeling the there are numerous examples of aberrant expression of Hox genes in extracellular matrix and cell migration. Here we investigated mammary tumors. Whereas HoxC6 is present in normal breast whether HoxD10 could also promote or maintain a differen- tissue, it is absent in mammary neoplasia (7, 13). HoxA5 expression is tiated phenotype in epithelial cells. We observed that HoxD10 also lost in breast cancer. Raman et al. (17) reported that methylation of the HoxA5 promoter correlates with loss of p53 expression in expression is progressively reduced in epithelial cells as malignancy increases in both breast and endometrial tumors. breast tumors, and Cheng et al. (20) showed that reexpression of Retroviral gene transfer to restore expression of HoxD10 in the HoxA5 can induce apoptosis through an apoptotic mechanism malignant breast tumor cells MDA-MB-231 significantly mediated by caspases 2 and 8. On the other hand, HoxA1 is detected impaired migration, and when these cells were cultured in a in mammary tumors, but not in the normal gland or in precancerous three-dimensional laminin-rich basement membrane outgrowths (7, 13, 21). It has been shown that forced expression of (3DlrBM) model, they formed polarized, acinar structures. HoxA1 in human mammary carcinoma cells results in oncogenic This phenotypic reversion was accompanied by decreased A3 transformation with in vivo tumor formation (22). To date, however, integrin expression and reduced proliferation. Importantly, no studies have shown that restoring normal Hox gene expression expression of HoxD10 in the MDA-MB-231 cells inhibited their can influence the course of breast tumorigenesis in vivo. Work from our laboratory showed that sustained expression of ability to form tumors in mouse xenografts. Taken together, our results suggest that HoxD10 has tumor-suppressive HoxD10 suppresses angiogenesis in vivo, suggesting that HoxD10 contributes to maintenance of a quiescent, differentiated phenotype functions for mammary epithelial cells. (Cancer Res 2005; in endothelial cells (3). Specifically, HoxD10 suppresses expression of 65(16): 7177-85) genes that directly affect remodeling of the extracellular matrix and cell migration during angiogenesis such as a3 integrin, matrix Introduction metalloproteinase 14 (MMP-14), and urokinase-type plasminogen Type I class Homeobox (Hox) genes encode master transcription activator receptor (uPAR). Interestingly, these genes have been factors that play an important role in organogenesis and shown to promote invasion, migration, and tumor progression in a morphogenesis during development (1). In addition to their role variety of cells, including breast tumors (23–27), raising the possibility in embryogenesis, Hox genes are expressed in adult cells where that HoxD10 functions as a general inhibitor of cell invasion. they regulate expression of genes involved in cell proliferation as In this study, we investigated whether HoxD10 expression is well as cell-cell and cell-extracellular matrix interactions (2–9). It is reduced in tumorigenic epithelial cells and whether restoring its now well established that several Hox genes display an altered expression can redirect normal tissue morphology and help maintain pattern of expression in certain malignancies as compared with a quiescent differentiated state in epithelial cells. We show that normal tissue including leukemia and solid tumors such as breast, HoxD10 is lost during tumor progression both in breast and endometrium, brain, colon, prostate, lung, and kidney (10–15). endometrial tumors. In addition, HoxD10 is capable of reverting the Although it was originally believed that up-regulation of Hox genes highly invasive breast tumor cell line MDA-MB-231 to a nonmalignant promoted oncogenesis, it is now becoming evident that loss of state by restoring normal cell-cell and cell-extracellular matrix various Hox genes is also associated with tumorigenesis (16, 17). interactions when culturing the cells in the three-dimensional Loss of normal breast tissue architecture and polarity, a pathologic laminin-rich basement membrane (3DlrBM) assay. Moreover, hallmark of breast cancer, is due to altered interactions between HoxD10 suppresses tumor growth in vivo in a mouse xenograft model. cancer cells and their surrounding extracellular matrix (18). In the normal mammary gland, Hox genes play an important role in Materials and Methods dictating cell specificity, proliferation, and differentiation during mamopoiesis, both in development and in pregnancy (19). Thus, Cell culture. The human breast epithelial cell line MDA-MB-231 and the human cervical epithelial (HeLa) cancer cell lines were obtained from the American Type Culture Collection (ATCC, Manassas, VA). Both cell lines were maintained in DMEM/F12 (Life Technologies, Rockville, MD) with 5% Requests for reprints: Nancy Boudreau, Department of Surgery, University of fetal bovine serum and 5 Ag/mL gentamicin sulfate (Life Technologies). California San Francisco, Box 1302, San Francisco, CA 94143. Phone: 415-206-6951; Three-dimensional cultures were prepared using a modified procedure E-mail: [email protected]. I2005 American Association for Cancer Research. described by Weaver et al. (28). Briefly, cells were grown to confluence as doi:10.1158/0008-5472.CAN-04-1717 monolayers, trypsinized, and plated (3 Â 104 cells in 24-well plates) on top www.aacrjournals.org 7177 Cancer Res 2005; 65: (16). August 15, 2005

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2005 American Association for Cancer Research. Cancer Research of a 300 AL of polymerized Matrigel (10.75 mg/mL; BD Biosciences, Bedford, After 3 hours of incubation at 37jC in the dark, MTT was removed and MA) in 0.5 mL DMEM/F12 media without serum and with 250 ng/mL 500 AL of solubilization solution (0.04 N HCl in isopropanol and 3% w/v insulin (Sigma, St. Louis, MO). After 30 minutes of incubation at 37jC, SDS) were added. After 1 hour at 37jC in the dark, 150 AL of each sample 0.5 mL media containing 10% Matrigel was added on top of the cells. were transferred into 96-well plates and the absorbance was determined at Cultures were analyzed after 4 days of cultivation. The morphology of the 600 nm using a Wallac Victor2 1420 Multilabel Counter (Perkin-Elmer three-dimensional cultures was assessed by visual examination of the Life Sciences, Boston, MA). In addition, we also did analysis of BrdUrd degree of colony organization with phase-contrast microscopy. incorporation as previously described (29). Retroviral vectors and transduction. The human 1,100 bp HoxD10 Transwell migration assay. Cell migration assays were done in modified cDNA (GenBank accession no. X59373) or a HoxD10 DNA binding mutant Boyden chambers. Briefly, 6.5 nm transwell chambers with 8 Am pores which contains a N316T [aspargine (N)-316 to threonine (T)] point mutation (Corning, Acton, MA) were coated with 10 Ag/mL of type I collagen (Collagen in the second helix of the homeodomain, abolishing DNA binding (SwissProt Biomaterials, Palo Alto, CA) for 2 hours at 37jC. Cells were serum starved for accession no. P28358, a gift from Dr. C. Largman, University of California San 16 hours and 5 Â 104 cells/well were plated to the upper chamber in 300 ALof Francisco, San Francisco, CA), was cloned into the EcoRI site of the pLXSN fibroblast basal medium (Clonetics, San Diego, CA) containing 0.5% bovine retroviral vector (Clontech, Palo Alto, CA). The resultant vectors were serum albumin (BSA). The lower chamber was filled with 500 AL of fibroblast confirmed by DNA sequencing at the Biomolecular Resource Center, basal medium. After 4 hours at 37jC, nonmigratory cells in the upper University of California San Francisco, and designated pLD10 and pLD10 chamber were removed with a cotton swab and cells that migrated to the N-T, respectively. Retroviral vectors were transfected by the calcium/ bottom of the membrane were fixed and stained with Diff-Quick (VWR phosphate-DNA precipitation method into the amphotropic packaging cell Scientific Products, West Chester, PA). The total number of migrated cells was line Phoenix Ampho (ATCC), and 48 hours after transfection, viral supernatant determined by counting five fields in each well per experimental condition was collected, passed through 0.45 Am filters, and used for transduction. using a phase-contrast microscope (magnification, Â20). MDA-MB-231 and HeLa cells were transduced with control plasmid When indicated, cells were preincubated for 2 hours at 37jC with (pLXSN), pLD10, or pLD10 N-T using 2 mL of virus-containing media in the 5 Ag/mL of control immunoglobulin G (IgG) or function-blocking presence of 8 Ag/mL of Polybrene (Sigma) for 16 hours at 37jC. Twenty-four antibodies against mouse anti-human integrin a3 (Chemicon) or mono- hours later, cells were selected in 800 Ag/mL G418 (Life Technologies) for 14 clonal antibody (mAb) against human uPAR (Clon CD87, American days, and the pooled cell population was used for subsequent experiments. Diagnostica, Inc., Greenwich, CT). After preincubation, cells were plated Release of cellular structures and RNA extraction from three- to the upper chamber and migration analysis was done as described earlier. dimensional cultures. Cultures grown in three-dimensions were released When indicated, 10 mg/AL of GM6001 (llomastat) MMP inhibitor from Matrigel using a modified procedure described by Weaver et al. (28). (Chemicon) was added to the cells just before plating them into the transwell. Briefly, cells were washed with cold PBS without Ca2+ and Mg2+ and Immunofluorescence. Three-dimensional cultures were fixed in 2% containing 5 mmol/L EDTA, and scraped into a centrifuge tube with a paraformaldehyde at room temperature for 1 hour, embedded in 20% sucrose, minimum volume of 30 mL cold PBS/EDTA. Cells were incubated on ice for and frozen in Tissue-Tek optimum cutting temperature compound (Sakura 45 minutes, until the Matrigel disassociated, and centrifuged at 115 Â g for Finetek USA, Inc., Torrance, CA). Five-micrometer-thick frozen sections were 2 minutes. The pellet was resuspended in an appropriate amount of RNA prepared for immunostaining as follows: sections were fixed with acetone for lysis buffer and RNA extraction was done using the RNeasy isolation kit 30 seconds at room temperature, washed with 50 mmol/L PBS/glycine thrice (Qiagen, Valencia, CA). for 20 minutes at room temperature, blocked with 10% goat serum in Reverse transcription-PCR. One microgram of total RNA extracted by immunofluorescence buffer (0.1% BSA, 0.2% Triton X-100, 0.05% Tween 20 in RNeasy isolation kit (Qiagen) was reverse transcribed using Moloney murine PBS) for 1 hour at room temperature, and with 2.5% BSA with 1:100 fab leukemia virus reverse transcriptase (Qiagen), and 1/25 of this reaction was fragment [anti-mouse IgG, F(ab) fragment of goat antibody, Sigma] in linearly amplified for 30 cycles (HoxD10 and h4 integrin), 25 cycles immunofluorescence buffer for 45 minutes at room temperature. Incubation (a3 integrin and uPAR), and 32 cycles (MMP-14) following denaturation with the primary antibody was done at 1:100 dilution of antibody in (30 seconds at 95jC), annealing (30 seconds at 58jC for HoxD10, h4 integrin, immunofluorescence buffer plus 1:100 fab fragment for 1 hour at room and a3 integrin; 60jC for MMP-14 and uPAR), and extension (30 seconds at temperature. Antibodies used were h4 integrin mAb (mAb 1964, Chemicon) 72jC) in a thermal cycler (PTC-200 Peltier Thermal cycler, M.J. Research, and a3 integrin mAb (Chemicon). After washing with immunofluorescence Waltham, MA). The following primers were used: HoxD10 forward buffer, 1:100 secondary goat anti-mouse FITC-conjugated IgG (Zymed 5VCTGTCATGCTCCAGCTCAACCC 3V, reverse 5VCTAAGAAAACGTGAGGTT- Laboratories, Inc., San Francisco, CA) was applied for 1 hour at room GGCGGTC 3V; h4 integrin forward 5VATGGACTTCTCCAACTCC 3V, reverse temperature. Nuclei were counterstained with 4V,6-diamidino-2-phenylindole 5VGAGTAGTTGGTGACAGCAAAG 3V; a3 integrin forward 5V CAAGTGG- (DAPI; Sigma). Control sections were stained with secondary antibodies only. CTGCTGTATCCCACG 3V, reverse 5VGTGCTGGTGCATCACGGATGG 3V; The slides were mounted in fluoromount (Southern Biotechnology MMP-14 forward 5VCAACACTGCCTACGAGAGGA 3V, reverse 5V GTTCTA- Associates, Inc., Birmingham, AL). The images were collected with Zeiss CCTTCAGCTTCTGG 3V; and uPAR forward 5VCCTTGCAGCTGTAACACTGG 410 LSM confocal microscope (Zeiss Pluar 63Â oil objective lenses, Carl Zeiss 3V, reverse 5VCAGGACCTCTGCAGGACCAC 3V. Total RNA was normalized MicroImaging, Inc., Thornwood, NY). Images for figures were colored with using 18S internal standards at a 1:4 ratio (Ambion, Austin, TX). Adobe Photoshop 7.0 Software. Immunoblot analysis. Cells were grown in 3DlrBM cultures for 4 days, Immunofluorescence for activated extracellular signal-regulated kinase and colonies were isolated using ice-cold PBS/EDTA as mentioned before. (ERK) 1/2 was done using the same protocol above using the antibody Pellets were formed from lysate in radioimmunoprecipitation assay buffer ERK1/2 (pT202/pY204) Phospho-Specific (BD Biosciences). Slides were [50 mmol/L Tris-HCl (pH 7.4), 150 mmol/L sodium chloride, 1% NP40, 0.5% analyzed under a Nikon Eclipse TE300 fluorescence microscope. deoxycholate, 0.2% SDS, 20 mmol/L sodium fluoride, and 1 mmol/L sodium Ki-67 proliferation index. The proliferative rate of cells grown in orthovanadate, and a cocktail of protease inhibitors]. Equal amounts of three-dimensional cultures was assessed by Ki-67 immunostainig. Five- protein were separated on a 7.5% nonreducing SDS-PAGE gels, immuno- micrometer-thick frozen sections (see above) were fixed in methanol/ blotted with a polyclonal a3 integrin antibody (AB1920, Chemicon, acetone (1:1, v/v) for 10 minutes at room temperature, washed in 50 mmol/L Temecula, CA) or polyclonal h4 integrin antibody (AB1922, Chemicon), PBS/glycine, blocked as described above, and incubated with a 1:50 dilution and detected with an enhanced chemiluminescence system (Amersham of Ki-67 mAb (Clon MB67, Neomarkers, Fremont, CA) in 10% goat serum Pharmacia BioTech, Piscataway, NJ). immunofluorescence buffer plus 1:100 fab fragment, overnight at 4jC. After Analysis of proliferation. Cells (5 Â 104) were plated in 12-well plates washing with immunofluorescence buffer, a 1:100 dilution of goat anti-mouse and serum starved for 16 hours. Proliferation was evaluated by adding 1 mL FITC conjugated IgG (Zymed) in immunofluorescence buffer containing 10% of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT)/ goat serum was applied for 40 minutes at room temperature. Slides were media (15% MTT/PBS 2 mg/mL and 85% media; Sigma) into each well. counterstained with DAPI and mounted in fluoromount (Southern

Cancer Res 2005; 65: (16). August 15, 2005 7178 www.aacrjournals.org

Biotechnology Associates). Proliferating cells were determined by visually cervical cancer cell line HeLa (Fig. 2A). In contrast, normal human counting nuclei stained with DAPI of at least 300 cells and thereafter scoring microvascular endothelial cells continue to express HoxD10. To Ki-67–positive cells as a percentage of the total cell number. determine the effects of restoring HoxD10 expression on tumor- In situ hybridization. Sense or antisense riboprobe against HoxD10 was igenic epithelial cells, we stably transduced both HeLa and the generated as previously described (3). Digoxigenin-labeled probes were highly invasive HoxD10-negative breast epithelial tumor cell line prepared using T7 or SP6 RNA polymerase (Roche, Indianapolis, IN). Seven- micrometer-thick sections of paraffin-embedded human breast or endome- MDA-MB-231 with a retroviral vector carrying the human HoxD10 trial tissue were hybridized with 800 ng/mL of probe as previously described cDNA (pLD10), a HoxD10 point mutant which does not bind DNA (3). Four of each fibrocystic, high-grade ductal carcinoma and invasive ductal (pLD10 N-T), or the control vector (pLXSN). RT-PCR showed that carcinoma human breast tumor specimens were examined. Five of each whereas no endogenous expression of HoxD10 could be detected in normal, grade 2, and grade 3 human endometrial carcinoma was examined. control-transduced HeLa and MDA-MB-231 cells (HeLapLXSN and Tumor formation in vivo. A total of 5 Â 106 MDA-MB-231pLXSN or MDA-MB-231pLXSN), HoxD10 expression was abundantly MDA-MB-231pLD10 was injected s.c. into the flank region of nu/nu mice expressed following selection of cells transduced with the HoxD10 (n = 4 mice/group, in each of two experiments; Charles River, Wilmington, retroviral vector (HeLapLD10, MDA-MB-231pLD10, and MDA-MB- MA). After 2 weeks, when tumors had developed, tumor volume was 231pLD10N-T; Fig. 2A). measured thrice a week for 6 weeks. Volume was calculated according to the following formula: V (mm3)=A (mm) Â B2 (mm2) / 2, where B is the Using the MTTassay to assess proliferation, we did not detect any smaller dimension. significant differences between control and HoxD10-transduced Statistical analysis was done using Student’s t distribution with cells either in MDA-MB-231 or HeLa cells (Fig. 2B). In addition, we significance reported when P < 0.05. also evaluated incorporation of BrdUrd in control and HoxD10- Procedures were done in accordance with regulations for the proper care transfected cells but did not observe any significant differences (data and use of laboratory animals. not shown). We had previously observed that HoxD10 did not induce Statistical analysis. In all experiments, statistical analysis was done growth arrest in cultured endothelial cells, yet in vivo HoxD10 was a using Student’s t distribution with significance reported when P < 0.05. potent inhibitor of angiogenesis. In tissue culture studies, we observed that HoxD10 impaired Results endothelial cell migration in response to basic fibroblast growth HoxD10 is lost during tumor progression. In previous studies, factor and vascular endothelial growth factor (3) and, therefore, we we examined the expression of HoxD10 in blood vessels in normal assessed migration in control or HoxD10-transduced tumor cells and preinvasive breast tumors by in situ hybridization (3). During the using modified Boyden chambers coated with type I collagen. course of these studies, we also noted diminished expression of HoxD10 significantly decreased migration (60%) of both HeLa and HoxD10 in breast epithelial cells in high-grade ductal carcinoma MDA-MB-231 cells compared with control cells (Fig. 2C). In in situ as compared with fibrocystic tissue. To fully evaluate contrast, a HoxD10 mutant, which does not bind DNA, did not expression of HoxD10 in normal, preinvasive, and invasive breast inhibit migration of MDA-MB-231 cells, indicating that binding to lesions, we did in situ hybridization on specimens of fibrocystic, DNA and subsequent effects on gene expression were required for ductal carcinoma in situ, and invasive human breast tissues. We HoxD10-mediated suppression of migration (Fig. 2C). To eliminate observed strong expression of HoxD10 in premalignant epithelial the possibility that changes in gene expression were altering the cells (Fig. 1A), whereas expression of HoxD10 was relatively reduced ability to adhere to the collagen matrix rather than influencing in ductal carcinoma in situ (preinvasive breast tissue; Fig. 1B) and migration per se, we also evaluated adhesion to type I collagen but largely absent in invasive breast carcinomas (Fig. 1C). To study did not observe any significant differences in control or HoxD10- whether the progressive loss of HoxD10 expression was restricted to transfected MDA-231 or HeLa cells (data not shown). breast tumors, we subsequently did in situ hybridization on normal Moreover, HoxD10 impaired migration in MDA-MB-231 cells to a and increasingly tumorigenic endometrial tissue. Again, we observed similar degree as treatment with function-blocking antibodies positive hybridization of HoxD10 mRNA in nonmalignant endome- against a3 integrin and uPAR, or with the MMP inhibitor GM6001 trial epithelial cells (Fig. 1E) with progressive loss of HoxD10 (Fig. 2D). expression in grade 2 endometrial tumors (Fig. 1F) and almost HoxD10 induces tumorigenic epithelial cells to acquire an complete absence of HoxD10 expression in epithelial cells of grade 3 organized polarized phenotype when cultured in the 3DlrBM endometrial carcinoma (Fig. 1G). Figure 1D and H shows control model. Interactions between epithelial cells and the microenvi- hybridization using a HoxD10 sense riboprobe for both fibrocystic ronment, in particular the extracellular matrix, are critical for breast (Fig. 1D) and normal endometrium (Fig. 1H). These results maintaining normal tissue architecture and differentiated function. show that HoxD10 expression decreases in both breast and When premalignant or invasive breast tumor cells are cultured in endometrial epithelial cells as malignancy progresses, suggesting 3DlrBM, they exhibit distinct phenotypic features, which cannot be that its expression may have an important role in maintaining a observed in conventional two-dimensional tissue culture plastic. nonmalignant phenotype. Whereas tumor cells form large disorganized, nonpolarized Sustained expression of HoxD10 impairs migration but not structures, nonmalignant epithelial cells form polarized, organized proliferation of tumor epithelial cells in conventional two- structures reminiscent of acinar structures observed in the normal dimensional tissue culture. We further analyzed expression of breast in vivo (28). We therefore cultured control and HoxD10- HoxD10 in at least eight invasive human breast cancer cell lines. expressing MDA-MB-231 cells in 3DlrBM. In contrast to control- Reverse transcription-PCR (RT-PCR) analysis revealed that HoxD10 transduced MDA-MB-231 cells, which formed large, stellate was not expressed by MDA-MB-231, MDA-MB-453, BT-549, T47D, disorganized clusters (Fig. 3A), MDA-MB-231 expressing HoxD10 or SkBr3 cell lines (Fig. 2A). Although HoxD10 was also lacking in cells formed small, compact structures similar to the acinar-like the invasive MDA-MB-468 and BT-20 breast cancer cell lines, low structures formed by nonmalignant breast cells (Fig. 3B; ref. 28). levels could be detected in the weakly malignant MDA-MB-436 cell The percentage of organized colonies increased by 2-fold in MDA- line (not shown). In addition, HoxD10 was also not expressed in the MB-231pLD10 compared with control cells (Fig. 3C). www.aacrjournals.org 7179 Cancer Res 2005; 65: (16). August 15, 2005

Figure 1. Expression of HoxD10 mRNA in breast (A-D) and endometrial (E-H) tissue. HoxD10 mRNA was detected by in situ hybridization as described in Materials and Methods. All sections were counterstained with fast green. A and E, normal human breast and endometrial tissue, respectively. Positive HoxD10 expression (black arrows) is seen in both breast and endometrial epithelium. Positive staining is also seen in capillaries within breast stromal connective tissue (red arrow). B and F, high-grade ductal carcinoma in situ and grade 2 endometrial carcinoma, respectively. HoxD10 expression is reduced in both breast and endometrial epithelium compared with normal tissue. C and G, invasive carcinoma of the breast and grade 3 endometrial carcinoma, respectively. HoxD10 expression is nearly absent in both breast and endometrial tissues. D and H, as negative controls, both fibrocystic breast (D) and normal endometrium (H) were hybridized with a HoxD10 sense riboprobe. Original magnifications, Â10 and Â40 (insets).

Subsequent immunofluorescent staining for h4 integrin con- mRNA and protein levels in MDA-231-pLD10 compared with firmed that this integrin was redistributed to the basolateral control cells were significantly reduced as shown by RT-PCR surface of HoxD10-transduced MDA-MB-231 but not control cells, analysis (Fig. 4B) and Western blot analysis (Fig. 4C). However, cultured in 3DlrBM cultures, consistent with formation of polarized mRNA levels of uPAR and MMP-14 did not significantly change in structures (Fig. 4A, top). However, neither expression of h4 integrin the presence or absence of HoxD10 (Fig. 4B). Figure 4B (bottom) mRNA nor protein changed between HoxD10-expressing cells and shows expression of HoxD10 in transduced MDA-MB-231 com- controls, as shown by RT-PCR (Fig. 4B) and Western blot analysis pared with control cells as determined by RT-PCR analysis. (Fig. 4C), respectively. Our previous studies showed that in Previous studies using the 3DlrBM assay also noted that when endothelial cells, HoxD10 impaired expression of a3 integrin, tumor cells were reverted to more organized, polarized structures, uPAR, and MMP-14, all genes known to promote cell migration and they also underwent growth arrest accompanied by a reduction in invasion in both endothelial and breast tumor cells (23–25). To mitogen-activated protein kinase (MAPK) signaling (29). We therefore determine whether similar target genes were being suppressed by did immunofluorescent staining for phospho-ERK1/2 in control or HoxD10 in breast tumor cells, we did immunofluorescence and HoxD10-expressing MDA-MB-231 cells in 3DlrBM and observed a RT-PCR analysis on control and HoxD10-expressing cells cultured marked reduction in intensity of phospho-ERK1/2 in HoxD10 in 3DlrBM. Immunofluorescence staining of MDA-MB-231 revealed expressing cells (Fig. 5A). Consistent with the formation of organized that a3 integrin expression was redistributed and reduced in polarized acini, and reduced phospho-ERK1/2, MDA-MB-231 cells HoxD10-expressing MDA-MB-231 cells compared with control cells expressing HoxD10 showed a significantly reduced percentage of Ki- (Fig. 4A, bottom). In contrast to h4 integrin, however, a3 integrin 67–positive cells, indicating induction of growth arrest (Fig. 5B and C).

Cancer Res 2005; 65: (16). August 15, 2005 7180 www.aacrjournals.org

Thus, restoring HoxD10 not only impairs migration but also addition, HoxA10 induces growth arrest in myelomonocytic cells promotes cell-cell interactions and cell-extracellular matrix inter- and thus might also function as a tumor suppressor (32). actions and induces growth arrest, which results in the formation A recent study by Cantile et al. (33) did not detect expression of of polarized acinar-like structures. Together these results suggest HoxD10 in normal human breast, but instead observed HoxD10 that HoxD10 may function as a tumor suppressor and prevent breast cancer progression in vivo. HoxD10 inhibits tumor growth in vivo. To determine the effects of HoxD10 on breast tumor growth in vivo, we inoculated MDA-MB-231pLXSN control and MDA-MB-231pLD10 into nude mice. Cells were injected s.c. into female nude mice and tumor volume was measured weekly for 6 weeks. Throughout the experiment, tumor volume was significantly smaller in mice that received MDA-MB-231pLD10 cells compared with the control group (Fig. 6). Furthermore, three of eight mice that received HoxD10- expressing cells were free of any measurable tumors 6 weeks after cell injection. Control mice were euthanized when tumor volume reached about 1 cm3, 6 weeks after cell injection, whereas the HoxD10 tumor-free mice were monitored for another 10 weeks and remained tumor-free. Thus, HoxD10 is a potent inhibitor of tumor growth in vivo.

Discussion We show that expression of HoxD10 is progressively lost in both breast and endometrial tissue with increasing tumorigenicity. However, if expression of HoxD10 is restored in malignant breast tumor cells using retroviral gene transfer, cell migration is impaired, a normal polarized, acinar morphology in three-dimensional cultures is restored, and tumor growth in vivo is inhibited. Taken together, our results suggest that HoxD10 normally functions as a tumor suppressor in breast tissue. Our results, together with an earlier study showing loss of HoxD10 expression in endometrial tumors relative to normal tissue (16), supports a hypothesis that HoxD10 may act to stabilize or maintain a differentiated phenotype in both breast and endometrial epithelial cells. Previous studies examining targeted disruption of HoxD10 showed that four of six females initially failed to undergo lactation (30). Subsequent analysis of these mice revealed that whereas alveolar development progressed normally through pregnancy, these mice failed in whole or in part to produce milk and expand their alveoli, a hallmark of functionally differentiated mammary epithelial cells (19). It is also worth noting that constitutive expression of the paralogous HoxA10 in BT20 cells, another aggressive human breast cancer cell line, also reduces their invasion through Matrigel (31). In

Figure 2. HoxD10 impairs migration, but not proliferation, of tumor epithelial cells in two-dimensional cultures. A, RT-PCR analysis showing relative levels of HoxD10 mRNA in control and HoxD10-transduced HeLa and MDA-MB-231 cells as well as levels of HoxD10 in nontransfected MDA-MB-453, BT-549, T47D, and SkBr3 human breast tumor cell lines. The human microvasculature endothelial cell line (HMVEC) serves as a positive control. The 493 bp product corresponds to HoxD10 mRNA. The relative RNA loading is shown by the 324 bp band corresponding to 18S rRNA determined using competitive primers. B, MTT assay showing relative proliferation of control and HoxD10-transduced HeLa and MDA-MB-231 cells under conventional two-dimensional culture conditions after 16 hours in serum-free medium. Columns, means from three independent experiments, each done in triplicate; bars, SD. C, relative migration of control and HoxD10-transduced cells on membranes coated with 10 Ag/mL type I collagen. Migration was assessed after 4 hours at 37jC. Columns, means from three independent experiments, each done in triplicate; bars, SD. **, P < 0.01, compared with control pLXSN; P values calculated using Student’s t test. D, relative migration of control MDA-MB-231pLXSN cells treated with function-blocking antibodies against a3 integrin and uPAR, or with GM6001 as indicated in Materials and Methods. Columns, means from three independent experiments, each done in triplicate; bars, SD. **, P < 0.01, compared with control pLXSN; P values calculated using Student’s t test. www.aacrjournals.org 7181 Cancer Res 2005; 65: (16). August 15, 2005

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2005 American Association for Cancer Research. Cancer Research expression in 7 of 14 breast cancer biopsies. The reason for this basement membrane is not entirely clear but a3 integrin also discrepancy is not clear. However, as their analysis was based on functions as a receptor for proteolyzed laminin fragments, generated RT-PCR using total tissue RNA, it is possible that activated stromal by high proteolytic activity of invasive breast tissue (36), which in cells may contribute to the increase in HoxD10 expression observed turn would promote migration and invasion. in tumor tissue, whereas the proportion of epithelial cells in normal Our results suggest that HoxD10 inhibits migration by reducing tissue, where we detected strong HoxD10 expression, comprises only expression of a3 integrin. This is consistent with the migration a minor proportion of total breast tissue RNA. Moreover, while this assay results showing that function-blocking antibody to a3 article was under review, a more recent study using real-time PCR integrin inhibited migration of MDA-MB-231 cells to the same reported significantly lower levels of HoxD10 in human invasive extent as HoxD10. Whether a3 integrin is a direct target for ductal breast cancer tissue compared with normal breast (34), HoxD10 is not known. The human promoter region for a3 integrin further supporting our in situ hybridization data. has not yet been identified, but the mouse a3 integrin promoter Our current results also support our previous observations for the contains at least two Hox binding recognition sites (TNAT/C; role of HoxD10 in inhibiting migration in vascular endothelial cells ref. 37). Chromatin immunoprecipitation analysis is currently (3) and previous studies showing that HoxD10 is strongly expressed under way in our laboratory to determine whether the promoter of in differentiated breast epithelial cells (19). It is of interest that we the a3 integrin is a direct or secondary target for HoxD10. could not detect significant levels of HoxD10 in any cultured We previously showed that HoxD10 inhibits expression of uPAR, epithelial cells, despite the strong expression we observed in normal MMP-14, and a3 integrin in endothelial cells (3). Although HoxD10 epithelium in vivo. These findings suggest that the normal breast inhibits expression of a3 integrin in MDA-MB-231 cells, we did not microenvironment plays a key role in maintaining HoxD10 detect differences in uPAR or MMP-14 mRNA expression levels expression. However, it is not clear what changes associated within between control and HoxD10-expressing cells, suggesting these the tumor microenvironment lead to the loss of HoxD10 expression. genes may not be direct transcriptional targets of HoxD10 in MDA- Our expression analysis data in three-dimensional cultures MB-231 cells. It is also worth noting that a3 integrin is required for indicate that HoxD10 suppresses expression of the a3 integrin proper uPAR localization and proteolytic and signaling functions subunit, member of the heterodimeric a3h1 integrin, which (38, 39). Thus, inhibition of a3 integrin expression may lead to recognizes a variety of extracellular matrix proteins including decreased plasminogen activity and lead to a reduction in laminin-5 and collagen, and mediates both cell-extracellular matrix migration. and cell-cell interactions. It has been reported in various tumors, In a recent article, Wang et al. (40) also reverted MDA-MB-231 including breast cancer, that the aberrant expression of a3h1 breast tumor cells to a normal phenotype using the 3DlrBM model. integrin is associated with a high invasive and metastatic potential Whereas inhibition of epidermal growth factor receptor, h1 integrin, (23, 26, 35). How a3 integrin disrupts normal interactions with the phosphatidylinositol 3-kinase, or MAPK alone was not sufficient

Figure 3. Influence of HoxD10 on morphology of MDA-MB-231 cells cultured in 3DlrBM. A, phase-contrast micrographs of MDA-MB-231pLXSN after 4 days on top of 3DlrBM (Matrigel). Top, original magnification, Â10; bar, 60 Am. Bottom, original magnification, Â40; bar, 10 Am. B, phase-contrast micrographs of MDA-MB- 231pLD10 cells. Transduction with HoxD10 induces formation of acinar structures in MDA-MB-231 cells after 4 days on top of 3DlrBM (Matrigel). Top, original magnification, Â10; bar, 60 Am. Bottom, original magnification, Â40; bar, 10 Am. C, percentage of organized colonies in MDA-MB-231pLXSN and MDA-MB-231pLD10 after 4 days on top of lrBM (Matrigel) as determined by counting at least 100 colonies in each of three independent experiments. Bars, SD. **, P < 0.01, compared with control pLXSN; P values calculated using Student’s t test.

Cancer Res 2005; 65: (16). August 15, 2005 7182 www.aacrjournals.org

Figure 4. HoxD10 reduces a3 integrin expression and induces a polarized phenotype in MDA-MB-231 cells cultured in 3DlrBM. A, immunofluorescence staining for h4 integrin (top) and a3 integrin (bottom) in control (C) and HoxD10-expressing cells (D10) after culturing for 4 days on 3DlrBM. Bar, 10 Am. B, relative mRNA levels of a3 integrin, h4 integrin, uPAR, MMP-14, and HoxD10 in control (C) or HoxD10-expressing cells (D10) cultured in 3DlrBM as determined by semiquantitative RT-PCR analysis. Densitometric analysis and correction for total 18S RNA levels in each sample are shown. Gray bars, controls; white bars, HoxD10-expressing cells. C, relative protein levels of a3 and h4 integrins in control and HoxD10-expressing cells cultured in 3DlrBM as determined by Western blot analysis. Equal amounts of protein lysates were loaded per lane. Hybridization with antibodies to actin serves as the loading control. to revert MDA-MB-231 cells, blocking h1 integrin or expressing Nonetheless, it is likely that the suppression of MAPK activity by E-cadherin in combination with inhibition of phosphatidylinositol HoxD10 occurs indirectly by promoting proper interactions with the 3-kinase or MAPK effectively reverted the cells to a nonmalignant basement membrane, perhaps via attenuation of a3integrin phenotype. Our results show that restoring HoxD10 alone was expression, which otherwise facilitates degradation of the extracel- sufficient to phenotypically revert the malignant phenotype of MDA- lular matrix and cellular invasion (23, 26). Although signaling via a3 MB-231, reduce signaling through the MAPK kinase pathway, and integrin has not been previously studied in breast tumor cells, recent induce growth arrest. Indeed, whereas the levels of total ERK protein studies show that signaling through this laminin receptor leads to in both control and HoxD10 expressing MDA-MB-231 cells were high levels of ERK activity in dermal epithelial cells (42). Nonetheless, similar, only the activity of ERKwas reduced with HoxD10. Moreover, our findings that restoring HoxD10 expression alone was sufficient to following transfection with HoxD10, we did not detect any revert the phenotype of the aggressive MDA-MB-231 cells further reexpression of E-cadherin or changes in the levels of h1orh4 supports the notion that HoxD10 affects key regulatory pathways and integrins in MDA-MB-231 cells which have previously been linked to is consistent with a role for Hox factors as master regulatory genes. reduced MAPK signaling in breast tumor cells (40, 41). Similarly, Normal tissue architecture depends on proper cell-cell and cell- using a combination of treatments to induce polarity and extracellular matrix interactions, and disruption of these inter- morphologic reversion in MDA-MB-231 cells, Wang et al. (40) did actions not only results in loss of tissue structure but also not observe any reexpression of E-cadherin. Whether other contributes to tumor progression. Thus, as tumors exhibit gross cadherins are induced and contribute to the polarized phenotype defects in normal tissue patterning, it is not surprising that acquired by MDA-MB-231 following transfection with HoxD10 alterations in expression of Hox genes, which direct normal tissue or treatment with a combination of inhibitors remains to be patterning, are common in many different solid tumors (for review, established. see refs. 19, 43). Moreover, using standard two-dimensional culture www.aacrjournals.org 7183 Cancer Res 2005; 65: (16). August 15, 2005

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2005 American Association for Cancer Research. Cancer Research conditions, where most cells exhibit a similar monolayer morphology and proliferation rate, it may not be possible for patterning genes to fully exert their influence in directing more complex tissue-type morphologies which they readily induce in vivo. Indeed, whereas previous studies have noted that loss of HoxA5 leads to impaired differentiation of gut and stomach epithelial cells, HoxA5 expression is primarily restricted to the adjacent mesenchyme and the subsequent effect on epithelial differentiation is derived by loss of inductive signals (44, 45). The utility of using the 3DlrBM approach to recapitulate phenotypic aspects of breast tumor cells in vivo has elegantly been shown by Bissell and others (28, 41, 46–51). Our results further emphasize that three-dimensional culture models or in vivo assays are required to fully appreciate the effect of morphoregulatory genes including HoxD10 on tissue patterning and morphology of breast epithelial cells. Moreover, we had previously observed that whereas HoxD10 does not induce growth arrest in endothelial cells Figure 6. HoxD10 inhibits tumor growth in vivo. Control or HoxD10-expressing cells (5 Â 106) were injected s.c. into nude mice at day 0. After 14 days, when tumors when cultured in conventional two-dimensional tissue culture developed, tumor volume was measured for 6 weeks. Points, means (n =8miceper methods, in vivo HoxD10 was a potent inhibitor of angiogenesis (3). group); bars, SD. **, P <0.01.P values calculated using Student’s t test. Similarly, whereas HoxD10 also does not affect proliferation of breast tumor cells in two-dimensional cultures, in three-dimen- Interestingly, previous studies have shown that Hox genes located sional cultures using lrBM, or in vivo, HoxD10 leads to growth near the opposite 3V end of the clusters are linked with oncogenic arrest. In previous studies, we observed that serum withdrawal was transformation and increased invasion. Forced expression of HoxA1 not sufficient to induce growth arrest in nonmalignant mammary in MCF-10A cells results in oncogenic transformation and the epithelial cells in two-dimensional culture, but culturing in 3DlrBM development of a rapidly growing carcinoma in vivo (22). HoxD3 led to suppressed expression of cyclin D1 and growth arrest (29). induces expression of metastasis-related genes in human lung Together these findings suggest that HoxD10 acts to restore the cancer cells (52, 53). Results from our laboratory show that HoxD3 tumor cells ability to respond to cues from the extracellular matrix also promotes expression of several genes which promote invasion and subsequently undergo growth arrest. in endothelial cells including integrins avh3, a5h1, and uPA (5, 54),

Figure 5. HoxD10 induces growth arrest in breast tumor cells cultured in 3DlrBM. A, immunofluorescence staining for phosphorylated ERK1/2 and DAPI nuclear staining in control (MDA-MB-231pLXSN) and HoxD10 (MDA-MB-231pLD10) expressing cells. Original magnification, Â20; bar, 10 Am. B, immunofluorescence staining for Ki-67 and DAPI. Original magnification, Â20; bar, 10 Am. C, quantitative analysis of Ki-67 labeling index as determined by counting nuclei stained with DAPI in at least 300 cells and thereafter scoring Ki-67–positive cells as a percentage of the total cell number. Columns, means; bars, SD. **, P < 0.01, compared with controls; P values calculated using Student’s t test. Cultures were analyzed after 4 days in 3DlrBM.

Cancer Res 2005; 65: (16). August 15, 2005 7184 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2005 American Association for Cancer Research. HoxD10 as a Tumor Suppressor Gene for Breast Cancer and when overexpressed in vivo, HoxD3 promotes a hemangioma- Acknowledgments like proliferation of blood vessels (5, 55). Taken together, these Received 6/4/2004; revised 5/4/2005; accepted 6/2/2005. results are consistent with the distinct roles predicted for 5Vand 3V Grant support: NIH grant CA 85249 (N. Boudreau). Hox genes based on their nonoverlaping expression patterns during The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance development and with the increasing expression of 5VHox genes in with 18 U.S.C. Section 1734 solely to indicate this fact. differentiating cells (56, 57). We thank Dr. Eduard Serra for useful comments on the manuscript.

References metastasis, invasion, and gelatinase B (MMP-9) activity. pathways in three-dimensional contexts. J Natl Cancer Int J Cancer 2000;87:336–42. Inst 2002;94:1494–503. 1. Botas J. Control of morphogenesis and differentiation 24. Hotary KB, Allen ED, Brooks PC, Datta NS, Long MW, 41. Weaver VM, Lelievre S, Lakins JN, et al. h4 integrin- by HOM/Hox genes. Curr Opin Cell Biol 1993;5:1015–22. Weiss SJ. Membrane type I matrix metalloproteinase dependent formation of polarized three-dimensional 2. Myers C, Charboneau A, Boudreau N. Homeobox B3 usurps tumor growth control imposed by the three- architecture confers resistance to apoptosis in normal promotes capillary morphogenesis and angiogenesis. dimensional extracellular matrix. Cell 2003;114:33–45. and malignant mammary epithelium. Cancer Cell 2002; J Cell Biol 2000;148:343–51. 25. Fisher JL, Field CL, Zhou H, Harris TL, Henderson 2:205–16. 3. Myers C, Charboneau A, Cheung I, Hanks D, Boudreau MA, Choong PF. Urokinase plasminogen activator 42. Manohar A, Shome SG, Lamar J, et al. a3h1 integrin N. Sustained expression of homeobox D10 inhibits system gene expression is increased in human breast promotes keratinocyte cell survival through activation angiogenesis. Am J Pathol 2002;161:2099–109. carcinoma and its bone metastases—a comparison of of a MEK/ERK signaling pathway. J Cell Sci 2004;117: 4. Boudreau N, Bissell MJ. Extracellular matrix signaling: normal breast tissue, non-invasive and invasive carci- 4043–54. integration of form and function in normal and noma and osseous metastases. Breast Cancer Res Treat 43. Abate-Shen C. Deregulated homeobox gene expres- malignant cells. Curr Opin Cell Biol 1998;10:640–6. 2000;61:1–12. sion in cancer: cause or consequence? Nat Rev Cancer 5. Boudreau N, Andrews C, Srebrow A, Ravanpay A, 26. Tawil NJ, Gowri V, Djoneidi M, Nip J, Carbonetto S, 2002;2:777–85. Cheresh DA. Induction of the angiogenic phenotype by Brodt P. Integrin a3h1 can promote adhesion and 44. Aubin J, Chailler P, Menard D, Jeannotte L. Loss of Hox D3. J Cell Biol 1997;139:257–64. 6. Edelman GM, Jones FS. Outside and downstream of spreading of metastatic breast carcinoma cells on the Hoxa5 gene function in mice perturbs intestinal the homeobox. J Biol Chem 1993;268:20683–6. lymph node stroma. Int J Cancer 1996;66:703–10. maturation. Am J Physiol 1999;277:C965–73. 7. Srebrow A, Friedmann Y, Ravanpay A, Daniel CW, 27. Grondahl-Hansen J, Christensen IJ, Briand P, et al. 45. Aubin J, Dery U, Lemieux M, Chailler P, Jeannotte L. Bissell MJ. Expression of Hoxa-1 and Hoxb-7 is regulated Plasminogen activator inhibitor type 1 in cytosolic Stomach regional specification requires Hoxa5-driven by extracellular matrix-dependent signals in mammary tumor extracts predicts prognosis in low-risk breast mesenchymal-epithelial signaling. Development 2002; epithelial cells. J Cell Biochem 1998;69:377–91. cancer patients. Clin Cancer Res 1997;3:233–9. 129:4075–87. 8. Leroy P, Berto F, Bourget I, Rossi B. Down-regulation of 28. Weaver VM, Petersen OW, Wang F, et al. Reversion of 46. Kirshner J, Chen CJ, Liu P, Huang J, Shively JE. Hox A7 is required for cell adhesion and migration on the malignant phenotype of human breast cells in three- CEACAM1-4S, a cell-cell adhesion molecule, mediates fibronectin during early HL-60 monocytic differentia- dimensional culture and in vivo by integrin blocking apoptosis and reverts mammary carcinoma cells to a tion. J Leukoc Biol 2004;75:680–8. antibodies. J Cell Biol 1997;137:231–45. normal morphogenic phenotype in a 3D culture. Proc 9. Care A, Silvani A, Meccia E, et al. HOXB7 constitutively 29. Boudreau N, Werb Z, Bissell MJ. Suppression of Natl Acad Sci U S A 2003;100:521–6. activates basic fibroblast growth factor in melanomas. apoptosis by basement membrane requires three- 47. Debnath J, Mills KR, Collins NL, Reginato MJ, Mol Cell Biol 1996;16:4842–51. dimensional tissue organization and withdrawal from Muthuswamy SK, Brugge JS. The role of apoptosis in 10. Peverali FA, D’Esposito M, Acampora D, et al. the cell cycle. Proc Natl Acad Sci U S A 1996;93:3509–13. creating and maintaining luminal space within normal Expression of HOX homeogenes in human neuroblas- 30. Carpenter EM, Goddard JM, Davis AP, Nguyen TP, and oncogene-expressing mammary acini. Cell 2002;111: toma cell culture lines. Differentiation 1990;45:61–9. Capecchi MR. Targeted disruption of Hoxd-10 affects 29–40. 11. Cillo C, Barba P, Freschi G, Bucciarelli G, Magli MC, mouse hind limb development. Development 1997;124: 48. Muthuswamy SK, Li D, Lelievre S, Bissell MJ, Brugge Boncinelli E. HOX gene expression in normal and 4505–14. JS. ErbB2, but not ErbB1, reinitiates proliferation and neoplastic human kidney. Int J Cancer 1992;51:892–7. 31. Chu MC, Selam FB, Taylor HS. HOXA10 Regulates induces luminal repopulation in epithelial acini. Nat Cell 12. Aberdam D, Negreanu V, Sachs L, Blatt C. The p53 Expression and matrigel invasion in human breast Biol 2001;3:785–92. oncogenic potential of an activated Hox-2.4 homeobox cancer cells. Cancer Biol Ther 2004;3:568–72. 49. Wang F, Weaver VM, Petersen OW, et al. Reciprocal gene in mouse fibroblasts. Mol Cell Biol 1991;11:554–7. 32. Bromleigh VC, Freedman LP. p21 is a transcriptional interactions between h1-integrin and epidermal growth 13. Friedmann Y, Daniel CA, Strickland P, Daniel CW. target of HOXA10 in differentiating myelomonocytic factor receptor in three-dimensional basement mem- Hox genes in normal and neoplastic mouse mammary cells. Genes Dev 2000;14:2581–6. brane breast cultures: a different perspective in epithelial gland. Cancer Res 1994;54:5981–5. 33. Cantile M, Pettinato G, Procino A, Feliciello I, biology. Proc Natl Acad Sci U S A 1998;95:14821–6. 14. De Vita G, Barba P, Odartchenko N, et al. Expression Cindolo L, Cillo C. In vivo expression of the whole 50. Boudreau N, Sympson CJ, Werb Z, Bissell MJ. of homeobox-containing genes in primary and meta- HOX gene network in human breast cancer. Eur J Suppression of ICE and apoptosis in mammary epithelial static colorectal cancer. Eur J Cancer 1993;29A:887–93. Cancer 2003;39:257–64. cells by extracellular matrix. Science 1995;267:891–3. 15. Tiberio C, Barba P, Magli MC, et al. HOX gene 34. Makiyama K, Hamada J, Takada M, et al. Aberrant 51. Liu H, Radisky DC, Wang F, Bissell MJ. Polarity and expression in human small-cell lung cancers xeno- expression of HOX genes in human invasive breast proliferation are controlled by distinct signaling path- grafted into nude mice. Int J Cancer 1994;58:608–15. carcinoma. Oncol Rep 2005;13:673–9. ways downstream of PI3-kinase in breast epithelial 16. Osborne J, Hu C, Hawley C, Underwood LJ, O’Brien 35. Koukoulis GK, Virtanen I, Korhonen M, Laitinen L, tumor cells. J Cell Biol 2004;164:603–12. TJ, Baker VV. Expression of HOXD10 gene in normal Quaranta V, Gould VE. Immunohistochemical localiza- 52. Miyazaki YJ, Hamada J, Tada M, et al. HOXD3 endometrium and endometrial adenocarcinoma. J Soc tion of integrins in the normal, hyperplastic, and enhances motility and invasiveness through the TGF-h- Gynecol Investig 1998;5:277–80. neoplastic breast. Correlations with their functions as dependent and -independent pathways in A549 cells. 17. Raman V, Martensen SA, Reisman D, et al. Compro- receptors and cell adhesion molecules. Am J Pathol 1991; Oncogene 2002;21:798–808. mised HOXA5 function can limit p53 expression in 139:787–99. 53. Okubo Y, Hamada J, Takahashi Y, et al. Transduction human breast tumours. Nature 2000;405:974–8. 36. Giannelli G, Falk-Marzillier J, Schiraldi O, Stetler- of HOXD3-antisense into human melanoma cells results 18. Bissell MJ, Radisky D. Putting tumours in context. Stevenson WG, Quaranta V. Induction of cell migration in decreased invasive and motile activities. Clin Exp Nat Rev Cancer 2001;1:46–54. by matrix metalloprotease-2 cleavage of laminin-5. Metastasis 2002;19:503–11. 19. Lewis MT. Homeobox genes in mammary gland Science 1997;277:225–8. 54. Boudreau NJ, Varner JA. The homeobox transcription development and neoplasia. Breast Cancer Res 2000;2: 37. Kato T, Katabami K, Takatsuki H, et al. Character- factor Hox D3 promotes integrin a5h1 expression and 158–69. ization of the promoter for the mouse a3 integrin gene. function during angiogenesis. J Biol Chem 2004;279: 20. Chen H, Chung S, Sukumar S. HOXA5-induced Eur J Biochem 2002;269:4524–32. 4862–8. apoptosis in breast cancer cells is mediated by caspases 38. Ghosh S, Brown R, Jones JC, Ellerbroek SM, Stack MS. 55. Zhong J, Eliceiri B, Stupack D, et al. Neovascularization 2 and 8. Mol Cell Biol 2004;24:924–35. Urinary-type plasminogen activator (uPA) expression of ischemic tissues by gene delivery of the extracellular 21. Chariot A, Castronovo V. Detection of HOXA1 and uPA receptor localization are regulated by a3h1 matrix protein Del-1. J Clin Invest 2003;112:30–41. expression in human breast cancer. Biochem Biophys integrin in oral keratinocytes. J Biol Chem 2000;275: 56. van Oostveen J, Bijl J, Raaphorst F, Walboomers J, Res Commun 1996;222:292–7. 23869–76. Meijer C. The role of homeobox genes in normal 22. Zhang X, Zhu T, Chen Y, Mertani HC, Lee KO, Lobie PE. 39. Wei Y, Eble JA, Wang Z, Kreidberg JA, Chapman HA. hematopoiesis and hematological malignancies. Leuke- Human growth hormone-regulated HOXA1 is a human Urokinase receptors promote h1 integrin function mia 1999;13:1675–90. mammary epithelial oncogene. J Biol Chem 2003; through interactions with integrin a3h1. Mol Biol Cell 57. Sauvageau G, Lansdorp PM, Eaves CJ, et al. 278:7580–90. 2001;12:2975–86. Differential expression of homeobox genes in function- 23. Morini M, Mottolese M, Ferrari N, et al. The a3h1 40. Wang F, Hansen RK, Radisky D, et al. Phenotypic ally distinct CD34+ subpopulations of human bone integrin is associated with mammary carcinoma cell reversion or death of cancer cells by altering signaling marrow cells. Proc Natl Acad Sci U S A 1994;91:12223–7.

www.aacrjournals.org 7185 Cancer Res 2005; 65: (16). August 15, 2005

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2005 American Association for Cancer Research. Homeobox D10 Induces Phenotypic Reversion of Breast Tumor Cells in a Three-Dimensional Culture Model

Meritxell Carrio, Gemma Arderiu, Connie Myers, et al.

Cancer Res 2005;65:7177-7185.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/65/16/7177

Cited articles This article cites 54 articles, 24 of which you can access for free at: http://cancerres.aacrjournals.org/content/65/16/7177.full#ref-list-1

Citing articles This article has been cited by 13 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/65/16/7177.full#related-urls

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/65/16/7177. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.