DEVELOPMENTAL DYNAMICS 219:261–276 (2000) Differential Mammary Morphogenesis Along the Anteroposterior Axis in Hoxc6 Gene Targeted Mice ALEJANDRA GARCIA-GASCA AND DEMETRI D. SPYROPOULOS* Center for Molecular and Structural Biology, Hollings Cancer Center, and Department of Cell Biology and Anatomy, Medical University of South Carolina, Charleston, South Carolina ABSTRACT Mammary epithelial cell proliferation, branching, and differentiation span from the appearance of the mammary bud in midgestation through to the cycling mammary gland in adulthood. Here, we show that females homozygous for a targeted disruption of the Hoxc6 homeobox gene produce thoracic mammary glands that are slightly under-developed at birth and completely cleared of epithelium by adulthood, and inguinal mammary ducts that are dilated and fail to regress in response to ovariectomy. Mammary buds are detected in E12.5 Hoxc6 homozygous embryos. However, in newborn Hoxc6 homozygous females, branching ductal structures and fat pad development are reduced. Whole-mount and histologic analyses of mammary glands from adult Hoxc6 homozygous females show the absence of mammary epithelium in thoracic glands and dilated ducts in inguinal glands at 100% penetrance. Histologic analysis of inguinal mammary glands from ovariectomized Hoxc6 homozygous females demonstrates no signs of the expected regression of epithelium, suggesting that these glands are not responsive to the loss of ovarian hormone signals. We further observe repression of Hoxc6 expression specifically within mammary stroma by estrogen and progesterone. Hoxc6 homozygous mice also exhibit a homeotic transformation of the second thoracic vertebra into the first (T2 to T1 conversion with 60% penetrance), corresponding to both the gene’s anterior boundary of expression and the most extreme appearance of mammary defects. The position-specific phenotypes observed and the potential role for Hoxc6 in mediating hormone-regulated ductal expansion and regression in the adult female are discussed. © 2000 Wiley-Liss, Inc. Key words: mammary; gland; epithelium; stroma; homeobox; Hox; estrogen; progesterone; skeletal; mutant; mice INTRODUCTION Mammary gland development in the mouse initiates with the formation of the mammary bud epithelium through interactions with surrounding mesenchyme. Invagination and not outward migration of proliferat© 2000 WILEY-LISS, INC. ing epithelial cells is an early feature that distinguishes the mammary bud from other proliferating epithelia, such as the apical ectodermal ridge of the developing limb bud (Tickle and Altabef, 1999). Mammary buds are first observed in the mouse on embryonic day E10 to E11. Key to their development is the activation of estrogen or testosterone receptors in the surrounding mesenchyme. In male embryos between days E13 and E15, androgens cause a cessation of mammary bud proliferation by means of condensation of surrounding mesenchyme. The fat pads lack mammary anlagen, and the glands remain rudimentary throughout life. In the absence of androgens, the mammary bud continues to proliferate and develop (Imagawa et al., 1994). By day E16, epithelial cell proliferation persists only in female embryos, and by the time of birth, the mammary gland has developed an epithelial component (parenchyma) embedded in fibrous and adipose connective (stroma) tissues (Imagawa et al., 1994; Sakakura, 1987). At this time, the parenchyma consists of epithelial cords that connect to the nipple by the primary duct. From birth to puberty, ductal growth is symmetrical and dependent on ovarian hormones (Daniel and Silberstein, 1987). At the onset of puberty, considerable proliferation occurs with asymmetric mammary growth and organization of the epithelial component into a branching ductal system surrounded by fibrous connective tissue that separates the ducts from subjacent adipose tissue (Haslam, 1987). Ovarian hormones play a significant role in breast development. However, the effect of other local tissue-specific factors and the interactions between hormones and target genes on epithelial cell proliferation and mammary development are not well understood. Hox homeobox genes encode transcription factors that establish spatial patterns by both creating positional identities and regulating cell proliferation. This integrated process involves the combinatorial actions of different Hox genes with overlapping boundaries of expression (Condie and Capecchi, 1994; Kostic and Capecchi, 1994; Horan et al., 1995a,b; Rancourt et al., *Correspondence to: Demetri D. Spyropoulos, Center for Molecular and Structural Biology, Hollings Cancer Center, and Department of Cell Biology and Anatomy, Medical University of South Carolina, 86 Jonathan Lucas Street, P.O. Box 250956, Charleston, SC 29403. E-mail: spyropdd@musc.edu Received 24 February 2000; Accepted 7 July 2000 262 GARCIA-GASCA AND SPYROPOULOS 1995; Boulet and Capecchi, 1996; Chen and Capecchi, 1999). Homeobox genes are known to be expressed in the mammary gland of adult female mice and in MCF-7 breast cancer cells (Castronovo et al., 1994; Friedmann et al., 1994; Chariot and Castronovo, 1996; Chariot et al., 1996; Friedmann and Daniel, 1996; Phippard et al., 1996; Srebrow et al., 1998; Chen and Capecchi, 1999). It has been demonstrated that Hoxc6 is expressed in the mouse mammary gland and that the level of Hoxc6 expression varies according to developmental stage (Friedmann et al., 1994). Because the developmental stage of the mammary gland is, in part, regulated by varying ovarian hormone levels, it has been suggested that Hoxc6 acts in response to ovarian hormones to mediate their effect on mammary tissue development (Friedmann et al., 1994). Epithelial-stromal interactions are essential for proper morphogenesis of mammary gland epithelium (Sakakura, 1987). Here, we used gene targeting to generate mice defective for Hoxc6 function. Mice homozygous for this targeted mutation exhibited a homeotic transformation in the thoracic vertebra at the anterior boundary of normal Hoxc6 expression. Neonatal lethality in pups of different genotypes from Hoxc6 homozygous females indicated a defect in the Hoxc6 homozygous mothers. Hoxc6 homozygous females were found to lack mammary epithelium in thoracic glands and to lack proper responsiveness to ovarian hormones in inguinal glands. Mammary buds appeared normal in homozygous E12.5 embryos, but branching and ductal development were reduced in thoracic glands of newborns. We further demonstrated that Hoxc6 expression is greatly up-regulated in mammary stroma of ovariectomized wild-type females and is normally down-regulated in stroma by estrogen and progesterone in intact females. Thus, differences between thoracic and inguinal gland phenotypes may be due to differences in the loss of homeotic and/or nonhomeotic Hoxc6 functions. The latter of which may occur in a hormone-regulated pathway involving epithelial-stromal interactions. RESULTS Targeted Disruption of the Hoxc6 Gene A targeting vector for the disruption of the Hoxc6 gene was generated using a genomic mouse DNA fragment encompassing both Hoxc5 and Hoxc6 coding sequences and flanked by two divergent copies of the HSV thymidine kinase genes (see Experimental Procedures section). Disruption of the Hoxc6 gene was accomplished by insertion of the pMC1neo gene cassette upstream of the first helix of the homeodomain (Fig. 1A). ES cells were electroporated with linearized targeting vector and subjected to double selection (see Experimental Procedures section). Eight of 189 drug-resistant ES cell lines were found to contain the Hoxc6 targeted allele, as identified by Southern blot analysis (Fig. 1B). Chimeric mice were generated by injection of ES cells into B6 blastocyst and germline chimeric males were identified by the appearance of Agouti pups from matings to B6 females. Het- erozygotes were initially identified by both Southern blot, using the same method for the identification of targeted cell lines, and by polymerase chain reaction (PCR). Subsequently, all mice were genotyped by PCR alone (Fig. 1A–C). Hoxc6-specific forward and reverse primers were used to identify the wild-type allele and Hoxc6-specific forward and pMC1neo gene reverse primers were used to identify the targeted allele, in a three-primer mix as described in Experimental Procedures section (Fig. 1A– C). The Targeted Hoxc6 Allele Is a True Null in the Mammary Gland Normally, at least two different promoters, the PRI and PRII promoters, drive transcription of the Hoxc6 gene. The distal PRI promoter is a common 59 regulatory unit for Hoxc4, Hoxc5, and Hoxc6 genes, and the proximal PRII promoter is responsive to retinoic acid activation (Cho et al., 1988; Sharpe et al., 1988; Simeone et al., 1988; Coletta et al., 1991; Mavilio, 1993; Shimeld et al., 1993). The larger Hoxc6 transcript (PRI mRNA) encodes the smaller 153 amino acid protein, whereas the smaller transcript (PRII mRNA) encodes the larger 235 amino acid protein. Both proteins have identical homeodomains but differ by 82 amino-terminal amino acids (Chariot and Gielen, 1998; Cho et al., 1988). Northern blot and reverse transcriptase (RT) PCR analyses of PolyA(1) mRNA from E12.5 Hoxc6 homozygous embryos demonstrated the complete absence of the PRII transcript and a reduced amount of a truncated PRI transcript consistent with the insertional disruption (data not shown). Whole-mount immunohistochemistry of E12.5 Hoxc6 homozygous embryos demonstrated a greatly reduced amount of truncated PRI protein with an anterior boundary of expression of pv12 (shifted posterior from the wild-type pv7/8 position; Fig. 1D). In vitro transcription assays using a luciferase reporter in conjunction with wildtype Hoxc6 PRI, the truncated (targeted) Hoxc6 PRI, and empty vector demonstrated negligible transcriptional activity for both the truncated (targeted) Hoxc6 PRI and empty vector in HeLa, NIH-3T3, and MCF7 cells (data not shown). Thus, although thoracic structures of the embryo lack Hoxc6 protein, more posterior embryonic structures contain a greatly reduced amount of a truncated form of Hoxc6 protein that is predicted to be defective for transcriptional activity. Northern blot analysis of PolyA(1) mRNA isolated from mammary glands was conducted on intact and ovariectomized females of both the wild-type and homozygous genetic backgrounds (Fig. 2). As in the whole embryo, PRII is the predominant transcript observed in the mammary glands of wild-type females and it is preferentially up-regulated in the mammary glands of ovariectomized females (Fig. 2). This result is consistent with those obtained by Friedmann and colleagues (Friedmann et al., 1994), demonstrating the hormoneregulated expression of the Hoxc6-PRII transcript. In contrast, neither PRI nor PRII transcripts are expressed MAMMARY DEFECTS IN Hoxc6 TARGETED MICE 263 Fig. 1. Generation and identification of the Hoxc6 targeted allele and immunohistochemistry of embryos. A: Targeted allele. The pMC1neo cassette (black triangle) was introduced into the BglII site upstream of the first helix of the homeobox of the Hoxc6 gene (black box). The targeting vector contained 12.8 kb of genomic DNA extending 39 from the EcoRI site through Hoxc6 and Hoxc5 genes to a BamHI site and was flanked by two divergent copies of the HSV-TK gene (not shown). B: Identification of targeted ES cells. Targeted ES cells were identified by Southern blot by using HindIII digested genomic DNA and a 700-bp EcoRI-HindIII Hoxc6 flanking probe. Targeted lines were identified by the appearance of both 6.4-kb wild-type (wt)and 5.2-kb mutant hybridizing bands. (Note: The 6.4-kb band was more intense than the targeted 5.2-kb band due to the contamination of ES cell DNA with that of feeder fibroblasts). C: Genotypic identification by polymerase chain reaction (PCR). Gene-specific PCR primers were used to amplify Hoxc6 and neo sequences in mouse tail DNAs. The targeted allele produced a 298-bp PCR product (neo) and the wild-type allele produced a 134-bp PCR product (wt). (The position of PCR primers is indicated by black arrows in A). D: Immunohistochemistry of embryonic day 12.5 embryos (incubated with anti-Hoxc6 antibody as described in Experimental Procedures section). Hoxc6 homozygous embryos demonstrated a greatly reduced amount of truncated PRI protein with anterior boundaries of expression shifted posterior from pv7/8 to pv12 (arrows). in Hoxc6 homozygous animals, and these transcripts remain absent even subsequent to ovariectomy (Fig. 2). Taken together, these results argue that the Hoxc6 mutation is acting as a null in the mammary gland. Sharpe et al., 1988; Shimeld et al., 1993). Skeletons from Hoxc6 targeted mice were analyzed to study the effect of the mutation in patterning of the axial skeleton (Table 1). These results demonstrate a homeotic transformation of the second thoracic vertebra (T2) into the likeness of the first (T1) with approximately 60% penetrance in the Hoxc6 homozygous mice (Fig. 3; Table 1). This conversion was associated with the deletion or misplacement of the dorsal process or processus spinosus (approximately 90% penetrance), which was usually found on T3 instead of T2, suggesting that Targeted Disruption of Hoxc6 Causes a Homeotic Transformation of Thoracic Vertebra T2 Into T1 Hoxc6 is expressed in the thoracic region of the mouse embryo with its anterior boundary of expression at the level of the prevertebra 7/8 (C7/T1; Fig. 1D; 264 GARCIA-GASCA AND SPYROPOULOS Fig. 2. Hoxc6 expression in the mammary glands of wild-type and Hoxc6 homozygous females. Northern blot analysis involving a Hoxc6specific probe was used to measure Hoxc6-specific gene expression. Ovariectomized wild-type animals demonstrate increased levels of Hoxc6 gene expression (1/1, ovx). No mRNA expression was detected in the mammary tissues of either intact or ovariectomized (ovx) Hoxc6 homozygous females. TABLE 1. Skeletal Malformations Observed in Hoxc6 Homozygotesa Phenotype/genotype (%) Deletion of one sternebra Sternebrae of different lengths Short manubrium First and second ribs articulated, fused, and/ or in close appositionb Dorsal process deleted or translocated to T3 Different left-right symmetry 1/1 (n 5 20) 0 0 c6/1 (n 5 30) 10 10 c6/c6 (n 5 33) 33.3 33.3 0 0 10 0 57.6 57.6 0 70 90.6 0 3.3 27.3 a Skeletal defects showing more that 50% penetrance are in boldfaced type. b The penetrance of the T2 to T1 homeotic transformation in Hoxc6 homozygous mice was 57.6%. the T2 to T1 conversion involved the entire vertebra and not just the ribs (Fig. 3B; Table 1). We also observed that Hoxc6 homozygous females frequently lost some or all their pups after delivery (regardless of pup genotype). These pups lacked obvious craniofacial defects, such as cleft palate, but still had not nursed properly as indicated by the absence of milk in their stomachs (data not shown). For these reasons, a mammary gland deficiency in the homozygous mothers was assessed. Hoxc6 Homozygous Mice Have Abnormal Mammary Development Along the A-P Axis To analyze the role of Hoxc6 in the mammary gland, whole-mount and histologic specimens were prepared from mammary glands of wild-type, heterozygous, and Fig. 3. Skeletal phenotypes observed in Hoxc6 homozygous mice. A homeotic transformation of the second thoracic vertebra (T2) into the first (T1) was observed in approximately 60% of the Hoxc6 homozygotes (B). The dorsal process usually located on T2 was deleted or translocated to T3 in 90% of the mutant animals, suggesting a complete transformation of the vertebra. A: Wild-type. B: Hoxc6 homozygotes. Hoxc6 homozygous mice. These results revealed profound and distinct phenotypes in both thoracic and inguinal mammary glands of the Hoxc6 homozygous mice at 100% penetrance. Thoracic mammary glands of both intact and ovariectomized Hoxc6 homozygous females were devoid of mammary epithelium, as demonstrated by the absence of ductal structures (Figs. 4E, 6E,G). Although less distinct, thoracic ductal structures were also absent in Hoxc6 homozygous males (Fig. 5E). Adipose fat pad aberrations (characterized by the presence of immature adipose cells) were also observed in both thoracic and inguinal glands of the homozygous females (Fig. 6E–H). Less severely affected than thoracic glands, inguinal glands from both intact and ovariectomized Hoxc6 homozygous females developed mammary ductal structures that were dilated, variably oriented, and unresponsive to changing hormone conditions (Fig. 6F,H). Wild-type mammary epithelium undergoes regression or involution in response to ovariectomy (Fig. 6A–D). The dilated ducts observed in histologic sections of inguinal glands from Hoxc6 homozygotes failed to regress in response to ovariectomy (Fig. 6F,H). Histologic sections of inguinal glands harvested from homozygotes at a variety of points in the estrous cycle displayed no differences in this distended ductal phenotype (Figs. 6F,H, 7D–F, and data not shown). No normal ducts were observed and the abnormal (dilated) ducts observed varied greatly in size (Figs. 6F,H, 7D–F, and data not shown). Although MAMMARY DEFECTS IN Hoxc6 TARGETED MICE 265 Fig. 4. Whole-mount preparations of mammary glands from wild-type and Hoxc6 mutant females (8 –10 weeks old). Mammary epithelium is not observed in the thoracic glands of Hoxc6 homozygous females (E). In contrast, thoracic mammary epithelium develops normally in heterozy- gous females (C). Mammary epithelium appears normal in whole-mount preparations of inguinal mammary glands from Hoxc6 homozygous females. A,B: Wild-type. C,D: Hoxc6 heterozygotes. E,F: Hoxc6 homozygotes. these variably oriented and dilated ducts were sectioned in the same plane and readily apparent in all histologic sections (Figs. 6F,H, 7D–F), they were not readily apparent in whole-mount preparations (Fig. 4F). These results could be interpreted to mean that differences in the processing of specimens for wholemount and histologic sections (such as the use of xylene and elevated temperatures in the latter) accentuated differences in cell-cell and/or cell-extracellular matrix interactions in mutant and wild-type tissues. Normal ducts consist of two epithelial cell layers, an inner layer surrounding the lumen (luminal epithelium) and an outer layer of myoepithelium (Fig. 7A–C). At higher magnifications, disorganization of ductal epithelium was observed in dilated inguinal ducts from Hoxc6 homozygotes (Fig. 7E,F). Together, whole-mount and histologic analyses suggest a position-specific role for Hoxc6 in mammary gland development and adult function. Mammary Buds Are Observed in Hoxc6 Homozygous Embryos at Day E12.5 Adult thoracic glands of males and females that are devoid of epithelium may have arisen from a defect in embryonic development. A retrograde analysis of mam- 266 GARCIA-GASCA AND SPYROPOULOS Fig. 5. Whole-mount preparations of mammary glands from wild-type and Hoxc6 mutant males (8 –10 weeks old). Consistent with results from Hoxc6 homozygous females, mammary epithelium is not present in the thoracic glands of Hoxc6 homozygous males (E). No abnormalities were observed in whole-mount preparations of inguinal glands of the mutants. A,B: Wild-type. C,D: Hoxc6 heterozygotes. E,F: Hoxc6 homozygotes. mary development in the mutant was pursued to clarify the origins of the mutant phenotypes. Mammary buds are detectable in the mouse embryo at day E12.5 (Fig. 8). To determine whether the mammary defects observed in adult Hoxc6 homozygous mice originate at the initial mammary bud stage, hematoxylin and eosin stained transverse sections of E12.5 embryos were generated and analyzed (Fig. 8). In general, mammary buds from the Hoxc6 homozygotes and wild-types were indistinguishable. Figure 8 displays smaller and more protruding thoracic mammary buds that were occasionally observed in the Hoxc6 homozygotes (Fig. 8D,E). This thoracic bud phenotype may be associated with delayed development. Nevertheless, the low penetrance of this thoracic bud phenotype (Fig. 8D,E) and the high penetrance of the postnatal “cleared” thoracic phenotype (Figs. 4E, 5E) argue that early determination and proliferation of mammary epithelium occur normally. Thoracic Mammary Epithelium Is Reduced in Hoxc6 Homozygous Newborn Females In the newborn female mouse, the mammary gland consists of mammary epithelium with limited ductal branching, all of which is embedded in adipose and fibrous connective (stromal) tissue. Hematoxylin and MAMMARY DEFECTS IN Hoxc6 TARGETED MICE 267 Fig. 6. Histology of mammary glands from intact and ovariectomized females. Hoxc6 homozygotes do not develop mammary ducts in the thoracic glands (E,G), and develop abnormal dilated ducts in the inguinal glands (F,H). Wild-type ovariectomized females demonstrate a great reduction in ductal number but not the complete absence that is observed in the thoracic glands of Hoxc6 homozygotes (C,D). Ductal regression, normally observed after ovariectomy (compare B and D), does not occur in inguinal glands of homozygotes (compare F and H). Du, ducts; AC, adipose connective tissue; FC, fibrous connective tissue; BV, blood vessels. Hematoxylin and eosin stain; 3100. eosin stained transverse sections of newborn females were generated and analyzed to determine the condition of mammary epithelium at the time of birth in mutant Hoxc6 females (Fig. 9). The nipple, nipple sheath, branching ducts, and fat pads were detected in the thoracic glands of homozygotes (Fig. 9E), but branching, ductal structures, and fat pad development were reduced compared with their heterozygous littermates (Fig. 9A,B,D,E). Primary ducts, which are derived by means of proliferation of mammary sprout anlagen, were detected in thoracic glands of all homozygous newborns examined (Fig. 9E). In contrast, the inguinal epithelium of both heterozygotes and homozygotes were indistinguishable (Fig. 9C,F,G). These Fig. 7. Dilated ductal morphology in cross-sections of inguinal mammary glands from Hoxc6 homozygous females. Comparison of crosssectioned inguinal ducts from wild-type (A–C) and Hoxc6 homozygous (D–F) females at various magnifications. A and D are lower magnification images. B and E, and C and F are two sets of higher magnification images. Abnormal ductal dilation is observed in all histologic sections of mammary tissue from inguinal glands of Hoxc6 homozygous females (D–F). M, myoepithelial cells; E, epithelial cells; L, lumen; Bv, blood vessels. A,D, 363; B,E, 3290; C,F, 3580. Fig. 8. Mammary buds in embryonic day (E) 12.5 embryos. Transverse sections of wild-type (A–C) and Hoxc6 homozygous (D–F) E12.5 embryos were generated and stained with hematoxylin and eosin. The left and right thoracic mammary buds from individual wild-type (A,B) and homozygous (D,E) embryos are shown. As depicted in D and E, thoracic mammary buds of Hoxc6 homozygous embryos occasionally appeared smaller and more protruding than their wild-type counterparts. In general, thoracic and inguinal mammary buds of Hoxc6 homozygotes were indistinguishable from those of wild-types. The heart (h) is labeled in thoracic sections to define position and orientation. 3200. MAMMARY DEFECTS IN Hoxc6 TARGETED MICE Fig. 9. Mammary epithelium in newborn females. In Hoxc6 homozygous newborn females (D,E), nipples and primary structures appear intact in thoracic glands; however, branching ductal structures and fat pads are reduced compared with their heterozygous counterparts (A,B). In some sections, only the empty fat pad is observed (D). Normal nipples 269 and ductal structures are observed in inguinal mammary glands from both heterozygous (C) and homozygous (F,G) females. A, B, D, and E, 2nd thoracic glands (#3 and #8); C, F, and G, 1st inguinal glands (#4 and #9). FP, fat pad. Serial sections are shown to confirm the presence or absence of mammary structures (A and B, D and E, F and G). 3100. Fig. 10. Hoxc6 expression in mammary glands of wild-type intact females. In situ hybridization using a Hoxc6 antisense probe demonstrates Hoxc6 mRNA expression primarily within the myoepithelium and secondarily within the luminal epithelium of ducts of both thoracic (A) and inguinal (C) glands. A,C: Hoxc6 antisense probe. B,D: Hoxc6 sense probe. 3167. Fig. 11. Hoxc6 expression in mammary glands of wild-type ovariectomized females. In situ hybridization using a Hoxc6 antisense probe demonstrates Hoxc6 mRNA expression primarily within the stroma (including adipose and fibrous tissues) of both thoracic (A,B) and inguinal (D,E) mammary glands of ovariectomized females. Expression in the remaining ducts was found to persist at reduced levels. A,B,D,E: Hoxc6 antisense probe. C,F: Hoxc6 sense probe. 3170. MAMMARY DEFECTS IN Hoxc6 TARGETED MICE 271 TABLE 2. Average Concentrations of Estrogen (E2) and Progesterone (P) in Serum of Mice at the Time of Sacrificea Animals/treatment Intact untreated Ovx untreated Ovx E2 treated Ovx P treated Ovx E2 1 P treated a E2 concentration (pg/ml) 35 6 14 863 21 6 6 7.6 6 3 19.8 6 5 P concentration (ng/ml) 964 3 6 0.6 2.9 6 0.9 12.5 6 5 8.8 6 1 n 5 5 for each treatment. results demonstrate that the thoracic mammary epithelium and stroma do form primary structures that fail to expand extensively by birth. These results also demonstrate that the embryonic phenotype is relatively subtle when compared with the profound “cleared” phenotype in the thoracic mammary glands of adults (Figs. 4E, 5E). Normal Hoxc6 Expression in the Mammary Gland Is Disrupted After Ovariectomy It has been previously shown that Hoxc6 is expressed in the mammary gland of adult female mice (Friedmann et al., 1994). We confirmed these observations by using Northern blot analysis (Fig. 2). To determine the spatial localization of Hoxc6 transcripts in the mammary gland before and after ovariectomy, in situ hybridization was performed (Figs. 10, 11). These results demonstrate that, in intact females, Hoxc6 is expressed primarily within the myoepithelium and secondarily within the luminal epithelium of ducts (Fig. 10A,C). In ovariectomized females, Hoxc6 expression within both thoracic and inguinal mammary glands is found primarily within the stroma (fibrous and adipose connective tissues) and secondarily within residual ductal epithelium (Fig. 11A, B, D, E). Together, these results suggest that ovarian hormones may be necessary to regulate both the level and spatial pattern of Hoxc6 expression. Estrogen (E2) and Progesterone (P) Are Two Ovarian Hormones Involved in the Downregulation of Hoxc6 Expression in Mammary Gland Stroma To determine whether Hoxc6 gene expression responds to specific ovarian steroid hormones, wild-type females were ovariectomized and treated with estrogen, progesterone, or both. As controls, the concentrations of hormones in the systemic circulation were measured by radioimmunoassay (Table 2). Both E2 and P, individually or in combination, were able to decrease the expression of Hoxc6 (Fig. 12). These results indicate that E2 and P are two significant ovarian hormones involved in the down-regulation of Hoxc6 expression in mammary stroma. Furthermore, because the concentrations of E2 and P used were near physiologic levels and the corresponding decrease in Hoxc6 Fig. 12. Ovarian hormones estrogen and progesterone repress Hoxc6 expression. Messenger RNA from mammary tissues of operated females was isolated and subjected to reverse transcriptase polymerase chain reaction analysis. Expression of the Hoxc6 transcript in the mammary glands of wild-type females is up-regulated after ovariectomy, and is down-regulated when ovariectomized females are treated with estrogen and progesterone. Ovx, ovariectomized; E2, b-estradiol; P, progesterone. n 5 5 for each group. expression was dramatic, it is possible that E2 and P are the primary ovarian hormones involved in Hoxc6 regulation in the stroma. DISCUSSION There is increasing evidence that Hox genes are involved in normal mammary gland development and adult functions (Friedmann et al., 1994; Srebrow et al., 1998; Chen and Capecchi, 1999). Here, we demonstrate roles for Hoxc6 in both embryonic development and adult functions. Mammary development initiates properly in Hoxc6 homozygotes with the formation of the mammary bud. By birth, the thoracic primary duct and the nipple sheath appear normal; however, fat pads are reduced in size and the mammary tissues are defective for branching of ductal structures. In adults, no epithelium at all is detected in the thoracic glands (“cleared” phenotype). From this, it is possible that mammary defects initiate with reduced proliferation and/or increased apoptosis late in gestation and become more extreme in adulthood with the regression of primary structures that had been established by birth. In the less extreme inguinal mammary phenotype, proliferation and branching occur to form the fully developed ductal tree; however, ducts are abnormal (dilated) and unresponsive to ovariectomy. Thus, mammary defects are more severe in the thoracic region at the level of the gene’s anterior boundary and highest level of expression. The observed homeosis of the axial skeleton in the thoracic region was consistent with the severity of mammary defects observed along the A-P axis. Hoxc5 272 GARCIA-GASCA AND SPYROPOULOS expression is perturbed but likely not to contribute to the observed phenotype. The perturbed expression pattern did not alter the expression level in the thoracic region. Instead, it mainly resulted in ectopic expression in the cervical region (data not shown). One interpretation of the results presented is that the thoracic skeletal and neonatal mammary phenotypes are due to the loss of homeotic Hoxc6 function in the thoracic region during embryogenesis, and that the later regression-defective inguinal gland phenotype is due to the loss of nonhomeotic Hoxc6 function in adult mammary expansion and regression. Along these lines, the “cleared” thoracic mammary phenotype in adults (Figs. 4E, 5E) would be due to the loss of both homeotic and nonhomeotic Hoxc6 functions. Regional differences in Hoxc6 expression during embryogenesis (Fig. 1D) and the absence of regional differences in Hoxc6 expression in adult glands (Fig. 10) support this interpretation, as does the correspondence between homeotic skeletal and mammary gland defects observed in the thoracic region (Figs. 3, 9). Hoxc6 is expressed in the epithelium of both thoracic and inguinal mammary glands. We demonstrated that the up-regulation of Hoxc6 gene expression observed after ovariectomy is the result of expression within fibrous and adipose tissues of the mammary stroma. Potentially, estrogen and progesterone are major ovarian hormones involved in this repression of Hoxc6 gene expression. These results suggest that the control of adult mammary expansion and regression may, in part, be controlled by differential hormone regulation of Hoxc6 expression in epithelia and stroma (Figs. 10, 11). Sex steroids have been demonstrated previously to regulate homeobox-containing genes. For example, Msx-2 expression, normally found in mammary epithelial cells, is down-regulated after ovariectomy (Friedmann and Daniel, 1996) and is induced by estrogen in MCF7 cells (Phippard et al., 1996). In other cases, treatment with estrogen and progesterone can increase the expression of Hoxa11 in the uterus (Taylor et al., 1999), whereas progestins are able to induce Hoxa1 transcripts in human mammary cells (Chariot and Castronovo, 1996). Our results demonstrate a role for Hoxc6 within both epithelial and stromal tissues of the mammary gland. Collectively, these results suggest a model for Hoxc6 function in adult mammary expansion and involution (Fig. 13). In this model, estrogen (E2) and progesterone (P) normally function to block Hoxc6 expression within the stroma and, thereby, prevent stromal-Hoxc6 mediated ductal regression (Fig. 13A). In the absence of E2 and P (e.g. ovariectomy), Hoxc6 expression within the stroma is derepressed and upregulated stromal Hoxc6 causes regression of the ductal epithelium, presumably through stromal-epithelial interactions (Fig. 13B). In knockout mice, Hoxc6 function is lost within both the epithelium and the stroma of the mammary gland. There is no longer a negative mediator of the effects of ovarian hormones; thus, ductal structures remain invariant regardless of whether Fig. 13. Model for Hoxc6-mediated hormone regulation of inguinal mammary duct expansion/regression. A: In intact females, estrogen and progesterone repress stromal but not ductal Hoxc6 expression, resulting in epithelial cell proliferation and ductal tree expansion. B: In ovariectomized females, Hoxc6 expression is derepressed in the stroma resulting in ductal regression. C,D: In the absence of Hoxc6, there is no longer a negative mediator of the effects of ovarian hormones; thus, dilated ducts develop regardless of whether the female is intact or ovariectomized. the female is intact or ovariectomized (Fig. 13C,D). In normal cycling females the oscillating levels of ovarian hormones are probably reflected by oscillating levels of Hoxc6 expression within the stroma and, thereby, more subtly regulate ductal expansion/regression (Friedmann et al., 1994). Mammary epithelial-stromal reconstitution experiments involving Hoxc6 homozygous and wild-type tissues are currently being used to test this model. The regulation of Hoxc6 gene expression by estrogen and progesterone may occur directly through steroid hormone binding of regulatory elements. A computer search using the GCG program Findpatterns and a consensus sequence for the estrogen response element (ERE) 59-GGTCAnnTGACC-39 (Augereau et al., 1994), detected a potential ERE binding site (59-GGTCAGCTGAC-39) in the proximal promoter (PRII; see Fig. 2) of Hoxc6, at position 1385, according to the sequence reported by Coletta and co-workers (1991). Despite the high variability of EREs in different genes, these re- MAMMARY DEFECTS IN Hoxc6 TARGETED MICE sults maintain the possibility that the regulation of Hoxc6 gene expression by steroid hormones is direct. Functional experiments involving the mutation of this element may resolve this issue. Ductal morphology is dependent on the extracellular matrix within and surrounding the ducts and the proliferative state of ductal cells. Ductal development and branching involve cellular proliferation and the production of extracellular matrix within the stroma (Bissell and Hall, 1987). In contrast, ductal regression involves basement membrane degradation by proteases and programmed cell death (Strange et al., 1992; Boudreau et al., 1995). Hoxc6 transcripts are localized within both the ductal epithelium and the stroma. In vitro evidence exists for the regulation of extracellular matrix and cell surface molecules by Hoxc6 (Jones et al., 1993; Shimeld and Sharpe, 1992). The abnormal ductal structure observed in the inguinal glands of Hoxc6 homozygotes may be the result of loss of Hoxc6 function in either or both of these structures. The Hoxc6 thoracic mammary phenotype is similar to, yet less profound, than that observed in partially rescued parathyroid hormone-related protein (PTHrP) knockout mice (Wysolmerski et al., 1998). PTHrP knockout mice die at birth from impaired endochondral bone development and inappropriate ossification of the coastal cartilage (Karaplis et al., 1994; Lanske et al., 1996). In PTHrP knockout mice that are rescued by transgenic expression of PTHrP in chondrocytes (ColII-PTHrP), the mammary buds develop, but epithelial cells fail to subsequently proliferate (Wysolmerski et al., 1998). Thus, it was demonstrated that PTHrP is necessary for the transformation of mammary buds into a branching ductal system. Our results indicate that homeotic Hoxc6 function is less crucial but may function downstream of PTHrP in this embryonic process. Further experiments are required to determine whether PTHrP and Hoxc6 function within the same pathway. In summary, we have demonstrated that Hoxc6 is necessary for normal mammary gland morphogenesis, especially in ductal growth late in gestation and ductal expansion and regression in the adult. In the latter process, the degree to which estrogen and progesterone repress stromal Hoxc6 expression may be a determinant in the expansion or regression of adult mammary epithelium. EXPERIMENTAL PROCEDURES Generation of Hoxc6 Targeted Mice The targeting vector was constructed using a 12.8-kb EcoRI-BamHI fragment of nonisogenic (C57Bl/6J) mouse genomic DNA encompassing Hoxc5 and Hoxc6 coding exons. The pMC1neo cassette was introduced into the BglII site of the Hoxc6 homeobox (Fig. 1A). The resulting construct was then inserted between HSV TK1 and TK2 thymidine kinase genes in the final targeting vector construct (not shown). The targeting vector was linearized with PvuI and introduced into CC1.2 (129/Sv) male ES cells by electroporation. The cells 273 were then subjected to positive and negative selection using G418 (230 mg/ml) and FIAU (1.25 mM), respectively. Targeted cell lines were identified by Southern blot, probing HindIII digested genomic DNA with a 700-bp EcoRI-HindIII Hoxc6 flanking probe (Fig. 1A). Targeted cell lines were identified by a hybridizing DNA band downshift from 6.4 kb (wild-type) to 5.2 kb (Fig. 1B). Chimeric mice were generated by injecting targeted ES cells into C57Bl/6J (B6) blastocyst, and chimeras were bred to B6 females to identify germline chimeras. The first filial generation heterozygotes were selected by the Agouti coat color and identified by PCR (Fig. 1C). DNA Preparation From Tails and PCR Tails were collected into Eppendorf tubes containing 400 ml of tail lysis buffer (100 mM NaCl, 50 mM Tris, pH 8.0, 100 mM EDTA, 1% SDS) and 20 ml of proteinase K (20 mg/mL), and incubated overnight at 56°C. Two hundred microliters of saturated NaCl (;6 M) was added, tubes were centrifuged, and the supernatant was transferred into tubes containing two volumes of absolute ethanol. The DNA from each tube was spooled and transferred into a new tube containing 100 ml of TE (10 mM Tris, 1 mM EDTA). Before PCR amplification, 7.5 ml of DNA in TE was mixed with 50 ml of PCR lysis buffer (50 mM KCl, 1.5 mM MgCl2, 10 mM Tris pH 8.5, 0.01% gelatin, 0.45% NP-40, 0.45% Tween-20), and heated at 95°C for 5 min. The PCR primers for Hoxc6, and neo were: FOR Hox3.3 59-GTC GGT TAC GGA GCG GAC CGG AG-39; REV Hox3.3 59-CAC AGA GCA TTG GCG ATC TCG ATG C-39; Rev neo263 59CGT GTT CGA ATT CGC CAA TGA CAA GCC-39. The cycle setup was: 94°C 1 min, 63°C 1 min, and 72°C 30 sec, for 35 cycles. PCR products were visualized on a 2% treva-gel (Trevigen). RNA isolation RNA was isolated from E12.5 embryos and mammary glands using the RNA STAT-60 method, following the instructions of the manufacturer (Tel-Test, Inc., Texas). Briefly, the tissue was homogenized in RNA STAT-60 and one-fifth volume of chloroform was added. Samples were mixed and then centrifuged for 15 min. The aqueous phase was transferred to a new tube. RNA was precipitated with isopropanol and washed with 75% ethanol. The pellet was resuspended in DEPC water and the RNA was treated with DNAse I before Northern blot and RT-PCR analyses. Northern blot Poly(A)1 mRNA was purified from total RNA using oligotex suspension following the instructions of the manufacturer (Qiagen, CA). Poly(A)1 mRNA (2 mg) was subjected to electrophoresis on a 1.2% agarose/ formaldehyde gel in MOPS buffer. The RNA was transferred to a Duralon-UV membrane (Stratagene, CA) and UV crosslinked with a Stratalinker (Stratagene). A 415-bp 59-coding region probe specific for Hoxc6 was 274 GARCIA-GASCA AND SPYROPOULOS generated by PCR, a 506-bp probe specific for the PRI exon of Hoxc6 was a gift from Dr. Paul Sharpe (Shimeld et al., 1993), and a 39 UTR probe specific for Hoxc8 was a gift from Dr. Alexander Awgulewitsch (Awgulewitsch et al., 1990). Radioactive probes were generated by random priming by using 32P (activity . 108 cpm/mg). The nylon membrane was prehybridized for 1 to 2 hr at 65°C with Quick-hyb solution (Stratagene, CA), and hybridized 3 to 4 hr at 65°C. High stringency washes were carried out in 0.23 SSPE, 0.1% SDS at 65°C. The membrane was exposed for 3 to 4 days using a Kodak MR intensifying screen. The membranes were stripped in 0.013 SSPE, 0.1% SDS and reprobed with Hoxc8 or b-actin as controls for RNA integrity. Skeletal Analysis Skeletons were prepared according to Kostic and Capecchi (1994). The animals were killed without damaging the skeletons. After removing the skin and other soft tissues, the skeletons were immersed in 95% ethanol for 5 days, transferred to acetone for 2 days, and then transferred to a solution of 5% acetic acid, 75% ethanol, 5% alizarin red, and 15% Alcian blue for 10 days. The skeletons were rinsed in distilled water and incubated in a solution of 1% trypsin, 30% sodium borate at 37°C for 6 –18 hr. Skeletons were transferred into a 1% KOH, 20% glycerol solution and finally transferred into 100% glycerol for storage and analysis. Histology Conventional histology and hematoxylin-eosin (H&E) staining were performed as described (Humason, 1979). E12.5 embryos, newborns, and the mammary glands from intact and ovariectomized females were dissected and fixed in PBS containing 4% paraformaldehyde overnight at 4°C. Tissues were washed in PBS, dehydrated, and embedded in paraffin. Eight micron sections were rehydrated, hematoxylin and eosin stained, mounted, and photographed with a Kodak Digital Science DC 120 zoom digital camera. Whole Embryo Immunohistochemistry Immunohistochemistry was performed as described by Oliver et al. (1988), and Boulet and Capecchi (1996) with slight modifications. Embryos were dissected in PBS, fixed overnight in 20% DMSO in methanol, bleached in 10% hydrogen peroxide in methanol for at least 6 hr, and stored in methanol at 220°C until use. Embryos were then rehydrated through decreasing concentrations of methanol in PBST (PBS 1 0.5% Triton-X 100) 30 min each, washed 3 3 30 min in PBST, and blocked 2.5 hr in PBSTMD (PBST, 1% DMSO, 10% goat serum). Embryos were incubated with anti-Hoxc6 antibody (antibody to the N-terminal arm of the homeodomain; Babco, CA) 1:100 in PBSTMD at 4°C overnight. After 6, 1-hr washes in PBST, embryos were incubated with goat anti-rabbit IgG-HRP 1:1,000 in PBSTMD at 4°C overnight. After 6, 1-hr washes in PBST, the signal was visualized by using DAB perox- idase substrate (Vector Laboratories, CA). Embryos were then dehydrated in absolute methanol, cleared in 1:2 benzyl alcohol/benzyl benzoate, and photographed with a Kodak Digital Science DC 120 zoom digital camera. Cotransfection Experiments and Reporter Gene Assays NIH3T3, HeLa, and MCF7 cells were cotransfected in triplicate with the coding regions of wild-type or mutant Hoxc6 genes in pSG5 expression vectors using Effectene transfection reagent following the instructions of the manufacturer (Qiagen, CA). The pTCBS reporter plasmid used contains an eight-fold homeodomain consensus-binding sequence upstream of the luciferase reporter gene (a gift from Dr. Vicenzo Zappavigna; Zappavigna et al., 1994). Either wild-type or mutant constructs (0.2 mg) and pTCBS (1.6 mg) were cotransfected into cells. DNA concentrations were maintained constant by the addition of pBS-KS plasmid (Stratagene). The pRL-CMV reporter plasmid (5 ng; Promega, WI) was used as an internal control. Cells were harvested 24 hr after transfection and luciferase activity was measured by using the Dual-Luciferase Reporter Assay System following the instructions of the manufacturer (Promega). Whole-Mount Preparations of Mammary Glands Thoracic and inguinal mammary glands were prepared as described in the NIH mammary web site: mammary.nih.gov. Glands were dissected and fixed for 2 to 4 hr in Carnoy’s fixative (ethanol:chloroform:acetic acid 6:3:1), washed 15 min in 70% ethanol, washed 5 min in distilled water, and stained in carmin alum overnight. The glands were washed for 15 min each in 70%, 95%, and 100% ethanol, cleared in Hemo-D (Sigma, MO) for 1 hr, mounted, and photographed with a Kodak Digital Science DC 120 zoom digital camera. In Situ Hybridization In situ hybridization was performed according to Wilkinson (1992). Signal detection was achieved using the TSA system following the instructions of the manufacturer (NEN Life Science Products, MA). Mammary glands from intact and ovariectomized wild-type females were dissected and fixed overnight in PBS containing 4% paraformaldehyde. Dehydration was carried out in increasing concentrations of ethanol/saline solution, absolute ethanol, and finally in xylene. Tissues were embedded in paraffin, and 8-micron sections were generated. The sections were rehydrated, treated with 1% hydrogen peroxide in PBS for 15 min, incubated in PBS containing 30 mg/ml proteinase-K for 12 min at 37°C, postfixed in PBS containing 4% paraformaldehyde for 20 min, acetylated in 1 mM triethanolamine, 0.25% acetic anhydride for 10 min, dehydrated, and hybridized overnight at 42°C in hybridization solution containing 50% formamide, 43SSC, 10% Dextran sulfate, 13 Denhardt’s, 0.5 mg/ml yeast tRNA, and 800 –1,000 ng/ml biotin-labeled ribo- MAMMARY DEFECTS IN Hoxc6 TARGETED MICE probe. A 400-bp EcoRI fragment of Hoxc6 was used as probe. The probe-containing plasmid was linearized with BamHI and RNA was synthesized in vitro using a biotin labeling kit and T3 RNA polymerase (NEN Life Science Products). Subsequent to hybridization, the first wash was performed at 42°C in 43SSC, 50% formamide for 30 min, then at 37°C in 23SSC and 13SSC each for 30 min, followed by incubation with 10 mg/ml RNAse A at 37°C for 30 min. Sections were blocked 30 min in blocking buffer, followed by incubation with SA-HRP for 30 min (NEN Life Science Products). The signal was amplified using TSA-Indirect amplification kit (NEN Life Science Products) and developed with DAB (Vector Laboratories). Slides were counterstained with methyl green, mounted, and photographed with a Kodak Digital Science DC 120 zoom digital camera. Hormone Treatment Twenty adult (8 –10 weeks old) CD-1 females were ovariectomized using survival surgery procedures. Four weeks later, 5 mg of 17-b-estradiol (Sigma) in PBS, 1 mg of progesterone (Sigma) in PBS, b-estradiol and progesterone together in PBS, or PBS alone were injected intraperitoneally each day for 20 days. Twenty-four hours after the final injection, the mice were anesthetized and 0.3 to 0.5 ml of blood were withdrawn from the eye sinus to measure the concentration of estrogen and progesterone using a radioimmunoassay kit following the instructions of the manufacturer (Diasorin, Inc., MN). RNA was then isolated from the mammary gland and analyzed by RT-PCR. Ovariectomy and treatments were performed as described (Friedmann and Daniel, 1996; Rajah et al., 1996; Hilakivi-Clarke et al., 1997; Humphreys et al., 1997; Rajan et al., 1997; Zysow et al., 1997). RT-PCR cDNA synthesis was performed at 52°C using 5 mg of total RNA, AMV (Invitrogen, CA), and M-MuLV (New England Biolabs, MA) reverse transcriptases, and oligo(dT). PCR amplification was then carried out using a Perkin-Elmer thermal-cycler, and Hoxc6 primers flanking the intron: RTC6F 59-GAG AAT GTC GTG TTC AGT TCC AGC-39 RTC6R 59-AAG TGA AAT TCC TTC TCC AGT TCC-39 PCR conditions were: 1 min at 94°C, 1 min at 65°C, and 30 sec at 72°C. PCR amplifications were performed for 30 cycles. b-Actin was used to normalize mRNA levels. The primers used for b-actin amplification were those described previously (Mansour et al., 1993). Animals The Hoxc6 targeted mouse strain was bred into the B6 inbred genetic background and used for the analyses described. For mammary gland analyses, virgin mice were selected randomly from different cages to obtain different stages of the estrous cycle. The mice were kept under standard conditions according to the NIH Guide for the Care and Use of Laboratory Animals. All procedures in- 275 volving mice in this work were approved by the AAALAC accredited institutional review board. ACKNOWLEDGMENTS We thank Dr. A. Darden, Dr. Y. Gong, M. Allen, S. Barnett, C. Lenz, E. Nakashima, and S. Tamowski for technical assistance. We also thank Drs. I. Maroulakou, T. Hsu, and A. Awgulewitsch for critical reading of this manuscript. A.G.-G. was supported by a FulbrightCONACyT graduate fellowship. REFERENCES Augereau P, Miralles F, Cavailles V, Gaudelet C, Parker M, Rochefort H. 1994. Characterization of the proximal estrogen-responsive element of human cathepsin D gene. Mol Endocrinol 8:693– 703. Awgulewitsch A, Bieberich C, Bogarad L, Shashikant C, Ruddle FH. 1990. Structural analysis of the Hox-3.1 transcription unit and the Hox-3.2— Hox-3.1 intergenic region. Proc Natl Acad Sci USA 87: 6428 – 6432. Bissell MJ, Hall HG. 1987. Form and function in the mammary gland. The role of extracellular matrix. In: MC Neville, CW Daniel, eds. The mammary gland: development, regulation and function, New York: Plenum Press. p 97–145. Boudreau N, Sympson CJ, Werb Z, Bissell MJ. 1995. Suppression of ICE and apoptosis in mammary epithelial cells by extracellular matrix. Science 267:891– 893. Boulet AM, Capecchi MR. 1996. Targeted disruption of hoxc-4 causes esophageal defects and vertebral transformations. Dev Biol 177: 232–249. Castronovo V, Kusaka M, Chariot A, Gielen J, Sobel M. 1994. Homeobox genes: potential candidates for the transcriptional control of the transformed and invasive phenotype. Biochem Pharmacol 47: 137–143. Chariot A, Castronovo V. 1996. Detection of HOXA1 expression in human breast cancer. Biochem Biophys Res Commun 222:292–297. Chariot A, Gielen J. 1998. The HOXC6 homeodomain-containing proteins. Int J Biochem Cell Biol 30:651– 655. Chariot A, Castronovo V, Le P, Gillet C, Sobel ME, Gielen J. 1996. Cloning and expression of a new HOXC6 transcript encoding a repressing protein. Biochem J 319(Pt 1):91–97. Chen F, Capecchi MR. 1999. Paralogous mouse Hox genes, Hoxa9, Hoxb9, and Hoxd9, function together to control development of the mammary gland in response to pregnancy. Proc Natl Acad Sci USA 96:541–546. Cho KWY, Goetz J, Wright CVE, Fritz A, Hardwicke J, De Robertis EM. 1988. Differential utilization of the same reading frame in a Xenopus homeobox gene encodes two related proteins sharing the same DNA-binding specificity. EMBO J 7:2139 –2149. Coletta PL, Shimeld SM, Chaudhuri C, Muller U, Clarke JP, Sharpe PT. 1991. Characterisation of the murine Hox-3.3 gene and its promoter. Mech Dev 35:129 –142. Condie BG, Capecchi MR. 1994. Mice with targeted disruptions in the paralogous genes hoxa-3 and hoxd- 3 reveal synergistic interactions. Nature 370:304 –307. Daniel CW, Silberstein GB. 1987. Postnatal development of the rodent mammary gland. In: MC Neville CW Daniel, eds. The mammary gland: develoment, regulation, and function. New York: Plenum Press. p 3–36. Friedmann Y, Daniel CW. 1996. Regulated expression of homeobox genes Msx-1 and Msx-2 in mouse mammary gland development suggests a role in hormone action and epithelial- stromal interactions. Dev Biol 177:347–355. Friedmann Y, Daniel CA, Strickland P, Daniel CW. 1994. Hox genes in normal and neoplastic mouse mammary gland. Cancer Res 54: 5981–5985. Haslam SZ. 1987. Role of sex steroid hormones in mammary gland function. In: MC Neville, CW Daniel, eds. The mammary gland: 276 GARCIA-GASCA AND SPYROPOULOS development, regulation, and function. New York: Plenum Press. p 499 –533. Hilakivi-Clarke L, Cho E, Raygada M, Kenney N. 1997. Alterations in mammary gland development following neonatal exposure to estradiol, transforming growth factor alpha, and estrogen receptor antagonist ICI 182,780. J Cell Physiol 170:279 –289. Horan GS, Kovacs EN, Behringer RR, Featherstone MS. 1995a. Mutations in paralogous Hox genes result in overlapping homeotic transformations of the axial skeleton: evidence for unique and redundant function. Dev Biol 169:359 –372. Horan GS, Ramirez-Solis R, Featherstone MS, Wolgemuth DJ, Bradley A, Behringer RR. 1995b. Compound mutants for the paralogous hoxa-4, hoxb-4, and hoxd-4 genes show more complete homeotic transformations and a dose-dependent increase in the number of vertebrae transformed. Genes Dev 9:1667–1677. Humason GL. 1979. Animal tissue techniques. San Francisco: WH Freeman & Company. p 692. Humphreys RC, Lydon J, O’Malley BW, Rosen JM. 1997. Mammary gland development is mediated by both stromal and epithelial progesterone receptors. Mol Endocrinol 11:801– 811. Imagawa W, Yang J, Guzman R, Nandi S. 1994. Control of mammary gland development. In: EN Knobil, JD Neill, eds. The physiology of reproduction. New York: Raven Press, Ltd. p 1033–1063. Jones FS, Holst BD, Minowa O, De Robertis EM, Edelman GM. 1993. Binding and transcriptional activation of the promoter for the neural cell adhesion molecule by HoxC6 (Hox-3.3). Proc Natl Acad Sci USA 90:6557– 6561. Karaplis AC, Luz A, Glowacki J, Bronson RT, Tybulewicz VL, Kronenberg HM, Mulligan RC. 1994. Lethal skeletal dysplasia from targeted disruption of the parathyroid hormone-related peptide gene. Genes Dev 8:277–289. Kostic D, Capecchi MR. 1994. Targeted disruptions of the murine Hoxa-4 and Hoxa-6 genes result in homeotic transformations of components of the vertebral column. Mech Dev 46:231–247. Lanske B, Karaplis AC, Lee K, Luz A, Vortkamp A, Pirro A, Karperien M, Defize LHK, Ho C, Mulligan RC, Abou-Samra AB, Juppner H, Segre GV, Kronenberg HM. 1996. PTH/PTHrP receptor in early development and Indian hedgehog-regulated bone growth. Science 273:663– 666. Mansour SL, Goddard JM, Capecchi MR. 1993. Mice homozygous for a targeted disruption of the proto-oncogene int-2 have developmental defects in the tail and inner ear. Development 117:13–28. Mavilio F. 1993. Regulation of vertebrate homeobox-containing genes by morphogens. Eur J Biochem 212:273–288. Oliver G, Wright CV, Hardwicke J, De Robertis EM. 1988. A gradient of homeodomain protein in developing forelimbs of Xenopus and mouse embryos. Cell 55:1017–1024. Phippard DJ, Weber-Hall SJ, Sharpe PT, Naylor MS, Jayatalake H, Maas R, Woo I, Roberts-Clark D, Francis-West PH, Liu YH, Maxson R, Hill RE, Dale TC. 1996. Regulation of Msx-1, Msx-2, Bmp-2, and Bmp-4 during foetal and postnatal mammary gland development. Development 122:2729 –2737. Rajah TT, Dunn ST, Pento JT. 1996. The influence of antiestrogens on pS2 and cathepsin D mRNA induction in MCF-7 breast cancer cells. Anticancer Res 16:837– 842. Rajan JV, Marquis ST, Gardner HP, Chodosh LA. 1997. Developmental expression of Brca2 colocalizes with Brca1 and is associated with proliferation and differentiation in multiple tissues. Dev Biol 184: 385– 401. Rancourt DE, Tsuzuki T, Capecchi MR. 1995. Genetic interaction between hoxb-5 and hoxb-6 is revealed by nonallelic noncomplementation. Genes Dev 9:108 –122. Sakakura T. 1987. Mammary embryogenesis. . In: MC Neville, CW Daniel, eds. The mammary gland: development, regulation, and function. New York: Plenum Press. p 37– 66. Sharpe PT, Miller JR, Evans EP, Burtenshaw MD, Gaunt SJ. 1988. Isolation and expression of a new mouse homeobox gene. Development 102:397– 407. Shimeld SM, Sharpe PT. 1992. Cell-surface changes induced by ectopic expression of the murine homeobox gene Hox-3.3. Biochim Biophys Acta 1136:253–258. Shimeld SM, Gaunt SJ, Coletta PL, Geada AM, Sharpe PT. 1993. Spatial localisation of transcripts of the Hox-C6 gene. J Anat 183(Pt 3):515–523. Simeone A, Pannese M, Acampora D, D’Esposito M, Boncinelli E. 1988. At least three human homeoboxes on chromosome 12 belong to the same transcription unit. Nucleic Acids Res 16:5379 –5390. Srebrow A, Friedmann Y, Ravanpay A, Daniel CW, Bissell MJ. 1998. Expression of Hoxa-1 and Hoxb-7 is regulated by extracellular matrix- dependent signals in mammary epithelial cells. J Cell Biochem 69:377–391. Strange R, Li F, Saurer S, Burkhardt A, Friis RR. 1992. Apoptotic cell death and tissue remodelling during mouse mammary gland involution. Development 115:49 –58. Taylor HS, Igarashi P, Olive DL, Arici A. 1999. Sex steroids mediate HOXA11 expression in the human peri-implantation endometrium. J Clin Endocrinol Metab 84:1129 –1135. Tickle C, Altabef M. 1999. Epithelial cell movements and interactions in limb, neural crest and vasculature. Curr Opin Genet Dev 9:455– 460. Wilkinson DG. 1992. In situ hybridization: a practical approach. New York: IRL Press at Oxford University Press. p 163. Wysolmerski JJ, Philbrick WM, Dunbar ME, Lanske B, Kronenberg H, Broadus AE. 1998. Rescue of the parathyroid hormone-related protein knockout mouse demonstrates that parathyroid hormonerelated protein is essential for mammary gland development. Development 125:1285–1294. Zappavigna V, Sartori D, Mavilio F. 1994. Specificity of HOX protein function depends on DNA-protein and protein- protein interactions, both mediated by the homeo domain. Genes Dev 8:732–744. Zysow BR, Kauser K, Lawn RM, Rubanyi GM. 1997. Effects of estrus cycle, ovariectomy, and treatment with estrogen, tamoxifen, and progesterone on apolipoprotein(a) gene expression in transgenic mice. Arterioscler Thromb Vasc Biol 17:1741–1745.