The pattern of β-catenin responsiveness within the mammary gland is regulated by progesterone receptor

The mammary gland is an epidermal appendage that undergoes both embryonic and postnatal development. During embryonic development, local signaling between the ectoderm and underlying mammary mesenchyme results in the formation of a rudimentary ductal tree embedded within a subdermal fat pad(Hens and Wysolmerski, 2005). Postnatal mammary development is temporally regulated by hormonal signaling(Lyons et al., 1958; Nandi, 1958). Puberty induces ductal elongation and bifurcation by stimulating cell proliferation within terminal end buds (TEBs) found at the ductal tips(Daniel and Silberstein, 1987; Bocchinfuso and Korach, 1997; Kleinberg et al., 2000). As puberty wanes, TEBs vanish and the ductal tree undergoes progressive estrus-induced maturation (Purnell and Saggers, 1974; Haslam,1987). During early pregnancy, tertiary side-branches sprout from the ductal tree and alveolar buds form at their tips(Russo and Russo, 1978; Brisken, 2002). Alveolar differentiation proceeds in two stages termed lactogenesis I and II. Lactogenesis I occurs during mid-pregnancy and involves the expression of genes encoding the milk proteins β-casein and WDNM1 and accumulation of lipid droplets inside alveolar luminal cells. During late pregnancy, a lactogenic switch occurs that permits differentiation to proceed to lactogenesis II (Neville et al.,2002; Brisken and Rajaram,2006). This stage involves expression of whey acidic protein(Wap) and α-lactalbumin genes, closure of tight junctions and secretion of milk and lipid droplets into the alveolar lumen(Robinson et al., 1995; Mather and Keenan, 1998; Nguyen et al., 2001). Lactation is followed by involution, which restores a pre-pregnant morphology to the parous gland (Lund et al.,1996; Wagner et al.,2002; Boulanger et al.,2005).

Experiments involving hormonal supplementation of oophorectomized and hypophysectomized mice first established the contributions of individual hormones to specific stages of mammary development. Growth hormone and estrogen (E) promote ductal elongation. Progesterone (P) induces side-branching and alveologenesis and prolactin (PRL), in combination with P promotes alveolar development (Lyons et al., 1958; Nandi,1958; Plaut et al.,1999; Shyamala,1999; Atwood et al.,2000; Hovey et al.,2002; Aupperlee and Haslam,2007). Recent studies on hormone receptor-null mice have confirmed and extended these findings. For example, mammary glands with targeted disruption of the estrogen receptor fail to undergo ductal extension(Bocchinfuso et al., 2000; Mallepell et al., 2006). Progesterone receptor (PR; Pgr - Mouse Genome Informatics)-null glands lack side-branches and alveoli(Lydon et al., 1995; Brisken et al., 1998). The PR-A isoform is dispensable to mammary development but PR-B^-/- mice have fewer side-branches and alveoli resulting in impaired lactation, despite complete alveolar differentiation(Mulac-Jericevic et al., 2000; Shyamala et al., 2000; Mulac-Jericevic et al., 2003). Prolactin receptor (PRLR) is also essential for alveologenesis and is required for complete alveolar differentiation. Prlr^-/- glands fail form alveoli when transplanted into pregnant hosts and Prlr^+/- glands show arrested alveolar development at day 15.5 of pregnancy (Ormandy et al.,1997a; Brisken et al.,1999). The hormonal ablation and supplementation studies taken together with the genetic studies show that PR and PRLR co-operate in early pregnancy to promote alveologenesis. However, in late pregnancy, PR activity restrains alveolar differentiation as demonstrated by the observation that declining P levels, in the presence of high PRL, trigger the lactogenic switch(Deis and Delouis, 1983).

A series of studies have indicated that PR activities during pregnancy are mediated by several paracrine factors, including Wnts(Brisken et al., 2000; Brisken et al., 2002; Mulac-Jericevic et al., 2003). Seven Wnt proteins are expressed during mammary development(Gavin and McMahon, 1992; Weber-Hall et al., 1994). However, their contributions to specific mammary developmental processes remain unclear, and whether they signal through canonicalβ-catenin-mediated or non-canonical pathways has not been determined.β-catenin has been strongly implicated in pregnancy-induced mammary development by both gain- and loss-of-function experiments. MMTV-ΔN89β-catenin induces precocious development in virgin females, whereas mice expressing suppressors of β-catenin signaling, such as axin or β-eng (in which the β-catenin C-terminal domain is replaced by the Drosophila Engrailed repressor domain), show impaired alveolar development in pregnancy (Hsu et al., 2001; Imbert et al.,2001; Hatsell et al.,2003; Rowlands et al.,2003; Tepera et al.,2003). These phenotypes suggest that hormonal and β-catenin pathways intersect to regulate postnatal mammary development.

To explore the relationship between PR and β-catenin signaling during mammary development, we tested the ability of MMTV-ΔN89β-catenin to rescue the PR^-/- phenotype. Our results show that stabilized β-catenin rescues alveologenesis at ductal tips in PR^-/- mammary glands. By contrast, PR is essential for the emergence or priming of a β-catenin-responsive subset of cells along the lateral borders of mammary ducts. In the presence of PR,ΔN89β-catenin-responding cells differentiate to lactogenesis I. In the absence of PR, ΔN89β-catenin-responding cell differentiation proceeds to lactogenesis II.

As PR^-/- mice are infertile and MMTV-ΔN89β-catenin mice cannot feed their pups after the third pregnancy owing to progressive alveolar hyperplasia, the following breeding strategy was taken to generate mice for analyses. MMTV-ΔN89β-catenin females(Imbert et al., 2001) on an FVB/n strain background were mated to PR^-/- males on a 129Sv X C57BL/6 strain background (Ismail et al., 2002), and the resulting F1 progeny were intercrossed. F2 PR^-/- males were crossed to wild-type FVB/n to generate large numbers of F3 PR^+/- female offspring, which were crossed to PR^-/-;ΔN89β-catenin males. All analyses were performed on the resulting F4 females of this cross. Mice were genotyped by PCR for the PR-lacZ allele using primers lacZF(5′-CATCCACGCGCGCGTACATC-3′) and lacZR(5′-CCGAACCATCCGCTGTGGTAC-3′) and for the wild-type allele using primers P1F (5′-TAGACAGTGTCTTAGACTCGTTGTTG-3′) and P2R(5′-GATGGGCACATGGATGAAATC-3′). Mice were screened for expression of stabilized β-catenin by Southern blot analysis(Imbert et al., 2001). Axin2dEGFP and conductin-lacZ mice were kind gifts of Dr Walter Birchmeier (Max Delbrueck Center, Berlin, Germany) and Dr Frank Costantini(Columbia University Medical Center, New York, NY). Conductin-lacZmice were generated by homologous recombination using a bacterial lacZ gene, with a nuclear localization signal, to replace the second exon of the conductin (Axin2) gene as described(Lustig et al., 2002). These mice were genotyped by PCR for the conductin-lacZ allele using the lacZF and lacZR primers described above.

Inguinal mammary glands were fixed in Carnoy's solution and stained in carmine alum overnight. Alternatively, glands were fixed in 4%paraformaldehyde (PFA) in PBS for 1 hour, washed three times in rinse buffer(2 mM MgCl₂, 0.2% sodium deoxycholate, 0.2% NP40 in PBS) and stained overnight at room temperature in 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal)staining solution (1 mg/ml X-Gal, 2 mM MgCl₂, 0.2% sodium deoxycholate, 0.2% NP40, 5 mM potassium ferrocyanide, 5 mM potassium ferricyanide). X-Gal-stained glands were post-fixed for 4 hours in 4% PFA and counterstained with carmine. Glands were then dehydrated in ethanol, cleared in Citrisolv (Fisher Scientific, Suwanne, GA) and mounted in Cytoseal (VWR,West Chester, PA).

Mammary glands were dissected, fixed in 10% phosphate-buffered formalin overnight, processed and embedded in paraffin. Histological analyses were performed on Hematoxylin and Eosin-stained sections of paraffin-embedded tissues. Alveolar luminal area was quantified by thresholding the image and computing the luminal area and the number of alveoli in each field using ImageJ software(http://rsb.info.nih.gov/ij/;National Institutes of Health, Bethesda, Maryland, USA). The graph represents an average of five fields that were examined in three PR^+/-;ΔN89β-catenin and five PR^-/-;ΔN89β-catenin mice. Immunofluorescence was performed on formalin-fixed tissues and immunohistochemistry on X-Gal-stained, PFA-fixed, paraffin-embedded tissues after antigen retrieval. Antigen retrieval was performed by microwaving sections in 6.52 mM sodium citrate (pH 6.0) for 30 minutes. Primary rabbit antibodies directed against NKCC1 (gift of Dr Jim Turner, NIH, Bethesda, MD; 1:2000), PR (A0098, DAKO,Carpinteria, CA; which recognizes both PR-A and PR-B isoforms; 1:500), casein(gift of Dr Margaret Neville, University of Colorado, Denver, CO; 1:4000) and Ki67 (Novocastra, Newcastle, UK, NCL-Ki67p; gift of Dr Susan Logan; 1:1000),and mouse antibodies recognizing the Myc epitope (9E10, 1:100; a gift of Dr Harold Varmus, Memorial Sloan Kettering, New York, NY; 1:500) and PCNA (M0879,DAKO; 1:1000) were used for these analyses. For histochemical analyses,biotinylated secondary antibodies (anti-rabbit or anti-mouse IgG, Vector Laboratories) were used in conjunction with streptavidin peroxidase (Fisher Scientific) that was colorimetrically detected using diaminobenzidine (K3466,DAKO). Alternatively, FITC-conjugated goat anti-mouse and Cy3-conjugated donkey anti-rabbit secondary antibodies (Cappel, Solon, OH; 1:100 and 1:200,respectively) were used for immunofluorescence.

Frozen sections (10 μm) were air-dried, fixed in 3.7% formaldehyde for 10 minutes and rinsed in deionized water and 60% triethylphosphate (TEP)(Fluka Chemie, Buchs, Switzerland). A working dilution of Oil Red O (Fluka Chemie, Buchs, Switzerland) was prepared fresh each time by diluting the stock solution (5 mg/ml in 60% TEP) to 36% with deionized water. Sections were stained for 25 minutes in the working solution of Oil Red O, counterstained with Hematoxylin 2 (Richard-Allan Scientific, Kalamazoo, MI), differentiated in running tap water, mounted in Gelmount (Fisher Scientific) and photographed immediately.

Total RNA was extracted from frozen mammary gland samples, stored in liquid nitrogen, using the TotallyRNA Isolation Kit (Ambion, Austin, TX). Total RNA(20 μg) from all samples was separated by electrophoresis on a 1%phosphate-glyoxal agarose gel using the NorthernMax-Gly Kit (Ambion). 28S rRNA was used as a loading control for the total amount of RNA in the sample. Northern blot analyses were performed using end-labeled oligonucleotide probes for β-casein (5′-GTCTCTCTTGCAAGAGCAAGGGCC-3′), Wap(5′-CAACGCATGGTACCGGTGTCA-3′) and WDNM1(5′-CAGAGCCCAGGCAGTAGTCATTGTC-3′)(http://mammary.nih.gov/tools/markers/molecular/MMD.html)and random-prime-labeled cDNA probes (Roche, Indianapolis, IN) forα-lactalbumin (GenBank BC069916) and K18 (Krt18)(Imbert et al., 2001) using the NorthernMax-Gly Kit (Ambion). All analyses were performed on a single membrane. RNA (2 μg) was reverse transcribed using Superscript III reverse transcriptase (18080-044, Invitrogen, Carlsbad, CA) and 1 μl of the cDNA thus obtained was used in the real-time PCR reaction. Real-time reverse transcriptase (RT) PCR was performed according to the instructions provided with the SYBR Green Quantitative RT-PCR Kit (Sigma Aldrich, St Louis, MO)using a Light Cycler (Roche). The following primers were used: β-casein,5′-GCCTTGCCAGTCTTGCTAAT-3′ and 5′-GGAATGTTGTGGAGTGGCAG-3′; WDNM1,5′-ACTGCCTGGGCTCTGTCTAA-3′ and 5′-TCTCCTGTGCATCGTTCATC-3′; α-lactalbumin 5′-CATAGCGTGTGCCAAGAAGA-3′ and 5′-CACATGGGCTTGTAGGCTTT-3′; Wap,5′-GTAGGACCCGCAAAACTCCT-3′ and 5′-TAGATTCCAAGGGCAGAAGC-3′; and 28S rRNA,5′-AAACTCTGGTGGAGGTCCGT-3′ and 5′-CTTACCAAAAGTGGCCCACTA-3′(Teuliere et al., 2005; Oxelmark et al., 2006). Analyses were performed in duplicate and values were normalized to those obtained for 28S rRNA. Fold changes were calculated relative to mRNA expression levels in a 12-week-old virgin mouse.

Three-week-old Rag1^-/- recipient mice were anesthetized by intraperitoneal injection of ketamine (1 mg/g body weight) and xylazine(0.5 mg/g body weight). A vertical, midline, abdominal incision was made to expose both inguinal mammary glands. The mammary fat pad was cleared of epithelium by resecting the gland lateral to the lymph node. Each recipient received 1 mm³ epithelial fragments either from 6- to 10-week-old donor mice (PR^-/- and PR^-/-;ΔN89β-catenin or PR^+/- and PR^+/-;ΔN89β-catenin) in contra-lateral cleared fat pads. The abdominal incision was closed with 4.0 Ethilon interrupted sutures(http://mammary.nih.gov/tools/mousework/index.html). Transplanted fragments were allowed to grow for 6 weeks to reconstitute the mammary ductal system. Recipient mice were then mated and sacrificed at days 8.5 and 14.5 of pregnancy. Transplanted glands were stained with X-Gal(Applichem, Cheshire, CT) to differentiate between ductal systems generated from transplanted fragments and those arising from incompletely removed endogenous epithelium. Endogenous glands from the Rag1^-/-mice were used to determine the extent of pregnancy-induced normal alveolar development and as negative controls for X-Gal staining.

The extent of branching in transplanted glands was analyzed by counting the number branches per 1000 pixels of ductal length in four to seven fields. Ductal length was measured using ImageJ software. Secondary ducts are smaller than primary ducts and were distinguished based on this criterion. All branches arising from secondary ducts were designated as side-branches.

The requirement for PR in pregnancy-induced mammary development has been unequivocally demonstrated by studies in which transplanted PR^-/- mammary glands fail to form side-branches or alveoli in response to E and P or to pregnancy(Lydon et al., 1995; Brisken et al., 1998). To investigate the intersections between β-catenin and PR signaling, we asked whether expression of ΔN89β-catenin could restore pregnancy-like morphogenesis to PR^-/- glands. As PR^-/- mice are infertile, these studies were performed in mammary glands transplanted into immunocompromised pregnant hosts.

As reported, PR^-/- glands fail to develop in response to pregnancy (Fig. 1A,E). By contrast, rounded protrusions were observed in PR^-/-;ΔN89β-catenin glands(Fig. 1B,F) that resembled structures that formed in control transplanted PR^+/-(Fig. 1C,G) and endogenous Rag1^-/- glands (Fig. 1D,H) but which were consistently more distended. Wholemount analyses showed a significant increase in these structures between 8.5 and 14.5 days of pregnancy (compare Fig. 1B-D with F-H). Histological analysis identified the presence of alveolar features, such as casein expression, in these protrusions(Fig. 2A,B). To determine whether ΔN89β-catenin also rescued ductal side-branching, we analyzed the expression of the Na-K-Cl co-transporter, NKCC1, in transplanted mammary glands during pregnancy(Shillingford et al., 2003). In wild-type glands, NKCC1 is highly expressed along the basolateral membrane of ductal cells of virgin animals (Fig. 2C), but is reduced within alveoli during pregnancy(Shillingford et al., 2002). High levels of NKCC1 expression, indicative of ductal differentiation, were maintained in PR^-/- glands(Fig. 2D). However, NKCC1 expression was reduced in the protrusions that develop in PR^-/-;ΔN89β-catenin glands(Fig. 2E) and in endogenous Rag1^-/- alveoli (Fig. 2F). These data suggest that that ΔN89β-catenin rescues alveologenesis, but not side-branching, in PR^-/- glands during pregnancy. Wholemounts of transplanted glands at 8.5 days of pregnancy showed no statistically significant difference in the number of secondary branches among the four genotypes of mice examined(Fig. 2G,H). Consistent with previous observations, PR^-/- mice showed significantly fewer side-branches than PR^+/- glands (t-test, P=0.008; Fig. 2H)(Lydon et al., 1995). Although on average ΔN89β-catenin glands contained more side-branches than their non-transgenic counterparts, these differences were not statistically significant (t-test, P=0.057).

We then asked whether MMTV-ΔN89β-catenin expression could induce precocious alveolar development in virgin mice in the absence of PR. Wholemounts of virgin mammary glands from 12-week-old adolescent(Fig. 3A-D) and 26-week-old mature adult (Fig. 3E-H) female PR^+/- and PR^-/- mice were compared with those from mice transgenically expressing ΔN89β-catenin. Pubertal ductal extension was completed normally in virgin PR^+/- and PR^-/- glands and their transgenic counterparts. In adolescents, PR^+/-(Fig. 3A) and PR^-/- (Fig. 3C) ducts remained smooth but 5/6 PR^+/-;ΔN89β-catenin(Fig. 3B) and 2/6 PR^-/-;ΔN89β-catenin(Fig. 3D) glands showed precocious mammary development. By 26 weeks of age, PR^+/-(Fig. 3E) and PR^-/- (Fig. 3G) ducts remained smooth but all transgenic glands displayed precocious development regardless of PR status(PR^+/-;ΔN89β-catenin, n=7, Fig. 3F; and PR^-/-;ΔN89β-catenin, n=7, Fig. 3H). These data show thatβ-catenin signaling induces precocious mammary development in the absence of PR activity.

Further examination of wholemounts from 26-week-old virgin PR^+/-;ΔN89β-catenin and PR^-/-;ΔN89β-catenin mammary glands revealed striking differences in the pattern of development(Fig. 3F,H and Fig. 4A,B). In PR^+/-;ΔN89β-catenin glands, development occurred at regular intervals along the lateral borders of the secondary ductal branches as well as at ductal tips(Fig. 4A). By contrast, PR^-/-;ΔN89β-catenin mammary development occurred exclusively at the ends of ductal branches(Fig. 4B). Thus, aΔN89β-catenin-responsive subset of cells exists at ductal tips regardless of PR activity, but the emergence of a second subset ofΔN89β-catenin-responsive cells along the lateral borders of ducts is strictly dependent upon PR activity within the virgin gland. As the MMTV-LTR is hormonally responsive, we examined whether PR produced these effects by altering ΔN89β-catenin expression(Shyamala and Dickson, 1976; Truss et al., 1992; Witty et al., 1995). However,indirect immunofluorescence detected Myc-tagged ΔN89β-catenin in all luminal cells regardless of PR expression(Fig. 4C,D). These data show that P/PR signaling has no effect on transgene expression but critically determines the pattern of response to ΔN89β-catenin in the adult mammary tree.

In other tissues it is known that only a subset of cells are capable of mounting a response to a uniformly expressed β-catenin transgene(DasGupta and Fuchs, 1999). To examine whether a similar phenomenon operates in the mammary gland, we investigated the expression of conductin-lacZ, a reporter ofβ-catenin signaling. Conductin (Axin2) is a target gene and negative-feedback regulator of the canonical Wnt signaling pathway. Expression of the conductin-lacZ knock-in allele provides a context-independent,reliable reporter of β-catenin signaling(Jho et al., 2002; Lustig et al., 2002; Yu et al., 2005). Analyses of PR^+/+;conductin^+/lacZ;ΔN89β-catenin mammary glands showed that, despite uniform ΔN89β-catenin transgene expression in all luminal cells (Fig. 5A,B), only a subset of cells expressed conductin-lacZ(Fig. 5B). Conductin-lacZ-positive cells were distributed at intervals along the ducts of PR^+/+;conductin^+/lacZ;ΔN89β-catenin glands (Fig. 5B,C) and were spatially distinct from cells expressing PR(Fig. 5C). Conductin-lacZ expression was prominent in all regions undergoing development, i.e. at ductal tips and along the ductal borders of PR^+/+;conductin^+/lacZ;ΔN89β-catenin glands (Fig. 5D). These results show that in PR^+/+ glands, the subset of cells capable of responding to β-catenin signaling is evenly patterned with respect to PR.

To determine which cell-types are proliferating in PR^+/+;conductin^+/lacZ;ΔN89β-catenin glands, we first analyzed the expression of PCNA and PR by double immunofluorescence. PR and PCNA were found in mutually exclusive expression patterns (Fig. 5E), reminiscent of previous reports in human breast and mouse mammary glands(Clarke et al., 1997; Seagroves et al., 2000). Thus,PR cells are quiescent in MMTV-ΔN89β-catenin virgin glands. By contrast, we observed that 34.9+11.3% of conductin-lacZ-positive cells express Ki67 (Fig. 5F,G). We conclude that the segregation between PR and proliferating cells is maintained in ΔN89β-catenin-expressing mice and that conductin-lacZ-positive cells are proliferating.

β-catenin-induced structures faithfully recapitulate many features of normal alveolar development. For example, they are composed of bilayered structures (Fig. 6A,B) that contain eosinophilic secretions, and they express casein(Fig. 6C,D) and lipid droplets(Fig. 6E,F). In addition to having alveolar morphology, β-catenin-induced structures showed downregulation of PR-lacZ in a manner similar to that observed in pregnancy-induced alveoli (data not shown). We observed that alveoli from PR^-/-;ΔN89β-catenin glands were consistently larger than their counterparts in PR^+/-;ΔN89β-catenin glands(Fig. 6A-G), suggesting that they have progressed to a more advanced stage of lactogenic differentiation. To determine the extent to which ΔN89β-catenin induces alveolar differentiation, we analyzed milk protein gene expression by northern blotting(Fig. 7) and real-time RT-PCR(Table 1) performed on RNA isolated from virgin glands. Early milk protein genes including WDNM1and β-casein were expressed in all PR^+/-;ΔN89β-catenin (n=3) and PR^-/-;ΔN89β-catenin (n=3) mice at levels above those detected in their non-transgenic virgin littermates. However, transcripts for late milk protein genes, includingα-lactalbumin and Wap, which were present in barely detectable amounts in PR^+/-;ΔN89β-catenin glands(n=3), were expressed at higher levels in 2/3 PR^-/-;ΔN89β-catenin glands(Fig. 7 and Table 1). We conclude thatΔN89β-catenin-induced alveoli undergo secretory differentiation but are restrained by PR activity from undergoing the lactogenic switch.

Mammary gland development is critically dependent on hormonal signaling. Previous studies have shown that β-catenin signaling is required for alveologenesis and alveolar survival (Hsu et al., 2001; Imbert et al.,2001; Tepera et al.,2003). However, connections between hormonal and β-catenin signaling have not been investigated. Here, we show that two distinct subsets of mammary cells are responsive to β-catenin. One subset, located at ductal tips, is intrinsically β-catenin-responsive. A second subset,distributed at intervals along the lateral ductal borders, requires PR signaling to become β-catenin-responsive.ΔN89β-catenin-induced alveoli commence secretory differentiation but are restrained by PR from undergoing the lactogenic switch(Fig. 8).

Our results show that although luminal expression ofΔN89β-catenin specifically restores alveologenesis to ductal tips,it does not rescue side-branching in PR^-/- mice. Side-branching and alveologenesis are abolished in PR^-/-glands but can be restored by placing PR^-/- cells in the proximity of PR^+/+ cells, thus indicating that PR exerts its effects by inducing paracrine factors(Brisken et al., 1998). Wnt and RANKL have been implicated as downstream paracrine effectors in these processes (Gavin and McMahon,1992; Weber-Hall et al.,1994; Bradbury et al.,1995; Humphreys et al.,1997; Brisken et al.,2000; Fata et al.,2000; Grimm et al.,2002; Mulac-Jericevic et al.,2003). β-catenin is the central transducer of the canonical Wnt pathway and can also synergize with RANKL signaling(Fig. 8)(Fantl et al., 1999; Shtutman et al., 1999; Tetsu and McCormick, 1999; Cao et al., 2001; Lamberti et al., 2001; Rowlands et al., 2003). The observation that ΔN89β-catenin rescues alveologenesis but not side-branching in PR^-/- mice is entirely consistent with previous findings that β-catenin signaling suppressors specifically impair alveologenesis (Hsu et al.,2001; Tepera et al.,2003). From these results, we conclude that PR induction of side-branching must be transduced through β-catenin-independent pathways in luminal cells, or, through effects on other cell types(Fig. 8). Alternatively,cell-fate choice could be determined in the mammary gland by specificβ-catenin levels as is the case for other epidermal appendages, where lowering the β-catenin activity promotes sebaceous and epidermal fate and raising β-catenin signaling induces hair-cell fate(Gat et al., 1998; Niemann et al., 2002; Niemann and Watt, 2002).

The most surprising finding of the current study is that PR signaling induces dramatic changes in the pattern of β-catenin responsiveness within the virgin gland. Using conductin-lacZ as a reporter ofβ-catenin signaling, we identified a subset of luminal cells that are capable of responding to β-catenin. These proliferative responder cells are regularly distributed with respect to PR-positive cells and appear to be precursors of conductin-lacZ-positive precocious alveoli. Alveolar development is found along the lateral ductal borders in PR^+/+;ΔN89β-catenin and PR^+/-;ΔN89β-catenin glands but is restricted to the ductal tips in PR^-/-;ΔN89β-catenin glands. One interpretation of our findings is that PR is essential for the spatial patterning of a β-catenin-responsive cell type along the lateral ductal borders (Fig. 8). Precedence for PR mediating patterning comes from analyses of PR expression in PR^-/- mice. PR^-/- ducts fail to undergo the transition from a juvenile, uniform pattern of PR expression to an adult, non-uniform PR expression pattern that is seen in wild-type and PR^+/- animals (Ismail et al., 2002; Shyamala et al.,2002). Our results suggest that PR activity plays a more global role in patterning the virgin gland during the process of ductal maturation and might regulate the positioning of multiple progenitors (bipotent, ductal and alveolar) along lateral borders. As alveoli emerge directly from lateral borders in PR^+/-;ΔN89β-catenin virgin glands(Fig. 4A), their formation is dependent upon PR activity but not upon the prior formation of a ductal side-branch.

An alternative interpretation of our data is that PR acts downstream ofβ-catenin. For example, temporal increases in PR activity during estrus and pregnancy could control β-catenin responsiveness by regulating the availability of a limiting factor required for β-catenin signaling(Fig. 8). At a molecular level,a number of factors are known to regulate β-catenin signaling. Regulated nuclear entry of β-catenin has been observed in Xenopus embryos and the β-catenin partners, Legless (LGS; BCL9)/BCL9-2 and Pygopus(PYGO), are required for this step in flies(Schneider et al., 1996; Kramps et al., 2002; Thompson et al., 2002; Hoffmans and Basler, 2004; Townsley et al., 2004; Stadeli et al., 2006). Once inside the nucleus, assembly of β-catenin transcription complexes is dependent upon the expression of DNA-binding partners, such as Tcf proteins/LEF1, which play important roles in mammary development. Lef1^-/- mice fail to develop mammary buds and Tcf1 and Tcf4 are expressed in adult mammary glands(van Genderen et al., 1994; Korinek et al., 1997; Barker et al., 1999; Roose et al., 1999). Intriguingly, recent studies have shown that P primes uterine cells forβ-catenin signaling by increasing Tcf/Lef and reducing GSK3β levels but E is required for nuclear entry (Rider et al., 2006).

β-catenin responsiveness of cells at ductal tips is PR-independent, as they undergo alveologenesis in PR^-/- as well as PR^+/- mice. We do not know why these cells are intrinsically responsive to β-catenin. It is possible that these cells are pre-existing alveolar progenitors. However, in rats it has been proposed that ductal tips contain a remnant of TEB stem cells(Russo and Russo, 1978). Previous studies have suggested that only stem cells are capable of responding to β-catenin signaling. For example, although keratin-14-ΔN87β-catenin is expressed throughout the basal epidermal layer, TOPGAL transcriptional response is restricted to epidermal and hair follicle stem cells (DasGupta and Fuchs, 1999). Similarly, despite elevation of β-catenin in all intestinal cells of Apc^min mice,conductin-lacZ expression is restricted to crypt stem cells and premalignant adenomas (Lustig et al.,2002; Maretto et al.,2003). These studies show that only stem cells and possibly early progenitors are β-catenin-responsive. Consistent with these observations,ΔN89β-catenin expression has been shown to expand the mammary stem/progenitor pool (Liu et al.,2003; Liu et al.,2004). Thus, the inherent ability of cells at ductal tips to respond to β-catenin, without prior priming by PR, is in keeping with their suggested designation as residual TEB stem cells, as proposed by Russo(Russo and Russo, 1978).

Our previous studies have shown that β-catenin stimulation results in the emergence of structures that are morphologically indistinguishable from normal alveoli (Imbert et al.,2001; Rowlands et al.,2003). Here we report that in addition to initiating alveologenesis and causing alveolar progenitor expansion and survival,β-catenin induces a significant degree of alveolar differentiation that progresses to lactogenesis II in the absence of PR(Fig. 8). It is not clear whether β-catenin initiates alveologenesis and subsequent differentiation proceeds by default, or, whether β-catenin acts at multiple steps to facilitate the differentiation process. During pregnancy, P and PRL intersect in complex ways to regulate alveologenesis, differentiation and expansion. PR and PRLR reciprocally regulate one another's expression and both pathways phosphorylate and activate STAT5A, an important mediator of alveolar development (Edery et al.,1985; Ormandy and Sutherland,1993; Gouilleux et al.,1995; Groner and Gouilleux,1995; Ormandy et al.,1997b; Richer et al.,1998). Moreover, both PR and PRLR increase the levels of RANKL and cyclin D1, which are essential for alveologenesis and alveolar maturation,respectively (Brisken et al.,2002; Brockman et al.,2002). However, numerous studies indicate that PRLR promotes and PR activity restrains the lactogenic switch(Neifert et al., 1981; Graham and Clarke, 1997; Ormandy et al., 1997a; Nguyen and Neville, 1998; Buser et al., 2007). For example, in the presence of a differentiated and secretion-competent mammary epithelium, P withdrawal triggers the lactogenic switch(Neville et al., 2002). Ovariectomy or treatment with the P antagonist RU486 in late pregnancy induces lactation (Nguyen et al.,2001). Decreasing P levels coincide with lactogenesis in many species and P secretion from retained placental fragments delays lactogenesis in women (Martin et al., 1978; Neifert et al., 1981). Some degree of PRLR activity, either through placental lactogen (PL) or PRL, is required for the lactogenic switch as abolishing PL, by hysterectomy, and PRL,by bromocryptine treatment, completely prevents tight-junction closure(Nguyen et al., 2001). The fact that β-catenin promotes the lactogenic switch in the absence of PR and is restrained from doing so by the presence of PR raises the possibility that β-catenin additionally functions downstream of PRL(Fig. 8). Future experiments will focus upon elucidating the relationship between β-catenin signaling and the PRL-JAK-STAT axis in secretory differentiation.

In summary, our data show that although β-catenin signaling is sufficient to induce alveologenesis at ductal tips, it is insufficient to stimulate alveolar development along lateral ductal borders. We propose that aβ-catenin-independent PR activity is required to generate and/or maintain stem/progenitor cell competence along lateral ductal borders. Once rendered competent by this mechanism, a progenitor subset is poised to undergo alveologenesis in response to β-catenin signals.

We thank Dr Rama Khokha (University of Toronto, Canada) and Carlos Hojilla for expert guidance with mammary transplant technology; Dr Peggy Neville(UCHSC Denver, CO) for antibodies to casein, Dr Jim Turner (NIH) for antibodies to NKCC1 and Dr Susan Logan for antibodies to Ki67; Jerome Nwachukwu and Dr Michael Garabedian for their assistance with real-time RT-PCR assays; members of the Cowin laboratory for critical comments and Brigitte Teissedre for thoughtful editing of the manuscript. This work was supported by a predoctoral fellowship DOD BCRP W81XWH-04-1-0373 (M.H.), a Susan G. Komen Grant BCR0504557 (P.C.), NIH-R01-GM47429 (P.C.) and by Philip Morris USA and Philip Morris International. PR^-/- mice were developed with the support of NIH CA-77530-01 and Susan G. Komen Grant BCTR0503763(J.P.L.).