Epithelial and non-epithelial Ptch1 play opposing roles to regulate proliferation and morphogenesis of the mouse mammary gland

Patched 1 (Ptch1) has epithelial, stromal and systemic roles in murine mammary gland organogenesis, yet specific functions remain undefined. Cre-recombinase-mediated Ptch1 ablation in mammary epithelium increased proliferation and branching, but did not phenocopy transgenic expression of activated smoothened (SmoM2). The epithelium showed no evidence of canonical hedgehog signaling, and hyperproliferation was not blocked by smoothened (SMO) inhibition, suggesting a non-canonical function of PTCH1. Consistent with this possibility, nuclear localization of cyclin B1 was increased. In non-epithelial cells, heterozygous Fsp-Cre-mediated Ptch1 ablation increased proliferation and branching, with dysplastic terminal end buds (TEB) and ducts. By contrast, homozygous Ptch1 ablation decreased proliferation and branching, producing stunted ducts filled with luminal cells showing altered ovarian hormone receptor expression. Whole-gland transplantation into wild-type hosts or estrogen/progesterone treatment rescued outgrowth and hormone receptor expression, but not the histological changes. Bone marrow transplantation failed to rescue outgrowth. Ducts of Fsp-Cre;Ptch1fl/fl mice were similar to Fsp-Cre;SmoM2 ducts, but Fsp-Cre;SmoM2 outgrowths were not stunted, suggesting that the histology might be mediated by Smo in the local stroma, with systemic Ptch1 required for ductal outgrowth and proper hormone receptor expression in the mammary epithelium. Summary: Systemic and tissue-specific depletion of patched 1 in epithelial and stromal compartments of the mammary gland defines functions in ductal patterning, proliferation and gene expression.


INTRODUCTION
Organogenesis is the developmental process by which organs are constructed from undifferentiated germ layers. This process requires coordinated interactions between cells and tissues, and, for endocrine-targeted organs, cellular responses to extrinsic hormonal signals. These developmental processes are studied extensively, as they are often perturbed in cancer and other diseases.
Some hedgehog network members function 'non-canonically', independent of the signaling cascade described above. For example, PTCH1 can sequester hedgehog ligand to restrict the range of signaling, sequester cyclin B1 in the cytoplasm to inhibit cell cycle progression, or induce caspase 9-or caspase 3-mediated apoptosis in the absence of hedgehog ligands (Barnes et al., 2001;Chen and Struhl, 1996;Mille et al., 2009). In mammary epithelial cells, SHHstimulated PTCH1 promotes ERK1 and ERK2 phosphorylation independently of SMO (Chang et al., 2010). In the mouse mammary epithelium, constitutively activated Smo (SmoM2) acts as a Gprotein-coupled receptor (GPCR) via G αi2 to induce proliferation independently of GLI activity, as hyperproliferation was not blocked by pharmacological inhibition of GLI1 or GLI2 (Villanueva et al., 2015), consistent with observations by Riobo et al. (Riobo et al., 2006). TGFβ induces Gli2 to regulate osteolysis independently of Smo (Johnson et al., 2011), whereas K-Ras inhibits GLI2 function and GLI3 processing in the context of Smo activation (Lauth et al., 2010). A long non-coding RNA induced by the Twist transcription factor upregulates Gli1 and Gas1 (canonical hedgehog target genes) in vitro (Zhou et al., 2015). These noncanonical functions necessitate the evaluation of multiple network genes to fully understand hedgehog network function in a given organ.
The murine mammary gland is an excellent model for organogenesis (Daniel and Smith, 1999). In this system, organogenesis is initiated in the embryo, yielding a rudimentary ductal tree at birth, which remains relatively growth quiescent until puberty begins at 3-4 weeks of age. With puberty, systemic hormones (e.g. estrogen, progesterone and other hormones) drive ductal outgrowth via terminal end buds (TEBs). TEBs are transient structures that migrate and proliferate to produce a branched ductal tree that fills the mammary fat pad by 8-10 weeks of age. With conception, pregnancy hormones induce alveolar development to prepare for lactation. After lactation, the gland involutes and remodels to resemble the adult virgin (Hennighausen and Robinson, 2005;Macias and Hinck, 2012).
Previously, analysis of mammary glands from mice heterozygous for a germline knockout allele (Ptch1 Δ/+ ), or homozygous for a hypomorphic Ptch1 allele (Ptch1 mes ), demonstrated distinct functions for Ptch1 in the mammary epithelium, local stroma and systemically (mammary gland extrinsic) during postnatal virgin development (Lewis et al., 1999;Moraes et al., 2009). Neither the specific functions of Ptch1, nor the association of these phenotypes with canonical hedgehog signaling was investigated. Here, we employ tissue compartment-specific ablation of Ptch1, transplantation and tissue-specific expression of an activated Smo allele, to specify epithelial, stromal and systemic Ptch1 functions in virgin mammary gland development.

Ptch1 inhibits proliferation and branching of mammary epithelium
To determine the null phenotype of Ptch1 in mammary epithelium, mTmG-tagged primary mammary epithelial cells homo-or heterozygous for a Ptch1 conditional ablation allele (Ptch1 fl ) were treated with Adenovirus-Cre (Ad-Cre) and transplanted into the mammary fat pads of SCID/bg recipients (wild-type for Ptch1). Ad-Cre-treated, Ptch1 +/+ , mTmG+ primary cells were transplanted to contralateral fat pads. This approach increased recombination compared with MMTV-Cre (Wagner et al., 2001).
To ensure that the phenotypes were not due to differences in Credependent recombination, we determined that GFP-positive cells contributed similarly to ductal outgrowths by immunofluorescence.  (unpaired t-test). Graphs show data as mean±s.e.m. Paired t-tests were used to compare Ptch1 fl/fl glands with contralateral Ptch1 +/+ controls. *P<0.05 and **P<0.01. Scale bars: 1 mm in A-C; 50 µm in E-G,I-K.
To investigate whether Ptch1 fl/fl outgrowths displayed activated canonical hedgehog signaling due to reduced Smo inhibition, Ptch1 +/+ and Ptch1 fl/fl epithelium was evaluated by qPCR for hedgehog network gene expression. Of the genes evaluated, only Ptch2 mRNA was slightly upregulated (Fig. 2F) (P<0.016), suggesting that canonical hedgehog signaling was not activated.
Increased proliferation in Ad-Cre;Ptch1 fl/fl ducts is not due to activated canonical hedgehog signaling Gene expression analysis indicated that phenotypes from Ptch1 loss may not be due to increased SMO activity (Fig. 1M), consistent with unique mammary gland phenotypes elicited by epithelium-limited ablation of Ptch1 and activation of Smo (Visbal et al., 2011). To test whether hyperproliferation requires Smo activity, we evaluated hyperproliferation due to Ptch1 loss in the context of pharmacological inhibition of SMO.
We tested whether the ERα and PR expression phenotypes of the  (Visbal et al., 2011), and elicited precocious alveolar buddingwhich are not the case with Ptch1 loss. Recently, we found that SmoM2-dependent hyperproliferation in the mammary gland requires G αi2 -dependent signaling (Villanueva et al., 2015). Hyperproliferation was blocked by inhibiting some G αi subunits, but not by inhibiting GLI1 and GLI2 (Villanueva et al., 2015). The differences between these models suggests that Ptch1 loss increases proliferation independently of Smo. However, we cannot exclude the possibility that divergent phenotypes could be due to different functions of SmoM2 [an allele identified in human basal cell carcinoma (Xie et al., 1998)] versus endogenous Smo. The phenotypic differences between SmoM2 conditional expression and Ptch1 loss in the mammary epithelium agree with the lack of canonical hedgehog target gene upregulation in Ad-Cre;Ptch1 fl/fl ducts, and the inability of IPI926 to block hyperproliferation (Figs 1,  2), suggesting that hyperproliferation is SMO independent. These data fit with reports that SMO (Moraes et al., 2009) and activated hedgehog signaling are absent from the normal mammary epithelium (Chang et al., 2010;Hatsell and Cowin, 2006). From our data, it is possible that Ptch1 loss-induced hyperproliferation is due to reduced sequestration of cyclin B1 outside the nucleus.
Data here confirm that non-epithelial Ptch1 regulates ductal histology. Analysis of Ptch1 Δ/+ (Lewis et al., 1999) and Ptch1 mes/mes animals (Moraes et al., 2009) indicated that Ptch1 mediates ductal development; virgin Ptch1 Δ/+ mice had dysmorphic TEBs and filled-in ducts (Lewis et al., 1999). Whole Ptch1 Δ/+ glands transplanted to a wild-type host displayed filled-in ducts, whereas transplanted epithelial fragments did not, indicating that local stromal Ptch1 controls histology. From the Fsp-Cre model and transplantation experiments, we conclude that Ptch1 in the mammary fat pad fibroblastsnot myeloid cellsregulates histology. Based on the similar histology of Fsp-Cre;Ptch1 fl/fl and Fsp-Cre;SmoM2 ducts, it seems that Ptch1 may regulate histology via Smo. Taken together, the Fsp-Cre and Ad-Cre studies indicate that most phenotypes of the Ptch1 mes/mes mice, including altered TEB and ductal histology, and defective ductal elongation, were due to non-epithelial functions of Ptch1.
Aside from defining local stromal Ptch1 function, we have uncovered a role for mammary extrinsic, non-epithelial Ptch1 in pubertal mammary ductal outgrowth and ER/PR patterning in the mammary epithelium. The Fsp-Cre;Ptch1 fl/fl mutant diverges from the Ptch1 mes/mes mutant (Moraes et al., 2009), which displayed reduced ER and PR expression in stunted ducts. The differences between the Ptch1 mes/mes and Fsp-Cre;Ptch1 fl/fl models could be due to conditional ablation versus a hypomorphic allele, and/or global genetic manipulation versus loss of Ptch1 in Fsp-positive cells. Altered ER/PR patterning may be due to abrogated hormone production by the ovary or pituitary, which may have been differentially affected in these models.
We have also further defined the 'systemic' function of Ptch1. As E+P rescued the stunted ducts, Ptch1 may regulate E+P production and ovarian function to regulate pubertal outgrowth and proliferation. Indeed, the Fsp-Cre;Ptch1 fl/fl mutants displayed functional defects, including abrogated cycling and fertility. As the stunted duct phenotype was not rescued by bone marrow transplantation, Ptch1 does not function in myeloid cells to control ductal elongation.
As Fsp-Cre-mediated Ptch1 loss reduced mammary gland mass, and the mammary fat pad consists primarily of adipocytes, it could be hypothesized that off-target Cre activity in adipocytes contributed to stunted ductal outgrowth. Mice with loss of adipocytes displayed stunted ducts (Landskroner-Eiger et al., 2010). Although we cannot exclude the possibility that changes in the mutant adipocytes contributed to the stunted ducts, we did not observe Cre-dependent GFP expression in adipocytes, consistent with previous reports (Cheng et al., 2005); thus, such effects would likely be due to paracrine signaling.
Data here show stroma-to-epithelium and epithelium intrinsic Ptch1 functions in mammary gland development. It would be pertinent to determine whether bi-directional hedgehog-mediated tissue interactions exist in other organs where only unidirectional signaling is reported, e.g. prostate and pancreas (Hebrok et al., 2000;Wang et al., 2003). Dissecting these tissue-tissue interactions is crucial, as these developmental programs are inappropriately reactivated in cancer, and correlate with poor prognosis, e.g. in prostate and pancreatic cancer (Bailey et al., 2009;Fan et al., 2004).

Implications for Ptch1 and Smo in breast cancer
The hedgehog network is misregulated in many cancers, including breast (Moraes et al., 2007;Rubin and de Sauvage, 2006). Although hedgehog network activation induces basal cell carcinoma and medulloblastoma, data connecting hedgehog signaling and breast tumorigenesis are largely correlative, although Gli1 overexpression in mice induces tumorigenesis (Fiaschi et al., 2009). PTCH1 protein levels are reduced in 50% of DCIS and invasive breast cancer (IBC), whereas 70% of DCIS and 30% of IBC display aberrant SMO, suggesting that hedgehog activation occurs frequently and early in human breast cancer (Moraes et al., 2007). Furthermore, PTCH1 underexpression correlated with Ptch1 promoter methylation (Wolf et al., 2007). However, neither Ptch1 Δ/+ nor MMTV-SmoM2 mice show mammary tumors (Moraes et al., 2007(Moraes et al., , 2009. Our data suggest that perhaps, in the case of Ptch1 Δ/+ , the opposing functions of epithelial and systemic Ptch1 offset one another. These observations may explain why breast cancer incidence in individuals with Gorlin syndrome (Gorlin, 1987), who are heterozygous for germline Ptch1 loss-of-function and display higher risk for other cancers, is not higher than in the general population. Our Ad-Cre;Ptch1 fl/+ data suggest that Ptch1 heterozygosity would not alter mammary epithelial histology or proliferation.
Previous data suggest that high hedgehog ligand expression in tumor epithelium induces GLI1 (which is indicative of activated hedgehog signaling) in the adjacent stroma, which correlates with invasiveness and poor patient prognosis (O'Toole et al., 2011). As local stromal loss of Ptch1 and non-epithelial activation of Smo promote a DCIS-like phenotype in mammary epithelium, perhaps Fig. 8. Ptch1 functions in mammary gland morphogenesis and histogenesis. PTCH1 in the mammary epithelium inhibits proliferation and branching, independently of SMO. PTCH1 is essential in a mammary gland extrinsic Fsp-positive cell (fibroblast) for mammary ductal ER/PR patterning and for pubertal outgrowth. Ptch1 acts locally in an Fsp-positive stromal cell (likely a fibroblast) to inhibit SMO and elicit normal TEB and ductal histology.
stromal Ptch1 loss promotes cancer-associated phenotypes. The data presented here suggest that loss of Ptch1 in fibroblasts may increase survival, reduce non-apoptotic cell death or alter lumen formation. It would be interesting to determine whether Ptch1 heterozygosity correlates with DCIS in patients.
For studies of Ptch1 fl , Fsp-Cre;Ptch1 fl/+ males were crossed to Ptch1 fl/+ or Ptch1 fl/fl females. Fsp-Cre;SmoM2 mice were obtained by crossing Fsp-Cre, mTmG-positive males to SmoM2 +/− females (Xie et al., 1998). Genotyping for Ptch1 fl , SmoM2 and Fsp-Cre was performed as previously described (Bhowmick et al., 2004;Ellis et al., 2003;Jeong et al., 2004). CB.17/IcrHsd-Prkdc-scid-Lyst-bg (SCID/beige) mice (Harlan Laboratories) used for transplantation were from a breeding colony at Baylor College of Medicine. Animals were maintained according to the NIH Guide for the Care and Use of Experimental Animals with approval from Baylor College of Medicine Institutional Animal Care and Use Committee. For some analyses, 5-Bromo-2′-deoxyuridine (BrdU) (Sigma, B5002) in PBS was administered intraperitoneally 2 h prior to harvest at 250 mg/kg.

Adenoviral transduction and transplantation
For epithelial ablation of Ptch1, mammary epithelial cells were harvested from glands 1, 3, 4 and 5 of 8-week-old Ptch1 +/+ and Ptch1 fl/fl females with the lymph nodes removed. Glands were minced, digested with collagenase A (Roche Applied Science) and 0.05% trypsin-EDTA, and strained into single cells (Visbal et al., 2011). Cells were infected at MOI 50 with Adenovirus-Cre (Ad-Cre) from the Vector Development Laboratory Core Facility at Baylor College of Medicine. Cells were recounted, resuspended in 50% PBS/50% Matrigel (BD Biosciences) and 100,000 Ptch1 +/+ and Ptch1 fl/+ or Ptch1 fl/fl cells were injected contralaterally into epithelium-free 'cleared' inguinal fat pads of 3-week-old SCID/beige recipient mice (Deome et al., 1959) using a Hamilton syringe. Outgrowths were harvested 8 weeks later.

Whole-mount analysis
For fluorescent whole-mount analysis, glands were agitated in 1 ml of 50% PBS/50% glycerol solution at 4°C overnight as described previously , and imaged using a Leica MZFL16 fluorescence stereomicroscope with a DFC300 FX camera. Branch points were counted manually using Metamorph software. Confocal microscopy was performed with a Leica TCS SP5 microscope. Non-fluorescent whole mounts were analyzed using Neutral Red (Sigma) staining and imaged with a Leica MZ12.5 stereomicroscope with a Lumenera Infinity 1 camera, as described previously .

Immunofluorescence
Tissues were fixed in 4% paraformaldehyde in PBS for 3 h at 4°C, embedded in paraffin wax and sectioned at 3 μm. Slides were rehydrated using decreasing concentrations of ethanol. Immunostaining was carried out using antigen retrieval in 0.1 M sodium citrate buffer ( pH 6.0) and heating to 120°C in a decloaker (Biocare Medical). Primary antibodies were incubated overnight at 4°C with 8% MOM protein reagent (Vector Labs, BMK2202) and 1.5% goat serum. See Table S1 for antibody information. Micrographs were taken with a Zeiss Leica Axioskop 2 Plus with an AxioCam MRm FX camera. Cells from ten 40× fields, or ∼1000 mammary epithelial cells were quantified per animal using Metamorph software. Each TEB was a data point, with ∼300 cells/TEB.

Whole-gland transplantation
Control (Ptch1 fl/fl only or Fsp-Cre only) and Fsp-Cre;Ptch1 fl/fl donor glands at 3 weeks of age were transplanted contralaterally into 3-week-old SCID/bg recipient mice as described previously (Lewis et al., 2001;Moraes et al., 2009). Glands were analyzed 8 weeks after transplantation.

Estrogen and progesterone treatment
Daily subcutaneous treatments of 1 µg β-estradiol (Sigma) and 1 mg (Sigma) progesterone in sesame oil, or sesame oil only, were administered for 14 days prior to animal harvest.

IPI926 treatment (inhibition of SMO)
Either IPI926 (Infinity) dissolved in 13% ethanol in Tween-20 (Sigma) or vehicle alone were administered by oral gavage. IPI926 doses were 40 mg/ kg. For the mammary gland experiment, three daily treatments of vehicle or IPI926 were given prior to harvest.

Uterine scratch
After ovarectomy post-weaning and a 1-week rest, a prescribed course of estrogen (0.1 µg in 100 µl sesame oil for 3 days), 2 days rest, then estrogen +progesterone (1 mg progesterone+6.7 ng estrogen daily until harvest) was administered prior to scratch of one uterine horn by blunted needle as described previously (Finn and Martin, 1972). Vehicle or IPI926 was administered for 7 days prior to, and the day of harvest 9 days after the first estrogen treatment. Hormone and IPI926 doses were timed as described previously (Villanueva et al., 2015).

QPCR
Tissues were collected into RNA Later (Qiagen) and frozen at −80°C. RNA was extracted with the Qiagen RNeasy Kit, and cDNA was synthesized with the Superscript III kit (Thermo Fisher) using random hexamers. The cDNA was analyzed using an Applied Biosystems 7500-Fast thermocycler for TaqMan quantitative PCR under standard conditions. Product accumulation was represented as 2 −ΔCt , with ANOVA of ΔCt values used for statistical comparison. 18S rRNA was used for normalization. See Table S2 for primers.

Bone marrow transplantation
Recipient animals 4-5 weeks of age received Baytril water 24 h prior to irradiation and up to 6 days post-irradiation. Recipients received a dose of 400 centigray, and 24 h later, bone marrow cells were harvested and isolated from 4-5-week-old donor mice. Irradiated recipients received 2 million donor cells injected retro-orbitally. Recipients were harvested 6 weeks posttransplantation.