Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2012 Jan 15.
Published in final edited form as: Dev Biol. 2010 Oct 23;349(2):125–136. doi: 10.1016/j.ydbio.2010.10.018

Adipocyte Derived Paracrine Mediators of Mammary Ductal Morphogenesis Controlled by Retinoic Acid Receptors

Christine V Marzan 1,1, Tara S Kupumbati 1,2, Silvina P Bertran 1, TraceyAnn Samuels 1, Boris Leibovitch 1, Rafael Mira y Lopez 1, Liliana Ossowski 1, Eduardo F Farias 1,3
PMCID: PMC3021183  NIHMSID: NIHMS248566  PMID: 20974122

Abstract

We generated a transgenic (Tg)-mouse model expressing a dominant negative-(DN)-RARα, (RARαG303E) under adipocytes-specific promoter to explore the paracrine role of adipocyte retinoic acid receptors (RARs) in mammary morphogenesis. Transgenic adipocytes had reduced level of RARα, β and γ, which coincided with a severely underdeveloped pubertal and mature ductal tree with profoundly decreased epithelial cell proliferation. Transplantation experiments of mammary epithelium and of whole mammary glands implicated a fat-pad dependent paracrine mechanism in the stunted phenotype of the epithelial-ductal tree. Co-cultures of primary adipocytes, or in vitro differentiated adipocyte cell line, with mammary epithelium showed that when activated, adipocyte RARs contribute to generation of secreted proliferative and pro-migratory factors. Gene expression microarrays revealed a large number of genes regulated by adipocyte-RARs. Among them, pleiotrophin (PTN) was identified as the paracrine effectors of epithelial cell migration. Its expression was found to be strongly inhibited by DN-RARα, an inhibition relieved by pharmacological doses of all-trans retinoic acid (atRA) in culture and in vivo. Moreover, adipocyte-PTHR, another atRA responsive gene, was found to be an up-stream regulator of PTN. Overall, these results support the existence of a novel paracrine loop controlled by adipocyte-RAR that regulates the mammary ductal tree morphogenesis.

Keywords: RARs, adipocytes, paracrine, mammary morphogenesis, pleiotrophin

Introduction

Retinoids and their receptors (RARs and RXRs) have been extensively studied in embryonic morphogenesis, but their role in post-natal tissue development has received sparse attention, partly because of the embryonic or peri-natal lethality of compound RAR mutants. The mammary gland is uniquely suited for post-natal study as a major morphogenic expansion occurs during the 2nd month after birth. One of several transgenic approaches we used to modulate RAR function was the expression, in mouse mammary epithelium, of a full length RARα with a point mutation in G303E which functions as a dominant negative (DN)-RARα. This approach produced rapidly developing B cell lymphoma in these mice (Wang et al., 2005),(Kupumbati et al., 2006), precluding a detailed examination of the RAR in normal mammary and tumors. Retinoic acid treatment slowed the growth of lymphomas suggesting that the construct functions as a DN-RARα (Wang et al., 2005).

Because adipocytes have been shown to affect mammary epithelial function, and since RARs, potent transcriptional regulators, are expressed in adipose tissue (Alvarez et al., 2000), including the mammary fat pad, we wondered whether they might influence mammary morphogenesis. Under the influence of pubertal hormones, the rudimentary mammary epithelial tree present at birth, begins extending and invading the mammary fat pad. Elongation and branching of the ducts, regulated by proliferation and migration of terminal end buds (TEB) cells, rely on both endocrine and local growth regulatory signals, ECM remodeling and stromal-epithelial interactions. Because the leading edge of the TEB is mostly devoid of myoepithelial layer and has fragmented basement membrane, it is directly juxtaposed to the surrounding fat pad adipocytes. The mammary fat pad is made mainly of white adipocytes, although some brown adipocytes have been identified (Gouon-Evans and Pollard, 2002). We propose that the directional ductal epithelial elongation might be influenced by the TEB contact with, and duct proximity to, the fat pad. To date, little is known about molecular interactions between these two compartments. In vitro, proliferation of mammary epithelium was shown to increase in presence of mammary fat pad explants or ECM and conditioned medium derived from 3T3-L1 adipocytes (Hovey et al., 1998; Levine and Stockdale, 1984; Rahimi et al., 1994). Primary mammary epithelium undergoes ductal morphogenesis when cultured on a monolayer of 3T3-L1 adipocytes or when co-cultured with fetal mammary fat pad precursor tissue (Kanazawa and Hosick, 1992; Wiens et al., 1987). Mammary adipocytes also induce alveolar morphogenesis and enhance functional differentiation of epithelial cells in a Transwell co-culture model (Darcy et al., 2000; Zangani et al., 1999). The virtual lack of ductal development in A-ZIP/F-1 fatless mouse (Couldrey et al., 2002) most convincingly demonstrates that adipocytes play and indispensable role in mammary development. Although, these studies demonstrate the importance of adipocytes in the development of mammary epithelial cells, they provide no clues to the mechanisms involved.

To analyze the molecular events that might be responsible for the adipocyte effects, we generated transgenic mice in which adipocytes-RAR function was inhibited by the expression of G303E point mutant of RARα. This dominant negative (DN) mutation in the ligand binding domain of RAR, which precludes the binding of physiological concentrations of ligand (Kupumbati et al., 2006; Saitou et al., 1994) gives rise to a severely stunted ductal mammary tree. A micro-array analysis showed that the presence of DN-RARα modulated the expression of a large number of genes, among them growth factors, proteins mediating cell adhesion and migration and proteins previously shown to modulate mammary development. We focused on 2 of these proteins, a membrane receptor PTHR and pleiotrophin, a secreted protein, both down-regulated in the transgenic glands and showing interconnected regulation and a paracrine effects on mammary epithelium. Overall, our results established adipocyte-RARs as indispensable regulators of a secreted growth factor (pleiotrophin), involved in paracrine regulation of epithelial ductal tree development.

MATERIALS AND METHODS

Generation of aP2.RARaG303E mice

A blunt ended RARαG303E fragment (Saitou et al., 1994) was cloned into Sma 1 digested PBks-AP2 vector (a gift from Dr. Bruce Spiegelman) (Graves et al., 1992). The fragment obtained by Sal 1 and Not 1 digestion and purified by agarose gel was used for microinjection. Three founders (Line 8, 10 and 16) were generated in FVB/N background by standard pronuclear injection. (Southern blot was used to determine amounts of transgene DNA relative to wt- RARα DNA using methods and reagents as previously described (Kupumbati et al., 2006). Hemizygous lines were established from all 3 founders. We found that the number of viable pups was somewhat greater when transgenic males were mated with wild type females; hence this approach was used for long term line maintenance. The Tg-females did not nurse their pups. PCR-based genotyping was routinely used. All 3 lines exhibited similar phenotypes and lines 8 and 16 were continuously maintained. Transgene expression was monitored by RT-PCR using the HA tag.

Mammary Glands Whole Mounts

Mammary glands were excised, fixed in Carnoy’s fixative and stained in Carmine Alum Solution. For details: http://mammary.nih.gov/tools/histological/Histology/index.html

Ductal tree proliferation assay

To compare proliferation of cells in the ductal mammary tree of the wt and aP2.RARaG303E mice, the Click-It EdU Alexa Fluor 488 Imaging Kit from Invitrogene was used. Briefly, 8 week-old female mice were injected with 100μg EdU (5-ethynyl-2′-deoxyuridine) in 100μl of PBS 3 hours prior to gland removal, the glands were fixed in 10% formalin in PBS, embedded in paraffin and 5μm sections were prepared and processed for staining as per manufacturer instructions. Alexa-488 green fluorescent nuclei (proliferating cells) were visualized under the fluorescence microscope and quantified by calculating the % of Alexa-488 positive nuclei per 300 DAPI positive epithelial cells at 200x magnification. Six wt and 6-Tg glands were used for quantification.

Epithelial and whole gland transplantation

Epithelium containing fragments of mammary glands of 6 weeks old virgin mice (wt or Tg) taken from the area between the nipple and the LN, were transplanted into epithelium pre-cleared glands as previously described (Kupumbati et al., 2006). The recipient glands were processed for whole mounts 8 weeks after transplantation. In total 24 epithelial fragments were transplanted; 7 mice were mock transplanted to ascertain the efficiency of epithelial removal. For whole gland transplantation, glands number 4 from a 4-week old wt or Tg-virgin mice were transplanted under the skin of 6 weeks old recipient, by placing the transplant under the skin between gland number 3 and 4. The glands were excised and examined as whole mounts 8 weeks after transplantation. In total 6 pairs of glands were transplanted.

Adipocyte isolation

Adipocytes of 7-week old virgin mice were prepared from left number 4 gland (#4L), and both number 3 glands (#3L/R); the number 4 right gland (#4R) was kept for whole mount analysis. Glands were digested in collagenase/BSA solution at 37°C for 2hrs and centrifuged for 10min at 900rpm, the floating pellet containing adipocytes was removed, examined under microscope for purity, stabilized with RNAlater solution (Ambion), and stored at −20°C until RNA isolation.

Induction of adipocyte differentiation in 3T3-L1 murine preadipocyte fibroblast cell line

3T3-L1 cells, (a gift from Dr. Philip Scherer’s Lab, Albert Einstein School of Medicine, NY) were propagated at 50–60% confluence in DMEM with 10% FBS (Benchmark FBS, Hyclone Laboratories, Inc) and 1x Penicillin-Streptomycin solution. For differentiation, 3T3-L1 cells were grown to confluence in DMEM, (Mediatech) with 10% FBS, then in DM1 (DMEM with 10% FBS, 160nM Insulin, 250μM Dexamethasone, 0.5 mM Isobutylmethylxanthine) for 48hrs, DM2 medium (DMEM with 10% FBS and 160nM insulin) for another 48hrs, and in DMEM containing 10% FBS for 2–6 more days, with medium change every 2 days. Intracellular lipids in adipocytes were visualized by staining with Oil Red O solution (60%).

Mouse mammary epithelial cell lines

HC-11 were grown in RPMI-1640 with 10% FBS, 5μg/ml insulin, 10ng/ml EGF, 2g/L sodium bicarbonate, Penicillin-Streptomycin and 2.42g/L HEPES; Comma 1D cells were grown in DMEM-F12 medium with 2% FBS, 10μg/ml insulin, 5ng/ml EGF, 0.3% BSA, Penicillin-Streptomycin and 2.42g/L HEPES.

Preparation of conditioned media (CM)

Serum-free CM was collected from 3T3-L1 cells on Day 10–14 of differentiation protocol overnight or after 48hrs, (6ml per 10cm dish), and centrifuged at 2000rpm for 10min. When transfected 3T3-L1 cells were used, differentiated 3T3-L1 cells were detached from the culture dish and transfected with plasmid or siRNA solution using Amaxa Cell Line Nucleofector Kit L as described by manufacturer, and inoculated into collagen I coated plates. For vector or DN-RARα expression, 2μg of vector or DN-RARα plasmid was added to 2×106 cells, for siRNA transfection experiments, 2μg of si-RNA targeting pool was transfected into 2×106 cells using Amaxa nucleofector. siRNAs used were siCONTROL (Non-targeting siRNA pool, Cat.# D-001206, Dharmacon), PTHR siRNA (siGENOME SMART pool Cat.# M-042524, Dharmacon) and PTN siRNA (siGENOME SMART pool Cat. #, Cat.# M-042740, Dharmacon). Following 48hrs incubation of the transfected cells in growth medium, CM was collected as above.

Proliferation assay

Comma 1D and HC-11 cells (1×104) seeded and grown in 24-well culture dishes were washed after 24hrs with PBS, and incubated in 1ml of serum-free DMEM-F12 medium for 48hrs. The medium was replaced with 3T3-L1 adipocyte conditioned medium, diluted 1:8, 1:16 or 1:32 in serum-free DMEM-F12 medium. After 48hrs the conditioned medium was refreshed and the cells were incubated for additional 48hrs (Comma 1D), washed, fixed with 1% glutaraldehyde for 15min, stained with 0.1% crystal violet solution for 30min, washed, photographed and the dye dissolved in 30% methanol/10% acetic acid solution. Absorbance was read at 590nm. HC-11 cells were incubated in CM for a total of 3 days, detached and counted.

Migration experiments

Transfected 3T3-L1 adipocytes were placed on collagen I-coated plates (5×105 cells per well in a 24-well plate) and allowed to attach overnight (for vector and DN-RAR transfected cells) or 48hrs (for siRNA transfected cells). The cells were washed with serum-free medium and 800μl of DMEM was added to each well. Subconfluent HC-11 cells were detached with 1mM EDTA in PBS, and 5.0 ×104 in 200μl DMEM was inoculated in the upper chamber of a modified Boyden Transwell (Pore size 8.0μm, Cat. # 353097, Becton Dickinson). The transfected 3T3-L1 adipocytes and HC-11 cells were co-cultured overnight, cells were removed from the top of the membrane, and the membrane was fixed and stained with 0.5% crystal violet and 1% formalin for 20min. Photographs were taken from 5 random sectors (at x10) of the membrane and the number of cells that have migrated was determined by manual counting. Values show average and standard deviation. Significance was determined using t-test.

Restoration of PTN in vitro and RARβ and PTN in vivo by atRA treatment

To show that the effects of the DN-RARα works through disruption of the RAR pathway, 2μg of DNA of vector or DN-RARα construct were transiently expressed in differentiated 3T3-L1 adipocytes by Amaxa nucleofection, as described above, and 24hrs after transfection the adipocytes were treated with pharmacological doses of atRA (200 and 500 nM for 24 hrs), the cells lysed and PTN expression was determined by immunoblotting. For in vivo experiments, transgenic aP2-DN-RARα1 mice (n=4 per group) were treated daily with 4mg atRA (stock solution in DMSO) in 100μl of olive oil by gavage for 48 and 72hrs, the #3 and 4 mammary glands were collected, RNA was extracted as described and the expression of RARβ2 (a known atRA target gene) and PTN were determine by Q-PCR analysis (see below). Glands from wt-mice receiving only oil served as positive control.

RT-PCR

Total RNA from mammary adipocytes, or from in vitro differentiated 3T3-L1 adipocytes, was isolated using RNeasy Lipid Tissue Mini Kit (Qiagen) using the manufacturer’s protocol. The RNA was purified in RNeasy Spin columns and 1 to 4μg of RNA was reverse transcribed using iScript cDNA Synthesis kit (Bio-Rad) and used for PCR reactions. The number of cycles used in each PCR reaction was dependent on the abundance of the transcript and ranged from 25–45 cycles. Primers used for PCR amplification: RARα1-S 5′ GACTGTTTGCCTGCTCTTCTG 3′ RARα1-AS 5′ CTCACAGGCGCTGACCCCAT 3′; RARα2-S, 5′GAACCGGGCCTGTTTGCTCC3′, RARα2-AS,5′ CTCACAGGCGCTGACCCCAT 3′; RARβ2-S5′ ATGGAGTTCGTGGACTTTTCTGTG 3′: RARβ2-AS 5′ CTCGCAGGCACTGACGCCAT 3′; RARγ1 -S 5′ TGGGGCCTGGATCTGGTTAC 3′; RARγ1 -AS 5′ TTCACAGGAGCTGACCCCAT 3′; RARγ2 -S 5′ GCCGGGTCGCGATGTACGAC 3′, RARγ2 –AS 5′ TTCACAGGAGCTGACCCCAT 3′; DN-RAR –S 5′ TACCCCTACGACGTGCCCGACTATGCCAGC 3′, DN-RAR-AS 5′ GTTTCTCACAGACTCCTTGGACATGCCCAC 3′; GAPDH -S 5′ CGTAGACAAAATGGTGAAGG 3′, GAPDH -AS 5′ GACTCCACGACATACTCAGC 3′; PTHR –S 5′ GGGCACAAGAAGTGGATCAT 3′, PTHR –AS 5′ GCCCATGAAGACGGTGTAGT 3′; Pleiotrophin –S 5′ GGAGAATGGCAGTGGAGTGT 3′, Pleiotrophin -AS 5′ TCAAGGCGGTATTGAGGTC 3′; Tubulin -S 5′ TGCCTTTGTGCACTGGTATG 3′ Tubulin-AS 5′ CTGGAGCAGTTTGACGACAC 3′.

Q-PCR

RNA was isolated from primary mammary epithelial cells as above. cDNA was synthesized from 2μg of RNA using SuperScriptRTM First-Strand Synthesis System for RT-PCR (Invitrogen). Q-PCR was performed using the QuantiTect SYBR® Green PCR Kit (Qiagen) with 100nM primers and 50ng of cDNA. To determine the fold change in method (also referred to as the 2−ΔΔCt method) was used expression, the comparative Ct with the Ct values averaged from triplicates for each sample for the RARβ2 and PTN expression normalized by the average of the triplicate for the RPL30 gene. The PCR primers used were: RARβ2 S: 3′ AAGCAGGAATGCACAGAGAG 5′, RARβ2 AS: 5′ TGGGCTTTCCGGATCTTCT 3′; PTN S: 5′ GGAGAATGGCAGTGGAGTGT 3′, PTN AS: 5′ TTCAAGGCGGTATTGAGGTC 3′; and RP30L30 S: 5′ GACAAGGCAAAGCGAAATTG 3′, RPL30 AS 5′ GTATTTTCCGCATGCTGTGC 3′.

RNA isolation from mammary glands for Affymetrix Array

Mammary glands (#4) of 7-wk old virgin mice (3-wild type and 3 transgenic) were excised, lymph nodes removed, stabilized in RNAlater solution (Ambion) and stored in −80°C until RNA isolation. Tissues were homogenized in 4M guanidine isothiocyanate (GTC) solution, layered on top of 5M Cesium Chloride solution, centrifuged at 38,000rpm at 18°C for 18hrs and the recovered RNA pellet was purified by Phenol/Chloroform extraction. Samples were further purified using RNeasy Mini Kit (Qiagen), as described by manufacturer’s protocol.

Affymetrix Array

RNA from individual animals was used to produce fluorescent labeled cRNAs and hybridized to Affymetrix GeneChip Mouse Expression Set 430 (done by Mount Sinai Microarray Facility). Samples from the 3-individual wt and Tg-glands were considered replicates. Array data was analyzed with GeneSpring Software (Agilent Technologies). Per-chip normalization was performed to adjust the intensity of each array, eliminating minor differences in probe preparation and hybridization conditions. Each gene had to be rated as present or absent in all three samples of a set before the average signal was determined. Increase or decrease in gene expression of ≥ 2.0 fold with a p value of < 0.05 (Wilcoxon-Mann-Whitney test; multiple testing correction: Benjamini and Hochberg False Discovery Rate) was considered statistically significant.

Immunohistochemistry

Formalin fixed mammary glands were embedded, sectioned, deparaffinized and hydrated. For detection of PTHR, antigen was retrieved by incubating samples in 4% paraformaldehyde for 15min, followed by incubation with Proteinase K (10μg/ml) for 10min and then again in 4% paraformaldehyde for 15min before boiling in 1mM EDTA solution. Sections were incubated in PTHR antibody (anti-PTHR, clone 3D1.1, Cat.# 05-517, Upstate Cell Signaling Solutions), (2μg/ml), overnight at 4°C. For HA-antigen retrieval, sections were boiled in 1mM EDTA solution (pH 7.4) for 6min, simmered for 15min, incubated with anti-HA antibody (2μg/ml (Cat. # 2367, Cell Signaling), processed using VectaStain ABC Elite Kit (Vector Laboratories), the signal was detected using Metal Enhanced DAB Substrate Kit (Cat. # 35065, Pierce Laboratories) and counterstained with Harris Hematoxylin Solution (Sigma Diagnostics).

Immunoblotting

Cells were washed twice with cold PBS and lysed with cell lysis buffer (50mM Tris-HCl-pH 7.4, 1% NP-40, 0.25% deoxycholate, 150mM NaCl, 1mM EDTA, 1mM EDTA, 1mM PMSF, 1mM orthovanadate, 1mM NaF, 5μg/ml of aprotinin, leupeptin, pepstain A and proteinase inhibitor cocktail (Complete, Cat. #11697498001, Roche Diagnostics). After centrifugation (14,000rpm for 30min), protein concentrations of the supernatants were determined using Bio-Rad Protein Assay Solution (Cat. # 500-0006, BIO-RAD). For immunoblots, 25 to 50μg of total protein, boiled in Laemmli buffer containing β-mercaptoethanol were subjected to electrophoresis on 6–15% SDS polyacrylamide gels. The proteins were then transferred to PVDF Membranes (Hybond-P, Amersham Biosciences), the membranes were blocked with 5% non-fat dried milk solution and incubated with the primary antibody in 3% non-fat dry milk solution diluted as specified; anti-HA tag 1:500 (Cell Signaling), anti-PTHR 1:200 (Upstate Cell Signaling); anti-PTN 1:250 (R and D Systems), anti-PTPRζ1 1:200 (Santa Cruz Biotechnology), mouse anti-Tubulin and anti-Vinculin (1:1000) (Sigma). Some membranes were stripped and re-probed.

Results

To examine the role of adipocyte-RARs on mammary morphogenesis we created a transgenic mouse in which DN-RARα was expressed in the adipocyte tissue. Using this mouse model, microarray analysis and in vitro model consisting of adipocytes and mammary epithelium we identified proteins active in the adipocyte-epithelial paracrine interactions.

Generation of aP2.RARαG303E mice

A targeting vector containing the transgene shown in Fig. 1A was generated and used to produce founders by standard pronuclear injection in FVB/N background; 3 founder lines were obtained, (Founder #8 a female and founders #10 and #16, males). Southern blotting (Fig. 1B), showed that the signal for the transgene was greater than the RARα signal. Hemizygous lines were established for all 3 founders and transgenic males were mated with wild type females. As shown for line #8 and #16, (Fig. 1C), the transgene was highly expressed in adipose tissue, including the forth mammary fat pad, abdominal white adipose tissue (WAT) and interscapular brown adipose tissue (BAT). Low level of expression in other organs was, likely, due to fat contamination. Staining of the transgenic mammary for HA-tag (Fig. 1D, right panel), showed that its expression was limited to adipose tissue and excluded from the mammary epithelium. As expected, the wt-mammary was negative for HA (Fig. 1D, left panel).

Figure 1. Expression of RARα1G303E in FVB/N mice.

Figure 1

Open in a new tab

A) RARα1G303E was subcloned into the PBks-AP2 vector and expressed in FVB/N mice by standard pronuclear injection. RARaG303E was HA-tagged. B) Southern blot analysis showing 3 founders (#8 female, and #10 and #16 males) with high levels of expression of the transgene. C) Transgene expression pattern determined by RT-PCR in organs of founder #8 and #16 (upper and middle panels). The weak expression detected in several organs (except brown and white fat and mammary) is most likely due to fat in the organ. D) Immunohistochemical analysis of the wt (left panel) and Tg- (right panel) mammary glands using anti-HA antibodies; arrows indicate positive staining in nuclei of adipocytes. E=epithelium. (Scale bar = 100μm).

Phenotypic changes in the mammary epithelial tree induced by DN-RARα expression in adipocytes

All 3 lines of the transgenic mice were visibly smaller, had coarser fur and their body weight, as well as the weight of their mammary fat pads, were ~10% lower than in the wild type, even at week 10. Other tissues, such as kidneys, heart and brain did not differ in weight (Results not shown). The transgenic and the wt-type mammary fat pad adipocytes had similarly appearing fat-filled cells with large, Oil Red O stained vacuoles suggesting that the decrease in weight is due to reduced number and not due to inhibited differentiation of adipocytes.

Whole mounts of mammary glands from control and Tg-mice at 6, 9 and 12 weeks after birth were examined. As shown in Fig. 2A, the epithelial duct outgrowth in the Tg-gland was drastically reduced at 6 weeks, and remained smaller even at 12 weeks when the wt epithelial ducts reached the end of the fat pad. Very similar changes were observed in the 2 additional Tg-lines (Results not shown). These changes were quantified by measuring the ductal length from the nipple area to the tip of the longest duct and the length of the fat pad from the nipple to the edge of the fat pad. As shown in Fig. 2B, the ducts in the transgenic glands were persistently (week 6 to 16) and significantly (p<0.001) shorter than in the wild type glands.

Figure 2. Mammary epithelial ductal tree phenotype.

Figure 2

Open in a new tab

A) Whole mounts of mammary glands from wt and DN-RARαG303E transgenic mice isolated at 6, 9 and 12 weeks. B) Quantification of ductal length shown as a ratio between the longest mammary duct measured from the nipple and the length of the fat pad (n=4 glands per group, p<0.001, t-test). C) Donor wt-mammary epithelium (8 weeks old mouse, gland #4, upper left panel) transplanted into wt mammary gland divested of epithelium, of a 3–4 week old mouse (lower left panel) and analyzed by whole mount 8 weeks later. Right upper and lower panels, as above, except the donor is Tg. The whole mount is representative of 12 individual transplants which were all successful and all showed complete repopulation of the fat pad. D) Whole mammary gland transplantations. The #4 wt (left panel) and Tg glands (right panel) of 3–4 weeks old mice were transplanted in contra lateral sites under the skin (between gland #3 and #4) of a wt, 6 week old recipient mouse and processed as in C. All of the 6 whole Tg-mammary gland (epithelium and adipocytes) transplants retained their stunted Tg-phenotype in a wt-mouse environment. Scale bars = 2 mm.

Transplantation experiments were performed to test whether microenvironment of the Tg-fat pad or a systemic effect of RAR-inhibition by the transgene in other fat tissue, was responsible for the mammary tree phenotype. First, epithelium-containing tissue fragments of the transgenic and wild type mammary glands were transplanted into a cleared fat pad of a wild type (wt)-mouse. Fig. 2C, shows that the donor epithelium taken from a severely stunted Tg-mammary tree (upper right panel) produced an outgrowth indistinguishable in morphology from the wt-transplant when transplanted into a wt-fat pad (lower left and right panels). This indicates that the fat pad is responsible for the defective development of the mammary tree in the DN-RARα Tg-mice. To exclude a role for the systemic effect of other transgenic fat depots we transplanted the entire intact Tg and wt-whole mammary glands (epithelium and fat pad) into FVB/N recipients. As shown in Fig. 2D, left panel, after 8 weeks of growth in the wild type environment, the ductal development remained stunted in the Tg-transplant relative to the contra lateral wt-transplanted gland. These results strongly suggest that systemic signals do not account for the striking ductal phenotype of aP2-RARαG303E glands.

While the underdeveloped Tg-mammary ductal tree was slightly enlarged during pregnancy (in mid-pregnant gland it extended slightly beyond the lymph node), its lobuloalveolar development was comparable to that of the wild type littermates. This shows that pregnancy hormones appear to only partially rescue the ductal development in Tg-glands, but that lobuloalveolar development is not affected by the transgene (Fig. S1). The Tg-females were fertile, although produced a smaller litter, but were unable to nurse their pups.

Reduced epithelial cell proliferation is the cause of the stunted mammary tree in the transgenic gland

As shown in Fig 2A, mammary tree development in the transgenic mice is severely impaired. Already at 6 weeks, when the wt-TEBs present as thick, club like structures, the TEBs in transgenic glands are smaller and remain small at weeks 7 to 9 (Fig. 3) and up to 12 weeks (Results not shown). To test if proliferation was affected, 6-wt and 6 Tg-mice, 6 weeks of age were injected with 5-ethynyl-2′-deoxyuridine (EdU) and analyzed ex-vivo by staining for EdU incorporation. EdU incorporation in the TEB cells, although difficult to quantify, (Fig. 3), by visual inspection appeared to be the highest in wt-TEBs. The percent of EdU-positive duct cells was reduced >5-fold in the transgenic glands (from 20% to <4%, Fig. 3A and B).

Figure 3. A. Terminal end buds (TEBs) phenotype.

Figure 3

Open in a new tab

Whole mounts of wt (upper panels) and Tg glands (lower panels) from 6, 7, and 9 weeks old mice. TEBs in Tg are much smaller and appear “mature” already at 7 weeks. (Magnification x63). EdU was injected 3hrs before isolation of the glands and EdU was detected by Alexa-488 fluorescence (green) in TEBs (inserts in the upper right hand panel, wt gland, and lower right hand panels, Tg-gland) and ducts (right hand upper panel, wt-gland, lower right hand panel, Tg-gland). The TEBs shown were selected for the highest and the lowest EdU incorporation for wt- and Tg- glands. Nuclei are visualized with DAPI (blue). (n=6 per group, magnification = 200x). B. EdU positive cells/300 ductal epithelial cells/section counted in 6 individual wt- and Tg-glands. Bars show mean and standard deviation. p<0.0001 by t-test.

In vitro modeling of adipocyte-epithelial interactions

The above in vivo studies suggest that the Tg-mammary fat pad is unable to support epithelial mammary tree elongation, possibly because it cannot support optimal proliferation of the epithelium. To examine this possibility further, we turned to a culture system consisting of an immortalized mouse mammary epithelial cell line, Comma 1D and conditioned medium produced by differentiated, vector-transfected (see Materials and Methods) 3T3-L1 cells. The growth of Comma 1D cells incubated for 4 days in serum-free conditioned medium (CM) from differentiated adipocytes, diluted 1:8 in fresh medium, was compared to cells grown in serum-free medium. Fig. 4A, upper panel shows that adipocyte-CM stimulated proliferation by approximately 4 fold. CM from DN-RARα expressing differentiated 3T3-L1 adipocytes, when incubated with Comma 1D (Fig. 4A, lower panel) or HC-11 cells (Fig. 4B), produced less robust stimulation of proliferation than CM from vector transfected cells. This reduction was not due to the DN-RARα effect on 3T3-L1 differentiation as determined by the unchanged level of oil Red O stained vacuoles, (Results not shown). Dilution of the DN-RARα-CM further reduced the stimulatory activity on both cell lines, suggesting that it contains less pro-proliferative factor(s).

Figure 4. Paracrine effect of adipocytes on epithelial cells.

Figure 4

Open in a new tab

A) Upper panel-mouse mammary epithelial cells, Comma-1D, were incubated in serum-free medium or in conditioned medium (CM) from differentiated (adipocytes) 3T3-L1 cells for 4 days. The cells were stained with crystal violet, photographed, the dye extracted and absorbance read at 570nm. Middle panel-CM was diluted 1:8, 16 and 32, Comma-1D incubated as above, stained and photographed. Lower panel-cell proliferation was quantified as in upper panel, t-test p<0.001. B. HC-11 cells incubated with CM as above for 3 days were detached and counted. (ADIP is adipocytes). ANOVA, p<0.001. C) Expression of DN-RARα in differentiated 3T3-L1 cells measured by RT-PCR. (Samples marked as “-” were not reverse-transcribed. D) The effect of co-culture (16hrs) of differentiated wt and DN-RARα expressing 3T3-L1 cells on migration of mammary epithelial HC-11 cells was measured in Boyden chambers. Upper panel, the underside of the filters fixed with 1% formalin and stained with crystal violet (x4 magnification). Lower panel, HC-11 cells that migrated in 25 random fields (Magnification × 10). Mean, SD, p < 0.001 (t-test). Similar results were obtained in 2 additional experiments.

Expression of DN-RARα in adipocytes reduces their ability to stimulate mammary epithelial cell migration

We next tested whether adipocytes provide chemotactic or pro-migratory stimuli required for the expansion of the mammary tree into the fat pad. The ability of vector and DN-RARα transfected adipocytes to stimulate HC-11 epithelial cell migration under co-culture conditions in a modified Boyden chamber was compared. We switched to these cells because Comma 1D migrated poorly even when stimulated with FBS. Fig. 4C showing stained undersides of Boyden chamber filters reveals that DN-RARα expressing adipocytes are much less effective in stimulating epithelial cells migration. Quantification of migration (lower panel) confirms that cell migration of HC-11 in co-culture with DN-RARα-transfected adipocytes is diminished by ~60% (*p<0.001) compared to co-cultures with vector controls. These results suggest that adipocytes release soluble mediators of epithelial cell migration that are up-regulated by active RARs and reduced by DN-RARα expression.

Gene expression profiles of transgenic and wild type mammary glands

Several stromal proteins have been identified as regulators of epithelial function needed for proper formation of mammary tree (Hinck and Silberstein, 2005; Sternlicht et al., 2006). However, no reports considered the role of RAR in their regulation. The profound effect of adipocyte DN-RARα expression on mammary gland morphogenesis in vivo and on proliferation and migration of mammary cells in culture prompted further analysis by microarray gene expression approach.

Using RT-PCR we first tested whether DN-RARα down-regulates the expression of known RAR target genes in adipocytes of wt and Tg-mammary glands of 8 weeks old mice. As shown in Fig. 5A, expression of the DN-RARα substantially (56 to 73%) reduced the mRNA levels of RARα2, β2, and γ2. This was not unexpected as DN-RARα is able to heterodimerize with RXR, bind to DNA and prevent RAR activity (Kupumbati et al., 2006; Saitou et al., 1994). For unknown reasons RARγ1 was also slightly reduced. For the RNA microarray analysis we opted to use whole mammary gland because adipocytes isolation takes several hours during which we feared the RNA profiles might be altered. Also, gene expression changes in epithelial cells due to altered paracrine interactions might also be of interest. RNA of mammary gland # 4 from 3-seven week old wt mice, and 3 DN-RARα Tg-mice, was isolated, providing 6 independent replicates which were used in hybridization to Affymetrix GeneChip Mouse Expression Set 430. We found that 215 genes were down-regulated, (p<0.05), in the transgenic glands, with 49 of those coding for secreted or ECM proteins (Table 1A), and 171 genes were up-regulated (p<0.05), with 47 coding for secreted or ECM proteins (Table 1B). Microarray results for 15 of the genes (11 down-regulated and 4 up-regulated in the transgenic glands) were confirmed by RT-PCR. Among 10 genes we characterized further, we focused on 2, which showed adipocyte-restricted expression and which were previously implicated as having a role in mammary development. One gene, strongly down-regulated (23.41 fold) in the microarray, codes for a secreted protein, pleiotrophin (PTN), and a second, down-regulated 2.45 fold, codes for parathyroid hormone receptor (PTHR).

Figure 5. RAR regulated genes in wt and Tg-adipocytes.

Figure 5

Open in a new tab

A) Adipocytes were recovered from mammary glands of 8 weeks old wt and Tg-mice, the RNA was isolated subjected to RT-PCR (see Materials and Methods) and analyzed by electrophoresis. The bands were scanned using Image J and a band intensity ratio of Tg/wt was calculated. B) Adipocytes were isolated from 8-wt and 8-Tg-mammary glands of 7-week old mice; RNA was extracted and subjected to RT-PCR for detection of PTHR and PTN expression as described in Methods. C) Sections of wt- and Tg-mammary glands of 7-week old mice were fixed and immunostained with anti-PTHR antibody. Brown stain, present in the thin wt adipocyte cytoplasm, indicates PTHR protein expression. Note that the stain is almost completely absent in the Tg-adipocytes. Scale bar = 100μm. D) Adipocytes isolated from wt-mammary glands were immobilized in Matrigel and incubated in medium with 10% FBS for 48hrs and then treated for 48hrs with 1μM atRA. Untreated adipocytes served as positive control. RNA was extracted and processed as above. Induction of RARβ2 served as positive control.

Table 1.

Table 1A. Genes downregulated in transgenic glands coding for secreted or ECM proteins.
Gene Fold Change Gene Fold Change Gene Fold Change
Von willebrand factor homolog 4.62 Midkine 2.58
Secreted Phosphoprotein 1 31.98 PERP, TP53 apoptosis effector 4.52 Parathyroid hormone receptor1 2.45
Pleiotrophin 23.41 Syndecan 1 4.43 Sclerostin domain ctng 1 2.44
Interleukin 17B 21.10 Serine protease inh (Spint2) 4.31 Fat globule-EGF factor 8 2.37
Serine 2 protease 10.58 Ceruloplasmin 4.25 Dickkopf homolog 3 2.37
Amphiregulin 10.36 Leukocyte protease inh. 3.85 TGFb 3 2.30
Serine proteinase inh. 7.97 Suppression of tumorigenicity 14 3.56 Thrombospondin 1 2.29
Transm. Serine protease 2 7.68 CTGF 3.49 Syndecan 4 2.26
Serine proteinase inh.2 7.62 Periostin 3.49 LIM/senescent antigen like dom 2 2.24
Inter-alpha trypsin inh. 7.24 Procollagen type VIII 3.40 Ficolin A 2.15
Insulin-like GF binding prot 6.57 Procollagen type XIII 3.38 V-erb-b2 homolog 3 2.13
Chemokine-like factor sf8 6.57 Latent TGFb binding prt 2 3.12 Opioid growth factor receptor-like1 2.12
Extracellular proteinase inh. 6.52 Tenascin C 2.80 Proprotein convertase 2.10
Chemokine CXC ligand 15 6.11 CEA-related CAM 1 2.77 Agrin 2.1
Cholecystokinin 5.66 TGFb receptor 1 2.77 Transglutaminase 2 2.09
Procollagen type IX 5.65 Adrenomedulin 2.67 Stanniocalcin 2 2.07
Prelp 5.40
Table 1B. Genes upregulated in transgenic glands coding for secreted or ECM proteins.
Gene Fold Change Gene Fold Change Gene Fold Change
Single WAP prt 2 16.92 Embigin 2.69 Inhibin beta-B 2.19
CXC ligand 9 4.10 Myc target 1 2.64 Interleukin 4 receptor 2.18
Delta-like homolog 4.08 Platelet/endothelial CAM 2.62 Procollagen C-proteinase enhc prt 2.18
Serine proteinase inh. Cld B 4.03 Calsyntenin 3 2.62 Biglycan 2.17
Melanoma antigen 4.00 Stanniocalcin 1 2.60 Glypican 2.17
CXC ligand 8 3.91 Necdin 2.52 Leptin 2.13
Embigin 3.31 Procollagen type IV 2.49 Suppressor of cytokine sgnl 3 2.13
ALCAM 3.19 Protein kinase C 2.47 Melanoma CAM 2.11
Type I transm. Receptor 3.16 Tyrosine kinase receptor 1 2.44 Integrin, beta like 1 2.07
Intercellular adh mol 2 3.13 Procollagen type XV 2.40 SPARC-like a (mast9) 2.06
Serine proteinase inh. Cld A 3.04 Interleukin 16 2.39 PKIf 2.06
CSF 2 receptor beta 1 2.88 Procollagen type VI 2.35 SPARC related Ca binding 2 2.05
Polydomain protein 2.87 Leucine-rich repeats (Lrig 1) 2.30 EGF-like domain 7 2.03
DNA-damage induc. Trnsrpt 2.85 Neurotrophic tk receptor 2 2.29 Secreted mod. Ca bdng prt 2.02
Small C-C ligand 11 2.74 Integrin alpha 7 2,29 Serine proteinase inh Cld 2.01
c-mer proto-oncogene t 2.72 Apelin 2.27
Open in a new tab

We quantified PTHR and PTN in RNA isolated from adipocytes purified from inguinal and thoracic glands of 2 wild type and 2 transgenic 7-week old mice using semi-quantitative RT-PCR. As shown in Fig. 5B, while wt-adipocytes expressed both PTHR and PTN, Tg-adipocytes had barely detectable levels of these two mRNAs. Sections of wt and Tg-glands were examined by immunohistochemistry with anti-PTHR antibodies. Fig. 5C shows that while PTHR (brown stain of adipocytes cytoplasm surrounding large lipid vacuoles) was detected in most cells of the wt gland, there was a marked decrease in PTHR protein in Tg-gland. An attempt to detect PTN by immunohistochemistry with the available antibodies was unsuccessful. These results indicated that PTHR and PTN are down-regulated when DN-RARα is expressed in adipocytes.

We next tested whether PTN and PTHR are under atRA regulation. Adipocytes isolated from wt-virgin mice and embedded in Matrigel, were incubated for 48hrs in medium with 10%FBS and then, for additional 48hrs, with medium with serum without or with 1μM of atRA. As a positive control for the culture conditions we used RARβ2, a known atRA target gene, and showed that it was strongly induced by this treatment. Under the same conditions PTHR and PTN expression were also strongly up-regulated (Fig. 5D).

Do PTN and PTHR have a role in mammary cell migration?

To answer this question we used co-cultures of differentiated 3T3-L1 adipocytes and HC-11 mouse mammary epithelial cells. Similarly to primary mammary adipocytes, 48hr treatment of differentiated 3T3-L1 cells with 1μM atRA in charcoal stripped serum induced RARβ2, PTHR and PTN (Fig. 6A). Moreover, transfection of differentiated 3T3-L1 cells, with DN-RARα strongly reduced the PTHR and PTN protein level (Fig. 6B, left and right panels). The dominant negative transcriptional effect of the DN-RARα can be overcome by pharmacological doses of atRA (Kupumbati et al., 2006; Saitou et al., 1994). We therefore treated differentiated 3T3-L1 adipocytes, transfected with DN-RARα or vector control, with 200 and 500nM atRA and showed (Fig. 6C) that at 500nM the PTN protein level reduced by the DN-RAR was restored to levels almost comparable to that of atRA-treated 3T3-L1-control. This indicates that PTN is inhibited through the dominant negative function of the mutant RAR. The effect of atRA on restoration of PTN expression was also tested in vivo in 3 groups of mice (4/group); wt-untreated, Tg-untreated and Tg-atRA treated by daily gavage with 4mg of atRA for 48 or 72hrs. Q-PCR analysis of the RNA extracted from glands #3 and #4 showed that RARβ, a known atRA target gene, and PTN were restored to the wt-level by 72hr treatment with the higher dose (4mg) of atRA (Fig. 6D). Whole mount analysis of several of the glands did not show phenotype rescue most likely because in vivo treatment of Tg-mice with this high dose of atRA inhibits the normal function/proliferation of mammary epithelium.

Figure 6. RAR regulated genes in differentiated 3T3-L1 adipocytes and in mammary adipocytes in vivo.

Figure 6

Open in a new tab

A) 3T3-L1 cells were allowed to fully differentiate and were incubated in charcoal-stripped serum for 4 days. Cells were either treated for 48hr with 1μM atRA or left untreated and RNA was extracted. PTHR and PTN (and β2, as positive control) expression was detected using RT-PCR. GAPDH served as loading control. B) Differentiated 3T3-L1 cells were transfected either with empty vector or with DN-RARα plasmid and following overnight incubation, cell lysates were prepared and examined by Western blotting using anti-HA antibody (to detect transgene expression), or anti-PTHR antibody (left panel), or anti-PTN antibody (right panel). Tubulin served as loading control. Both PTHR and PTN levels were reduced in DN-RAR transfected adipocytes. C) Vector or DN-RARα-transfected, differentiated 3T3-L1 cells were treated with 200 and 500nM of atRA for 24hrs, the protein was extracted and tested for PTN expression. Tubulin served as loading control. D. Mammary glands #3 and #4 of wt-mice or Tg-mice untreated or treated daily with 4mg atRA for 72hrs, were removed, the RNA was extracted and analyzed for RARβ and PTN using Q-PCR. The bar represents an average of glands from 4 mice.

To explore PTN functions as a paracrine mediator of epithelial cell migration we first tested whether it was secreted exclusively by the adipocytes and whether epithelial cells express PTN-receptor on their surface. As shown in Fig. 7A, left panel, PTN was present exclusively in adipocytes, and not in the epithelial HC-11 cells, while its receptor, PTPRζ1 was very highly expressed in HC-11 cells and barely detectable in the adipocytes. Moreover, PTN was easily detected in 3T3-L1 conditioned medium (Fig. 7A, right panel), suggesting that it is available for paracrine interactions. We next transfected differentiated 3T3-L1 cells with pooled PTN-siRNA or scrambled- siRNA and found (Fig. 7B, middle panel) that 48hrs after transfection the PTN protein was undetectable. The loss of PTN did not affect the state of 3T3-L1 differentiation (Fig. 7B, lower panels).

Figure 7. PTN and PTHR role in mammary epithelial cell migration.

Figure 7

Open in a new tab

A) Mammary epithelial cells, HC-11 and differentiated 3T3-L1 adipocytes were lysed and tested for the presence of PTN, and its receptor PTPRζ1 by Western blotting. Left panel: PTN is highly expressed (upper panel) by 3T3-L1 adipocytes, it is undetectable in HC-11 cells; PTN-receptor PTPRζ1 is highly expressed in HC-11 cells but minimally expressed in the 3T3-L1 adipocytes (middle left panel). Both tubulin and vinculin serve as loading control because of difference in their expression between the cell types. Right panel: differentiated cell lysates (1/5 of total) and serum-free conditioned medium (1/100 of total), collected after 48hrs of incubation, tested for PTN by Western blotting. B) PTN was knocked down in differentiated 3T3-L1 adipocytes by pooled siRNA and C) PTHR was knocked down in 3T3-L1 using pooled PTHR-siRNA as described in Methods (middle panel). The treatments impaired cell migration of HC-11 cells in Boyden chambers (upper panels) (p<0.01 t-test) but did not affect adipocyte differentiation (lower panels). D) PTHR- siRNA induced knockdown of PTHR reduced PTN protein levels. (scrbl in B, C and D means scrambled).

In co-cultures with HC-11 cells, inhibition of PTN production in adipocytes by PTN-siRNA significantly (p<0.001) inhibited HC-11 cell migration (Fig. 7A, upper panel). A similar experiment, but with pooled PTHR-siRNA, was also conducted. This siRNA effectively reduced PTHR protein level, (Fig. 7C, middle panels), did not affect 3T3-L1 differentiation (lower panels) and significantly reduced HC-11 cell migration (Fig. 7C, upper panel). This result created a dilemma because PTHR is a cell surface protein yet it seems to affect, in a paracrine fashion, in physically separated co-culture, migration of mammary epithelial cells. Because, in some cells cAMP was shown to regulate processing of PTN (Mourlevat et al., 2005) and since activated PTHR activates adenyl cyclase and increases cAMP, we asked whether PTHR might be regulating migration by regulating PTN. 3T3-L1 adipocytes were transfected with PTHR-siRNA or its scrambled form and 48hrs later cell lysates were tested by immunoblotting for PTHR and PTN proteins. As shown in Fig. 7D, the reduced level of PTHR protein was accompanied by a strong reduction in full length, 18kDa PTN level. Thus, it is likely, that DN-RARα inhibits directly PTN expression through RA-mediated pathway but that there is further regulation of PTN which is mediated through the effect of DN-RARα on PTHR.

In summary, we show that one of the functions of RARs in the fat pad adipocytes is to maintain the proper ductal elongation and that this effect is paracrine in nature. We also identify 2 proteins expressed by the adipocytes, a membrane receptor, PTHR and a secreted protein, PTN, which might be downstream of PTHR, which are regulate by RAR and have proven function in epithelial cell migration.

Discussion

Using a novel transgenic mouse model we determined that signaling regulated by adipocyte RARs plays an important paracrine role in morphogenesis of the mammary epithelial tree. By expressing a DN-RAR in the adipocytes and using microarray analysis we showed that RAR regulates the expression of several hundreds of mammary gland genes and, subsequently, we identified and characterized two of the RAR up-regulated proteins as participant in adipocyte-epithelial interactions. Few transcription factors have been implicated in this context, lending greater significance to our finding.

Can the effect on mammary ductal tree be traced to mammary adipocytes?

An obvious limitation of the transgenic approach adopted here is that the dominant negative RARα transgene under the aP2 promoter is expressed in all adipose tissues. However, transplantation experiments allowed us to exclude with high degree of certainty systemic effects on mammary phenotype. Loss of Tg-effect when epithelium from Tg-gland was transplanted into epithelium-divested wt-fat pad, combined with preservation of the phenotype when whole Tg-glands were transplanted into wt-mice strongly indicate that the effect is generated locally by the fat pad and not by systemic effects.

The predominant effect exerted by the DN-RARα expression, which causes down-regulation of all 3 RAR isotypes in adipocytes, is the profound impairment of ductal elongation in the virgin mammary gland. It is critical to note that this effect is not due to the reduction in adipose tissue mass or the mammary fat pad. Thus, as made patent by the whole mounts and the quantification of ductal growth (Figs.2A and B), ductal development is curtailed in Tg-glands to a much greater degree than the relatively modest decrease in mammary fat pad size, i.e., ducts fail to elongate despite an expanse of fat pad ahead of the ductal tips.

In a comparative whole mount analysis of glands from Tg and wild type mid-pregnant littermates we found that the Tg-mammary tree, although somewhat extended, did not fill the fat pad but that its lobuloalveolar development was similar to the wt (Fig. S1). Thus, adipose RARs do not appear to contribute to this developmental phase. This partial rescue of the ductal development by the pregnancy hormones was transient as shown by its regression to a primitive state following lactation and gland involution (Results not shown). This is reminiscent of the effect of other genes shown to influence pubertal duct development but not lobuloalveolar differentiation (Howlin et al., 2006; Watson and Khaled, 2008), findings that underline the different programs required for the execution of these mammary developmental milestones.

The DN-RARα transgene also decreased the interscapular brown adipose tissue mass. Brown adipocytes, which can be found in the fat pad, have been implicated in ductal development, but their contribution was traced to a systemic mechanism (Gouon-Evans and Pollard, 2002). In contrast, both our transplantation and in vitro modeling experiments demonstrated that a paracrine mechanism attributed to the fat pad adipocytes is responsible for the aP2.RARaG303E mammary phenotype, thus arguing against a significant role of systemic brown fat. The fact that the population of mammary brown adipocytes declines at a time when the ductal phenotype remains strong, and that the 3T3-L1 cells differentiate in culture into white adipocytes yet provide both pro-migratory and pro-proliferative stimulus, enforces this argument.

Microarrays point to extensive transcriptional regulation by RARs in adipocytes

There might be multiple reasons for the curtailment of the duct length and TEBs regression before they reach the end of the transgenic fat pad. Our data show that when RAR signaling is blocked, adipocytes provide a weaker stimulus for proliferation and migration of epithelial cells. The comparison of gene expression profiles of pubertal wild type and DN-RARα transgenic mammary glands using microarrays, even after applying strict filter criteria, yielded 386 genes whose expression was significantly altered. This indicates the complexity of functions of adipose RARs. Reduction in genes such as cellular retinol-binding protein 1, transglutaminase 2, and midkine, known targets of RA suggest that DN-RAR blocked RAR signaling. Other down-regulated genes with important role in mammary cell proliferation, such as amphiregulin, although validated by RT-PCR, were not pursuit further because they were also expressed in HC-11, Comma 1D and NmuMG epithelial cells, (Results not shown) and thus not appropriate for our aim of testing paracrine effects. Of the 10 validated genes, pleiotrophin and parathyroid hormone receptor (PTHR) fit the criteria of paracrine regulators and were studied in detail. PTHR and its ligand PTHrP, have been shown to be critical for proper mammary gland development (Dunbar et al., 2001; Wysolmerski et al., 1995). PTHR is specifically expressed in the stromal compartment of the mammary gland, surrounding the neck of the TEBs and PTHrP, the ligand for PTHR is expressed in the epithelial compartment and, in the postnatal stage of mammary development and it was shown to be restricted to the TEB cells (Dunbar et al., 1998). PTHR gene knockout is lethal but mice rescued by expression of PTHR in chondrocytes do not form mammary ducts. It is possible that, as published (Kobayashi et al., 2005), it is the level of PTHR expressed that determines the phenotype. It appears that both loss and excess of PTHR affect ductal proliferation (Dunbar and Wysolmerski, 2001; Wysolmerski et al., 1998).

The second protein, pleiotrophin (PTN), was decreased about 23 fold in transgenic glands, the second most downregulated secreted protein from the array analysis. PTN is a heparin-binding growth factor previously shown to stimulate both cell proliferation and migration of numerous cell types. In one study, (Kouros-Mehr and Werb, 2006) PTN was shown, by in situ hybridization, to be localized to TEBs in 5-week old mice but protein presence was not examined. PTN null animals have been produced but their mammary phenotype has not been described (Ochiai et al., 2004). PTN, however, has been shown to regulate branching morphogenesis of the uteric bud during kidney development (Sakurai et al., 2001), while midkine, a close relative of PTN, also down-regulated by DN-RARα expression, has been implicated in the branching morphogenesis of the lung (Kaplan et al., 2003).

Potential mechanism of action of PTHR and PTN

We have shown that PTN and PTHR-mRNA and protein are highly expressed in wt-mammary fat pad adipocytes and that these two proteins play a role in regulation of mammary morphogenesis by adipose RARs. Compared to wt, Tg-adipocytes have lower expression of PTHR and PTN (Fig. 5B). Transcription of these genes is induced in adipocytes isolated from wt-mammary gland or in in vitro atRA-differentiated 3T3-L1 adipocytes and is downregulated by transfection of DN-RARα construct into 3T3-L1 adipocytes (Fig. 6B), confirming that they are regulated by retinoids. The DN-RARα-mediated decrease in PTHR or PTN expression, or their knockdown with siRNA decreases mammary epithelial cell migration.

Because PTHR is a transmembrane, G-protein-coupled receptor, and as such cannot directly be involved in paracrine interactions, we suspected that PTHR might be regulating a downstream secreted protein that can in turn affect epithelial cell migration. Indeed, we found that knockdown of PTHR reduces the level of PTN (Fig. 7D). Thus, adipose RAR induces PTHR expression and this leads to increased expression of PTN which in turn regulates mammary epithelial migration. Although, several receptors for PTN have been described, cell migration has been shown to be induced through PTN binding to a protein tyrosine phosphatase receptor ζ1 (also referred to as PTPRβ/ζ) (Ulbricht et al., 2003) with a second receptor ALK, being now considered a downstream substrates for PTPRζ1 (Perez-Pinera et al., 2007). Binding of PTN to PTPRζ1 receptor inactivates the receptor phosphatase activity, causing increased phosphorylation of target proteins. We have shown that PTPRζ1 receptor is highly expressed in mammary epithelial cells (Fig. 7A) and that conditioned medium derived from adipocytes, which contains PTN (Fig. 7A, right panel) increases the phosphorylation intensity of epithelial cell lysates (Results not shown). PTN-PTPRζ1 signaling substrates include proteins involved in cytoskeleton organization and cell-cell adhesion, among them P190-B RhoGAP, shown to be highly expressed in TEBs, and to lesser degree in ducts, alveolar buds and the stroma. Mice deficient in p190-B RhoGAP display delayed ductal outgrowth from 3–5 weeks of age (Chakravarty et al., 2003). Recent work (Perez-Pinera et al., 2006) has shown that PTN-PTPRζ1 interaction might promote EMT and cell migration. In context of these observations, we propose that at physiological level, atRA through PTHR, and directly, controls PTN level, the level of PTPRβ/ζ activation and thus mammary epithelial cells migration and morphogenesis. The microarray analysis suggests that other genes might control TEB and duct cell proliferation and might have even more profound effects on mammary morphogenesis.

The DN-RARα is over-expressed in all adipose tissue and it can heterodimerize with RXRs (Saitou et al., 1994), and possibly reduce the activity of other nuclear receptors. We found that, although PPARγ expression, a factor crucial for adipose differentiation, was not reduced in the microarray in Tg- glands, the transfection of 3T3-L1 cells with the DN-RARα construct reduced its trans-activating activity in a promoter-reporter assay. (The activity of VDR and TR promoters was unaffected, results not shown). Published reports show that RA treatment early during differentiation can down-regulate PPARγ and inhibit 3T3-L1 differentiation (Xue et al., 1996). There are also reports that PTHR (Rickard et al., 2006) and PTN (Gu et al., 2007) inhibit adipogenesis. However, we found no indication of change in adipose differentiation determined by lipid vacuole accumulation, in adipocytes isolated from the Tg mouse or in 3T3-L1 pre-adipocytes transfected with DN-RARα. Importantly, 3T3-L1 adipocytes treated with a PPARγ antagonist, T0070907, showed no decrease in PTN expression (Results not shown), suggesting that the regulation of PTN levels by DN-RARα expression is mainly due to inhibition of RAR and not PPARγ signaling.

Our current working model proposes that TEB cells express PTPRζ1 and respond to PTN produced by adipocytes that surround TEBs. These adipocytes also express PTHR and both PTN and PTHR are under the control of RAR mediated pathway. It is likely that, as described by others (Dunbar et al., 1998), TEB cell secrete PTHrP that binds to adipocytes PTHR, thus closing the epithelial-adipocyte-epithelial circuitry. Which additional molecules are induced as a result of these interactions remains to be discovered.

In summary we describe a new model which indicates the existence of a paracrine effect of adipocytes on mammary morphogenesis. Mammary ductal elongation in DN-RARα expressing fat pads is severely curtailed during pubertal and adult growth but lobuloalvolar development of pregnancy is unaffected, indicating that this effect is stage and tissue specific. Inhibition of RAR activity in adipocytes has a profound effect on regulation of several hundreds of genes suggesting that there are multiple coordinately RAR-regulated programs that wait to be discovered. For now we show that PTHR up-regulated directly by RAR, and PTN, up-regulated by RAR and, most likely by PTHR signaling, stimulate epithelial cell migration in culture, and most likely contribute to the proper development of the epithelial mammary tree.

Supplementary Material

01
02

Acknowledgments

We are grateful to Dr. Akira Kakizuka (Kyoto University, Japan) for the gift of pCMX-RAR-E, Dr. Bruce Spiegelman (Dana-Farber Cancer Institute, Boston, MA) for the PBks-AP2 vector, to Dr. Phillip Scherer (Albert Einstein College of Medicine, NY) for 3T3-L1 cells and to Yeriel Estrada for help with some experiments. This work was supported by the NCI grant CA54273 and the Samuel Waxman Cancer Research Foundation.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Alvarez R, Checa M, Brun S, Vinas O, Mampel T, Iglesias R, Giralt M, Villarroya F. Both retinoic-acid-receptor- and retinoid-X-receptor-dependent signalling pathways mediate the induction of the brown-adipose-tissue-uncoupling-protein-1 gene by retinoids. Biochem J. 2000;345(Pt 1):91–7. [PMC free article] [PubMed] [Google Scholar]
  2. Chakravarty G, Hadsell D, Buitrago W, Settleman J, Rosen JM. p190-B RhoGAP regulates mammary ductal morphogenesis. Mol Endocrinol. 2003;17:1054–65. doi: 10.1210/me.2002-0428. [DOI] [PubMed] [Google Scholar]
  3. Couldrey C, Moitra J, Vinson C, Anver M, Nagashima K, Green J. Adipose tissue: a vital in vivo role in mammary gland development but not differentiation. Dev Dyn. 2002;223:459–68. doi: 10.1002/dvdy.10065. [DOI] [PubMed] [Google Scholar]
  4. Darcy KM, Zangani D, Shea-Eaton W, Shoemaker SF, Lee PP, Mead LH, Mudipalli A, Megan R, Ip MM. Mammary fibroblasts stimulate growth, alveolar morphogenesis, and functional differentiation of normal rat mammary epithelial cells. In Vitro Cell Dev Biol Anim. 2000;36:578–92. doi: 10.1007/BF02577526. [DOI] [PubMed] [Google Scholar]
  5. Dunbar ME, Dann P, Brown CW, Van Houton J, Dreyer B, Philbrick WP, Wysolmerski JJ. Temporally regulated overexpression of parathyroid hormone-related protein in the mammary gland reveals distinct fetal and pubertal phenotypes. J Endocrinol. 2001;171:403–16. doi: 10.1677/joe.0.1710403. [DOI] [PubMed] [Google Scholar]
  6. Dunbar ME, Wysolmerski JJ. Mammary ductal and alveolar development: lesson learned from genetically manipulated mice. Microsc Res Tech. 2001;52:163–70. doi: 10.1002/1097-0029(20010115)52:2<163::AID-JEMT1002>3.0.CO;2-R. [DOI] [PubMed] [Google Scholar]
  7. Dunbar ME, Young P, Zhang JP, McCaughern-Carucci J, Lanske B, Orloff JJ, Karaplis A, Cunha G, Wysolmerski JJ. Stromal cells are critical targets in the regulation of mammary ductal morphogenesis by parathyroid hormone-related protein. Dev Biol. 1998;203:75–89. doi: 10.1006/dbio.1998.9029. [DOI] [PubMed] [Google Scholar]
  8. Gouon-Evans V, Pollard JW. Unexpected deposition of brown fat in mammary gland during postnatal development. Mol Endocrinol. 2002;16:2618–27. doi: 10.1210/me.2001-0337. [DOI] [PubMed] [Google Scholar]
  9. Graves RA, Tontonoz P, Platt KA, Ross SR, Spiegelman BM. Identification of a fat cell enhancer: analysis of requirements for adipose tissue-specific gene expression. J Cell Biochem. 1992;49:219–24. doi: 10.1002/jcb.240490303. [DOI] [PubMed] [Google Scholar]
  10. Gu D, Yu B, Zhao C, Ye W, Lv Q, Hua Z, Ma J, Zhang Y. The effect of pleiotrophin signaling on adipogenesis. FEBS Lett. 2007;581:382–8. doi: 10.1016/j.febslet.2006.12.043. [DOI] [PubMed] [Google Scholar]
  11. Hinck L, Silberstein GB. Key stages in mammary gland development: the mammary end bud as a motile organ. Breast Cancer Res. 2005;7:245–51. doi: 10.1186/bcr1331. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Hovey RC, MacKenzie DD, McFadden TB. The proliferation of mouse mammary epithelial cells in response to specific mitogens is modulated by the mammary fat pad in vitro. In Vitro Cell Dev Biol Anim. 1998;34:385–92. doi: 10.1007/s11626-998-0020-2. [DOI] [PubMed] [Google Scholar]
  13. Howlin J, McBryan J, Martin F. Pubertal mammary gland development: insights from mouse models. J Mammary Gland Biol Neoplasia. 2006;11:283–97. doi: 10.1007/s10911-006-9024-2. [DOI] [PubMed] [Google Scholar]
  14. Kanazawa T, Hosick HL. Transformed growth phenotype of mouse mammary epithelium in primary culture induced by specific fetal mesenchymes. J Cell Physiol. 1992;153:381–91. doi: 10.1002/jcp.1041530218. [DOI] [PubMed] [Google Scholar]
  15. Kaplan F, Comber J, Sladek R, Hudson TJ, Muglia LJ, Macrae T, Gagnon S, Asada M, Brewer JA, Sweezey NB. The growth factor midkine is modulated by both glucocorticoid and retinoid in fetal lung development. Am J Respir Cell Mol Biol. 2003;28:33–41. doi: 10.1165/rcmb.2002-0047oc. [DOI] [PubMed] [Google Scholar]
  16. Kobayashi T, Kronenberg HM, Foley J. Reduced expression of the PTH/PTHrP receptor during development of the mammary gland influences the function of the nipple during lactation. Dev Dyn. 2005;233:794–803. doi: 10.1002/dvdy.20406. [DOI] [PubMed] [Google Scholar]
  17. Kouros-Mehr H, Werb Z. Candidate regulators of mammary branching morphogenesis identified by genome-wide transcript analysis. Dev Dyn. 2006;235:3404–12. doi: 10.1002/dvdy.20978. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Kupumbati TS, Cattoretti G, Marzan C, Farias EF, Taneja R, Mira-y-Lopez R. Dominant negative retinoic acid receptor initiates tumor formation in mice. Mol Cancer. 2006;5:12. doi: 10.1186/1476-4598-5-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Levine JF, Stockdale FE. 3T3-L1 adipocytes promote the growth of mammary epithelium. Exp Cell Res. 1984;151:112–22. doi: 10.1016/0014-4827(84)90361-6. [DOI] [PubMed] [Google Scholar]
  20. Mourlevat S, Debeir T, Ferrario JE, Delbe J, Caruelle D, Lejeune O, Depienne C, Courty J, Raisman-Vozari R, Ruberg M. Pleiotrophin mediates the neurotrophic effect of cyclic AMP on dopaminergic neurons: analysis of suppression-subtracted cDNA libraries and confirmation in vitro. Exp Neurol. 2005;194:243–54. doi: 10.1016/j.expneurol.2005.02.015. [DOI] [PubMed] [Google Scholar]
  21. Ochiai K, Muramatsu H, Yamamoto S, Ando H, Muramatsu T. The role of midkine and pleiotrophin in liver regeneration. Liver Int. 2004;24:484–91. doi: 10.1111/j.1478-3231.2004.0990.x. [DOI] [PubMed] [Google Scholar]
  22. Perez-Pinera P, Alcantara S, Dimitrov T, Vega JA, Deuel TF. Pleiotrophin disrupts calcium-dependent homophilic cell-cell adhesion and initiates an epithelial-mesenchymal transition. Proc Natl Acad Sci U S A. 2006;103:17795–800. doi: 10.1073/pnas.0607299103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Perez-Pinera P, Garcia-Suarez O, Menendez-Rodriguez P, Mortimer J, Chang Y, Astudillo A, Deuel TF. The receptor protein tyrosine phosphatase (RPTP)beta/zeta is expressed in different subtypes of human breast cancer. Biochem Biophys Res Commun. 2007;362:5–10. doi: 10.1016/j.bbrc.2007.06.050. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Rahimi N, Saulnier R, Nakamura T, Park M, Elliott B. Role of hepatocyte growth factor in breast cancer: a novel mitogenic factor secreted by adipocytes. DNA Cell Biol. 1994;13:1189–97. doi: 10.1089/dna.1994.13.1189. [DOI] [PubMed] [Google Scholar]
  25. Rickard DJ, Wang FL, Rodriguez-Rojas AM, Wu Z, Trice WJ, Hoffman SJ, Votta B, Stroup GB, Kumar S, Nuttall ME. Intermittent treatment with parathyroid hormone (PTH) as well as a non-peptide small molecule agonist of the PTH1 receptor inhibits adipocyte differentiation in human bone marrow stromal cells. Bone. 2006;39:1361–72. doi: 10.1016/j.bone.2006.06.010. [DOI] [PubMed] [Google Scholar]
  26. Saitou M, Narumiya S, Kakizuka A. Alteration of a single amino acid residue in retinoic acid receptor causes dominant-negative phenotype. J Biol Chem. 1994;269:19101–7. [PubMed] [Google Scholar]
  27. Sakurai H, Bush KT, Nigam SK. Identification of pleiotrophin as a mesenchymal factor involved in ureteric bud branching morphogenesis. Development. 2001;128:3283–93. doi: 10.1242/dev.128.17.3283. [DOI] [PubMed] [Google Scholar]
  28. Sternlicht MD, Kouros-Mehr H, Lu P, Werb Z. Hormonal and local control of mammary branching morphogenesis. Differentiation. 2006;74:365–81. doi: 10.1111/j.1432-0436.2006.00105.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Ulbricht U, Brockmann MA, Aigner A, Eckerich C, Muller S, Fillbrandt R, Westphal M, Lamszus K. Expression and function of the receptor protein tyrosine phosphatase zeta and its ligand pleiotrophin in human astrocytomas. J Neuropathol Exp Neurol. 2003;62:1265–75. doi: 10.1093/jnen/62.12.1265. [DOI] [PubMed] [Google Scholar]
  30. Wang YA, Shen K, Wang Y, Brooks SC. Retinoic acid signaling is required for proper morphogenesis of mammary gland. Dev Dyn. 2005;234:892–9. doi: 10.1002/dvdy.20570. [DOI] [PubMed] [Google Scholar]
  31. Watson CJ, Khaled WT. Mammary development in the embryo and adult: a journey of morphogenesis and commitment. Development. 2008;135:995–1003. doi: 10.1242/dev.005439. [DOI] [PubMed] [Google Scholar]
  32. Wiens D, Park CS, Stockdale FE. Milk protein expression and ductal morphogenesis in the mammary gland in vitro: hormone-dependent and -independent phases of adipocyte-mammary epithelial cell interaction. Dev Biol. 1987;120:245–58. doi: 10.1016/0012-1606(87)90122-9. [DOI] [PubMed] [Google Scholar]
  33. Wysolmerski JJ, McCaughern-Carucci JF, Daifotis AG, Broadus AE, Philbrick WM. Overexpression of parathyroid hormone-related protein or parathyroid hormone in transgenic mice impairs branching morphogenesis during mammary gland development. Development. 1995;121:3539–47. doi: 10.1242/dev.121.11.3539. [DOI] [PubMed] [Google Scholar]
  34. Wysolmerski JJ, Philbrick WM, Dunbar ME, Lanske B, Kronenberg H, Broadus AE. Rescue of the parathyroid hormone-related protein knockout mouse demonstrates that parathyroid hormone-related protein is essential for mammary gland development. Development. 1998;125:1285–94. doi: 10.1242/dev.125.7.1285. [DOI] [PubMed] [Google Scholar]
  35. Xue JC, Schwarz EJ, Chawla A, Lazar MA. Distinct stages in adipogenesis revealed by retinoid inhibition of differentiation after induction of PPARgamma. Mol Cell Biol. 1996;16:1567–75. doi: 10.1128/mcb.16.4.1567. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Zangani D, Darcy KM, Masso-Welch PA, Bellamy ES, Desole MS, Ip MM. Multiple differentiation pathways of rat mammary stromal cells in vitro: acquisition of a fibroblast, adipocyte or endothelial phenotype is dependent on hormonal and extracellular matrix stimulation. Differentiation. 1999;64:91–101. doi: 10.1046/j.1432-0436.1999.6420091.x. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

01
NIHMS248566-supplement-01.tif (2.9MB, tif)
02
NIHMS248566-supplement-02.doc (19KB, doc)

RESOURCES