Introduction
Retinoic acid receptors (RARs) are the main mediators of the biologic effects of vitamin A, with a long established essential role in the maintenance of the differentiated state of epithelial tissues [
1]. More recently, retinoic acid (RA) and other RAR agonists were found to be growth inhibitory for cancer cell lines
in vitro [
2‐
7], in carcinogen-induced rodent mammary cancer models, [
8‐
10] and in xenograft models of human cancer cell [
11‐
13].
Because pharmacological doses of retinoids were used in the carcinogenesis studies, the question of the ability of physiological retinoid levels of vitamin A-replete animals to exert growth suppressive effects and protect epithelia from neoplastic transformation remained unanswered. It is, however, well established that the RAR-signaling pathway is defective in carcinomas of several organs, including breast, mostly due to reduced expression of
RARβ or
CRBP-1 [
14,
15]. Whether these alterations affect oncogenesis or tumor maintenance, and what might be the mechanism of these effects remains unresolved.
To address the potential role of RAR both in mammary gland morphogenesis and in modifying cancer susceptibility at physiological levels of vitamin-A, we used
RARα1 homozygous null (
RARα1/KO) mice and bi-genic mice generated by crossing
RARα1/KO with MMTV-
wnt1 transgenic mice. We found that loss of
RARα1 produced, in pubertal glands, a highly branched ductal epithelial tree phenotype, which was epithelial cell autonomous. Because retinoids are well known for regulation of embryonic stem cells [
12,
16‐
18], and, in one case, adult HSCs [
19], and because adult stem cells are known to be involved in mammary gland morphogenesis [
20‐
27], we hypothesized that loss of
RARα might affect the mammary stem cell compartment. Moreover, because wnt1 oncogenesis is believed to target mammary stem cells or bi-potent progenitors and might be responsible for progenitor amplification [
20,
28‐
30], we predicted that the
RARα1/KO glands with the complex epithelial tree phenotype, will be more susceptible to wnt1-tumorigenesis.
We now show that epithelial cells derived from the highly branched ductal mammary tree of the
RARα1/KO glands contain higher percentage of luminal progenitor cells, that they form larger primary mammospheres when cultured under adhesion-free and serum-free conditions, and that their MaSC-containing compartment is smaller than that of the wild type (wt)-glands. We further show that activation of RARα by a specific agonist inhibits primary wt-mammospheres growth. Our published work [
31], showed that chronic treatment of MMTV-
wnt1 mice with the same RARα agonist inhibited mammary tumor formation and growth. In spite of all these inhibitory effects of activated RARα we found a significant increase in tumor-free survival when mice null for
RARα1 were crossed with MMTV-
wnt1 transgenic mice. We propose that in vitamin-A replete conditions,
RARα guards normal morphogenesis and influences wnt-induced tumorigenesis at least in part by maintaining a proper hierarchy of the mammary epithelial compartments.
Materials and methods
Animals
RARα1-/- mice (129/Bl-6 background) were generated in Pierre Chambon's laboratory (IGBMC, Strasbourg, France), and MMTV-wnt1 mice (FVB, SJL, and Bl/6 mixed background, with FVB prevalence) were generously provided by Dr. Yi Li (Baylor College of Medicine, Houston, TX, USA). The RARα1-/+ female mice were crossed with hemizygous male MMTV-wnt1 mice followed by intercrossing of F1 progeny that was RARα1+/- and MMTV-wnt1 transgenic, until sufficient mice for study were obtained. Genotyping was carried out by PCR. To obtain RARα1 -/- in FVB background RARα1+/- females were backcrossed eight times with FVB males. All animal experiments were conducted in accordance with the IACUC approved protocols following the Mount Sinai Guidelines.
Whole mounts of mammary glands and quantification of side branching and mature terminal end buds
Mammary glands were excised, fixed in Carnoy's fixative and stained in carmine alum solution as described in reference 32 [
32]. The neonatal glands were photographed and JPG files analyzed using the ImageJ software. The entire mammary tree in the abdominal number four gland/group was circumscribed and the Integrated Density of the area was measured in pixels. The branch points were counted in the three largest ducts in seven pairs (a total of 14 glands) of number four glands from the nipple to the periphery of the fat pad and the mature terminal end buds along the periphery of the fat pad.
Mammary gland transplantation
Fragments (approximately 2 mm
3, approximately 30,000 cells) of mammary glands of 8- to 10-week-old virgin mice (wt or
RARα1/KO) taken from the area between the nipple and the LN, were transplanted into epithelium pre-cleared glands of three-week-old wild type or
RARα1/KO animals, as previously described [
33]. The recipient glands were excised and processed as for whole mounts eight weeks after transplantation. Seven mice were sham transplanted.
Quantification of stem/progenitors using FACS analysis
Mammary gland numbers four and five of seven- to eight-week-old female mice were isolated, minced and digested in trypsin/collagenase. The epithelium was freed of adipocytes by centrifugation and red blood cells were lysed. The remaining cells were treated exactly as described [
34]. For the characterization of MaSC-containing and luminal progenitor containing compartments, cells (1 × 10
6 cells/sample in duplicate) were incubated with PE or APC-conjugated anti-CD24 Ab (BD Pharmingen, San Jose, CA, USA), APC-conjugated anti-CD29 Ab (Invitrogen, Camarillo, CA, USA) and FITC-conjugated anti-CD61 Ab (eBioscience, San Diego, CA, USA). In all other flow cytometry experiments, ALDH activity was determined by the Aldefluor assay according to manufacturer's directions (Aldegen, Durham, NC, USA), followed by CD24 antibody (1:100, biotin conjugated, BD Pharmingen, secondary antibody 1:1000 streptavidin-Alexa 633 and incubation on ice for 30 minutes). A pool of cells from approximately 10 mice was routinely used for each experiment. FACS analysis was carried out using a FACScanto flow cytometer, DIVA software program for acquisition (BD Biosciences) and Flowjo (Treestar, Inc. Ashland, OR, USA) software for analysis.
Growth of mammospheres
Mammary epithelial cells prepared as described for FACS analysis were further dissociated by pipetting and filtering through a 40 μ pore cell strainer. (Occasional small clumps of up to six cells remained). Between 2.5 to 5.0 × 104 cells were plated in ultra low adhesion 24-well plates (Corning, Corning, NY, USA) and incubated in serum-free F12/DMEM 50:50 medium (Cellgro, Mediatech, Inc., Manassas, VA, USA) supplemented with 20 ng/ml EGF and 1:50 B27 Supplement (Invitrogen, Carlsbad, CA, USA). When treatment was indicated, the agonist and antagonists were added after 8 to 10 days of incubation at 10 nM for Am580, 200 nM for at RA, and 10-fold excess of Ro41-5253 RARα antagonist, 100 nM or 2 μM, respectively, and the mammospheres were incubated for approximately eight days with partial medium and drug replenishment every two days. Mammospheres from four to eight wells were combined, allowed to settle at 1 g, dissociated with EDTA, Joklik's medium and a brief trypsin treatment, followed by Soybean trypsin inhibitor (Sigma, St. Louis, MO, USA) and cells from at least duplicate culture pools were counted either in hemacytometer or in 5 μl droplets in duplicates. For second passage the dissociated cells were re-plated as above. In some experiments mammospheres were counted before dissociation.
Expression of RAR by Q-PCR
RNA was isolated from primary mammary epithelial cells using RNeasy kits (Qiagen, Valencia, CA, USA), according to the manufacturer's protocol. cDNA was synthesized from 2 μg of RNA using RevertAid M-MuLV reverse transcriptase (Fermetas, Geln Burnie, MA, USA). Q-PCR was performed using the 2 × SYBR Green master mix (Applied Biosystems, Carlsbad, CA, USA) with 300 nM primers and 40 ng of cDNA. To determine the fold change in expression, the Ct value was averaged from triplicates for each sample. To normalize the RARs expression, triplicates for Ct from the RAR gene was averaged and divided by the average of the triplicate from the GAPDH gene. The PCR primers used were: RARβ2 for: 3' CTT CCT CCT GCA TGC TGC AG 5', RARβ2 rev: 5' GG CAC TGA CGC CAT AGT GGT A 3'; RARγ1 for: 3;' TGG GGC CTG GAT CTG GCT A 5', RARγ1 rev: 5' AT CTC CTC CGA GCT GGT GCT 3'; RARγ2 for: 3' CGG ACT TGA GTC TTT TGC CTG 5', RARγ2 rev: 5' GCT CTG TGT CTC CAC CGA TT 3'.
Activation of wnt-pathway
Cells from three individual MMTV-wnt1 tumors in the first or second passage in culture, were transfected with the β-catenin reporter TOP-FLASH plasmid and renilla luciferase plasmid (pRL-SV4) in triplicates using Lipofectamine LTX (Invitrogen). After approximately 20 hr incubation, 4 μM Ro41-5253 was added and 24 hr later the cells were lysed in 1 × passive lysis buffer (Dual-Luciferase Reporter Assay System, Promega, Madison, WI, USA) and assayed for luciferase using the Dual-Luciferase reporter (DLR) Assay System (Promega).
Identification of bi-potential cells in mammospheres
Eight-day-old mammospheres were dissociated and plated at one cell/20 μl of medium in 96-well plates (400 wells). Wells with single cells (118 total) were marked and incubated in medium with serum with weekly medium changes for a month. Of these only 21 produced colonies; these were stained with anti-CK14 (Neomarkers, Fremont, CA, USA; 1:200, secondary anti-rabbit-Alexa 488, 1:500, Molecular Probes, Invitrogen) and anti-CK18 (Sigma, St. Louis, MO, USA; 1:400, anti-mouse-Alexa 568, 1:500, Molecular Probes) and DAPI, examined under fluorescent Nikon Eclipse E600 microscope and photographed with SPOT-RTTM camera, Spot Diagnostic Instruments (Sterling Height, MI, USA) (× 200).
Kaplan-Meier disease-free survival curves
MMTV-wnt1 and MMTV-wnt1-RARα1/KO mice were palpated weekly (all five-gland pairs were examined) by the same investigator. The appearance of the first palpable tumor was recorded. Mice were euthanized when tumors reached a size of approximately 1 cm3 and parts of the tumors were taken for histological analysis (formalin fixation), RNA analysis (RNAlater), and in some cases for preparation of primary cultures and/or transplantation. Histological analysis of the tumors was performed by the Mutant Mouse Pathology Laboratory at UC Davis and by Mount Sinai's Center for Comparative Medicine and Surgery. Statistical analysis was done using the Log-rank test for the Kaplan-Meier survival studies and by two-way ANOVA test for the tumor growth studies. Significant differences were considered at P < 0.05. Specific methods are indicated in figure legends.
Tumor transplantation
For transplantation, fragments (approximately 1 mm3) of MMTV-wnt1 and MMTV-wnt1-RARα1/KO tumors were transplanted with an implant needle (Fisher, Pittsburgh, PA, USA) into opposite sides between gland numbers three and four of anesthetized eight-week old FVB mice (four mice per experiment, repeated three times). Each transplant consisted of three randomly chosen tumor fragments from a mince. Tumor diameter was measured twice a week with a caliper. For determination of tumor latency and tumor take, primary wnt or RARα1/KO-wnt tumors were dissociated, the single cells counted and 105 or 104 viable cells were injected into FVB mice as above.
Discussion
Our results provide evidence that at physiological levels of vitamin A, a single
RAR isoform,
RARα1 participates in the control of normal branching morphogenesis of the pubertal mammary epithelial tree. This is evidenced by the excessive side budding and secondary branching observed in mice of two genetic backgrounds when
RARα1 is knocked out (Figure
1). Although, control of mammary morphogenesis is complex and can be driven by both systemic [
43‐
45] and local [
46‐
48] effects, our mammary transplantation experiments results (Figure
2 and Table
1) indicate that in this capacity,
RARα1 functions in a predominantly epithelial-cell autonomous fashion. In addition to the effects in the pubertal gland, we found that loss of
RARα1 causes doubling of the rudimentary mammary tree in the neonatal gland (Figure
1), suggesting a possible role for this receptor in embryonic mammary development. A similar branching phenotype has been described in transgenic mice overexpressing a DN-
RAR mutant, which blocks all retinoid signaling [
49], but, to the best of our knowledge, ours is the first described link between a single
RAR isoform and mammary epithelial growth and morphogenesis in vitamin A replete animals.
Based on the increased epithelial cellularity of the KO-glands and published data implicating
RARα in anti-cancer activity, we expected that the
RARα1/KO mice crossed with the MMTV-
wnt1 mice will be more susceptible to wnt1-induced tumorigenesis. Moreover, activation of retinoid signaling has been shown to inhibit the wnt1 pathway in cell culture [
50,
51]. Indeed, treatment of cells derived from three individual wnt1-tumors and transfected with a reporter for wnt1 pathway activation (TCF/β-catenin activity), when treated with an RARα antagonist, Ro415253 produced a significant (
P = 0.01) two-fold increase in the reporter (luciferase) activity (results not shown). However, unexpectedly, we found that MMTV-
wnt1-
RARα1/KO mice had significantly longer tumor-free survival than MMTV-
wnt1 wt-mice (Kaplan Meier analysis, Figure
6). Moreover, in three independent experiments fragments of
wnt1-
RARα1/KO tumors grew much more slowly than fragments of wnt1 tumors when implanted into contra lateral sides of the same mouse, and single cells suspensions obtained from
wnt1-
RARα1/KO tumors showed much longer latency.
How is it possible that active RARα1 which prevents apparent epithelial hyperplasia and blocks wnt pathway activity, also allows a more efficient wnt1-induced oncogenesis? The oncogenic targets of wnt1 are believed to be progenitor/stem cells [
28,
29] and wnt signaling might have a role in mammary stem cells self-renewal [
24]. The uncertainty regarding the precise target comes from the difficulty in comparing stem cell compartments in normal mammary and in wnt1 induced tumors. The situation is further complicated by the findings [
42] that in the pre-neoplastic stage constitutive wnt1 signaling perturbs the epithelial hierarchy, leading to the emergence of aberrant multipotential stem-like cells in the committed luminal cell fractions. How can then our result fit into this complex scheme? We showed that pre-malignant
wnt1-RARα1/KO-glands have the highest content of ALDH
high/CD24
low cells and that pubertal
RARα1/KO mammary gland (without wnt1-expression) contains 1.7 times more of these cells than wt-mammary (Figures
4 and
5). Cells isolated from
RARα1/KO and mammospheres produced by the
RARα1/KO cells form larger mammospheres (Figure
3), possibly because the progenitors might proliferate more rapidly. These progenitors were shown to be elevated during puberty, concomitant with an increased ductal branching and elongation, a phenotype that is enhanced in the
RARα1/KO mice [
52]. At the same time, however, we found a reduction in MaSc-enriched compartment and a reduction in repopulation efficiency of
RARα1/KO mammaryfragments (Figure
2 and Table
1). Although, we did not perform consecutive
in vivo passages of the transplants, the diminished capacity of the
RARα1/KO fragments to repopulate, suggests that they contain fewer stem cells or, at the least, they contain stem cells with diminished activity.
That RARα1 keeps in check the proliferative capacity of the progenitors in the pubertal gland and, that its loss leads to the enlargement of the progenitor compartment, fits with the established anti-proliferative role of RAR and retinoids. For example, p27
kip, a protein that accumulates in response to RAR activation, can limit the self-renewal of some adult tissue progenitors [
53]. It is also possible, that in addition to removing a block to proliferation, loss of
RARα1/KO provides an indirect proliferative stimulus mediated through RARγ1. We showed that
RARα1/KO epithelium has higher levels of
RARγ1-mRNA (Figure
3D), and that this isoform has pro-proliferative activity in mammary cells [
31]. Moreover, wt-mammospheres treated simultaneously with atRA and an RARα antagonist, a combination that allows RARγ (and β) activation (Figure
3C), yielded the highest numbers of cells.
Overall, our current data suggest that the profound reduction in mammary repopulating activity combined with the predominance of progenitors over MaSC-containing compartment in the
RARα1/KO, by reducing wnt1 target might contribute to the delay in wnt1-tumor development. The mechanism for this is unknown, but it has been shown that PTEN loss in HSCs causes a transient expansion of the stem cell pool followed by its depletion [
54,
55]. We have shown that RAR activation inhibits PI3K activity, thus loss of RARα1 might possibly cause a similar effect to that of PTEN deletion. In an elegant study Purton
et al. [
19] have shown that loss of
RARα1, but not RARα, results in reduced numbers of HSCs and increased numbers of more differentiated progenitor cells in the mutant mice. Although, in the mammary it is the RARα that appears to be involved, different RAR isotypes are known to have tissue-specific roles [
17,
56].
Because wnt1 pathway activation is involved in human breast cancer [
57], and RARα has a proven role in breast cancer, the mechanism connecting RARα1 to control of wnt-tumorigenicity is worthy of further study. It remains to be determined whether a reduction in stem cell level in
RARα1/KO mice translates into reduced cancer stem cell level, what is the mechanism responsible for this reduction and whether it is the cause of increased tumor-free survival and a delay in growth of transplantable tumors.
In summary, we showed that loss of RARα1 leads to reduced mammary stem cell content and an increase in wnt1-tumor free survival in mice. Under physiological conditions RARα1 signaling appears to control the content of mammary stem/progenitor cell compartments and affect the proper morphogenesis of neonatal and pubertal mammary gland.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
EC assisted in collection and assembly of data, data analysis and editing of the manuscript. LO assisted in conception and design, collection and assembly of data, data analysis, and manuscript writing. SB assisted in collection and assembly of in vivo data. CM assisted in collection, assembly and analysis of morphogenesis data. EFF assisted in conception and design, collection and assembly of data, and data analysis. All authors read and approved the final manuscript.