Author contributions: Y.W.: data collection, data analysis, and manuscript writing; J.D. and D.L.: data collection and data analysis; L.L.: data collection; S.S.: data analysis and manuscript writing; Y.L.: conception and design, data analysis, and financial support; M.L.: conception and design, data analysis, financial support, and final approval. Y.W. and J.D. contributed equally to this article.
Lester and Sue Smith Breast Center, Baylor College of Medicine, Houston, Texas, USA
The key signaling networks regulating mammary stem cells are poorly defined. The leucine-rich repeat containing G protein-coupled receptor (Lgr) family has been implicated in intestinal, gastric, and epidermal stem cell functions. We investigated whether Lgr4 functions in mammary gland development and mammary stem cells. We found that Lgr4−/− mice had delayed ductal development, fewer terminal end buds, and decreased side-branching. Crucially, the mammary stem cell repopulation capacity was severely impaired. Mammospheres from Lgr4−/− mice showed decreased Wnt signaling. Wnt3a treatment prevented the adverse effects of Lgr4 loss on organoid formation. Chromatin immunoprecipitation analysis indicated that Sox2 expression was controlled by the Lgr4/Wnt/β-catenin/Lef1 pathway. Importantly, Sox2 overexpression restored the in vivo mammary regeneration potential of Lgr4−/− mammary stem cells. Therefore, Lgr4 activates Sox2 to regulate mammary development and stem cell functions via Wnt/β-catenin/Lef1. Stem Cells2013;31:1921-1931
Deciphering the complex signal transduction pathways regulating stem cell activity remains an ongoing challenge. Key pathways controlling mammary gland stem cells include the Wnt, Notch, and Hedgehog pathways [1-3]. However, crucial components of these pathways and how they determine mammary stem cell behavior remain unexplored.
The leucine-rich repeat containing G protein-coupled receptors (LGRs) 4, 5, and 6 are vital for stem cell maintenance in several systems. Lgr5 marks stem cells in the intestine [4, 5], the stomach , and the hair follicle . Likewise, Lgr6 is a stem cell marker in the skin . Lgr4 (also called GPR48) is required for intestinal stem cell maintenance ex vivo [9, 10]. Lgr4 loss causes developmental defects in multiple organs, including gall bladder , male reproductive tracts [12-14], and hair follicle development  as well as reduced embryonic growth . Recently, three separate groups have identified members of the R-spondin family of Wnt potentiators as ligands for LGR4–6, directly linking these receptors to stem cell signaling [10, 17, 18]. Since disruption of Lgr4 causes developmental defects in multiple organs, a role in mammary stem cell regulation is likely; however, the function of Lgr4 in mammary stem cells has not been defined.
We have previously described the roles of Lgr4 in early eye development [19, 20], midgestational erythropoiesis , osteoblast differentiation and postnatal bone modeling , spermatogenesis , and dextran sodium sulfate-induced inflammatory bowel disease . Here, we report that the loss of Lgr4 compromised mammary development and stem cell self-renewal and function. Lgr4−/− mouse mammary glands have impaired terminal end buds (TEBs) and delayed ductal tree development. Defects in the basal epithelium were observed in both ducts and TEBs in Lgr4−/− mice. Crucially, the mammary stem cell pool was reduced in Lgr4−/− mice, and Lgr4−/− mammary stem cells had diminished regenerative potential. Finally, we discover Sox2 as a key transcription factor regulated by Lgr4 signaling in mammary stem cells.
Materials and Methods
Lgr4 homozygous mutant mice (Lgr4−/−) were generated by microinjecting gene trap-mutated Lgr4 embryonic stem cells into blastocysts of C57BL/6 mice . Mice were backcrossed to C57BL/6 for 5–10 generations or to FVB/NJ for 6–12 generations. Experimental procedures were approved by the animal care and use committee of Texas A&M University.
BrdU Incorporation, Tissue Harvest, Histology, and Whole-Mount
Two hours before euthanasia, BrdU 5-bromo, 2′deoxyuridine (100 μg/g b.wt., Sigma, St. Louis, MO, www.sigmaaldrich.com) was injected intraperitoneally into some mice to assay cell proliferation. Mammary glands were excised and processed as described in , with the following changes: formalin fixation for histology was for 4 hours to overnight at 4°C. Fixation for whole mount was done in acetic acid/ethanol for 2–4 hours at room temperature, and staining was with either carmine alum or Neutral Red. After taking whole-mount pictures, tissues were embedded in paraffin for sectioning and analysis. Whole-gland β-galactosidase staining was performed as described . A minimum of three animals per genotype and time point were analyzed.
Immunohistochemistry, Immunofluorescence, and TUNEL Assay
Immunohistochemistry and immunofluorescence were performed as described previously . MOM, Vectastain Elite ABC Rabbit, or Rat Kits (Vector Laboratories, Burlingame, CA, vectorlabs.com, cat nos. PK-2000, PK-6101, and PK-6104) were used according to manufacturer instructions. Apoptotic cells were determined by the DeadEnd Fluorometric TUNEL (Terminal deoxynucleotidyl transferase dUTP nick end labeling) System (Promega, Madison, WI, www.promega.com). 4′6-diamidino-2-phenylindole (DAPI) counterstain was used to visualize nuclei. TUNEL-positive cells were scored in at least five fields per section, and at least 1,000 cells were counted for each section.
Isolation of Primary Mammary Epithelial Cells and Primary Mammary Organoids
Primary mammary epithelial cells (MECs) were isolated from 12- to 16-week-old virgin female FVB, C57BL/6 mice as described , with the following modifications: initial digestion in Dulbecco's modified Eagle's medium (DMEM)/F-12 medium containing 0.2% Fetal Bovine Serum (FBS), 30,000 U/ml collagenase A (Worthington, Lakewood, NJ), and 1,000 U/ml hyaluronidase (Sigma, St. Louis, MO) for 1–2 hours followed by 5–10-minute digestion in 0.25% Trypsin-EDTA. Primary mammary organoid preparation was performed as previously described .
Mammosphere culture was performed as described in Dontu et al. , except that the culture medium (serum-free DMEM/F12 supplemented with B27 [Invitrogen, Carlsbad, CA, www.invitrogen.com], 20 ng/ml epidermal growth factor [EGF] [Invitrogen], and 20 ng/ml basic fibroblast growth factor [bFGF] [R&D systems, Minneapolis, MN, www.rndsystems.com]) contained 1% methyl cellulose to prevent cell aggregation. Briefly, primary MECs were isolated as described above, and single cells were plated at 10,000 cells per well in six-well, ultra-low attachment dishes. After culturing for 7–10 days, mammospheres with diameter ≥50 μm were counted. Mammospheres were then harvested for serial passage using 70 μm cell strainers, dissociated to single cells with trypsin, and 5,000 cells per well were plated in a six-well or 24-well ultra-low attachment plate and cultured for 10 days before counting and passage.
In Vitro Branching Morphogenesis Assays (Mammary Organoid Culture)
Organoid assays were performed in 24-well plates. Three-dimensional primary cultures were generated as described . Briefly, we embedded 5,000 MECs in 50 μl of growth factor-reduced Matrigel. Fragment organoids were obtained by embedding purified epithelial fragments into Matrigel and stimulated with 2.5 nM bFGF (R&D systems) according to .
Mammary Transplantation and Analysis
Primary MECs (100–10,000) from 13- to 15-week-old Lgr4−/− and wild-type (WT) glands were suspended in a 1:1 mix of Matrigel and DMEM/F-12 medium and injected into cleared mammary fat pads of 3-week-old females. After 2 months, the fat pads were dissected, processed, and stained with Neutral Red as described above, then analyzed to evaluate the extent of mammary outgrowth. A minimum cutoff of 5% fat pad filling was used to indicate positive ductal outgrowths.
Cell Labeling and Flow Cytometry
MECs were isolated as described above. We stained Lgr4−/− and WT MECs with lineage antibodies (biotinylated CD45/CD31/TER119 and Streptavidin-APC), CD24-PE, and CD49f-FITC on ice for 20 minutes in Hanks' balanced saline solution with 5% FBS, and established the fluorescence-activated cell sorting (FACS) gating and positioning of the mammary repopulating unit (MRU), colony forming cell (CFC), and myoepithelial (MYO) populations ; gates were set according to the isotype control antibody labeled with the corresponding fluorochromes. All cell sorts were performed using FACSAria (Beckton Dickinson, Franklin Lakes, NJ, www.bd.com).
Quantitative Real-Time PCR Analysis
Mammosphere culture was performed as described above, and total RNA was extracted using the RNeasy Mini Kit (Qiagen, Valencia, CA). Following reverse transcription to generate single-stranded cDNA, quantitative PCR (qPCR) was performed using the SYBR Green-based system on the ABI 7900HT according to the manufacturer's instructions (Applied Biosystems, Carlsbad, CA). ΔΔCt values were calculated. 18s as a reference was used for normalization. Primers are described in detail in Supporting Information Table S1.
Human SOX2 Reporter Constructs
The Human SOX2 proximal promoter region was obtained by PCR using human genomic DNA, then subcloned to pGL3-Basic Vector (Promega). For generation of the promoter region, forward primer: 5′-TTGGAGCTCGTGAGTTTGACAGTAACAGGCTAGG-3′ and reverse primer: 5′-TGCACATCTCAGCCACGTAGAAACCTTTGTATC-3′ were used. The constructs were verified by sequencing.
Cell Transfection and Luciferase Assays
Human MCF10A cells were maintained in DMEM/F-12 supplemented with 15 mM Hepes buffer, 5% horse serum, 10 μg/ml insulin, 20 ng/ml EGF, 100 ng/ml cholera toxin, 0.5 μg/ml hydrocortisone, 100 μg/ml penicillin, and 100 μg/ml streptomycin. For the luciferase reporter assay, 48 hours after plasmid transfection, cells were lysed and harvested in Reporter Lysis Buffer (Promega, Madison, WI). Extracts were normalized by β-galactosidase activity using the Galacto-Light plus β-Galactosidase Reporter Gene Assay System (Promega).
Chromatin Immunoprecipitation Assays
The chromatin immunoprecipitation (ChIP) assay was performed on lysates of MECs derived from Lgr4+/+ or Lgr4−/− mice grown in organoid culture using the Chip Assay kit (Cell Signaling Technologies, Beverly, MA, www.cellsignal.com), according to the manufacturer's instructions. The regions of LEF1 sites of Sox2 in both human and mouse species were amplified from the immunoprecipitated chromatin using the primers in Supporting Information Table S1.
The mouse Sox2 lentivirus plasmid was constructed in pLVX-IRES-ZsGReen vector (Clontech, Mountain View, CA, www.clontech.com). Lentivirus was produced by transient transfection in 293T cells. Mammary cells were isolated from 12- to 16-week-old virgin female glands as described above, followed by plating for 30 minutes to deplete stromal cells. The suspended MECs were collected and plated in 5% FBS Epicult medium (Stem Cell Technologies, Vancouver, Canada, www.stemcell.com) with virus. Twenty-four hours after infection, cells were washed twice with DMEM/F-12 medium. Rescue by overexpression of Sox2 was tested by in vitro colony assay and in vivo transplantation assay.
In Vitro Colony Assay
Primary MECs were mixed with 100 μl of Matrigel mix (a 1:1 mixture of 5% FBS Epicult medium [Stem Cell Technologies, Vancouver, Canada, www.stemcell.com] and Matrigel) and plated into Matrigel-coated eight-well chambers. After 8–10 days, Matrigel culture was fixed in 4% paraformaldehyde and photographed.
Primary antibodies used in these experiments were against: Lgr4 (Abcam, Cambridge, MA, www.abcam.com), keratin 8 (TROMA1, DSHB, Iowa City, Iowa, www.dshb.biology.uiowa.edu), keratin 5 (Covance, Princeton, NJ, www.covance.com), keratin 14 (Covance), Ki67 (Novocastra, Newcastle upon Tyne, UK, www.novocastra.co.uk), Lgr6 (Epitomics, Burlingame, CA, www.epitomics.com), BrdU (Beckon Dickinson, Franklin Lakes, NJ), LEF1 (Cell Signaling Technologies, Beverly, MA, www.cellsignal.com), CD45/CD31/TER119 (Ebioscience, San Diego, CA, www.ebioscience.com), Streptavidin-APC, CD24-PE, and CD49-FITC (BD Biosciences, San Jose, CA, www.bdbiosciences.com). Secondary antibodies were from Cell Signaling Technologies except for secondary antibodies for FACS (BD Biosciences).
Apart from mammary transplantation experiments, all comparisons were analyzed using two-tailed Student's t tests. p-Values of .05 or less were defined as statistically significant. Error bars represent SD. Limiting dilution analysis was performed according to  to estimate mammary stem cell frequency, and the Sox2 rescue of mammary outgrowth potential was analyzed using Fisher's exact test.
Lgr4 Loss Impairs Mammary Development
To explore the role of Lgr4 in the mammary gland, we used Lgr4 mutant mice in which a gene-trapping cassette encoding the β-galactosidase (β-gal) protein was inserted into the first intron of the Lgr4 gene . We assessed β-gal activity in Lgr4+/− mammary glands to track Lgr4 expression. In 6-week-old virgin mice, β-gal activity was detected in most terminal end bud (TEB) cap cells and in some body cells (Fig. 1A) as well as in mammary duct basal cells (Fig. 1B). Using immunohistochemistry staining to detect Lgr4 in age-matched WT mice, we confirmed the Lgr4 expression pattern reported by β-gal (Fig. 1C, 1D). In addition, we detected occasional staining of luminal epithelial cells (Fig. 1D). In mature ducts of 13-week-old virgin mice, β-gal activity and anti-Lgr4 immunohistochemical staining were limited to the basal layer (Supporting Information Fig. S1), where mammary stem cells have been suggested to reside [27, 32]. The Lgr4 family members Lgr5 and Lgr6 are also receptors for R-spondins. Lgr5 has been reported in rare basal MECs [33, 34], a pattern we also observed using lacZ staining of Lgr5+/− mice (data not shown). Lgr6 is expressed in occasional clusters of luminal epithelial cells; we did not observe Lgr6 expression in basal cells, and the Lgr6 expression pattern was unaffected by Lgr4 genotype (Supporting Information Fig. S2).
The mammary glands of 6-week-old Lgr4−/− mice exhibited marked developmental defects. The number of TEBs was reduced by 41% in Lgr4−/− mice compared to WT littermates (p = .0016, n = 5; Fig. 1E, 1F). Ductal elongation and branching morphogenesis were also impaired: 6-week-old WT ducts extended past the inguinal lymph node into the distal fat pad, while Lgr4−/− ducts ended approximately at the lymph node (p < .001, n = 5; Fig. 1E, 1G). Furthermore, the number of side branches per gland was significantly reduced (p = .0006, n = 5; Fig. 1E, 1H). The decrease in side-branching persisted into adulthood (Supporting Information Fig. S1C–S1F). Heterozygous mice displayed normal ductal outgrowth similar to WT (Fig. 1F–1H), suggesting that one copy of Lgr4 is sufficient for normal mammary development.
To investigate potential causes of the reduced TEBs and delayed ductal elongation, we examined cell proliferation and apoptosis in 6-week-old WT or Lgr4−/− mammary glands. While body cell proliferation was similar to WT, cap cell proliferation in Lgr4−/− TEBs was markedly decreased (Fig. 1I and Supporting Information Fig. S3A). In addition, Lgr4−/− TEBs displayed increased apoptosis (Supporting Information Fig. S4). Finally, while luminal cell proliferation in mature Lgr4−/− ducts was similar to WT, basal Lgr4−/− cells had a sharp decrease in proliferation (25% ± 10% vs. 10% ± 5%, p = .028, n = 5; Fig. 1J and Supporting Information Fig. S3B), suggesting a potential effect of Lgr4 loss on mammary stem cell and basal progenitor cell proliferation.
Lgr4 Loss Impairs Lineage Specification, Branching Morphogenesis, and Basal Cell Colony Formation
To test whether these developmental defects resulted from impaired stem cell functions and altered cellular hierarchy and lineage specification, we first costained cell lineage markers cytokeratin 14 (K14) and cytokeratin 8 (K8) in 4–6-week-old Lgr4−/− mammary epithelia. Lgr4+/+ TEBs showed a distinct, organized K5+ cap cell layer (proposed to contain the mammary stem cells) surrounding well-ordered body cells; however approximately 37% of Lgr4−/− TEBs exhibited a more disorganized body cell layer occasionally breaking through a discontinuous cap cell layer to contact the stroma (n = 8, Fig. 2A, arrowheads indicate gaps). Interestingly, a large number of Lgr4−/− ducts contained cells that were double positive for K14 and K8, whereas no double-positive cells were detected in WT ducts of the same age (p = .026, n = 6; Fig. 2B). These data suggest a role for Lgr4 in lineage specification and ductal morphogenesis. In accord with this suggestion, Lgr4−/− mammary ducts occasionally exhibited multiple layers of luminal cells as well as mis-localization of basal cells in the lumen (Supporting Information Fig. S5A). These defects in ductal morphogenesis and lineage specification were also observed in outgrowths of Lgr4−/− MECs transplanted into cleared mammary fat pads (Supporting Information Fig. S5B), indicating that these Lgr4 functions are intrinsic to the mammary epithelium per se. Of note, some of the Lgr4−/− ductal growths lacked a basal cell layer (Supporting Information Fig. S5B right panel), further confirming a crucial role of Lgr4 in the mammary basal compartment.
We then examined the colony forming abilities of cells expressing Lgr4 compared to cells not expressing Lgr4 from the same mammary gland. According to Stingl et al. , the mammary stem cell-enriched cell population generates small, solid colonies when grown in Matrigel, whereas the luminal-enriched cell population forms hollow, acinar-like colonies. When we FACS sorted mammary cells from Lgr4+/− mice on the basis of lacZ staining, we found an increase in small, solid colony formation and decreased acinar-like colony formation in the lacZ+ (i.e., Lgr4+) population (Supporting Information Fig. S6). This further supports a role for Lgr4 in basal mammary cells and mammary stem cells.
Next we used a mammary organotypic culture model to study the role of Lgr4 in regulating mammary stem cell functional differentiation [28, 35, 36]. Lgr4−/− MECs generated much fewer branching organoids than WT controls (p = .0142, n = 4, Fig. 2C, 2D), and the average number of branches was one third of that of WT controls (p < .0001; Fig. 2E). These in vivo branching morphogenesis defects further suggest a role for Lgr4 in regulating mammary stem cells. In addition, WT organoids showed an outer layer of K14+ cells surrounding an inner population of K8+ luminal cells, with the appearance of a hollow lumen; in contrast, Lgr4−/− organoids had only sparse K14+ basal cells and a solid mass of K8+ cells filling the entire interior (Fig. 2C, right). This organoid defect resembles the mammary ductal phenotype in the Lgr4−/− mice (Supporting Information Fig. S4A). Together, these observations suggest that loss of Lgr4 impairs mammary morphogenesis, lineage specification, and stem cell function.
The above experiments demonstrated an impaired basal compartment in Lgr4−/− mice. In agreement with this, by staining single cell preparations of mammary glands for CD24 and CD49f, FACS analysis showed a decreased ratio of basal (CD24lo/CD49f+) versus epithelial cells (CD24+/CD49flo Fig. 3A–3C). Importantly, these Lgr4−/− CD24lo/CD49f+ basal cells resulted in fewer and smaller colonies in Matrigel than their WT counterparts (Fig. 3D–3F), again suggesting that Lgr4 is important for stem cell proliferation and differentiation.
Lgr4 Ablation Impairs Stem Cell Self-Renewal and Regenerative Capacity
To further investigate the role of Lgr4 in mammary stem cells, we isolated MECs from 12- to 16-week-old Lgr4+/+ and Lgr4−/− mice, stained for CD24 and CD49f as described by Stingl et al. , and observed a similar distribution of WT mammary cells into the MRU, CFC, and MYO populations based on relative CD24 and CD49f expression. When we used FACS to quantify the proportion of cells in each population, the CFC and MYO populations were not significantly affected by Lgr4 loss, but the MRU population in Lgr4−/− mice was less than half of that in WT (p = .006, n = 10; Fig. 4A, 4B), suggesting that the stem cell pool is sharply decreased in Lgr4−/− mammary glands.
Next, we performed a mammosphere formation assay to quantify self-renewal and differentiation. Loss of Lgr4 had no apparent effect on either the number (n = 5; Fig. 4C) or size (data not shown) of primary mammospheres relative to WT, suggesting that short-term proliferation (i.e., the population of cells with limited replicative potential) is not significantly affected by Lgr4 loss. Lgr4−/− epithelial cells gave rise to significantly smaller secondary mammospheres than WT cells (61 μm vs. 72 μm p = .0002, Fig. 4D, 4E), without significant change in the number of secondary mammospheres (n = 5; Fig. 4C). Lgr4−/− cells initiated tertiary mammospheres at a much lower rate than WT cells (p = .0034, n = 5; Fig. 4C). This decline was accelerated in successive generations, and by the fifth passage there were almost no mammospheres observed in the Lgr4−/− group (p = .0011, n = 5; Fig. 4C). These data suggest a defect in long-term mammary stem cell self-renewal in Lgr4−/− mice.
To functionally assess Lgr4−/− mammary stem cell defects in vivo, we transplanted MECs from adult Lgr4−/− and WT mice into cleared fat pads of 3-week-old WT recipients in a limiting dilution assay. After 2 months, transplanted mammary tissue was harvested to evaluate the extent of mammary outgrowth. In three independent transplantation experiments, cells from Lgr4−/− mammary glands exhibited an 80% decrease in repopulating frequency compared to WT cells (Table 1, p < .001). Furthermore, the extent of Lgr4−/− ductal outgrowth was decreased (Table 1), strongly suggesting that Lgr4 loss leads to compromised stem cell activity.
Table 1. Mammary outgrowths derived from Lgr4+/+ and Lgr4−/− mammary cells
Number of injected cells
p < 0.001 100% 80%∼100% 50%∼80% 30%∼50% 5%∼30% 0%
Estimated repopulating frequency
Canonical Wnt Signaling Rescues the Lgr4-Null Mammary Organoid Phenotype
Lgr4 has recently been reported to potentiate Wnt signaling after binding to its ligand R-spondin [10, 17, 18]. To test whether the Wnt pathway mediates Lgr4 in mammary stem cell maintenance, we generated mammary organoids from WT or Lgr4−/− MECs and treated them with Wnt3a or R-spondin1 alone or in combination. In the WT culture, treatment of R-spondin1 or Wnt3a alone increased the fraction of branching organoids (Fig. 5A, 5B), suggesting that R-spondin1 or Wnt3a can promote mammary stem cell function as previously reported . Combinatorial treatment with both Wnt3a and R-spondin1 resulted in a significant increase in the percentage of branching organoids arising from WT MECs (Fig. 5A, 5B), suggesting that R-spondin1 and Wnt3a can synergize in stimulating mammary gland stem cell functions. However, in the Lgr4−/− culture, R-spondin1 treatment alone did not affect the fraction of branching organoids (Fig. 5A, 5B), confirming the role of Lgr4 as a potential receptor for R-spondin1-mediated response. Wnt3a treatment significantly increased the frequency of branching organoids from Lgr4−/− MECs, but this new level did not reach that of WT cells treated with Wnt3a, suggesting that Wnt can partially activate its downstream components in the absence of Lgr4 and that Lgr4 potentiates Wnt signaling in the mammary gland. Adding R-spondin1 to Wnt3a treatment did not increase the fraction of branching organoids from the Lgr4−/− culture, confirming that Lgr4 is obligatory for the R-spondin1 effect on Wnt signaling and mammary stem cells.
To identify key transcriptional targets by which Lgr4 and Wnt signaling regulate mammary stem cell functions, we used qPCR to compare Lgr4−/− and WT primary mammospheres for expression differences in genes that are known to be critical for stem cell maintenance in breast and other organs. These include known and potential Wnt signaling targets , and members of the Notch  and Hedgehog  signaling pathways, which have been reported to be critical for the maintenance of stem cells in the mammary gland and in other organs. Hedgehog pathway component expression was unchanged, along with most Notch pathway members, although two Notch target genes (Hey1 and Gata3) were altered in Lgr4−/− mammospheres (Supporting Information Fig. S7A, S7B). However, many Wnt target genes including Axin2, Sox9, Nanog, Cyclin D1, and Cyclin E were downregulated in Lgr4−/− mammospheres (p < .05; Fig. 5C). These data imply that the Wnt/β-catenin signaling pathway is reduced in Lgr4−/− mammospheres and suggest a role for the Wnt pathway in mediating Lgr4 signaling in mammary stem cells.
Lgr4 Regulates Mammary Stem Cells Through Sox2
Sox2 was reported to be upregulated by Wnt in the Xenopus retina , but whether Sox2 is direct target of Wnt signaling is still unknown. Our gene expression screening indicated that Sox2 was downregulated by more than 50% in Lgr4−/− mammospheres (Fig. 5C), suggesting that Sox2 is a target gene of Lgr4-mediated signaling. Since Lgr4 is involved in R-spondin-mediated Wnt signaling, we examined whether Sox2 is a direct target of Wnt signaling by analyzing the Sox2 promoter regions using rVista 2.0 (http://rvista.dcode.org/). Two consensus LEF1 sites (CTTTGTT, between −1,484 and −1,490 bp, and TACAAAG, between −214 and −220 bp) were found in approximately 2 kb of the human and mouse Sox2 promoter regions. To test whether Lgr4 trans-activates Sox2 through those LEF1-binding sites, we constructed vectors for either overexpression or knockdown of human Lgr4 and performed Sox2 luciferase reporter assays in MCF10A cells. As shown in Figure 6A, overexpression of Lgr4 caused an eightfold increase in luciferase reporter activity compared to the control vector. Conversely, knockdown of Lgr4 led to a threefold decrease in luciferase reporter activity (Fig. 6A). These data suggest that Sox2 transcriptionally responds to Lgr4 expression. To test whether the putative LEF1-binding sites in the Sox2 promoter are indeed occupied by LEF1, we performed a ChIP of primary MEC lysates. In WT cells, LEF1 bound to the −220 to −214 bp site (but not to the −1,484 to −1,490 bp site), and this binding was significantly reduced in Lgr4−/− MECs (Fig. 6B). Together, these data suggest that Sox2 is a direct target of Wnt signaling and that Lgr4 regulates Sox2 expression through Wnt signaling in the mammary gland.
To investigate whether Sox2 is the crucial target gene whose protein product mediates Lgr4 and Wnt signaling in regulating mammary stem cells, we constructed a lentiviral vector overexpressing mouse Sox2 cDNA, used it to infect Lgr4−/− primary MECs, and performed in vitro colony assays and in vivo regeneration assays. In Lgr4−/− MECs, overexpression of Sox2 significantly enhanced colony formation relative to green fluorescent protein (GFP)-infected MECs (p = .018; Fig. 6C), effectively restoring the colony formation potential to the level of WT MECs that also overexpressed Sox2 (p = .139; Fig. 6C). To further confirm the role of Sox2 in regulation of mammary stem cell self-renewal, we injected 10,000 Lgr4−/− MECs infected with lentivirus overexpressing Sox2 or control GFP into the cleared fat pads of syngeneic mice. Eight weeks after transplantation, we quantified the ductal outgrowths. In 11 mice transplanted with Lgr4−/− MECs infected by control virus, only three exhibited ductal outgrowths (Fig. 6D). However, of 11 mice transplanted with Lgr4−/− MECs infected by Sox2-overexpressing virus, 10 displayed ductal outgrowths (p = .0075; Fig. 6D). This Sox2-mediated rescue of ductal growth potential was robust since the rescued cell outgrowth frequency was similar to that of transplantation of 10,000 WT MECs (p = 1.0; Fig. 6D). Taken together, these observations strongly suggest that Sox2 mediates Lgr4 and Wnt signaling to control mammary stem cell self-renewal and functions.
We report here that genetic ablation of Lgr4 causes mammary stem cell depletion and functional impairment. There was an 80% decrease in the number of functional mammary stem cells in Lgr4−/− mammary glands as determined by mammary regeneration following transplantation, and the mammosphere formation ability of Lgr4−/− MECs diminished rapidly after three passages, indicating a crucial role for Lgr4 in mammary stem cell self-renewal. Branching morphogenesis was impaired in the absence of Lgr4, resulting in a persistent decrease in ductal tree side branching in Lgr4−/− mice. Mechanistically, Lgr4 loss resulted in decreased expression of Wnt target genes, especially Sox2, which encodes a transcriptional factor that is also a key factor in cell reprogramming. Sox2 overexpression in Lgr4−/− mammary cells restored mammary regenerative potential, demonstrating Sox2 as a critical effector of Lgr4 in mammary stem cell self-renewal. We also found that LEF binding to the Sox2 promoter was decreased in Lgr4−/− mammary cells, and that treatment with Wnt3a rescued the impaired organoid branching morphogenesis of Lgr4−/− mammary cells. Altogether, our data support a model where R-spondin binding to Lgr4 activates Wnt/β-catenin signaling, resulting in Sox2 transcriptional activation, which controls mammary stem cell self-renewal and functions (Supporting Information Fig. S8).
A role for Lgr4 in stem cell maintenance fits well with the existing literature on developmental consequences of Lgr4 loss. Several groups including ours have reported that Lgr4 regulates embryonic growth , postnatal male reproductive duct system development [12–14], gall bladder formation , hair follicle development , early eye development [19, 20], midgestational erythropoiesis , bone formation , and homeostasis of electrolytes and blood pressure . The breadth of organs affected by Lgr4 loss, and their varying origins from all three germ layers, is initial evidence for a potential role of Lgr4 in stem cell function. Furthermore, the Lgr4 homologs Lgr5 and Lgr6 have established roles regulating stem cells of various tissues, including Lgr5 in small intestine [4, 5], stomach, and hair follicle stem cells [6, 7], and Lgr6 as a marker for the most primitive type of epidermal stem cells responsible for both hair follicle and skin renewal . Finally, several groups have identified the R-spondin family of Wnt potentiators as ligands for LGRs 4–6 [10, 17, 18], with increased Wnt-induced LRP6 phosphorylation and downstream β-catenin and planar cell polarity signaling  or decreased ZNRF3 ubiquitination of Frizzled and LRP6  following R-spondin binding to LGRs. These findings provide a mechanistic model connecting Lgr4 to Wnt signaling and make plausible a role for Lgr4 in mammary stem cell regulation.
Extensive data support a critical role for Wnt signaling in mammary stem cells [40-43] and branching morphogenesis . Exogenous Wnt3a treatment expanded the mammary stem cell pool in vitro and maintained their functional capacity to reconstitute mammary glands upon transplantation . Sox2 is a key pluripotency transcription factor regulating embryonic and neuronal development, and is implicated in adult stem and progenitor cells in a range of epithelial tissues, including mammary epithelium . Sox2-expressing cells are essential for tissue renewal throughout life . A considerable body of literature describes connections between Wnt signaling and Sox2 in various systems, with complementary or opposing effects depending on the tissue examined [37, 47-51]. Our finding that Lgr4 regulates Sox2 expression through R-spondin/Wnt/β-catenin builds upon the emerging theme of context- and tissue-dependent interactions between Wnt and Sox2. While we have not ruled out alternative pathways as mediators of Lgr4 signaling in mammary stem cells (e.g., Gα/PKA/CREB), to our knowledge, this is the first report of Lgr4 as a regulator of mammary stem cells through Sox2.
A widely accepted model of the mammary cell hierarchy posits that a rare population of mammary stem cells is able to generate multipotent progenitor cells that individually have more limited lineage generation capability, whereas larger populations of lineage-committed progenitor cells are able to divide and produce daughter cells of only that specific lineage . Corresponding to this hierarchy is the replicative potential of different cell populations, with mammary stem cells having potentially unlimited replicative potential, whereas committed progenitors are able to divide a limited number of times before differentiating. The gold standard for mammary stem cell function is ductal outgrowth formation following transplantation; in contrast, assays such as primary mammosphere formation likely measure a combined effect of stem cell and progenitor cell division (at least until progenitor cells cease dividing after several passages). Our data suggest that Lgr4 is required for mammary stem cell self-renewal, as illustrated by both decreased outgrowth formation and diminished late-passage mammosphere formation, whereas the lack of effect of Lgr4 loss in primary mammosphere formation argues that Lgr4 plays a different role in basal progenitor cells, perhaps contributing to preventing a luminal cell fate in this population.
Apart from impaired mammary stem cells, we found several additional defects in Lgr4−/− mammary glands. These included delayed mammary duct branching morphogenesis, frequent deficiencies in the TEB cap cell layer (Fig. 2A), persistence of K14+K8+ epithelial cells (Fig. 2B), and occasional luminal filling of mammary ducts (Supporting Information Fig. S4). The delayed branching and the defects in TEB cap cells may reflect the diminished mammary stem cell pool resulting from Lgr4 loss, while the persistence of double K14+K8+ cells suggests a potential role for Lgr4 in the lineage commitment of multipotent progenitor cells. Several defects (fewer TEBs, diminished ductal branching, and decreased mammary stem cell pool) were similar to those reported in Lrp5−/−  and Lrp6−/−  mice, emphasizing the connection between Lgr4 and Wnt signaling in mammary stem cell function. A recent report described Lgr4K5KO mice having a similar delay in branching morphogenesis , but otherwise normal mammary glands, perhaps due to loss of Lgr4 exclusively in cells expressing keratin 5 in Lgr4K5KO mice. Our more severe phenotype in germline knockout mice may result from loss of Lgr4 expression in a broader range of MECs and basal cells and especially in the stem cell pool. At least some of the mammary stem cells lack the expression of keratin markers such as K5 [33, 54]. At any rate, to our knowledge, this is the first report of Lgr4 as a regulator of mammary stem cell function.
In conclusion, we here identify Lgr4 as a key protein in maintenance of mammary stem cell self-renewal, with a significant role in mammary gland development and branching morphogenesis. Furthermore, we characterize the pluripotency transcription factor Sox2 as a target of Lgr4 signaling and identify Wnt/β-catenin signaling as a mediator of Lgr4 signal transduction.
We thank Sung-Gook Cho and Weijia Luo for invaluable technical assistance, Ms. Svasti Haricharan for statistical analysis, and Ms. Vidya Sinha for stimulating discussion and critical review of this manuscript. We acknowledge the assistance of the BCM Cytometry and Cell Sorting Core. We thank Dr. Qingyun Liu for the antibody against Lgr6. This work was supported in part by grants from the National Basic Research Program of China (2012CB910402 to M.L.), the National Natural Science Foundation of China (30930055 to M.L.), the DOD (BC085050 to Y.L. and PC093061 to M.L.), and NIH (R01CA124820 and NIH U54CA149196 to Y.L. and R01CA106479 to M.L.). J. D. was supported by a SPORE career development award (P50-CA058183). Ying Wang is a Ph.D. student in the Graduate Program of College of Life Sciences, Hunan Normal University, Changsha, Hunan 410081, China.
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.