Stem Cell Antigen-1 Deficiency Enhances the Chemopreventive Effect of Peroxisome Proliferator–Activated Receptorγ Activation Free

Stem cell antigen-1 (Sca-1, Ly6A) is a glycerophosphatidylinositol (GPI)-anchored protein that was first identified as a murine marker of bone marrow stem cells (1, 2), and has been used subsequently to enrich for stem and progenitor cells from a variety of tissues (3–10). In the mammary gland, Sca-1–expressing epithelial cells have the capacity to reconstitute the mammary gland in the cleared fat pad (5, 11) and to generate alveolar and ductal structures (12). These findings are consistent with enhanced fluorescence in the alveolar end buds (13) in heterozygous Sca-1^+/EGFP mice (14), where the stem cell niche is believed to reside (15, 16). Even though, these studies suggest that Sca-1–expressing cells behave like stem and progenitor cells, it is unclear whether Sca-1 is a bona fide stem cell marker or serves as a growth modifier through its ability to prime multipotent cell populations (17, 18). In support of the latter possibility, Sca-1 expression is increased in mammary tumors arising in MMTV-Wnt1 (19) and MMTV-erbB2 (20) transgenic mice, and serves as a tumor-initiating factor in murine mammary tumor cells by suppressing TGF-β signaling (21).

PPARs are ligand-activated nuclear receptors that regulate a plethora of signaling pathways controlling fatty acid β-oxidation, glucose utilization, cholesterol transport, adipocyte differentiation, cell proliferation, and survival (22–28). In the context of mammary tumorigenesis, PPARγ agonists inhibit carcinogen-induced preneoplastic mammary lesions (29) and delay mammary tumor development (30–32), whereas the opposite phenotype results from PPARγ haploinsufficiency (33), expression of dominant-negative Pax8PPARγ (34) and by treatment with PPARδ agonists (35, 36). PPARδ is also an etiologic factor in gastric carcinogenesis (37) and colon tumorigenesis (38, 39), although interpretation of the role of PPARδ in the latter process is dependent on the derivation of the knockout model (40).

The objectives of the present study were to determine if Sca-1 deficiency affected mammary gland development and tumorigenesis in response to the potent and highly selective PPARγ agonist GW7845 (35). Here, we report that Sca-1 deficiency reduced the proliferation of mammary epithelium and markedly enhanced the chemopreventive effects of GW7845. This outcome correlated with increased PPARγ and PTEN expression, and a reduction in the ratio of pSer84PPARγ/PPARγ, as well as the levels of PPARδ and pERK1/2 in the mammary gland and tumors of Sca-1–deficient animals. Reduction of Sca-1 in mammary tumor cells by RNA interference produced a similar PPARγ phenotype, and further showed reduced proteasomal turnover and increased the transcriptional activity of PPARγ. These results suggest that Sca-1 attenuates the tumor suppressor activity of PPARγ through inhibition of posttranslational processes that modulate its activity.

Sca-1 null (KO) mice were bred from heterozygous Sca-1 mice kindly provided by Dr. William L. Stanford, University of Toronto (41). Mice were screened by PCR, and Sca-1 deficiency in the mammary gland was confirmed by fluorescence-activated cell sorting (FACS) and Western blotting.

Five week-old wild-type (WT) C57BL/6 mice and Sca-1 KO mice of the same background were treated with medroxyprogesterone (MP) and 7,12dimethylbenz(a)anthracene (DMBA) as previously described (30, 35). Briefly, mice were injected subcutaneously with 15 mg MP suspension (Depo-Provera), followed by 4 weekly doses of 1 mg DMBA/0.1 mL cottonseed oil administered by gavage (8, 21). All animals were fed diets of Purina rodent chow 5001 supplemented with either 0.005% (w/w) GW7845 (provided by Dr. Brad Henke, GlaxoSmithKline) or 0.005% GW501516 (provided by the Chemoprevention Branch, NCI) beginning 1 day after the last dose of DMBA.

The source, use, and dilutions of antibodies were the following: FITC-mouse anti-Sca-1 (553335, BD Biosciences, FACS, 1:200), rat anti-Sca-1 (553333, BD Biosciences, Western blotting, 1:1,000), mouse anti-PPARγ (sc-7273, Santa Cruz Biotechnology, Western blotting, 1:300), mouse anti-pSer84PPARγ (sc-28001-R, Santa Cruz Biotechnology, Western blotting, 1:1,000), rabbit anti-PPARδ (48-6600, Zymed Laboratories, Western blotting, 1:1,000), mouse anti-PTEN (7974, Santa Cruz Biotechnology, Western blotting, 1:1,000), pERK1/2 (4376S, Cell Signaling, Western blotting, 1:1,000), ERK1/2 (9102, Cell Signaling, Western blotting, 1:1,000), anti-pRB (9308, Cell Signaling, IHC, 1:100), anti-Ki-67 (M7249, Dako, IHC, 1:50), and mouse anti-β-actin (A-5441, Sigma-Aldrich, Western blotting, 1:5,000).

Cell line 34T was generated from a mammary adenocarcinoma induced by MP/DMBA in heterozygous Sca-1^+/EGFP mice (34). Briefly, tissue was excised and minced into 1 mm pieces under sterile conditions, and digested in Dulbecco's Modified Eagle's Media (DMEM)/F12 medium supplemented with 5% FBS, 100 U/mL collagenase I, 100 U/mL hyaluronidase, 100 U/mL penicillin, 100 U/mL streptomycin, 10 μg/mL gentamicin, and 2.5 μg/mL amphotericin B. Cell lines were maintained in DMEM medium supplemented with 5% FBS (34). 34T cells and 34T cells in which Sca-1 was reduced by transduction with a lentivirus-expressed as Sca-1 shRNA (D8 cells) were generated as previously described (21).

Western blotting was carried out as previously described (34). Briefly, either the cell pellet at 4°C or tissue was frozen in liquid nitrogen and pulverized in a mortar and pestle, were mixed with lysis buffer containing: 0.1% SDS, 0.5% NP-40, phenylmethylsulfonyl fluoride, 1 mmol/L sodium vanadate, 50 mmol/L sodium fluoride, 10 mmol/L β-glycerophosphate, 5 mmol/L sodium pyrophosphate, and protease inhibitor cocktail (Roche Diagnostics). Following incubation on ice for 30 minutes, lysates were cleared by centrifugation for 15 minutes at 13,000 × g at 4°C. Protein concentrations were determined by the Coomassie Plus Protein Assay (Pierce), and 50 μg of lysate was separated in a 4% to 12% NuPAGE Bis-Tris gel (Invitrogen). After wet transfer, membranes were blocked for one hour at room temperature in 5% nonfat dry milk in TBS (pH 7.4) containing 0.1% Tween 20. Primary antibody was incubated overnight at 4°C. Secondary antibody was incubated for one hour at room temperature, and proteins were visualized with either SuperSignal West Pico or SuperSignal West Dura (Pierce).

Immunohistochemical analysis was carried out as previously described (30, 34, 35).

Single-cell suspensions of primary mammary epithelial cell cultures were prepared from WT and Sca-1 KO mice as previously described (34). Briefly, cells were detached from the flask with 0.05% trypsin, filtered through a 45 μm cell strainer, washed twice with 1× PBS and blocked for 15 minutes on ice with 1× PBS supplemented with 3% FBS. Cells were washed twice with 1× PBS, fixed with 1% paraformaldehyde in PBS and stored in the dark at 4°C until sorting. Cells (5 × 10⁶/mL) were incubated for one hour at room temperature with a 1:400 dilution of either fluorescein isothiocyanate (FITC)–Sca-1 (clone E13-161.7, BD Biosciences) or a conjugated IgG of the same isotype as a negative control. Sorting was carried out on a FACSAria flow cytometer (Becton Dickinson) and analyzed by FACExpress Denovo software by the Flow Cytometry/Cell Sorting Shared Resource, LCCC.

Total RNA was extracted using the RNAeasy Mini Kit (Qiagen) according to the manufacturer's protocol. One microgram of RNA was reverse transcribed in a total volume of 20 μL using the Cloned AMV First-Strand cDNA Synthesis kit (Invitrogen). PCR was carried out in triplicate in an ABI 7900 instrument (Applied Biosystems) using SYBRGreen detection (Applied Biosystems) according to the manufacturer's protocol. qRT-PCR primers were designed using the Integrated DNA Technologies primer design tool (42). Efficiencies of all primer sets were validated using a standard curve of 5 serial cDNA dilutions in water in duplicate. Primers were acceptable if the deviation from the slope of the standard curve was less than 0.3, and if the melting curve showed only one product. The expression of each target gene was normalized to the expression of GAPDH, and the relative quantification method was applied using SDS2.3 software (Applied Biosystems). Primers were the following:

PTEN: forward, TGTAAAGCTGGAAAGGGACG, reverse, CCTCTGACTGGGAATTGTGAC

FABP3: forward, TCATCCATGTGCAGAAGTGG, reverse, CTTCTCATAAGTCCGAGTGCTC.

Total RNA was extracted from tumors of WT and Sca-1 KO mice using an RNeasy Mini Kit (Qiagen) following the manufacturer's protocol. cRNA synthesis was carried out using the Affymetrix (Santa Clara) protocol with minor modifications as described (30). Biotin-labeled cRNA was fragmented for 35 minutes at 94°C and hybridized overnight to an Affymetrix mouse 430A 2.0 GeneChip representing approximately 14,000 annotated mouse genes by the Macromolecular Analysis Shared Resource, Lombardi Comprehensive Cancer Center, Georgetown University. Hybridization signals were detected with an Agilent Gene Array scanner, and grid alignment and raw data generation done with Affymetrix GeneChip Operating software 1.1. Gene expression data with a signal 300 or above and with ≥2.5-fold change are listed in Supplementary Table S1. Gene ontology analysis categorizing sets of 4 of more genes at P < 0.05 was conducted with Pathway Studio 7.1 (Aridane Genomics, Inc.; Supplementary Table S2). Array data have been deposited in the GEO database under accession no. GSE31954.

The significance (P < 0.05) of data presented as the mean ± SEM was determined using the 2-tailed Student t test. Differences between groups of 2 or more variables were determined by the 2-tailed Fisher Exact test, and survival curves were analyzed by the Pearson log-rank test.

Sca-1 KO mice were bred from heterozygous mice and confirmed to have loss of Sca-1 by FACS analysis (Fig. 1A). The effect of Sca-1 deficiency on mammary gland development was initially measured in whole mounts. Sca-1 KO mice exhibited a reduction in ductal invasion of the fat pad at 7 weeks of age (Ctl; Fig. 1B), and responded less vigorously to MP treatment in comparison to WT mice (MP; Fig. 1B). The phenotype of the Sca-1–deficient gland was further assessed by determining expression of the antiproliferative marker p21^Cip1 and proliferative marker phosphoRB (pRB). Sca-1 KO mice expressed increased p21^Cip1 (Fig. 1C) and reduced pRB (Fig. 1C), suggesting that the ductal epithelium was less poised to respond to a stimulus. This was confirmed, in part, by the reduced expression of the proliferative marker Ki-67 in Sca-1 KO mice in response to MP treatment (Fig. 1D), although there was no significant difference in Ki-67 between untreated WT and Sca- KO mice. This suggests that p21^Cip1 and pRB are sensitive predictors of the proliferative potential of the tissue.

To further characterize the mammary gland of Sca-1 KO mice, tissue lysates were probed by Western blotting for PPARγ, pSer84PPARγ, PPARδ, PTEN, pERK1/2, and ERK1/2 (Fig. 2A). Sca-1–deficient tissue expressed a 3-fold higher level of PPARγ than WT tissue (Fig. 2B), as well as a reduction in the ratio of pSer84PPARγ/PPARγ (Fig. 2C). These changes were accompanied by a marked reduction in PPARδ (Fig. 2D), a 50% reduction in the percentage of pERK1/2/ERK (Fig. 2E), and an increase in PTEN expression (Fig. 2A). These results are consistent with reduced proteasome-dependent turnover of hypophosphorylated Ser84PPARγ (43).

Because Sca-1 KO mice expressed a significantly higher ratio of PPARγ/PPARδ, mammary carcinogenesis (30) was evaluated in animals administered diets supplemented with either PPARγ agonist GW7845 or PPARδ agonist GW501516 beginning 1 day after administration of the last dose of DMBA (35; Fig. 3). Tumorigenesis developed more slowly in Sca-1 KO mice than in WT mice, and KO mice treated with GW7845 exhibited increased survival and an 80% reduction in tumor formation in comparison with WT mice (Fig. 3A and B). There was no significant delay in tumor formation in WT mice treated with GW7845 (Fig. 3A and B). In contrast, GW501516 produced an increased rate of tumorigenesis in WT mice, whereas, Sca-1 KO mice were unaffected by this treatment (Fig. 3C and D). Sca-1 KO mice tended to present with a higher percentage of adenocarcinomas and a lower percentage of squamous cell carcinomas than WT mice (30), although it did not reach statistical significance (Table 1). Analysis of the adenocarcinomas from WT and KO mice (Fig. 4A) generally phenocopied the changes found in the mammary gland (Fig. 2A), as shown by the increase in PPARγ levels (Fig. 4B) and reduction in the ratio of pSer84PPARγ/PPARγ (Fig. 4C), as well as decreased expression of PPARδ (Fig. 4D). Although the percentage of pERK1/2/ERK was reduced in tumors from Sca-1 KO mice (Fig. 4E), it did not approach the changes observed in the mammary gland.

Adenocarcinomas from WT and KO mice were also assessed for changes in gene expression. Tumors from KO mice expressed ≥2.5-fold changes in 182 genes (Supplementary Table S1), of which 108 genes exhibited increased expression and 74 genes reduced expression. Reduced gene expression in tumors from KO mice was associated predominantly with proteolytic and developmental processes, whereas increased gene expression was associated with the cell cycle, mitosis, and cell division (Supplementary Table S2).

The mechanism by which Sca-1 deficiency increased PPARγ expression was further examined in mammary tumor cell line 34T using RNA interference. Cells were transduced with a lentivirus expressing either an EGFP shRNA as a control or a Sca-1 shRNA (D8 cells) to reduce Sca-1 expression (ref. 21; Fig. 5A). Reduction of Sca-1 increased PPARγ 2-fold and reduced the ratio of pSer84PPARγ/PPARγ by 70% (Fig. 5B). These changes were accompanied by a 50% reduction in the ratio of pERK1/2/ERK (Fig. 5B), and an increase in PTEN expression (Fig. 5A). Treatment of cells with GW7845 further enhanced PPARγ levels in both cell lines (Fig. 5C), which was consistent with ligand-induced stability and/or autoactivation. To determine if increased PPARγ expression was due to reduced turnover, cells were treated with the proteasome inhibitor lactacystin (Fig. 5D). Lactacystin increased PPARγ levels to a greater extent in 34T cells than in D8 cells (Fig. 5D), and correlated with the reduced ratio of pSer84PPARγ/PPARγ in these cells. Higher expression of PPARγ in D8 cells correlated with the increased expression of 2 PPARγ-dependent genes, PTEN and FABP3, which was consistent with the higher ratio of the activated hypophosphorylated species of Ser84PPARγ in these cells (ref. 44–46; Fig. 5E). Overall, these results suggest that Sca-1 deficiency in the mammary gland increases PPARγ expression and activity through a combination of effects resulting from ligand stabilization, decreased ERK1/2-dependent phosphorylation and reduced susceptibility to proteasomal degradation.

Although Sca-1 is commonly used as a marker of stem and early progenitor cells in the mammary gland and other organs, its function in normal and malignant tissues has remained largely undefined. In the present study, Sca-1 deficiency enhanced the chemopreventive action of PPARγ agonist GW7845, and negated the tumor-promoting action of PPARδ agonist GW501516. These opposing actions have been previously described in WT FVB mice (35), but the almost complete protection against tumorigenesis by GW7845 in Sca-1 KO mice is most remarkable. The enhanced chemopreventive effect of GW7845 was associated with increased PPARγ and a reduced ratio of pSer84PPARγ/PPARγ, and correlated with increased expression of the PPARγ-driven gene, PTEN (47). However, the latter finding could also be explained by alterations in posttranslational processes. We have found no evidence by gene profiling of either the KO mammary gland (unpublished results) or tumors from KO mice (Supplementary Tables S1 and S2) for enhanced PTEN mRNA expression. The Sca-1 “knockdown” phenotype in mammary tumor cell line D8 correlated with increased tumor suppressor signaling, in part through TGFβ ligand GDF10 and possibly increased PTEN levels (21). In the present study, D8 cells also expressed moderately higher PPARγ levels and a lower relative percentage of pSer84PPARγ and pERK1/2, which are consistent with the phenotype of the KO mammary gland. Because ERK-dependent phosphorylation at Ser84PPARγ has been reported to inhibit transcriptional activity (44–46) and promote proteasomal degradation (43), our data suggest that PPARγ is regulated through a posttranslational rather than by a transcriptional mechanism (47). One explanation for this effect is that ERK1/2 is activated downstream of Ras signaling by dominant-negative regulation of PPARγ (34), an alteration that may set into motion repression of PPARγ signaling and negate some of the effects of PPARγ agonists, whereas, the opposite occurs in KO mice. A second level of regulation may also result from changes in the ratio of PPARγ to PPARδ as was evident in both the KO mammary gland and tumors. Increased expression of PPARδ leads to inhibition of PPARγ target genes, and has been postulated to serve as a “gateway receptor” that can modulate PPARγ-mediated processes, such as adipogenesis (48). These opposing actions of PPARγ and PPARδ agree with their opposite functions in mammary tumorigenesis (25, 34, 35). Because there are more than 1,000 genes in the human genome containing PPAR response elements, which encompass 25 biological processes, including mitogen-activated protein kinase signaling (49), the mechanism by which PPAR signaling ultimately affects tumorigenesis is likely to occur directly, as well as through modulation of ancillary processes, such as angiogenesis, inflammation, and immune function (25, 50).

In addition to the negative regulation of the tumor suppressors PPARγ and PTEN in Sca-1 KO mice, Sca-1 was recently shown to block TGF-β signaling through a unique GDF10/Smad3 signaling pathway (21). It is interesting that PPARγ also regulates the tumor suppressor gene RASSF5 (51), a suppressor of the ERK pathway (52), which could additionally contribute to the antiproliferative effects of Sca-1 deficiency. Overall, these actions may account for the association of Sca-1 with metastasis (53–55), mammary tumor–initiating cell capacity in MMTV-Wnt1 and MMTV-ErbB2 mice (19, 20), and BCR-ABL–induced chronic myeloid leukemia (56), as well as radiation resistance (57, 58). Although, the Sca-1/Ly6A gene is missing in the human genome (59), there are Ly6 family members that may serve a similar purpose in breast cancer. In this context, the 8q24.3 amplicon identified in breast cancer includes the Ly6 locus (60), Ly6D is a marker of basal-type breast cancer (61) and micrometastases in head and neck cancer (62), and Ly6K is a marker of breast (63) and head and neck cancer (64). However, a comprehensive analysis of Ly6 gene expression and their function in breast and other cancers has not been reported.

Another intriguing finding was that Sca-1 deficient mammary gland exhibited a reduced basal proliferative state as determined by p21^Cip1, pRB and Ki-67 expression, as well as a reduced response to progestin stimulation (Fig. 1). This phenotype likely contributed to the delay in tumor formation following carcinogenesis, and to the enhanced response to GW7845. However, Sca-1 deficiency did not alter lactation and involution in multiparous animals (results not shown), suggesting that Sca-1 likely functions in a developmentally and physiologically stochastic manner. This agrees with its role in modulating the proliferation of multipotent hematopoietic cells (17), and the tumorigenicity of mammary tumor cells (21). Thus, Sca-1 may act as a rheostat of proliferation under specific conditions and cooperate with other signaling pathways. In this regard, Sca-1 KO mice were deficient in the stem and progenitor cell marker, Musashi1 (65), suggesting a link to the Notch and Wnt pathways under its control in mammary epithelial (66) and breast cancer cells (67). Thus, Sca-1 seems to be a master modulator of tumor suppressor function that ultimately impinges on tumorigenesis and the action of chemopreventive agents.

The content does not necessarily represent the official views of the National Cancer Institute or the NIH. No potential conflicts of interest were disclosed.

This investigation was conducted using the Animal Research, Flow Cytometry, Genomics and Epigenomics, and Microscopy and Imaging Shared Resources.

The work was supported by NIH grant 1NO1 CN43302-WA19 (R.I. Glazer). This study was supported by Award P30CA051008 from the National Cancer Institute, NIH, and by Grant 1G20RR025828-01 from the NIH for Rodent Barrier Facility Equipment.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.