The epithelial components of the mammary gland are thought to arise from stem cells with a capacity for self-renewal and multilineage differentiation. Furthermore, these cells and/or their immediate progeny may be targets for transformation. We have used both in vitro cultivation and a xenograft mouse model to examine the role of hedgehog signaling and Bmi-1 in regulating self-renewal of normal and malignant human mammary stem cells. We show that hedgehog signaling components PTCH1, Gli1, and Gli2 are highly expressed in normal human mammary stem/progenitor cells cultured as mammospheres and that these genes are down-regulated when cells are induced to differentiate. Activation of hedgehog signaling increases mammosphere-initiating cell number and mammosphere size, whereas inhibition of the pathway results in a reduction of these effects. These effects are mediated by the polycomb gene Bmi-1. Overexpression of Gli2 in mammosphere-initiating cells results in the production of ductal hyperplasia, and modulation of Bmi-1 expression in mammosphere-initiating cells alters mammary development in a humanized nonobese diabetic-severe combined immunodeficient mouse model. Furthermore, we show that the hedgehog signaling pathway is activated in human breast “cancer stem cells” characterized as CD44+CD24−/lowLin. These studies support a cancer stem cell model in which the hedgehog pathway and Bmi-1 play important roles in regulating self-renewal of normal and tumorigenic human mammary stem cells. (Cancer Res 2006; 66(12): 6063-71)

Stem cells are characterized by their ability to self-renew as well as generate differentiated cells within each organ. There is increasing evidence that these cells or their immediate progeny may be targets for transformation. We have hypothesized that an early event in carcinogenesis may involve dysregulation of stem cell self-renewal leading to a clonal expansion of initiated stem cells (1, 2).

A number of developmental signaling pathways, such as Wnt, Notch, and hedgehog, have been found to play a role in regulating the self-renewal of normal stem cells in the hematopoietic system, the skin, the nervous system, and the breast (1, 3, 4). In normal breast development, the epithelial components of the mammary gland are generated by a stem cell able to give rise to the lineages found in the adult gland, including myoepithelial cells, ductal epithelial cells, and alveolar epithelial cells (5). In the past, characterization of the pathways that regulate self-renewal of mammary stem cells has been limited by the lack of systems that support propagation of these cells in an undifferentiated state in vitro. When primary cultures of mammary epithelium from rodents or humans are cultured on solid substrata, they undergo limited replication and terminally differentiate (68). Moreover, the in vivo study of human mammary stem cells has been precluded by the lack of xenotransplantation mouse models. We have recently described an in vitro system for the propagation of human mammary stem and progenitor cells in suspension culture. We showed that human mammary stem cells isolated from reduction mammoplasties generate spherical colonies in suspension culture. These colonies, which we have termed nonadherent mammospheres, are highly enriched in mammary stem and progenitor cells capable of both self-renewal and multilineage differentiation (9). We have previously used this culture system to show that the Notch pathway plays a role in cell fate determination of human mammary stem cells (10).

The characterization of mouse mammary stem cells and study of mammary development has been greatly facilitated by the use of transplantation models in which mammary cells can be transplanted into the cleared mammary fatpads of syngenic mice (5, 11). Recently, Kuperwasser et al. (12) described a system in which the fatpads of nonobese diabetic-severe combined immunodeficient mouse (NOD-SCID) mice, “humanized” by implantation of immortalized human mammary fibroblasts, were able to support the growth of human mammary cells. The use of in vitro human mammosphere cultures and their transplantation into humanized NOD-SCID mouse fatpads has allowed us to further elucidate the pathways that regulate self-renewal of normal human mammary stem cells.

In addition to addressing the cell involved in tumor initiation, the “cancer stem cell hypothesis” postulates that tumors are driven by a cellular subpopulation retaining stem cell properties (2, 3, 13). Consistent with this model, we recently identified a subpopulation of cells in human breast cancers with the phenotype CD44+CD24−/lowlineage that display stem cell properties. As few as 200 cells that display this phenotype were capable of generating tumors in NOD-SCID mice, whereas the bulk of the tumor population was not tumorigenic. Furthermore, consistent with a stem cell model, these tumor-initiating cells produce tumors that recapitulate the phenotype of the initial tumor. Thus, these tumor-initiating cells display the stem cell characteristics of self-renewal and differentiation. Over the past several years, tumorigenic stem cells have been detected in myeloma, brain cancers, sarcoma, and prostate cancers (1416), lending support to the cancer stem cell hypothesis. However, it remains unclear how pathways such as Hh regulate the self-renewal of normal stem cells and the role that deregulation of these pathways plays in carcinogenesis.

In the present studies, we have used both in vitro and mouse model systems to elucidate the role of hedgehog signaling and the polycomb gene Bmi-1 in regulating the self-renewal of normal human mammary stem cells. Furthermore, we have examined the activation of these pathways in breast cancer stem cells. These studies provide support for the cancer stem cell hypothesis in which dysregulation of normal stem cell self-renewal pathways generates tumors driven by cells that maintain stem cell characteristics.

Dissociation of mammary tissue and mammosphere culture. One hundred to 200 g normal breast tissue from reduction mammoplasties were minced and dissociated, and single cells were cultured in suspension as described previously (9). Primary mammospheres were dissociated enzymatically and mechanically, and then cultured in suspension to produce mammospheres or on a collagen substratum, as described previously (9). After mammospheres were formed in suspension culture or cells reached 85% confluency on the collagen plate (∼7 days), total RNA was isolated using RNeasy Mini kit (Qiagen, Valencia, CA) and used for real-time quantitative reverse transcription-PCR (qRT-PCR) assays in a ABI PRISM 7900HT sequence detection system with 384-well block module and automation accessory (Applied Biosystems, Foster City, CA) as described in Supplementary Data.

Treatments of mammospheres with hedgehog agonists and antagonist. Single cells from epithelial organoids were plated in six-well ultra-low attachment plates (Corning, Acton, MA) as described previously (9). Biologically active, unmodified amino-terminal recombinant human Sonic hedgehog (Shh) and mouse Indian hedgehog (Ihh; R&D Systems, Minneapolis, MN), cyclopamine (TRC, Inc., North York, Ontario, Canada) were used. We tested different concentrations of Shh and determined the optimum stimulation or inhibition was obtained with 3 μg/mL Shh (17) or 300 nmol/L cyclopamine (18) in our studies. Tomatidine was used as a negative control for cyclopamine. Mammospheres were then collected at days 1, 3, 5, or 7. All of these collected mammospheres were used for RNA extraction and qRT-PCR and the mammospheres treated for 7 days were also used for in vitro self-renewal assays as described in Supplementary Data.

Immunostaining. To assess lineage composition of the colonies, single-cell suspensions were plated on collagen-coated dishes and cultured as described previously (9) for 7 days. Cells were fixed on plates in −20°C methanol for 20 minutes and stained using Peroxidase Histostain-Plus and Alkaline-Phosphatase Histostain-Plus kits (Zymed, South San Francisco, CA), according to the protocol of the manufacturer. The primary antibodies, cytokeratin 18 for epithelial cells and cytokeratin 14 (Novocastra, Norwell, MA) for myoepithelial cells, were used at the dilutions indicated by the manufacturer. AEC and 3,3′-diaminobenzidine (Zymed) were used as substrates for peroxidase and nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (Life Technologies, Gaithersburg, MD) for alkaline phosphatase.

Virus production, infection, and cell culture. The retroviral plasmid DNAs for Vector only (SIN-IP-EGFP), Gli1 (SIN-GLI1-EGFP; ref. 19), and Gli2 (SIN-GLI2-EGFP; ref. 20) were generous gifts from Dr. Graham W. Neil. Retroviruses for SIN-IP-EGFP, SIN-GLI1-EGFP, and SIN-GLI2-EGFP were produced by stable transfection of 293 cells and were used to infect the single cells isolated from primary mammosphere (see Supplementary Data for details). A highly efficient lentiviral expression system (pLentiLox 3.7)1

was used to generate Bmi-1-expressing (hBmi-1-GFP) and green fluorescent protein (GFP)–expressing (GFP alone) lentiviruses in University of Michigan Vector Core Facility.

Small interfering RNA constructions. Three human Bmi-1 (hBmi-1) siRNA oligos were purchased from Ambion, Inc. (Austin, TX; Silencer Predesigned siRNAs) and were used to confirm the knockdown of Bmi-1 expression in primary human mammary epithelial cells. All the siRNA sequences were converted to small hairpins (shRNA) and inserted into lentivirus vector LentiLox 3.7. The GFP is expressed in lentivirus-infected cells as the marker to indicate that the cells express the shRNA for hBmi-1. In our experiments, >90% of cells were infected with the control (GFP alone) or siRNA lentiviruses (hBmi-1-siRNA1-GFP, hBmi-1-siRNA2-GFP, and hBmi-1-siRNA3-GFP).

Mammosphere implantation into the cleared fatpads of NOD-SCID mice. Three-week-old female NOD-SCID mice were anesthetized by an i.p. injection (21). The no. 4 inguinal mammary glands were cleared and humanized with 2.5 × 105 nonirradiated telomerase immortalized human mammary fibroblasts (a generous gift from John Stingl and Connie Eaves, Terry Fox Laboratory, Vancouver, British Columbia, Canada) and 2.5 × 105 irradiated (4 Gy) fibroblasts as previously described (12), following a previously established protocol (22). A 60-day release estrogen pellet (0.72 mg/pellet, Innovative Research of America, Sarasota, FL) was placed s.c. on the back of the neck of the mouse by using a trocar, and 400 mammospheres were mixed with 2.5 × 105 normal human mammary fibroblasts and resuspended in 10 μL of 1:1 Matrigel: 5% serum Ham's F-12 and injected into each of the cleared fatpads. All of the implantation experiments were repeated five times using mammospheres from different patients with three mice implanted per patient sample.

Preparation of mammary fatpad sections. Approximately 8 weeks after the implantation, the fatpads were removed and fixed in Carnoy's solution for 1 hour and subsequently stained with carmine alum overnight. The tissue was then defatted through graded ethanol and cleared in 5 mL of xylene for 1 hour. The tissue was then embedded in the paraffin and sectioned for H&E staining.

Preparation of single-cell suspensions of tumor cells, xenografts, and flow cytometry. Human mammary tumors were passaged in NOD-SCID mice as previously described (21). Following tumor growth, which took 1 to 2 months, tumors were removed and single cells were obtained by collagenase digestion as described previously (21). One part of the single cells was used for flow cytometry to sort out the H2Kd-CD44+CD24−/lowlineage population and H2Kd−CD44+CD24+lineage population as described previously (21). RNA was extracted from these two populations and real-time RT-PCR was used to quantitate gene expression.

Statistical analysis. Results are presented as the mean ± SD for at least three repeated individual experiments for each group. Analysis was done using Minitab statistical software for Windows (Minitab, Inc., State College, PA). Statistical differences were determined by using one-way ANOVA for independent samples. P < 0.05 was considered statistically significant.

Components of the hedgehog pathway are highly expressed in mammary stem/progenitor cells. We have previously described the development of an in vitro culture system and a xenograft mouse model for the propagation of mammary stem/progenitor cells. This system is outlined in Fig. 1A. When primary human mammary epithelium isolated from reduction mammoplasties are cultured in nonadhering conditions, the vast majority of cells undergo anoikis. However, a small number (mammosphere-initiating cells; ∼4 per 1,000 cells) are able to form floating spherical colonies (mammospheres). Utilizing retroviral marking studies, we showed that these mammospheres could be dissociated and serially passaged at clonal density, with secondary and subsequent generation of mammospheres generated from single cells (9), maintaining a relatively constant number of mammospheres over a number of generations. The lineage-specific differentiation potential was assessed by plating these cells at clonal density on collagen substrata (9). These studies suggest that mammospheres are composed of a small number of stem cells capable of mammosphere formation and progenitors capable of multilineage differentiation, but not sphere formation. Similar findings have been reported for neural stem cells in neurospheres (1, 3, 4, 6). Attachment of cells to collagen substrata induces irreversible differentiation of these cells (9).

We compared expression of the genes in the hedgehog pathway in mammary stem/progenitor cells to that of differentiated mammary cells, using mammosphere-derived cells grown in suspension culture versus mammosphere-derived cells cultured on a collagen substratum. As shown in Fig. 1B, Ihh is the major Hh ligands expressed in mammary epithelial cells and its expression level is ∼9-fold higher in stem/progenitor cells in mammospheres than in differentiated cells cultured on a collagen substratum. During normal mammary development, hedgehog signaling is present in the stroma as well as the epithelium (23). We found that mammary fibroblasts produce Hh ligands although at lower level than in mammospheres. Figure 1C shows that hedgehog receptors PTCHs and SMO are expressed in both cell populations. However, mammosphere-derived mammary stem/progenitor cells express ∼4-fold higher levels of PTCH mRNA, and 3-fold higher levels of SMO mRNA than differentiated mammary cells. We measured the expression of transcription factors Gli1 and Gli2, which are downstream components of the hedgehog pathway, and found that mammary stem/progenitor cells have almost 25-fold higher levels of Gli1 mRNA and 6-fold higher levels of Gli2 mRNA than differentiated mammary cells (Fig. 1D). Taken together, these results indicate that hedgehog signaling pathway is activated in mammary stem/progenitor cells and is down-regulated during differentiation.

Hedgehog signaling agonists and antagonist regulate mammary stem cell self-renewal and multilineage differentiation. We have previously shown that mammosphere number upon multiple passages reflects stem cell self-renewal, whereas mammosphere size reflects progenitor cell proliferation (9, 10). We examined the effects of the hedgehog ligand Shh and hedgehog signaling inhibitor cyclopamine on primary and secondary mammosphere formation. We found that 3 μg/mL Shh increased primary mammosphere formation by 57% and the average cell number in these mammospheres by 62% (Fig. 2A). In contrast, cyclopamine decreased primary mammosphere formation by 45% and the average cell number in the primary mammospheres by 51% (Fig. 2A). The specificity of Shh stimulation was shown by reduction of this effect by cyclopamine (Fig. 2A), but not by tomatidine, an inactive cyclopamine analogue (Supplementary Fig. S1). The degree of reversal was dependent on the concentrations of cyclopamine and Shh. This suggests that at low concentration of cyclopamine, inhibition of Smoothened was incomplete.

Modulation of hedgehog signaling had an even greater effect on secondary mammosphere formation. Shh-treated primary mammospheres formed 100% more secondary mammospheres and the average cell numbers per secondary mammosphere were increased 67% (Fig. 2A). This stimulation could be reversed by addition of 300 nmol/L cyclopamine (Fig. 2A). Single cells from primary mammospheres treated with cyclopamine generated 54% less secondary mammospheres and the average cell numbers per secondary mammosphere were decreased 56% (Fig. 2A) compared with controls.

To show that Hh stimulated self-renewal of undifferentiated cells, we examined the differentiation potential of cells treated with Hh ligand. If the hedgehog pathway acts on primitive cells, then stimulation of this pathway should increase the number of primitive mammary cells capable of multilineage differentiation. To assess this, we plated mammosphere-derived cells at clonal density on collagen plates in the presence of FCS, conditions that we have previously determined promote cell differentiation (9). We used cytokeratin 14 as a marker of myoepithelial cells and cytokeratin 18 as marker of epithelial cells. Similar results were obtained with the markers ESA and CD10, respectively (data not shown). Addition of Shh resulted in a 3.5-fold increase in the number of multipotent cells, whereas cyclopamine decreased the number of these cells by 1.8-fold (Fig. 2B), demonstrating that Hh activation increased the generation of undifferentiated cells.

Because Ihh was the main hedgehog ligand expressed in the mammospheres as assayed by real-time quantitative RT-PCR, we also determined the effects of recombinant Ihh on the system. Consistent with the previously reported interchangeability of hedgehog ligands, Ihh had same effects as Shh on mammosphere formation and production of multilineage progenitors (data not shown).

In contrast to the effect on mammospheres, addition of 3 μg/mL Shh or 300 nmol/L cyclopamine to mammary epithelial cells cultured on a collagen substratum had no effect on the proliferation of these cells (Fig. 2C). This suggests that the Hh pathway primarily affects undifferentiated cell proliferation.

Mammary stem cell self-renewal is regulated by Gli transcription factors. To determine whether the effects of Hh signaling on self renewal of stem cells and proliferation of progenitor cells was mediated by the Gli transcription factors, we infected mammosphere-initiating cells with retroviral vectors containing Gli1 or Gli2 and determined the effect of constitutive expression of these transcription factors on mammosphere formation.

A highly efficient retroviral expression system (19, 24) was used to generate Gli1-, Gli2-, and EGFP-expressing human mammospheres. We found that compared with uninfected controls or the EGFP-expressing cells, overexpression of Gli1 and Gli2 in mammary epithelial cells in primary suspension culture stimulated mammosphere formation by 49% and 66%, respectively (Fig. 2D). Furthermore, overexpression of Gli1 and Gli2 increased the number of cells per mammosphere by 77% and 100%, respectively (Fig. 2D). Thus, Gli1 or Gli2 overexpression recapitulates the effects of hedgehog activation in this system.

Hedgehog effects on mammary stem cell self-renewal are mediated by the polycomb gene Bmi-1.Bmi-1 is a polycomb gene, which has recently been shown to play a role in the regulation of hematopoietic (25) and neural stem cell self-renewal (26). Interestingly, we found that Bmi-1 mRNA levels are increased ∼3.5-fold in mammospheres compared with differentiated mammary cells (Fig. 3A). We hypothesized that Bmi-1 might function as a downstream target of the hedgehog pathway. To test this hypothesis we investigated the effect of hedgehog activation on Bmi-1 expression. We found that activation of the hedgehog pathway by addition of Shh resulted in a 6-fold increase in expression of Bmi-1 in mammospheres, an effect that was blocked by the hedgehog pathway specific inhibitor cyclopamine (Fig. 3B). Furthermore, both Gli1-overexpressing and Gli2-overexpressing mammospheres displayed a 6-fold higher Bmi-1 expression compared with control cultures (Fig. 3B). Together, these results suggest that Bmi-1 expression can be up-regulated by Hh signaling in human mammary stem/progenitor cells.

To test whether Bmi-1 plays a role in regulating mammary stem cell self-renewal, we infected mammosphere-initiating cells with lentiviral vectors containing Bmi-1. We found that compared with uninfected controls or GFP-expressing cells, overexpression of Bmi-1 stimulated mammosphere formation by 80% and increased the number of cells per mammosphere by 67% (Fig. 3C).

To provide further evidence that Hh effects on stem cell self-renewal are mediated by Bmi-1, we used siRNA delivered in a lentiviral vector tagged with GFP to down regulate Bmi-1 expression in mammospheres. Two different siRNA lentiviruses significantly reduced the Bmi-1 expression at both the mRNA (over 80% reduction) and protein levels (over 70% reduction; Supplementary Fig. S2). We used these vectors to examine the effect of down-regulation of Bmi-1 on mammosphere formation in the presence or absence of Hh activation. Down-regulation of Bmi-1 expression reduced primary and secondary mammosphere formation by 80% (Fig. 3D) and 70% (Fig. 3D), respectively; and reduced the primary and secondary mammosphere size by 60% (Fig. 3D) and 70% (Fig. 3D), respectively. Furthermore, the effects of Hh activation on both primary and secondary mammosphere formation were significantly reduced by Bmi-1 down-regulation (Fig. 3D, Supplementary Fig. S3). Taken together, these studies suggest that Hh effects on mammary stem/progenitor cells are mediated by the polycomb gene Bmi-1.

Effects of Gli2 and Bmi-1 expression on mammary development in humanized NOD-SCID xenotransplants. Because Hh signaling modulates both Glis and Bmi-1, we determined the effects of Gli and Bmi-1 expression on mammary development. This was accomplished using a modification of the model described recently by Kuperwasser et al. (12) in which irradiated and nonirradiated human mammary fibroblasts are implanted into the cleared fatpads of NOD-SCID mice to support the growth of normal human mammary epithelial cells. The cleared fatpads of 3-week-old NOD-SCID mice were humanized with telomerase immortalized human mammary fibroblasts. Subsequently, they were implanted with control mammospheres, mammospheres overexpressing Gli2, or mammospheres with Bmi-1 overexpression or Bmi-1 down-regulation. After 8 weeks, the mammary glands were removed and examined by histologic analysis. Dense human mammary stroma was apparent in the humanized NOD-SCID mouse fatpad that expressed GFP (data not shown). Control mammospheres (SIN-IP-EGFP, GFP alone) produced limited ductal growth in these areas (Fig. 4A and C). In contrast, both Gli2-overxpressing mammospheres (SIN-GLI2-EGFP) and Bmi-1-overxpressing mammospheres (hBmi-1-GFP) developed substantially more outgrowths (Fig. 4B and D) than control mammospheres. Furthermore, down-regulation of Bmi-1 expression in mammospheres by siRNAs inhibited the mammary development (Fig. 4E). Microscopic examination indicated that Gli2-transfected mammospheres but not control mammospheres produced ductal hyperplasia (Fig. 4B). Gli2-transfected mammospheres produced ductal hyperplasia in ∼90% of ductal structures, and these were not detected with the implantation of control mammospheres. The human origin of these cells was confirmed by immunostaining with human specific antibodies, such as ESA and cytokeratins (data not shown). These results show that mammospheres can generate human ductal/alveolar structures when implanted into the humanized cleared fatpad of NOD-SCID mice. Furthermore, generation of these mammary outgrowths is modulated by the expression of Gli2 and Bmi-1.

The hedgehog pathway and Bmi-1 are activated in breast tumor-initiating cells. We have recently reported that human breast cancers are driven by a small subset of cancer stem cells that are characterized by the cell surface phenotype CD44+CD24−/lowlin. These cells functionally resemble normal stem cells in that they are able to self-renew generating tumors in NOD-SCID mice, as well as to differentiate into nontumorigenic cells that form the bulk of tumors (21). To determine whether the Hh pathway is activated in these cells, we used flow cytometry to isolate CD44+CD24−/lowlineage cells from a metastatic human breast carcinoma xenografted in NOD-SCID mice. CD44 is an adhesion receptor for extracellular matrix ligands, such as hyaluronic acid, whose expression has been linked to aggressive behavior and tumor metastasis (27). CD24 may regulate cell adhesion by down-regulation of CXCR4 an important receptor in stem cell homing and tumor metastasis (28). Mouse cells are eliminated in these studies by eliminating H2K-positive cells. Lin cells were depleted of cells displaying mammary differentiation antigens using a cocktail of monoclonal antibodies as previously described (21). The levels of mRNAs for Hh pathway components and Bmi-1 were measured by qPCR. As indicated in Fig. 5A, CD44+CD24−/lowlin cells displayed increased expression of Hh pathway components PTCH1, Gli1, and Gli2 by ∼1.7-fold, 30-fold, and 6-fold, respectively, as well as 5-fold increase in Bmi-1 compared with the cells isolated from the same tumor, which lacked these cancer stem cell markers (Fig. 5).

There is increasing evidence that stem cells or their immediate progeny may be the targets of transformation during carcinogenesis. Carcinomas are believed to arise through a series of mutations that occur over many years. Adult stem cells are slowly dividing long-lived cells, which by their very nature are exposed to damaging agents over long periods of time. Therefore, they may accumulate mutations that result in transformation (1). Stem cells are characterized by their ability to undergo self-renewal divisions, as well as to differentiate into cell lineages that form adult organs. The property of self-renewal in which a stem cell can produce one or two exact copies of itself is a property that is unique to stem cells. The development of in vitro culture system that maintain human mammary stem and progenitor cells in an undifferentiated state as well as NOD-SCID mouse models has permitted a more direct analysis of these pathways in normal and tumorigenic mammary stem and progenitor cells. We have shown that components of Hh signaling (PTCH1, Gli1, and Gli2) are highly expressed in normal mammary stem/progenitor cells compared with differentiated cells on a collagen substratum. Furthermore, we show that activation of this pathway with Hh ligands promotes the self-renewal of mammary stem cells, as evidenced by increased number of mammosphere-initiating cells. This effect was blocked by cyclopamine, a specific inhibitor of this pathway. Hh activation also increases the proliferation of mammary progenitor cells as reflected by increased mammosphere size.

We have used this system to investigate the downstream targets of Hh signaling responsible for mediating these effects. Addition of Hh ligands increases the expression of the transcription factors Gli1 and Gli2, which was inhibited by cyclopamine. Because Gli1 and Gli2 are positive mediators of Hh signaling, whereas Gli3 functions as a negative regulator of this pathway (29, 30), we focused on Gli1 and Gli2 in the current studies. Forced overexpression of Gli1 or Gli2 in mammosphere-initiating cells by retroviral transduction recapulated the effects of Hh ligands. These effects were unaffected by cyclopamine, an observation consistent with previous reports that Gli1 and Gli2 act downstream of smoothened, the target of cyclopamine (23).

It has recently been reported that the polycomb gene Bmi-1 plays an important role in the regulation of self-renewal of hematopoitic (25) and neuronal stem cells (26). Bmi-1 is a transcriptional repressor that may regulate stem cell self-renewal through the repression of important cell cycle regulatory genes in the INK-4A/ADP ribosylation factor (ARF) complex, p16 INK-4A, and p19 ARF (25, 26). These studies have recently been confirmed and extended using mouse knockout models of INK-4A and p19 ARF (31). We have shown that Bmi-1 is expressed at increased levels in undifferentiated compared with differentiated mammary cells. Activation of Hh signaling increases Bmi-1 expression, and Bmi-1 overexpression promotes mammary stem cell self-renewal and proliferation as indicated by an increase in mammosphere number and size in vitro and increases ductal/alveolar development in humanized NOD-SCID mammary fatpads. In contrast, down-regulation of Bmi-1 using siRNA abrogates the effects of Hh signaling on mammosphere formation in vitro and inhibited ductal/alveolar development in NOD-SCID mice. These studies suggest that the effects of Hh signaling on mammary stem cell self-renewal may be mediated by Bmi-1.

We have previously proposed that deregulation of self-renewal may be one of the key events involved in the initial stages of carcinogenesis (1). Activation of the Hh signaling pathway as well as Bmi-1 has been shown to result in the generation of mammary carcinomas in vitro or in transgenic models (32, 33). PTCH mutations have been found in a subset of human breast cancers (33). A specific mutation in PTCH1 was linked to increased risk of breast cancers with oral contraceptives (34). Hh signaling was also shown to be activated in a subset of human cancers based on immunohistochemical staining of a set of 52 invasive breast cancers (35). We have found that overexpression of the Hh target Gli2 in mammospheres produces ductal hyperplasias when these cells are implanted into the humanized fatpads of NOD-SCID mice. These findings are consistent with a stem cell model of carcinogenesis in which early events involve deregulation of Hh signaling resulting in clonal expansion of stem or progenitor cells. These cells in turn may undergo further mutation to acquire a fully malignant phenotype.

We have recently described the existence of a cancer stem cell population in human breast cancers (21). In the present study, we show that these cancer stem cells display activation of Hh signaling components as well as increased expression of Bmi-1.

Taken together, these studies lend support to the cancer stem cell hypothesis in which carcinogenesis results from deregulation of self-renewal pathways in normal stem cells generating a cancer stem cell population that drives tumorigenesis. In normal mammary development, Hh and the downstream transcription factor Bmi-1 play an important role in regulating stem cell self-renewal. These processes are tightly regulated by factors in the stem cell niche. Deregulation of these processes during carcinogenesis may result in stem cell expansion, a key event in carcinogenesis. A hypothetical model depicting the role of Hh and Bmi-1 in the regulation of mammary stem cell self-renewal and deregulation of this pathway in cancer stem cells is shown in Fig. 6. The clinical importance of this is highlighted by a recent report demonstrating a strong correlation between the expression of an 11-gene Bmi-1 stem cell signature and poor prognosis in patients with a wide variety of malignancies (36). Recently, inhibitors of Hh signaling, such as cyclopamine and related compounds, have been shown to have antitumor activity with minimal systemic toxicity in mouse tumor models (37, 38). Our studies highlight the importance of the Hh signaling pathway and Bmi-1 in the regulation of normal and malignant stem cell self-renewal and suggest that strategies aimed at inhibiting these pathways represent a rationale therapeutic approach to target cancer stem cells.

Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).

M.S. Wicha is a consultant for and has financial holdings in OncoMed Pharmaceuticals.

Grant support: NIH grant CA101860 (M.S. Wicha), Department of Defense grant BC030214 (M.S. Wicha), University of Michigan Cancer Center NIH Support grant 5 P 30 CA46592 (M.S. Wicha), and Susan G. Komen Postdoctoral Fellowship grant PDF0503599 (S. Liu).

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.

We thank Dr. Thomas Giordano for tissue procurement, Dr. Michael Clarke for technical advice, Dr. Celina G. Kleer for examining the H&E staining slides, Zhifen Wu for helping with the Western blotting, the University of Michigan Cancer Center Flow Cytometry and Vector Core Facilities, and Dr. Graham W. Neill (Centre for Cutaneous Research, Barts and the London, Queen Mary's School of Medicine and Dentistry, London, United Kingdom) for generously providing the retroviral plasmid DNAs.

1
Dontu G, Al-Hajj M, Abdallah WM, Clarke MF, Wicha MS. Stem cells in normal breast development and breast cancer.
Cell Prolif
2003
;
36
:
59
–72.
2
Wicha MS, Liu S, Dontu G. Cancer stem cells: an old idea—a paradigm shift.
Cancer Res
2006
;
66
:
1883
–90.
3
Kopper L, Hajdu M. Tumor stem cells.
Pathol Oncol Res
2004
;
10
:
69
–73.
4
Taipale J, Beachy PA. The hedgehog and Wnt signalling pathways in cancer.
Nature
2001
;
411
:
349
–54.
5
Smith GH, Chepko G. Mammary epithelial stem cells.
Microsc Res Tech
2001
;
52
:
190
–203.
6
Reynolds BA, Weiss S. Clonal and population analyses demonstrate that an EGF-responsive mammalian embryonic CNS precursor is a stem cell.
Dev Biol
1996
;
175
:
1
–13.
7
Muschler J, Lochter A, Roskelley CD, Yurchenco P, Bissell MJ. Division of labor among the α6β4 integrin, β1 integrins, and an E3 laminin receptor to signal morphogenesis and β-casein expression in mammary epithelial cells.
Mol Biol Cell
1999
;
10
:
2817
–28.
8
Romanov SR, Kozakiewicz BK, Holst CR, Stampfer MR, Haupt LM, Tisty TD. Normal human mammary epithelial cells spontaneously escape senescence and acquire genomic changes.
Nature
2001
;
409
:
633
–7.
9
Dontu G, Abdallah W, Foley J, et al. In vitro propagation and transcriptional profiling of human mammary stem/progenitor cells.
Genes Dev
2003
;
17
:
1253
–70.
10
Dontu G, Jackson KW, McNicholas E, Kawamura MJ, Abdallah WM, Wicha MS. Role of Notch signaling in cell-fate determination of human mammary stem/progenitor cells.
Breast Cancer Res
2004
;
6
:
605
–15.
11
Stingl J, Eirew P, Ricketson I, et al. Purification and unique properties of mammary epithelial stem cells.
Nature
2006
;
439
:
993
–7.
12
Kuperwasser C, Chavarria T, Wu M, et al. Reconstruction of functionally normal and malignant human breast tissues in mice.
Proc Natl Acad Sci U S A
2004
;
101
:
4966
–71.
13
Reya T, Morrison SJ, Clarke MF, Weissman IL. Stem cells, cancer, and cancer stem cells.
Nature
2001
;
414
:
105
–11.
14
Pellat-Deceunynck C, Bataille R. Normal and malignant human plasma cells: proliferation, differentiation, and expansions in relation to CD45 expression.
Blood Cells Mol Dis
2004
;
32
:
293
–301.
15
Singh SK, Clarke ID, Terasaki M, et al. Identification of a cancer stem cell in human brain tumors.
Cancer Res
2003
;
63
:
5821
–8.
16
Xin L, Lawson DA, Witte ON. The Sca-1 cell surface marker enriches for a prostate-regenerating cell subpopulation that can initiate prostate tumorigenesis.
Proc Natl Acad Sci U S A
2005
;
102
:
6942
–7.
17
Kenney AM, Rowitch DH. Sonic hedgehog promotes G(1) cyclin expression and sustained cell cycle progression in mammalian neuronal precursors.
Mol Cell Biol
2000
;
20
:
9055
–67.
18
Chen JK, Taipale J, Cooper MK, Beachy PA. Inhibition of Hedgehog signaling by direct binding of cyclopamine to Smoothened.
Genes Dev
2002
;
16
:
2743
–8.
19
Regl G, Neill GW, Eichberger T, et al. Human GLI2 and GLI1 are part of a positive feedback mechanism in basal cell carcinoma.
Oncogene
2002
;
21
:
5529
–39.
20
Ikram MS, Neill GW, Regl G, et al. GLI2 is expressed in normal human epidermis and BCC and induces GLI1 expression by binding to its promoter.
J Invest Dermatol
2004
;
122
:
1503
–9.
21
Al-Hajj M, Wicha MS, Benito-Hernandez A, Morrison SJ, Clarke MF. Prospective identification of tumorigenic breast cancer cells.
Proc Natl Acad Sci U S A
2003
;
100
:
3983
–8.
22
Ip MM, Asch BB. Introduction: an histology atlas of the rodent mammary gland and human breast during normal postnatal development and in cancer.
J Mammary Gland Biol Neoplasia
2000
;
5
:
117
–8.
23
Lewis MT, Veltmaat JM. Next stop, the twilight zone: hedgehog network regulation of mammary gland development.
J Mammary Gland Biol Neoplasia
2004
;
9
:
165
–81.
24
Eichberger T, Regl G, Ikram MS, et al. FOXE1, a new transcriptional target of GLI2 is expressed in human epidermis and basal cell carcinoma.
J Invest Dermatol
2004
;
122
:
1180
–7.
25
Park IK, Qian D, Kiel M, et al. Bmi-1 is required for maintenance of adult self-renewing haematopoietic stem cells.
Nature
2003
;
423
:
302
–5.
26
Molofsky AV, Pardal R, Iwashita T, et al. Bmi-1 dependence distinguishes neural stem cell self-renewal from progenitor proliferation.
Nature
2003
;
425
:
962
–7.
27
Avigdor A, Goichberg P, Shivtiel S, et al. CD44 and hyaluronic acid cooperate with SDF-1 in the trafficking of human CD34+ stem/progenitor cells to bone marrow.
Blood
2004
;
103
:
2981
–9.
28
Zhang M, Rosen JM. Stem cells in the etiology and treatment of cancer.
Curr Opin Genet Dev
2006
;
16
:
60
–4.
29
Stecca B, Mas C, Ruiz i Altaba A. Interference with HH-GLI signaling inhibits prostate cancer.
Trends Mol Med
2005
;
11
:
199
–203.
30
Ingham PW, McMahon AP. Hedgehog signaling in animal development: paradigms and principles.
Genes Dev
2001
;
15
:
3059
–87.
31
Molofsky AV, He S, Bydon M, Morrison SJ, Pardal R. Bmi-1 promotes neural stem cell self-renewal and neural development but not mouse growth and survival by repressing the p16Ink4a and p19Arf senescence pathways.
Genes Dev
2005
;
19
:
1432
–7.
32
Dimri GP, Martinez JL, Jacobs JJ, et al. The Bmi-1 oncogene induces telomerase activity and immortalizes human mammary epithelial cells.
Cancer Res
2002
;
62
:
4736
–45.
33
Xie J, Johnson RL, Zhang X, et al. Mutations of the PATCHED gene in several types of sporadic extracutaneous tumors.
Cancer Res
1997
;
57
:
2369
–72.
34
Strange RC, El-Genidy N, Ramachandran S, et al. Susceptibility to basal cell carcinoma: associations with PTCH polymorphisms.
Ann Hum Genet
2004
;
68
:
536
–45.
35
Kubo M, Nakamura M, Tasaki A, et al. Hedgehog signaling pathway is a new therapeutic target for patients with breast cancer.
Cancer Res
2004
;
64
:
6071
–4.
36
Glinsky GV, Berezovska O, Glinskii AB. Microarray analysis identifies a death-from-cancer signature predicting therapy failure in patients with multiple types of cancer.
J Clin Invest
2005
;
115
:
1503
–21.
37
Karhadkar SS, Bova GS, Abdallah N, et al. Hedgehog signalling in prostate regeneration, neoplasia and metastasis.
Nature
2004
;
431
:
707
–12.
38
Romer JT, Kimura H, Magdaleno S, et al. Suppression of the Shh pathway using a small molecule inhibitor eliminates medulloblastoma in Ptc1(+/−)p53(−/−) mice.
Cancer Cell
2004
;
6
:
229
–40.

Supplementary data