THE ROLE OF WNT5A IN MAMMARY GLAND DEVELOPMENT SARAH BAXLEY

Transcription

1 THE ROLE OF WNT5A IN MAMMARY GLAND DEVELOPMENT by SARAH BAXLEY ROSA SERRA, COMMITTEE CHAIR ANUPAM AGARWAL CHENBEI CHANG ANDRA FROST CHRISTOPHER KLUG JIANBO WANG A DISSERTATION Submitted to the graduate faculty of The University of Alabama at Birmingham, in partial fulfillment of the requirements for the degree of Doctor of Philosophy BIRMINGHAM, ALABAMA 2010

2 THE ROLE OF WNT5A IN MAMMARY GLAND DEVELOPMENT Sarah Baxley DEPARTMENT OF CELL BIOLOGY ABSTRACT Transforming growth factor β (TGF-β) negatively regulates mammary gland development and requires Wnt5a to exert some of these effects on mammary gland development. Wnt5a is a non-canonical signaling Wnt that is expressed in all stages of mammary gland development except lactation. Using slow release pellets containing Wnt5a, as well as Wnt5a null tissue, we previously showed that Wnt5a also acts to limit mammary development. Initial studies revealed a potential role for TGF-β and Wnt5a in regulating mammary gland progenitor cells, indicating they may act to regulate the stem and progenitor cell population. In order to study the role of Wnt5a on mammary stem and progenitor cell maintenance, we wanted to create a new mouse model. Here, we generated transgenic mice that overexpress Wnt5a in the mammary epithelium using the MMTV promoter (M5a mice). Lactation was impaired in two high Wnt5a expressing lines. Lactation defects could not be explained by differences in apoptosis, lineage differentiation, milk synthesis or secretion. Instead, Wnt5a overexpression led to a failure in oxytocin response and milk ejection, similar to mice with a mutation in Connexin43 (Cx43), we examined Cx43 phosphorylation and localization in M5a mice. In wild type mice, Cx43 switched from a phosphorylated to a more hypophosphorylated form after parturition. In contrast, Cx43 was maintained in the phosphorylated form after parturition in M5a mice. Similarly, mammary myoepithelial cells grown in culture and treated with Wnt5a exhibited altered Cx43 phosphorylation ii

3 relative to vehicle treated control cells. We propose that Wnt5a inhibits the response to oxytocin and prevents milk ejection through regulation of Cx43 phosphorylation. Additionally, we examined the effects of Wnt5a and TGF-β on mammary stem cell populations and found no direct, significant effects on stem cell number. Although Wnt5a and TGF-β may not regulate stem cell number, we hypothesize they may direct other aspects of stem or progenitor cell maintenance, such as progenitor proliferation or stem cell self-renewal. Overall, we conclude that Wnt5a and TGF-β may regulate the mammary progenitor cell population. Additionally, Wnt5a overexpression can alter mammary Cx43 phosphorylation at parturition, inhibiting the milk ejection response. Keywords: Wnt5a, Mammary gland, Cx43, Breast cancer, Lactation, Stem cells iii

4 DEDICATION I would like to dedicate this dissertation to my grandmother, Emily Vogt Postma. Her brave fight against cancer has inspired me to be part of the cure. iv

5 ACKNOWLEDGEMENTS I would first like to acknowledge my mentor, Dr. Rosa Serra, who guided me patiently throughout my research and imparted much wisdom about many aspects of scientific inquiry. I hope to follow her example and become a strong, independent investigator with my own future laboratory. Also, thank you to all the members of the Serra lab, both past, present, and future, who have made my journey all the more enjoyable and educational. I would like to thank my committee members, Dr. Anupam Agarwal, Dr. Chenbei Chang, Dr. Andra Frost, Dr. Christopher Klug, and Dr. Jianbo Wang, for helping to shape and refine my research questions, leading me to be a better scientist. I would like to thank my mother, father, sister, brothers, grandparents, uncles and aunts. My family has been the strength behind my entire career and I know they will continue to support me as I move forward. Also, thank you to all my friends who have helped me weather the challenges and celebrate the victories. v

6 TABLE OF CONTENTS Page ABSTRACT... ii DEDICATION... iv ACKOWLEDGMENTS...v LIST OF FIGURES... viii LIST OF TABLES...x INTRODUCTION...1 Breast Cancer...1 Rodent models of breast development and breast cancer...2 Mammary gland development...2 Stem and progenitor cells within the mammary gland...8 Wnt signaling...10 Transforming growth factor-β...13 Wnt5a...15 Wnt5a and TGF-β in mammary gland progenitor cell maintenance...17 Aims...20 Connexin43 and gap junctions...21 OVEREXPRESSION OF WNT5A IN MOUSE MAMMARY GLAND INHIBITS THE MILK EJECTION RESPONSE AND REGULATES CX43 PHOSPHORYLATION...23 DISCUSSION...58 Summary...58 Cx43 modifications and function...58 Calcium and ATP regulation in myoepithelial contractions...60 Potential Cx43/Wnt5a interactions: organogenesis...62 Potential Cx43/Wnt5a interactions: tumorigenesis...63 Potential Cx43/Wnt5a interactions: canonical Wnt signaling...65 Stem cells...69 Future directions...73 vi

7 GENERAL LIST OF REFERENCES...75 APPENDIX: IACUC Approval Form...86 vii

8 LIST OF FIGURES Figure Page INTRODUCTION 1 Mammary gland development from puberty to lactation Mammary gland lineages and fate mapping Wnt signaling TGF-β and Wnt5a may regulate mammary progenitor or stem cells...19 OVEREXPRESSION OF WNT5A IN MOUSE MAMMARY GLAND INHIBITS THE MILK EJECTION RESPONSE AND REGULATES CX43 PHOSPHORYLATION 1 M5a mice overexpress Wnt5a in the mammary gland M5a3 mice display only mild delays in mammary gland development Wnt5a overexpression leads to lactation deficits that correlate to differences in Wnt5a expression Luminal, alveolar, and myoepithelial cell differentiation is normal in M5a3 mammary glands M5a3 mammary glands fail to respond to oxytocin despite normal oxytocin receptor levels M5a2 mammary glands have attenuated responses to oxytocin Wnt5a overexpression alters Cx43 phosphorylation but not localization at parturition...47 viii

9 8 Wnt5a treatment can shift Cx43 phosphorylation in vitro without altering Cx43 localization Wnt5a is expressed in M5a uterus at 17.5 dpc...53 DISCUSSION 1 Wnt5a inhibits canonical Wnt signaling in mammary myoepithelial cells and mildly rescues Wnt1 phenotype in mammary glands TGF-β and Wnt5a do not directly regulate mammary stem cell number...71 ix

10 LIST OF TABLES Table Page OVEREXPRESSION OF WNT5A IN MOUSE MAMMARY GLAND INHIBITS THE MILK EJECTION RESPONSE AND REGULATES CX43 PHOSPHORYLATION 1 Delayed parturition in M5a females by strain...52 x

11 INTRODUCTION Breast Cancer Although medical advances have lead to increased early diagnosis and treatment of breast cancer, it is still a major cause of morbidity and mortality worldwide (Kamangar et al., 2006). Cancer has a strong genetic component, as women with an immediate family member who developed breast or ovarian cancer at a young age have increased risk of developing breast cancer (Claus et al., 1996). However, the underlying mutations leading to this increased risk are poorly understood. BRCA1 and BRCA2 are the most welldefined germline mutations leading to increased risk of breast and ovarian cancer, although these hereditary mutations can only account for a small portion of breast cancer diagnoses. For the most part, the mutations underlying the remaining familial breast cancer cases are undefined. In addition, the majority of diagnosed breast cancer cases are classified as sporadic, meaning the cancer was not hereditary. Sporadic breast cancers are thought to result from genetic mutations accumulated in the breast tissue over time (Martinez-Climent et al., 2006). Currently, both sporadic and hereditary cancers are treated similarly, although their underlying cause may be vastly different. Consequently, new targeted therapies need to be developed to tailor treatment to the individual breast cancer subtypes. The breast is a dynamic organ that responds to hormonal stimulus and, unsurprisingly, many breast cancer subtypes are also hormone responsive. The development of 1

12 targeted therapies such as tamoxifen and herceptin were monumental advancements in breast cancer treatment (Spigel and Burstein 2002). Both tamoxifen and herceptin therapies block endogenous signaling pathways that were usurped for breast cancer development. Understanding more about the signaling pathways that regulate mammary gland development can help us develop future therapies to treat and prevent breast cancer. Rodent models of breast development and breast cancer The mouse is a convenient mammal in which to study mammary gland development and tumorigenesis (reviewed in Serra and Crowley 2003 and 2005). Each mouse develops five paired mammary glands, and the stages of rodent mammary development are well characterized. In addition, many mouse models of breast cancer have been developed to study both genetic and environmental causes of breast cancer (Russo and Russo 1996). Murine research frequently utilizes a mammary gland specific promoter, the mouse mammary tumor virus (MMTV) promoter. This hormone-responsive promoter was cloned from a natural virus that causes murine mammary gland tumors. However, only the promoter and enhancer elements of the virus have been manipulated to direct protein expression specifically to the mammary epithelium. MMTV-driven mouse models are used for developmental studies as well as tumor research (Serra and Crowley 2003). Using these models within the rodent mammary gland has allowed researchers to tease out the effects of signaling pathways and genetic mutations on breast development and tumorigenesis. 2

13 Mammary gland development The mammary gland is a unique organ in that most of its development occurs post-natally. A rudimentary ductal tree forms during embryogenesis (Sakakura 1998). The mammary epithelial bud undergoes isometric growth until the onset of puberty when ovarian and pituitary hormones such as estrogen and growth hormone act upon the mammary epithelium to induce rapid development (Silberstein et al 1994; Hennighausen and Robinson 2005; Figure 1). During puberty the mammary gland develops terminal end buds (TEB) at the ductal tips. These TEBs drive extension of the epithelial ducts through the fat pad to form the adult mammary gland. After puberty, cycles of estrogen and progesterone lead to small amounts of epithelial branching that regress at the end of each cycle (Richert et al 2000). Pregnancy induces another stage of rapid mammary gland development. Again, hormonal stimulus from estrogen and progesterone leads to growth and expansion of the epithelial population. During early pregnancy, the epithelial cells proliferate and branch extensively (Richert et al 2000). At mid-gestation, lobulo-alveolar structures begin to appear. These grape-like structures are vital for milk production and secretion during lactation. Lobulo-alveolar structures fill the mammary fat pad at the end of pregnancy (Neville 1999). At parturition, the mammary gland must switch to active production, secretion, and ejection of milk. This is called the milk let down response. The mammary gland reaches its full potential at lactation. Milk proteins are synthesized and secreted in mass quantities. Milk protein production and secretion relies on alveolar and luminal cells, while milk ejection during feeding requires myoepithelial cell function (Richert et al 2000). Alveolar cells differentiate during pregnancy and lactation. These 3

14 Figure 1. Mammary gland development from puberty to lactation. Embryonicly, the mammary gland develops as a rudimentary ductal tree that grows isometrically until the onset of puberty. Hormones drive extension of the mammary epithelium through the surounding stroma or fat pad. Estrogen directs the formation of terminal end buds (TEB), highly proliferative structures containing the stem cells needed to extend the epithelium. At the end of branched epithlial ducts fill the mammary fat pad. During pregnancy grape-like lobulo-alveolar structures, or alveolar buds, form for the first time. Alveoli produce and secrete milk during lactation, when the mammary gland has reached its full potential. After weaning, the mammary gland undergoes waves of involution and remodeling to return it to a resting state until the next cycle of pregnancy and lactation. Figure adapted from Hennighausen and Robinson, 2005, with permission from Nature Reviews Molecular Cell Biology. 4

15 cells line the lobulo-alveolar structures (Figure 2A). Luminal cells are present in both virgin and lactating glands and form the ductal lining that conducts milk from the lobuloalveolar structures to the nipple (Sonnenberg et al 1986). The third and last mammary epithelial cell type is the myoepithelial cell. Myoepithelial cells form an exterior network around the luminal and alveolar structures (Sonnengerg et al 1986). Expression of proteins such as alpha-smooth muscle actin (asma) allows mammary myoepithelial cells to contract, propelling secreted milk from alveolar lumens into the larger ducts. Disrupted differentiation of alveolar, luminal, or myoepithelial cells can impair lactation (Manhes et al 2006; Gonzalez-Suarez et al 2007). Milk production, secretion, and ejection are hormonally regulated through the action of two main hormones: prolactin and oxytocin. Lactotrophs in the anterior pituitary synthesize prolactin, although the human breast can also produce its own prolactin (Cassoni et al 2006). Mammary luminal and alveolar cells express the prolactin receptor and prolactin signaling initiates milk production and secretion within these cells (McManaman and Neville 2003). Milk ejection is regulated through the action of oxytocin on mammary myoepithelial cells, the only mammary cell type to express the oxytocin receptor (OXTR; Adan et al 2005). Oxytocin is small nonapeptide synthesized in the periventricular nuclei of the hypothalamus, but secreted into the bloodstream though the posterior pituitary (Meister et al 1990). Suckling of the pup or infant stimulates oxytocin release from the posterior pituitary through a neural reflex arc (Crowley and Armstrong 1992). Oxytocin in turn stimulates contraction of the mammary myoepithelial cells, leading to milk ejection. 5

16 Figure 2. Mammary gland lineages and fate mapping. (A) Luminal cells expressing ECadherin line the mammary gland ducts. Alveoli form during pregnancy to produce milk proteins such as Whey acidic protein (WAP) and β-casein. These structures regress after weaning. Myoepithelial cells compose the basal compartment and surround ducts and alveoli, forming a meshwork of contractile tissue to facilitate milk ejection. Myoepithelial cells express oxytocin receptor, α-smooth muscle actin (αsma), and p63. Both adipocytes and fibroblasts fill the surrounding stroma. (B) Mammary gland stem cells are Lineage - ;CD24 + ;CD49f + by flow cytometry. Bi-potent progenitor cells express stem cell antigen 1 (Sca1) and can divide to form more committed progenitors for the luminal and myoepithelial cell lineages. Panel (A) adapted from Baxley and Serra, 2010, with permission from Bentham. 6

17 After weaning, the mammary gland undergoes a restructuring process called involution. Milk stasis within the ducts and lobulo-alveolar structures, indicative of weaning, initiates a wave of apoptosis (Watson 2005; Stein et al 2007). Alveolar structures regress and the mammary gland returns to a virgin-like state. The mammary gland remains quiescent until the next pregnancy stimulates another cycle of development. Lactation provides milk, a vital source of nutrition for young mammals. Improper lactation can lead to retarded growth rates or death of the young (Palmer et al 2006). Therefore, understanding the signaling pathways regulating lactation is of utmost importance. Additionally, early pregnancy decreases risk for breast cancer development (reviewed in Britt et al 2007). Multiparous women and women who complete a full-term pregnancy before age 20 (MacMahon et al 1970) have decreased risk of breast cancer over their lifetime. Alternatively, nulliparous women (MacMahon et al 1970) and women with a full-term birth after the age of 35 (Trichopoulos et al 1983) have increased risk of breast cancer development. Parity-specific effects on breast cancer development apply mostly to the risk of developing hormone-responsive breast cancer (Ma et al 2006). This emphasizes the protective role that pregnancy and hormonal alteration of the mammary gland may play in preventing breast cancer development (Russo and Russo 2007). Although the exact mechanisms are unclear, it has been proposed that the protective effects of pregnancy result from alterations in the mammary stem cell compartment. Perhaps post-partum mammary stem cells develop resistance to carcinogenesis relative to stem cells that have not undergone the hormone-induced growth and regression that occurs in pregnancy, lactation, and weaning (Siwko et al 2008). 7

18 Stem and progenitor cells within the mammary gland Understandably, the postnatal growth of the mammary gland during puberty and pregnancy relies on epithelial stem and progenitor cell populations within the gland itself. A stem cell is defined as a cell that can undergo self-renewal to sustain the stem cell population and asymmetric division to generate a daughter cell that is different from the stem cell itself. Progenitor cells, on the other hand, are cells that can give rise to many other cells but have lost the ability to differentiate into all of the lineages of that organ. In the breast, for example, a stem cell can give rise to all three epithelial lineages luminal, myoepithelial (or basal), and alveolar while committed progenitor cells can give rise to only a subset of these lineages. Existence of mammary stem cells has been known for some time through experiments demonstrating that mammary gland fragments retain the ability to regenerate a functional new mammary gland even after serial transplantation (DeOme et al 1959; Smith 1996; Kordon and Smith 1998). In this type of experiment, a small mammary gland segment is transplanted into a cleared fat pad and allowed to fill the fat pad. Then a portion of that transplanted tissue is again transplanted into another cleared fat pad and so on. The ability of a small portion of tissue to form a new gland, with each portion of the new gland retaining that same ability, implies the existence of self-renewing stem cells within each portion of the mammary epithelium. Recently, several independent groups have isolated populations of cells that have stem cell characteristics. Shackleton et al (2006) showed that a single transplanted Lin - ;CD24 + ;CD29 + /CD49f + cell could form a functional mammary gland (Figure 2B). Other groups have isolated breast stem and progenitor cells using different criteria such as CD24 lo ;Estrogen Receptor α-negative 8

19 (ERα-) (Sleeman et al 2007), expression of aldehyde dehydrogenase (Ginestier et al 2007), and formation of a side population by Hoescht dye exclusion (reviewed in Smalley and Clarke 2005). Mammary gland development and maintenance requires proper regulation of this stem cell population. In addition to their role in normal mammary gland development, stem cells have been isolated from human cancers, although the origin of these cells is still under debate (Al-Hajj et al 2003). Existence of cancer stem cells was first proposed when stem-like cells were isolated from acute myeloid leukemia (AML) (Lapidot et al 1994; Bonnet and Dick 1997). This idea that cancers recur from stem-like cancer progenitor cells was coined the cancer stem cell (CSC) hypothesis. Similar to stem cells from normal breast tissue, CSCs must retain self-renewal capacity and the ability to generate all the cell types of the original tumor from which it was isolated. This ability to retain stem-like characteristics and generate tumor heterogenetity requires both symmetric and asymmetric divisions (Agnieszka et al 2008). Multiple hypotheses have been postulated to explain the origin of CSCs. The most obvious is that CSCs result from deregulation of normal stem cells, including those in the mammary gland (Costa et al 2006; (Kakarala and Wicha, 2008)). In this case, CSCs retain the stem cell characteristics of their normal counterparts. Stem cells are long-lived, a fact demonstrated within the mammary gland through long-term DNA strand labeling techniques (Smith 2005). Longevity of stem cell populations suggests that mutations may be able to accumulate in these long-lived, division-competent cells (Costa et al 2006). Therefore, they may be a target for age and genetic induced tumorigenesis. Alternatively, some researchers suggest CSCs are derived from cancer cells that 9

20 have dedifferentiated to an extreme degree. In this case, the cancer-initiating cell may have been derived from any cell lineage within the mammary epithelium from fully differentiated cells to progenitor cells. Recently, luminal progenitor cells rather than stem cells themselves were identified as tumor-initiating cells in Brca1 deficient tumors (Molyneux et al 2010). Both of these hypotheses have supporting material, but neither has provided sufficient direct evidence to prove their hypothesis. Regardless, understanding the mechanisms that regulate normal mammary stem cells can provide insight and potential therapeutic pathways to target CSCs with chemotherapy. Control of the mammary gland stem cell population exhibits many similar mechanisms to those seen in the other organ systems (Alonso and Fuchs 2003). Currently, there are many known regulators of the stem cell maintenance pathway, notably wingless-related (Wnt) family members, progesterone, and Transforming Growth Factor-β. Wnt signaling Wnt proteins are members of a family of 19 secreted growth factors, which are involved in a wide range of cellular processes. Wnt signaling can be broadly divided into two categories (1) the canonical, β-catenin-dependent pathway and (2) the non-canonical β-catenin-independent pathway (Figure 3; Kikuchi et al 2006; Kuhl et al 2000; Veeman et al 2003; Widelitz 2005). The canonical Wnt pathway is involved in many cell processes both during development and in tissue maintenance such as cell proliferation, differentiation, and growth. β-catenin-dependent Wnt signaling is required to maintain the stem cell populations in various tissues, including the breast, and increased or inappropriate canonical signaling is associated with the formation of breast and colon tumors 10

21 Figure 3. Wnt signaling. Activation of canonical Wnt signaling through the frizzled (Fzd) receptor leads to β-catenin release from the inhibitory complex of GSK3β and Axin. Free β-catenin accumulates in the cytoplasm and translocates to the nucleus where it bind s to transcription factors TCF and LEF to induce transcription of Wnt responsive gene. Non-canonical Wnt signaling activates either the planar cell polarity (PCP) pathway or the Wnt/Ca2+ pathway. Alternatively, some non-canonical Wnts can inhibit canonical Wnt signaling, although the exact mechanism of this inhibition is tissue specific. 11

22 (Alonso and Fuchs 2003; Grigoryan et al 2008). Alternatively, non-canonical Wnt signals control processes such as cell movement, motility, and polarity, and some have been shown to block canonical Wnt signals (Topol et al 2003; Westfall et al 2003; Mikels and Nusse 2006). Canonical Wnts transmit their signals by binding to a subset of members within a family of seven-pass-transmembrane-spanning receptors, termed Frizzled receptors, in addition to co-receptors that belong to the LDL-receptor-related protein family, Lrp5 and Lrp6 (Grigoryan et al 2008). In the absence of Wnt, glycogen synthase kinase-3β (GSK- 3β) is active and phosphorylates β-catenin, targeting it for degradation. In the presence of Wnt, the cytoplasmic protein Dishelleved (Dsh) acts to inhibit the activity of GSK-3β, which is in a complex with the Adenomatous Polyposis Coli protein (APC) and Axin, resulting in stabilization of the β-catenin protein and its subsequent translocation to the nucleus. Nuclear β-catenin associates with the Lymphoid Enhancer Factor/T-Cell- Specific Transcription Factor (LEF/TCF) family of transcription factors and activates transcription of Wnt target genes (Figure 3B; Kikuchi et al 2006). The non-canonical signaling pathways are not as well characterized, but act through two main pathways: planar cell polarity (PCP) and Wnt/Ca 2+. Some non-canonical signaling Wnts, including Wnt5a, can antagonize canonical signaling (Mikels and Nusse, 2006; Topol et al., 2003; Westfall et al., 2003), although the mechanism by which the antagonism occurs is not clear. As discussed above, Wnt signals can be utilized to deliver paracrine messages downstream of other messengers such as progesterone. Within the mammary gland both canonical Wnts, such as Wnt4, and non-canonical Wnts, such as Wnt5a, are expressed 12

23 (Gavin and McMahon 1992). The function of these Wnt signals needs to be elucidated in order to properly understand the role of paracrine signaling in the mammary gland. Transforming Growth Factor-β TGF-β is the prototypical member of the TGF-β superfamily, a large group of evolutionarily conserved polypeptide growth factors including activins, inhibins, growth and differentiation factors (GDFs), and bone morphogenetic proteins (BMPs). Each of these proteins plays important and distinct roles in regulating different aspects of development from cell growth and differentiation to apoptosis and migration. TGF-β itself has been shown to negatively regulate mammary gland development during puberty. Implantation of TGF-β slow release pellets into the mammary gland during puberty decrease ductal extension and TEB size (Silberstein and Daniel 1987; Roarty and Serra 2007). In addition, cellular proliferation within the TEB was decreased, indicating TGF-β can negatively regulate mammary epithelium during puberty. Furthermore, it has previously been demonstrated that TGF-β regulates epithelial cell differentiation and milk secretion during pregnancy and lactation (Jhappan et al 1993; Robinson et al 1993;Gorska et al 1998). Activation of TGF-β within mammary epithelium stimulates premature involution through the induction of apoptosis (Ngyuen and Pollard 2000), while disruption of TGF-β signaling leads to precocious alveolar development (Gorska et al 1998; reviewed in Serra and Crowley 2003). Therefore, TGF-β can negatively regulate mammary epithelium during multiple stages of development. As well as its role in regulating mammary gland development, TGF-β can influence breast tumorigenesis. Activation of the TGF-β pathway inhibits primary tumor for- 13

24 mation, while inhibition of TGF-β signaling enhanced tumor incidence in murine breast cancer models (Pierce et al 1995, Bottinger et al 1997, reviewed in Serra and Crowley 2005 and Baxley and Serra 2010). Additionally, mice with disrupted TGF-β signaling in the mammary gland developed spontaneous tumors with a long latency (Gorska et al 2003). Alternatively, TGF-β enhances breast cancer metastasis and invasiveness in some models (Welch et al 1990; Gorska et al 2003; Siegel et al 2003; Muraoka-Cook et al 2004; Muraoka-Cook et al 2006). Thus, although TGF-β suppresses early tumorigenesis within the mammary gland, it enhances later stages of cancer development by increasing metastatic potential of breast cancer cells. Therapies directed toward TGF-β signaling components can therefore be very broad in spectrum, and, indeed, several potential drugs are already in development (reviewed in Baxley and Serra 2010). The effects of TGF-β within the mammary gland were of particular importance to us because of its tumor suppressive actions and its ability to negatively regulate development of the mammary gland. Previously, we generated transgenic mice that express a truncated, kinase-defective TGF-β type II receptor (DNIIR) in the mammary gland (Joseph et al., 1999). Mice expressing the DNIIR transgene demonstrated increased ductal elongation and lateral branching in the mammary gland as well as alterations in the progression of mammary tumors (Crowley et al., 2005; Crowley et al., 2006; Joseph et al., 1999). To find specific genes in the mammary gland that are regulated by TGF-β and mediate TGF-β's effects on tumor progression, we performed cdna based microarray assays comparing gene expression in wild type and DNIIR glands. Wnt5a was identified and direct regulation by TGF-β was verified (Roarty and Serra, 2007). 14

25 Wnt5a Wnt5a signals through the non-canonical Wnt pathway and is expressed during all stages of development except during lactation (Gavin and McMahon 1992; Weber-Hall et al 1994). Recently our lab showed that TGF-β can positively regulate the expression of Wnt5a within the mammary gland and that Wnt5a is required for some of the effects of TGF-β in the mammary gland (Roarty and Serra 2007). We also demonstrated that a loss of Wnt5a within the mammary epithelium leads to overgrowth of the gland. This phenotype is similar to that seen with a loss of TGF-β signaling (reviewed in Serra and Crowley 2005). TGF-β and Wnt5a exert at least part of their effects on mammary gland development via inhibition of canonical Wnt signaling. Since canonical Wnt signaling is involved in stem cell maintenance, it is likely that Wnt5a plays a role in regulating stem or progenitor cells. As TGF-β suppresses early tumorigenesis, we suspect Wnt5a, as a downstream effector of TGF-β signaling, will have similar effects. Members of the Wnt family can be subdivided by their ability to transform C57MG mammary epithelial cells to a more aggressive, cancer-like phenotype (Wong et al., 1994). For example, Wnt1 and Wnt3a have very strong transforming capacity whereas Wnt5a is non-transforming. Suppression of Wnt5a expression leads to transformation similar to that induced by Wnt1, whereas ectopic expression of Wnt5a can reverse the tumorigenic phenotype in many tumor cell lines. Thus, it has been proposed that Wnt5a can act as a tumor suppressor protein (Jonsson and Andersson, 2001; Kremenevskaja et al., 2005; Olson and Gibo, 1998; Olson et al., 1997). 15

26 In support of this hypothesis, loss of Wnt5a is associated with early relapse of invasive breast cancer, and immunohistochemical detection of Wnt5a in tumors inversely correlates with metastasis and survival of patients (Dejmek et al., 2005; Jonsson et al., 2002; Leris et al., 2005). Furthermore, a screen of Wnt expression in various established tumor cell lines showed that, in general, canonical Wnts were up-regulated in cancer cell lines relative to normal human mammary epithelial cells while the expression of noncanonical Wnts, including Wnt5a, was down-regulated (Benhaj et al., 2006). In contrast, Wnt5a is up-regulated in many other tumor types, most notably melanoma, relative to the normal tissue, and is critical for macrophage-induced invasion of breast cancer cell lines (Pukrop et al., 2006; Weeraratna et al., 2002). Together, the results suggest that Wnt5a can act as a tumor suppressor in certain cases but also promote invasion and metastasis in others. It was proposed that specific intracellular signaling pathways, as well as intercellular interactions, modulate whether Wnt5a suppresses or promotes tumor progression (Pukrop et al., 2006). Therefore, there is a necessity for a more in-depth understanding of Wnt5a s role in vivo in order to offer specific mechanisms for its role in glandular development and the development of cancer. Knowing that TGF-β could function as a tumor suppressor, we tested the hypothesis that loss of Wnt5a could lead to increases in tumorigenesis. Polyoma virus middle T antigen (PyVmT) can transform cultured cells (Guy et al 1994) and MMTV-PyVmT is a common model of breast cancer where overexpression of PyVmT leads to rapid development of tumors. In this background we added loss of Wnt5a to examine the role of Wnt5a in tumor development in the breast. We found that loss of Wnt5a led to increased volume of primary tumors and increased proliferation, similar to that seen with a loss of 16

27 TGF-β (Roarty et al 2009). Importantly, loss of Wnt5a or TGF-β led to an increase in basal cell marker expression within tumors, indicating an altered tumor phenotype as basal cell tumors rarely develop in MMTV-PyVmT mice. Histologically, the loss of Wnt5a or TGF-β led to tumor profiles more similar to those seen with increased canonical Wnt signaling. Indeed, β-catenin, the downstream effector of canonical Wnt signaling, localized more strongly to the nucleus in TGF-β and Wnt5a deficient tumors than in control tumors. Non-tumorigenic ductal tissue from Wnt5a deficient glands also demonstrated increased activation of canonical Wnt signaling. Similarly, tissue fractionation reveals stronger β-catenin localization to the nuclear fraction of Wnt5a null mammary tissue compared to wildtype control tissue. Taken together, the data suggests that Wnt5a and TGF-β can inhibit canonical Wnt signaling within the mammary epithelium and redirect mammary tumor phenotype to more basal characteristics. Wnt5a and TGF-β in mammary gland progenitor cell maintenance Canonical Wnt signaling maintains the stem cell population and activation of canonical Wnt signaling, especially via overexpression of Wnt1 (MMTV-Wnt1), leads to development of heterogeneous tumors, similar to those seen with a loss of TGF-β or Wnt5a (Li et al 2003; Teissedre et al 2009). Knowing that TGF-β and Wnt5a can inhibit canonical Wnt signaling and that loss of either leads to altered tumor phenotype, we hypothesized that TGF-β and Wnt5a may regulate the mammary stem cell population. Sca-1 is used as a marker for stem/progenitor cells in multiple organ systems (reviewed in Holmes and Stanford 2007). It was initially characterized in the hematopoietic system but has been shown to enrich for cells with stem cell properties in other systems 17

28 as well. In the mammary gland Sca-1 is generally thought to be a marker for progenitor cells (Welm et al 2002; Alvi et al 2003). Both RNA and protein levels of Sca-1 were increased in DNIIR glands suggesting that loss of responsiveness to TGF-β results in an increase in the population of progenitor cells present in the mammary gland (Roarty et al 2009; Figure 4). In support of this hypothesis, it was previously shown that when overexpressed, TGF-β limits the life-span of mammary stem cells correlating with reduced tumor formation (Boulanger and Smith, 2001; Boulanger et al., 2005). Knowing that TGF-β could regulate Wnt5a, we wanted to examine expression of Sca-1 in Wnt5a null tissue. As loss of Wnt5a is peri-natal lethal (Yamaguchi et al 1999) and no floxed Wnt5a mouse is yet available, the mammary bud is rescued from E16.5- E18.5 day Wnt5a null embryos and transplanted into the cleared fat pad of pre-pubertal SCID mice (Roarty et al 2007). This transplant allows the growth of the mammary epithelium within its normal microenvironment and wild type tissue can be transplanted as a control. Sca-1 expression was visibly increased in lysate from Wnt5a null tissue (Figure 4E), indicating that loss of Wnt5a increases mammary progenitor cell marker expression. In order to examine the stem and progenitor cell compartment more directly, we utilized the mammosphere assay. Stem and progenitor cells can survive and proliferate in single cell suspension culture, whereas normal, differentiated cells require cell-cell contact for continued survival cues. In suspension mammary stem/progenitor cells proliferate but remain tightly associated with daughter cells. Balls of cells form in this method and these semi-round structures are termed mammospheres (Dontu et al 2003). The number of stem/progenitor cells isolated from the initial tissue correlates to the number of 18

29 Figure 4. TGF- β and Wnt5a may regulate mammary progenitor or stem cells. (A) Loss of TGF-β in mammary gland leads to increased protein expression of progenitor cell marker Sca-1. (B) DNIIR tissue contains proportionally higher numbers of Sca-1 positive cells than control tissue by flow cytometry. (C) TGF-β treatment downregulates Sca-1 protein expression in MMTV-Wnt1 mammospheres. Protein was isolated from Wnt1 mammospheres treated with 5 ng/ml TGF-β versus untreated mammospheres. (D) Mammosphere formation correlates to stem cell number in the original tissue. Wnt1 tissue contains greater numbers of stem cells and forms a greater number of mammospheres than WT tissue. TGF-β treatment in mammosphere culture reduces mammosphere number in both Wnt1 and WT tissue, indicating TGF-β may negatively regulate the mammary epithelial stem or progenitor cell population. (E) Wnt5a null tissue (-/-) expresses a greater amount of Sca-1 protein than control tissue, similar to that seen in DNIIR tissue. Protein was isolated from transplanted mammary gland tissue, as Wnt5a-/- mice die at birth. (F) Wnt5a null tissue forms a greater number of mammospheres than WT control, suggesting a role for Wnt5a in regulating mammary stem and progenitor cells. Panels (A) and (E) adapted from Roarty et al 2009 with permission from Breast Cancer Research. 19

30 mammospheres formed in vitro. Thus, culturing a single cell suspension of mammary epithelial cells allows isolation and quantification of mammary stem and progenitor cells. As previously mentioned, MMTV-Wnt1 mice overexpress a canonical Wnt and develop tumors rapidly. Characterization of these mice has demonstrated increased mammary stem and progenitor cell populations in MMTV-Wnt1 glands due to overactivation of the canonical Wnt pathway (Li et al 2003). If our hypothesis is correct, TGF-β and Wnt5a should inhibit mammosphere formation with both wildtype and MMTV-Wnt1 cells due to inhibition of the canonical Wnt pathway. Indeed, MMTV-Wnt1 mammary glands form a greater number of mammospheres per initial plated cell in culture than do wildtype controls (Figure 4D), confirming the fact that MMTV-Wnt1 glands contain a higher percentage of stem and progenitor cells. Treatment of mammospheres with TGFβ leads to a decrease in mammosphere formation both with wildtype cells and MMTV- Wnt1 cells (Figure 4D). Similarly, Wnt5a-/- tissue generates increased mammospheres compared to control tissue (Figure 4F), supporting the idea that TGF-β and Wnt5a regulate mammary stem/progenitor cells. Aims In this study we aimed to examine the role of TGF-β and Wnt5a in mammary stem and progenitor cell maintenance both in vivo and in vitro. In order to achieve this aim, we generated mice that overexpress Wnt5a in the mammary epithelium utilizing the mammary gland specific promoter MMTV. By overexpressing Wnt5a we hoped to characterize defects in mammary gland development caused by inhibition of mammary stem and/or progenitor cells. However, while characterizing the mice we discovered that mi- 20

31 sexpression of Wnt5a at lactation led to failure in the oxytocin-induced milk ejection response and that this effect is likely mediated by alterations in gap junction proteins. Connexin43 and gap junctions Connexin proteins form intercellular connections called gap junctions, allowing quick transmission of small molecule messengers (<1 kda) between adjacent cells. Each connexin protein assembles with five others to form a hemi-channel, also called a connexon. Hemi-channels integrate into the cell membrane and dock with a connexon from an adjoining cell to form a complete channel. Multiple gap junctions coalesce into gap junction plaques at the plasma membrane, where up to 50 connexin-mediated channels congregate to maximize intercellular communication. Gap junction connections mediate coordinated contractions in tissues such as heart and uterus, and altered gap junction function has been linked with a variety of diseases including neuropathy, hereditary deafness, cataracts, heart disease, and cancer (reviewed in Saez et al 2003, Solan and Lampe 2005). The variety of diseases associated with altered gap junction proteins points to their necessity for normal function in many organ systems. Connexin nomenclature identifies connexin proteins based on their molecular weight. For instance, Connexin 43 (Cx43), also called gap junction protein alpha 1 (Gja1) but hereafter referred to as Cx43, migrates on an SDS-PAGE gel with the 43 kda marker. The human genome encodes at least 20 connexin proteins, from Cx23 to Cx62, each with distinct function and expression patterns throughout development. In this study we have focused on Cx43 and its role in lactation. 21

32 Mammary myoepithelial cells strongly express Cx43, which localizes to myoepithelial cell junction sites. During lactation myoepithelial cells must contract to eject milk (Moore et al 1987). Improper gap junction function leads to defects in milk ejection, presumably due to lack of coordination during myoepithelial contractions (Plante and Liard 2008; Plante et al 2010). Numerous mechanisms regulate gap junction function, one of the most important and well studied being phosphorylation (reviewed in Solan and Lampe 2005; King and Lampe 2005). Phosphorylation of connexin proteins affects protein localization, degradation, connexon assembly, turn-over rates in the plasma membrane, and gap junction gating. A Cx43 mutant with decreased phosphorylation, Gja1 Jkt, leads to decreased gap junction communication (Plante and Laird 2008; Tong et al 2009). In addition, Gja1 Jkt mutant mammary glands fail to respond to oxytocin, leading to milk ejection failure (Plante and Laird 2008; Plante et al 2010). Surprisingly, mammary glands containing this Cx43 phosphorylation mutation develop with only a mild visible phenotype (Plante and Laird 2008). The main defect is the failure in milk ejection seen at lactation, indicating the importance of proper Cx43 phosphorylation, and thereby gap junction communication, to milk ejection. 22

33 OVEREXPRESSION OF WNT5A IN MOUSE MAMMARY GLAND INHIBITS THE MILK EJECTION RESPONSE AND REGULATES CX43 PHOSPHORYLATION by SARAH E. BAXLEY and ROSA SERRA Submitted to Development,

34 SUMMARY Wnt5a is a non-canonical signaling Wnt that is expressed in all stages of mammary gland development except lactation. Using slow release pellets containing Wnt5a, as well as Wnt5a null tissue, we previously showed that Wnt5a acts to limit mammary development. Here, we generated transgenic mice that overexpress Wnt5a in the mammary epithelium using the MMTV promoter (M5a mice). Lactation was impaired in two high Wnt5a expressing lines. Lactation defects could not be explained by differences in apoptosis, lineage differentiation, milk synthesis or secretion. Instead, Wnt5a overexpression led to a failure in oxytocin response and milk ejection. Noting the similarity between the M5a phenotype and that of mice with a mutation in Cx43, we examined Cx43 phosphorylation and localization in M5a mice. In wild type mice, Cx43 switched from a phosphorylated to a more hypophosphorylated form after parturition. In contrast, Cx43 was maintained in the phosphorylated form after parturition in M5a mice. Alterations in Cx43 localization were not detected between wild type and M5a mice. Similarly, mammary myoepithelial cells grown in culture and treated with Wnt5a exhibited altered Cx43 phosphorylation relative to vehicle treated control cells. In summary, we propose that Wnt5a inhibits the response to oxytocin and prevents milk ejection through regulation of Cx43 phosphorylation. 24

35 INTRODUCTION Lactation defines the ability of mammals to feed their young. The mammary gland fully develops during pregnancy and is at its most productive during lactation (Peaker, 2002; Richert et al., 2000). Milk production and secretion are vital for nutrition and survival of neonatal mammals. Despite the obvious importance of lactation, the local mechanisms regulating the onset of lactation are not well understood. During pregnancy hormonal stimulation, mostly through estrogen, progesterone, and prolactin, drives formation of lobulo-alveolar structures, the milk producing unit of the mammary gland (Brisken and Rajaram, 2006; Richert et al., 2000). In murine mammary glands, lobuloalveolar structures develop during pregnancy. Communication between mammary epithelial cells is vital for lactation (Bry et al., 2004; Plante and Laird, 2008; Plante et al., 2010). Connexin43 (Cx43) is a gap junction protein expressed in myoepithelial cells of the mammary gland. When Cx43 function is disrupted, mammary gland development, including development of lobulo-alveolar structures, is only mildly impaired but, lactation does not occur. In this instance, the failure in lactation is due to an inability of myoepithelial cells to respond to oxytocin, the hormone responsible for milk ejection(plante and Laird, 2008). Oxytocin response stimulates contraction of the myoepithelial cells allowing ejection of milk from the gland (Nakano et al., 1997). When gap junction connections are disrupted via a decrease in Cx43, oxytocin responses are greatly attenuated (Plante and Laird, 2008), indicting the importance of myoepithelial cell communication during milk ejection and, therefore, lactation. Wnt5a is a member of the wingless-related (Wnt) family of secreted glycoproteins. Wnt proteins activate two general signaling pathways (Widelitz, 2005). The canonical 25

36 Wnt pathway involves stabilization and nuclear localization of β-catenin (Kikuchi et al., 2006). Canonical Wnt signaling regulates growth, differentiation, and stem cell maintenance. Non-canonical Wnt signaling is independent of ß-catenin and activates either the planar cell polarity (PCP) or the Wnt/Ca 2+ pathway (Kuhl et al., 2000; Veeman et al., 2003). In addition, certain non-canonical Wnts inhibit canonical Wnt signaling. Wnt5a is classified as a non-canonical signaling Wnt. Both canonical and non-canonical signaling Wnt proteins are known to regulate development and cancer progression (Grigoryan et al., 2008). It has been previously demonstrated that Wnt5a is expressed in all stages of mammary gland development except lactation (Gavin and McMahon, 1992). Our lab recently demonstrated that lack of Wnt5a in murine mammary glands accelerates development during puberty and enhances tumorigenesis (Roarty et al., 2009; Roarty and Serra, 2007). However, the reason for the decrease in Wnt5a expression during lactation remains unclear. Although previous research has defined possible roles for Wnt5a during puberty and tumor progression, investigation into the role of Wnt5a during pregnancy and lactation has been hampered by a lack of available mouse models. Wnt5a null mice die at birth. Studies examining Wnt5a have therefore utilized mammary transplant models, which, by their very nature, prevent examination of normal lactation and specifically milk ejection. In order to characterize the effects of Wnt5a on pregnancy and lactation, we generated mice that overexpress Wnt5a in the mammary gland using the mouse mammary tumor virus (MMTV) promoter. In this study, we demonstrate that Wnt5a can regulate the milk-let down response to oxytocin at parturition, potentially through regulation of Cx43 phosphorylation in myoepithelial cells. 26

37 MATERIALS AND METHODS Cloning and transgenic mouse generation Human Wnt5a cdna was obtained from Open Biosystems (Huntsville, AL, USA) and was cloned into the MKbpAII vector containing the MMTV-LTR promoter/enhancer and KCR intron to increase transgene expression (gift from Jeff Rosen, Baylor College of Medicine). The human Wnt5a cdna was amplified using primers with engineered restriction enzyme sites on either side. Wnt5a cdna was then inserted into the MKbpAII plasmid using BamHI and XhoI sites (Figure 1A). The final transgenic expression vector was termed M5a. Linearized M5a plasmid DNA was injected into hybrid C57Bl6/SJL embryos by the UAB Transgenic Mouse Facility. Genotyping primers were designed to amplify a segment from the end of the KCR intron through a portion of the hwnt5a transgene (M5aF 5 TCCTGGTCATCATCCTGCCTTTCT; M5aR 5 GCGACCAC- CAAGAATTGGCTTCAA). Mouse Husbandry All mice utilized in this study were maintained under the guidelines of the Institutional Animal Care and Use Committee of the University of Alabama at Birmingham. Seven MMTV-Wnt5a founder mice were generated and each was used to generate a separate line, termed M5a1-M5a7. Each founder mouse was backcrossed into C57Bl6 obtained from Jackson Laboratories (Bar Harbor, MI, USA). Gestational age was determined with vaginal plugs. The morning when a plug was seen was counted as 0.5 days post coitus (dpc). For pregnancy studies female mice were sacrificed at 17.5 dpc. The morning when pups were first seen in the cage was counted as 1 day post-partum (dpp). 27

38 RNA isolation and semi-quantitative RT-PCR RNA was extracted from whole mammary gland tissue (after removal of the lymph node) homogenized in Trizol (Invitrogen, Carlsbad, CA, USA) and resuspended in water. cdna was synthesized from 2 µg total RNA using a reverse transcription kit (Qiagen, Valencia, CA, USA). Semi-quantitative PCR was set up using approximately 50 ng per sample. Each sample was analyzed at the linear range of amplification, as determined by analysis at three different cycles. cdna levels were normalized to β2- microglobulin (b2mf 5 TTCTGGTGCTTGTCTCACTGA, b2mr 5 CAG- TATGTTCGGCTTCCCATTC). Transgene levels were assessed using primers specific for human Wnt5a (hwnt5af 5 CCGCGAGCGGGAGCGCAT, hwnt5ar 5 GCCA- CATCAGCCAGGTTGTACACC). Protein isolation and western blotting The whole number 3 mammary gland from 8 week old mice was homogenized into 500 µl of standard RIPA buffer containing phosphatase (Sigma) and protease inhibitors (Roche, Indianapolis, IN, USA). For pregnant and lactating glands, protein lysates were generated from purified mammary epithelial cells. Mammary epithelial cell isolation was performed as previously described (Roarty and Serra, 2007). Briefly, mammary glands were digested in collagenase/pronase for 2 hours at 37 C with rotation. Cells were spun at 850xg for 5 min. The pellet containing mammary epithelial cells was washed five times with HBSS+2%FBS (Invitrogen) and spun for 30 seconds at 850xg. The resulting cell pellet contains purified mammary epithelial cells. This pellet was resuspended in 300µl RIPA containing phosphatase and protease inhibitors and homogenized 28

39 to ensure complete disruption of cell membranes. Protein concentration was determined using a Bradford assay (Biorad, Hercules, CA, USA). Proteins were analyzed by Western blot analysis. Protein samples were run on 10% SDS Polyacrylamide gel (PAGE) with 30 ug of protein per lane. To assess Cx43 phosphorylation, protein was run on a 12% gel until the 37 kda marker reached the edge of the gel. Protein was transferred onto PVDF membrane (Biorad). Membranes were blocked and stained in 6% non-fat milk in TBST using anti-beta-tubulin (1:1000; SCBT, Santa Cruz, CA, USA), anti-wnt5a (1:1000; Cell Signaling Technology), anti-alpha smooth muscle actin (1:2000; Abcam), anti-oxytocin receptor (1:1000; MBL International, Woborn, MA, USA), anti-beta-casein (1:1000; SCBT), anti-p63 (1:1000; ThermoScientific), anti-pan cytokeratin (1:2000; Novus Biologicals, Littleton, CO, USA), anti- Cx43 (1:1000; Sigma) and secondary HRP-conjugated anti-rabbit or anti-mouse (1:1000; SCBT). Bands were visualized using either standard ECL reagents or SuperSignal ECL for weaker signals (ThermoScientific, Rockford, IL, USA). Tissue processing The number 4 mammary glands were removed and fixed in 4% PFA for 1 hour at room temperature. Tissue was dehydrated at 4 C, cleared in xylene, and paraffin embedded for sectioning. For whole mount staining, glands were fixed in Carnoy s fix for 1 hour and stained overnight with carmine. After dehydration and clearing in xylene, glands were mounted in high-viscosity toluene based medium (ThermoScientific) 29

40 Immunohistochemistry/Immnofluorescence Hematoxylin and eosin staining was performed according to manufacturer s instructions (Sigma). Roche TUNEL kit was used to assess apoptotic cells. Sections were treated according to manufacturer s guidelines. Slides were developed in DAB for 8 minutes and were counterstained with methyl green. Toluene based media was used for mounting. A section of involution day 9 mammary gland was used as a positive control. A -TdT enzyme control was used as a negative control to rule out background staining. For immunofluorescence 5 µm paraffin sections of mammary gland tissue were deparaffinized and rehydrated. Retrivagen A (Roche) or 0.1% pepsin (Sigma) were used for antigen retrival. For immunocytochemistry, cultured cells were fixed with ice-cold methanol for 30 min on ice. Immunofluorescence or immunocytochemistry were performed using anti-alpha smooth muscle actin-cy3(1:1000; Sigma), anti-connexin43 (1:200; Sigma), anti-keratin14 (1:150; Covanace), anti-wap (1:50, SCBT), and anti-ecadherin (1:100, Cell Signaling Technology) in 10% normal goat serum (NGS). Secondary goat-anti-rabbit Alexa488 or goat-anti-rabbit Cy3 (Invitrogen) were used for 15 min at 1:200. Nuclei were counterstained for 5 minutes in 0.6 mm DAPI. For double Cx43 and K14 staining, sections were blocked for 30 min in 10% NGS and incubated in Cx43 antibody solution at 1:150 in 10% NGS for 30 min. Slides were rinsed in TBST 3x2min and incubated in goat-anti-rabbit-alexa568 (Invitrogen) at 1:200 for 15 min. Slides were rinsed 3x2min and incubated in normal rabbit serum for 30 min to remove extraneous goat-anti-rabbit. Sections were rinsed in TBST briefly and incubated in primary K14 at 1:200 for 30 min. Sections were again rinsed and incubated in goat-anti-rabbit-alexa488 for 15 min followed by DAPI counterstaining. Aqueous polymount was used to mount 30

41 stained sections. Pictures were taken using a Olympus microscope and camera. Cx43/K14 double stains were visualized using a scanning laser confocal microscope. Oxytocin response This assay was performed similarly to Plante and Laird (Plante and Laird, 2008). The number 3 mammary glands were exposed one at a time. One side was treated with oxytocin and then the other side was treated with PBS as a control. An initial picture was taken prior to addition of PBS or 1 mg/ml Oxytocin (Sigma,St Louis, MO, USA) in PBS. After 1 minute, the gland was photographed again. Pictures were taken using a Zeiss dissecting scope and Olympus camera. Myoepithelial cell isolation and cell culture Mammary glands from 1day post partum female mice were removed and digested according to the same procedure described above for mammary epithelial cell isolation. After isolation of purified mammary epithelial cells, cells were plated on collagen-coated 10 cm dishes at a density of 1x10 5 cells/cm 2 in differentiation media containing 5% FBS, 5 µg/ml prolactin, 100 ng/ml oxytocin, 0.5 µg/ml hydrocortisone, 100 ng/ml cholera toxin, 10 µg/ml insulin, 100 µg/ml penicillin, and 5 µg/ml streptomycin in DMEM/F12. Cells were grown for 3 days until 80% confluent. Cells were then trypsinized and resuspended in 1 ml HBSS+2% FBS. 10 µl anti- ECadherin antibody was added to each and mixed gently. Cells were incubated on ice for 30 min and briefly rinsed and centrifuged to remove excess primary antibody. Cells were resuspended in 0.5 ml of HBSS+2%FBS with 2.5 µl of biotinylated goat anti-rabbit anti- 31

42 body (Vector) and incubated on ice for 15 min. 50 µl of biotin selection cocktail (Stem Cell Technologies) was added and incubated on ice for 15 min, followed by addition of 25 µl of magnetic nanoparticles (Stem Cell Technologies) with another 15 min incubation. Magnetic selection was used according to standard protocol for negative selection (Stem Cell Technologies). Remaining ECadherin negative cells were termed isolated myoepithelial cells. These cells were plated at a density of 5x10 4 /cm 2 on collagen coated 4-well chamber slides or 6 well plates. Myoepithelial cell media contains 5% FBS, 100 ng/ml oxytocin, 0.5 µg/ml hydrocortisone, 100 ng/ml cholera toxin, 10 µg/ml insulin, 100 µg/ml penicillin, and 5 µg/ml streptomycin in DMEM/F12. All in vitro experiments were performed using these myoepithelial cells with 1µg/ml oxytocin. After selection against ECadherin, cells containing large lipid vesicles characteristic of alveolar cells were no longer detected by phase contrast (data not shown). In addition, the cells stained with keratin14 (K14), confirming myoepithelial cell identity. Conditioned media was used to verify in vivo results withwnt5a treatment to cells in culture. Standard AATC protocol was used to generate conditioned media with both L-Wnt5a cells (AATC # CRL-2814 ; Wnt5a conditioned media) and L-Parental cells (AATC# CRL-2648; control conditioned media). Cells were treated with control- or Wnt5a-conditioned media for 2 hours prior to addition of oxytocin at a concentration of 1 µg/ml in the conditioned media. A 24 hour incubation followed oxytocin treatment before protein isolation or immunostaining procedures were begun. Both protein isolation and immunocytochemistry were performed as described above. 32

43 RESULTS Generation of MMTV-Wnt5a transgenic mice. The MMTV promoter/enhancer element is frequently used to drive protein expression in murine mammary epithelial tissue. We utilized this promoter to overexpress human Wnt5a in the mammary gland (Figure 1A). Human and murine Wnt5a have 99% amino acid homology, and recombinant human Wnt5a has been used to stimulate murine cells (Clark et al., 1993; Roarty et al., 2009; Roarty and Serra, 2007). Out of seven initial founders, three lines were chosen for characterization: M5a2, M5a3, and M5a4. Most of the results shown here are from the M5a3 line because it had the highest level of expression. Expression of the hwnt5a transgene was detectable at 8 weeks of age at both the RNA and protein level in M5a3 transgenic females (Figure 1B,C). Due to the hormone responsive nature of the MMTV promoter, the levels of Wnt5a increased dramatically during pregnancy and lactation (Figure 1C). Strong expression of Wnt5a was seen by western blot during late pregnancy (17.5 days post-coitus; dpc), and at 1 day post-partum (1dpp). Mammary gland development is not affected by overexpression of Wnt5a. Previous data demonstrated that loss of Wnt5a results in increased proliferation in mammary epithelium and increased branching in the mammary gland during puberty (Roarty and Serra, 2007). Additionally, implantation of slow release pellets containing Wnt5a into mammary glands of developing mice resulted in reduced proliferation and extension of terminal end buds as well as inhibition of growth of lateral branching. 33

44 Figure 1. M5a mice overexpress Wnt5a in the mammary gland. (A) Human Wnt5a cdna was cloned into the MKbpAII vector containing the MMTV-LTR promoter, KCR intron, and polya tail (pa). A 350 bp segment is amplified for genotyping (primer sequences listed above diagram). A representative gel shows specificity of the primers for the transgene in M5a mice versus their wildtype (WT) littermates. (B) Semi-quantitative RT-PCR confirms expression of the human Wnt5a transgene (hwnt5a) in the M5a3 line at 8 weeks of age. Cycle numbers used for expression analysis are listed below the gel picture. Beta2microglobulin (b2m) was used to normalize. (C) Transgenic Wnt5a is detectable by western blot at 8 weeks (8 wk), 17.5 days into pregnancy (17.5 dpc), and 1 day post-partum (1d pp). Beta-tubulin is used as a loading control. 34

45 However, M5a3 mammary glands developed normally during puberty with only mild effects on end bud formation at 5 weeks of age and lateral branching at 8 weeks of age (Figure 2). M5a3 mammary glands at 20 weeks of age were indistinguishable from ageand litter-matched wildtype controls (Figure 2). However, the levels of Wnt5a as measured by western blot were fairly low in non-pregnant females (Figure 1C) and it is possible that the level of expression in these mice may not be high enough to generate a strong phenotype in the transgenic animals at this stage. Wnt5a inhibits normal lactation without affecting alveolar development. Since levels of endogenous Wnt5a are known to drop at lactation (Gavin and McMahon, 1992), we examined the role of Wnt5a during pregnancy and lactation to determine if continued expression of Wnt5a would affect lactation. We measured pup survival rates for M5a2, M5a3, M5a4, and wild type dams. Although all pups born to wildtype mothers survived, none of the pups from M5a3 mothers survived past 3 days post partum. (Figure 3A). In a second line, M5a2, pup survival was only 62%. Interestingly, the pup survival rate correlated to the level of Wnt5a transgene expression (Figure 3B). Wnt5a protein, as measured by Western blot, was expressed strongly in M5a3 glands whereas M5a2 expressed a moderate level. Wnt5a protein was not detectable by Western blot in M5a4 glands, the only line with 100% pup survival. Pup survival depended on the lactation ability of the mother, as pups fostered from transgenic dams to wildtype dams survived without complication. Litter size was similar for wildtype, M5a2, M5a3, and M5a4 mothers, indicating differences in pup survival rates were not due to differences in pup number (data not shown). 35

46 Figure 2. M5a3 mice display only mild delays in mammary gland development. Whole mount staining of M5a3 glands was performed at several stages of development: 5 weeks, 8 weeks, 20 weeks, and 17.5 dpc. At 5 weeks of age, M5a3 epithelium extends normally, although TEB size may be slightly reduced. 8 week M5a3 mammary glands are virtually indistinguishable from WT glands. By 20 weeks, M5a3 glands may have slightly reduced branching. Lobulo-alveolar structures are apparent at 17.5 dpc, although the may be sparser than wildtype levels, perhaps due to mildly decreased branching. Overall, development of M5a3 glands is only mildly impaired by Wnt5a overexpression. 36

47 37

48 Figure 3. Wnt5a overexpression leads to lactation deficits that correlate to differences in Wnt5a expression. (A) Pup survival after live birth decreases with M5a3 and M5a2 dams. Pups born to wildtype or M5a4 dams have 100% survival, while pups born to M5a3 dams have 0% survival. Pups born to M5a2 dams have 62% chance of survival. Both M5a3 and M5a2 lines are statistically different from WT (p<0.001 by χ 2 analysis). (B) Wnt5a expression levels in each line correlate to pup survival. M5a3 has the highest level of expression by western blot, with M5a2 having a moderate level of expression and M5a4 having no detectable expression. 38

49 Gross mammary gland morphology, as measured by whole mount Carmine staining at late pregnancy, appeared normal in Wnt5a overexpressing lines (Figure 2). Lobulo-alveolar structures formed properly, as evidenced by both whole mount and H&E staining (Figure 4A, Figure 2), indicating that lactation failure in M5a dams did not result from an overall retardation of mammary development. Failure to lactate is not due to alterations in apoptosis, cell differentiation, or milk synthesis and secretion. Differentiation of all cell types within the mammary gland, luminal, alveolar, and basal/myoepithelial, are necessary for proper lactation. Initially, to examine morphology of luminal cells, we stained pregnant and lactating mammary glands with ECadherin. Luminal cell morphology was normal in the transgenic mammary glands compared to controls (Figure 4C). ECadherin staining intensity was unaffected by Wnt5a overexpression, indicating luminal cells differentiated normally. Alveolar cell development is more specific to lactation. This cell type begins to develop at the onset of pregnancy. During lactation alveolar cells produce milk proteins that are then secreted into the lumen. Beta-casein and whey acidic protein (WAP) are two major milk proteins used to examine milk production. Beta-casein expression was normal in M5a3 glands, as assayed by western blot (Figure 4F). WAP localized to the alveolar and luminal cells in both wildtype and M5a glands and was clearly secreted into alveoli and ducts so we conclude that failure to make or secrete milk is not the primary defect leading to a failure to lactate in the M5a3 mice (Figure 4D). Interestingly, H&E staining showed that ducts in M5a3 and M5a2 mammary glands contained copious 39

50 Figure 4. Luminal, alveolar, and myoepithelial cell differentiation is normal in M5a3 mammary glands. (A) M5a3 histology by H&E staining resembles wildtype (WT) controls both during late pregnancy (17.5 dpc) and in early lactation (1dpp). Ducts in M5a females are dilated with milk at 1dpp, potentially indicating milk stasis (bottom right panel). (B) Mammary glands in M5a3 females (bottom panel) contain dilated ducts that are visible grossly. Wild type glands rarely demonstrate such extreme dilation (top panel). (C) ECadherin staining indicates no difference in luminal cell morphology, differentiation, or localization in transgenic animals versus wildtype controls. (D) Milk protein staining (WAP) in M5a3 mammary glands mirrors wildtype staining during pregnancy. Milk appears in both wildtype and M5a3 alveolar structures at 1 dpp. (E) Smooth muscle actin stains mammary myoepithelial cells. Staining is indistinguishable between wildtype and M5a3 mice at 17.5 dpc and 1dpp. The similarity in staining patterns signifies normal differentiation and histology of myoepithelial cells in transgenic animals. (F) Protein isolated from mammary epithelial cells reveals normal expression of myoepithelial cells markers smooth muscle actin and p63. Additionally, M5a3 dams produce normal amounts of milk (beta-casein). (G) TUNEL staining to identify apoptotic cells reveals no difference in cell death between WT and M5a3 epithelium during pregnancy at 17.5 dpc. However, increased TUNEL staining occurs at 1dpp in M5a3 glands, possibly secondary to milk stasis. 40

51 41

52 amounts of milk compared to wildtype controls and were therefore dilated (Figure 4B). Furthermore, ducts in M5a3 and M5a2 glands contain more milk than do wild-type glands (Figure 4A,B). Dilation of ducts filled with milk suggested a defect in milk ejection. Since mild ejection depends on the function of myoepithelial cells, we examined myoepithelial cell histology using alpha-smooth muscle actin (asma) immunostaining. Myoepithelial cell localization and asma expression levels appeared normal by immunofluorescent staining (Figure 4E). Additionally, western blot demonstrated comparable protein expression levels of the myoepithelial cell markers p63 and asma in M5a and wildtype mammary tissue (Figure 4F). Therefore, we conclude that neither luminal, alveolar, nor myoepithelial cell differentiation is affected by Wnt5a overexpression. Lactation defects could be a result of premature involution so we examined apoptosis in control and M5a3 glands. A relatively easy method to examine apoptosis in tissue sections is TUNEL staining (Gavrieli et al., 1992). TUNEL staining during late pregnancy indicated no differences in apoptosis between transgenic glands and wild type controls (Figure 4G). Very few apoptotic cells were observed at this stage of development. At 1dpp, TUNEL staining was slightly increased in M5a glands relative to controls, however, the small level of apoptosis seen is unlikely to result in a total failure of lactation. In addition, it is known that milk stasis can induce apoptosis in mammary epithelium within hours. H&E staining, described above, indicated accumulation of milk in the ducts, therefore, the slight increase in TUNEL positive cells was likely a consequence of the accumulation of milk. We conclude that the increase in apoptosis at 1dpp is most 42

53 likely not the primary cause of failure to lactate but rather a consequence of the failure in milk ejection. Milk ejection is impaired in M5a mammary glands. Since milk production, and therefore luminal and alveolar cell function, is clearly unaffected by Wnt5a overexpression, we wanted to examine myoepithelial cell function. Myoepithelial cells respond to the hormone oxytocin by contracting and forcing milk from the alveolar space into the ducts. Grossly, the myoepithelial cell responses can be visualized as increase in the amount of milk in larger ducts after a 1 minute treatment with oxytocin (Figure 5A). Although wild type glands respond readily to oxytocin administration, M5a3 mammary glands fail to respond, and ducts do not accumulate more milk (Figure 5A). Similarly, M5a2 mammary glands lack a robust response to oxytocin (Figure 6). In the mammary gland, only myoepithelial cells express the oxytocin receptor (OXTR), allowing them to respond to oxytocin. The failure of M5a3 myoepithelial cell function could potentially be due to changes in OXTR expression; however, protein levels of OXTR, as measured by Western blot, were unaltered by Wnt5a overexpression (Figure 5B), indicating that Wnt5a blocks oxytocin responses downstream of the receptor. Wnt5a alters Cx43 phosphorylation in mammary myoepithelial cells. Mutations in a protein called connexin43 (Cx43; also termed gap junction protein alpha, Gja1) lead to a phenotype strikingly similar to that in the M5a mice. In the mammary gland,cx43 is specifically expressed in myoepithelial cells and expression of a Cx43 phosphorylation mutant, Gja1 Jkt, impairs milk ejection and parturition in mice 43

54 Figure 5. M5a3 mammary glands fail to respond to oxytocin despite normal oxytocin receptor levels. (A) Milk protein accumulates in the ducts of wildtype mammary glands in response to oxytocin treatment (+OXT; top left panels). However, M5a3 glands lack any visible response to oxytocin addition (top right panels). PBS negative controls are shown below. (B) Western blots demonstrate no differences in oxytocin receptor protein levels between wildtype and M5a3 mice. Two wildtype and M5a3 samples are shown, and pan-cytokeratin is used as a loading control. 44

55 Figure 6. M5a2 mammary glands have attenuated responses to oxytocin. Wildtype (WT) mammary glands (left, top panels) treated with oxytocin have clear accumulation of milk in the ducts following oxytocin treatment, while M5a2 glands (right, top panels) have a greatly attenuated response to oxytocin. PBS treated controls are included in the bottom panels. 45

56 without apparent affects on development through puberty and pregnancy (Plante et al 2010; Tong eta al 2009). Gja1 Jkt mutant mammary glands demonstrated an accumulation of milk in mammary ducts as well as failure to respond to oxytocin despite normal levels of OXTR, phenotypes we also observed in M5a females. Thus, we examined Cx43 expression and phosphorylation in myoepithelial cells isolated directly from M5a and wild type glands. Differences in Cx43 phosphorylation can be visualized by Western blot as slight differences in the banding pattern of the protein. Three bands were previously described. The lowest band has been termed P0 because it was initially thought to be the unphosphorylated form of Cx43. P1 is the middle band and P2 is the upper band, representing differentially phosphorylated forms of the protein. We examined Cx43 phosphorylation both during late pregnancy and early lactation. In late pregnancy, most of the Cx43 was in the P1 form and there was no difference in wild type and M5a3 mice (Figure 7A ). At lactation, there was a shift in wild type mice from the P1 form to a predominance of the P0 form of Cx43. In contrast, M5a mice at 1dpp demonstrated a predominance of the P1 form of Cx43 (Figure 7A). The ratio of P1/P0 by densitometry is significantly increased in M5a3 glands relative to wildtype at 1dpp (p=0.03). The results suggest that Wnt5a regulates the phosphorylation status of Cx43, which is necessary for the response to oxytocin and milk ejection. To determine if the alterations in phosphorylation resulted in changes in localization of Cx43 and to rule out any effects on Cx43 trafficking, we performed a double immunostaining using both Cx43 and K14 antibodies, allowing us to look at Cx43 in myoepithelial cells in sections from wild type and M5a glands. The punctate pattern of expression on the surface of myoepithelial cells suggested that Cx43 localized to presumptive 46

57 Figure 7. Wnt5a overexpression alters Cx43 phosphorylation but not localization at parturition. (A) Western blot for Cx43 phosphorylation bands shows three distinct bands, P0, the presumptive hypo-phosphorylated protein, and P1 and P2 the more phosphorylated Cx43 forms. At 17.5 dpc both WT and M5a3 glands have an abundance of P1. At 1dpp, wild type glands switch to predominately the P0 form, while M5a3 glands maintain the P1 form, similar to the pregnancy pattern. M5a2 mammary glands also have stronger P1 staining. (B) Cx43 and keratin14 (K14) staining in WT and M5a3 glands shows no difference in either K14 or Cx43 localization. (C) Strong, punctate Cx43 staining marking gap junctions is visible in both WT and M5a3 glands. 47

58 gap junction plaques similarly in transgenic and wild type glands (Figure 7B,C), indicating that Wnt5a overexpression did not affect Cx43 localization in the myoepithelial cells. Next, we wanted to determine if exogenous treatment with Wnt5a could directly alter Cx43 phosphorylation in myoepithelial cells in culture. Therefore, we isolated lactation stage mammary myoepithelial cells and treated them with Wnt5a conditioned media or control media. After 48 hours of treatment, protein was collected and changes in Cx43 phosphorylation were determined by western blot (Figure 8A). Untreated cells demonstrated a predominance of the P0 form of Cx43. Wnt5a treatment shifted the Cx43 band to the more phosphorylated P1 form suggesting Wnt5a acts directly on myoepithelial cells to regulate Cx43 phosphorylation. Localization of Cx43 in vitro was unaffected in wildtype versus M5a3 myoepithelial cells (Figure 8B). Similarly, treatment of myoepithelial cells with Wnt5a conditioned media did not affect Cx43 localization (Figure 8B). K14 staining in both mutant and wildtype cells demonstrates normal cytosekeletal organization, suggesting that Wnt5a does not alter the cytoskeleton. These results confirm that Wnt5a regulates Cx43 phosphorylation in mammary myoepithelial cells without affecting its localization. DISCUSSION Wnt5a is expressed at all stages of mammary gland development except for lactation (Gavin and McMahon, 1992). The reason for the dramatic reduction in Wnt5a expression at lactation was not clear until now. For this study, we generated mice that over express Wnt5a (M5a) in the mammary gland. This mouse model allowed us the unique opportunity to examine the effects of Wnt5a on lactation. We show here that Wnt5a 48

59 Figure 8. Wnt5a treatment can shift Cx43 phosphorylation in vitro without altering Cx43 localization. (A) Wildtype (WT) mammary myoepithelial cells were cultured in vitro and treated with control conditioned media or Wnt5a conditioned media. Western blot demonstrated increased P1 intensity with Wnt5a treatment, similar to the banding patterns seen in vivo. (B) Cultured myoepithelial cells from WT and M5a3 mammary glands show no alteration in Cx43 localization. Cx43 can still be visualized in gap junction plaques at sites of cell contact in both M5a3 and WT myoepithelial cells. (C) WT cells were treated with either control media or Wnt5a conditioned media. No changes in Cx43 localization can be seen with Wnt5a treatment. K14 staining reveals normal cytoskeletal structures with Wnt5a treatment, indicating Wnt5a does not affect the cytoskeletal organization. 49

60 overexpression prevents transgenic dams from feeding their pups, and that this lactation phenotype results from failed milk ejection in response to oxytocin. These findings suggest that myoepithelial cell function may be strongly impacted by Wnt5a signaling, although myoepithelial cells were present and appeared to be fully differentiated as measured by expression of asma, K14, and p63. The results indicate that Wnt5a regulates myoepithelial cell function in M5a mice rather than regulating early stages of differentiation. Unexpectedly, expression of Wnt5a in M5a mice resulted in only very mild effects on development during puberty. Since previous reports from our lab showed that implantation of Wnt5a slow release pellets can inhibit ductal extension and end bud proliferation during puberty (Roarty and Serra, 2007), we expected to observe a more robust branching phenotype in Wnt5a mice. It is possible that the level of Wnt5a transgene expression during puberty in the M5a mice is not sufficient to generate more robust alterations in mammary gland morphology (Figure 1C). Nevertheless, Wnt5a transgene expression in M5a mice increases dramatically during pregnancy and into lactation due to the hormone responsive nature of the MMTV promoter. Therefore, we were able to bypass any effects Wnt5a may have on early mammary gland development and focus on the role of Wnt5a in pregnancy-associated development and the onset of lactation. The milk-let down response that occurs at parturition requires oxytocin stimulation and contraction of myoepithelial cells (Uvnas-Moberg and Eriksson, 1996). However, very little is known about the paracrine and autocrine factors regulating oxytocin responses in the mammary gland at the time of parturition. Milk ejection requires Cx43, as a functional mutation in Cx43 abrogates oxytocin response (Plante and Laird, 2008). 50

61 Cx43 disruption only mildly impairs mammary gland development, but lactation is inhibited and glands fail to respond to oxytocin. In this study we provide evidence that Wnt5a may be one of the paracrine factors acting to mediate the oxytocin response and maintain the mammary myoepithelial cells in an unresponsive state until parturition has occurred. The results suggest that changes in Cx43 phosphorylation between late pregnancy and early lactation may be an important control mechanism for the milk-let down response that occurs at parturition, and that Wnt5a can prevent changes in Cx43 phosphorylation at this point. This data coupled with the data showing impaired oxytocin responses in M5a mammary glands suggests that Wnt5a inhibits milk ejection through altered Cx43 phosphorylation. Cx43 is widely expressed and is also important for myometrial cell contraction at parturition in response to oxytocin (Reversi et al., 2005). When Cx43 is disrupted in the uterus, parturition is delayed and not all pups are delivered (Tong et al., 2009). Likewise, we observed a mild parturition phenotype in the M5a mice. Almost 20% of M5a3 dams and 15% of M5a2 dams failed to deliver their pups by 21.5 dpc (Table 1). Expression of MMTV transgenes in the uterus has been documented (Marozkina et al., 2008; Strizzi et al., 2007). Using the Rosa26 LacZ reporter and MMTV-Cre mice, our lab verified uterine MMTV expression, observing staining in the uterine glands of 5 week old mice (data not shown). We also examined expression of Wnt5a in protein isolated from whole uterus both during late pregnancy and at 1d pp. Wnt5a protein was detected in M5a uteri, but weakly or not at all in wild type uteri (Figure 9). As both parturition and lactation are facilitated by oxytocin, we hypothesize that parturition may be impaired in transgenic females due to expression of Wnt5a. This phenotype is incompletely penetrant and re- 51

62 Table 1. Delayed parturition in M5a females by strain. Normal gestation in C57Bl/6 backgroud is 19.5 dpc. M5a3 and M5a2 females display delayed parturition, albeit with incomplete penetrance. Almost 20% of M5a3 and 14% of M5a2 females fail to deliver pups by 21 dpc compared to 0% of wildtype females. Strain <21 dpc n (%) >21 dpc n (%) WT 13 (100%) 0 (0%) M5a3 9 (81.8%) 2 (19.2%) M5a2 6 (85.7%) 1 (14.3%) 52

63 Figure 9. Wnt5a is expressed in M5a uterus at 17.5 dpc. Western blot analysis of protein lysate from 17.5 dpc wildtype (WT) or M5a3 uteri shows the presence of Wnt5a overexpression in M5a uterus, potentially impacting parturition. 53

64 quires further investigation. However, it further supports the hypothesis that Wnt5a may be regulating oxytocin responses through regulation of Cx43. The primary function of Cx43 is to mediate gap junction intercellular communication (GJIC). We propose disruption of GJIC leads to the inability of myoepithelial cells to contract together thereby preventing milk ejection. However, Wnt5a could also regulate GJIC-independent function of Cx43. In some cells, Cx43 has been linked to intracellular functions that can regulate cell buffering of Ca 2+, Na 2+, and ATP independently of its function in intercellular communication (Bruzzone et al., 2001; Cotrina et al., 1998; Song et al., 2010; Yi et al., 2010). Unknown GJIC-independent pathways could also have effects on myoepithelial cells that could influence contractility or response to oxytocin treatment. The importance of lactation to mammals is indisputable. However, the interaction between Cx43 and Wnt5a that we have described here for the first time may have broader implications. As yet, no papers have been published examining the effects of Wnt5a on Cx43 regulation at any level, transcriptional, translational, or post-translational. However, both Wnt5a and Cx43 are co-expressed in several tissues, such as limb bud (Laird et al., 1992; Yamaguchi et al., 1999) and heart (Clark et al., 1993; Lo et al., 1997). 54

65 Acknowledgements Breast cancer research in R. Serra s laboratory is supported by NIH R01 CA S.E. Baxley is supported by the UAB Cancer Prevention and Control Training Program, NCI R25 CA and MSTP training grant T32 GM We would like to thank the UAB transgenic mouse core facility, especially Bob Kesterson, for their help in generating the transgenic mice. We would like to acknowledge Renee Desmond for her help with statistical analysis and Brad Yoder and Zak Kosan for their help with confocal microscopy. 55

66 LIST OF REFERENCES Brisken, C. and Rajaram, R. D. (2006). Alveolar and lactogenic differentiation. J Mammary Gland Biol Neoplasia 11, Bruzzone, S., Franco, L., Guida, L., Zocchi, E., Contini, P., Bisso, A., Usai, C. and De Flora, A. (2001). A self-restricted CD38-connexin 43 cross-talk affects NAD+ and cyclic ADP-ribose metabolism and regulates intracellular calcium in 3T3 fibroblasts. J Biol Chem 276, Bry, C., Maass, K., Miyoshi, K., Willecke, K., Ott, T., Robinson, G. W. and Hennighausen, L. (2004). Loss of connexin 26 in mammary epithelium during early but not during late pregnancy results in unscheduled apoptosis and impaired development. Dev Biol 267, Clark, C. C., Cohen, I., Eichstetter, I., Cannizzaro, L. A., McPherson, J. D., Wasmuth, J. J. and Iozzo, R. V. (1993). Molecular cloning of the human proto-oncogene Wnt-5A and mapping of the gene (WNT5A) to chromosome 3p14-p21. Genomics 18, Cotrina, M. L., Lin, J. H., Alves-Rodrigues, A., Liu, S., Li, J., Azmi-Ghadimi, H., Kang, J., Naus, C. C. and Nedergaard, M. (1998). Connexins regulate calcium signaling by controlling ATP release. Proc Natl Acad Sci U S A 95, Gavin, B. J. and McMahon, A. P. (1992). Differential regulation of the Wnt gene family during pregnancy and lactation suggests a role in postnatal development of the mammary gland. Mol Cell Biol 12, Gavrieli, Y., Sherman, Y. and Ben-Sasson, S. A. (1992). Identification of programmed cell death in situ via specific labeling of nuclear DNA fragmentation. J Cell Biol 119, Grigoryan, T., Wend, P., Klaus, A. and Birchmeier, W. (2008). Deciphering the function of canonical Wnt signals in development and disease: conditional loss- and gain-offunction mutations of beta-catenin in mice. Genes Dev 22, Kikuchi, A., Kishida, S. and Yamamoto, H. (2006). Regulation of Wnt signaling by protein-protein interaction and post-translational modifications. Exp Mol Med 38, Kuhl, M., Sheldahl, L. C., Park, M., Miller, J. R. and Moon, R. T. (2000). The Wnt/Ca2+ pathway: a new vertebrate Wnt signaling pathway takes shape. Trends Genet 16, Laird, D. W., Yancey, S. B., Bugga, L. and Revel, J. P. (1992). Connexin expression and gap junction communication compartments in the developing mouse limb. Dev Dyn 195, Lo, C. W., Cohen, M. F., Huang, G. Y., Lazatin, B. O., Patel, N., Sullivan, R., Pauken, C. and Park, S. M. (1997). Cx43 gap junction gene expression and gap junctional communication in mouse neural crest cells. Dev Genet 20, Marozkina, N. V., Stiefel, S. M., Frierson, H. F., Jr. and Parsons, S. J. (2008). MMTV-EGF receptor transgene promotes preneoplastic conversion of multiple steroid hormone-responsive tissues. J Cell Biochem 103, Nakano, H., Furuya, K., Furuya, S. and Yamagishi, S. (1997). Involvement of P2- purinergic receptors in intracellular Ca2+ responses and the contraction of mammary myoepithelial cells. Pflugers Arch 435,

67 Peaker, M. (2002). The mammary gland in mammalian evolution: a brief commentary on some of the concepts. J Mammary Gland Biol Neoplasia 7, Plante, I. and Laird, D. W. (2008). Decreased levels of connexin43 result in impaired development of the mammary gland in a mouse model of oculodentodigital dysplasia. Dev Biol 318, Plante, I., Wallis, A., Shao, Q. and Laird, D. W. (2010). Milk secretion and ejection are impaired in the mammary gland of mice harboring a Cx43 mutant while expression and localization of tight and adherens junction proteins remain unchanged. Biol Reprod 82, Reversi, A., Cassoni, P. and Chini, B. (2005). Oxytocin receptor signaling in myoepithelial and cancer cells. J Mammary Gland Biol Neoplasia 10, Richert, M. M., Schwertfeger, K. L., Ryder, J. W. and Anderson, S. M. (2000). An atlas of mouse mammary gland development. J Mammary Gland Biol Neoplasia 5, Roarty, K., Baxley, S. E., Crowley, M. R., Frost, A. R. and Serra, R. (2009). Loss of TGF-beta or Wnt5a results in an increase in Wnt/beta-catenin activity and redirects mammary tumour phenotype. Breast Cancer Res 11, R19. Roarty, K. and Serra, R. (2007). Wnt5a is required for proper mammary gland development and TGF-beta-mediated inhibition of ductal growth. Development 134, Song, D., Liu, X., Liu, R., Yang, L., Zuo, J. and Liu, W. (2010). Connexin 43 hemichannel regulates H9c2 cell proliferation by modulating intracellular ATP and [Ca2+]. Acta Biochim Biophys Sin (Shanghai) 42, Strizzi, L., Bianco, C., Hirota, M., Watanabe, K., Mancino, M., Hamada, S., Raafat, A., Lawson, S., Ebert, A. D., D'Antonio, A. et al. (2007). Development of leiomyosarcoma of the uterus in MMTV-CR-1 transgenic mice. J Pathol 211, Tong, D., Lu, X., Wang, H. X., Plante, I., Lui, E., Laird, D. W., Bai, D. and Kidder, G. M. (2009). A dominant loss-of-function GJA1 (Cx43) mutant impairs parturition in the mouse. Biol Reprod 80, Uvnas-Moberg, K. and Eriksson, M. (1996). Breastfeeding: physiological, endocrine and behavioural adaptations caused by oxytocin and local neurogenic activity in the nipple and mammary gland. Acta Paediatr 85, Veeman, M. T., Axelrod, J. D. and Moon, R. T. (2003). A second canon. Functions and mechanisms of beta-catenin-independent Wnt signaling. Dev Cell 5, Widelitz, R. (2005). Wnt signaling through canonical and non-canonical pathways: recent progress. Growth Factors 23, Yamaguchi, T. P., Bradley, A., McMahon, A. P. and Jones, S. (1999). A Wnt5a pathway underlies outgrowth of multiple structures in the vertebrate embryo. Development 126, Yi, F. X., Boeldt, D. S., Gifford, S. M., Sullivan, J. A., Grummer, M. A., Magness, R. R. and Bird, I. M. (2010). Pregnancy enhances sustained Ca2+ bursts and endothelial nitric oxide synthase activation in ovine uterine artery endothelial cells through increased connexin 43 function. Biol Reprod 82,

68 DISCUSSION Summary We generated mice (M5a) that overexpress Wnt5a in the mammary epithelium and demonstrated that these mice exhibit defects in the milk ejection response during lactation. Wnt5a overexpression did not affect lineage differentiation, as markers of all mammary epithelial subtypes appeared both histologically and grossly similar between transgenic and wildtype females. Localization of gap junctions appeared normal. However, at parturition M5a females displayed an altered Cx43 phosphorylation pattern compared to wildtype controls, indicating a modification in Cx43 function. Cx43 modifications and function Connexin proteins possess four hydrophobic membrane spanning domains with two conserved, extracellular domains that allow docking to adjacent cells (reviewed in Solan and Lampe 2005 and Laird 2005). Within the membrane, connexins have a single intracellular loop and two tails. The carboxy-terminal tail region is variable in length and contains phosphorylation sites. In the case of Cx43, the carboxy-terminal tail contains 12 known serine and two tyrosine phosphorylation sites. The high turnover rate of Cx43 at the plasma membrane (1-5 hours) suggests a high level of post-translational modifications like phosphorylation (Laird et al 1991; Beardslee et al 1998). Many proteins are known to phosphoryate Cx43 including v-src, PKA, Casein kinase 1 (CK1), PKC, and 58

69 MAPK. Cx43 is differentially phosphorylated throughout the cell cycle, and in cultured cells phosphorylation of Cx43 can occur within 15 minutes of synthesis (Crow et al 1990; reviewed in Solan and Lampe 2005). Phosphorylation on these different sites by different enzymes regulates a wide variety of actions including Cx assembly into hemichannels, gating, incorporation into plaques, endocytosis, and degradation (reviewed in Solan and Lampe 2007). Most studies examining the role of connexins in development or disease focus on gap junction intercellular communication (GJIC), or the passive diffusion of intracellular molecules through gap junctions to a neighboring cell. A variety of cellular processes have been associated with GJIC, such as buffering of cytoplasmic ions, cell migration and proliferation, differentiation, electrical coupling, metabolism, and carcinogenesis (Krysko et al 2005; Mesnil et al 2005; Wong et al 2008). Within the mammary gland GJIC has been shown to induce differentiation of epithelial cells through decreased canonical Wnt signaling (El-Sabban et al 2003; Talhouk et al 2008). Additionally, gap junction permeability is necessary for proper milk secretion through Cx26 and Cx32 during lactation (Locke et al 2004). GJIC plays a role in breast cancer development as well (Carystinos et al 2001), a topic to be discussed in more detail later in this section. One of the major roles of GJIC is to transmit signals to neighboring cells through the use of second messengers and small molecule messengers such as Ca 2+, ATP, and Na +. Particularly for this study, communication between cells in contractile tissue such as the myoepithelial cells of the mammary gland requires functional gap junctions. GJIC is necessary for Ca 2+ wave propragation and mesangial cell contraction (Yao et al 2002). Within the uterus, a dominant negative Cx43 mutation, Gja1 Jkt, leads to decreases in elec- 59

70 trical coupling between myometrial cells and is associated with decreased response to oxytocin stimulation, preventing proper parturition (Tong et al 2009). Additionally, Gja1 Jkt mammary epithelial cells display reduced cell coupling, as assessed with a dye transfer assay (Plante and Laird 2008). This disruption in GJIC leads to failure in the milk ejection response, indicating the importance of GJIC on myoepithelial cell function during lactation. As Wnt5a overexpression altered Cx43 phosphorylation and inhibited the milk ejection response, GJIC may be inhibited in M5a glands, leading to lactation failure. Calcium and ATP regulation in myoepithelial contractions Ca 2+ transmission through gap junctions, including those mediated by Cx43, can facilitate synchronized contractions in several cell types. Within the mammary gland, Moore et al (1987) demonstrated that mammary myoepithelial cells require extracellular calcium for proper contraction and that treatment with a calmodulin antagonist also attenuates contractility, indicating that intra- and extra-cellular calcium is required for myoepithelial cell contraction. Additionally, oxytocin increases intracellular calcium levels in conjunction with ATP signaling to induce myoepithelial cell contraction (Nakano et al 1997). ATP has also previously been shown to aid in oxytocin-induced responses in myoepithelial cells (Nakano et al 1997) and in renal epithelium (Chen et al 2008; Geyti et al 2008). Myoepithelial cells in other organ systems such as skin, salivary gland, and C. elegans sheath cells, similarly require calcium during contractions (Balls and Holmes 1978; Nishiyama et al 1980; Yin et al 2004). These data suggest that myoepithelial cells require well-regulated intracellular calcium concentrations for proper contractility and 60

71 that ATP release can facilitate Ca 2+ induced contractions, implying the importance of ATP for contractility. Based on this data, myoepithelial cells depend on Ca 2+ release for contraction, and oxytocin induces mammary myoepithelial contraction through a Ca 2+ - dependent, ATP-facilitated mechanism. Connexin hemichannels have also been shown to regulate intracellular Ca 2+ levels, some specifically in a GJIC-independent manner. Initial studies in fibroblasts demonstrated that ADP-metabolism and intracellular Ca 2+ were regulated by cross-talk between Cx43 and CD38 (Bruzzone et al 2001). During pregnancy uterine endothelial cells exhibit cyclic intracellular Ca 2+ and nucleotide waves that are associated with increased Cx43 expression and enhanced uterine circulation (Yi et al 2010), implying that Cx43 can regulate intracellular Ca 2+ in endothelium. Blockade of Cx43 hemichannel function leads to decreased intracellular Ca 2+ and ATP concentrations (Song et al 2010). Glioma, HeLa, and glioblastoma cells displayed increased ATP release upon forced connexin expression (Cotrina et al 1998). Ca 2+ wave propagation in these cells was sensitive to purinergic receptor antagonists but insensitive to gap junction inhibition, suggesting the functions of ATP and Cx in regulating intracellular calcium are gap-junction independent. Thus, connexins, including Cx43, regulate intracellular calcium and ATP release. Although the main function of connexins is generally thought to be gap junction communication as alluded to above, a large body of evidence also reveals the importance of connexins in gap junction-independent calcium regulation. Mammary myoepithelial cells require Cx43 for proper oxytocin-induced milk ejection (Plante and Laird 2008; Baxley and Serra, submitted). Gap junction communication likely mediates the majority of the oxytocin response; however, we suggest that calcium regulation may also strongly 61

72 influence myoepithelial contractility, and, therefore, milk ejection. Our data suggests that Wnt5a overexpression may inhibit lactation by downregulating Cx43 function, either through inhibition of GJIC or intracellular Ca 2+ release. In the future, we hope to test both of these mechanisms in Wnt5a inhibition of lactation. Potential Cx43/Wnt5a interactions: organogenesis To our knowledge, this is the first report of any interaction between Cx43 and Wnt5a. We have introduced the idea that Wnt5a regulates Cx43 function through alterations in phosphorylation. Although our research focuses on the role of Wnt5a and Cx43 specifically in mammary gland development, both Cx43 and Wnt5a are known to be expressed in other organs. Cx43 plays a major role in heart development and over a quarter of all articles on Cx43 available through the national library of medicine discuss its role in development or disease of the heart. Myocardium requires constant and consistent cell coupling to coordinate contractions. Cardiomyocytes express high levels of Cx43 in the intercalated discs between cells (Fromaget et al 1992), allowing a high degree of functional coupling between adjacent cells. In addition, Cx43 is expressed in neural crest cells, which migrate to the primitive cardiac streak to initiate heart development (Lo et al 1997). Cx43 knockout mice die at birth of cyanosis due to heart malformation (Reaume et al), specifically right ventricular outflow tract obstruction leading to failure of blood flow from the right ventricle into the pulmonary arteries. Closer examination revealed aberrant septum formation, leading to increased septum in the conus of the right ventricle and outflow tracts. In addition to its role in cardio-myocyte contraction, Cx43 clearly regulates heart forma- 62

73 tion, as loss of Cx43 leads to increased ventricular septum formation. Inhibition of canonical Wnt signaling, a known function of Wnt5a (Roarty and Serra 2007), can initiate Xenopus cardiogenesis (Schneider VA, Mercola M 2001). Human neonatal heart tissue expresses Wnt5a (Clark et al 1993), and loss of Wnt5a in heart tissue results in a loss of septum formation (Schleiffarth et al 2007), indicating the importance of Wnt5a in heart formation. The limb bud also expresses both Cx43 and Wnt5a, specifically in the apical ectodermal ridge (AER) (Yancey et al 1992; Yamaguchi et al 1999). Within the limb bud, Wnt5a activates the PCP pathway to coordinate cell movement and convergent extension, and mutations in Wnt5a and its receptor ROR2 have been associated with Robinow syndrome, characterized by short stature and craniofacial abnormalities (Gros et al 2010; Person et al 2010; Wang et al 2010). Similarly, mutations in Cx43 are linked with oculodentodigital dysplasia, characterized by craniofacial abnormalities and limb dysmorphisms (Paznekas WA et al 2003). Potential Cx43/Wnt5a interactions: tumorigenesis Cancer cells frequently usurp normal developmental pathways to gain functions they did not previously possess or lose the inhibitory effects of tumor suppressors. As discussed in the introduction, Wnt5a can act as a tumor suppressor. Cx43 also plays a role in breast cancer progression. Addition of Cx43 to breast cancer cells decreases growth and tumor formation rates and restores differentiation potential (Hirschi et al 1996), while knockdown of Cx43 using sirna increases proliferation and migration (Shao et al 2005). Knockdown of Cx43 in breast cancer cells also enhances lung metas- 63

74 tasis (Li et al 2008). Examination of breast tumors demonstrated increased cytoplasmic staining patterns for Cx43 in tumors as compared to the plasma membrane staining seen in surrounding normal tissue (Kańczuga-Koda et al 2003). Retroviral infection of MDA- MB-231 cells with human Cx43 suppressed tumor growth when transplanted into the mammary fat pad (Qin et al 2002). Cx43 can also act as a tumor suppressor in other cancer types. Ovarian carcinoma cells have decreased GJIC and Cx43 expression relative to their normal ovarian counterpart (Hanna et al 1999), and ovarian carcinoma samples taken from patients reveal decreased Cx43 staining in tumor tissue (Umhauer et al 2000). Tumor grade and Cx43 expression were inversely correlated in both endometrial cancer (Schlemmer et al 1999) and glioma (Pu et al 2004). Alternately, Cx43 and GJIC can increase growth, invasiveness, and migration of cancer cells. Cx43 expression in mammary stroma increased in DCIS and invasive carcinomas, while normal breast stroma did not express Cx43 (Jamieson et al 1998). Additionally, 50% of invasive carcinomas stained positively for Cx43, albeit most staining was cytoplasmic. Kanczuga-Koda et al (2005) demonstrated a positive correlation between Cx43 expression and tumor grade in breast cancer samples. Again, 90% of tumors showed Cx43 expression predominantly in the cytoplasm, suggesting non-gjic functions. Increases in the phosphorylated forms of Cx43 were shown in epithelial cells of in situ carcinomas and invasive breast carcinomas (Gould et al 2005). Overexpression of Cx43 increased breast cancer homing to the lung, while knockdown of Cx43 decreased adhesion to pulmonary endothelium, suggesting Cx43 is required for metastasis (Elzarrad et al 2008). Lymph node breast cancer metastases often stain positive for Cx43, both in the 64

75 cytoplasm and at the plasma membrane, even if the primary tumor did not express Cx43 (Kanczuga-Koda et al 2006). This data suggests that Cx43 can influence tumorigenesis either negatively or positively, depending on tumor context. TGF-β, an upstream regulator of Wnt5a in the mammary gland, displays a dualnature as well, acting as a tumor suppressor in early tumorigenesis but enhancing invasion and metastasis at later stages. Similarly, Wnt5a has also been implicated as a tumor suppressor in the mammary gland and enhancer in other cancers such as melanoma. As both Wnt5a and Cx43 can regulate tumorigenesis in a context and organ specific manner, they may functionally interact to inhibit tumor formation, enhance metastasis, or counterbalance each other. Perhaps some of the organ specific effects of Wnt5a may be influenced by function and/or levels of Cx43 in that tissue. Potential Cx43/Wnt5a interactions: canonical Wnt signaling Interestingly, Cx43 was identified as a target of canonical Wnt signaling more than a decade ago (van der Heyden et al 1998) and is frequently used as a read out for canonical Wnt activation in cell lines (Samarzija et al 2009; Robinson et al 2006). Wnt1 upregulates Cx43 at both the transcriptional and translational level and this increase in Cx43 enhances gap junction communication (van der Heyden et al 1998; Ai et al 2000). Our lab and other labs previously demonstrated that Wnt5a can inhibit canonical Wnt signaling (Roarty et al 2009), so Wnt5a could potentially inhibit canonical Wnt signaling to regulate Cx43. However, Wnt5a overexpression did not alter Cx43 expression levels, implying that Wnt5a may be acting downstream of canonical Wnt-regulated Cx43 transcription. 65

76 Based on the data presented, we propose that Wnt5a may be regulating canonical Wnt effects through post-translational modification of canonical Wnt regulated genes such as Cx43. In order to examine the role of canonical Wnt inhibition in M5a glands, we examined the levels of active-β-catenin by western blot. The antibody recognizes the unphosphorylated, active form of β-catenin but not the phosphorylated form normally bound to ECadherin or Axin/GSK3β. Wildtype and M5a3 primary mammary epithelial cells (PMECs) were grown in culture for three days and then treated for 48 hours with 1 mg/ml oxytocin (Figure 1B). Active β-catenin, an indicator of canonical Wnt activation increased in the wildtype cells compared to the M5a3 cells. Interestingly, M5a3 PMECs expressed less K14, a myoepithelial cell marker, potentially implying regulation of myoepithelial cell growth. However, we have not observed any other alteration in the myoepithelial cell population in vivo. Myoepithelial markers are normal in M5a3 tissue and K14 expression, and localization appears normal in isolated myoepithelial cells both with M5a3 tissue and with Wnt5a treatment in vitro (Chapter 2, Figures 7,8). To test specifically for myoepithelial cell activation, we isolated myoepithelial cells as previously described (Chapter 2), pre-treated with control or Wnt5a conditioned media for 2 hours, and treated with oxytocin for 48 hours. Wnt5a treatment strongly downregulated active-β-catenin without affecting total β-catenin levels (Figure 1C). However, oxytocin treatment downregulated active-β-catenin in the control treatment, but upregulated it with Wnt5a co-treatment. This indicts that Wnt5a downregulates canonical Wnt signaling in mammary myoepithelial cells, but the connection between Wnt5a, canonical Wnt, Cx43, and oxytocin remains unclear. 66

77 Figure 1. Wnt5a inhibits canonical Wnt signaling in mammary myoepithelial cells and mildly rescues Wnt1 phenotype in mammary glands. (A) Whole mount analysis of Wnt1, Wnt1/M5a3, and M5a3 littermates at 8 weeks shows that Wnt1/M5a3 mammary glands closely resemble Wnt1 glands. However, extension of the epithelium appears slightly greater in the double transgenic compared to Wnt1 alone, indicating Wnt5a may be partially inhibiting the canonical Wnt signal. (B) PMECs were cultured from WT and M5a3 glands and either treated or not treated with oxytocin. Active β-catenin is increased in the WT cells and decreases slightly with oxytocin treatment. M5a3 PMECs do not have strong staining for active β-catenin, indicating lower canonical Wnt signaling in these cells. M5a3 PMECs show reduced expression of K14, indicating potentially fewer myoepithelial cells in the culture. (C) WT myoepithelial cells were treated with control or Wnt5a conditioned media in the presence or absence of oxytocin. Active β-catenin decreases in control conditions with oxytocin treatment. However, cells treated with Wnt5a demonstrate the opposite. Treatment with Wnt5a decreases basal levels of active β-catenin. Combined Wnt5a and oxytocin treatment augments canonical Wnt signaling, leading to increased β-catenin activation without affecting total β-catenin levels. K14 is used as a loading control. 67

78 68

79 Knowing that Wnt5a could inhibit canonical Wnt signaling, we crossed M5a3 mice to Wnt1 mice which exhibit overactive canonical Wnt signaling in the mammary gland. Wnt1 mammary glands develop precocious alveolar structures during puberty and branch so expansively that extension of the epithelium through the fat pad is severely retarded (Figure 1A). A M5a3/Wnt1 double positive mammary gland at 8 weeks exhibited a similar phenotype to Wnt1 alone, with only mildly increased extension of the epithelium past the lymph node, indicating that although Wnt5a can inhibit canonical Wnt signaling, the effects are not strong enough to completely reverse the Wnt1 phenotype. Wnt5a may be inhibiting a subset of canonical Wnt signaling that does not include epithelial proliferation and branching. In this study, we did not examine markers of differentiation within these glands, nor did we examine Cx43 phosphorylation, both of which may be affected by Wnt5a inhibition of canonical Wnt signaling. Stem cells The initial aim of this study was to examine the role of Wnt5a and TGF-β on mammary gland stem cell maintenance. Previously, we demonstrated increased progenitor cell marker expression and altered tumor phenotype with loss of TGF-β or Wnt5a, indicating a role in progenitor cell maintenance and suggesting a potential role in stem cell regulation. To that end we directly examined the mammary gland stem cell population through flow cytometry. To isolate the stem cell population, mammary glands were digested to single cell suspension and negatively selected against immune and endothelial cell markers TER119, CD45, and CD31, respectively, to remove non-epithelial Lin + cells. The remaining so called Lin - population were stained with CD24-PE and CD49f-FITC to 69

80 identify the subpopultions of mammary cells. Stem cells in the mammary gland are Lin - ;CD24 + ;CD49f + (Shackleton et al 2006). DNIIR tissue exhibits loss of TGF-β signaling, but flow cytometry for the double positive stem cell population shows no difference between DNIIR and wildtype controls (Figure 2A). Utilizing the mammosphere assay, as described in the introduction, we tested mammosphere formation in DNIIR glands versus wildtype controls. In wildtype cells, mammosphere formation was reduced with TGF-β treatment, but, as expected, no change was observed upon treatment with TGF-β in mammosphere formation with DNIIR cells (Figure 2B). If TGF-β signaling decreases the stem and/or progenitor cell population, we would expect to see an increase in mammosphere formation with a loss of TGF-β signaling. However, DNIIR mammary glands formed mammospheres at a similar rate compared to controls. This mammosphere data combined with the stem cell data suggests that TGF-β does not act directly to regulate stem cell number in the mammary gland. The data presented here needs to be verified through repetition in order to ensure accuracy, as only two biological replicates were tested in each case. As loss of Wnt5a similarly regulated markers of progenitor cells, we wanted to examine the role ofwnt5a in mammary stem cell maintenance. Wnt5a null tissue requires transplantation, as pups die peri-natally. Consequently, the tissue used for flow cytometry had been transplanted twice to increase the amount of tissue available for experiments. Although only one sample has been tested, we observed no difference in the stem cell population between Wnt5a null tissue and transplanted control tissue. Utilizing the M5a mouse model, we crossed M5a3 mice into the Wnt1 background and performed mammosphere on Wnt1 versus Wnt1/M5a3 double positive tissue. Wnt1 tissue contains 70

81 Figure 2. TGF-β and Wnt5a do not directly regulate mammary stem cells. (A) Stem cells in the mammary gland are CD24 + ;CD49f +. By flow cytometry, loss of TGF- β does not change stem cell numbers relative to WT (Bl/6) control. (B) Mammosphere formation in WT cells is reduced upon treatment with TGF- β. Mammospheres formed from DNIIR epithelium fail to respond to TGF- β administration. WT and DNIIR mammosphere formation are indistinguishable. (C) Similar to results seen in DNIIR tissue, loss of Wnt5a does not change the number of stem cells by flow cytometry relative to control (Bl/6). (D) Mammosphere forming capacity of Wnt1 and Wnt/M5a3 appears similar (blue bars). TGF- β treatment reduces mammosphere formation in Wnt1/M5a3 epithelium. 71