Investigation of the Role of Novel Hormone Regulated Genes in Mammary Gland Development and Carcinogenesis
Heidi Nicole Hilton
A thesis submitted for the degree of Doctor of Philosophy in the Faculty of Medicine at the University of New South Wales
Acknowledgements
Having reached this point where I am almost finished and about to disembark the PhD rollercoaster, I am all too aware of the people that have helped me get here. First and foremost I must thank my supervisor, Chris Ormandy. Thank you for giving me enough freedom to learn and try out new things on my own, yet providing good and solid support whenever it was required. Your acceptance of me having a life outside of my PhD has made this experience far closer to enjoyable than traumatic! What I have learned, and the experiences I have gained both inside and outside of the lab, probably could not have been matched had I not worked in your group over the past 6 years, so thank you.
Roger Daly – thanks for stepping up as my co-supervisor mid-way through the event. You were definitely the one for the job! Thanks for all your help and support, and especially with getting that Goblin paper over the line in the end.
Prue Stanford – thank you for teaching me so much when I first started working at the Garvan and onwards, for being a fantastic part-time co- supervisor, and for giving me such good support and encouragement when it was sorely needed! On top of that you have also been a great friend to me. Thank you!
Sam Oakes – what can I say Sammy? It’s been extreme! Thank you so much for all your advice and encouragement with my work, not to mention all the fun times we have had in the lab, at the Green Park, and of course at conferences! I don’t think living in different cities (or countries) will ever make us lose touch!
I am indebted to Assoc Prof Liz Musgrove for helpful discussion and use of reagents, to Chehani Alles for doing such a great job with the microarray analyses in a short amount of time, to Maria Kalyuga for generating the Elf5
pHUSH and ProEx constructs, and Warren Kaplan for his great bioinformatic work on Goblin. Thank you!
Many thanks must go to all the past and present Development Group members – Matt, Jess, Fi, Cara, Heather, Wendy, Renee and Anita, and particularly Katrina, for being my ABI fairy and close friend.
And of course thank you to a number of other past and present Garvanites – Gillian for all her help in TC plus more, Tilman, Darren, Nick and Ashleigh for making work so fun, and especially Liz C, the genius with a gold heart, who has given me so much help and advice and answered my endless questions over the years!
To my family who are always there for me no matter what. Mum and Dad, thank you so much for your endless support in everything I do. I know you would be equally proud no matter what I had done. Thank you.
Ryan – perhaps my personal highlight of this PhD was that I met you during it. Thank you so much for keeping me relaxed and giving me perspective during this time, as well as countless other things. I look forward to whatever lies ahead for us.
Financial support
University Postgraduate Award (UNSW); August 2004 – January 2008
Manuscripts arising from this thesis
Oakes SR, Hilton HN, Ormandy CJ. The alveolar switch: coordinating the proliferative cues and cell fate decisions that drive the formation of lobuloalveoli from ductal epithelium. Breast Cancer Res. 8(2):207 (2006) (see Appendix I)
Hilton HN, Stanford PM, Harris J, Oakes SR, Kaplan W, Daly RJ, Ormandy CJ. KIBRA interacts with discoidin domain receptor 1 to modulate collagen-induced signalling. Biochim Biophys Acta. 1783(3);383-93 (2008) (see Appendix II)
Oakes SR, Naylor MJ, Asselin-Labat ML, Blazek KD, Gardiner-Garden M, Hilton HN, Kazlauskas M, Pritchard MA, Chodosh LA, Pfeffer PL, Lindeman GJ, Visvader JE, Ormandy CJ. The Ets transcription factor Elf5 specifies mammary alveolar cell fate. Genes Dev. 22(5):581-6 (2008) (see Appendix III)
Abbreviations -/- Knockout +/+ Wildtype 2D Two-dimensional 3D Three-dimensional AR Androgen receptor AREG Amphiregulin BLAST Basic local alignment search tool oC Degrees Celcius Ccnd1 Cyclin D1 ChIP Chromatin immunoprecipitation cDNA Complementary DNA DDR1 Discoidin domain receptor 1 DNA Deoxyribonucleic acid Dox Doxycycline dpc Days post coitus dpp Days post partum DSP Desmoplakin ECM Extracellular matrix E Estrogen EcoR Ecotrophic receptor EGF Epidermal growth factor EGFR Epidermal growth factor receptor Elf3 E74-like factor 3 Elf5 E74-like factor 5 ER Estrogen receptor ER Estrogen receptor ErbB2 v-erb-b2 erythroblastic leukemia viral oncogene homolog 2 EST Expressed sequence tag Ets E twenty-six
7
FACS Fluorescence activated cell sorting FAK Focal adhesion kinase FASN Fatty acid synthase FBS Foetal bovine serum FGF Fibroblast growth factor Gal Galanin GAPDH Glyceraldehyde-3-phosphate dehydrogenase GATA-3 GATA binding protein 3 GFP Green fluorescent protein GH Growth hormone GHR Growth hormone receptor h Hours HEK-293 Human embryonic kidney-293 HELU Hyperplastic enlarged lobular unit HEPES 4-(2-hydroxyethyl)-1- piperazineethanesulfonic acid HMEC Human mammary epithelial cell HRT Hormone replacement therapy IF Immunofluorescence IGF-1 Insulin-like growth factor 1 IHC Immunohistochemistry IgG Immunoglobulin G IKK Kinase IkB IP Immunoprecipitation IRES Internal ribosome entry site Jak2 Janus kinase 2 kDa Kilodalton MAPK Mitogen activated protein kinase MEC Mammary epithelial cell
8
min Minutes MMTV Mouse mammary tumour virus MMP Matrix metalloproteinase NaCl Sodium chloride NF1 Nuclear factor 1 NF- B Nuclear Factor-kappa B NLB Normal lysis buffer Nrg Neuregulin PBS Phosphate buffered saline PCR Polymerase chain reaction Pg Progesterone Pgr Progesterone receptor PgrA Progesterone receptor A PgrB Progesterone receptor B PI3K Phosphatidylinositol 3-kinase PKC Protein kinase C PKC Protein kinase C zeta PL Placental lactogen PRE Progesterone receptor element PTH1R Parathyroid hormone-related protein receptor PTHrP Parathyroid hormone-related protein Prl Prolactin Prlr Prolactin receptor PVDF Polyvinylidene Fluoride RACE Rapid amplification of complementary DNA ends RankL Receptor activator of NF-kB-ligand RNA Ribonucleic acid rpm Revolutions per minute
9
SDS Sodium dodecyl sulfate Shc Src homology 2 domain containing transforming protein SH2 Src homology 2 siRNA Small interfering RNA Socs Suppressor of cytokine signalling Sos Son of sevenless Srebp1 Sterol regulatory element-binding protein 1 Stat5 Signal transducer and activator of transcription 5 TDLU Terminal ductal lobular unit TEB Terminal end bud TGF Transforming growth factor, alpha TGF 1 Transforming growth factor, beta 1 Tyr Tyrosine WAP Whey acidic protein WB Western blot Wnt Wingless-related MMTV integration site Wt Wildtype
10
Abstract
Mammary gland development is controlled by hormones such as progesterone and prolactin, which activate a genomic regulatory network. Identification of the components and regulatory links that comprise this network will provide the basis for defining the network's dynamic response during normal development and its perturbation during breast carcinogenesis. This thesis investigates two molecules in detail, Elf5 and KIBRA, which were identified as potential prolactin targets in a transcript profiling screen for key members in this genetic program of mammary morphogenesis.
We examined the effect of expression of Elf5, a transcription factor critical in alveolar differentiation, in a 3D culture model of non-transformed mammary epithelial MCF-10A cells. We discovered that Elf5 expression was selectively repressed over time in these cells when cultured on a basement membrane, and that Elf5 overexpression disrupted the architecture of acini resulting in luminal filling. This occurred due to an increase in the expression of epidermal growth factor receptor (EGFR) with repressed the induction of the pro- apoptotic molecule, Bim. We also observed that Elf5 is up-regulated with progesterone treatment, and that suppression of Elf5 expression in T47D breast cancer cells inhibits proliferation. Data obtained from the suppression of Elf5 expression in the presence of progesterone suggested that the role played by Elf5 in the Pg signalling pathway in T47D cells is relatively minor, and that rather than being a major downstream factor, the induction of Elf5 expression is utilised more to influence and potentiate other signalling pathways, such as the Prl pathway.
We characterised expression of KIBRA in the mammary gland and breast cancer cell lines, and observed that KIBRA was also up-regulated with progesterone treatment. Using a bioinformatic approach, we identified the tyrosine kinase receptor DDR1 as a binding partner of KIBRA. We have demonstrated that the WW domains of KIBRA bind to a PPxY motif in DDR1,
11
and that these molecules dissociate upon treatment with the DDR1 ligand, collagen. Finally, overexpression and knockdown studies demonstrate that KIBRA promotes the collagen-stimulated activation of the MAPK cascade. Thus KIBRA may play a role in how the reproductive state influences the mammary epithelial cell to respond to changing cell-context information, such as experienced during the tissue remodelling events of mammary gland development. Overall, the data presented in this thesis contributes to our growing knowledge of the genetic program responsible for mammary development and carcinogenesis.
12
Table of Contents
Abbreviations.……………………………………………………………………………..7
Abstract…………………………………………………………………………………….11
CHAPTER 1: INTRODUCTION ...... 17
Background ...... 18
Overview of Mammary Gland Cellular Morphology...... 19
Key Stages of Post-natal Mammary Gland Morphogenesis ...... 21 Ductal branching morphogenesis ...... 21 Alveolar morphogenesis...... 22 Involution...... 24
Hormonal Regulation of Mammary Development ...... 25 Regulation of ductal morphogenesis ...... 26 Regulation of alveolar morphogenesis ...... 29 Regulation of lactation and involution ...... 31
Mammary development and stem cells ...... 33
Signalling Pathways Involved in Alveolar Morphogenesis ...... 34 Molecular modulators of prolactin and progesterone induced alveolar morphogenesis ...... 34 The MAPK pathway ...... 36 Other factors involved in alveolar morphogenesis ...... 42
The Ets Transcription Factor Family and Elf5 ...... 46
Prolactin and Progesterone in Mammary Carcinogenesis ...... 50
Project Aims...... 52
CHAPTER 2: MATERIALS AND METHODS ...... 55
DNA Methods...... 56 Expression vector construction ...... 56 Site-directed mutagenesis ...... 57
Cell Culture Methods ...... 58
13
Cell lines ...... 58 HEK-293 cells ...... 58 MCF-10A cells...... 58 T47D and BT474 cells...... 58 Other cell lines ...... 59 Transient transfections ...... 59 Production of retrovirus ...... 59 Infection of MCF-10A cells ...... 59 Three-dimensional (3D) culture...... 60 siRNA duplexes and transfection...... 60 Ligand stimulation of cells...... 61 Progesterone receptor transactivation assay...... 61 Cell proliferation assays...... 62 Growth curve analysis ...... 62 MTT assays ...... 62 Flow Cytometry ...... 62
RNA Methods...... 63 Extraction, quantitation and analysis ...... 63
Protein Methods ...... 64 Antibody preparation...... 64 Cell lysate preparation and protein extraction...... 64 From monolayer cultures ...... 64 From 3D cultures ...... 65 From mammary glands ...... 65 Immunoprecipitation ...... 66 Western blot analysis ...... 66 Immunohistochemistry ...... 68
Immunofluorescence and microscopy ...... 68 Fixation and staining protocol of 2D cultures ...... 68 Fixation and staining protocol of 3D cultures ...... 69 Microscopy ...... 69
Bioinformatic Methods ...... 70 Statistical analysis ...... 70 Microarray analysis ...... 70 Functional Annotation Analysis ...... 71
CHAPTER 3: EFFECT OF ELF5 EXPRESSION ON MCF-10A CELLS IN 3D CULTURE...... 72
Introduction ...... 73
Results ...... 77 Generation of constitutive Elf5 overexpressing stable cell lines...... 77
14
Investigation of the effect of Elf5 expression on the proliferation of MCF-10A cells in two-dimensional culture...... 79 Investigation of the effect of Elf5 expression on acinar morphogenesis in three-dimensional culture ...... 81 Mechanism of Elf5 action on lumen formation in 3D culture ...... 89 Generation of inducible Elf5 overexpressing stable cell lines ...... 93 Effect of inducible Elf5 overexpression on levels of EGFR and Bim expression...... 96 Role of the ERK MAPK pathway in repression of Bim expression by Elf5 in acinar morphogenesis ...... 98
Discussion ...... 99
CHAPTER 4: INVESTIGATION OF THE ROLE OF ELF5 IN THE HUMAN BREAST CANCER CELL LINE, T47D...... 102
Introduction ...... 103
Results ...... 105 Hormonal regulation of Elf5 ...... 105 Regulation of Elf5 expression by sex-steroid hormones in the mouse mammary gland ...... 105 Regulation of human Elf5 expression by progestins in breast cancer cells ...... 106 Regulation of Elf5 expression by co-treatment of T47D breast cancer cells with estrogen and progestin...... 109 Regulation of Elf5 expression by growth hormone in T47D cells ...... 110 Effects of down-regulation of Elf5 expression in T47D breast cancer cells by siRNA...... 111 Changes in cellular morphology in T47D cells with suppressed Elf5 expression...... 112 Effect of suppression of Elf5 expression on cell cycle progression in T47D cells...... 112 Role of Elf5 in progestin-induced differentiation in T47Ds ...... 117 Identification of potential downstream targets of Elf5 in T47Ds ...... 119
Discussion ...... 137
CHAPTER 5: EXPRESSION PROFILE AND HORMONAL REGULATION OF KIBRA ...... 140
Introduction ...... 141
Results ...... 143 Generation of reagents ...... 143 Cloning of KIBRA ...... 143 Validation of KIBRA antibody ...... 144
15
Downregulation of KIBRA expression by RNA interference ...... 145 KIBRA expression in mammary gland and breast ...... 146 Expression of KIBRA protein in human breast cancer cell lines ...... 146 Localisation of KIBRA as visualised by immunofluorescence ...... 146 Localisation of KIBRA as visualised by immunohistochemistry...... 150 Effect of hormonal stimulation on KIBRA gene expression ...... 152 KIBRA is not a co-activator of the progesterone receptor...... 155 Effect of KIBRA overexpression levels on cell growth...... 156
Discussion ...... 161
CHAPTER 6: IDENTIFICATION AND CHARACTERISATION OF KIBRA INTERACTION WITH DDR1 ...... 163
Introduction ...... 164
Results ...... 166 Use of bioinformatic approach to identify KIBRA interacting partners ...... 166 KIBRA interacts with DDR1 in a collagen-regulated manner ...... 168 The WW domains of KIBRA bind the PPxY motif in DDR1 ...... 171 KIBRA interacts with DDR1 and PKC in a trimeric complex ...... 172 Modulation of KIBRA expression has downstream effects on the collagen- stimulated activation of ERK MAPK cascade ...... 173
Discussion ...... 177
CHAPTER 7: DISCUSSION ...... 181
Overview ...... 182
A Role for Elf5 in Proliferation and Survival of Mammary Epithelial Cells ...... 182
A Potential Role for Elf5 in Breast Cancer?...... 189
KIBRA and Modulation of ECM Signalling ...... 192
Summary and Conclusions...... 196
REFERENCES ...... 197
APPENDICES……………………………………………………………………233
16
Chapter 1: Introduction
17
Background
According to recent statistical reports, incidence rates of breast cancer in New South Wales, Australia have remained constant over the past 10 years, and mortality rates have fallen (Tracey et al. 2007). Breast cancer represents 30% of female cancer diagnoses, and remains the most common malignancy and cause of cancer-related death in Australian women today. Encouragingly, improvements in detection and treatment have seen survival rates continue to increase (Figure 1.1), indicating that in many cases cancer has been transformed from a critical disease into a chronic one.
Figure 1.1 Breast cancer survival rates. Age standardised relative survival (%) of female breast cancer patients at one, five, ten and twenty years post-diganosis in England and Wales (1971-2003) (Cancer Research UK www.cancerresearchuk.org 2008; Coleman et al. 1999; Coleman et al. 2004)).
Although survival rates are improving, incidence rates and development of resistance to treatment are increasing, demonstrating the need for new therapies that are truly curative. The rational design of new therapies requires an understanding of the etiology of breast cancer. Cancer can be viewed as
18
aberrant development, or neo-organogenesis, in which normal developmental mechanisms are perturbed by oncogenic insult. A key approach to understanding breast cancer is to define the normal developmental mechanisms that operate in the mammary gland and to determine how these mechanisms are corrupted during carcinogenesis. Development of the mammary gland occurs in defined stages, involving cellular processes such as proliferation, differentiation and apoptosis, as a result of hormonal cues that regulate reproductive events. Mammary carcinogenesis involves the mutation or dysregulation of genes required for normal mammary gland development. This may result in uncontrolled cell proliferation, loss of differentiation, sustained cell division, loss of apoptosis and a capacity to migrate and invade to other tissues and organs, the hallmark features of a cancerous cell. Therefore a detailed understanding of the complex network of the signalling systems that drive normal development of the mammary gland will provide the foundation to understand how these processes are perturbed during tumorigenesis.
Overview of Mammary Gland Cellular Morphology
The mammary gland is unique in that it only reaches complete development post-natally, following pregnancy and lactation. It is comprised of two major tissue compartments – the epithelium, which includes the ducts and milk- secreting alveolar cells, and the stroma, which consists of adipocytes, fibroblasts, blood vessels, inflammatory cells and the extracellular matrix (ECM) (Richert et al. 2000). Growing ducts end in bulbous terminal end bud (TEB) structures, which extend to completely fill the fat pad, and eventually give rise to basal myoepithelial cells and luminal alveolar cells. The alveolar cells form a sphere-like single layer of epithelial cells that envelopes a circular lumen, termed alveolus, which is connected to the ductal network via a single small duct. Upon functional differentiation of these luminal alveolar cells during pregnancy, they synthesise and secrete milk into the lumen of the alveolus where it exits the body via the nipple. The architecture of this extensive
19
system of TEBs and ducts embedded in the stroma is depicted in Figure 1.2. Terminal differentiation of the luminal cells into milk-secreting lobuloalveolar structures occurs in distinct stages tightly regulated by hormonal cues, as described in detail in the following sections.
Figure 1.2 Terminal end bud and duct morphology. TEBs are composed of an outer layer of myoepithelial cells surrounding a mass of luminal epithelial cells. This TEB is encased by a sheet of basement membrane, which is in turn surrounded by the fatty stroma, consisting of adipocytes, fibroblasts, immune cells and ECM components. The cap cells (green) at the tip of the TEB are putative progenitor stem cells which can give rise to myoepithelial (blue) or luminal (red) cells.
20
Key Stages of Post-natal Mammary Gland Morphogenesis
Ductal branching morphogenesis
In the mouse, mammary gland morphogenesis commences before birth when the mammary anlage develops during the embryonic and foetal periods (Hovey et al. 2002). Following birth, the anlage of mammary epithelium consists of a small, branched ductal tree connected to the nipple. These ducts then undergo isometric growth until the onset of puberty when secretion of ovarian hormones induces allometric growth. This is when the TEB structures appear at the end of the epithelial ducts, and begin the process of ductal elongation and bifurcation in order to reach the periphery of the fat pad (Richert et al. 2000). As depicted in Figure 1.2, the TEB is a unique and specialised structure with rapidly dividing cells arranged into a single outer layer of undifferentiated cap cells, and multiple inner layers of luminal epithelial cells (Hovey et al. 2002). The highly proliferative luminal cells also undergo an extensive amount of concurrent apoptosis, which is important for the formation of hollow ducts enabling milk delivery during lactation (Humphreys et al. 1996; Humphreys 1999). The mammary gland branches by two processes: bifurcation of the TEBs directly into adipose tissue, and sprouting of side branches from mature ducts through basement membrane (Wiseman and Werb 2002). Each side branch terminates in a small alveolar bud. As branching proceeds the mammary fat pad is eventually filled with alveolar buds attached to the sides of ducts. Despite this, ducts never come into very close contact with each other, suggesting the diffusion of local inhibitory signals from adjacent ducts (Hovey et al. 2002). Once the elongated ducts have reached the edge of the fat pad, growth ceases, the cap cell layer vanishes, and the TEBs disappear (Brisken 2002).
Due to the limited availability of human breast tissue, the vast majority of our knowledge of mammary development has evolved from mouse models. The
21
morphology of the human breast differs substantially from the mouse mammary gland. Whereas humans have the one pair of nipples and glands, most mouse strains have five pairs. The human mammary gland has a more complex segmented structure starting with the nipple that contains the outlet for 5 to 10 lactiferous ducts (Cardiff and Wellings 1999). In contrast, the mouse mammary gland generally terminates in a single lactiferous duct that forms 5 to 10 secondary ducts that grow in a more linear pattern than in humans (Cardiff and Wellings 1999). The human counterpart of the mouse side branch and alveolar bud is the terminal ductal lobular unit (TDLU), and is a small unit resembling a cluster of grapes at the end of a stem, encased in a loose intra-lobular connective tissue surrounded by denser stroma (Cardiff and Wellings 1999; Brisken 2002). However despite these morphological differences, there are also many similarities between the human breast and the mouse mammary gland. In the human breast, ductal elongation and side branching also occurs during puberty, and these ducts are also lined by a bilayer of myoepithelial and luminal epithelial cells (Monaghan et al. 1990). Both the alveolar buds and TDLUs have similar compositions of epithelial cells, which are hormone responsive throughout the murine estrous or human menstrual cycle, respectively, and are the source of milk production (Cardiff and Wellings 1999). Due to these parallels, the mouse is thus an ideal model for studying mammary morphogenesis and cancer.
Alveolar morphogenesis
Alveolar morphogenesis or the development of the mouse mammary epithelium during pregnancy results in the formation of alveolar structures primed for milk secretion during lactation. This involves rapid and global proliferation of the epithelial cells within the ductal branches and developing alveoli, which increases both epithelial cell number and surface area, actions essential for sufficient milk production during lactation. This initial proliferative phase is instigated by an increase in the level of serum prolactin (Prl) and progesterone (Pg) (Brisken 2002). These hormones activate the ‘alveolar
22
switch’, a genetic program that coordinates changes in epithelial cell proliferation, migration, differentiation and deletion within the many tissue types of the mammary gland.
Following this massive proliferation phase, cell differentiation becomes dominant from mid-pregnancy as the gland begins to move into the secretory initiation phase (Richert et al. 2000). The developing alveoli cleave and polarise to form a spherical layer of epithelial cells surrounding a hollow lumen, connected to the ductal network via a single small duct. Each individual alveolus is surrounded by a basket-like architecture of contractile myoepithelial cells. These contractile cells are responsible for the movement of milk out of the alveoli and through the ducts during lactation (Richert et al. 2000). The myoepithelium of the alveoli is discontinuous so that the luminal cells can directly contact the underlying basement membrane, which forms part of the ECM. Some cells of the ductal network also contact the basement membrane. This contact is required in order for complete lobuloalveolar differentiation to occur (Streuli et al. 1995; Fata et al. 2004), seen morphologically by the appearance of lipid droplets (Neville et al. 2002) and by the initiation of gene expression in a defined order (Rudolph et al. 2003). Nearing parturition, alveolar tight junctions close, and milk and colostrum proteins move into the alveolar lumen. This occurs in preparation for active milk secretion post partum, which marks the onset of the secretory activation phase (Nguyen et al. 2001).
The epithelial expansion is paralleled by equally dramatic changes in other tissue compartments. Adipocytes lose their lipid content and remain as long projections scattered throughout the alveolar epithelium (Neville et al. 1998). A huge expansion of the vasculature also occurs within the stroma, to provide the large quantities of energy, sugars, amino acids and solutes required for milk production (Djonov et al. 2001). Developmental events are also elicited elsewhere in the animal, for example the gut and liver enlarge dramatically to cope with the energy needs of gestation and lactation. The brain is
23
programmed for correct maternal behaviour by Prl (Lucas et al. 1998). Thus the alveolar switch is part of a larger mechanism controlling all aspects of adaptation to pregnancy and lactation.
Involution
Following lactation, further tissue remodelling occurs during involution when the milk-secreting epithelial cells undergo cell death once they are no longer required. It is a two-step process whereby the first phase is marked by cells undergoing apoptosis and shedding into the lumen of the alveoli, followed by the second phase, which is marked by remodelling of the surrounding stroma and replacement of the secretory epithelium by adipocytes (Watson 2006). During the second phase of involution, further apoptosis is induced by degradation of the basement membrane and ECM by matrix metalloproteases (MMPs) (Pullan et al. 1996; Green and Lund 2005). Following this, the involuted mammary gland returns to a predominantly quiescent state that resembles the pre-pregnant state, only to redevelop with the next pregnancy. A schematic diagram representing the different developmental stages of the mammary gland is depicted in Figure 1.3 over the page.
24
Figure 1.3 Post-natal organogenesis of the mammary gland. The pink ovals depict the mammary fat pad (stroma) and the solid black circle represents the nipple from which the ducts originate. The growing ducts end in terminal end buds (TEBs) during puberty. In the mature virgin, the entire fat pad is filled with an extensive ductal system with side branches. The onset of pregnancy induces increased cell proliferation and the formation of alveolar buds which grow and differentiate into milk-secreting alveoli at the end of pregnancy. During lactation, the luminal cells of fully mature alveoli synthesise and secrete milk components into the lumen. After weaning the mammary gland undergoes apoptosis (involution), returning the gland to a pre-pregnant state.
Hormonal Regulation of Mammary Development
The initial and very restricted development of the mammary gland in the embryo is independent of hormonal regulation, demonstrated by a lack of any obvious embryonic mammary phenotype in either of the knockout mouse models of the estrogen receptor (ER), prolactin receptor (Prlr), growth hormone receptor (GHR) or progesterone receptor (Pgr) (Hennighausen and Robinson 2001; Hovey et al. 2002; Hens and Wysolmerski 2005). The anlage
25
is, however, sensitive to maternal hormonal stimulation as lactation is observed in some newborns, both male and female. These secretions are termed “witches’ milk” following English medieval folklore where they were said to be stolen by witches to feed their familiar spirits, or as an extension of the ability of witches to milk inanimate objects. The interesting paradox raised by this observation is that these hormones do not elicit morphological development (as they will during puberty) but can initiate lactation from the anlage.
Embryonic mammary development requires the expression of parathyroid hormone-related protein (PTHrP) in the embryonic mammary bud, and of its receptor (PTH1R) in the mammary mesenchyme, which are essential in order for the mesenchyme to acquire a specialised mammary fate (Wysolmerski et al. 1998; Foley et al. 2001b). In the absence of PTHrP signalling, mammary epithelial cells (MECs) revert to an epidermal fate and no mammary ducts or nipple are formed (Foley et al. 2001b). PTHrP is also critical in maintaining sexual dimorphism, as it induces androgen receptor (AR) expression within the mammary mesenchyme (Dunbar and Wysolmerski 1999). Thus in males, the expression of androgens from the fetal testes bind to the AR in the mammary mesenchyme, resulting in the destruction of the nipple and mammary bud (Dunbar and Wysolmerski 1999). In addition to PTHrP, signalling pathways downstream of wingless-type MMTV integration site family (Wnt) and fibroblast growth factor 10 (FGF10) are also necessary for mammary gland development in the mouse embryo (Mailleux et al. 2002; Chu et al. 2004). In contrast to embryonic development, post-natal mammary development is tightly controlled by hormonal signals elicited by the ovaries and pituitary gland, which will now be discussed in detail.
Regulation of ductal morphogenesis
Ductal elongation and TEB growth is dependent upon estrogen (E) and growth hormone (GH) (Daniel et al. 1987; Silberstein et al. 1994; Kleinberg 1997).
26
This development is triggered by an increase in ovarian production of E, acting through the estrogen receptor alpha (ER ). It also requires secretion of GH by the pituitary gland, which acts through the GHR in the stroma to induce production of insulin-like growth factor 1 (IGF-1), which, in turn, stimulates mammary development in a paracrine manner (Ruan et al. 1992; Kleinberg et al. 2000). Indeed, ER -/-, GHR-/- and IGF-1-/- mouse models display impaired ductal elongation and reduced numbers of TEBs (Ruan and Kleinberg 1999; Curtis Hewitt et al. 2000; Gallego et al. 2001). IGF-1-/- mice fail to develop TEBs and ducts when treated with E and GH, and in addition, long-term treatment of IGF-1-/- mice with IGF-1 and E increases TEB formation and ductal elongation, demonstrating the synergy between these hormones and growth factors that is required during ductal morphogenesis (Ruan and Kleinberg 1999).
Tertiary side branches form in response to stimulation by Pg, further filling out the mature ductal tree (Atwood et al. 2000). Pg mediates its effects on the mammary gland by binding to the Pgr, of which there are two isoforms, PgrA and PgrB, which are expressed from a single gene. Loss of both Pgr isoforms in mice blocks tertiary side branching in the mammary gland following puberty (Lydon et al. 1995), and it was subsequently shown that loss of PgrB was necessary and sufficient to mediate these effects (Mulac-Jericevic et al. 2003). Not every cell in the mammary epithelium expresses Pgr, but those that do are segregated from, but located adjacent to proliferating cells (Silberstein et al. 1996; Seagroves et al. 2000). Thus, in the normal mammary gland, Pgr- positive cells do not appear to proliferate, but instead may be responsible for the expression and/or secretion of locally acting growth factors that stimulate proliferation of adjacent Pgr-negative cells (Brisken et al. 1998; Seagroves et al. 2000). Signalling by Wnt family proteins are important in mediating the paracrine effects of Pg that result in epithelial cell proliferation and side- branching of the adult mammary gland. This was demonstrated by Brisken et al. in 2000, when Pg was shown to directly induce Wnt4 expression, and that
27
Wnt4 was also required for side-branching (Brisken et al. 2000). Pg may also indirectly influence normal primary and secondary branching by synergising with IGF-1 action (Ruan et al. 2005).
Influences of the mammary stromal environment are crucial during ductal development (reviewed in Silberstein 2001). Indeed, the extent of ductal side branching differs between stroma of different mouse strains. This has been demonstrated by mammary stromal-epithelial recombination, whereby epithelium from a strain such as C57BL/6 which shows no side branching, results in a highly side branched morphology when recombined with the stroma of the highly side branched 129 strain (Naylor and Ormandy 2002). The important influence of the stroma is further demonstrated by epidermal growth factor receptor (EGFR) expression being required in the stroma, and not the epithelium, in order for normal mammary ductal growth and branching morphogenesis to occur (Wiesen et al. 1999). Although the stromal background is strongly stimulatory, neighbouring ductal branches avoid one another. This maintenance of proper ductal spacing is due to local inhibition of both ductal elongation and lateral branching elicited by transforming growth factor -1 (TGF 1) (Silberstein et al. 1990; Pierce et al. 1993).
A number of knockout and transgenic mouse models have demonstrated a role for other molecules, both epithelial and stromal, in mediating ductal side- branching of the mammary epithelium. These include amphiregulin (AREG) (Luetteke et al. 1999), the disintegrin and metalloproteinase, ADAM17 (Sternlicht et al. 2005), C/EBP (Grimm and Rosen 2003), activin/inhibin (Robinson and Hennighausen 1997), DDR1 (Vogel et al. 2001), and the matrix metalloproteases, MMP2 and MMP3 (Wiseman et al. 2003). Once the complex network of ducts has been formed and the TEBs have reached the periphery of the mammary fat pad, the TEBs of the mature adult virgin regress and differentiate into quiescent alveolar buds, ready for a second wave of hormonal stimulation upon pregnancy.
28
Regulation of alveolar morphogenesis
The formation of the milk secreting structures during pregnancy is dependent on a synergy between Prl and Pg signalling (Neville et al. 2002). These hormones trigger an initial wave of cell proliferation during early pregnancy (Traurig 1967). The Pgr knockout mouse demonstrated that Pg is required for alveolar morphogenesis, and epithelial recombination experiments demonstrated that Pgr in the mammary epithelium, not the stroma, was essential for epithelial cell proliferation (Brisken et al. 1998). Mammary gland chimeras made from wildtype and Pgr-/- MECs, demonstrated that Pgr-/- epithelial cells proliferate in response to Pg (Brisken 2002). As not all mammary epithelial cells express Pgr, and are unable to respond to Pg directly, they must therefore respond to a paracrine factor from Pgr+/+ cells. Indeed, in the epithelium, proliferating cells segregate with Pgr positive cells (Conneely et al. 2003). This is also true for ER positive cells (Clarke et al. 1997). Further, steroid receptor positive cells are in close proximity to proliferating cells, indicating that proliferation is mediated, at least in part, by a paracrine mechanism. This heterogeneous receptor patterning observed in the luminal epithelium is required for complete lobuloalveolar development (Grimm et al. 2002).
As described in the previous section, Pgr is essential in ductal branching morphogenesis, and this role continues in the extensive branching that continues throughout pregnancy. Wnt4 continues to act in a paracrine fashion to stimulate epithelial ductal side branching during early pregnancy (Brisken et al. 2000). Although Wnt4-/- mammary epithelial transplants displayed impaired side branching in these experiments, normal lobuloalveolar proliferation was observed during the later half of pregnancy, indicating that other factors mediating proliferation in late pregnancy are likely to be involved.
In addition to Wnt4, Receptor activator of NF- B-ligand (RankL) is also a target of the Pgr signalling pathway and may be another paracrine factor responsible
29
for cellular proliferation in steroid receptor negative cells. The RankL target, Nuclear Factor-kappa B (NF- B), is required for Cyclin D1 (Ccnd1) activation via the kinase IkB (IKK ) in neighbouring proliferating cells (Cao et al. 2001). Germ-line deletion of both RankL and its receptor (Rank) in mice resulted in failed alveolar morphogenesis due to reduced proliferation and increased apoptosis of alveolar epithelial cells (Fata et al. 2000). These effects were mediated by PI3K/Akt, demonstrating that this pathway is also essential for the formation of lobuloalveolar structures (Fata et al. 2000). The RankL/NF- B/Ccnd1 pathway is now known to be crucial for the formation of alveolar structures during pregnancy (Cao and Karin 2003), and NF- B is essential for Pg driven proliferation within alveoli (Conneely et al. 2003). RankL also co- localises with Pgrs in response to pregnancy levels of E and Pg, indicating this is an important part of the response. In primary MEC cultures, Pg acts in synergy with E to increase Ccnd1 transcription resulting in increased proliferation (Said et al. 1997). Together these data indicate that Pg may drive the proliferation of neighbouring cells via RankL/NF- B resulting in Ccnd1 transcription. In addition to the impaired ductal proliferation displayed in the PgrB-/- mouse model, alveoli also failed to develop during pregnancy, possibly an effect mediated via activation of RankL (Mulac-Jericevic et al. 2003).
In mice, pituitary Prl stimulation of ovarian Pg assists in maintaining the required levels of Pg during early pregnancy (Binart et al. 2000). In addition, up-regulation of Pgr expression by Prl, and Prlr expression by Pg, suggests that these hormones may interact in a synergistic manner to control alveolar development. Prlr knockout mice (Prlr-/-) have demonstrated the importance of this receptor during mammary development (Ormandy et al. 1997a). Similarly to Pgr, experiments with Prlr-/- mice have shown that the presence of Prlr in the epithelial cells, not the stroma, is essential for normal lobuloalveolar differentiation (Brisken et al. 1999). Prlr-/- mammary transplants fail to develop lobuloalveoli and produce milk proteins during pregnancy, illustrating that Prlr is essential in the mammary epithelium during alveolar morphogenesis.
30
Other hormones and growth factors can influence alveolar morphogenesis. The neuronal peptide galanin (Gal) regulates Prl secretion from the pituitary lactotrophs (Wynick et al. 1993). In addition, the mammary epithelium is responsive to Gal, as it augments alveolar morphogenesis in mammary explants in the presence of Prl (Naylor et al. 2003). Gal-/- mice show increased levels of the inhibitory phosphorylated form of Prl (Naylor et al. 2005), and are unable to nurse pups due to failed secretory activation (Wynick et al. 1998). Therefore Gal has dual actions: firstly an indirect role by modulating pituitary Prl and phosphorylated Prl release; and secondly a direct cell autonomous role in the formation of lobuloalveoli during pregnancy. GH may act in combination with Prl to mediate alveolar proliferation. GH treatment restores alveolar morphogenesis but inhibits lactation in Prlr+/- mammary glands (Allan et al. 2002). Placental lactogen (PL) is released from the placenta during pregnancy and can fully compensate for Prl, allowing alveolar morphogenesis in Prl-/- mice (Horseman 1999). Some of the key hormones and their mediators which control ductal and alveolar morphogenesis are depicted in Figure 1.4. The downstream targets and molecular modulators of Prl and Pg signalling will be discussed in more detail later in this chapter.
Regulation of lactation and involution
Just prior to lactation, Pg suppresses active milk secretion. Withdrawal of Pg in the presence of high levels of Prl induces the onset of secretory activation, characterised by closure of tight junctions and the movement of milk and lipids into the alveolar lumina for transport down the ducts and out through the nipple (Neville et al. 2002). Milk ejection maintains lactation, together with Prl stimulation of the luminal epithelial cells to continue milk secretion, and oxytocin stimulation of the contractile myo-epithelial cells to continue milk ejection (Neville et al. 2002). After termination of the suckling stimulus and withdrawal of Prl stimulation, the mammary gland undergoes rapid apoptosis, and together with remodelling of the ECM, the ductal framework is left remaining in readiness for a subsequent pregnancy.
31
Figure 1.4 Hormonal regulation of mammary morphogenesis. A Representative mammary wholemount images (Carmine alum stain) of (i) virgin, (ii) 12 days post coitus (dpc), (iii) 18 dpc and (iv) 1 day post partum. B Schematic diagrams of corresponding mammary gland development stages. C Key hormones and their mediators which regulate mammary morphogenesis. Solid arrows indicate direct interactions; broken arrows indicate indirect interactions. Wholemount images were taken from Oakes et al. 2006.
32
Mammary development and stem cells
During pregnancy, ductal branches grow and proliferate to yield differentiated and functional alveoli, which following weaning are then removed by apoptotic cell death during the involution phase, only to re-develop with the next pregnancy. The cyclical nature of this tissue remodelling is striking. This observation first led researchers to hypothesise that mammary tissue must contain persistent self-renewing mammary stem cells (reviewed in Smalley and Ashworth 2003). The ability of small epithelial transplants to recapitulate a complete and fully functional epithelial mammary gland reinforced this view (DeOme et al. 1959). The presence of a single mammary stem cell was indicated by limiting dilution experiments and the existence of committed progenitor cells was demonstrated by transplants that showed limited developmental capacity (Smith and Boulanger 2003). This cell was recently isolated and elegantly demonstrated to be capable of producing a renewable and complete mammary epithelium (Shackleton et al. 2006). Thus it is hypothesised, based on a paradigm developed in the hematopoietic system, that a primary mammary epithelial stem cell gives rise to a hierarchy of epithelial progenitor cell lineages to ultimately produce the different cells found in the mammary epithelium (Smalley and Clarke 2005; Stingl et al. 2005). The flux of cells through these lineages is likely to be controlled by, and in turn control, the patterns of gene expression that comprise the alveolar switch. Integrating our knowledge of gene expression patterns with the emerging knowledge regarding stem cell lineages and their interactions offers us an unprecedented opportunity to understand this phase of mammary development.
33
Signalling Pathways Involved in Alveolar Morphogenesis
Molecular modulators of prolactin and progesterone induced alveolar morphogenesis
Members of the prolactin- and progesterone-signalling pathways are essential for normal alveolar morphogenesis (Hennighausen and Robinson 2005). Prlr dimerisation occurs after Prl binding and leads to the phosphorylation of the associated Janus kinase (Jak2) (Han et al. 1997; Pezet et al. 1997), which in turn phosphorylates specific residues on the Prlr (Lebrun et al. 1995). Stat5 is then recruited to the receptor and is phosphorylated by Jak2 (DaSilva et al. 1996). Phosphorylated Stat5 is then translocated to the nucleus where it can activate transcription of multiple genes (Wakao et al. 1995) involved in a variety of processes during alveolar morphogenesis, including establishment of epithelial polarity and cell-cell interactions, stromal epithelial interactions and milk protein expression during lactation. Both isoforms of Stat5, Stat5a and Stat5b, result in lobuloalveolar defects when knocked out in mice, and this phenotype is more severe in combined Stat5a/Stat5b knockout animals (Liu et al. 1995; Liu et al. 1997; Teglund et al. 1998).
Pg is also a critical regulator of Stat5 in the mammary gland. Stat5 expression increases with E and Pg treatment in ovariectomized mice, and co-localises extensively with PgR positive cells, suggesting it may be a target of PgR (Santos et al. 2008). Stat5 is up-regulated and translocated to the nucleus with Pg treatment in human breast cancer cells (Lange et al. 1998; Richer et al. 1998), and an interaction has been reported between PgR and Stat5, either directly, or via binding on the beta-casein promoter (Richer et al. 1998; Buser et al. 2007). This interaction may be pivotal in the mechanism of how Pg antagonises the lactogenic effect of Prl observed in the mammary gland during pregnancy.
34
One class of genes activated by the Prl-signalling pathway are the suppressor of cytokine signalling (Socs) members, which act to shut down the Prl- signalling pathway. In a negative feedback loop, Prl induces expression of Socs1 and Socs2, which attenuates Stat5 activation by direct binding of Jak2 and preventing its activation of Stat5 (Hanada et al. 2001; Harris et al. 2006). Socs1 knockout mice show precocious development during pregnancy, and heterozygous Socs1+/- mice can restore the lobuloalveolar defects present in Prlr+/- mice (Lindeman et al. 2001). Socs2-/- mice show no developmental defects and complete loss of Socs2 is required to rescue lactation in Prlr+/- females (Harris et al. 2006). Socs2 shows the most robust transcriptional response to modulation of prolactin (Harris et al. 2006), with little change in Socs1 detectable.
Transcript profiling of prolactin receptor knockout mammary glands identified a panel of genes that require Prlr-mediated signalling for increased expression during early pregnancy (Ormandy et al. 2003; Harris et al. 2004). Laminin and two members of the collagen family were identified. These molecules are cell adhesion components of the ECM and play an important part in the epithelial- stromal signalling required for full lobuloalveolar differentiation and gene expression (Streuli et al. 1995; Rudolph et al. 2003). Alveolar morphogenesis induced by Prl involves the establishment of polarity and cell-cell communication. The maintenance of cellular polarity is regulated by the closure of tight junctions, and the expression of tight junction proteins Claudin- 3 and Claudin-7 was reduced in Prlr-/- mammary transplants. The gap junction protein Connexin-26 was also identified and is involved in the exchange of small ions and metabolites. Recently, Connexin-26 was shown to be important in full lobuloalveolar development and in the prevention of alveolar cellular apoptosis (Bry et al. 2004).
As discussed earlier, Wnt4 and RankL are targets of the Pgr signalling pathway. Wnt4 was also down regulated in Prlr-/- transplants, indicating that it is potentially a target of Prlr signalling also. The downstream target of Wnt, -
35
catenin, has specific actions in both the luminal and myoepithelial compartments of the epithelium, and as a component of cell-cell junctions, appears to have a role in signalling to luminal epithelial cells (Imbert et al. 2001; Teulière et al. 2005). Indeed, activation of -catenin within the basal epithelial cells results in premature differentiation of the luminal epithelium during pregnancy and persistent proliferation resulting in tumors. These tumors consisted predominantly of undifferentiated basal cells, which were amplified in response to -catenin activation, thereby implicating this molecule in cell fate decisions in the mammary gland (Teulière et al. 2005).
RankL was also identified as a potential Prl-regulated gene (Ormandy et al. 2003; Srivastava et al. 2003). Ccnd1 null mutants exhibit significantly delayed alveolar cell proliferation and impaired lactation, which was shown to be epithelial cell autonomous (Fantl et al. 1999). Interestingly, Prl can induce Ccnd1 expression via induction of Igf2, independent of RankL induction (Brisken et al. 2002). The similarities between Prl and Pg mediated effects on both RankL and Wnt signalling is further evidence of the co-operation of these pathways for alveolar cell proliferation during early pregnancy.
Gene expression profiling of Prl-/- mice has also identified unique targets of mammary development. Expression of tryptophan hydroxylase, the rate- limiting enzyme in serotonin biosynthesis, is increased by prolactin during pregnancy and lactation. Accumulation of serotonin due to milk engorgement experienced during weaning or experimentally via teat sealing, inhibits milk gene expression and can induce involution, providing a mechanism that is put in place by prolactin to stop lactation at weaning (Matsuda et al. 2004).
The MAPK pathway
In addition to activating the Jak2/Stat5 signalling cascade, binding of Prl to the Prlr also stimulates the MAPK cascade. Dimerisation of ligand-bound Prlr activates Jak2, which generates docking sites for Src homology domain 2
36
(SH2)-containing molecules, like SHC, and stimulates the Grb2/SOS-Ras-Raf- MAPK cascade (Erwin et al. 1995). Upon Prl treatment of MECs, Raf is rapidly activated, quickly followed by activation and phosphorylation of MEK and MAPK, mediating the Prl-induced proliferation that is observed in MECs (Das and Vonderhaar 1996). Although Jak2 is crucial for activation of Stat5 signalling, it is dispensable for the Prl-induced activation of the MAPK cascade, which instead may be mediated by the tyrosine kinase, c-Src, and focal adhesion kinase (FAK) (Acosta et al. 2003; Sakamoto et al. 2007).
Progesterone and synthetic progestins also rapidly activate the Src/Ras/MAPK cascade in MECs (Skildum et al. 2005). Progestins can activate c-Src either by direct binding by Pgr (Boonyaratanakornkit et al. 2001), or via an interaction between Pgr and ER (Ballaré et al. 2003). This cross-talk with ER is essential for Pg-induced proliferation of breast cancer cells (Migliaccio et al. 1998; Vicent et al. 2008). Pg activation of the PI3K/Akt pathway has been shown to be required for Pg-induced cell proliferation in the mammary gland (Ballaré et al. 2006). Prl can also activate the PI3K/Akt pathway, which has been demonstrated to promote proliferation (Fresno Vara et al. 2001; Chakravarti et al. 2005), and survival in the mammary gland (Bailey et al. 2004).
In addition to cross-talk with other steroid hormone signals, Pg also can enhance growth factor signalling, as demonstrated by the selective potentiation of EGF-stimulated MAPK activation by pre-treatment of human breast cancer cells with Pg (Lange et al. 1998). The ultimate targets of activation of this signalling pathway are not well-defined, but likely to include transcription factors and co-factors involved in cell cycle control and proliferative responses (Ballaré et al. 2006). It has been demonstrated that Prl and Pg synergistically stimulate the proliferation of mouse MECs in vivo (Hovey et al. 2001), and although the mechanisms underlying their co-operative effect is not known, it is likely that cross-talk between the Jak2-Stat5 pathway and mitogenic stimulation of the MAPK cascade is involved. A summary of the key regulators in these signalling cascades is depicted in Figure 1.5.
37
Figure 1.5 Prolactin and progesterone signalling pathways. Binding of Prl to Prlr induces receptor dimerisation and activation of Jak2. Stat5 is then recruited and phosphorylated, and then dimerises and translocates to the nucleus to activate the transcription of genes involved in alveolar morphogenesis, such as milk protein genes. Jak2 activation can also mediate phosphorylation of Shc, followed by association of Grb2 and SOS resulting in the binding and activation of Ras and Raf and subsequently, the MAPK signalling pathway. Attenuation of Prlr signalling is mediated by members of the Socs gene family through inhibition of Jak2 tyrosine kinase activity. Alternatively, Prl binding of Prlr can directly induce activation of c-Src, which mediates the activation of the MAPK cascade, required for cell proliferation. Progesterone regulates several intracellular effectors by up-regulating and inducing nuclear localisation of Stat5, increasing the phosphorylation of Jak2 and Shc, and potentiating EGF-stimulated MAPK activity. Prl and Pg can also promote mammary proliferation and survival, through activation of the PI3K/Akt pathway.
38
Transcription factors involved in alveolar morphogenesis Prl and Pg and other factors induce the transcription of genes via activation of target transcription factors. These include Stat5 and the steroid hormone receptors as discussed previously, which bind to DNA and result in the transcription of genes involved in many aspects of alveolar morphogenesis. Further, some of these target genes are transcription factors also, which act to induce the expression of genes or groups of genes involved in lobuloalveolar development. An example includes the transcription factor Srebf1, which was identified from transcript profiling experiments on 3 mouse models of failed secretory activation (Naylor et al. 2005). Srebf1 controls the expression of a number of key lipid metabolism genes (Horton et al. 2002), that showed reduced expression concomitantly with decreased Srebf1 expression (Naylor et al. 2005). Some transcription factors which appear to be involved in alveolar morphogenesis and regulation of cellular proliferation include the Homeobox genes, Helix-loop-helix genes, Stats, Tcf/Lef family, NF- B, C/EBP family, Nuclear factor family and the Ets transcription factors (Coletta et al. 2004).
Pg and Prl are hypothesised to influence the expression of -catenin, via induction of the Wnt pathway as discussed earlier. -catenin regulates the activity of the Tcf/Lef family of transcription factors which appear to mediate - catenin signalling and therefore may play a role during alveolar morphogenesis (Hatsell et al. 2003). Inhibition of -catenin results in alveolar apoptosis and greatly reduced milk production capacity. Mice lacking Lef-1 demonstrate a failure to form the alveolar bud at embryonic day 13. The expression of Lef-1 was co-expressed with -catenin, and shows a similar expression pattern in response to PTHrP (Foley et al. 2001a). Thus Lef-1 may act to mediate the actions of -catenin, although its effects during alveolar morphogenesis are still unclear.
The nuclear factor (NF1) family of transcriptions factors also play a role in functional differentiation as they regulate the transcription of milk protein
39
genes such as those encoding whey acidic protein (WAP), -lactalbumin and - lactoglobulin (Murtagh et al. 2003). The NF1-C2 isoform member of this family induces the expression of the milk genes caboxyl ester lipase and WAP. Prl regulates the protein expression of NF1-C2 in NMuMG cells, and its expression is reduced in the nucleus of Prlr-/- luminal cells at mid-pregnancy, indicating that NF1-C2 may be regulated by Prl signalling during pregnancy and involved in expression of milk genes in preparation for lactation (Johansson et al. 2005).
The helix-loop-helix transcription factors Id1 and Id2 have varying expression in the mammary gland. Id1 expression is increased during early pregnancy, remains low during lactation and rises again at involution. Unlike Id1, Id2 remains high during lactation, indicating that these isoforms have specific functional roles during alveolar morphogenesis (Desprez et al. 2003). Id1 is specifically expressed by the expanding epithelium during the alveolar proliferative phase and is inversely correlated with the expression of -Casein, therefore it appears to be an important factor during early alveolar proliferation. Id1 also regulates Clusterin, a gene involved in the regulation of cell-cell interactions. Additionally, lobuloalveolar development is severely impaired in Id2 knockout mice. Reduced proliferation and increased apoptosis has been observed in mammary epithelium lacking Id2, resulting in the failure to form alveolar structures and consequently failure of lactation (Mori et al. 2000). Id2 also promotes differentiation in MEC cultures, indicating Id2 is essential for the differentiation of the mammary epithelium (Desprez et al. 2003)
The transcription factor NF-kB discussed earlier in this review is essential for Pg induced alveolar cell proliferation resulting in Cyclin D1 transcription (Cao and Karin 2003; Conneely et al. 2003). NF- B can also induce the transcription many genes involved in the regulation of apoptosis. NF- B levels are induced during pregnancy, decline during lactation and are re-induced during lactation
40
implying a role in mammary gland remodelling. It is also hypothesised that NF- B is an essential “checkpoint” of apoptosis, whose actions are dependent on association with specific transcriptional regulators. Thus NF- B is an important transcription factor controlling both proliferation and apoptosis in the epithelium during pregnancy (Clarkson and Watson 1999).
The C/EBP family of proteins appear to be important regulators of alveolar morphogenesis. (For review see Grimm and Rosen 2003). C/EBP and C/EBP isoforms are increased during pregnancy and decline during lactation, indicating that they play a critical role in alveolar morphogenesis, and early milk gene expression. Transplantation experiments have revealed that C/ebp is required in epithelial cells for normal lobuloalveolar development during pregnancy, and C/EBP knockout mice display phenotypes similar to Pgr, Prlr, Stat5a/b, Ccnd1, Id2 and RankL knockouts (Grimm and Rosen 2003). Interestingly Pgr expression was dramatically increased in the mammary glands of C/EBP null mice, and in addition, the expression of Pgr was unusually uniform within the epithelium (Seagroves et al. 2000). These effects were associated with a 10-fold decrease in the rate of proliferation. There was however no change in the expression of C/EBP in the mammary glands of Pgr knockout mice, indicating that C/EBP is upstream of Pgr, and possibly controls the spatial distribution of epithelial cells, which influence proliferation in alveolar progenitors (Seagroves et al. 2000). C/EBP null epithelium significantly increased TGF and Smad2 signalling, and this pathway is known to inhibit cellular proliferation (Barcellos-Hoff and Ewan 2000). Cell cycle progression in C/EBP null MECs was blocked at the G1/S transition, preventing these cells from proliferating in response to early pregnancy levels progesterone and estrogen (Grimm et al. 2005). Therefore, C/EBP is essential for controlling cell fate decisions within the mammary gland, including attenuating Pgr expression resulting in mammary epithelial cell differentiation during pregnancy.
41
The expression of the Ets transcription factor sub family Pea3, are elevated at the onset of pregnancy but decline during mid pregnancy to low levels at lactation and involution suggesting a role in early pregnancy induced ductal outgrowth. There are 3 members of the Pea3 subfamily, which are expressed by both the myoepithelium and the luminal cells, however their expression varies during pregnancy suggesting multiple signalling roles during alveolar morphogenesis. The expression of all members of the family remains in the myoepithelium during pregnancy, however the expression of the ER81 member declines in the luminal epithelium 7 days after impregnation. Increased numbers of dividing cells was observed in the terminal end buds of Pea3 knockout mice, and mammary gland transplants of Pea3 knockout epithelium displayed reduced mammary branching during pregnancy suggesting a role for Pea3 in progenitor cell differentiation (Kurpios et al. 2003).
Other factors involved in alveolar morphogenesis
The receptor tyrosine kinase, ErbB (epidermal growth factor) family and their ligands, are important mediators of all aspects of mammary development. There are four receptors: EGFR/ErbB/Her1, ErbB2/Her2/neu, ErbB3/Her3 and ErbB4/Her4, which are activated by a variety of ligands inducing activation via dimerisation and cross phosphorylation. ErbB ligands share a 50 amino acid domain, which is homologous to EGF. Mice expressing a truncated dominant negative allele of ErbB2 did not exhibit a phenotype until late pregnancy where alveoli failed to expand and distend, indicating that ErbB2 is critical for secretory activation. Conditional deletion of ErbB4 within the mammary gland at pregnancy demonstrated a critical role for this receptor during alveolar morphogenesis (Long et al. 2003). Alveolar expansion was reduced from 13.5 days post coitus in mammary epithelium lacking ErbB4 resulting in incomplete alveolar development and failure to nurse pups due to reduced milk gene expression. Alveolar proliferation was attenuated and Stat5 phosphorylation was abolished. The ErbB4 ligand Neuregulin/Heregulin-1 (Nrg) promotes lobuloalveolar development and the expression of milk genes when used in
42
mammary gland explants (Yang et al. 1995), indicating a role for this ligand in lobuloalveolar development. In addition, mice that lack the alpha form of Nrg show a similar phenotype to ErbB4 knockout with reduced alveolar proliferation and differentiation, demonstrated by reduced -casein expression in reduced alveoli expansion (Li et al. 2002).
Other ErbB ligands also appear to have overlapping functions for mammary gland development. Amphiregulin (AREG) null animals have reduced alveolar development, however the phenotype was much more severe in a triple mutant including knockouts of TGF and EGF (all ligands of the ErbB family), indicating overlapping and compensatory roles for these ligands during alveolar morphogenesis (Luetteke et al. 1999). Triple mutants developed poorly organised and differentiated alveoli, had reduced milk protein expression and often pups born to these mice did not survive. AREG loss was also associated with reduced Stat5 phosphorylation. Our transcript profiling experiments demonstrated that AREG was down-regulated in Prlr-/- epithelium (Ormandy et al. 2003), indicating that AREG may be modulated by Prlr signalling. These data together indicate important roles for the ErbB receptors and ligands during alveolar morphogenesis. The overlapping phenotypes observed in Prlr, Pgr and ErbB knockout mice suggest there may be some cross talk between these receptors, which is yet to be fully understood.
The cell surface receptors, 1-integrins, which are present on luminal epithelial cells, are essential mediators of ECM signalling via its ligands collagen and laminin (Zutter et al. 1998). Mammary epithelium in mice lacking 1-integrin in the luminal cells, displayed reduced proliferation and alveolar disorganisation (Li et al. 2005). The focal adhesion kinase, which is important in protein complexes that connect the ECM to the actin cytoskeleton was also reduced in these mice. Conditional deletion of 1 integrin during early pregnancy and late pregnancy demonstrates that this molecule was important for both the formation of lobuloalveolar structures and for functional
43
differentiation (Zutter et al. 1998). In these mammary glands, luminal epithelium becomes dissociated from the basement membrane, and cellular polarity is compromised as luminal epithelial cells protrude into the alveolar luminal space. In addition, Prl stimulated milk protein expression via phosphorylation of Stat5 was largely absent in primary mammary epithelial cells lacking 1-integrin, indicating that it is essential for Prl-induced activation of Stat5 (Naylor and Streuli 2006).
The cytokine transforming growth factor- 1 (TGF 1) is an important regulator of mammary cell proliferation during pregnancy (Barcellos-Hoff and Ewan 2000). TGF 1 is restricted to the luminal epithelial cells and can control cell proliferation via phosphorylation of Smad following Tgf- receptor activation (Massagué and Chen 2000). TGF 1 heterozygote mice display accelerated lobuloalveolar development due to increased proliferation, indicating that the expression of TGF 1 restricts alveolar cell proliferation. Epithelial cell proliferation was increased more than 15-fold in TGF 1 null ovariectomised animals treated with E and Pg compared to wildtype mice (Ewan et al. 2002). In animals treated with E and Pg, TGF 1 expression was restricted to the steroid receptor positive epithelial cells, indicating that TGF 1 may play an important role in restricting epithelial cell proliferation in these cells (Ewan et al. 2005). Some of the major proteins and pathways involved in the molecular control of alveolar morphogenesis are depicted in Figure 1.6.
44
Figure 1.6 Molecular control of alveolar morphogenesis. Signalling from the progesterone receptor (Pgr) and prolactin receptor (Prlr) is essential for alveolar morphogenesis in pregnancy. Increases in serum progesterone (Pg) and prolactin (Prl) result in luminal proliferation during early pregnancy, which continues throughout gestation. Wnt4 and RankL are transcribed in response to Pgr signalling in co-operation with Prl signalling, and appear to stimulate proliferation of neighbouring cells via paracrine mechanisms. RankL binds to its receptor Rank in a neighbouring cell and activates the RankL/NF- B pathway, resulting in cyclin-D1 (Ccnd1) transcription and proliferation. Wnt4 binds and activates its target -catenin, which has specific roles for both luminal and myo-epithelium for cell fate decisions involving proliferation and differentiation. Prl binds to Prlr and activates Jak2/Stat5 cascade, resulting in transcription of various genes, including milk proteins essential in lactation and proteins that regulate their own pathway (Socs proteins). Laminin in the extracellular matrix binds to 1-integrin when contact between the basement membrane and the luminal epithelium is established, and is essential for the maintenance of alveolar cell polarity and differentiation. ErbB receptors and their ligands complement Prlr signalling as activation of ErbB4 results in Stat5 phosphorylation and translocation to the nucleus. (GJ = gap junction; TJ = tight junction; L = lipid droplet).
45
The Ets Transcription Factor Family and Elf5
The E26 transformation-specific (Ets) family is one of the largest families of transcription factors, and is unique to metazoans. Family members are identified by a highly conserved DNA binding domain (the Ets domain), which is a winged helix-turn-helix structure, and binds to sites containing a central GGA motif (Sharrocks et al. 1997). Many Ets transcription factors contain a pointed domain, which serves as a protein-protein interaction motif, most contain a transcription activation domain, and some possess a repressor domain (Sharrocks 2001). The Ets domain has been highly conserved in both sequence and function between species, and it is the regions outside of this domain that vary more widely between family members, and which are likely to contribute to functional differences between Ets proteins. For example, most Ets proteins transactivate target promoters, while some are transrepressors. Overall, the Ets family is divided into 11 subfamilies based on their structure, which are listed in Table 1.1 on the following page.
Ets functional activity is modulated at multiple levels. It is dependent on specific protein interactions, and signalling pathways, such as the MAP kinases (ERK, p38 and JNK), converging on the Ets transcription factors to regulate their DNA-binding activity, subcellular localisation, transactivation potential and specification of downstream target genes (Seth and Watson 2005). The end result is that with over 400 genes identified as potential targets of the 27 different human Ets genes (Seth and Watson 2005), this family of transcription factors plays a critical role in the regulation of normal biological processes. In particular, these include differentiation, proliferation, migration, hormone responses and tumorigenesis of multiple tissues (reviewed in (Oikawa and Yamada 2003; Hsu et al. 2004; Gutierrez-Hartmann et al. 2007)).
46
Table 1.1 The Ets family of transcription factors. The subfamilies, their respective members and a schematic of the arrangement of their functional domains are depicted. (AD – transcriptional activation domain; ETS – DNA binding domain; RD – transcriptional repressor domain). Adapted from (Gutierrez-Hartmann et al. 2007).
Subfamily Members Structure ETS Ets-1, Ets-2, Pointed
ERG Erg, Fli1, Fev
ELG Gabp , Elg
ELF Elf1, Nerf/Elf2, Mef/Elf4
ESE ESE-1/ESX/Elf3, ESE-2/Elf5, ESE-3/EHF ERF ERF/PE2, ETV-3/PE1
TEL Tel/ETV-6, Tel-2/ETV-7
PEA3 PEA3/E1AF/ETV-4, ERM/ETV- 5, ER81/ETV-1, ER71/Etv-2 SPI PU.1/SP1, SPIB, SPIC
TCF Elk-1, SAP1/Elk-4, NET/SAP2/Elk-3, LIN PDEF PDEF/SPDEF/PSE
47
Our lab became interested in Ets transcription factors after identifying Elf5 (e74-like factor 5 or ESE-2) as showing reduced expression in response to a loss of Prlr, in addition to up-regulation in a cell based model of positive prolactin action (Harris et al. 2006). Elf5 is an epithelial specific member of the Elf subfamily of Ets transcription factors, and is closely related to the epithelial specific Elf3 (ESE-1) and Ehf (ESE-3) (see Table 1.1; (Zhou et al. 1998; Oettgen et al. 1999). The predicted protein products of mouse Elf5 and human ESE-2 are 95% identical and are expressed as 2 isoforms produced by alternative start sites. Such high conservation of sequence implies similar conservation of function (Zhou et al. 1998).
Elf5 is expressed specifically in the luminal cells of mammary tissue (Harris et al. 2006), and its expression is increased dramatically during pregnancy, to levels that far exceed those seen in other tissues. Elf5 can also bind to an Ets like domain in the proximal promoter of Whey acidic protein (Wap) and induce its expression independently of lactogenic hormones, indicating that Elf5 may be an important mediator of alveolar differentiation during mid-pregnancy (Thomas et al. 2000). Elf5 knockout (Elf5-/-) mice die in utero due to a placentation defect (Donnison et al. 2005). Elf5 heterozygous (Elf5+/-) mice did not lactate due to failed alveolar development, and in some mice where alveoli had formed, differentiation into functional secretory units was severely impaired (Zhou et al. 2005). Mammary epithelial cell proliferation was reduced throughout alveolar morphogenesis and secretory activation, and mammary epithelial transplants demonstrated that this effect was cell autonomous. The level of Elf5 expression is reduced in Prlr+/- glands, however there is no similar reduction in the expression of Prlr in Elf5+/- glands, indicating that Elf5 is downstream of the Prlr (Zhou et al. 2005). MECs from Prlr-/- mammary glands fail to form lobuloalveoli during pregnancy when transplanted into the cleared fat pad of hosts with a normal endocrine milieu. Retroviral re-expression of Elf5 in Prlr-/- MECs followed by transplantation to a cleared fat pad resulted in a rescue of alveolar morphogenesis. MECs expressing high levels of Elf5 proliferated and differentiated into distended, milk filled alveoli (Harris et al.
48
2006). Thus re-expression of Elf5 in Prlr-/- MECs could completely compensate for the loss of the Prlr signalling cascade. Prlr-/- MECs expressing lower levels of Elf5 showed development that passed alveolar formation but failed during secretory initiation, mimicking the situation seen in Elf5+/- and Prlr+/- mice. Elf5 is a key mediator of structural and functional development of lobuloalveoli (Harris et al. 2006). Elf5 would thus appear to be a master-regulator of the alveolar switch required for alveolar morphogenesis.
So far, the mechanism of Elf5 action remains largely unknown. Elf5 is a critical regulator of milk protein synthesis, as recently demonstrated where the knockout of Elf5 in the epithelium results in almost no expression of milk proteins during pregnancy, and conversely, overexpression of Elf5 in an inducible transgenic model caused precocious development and milk secretion in virgin mice (Oakes et al. 2008 – See Appendix II), however how this occurs remains to be determined. In the murine mammary epithelial cell line, HC11, Stat5 has been shown to modulate Elf5 transcription (Oakes 2006), and Elf5 appeared to be downstream of Stat5. The Elf5 promoter has several Stat5 binding sites, or ‘GAS elements’, and it likely that Stat5 dimers translocate to the nucleus and bind to the GAS element of the Elf5 promoter to induce its transcription, although this has not yet been demonstrated explicitly (Hennighausen and Robinson 2008). The molecular mechanisms downstream of the Prlr which lead to Elf5 expression, and subsequently, those downstream of Elf5 which lead to milk protein synthesis requires further investigation. One of the specific aims of this thesis is to further understand the role and mechanism of Elf5 action in human normal mammary and breast cancer cell lines.
49
Prolactin and Progesterone in Mammary Carcinogenesis
Prolactin has often been implicated in the initiation and/or progression of breast cancer (Clevenger et al. 2003). For example in human studies, it has been shown that high serum Prl levels positively correlate with a 30% increased risk of breast cancer in women, and an 80% increased risk if the tumors were ER/Pgr positive (Hankinson et al. 1999; Tworoger et al. 2004; Tworoger et al. 2007). In a separate study, elevated serum Prl levels correlated with tumor size and stage, and risk of developing metastatic disease (Bhatavdekar et al. 2000). Studies carried out in rodent models have shown that raised serum prolactin is a complete carcinogen (Welsch and Nagasawa 1977). More recent approaches include the generation of mice which overexpress Prl within MECs under the control of a hormonally non-responsive promoter, neu-related lipocalin, and the demonstration that the majority of these transgenic mice developed mammary carcinomas, compared to only 5% of non-transgenic controls (Rose-Hellekant et al. 2003). Furthermore, the presence of the Prlr in the mammary epithelium is important for the early progression of pre-invasive lesions to invasive carcinoma (Oakes et al. 2007). In addition to this, in vitro studies have shown that Prl increases cell proliferation (Liby et al. 2003; Chakravarti et al. 2005), and inhibits apoptosis (Clevenger et al. 2003) in breast cancer cell lines. In contrast, local over- expression of human Prl in a differentiating murine mammary gland has been shown to induce morphological and functional defects, but not carcinoma (Manhès et al. 2006), and Prl can suppress invasion capacity of breast cancer cells (Nouhi et al. 2006). The mechanisms behind these observations are not clear, and so although Prl was at one time thought to directly induce breast cancer, the involvement of Prl in mammary carcinogenesis remains a controversial subject (Goodman and Bercovich 2008).
Produced in many tissues and acting in various mechanisms and roles, the involvement of Prl in carcinogenesis could occur by any number of ways. One
50
direct mechanism could involve the stimulation of cell proliferation and induction of cyclin D1 expression and MAPK activation by Prl, as previously described. This mitogenic action of Prl, combined with its role as a pro-survival factor and inhibitor of apoptosis (Clevenger et al. 2003; Perks et al. 2004), indicate possible methods by which Prl may contribute to carcinogenesis. Alternatively, perhaps Prl has more of an indirect role in carcinogenesis via modulation of the action of other hormones. For example, exogenous Prl modulates the expression of Pgr by human breast cancer cells (Ormandy et al. 1997b), while endogenous Prl can modulate ER levels (Gutzman et al. 2004). Prl may also act by transactivation or interaction with growth factor or receptor tyrosine kinase signalling pathways during carcinogenesis. Much attention has been paid to the role of EGF signalling in breast tumorigenesis (reviewed in (Gschwind et al. 2004)), and synergy between this pathway and Prl signalling has recently been demonstrated. Prl treatment has been shown to induce tyrosine phosphorylation of EGFR and ErbB2, leading to the activation of ERK/MAPK in breast cancer cell lines (Huang et al. 2006). Also, Prl potentiated the ability of TGF , a ligand of EGFR, to induce mammary neoplasia in a transgenic mouse model co-expressing both genes (Arendt et al. 2006). This observation was associated with increased ERK/MAPK phosphorylation, compared to TGF alone, suggesting that these two important pathways can synergise during the development of disease.
Progesterone and its synthetic analogs has also been implicated in the etiology and pathogenesis of human breast cancer, and too, remains a controversial subject. Depending on the time of administration and the dosage used, Pg can either promote, or provide protection, from mammary tumorigenesis. This most likely reflects whether Pg is exerting proliferative or anti-proliferative effects, dependent on the cell context (Musgrove et al. 1991; Groshong et al. 1997; Lange et al. 1999). For example, the addition of progestin to estrogen in hormone replacement therapy (HRT) can increase the risk of breast cancer, compared with E alone (Ross et al. 2000; Schairer et al. 2000; Campagnoli et
51
al. 2005). Also, Pg can reverse the anti-tumorigenic effects of the antiestrogen, tamoxifen, via a Pgr-mediated mechanism (Robinson and Jordan 1987), and ablation of Pgr is associated with a protective effect against breast carcinogenesis in mice (Lydon et al. 1999). Conversely, the strong protective factor against breast cancer provided from a first full-term pregnancy occurring at an early age, can be recapitulated in rodents by low dose treatments of E plus Pg (Sivaraman and Medina 2002). Neither E nor Pg will induce this response alone. In the normal adult mammary gland, Pgr is expressed by approximately 10% of epithelial cells, which are non-proliferating (Clarke et al. 1997), however in breast cancer proliferating cells can also be Pgr-positive, suggesting that signalling pathways involved in normal mammary development are reactivated during cancer progression (Lange 2008).
In summary, accumulating data indicate that Prl and Pg are involved in the etiology of at least some breast tumors, although at which point in the carcinogenic process they exert their effects is not known, and further studies must be conducted in order to confirm this. Therefore, it is of importance that we continue to investigate the target genes and signalling networks downstream of these hormones, in order to further understand the precise mechanisms which regulate normal mammary development, and potential carcinogenesis.
Project Aims
As outlined above, a combination of pituitary and ovarian hormones tightly regulate normal mammary gland development and influence breast tumorigenesis, however the mechanisms that underlie these actions remain to be elucidated. The challenge now is to dissect the various signalling networks and delineate the different mechanisms downstream of these hormones, in order to increase our overall understanding of how they regulate mammary gland function. The broad aim of this thesis is to therefore investigate the role
52
of novel genes downstream of the Prl- and Pg-signalling pathways in normal mammary gland development and human breast cancer.
From transcript profile experiments previously carried out in our lab (Harris et al. 2006), I have chosen to investigate two putative Prl-regulated genes, which were identified as displaying decreased expression in Prlr-/- mammary epithelium, and increased expression in response to Prl in a cell culture model. One of these genes is the Ets transcription factor, Elf5, which has become the focus of a large body of work within our laboratory. Elf5 is required by the epithelium for full lobuloalveolar development (Zhou et al. 2005), and work conducted in our lab has since demonstrated that (1) Re-expression of Elf5 in Prlr-/- MECs can rescue failed alveolar development, (2) Elf5-/- mice fail to lactate during lactation, and (3) Elf5 overexpression in an inducible transgenic model caused precocious development and milk secretion in virgin mice (Harris et al. 2006; Oakes et al. 2008). These observations, amongst others, illustrate a critical role for Elf5 in the Prl-induced differentiation of the mammary gland.
The work in our laboratory has so far revolved around the investigation of Elf5 function in the normal development of the murine mammary gland. In this thesis, I have investigated the role of Elf5 in human breast cancer and normal mammary epithelial cell lines, in order to determine whether this essential function of Elf5 in the mouse can be translated to the human, and importantly, whether this has any implications in the progression of mammary carcinogenesis. In Chapter Three, we examine the role of Elf5 in the immortal human mammary epithelial cell line, MCF-10A. When cultured on basement membrane, this cell line forms spherical acini and is useful for modelling of normal in vivo mammary development. We investigated the effect of overexpression of Elf5 in MCF-10A cells, and the mechanism involved. In Chapter Four, we examine the role of Elf5 in a human breast cancer cell line, T47D. This steroid receptor-positive cell line has high levels of endogenous Elf5 expression, and we utilise it to further characterise the hormonal regulation of
53
Elf5, and the effect of suppressing Elf5 expression by RNA interference technology.
Another specific aim of this thesis was to characterise a completely novel gene identified from the transcript profile screen. The second molecule I chose to investigate was the EST, AI850846. This EST corresponded to a protein that was later identified and named KIBRA. In Chapter Five, we describe the generation and construction of reagents, which we subsequently used to characterise the expression of KIBRA in the mammary gland and mammary cell lines. In this chapter, we then go on to investigate the effect of hormonal stimulation on KIBRA expression, and the effect of KIBRA overexpression on the growth of normal human mammary cells and breast cancer cell lines. Following on from this work, the aim of Chapter Six was to investigate the function of KIBRA in the mammary gland. We did this by identifying putative interacting partners of KIBRA using a bioinformatic approach, followed by validation of these interactions and further characterisation of the functional consequences. Overall, the work presented in this thesis aimed to further characterise the role and function of two genes downstream of Prl-signalling in the mammary gland, in order to improve our understanding of the mechanisms behind the hormonal influences of normal development and potential tumorigenesis of the mammary gland.
54
Chapter 2: Materials and Methods
55
DNA Methods
Expression vector construction
The human KIBRA cDNA was cloned from HEK-293 cells by PCR into the Gateway entry vector pDONR221 (Invitrogen Life Technologies, Carlsbad, CA) using Forward Primer 5’-GGGG ACA AGT TTG TAC AAA AAA GCA GGC TGG AAG ATG CCC CGG CCG GAG C-3’ and Reverse Primer 5’-GGGG AC CAC TTT GTA CAA GAA AGC TGG GTA GAC GTC ATC TGC AGA GAG AGC TGG -3’. Similarly, two overlapping fragments (WW and C2) of the KIBRA cDNA were cloned into pDONR221 by PCR. Fragment 1, containing the two WW domains, was amplified using the Forward Primer described above, and 5’- GGGG AC CAC TTT GTA CAA GAA AGC TGG GTA TTA CGA TAC GGC GGC TGA GAC ACA GG-3’. Fragment 2, containing the C2 domain, was amplified using 5’- GGGG ACA AGT TTG TAC AAA AAA GCA GGC TTC ACC ATG GAG AGG GAC CGG CTG ATC CTT ATC-3’ and the Reverse Primer described above. The KIBRA cDNA and WW and C2 fragments were then cloned into the mammalian Gateway expression vector pcDNA-DEST40 (CMV/C-terminal V5-6xHis) by recombination (Invitrogen). Human KIBRA cDNA was also subcloned into the NotI site of the untagged expression plasmid pcDNA3.1(+) via the pGEM-T Easy Vector System 1 (Promega) using primers KIBRA5’Forward (GG AAG ATG CCC CGG CCG GAG C) and KIBRA3’Reverse (C TTT TCT GGC GAT TAG ACG TCA TCT GC).
The retroviral vector pMIG has an LTR promoter and expresses a bi-cistronic messenger RNA encoding both the gene of interest and green fluorescent protein (GFP), using an IRES (internal ribosome entry site) (Van Parijs et al. 1999). The Elf5 gene was amplified by PCR with a C-terminal V5 tag using primers hElf5-V5 F (GGA TCC ATG TTG GAC TCG GTG ACA CAC) and hElf5-V5 R (GAA TTC TCA CGT AGA ATC GAG ACC GAG) and was cloned into the pMIG
56
vector. The KIBRA cDNA was also cloned by recombination (Invitrogen) into the retroviral vector, pMIG, which was converted into a Gateway expression construct by Dr Liz Caldon (Cancer Program, Garvan Institute). A retroviral single vector system that enables Doxycycline regulated RNAi (pHUSH) or transgene (ProEx) expression was obtained from Genentech Inc., San Francisco, USA (Gray et al. 2007). A fellow student in our lab, Maria Kalyuga, generated the inducible transgenic Elf5-V5-ProEx and shRNA Elf5-pHUSH constructs according to the methods detailed in Gray et al., 2007.
All new constructs generated were verified by sequencing (Australian Genome Research Facility, University of Queensland, St Lucia, Qld, Australia). The pRK5-DDR1b expression plasmid was kindly provided by Dr A. Ullrich, Max Planck Institute, Germany. The pRcCMV-PKC expression plasmid was kindly provided by Dr T. Biden, Garvan Institute, Australia. The hPRB-1, pMSGluc and pRcCMV-SRC-1 vectors were kindly provided by A. Russell, Garvan Institute, Australia, and pCMV was a kind gift from Dr T. Brummer, Garvan Institute, Australia.
Site-directed mutagenesis
The DDR1 mutants were constructed by site-directed mutagenesis with the pRK5-DDR1b plasmid as a template using the QuikChange II Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) according to the manufacturer’s instructions. For each mutant, two complementary synthetic oligonucleotides were designed of which one is indicated: AAPY mutant, 5’-GGT CCT AGA GAG GCA GCC CCG TAC CAG GAG-3’, AAAA mutant, 5’-GA GAG GCA GCC GCG GCC CAG GAG CCC CGG-3’ and PPPF mutant, 5’GGT CCT AGA GAG CCA CCC CCG TTC CAG GAG-3’. The mutated plasmids were transformed into E. coli XL1 Blue Supercompetent cells (Stratagene), and the mutants were verified by sequencing of the plasmid DNA.
57
Cell Culture Methods
Cell lines
HEK-293 cells
HEK-293 (human embryonic kidney) cells were maintained in Minimum Essential Media (MEM) supplemented with 10% foetal bovine serum (FBS) (15- 010-0500V, Thermotrace, Melbourne, Australia) and 0.5 g of sodium bicarbonate per litre.
MCF-10A cells
MCF-10A cells and MCF-10A-EcoR cells (expressing the ecotropic retroviral receptor; a generous gift of Drs. Danielle Lynch and Joan Brugge) were maintained in monolayer cultures, as previously described (Debnath et al. 2003). The primary culture medium consisted of DMEM/F12 (Invitrogen) with 5% horse serum (Invitrogen), 20 ng/mL recombinant human EGF (236-EG; R&D Systems, Minneapolis, MN, USA), 0.5 g/mL hydrocortisone (H-0888; Sigma, Castle Hill, NSW, Australia), 100 ng/mL cholera toxin (C-8052; Sigma), 10 g/mL bovine insulin (I-1882; Sigma), 50 units/mL penicillin G (Invitrogen), and 50 g/mL streptomycin sulfate (Invitrogen).
T47D and BT474 cells
T47D cells were cultured in RPMI 1640 medium supplemented with 10% fetal calf serum (Thermotrace, Noble Park, Vic., Australia) and insulin (10 mg/mL, Actrapid, Novo Nordisk, Baulkham Hills, NSW, Australia). T47D-EcoR cells for retroviral infection were generated as described in (Musgrove et al. 2001).
58
Other cell lines
A panel of normal and immortalised breast epithelial cell lines and breast cancer cell lines were cultured and maintained within the Cancer Research Program, Garvan Institute of Medical Research.
Transient transfections
Semiconfluent HEK-293 cells were transiently transfected using FuGENE reagent (Roche Diagnostics, Castle Hill, NSW, Australia) according to the manufacturer’s instructions. After 48 h cells were harvested for western blot analysis or immunoprecipitation experiments, as appropriate.
Production of retrovirus
Sixty micrograms of Elf5-V5-pMIG or control pMIG plasmid DNA was transfected into PlatE-EcoR cells, an ecotropic packaging cell line (Morita et al. 2000), using Fugene reagent. The cells were incubated at 37°C and the medium changed 24 h post-transfection. Retroviral supernatants were harvested 24 h later, passed through a 0.45 m filter, and either used for infection immediately or snap frozen and stored at -80°C for later use.
Infection of MCF-10A cells
Low passage MCF-10A/EcoR cells were plated at low density (5 x 105 cells per 10 cm dish) one day prior to infection. The next day the medium was removed, and Polybrene (8 g/mL; Sigma, Castle Hill, NSW, Australia) was added to a 1:3 dilution of the virus, which was added to the cells. The infection was repeated 24 h later. These cells were then expanded and expression confirmed by Western blotting. Cells infected with the pMIG, Elf5-V5-pMIG,
59
KIBRA-pMIG virus were sterile sorted by flow cytometry to obtain a GFP positive population, and these populations were subsequently sorted into low, medium and high expressing populations. Cells infected with the Elf5-V5-ProEx virus were maintained in culture a further 48 h before 7 days of selection using 2 μg/mL puromycin. Efficiency of selection was tested by Western blotting of protein lysates, which were harvested from cells that had been treated with Doxycycline (Clontech, Palo Alto, CA, USA).
Three-dimensional (3D) culture
The establishment of 3D cultures of MCF-10A cells was carried out as described in Debnath et al. 2003. Pre-chilled chamber slides (BD Falcon, Castle Hill, NSW, Australia) were thinly coated with Growth Factor Reduced Matrigel (Product #354230; BD Biosciences, North Ryde, NSW, Australia). MCF-10A cells were resuspended in Assay Medium (the same as the primary culture medium previously described, but with 5 ng/mL EGF and 2% Matrigel). This cell resuspension was syringed and overlaid at a density of 3000 cells per well of an 8-well chamber slide. This time point was taken as Day 0, and the medium was changed every 3-4 days.
siRNA duplexes and transfection
For knockdown of mKibra in HC11 cells an siRNA duplex was designed using the algorithm as described in Reynolds et al, 2004 (Reynolds et al. 2004). Double-stranded RNA oligonucleotides against the target sequence GCUUCACUGACCUCUAUUA were synthesised by Dharmacon Research Inc. (Layfayette, Colorado, USA). For knockdown of hKibra in MCF-10A and T47D cells, four siRNA duplexes (KIBRA siRNA #1; D-014058-1, KIBRA siRNA #2; D- 014058-02, KIBRA siRNA #3; D-014058-03 and KIBRA siRNA #4; D-014058- 04) were purchased from Dharmacon. For knockdown of hElf5 in T47D cells, four siRNA duplexes (Elf5 siRNA #1; D-011265-1, Elf5 siRNA #2; D-011265-
60
02, Elf5 siRNA #3; D-011265-03 and Elf5 siRNA #4; D-011265-04) were also purchased from Dharmacon. Control siRNAs used were siGFP and RISC-Free siRNA (Dharmacon).
The day before transfection cells were plated in 6-well plates at approximately 1.5 x 105 cells per well. siRNA duplexes were transiently transfected using Lipofectamine 2000 (Invitrogen), according to the manufacturer’s instructions. Cells were harvested at 2 days post-transfection to check for mRNA and/or protein levels, unless otherwise specified.
Ligand stimulation of cells
Progesterone stimulation experiments involved treating T47D cells with 10 nM ORG2058 (16a-ethoxy-21-hydroxy-19-norpregn-4-en-3,20-dione), which was obtained from Amersham Biosciences Rydelmere, NSW, Australia) or 100 nM RU486 [17 -hydroxy-11-beta-(4-methylaminophenyl)-17- -(1-propynyl)- estra-4,9-diene-3-one-6-7], which was generously provided by Dr J-P Raynaud of Roussel-Uclaf, Romainville, France. Estradiol (3,17 -Dihydroxy-1,3,5(10)- estratriene) was obtained from Sigma. Human growth hormone was kindly provided by Prof Ken Ho (Garvan Institute of Medical Research). For collagen treatment, HEK-293 and MCF-10A cells were serum starved overnight, then treated with 10 μg/mL of collagen type IV (human placenta; Sigma, Castle Hill, NSW, Australia) or collagen type I (rat tail; Sigma) for 90 min, unless otherwise specified.
Progesterone receptor transactivation assay
HEK-293 cells were plated in six-well plates (2 x 105 cells/well) and 24 h later were transfected with Fugene with 90 ng of hPRB-1, 450 ng of pMSGluc, and 1.2 g of KIBRApcDNA3.1, pcDNA3.1 or SRC-1, and 200 ng of pCMV . The following day cells were treated with 10 nM ORG2058 and harvested 48 h later in 50 L -galactosidase lysis buffer (Galacto-Star, Tropix, Bedford, MA, USA)
61
and assayed for luciferase activity (LucLite reagent, Packard Bioscience, Meriden, CT), and corrected for transfection efficiency by carrying out - galactosidase assays (Galacto-Star, Tropix), according to the manufacturers' instructions.
Cell proliferation assays
Growth curve analysis
Cells were plated at low density in 12-well or 6-well plates and trypsinised at the timepoints indicated. After neutralisation in harvest medium, the cells were washed with PBS and resuspended in 1 mL of harvest medium (or as appropriate). Cell number was quantitated using haemocytometer counting. Relative proliferation of cultures was analysed by plotting cell number on a log scale.
MTT assays
Cell proliferation was determined by the MTT CellTiter96 Assay (Promega Corporation, Madison, WI, USA) following the manufacturer’s instructions. Cells were plated in 96-well plates at 5000 cells per well with 6 replicates per sample. Proliferation was assessed by measuring the absorbance of cells relative to that of control wells (without cells) using 490-nm wavelength on a spectrophotometer.
Flow Cytometry
Cells grown in monolayers were trypsanised and then neutralised with harvest medium. Cells that were to be sorted for GFP expression levels were analysed immediately at the Flow Cytometry Core Facility (Garvan Institute) on a BD FACS AriaTM Cell Sorter (BD Biosciences). Cells that were to be analysed for
62
DNA content were collected by centrifugation at 250 g (5 min at room temperature) and resupended in 1 mL of cold PBS. The resuspended cells were added to 9 mL of 70% ethanol for fixation and then incubated at 4oC at least overnight prior to FACS analysis. On the day of analysis, cells were spun down (250 g; 5 min) and the ethanol was aspirated so as to leave 1 mL remaining. Cells were resuspended in this 1 mL of ethanol and counted using a haemocytometer. Approximately 7 x 105 cells of each sample were aliquotted into a fresh tube and collected by centrifugation, as before, and the ethanol aspirated. Cells were resuspended in 500 L cold PBS/1% Tween while gently vortexing, followed by staining with approximately 5 L of 1 mg/mL propidium iodide (PI) in PBS. Twenty-five microlitres of Ribonuclease A (Sigma) was also added, to a final concentration of 500 g/mL, and gently vortexed to mix. Samples were incubated in the dark at room temperature for up to 5 hours. Samples were syringed gently but thoroughly and then analysed using the FL- 2 channel on a FACSCalibur (BD Biosciences) using CELLQuest 2.0 software. The proportion of cells in the G1, S and G2/M phases of the cell cycle were calculated from the resulting DNA histograms using ModFit LT analysis software (Verity Software House, Inc., Topsham, ME, USA).
RNA Methods
Extraction, quantitation and analysis
Total RNA was extracted using the RNeasy Minikit (Qiagen, Doncaster, Vic., Australia) as described in the manufacturer’s instructions. For Affymetrix GeneChip analysis, RNA quality was assessed by RNA Nano LabChip analysis on an Agilent Bioanalyzer 2100 (Agilent Technologies, Palo Alto, CA, USA) and taken to the Ramaciotti Centre for Gene Function Analysis (UNSW) for Affymetrix GeneChip Array processing. For real-time quantitative PCR, single- stranded cDNA was produced by reverse transcription using 1 μg of RNA in a 20 L reaction using the Reverse Transcription System (Promega, Annadale, NSW, Australia) according to manufacturer’s instructions. cDNA was diluted to
63
o 1:5 with H20 and stored at -20 C. Quantitative PCR was performed using the TaqMan probe-based system on the ABI Prism 7900HT Sequence Detection System (Applied Biosystems, Scoresby, Vic., Australia). All reagents and Gene Expression Assays (Table 2.1) were purchased from Applied Biosystems.
Figure 2.1 Gene expression assays Gene Gene Expression Assay Elf5 (human) Hs00154971_m1 Elf5 (mouse) Mm00468732_m1 Fatty acid synthase (human) Hs00188012_m1 Desmoplakin (human) Hs00189422_m1 KIBRA (human) Hs00392086_m1 GAPDH (mouse) 4352339E
Protein Methods
Antibody preparation
Anti-KIBRA antiserum was prepared in rabbit (Invitrogen) using the synthetic peptide sequence, SAQERYRLEEPGTEGKQ, derived from the human KIBRA central portion, conjugated to keyhole limpet haemocyanin (KLH). The KIBRA IgG polyclonal antibody was purified on a peptide affinity column.
Cell lysate preparation and protein extraction
From monolayer cultures
Protein samples were prepared from cell lines lysed in ice-cold normal lysis buffer (NLB; 50 mM HEPES pH 7.5, 150 mM NaCl, 10% glycerol, 1% Triton X-
100, 1.5 mM MgCl2, 1 mM EDTA, 10 mM pyrophosphate, 100 mM NaF) containing protease and phosphatase inhibitor cocktail tablets (Roche).
64
Samples were incubated in lysis buffer on ice for 20 min, before being briefly vortexed and centrifuged at 17,000 g at 4oC. Supernatants were stored at - 20oC or -80oC. For analysis of KIBRA phosphorylation T47D lysates were incubated with lambda protein phosphatase (New England Biolabs) according to the manufacturer’s instructions.
From 3D cultures
To recover whole cell lysates from acini cultured in Matrigel, media was gently aspirated from culture slides, and 300 mL of BD Cell Recovery Solution (BD Biosciences) was added per well of a 4-well chamber slide and incubated for 75 min at 4°C to depolymerise the Matrigel. The cell/Matrigel suspension was then resuspended and transferred to a fresh tube, and each well washed with 200 mL cold PBS to collect any remaining cells. The suspension was centrifuged at 400 g for 10 min at 4°C and the supernatant sucked off. Residual Matrigel was removed by washing the pellet with cold PBS and centrifuging again at 400 g for 10 min at 4°C. Cell pellets were resuspended in an appropriate volume of normal lysis buffer and prepared as described above for monolayer culture protein isolation.
From mammary glands
The fourth inguinal mammary glands from Elf5-/- mice was frozen in liquid nitrogen before storage at -80°C prior to use (Oakes et al. 2008). Total RNA and DNA was extracted using TRIZOL Reagent (Gibco/Invitrogen Vic, Australia), following which protein was isolated from the phenol-ethanol supernate obtained, according to the manufacturers instructions. Protein lysates extracted using all methods were quantitated using Bradford Protein Assay Reagent (Bio-Rad), according to the manufacturer’s instructions.
65
Immunoprecipitation
One milligram of protein from cell lysates in a total volume of 1 mL of NLB was pre-cleared by rotating at 4oC for 1 h with 40 L of Protein-G Sepharose beads/PBS (1:1; Zymed, San Francisco, CA, USA). The beads were then removed by centrifugation (4oC, 1200 g, 10 min) and each lysate transferred to a fresh tube and incubated with antibodies (approximately 10 L) overnight at 4oC. The following morning Protein-G Sepharose beads/PBS (1:1) were added to the lysates and incubated for 1 h at 4oC. The resulting immunoprecipitates were collected by centrifugation and washed four times with NLB and 1 M NaCl, and resuspended in buffers as described below.
Western blot analysis
Immunoprecipitates and lysates (10 – 40 g) were prepared in NuPAGE LDS Sample Buffer and NuPAGE Sample Reducing Agent (Invitrogen) and separated on NuPAGE Bis-Tris acrylamide gels run in MOPS buffer (Invitrogen). Precision Plus Protein (Dual Colour) (Bio-Rad Laboratories, Hercules, CA, USA) standards were used as a molecular marker. Gels were then transferred to polyvinylidene difluoride (PVDF) membranes (Bio-Rad). The membranes were incubated overnight with the primary antibodies listed in Table 2.2. The secondary antibodies used were horseradish peroxidase-conjugated sheep anti-mouse, donkey anti-rabbit (Amersham Biosciences), donkey anti-goat (Santa Cruz) and goat anti-rat (Pierce Biotechnology, Rockford, IL, USA). Proteins were visualised by chemiluminescence (Perkin-Elmer, Rowville, Vic., Australia). Densitometry was performed using IPLab Gel software (Scanalytics Inc., Fairfax, VA, USA).
66
Antibodies
The following table lists all primary antibodies used, for Western analysis (WB) and immunofluorescence (IF), as well as their dilution and source. Name Source Application Dilution Elf5 Santa Cruz Biotechnology WB; IF 1:100 (Santa Cruz, CA, USA) (WB); 1:50 V5 (R960-25) Invitrogen WB; IF 1:5,000 (WB) 1:150 (IF) GFP (ab6658) Abcam (Cambridge, UK) WB 1:1,000 Laminin V (MAb Chemicon (Temecula, CA, IF 1:100 19562) USA) Bcl-xL BD PharMingen WB 1:1,000 Bcl-2 (124) DakoCytomation WB 1:100 (Carpinteria, CA. USA) Activated Cell Signaling Technologies WB 1:750 caspase-3 (Beverly, MA, USA) Bim Calbiochem (La Jolla, CA, WB 1:100 USA) EGFR (1005): Santa Cruz WB 1:1,000 sc-03 Phospho-ERK Cell Signaling Technologies WB 1:1,000 Total-ERK Cell Signaling Technologies WB 1:1,000 KIBRA N/A WB; IF; IHC 1:500 DDR1 (C-20) Santa Cruz WB 1:500 nPKC (C-20) Santa Cruz WB 1:1,000 Cyclin-D1 (clone Novacastra Laboratories WB 1:100 DCS6) Ltd., United Kingdom -actin (AC-15) Sigma WB 1:40,000
67
Progesterone Kindly donated by Dean WB 1:1,000 receptor (MAb Edwards, Houston, TX, 1294) USA Phospho-Tyr Upstate Biotechnology WB 1:1,000 (4G10) (Lake Placid, NY, USA)
Immunohistochemistry
4 μm tissue sections were mounted on Superfrost Plus adhesion slides (Lomb Scientific, Sydney, Australia) and heated in a convection oven at 75°C for 2 h to promote adherence. Sections were de-waxed and re-hydrated according to standard protocols. Antigen retrieval was performed using citrate EDTA buffer boiled under pressure (human cells) or Proteinase K enzymatic digestion (human and mouse tissue). Endogenous peroxidase activity was inhibited with
3% H2O2. Sections were incubated with the anti-KIBRA primary antibody for 60 min (1:200, human tissue and cells) or 30 min (1:150, mouse tissue) and bound antibody was detected using LINK/LABEL and 3,3'-diaminobenzidine Plus as substrate. With each run, diluent and rabbit IgG isotype control were the negative technical controls. Counterstaining was performed with haematoxylin and 1% acid alcohol. Cell lines were processed at the same time to enable comparison of signal strength. All immunohistochemistry reagents were obtained from DAKO Corporation (Carpinteria, CA) unless otherwise specified.
Immunofluorescence and microscopy
Fixation and staining protocol of 2D cultures
Cells were plated onto chamber slides, washed with PBS once prior to fixation, and then fixed in 4% paraformaldehyde/PBS for 15 min at room temperature.
68
The slides were then rinsed in PBS, permeablised in 0.2% Triton X-100, rinsed again in PBS and blocked with 1% BSA in PBS for at least one hour. Slides were then incubated in primary antibody diluted in 1% BSA/PBS overnight at 4oC. The following morning the slides were rinsed in PBS 2-3 times, cells were stained using the appropriate Cy2- or Cy3-labelled secondary antibodies (Jackson ImmunoResearch Laboratories, Pennsylvania, USA). Where required cells were counterstained with either FITC-phalloidin or TRITC-phalloidin (Sigma) as an actin counterstain, and TOPRO-3 (Molecular Probes OR, USA) as a DNA counterstain. Following 2-3 washes with PBS, the slides were then mounted in 90% glycerol/PBS ready for visualisation.
Fixation and staining protocol of 3D cultures
Cells cultured on Matrigel were rinsed with PBS once, then fixed with 2% paraformaldehyde/PBS (pH 7.4) for 20 min at room temperature. Slides were then permeabilised in 0.5% Triton X-100/PBS for 10 min at 4oC, followed by rinsing 3 times in 100 mM glycine/PBS (10-15 min per wash at room temperature). Slides were then blocked in IF Buffer (7.7 mM azide, 0.1% BSA, 0.2% Triton X-100, 0.05% Tween-20 in PBS) plus 10% goat serum for 1-1.5 h at room temperature. Following this, slides were incubated with primary antibody, where appropriate, diluted in IF Buffer plus 10% goat serum at 4oC overnight. The next morning slides were rinsed with 3 washes of IF Buffer then stained with secondary antibodies and counterstains diluted in IF Buffer plus 10% goat serum for approximately 1 h at room temperature. Following 2 rinses with IF Buffer and 1 rinse with PBS (5 min), slides were mounted with Prolong Antifade Reagent (Molecular Probes, Invitrogen) and allowed to dry overnight at room temperature. Slides were visualised the following day.
Microscopy
Immunofluorescent microscopy was carried out on a Leica DMRBE upright microscope. Imaging of cells grown as monolayers was performed using either
69
a 63x or 100x PL APO oil objective. Imaging of cells grown in 3D was performed using a 40x HCX PL APO oil or immersion objective.
Bioinformatic Methods
Build 33 of the human proteome, containing 37490 sequences, was downloaded from the NCBI (http://www.ncbi.nlm.nih.gov) and the protein regular expression program 'Preg' (Rice et al. 2000) was used to identify all sequences in the proteome with the PPxY and RxPPxY motifs (where P is proline, R is arginine, Y is tyrosine and x is any amino acid). The GI accession numbers from the matched sequences were used to extract annotations from the 15 June 2003 build of the Affymetrix HGU-133A annotation file. The annotations were then extended by searching batch SOURCE (Diehn et al. 2003) with the LocusLink identity numbers derived from the matched Affymetrix HGU-133A annotations. Protein expression in the mammary gland was determined from the NCBI Unigene database (http://www.ncbi.nlm.nih.gov/unigene).
Statistical analysis
Statistical analyses were performed using StatView 4.5 (Abacus Concepts Inc., Berkely, CA, USA). Differences between groups were evaluated by Fisher’s protected least significant difference test after ANOVA analysis where appropriate.
Microarray analysis
Quality of the arrays was assessed by the Ramaciotti Centre for Gene Function Analysis by measuring 5 different parameters, including the mean of the raw intensity for all of the probes on the array, the mean probeset signal for all
70
probesets analysed, and the mean over all probesets analysed of the absolute log expression value relative to all other arrays). The arrays were found to be of similar high quality. Normalization of the arrays was performed using the Probe Logarithmic Intensity ERror (PLIER) algorithm as implemented in the Affymetrix Power Tools environment. CEL files from all nine arrays were normalized together. All statistical analyses were carried out in the R statistical programming language (R website; www.R-package.org). The data was analysed for differential expression using LIMMA (Smyth 2005), comparing Pg- treated and Baseline cells, Elf5-KD and Baseline cells, and finally Elf5-KD and Pg-treated cells. The p-values obtained for each comparison were adjusted for multiple testing using the Benjamini-Yekutieli (BY) correction, implemented in the multtest package in R (Benjamini and Yekutieli 2001; Pollard et al. 2005).
Functional Annotation Analysis
Functional annotation analysis of sets of selected genes was carried out using the gene-enrichment annotation analysis tools within DAVID version 2008 (DAVID website; http://david.abcc.ncifcrf.gov/home.jsp). Categories analyzed included GO categories (Biological Process, Molecular Function, and Cellular Component), protein domain categories (InterPro Name, Superfamily Name, SMART Name), pathways databases (BBID, BioCarta, KEGG Pathways), functional categories (COG/KOG Ontology, Sp Pir Keywords, Up Seq Feature) and a disease category (Genetic Association Database). The BY correction for multiple testing was applied to the EASE scores, and the significance threshold set at adjusted P 0.05.
71
Chapter 3: Effect of Elf5 Expression on MCF-10A Cells in 3D Culture
72
Introduction
The genes that are important for the control of mammary cell proliferation and differentiation during normal mammary gland development are also targets for mutation and dysregulated expression that can result in hyperplasia and cancer. New techniques that define this genomic regulatory network offer an unprecedented opportunity to understand the processes that build a mammary gland, and which go awry in cancer.
To investigate what genes are important in the control of mammary cell proliferation and differentiation, our lab transcript profiled mammary gland epithelial cells from the prolactin receptor knockout (Prlr-/-) mouse (Ormandy et al. 2003). In addition to a number of reproductive defects such as sterility, irregular cycles and failed blastocyst implantation, these mice display a lack of alveolar differentiation and milk secretion (Ormandy et al. 1997a). For this study, Prlr-/- and wildtype (Prlr+/+) mammary epithelia were transplanted into the cleared mammary fat pads of 4-week-old immunocompromised Rag1-/- animals. We transcript profiled the Prlr+/+ and Prlr-/- mammary epithelial transplants during early time points of pregnancy to overcome reduced epithelial cell numbers observed in Prlr-/- mammary glands at later time points and to focus our attention on early events (Harris et al. 2006). To exclude gene expression contributed by the fat pad we also transcript profiled Rag1-/- mammary fat pads cleared of endogenous epithelium. In addition, we profiled prolactin treated and control SCp2 cells, a murine mammary epithelial cell line, that secretes -casein when cultured on extracellular matrix (ECM) and treated with lactogenic hormones. A combined analysis of these different models identified genes that showed decreased expression in the Prlr-/- epithelium compared to wild type during early pregnancy and showed upregulation by prolactin treatment of SCp2 cells (Harris et al. 2006). Many of these genes were known to be important for mammary development, but we also identified
73
a number of novel epithelial-specific prolactin-regulated genes and expressed sequence tags (ESTs). One of these molecules was Elf5.
Elf5 is an epithelial-specific member of the large Ets transcription factor family characterised by a highly conserved DNA binding domain, the Ets domain, which binds to sites containing a central GGA motif (Sharrocks et al. 1997). Elf5 (e74-like factor 5 or ESE-2) was originally described by Zhou et al. (1998) as being isolated from an adult mouse lung cDNA library following screening with a probe containing the Ets domain of Elf3 (e74-like factor 3 or ESE-1), thus it is most closely related to Elf3. In fact, the divergent Ets domain of Elf5 shares 67% amino acid identity with that of Elf3 (Zhou et al. 1998), and is also 65% similar to ESE-3/EHF, indicating that these three Ets family members form a separate subfamily of genes (Kas et al. 2000). Mouse Elf5 and human ESE-2 are highly conserved, sharing approximately 95% identity, suggesting conservation of function (Zhou et al. 1998).
Elf5 is expressed predominantly in secretory epithelium of organs such as the mammary gland and salivary gland, as well as displaying moderate levels in the prostate, lung, and kidney (Oettgen et al. 1999). In addition, work in our laboratory has shown that in the mammary gland Elf5 expression is restricted to the luminal epithelial cells, and its expression increases dramatically during pregnancy to far higher levels than observed in other tissues (Harris et al. 2006). Elf5 has subsequently been shown to be important in mammary gland development by a number of experiments. Firstly, Elf5 heterozygote (Elf5+/-) mice show defective lobuloalveolar development and decreased milk protein gene expression during pregnancy (Zhou et al. 2005). Secondly, retroviral re- expression of Elf5 in Prlr-/- mammary glands rescued failed alveolar development (Harris et al. 2006). Elf5 can induce transcription of the milk protein gene whey acidic protein (Wap) by binding to its promoter (Thomas et al. 2000). Finally, in Elf5 knockout (Elf5-/-) mice, lobuloalveolar development is severely impaired and milk production is lost. Conversely, forced Elf5 expression in an inducible transgenic model resulted in precocious milk
74
secretion during pregnancy (Oakes et al. 2008). These data all lend support to the critical role of Elf5 in the differentiation of the alveolar epithelium during pregnancy.
The use of three-dimensional (3D) cell culture models are useful tools for studying the biochemical and cellular processes which occur in mammary epithelial cell morphogenesis in vitro, and which recapitulates numerous features of breast epithelium in vivo (Petersen et al. 1992; Debnath and Brugge 2005). These cell culture systems also allow for the identification and investigation of potential oncogenes and pathways in a biologically relevant context. MCF-10A cells are a widely used cell line for this in vitro modelling of developmental pathways. When grown in 3D culture, these cells form acini-like spheroids of a single outer layer of cells surrounding a hollow lumen, with apicobasal polarisation of cells and deposition of basement membrane components (Debnath et al. 2003). A schematic diagram of the biological processes that contribute to this morphogenesis is depicted in Figure 3.1. The work in our laboratory so far has focussed on investigating the role of Elf5 in normal development of the mouse mammary gland, and so the aim of the work presented here was to examine the role of Elf5 expression in human mammary epithelial cells, and to identify any potential role in carcinogenesis. For clarity, human ESE-2 will be referred to as Elf5 throughout this chapter.
The specific aims of this chapter are to: (1) Generate constitutive and inducible Elf5 overexpressing human mammary cell lines, (2) Investigate the effect of Elf5 overexpression on acinar morphogenesis in three-dimensional culture, and (3) Investigate the mechanism of Elf5 action in MCF-10A cells.
75
proliferation
polarisation of outer cells
survival signalling in outer cells
luminal cell death
Figure 3.1 Schematic diagram of MCF-10A acinar morphogenesis. The process is depicted over 20 days, and shows the stages of proliferation, polarisation and luminal apoptosis (Figure adapted from Debnath et al. 2003).
76
Results
Generation of constitutive Elf5 overexpressing stable cell lines
To investigate the effect of Elf5 overexpression in human mammary cells, we used the non-tumorigenic human breast epithelial cell line, MCF-10A, which retains many of the characteristics of normal breast epithelium and fails to exhibit growth in immune-deficient mice (Lane et al. 1999). Together with fellow PhD student, Maria Kalyuga, retroviral vectors expressing Elf5 were constructed and used to infect the MCF-10A-EcoR cell line (which stably expressed the ecotropic retroviral receptor, EcoR). This particular construct consisted of a V5-tagged Elf5 cDNA in the pMIG-IRES-GFP vector (Van Parijs et al. 1999) for constitutive Elf5 expression (Figure 3.2).
Figure 3.2 Schematic representations of Elf5-V5-pMIG vector for constitutive Elf5-V5 overexpression. LTR = long terminal repeat; IRES = internal ribsome entry site; GFP = green fluorescent protein.
The presence of the internal ribosome entry site (IRES) in the pMIG vector enables the expression of both Elf5 and GFP (Green Fluorescent Protein) controlled by one promoter, and this GFP tag allowed us to sort by FACS (fluorescence activated cell sorting) for GFP-positive populations into high, medium and low expressing populations. This corresponded to the level of Elf5 expression, as shown by the relative Elf5 mRNA levels in these sorted cell populations (Figure 3.3A). It should be noted that MCF-10A cells do not express any endogenous Elf5. Immunofluorescence and fluorescence analysis of -Elf5 and GFP respectively, showed that Elf5-V5-pMIG infected cells, which displayed higher levels of GFP did indeed express higher levels of Elf5, demonstrating the direct correlation between GFP and Elf5 levels (Figure 3.3B).
77
A
B GFP -Elf5 TOPRO overlay
(a)
(b)
(c)
(d)
Figure 3.3 A Relative levels of Elf5 mRNA transcripts in sorted Elf5-overexpressing MCF- 10A populations B Immunofluorescent images of MCF-10A cells infected with Elf5-V5-pMIG (a – c) or control (d) retrovirus, showing GFP fluorescence and staining for Elf5 (red) and TOPRO (nucleus; blue) as labelled. Scale bars represent 40 m. 78
Investigation of the effect of Elf5 expression on the proliferation of MCF-10A cells in two-dimensional culture
To determine whether Elf5 overexpression had any effect on the growth of MCF-10A cells in 2D culture, we carried out a number of proliferation assays on cells that had been sorted for high Elf5 expression levels. These assays included growth curves, MTT assays and FACS analysis of the proportion of cells in the proliferative S phase. As illustrated in Figure 3.4, Elf5 expression had no significant effect on proliferation in the assays performed. However, as the three-dimensional culture of this cell line on a reconstituted basement membrane is commonly used for in vitro modelling of several aspects of glandular architecture in vivo, we were interested in determining whether Elf5 expression had any effect on acinar morphogenesis in these cells.
79
A
B
C
Figure 3.4 Elf5 overexpression has no significant effect on the proliferation of MCF- 10A cells in 2D culture. A Growth curve B MTT assay C Percentage of cells in proliferative S- phase as determined by FACS analysis. Error bars represent mean±SE of at least 2 independent experiments, each which consisted of 3 replicates.
80
Investigation of the effect of Elf5 expression on acinar morphogenesis in three-dimensional culture
The MCF-10A cells which had been infected with Elf5-V5-pMIG retrovirus (Elf5) or corresponding control virus (pMIG) were plated onto Matrigel, a commercially available reconstituted basement membrane, to examine the effect of Elf5 overexpression on 3D culture. As MCF-10A cells normally do not express any endogenous Elf5, the first thing we examined was the expression of Elf5 over the twenty-day time course of the experiment. Figure 3.5 shows that Elf5 levels were repressed over time to low levels. Expression levels of V5 and GFP also decreased accordingly. This phenomenon was due specifically to the expression of Elf5, as levels of GFP in the pMIG control cells remained high throughout the experiment (Figure 3.5A). Figure 3.5B depicts V5 protein levels in the Elf5-overexpressing cells normalised to -actin levels, as measured by densitometric analysis. Repression of expression from transgenic cassettes via methylation and other epigenetic mechanisms acting on cassette promoter regions is an established phenomenon (Gebara et al. 1987). This could occur globally to reduce Elf5 expression, or could confer a growth advantage on a subset of cells expressing low Elf5 levels. This suppression of Elf5 expression over time was a 3D-culture specific phenomenon, as when we maintained Elf5- V5-pMIG MCF-10A cells which had been sorted for high expression in normal 2D culture over a similar time course, there was no decrease in V5 expression in these cells (Figure 3.5C).
81
A Day 0 Day 7 Day 10 Day 15 Day 20 P E P E P E P E P E