UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Molecular regulators of apical/basal polarity during mammary epithelial morphogenesis and invasive tumor… Somasiri, Aruna Mahendra 2004

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2004-931803.pdf [ 20.13MB ]
Metadata
JSON: 831-1.0091859.json
JSON-LD: 831-1.0091859-ld.json
RDF/XML (Pretty): 831-1.0091859-rdf.xml
RDF/JSON: 831-1.0091859-rdf.json
Turtle: 831-1.0091859-turtle.txt
N-Triples: 831-1.0091859-rdf-ntriples.txt
Original Record: 831-1.0091859-source.json
Full Text
831-1.0091859-fulltext.txt
Citation
831-1.0091859.ris

Full Text

MOLECULAR REGULATORS OF APICAL/BASAL POLARITY DURING MAMMARY EPITHELIAL MORPHOGENESIS AND INVASIVE TUMOR PROGRESSION By Aruna Mahendra Somasiri B.Sc , The University of British Columbia, 1996 M.Sc., The University of British Columbia, 1999 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE R E Q U I R E M E N T S FOR THE D E G R E E OF DOCTOR OF PHILOSOPHY in THE F A C U L T Y OF G R A D U A T E STUDIES (Department of Anatomy and Cell Biology) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA June 2004 © Aruna Mahendra Somasiri, 2004 ABSTRACT During breast cancer progression, the mammary gland undergoes architectural changes marked by a disruption of epithelial apical/basal polarity. In infiltrating lobular carcinomas (ILC) this disruption is marked by a loss of adherens (AJ) and tight (TJ) junctions. However, in the more prevalent infiltrating ductal carcinomas (IDC), A J often remain while TJ are lost or disorganized. Thus, the loss of TJ polarity may be an important component of invasive, cancer progression. Because TJs and the basement membrane (BM) are critical components of apical and basal polarity, I chose to study the importance of polarity changes during breast cancer progression. To identify the molecular regulators of TJ dynamics in the normal breast I used a 3D hierarchical model of mammary epithelial cell morphogenesis in vitro. In this hierarchy, functional A J were initially formed and then the TJs were formed. These TJs were not polarized and the TJ scaffolding protein ZO-1 co-localized with the A J protein [3-catenin at AJs. Upon addition of a soluble B M , ZO-1 was released from the AJ complex and the TJ complex migrated apically. This apical polarization of the TJ was mediated, at least in part, by interactions of the a6p4 integrin with laminin in the B M . The ETV6-NTRK3 (EN) fusion protein is expressed in secretory breast carcinomas (SBC) and two proteins, Podocalyxin and the Integrin-linked kinase (ILK), are involved in modulating cell-cell junctions. Thus, I examined the effects of these three genes on cell polarity. Forced expression of E N in normal epithelial cells induced cell proliferation without affecting TJ polarity, further explaining the non-metastatic phenotype of SBC. Podocalyxin was highly expressed in a subset of ductal tumors that become metastatic. When podocalyxin was expressed in a well-differentiated breast cancer cell line, TJs and polarity were perturbed, resulting a phenotype similar to IDC. When ILK was expressed in normal epithelial cells, they completely lost both TJs and AJs initiating an epithelial to mesenchymal transformation, hence an ILC phenotype. These data suggest that the establishment of polarized TJs is critical for normal mammary epithelial architecture, while changes to this architecture, at least in part, contribute to invasive breast tumor progression. 11 T A B L E O F C O N T E N T S A B S T R A C T i i T A B L E CONTENTS i i i LIST OF FIGURES viii LIST OF T A B L E S xi LIST OF ABBREVIATIONS xii A C K N O W L E D G M E N T S xv CHAPTER 1: INTRODUCTION 1 1.1 Breast Cancer Problem 1 1.2 Breast Cancer Progression 2 1.2.1 Hyperplasia of Usual Type 3 1.2.2 Atypical Hyperplasia 8 1.2.3 Carcinoma In Situ 9 1.2.3.1 Ductal Carcinoma In Situ 10 1.2.3.2 Lobular Carcinoma In Situ 11 1.2.4 Infiltrating Carcinoma 12 1.2.4.1 Infiltrating Ductal Carcinoma 12 1.2.4.1a Secretory Breast Carcinoma 13 1.2.4.2 Infiltrating Lobular Carcinoma 13 1.3 Normal Mammary Gland Development 15 1.3.1 Structure of the Terminal Duct Lobular Unit 18 1.4 Apical/Basal Polarity in Epithelial Cells 18 1.4.1 What is Apical-Basal Polarity? 18 1.4.2 Apical Junction Complex 19 1.4.2.1 Adherens Junction 20 1.4.2.2 Tight Junction 21 1.4.2.3 Cell Polarity Proteins 25 1.4.2.4 M A G U K Proteins 26 1.4.3 Interactions with the Basement Membrane 27 ii i 1.4.3.1 Basement Membrane Proteins 27 1.4.3.2 Integrin and Dystroglycan 28 1.4.4 Assembly of Apical Junction Complex and Cell Polarization 30 1.5 Thesis Problem 31 1.6 References 46 CHAPTER 2: C E L L SHAPE A N D B A S E M E N T M E M B R A N E - D E P E N D E N T TIGHT JUNCTION FORMATION A N D POLARIZATION OF M A M M A R Y EPITHELIAL C E L L S 78 2.1 Summary 78 2.2 Introduction 79 2.3 Materials and Methods 81 2.3.1 Antibodies 81 2.3.2 Cell Culture 81 2.3.3 Immunofluoresence Microscopy 82 2.3.4 Cell Fractionation and Western Blotting 83 2.3.5 Integrin Blocking 85 2.4 Results , 86 2.4.1 Basement Membrane Gel Culture Induces Tight Junction Formation and Polarization in Mammary Epithelial Spheroids...86 2.4.2 Differential Regulation of Tight junction Formation and Polarization 87 2.4.3 Adherens and Tight Junction Proteins Interact in Naked Cell Clusters 89 2.4.4 Integrin Signaling Initiates Tight Junction Polarization in Basement Membrane Overlaid Clusters 90 2.4.5 Integrin Signaling Initiates Tight Junction Polarization in Basement Membrane Overlaid Clusters 91 2.5 Discussion 91 2.6 References 112 iv CHAPTER 3: ETV6-NTRK3 FUSION PROTEIN INDUCES HYPER-PROLIFERATION OF M A M M A R Y EPITHELIAL C E L L S IN 3D C U L T U R E BUT FAILED TO DISRUPT EPITHELIAL C E L L POLARITY 117 3.1 Summary 117 3.2 Introduction 117 3.3 Materials and Methods 119 3.3.1 Cell Culture 119 3.3.2 Immunofluoresence Microscopy 121 3.3.3 Western Blotting 121 3.4 Results 121 3.4.1 Expression of E N did not cause a phenotypic change in mammary epithelial cell monolayers 124 3.4.2 Expression of E N did not prevent extracellularmatrix dependent spheroid formation and differentiation 122 3.4.3 Expression of E N induced IGF-1/Insulin dependent cell proliferation in 3D culture 123 3.4.4 Cyclin Dl /2 levels remained high in E N expressing spheroids 123 3.4.5 PI3K is required for E N induced cell proliferation in 3D spheroids 124 3.4.6 E N expressing spheroids are able to undergo normal junction polarization 125 3.5 Discussion 125 3.6 References 142 CHAPTER 4: THE ANTI-ADHESION P O D O C A L Y X I N DISRUPTS BREAST C A R C I N O M A C E L L JUNCTION A N D ITS OVEREXPRESSION INDEPENDENTLY PREDICTS BREAST C A N C E R PROGRESSION 148 v 4.1 Summary 148 4.2 Introduction 149 4.3 Materials and Methods 151 4.3.1 T M A Construction 151 4.3.2 T M A Immunohistochemistry, Scoring and Correlation Analysis 152 4.3.3 Cell Culture, Transfection and Immunostaining 152 4.3.4 Transepithelial Resistance 154 4.4 Results 154 4.4.1 Podocalyxin is Highly Expressed in a Subset of Invasive Breast Carcinomas 154 4.4.2 High Podocalyxin Expression is an Independent Marker of Poor Outcome 155 4.4.3 Ectopic Podocalyxin Overexpression Initiates MCF-7 Breast Carcinoma Cell Delamination in Monolayer Culture 156 4.4.4 Ectopic Podocalyxin Expression Disrupts Breast Carcinoma Cell Junctions 157 4.4.5 Ectopic Podocalyxin Expression Perturbs Basement Membrane-Dependent Polarization and Spheroidal Morphogenesis 158 4.5 Discussion 159 4.6 References 183 CHAPTER 5: OVEREXPRESSION OF THE INTEGRIN-LINKED KINASE L E A D S TO LOSS OF POLARITY, C E L L JUNCTIONS A N D M E S E N C H Y M A L L Y TRANSFORMS M A M M A R Y EPITHELIAL CELLS 192 5.1 Summary 192 5.2 Introduction 193 5.3 Materials and Methods 195 vi 5.3.1 Cell Culture 195 5.3.2 Viral Infection and Transfection 195 5.3.3 Immunofluorescence 196 5.3.4 Western Blotting 196 5.3.5 ILK Kinase Activity 197 5.3.6 RT-PCR 197 5.4 Results 198 5.4.1 Endogenous ILK Kinase Activity in Mammary Epithelial Cells is Upregulated by Interactions with the Basement Membrane 198 5.4.2 Overexpression of wtILK Inhibits Basement Membrane-Dependent Morphogenesis and Differentiation 199 5.4.3 wtILK Overexpression Disrupts Adherens Junctions 199 5.4.4 wtILK Overexpression Initiates an Epithelial to Mesenchymal Transformation 200 5.4.5 Subcellular ILK Localization is Altered in Mesenchymally Transformed Cells ; 201 5.4.6 Forced E-cadherin Expression Causes an Epithelial Reversion in Mesenchymal wtELK-overexpressing Cells 202 5.5 Discussion 203 5.6 References 228 CHAPTER 6: G E N E R A L DISCUSSION A N D FUTURE DIRECTION 239 6.1 References 249 vii LIST OF FIGURES Figure 1.1 Epithelial Organization 34 Figure 1.2 Hypothetical Linear Model of Breast Cancer progression 36 Figure 1.3 IGF-1 signaling pathways 38 Figure 1.4 Role of 14-3-3 in apoptosis 40 Figure 1.5 Apical Junction Complex 42 Figure 1.6 PDZ Domain Containing Proteins of Tight Junctions 44 Figure 2.1 Mammary Epithelial Morphogenesis Model 96 Figure 2.2 Basement Membrane Gel Culture Induces the Formation of Differentiated and Polarized Mammary Epithelial Spheroids 98 Figure 2.3 Differential Regulation of Tight Junction Formation and Polarization. 100 Figure 2.4 Sequential and Relative Distribution of Adherens and Tight Junction Proteins During Spheroid Formation 102 Figure 2.5 The Ratios of Soluble vs Insoluble Junction proteins During Spheroid Formation 104 Figure 2.6 Adherens and Tight Junction Proteins Colocalized in Naked Clusters and Separate During Spheroid Formation 106 Figure 2.7 ZO-1 and P-Catenin Physically Interacts in Naked Clusters 108 Figure 2.8 a6 Integrin Blocking Antibody Disrupts Tight Junction Polarization 110 Figure 3.1 Expression of E N did not cause a phenotypic change in mammary epithelial cell monolayers 130 Figure 3.2 E N expressing cells are able to undergo spheroid formation and functional differentiation of mammary epithelial cells 132 Figure 3.3 Expression of E N induced IGF-1 /Insulin dependent cell proliferation in 3D culture 134 Figure 3.4 Cyclin Dl /2 levels remained high in E N expressing spheroids 136 Figure 3.5 PI3K is required for E N induced cell proliferation 138 vi i i Figure 3.6 E N expressing cells form multi-layered spheroids without disrupting the spheroid polarity : 140 Figure 4.1 Podocalyxin is highly expressed in a subset of invasive breast tumors 167 Figure 4.2 High Podocalyxin expression is associated with poor outcome 169 Figure 4.3 Expression of Podocalyxin in human breast carcinoma cell lines 171 Figure 4.4 Podocalyxin overexpression had a subtle effect on E-cadherin and occludin localization in MCF-7 cell monolayers 173 Figure 4.5 Podocalyxin overexpression has no effect on steady state levels of E-cadherin and Occludin 175 Figure 4.6 Podocalyxin overexpression alters the localization of peripheral membrane cell junction proteins in MCF-7 cell monolayers 177 Figure 4.7 Podocalyxin overexpression does not disrupt adherens junctions in basement membrane gel culture 179 Figure 4.8 Podocalyxin ovrexpression disrupts apical/basal polarity and spheroid architecture 181 Figure 5.1 Endogenous ILK activity is upregulated by cellular interactions with the basement membrane 209 Figure 5.2 wtILK overexpression in viral infectants prevents basement membrane-dependent differentiation 211 Figure 5.3 wtILK overexpression in viral infectants disrupts cell-cell junctions... .213 Figure 5.4 wtILK overexpression in viral infectants alters the cytoskeleton 215 Figure 5.5 ILK overexpression in scp2 cells leads to loss of E-cadherin and increased nuclear (3-catenin activity 217 Figure 5.6 Increasing overexpression of wtILK in stable transfectants induces an epithelial to mesenchymal transformation 219 Figure 5.7 Increasing overexpression of wtILK in stable transfectants causes a loss of cytokeratin proteins, an increase in vimentin protein, and a decreased ability to undergo basement membrane-dependent differentiation 221 ix Figure 5.8 ILK localization is altered in wtILK overexpressing, mesenchymally transformed cells 223 Figure 5.9 wtILK overexpression downregulates E-cadherin 225 Figure 5.10 Forced E-cadherin expression rescues the epithelial phenotype in wtILK overexpressing cells 227 Figure 6.1 Hypothetical linear model of breast cancer progression showing the findings of this thesis 247 x L I S T O F T A B L E S Table 4.1 Characteristics of the Invasive Breast Carcinoma Tissue Microarray.... 164 Table 4.2 Cox Regression Multi-Variant Analysis of Disease Specific Survival... 165 Table 4.3 Marker Correlation with High Podocalyxin Expression 166 Table 5.1 ILK overexpression 208 xi List of Abbreviations 3D Three dimensional 2D Two dimension a6 a6 integrin aPKC Atypical protein kinase C A D H Atypical ductal hyperplasia A H Atypical hyperplasia A J Adherens junction A L H Atypical lobular hyperplasia B M Basement membrane B S A Bovine serumalbumin CALX Carbonic anhydrase IX Cdk2 cycle-dependent kinase2 C G H Comparative genomic hybridization CHO Chinese hamster ovary cells CIS Carcinoma in situ DCIS Ductal carcinoma in situ D G Dystroglycan DMEM/F12 Dubecco Modified Eagle's Medium/F12 (Ham's F12) DOC Sodium deoxycholate DTDL Asp-Thr-Asp-Leu DTHL Asp-Thr-His-Leu E2 17P-estradiol E C L Enhanced chemiluminescent E C M Extracellular Matrix EDTA Ethylenediaminetetraacetic acid EGFP Enhanced green fluorescence protein EHS Engelberth-Holm-Swarm E M T Epithelial to Mesenchymal Transformation ER Estrogen receptor ERE Estrogen response elements E R M Ezrin-radixin-moesin F A K Focal adhesion kinase FBS Fetal bovine serum FITC Fluorescein isothiocyanate FRET Fluorescence resonance energy transfer GSK3 Glycogen synthase kinase3 HGF Hepatocyte growth factor HUT Hyperplasia of usual type IC Invasive carcinoma IDC Infiltrating ductal carcinoma IgG Immunoglobulin G IGF-1 Insulin like growth factor-1 IGF-1R Insulin like growth factor-1 receptor xii V ILK Integrin linked Kinase IRS-1 Insulin receptor substrate-1 J A M Junctional adhesion molecules kdILK Kinase dead Integrin linked Kinase L - C A M Liver-cell adhesion molecule LEF-1 Lymphoid enhancing factor-1 L N Laminin L O H Loss of heterozygosity LCIS Lobular carcinoma in situ M A P K Mitogen activated protein kinase M A G U K Membrane associated guanylate kinase M D C K Madin Darby Canine Kidney MeOH Methanol MEP-21 Myb-Ets-transformed progenitor M R E M A G U K recruitment domain M M T V Mouse mammary tumor virus M T A metastasis-associated gene MUPP1 Multi-PDZ domain protein 1 NDF Neu differentiation factor NGS Normal Goat Serum NHERF-2 Na(+)/H(+) exchange regulatory factor-2 NTRK3 Neurotrophin-3 receptor P A R Partitioning-defective protein PBS Phosphate Buffered Saline PCR Polymerase Chain Reaction PDZ PSD95-dlg-ZO-l PI3-K Phosphotidylinositol 3 kinase P K B Protein kinase B PL Plastic PMSF Phenylmethanesulfonyl Fluoride PRL Prolactin PTK Protein tyrosine kinase RT-PCR Reverse transcriptase polymerase chain reaction SAGE Serial analysis of gene expression S A M Sterile a motif SBC Secretory breast carcinoma SDS-PAGE Sodium Dodecyl Sulphate Polyacrylamide Gel Electrophoresis SH3 Src hormology-3 TBS Tris Buffered Saline T D L U Terminal duct lobular unit TER Transepithelial resistance Tcf T cell factor TJ Tight junction T M A Tissue microrray TGF-(3 Transforming growth factor-beta WAP Whey acidic protein Xll l WtILK Wild-type Integrin linked Kinase ZO Zonula Occludens xiv A c k n o w l e d g e m e n t s I would like to take this opportunity to thank my supervisor Dr. Calvin Roskelley for providing an independent learning environment and for supporting me throughout my graduate career. It has been an eye opening experience working in his lab, and I have been able to explore many different research areas as well as gain lot of technical experiences. I must also thank my supervisory committee, Dr. Linda Matsuuchi, Dr. Kelly McNagny, and Dr. Wayne Vogl for support and guidance as well as friendships throughout the years without which this work could not have been completed. A special thanks to past and present members of the Roskelley lab, especially Colleen Wu and Marcia McCoy for creating a fun and supportive environment to work in. I would also like to thank the faculty & staff and fellow students in the department for making my stay in the department a memorable one. A special thank to Arthur Legg, Andrea Feldman, Lynn Bechberger and Eliza Leung for their help in proof reading numerous drafts of this thesis. I would also like to thank Dr. Elaine Humphrey and the Bio Imaging facility for the continuous support and unrestricted access to the facility Finally, I would like to specially thank my mom, dad and brother, without whose support and tolerance I would not have been able to start or finish my studies far away from home. This work was supported by studentships from the National Cancer Institute of Canada and the Michael Smith Foundation for Health Research. xv CHAPTER 1 Introduction 1.1 Breast Cancer Problem Breast cancer is one of the most common female cancers in developed countries. It has become a major health problem over the last 50 years, affecting as many as one in eight women during their lifetime. The burden of breast cancer worldwide in both developed and developing countries is. increasing, and evidence suggests that unless action is taken it will continue to grow for the foreseeable future. It is estimated that each year the disease is diagnosed in over one million women worldwide and is the cause of death of over 400,000 women. Canadian cancer statistics estimate that in Canada alone, in 2003, 30.7% of all the new cancer cases and 17% of all the cancer related deaths are attributed to breast cancer (Canadian Cancer Statistics 2003, National Cancer Institute of Canada). In the United States, for instance, the incidence rate for breast cancer has increased steadily by about 1-2% per year since 1960 (Forbes, 1997a and 1997b). Similar incidence rates for breast cancer to those in United States are found in the majority of other western industrialized countries (Parkin et al., 1999). The majority of breast cancer-related deaths occur due to aggressive secondary tumor formation at distant sites via a process known as metastasis (Parker and Sukumar, 2003). Metastasis begins with alterations in cell-cell adhesions that allow cells to break away from the primary tumor as it expands. In this introduction, I will address our current understanding of how benign proliferative lesions in the breast undergo a progressive disruption of the normal glandular architecture during the early stages of invasive pre-metastatic breast carcinoma progression. 1.2 Breast Cancer Progression The target cells for breast cancer formation are located in the epithelial component of the exocrine mammary gland. The organizational structure of this 1 epithelium resembles a cluster of grapes, where the lobular components (also known as alveolar units) are analogous to the fruit and the connecting terminal ducts are analogous to the attached stems. These alveolar units and continuous ducts are known as the terminal duct lobular unit of the mammary gland (TDLU; Figure 1.1). Lobules and terminal ducts are both composed of two epithelial cell layers (McManaman and Neville, 2003). The outer-layer of spindle-shaped myoepithelial cells is contractile in nature. The elongated cell processes of these myoepithelial cells contact the extracellular glycoprotein-rich basement membrane (BM) and they wrap themselves around the cells of the inner luminal layer (Figure 1.1). The luminal epithelial cells are polarized and become secretory cells when they differentiate during developmental cycles of pregnancy and lactation (McManaman and Neville, 2003). The basal surfaces of these luminal cells interact with the B M via the spaces between myoepithelial cell processes while the apical surfaces surround the central lumen of the lobules and ducts (Slade et al., 1999). At the apical surface these luminal cells interact with each other via the apical junction complex, which includes both adherens (AJ) and tight (TJ) junctions (Nguyen and Neville, 1998). This junction complex is often disrupted in metastatic breast carcinoma progression (Parker and Sukumar, 2003). Newly developed techniques such as intravital microscopy, may soon allow direct examination of breast cancer development in animal model systems. However, because of the invasive nature of these procedures, it will be difficult to examine how breast cancer develops from its presumed precursors, at least in humans. Thus it is likely that we will continue to use molecular, cytological and histopathological changes of the breast epithelial cells to further refine the existing hypothetical model of linear progression (Lakhani 1999; Figure 1.2). This model is briefly outlined below and described in detail in next sections. For some time, benign epithelial lesions have come under suspicion as initiators of breast cancer development. "Hyperplasia of usual type" (HUT) lesions are the mildest form of benign lesion where an increase in the number of epithelial cells of the lobules and ducts is observed. Long-term follow-up studies have confirmed that these lesions double the cancer risk (Dupont and Page, 1985; Fitzgibbons et al., 1998). Thus, it has been proposed that HUT lesions progress into clearly pre-malignant "atypical 2 hyperplasias" (AH) where the proliferating cells begin to lose some of the phenotypic characteristics of the normal luminal epithelium (Going, 2003). Histopathology and common molecular changes indicate that A H lesions can further develop into "carcinoma in situ" (CIS), where cells with true malignant potential fill the lumen of the T D L U (Going, 2003). In the next stage of progression, malignant cells break away from the CIS and invade the surrounding stroma. This is known as "invasive carcinoma" (IC). Similar genetic alterations and cytological features often co-exist in CIS and IC (Reis-Filho and Lakhani, 2003). These observations further support the idea that invasive carcinoma does evolve from presumed pre-malignant precursors. However, it is also important to recognize that all the breast carcinomas may not always fit into this linear hypothetical model. There is evidence showing the existence of less genetic alterations in more advanced stages of the disease indicating that these lesions may not have derived in a linear progression. For example Fujii et al (1996) found that 40% of cases with synchronous DCIS and IDC contained heterogeneous patterns of allelic lost at one or more loci in the DCIS components. In contrast to the linear progression model, the parallel progression model indicate the formation of advanced disease from a morphologically normal epithelium. This model is supported by findings of many small invasive cancers that do not accompany atypical components (Deng et al., 1996). Some molecular observations further indicate that breast cancer can potentially follow several different tumorgeneic pathways resulting more complex and heterogeneous pathology. In the following sections I will describe some of the known features of different stages of breast cancer progression as outlined in the above linear model. Specifically, I will focus on the molecular basis of architectural changes that contribute to primary tumor formation and dissemination during malignant progression. 1.2.1 Hyperplasia of Usual Type Hyperplasia of usual type (HUT; Figure 1.2) is an intraluminal proliferation that leads to an increase in the number of cell layers within the ducts and lobules of the T D L U (Page et al., 1995). Because of the cyclical regulation of the mammary gland, T D L U confers vulnerability to processes contributing to the development of breast carcinoma. 3 Regular proliferative bursts require the maintenance of a population of cells with renewal properties, and the gain of growth regulatory factor-independence within this cell population provides a source for hyperplastic growth. Thus, increased rates of cell division and proliferation are predisposed for developing hyperplastic lesions and the further progression into malignant breast carcinoma. There are a number of epidemiological studies consistently showing that high circulating levels of insulin like growth factor-1 (IGF-1) are associated with increased risk of breast cancer development (Peyrat et al., 1993; Enriori et al., 2003; Schairer et al., 2004). The IGF-1 in serum can stimulate cell proliferation in an endocrine fashion while IGF-1 produced by stromal cells adjacent to the T D L U can function in a paracrine fashion to induce cell proliferation (Hankinson et al., 1998; Yee et al., 1989). During normal development, IGF-1 is required for the morphogenesis of the T D L U (Ruan and Kleinberg, 1999). However, mammary glands in transgenic mice that overexpress IGF-1 fail to undergo involution following weaning and remain proliferative (Hadshell et al., 1996). Furthermore, these transgenic mice exhibit both mammary epithelial hyperplasia and increased frequency of mammary tumor formation (Tornell et al., 1992; Bates et al., 1995). In another study, mutation of a growth hormone releasing hormone receptor (lit) results in reduction of IGF-1 to approximately 10% of the normal circulating levels. This leads to a significant reduction of the growth of human MCF-7 breast cancer cells transplanted into lit mutant immunocompromised mice (Yang et al., 1996). A l l of these experimental results clearly suggest that increased IGF-1 signaling may contribute to the transformation of mammary epithelial cells. The function of IGF-1 is mediated via its cell surface receptor, IGF-1R. Blocking the interaction of IGF-1 with IGF-1R, eliminating IGF-1R from the cell membrane, or interrupting the downstream signaling pathways of IGF-1 R, can all block the mitogenic activity of IGF-1 (Khandwala et a l , 2000). Binding of the ligand to the IGF-1R leads to the receptor's intrinsic tyrosine kinase activity, which triggers downstream signaling pathways (Figure 1.3). Two distinct signaling pathways activated by IGF-1R have been identified. One involves the ras/raf/mitogen-activated protein kinase (MAP-K) pathway, while the other involves the phosphoinositol-3-kinase pathway (PI3-K; LeRoith and Roberts Jr, 2003). Insulin receptor substrate 1 (IRS-1) is an intracellular substrate of IGF-4 1, which is also up regulated in hyperplastic lesions (Rocha et al., 1997). IRS-1 binds to the activated IGF-1 receptor and activates the PI3-K pathway. The p85 regulatory subunit of PI3-K can associate with the phosphotyrosine of IRS-1. Once activated, the catalytic p i 10 subunit of PI3-K phosphorylates phosphoinositides to trigger the activation of down-stream effectors such as Akt/PKB. The activation of this pathway plays an important role in cell proliferation as seen in the dependency of PI3-K to induce the proliferation of IGF-1-stimulated MCF-7 cells (Hamelers and Steenbergh, 2003). Furthermore, IGF-1 dependent activation of the PI3-K pathway can also lead to alterations in cell adhesion (Zhao et al., 2003; Guvakova et al., 2003). In one study, stimulation of breast cancer cells with IGF-1 resulted in cell rounding and detachment from the B M in a PI3-K dependent manner (Guvakova et al., 2003). In other cell culture systems such as colorectal cells, abnormally high IGF-1 signaling is also able to destabilize A J (Andre et al., 1999). Thus, the interplay between increased cell proliferation and decreased cell adhesion in IGF-1 signaling may initiate alterations to the architecture of the T D L U that may help to initiate tumor formation. In addition to growth factor-stimulated proliferation, transcription factors also can impinge upon the cell cycle to stimulate cell proliferation. Cyclin DI is a transcription factor that stimulates cell cycle progression during the G l phase of the cell cycle. Several studies show that 11-19% of HUT lesions exhibit increased levels of cyclin DI protein and the frequency further increases with malignant progression (Alle et al., 1998; Zhu et al., 1998; Heffelfinger et al., 2000; Ormandy et al., 2003). Another transcription factor that impinges upon cell cycle regulation is c-myc. c-Myc binds to a specific D N A sequence known as the E-box (CACGTG) to activate gene transcription (Steiner et al., 1995; Amatet et al., 1998). In one study Pechoux et al., (1994), examined 19 HUT lesions and found 15 lesions with increased c-myc protein expression and 17 lesions with c-myc gene amplification. These observations clearly show the role of transcription factor up-regulation in proliferative lesions. In transgenic mice, targeted overexpression of both cyclin DI and c-myc in the virgin mammary glands lead to increased proliferation and an architectural expansion of the T D L U . These mice eventually develop mammary carcinomas (D'Cruz et al., 2001; Rose-Hellekant and Sandgren, 2000; Wang et al., 1994; Sinn etal., 1987). 5 Altered hormone levels and hormone responses can affect cell number and tissue architecture. A number of studies have identified women with early menarche and late menopause (Russo and Russo, 1998) or increased long-term exposure to estrogen (Thomas et al., 1997; Hankinson et al., 1998) to be at increased risk for tumor development. The most active estrogen in breast tissue is 17p-estradiol (E2), which originates mainly from the ovaries. The action of E2 in cells is mediated via estrogen receptors (ER), which belong to a family of nuclear receptors (McDonnell and Norris, 2002). There are two types of ER, ERcc and ERP, which share high sequence homology, specifically in the ligand and D N A binding domains. These two receptors are able to form both homodimers and heterodimers suggesting that three possible complexes could activate downstream signaling pathways (McDonnell and Norris, 2002). Upon ligand binding, ER dimers translocate to the nucleus and then activate the genes containing estrogen response elements (ERE). However, ER can also promote expression of genes that do not have ERE by interacting with other transcription factors such as AP-1 and Spl (Bjornstrom and Sjoberg, 2002; Poter et al., 1997). In the normal mammary epithelium, ER positive cells account for 4-15% of the total and they are scattered among the ER-negative cells (Petersen et al., 1987). In contrast, Shoker et al (1996), found that the percentage of ER positive cells in HUT increases up to 36%, which correlates with the hyperproliferative activity of the lesion. When stimulated by E2, these ER positive cells modulate proteins implicated in HUT lesion development. Specifically, IGF-1, Cyclin D l and c-myc expression are all up regulated in an estrogen dependent manner (Dubik and Shiu, 1992; Planas-Silva and Weinberg, 1997; Westley et al., 1998; Doisneau-Sixou et al., 2003). Thus, estrogen signaling likely exerts a global cell proliferation effect by affecting multiple targets. Furthermore, in vitro studies demonstrate that E2 treated normal mammary epithelial cells on collagen gels form structures that are phenotypically similar to hyperplastic lesions (Russo et al., 2003). Therefore an increase in the percentage of ER positive cells in hyperplastic lesions represents an increase in cell proliferation and initiation of architectural disruption of the T D L U . In addition to proliferation, increased cell survival can also contribute to hyperplasia. Developing tumors can increase survival by inhibiting programmed cell death, a process also known as apoptosis. The tumor suppressor protein, p53, plays a 6 major role in the control of cell cycle progression, D N A maintenance and repair, and the induction of apoptosis (Lundberg and Weinberg, 1999). p53 is a phosphoprotein with an N-terminal transactivation domain, a central D N A binding domain, and a C-terminal negative regulatory domain. Damage to the genome that can occur during G l phase of the cell cycle can lead to a rapid cell cycle arrest, which is mainly regulated by p53 (Ziyaie et al., 2000). One of the target genes of p53 is the p21 gene, the product of which mediates the tumor-suppressing effects of p53 by inhibiting cycle-dependent kinase (Cdk) activity to block G l to S phase transition in the cell cycle (Harper et al., 1993). Once the cells with damaged D N A are growth arrested, p53 is further able to activate the repair mechanisms. If there is a failure to repair DNA-damage, p53 induces apoptosis by down regulating bcl-2, an inhibitor of the process, and this clears cells with damaged D N A (Ziyaie et al., 2000; Miyashita and Reed, 1995; Miyashita et al., 1994). Therefore, p53-dependent apoptosis prevents the further accumulation of mutations that can contribute to the development of cancer. The wild-type p53 protein has a short half-life of about 20min and is virtually impossible to detect in the normal cells. However, mutations that increase the half-life of the p53 often generate a non-functional form of the protein. There are examples of such inactivating p53 mutations in 26.7% of HUT lesions while other studies show 5%-8% of HUT lesions with accumulating p53 protein (Kandel et al., 2000; Thor et al., 1992; Schmitt et a l , 1995; Mommers et al, 1998). In addition to direct defects in p53, there is a report showing anti-apoptotic bcl-2 down regulation in 16% of the HUT they examined (Visscher et al., 1996). The histopathology of these lesions shows an accumulation of cells that are not in contact with the B M suggesting that the cells have lost anchorage-independent cell growth. The form of apoptosis triggered by the loss of cell anchorage is known as anoikis and several studies have determined that loss of p53 inhibits anoikis and enhances cell survival (Bachelder et al., 2001; Ilic et al., 1998; Nikiforov et al., 1996). Therefore, an inhibition of anoikis facilitates the disruption of T D L U architecture by initiating the movement of proliferative cells into the luminal space. 7 1.2.2 Atypical Hyperplasia Histopathologically, atypical hyperplastic lesions are characterized by excessive proliferation of uniform looking cells. These cells form streams, which protrude into the lumen and connect to the other side of the luminal wall. These bridge-like protrusions of cells give a fenestrated appearance to the lesion (Harris et al., 2000; Figure 1.2). In long-term outcome studies, Dupont et al., (1993), have shown that atypical hyperplastic lesions are 5.3 times more likely to develop into invasive carcinomas than typical hyperplasias. This increase in risk further supports the linear progression model. Depending on their site of origin within the T D L U , these atypical lesions can be divided into "atypical ductal hyperplasia" (ADH) and "atypical lobular hyperplasia" (ALH). These lesions are believed to be the precursors of ductal carcinoma in situ and lobular carcinoma in situ respectively (Lakhani, 2003). Some of the cytogenetic abnormalities observed in A D H are also common to those of HUT and invasive carcinoma, further validating the linear model of breast carcinoma progression. Specifically, loss of heterozygosity (LOH) at 16q and 17p is common to all of these lesions (Lakhani et al., 1995; Amari et al., 1999). Because tumor suppressor genes are recessive, cells that contain one normal and one mutated gene leads to heterozygosity. Thus the loss of the normal gene is known as L O H . In 15 polymorphic loci examined, O'Connell's group found that 42% of the A D H cases have at least one L O H event (O'Connell et al., 1998), which is an increase of 5% from HUT. In addition to the 16q and 17p common sites, other L O H events observed include gain of 3p, 8q, 15q and loss of 20q and 13q (Aubele et al., 2000). Although these genetic events are found at high frequencies, one of the major problems with these studies is the small sample size, making it difficult make to valid comparisons between HUT and A D H lesions. As with the L O H events, most of changes in protein expression seen in HUT are also seen in A D H , but at an increased frequency. Thus, the increases in proliferative index observed in atypical lesions may occur because of increases in cell cycle regulator proteins. For example, 43% of the A D H lesions overexpress Cyclin D l compared to the 11% in HUT lesions (Heffelfinger et al., 2000). In addition to increased proliferation, protection from apoptosis is also further enhanced in A D H . Increased accumulation of p53 was observed in a cohort of 4888 women who were diagnosed with benign lesions 8 that later developed in to carcinoma in situ or invasive carcinoma. This led Rohn et al., (1998) to conclude that p53 mutations increase the risk that benign lesions will progress into breast cancer. The 14-3-3-epsilon protein is a recently identified substrate in the apoptosis pathway. In normal cells 14-3-3 is in constant association with the cytoskeleton and with another protein known as Bad. During apoptosis, Bad is released from 14-3-3, which results in its translocation into the mitochondria where it interacts with Bcl-2 to induce apoptosis (Figure 1.4; Won et al., 2003). Umbricht et al., (2001) examined a series of breast lesions and found that 38% of the atypical hyperplasias have lost 14-3-3 by methylation-mediated gene silencing. From this study they conclude that loss of 14-3-3 is an early event in neoplastic transformation that occurs in atypical hyperplasia. Thus, the loss of 14-3-3 may further help to inhibit apoptosis and increase cell accumulation in the atypical hyperplastic lesions. As it also associates with the cytoskeleton, 14-3-3 may also increase architectural disruption in atypical lesions. 1.2.3 Carcinoma in situ With the introduction of mammographic screening, there has been an increase in the identification of pre-malignant lesions, particularly carcinoma in situ (Ernster et al., 1997). These lesions are primary tumors and are considered as precursors that progress into invasive breast tumors (Fitzgibbone et al., 1998). At the histopathological level, carcinoma in situ can be defined as an architectural disruption caused by a complete obliteration of the central lumen of the T D L U (Harris et al., 2000; Figure 1.2). Thus, this tumor phenotype can result from a continuation of atypical hyperplasia. There are two types of carcinoma in situ: ductal carcinoma in situ (DCIS) and lobular carcinoma in situ (LCIS). DCIS resembles epithelial ducts, while LCIS resembles lobules (Harris et al., 2000). However, whether these lesions actually originate from ducts vs lobules respectively still remains unknown. 9 1.2.3.1 Ductal Carcinoma in situ DCIS is the most important non-metastatic lesion, which accounts for up to 25% of newly diagnosed breast cancer cases (Porter et al., 2003). Identification of this leasion results in significantly better prognosis. The histopathology of DCIS represents a heterogeneous group of cell proliferations, with cells often organizing into a cribriform pattern. This pattern is made up of cells separated by oval-shaped, multiple spaces within the lesion (Harris et al., 2000). These cells exhibit variations in shape and nuclear size, as well as a marked increase in the nuclear/cytoplasmic ratio. Based on the cytonuclear features and overall tissue-architectural features, DCIS lesions are classified into low, intermediate and high-grade lesions (Holland et al., 1994). Even with these classifications, DCIS lesions remain controversial because of the difficulties in distinguishing them from A D H lesions. Specifically, the histopathological similarities in low-grade DCIS and A D H lesions make it difficult to separate them with certainty. However, recent advances in molecular genetic techniques have begun to shed light on some of the differences between these lesions. In one study, microdissected lesions show that 50% of L O H events found in DCIS were also commonly found in A D H (O'Connell et al., 1994). This observation suggests not only that there are differences between A D H and DCIS, but also that A D H lesions may accumulate additional genetic changes in order to progress into DCIS. This latter idea is further supported by the L O H , comparative genomic hybridization (CGH) and serial analysis of gene expression (SAGE). These studies show a sharp increase in genetic alterations in DCIS lesions in addition to those already observed in A D H (Adeyinka et al., 2002; Porter et al., 2003). Now one of the major challenges is to accurately identify the genes that are localized at these affected genetic loci, and to characterize the normal protein function of these genes. This could further expand our understanding of the genes that are responsible for architectural disruption of T D L U during breast carcinoma progression. In DCIS lesions, though the normal tissue organization of the ductal epithelium is disrupted during lumen obliteration, malignant cells often manage to maintain cell-cell interactions and remain within the early lesion. For example, there is evidence that early DCIS lesions maintain a modicum of A J and TJ-mediated interactions (Jacobs et al., 2001; Maluf and Koerner, 2001; Asc et al., 2001; Kominsky et al., 2003). However, in 10 the later stages of DCIS, TJs are frequently selectively down-regulated. This down-regulation is clearly evident by loss of the TJ proteins claudin-7 and ZO-1 (Hoover et al., 1997; Cleton-Jansen, 2002; Kominsky et al., 2003). These observations suggest that across the spectrum of DCIS lesions, AJs are often maintained while TJs are progressively lost. 1.2.3.2 Lobular Carcinoma in situ. Lobular carcinomas are often diagnosed in women between the ages of 40 and 50 years, a decade earlier than the diagnosis of DCIS (Reis-Filho and Lakhani, 2003). Because of the lack of clinical abnormalities and detectable lumps in mammography, LCIS is often harder to detect (Sonnenfeld et al., 1991). Histologically, LCIS is often composed of a monomorphic population of cells, which are usually small, round or polygonal-shaped and contain a clear cytoplasmic rim around the nucleus. These loosely cohesive, regularly-spaced cells fill and distend the lobules while maintaining the lobular structure (Middleton et al., 2003; Harris et al., 2000). Unfortunately, cytogenetic analysis on LCIS has been limited. Several studies examining these lesions report losses at loci 16p, 16q, 17p, 22q and gain of 6q in both LCIS and A L C (Lu et al., 1998; Nishizaki et al., 1997; Lakhani et al., 1995). The intermediate or precursor nature of these tumors, in the formation of invasive tumors, is evident in the finding that L O H events, common in LCIS, are also seen in the invasive lobular tumors (Lakhani et al., 1995; Lakhani, 1999). This observation further supports the linear progression model of breast cancer. Contrary to the DCIS where TJs are often lost while AJs are maintained, in LCIS both AJs and TJs are frequently lost (Sneige et al., 2002; Jacobs et al., 2001; Maluf and Koerner, 2001).The AJ protein E-cadherin is a presumed tumor-suppressor gene localized on 16q22.1, a chromosomal region often affected in LCIS. Immunohistochemistry studies show that E-cadherin staining is often absent in LCIS, while it is unaffected in DCIS (Droufakou et a l , 2001; Roylance et al., 2003). Vos et al., (1997) examined LCIS and invasive lobular carcinomas adjacent to each other and found the same E-cadherin truncated mutations in both lesions. This study points out that the loss of functional E-cadherin is critical for LCIS progression, and that these lesions may be direct precursors 11 for invasive carcinoma. It is likely that suppression of E-cadherin is a critical regulator of epithelial-to-mesenchymal transformation (EMT; Hay, 1995). Thus, the loss of E-cadherin expression and/or function seen in LCIS may induce E M T and the subsequent progression from in situ lesion to infiltrating carcinoma. 1.2.4 Infiltrating Carcinoma Infiltrating breast cancers are defined by the ability of the cells to break away from the primary tumor and move into the surrounding stroma. These lesions are invasive and they are considered precursors of distant metastasis (Going, 2003). Cancer metastasis is the primary causes of death by cancer, and for that reason I am interested in the tumor phenotypes of the early-infiltrating breast carcinoma which leads to complete destruction of the T D L U architecture. There are two main types of infiltrating breast carcinomas: Infiltrating-ductal carcinoma (IDC) accounts for about 75% of the currently-detected invasive breast carcinomas and they are believed to have originated from DCIS (Tavassoli and Devilee, 2003). Secretory breast carcinoma (SBC) is a rare subtype of IDC. A less prominent type, invasive lobular carcinoma (ILC) accounts for about 5-15% of the invasive carcinomas and is believed to have originated from LCIS (Tavassoli and Devilee, 2003). These carcinomas show the highest degree of architectural disruption of the T D L U . 1.2.4.1 Infiltrating Ductal Carcinoma In infiltrating-ductal carcinoma, cells break away from the primary tumor as small, disorganized clusters. Architecturally, these tumor cells may be arranged into cords, clusters or trabeculae surrounded by stroma. In some cases, glandular differentiation may be seen with tubular structures that possess internal, luminal spaces (Harris et al., 2000; Figure 1.2). There is neither basement membrane nor myoepithelial cells that surrounds these structures (Barsky et al., 1983). The frequent finding of A D H and DCIS, along with IDC further provides evidence to the linear progression model. These primary tumor cells that invaded the stroma still manage to maintain some of their cell-cell adhesions, however the role of E-cadherin in IDC is not clear. A number 12 of imunohistochemical studies have reported some loss or down-regulation of E-cadherin, although it has been shown to occur mainly in late-stage ductal carcinomas (Gamallo et al., 1993; Berx et al., 1998). In contrast, tight junction proteins as well as epithelial polarity are lost (Hoover et al., 1997; Maluf and Koerner, 2001; Kominsky et al., 2003). I have determined that the anti-adhesive molecule podocalyxin is overexpressed in a subset of mammary tumors that become metastatic. Thus, in this thesis I have attempted to determine whether podocalyxin is capable of perturbing mammary epithelial cell junctions thereby inducing a phenotype similar to that of IDC. 1.2.4.l.a Secretory Breast Carcinoma Secretory breast carcinoma (SBC; Figure 1.2) is characterized by an infiltrating pattern of neoplastic epithelial cells, forming well-differentiated glandular structures. Tumor cells of SBC are typically vacuolated with abundant extracellular and intracellular material that stains strongly with PAS and mucicarmine stains (Siegel et al., 1999; Rosen and Cranor, 1991; Oberman 1980). The tumor cells manage to maintain cell-cell junctions and form multilayered clusters of cells that are surrounded by a basal lamina (Suzuki et al., 1999). These disorganized cell clusters also contain well-developed microvilli, that protrude into a lumen (Suzuki et al., 1999). Furthermore, these cells maintain epithelial characteristics by maintaining keratin expression (Suzuki et al., 1999). The recently identified fusion oncoprotein ETV6-NTRK3 (EN) is found in the majority of SBC (Tognon et al., 2002). Due to the high frequency of fusion events, E N may be functionally important in generating the SBC phenotype, which is characterized by multilayering and simultaneous differentiation. This is an issue I have addressed in my thesis. 1.2.4.2 Infiltrating Lobular Carcinoma ILC is characterized by single cells breaking away from the primary tumor and invading through the basement membrane into surrounding connective tissue stroma in a single-file pattern. As with IDC, ILC often associates with a LCIS component. The great majority of these infiltrating cells have completely lost their cell-cell adhesion and many of their epithelial characteristics. This loss of cell-cell adhesion is a continuation of a 13 process began in LCIS. Thus, it is ILC that appears to undergo EMT, and this is further supported by the loss of E-cadherin that often occurs in ILC (Moll et al., 1993; Berx et al., 1996;Cleton-Jansen, 2002; Becker et al., 2002). The mechanisms that are responsible for the loss of E-cadherin expression are now becoming clear. The gene-encoding E-cadherin protein is known as CDH1, and the loss of this protein can occur through mutation (Berx et al., 1998; Droufakou et al., 2001), somatic L O H and epigenetic down-regulation of protein expression (Sarrio et al., 2003; Cleton-Jansen, 2002; Becker et al., 2002). The mechanisms of epigenetic down-regulation may involve the CHD1 promoter hypermethylation (Hajra and Fearon, 2002; Cheng et al., 2001) and/or transcription repressors binding to the promoter region. The CHD1 promoter has three E-box elements that may be involved in silencing gene transcription during cancer progression. Three transcriptional repressors that are known to bind to the E-box of CHD1 include: the zinc finger protein Snail (Blanco et al., 2002; Cano et al., 2000), Slug (Hajra et al., 2002) and SIP1 (Comijn et al., 2001). It has been shown that up-regulation of these transcription factors and loss of E-cadherin is closely associated with invasive breast carcinoma progression (Blanco et al., 2002). The recently-found metastasis-associated gene (MTA) family of transcription factors (Kumar et al., 2003) are also linked to CHD1 gene regulation. In a recent study, Fujita et al., (2003) demonstrated that CHD1 gene expression is regulated in an estrogen-dependent manner via MTA3 and Snail transcription factors. In the model proposed by Fujita's group, ER activation induces the expression of M T A 3 , which in turn participates in a transcription-repression complex to inhibit Snail expression. In the absence of Snail, E-cadherin expression is maintained. The data presented in this study, however, failed to show direct activation of MTA3 by ER. Therefore, the relationship between ER and MTA3 still remains to be identified. Regardless, they were able to show a positive correlation between ER, MTA3 and E-cadherin protein levels in primary breast cancer specimens. This data can further explain the more aggressive and invasive phenotype seen in ER-negative breast carcinomas. Integrin-linked kinase (ILK) is another molecule that is linked indirectly to suppressing E-cadherin expression. Tan et al., (2001) have shown that overexpression of ILK is able to up-regulate Snail expression in colon carcinoma cells. In the breast, ILK overexpression has been shown to generate invasive mesenchyme-like spindle cell 14 tumors in transgenic mice in vivo (White et al., 2001). Therefore, in this thesis I set out to determine whether ILK is capable of initiating an E M T in mammary epithelial cells. As breast cancer develops and progresses the linear progression model predicts that there is a gradual increase in the disruption of epithelial cell adhesion leading to an architectural disruption of T D L U . The regulation of epithelial cell adhesion in the normal mammary gland has been shown to be relatively dynamic, and it is tightly regulated during the development of the gland. Therefore, I was further interested in exploring the epithelial architecture that is regulated in this developmental context. 1.3 Normal Mammary Gland Development The mammary gland is a highly specialized and modified apocrine sweat gland. At birth, the parenchyma of the gland consists of branching cords of ectodermally derived epithelial cells connected to the nipple through a single primary duct (Sakakura, 1987). Prior to puberty, these epithelial cords cavitate to form ducts with a central lumen. Throughout their length, a sleeve of fibroelastic stroma, except at their terminal ends, surrounds these ducts. The terminal ends are known as end buds and they are in direct contact with the gland's highly-adipocytic stroma. Interactions between the epithelium and the mesenchymally-derived stroma are major determinants of the tissue-specific mammary gland morphogenesis. For example, when the mouse mammary epithelium is combined with the salivary gland mesenchyme, it develops salivary gland ductal patterns (Kratochwil, 1969). Co-cultures of mammary epithelial cells and stromal fibroblasts cause extracellular matrix (ECM) deposition (Reichman et al., 1986). Therefore, it is possible that the E C M deposition induced by the stroma may regulate epithelial cell fate. Prior to puberty, the rate of mammary gland growth parallels the growth of the rest of the body. During puberty however, end buds and connective tissue increase their growth rates under the influence of estrogen produced by the ovaries (Daniel and Silberstein, 1987). Increased cell proliferation at the end buds leads to the elongation of the ductal system. At the onset of pregnancy, as levels of estrogen and progesterone increase, the end buds experience massive cell proliferation, which leads to the 15 displacement of surrounding adipocytic stroma to the gland margins (Topper and Freeman, 1980; Conneely et al., 2003). When the estrogen receptor in mice is deleted, only a rudimentary gland is present at birth and it fails to further develop (Bocchinfuso et al., 2000). Likewise, progesterone injection into the gland induces lateral branching (Atwood et al., 2000), further confirming that ovarian hormones are regulating the development of the gland. Lateral branching of the end buds is responsible for the tree-like pattern of the parenchyma of the adult mammary gland. Dominant-negative mutants of the transforming growth factor-beta (TGF-P; Joseph et al., 1999) or TGF-P receptor (Gorska et al., 1998) in transgenic mice leads to increased lateral branching and hyperplasia of the end buds in the virgin mammary ductal system. Thus, TGF-P functions as a localized epithelial-growth inhibitor to prevent chronic lateral budding and to maintain the normal pattern of branching during ductal morphogenesis. TGF-P expression is regulated at the transcriptional level, by the specialized basement membrane extracellular matrix (ECM) that surrounds the mammary epithelium (Streuli et al., 1993). In transgenic mice in which the expression of an autoactivated form of the ECM-degrading metalloproteinase stromelysin-1 is targeted to the mammary epithelium, the basement membrane is constantly remodeled and end bud-hyperbranching occurs (Sympson et al., 1994; Sternlicht et al., 1999). Therefore, an intact basement membrane is a critical regulator of early mammary-specific morphogenetic events. Early in the pregnancy, the proliferating cells of the end buds undergo a secondary spatially-restricted branching that is regulated by hepatocyte growth factor (HGF; Nermann et al., 1998; Kamalati et al., 1999). HGF is another locally-acting growth factor produced by mammary mesenchymal cells, that is sequestered in the E C M and acts by binding to the c-met receptor tyrosine kinase on the surface of the epithelial cells. In organ cultures, antisense-oligonucleotide-suppression of HGF completely abolishes the branching morphology (Yang et al., 1995). Conversely, overexpression of HGF and or the c-Met receptor in transgenic animals leads to enhanced end bud size and number, inducing hyperplastic-branching morphology (Yant et al., 1998). The cell-cell adhesion in these proliferating mammary epithelial cells is shown to be dynamic during the early stages of development. Immunofluoresence and electron 16 microscopy studies of the mammary bud have shown that cell-cell adhesion complexes are hardly detected in the early proliferative stages (Nanba et al., 2001). Furthermore, "spot-weld" like cell-cell adhesion junctions, desmosomes, are down-regulated in the early stages (Nanba et al., 2001). This lack of adhesion is conceivably necessary for cellular reorganization during morphogenesis. As the cell masses grow to elongate and form lumen, junction proteins are re-expressed. For example, TJ protein occludin is expressed at the same time that the lumen forms (Nanba et al., 2001). Later in pregnancy, the end buds expand to form alveoli, a process which is under the control of the neu differentiation factor (NDF; Neimann et al., 1998). NDF, another morphogen produced by the mammary mesenchyme, is sequestered in the E C M and acts on epithelial receptors, in this case, the erbB family of tyrosine kinases receptors. In mammary organ cultures derived from pregnant mice, NDF-antisense-oligonucleotides inhibit alveolar morphogenesis, a phenotype that can be rescued by the addition of recombinant-NDF. Moreover, transgenic virgin mice overexpressing mammary-targeted NDF undergo premature end bud expansion and alveolar morphogenesis (Krane and Leder, 1996). Importantly, in all cases, when alveolar morphogenesis is affected, epithelial differentiation is similarly affected. Thus, when NDF-mediated morphogenesis is blocked, differentiation does not occur. An extensive branching and terminal differentiation of the mammary epithelial cells finally occurs during the full-term pregnancy. Mammary epithelial differentiation is marked by the induction of milk protein gene expression, which is regulated by NDF, lactogenic hormones and the basement membrane (Simpson et al., 1998; Howlett and Bissell, 1993; Teng et al., 1989; Lee et al., 1998). The major mouse milk proteins are: firstly, the iron-binding proteins, including lactoferrin, that are expressed in early to mid pregnancy; secondly, the caseins, including P-casein, that are first expressed in mid to late pregnancy; and thirdly, the whey proteins, including whey acidic protein (WAP) which is first expressed in late pregnancy. In a less dramatic manner, similar changes occur during the adult estrus cycle. Because of this cyclic nature in the remodeling of the mammary epithelium, there is a greater probability of alterations that can occur to these normal remodeling mechanisms. Hormones, extracellular matrix molecules, cytokines and growth factors play a major role 17 in each of these remodeling stages and changes to any of these pathways can lead to malignancy. Thus, the mammary gland is specifically vulnerable for initiation of cancer progression. 1.3.1 Structure of the Terminal Duct Lobular Unit The branching terminal ducts and terminal alveolar units of a mammary gland are known as the terminal duct lobular unit (TDLU; Figure 1.1). They consist of two epithelial cell layers: the inner luminal epithelial layer and outer myoepithelial layer. Secretory and lactational changes take place primarily in the cuboidal to columnar shaped luminal epithelial cells. Myoepithelial cells do not contact with the luminal surface, instead they interact with the B M and wrap around the luminal cells. These myoepithelial cells are not well developed in the ductal areas compared with the lobules. The function of myoepithelial cells is well recognized, as assisting milk ejection during suckling, in response to oxytocin (Lakhani and O'Hare, 2000). In addition to this function, myoepithelial cells are also known to assist in growth and differentiation of the T D L U . Activin, a member of TGF-P family of growth factors, is known to be exclusively expressed by myoepithelial cells and assists in ductal formation in vivo and further, it inhibits growth of the MCF-7 mammary tumor line (Liu et al, 1996). More recently Jones et al., (2003) demonstrated that human primary myoepithelial cells are able to inhibit breast cancer cell invasion. These observations demonstrate that myoepithelial cells are involved in maintaining the normal-polarized epithelial structure of the T D L U . 1.4 Apical-Basal Polarity in Epithelial Cells. 1.4.1 What is apical-basal polarity? Cell polarity is defined as asymmetry in a cell or in a tissue and the most fundamental type of tissue organization is the formation of a polarized epithelium. A polarized epithelium participates in two major functions; 1) It delineates separate compartments, and 2) maintains homeostasis by regulating the exchange of molecules 18 and ions between the two compartments. In order to generate these special compartments, polarized epithelial cells must have a distinctive cell shape, organization of membrane proteins, oriented alignment of the cytoskeleton and uneven distribution of organelles. Most of these compartments organize into ducts, acini or sheets of cells facing a free apical surface that is continuous with the environment and a basal surface in contact with a B M . Thus, these cells are apically/basally polarized. This polarization is achieved by generating three different types of plasma membrane surfaces; 1) a free, apical surface that faces the lumen or out side, 2) a lateral surface where cells interact with each other, and 3) a basal surface where cells interact with the B M (Figure 1.1; O'Brien et al., 2002). There are two main control points that regulate the formation of these membrane domains. First the apical junction complex, which is conserved throughout evolution and is responsible for initiating and maintaining cell-cell contacts as well as regulating the composition of different membrane domains. These cell-cell adhesions are essential in restricting the movement of membrane proteins and initiating an asymmetry in the distribution of membrane proteins. Second, cell-BM interactions where the integrin superfamily of cell surface adhesion receptors interact with the B M glycoproteins to define the basal membrane. Cell adhesion to the B M is particularly important in organizing the apical to basal axis of polarity. This axis also defines the orientation of the apical and basolateral surface and provides the direction for secretions and solute transport. 1.4.2 Apical Junction Complex The assembly of the apical junction complex is one of the early events in generating a polarized epithelium. The main components of the apical junction complex consist of AJs and TJs (Figure 1.5), which are localized at the boundary of the apical and lateral membrane domains. Both of these specialized junctions have a common organization with transmembrane proteins linking to the actin cytoskeleton via modules of cytoplasmic scaffolding protein complexes. These scaffolding complexes consist of proteins that contain multiple protein-protein interacting domains, which will be discussed later. •J 19 1.4.2.1 Adherens Junctions Initial epithelial cell-cell interactions lead to the formation of AJ . At the ultrastructural level, AJs are seen as closely apposed plasma membrane domains with electron dense cytoplasmic plaques where actin filaments are attached to the membrane. At the molecular level, E-cadherin mediates AJ formation. E-cadherin is a transmembrane glycoprotein that mediates calcium dependent, homophilic binding at A J (Takeichi, 1988). The essential role of E-cadherin in the formation of epithelium was first identified in the chicken liver, and was named liver-cell adhesion molecule ( L - C A M ; Gallin et al., 1987). E-cadherin is a 120 kDa protein with a large extracellular domain, a single transmembrane domain and a short cytoplasmic domain. The extracellular domain consists of five tandem repeats of 100 amino acid motifs with CSL^ binding pockets localized between the repeats. Cell-cell adhesions are formed by homotypic interactions of these extracellular domains in an interdigitating manner to give a zipper-like appearance. The cytoplasmic domain of E-cadherin is linked to the actin cytoskeleton via a catenin protein complex to strengthen the cell-cell adhesion. This multiprotein complex is thought to provide the stable framework for AJs. E-cadherin expression plays a major role in epithelial differentiation and polarization during development. When the E-cadherin gene is deleted in mouse embryonic stem cells, heterozygous mutant animals grow normally while homozygous mutants fail to maintain polarized compacted pre-implantation embryos (Rietmacher et al., 1995). (3-Catenin was discovered as a co-precipitating partner of E-cadherin, and has recently gained more attention as a critical member of the Wnt signaling pathway (Gumbiner, 1996; Cox et al., 1999). In this signaling pathway, cytoplasmic P-catenin is translocated to the nucleus where it functions as a co-transcriptional regulator. In epithelial cells, at the A J , P-catenin binds directly to the cytoplasmic tail of E-cadherin. a-Catenin is able to link the actin cytoskeleton to P-catenin (Imamura et al., 1999). PI 20-catenin binds to a membrane proximal cytoplasmic domain of E-cadherin and is believed to be involved in E-cadherin clustering at the adherens junction (Yap et al., 1998). The fully functional adhesive properties of A J depend on the integrity of E-cadherin-catenin-actin complexes, and loss of this integrity can lead to loss of cell polarity and an increase 20 in motility and invasive properties of the cells. Analysis of many rumors derived from epithelia suggests that the loss of E-cadherin correlates with the tumor cell invasion (Rasbridge et al., 1993; Sommers et al., 1991) and ectopic expression of E-cadherin in invasive cells is sufficient to reverse this phenotype (Chen et al., 1997). Thus, loss of E-cadherin could be a major event in enhancing tumor cell invasiveness. In addition to E-cadherin mediated A J formation, there is a newly discovered adhesion system mediated by the immunoglobulin-like molecule, nectin. In contrast to E-cadherin, nectin is a Ca** independent transmembrane receptor. Currently there are four known members (1-4) in the nectin family. The molecular structure consists of three extracellular IgG domains, a single transmembrane domain and a cytoplasmic domain (Takai and Nakanishi, 2003). The extracellular domains form cis-dimers and then trans-dimmers to form a zipper like interaction as seen with E-cadherin (Miyahara et al., 2000; Satoh-Horikawa et al., 2000). Unlike E-cadherin, nectin is capable of forming heterodimmers with other family members (Satoh-Horikawa et al., 2000). The cytoplasmic domain of nectin interacts with the f-actin binding protein afadin providing a link to the actin cytoskeleton (Takahashi et al., 1999). This nectin-afadin interaction is mediated through the PDZ domain (see below) of afadin (Takahashi et al., 1999; Reymond et al., 2001). Nectin overexpression in M D C K cells leads to the recruitment of E-cadherin to the nectin-based interactions (Honda et al., 2003). This suggests a close relationship between E-cadherin and nectin mediated adhesions. 1.4.2.2 Tight Junctions Tight junctions (TJ) are generally the most apical structures of the apical junction complex. TJs form a continuous belt at the boundary between the apical and basolateral membrane domains of neighboring epithelial cells which link to the actin cytoskeleton. At the ultrastructure level, they are characterized by the fusion of the exoplasmic leaflets of the plasma membrane to form a seal between cells (Farquhar and Palade, 1963). TJs regulate the passage of small molecules and ions across the paracellular boundaries by functioning as a barrier and they maintain different compositions of the apical and basolateral membranes by functioning as a fence (Gumbiner, 1987, 1993; Rodriguez-Boulan and Nelson, 1989). 21 At the molecular level TJs are composed of at least three transmembrane proteins: occludin, claudin and junctional adhesion molecule (JAM). Occludin and the family of claudin proteins contain four transmembrane domains and have both amino and carboxy terminal ends oriented towards the cytoplasm. Unlike other transmembrane TJ proteins, the last 150 amino acids of the carboxy terminus of occludin directly interacts with the F-actin cytoskeleton (Wittchen et al., 1999). Occludin contains possible phosphorylation sites on serine, threonine and tyrosine groups, which results in a molecular weight range of 62-82 kDa (Sakakibara et al., 1997; wong, 1997; Wong and Gumbiner, 1997; Chen at al., 2002). In cultured epithelial cells, highly phosphorylated occludin localizes at the TJ while less or unphosphorylated protein is observed in the cytoplasm, which could be associated with vesicles (Andreeva et al., 2001; Tsukamoto and Nigam, 1999; Sakakibara et al., 1997). In contrast, in the Xenopus embryos, dephosphorylation of occludin correlates with the TJ assembly (Cordenonsi et al., 1997). These opposing effects may be due to different roles for occludin in TJ assembly as well as the difference in phosphorylated groups. Interestingly, occludin knockout mice display well-developed TJs indicating a possible redundancy of the transmembrane TJ proteins (Saitou et al., 2000). The second group of transmembrane proteins, claudins, consists of 24-family members (Tsukita et al., 2001). When claudins are expressed in fibroblasts (L-cells), which lack TJs, the cells then form TJs which display a freeze-fracture replica pattern that is similar to normal TJs (Furuse et al., 1998; Morita et al., 1999). Furthermore, expression of different claudins leads to different freeze-fracture patterns indicating possible tissue specificity. Hoevel et al., (2002) have demonstrated that re-expression of claudin-1 is sufficient to restore the TJs in human mammary tumor cell lines. Claudin-1 knockout mice lose the skin epithelial cell (keratinocyte) TJs and thereby lose cell polarity (Furuse at al., 2002). These gain-of-function and loss-of-function studies further confirm the integral role of claudins in TJ formation. A third transmembrane protein, junctional adhesion molecule (JAM) is a 43 kDa glycosylated protein with two extracellular immunoglobulin domains, a transmembrane domain and a short cytoplasmic domain (Martin-Padura et al., 1998). When exogenously expressed in CHO cells, J A M localizes to the newly formed TJs (Martin-Padura et al., 1998; Aurrand-Lions et al., 2001). Furthermore, blocking antibodies against J A M block 22 TJ formation while failing to effect fully formed TJs (Liu et al., 2000). This suggests that J A M may play a major role in junction assembly but not specifically in maintaining the junction. The cytoplasmic domains of occludin, claudin and JAMs form interactions with the underlying scaffolding protein complexes (Figure 1.5). In addition above described transmembrane proteins, number of TJ-associated cytoplasmic proteins also have been identified. The 225 kDa phosphoprotein ZO-1, was the first TJ protein to be identified (Stevenson et al., 1986). cDNA sequence analysis of ZO-1 indicates homology to lethal discs-large (dig) tumor suppressor gene of Drosophila (Itoh et al., 1993) and postsynaptic density protein PSD-95/SAP-90 from rat brain (Willott et al., 1993). Hence, they were named as PDZ domain proteins (see section 1.4.2d; See Figure 1.6 for PDZ domain containing TJ proteins). Since the initial identification, two other ZO proteins ZO-2 (Jesaitis and Goodenough, 1994) and ZO-3 (Haskins et al., 1998) have been identified as members of membrane associated guanylated kinase ( M A G U K ) family of proteins that contain structurally conserved PDZ, SH3 and G K domains (See section 1.4.2d). ZO-1 directly interacts with the carboxy terminal of occludin through its G K domain (Fanning et al., 1998; Schmidt et al., 2001), with claudin through its PDZ1 domain (Itoh et al., 1999) and with J A M through its PDZ1/PDZ2 domains (Ebnet et al., 2000). Furthermore, the cytoplasmic tail of ZO-1 interacts with the actin cytoskeleton as well as the actin binding protein, 4.1. In the light of these interactions with three transmembrane TJ proteins and its ability to link with the cytoskeleton, it is reasonable to consider ZO-1 as a tumor suppressor protein that maintains a normal polarized epithelium. This model further supported by the observation that the loss of ZO-1 is coupled to breast cancer progression (Hoover et al., 1998) and that IGF-1R induced cell-cell adhesion in breast carcinoma cell line MCF-7 depends on ZO-1 expression (Mauro et al., 2001). ZO-2 is a 160 kDa protein and was first identified as a ZO-1 binding protein (Gumbiner et al., 1991). It is now known that ZO-2 can interact with claudin by its PDZ1 domain (Itoh et al., 1999) and with occludin and another TJ protein, Cingulin, by its G K domain (Itoh et al., 1999; D'Atri et al., 2002). The proline-rich carboxy terminus interacts with the actin cytoskeleton (Wittchen et al., 1999) and actin binding protein, 4.1 (Mattagajasingh et al., 2000). Furthermore, ZO-2 can also interact with the A J protein, a-23 catenin (Itoh et al., 1999). Several groups have suggested a possible role of ZO-2 as a tumor suppressor protein. This suggestion is supported by ZO-2's ability to inhibit the. neoplastic growth induced by polyomavirus middle T protein and adenovirus type 9 oncogenic determinent E4 (Glaunsinger et al., 2001). Furthermore, there is significant reduction of ZO-2 levels in breast cancer cell lines as well as adenocarcinomas (Chlenski et al., 2000). ZO-3 is a 130kD protein initially identified as a ZO-1 /ZO-2 binding protein (Balda et al., 1993). This M A G U K protein interacts with claudin via the first PDZ domain (Itoh et al., 1999) and with occludin via both its amino and carboxy termini (Haskins et al., 1998). Unlike ZO-1 or ZO-2, the amino terminal of ZO-3 interacts with the actin cytoskeleton (Wittchen et al., 2000). When the amino terminal half of the protein is transfected into cells, the TJ assembly is delayed, indicating its crucial role in junction assembly (Wittchen et al., 2000). Another member of the M A G U K family that associates with TJ is Palsl. This protein was originally identified as a PDZ-domain-containing molecule that interacts with Lin-7. Lin-7 is a C. elegans growth-factor-receptor-basolateral-membrane-targeting molecule that localizes at the TJ (Kamberov et al., 2000; Kaech et al., 1998; Roh et al., 2002). Pals is targeted to the TJ by binding to PATJ and MUPPI, two recently identified mult-PDZ domain containing scaffolding proteins. PATJ contains 10 PDZ domains and is found to be concentrated at TJs. When overexpressed in M D C K cells, PATJ leads to mislocalization of ZO-1 and ZO-3 indicating a possible role in TJ integrity. The Drosophila homolog of PATJ interacts with Crumbs to regulate cell polarity. Like PATJ, MUPPI is a multi-PDZ domain protein, which has 13 PDZ domains. This large number of PDZ domains makes MUPPI a very likely candidate to function as a scaffolding protein. Its exclusive localization to the TJ and interaction through its PDZ domain 10 with claudin and the PDZ domain 9 with J A M (Hamazaki et al., 2002) suggests a possible role in cross-linking claudin and J A M based junction strands. Overexpression of MUPPI in M D C K cells increases the transepithelial resistance suggesting that it may increase the strength and stability of the TJ (Jeansonne et al., 2003). The newly identified protein-protein interaction domain of MUPPI , M A G U K recruitment domain (MRE) is involved in its interaction with Pals (Roh et al., 2002). At the TJ, Pals forms a complex 24 with PATJ, MUPP1 and human Crumbs homologue CRB3 which mediate their interaction. In Drosophila, Crumbs and Disk lost (Dit) play a major role in apical polarity determinants (Roh et al., 2002). The interaction of ZO-3 and PATJ is necessary to localize PATJ to the TJ. Thus, ZO-3 could serve to recruit CRB3/Pals/PATJ complex to the TJ. Alternatively CRB3/Pals could be recruited to the TJ via MUPP1 interaction with J A M and claudin. This further emphasizes the role of the PDZ-domain scaffolding proteins in assembling the junction as well as organizing other cell polarity determinants at the junction. 1.4.2.3 Cell polarity proteins Partitioning-defective proteins (PAR) are required for the proper establishment of embryonic cell polarity in C. elegans. In epithelial cells, PAR-3 is a three PDZ-domain-containing protein that is localized to the TJs and interacts with the carboxy terminus of J A M (Itoh et al., 2001). Furthermore, PAR-3 also interacts with the other P A R family protein PAR-6, which contains a single PDZ domain (Johansson et al., 2000) and with atypical protein kinase C (aPKC; Izumi et al., 1998). When a dominant negative aPKC is expressed in M D C K cells, PAR3 mislocalizes and disrupts cell polarity, indicating the significance of this polarity protein complex (Suzuki et al., 2001). Yamanaka et al., 2001, demonstrate that the complex formation between PAR3-aPKC and PAR6 is required for initiation of TJ formation in epithelial cells. Furthermore, they demonstrate that a PDZ-domain mutant form of PAR-6-expressing M D C K cells are unable to form cell-cell contacts and thereby induce TJ formation or development of TER. Contradictory to these observations Gao et al., 2002 find that PAR-6 overexpression blocks the reassembly of TJs without affecting A J assembly after the junctions are disrupted by Cdt* depletion. These observations suggest further studies are required to determine the significance of these protein interactions. Interestingly, studies in Drosophila tell us that the localization of three proteins at the junctions is co-dependent on each other (Tabuse et al., 1998). PAR-6 is also a binding partner for Rho GTPases Cdc42 and Racl , which are major players in organization of the actin cytoskeleton (Johansson et al., 2000). Activated Cdc42 and Rac directly bind to the Cdc42/Rac interactive binding domain (CRIB) of PAR-6 (Lin et al., 2000). This observation suggests that PAR-6 is the major adaptor 25 linking Cdc42, Racl and aPKC to PAR-3 thereby localizing the complex at the TJ. Cdc42 interaction with PAR3-aPKC-PAR6 leads to increased activity of aPKC (Yamanaka et al., 2001), which supports the observation that activated Cdc42, is able to disrupt TJs (Gao et al., 2002). 1.4.2.4 M A G U K Proteins The Membrane-associated-guanylate-kinase family of scaffolding proteins is an important regulator of epithelial cell polarity. This family of proteins is distinguished based on the presence of three conserved protein-protein interaction domains PDZ, SH3 and GK. PDZ domains PDZ-domains are one of the most abundant protein-protein interaction domains in the eukaryotic genome. Initially these domains were identified as 80-90 amino acid residues in three proteins; the postsynaptic density protein PSD-95, Drosophila septate junction protein disc-large, and zona occludins protein ZO-1 (PDZ). Although PDZ-domain containing proteins do not have transmembrane domains, the majority of them are associated with other cell membrane proteins to organize and localize scaffolding protein complexes for cell adhesion and signal transduction. Class I PDZ domains recognize proteins with carboxyl terminal motifs that contain the S/TXIV amino acid sequence while class II PDZ domain recognize carboxyl domains with hydrophobic amino acids at the -2 position (Gonzalez-Mariscal et al., 2000). Furthermore, PDZ-domains are also able to form dimers increasing the scaffolding potential of PDZ-proteins. SH3 domain SH3 domains, also known as Src-homology-3 domains, consist of noncatalytic , 50-70 amino acid segments that facilitate protein-protein interactions. Originally, they were found in the tyrosine kinase oncoprotein V-Src (Dalgarno et al., 1997). SH3 domains specifically interact with G K modules or with four amino acid regions that have P X X P ( proline rich) sequences. 26 The Guanvlate kinase (GK) domain The G K modules are homologues of the enzyme that is responsible for catalyzing the ATP dependent GMP to GDP conversion reaction. However, in the M A G U K proteins there is no detected catalytic activity nor predicated GMP or ATP binding sites. Instead the GK-domain functions as a protein-protein interaction domain to provide a scaffolding function. The ability of these domains of M A G U K proteins to interact with each other as well as with proteins containing binding partners provides great flexibility in assembling the TJ. These domains not only provide structural functions, they also are able to recruit and facilitate the interaction of signaling molecules at the TJ. 1.4.3. Interactions with the Basement Membrane Generation of the appropriate apical junction complexes, distinctive membrane domains, and cell polarity must also be coupled to the overall polarized organization of the resulting tissue. In this context, epithelial cells must also orient themselves with the apical membrane at tissue lumens and the basal membrane towards the B M . Thus, the cell-BM contacts play a critical role in establishing apical/basal polarity in epithelial cells. There are several studies demonstrating the significance of B M proteins and cell surface integrin receptors in normal epithelial morphogenesis (Wang et al., 1990; Weaver etal., 1997, Schuger, 1997). 1.4.3.1 Basement Membrane The majority of normal epithelial cells are in continuous contact with the B M . The B M is a specialized extracellular matrix (ECM) that forms between epithelial cells and the surrounding stroma. The B M also provides a major barrier for cell invasion into surrounding stroma. Major components of the B M include laminin, collagen IV, perlecan, nidogen and entactin (Engbring and Kleinman, 2003). Over a decade ago, Wang et al., (1990) demonstrated that culturing M D C K cells in collagen gels induce the endogenous secretion of E C M proteins at the same time they form the central lumen and 27 localize TJ proteins at the apical membrane domains. Furthermore, it is known that M D C K cultures deposit laminin just prior to localizing TJs at the apical membrane (O'Brien et al., 2002). In addition, laminin is a major ligand that binds to integrins and has been demonstrated to be crucial for normal mammary epithelial morphogenesis (Roskelley et al., 1995; Streuli et al., 1995; Stahl et al., 1997). These observations suggest that laminin may be a major player in the formation of a polarized epithelium. Laminin is a 200-400 kDa glycoprotein. It was discovered over 20 years ago to be localized in the matrix of a murine sarcoma (Mouse EHS sarcoma). This molecule is composed of three disulphide linked chains a,p and y organized into a cross shape. There are five a, three P and three y chains forming 12 distinct isoforms of laminin (Colognato and Yurchenco, 2000). These contain common globular domains that allow chain polymerization, epidermal growth factor-like repeats with nidogen-binding sites and carboxy terminal a chains that allow binding to cell surface integrin receptors. The interaction with nidogen facilitates the cross linking of laminin to the type IV collagen network (Engbring and Kleinman, 2003). Laminin isotypes are expressed in a wide range of tissues in a tissue-specific manner. It is important to point out that laminin is expressed in virtually all types of epithelial cells and the primary role of laminin is cell attachment to the B M (Aumailley and Smyth, 1998). 1.4.3.2 Integrins and Dystroglycan Integrins are a family of heterodimeric transmembrane receptors that have overlapping and competing affinities for extracellular matrix proteins (Hynes, 2002). Combinations of 16a and 8P chains allow for the formation of at least 24 different heterodimeric isoforms of the receptor. Integrins generally contain large extracellular domains formed by a (~ 1000 amino acids) and P (~750 amino acid residues) subunits and these contain relatively short cytoplasmic domains with the exception of P4 integrin with has a cytoplasmic domain that contains > 1000 amino acid residues (Hogervorst et al., 1990; Suzuki and Naitoh, 1990; Tamura et a l , 1990). a2p i , a3pi , a6pl and a6p4 are laminin-binding integrin heterodimers that are all expressed in the normal breast tubule (Gui et al., 1997). In response to ligand binding, p i integrins cluster and form a multiunit plaque of signaling molecules and specialized cytoskeletal components 28 accumulate at the cytoplasmic tail of the integrin (Zamir and Geiger, 2001; Gimond et al., 1999). Formation of this structure is known as "outside-in" signaling as it is dependent on the binding ligand to the extracellular domain of integrin. The signaling molecules in the adhesion plaque are involved in transducing subsequent post-receptor signals to the cell as well as rearrangements of the cytoskeleton. These signaling events include activation of focal adhesion kinase, src family kinases, M A P kinases, PI3-K and integrin-linked kinase, all of which are involved in ECM-dependent processes such as cell migration, cell proliferation, suppression of apoptosis and morphogenesis. Integrins are also involved in "inside-out" signaling, where changes in the cytoskeleton influence integrin-ligand binding and thereby regulating integrin function. Integrin engagement reinforces the f-actin cytoskeleton to further strengthen the adhesion (Calderwood et al., 2000). When blocking antibodies against P1 and ct6 integrin receptors are applied to the colon cancer line, Caco-2, a disruption of AJs and polarized actin organization are observed (Schreider et al., 2002). During breast tumor progression it is believed that loss or alteration of signals mediated by a6p4-integrin lead to tissue disorganization and increased invasiveness (Weaver et al., 1997; Shaw et al., 1997). Almost all the integrins engage the actin cytoskeleton, with the exception of P4. Interestingly, ct6p4 integrin initiates the formation of hemidesmosomal junctions that link laminin to the intermediate filament cytoskeleton. In the normal mammary gland, hemidesmosomes anchor epithelial cells to the B M , prevent cell migration and induce differentiation (Muschler et al., 1999). When a cytoplasmic tail truncated form of p4 integrin is expressed in mammary epithelial cells they fail to undergo normal morphogenesis and polarization (Weaver et al., 2002). Since these cells do not express ct6pi integrin, it possible that a6p4 is playing an essential role in epithelial polarization. However, the downstream events following a6p4 integrin engagement still need to be elucidated. These results further confirm the role of B M -integrin interaction in generating the apical/basal polarity of epithelial cells. A second cell-BM attachment process is mediated through the cell surface glycoprotein, dystroglycan (DG). Like a6p4 integrin, this transmembrane receptor also binds to the B M through its interaction with laminin. D G links the B M to the actin cytoskeleton via an interaction with the actin binding protein dystrophin. Until recently, the role of D G was not well understood. R N A i studies show that D G is required for the 29 establishment of apical/basal polarity in the Drosophila epithelium (Deng et al., 2003). In addition, D G expression is down regulated in breast and prostate tumors. Furthermore, D G negative mammary tumor cells that fail to undergo polarization in 3D cultures, regain the capability to undergo polarization when D G is reintroduced (Muschler et al., 2002). Interestingly, these cells have significantly less potential to form tumors when injected into nude mice. Thus, these observations suggest that D G is an inducer of epithelial polarity in response to interactions with the B M . As with integrins, the down stream mechanisms leading D G to induce polarity need to be identified. 1.4.4 Assembly of the Apical Junction complex and Cell Polarization. With a basic understanding of how cell junctional and polarity proteins function, we can begin to build a hypothetical model of the events that lead to TJ formation and cell polarization. The assembly of the apical junction complex is generally triggered by cell-cell interactions that are mediated by nectin and E-cadherin. This initial interaction leads to the recruitment of other AJ and TJ components to "primordial junctions". This recruitment can be mediated by numerous protein-protein interactions that are not just limited to each specific junction. At the initial stages, the TJ protein, ZO-1, is able to interact with the A J protein, a-catenin, as well as nectin (Yokoyama et al., 2001). ZO-1 acts as a major scaffolding protein that is capable of recruiting other ZO-1 proteins to these interaction sites. Specifically, occludin, claudin and J A M can localize to the cell membrane via their association with ZO-1. Furthermore, nectin is also involved in recruiting JAMs to these contact sites (Fukuhara et al., 2002). With the maturation of these junctions, occludin, ZO-1, ZO-2 and ZO-3 gradually accumulate at spot-like r • junctions which then separate out from E-cadherin based adherens junction to form two independent junction complexes, AJs and TJs (Ando-Akatsuka et al., 1999). Interestingly, CRB3 is only observed in the cytoplasmic regions during the primordial junction formation (Rho et al., 2003). As the junctions mature, Pals/PATJ become localized to the junction (Rho et al., 2003). The interaction between JAMs and PAR3 facilitates the localization of PAR3-aPKC-PAR6 complexes at the junction. This assembly of the PAR3-aPKC-PAR6 complex at the junction and interaction with J A M 30 (Ebnet et al., 2001) is required for the formation of a polarized TJ. When J A M is ectopically expressed in CHO cells they recruit PAR3 to the cell-cell contacts while dominant-negative J A M results in redistribution of PAR3 away from the membrane. In M D C K cells Cdc42 plays an essential role for generating a polarized epithelium (Kroschewski et al., 1999). The activity of Cdc42 allows A J stabilization by actin cytoskeleton organization as well as activation of aPKC (Joberty et al., 2000). In wound healing assays, however, aPKC activity is not required for the primordial junction formation, but is necessary for the maturation of the apical junction complex in separating AJs and TJs (Suzuki et al., 2001). Thus, the activation of aPKC-PAR3-PAR6 complex via Cdc42 activation of aPCK may well be the promoter of TJ polarization. Finally, CRB3 is localized to the apical region and interacts with the PATJ/Pals complex to generate the fully formed and stabilized TJ complex in the polarized epithelium. Finally, integrin-laminin interactions are essential for generating a fully polarized TJ (Weaver et al, 2002), although it is unclear at what point of polarization these interactions are essential. In such a model, E-cadherin together with nectin-based primordial contacts might represent positional cues, leading to the appropriate subcellular localization of the TJs and further, functioning as a scaffold to localize and subsequently activate the cell polarity associated proteins to polarize the TJ. 1.5 Thesis Problem During metastatic tumor progression, cell-cell adhesion is often disrupted and cell polarity is lost. In lobular breast tumors, E-cadherin expression is often suppressed allowing single cells to break away and invade into surrounding stroma. Interestingly, ductal carcinomas display a different scenario, where E-cadherin may or may not be present during loss of cell polarity. Thus I hypothesized that loss of cell polarity, not cell-cell adhesion per se, is a common, unifying, feature of malignant breast tumor progression. Therefore, the goals of this thesis were: 1) Tto use a 3-dimensional culture model system (Roskelley et al., 1994) that closely mimics alveolus-like hierarchical and developmentally regulated morphogenesis of the mammary epithelium to gain a better understanding of how polarized TJs are formed in this tissue. 2. To use use this model to 31 identify possible candidate architectural genes that can disrupt cell adhesion and or TJ polarity and thereby contribute to metastatic breast cancer progression. In this culture system, when functional mouse mammary epithelial cells (Scp2) are placed on a tissue culture substratum they form an epithelial monolayer with A J but lacking TJs. When these cells are placed on a reconstituted-BM gel (Matrigel), they form small aggregates and pull the matrix around them. Over time these aggregates cavitate and become apically/basally polarized with TJ that surround and cage the central lumen. The result is a fully functional morphogenetic unit, which closely resembles the lactating alveolus in vivo. Similarly, when M D C K cells are placed on a collagen gel or on Matrigel, they form polarized spheroids with apically localized TJs. However, unlike the M D C K model where cells are also able to synthesis their own laminin, Scp2 mammary epithelial cells do not deposit endogenous E C M proteins (Desprez et al., 1993). Instead, they are completely dependent on exogenously added E C M for morphogenesis. This has allowed me to further dissect this model to examine TJ formation independent of TJ polarization. As I have described previously, architectural disruption of the epithelium is increased with the progression of breast cancer. It's also known that ETV6-NTRK3 fusion protein (fusion between ETS transcription factor ETV6 and protein tyrosine kinase domain of the neurotrophin-3 receptor) is expressed in SBC, while Podocalyxin and Integrin-linked kinase are involved in modulating cell-cell junctions. Thus, I studied the effects of these three proteins on epithelial polarity. ETV6-NTRK3 (EN) is a fusion protein that is highly expressed in the majority of highly differentiated secretory breast carcinomas (SBC; Tognon et al., 2002), which are also rarely metastatic. When the E N gene is expressed in the non-tumorgeneic mouse mammary epithelial cell line EpH4, cells gain the capability to induce tumor formation in nude mice. These tumors are histopathologically quite similar to that of SBC (Tognon et al., 2002). Based on this observation I hypothesized that E N is functionally important in generating the SBC phenotype and these tumors are non-metastatic because their cell polarity is unaffected. To test this hypothesis, I have used the 3D culture system to examine the effects of E N expression in the normal mammary epithelial cell polarity. 32 Podocalyxin is a highly sialyated and sulfated cell surface glycoprotein (Kerjaschki et al., 1984). This CD34-related protein acts as an anti-adhesion molecule in normal development. Furthermore, a loss-of-function study has indicated that in normal kidney development, Podocalyxin is important in breaking down cell-cell junctions (Doyonnas et al., 2001). I hypothesize that Podocalyxin is involved in the disruption of epithelial junction polarity, inducing a phenotype similar to invasive ductal carcinoma. To test this hypothesis I have used a patient outcome-linked invasive breast tumor tissue array to examine whether Podocalyxin is indeed overexpressed in invasive breast tumors. In addition, I ectopically overexpressed podocalyxin in a well-behaved MCF-7 breast carcinoma cell line and examined the effects on cell-cell junctions and polarity. The interaction of cells within the extracellular matrix regulates cell shape, and motility through integrin-mediated signal transduction. ILK is known to interact with the p i integrin cytoplasmic domain (Hannigan et al., 1996). Initial studies show that overexpression of ILK can suppress E-cadherin, can generate invasive mesenchymal like spindle cell breast tumors, and can induce anchorage-independent cell growth (Attwell et al., 2000; White et al., 2001). Based on these observations I hypothesize that ILK overexpression in mammary epithelial cells would lead to a complete disruption of epithelial polarity and cell-cell interactions, inducing an epithelial to mesenchymal transformation. This would generate a phenotype that is similar to ILC. To test this, I have overexpressed ILK in mammary epithelial cells and examined the effects on cell-cell junctions and morphogenesis. In summary, the work presented in this thesis examines how mammary epithelial TJs are formed and polarized during normal development. Furthermore, examines the role of three candidate proteins on their effects on cell adhesion and polarity, which plays a mojor role during breast cancer progression. 33 Figure 1.1 Epithelial Organization. Schematic representation of the mammary epithelial organization; The mammary epithelium is organized into two main types of epithelial compartments, ducts and acini/lobules. These two types of compartments in combination form the "terminal duct lobular units" of the mammary gland. Both, ducts and acini, posses a polarized architecture, where, two layers of epithelial cells surround a central lumen. These two layers of cells are the inner layer of luminal epithelial cells and the outer layer of myoepithelial cells. The basal surface of these epithelial cells interacts with the basement membrane while the apical surface faces the central lumen. The apical junction complex is localized at the apical membrane where cell-cell interactions are taking place. This apical junction complex consists of both adherens junctions and tight junctions. 34 35 Figure 1.2 Hypothetical linear model of breast cancer progression. Schematic representation of the hypothetical linear model of breast cancer progression; the normal epithelial architecture of the T D L U consists of two layers of polarized epithelial cells surrounding a central lumen (Figure 1.1). This normal architecture is gradually altered during breast cancer progression. In the initial benign, "hyperplastic" stage, reperents an increase in cell proliferation, which can lead to an increase in the number of cells in the ducts or lobules. This increase in cell proliferation and cell accumulation leads to a morphological disruption of T D L U architecture, which is reffered to as "Atypical hyperplasia". These hyperplasias are of both lobule and ductal origian and depend on their origin they can progress into lobular or ductal type carcinomas respectively. In the "lobular carcinoma in situ" stage, lumens are filled with a solid mass of cells, which subsequently progress into "Invasive lobular carcinoma" where the cells break away as single cells and invade into the surrounding stroma. "Ductal carcinoma in situ" is marked by an increase in ductal architectural disruption leading to multiple oval shaped spaces within the lesion. These lesions can further progress into "Invasive ductal carcinoma" where cells break away as groups of cells and invade into the surrounding stroma. A subtype of this infiltrating ductal carcinoma is "secretory breast carcinoma", where cells are organized into multilayered ductal looking structures. 36 Secretory Breast Carc inoma f Multilayering Differentiated Clusters Figure 1.3 IGF-1 signaling pathways. This simplified schematic depicts two down-stream signalling pathways, the PI3-K pathway and the M A P - K pathway, that are activated upon binding of IGF-1 to IGF-1R. Activation of these pathways leads to the regulation of a number of cellular effects including cell proliferation, cell adhesion, differentiation and apoptosis. IRS-1 have no intrinsic enzymatic activity, but are thought to act as linkers between the activated receptor and down stream signaling molecules. In the PI3-K pathway, the p85 regulatory subunit can associate with IRS-1. Once activated, the p i 10 catalytic subunit phosphorylates phosphoinositides, which triggers the phosphorylation of PKB/Akt. The Grb-2/SOS complex binding to IRS-1 initiates the M A P - K pathway, and this is followed by the activation of down stream Ras, Raf and M A P kinases. 38 Proliferation, Differentiation, Adhesion, Apoptosis <*=p 3 9 Figure 1.4 Role of 14-3-3 in apoptosis. 14-3-3 is in constant association with B A D protein. In the presence of apoptotic signals B A D is released from 14-3-3 allowing it to associate with Bcl-2. The newly formed BAD-Bcl-2 complex can now activate downstream apoptotic pathways. 40 Anti-Apoptotic Pro-Apoptotic Figure 1.5 Apical Junction Complex. The apical junction complex consists of tight junctions (TJs) and adherens junctions (AJs). The most apically localized TJ is composed of the transmembrane proteins claudin, occludin and J A M . At the cytoplasmic domains, these transmembrane proteins interact with other cytoplasmic TJ proteins ZO-1, ZO-2, ZO-3, cingulin, PATJ and MUPP1. The cell polarity protein complex PAR3-aPKC-PAR6 interacts with J A M at the junction. The TJ is linked to the actin cytoskeleton via the cytoplasmic TJ protein interactions with the actin cytoskeleton. AJs are localized just basal to the TJs. The transmembrane proteins at the A J are nectin and E-cadherin. The cytoplasmic domain of nectin binds to afadin and links the junction to the actin cytoskeleton. E-cadherin binds to P-catenin and pi20 by its cytoplasmic tail. Further, P-catenin links E-cadherin to the actin cytoskeleton via binding to ct-catenin. 42 Act i i Tight Junction Claudi i i P A T . MUPP1 J A M Cingulir Adherens Junction 43 Figure 1.6 PDZ Domain Containing Proteins of Tight Junctions PDZ domains are represented by the ovals while the boxes represent other common domains. Intermolecular interactions and cytoskeletal associations are indicated below each domain. PDZ - Psd95, Dig, ZO-1 homology domain SH3 - Src homology domain 3 G K - Guanylate kinase PR - Proline rich domain U l - Unknown domain 1 L27 - Two Lin-7 binding domains M R E - M A G U K recruitment domain CR - Conserved region CRIB - Cdc42/Racl interactive binding domain 44 ZO-1 ZO-2 ZO-3 Palsl i i C l a u d i n Z O - l / Z O - 3 C l a u d i n Z O - 1 C P D ^ > < P D Z ^ H P R I C l a u d i n Z O - 1 A c t i n GK P R O c c l u d i n A c r i n / 4 . 1 GK O c c l u d i n A c t i n / 4 . 1 : P D Z ; SH3 GK O c c l u d i n P A T J P A T J / M U P P 1 r ^ P D Z > ^ S I CRB3 GK MAGUK Proteins PATJ 1 PAR3 J'j>p«Cprjz>^g 1AM a P K C PAR6 H C R l f ^ R ^ C R I B V P D Z aPKC Cdc42/Racl References Acs G, Lawton TJ, Rebbeck TR, LiVolsi V A and Zhang PJ. (2001) Differential expression of E-cadherin in lobular and ductal neoplasms of the breast and its biologic and diagnostic implications. A m J Clin Pathol. 115(l):85-98. Adeyinka A , Emberley E, Niu Y , Snell L, Murphy L C , Sowter H , Wykoff CC, Harris A L , Watson PH. (2002) Analysis of gene expression in ductal carcinoma in situ of the breast. Clin Cancer Res. 8(12):3788-95. Alle K M , Henshall SM, Field AS and Sutherland RL. (1998) Cyclin D l protein is overexpressed in hyperplasia and intraductal carcinoma of the breast. Clin Cancer Res. 4(4):847-54 Amari M , Suzuki A , Moriya T, Yoshinaga K , Amano G, Sasano H, Ohuchi N , Satomi S and Horii A . (1999) L O H analyses of premalignant and malignant lesions of human breast: frequent L O H in 8p, 16q, and 17q in atypical ductal hyperplasia. Oncol Rep. 6(6), 1277-80. Amati B,Alevizopoulos K and Vlach J. (1998) Myc and the cell cycle. Frontiers of Bioscience 3 D250 -D268. Ando-Akatsuka Y , Yonemura S, Itoh M , Furuse M and Tsukita S. (1999) Differential behavior of E-cadherin and occludin in their colocalization with ZO-1 during the establishment of epithelial cell polarity. J Cell Physiol. 179(2): 115-25. Andre F, Rigot V , Thimonier J, Montixi C, Parat F, Pommier G, Marvaldi J and Luis J. (1999) Integrins and E-cadherin cooperate with IGF-I to induce migration of epithelial colonic cells. Int J Cancer. 83(4):497-505. 46 Andreeva A Y , Krause E, Muller EC, Blasig IE and Utepbergenov DI. (2001) Protein kinase C regulates the phosphorylation and cellular localization J Biol Chem. 276(42), 38480-6. Atwood CS, Hovey RC, Glover JP, Chepko G, Ginsburg E, Robison W G and Vonderhaar B K . (2000) Progesterone induces side-branching of the ductal epithelium in the mammary glands of peripuberta J Endocrinol. 167(l):39-52. Aubele M , Cummings M , Walsch A , Zitzelsberger H , Nahrig J, Hofler H and Werner M . (2000) Heterogeneous chromosomal aberrations in intraductal breast lesions adjacent to invasive carcinoma. Anal Cell Pathol. 20(1): 17-24. Aumailley M and Smyth N . (1998). The role of laminins in basement membrane function. J Anat 1998, 1-21. Aurrand-Lions M , Johnson-Leger C, Wong C, Du Pasquier L and Imhof B A . (2001) Heterogeneity of endothelial junctions is reflected by differential expression and specific subcellular localization of the three J A M family members. Blood. 98(13):3699-707. Bachelder RE, Wendt M A , Fujita N , Tsuruo T and Mercurio A M . (2001) The cleavage of Akt/protein kinase B by death receptor signaling is an important event in detachment-induced apoptosis. J Biol Chem. 276(37):34702-34707. Balda MS, Gonzalez-Mariscal L, Matter K , Cereijido M and Anderson JM. (1993) Assembly of the tight junction: the role of diacylglycerol. J Cell Biol. 123(2):293-302. Barsky SH, Siegal GP, Jannotta F, Liotta L A . (1983) Loss of basement membrane components by invasive tumors but not by their benign counterparts. Lab Invest. 49(2): 140-7. 47 Becker K F , Kremmer E, Eulitz M , Schulz S, Mages J, Handschuh G, Wheelock MJ , Cleton-Jansen A M , Hofler H and Becker I. (2002) Functional allelic loss detected at the protein level in archival human tumours using allele-specific E-cadherin monoclonal antibodies. J Pathol. 197(5):567-74. Berx G, Becker K F , Hofler H and van Roy F. (1998) Mutations of the human E-cadherin (CDH1) gene. Hum Mutat. 12(4):226-37. Berx G, Cleton-Jansen A M , Strumane K , de Leeuw WJ, Nollet F, van Roy F, Cornelisse C. (1996) E-cadherin is inactivated in a majority of invasive human lobular breast cancers by truncation mutations throughout its extracellular domain. Oncogene. 13(9):1919-25. Birchmeier C, Birchmeier W and Brand-Saberi B. (1996). Epithelial-mesenchymal transitions in cancer progression. Acta Anat. 156, 217-226. Bjornstrom L and Sjoberg M . (2002) Mutations in the estrogen receptor DNA-binding domain discriminate between the classical mechanism of action and cross-talk with Stat5b and activating protein 1 (AP-1). J Biol Chem. 277(50):48479-83. Blanco M J , Moreno-Bueno G, Sarrio D, Locascio A , Cano A , Palacios J, Nieto M A . (2002) Correlation of Snail expression with histological grade and lymph node status in breast carcinomas. Oncogene. 21(20):3241-6. Bocchinfuso WP, Lindzey JK, Hewitt SC, Clark JA, Myers PH, Cooper R, Korach KS . (2000) Induction of mammary gland development in estrogen receptor-alpha knockout mice. Endocrinology. 141(8), 2982-94. Brown A M (2001) Wnt signaling in breast cancer: have we come full circle? Breast Cancer Res. 3(6):351-5. 48 Calderwood DA, Shattil SJ and Ginsberg M H . (2000) Integrins and actin filaments: reciprocal regulation of cell adhesion and signaling. J Biol Chem. 275(30):22607-10. Canadian Cancer Statistics. (2003). National Cancer institute of Canada, Toronto Canada, April 2003. Cano A , Perez-Moreno M A , Rodrigo I, Locascio A , Blanco MJ , del Barrio M G , Portillo F, Nieto M A . (2000) The transcription factor snail controls epithelial-mesenchymal transitions by repressing E-cadherin expression. Nat Cell Biol. 2(2):76-83. Chen H , Paradies NE, Fedor-Chaiken M , Brackenbury R (1997). E-cadherin mediates adhesion and suppresses cell motility via distinct mechanisms. J Cell Sci. 110 ( Pt 3):345-56. Cheng CW, Wu PE, Y u JC, Huang CS, Yue CT, Wu CW, Shen C Y . (2001) Mechanisms of inactivation of E-cadherin in breast carcinoma: modification of the two-hit hypothesis of tumor suppressor gene. Oncogene. 20(29):3814-23. Chlenski A , Ketels K V , Korovaitseva G l , Talamonti MS, Oyasu R and Scarpelli DG. (2000) Organization and expression of the human zo-2 gene (tjp-2) in normal and neoplastic tissues. Biochim Biophys Acta. 1493(3):319-24. Cleton-Jansen A M . (2002) E-cadherin and loss of heterozygosity at chromosome 16 in breast carcinogenesis: different genetic pathways in ductal and lobular breast cancer? Breast Cancer Res. 4(l):5-8. Colognato H and Yurchenco PD. (2000) Form and function: the laminin family of heterotrimers. Dev Dyn. 218(2):213-34. 49 Comijn J, Bene G, Vermassen P, Verschueren K, van Grunsven L, Bruyneel E, Mareel M , Huylebroeck D, van Roy F. (2001) The two-handed E box binding zinc finger protein SIP1 downregulates E-cadherin and induces invasion. Mol Cell. 7(6): 1267-78. Conneely O M , Jericevic B M and Lydon JP. (2003) Progesterone receptors in mammary gland development and tumorigenesis. J Mammary Gland Biol Neoplasia. 8(2), 205-14. Cordenonsi M , Mazzon E, De Rigo L , Baraldo S, Meggio F and Citi S. (1997) Occludin dephosphorylation in early development of Xenopus laevis. J Cell Sci. 110 ( Pt 24), 3131-9. D'Cruz C M , Gunther EJ, Boxer RB, Hartman JL, Sintasath L, Moody SE, Cox JD, Ha SI, Belka GK, Golant A , Cardiff RD, Chodosh L A . (2001) c -MYC induces mammary tumorigenesis by means of a preferred pathway involving spontaneous Kras2 mutations. Nat Med 7:235-239. Daniel C W and GB Silberstein. (1987) "Postnatal development of the rodent mammary gland". In The mammary gland: development, regulation and function. Plenum press, New York. 3-36. D'Atri F, Nadalutti F and Citi S. (2002) Evidence for a functional interaction between cingulin and ZO-1 in cultured cells. J Biol Chem. 277(31):27757-64. Dalgarno DC, Botfield M C and Rickles RJ. (1997) SH3 domains and drug design: ligands, structure, and biological function. Biopolymers. 43(5):383-400. Deng W M , Schneider M , Frock R, Castillejo-Lopez C, Gaman EA, Baumgartner S and Ruohola-Baker H . (2003) Dystroglycan is required for polarizing the epithelial cells and the oocyte in Drosophila. Development. 130(1): 173-84. 50 Desprez PY, Roskelley CD and Bissell MJ . (1993) Isolation of functional cell lines from a mouse mammary epithelial cell strain: the importance of basement membrane and cell-cell interactions. Mol . Cell. Differ. 1, 99-100. G. Deng, Y . Lu, G. Zlotnikov, A .D . Thor and H.S. Smith. (1996) Loss of heterozygosity in normal tissue adjacent to breast carcino-mas, Science 274, 2057-2059. Dmytrasz K, Tartter PI, Mizrachy H, Chinitz L, Rosenbaum Smith S and Estabrook A . (2003) The significance of atypical lobular hyperplasia at percutaneous breast biopsy. Breast J. 9(l):10-2. Doisneau-Sixou SF, Sergio C M , Carroll JS, Hui R, Musgrove E A and Sutherland RL. (2003) Estrogen and antiestrogen regulation of cell cycle progression in breast cancer cells. Endocr Relat Cancer. 10(2): 179-86. Doyonnas, R., Kershaw, D.B., Duhme, C , Merkens, H. , Chelliah, S., Graf, T. and McNagny, K . M . (2001). Anuria, omphalocele, and perinatal lethality in mice lacking the CD-34-related protein Podocalyxin. J Exp Med, 194,13-27. Droufakou S, Deshmane V, Roylance R, Hanby A , Tomlinson I, Hart IR. (2001) Multiple ways of silencing E-cadherin gene expression in lobular carcinoma of the breast. Int J Cancer. 92(3):404-8 Dubik D and Shiu RP (1992) Mechanism of estrogen activation of c-myc oncogene expression.Oncogene 7 1587 -1594. Dupont WD, Pari FF, Hartmann W H , Brinton L A , Winfield A C , Worrell JA, Schuyler PA and Plummer WD. (1993) Breast cancer risk associated with proliferative breast disease and atypical hyperplasia. Cancer. 71(4):1258-65. 51 Dupont WD and Page DL. (1985) Risk factors for breast cancer in women with proliferative breast disease. N EnglJ Med 312: 146-151. Ebnet K, Suzuki A , Horikoshi Y , Hirose T, Meyer Zu Brickwedde M K , Ohno S and Vestweber D. (2001) The cell polarity protein ASIP/PAR-3 directly associates with junctional adhesion molecule (JAM). E M B O J. 20(14):3738-48. Ebnet K , Schulz C U , Meyer Zu Brickwedde M K , Pendl G G and Vestweber D. (2000) Junctional adhesion molecule interacts with the PDZ domain-containing proteins AF-6 and ZO-1. J Biol Chem. 275(36):27979-88. Engbring JA and Kleinman HK. (2003) The basement membrane matrix in malignancy. J Pathol. 200(4):465-70. Enriori PJ, Fischer CR, Gori JR, Etkin A E , Calandra RS and Luthy IA. (2003) Augmented serum levels of the IGF-I/IGF-binding protein-3 ratio in pre-menopausal patients with type I breast cysts. Eur J Endocrinol. 148(2): 177-84. Ernster V L , Barclay J. (1997) Increases in ductal carcinoma in situ (DCIS) of the breast in relation to mammography: a dilemma.J Natl Cancer Inst Monogr. (22):151-6. Fanning AS, Jameson BJ, Jesaitis L A and Anderson J M . (1998) The tight junction protein ZO-1 establishes a link between the transmembrane protein occludin and the actin cytoskeleton. J Biol Chem. 273(45), 29745-53. Farabegoli F, Champeme M H , Bieche I, Santini D, Ceccarelli C, Derenzini M and Lidereau R. (2002) Genetic pathways in the evolution of breast ductal carcinoma in situ. J Pathol. 196(3):280-6. Farquhar M and Palade GE. (1963). Junctional complexes in various epithelia. J. Call Biol. 17, 375-412. 52 Fitzgibbons PL, Henson DE, Flutter RV. (1998) Benign breast changes and the risk for subsequent breast cancer: an update of the 1985 consensus statement. Cancer Committee of the College of American Pathologists. Arch Pathol Lab Med. 122(12):1053-5. Forbes JF. (1997a) The incidence of breast cancer: the global burden, public health considerations. Semin Oncol. 24(1 Suppl l):Sl-20-Sl-35 Forbes JF. The control of breast cancer: (1997b) The role of tamoxifen. Seminars in Oncol 24(1) Suppl 1: S1-5-S1-19 Fujita N , Jaye DL, Kajita M , Geigerman C, Moreno CS, Wade PA. (2003) MTA3, a M i -2/NuRD complex subunit, regulates an invasive growth pathway in breast cancer. Cell. 113(2):207-19. Fukata M , Nakagawa M , Itoh N , Kawajiri A , Yamaga M , Kuroda S and Kaibuchi K . (2001) Involvement of IQGAP1, an effector of Rac 1 and Cdc42 GTPases, in cell-cell dissociation during cell scattering. Mol Cell Biol. (6):2165-83. Fukuhara A , Irie K , Nakanishi H , Takekuni K, Kawakatsu T, Ikeda W, Yamada A, Katata T, Honda T, Sato T, Shimizu K , Ozaki H , Horiuchi H , Kita T and Takai Y . (2002a) Involvement of nectin in the localization of junctional adhesion molecule at tight junctions. Oncogene. 21(50), 7642-55. Fukuhara A, Irie K , Yamada A, Katata T, Honda T, Shimizu K , Nakanishi H and Takai Y . (2002b) Role of nectin in organization of tight junctions in epithelial cells. Genes Cells. 7(10): 1059-72. Furuse M , Hata M , Furuse K, Yoshida Y , Haratake A , Sugitani Y , Noda T, Kubo A and Tsukita S. (2002) Claudin-based tight junctions are crucial for the mammalian epidermal barrier: a lesson from claudin-1-deficient mice. J Cell Biol. 156(6):1099-1 111. 53 Furuse M , Sasaki H, Fujimoto K, Tsukita S. (1998) A single gene product, claudin-1 or -2, reconstitutes tight junction strands and recruits occludin in fibroblasts. J Cell Biol. 143(2), 391-401. Gallin WJ, Sorkin BC, Edelman G M and Cunningham B A . (1987) Sequence analysis of a cDNA clone encoding the liver cell adhesion molecule, L - C A M . Proc Natl Acad Sci 84(9), 2808-12. Gamallo C, Palacios J, Suarez A , Pizarro A , Navarro P, Quintanilla M and Cano A . (1993) Correlation of E-cadherin expression with differentiation grade and histological type in breast carcinoma. A m J Pathol. 142(4), 987-93. Gao L, Joberty G and Macara IG. (2002) Assembly of epithelial tight junctions is negatively regulated by Par6. Curr Biol. 2002 Feb 5;12(3):221-5. Gimond C, van Der Flier A , van Delft S, Brakebusch C, Kuikman I, Collard JG, Fassler R and Sonnenberg A. (1999) Induction of cell scattering by expression of betal integrins in betal-deficient epithelial cells requires activation of members of the rho family of GTPases and downregulation of cadherin and catenin function. J Cell Biol . 147(6): 1325-40. Glaunsinger B A , Weiss RS, Lee SS and Javier R. (2001) Link of the unique oncogenic properties of adenovirus type 9 E4-ORF1 to a select interaction with the candidate tumor suppressor protein ZO-2. E M B O J. 20(20):5578-86. Going JJ. (2003) Stages on the way to breast cancer. J Pathol.199(l):l-3. Gong G, DeVries S, Chew K L , Cha I, Ljung B M and Waldman F M . (2001) Genetic changes in paired atypical and usual ductal hyperplasia of the breast by comparative genomic hybridization. Clin Cancer Res. 7(8):2410-4. 54 Gonzalez-Mariscal L, Betanzos A and Avila-Flores A . (2000) M A G U K proteins: structure and role in the tight junction. Semin Cell Dev Biol. 11(4):315-24. Gorska, A . E., Joseph, H. , Derynck, R., Moses, H . L. and R Serra. (1998) Dominant-negative interference of the transforming growth factor beta type II receptor in mammary gland epithelium results in alveolar hyperplasia and differentiation in virgin mice. Cell Growth Differ. 9(3):229-38. Gui GP, Puddefoot JR, Vinson GP, Wells C A and Carpenter R. (1997) Altered cell-matrix contact: a prerequisite for breast cancer metastasis? Br J Cancer. 75(5):623-33. Gumbiner B (1996) Cell adhesion. The molecular basis of tissue architecture and morphogenesis. Cell 84, 345-357. Gumbiner B, Lowenkopf T, Apatira D. (1991) Identification of a 160-kDa polypeptide that binds to the tight junction protein ZO-1. Proc Natl Acad Sci U S A . 88(8), 3460-4. Gumbiner B. (1987) Structure, biochemistry, and assembly of epithelial tight junctions. A m J Physiol. 253(6 Pt l):C749-58. Guvakova M A , Adams JC and Boettiger D. (2002) Functional role of alpha-actinin, PI 3-kinase and MEK1/2 in insulin-like growth factor I receptor kinase regulated motility of human breast carcinoma cells. J Cell Sci. 115(Pt 21):4149-65. Hajra K M , Chen D Y , Fearon ER. (2002) The SLUG zinc-finger protein represses E-cadherin in breast cancer. Cancer Res. 62(6): 1613-8. Hamelers IH and Steenbergh PH. (2003). Interactions between estrogen and insulin-like growth factor signaling pathways in human breast tumor cells. Endocr Relat Cancer. 10: 331-345. 55 Hankinson SE, Willett WC, Colditz GA, Hunter DJ, Michaud DS, Deroo B, Rosner B, Speizer FE and Pollak M . (1998) Circulating concentrations of insulin-like growth factor-I and risk of breast cancer. Lancet. 351(9113): 1393-6. Hannigan, G. E., Leung-Hagesteijn, C , Fitz-Gibbon, L., Coppolino, M . G., Radeva, G., Filmus, J., Bell, J. C. and Dedhar, S. (1996). Regulation of cell adhesion and anchorage-dependent growth by a new bl-integrin-linked protein kinase. Nature 379, 91-96. Harper JW, Elledge SJ, Keyomarsi K , Dynlacht B, Tsai L H , Zhang P, Dobrowolski S, Bai C, Connell-Crowley L, Swindell E. (1995) Inhibition of cyclin-dependent kinases by p21. Mol Biol Cell. 6(4):387-400. Harris JR, Lippman M E , Morrow M and Osborne C K . (2000). Diseases of the breast. Second edition. Lippincott Williams and Wilkins, Philadelphia, USA. Haskins J, Gu L, Wittchen ES, Hibbard J and Stevenson BR. (1998) ZO-3, a novel member of the M A G U K protein family found at the tight junction, interacts with ZO-1 and occludin. J Cell Biol. 141(1), 199-208 Hay ED. (1995) An overview of epithelial-mesenchymal transformation. Acta Anat 154, 8-21. Hadsell DL, Greenberg N M , Fligger J M , Baumrucker CR, Rosen J M . (1996) Targeted expression of des(l-3) human insulin-like growth factor I in transgenic mice influences mammary gland development and IGF-binding protein expression. Endocrinology. 137(l):321-30. Heffelfinger SC, Yassin R, Miller M A and Lower EE. (2000) Cyclin D I , retinoblastoma, p53, and Her2/neu protein expression in preinvasive breast pathologies: correlation with vascularity. Pathobiology. 68(3), 129-36. 56 Hamazaki Y , Itoh M , Sasaki H, Furuse M and Tsukita S. (2002) Multi-PDZ domain protein 1 (MUPP1) is concentrated at tight junctions through its possible interaction with claudin-1 and junctional adhesion molecule. J Biol Chem. 277(1):455-61 Hoevel T, Macek R, Mundigl O, Swisshelm K and Kubbies M . (2002) Expression and targeting of the tight junction protein CLDN1 in CLDN1-negative human breast tumor cells. J Cell Physiol. 191(1), 60-8. Hogervorst F, Kuikman I, von dem Borne A E and Sonnenberg A . (1990) Cloning and sequence analysis of beta-4 cDNA: an integrin subunit that contains a unique 118 kd cytoplasmic domain. E M B O J. 9(3):765-70. Holland R, Peterse JL, Millis RR, Eusebi V , Faverly D, Vandevijver M J , Zafrani B: Ductal carcinoma in situ: a proposal for a new classification. Semin Diagn Pathol 1994, 11:167-180. Honda T, Shimizu K , Kawakatsu T, Yasumi M , Shingai T, Fukuhara A , Ozaki-Kuroda K, Irie K , Nakanishi H and Takai Y . (2003) Antagonistic and agonistic effects of an extracellular fragment of nectin on formation of E-cadherin-based cell-cell adhesion. Genes Cells. 8(1), 51-63. Hoover K B , Liao SY and Bryant PJ (1998) Loss of the tight junction M A G U K ZO-1 in breast cancer: relationship to glandular differentiation and loss of heterozygosity. A m J Pathol 153(6): 1767-73 Howlett AR, and M J Bissell. (1993) The influence of tissue microenvironment (stroma and extracellular matrix) on the development and function of mammary epithelium. Epithelial Cell Biol. 2(2):79-89. 57 Hynes RO. (1992) Integrins: versatility, modulation, and signaling in cell adhesion. Cell. 69(l):ll-25. Ilic D, Almeida EA, Schlaepfer DD, Dazin P, Aizawa S and Damsky CH. (1998) Extracellular matrix survival signals transduced by focal adhesion kinase suppress p53-mediated apoptosis. J Cell Biol. 143(2):547-60. Imamura Y , Itoh M , Maeno Y , Tsukita S and Nagafuchi A. (1999) Functional domains of alpha-catenin required for the strong state of cadherin-based cell adhesion. J Cell Biol. 144(6), 1311-22. Itoh M , Furuse M , Morita K, Kubota K , Saitou M and Tsukita S. (1999) Direct binding of three tight junction-associated M A G U K s , ZO-1, ZO-2, and ZO-3, with the COOH termini of claudins. J Cell Biol. 147(6):1351-63. Itoh M , Nagafuchi A , Yonemura S, Kitani-Yasuda T, Tsukita S and Tsukita S. (1993) The 220-kD protein colocalizing with cadherins in non-epithelial cells is identical to ZO-1, a tight junction-associated protein in epithelial cells: cDNA cloning and immunoelectron microscopy. J Cell Biol. 121(3):491-502. Itoh M , Sasaki H, Furuse M , Ozaki H , Kita T and Tsukita S. (2001) Junctional adhesion molecule (JAM) binds to PAR-3: a possible mechanism for the recruitment of PAR-3 to tight junctions. J Cell Biol. 154(3):491-7. Izumi Y , Hirose T, Tamai Y , Hirai S, Nagashima Y , Fujimoto T, Tabuse Y , Kemphues K J and Ohno S. (1998) A n atypical P K C directly associates and colocalizes at the epithelial tight junction with ASIP, a mammalian homologue of Caenorhabditis elegans polarity protein PAR-3. J Cell Biol. 143(1):95-106. 58 Jacobs TW, Pliss N , Kouria G and Schnitt SJ. (2001) Carcinomas in situ of the breast with indeterminate features: role of E-cadherin staining in categorization. A m J Surg Pathol. 25(2), 229-36. Jeansonne B, Lu Q, Goodenough D A and Chen Y H . (2003) Claudin-8 interacts with multi-PDZ domain protein 1 (MUPPI) and reduces paracellular conductance in epithelial cells. Cell Mol Biol. 49(1), 13-21. Jesaitis L A and Goodenough DA. (1994) Molecular characterization and tissue distribution of ZO-2, a tight junction protein homologous to ZO-1 and the Drosophila discs-large tumor suppressor protein. J Cell Biol. 6):949-61. Joberty G, Petersen C, Gao L and Macara IG. (2000) The cell-polarity protein Par6 links Par3 and atypical protein kinase C to Cdc42. Nat Cell Biol. 2(8):531-9. Johansson A, Driessens M and Aspenstrom P. (2000) The mammalian homologue of the Caenorhabditis elegans polarity protein PAR-6 is a binding partner for the Rho GTPases Cdc42 and Racl . J Cell Sci. 2000 Sep;113 ( Pt 18):3267-75. Jones JL, Shaw JA, Pringle JH and R A Walker. (2003) Primary breast myoepithelial cells exert an invasion-suppressor effect on breast cancer cells via paracrine down-regulation of M M P expression in fibroblasts and tumor cells. J Pathol. 201, 562-572. Joseph, H. , Gorska, A . E., Sohn, P., Moses, H. L. and R. Serra. (1999) Overexpression of a kinase-deficient transforming growth factor-beta type II receptor in mouse mammary stroma results in increased epithelial branching. Mol . Biol. Cell 10(4): 1221-34. Kaech SM, Whitfield C W and Kim SK. (1998) The LrN-2/LLN-7/LIN-10 complex mediates basolateral membrane localization of the C. elegans EGF receptor LET-23 in vulval epithelial cells. Cell. 94(6):761-71. 59 Kamalati, T. et al. (1999) HGF/SF in mammary epithelial growth and morphogenesis: in vitro and in vivo models. J. Mammary Gland Biol. Neoplasia 4, 69-77 Kamberov E, Makarova O, Roh M , Liu A , Karnak D, Straight S and Margolis B. (2000) Molecular cloning and characterization of Pals, proteins associated with mLin-7. J Biol Chem. 275(15): 11425-31. Kandel R, L i SQ, Ozcelik H and Rohan T. (2000) p53 protein accumulation and mutations in normal and benign breast tissue. Int J Cancer. 87(l):73-8. Kerjaschki, D., Sharkey, D.J., and Farquhar, M . G . (1984) Identification and characterization of podocalyxin-the major sialoprotein of the renal glomerular epithelial cell. J. Cell Biol. 98, 1591-1596. Khandwala H M , McCutcheon IE, Flyvbjerg A and Friend K E . (2000) The effects of insulin-like growth factors on tumorigenesis and neoplastic growth. Endocr Rev. 21(3):215-44. Kominsky SL, Argani P, Korz D, Evron E, Raman V , Garrett E, Rein A , Sauter G, Kallioniemi OP and Sukumar S. (2003) Loss of the tight junction protein claudin-7 correlates with histological grade in both ductal carcinoma in situ and invasive ductal carcinoma of the breast. Oncogene. 22(13), 2021-33 Krane, I. M . and P. Leder. (1996) NDF/heregulin induces persistence of terminal end buds and adenocarcinomas in the mammary glands of transgenic mice. Oncogene 12(8):1781-1788. Kratochwil, K . (1969) Organ specificity in mesenchymal induction demonstrated in the embryonic development of the mammary gland of the mouse. Dev. Biol. 20: 46-71. 60 Kroschewski R, Hall A and Mellman I. (1999) Cdc42 controls secretory and endocytic transport to the basolateral plasma membrane of M D C K cells. Nat Cell Biol. 1(1):8-13. Kuerbitz S,Plunkett B,Walsh and W,Kastan M . (1992)Wild-type p53 is a cell checkpoint determinant following irradiation.Proc Natl Acad Sci U SA . 89:7491 -7495. Kumar R, Wang RA, Bagheri-Yarmand R. (2003) Emerging roles of M T A family members in human cancers. Semin Oncol. 30(5 Suppl 16):30-7. Lakhani SR and O'Hare MJ . (2001) The mammary myoepithelial cell—Cinderella or ugly sister? Breast Cancer Res. 3(1), 1-4. Lakhani SR, Collins N , Stratton M R and Sloane JP. (1995) Atypical ductal hyperplasia of the breast: clonal proliferation with loss of heterozygosity on chromosomes 16q and 17p. J Clin Pathol 48: 611-615. Lakhani SR. (2003) The diagnosis and management of pre-invasive breast disease-current challenges, future hopes. Breast Cancer Res. 5(5):248-9. LeRoith D, Roberts CT Jr. The insulin-like growth factor system and cancer.Cancer Lett. 2003 Jun 10; 195(2): 127-37. Lin D, Edwards AS, Fawcett JP, Mbamalu G, Scott JD and Pawson T. (2001) A mammalian PAR-3-PAR-6 complex implicated in Cdc42/Racl and aPKC signalling and cell polarity. Nat Cell Biol . 2(8):540-7. Liu QY, Niranjan B, Gomes P, Gomm JJ, Davies D, Coombes RC and Buluwela L. (1996) Inhibitory effects of activin on the growth and morpholgenesis of primary and transformed mammary epithelial cells. Cancer Res. 56(5), 1155-63. 61 Liu Y , Nusrat A , Schnell FJ, Reaves TA, Walsh S, Pochet M , Parkos CA. (2000) Human junction adhesion molecule regulates tight junction resealing in epithelia. J Cell Sci. 113 (Pt 13):2363-74. Lu Y J , Osin P, Lakhani SR, Di Palma S, Gusterson B A , Shipley JM. (1998) Comparative genomic hybridization analysis of lobular carcinoma in situ and atypical lobular hyperplasia and potential roles for gains and losses of genetic material in breast neoplasia. Cancer Res. 58(20):4721-7. Lundberg AS and Weinberg RA. (1999) Control of the cell cycle and apoptosis. Eur J Cancer. 35(14):1886-94. Maluf H and Koerner F. (2001) Lobular carcinoma in situ and infiltrating ductal carcinoma: frequent presence of DCIS as a precursor lesion. Int J Surg Pathol. 9(2), 127-31. Martin-Padura I, Lostaglio S, Schneemann M , Williams L, Romano M , Fruscella P, Panzeri C, Stoppacciaro A, Ruco L, Vi l la A , Simmons D and Dejana E. (1998) Junctional adhesion molecule, a novel member of the immunoglobulin superfamily that distributes at intercellular junctions and modulates monocyte transmigration. J Cell Biol. 142(1), 117-27. Mattagajasingh SN, Huang SC, Hartenstein JS and Benz EJ Jr. (2000) Characterization of the interaction between protein 4.1R and ZO-2. A possible link between the tight junction and the actin cytoskeleton. J Biol Chem. 275(39):30573-85. Mauro L, Bartucci M , Morelli C, Ando' S and Surmacz E. (2001) IGF-I receptor-induced cell-cell adhesion of MCF-7 breast cancer cells requires the expression of junction protein ZO-1. J Biol Chem. 276(43):39892-7. 62 McDivitt RW, Stevens JA, Lee NC, Wingo PA, Rubin G L and Gersell D. (1992) Histologic types of benign breast disease and the risk for breast cancer. The Cancer and Steroid Hormone Study Group. Cancer. 69(6): 1408-14 McDonnell DP &Norris JD (2002) Connections and regulation of the human estrogen receptor.Science 296 1642 -1644. McManaman JL and Neville M C . (2003) Mammary physiology and milk secretion. Adv Drug Deliv Rev. 55(5):629-41. Middleton LP, Grant S, Stephens T, Stelling CB, Sneige N and Sahin A A . (2003) Lobular carcinoma in situ diagnosed by core needle biopsy: when should it be excised? Mod Pathol. 16(2): 120-9. Miyahara M , Nakanishi H , Takahashi K, Satoh-Horikawa K, Tachibana K and Takai Y . (2000) Interaction of nectin with afadin is necessary for its clustering at cell-cell contact sites but not for its cis dimerization or trans interaction. J Biol Chem. 275(1), 613-8. Miyashita T and Reed JC. (1995) Tumor suppressor p53 is a direct tran-scriptional activator of the human bax gene. Cell 80: 293-299. Miyashita T,Harigai M,Hanada M and Reed JC. (1994) Identification of a p53-dependent negative response element in the bcl-2 gene.Cancer Res . 54, 3131-3135. Moll R, Mitze M , Frixen U H , Birchmeier W. (1993) Differential loss of E-cadherin expression in infiltrating ductal and lobular breast carcinomas. A m J Pathol. 143(6): 1731-42. Mommers EC, van Diest PJ, Leonhart A M , Meijer CJ and Baak JP. (1998) Expression of proliferation and apoptosis-related proteins in usual ductal hyperplasia of the breast. Hum Pathol. 29(12): 1539-45. 63 Morita K , Furuse M , Fujimoto K, Tsukita S. (1999) Claudin multigene family encoding four-transmembrane domain protein components of tight junction strands. Proc Natl Acad Sci U S A. 96(2), 511-6. Muschler J, Levy D, Boudreau R, Henry M , Campbell K and Bissell M J . (2002) A role for dystroglycan in epithelial polarization: loss of function in breast tumor cells. Cancer Res. 62(23):7102-9. Muschler J, Lochter A , Roskelley CD, Yurchenco P and Bissell MJ . (1999) Division of labor among the alpha6beta4 integrin, betal integrins, and an E3 laminin receptor to signal morphogenesis and beta-casein expression in mammary epithelial cells. Mol Biol Cell. 10(9):2817-28. Nanba D, Nakanishi Y and Hieda Y . (2001) Changes in adhesive properties of epithelial cells during early morphogenesis of the mammary gland. Dev Growth Differ. 43(5):535-44. Nguyen D A and Neville M C . (1998) Tight junction regulation in the mammary gland. J Of Mammary Gland Biology And Neoplasia. 3(3): 233-246. Niemann, C , Brinkmann, V . , Spitzer, E., Hartmann, G., Sachs M . , Naundorf, H . and W. Birchmeier. 1998. Reconstitution of mammary gland development in vitro: requirement of c-met and c-erbB2 signaling for branching and alveolar morphogenesis. J. Cell Biol. 143(2):533-545. Nikiforov M A , Hagen K, Ossovskaya VS, Connor T M , Lowe SW, Deichman G l and Gudkov A V . (1996) p53 modulation of anchorage independent growth and experimental metastasis. Oncogene; 13(8):1709-19. 64 Nishizaki T, Chew K, Chu L, Isola J, Kallioniemi A , Weidner N , Waldman F M . (1997) Genetic alterations in lobular breast cancer by comparative genomic hybridization. Int J Cancer. 74(5):513-7. Obermah HA. (1980) Secretory carcinoma of the breast in adults. A m J Surg Pathol. 4(5):465-70. O'Brien L E , Zegers M M and Mostov K E . (2002) Opinion: Building epithelial architecture: insights from three-dimensional culture models. Nat Rev Mol Cell Biol. 3(7):531-7. O'Connell P, Pekkel V , Fuqua S, Osborne C K , Allred DC. (1994) Molecular genetic studies of early breast cancer evolution. Breast Cancer Res Treat. 32(1):5-12. O'Connell P, Pekkel V , Fuqua SA, Osborne CK, Clark G M and Allred DC. (1998) Analysis of loss of heterozygosity in 399 premalignant breast lesions at 15 genetic loci. J Natl Cancer Inst. 90(9):697-703. Ormandy CJ, Musgrove EA, Hui R, Daly RJ and Sutherland RL. (2003) Cyclin DI , EMS1 and 1 lq l3 amplification in breast cancer. Breast Cancer Res Treat. 78(3):323-35. Page DL, Ellis IO, Elston CW. (1995) Histologic grading of breast cancer. Let's do it. Am J Clin Pathol. 103(2): 123-4. Page DL, Schuyler PA, Dupont WD, Jensen RA, Plummer W D Jr and Simpson JF. (2003) Atypical lobular hyperplasia as a unilateral predictor of breast cancer risk: a retrospective cohort study. Lancet. 361(9352): 125-9. Parker B and Sukumar S. (2003) Distant metastasis in breast cancer: molecular mechanisms and therapeutic targets. Cancer Biol Ther. 2(1): 14-21. 65 Parkin D M , Pisani P and Ferlay J. (1999) Global cancer statistics. C A Cancer J Clin. 49(1): 33-64. Porter W, Saville B, Hoivik D, Safe S. (1997) Functional synergy between the transcription factor Spl and the estrogen receptor. Mol Endocrinol. 11(11):1569-80. Pechoux C, Chardonnet Y , Chignol M C and Noel P. (1994) Heterogeneous immunoreactivity of frozen human benign and malignant breast lesions to C - M Y C and C-Ha-ras cellular oncogenes. Histol Histopathol. 9(1): 35-44. Petersen OW, Gudjonsson T, Villadsen R, Bissell M J and Ronnov-Jessen L. (2003) Epithelial progenitor cell lines as models of normal breast morphogenesis and neoplasia. Cell Prolif. 36 Suppl 1: 33-44. Petersen OW, Hoyer PE and van Deurs B. (1987) Frequency and distribution of estrogen receptor-positive cells in normal, nonlactating human breast tissue. Cancer Res. 47(21):5748-51. Peyrat JP, Bonneterre J, Hecquet B, Vermin P, Louchez M M , Fournier C, Lefebvre J and Demaille A . (1993) Plasma insulin-like growth factor-1 (IGF-1) concentrations in human breast cancer. Eur J Cancer. 29A(4), 492-7. Pinder SE and Ellis IO (2003) The diagnosis and management of pre-invasive breast disease: ductal carcinoma in situ (DCIS) and atypical ductal hyperplasia (ADH)—current definitions and classification. Breast Cancer Res. 5(5):254-7. Planas-Silva M D and Weinberg R A (1997) Estrogen-dependent cyclin E-cdk2 activation through p21 redistribution. Molecular and Cellular Biology. 17 4059 -4069. Porter D, Lahti-Domenici J, Keshaviah A , Bae Y K , Argani P, Marks J, Richardson A , Cooper A , Strausberg R, Riggins GJ, Schnitt S, Gabrielson E, Gelman R, Polyak K. 66 (2003) Molecular markers in ductal carcinoma in situ of the breast. Mol Cancer Res. l(5):362-75. Rasbridge SA, Gillett CE, Sampson SA, Walsh FS, and RR Millis. (1993) Epithelial (E-) and placental (P-) cadherin cell adhesion molecule expression in breast carcinoma. J Pathol. 169(2):245-50. Reichmann E, Ball R, Groner B and Friis RR. (1989) New mammary epithelial and fibroblastic cell clones in coculture form structures competent to differentiate functionally. J Cell Biol. 108(3):1127-38. Reis-Filho JS, Lakhani SR. (2003) The diagnosis and management of pre-invasive breast disease: genetic alterations in pre-invasive lesions. Breast Cancer Res. 5(6):313-9. Reymond N , Borg JP, Lecocq E, Adelaide J, Campadelli-Fiume G, Dubreuil P and Lopez M . (2000) Human nectin3/PRR3: a novel member of the PVR/PRR/nectin family that interacts with afadin. Gene. 255(2), 347-55. Rietmacher D, Brinkmann V and Birchmeier CA. (1995). Targeted mutation in the mouse E-cadherin gene results in defective preimplantation development. Proc Natl Acad Sci. 92, 855-859 Rocha RL, Hilsenbeck SG, Jackson JG, VanDenBerg C L , Weng C, Lee A V and Yee D. (1997) Insulin-like growth factor binding protein-3 and insulin receptor substrate-1 in breast cancer: correlation with clinical parameters and disease-free survival. Clin Cancer Res. 3(1): 103-9. Rodriguez-Boulan E and Nelson WJ. (1989) Morphogenesis of the polarized epithelial cell phenotype. Science. 245(4919), 718-25. 67 Roh M H , Fan S, Liu CJ and Margolis B. (2003) The Crumbs3-Palsl complex participates in the establishment of polarity in mammalian epithelial cells. J Cell Sci. 116(Pt 14):2895-906. Roh M H , Liu CJ, Laurinec S and Margolis B. (2002) The carboxyl terminus of zona occludens-3 binds and recruits a mammalian homologue of discs lost to tight junctions. J Biol Chem. 277(30), 27501-9. Roh M H , Makarova O, Liu CJ, Shin K, Lee S, Laurinec S, Goyal M , Wiggins R and Margolis B. (2002) The Maguk protein, Palsl, functions as an adapter, linking mammalian homologues of Crumbs and Discs Lost. J Cell Biol. 157(1), 161-72. Rohan TE, Hartwick W, Miller A B , Kandel RA. (1998) Immunohistochemical detection of c-erbB-2 and p53 in benign breast disease and breast cancer risk. J Natl Cancer Inst 90:1262-9. Rose-Hellekant T A and Sandgren EP. (2000) Transforming growth factor alpha- and c-myc-induced mammary carcinogenesis in transgenic mice. Oncogene. 19(8): 1092-6. Rosen PP and Cranor M L . (1991) Secretory carcinoma of the breast. Arch Pathol Lab Med. 115(2): 141-4. Roskelley CD, Desprez P Y and Bissell MJ . (1994) Extracellular matrix-dependent tissue-specific gene expression in mammary epithelial cells requires both physical and biochemical signal transduction. Proc Natl Acad Sci U S A . 91(26): 12378-82. Roskelley CD, Srebrow A and Bissell MJ . (1995) A hierarchy of ECM-mediated signalling regulates tissue-specific gene expression. Curr Opin Cell Biol. 7(5):736-47. Roylance R, Droufakou S, Gorman P, Gillett C, Hart IR, Hanby A , Tomlinson I. (2003) The role of E-cadherin in low-grade ductal breast tumourigenesis. J Pathol. 200(1 ):53-8. 68 Ruan W and Kleinberg DL. (1999) Insulin-like growth factor I is essential for terminal end bud formation and ductal morphogenesis during mammary development. Endocrinology. 140: 5075-5081. Russo IH and Russo J. (1998) Role of hormones in mammary cancer initiation and progression. J Mammary Gland Biol Neoplasia. 3(1):49-61. Russo J, Hasan Lareef M , Balogh G, Guo S and Russo IH. (2003) Estrogen and its metabolites are carcinogenic agents in human breast epithelial cells. J Steroid Biochem Mol Biol. 87(l):l-25. Saitou M , Furuse M , Sasaki H, Schulzke JD, Fromm M , Takano H, Noda T and Tsukita S. (2000) Complex phenotype of mice lacking occludin, a component of tight junction strands. Mol Biol Cell. 11(12), 4131-42. Sakakibara A , Furuse M , Saitou M , Ando-Akatsuka Y and Tsukita S. (1997) Possible involvement of phosphorylation of occludin in tight junction formation. J Cell Biol. 137(6), 1393-401. Sakakura, T. 1987. Mammary embryogenesis in "Mammary gland development, regulation, and function". Plenum press, New york. 37-63. Sarrio D, Moreno-Bueno G, Hardisson D, Sanchez-Estevez C, Guo M , Herman JG, Gamallo C, Esteller M , Palacios J. (2003) Epigenetic and genetic alterations of A P C and CDH1 genes in lobular breast cancer: relationships with abnormal E-cadherin and catenin expression and microsatellite instability. Int J Cancer. 106(2):208-15. Satoh-Horikawa K , Nakanishi H , Takahashi K , Miyahara M , Nishimura M , Tachibana K , Mizoguchi A and Takai Y . (2000) Nectin-3, a new member of immunoglobulin-like cell 69 adhesion molecules that shows homophilic and heterophilic cell-cell adhesion activities, J Biol Chem. 275(14), 10291-9. Schmitt FC, Leal C, Lopes C. (1995) p53 protein expression and nuclear D N A content in breast intraductal proliferations. J Pathol. 176(3):233-41. Schreider C, Peignon G, Thenet S, Chambaz J and Pincon-Raymond M . (2002) Integrin-mediated functional polarization of Caco-2 cells through E-cadherin~actin complexes. J Cell Sci. 115(Pt 3):543-52. Schairer C, Hi l l D, Sturgeon SR, Fears T, Pollak M , Mies C, Ziegler RG, Hoover RN, Sherman M E . (2004) Serum concentrations of IGF-I, IGFBP-3 and c-peptide and risk of hyperplasia and cancer of the breast in postmenopausal women. Int J Cancer. 108(5):773-9. Schuger L. (1997) Laminins in lung development. Exp Lung Res. 23(2):119-29. Shoker BS, Jarvis C, Sibson DR, Walker C and Sloane JP. (1999) Oestrogen receptor expression in the normal and pre-cancerous breast. J Pathol. 188(3):237-44. Siegel JR, Karcnik TJ, Hertz M B , Gelmann H and Baker SR. (1999) Secretory Carcinoma of the Breast. Breast J. 5(3):204-207. Simpson K J , Bird P, Shaw D, and K Nicholas. (1998) Molecular characterisation and hormone-dependent expression of the porcine whey acidic protein gene. J Mol Endocrinol. 20(l):27-35. Simpson PT, Gale T, Fulford L G , Reis-Filho JS and Lakhani SR. (2003) The diagnosis and management of pre-invasive breast disease: pathology of atypical lobular hyperplasia and lobular carcinoma in situ. Breast Cancer Res. 5(5):258-62. 70 Sinn E, Muller WJ, Pattengale P, Tepler I, Wallace R, Leder P. (1987) Coexpression of MMTV/v-Ha-ras and MMTV/c-myc genes in transgenic mice: synergistic action of oncogenes in vivo. Cell 49:465-475. Slade MJ , Coope RC, Gomm JJ and Coombes RC. (1999) The human mammary gland basement membrane is integral to the polarity of luminal epithelial cells. Exp Cell Res. 247(l):267-78. Sneige N , Wang J, Baker B A , Krishnamurthy S and Middleton LP. (2002) Clinical, histopathologic, and biologic features of pleomorphic lobular (ductal-lobular) carcinoma in situ of the breast: a report of 24 cases. Mod Pathol. 15(10): 1044-50. Sommers CL, Thompson EW, Torri JA, Kemler R, Gelmann EP, Byers SW (1991). Cell adhesion molecule uvomorulin expression in human breast cancer cell lines: relationship to morphology and invasive capacities. Cell Growth Differ. 2(8):365-72. Sonnenfeld MR, Frenna TH, Weidner N , Meyer JE. (1991) Lobular carcinoma in situ: mammographic-pathologic correlation of results of needle-directed biopsy. Radiology. 181(2):363-7. Stahl S, Weitzman S and Jones JC. (1997) The role of laminin-5 and its receptors in mammary epithelial cell branching morphogenesis. J Cell Sci. 110 ( Pt l):55-63. Sternlicht M D , Lochter A , Sympson CJ, Huey B, Rougier JP, Gray JW, Pinkel D, Bissell MJ , and Z Werb. (1999) The stromal proteinase MMP3/stromelysin-l promotes mammary carcinogenesis. Cell. 98(2): 137-46. Stevenson BR, Siliciano JD, Mooseker MS, Goodenough DA. (1986) Identification of ZO-1: a high molecular weight polypeptide associated with the tight junction (zonula occludens) in a variety of epithelia. J Cell Biol . 103(3):755-66. 71 Streuli C H , Schmidhauser C, Bailey N , Yurchenco P, Skubitz AP, Roskelley C, and Bissell MJ . (1995) Laminin mediates tissue-specific gene expression in mammary epithelia. J Cell Biol. 129(3):591-603 Streuli, C. H. , Schmidhauser, C , Kobrin, M . , Bissell, M . J. and R Derynck. (1993) Extracellular matrix regulates expression of the TGF-beta 1 gene. J. Cell Biol . 120(l):253-60. Suzuki A , Yamanaka T, Hirose T, Manabe N , Mizuno K, Shimizu M , Akimoto K, Izumi Y , Ohnishi T and Ohno S. (2001) Atypical protein kinase C is involved in the evolutionarily conserved par protein complex and plays a critical role in establishing epithelia-specific junctional structures. J Cell Biol. 152(6): 1183-96. Suzuki F, Saito A , Ishi K, Okazaki T, Kina K, Koyatsu J and Sugiyama K. (1999) Secretory carcinoma of the breast: an immunohistochemical and ultrastructural study. Med Electron Microsc. 32(l):50-56. Sympson, C. J., Talhouk, R. S., Alexander, C. M . , Chin, J. R., Clift, S.M., Bissell M . J and Z. Werb. 1994. Targeted expression of stromelysin-1 in mammary gland provides evidence for a role of proteinases in branching morphogenesis and the requirement for an intact basement membrane for tissue-specific gene expression. J. Cell Biol. 125(3):681-93. Tabuse Y , Izumi Y , Piano F, Kemphues KJ , Miwa J and Ohno S. (1998) Atypical protein kinase C cooperates with PAR-3 to establish embryonic polarity in Caenorhabditis elegans. Development. 125(18):3607-14. Takahashi K , Nakanishi H , Miyahara M , Mandai K, Satoh K, Satoh A, Nishioka H , Aoki J, Nomoto A , Mizoguchi A and Takai Y . (1999) Nectin/PRR: an immunoglobulin-like cell adhesion molecule recruited to cadherin-based adherens junctions through interaction with Afadin, a PDZ domain-containing protein. J Cell Biol. 145(3), 539-49. 72 Takai Y and Nakanishi H. (2003) Nectin and afadin: novel organizers of intercellular junctions. J Cell Sci. 116(Pt 1), 17-27. Takeichi M . (1988) The cadherins: cell-cell adhesion molecules controlling animal morphogenesis. Development 102(4), 639-55. Tan C, Costello P, Sanghera J, Dominguez D, Baulida J, de Herreros A G , Dedhar S. (2001) Inhibition of integrin linked kinase (ILK) suppresses beta-catenin-Lef/Tcf-dependent transcription and expression of the E-cadherin repressor, snail, in APC-/ -human colon carcinoma cells. Oncogene. 20(1): 133-40. Tavassoli F A and Devilee P. (2003) Pathology and Genetics. Tumor of the breast and female genital organs. IARC Press, International Agency for Research on Cancer, Lyon France. Tavassoli F A and Norris HJ. (1990) A comparison of the results of long-term follow-up for atypical intraductal hyperplasia and intraductal hyperplasia of the breast. Cancer 65: 518-529. Teng, C. T., Pentecost, B. T., Chen, Y . H. , Newbold, R. R., Eddy, E. M and J. A . McLachlan. 1989. Lactotransferrin gene expression in the mouse uterus and mammary gland. Endocrinology. 124(2):992-999. Thor A D , Liu S, Moore D H 2nd, Shi Q and Edgerton SM. (2000) p(21WAFl/CIPl) expression in breast cancers: associations with p53 and outcome. Breast Cancer Res Treat. 61(1), 33-43. Tognon C, Knezevich SR, Huntsman D, Roskelley CD, Melnyk N , Mathers JA, Becker L, Carneiro F, MacPherson N , Horsman D, Poremba C and Sorensen PH. (2002). 73 Expression of the ETV6-NTRK3 gene fusion as a primary event in human secretory breast carcinoma. Cancer Cell., 2(5), 367-76. Topper, Y . J. and C. S. Freeman. (1980) Multiple hormone interactions in the developmental biology of the mammary gland. Physiol. Rev. 60(4), 1049-106. Tsukamoto T, Nigam SK. (1999) Role of tyrosine phosphorylation in the reassembly of occludin and other tight junction proteins. A m J Physiol. 276(5 Pt 2):F737-50. Tsukita S, Furuse M and Itoh M . (2001) Multifunctional strands in tight junctions. Nat Rev Mol Cell Biol. 2(4), 285-93. Umbricht CB, Evron E, Gabrielson E, Ferguson A , Marks J, Sukumar S. (2001) Hypermethylation of 14-3-3 sigma (stratifin) is an early event in breast cancer. Oncogene. 20(26):3348-53. Vos CB, Cleton-Jansen A M , Berx G, de Leeuw WJ, ter Haar NT, van Roy F, Cornelisse CJ, Peterse JL and van de Vijver MJ . (1997) E-cadherin inactivation in lobular carcinoma in situ of the breast: an early event in tumorigenesis. Br J Cancer. 76(9): 1131-3. Wang A Z , Ojakian G K and Nelson WJ. (1990) Steps in the morphogenesis of a polarized epithelium. I. Uncoupling the roles of cell-cell and cell-substratum contact in establishing plasma membrane polarity in multicellular epithelial (MDCK) cysts. J Cell Sci. 95 ( Pt 1):137-51. Wang TC, Cardiff RD, Zukerberg L, Lees E, Arnold A , Schmidt EV. (1994) Mammary hyperplasia and carcinoma in MMTV-cycl in D l transgenic mice. Nature 369:669-671. Weaver V M , Lelievre S, Lakins JN, Chrenek M A , Jones JC, Giancotti F, Werb Z and Bissell MJ . (2002) beta4 integrin-dependent formation of polarized three-dimensional 74 architecture confers resistance to apoptosis in normal and malignant mammary epithelium. Cancer Cell. 2(3):205-16. Weaver V M , Petersen OW, Wang F, Larabell C A , Briand P, Damsky C and Bissell MJ . (1997) Reversion of the malignant phenotype of human breast cells in three-dimensional culture and in vivo by integrin blocking antibodies. J Cell Biol. 137(l):231-45 Westley BR, Clayton SJ, Daws MR, Molloy C A and May FE. (1998) Interactions between the oestrogen and insulin-like growth factor signaling pathways in the control of breast epithelial cell proliferation. Biochem. Soc. Symp., 63: 35-44. White DE, Cardiff RD, Dedhar S and Muller WJ. (2001) Mammary epithelial-specific expression of the integrin-linked kinase (ILK) results in the induction of mammary gland hyperplasias and tumors in transgenic mice. Oncogene. 20(48):7064-72. Willott E, Balda MS, Fanning AS, Jameson B, Van Itallie C and Anderson JM. (1993) The tight junction protein ZO-1 is homologous to the Drosophila discs-large tumor suppressor protein of septate junctions. Proc Natl Acad Sci U S A . 90(16), 7834-8. Wittchen ES, Haskins J and Stevenson BR. (1999) Protein interactions at the tight junction. Actin has multiple binding partners, and ZO-1 forms independent complexes with ZO-2 and ZO-3. J Biol Chem. 274(49), 35179-85. Wittchen ES, Haskins J and Stevenson BR. (2000) Exogenous expression of the amino-terminal half of the tight junction protein ZO-3 perturbs junctional complex assembly. J Cell Biol. 151(4):825-36. Wittchen ES, Haskins J and Stevenson BR. (2003) NZO-3 expression causes global changes to actin cytoskeleton in Madin-Darby canine kidney cells: linking a tight junction protein to Rho GTPases. Mol Biol Cell. 14(5): 1757-68 75 Won J, K im D Y , La M , Kim D, Meadows GG, Joe CO. (2003) Cleavage of 14-3-3 protein by caspase-3 facilitates bad interaction with Bcl-x(L) during apoptosis. J Biol Chem. 278(21):19347-51. Wong V and Gumbiner B M . (1997) A synthetic peptide corresponding to the extracellular domain of occludin perturbs the tight junction permeability barrier. J Cell Biol. 136(2), 399-409. Wong V . (1997) Phosphorylation of occludin correlates with occludin localization and function at the tight junction. A m J Physiol. 273(6 Pt 1), C l 859-67. Yamanaka T, Horikoshi Y , Suzuki A , Sugiyama Y , Kitamura K, Maniwa R, Nagai Y , Yamashita A , Hirose T, Ishikawa H and Ohno S. (2001) PAR-6 regulates aPKC activity in a novel way and mediates cell-cell contact-induced formation of the epithelial junctional complex. Genes Cells. 6(8):721-31. Yang, Y . , Spitzer, E., Meyer, D., Sche, M . , Niemann, C , Hrtmann, G., Weidner, K . M . , Birchmeier, C and W. Birchmeier. (1995) Sequential requirement of hepatocyte growth factor and neuregulin in the morphogenesis and differentiation of the mammary gland. J. Cell Biol . 131(1), 215-226. Yang X F , Beamer WG, Huynh H , Pollak M . (1996) Reduced growth of human breast cancer xenografts in hosts homozygous for the lit mutation. Cancer Res. 56(7): 1509-11. Yant J, Buluwela L, Niranjan B, Gusterson B and Kamalati T. (1998) In vivo effects of hepatocyte growth factor/scatter factor on mouse mammary gland development. Exp Cell Res. 241(2), 476-81. Yap AS, Niessen C M and Gumbiner B M . (1998) The juxtamembrane region of the cadherin cytoplasmic tail supports lateral clustering, adhesive strengthening, and interaction with pl20ctn. J Cell Biol. 141(3), 779-89. 76 Yee D, Paik S, Lebovic GS, Marcus RR, Favoni RE, Cullen K J , Lippman M E and Rosen N . (1989) Analysis of insulin-like growth factor I gene expression in malignancy: evidence for a paracrine role in human breast cancer. Mol Endocrinol. 3(3): 509-17. Yokoyama S, Tachibana K, Nakanishi H , Yamamoto Y , Irie K , Mandai K, Nagafuchi A , Monden M and Takai Y . (2001) alpha-catenin-independent recruitment of ZO-1 to nectin-based cell-cell adhesion sites through afadin. Mol Biol Cell. 12(6): 1595-609. Zhao JJ, Gjoerup OV, Subramanian RR, Cheng Y , Chen W, Roberts T M and Hahn WC. (2003) Human mammary epithelial cell transformation through the activation of phosphatidylinositol 3-kinase. Cancer Cell. 3(5):483-95. Zhu X L , Hartwick W , Rohan T , Kandel R (1998). Cyclin D l gene amplification and protein expression in benign breast disease and breast carcinoma . Mod Pathol 11: 1082-1088. Zamir E and Geiger B. (2001) Molecular complexity and dynamics of cell-matrix adhesions. J Cell Sci. 114(Pt 20):3583-90. Ziyaie D, Hupp TR and Thompson A M . (2000) P53 and breast cancer. Breast. 9(5):239-46. 77 CHAPTER 2: Cell Shape and Basement Membrane-Dependent Tight Junction Formation and Polarization of Mammary Epithelial Cells 2.1 Summary Although cell-cell and cell-basement membrane interactions are known to contribute to epithelial cell polarity, the specific roles they play are less well defined. In the work presented here, I used a 3-dimensional culture system that closely mimics the formation of alveoli in the normal mammary gland to examine the assembly of TJs and polarization during morphogenesis. The ability to dissect the culture model into three levels of hierarchy allowed me to examine TJ formation and polarization independent of each other. At the first level of hierarchy, cells were maintained as monolayers where they formed AJs. In these monolayers, TJ proteins were present, but did not associate with the actin cytoskeleton. At the second level of hierarchy, changes in the cell shape and clustering allowed for the initial recruitment of TJ proteins to the membranes where they interacted with the actin cytoskeleton. Furthermore, TJ proteins co-localized with E-cadherin and P-catenin. Observed co-localization of ZO-1 with p-catenin supports the idea that pre-formed AJ function as scaffolds for the TJ formation. At the third level of hierarchy, the presence of a B M caused TJ proteins to separate from AJs and localize exclusively to the apical domains where a lumen was formed. This induction of polarity was partially disrupted by ct6 integrin blocking antibodies, which indicated that TJ apical polarization depended, at least in part, on a6 integrin interaction with the B M . Thus, it appears that the formation and polarization of TJs can be separated functionally, on the basis of cell shape and integrin-dependency. 78 2.2 Introduction The epithelium of the mammary gland is composed of individual acinar units and their connecting ducts, both of which consist of hollow lumens surrounded by polarized epithelial cells (McManaman and Neville, 2003). The development and maintenance of this polarized epithelium is essential for the form and the function of the gland. Furthermore, this polarized organization has to be continuously maintained in the differentiated gland, while accommodating cell proliferation and cell death. During breast cancer development, the mammary gland undergoes architectural changes that are marked by a disruption of cell polarity (Bissell and Radisky, 2001). Specifically, in infiltrating lobular carcinomas this disruption is marked by a loss of AJs and TJs. In contrast, in the much more prevalent infiltrating ductal carcinomas, AJs often remain while TJs are lost. Therefore, the loss of TJs may be an early and important event in ductal cancer progression. Little is known about how the TJs are formed and polarized in the glandular epithelium during development of the mammary gland, nor it is understood how TJs are disrupted during carcinoma progression. Clarification of this process may lead to the elucidation of new therapeutic targets as well as diagnostic markers. The use of 3D culture models gives us the opportunity to study the epithelial polarized architecture in vitro. Unlike the M D C K monolayer cultures that are commonly used to study epithelial polarity (Pollack et al., 2004), mammary epithelial cells grown on matrices, as 3D-spheroids, recapitulate features of the luminal epithelium of T D L U in vivo. These spheroids are polarized with central hollow lumens and are composed of non-proliferative cells surrounded by a B M (Petersen et al., 1992). They also commonly induce the deposition of B M components at the basal surface (Streuli and Bissell, 1990). In the presence of lactogenic hormones these spheroids induce the production of milk proteins. Therefore, these spheroids not only undergo glandular morphogenesis, but also are also functionally differentiated. Due to the ability to carry out biochemical and cell biological manipulations, while maintaining a similar 3D microenvironment to that of the in vivo T D L U , we are able to study genes that may help maintain or disrupt the polarized architecture. Thus, 3D-mammary epithelial cultures provide a unique opportunity to study the fundamental aspects of glandular morphogenesis such as cell-cell junction formation, cell-BM junction formation and lumen formation. These cultures also provide 79 an opportunity to study the architectural changes that take place during breast cancer progression. Several recent studies suggest that disruption or loss of TJ components may be linked to cancer progression (Kominsky et al., 2003; Hoevel et al., 2002; Swisshelm et al., 1999; Hoover et al., 1998), however, the molecular mechanisms that are responsible for modulation of the TJs are not determined. This prompted me to investigate how TJs form and polarize during morphogenesis. For this, I used a 3D culture model of mouse mammary epithelial morphogenesis (Figure 2.1; Roskelley et al., 1994; Somasiri and Roskelley, 1999; Somasiri et al., 2000), which allowed me to examine TJ formation and polarization in a hierarchical manner. In the first level of this hierarchical model, when functional mouse mammary epithelial cells (Scp2) are cultured on tissue culture plastic they do not undergo differentiation or morphogenesis, instead they grow as flat monolayers. In the second level, when the cells are placed on a non-adhesive substratum, cells round up and form small solid clusters to which I refer to as 'naked' clusters. In contrast to cell rounding, in the third level, when the naked clusters are mixed with a reconstituted B M gel, these rounded cell clusters undergo full morphogenesis to form polarized spheroids with hollow lumens (Figure 2.1). Furthermore, at the electron microscopic level, TJs are localized at the apical domains where cell-cell interactions are taking place (data not shown). The result is a fully-functional morphogenic unit, which closely resembles the lactating alveolus in vivo, and is able to differentiate and express milk proteins in the presence of lactogenic hormones. This culture model allowed me to examine the precise localization of TJ proteins ZO-1 and occludin during morphogenesis. The results from this study indicated that TJ proteins were not localized to the cell-cell interaction sites of the monolayer. During the morphogenic process, however, TJ proteins were gradually localized to the cell membranes where they formed the rudimentary TJs. At this stage, TJ proteins may use AJs as scaffolds to position themselves to the cell-membrane. Finally, in the presence of a B M , TJs separated from A J sites and were localized at the apical domains of the cells caging a central lumen. This polarized localization of TJs was partially disrupted by treatment with a ct6 integrin-blocking antibody. Based on these observations, it appears that the process of TJ formation and of TJ polarization can be separated on the basis of integrin-dependency. 80 2.3 Materials and Methods 2.3.1 Antibodies The E-cadherin and P-catenin mouse monoclonal antibodies were obtained from Transduction Laboratories (250 pg/ml; Lexington, K Y ) . Rat polyclonal ZO-1 antibody was purchased from Chemicon International (1000 pg/ml; Temecula, CA) while both mouse monoclonal occludin and rabbit polyclonal ZO-1 antibodies were purchased from Zymed Laboratories (500 pg/ml; 250 p/ml; South San Francisco, CA). Mouse, rat and rabbit IgGs were purchased from Jackson Immunoresearch Laboratories (West Grove PA) and were used as controls in all the studies. The function-blocking antibody against the a6 integrin subunit was purchased as an endotoxin- and azide-free form from BD Pharmingen (1 mg/ml; San Diego, CA). The monoclonal anti-mouse P-casein antibody was a gift from Dr. Kaetzel (University of Kentucky, Lexington, K Y ; Kaetzel and Ray, 1984). A l l the secondary antibodies were obtained from Jackson ImmunoResearch Laboratories. 2.3.2 Cell Culture The functional mammary epithelial cell strain C O M M A - I D is a cell strain that was initially derived from the mammary-alveolar epithelium of a mid pregnant mouse (Danielson et al., 1984). In the presence of exogenously added extracellular matrix some of the COMMA-D1 heterogenous cells respond to lactogenic hormones and are induced to express the milk protein P-casein. CID-9 is a population of cells that originated from the P-casein expressing cells of C O M M A - I D strain (Schmidhauser et al., 1990). The Scp2 line consists of a homogeneous, functional population of cells that was isolated by limited dilution cloning of the heterogeneous CID-9 line. Scp2 cells are composed of small cuboidal epithelial cells that grow as flat monolayers on tissue culture plastic and absolutely require exogenously-added E C M to undergo morphogenesis and induce P-casein expression (Disprez et al., 1993). 81 For routine monolayer culture, the cells were maintained in DMEM/F12 medium (1:1 v/v; Sigma, St Louis, MO) supplemented with 5% Fetal bovine serum (FBS; Hyclone, Logan, UT) and insulin (5 pg/ml; Sigma). To generate 'naked' clusters, cells were plated on poly (2-Hydroxyethyl Methacrylate) (polyHEMA; 1-4 mg/ml; Sigma, St Louis, MO) coated dishes or glass cover slips in serum-free DMEM/F12 medium supplemented with a full complement of lactogenic hormones (5 pg/ml insulin, 1 pg/ml hydrocortisone, 3 pg/ml prolactin). PolyHEMA is used as a surface coating agent to reduce or eliminate the adhesion of cells to the growth surfaces allowing floating cluster formation. For 3D B M culture, dishes were coated with Matrigel (Collaborative Research, Bedford M A ) and were allowed to gel for one hour in a 37°C incubator. Then the cells were plated on top of the gel. For overlay cultures cells were pre-clustered by plating on polyHEMA coated dishes for 8 hrs. These clusters were then collected and cooled down on ice for 15min. Clusters were allowed to settle to the bottom of the centrifuge tube or were spundown at 1 OOOrpm on a bench top clinical centrifuge. Then the clusters were mixed in cold serum-free DMEM/F12 medium supplement with lactogenic hormones as well as 1% Matrigel. Then these Matrigel coated clusters were plated on pre-cooled dishes or glass cover slips. Media changed every other day. MATRIGEL® Basement Membrane Matrix M A T R I G E L Basement Membrane is a solubilized basement membrane extracted from the Engelbreth-Holm-Swarm (EHS) mouse sarcoma, a tumor rich in extracellular matrix proteins. The major component is laminin, followed by collagen IV, heparan sulfate proteoglycans, entactin and nidogen. It also contains TGF-(3, fibroblast growth factor, tissue plasminogen activator and other naturally occurring growth factors of EHS tumor. 2.3.3 Immunofluoresence microscopy Monolayer, naked and 3D B M cultures were maintained on glass cover slips, fixed with -20°C MeOH for 15min, rinsed three times in phosphate buffered saline 82 (PBS). The cells were blocked (10 % normal goat serum (NGS)/ 1% bovine serum albumin (BSA) in PBS) for 20 min to reduce non-specific antibody binding, and incubated with antibodies against E-cadherin (1:500), P-catenin (1:100), claudin (1:250), occludin (1:250), and ZO-1 (1:250) at room temperature (60 min for monolayers; overnight for naked and 3D cultures). Cells were washed 3 times with PBS to remove unbound antibodies. Primary antibody binding was detected using FITC- or Texas red-conjugated secondary antibodies (1:100; Jackson ImmunoResearch). Cells were incubated with the secondary antibodies for 1 hr at room temperature. Then they were washed 3 times with PBS and were mounted on glass slides using 1% D A B C O in 95% glycerol. Coverslips were sealed using nail polish. Antibody concentrations were optimized by limited dilution series and IgG's were used as controls in the place of primary antibodies. The slides were stored at 4°C till imaged. To examine the cytoskeletal/junction-associated fractions of ZO-1, P-catenin and E-cadherin, live cell cultures were treated with the cross-linking agent dithiobis-succinimidylpropionate (200 mg/ml in PBS or DMEM/F12) for 10 min at room temperature. Then these cells were extracted with 0.25% TritonX-100 in PBS for 10 min at room temp. Extracted cells were rinsed 4 times with culture media and then once with PBS before fixation for 20min in ice cold MeOH. They were then antibody labeled and detected by immunofluoresence as described above. Images were collected using a BioRad Radience plus a confocal unit connected to a Zeiss Axiophot microscope. Unless specified, 63 X oil lens was used for imaging. When dual stained images were collected, the pinhole size of the confocal was maintained to be same in both channels and images were collected sequentially to avoid any bleed through. The images were analyzed using NIH image (v 1.62) software (http://rsb.info.nih.gov/nih-image/Default.html), and the plates were generated using Photoshop 6 imaging software. 2.3.4 Cell fractionation and Western blotting and immunoprecipitation The cytoskeletal and membrane associated fraction of ZO-1 and P-catenin was determined by serial fractionation (Nusrat et al., 2000; Piepenhagen and Nelson, 1998). Monolayer, naked cluster and 3-D cultures were lysed on ice in 0.25% TritonX-100 83 buffer (0.25% TritonX-100 in PBS and protease inhibitors; 2 pg/ml aprotinin, 10 m M PMSF, 1 pg/ml leupeptine, 1 pg/ml pepstatin A) and the lysates were gently rocked at 4°C for 15 min and centrifuged at 10,500 rpm for 10 min in a microfuge. The supernatant was transferred to a new tube and the pellet was re-extracted with 1% TritonX-100 buffer (1% TritonX-100 in PBS and protease inhibitors). The lysates was gently rocked at 4°C for 15 min and centrifuged at 15,000 rpm on a microfuge for 30min. The remaining pellet was re-extracted with RIPA/SDS (1:3 v/v; 150 m M Nacl, 50 m M Tris pH 7.4, 5 m M EDTA, 5% NP-40, 1% DOC and 0.1% SDS; 1% SDS in PBS) buffer supplemented with 1 m M sodium vanadate, 2 m M leupeptin, 1 m M PMSF, 10 m M NaF and 2 m M pepstatin A . This extract was then, sonicated and re-centrifuged for 30 min in a microfuge. To examine the total protein levels, cells were directly lysed in RIPA/SDS buffer and centrifuged for 30 min at 15,000 rpm on a microfuge. For ZO-1, 20 ug of total protein was separated on 7% SDS-PAGE gels and transferred to PVDF membranes (BioRad, Hercules, CA). For P-catenin, 5 ug of the lysates were separated on a 10% SDS-PAGE gel. The membranes were then blocked (4% BSA, 5% FBS and 0.5% NGS in TBS with 0.1% Tween-20) for minimum of 8 hours at 4°C. Blots were then rinsed once and probed with a rat anti ZO-1 antibody or mouse monoclonal anti P-catenin antibody over night at 4°. They were rinsed 3 times 5 min each at room temperature and antibody binding was visualized using HRP-labelled, species-specific IgG followed by enhanced chemiluminescence substrate (Amersham, Arlington Heights, IL). The secondary antibody binding was at room temperature for 1 hour. Quantification of protein bands was done by densitometry with the use of NIH image software. For this quantification the blots were scanned and the area under of each band was measured. For milk protein analysis, cells were released from Matrigel by treating with Dispase for 30min at 37°C incubater. Cells were collected and washed 3 times with base media (DMEM/F12) to remove the proreases and the Matrigel. Cells were maintained on ice during the washing steps. Cells were then lysed on ice in RIPA buffer containing protease inhibitors and 20 ug of total protein were separated on a 13% SDS-PAGE gel and was subjected to immunoblotting using a mouse monoclonal antibody against P-casein as described above. 84 For co-immunoprecipitation studies, 800 ug of protein from each fraction was diluted to 1 ml final volume with lysis buffer. Each lysates was cleared for 1 hr on a rocker with 25 pi of Protein A agarose beads (Sigma, St Louis MO) and 5 pg of rabbit IgG. Lysates were spundown at 6000 rpm the beads were discarded. Rabbit anti ZO-1 was added to each supernatent and placed on the rocker for 4 hours. After 4 hrs of incubation 25 pi of Protein A sepharose beads added to each and incubated over night. Beads were spundown and washed with lysis buffer 4 times at 4°C. A l l the samples were maintained at 4 °G. Finally, the beads were boiled for 10 min with gel loading buffer, spundown at 14,000 rpm on a microfuge and supernatants were separated on 10% SDS-P A G E gels. Western blotting was performed for P-catenin and ZO-1 as described above. In the protein cross-linking experiments, cells were treated with 2 pg/ml 3,3-Dithio-bis-(sulfosuccinimidyl)propionate (DSP; Jethmalani and Henle, 1998; Calbiochem, La Jolla, CA) for 20 min at room temperature, and then rinsed with media prior to lysing the cells. 2.3.5 Integrin blocking Scp2 cells were cultured on 4 mg/ml polyHEMA coated dishes for 8 hrs and were allowed to form naked clusters. Then, these clusters were incubated with the rat IgG-anti-ct6 integrin blocking antibody GoFD (5 pg/ml) or with normal rat IgG for 30 min at 37°C before they were cooled on ice for 5 min. The cells were initially plated on polyHEMA with half the final volume. The clusters were mixed with equal volume of cold serum-free media containing lactogenic hormones and 1% Matrigel and were then plated on 35 mm dishes for immunoblotting or on glass cover slips for immunofluoresence. The final concentration of antibody and IgG was 5pg/ml. The clusters were allowed to undergo differentiation for 3 days as described previously. Cells were collected by treating with Dispace for 30 min at 37°C, washed 3 times with culture media and were lysed in RIPA buffer for P-casein western blotting. To examine the cell polarity disruption, 3D-speroids were counted based on apical or basal localization of ZO-1. In each experiment, all the spheroids with apically localized ZO-1 were counted as polarized and spheroids with basally localized ZO-1 were counted as polarity disrupted. 85 2.4 Results 2.4.1 Basement Membrane Gel Culture Induces Tight Junction Formation and Polarization in Mammary Epithelial Spheroids. Scp2 mammary epithelial cells absolutely require an exogenously added B M (extracellular matrix; ECM) for functional differentiation (ie. milk protein gene expression; Roskelley et al., 1994). Thus, when cultured as flat monolayers, these cells did not express P-casein milk protein, even in the presence of lactogenic hormones (Figure 2.2A). In contrast, when scp2 cells were maintained on a reconstituted B M gel they formed functionally differentiated spheroids that mimicked the alveolar structures of the lactating mammary glands and express P-casein (Figure 2.2A). AJs were present in both the scp2 cell monolayers and in the spheroids. Specifically, E-cadherin and p-catenin localized continuously at points of cell-cell contact, even after Triton extraction (Figure 2.2B). This indicated that E-cadherin and P-catenin both associated with the cytoskeleton in forming functional junctions. Confocal analysis indicated that the A J proteins were evenly distributed along the z-axis (i.e. basolateral localization) between the cells in the monolayers and localized at all points of cell contact in the spheroids (see Figure 2.4 below). Contrary to AJs, there is a significant difference in cytoskeletal-associated TJ protein distribution between monolayers and spheroids on Matrigel. In the monolayers, ZO-1 and occludin were only occasionally localized at sites of cell-cell contact after Triton extraction and often this localization was discontinuous (Fig 2.2C). This indicates that flat, undifferentiated scp2 monolayers are unable to form continuous functional TJs, which agrees with previously published findings where TER, a measure of TJ function, is much reduced under these conditions (Woo et al., 2000). However, the inability of monolayers to form functional TJs in my experiments was not due to a deficit in the plasmamembrane-associated f-actin as f-actin mimicked the staining pattern of Triton extracted AJ proteins (Figure 2.2B). 86 In contrast to what was seen in the monolayers, in fully differentiated spheroids, ZO-1 and occludin invariably localized to sites of cell-cell interaction after Triton extraction (Figure 2.2C). Unlike the A J proteins, these cytoskeleton-associated TJ proteins were unevenly distributed in the spheroids. Specifically, ZO-1 and occludin were localized to the apical domains that lined the central lumina of the spheroids (Figure 2.2C and Fig 2.3C). Under these conditions the majority of the f-actin was in these apical regions of cell-cell contact. Furthermore, at the electron microscopic level, TJs are localized at the apical domains where cell-cell interactions are taking place. Taken together, these, observations indicate that the TJs both form and polarize in Matrigel-induced spheroids. This is different from other commonly used mammalian culture models shuch as M D C K , where cells form polarized TJs in monolayers and it is difficult to separate TJ formation from TJ polarization. Therefore, I further characterized the B M dependent TJ formation and polarization. 2.4.2 Differential Regulation of Tight junction Formation and Polarization The BM-dependent differentiation of scp2 cells is regulated in two sequential steps (Roskelley et al., 1994). First, cell rounding initiates a pre-lactation state, which is reflected by the transcriptional regulated induction of lactofferrin expression (Close et al., 1997). Secondly, laminin-dependent integrin signalling initiates lactational, terminal differentiation of these rounded cells. This is reflected by the induction of transcription of the P-casein gene and expression of the milk protein (Muschler et al., 1999). The cell rounding state can be reached in the absence of exogenously added B M by pre-clustering the cells in suspension, followed by attachment to tissue culture plastic in the absence of serum, which prevents cells from spreading ('naked' clusters; Somasiri and Roskelley, 2000). When assayed for milk protein expression, these 'naked' clusters did not functionally differentiate to express P-casein (Figure 2.3A). Furthermore, these clusters did not contain apically localized actin, which strongly suggests they were not polarized. When these naked clusters were subsequently overlaid with a soluble B M matrix, the majority of the actin cytoskeleton were reorganized, localizing at the apical 87 domain surrounding the newly formed central lumen. A J , were formed in both naked clusters and overlay clusters, localizing both E-cadherin and P-catenin at cell-cell interaction sites. This was observed throughout the clusters. Cell rounding by culturing them on polyHEMA led to the localization of TJ proteins to the outer membranes of the naked clusters. Thus, changes in the cell shape and clustering were sufficient to localize TJ proteins ZO-1 and occludin to the cell membrane. Terminal differentiation can be achieved by overlaying these naked clusters with a soluble B M matrix (Somasiri and Roskelley, 2000). These 'overlaid' clusters express P-casein (Figure 2.3A). Cells within these overlaid structures form AJs and TJs similar to the naked clusters, however, under these conditions, the B M overlay clearly induced polarizing signals since TJ proteins and f-actin became apically targeted (Figure 2.3B and Figure 2.3C). These data suggest that cell rounding and clustering induce TJ formation, while the signals from the cell-BM initiate TJ polarization. The data presented in Figures 2.2 and 2.3 suggests that the morphogenic manipulation of scp2 cells can be used to study TJ dynamics in regulatable, sequential fashion. This is potentially important as other culture models rely on kinetics alone to examine TJ dynamics. To ensure that this tentative conclusion, based on immunofluoresence staining, was indeed correct, I compared the relative amounts of AJ and TJ proteins in whole cell lysates as well as in Triton-soluble and Triton insoluble fractions using Western blotting techniques. Clearly, morphogenic manipulation of scp2 cells had little or no effect on the steady state levels of E-cadherin and P-catenin as the steady state levels of these A J proteins were the same in whole cell lysates of monolayers, naked clusters and overlaid clusters (Figure 2.4A). The steady state levels of the TJ protein ZO-1, however, were possibly affected by the morphologic manipulation; relative to the monolayers, naked. clusters showed possible increased levels of ZO-1 r while overlaid clusters showed possible decreased levels. The steady state levels of occludin showed a possible increase from monolayers to overlaid clusters. This observation was further supported by immunofluoresence data, where monolayers had little or no TJ protein staining upon Triton extraction (Figure 2.2). Thus, to examine the distribution of junction proteins during hierarchy of morphogenesis I isolated different Triton soluble and insoluble protein fractions from these cells. The morphogenic 88 manipulation had little or no effect on the association of AJ proteins; in monolayers, in naked clusters and in overlaid clusters, significant proportions of both E-cadherin and P-catenin were present in the 1% Triton insoluble pool (Figure 2.4B) and the ratio of soluble to insoluble pools did not change under each of these morphological conditions (Figure 2.5A). In contrast, morphological manipulation affected the association of TJ proteins with the cytoskeleton. There was an increase in the soluble ZO-1 in monolayers compared to naked and overlaid clusters (Figure 2.4B). Specifically the 0.25% Triton soluble fraction of ZO-1 was significantly higher in monolayers. This was further reflected by an increase in the ratio of Triton insoluble to soluble fractions in naked clusters (Figure 2.5B). I should further point out that in monolayers, 0.25% Triton soluble fractions of E-cadherin and P-catenin pools were very small relative to the ZO-1 pool (Figure 2.4B). This suggests that in monolayers, the majority of A J proteins are already associated with the membranes forming AJs. 2.4.3 Adherens and Tight Junction Proteins Interact in Naked Cell Clusters AJs and TJs associate with the actin cytoskeleton, at least in part, via direct interactions with ot-catenin/p-catenin complex and ZO-1 respectively. Thus, I tested whether the cytoskeletal-associated pools of these proteins co-localized under various morphological conditions. As shown in Figure 2.6, a Z-series analysis of dual P-Cat/ZO-1 immunofluoresence staining indicated that there was no colocalization in flat monolayers. This was expected, given the reduced level of ZO-1 association with the cytoskeleton under these conditions. In naked clusters, cytoskeletal P-catenin was found at all points of cell-cell interaction; cytoskeletal ZO-1 was more restricted as the majority of the protein was at sites of cell-cell contact at the periphery of the spheres. Regardless, there was a considerable co-localization of the two proteins in naked clusters (ie. yellow; Figure 2.6). In overlaid clusters, cytoskeletal P-catenin was still found at the point of cell-cell interactions, while cytoskeletal ZO-1 was more tightly restricted to the central, apical domains of cell-cell interactions. Under these conditions there was little co-localization of the two proteins. Instead, it appeared that ZO-1 was restricted to the apical domain. With 89 the use of the confocal microscope I was able to complete image the spheroids eliminating the need of physically sectioning each spheroid and imaging individual sections. These observations suggest that the AJs, which functionally associate with the cytoskeleton, regardless of morphology, might act as a scaffold for TJ formation in unpolarized cells. Subsequently, when the clusters are overlaid with B M , the two junctional domains physically separate as apical/basal polarity is generated. 2.4.4 ZO-1 and p-Catenin Physically Interacts During Tight Junction Formation. In M D C K cells, ZO-1 has been shown to interact with the adherens junction complex (Rajasekeran et al., 1996). Therefore, I determined i f such an interaction occurs in any of the morphological conditions in our model. This was done by co-immunoprecipitations of ZO-1 and B-catenin in the Triton soluble and insoluble fractions (Figure 2.7). As was expected, there was no interaction between the two proteins in flat monolayers. In contrast, in the naked clusters there was considerable interaction between ZO-1 and 13-catenin, which agrees with the immunofluoresence co-localization data described above (see Figure 2.6). The strength of this interaction was evident from the fact that protein cross-linking was not necessary to maintain the interaction during the cell lysis. Furthermore, the fact that that there was little interaction between the two proteins in the basement membrane clusters also agrees with the immunofluoresence data. However, because of the high SDS concentrations needed to extract the Triton insoluble fractions the complex dynamics might have changed and the ZO-l/p-catenin interaction disrupted. Furthermore, the SDS treatment could have also led to the disruption or masking the epitope leading to a negative antibody binding in this fraction. Taken together immunofluoresence data and triton soluble fraction interactions suggest that the adherens junctions, which are functionally associated with the cytoskeleton regardless morphology, may act as a scaffold for tight junction formation in unpolarized cells. Subsequently, when B M driven apical/basal polarization occurs the two junctional domains become physically separated (Figure 2.6 above). 90 2.4.5 Integrin Signaling Initiates Tight Junction Polarization in Basement Membrane Overlaid Clusters cc6 Integrin signaling is required for lactogenic differentiation in B M overlaid clusters of scp2 cells (Muschler et al., 1999). ct6 signaling is also implicated in mammary epithelial cell apical/basal polarization. Therefore, I investigated whether this signaling was involved in TJ polarization. As seen in Figure 2.8A, the ct6 function blocking antibody, G0H3, appiered to partially inhibit lactogenic P-casein expression in overlaid clusters. However, the inhibition of P-casein was not as significant as previously observed (Muschler et al., 1999) Therefore, in my experiments the GoH3 antibody may not have been efficient in blocking ct6 integrin. While the GoH3 treatment had no effect on p-catenin localization, it partially disrupted ZO-1 apical localization (Figure 2.8B). Thus, GoH3 treatment re-elicited ZO-1 co-localization with P-catenin at the periphery of the overlaid clusters as is normally seen in naked clusters (Z-plane; Figure 2.8B). The number of clusters that weren't completely polarized increased dramatically from IgG control to GoH3 treated clusters (Fig 2.8C). These data are representative of four independent blocking experiments. Therefore, blocking BM-dependent a6 integrin signaling partially disrupts TJ polarization. 2.5 Discussion Although cell-cell interactions and cell-BM interactions are known to contribute to the epithelial cell polarity, the specific roles they play have been less well defined. In the work presented here, I used a 3D culture system (Roskelley et al., 1994) that closely resembles alveoli (lobular units) of the normal mammary gland to examine epithelial TJ formation and polarization during morphogenesis. First, I examined how TJs are formed in highly ordered 3D epithelial spheroids and subsequently the role of cell-BM interactions in the polarization and differentiation of these spheroids. The unique ability to dissect this culture model into three levels of hierarchy allowed me to examine TJ formation and polarization independently of each other. 91 Previous work by Gumbiner et al., (1988) and Rajasekaran et al., (1996) demonstrates that when M D C K cell monolayers are kept in micromolar C a 2 + concentrations, there is very little to no surface localization of E-cadherin and ZO-1. When Ca is added back to culture media, both E-cadherin and ZO-1 localize to the membrane. Furthermore, with the use of C a 2 + switch experiments on M D C K cells, Gumbiner and Simons, (1986) demonstrate that TJ formation is blocked and transepithelial resistance decreases when the cells are treated with an E-cadherin blocking Ab prior reintroducing Ca 2 + . These observations indicate a close relationship between AJs and TJs during the formation of TJs. In the first level of hierarchy, when scp2 normal mammary epithelial cells are placed on tissue culture plastic, they form epithelial monolayers (Somasiri and Roskelley, 2000; Roskelley et al., 1994) with AJs forming at cell-cell interaction sites. Normally, at these interaction sites, underlying actin is concomitantly arranged with AJs to reinforce and strengthen the adhesion. These monolayers showed only minimal localization of the TJ proteins to the cell-cell interaction sites. The proteins that localized to the membranes were weakly associated with the actin cytoskeleton and were easily extracted by treatment with low levels of detergent while E-cadherin and (3-catenin were unaffected by this treatment. Even though there was very little to no staining of ZO-1 in these monolayers, Western blot analysis indicated a substantial steady state level of ZO-1. However, this lack of immunofluoresence staining may be due to the diffuse distribution of ZO-1 throughout the cytoplasm. This was evident by the increase in ZO-1 levels even at the low triton soluble pool. These observations agree with Woo et al., (2000) where they demonstrate that when cultured on filter membranes as monolayers, Scp2 cells do not form proper functional TJs. In the second level of hierarchy, naked clusters formed AJs as observed in the monolayers. Interestingly, at this stage, TJ proteins, ZO-1 and occludin, were also localized to the membrane where cell-cell interactions take place. This observation further confirms that TJ formation depends on prior A J formation. The naked clusters did not contain a lumen and TJs did not polarize. The changes in the cell shape and cell clustering were sufficient to localize ZO-1, occludin and claudin-1 to the cells membrane where much of the actin cytoskeleton was localized. This observation further supports the 92 fact that cell shape changes can initiate morphogenic signals. In different culture systems such as keratinocytes (Watt, 1987), retinal pigment epithelial cells (Opas, 1989), steroidogenic cells (Roskelley and Auersperg, 1993) and hepatocytes (Mooney et al., 1992) cellular differentiation initiates by cell rounding alone. Thus, changes in the mechanical properties alone allows the formation of TJs, although they are not at the proper polarized locations and there is no end point differentiation with milk protein induction. One of the major advantages in our hierarchical culture system is that it allows me to examine TJ formation and polarization independent of each other, which is difficult in M D C K models where cells only form fully polarized monolayers that can not be regulated actively (Pollack et al., 2004). These M D C K monolayers form both AJs and TJs making it difficult to examine the formation these junctions independent from each other. Furthermore, in 3D cultures M D C K cells are able to deposit it's own laminin making it further difficult to examine the role of the B M in TJ polarization. However, in the third level of hierarchy, introducing a B M matrix onto the naked mammary epithelial clusters further modifies the 3D architecture. This induced TJ localization at an apical position surrounding the newly formed central lumen. Interestingly, at this stage the majority of the actin also was relocalized to the apical domain caging the lumen. This association of the actin cytoskeleton agrees with the evidence implicating its role in regulating paracellular permeability. In M D C K cells, expression of dominant-negative mutants or constitutively active mutants of actin organizing small G proteins RhoA, Rac and Cdc42 dramatically affects both barrier and fence functions of TJs and affects the establishment of cell polarity (Rojas et al., 2001; Jou et al., 1998). Arf6 is another distant member of the family of small G proteins known to be involved in organizing the actin cytoskeleton in the apical domains of polarized M D C K cells (Altschuler et al., 1999). Overexpression of a constitutively active Arf6 mutant in M D C K cells leads to the formation of membrane ruffles, actin rearrangement and an increase in the turnover of AJs (Palacios et al., 2002; Palacios et al., 2001). These observations imply that actin organization is important in stabilizing the TJ at its polarized location. Furthermore, the actin-binding domain of ZO-1 (Fanning et al., 2002) may also facilitate the recruitment of actin to the junction complex. 93 The ratio of Triton insoluble and soluble fractions steadily increased from monolayers to naked clusters to overlaid clusters. This increase in the ratio further confirmed ZO-1 association with the cytoskeleton and membrane localization leading to junction formation. I also observed an increase in the level of ZO-1 expression upon cell rounding, which might suggest the necessity of an increase in the expression levels of TJ components during morphogensis. The co-localization of P-catenin and ZO-1 in the naked clusters suggests an interaction between the two molecules, however the functional significance of this interaction is still remains to be tested. To closely examine the relative distribution of ZO-1 and P-catenin, I created 3D reconstructions of confocal images of flat, naked and overlaid clusters. Analysis of these data sets from different angles clearly indicated a co-localization of ZO-1 and P-catenin in the naked clusters. Furthermore, since each of these confocal images were collected at 0.4 p-thickness z-step size, I was able to precisely examine the co-localization of ZO-1 and P-catenin at immunofluoresence level. Thus, I propose that P-catenin may be involved in shuttling ZO-1 to the membrane and that AJs function as scaffolds to assemble TJs. Similar localizations have been shown between ZO-1 and cc-catenin in other systems (Itoh et al., 1997), although I failed to detect it in our system. I have shown that TJs are able to undergo polarization only in the presence of an exogenously added B M . In the normal gland, ct6p4 integrin initiates the formation of hemidesmosomes, which anchor epithelial cells to the B M , preventing cell migration and inducing differentiation (Muschler et al., 1999). Previously it was shown that blocking a6 integrin lead to disruption of functional differentiation i.e. induction of P-casein in mammary epithelial spheroids (Muschler et al., 1999). Moreover, during breast tumor progression it is believed that loss or alteration of signals mediated by a6p4-integrin leads to tissue disorganization and increased invasive properties (Weaver et al., 1997). M y results on a6 integrin blocking studies further corroborate with these previous studies and suggest that integrin-BM interaction is a major contributor to TJ polarity. However, the signaling mechanisms that may regulate this event are still to be elucidated. 94 A Model for Tight Junction Formation and Polarization. Initial cell-cell interaction mediated by E-cadherin, establishes an intracellular signaling cascade that mobilizes TJ proteins to the plasma membrane. This is followed by accumulation of ZO-1 and P-catenin at the plasma membrane. Thus, AJs function as a scaffold to bring TJ proteins to the membrane and subsequent TJ assembly. Initially, these junctions are not functionally polarized and reside on the outer surface (future basal surface) of the epithelial cells. Upon cell-BM interactions, a6 integrin engages and TJs now translocate to the apical membrane domains, which surround the central lumen. The maturation and functional polarization of the TJ depend on integrin engagement, which leads to the dissociation of ZO-1 from P-catenin and formation of an independent junction complex, localizing apical to the A J complex. 95 Figure 2.1 Mammary Epithelial Morphogenesis Model First level Flat; When mammary epithelial cells are placed on tissue culture plastic they form 2D monolayers. Second level Naked; When cells are placed on polyHEMA coated dishes, cells aggregate to form clusters of cells with no central lumen. Third level Overlay; When naked clusters are mixed with soluble B M , cell clusters cavitate and become apically/basally polarized. Stars represent the soluble B M . On Gel; When cells are placed on a reconstituted B M gel, cells aggregate and pull the gel around them to form polarized spheroids with a central lumen. 96 Flat On Gel Figure 2.2 Basement Membrane Gel Culture Induces the Formation of Differentiated and Polarized Mammary Epithelial Spheroids. (A) When Scp2 cells were cultured on reconstituted basement membrane (Matrigel) gel culture they differentiated to express P-casein milk protein (F = flat monolayers, G = B M gel culture). (B) In flat monolayer culture, f-actin and A J proteins E-cadherin and P-catenin were localized at cell-cell interaction sites. When these cells were placed on a reconstituted basement membrane gel, A J proteins remained localized at cell-cell interaction sites while the majority f-actin was localized in the apical region surrounding the central lumen. (C) Flat monolayers did not efficiently localize the TJ proteins ZO-1 and occludin to the membrane. However, in basement membrane gel culture, ZO-1 and occludin were localized to the apical region of the spheroid caging the central lumen. 98 99 Figure 2.3 Differential Regulation of Tight Junction Formation and Polarization. (A) When the scp2 cells were forced to form round clusters ('naked' clusters) by placing them on polyHEMA-coated dishes, they did not induce P-casein. However, when these naked clusters were overlaid with a matrigel, they induced P-casein (N = naked clusters, O = B M overlay). (B) Immunofluoresence staining and confocal microscopy showed that A J proteins were localized at cell-cell interaction sites of these naked clusters and overlaid clusters. (C) However, TJ proteins occludin, ZO-1 and claudin-1 were localized to the outer surface of naked clusters and when a basement membrane was overlaid on these naked clusters, TJs were driven to the apical membrane region of the spheroid where they caged the lumen. Thus the overlay induced TJ polarization. Thus, the combined observations from Figure 2.2 and 2.3 show the hierarchy of morphogenesis; flat monolayer, naked cluster and overlaid clusters. Bar =10 um. 100 101 Figure 2.4 Sequential and Relative Distribution of Adherens and Tight Junction Proteins During Spheroid Formation. (A) Monolayer, naked cluster and overlaid spheroid whole cell lysates were assayed for both A J and TJ total proteins levels. There was no effect on the total E-cadherin and B-catenin levels by morphogenic manipulation. However, there was a slight increase in ZO-1 and occludin total protein levels during the process of naked cluster formation. (B) Cells were subjected to sequential Triton-XlOO extraction to collect triton soluble and insoluble proteins. Morphogenic manipulation had no effect on E-cadherin and (3-catenin levels (B) and this was further evident by the unchanged ratio of triton soluble to insoluble protein levels (Figure 2.5). In contrast, morphogenic manipulation did affect the soluble fractions of tight junction proteins. There was less ZO-1 or occludin associated in the insoluble fractions of monolayers compared to naked clusters (B). This was further evident by the increased ratio in the naked clusters (Figure 2.5). (F= flat monolayer; N= Naked cluster; 0= Overlaied spheroid). This data is a representation of four independent trials. 102 Total .25%TX-100 1%TX-100 TX-100 Soluble Soluble Insoluble F N O F N O F N O 103 Figure 2.5 The Ratios of Soluble vs Insoluble Junction proteins During Spheroid Formation. The relative amount of proteins in each fractions were calculated by using NIH image softwear to scan the area under each band of the Western blot in Figure 2.4B. Then the ratios between Triton soluble and insoluble fractions were calculated. (A) Morphogenic manipulation had no effect on A J proteins while (B) TJ proteins were increasingly localized to the Triton insoluble fraction. 104 Ratio of Insoluble vs Soluble Fraction of E-Cadherin and B-Catemn 6 1 • E-Cadherin • ft-Catenin 5 <' 4 A * 3 FLAT NAKED OVERLAY 7 i Ratio of Insoluble vs Soluble Fraction of ZO-1 and Occludin • ZO-1 • Occludin FLAT NAKED OVERLAY 105 Figure 2.6 Adherens and Tight Junction Proteins Colocalized in Naked Clusters and Separate During Spheroid Formation. Monolayers, naked clusters and overlaid spheroids were fixed and were dual immunostained for p-catenin (green) and ZO-1 (red). Confocal stacks of each image were collected with the same pinhole size on both channels. 0 .4p Sections of top, middle and bottom plains of each channel were merged to examine the colocalization. There was no colocalization in the flat monolayer. In the naked cluster, ZO-1 was localized to the outer membrane where it colocalized with P-catenin (yellow). Note, there was no lumen or apically localized ZO-1. In the overlaid spheroid, ZO-1 was only localized to the apical domain as seen in the mid plane. There was little or no colocalization of ZO-1 and P-catenin. Optical sectioning through the Z-plane further demonstrates the colocalization and separation. The Z-plane image was generated by digitally sectioning the confocal stack in a vertical plane. 106 Top Plane Mid Plane Bottom Plane Z plane FLAT Top Plane Mid Plane Bottom Plane N A K E D OVERLAY 107 Figure 2.7 ZO-1 and P-Catenin Physically Interacts in Naked Clusters Monolayers, naked clusters and overlaid clusters were sequentially extracted with Triton buffer and immunoprecipitated with anti-ZO-1 antibody and blotted against anti-P-catenin antibody. There was a strong interaction between ZO-1 and P-catenin in the naked clusters, which agreed with immunofluoresence co-localization data (Figure 2.7). (A) There was very little interaction observed in overlaid clusters. (B) Protein cross-linking prior to lysis had no increased advantage during the immunoprecipitation, which indicated that this was relatively a strong interaction. 108 A .25%TX-100 1%TX-100 TX-100 Soluble Soluble Insoluble F N O F N O F N O ZO-1 S-Cat B 1% TX-100 Soluble F N F N ZO-1 li-Cat No X-link X-linked 109 Figure 2.8 a6 Integrin Blocking Antibody Disrupts Tight Junction Polarization. Naked clusters were mixed with a a6 integrin blocking antibody just prior to overlaying with a basement membrane. IgG was used as the control condition, a.6 blocking antibody led to a partial disruption of P-casein induction (A). Furthermore, apical localization of ZO-1 was also disrupted. Note the basal localization of ZO-1 (B). However, there was no effect on E-cadherin localization. The spheroids with mislocalized ZO-1 were counted for stastical analysis. There was an increased % of spheroids that were unable to polarize in the presence of a6 blocking Ab. Also note that there was slight effect from the IgG control (C). This is representation of three independent trials. 110 IgG GoH3 G-Casein B IgG Control a6 Block (GoH3) G-Catenin ZO-1 V . Merged Z Plane GoH3 Treatment Partially Dtsruptes ZO-1 Polarity • 20-) Polarised DZO-t Polarity Disrupted IgG Control 111 2.6 References Altschuler, Y . , Liu, S.-H., Katz, L., Tang, K. , Hardy, S., Brodsky, F., Apodaca, G., and Mostov, K . (1999) ADP-ribosylation factor 6 and endocytosis at the apical surface of Madin-Darbycanine kidney cells. J. Cell Biol. 141, 7-12. Bissell M J and Radisky D. (2001) Putting tumours in context. Nat Rev Cancer. l(l):46-4. Close MJ , Howlett AR, Roskelley CD, Desprez PY, Bailey N , Rowning B, Teng CT, Stampfer M R and Yaswen P. (1997) Lactoferrin expression in mammary epithelial cells is mediated by changes in cell shape and actin cytoskeleton. J Cell Sci. 110 ( Pt 22):2861-71. Fanning AS, Ma T Y and Anderson JM. (2002) Isolation and functional characterization of the actin binding region in the tight junction protein ZO-1. FASEB J. 16(13): 1835-7. Gumbiner B, Simons K. (1986) A functional assay for proteins involved in establishing an epithelial occluding barrier: identification of a uvomorulin-like polypeptide. J Cell Biol . 102(2):457-68. Gumbiner B, Stevenson B, Grimaldi A . (1988) The role of the cell adhesion molecule uvomorulin in the formation and maintenance of the epithelial junctional complex. J Cell Biol . 107(4):1575-87 Hoevel T, Macek R, Mundigl O, Swisshelm K and Kubbies M . (2002) Expression and targeting of the tight junction protein CLDN1 in CLDN1-negative human breast tumor cells. J Cell Physiol. 191(1), 60-8. Hoover K B , Liao SY and Bryant PJ (1998) Loss of the tight junction M A G U K ZO-1 in breast cancer: relationship to glandular differentiation and loss of heterozygosity. A m J Pathol 153(6): 1767-73 112 Itoh M , Nagafuchi A , Moroi S, Tsukita S. (1997) Involvement of ZO-1 in cadherin-based cell adhesion through its direct binding to alpha catenin and actin filaments. J Cell Biol. 138(l):181-92. Jethmalani S M and Henle K J (1998). Calreticulin associates with stress proteins: implications for chaperone function during heat stress. J Cell Biochem. 69(l):30-43. Jou, T and Nelson, W.J. (1998b) Effects of regulated expression of mutant Rho A and Racl small GTPases on the development of epithelial (MDCK) cell polarity. J. Cell Biol. 142, 85-100. Kaetzel CS and Ray DB (1984) Immunochemical characterization with monoclonal antibodies of three major caseins and alpha-lactalbumin from rat milk. J Dairy Sci. 67(l):64-75. Kominsky SL, Argani P, Korz D, Evron E, Raman V , Garrett E, Rein A , Sauter G, Kallioniemi OP and Sukumar S. (2003) Loss of the tight junction protein claudin-7 correlates with histological grade in both ductal carcinoma in situ and invasive ductal carcinoma of the breast. Oncogene. 22(13), 2021-33 McManaman JL and Neville M C . (2003) Mammary physiology and milk secretion. Adv Drug Deliv Rev. 55(5):629-41. Mooney D, Hansen L, Vacanti J, Langer R, Farmer S and Ingber D. (1992) Switching from differentiation to growth in hepatocytes: control by extracellular matrix. J Cell Physiol. 151(3):497-505. Muschler J, Lochter A , Roskelley CD, Yurchenco P, Bissell MJ . (1999). Division of labor among the alpha6beta4 integrin, betal integrins, and an E3 laminin receptor to 113 signal morphogenesis and beta-casein expression in mammary epithelial cells. Mol Biol Cell. 10(9):2817-28. Nusrat A , Parkos CA, Verkade P, Foley CS, Liang TW, Innis-Whitehouse W, Eastburn K K , Madara JL. (2000) Tight junctions are membrane microdomains. J Cell Sci. 113 ( Pt 10):1771-81. Opas M (1989) Expression of the differentiated phenotype by epithelial cells in vitro is regulated by both biochemistry and mechanics of the substratum. Dev Biol . 131(2):281-93. Palacios, F., Price, L., Schweitzer, J., Collard, J.G., and D'Souza-Schorey, C. (2001). A n essential role for ARF6-regulated membrane traffic in adherens junction turnover and epithelial cell migration. E M B O J. 20, 4973-4986. Palacios, F., Schweitzer, J.K., Boshans, R.L., and D'Souza-Schorey, C. (2002). ARF6-GTP recruits Nm23-Hl to facilitate dynamin-mediated endocytosis during adherens junctions disassembly. Nat. Cell Biol. 11,11. Piepenhagen, P., and Nelson WJ. (1998) Biogenesis of polarized epithelial cells during kidney development in situ: roles of E-cadherin-mediated cell-cell adhesion and membrane cytoskeleton organization. Mol Biol Cell. 9(11):3161-77. Petersen OW, Ronnov-Jessen L, Howlett A R and Bissell MJ . (1992) Interaction with basement membrane serves to rapidly distinguish growth and differentiation pattern of normal and malignant human breast epithelial cells. Proc Natl Acad Sci U S A . 89(19):9064-8. Pollack A L , Apodaca G and Mostov K E . (2004) Hepatocyte growth factor induces M D C K cell morphogenesis without causing loss of tight junction functional integrity. A m J Physiol Cell Physiol. 286(3):C482-94. 114 Rajasekaran A K , Hojo M , Huima T, Rodriguez-Boulan E. (1996) Catenins and zonula occludens-1 form a complex during early stages in the assembly of tight junctions. J Cell Biol. 132(3):451-63. Rojas, R., Ruiz, W.G., Leung, S.-M., Jou, T.-S., and Apodaca, G. (2001). Cdc42-dependent modulation of tight junctions and membrane protein traffic in polarized Madin-Darbey Canine kidney Cells. Mol . Biol. Cell 72, 2257-2274. Roskelley C and Auersperg N . (1993) Mixed parenchymal-stromal populations of rat adrenocortical cells support the proliferation and differentiation of steroidogenic cells. Differentiation. 55(l):37-45. Roskelley CD, Desprez PY and Bissell MJ . (1994) Extracellular matrix-dependent tissue-specific gene expression in mammary epithelial cells requires both physical and biochemical signal transduction. Proc Natl Acad Sci U S A . 91(26):12378-82. Schmidhauser C, Bissell MJ , Myers C A and Casperson GF. (1990) Extracellular matrix and hormones transcriptionally regulate bovine beta-casein 5' sequences in stably transfected mouse mammary cells. Proc Natl Acad Sci U S A . 87(23):9118-22. Somasiri A , Roskelley CD. (1999) Cell shape and integrin signaling regulate the differentiation state of mammary epithelial cells. Methods Mol Biol. 129:271-83. Somasiri A , Wu C, Ellchuk T, Turley S, Roskelley CD. 2000. Phosphatidylinositol 3-kinase is required for adherens junction-dependent mammary epithelial cell spheroid formation. Differentiation. 66(2-3): 116-25. Streuli C H and Bissell M J (1990) Expression of extracellular matrix components is regulated by substratum. J Cell Biol. 110(4): 1405-15. 115 Swisshelm K, Machl A , Planitzer S, Robertson R, Kubbies M and Hosier S. (1999) SEMP1, a senescence-associated cDNA isolated from human mammary epithelial cells, is a member of an epithelial membrane protein superfamily. Gene. 226(2):285-95. Watt F M (1987) Influence of cell shape and adhesiveness on stratification and terminal differentiation of human keratinocytes in culture. J Cell Sci Suppl. 8:313-26. Weaver V M , Petersen OW, Wang F, Larabell CA, Briand P, Damsky C and Bissell MJ . (1997) Reversion of the malignant phenotype of human breast cells in three-dimensional culture and in vivo by integrin blocking antibodies. J Cell Biol. 137(l):231-45. Woo PL, Cercek A , Desprez P Y and Firestone GL. (2000) Involvement of the helix-loop-helix protein Id-1 in the glucocorticoid regulation of tight junctions in mammary epithelial cells. J Biol Chem. 275(37):28649-58. 116 CHAPTER 3: ETV6-NTRK3 fusion protein induces hyper-proliferation of mammary epithelial cells in 3D-culture without disrupting epithelial cell polarity 3.1 Summary ETV6-NTRK3 (EN) is a fusion protein generated by a t(12;15)(pl3;q25) chromosomal translocation which results in the expression of constitutively active tyrosine kinase. E N is expressed in a high proportion of highly differentiated secretory breast carcinomas (SBC), which are rarely metastatic. When the E N gene is expressed in a non-tumorgenic mouse mammary epithelial cell line, EpH4, these cells gain the ability to induce tumors in nude mice. These tumors are histopathologically quite similar to SBC. In this study, I determined the effects of E N expression on the normal mammary epithelial differentiation and morphogenesis. When E N expressing EpH4 cells were cultured on a B M , they formed differentiated spheroids that expressed P-casein. However, these spheroids were much larger than controls. Unlike control spheroids, cells in these EN-expressing spheroids showed increased levels of cyclin DI in response to insulin/IGF-1 and continued to undergo cell proliferation. Thus they are able to over ride the anti-proliferative signals generated by cell-BM interactions. E N expressing spheroids were multi-layered and were able to undergo proper polarization by forming apical TJs caging the central lumen. Therefore, I conclude that E N is capable of inducing cell proliferation while maintaining proper differentiation. This is consistent with the SBC phenotype. 3.2 Introduction The ETV6-NTRK3 (EN) is a fusion oncoprotein found in a majority of secretory breast carcinomas (SBC; Tognon et al., 2002). This fusion oncoprotein was first 117 described in congenital fibrosarcoma, which resulted from chromosomal translocation break points of t (12;15)(pl3;q25) (Knezevich et al., 1998). This translocation leads to the fusion between the sterile a motif (SAM), also known as the helix-loop-helix protein dimerization domain of the ETS transcription factor, ETV6, and the protein tyrosine kinase (PTK) domain of the neurotrophin-3 receptor NTRK3. The S A M domain allows this protein to undergo homodimerization, which facilitates ligand-independent constitutive activation of the PTK (Wai et al., 2000). When the E N gene is expressed in a normal mouse mammary epithelial cell line EpH4, these cells gain the ability to form differentiated tumors in nude mice (Tognon et al., 2002). These tumors are histopathologically quite similar to that of SBC. SBC is a rare type of IDC, which is characterized by an infiltrating pattern of neoplastic epithelial cells, forming well-differentiated glandular structures. Tumor cells of SBC are typically vacuolate with abundant extracellular and intracellular material, which is strongly positive for PAS and mucicarmine stains (Siegel et al., 1999; Rosen and Cranor, 1991; Oberman 1980). In these tumors, cells manage to maintain a phenotype similar to the T D L U architecture by forming cell clusters that are surrounded by a basal lamina (Suzuki et al., 1999). These organized cell clusters contain well-developed microvilli that protrude into a lumen, (Suzuki et al., 1999) indicating a polarized phenotype. These cells show epithelial cell characteristics by maintaining keratin expression (Suzuki et al., 1999). These glandular structures, furthermore, resemble salivary gland acinic cell carcinoma (Hirokawa et al., 2002). SBC was first described by McDivitt and Stewart, (1966), and is known to be the most common variant of childhood breast cancer. However, it has now been documented that it can also occur in all age groups and in both sexes (Page et al., 1987; Richard et al., 1990; de Bree et al., 2002; Paeng et al., 2003). In children this tumor is characterized by very high secretory activity. This rare type of IDC is known to have a better prognosis than most other IDC because of its slow progressing behavior (de Bree et al., 2002). There is new evidence, however, indicating that this prognosis depends on patient age; older patients tend to have a poor outcome. Thus, identification of E N fusion gene expression as the primary event in SBC (Tognon et al., 2002) may facilitate the identification of important downstream signaling pathways of therapeutic significance. The phenotype of tumors derived from E N -118 expressing EpH4 cells is that the cells are transformed, but they also retain epithelial differentiation as seen in SBC. Thus, E N may modulate the downstream signaling pathways of cell survival and proliferation without affecting the differentiation pathways. In this study, I have used a 3D culture model (see chapter 2) to determine the effects of E N expression in normal mammary epithelial differentiation and morphogenesis. As I have demonstrated, normal mammary epithelial cells in 3D, form acini structures with polarized epithelium caging a central lumen. Furthermore, these spheroids are grown arrested (Petersen et al., 1992). Therefore, these cultures are useful to examine the effects of molecules that could effect both normal mammary epithelial differentiation and luminal filling which is a critical initial step in breast tumor formation (Debnath et al., 2002; Muthuswamy et al., 2001). Here I have demonstrated that when E N was overexpressed in EpH4 cells, they underwent normal epithelial differentiation. However, unlike controls, E N spheroids were much larger in size and did not respond to anti-proliferative signals from the B M . There is increased Cyclin DI levels in E N transformed fibroblasts (Tognon et al 2001) and here I find in 3D culture, mammary epithelial cells also have increased levels of cyclin DI , and an induction.of cell proliferation forming multi-layered spheroids. Furthermore, this cell proliferation in 3D culture was absolutely dependent on both insulin/IGF-1 signaling and PI3-K activity. Strikingly, these spheroids still formed polarized epithelial cell-cell junctions while maintaining an architecture similar to that of SBC. These data indicate that E N is able to alter anti-proliferative signals induced by the B M in a growth factor dependent manner, inducing a SBC like phenotype. The ability to maintain normal cell polarity further explains the rare metastatic nature of SBC. 3.3 Materials and Methods. 3.3.1 Cell culture The EpH4 cell line consists of a homogenous, functional population of cells derived from the IM-9 mouse mammary epithelial cell line (Reichmann et al., 1989). 119 These cells were also originally derived from the mammary epithelium of a mid pregnant mouse mammary gland. They also do not deposit a B M and depends on exogenously added B M material. These cells are similar to the scp2 cells described in chapter 2. These cells were routinely maintained as described previously (Chapter 2, page 82). EpH4 parent cells were retrovirally transformed with MSCV-puromycin or E N - M S C V vectors to generate stable cell lines. Replication-defective ecotropic retrovirus was produced by transient transfection of the latter construct into BOSC-29 packaging cells (Pear et al., 1993). Culture meda from the transfected BOSC-29 cells were collected and used for infection or snap frozen on dry ice and stored at -80°C. Forty eight hours after retroviral infection, EpH4 cells were put under selection for 5 days (puromycin, 2 mg/ml) and then pooled. For 3D cultures, cells were serum and insulin starved for 24 hours prior to placing them on Matrigel coated dishes or coverslips. Then these cells were allowed to form spheroids as described in chapter 2 (page 82). In order to examine BrdU incorporation and cyclin DI expression levels, cells were serum and insulin starved for 24 hours and then plated on polyHEMA (4 pg/ml) coated dishes for 5 hours in DMEM/F12 media supplemented with 1 pg/ml hydrocortisone and 3 pg/ml prolactin, and were allowed to form naked clusters. These cell clusters were collected and mixed with media containing 1 % Matrigel with 5 pg/ml insulin or 200ng/ml IGF-1, and were plated on tissue culture dishes or glass coverslips. To assay for cell proliferation in 3D, cells plated on coverslips were incubated with BrdU labeling reagent (Roche Applied Science) for 1 hour at 37°C incubator. Then the cells were rinsed 3 times with culturemedia and once with PBS prior to fixing with EtOH at -20°C for 30 min. Cells were rinsed 3 times with washing buffer and were incubated with anti-BrdU working solution for 30 min at 37°C. After rinsing the cells 3 times with washing buffer, they were incubated with anti-mouse-fluorescine for 30 min at 37°C. Cells were then rinsed 3 times and mounted on glass slides as previously described. For assessing cyclin DI levels, cells were lysed at the times indicated in Figure 4 and 5. For PI3-K inhibition experiments, cells were further supplemented with LY294002 (25 pM; 120 Sigma) when adding Matrigel containing media. Since DMSO was used to dilute LY294002, DMSO was used as the control condition. 3.3.2 Immunofluoresence Microscopy Immunofluoresence microscopy was performed using antibodies against E-cadherin and ZO-1 as described in chapter 2 (page 83). Mouse and rabbit IgGs were used as controls. 3.3.3 Immunoprecipitation and Western blotting Lysates of EpH4-EN and EpH4-MSCV monolayers were prepared with RIPA buffer and were cleared by centrifugation at 15,000 rpm for 30 min in a micro centrifuge. 500 pg of total protein was immunoprecipitated with 5 pi of a-TrkC (C-14; Santa Cruz) antibody and probed with the same antibody as described previously (chapter 2, page 84). As described previously (Chapter 2, page 84), 3D cultures were lysed with RIPA buffer and 30 ug of total protein were assayed by Western blotting using antibodies against P-casein, cyclin D l (Clone 504; 1:2000; Upstate) and Grb-2 (1:5000; B D Biosciences). 10% SDS-PAGE was used for cyclin D l and Grb-2. 3.4 Results 3.4.1 Expression of EN did not cause a morphologic change in mammary epithelial cell monolayers. Normal mouse EpH4 mammary epithelial cells are polygonal in shape and form intimate contacts with each other in monolayers culture. At confluence, these cells form classical cobblestone shape monolayers that completely cover the bottom of the tissue culture dish. A retroviral infection system was used to introduce E N fusion oncoprotein into these cells. This produced a heterogeneous, stable pool of cells that expressed E N 121 (EpH4-EN). Similarly, a pool of control cells was generated by retrovirally introducing the empty vector into the parent cells (EpH4-C). When both EpH4-EN and EpH4-C cells were placed on a plastic tissue culture dish, they formed confluent monolayers that had a similar morphology to the parental cells (Figure 3.1 A). Expression of the transgene was detected by immunoprecipitation (Figure 3.IB), however, I was unable to demonstrate the E N expression using immuocytochemistry. This was due to the lack of an antibody that recognized the fusion gene at immunocytochemical level. In future studies epitope tagging E N will be useful to detect the localization in the transformed cells. 3.4.2 Expression of EN did not prevent extracellular matrix-dependent spheroid formation and differentiation. When EpH4 cells are plated on a B M , cells aggregate to form small spheroids that differentiate and express p-casein in a fashion similar to the scp2 line described in chapter 2. In addition EpH4 cell spheroids have a large central lumen lined by a single layer of polarized cells. Therefore, this cell line is useful to examine the effects of molecules that could effect both normal mammary epithelial differentiation and luminal filling which is a critical initial step in breast tumor formation (Debnath et al., 2003; Debnath et al., 2002; Muthuswamy et al., 2001). In order to analyze the functional effects of mammary epithelial morphogenesis, EpH4-EN and EpH4-C cells were allowed to undergo spheroid formation. As seen in Figure 3.2a, control cells formed spheroids of a relatively uniformed diameter of 50-100 um, while EpH4-EN cells were also able to form spheroids with a similar morphology to that of the controls. These spheroids were, however, two to three times larger than the control spheroids. This observation was consistent in multiple trials while there were no other morphological differences observed between the E N expressing and control spheroids. Furthermore, the spheroid formation was similar in both gel and overlay conditions (Figure 3.2). The ability to undergo functional differentiation of the cells was examined by the expression of milk protein P-casein. As seen in Figure 3.2c, Western blot analysis indicated that both EpH4-EN and EpH4-C spheroids induced p-casein. Thus E N expression does not disrupt the functional differentiation of the cells. 122 3.4.3 Expression of EN induced IGF-1/lnsulin dependent cell proliferation in 3D culture Unlike monolayer cultures, mammary epithelial cells grown as 3D cultures recapitulate the mammary acini, forming growth-arrested polarized spheroids with hollow lumens (Debnath et a l , 2002; Muthuswamy et a l , 2001; Weaver et al., 1997; Petersen et al., 1992). Reinitiation of cell proliferation in these growth arrested acini leads to the formation of altered structures with characteristics of early stage tumors (Muthuswamy et al., 2001). Furthermore, integrin mediated signals generated by cell-BM interactions may allow IGF signaling to proceed through PI3-K pathway (Farrelly et al., 1999; Lee and Streuli, 1999). In 3D culture, EpH4-EN spheroids demonstrated an altered structure by forming spheroids that were much larger than the controls. Therefore, I asked whether this increase in size was due to an increase in cell proliferation. To assay for this, I allowed cells to form spheroids for 72 hours and then labeled them for an hour with BrdU. As seen in Figure 3.3, in the absence of the growth factors insulin and IGF-1, there was no cell proliferation in the control cells or E N cells. In presence of insulin or IGF-1, however, EpH4-EN had a high number of labeled cells, while there was very little or no labeling in control spheroids. Thus, the proliferative effects of E N depend on growth factor signaling pathways. When I counted the number of spheroids with BrdU positive cells, ~ 2% of the controls had one or two positive cells per spheroid in the presence of insulin. In contrast, in the presence of insulin or IGF-1, ~ 40% of the E N -spheroid had four or more BrdU positive cells per spheroid. The number of cells that are undergoing cell proliferation in the short one-hour labeling reaction indicates the proliferative influence that E N has on these cells. Thus, this proliferation can account for the increase in size of these E N expressing spheroids. 3.4.4 Cyclin D1/2 levels remained high in EN expressing spheroids. Cyclin DI is up regulated during cell cycle progression through the Gi phase. It is also known that E N up regulates cyclin DI levels during fibroblast transformation (Wai 123 et al., 2000). Furthermore, E N expressing fibroblasts has increased levels of cyclin D l (Tognon et al., 2001). Hence, to explore whether the cyclin D l levels are associated with observed cell proliferation I examined total Cyclin D l levels during spheroid morphogenesis. Total lysates were prepared at 2, 5, and 48 hours of exposure to exogenously added B M . As seen in Figure 3.4, the expression levels of cyclin D l decreased when the control cells underwent spheroid formation. The expression levels detected up to 5 hrs could be the residual protein, since cells were collected from proliferative monolayer cultures prior to transfer onto a 3D culture. However, E N expressing spheroids managed to maintain high levels of cyclin D l throughout the 48 hr period. These high expression levels correlated with the cell proliferation observed above. Grb2 total protein levels were used as a loading control. 3.4.5 PI3-K is required for EN induced cell proliferation in 3D spheroids The ability of E N to transform fibroblasts depends on the presence of functional IGF-1 (Morrison et al., 2000). Furthermore, upon activating IGF-1 receptors, PI3-K signaling pathway can be activated through IRS-1 (LeRoith and Roberts Jr, 2003). E N is known to constitutively activate the PI3-K signaling pathway in NIH3T3 fibroblast cells (Morrison et al., 2000). Therefore, I next tested whether E N induced cell proliferation observed in mammary epithelial spheroids was also dependent on the activity of PI3-K. EpH4-EN and EpH4-C cells were allowed to undergo spheroid formation in the presence of a B M for 72 h. The cells were treated with the PI3-K inhibitor LY294002 at a 25 p M final concentration for the duration for the experiment. In these spheroids, the effect of PI3-K inhibition on the cell proliferation was measured by a one-hour BrdU incorporation assay (Figure 3.5A). LY294002 treatment completely suppressed cell proliferation that was observed in the presence of insulin or IGF-1, which was expected since both have been shown to signal through IGF-1R (Marshman and Streuli, 2002). Furthermore, LY294002 treatment completely inhibited the cyclin D l expression (Figure 3.5B) in the spheroids correlating with inhibition of cell proliferation. 124 3.4.6 EN expressing spheroids are able to undergo normal junction polarization. Next, I analyzed whether strong hyperproliferation exhibited by EpH4 spheroids was also associated with the disruption of spheroid polarity. The control EpH4 cells form single layered spheres where the A J protein E-cadherin was localized basolateraly at cells-cell interaction sites throughout the spheroid giving a ring like appearance. As expected (see chapter 2 above), TJs were specifically localized to the apical domains caging the central lumen as seen by ZO-1 localization (Figure 3.6). In contrast, the EpH4-E N spheroids were often multi-layered and there were regions where cells protruded into the lumen. This is resemblance of early stages of CIS formation. Interestingly, these spheroids were still able to localize E-cadherin at cell-cell contact sites similar to the controls. Furthermore, TJs properly formed at the most apical regions of the apical layer of cells. Thus, E N does not disrupt the polarity of the mammary epithelial spheroids. The multilayered but differentiated phenotype of E N spheroids is reminiscent of that which occurs in true SBC (Figure 3.6e). 3.5 Discussion The identification of E N gene fusion as a primary event in SBC (Tognon et al., 2002) illustrates an underlying concept of a dominant acting oncogene possibly functioning as the main initiator of breast carcinoma. When E N expressing mammary epithelial cells are injected into mice, they form differentiated non-metastatic tumors (Tognon et al., 2002) indicating E N can not only able to transform cells but also manages to maintain epithelial differentiation. In the work presented here, I demonstrated that tumor formation in vivo by E N expressing EpH4 cells correlated with the ability of these cells to hyper proliferate in 3D spheroid culture. Moreover, these spheroids underwent normal differentiation and polarization further describing the well-differentiated nature of SBC. Since these cells formed tumors in mice, I expected them to change to an invasive phenotype. Surprisingly, in 2D culture there was no phenotypic difference in the cells that expressed E N . Loss of tissue polarity and increased cell proliferation are characteristics of breast tumor progression, which can be studied in a 3D environment 125 that could recapitulate the tissue microenvironment (Bissell and Radisky, 2001). In 3D culture, EN-expressing cells formed spheroids with a phenotype that was similar to that of the EN-expressing tumors in mice as well as in SBC (Figure 3.6). The extracellular matrix components in the B M play a major role in the tissue architecture and homeostasis. When normal mammary epithelial cells are placed on a reconstituted B M gel, cells are able to undergo cell-cell and cell-BM dependent spheroid morphogenesis (Roskelley et al., 1994; Roskelley et al., 2000; Somasiri et al., 2000; Chapter 2). In the presence of proper lactogenic signals these spheroids also express the milk protein P-casein. When EpH4-EN and EpH4-C cells were placed on a reconstituted B M gel, both formed spheroids indicating that the E N expression alone does not inhibit the morphorgenic signal. Further, these spheroids expressed P-casein indicating that they underwent proper functional differentiation. However, the observation that E N expressing spheroids were always larger in size implied that they contained more cells per spheroid. Thus, I hypothesized that E N was inducing cell proliferation in these spheroids to account for the increase in size. When normal mammary epithelial cells are placed on a B M they stop proliferation and induce differentiation signaling (Debnath et al., 2003). It is well documented that when cultured on a laminin and collagen IV rich matrix (Matrigel) mammary epithelial cells are able to form acini-like structures with a single layer of polarized, growth-arrested cells (Muthuswamy et al., 2001; Weaver et al., 1997; Petersen et al., 1992). With the use of BrdU incorporation as an indicator of cell proliferation I demonstrated that E N expression alone was enough to induce a proliferation signal. Thus, E N is able to initiate a cell proliferation signal that is maintained in 3D culture. This induction in cell proliferation depended on the presence of growth factors insulin or IGF-1. The insulin receptor and IGF-1 receptor belong to a family of growth factor receptors with intrinsic tyrosine kinase activity. These receptors are known to stimulate a number of common intracellular signaling pathways. Upon ligand binding to their respective receptors, each of the hormones induces autophosphorylation of the receptor leading to the activation of intrinsic kinase activity (Reviewed by Van Obberghen et al., 2001). This receptor kinase can then phosphorylate a number of other proteins including insulin receptor substrate (IRS) family of proteins, She, Grb-associated 126 binder-1 and ppl20. Upon phosphorylation, these proteins bind to SH2 domain containing proteins and activate downstream signaling pathways. Specifically, IRS-1 and IRS-4 contain binding sites for Grb2 and PI3-K. Interestingly, the ability of E N to transform fibroblasts depends on the presence of functional IGF-1 receptors, which is evident by EN's inability to transform IGF-1 receptor null cells (Morrison et al., 2000). This would further explain the dependence of insulin or IGF-1 to induce cell proliferation in E N expressing EpH4 spheroids. One of the well-studied signaling pathways activated by insulin/IGF-1 is the PI3-K pathway. Distinctive and complementary roles of IGF-1 stimulated pathways have been reported. In fetal sheep cardiomyocytes (Sundgren et al., 2003) and in vascular smooth muscle cells (Jung et al., 2000), IGF-1 induces cell proliferation through the PI3-K pathway while; in skeletal myoblasts IGF-1 induces differentiation through PI3-K (Coolican et al., 1997). In a neuroblastoma cell line, however, the PI3-K pathway is required for both growth and differentiation. Furthermore, the PI3-K signaling is important for both proliferation and differentiation in mammary epithelial cells (Somasiri et al., 2000). Thus, cell or tissue specific activation of these pathways is clearly evident. The PI3-K pathway is constitutively activated in E N transformed 3T3 fibroblast cells (Wai et al., 2000). Furthermore, IGF receptor knockout fibroblast cells are not transformed by E N until IGF receptor is re-introduced to the cells (Morrison et al., 2002). When EpH4-EN spheroid cultures were treated with LY294002, a well-characterized PI3-K inhibitor, cell proliferation was completely inhibited. This was true in the presence of both insulin and IGF-1, further suggesting that this proliferative activity absolutely depended on the PI3-K activity and the insulin/IGF-1 receptor-signaling pathway. Inhibition studies indicated that increase in cyclin DI activity was dependent on PI3-K activity. Cyclin DI is a known oncogene and is an important key element in cell cycle progression (Sherr, 1996). In response to numerous mitogenic stimuli, Cyclin DI is able to activate its catalytic partners CDK4 and CDK6 and kick-starts the cell cycle. Amplification and overexpression of the cyclin DI gene can overcome the dependency of mitogenic stimulation and play a key role in the process of oncogenic transformation (Bartkova et al., 1997; Sherr and Roberts 1999). Therefore, it may not seem surprising that cyclin DI plays a pivotal role in breast cancer, since it is overexpressed at a 40% 127 incidence. Furthermore, the role of cyclin D in rapid cell growth is implicated by one major phenotype detected in cyclin Dl-knockout mice, where alveolar lobular cells fail to expand during pregnancy (Fantl et al., 1995). In fibroblast cells E N transformation leads to an upregulation of cyclin D l levels (Tognon et al., 2001). When mammary cells were placed in 3D culture, the residual cyclin D l levels remained high and began to decrease when cells underwent spheroid formation, which indicated a growth-arrest stage. When E N was expressed, however, cyclin D l levels remained high and increased over time. These results are consistent with recent studies indicating that PI3K signaling is required for the induction of cyclin D l (Gao et al., 2003; Muise-Ffelmericks et al., 1998; Brennan etal., 1997). The perturbations of epithelial junctions have been long implicated in tumor progression and invasion. The A J protein E-cadherin plays a major role as an invasion/tumor suppressor protein. Transcriptional down regulation or somatic mutations (Brex et al., 1998) of E-cadherin have been observed in many invasive cells and tumors. In pancreatic (3 tumorigenesis, E-cadherin mediated cell-cell adhesion is crucial in preventing the progression from well-differentiated to invasive carcinoma (Perl et al., 1998). In addition, TJ proteins play a major role in maintaining the normal mammary epithelial architecture. In this study I found that E N expression alone was not enough to disrupt cell-cell junctions, which led to the maintenance of polarized phenotype. Activation of ErbB2 receptors in differentiated acini also leads to reinitiation of cell proliferation to form non-invasive multi-layered acini (Muthuswamy et al., 2001) generating a phenotype that is quite similar to that of E N spheroids. However, unlike E N spheroids, activation of ErbB2 also leads to loss of polarized organization of the acini. Early stages of breast cancer such as hyperplasia and ductal carcinoma in situ are characterized by an increase in luminal epithelial cell proliferation, a loss of acinar organization and filling of the luminal space. This lack of organization and later acquisition of invasive properties can lead to cancer progression. In well-differentiated and well-behaved SBC, however, tumor cells organize into multi-layered glandular structures, which often do not become invasive. Thus, it is likely that E N alone is not sufficient to induce an invasive phenotype in EpH4 cells. Instead, additional events that allow the disruption of polarity be involved. Furthermore, it seems that E N expression 128 does not have an effect on the normal differentiation pathways in spheroid formation. Deregulation of both growth factor signaling pathways and PI3-K signaling pathways help to disrupt cell-cell junctions and induce an invasive phenotype in many cell types (Qiang et al., 2003; Zeng et al., 2003; Zhang et al., 2003; Ellerbroek et al., 2001), however, this deregulation does not occur in E N expressing EpH4 spheroids. In summary, these studies provide evidence that E N is able to induce cell proliferation in growth-arrested spheroids, generating multi-layered spheroids, which resembles the SBC phenotype. Furthermore, E N has no affect on the normal spheroid differentiation and polarization signals. The lack of polarity disruption may explain the non-metastatic nature of SBC. Therefore, my data are consistent with E N expression being an initiating event in SBC and this in vitro model provides a useful 3D culture model to further investigate the additional signals that may be required for SBC to become metastatic. 129 Figure 3.1 Expression of EN did not cause a phenotypic change in mammary epithelial cell monolayers. (A) EpH4 mouse mammary epithelial cells expressing E N (EpH4-EN) and vector control (EpH4-C) cells were cultured as monolayers. Note the classical 'cobblestone' epithelial morphology in both cell lines. (B) Whole lysates were subjected to immunoprecipitation with anti-TrkC antibody and probed with the same antibody. E N is seen as the 73 kDa and 68 kDa doublet. (Bar=10pm) 130 131 Figure 3.2 EN expressing cells are able to undergo spheroid formation and functional differentiation of mammary epithelial cells. Cells were cultured on reconstituted B M gels for 72 hours. Both E N expressing cells (a and b) and control cells (a' and b') were able to undergo normal spheroid formation. They were similar in both a gel and overlay culture. However, E N spheroids were much larger in size than the controls. Whole lysates of spheroids were subjected to immunoblotting for 30 kDa milk protein P-casein (c; P-Cas). Thus, these cells were able to undergo functional differentiation by inducing P-casein. The lower bands on the blot resulted from protein degradation by the proteases (Dispase) used to digest the Matrigel prior to lysing the cells. (Bar= 100pm) 132 133 Figure 3.3 Expression of EN induced IGF-l/Insulin dependent cell proliferation in 3D culture. (A) Cells were serum and insulin starved for 24 hrs and were allowed to form cell clusters on polyHEMA-coated dishes. These clusters were collected and plated with 1% E C M containing media supplemented with insulin (5 pg/ml) or IGF-1 (200 ng/ml) and were allowed to undergo spheroid formation for 71 hrs. The cell proliferation in these spheroids was detected by a one hour BrdU incorporation assay. Cells were fixed and immunostained. Both E N expressing and control cells had no proliferating cells in the absence of insulin or IGF-1. In the presence of growth factors insulin or IGF-1 only the E N expressing cells were able to induce cell proliferation. (Bar =100 um) (B) ~ 35% of the E N spheroids underwent cell proliferation in the presence of insulin or IGF-1, while less than 2.5% of the controls had proliferating cells. In the presence of IGF-1 there was not a single spheroid that had BrdU positive cells. For statistical analysis, at 400X magnification, all the spheroids with two or more BrdU positive cells in 10 fields were counted and averaged. The data shown here is a representation of three trials. 134 A 135 Figure 3.4 Cyclin Dl/2 levels remained high in EN expressing spheroids. Cells were serum and insulin starved for 24 hrs and were allowed to form cell clusters on polyHEMA-coated dishes. These clusters were collected and plated with 1% E C M containing media supplemented with 5 pg/ml insulin or 200 ng/ml IGF-1. At 2, 5 and 48 hr time points total cellular protein lysates were prepared and 30 pg of total protein of each sample was subjected to Western blotting. In the control cells, cyclin D l expression levels were down regulated during spheroid formation while in E N expressing spheroids cyclin D l levels remained high. Total Grb2 levels were used as a loading control. Originally the samples were run in a different order, therefore, the blot had to be cut to assemble the figure. 136 EpH4-C EpH4-EN Time (hrs) 2 5 48 2 5 48 Cyclin DI § M m*>-35 kD* Grb-2 23kD« Insulin Cyclin D I mm M 35kD» IGF-1 Grb-2 28 kD* F i g u r e 3.5 P I 3 - K is r e q u i r e d for E N i n d u c e d cell pro l i f era t ion . EpH4-EN cells were serum and insulin starved for 24 hrs and were allowed to form cell clusters on polyHEMA-coated dishes. These clusters were collected and plated with 1% E C M containing media supplemented with insulin or IGF-1. They were treated with 25 pg/ml PI3-K inhibitor LY294002 (I have previously demonstrated effectiveness of 25 pm/ml LY294002 on mammary epithelial cells; Somasiri et al., 2000) or with DMSO carrier for 71 hrs and labeled with BrdU for one hour. Treatment with LY294002 completely inhibited the E N induced cell proliferation in these spheroids (A). PI3-K inhibition also completely inhibited the cyclin DI expression in these spheroids (B). Total Grb2 levels were used as a loading control. (Bar = 100pm) 138 B EpH4-C EpH4-EN Insulin IGF- 1 LY294002 D M S O C y c l i i i D l G i b 2 + + + + 35 kDa 28 kDa 139 Figure 3.6 E N expressing cells form multi-layered spheroids without disrupting the spheroid polarity. Cells were cultured on reconstituted basement membrane gels for 72 hours and were fixed and immunostained for adherens junction protein E-cadherin and tight junction protein ZO-1. Images represent a 0.8 pm slices taken through the center of the spheroids. Both EpH4-C (a) and EpH4-EN (b) spheroids were able to localize E-cadherin at cells-cell interaction sites throughout the spheroid. ZO-1 was localized to the most apical domains caging the central lumen (c and d). E N expression did not disrupt the normal polarity of the spheroid. Arrows indicate the areas where EpH4-EN cells are forming multi-layered lumens. (Bar = 20pm). These E N expressing spheroids are similar in phenotype to that of SBC (e). The SBC tumor section was stained with haematoxylin and eosin for histological observation (e; this image was obtained from C. Tognon, Department of Pathology & Laboratory Medicine, Children's and Women's Health Center of British Columbia). The nuclei stained in blue show the multi-layering. 140 141 3.6 R e f e r e n c e s Bartkova J, Lukas J and Bartek J. (1997) Aberrations of the G l - and Gl/S-regulating genes in human cancer. Prog Cell Cycle Res. 3, 211-20. Berx G, Becker K F , Hofler H and van Roy F. (1998) Mutations of the human E-cadherin (CDH1) gene. Hum Mutat. 12(4), 226-37. Bissell, M.J. , and D. Radisky. 2001. Putting tumours in context. Nat. Rev. Cancer. 1:46-54. Brennan P, Babbage JW, Burgering B M , Groner B, Reif K and Cantrell DA. (1997) Phosphatidylinositol 3-kinase couples the interleukin-2 receptor to the cell cycle regulator E2F. Immunity. 7(5), 679-89. de Bree E, Askoxylakis J, Giannikaki E, Chroniaris N , Sanidas E and Tsiftsis DD. (2002) Secretory carcinoma of the male breast. Ann Surg Oncol. 9(7), 663-7. Coolican SA, Samuel DS, Ewton DZ, McWade FJ and Florini JR. (1997) The mitogenic and myogenic actions of insulin-like growth factors utilize distinct signaling pathways. J Biol Chem. 272(10), 6653-62. Debnath J, Walker SJ, Brugge JS. (2003) Akt activation disrupts mammary acinar architecture and enhances proliferation in an mTOR-dependent manner. J Cell Biol. 163(2):315-26. Debnath, J., K . Mills, N . Collins, M . Reginato, S. Muthuswamy, and J. Brugge. 2002. The role of apoptosis in creating and maintaining luminal space within normal and oncogene-expressing mammary acini. Cell. 111:29-40. 142 Eguchi M , Eguchi-Ishimae M , Tojo A , Morishita K, Suzuki K, Sato Y , Kudoh S, Tanaka K , Setoyama M , Nagamura F, Asano S and Kamada N . (1999). Fusion of ETV6 to neurotrophin-3 receptor T R K C in acute myeloid leukemia with t(12;15)(pl3;q25). Blood, 93(4), 1355-63. Ellerbroek SM, Halbleib JM, Benavidez M , Warmka JK, Wattenberg EV, Stack MS and Hudson L G . (2001). Phosphatidylinositol 3-kinase activity in epidermal growth factor-stimulated matrix metalloproteinase-9 production and cell surface association. Cancer Res. 61(5), 1855-61. Farrelly N , Lee Y J , Oliver J, Dive C, Streuli CH. (1999) Extracellular matrix regulates apoptosis in mammary epithelium through a control on insulin signaling. J Cell Biol 144: 1337-1347. Fantl V , Stamp G, Andrews A, Rosewell I and Dickson C. (1995) Mice lacking cyclin D l are small and show defects in eye and mammary gland development. Genes Dev. 9(19), 2364-72. Gao N , Zhang Z, Jiang B H and Shi X . (2003) Role of PI3K/AKT/mTOR signaling in the cell cycle progression of human prostate cancer. Biochem Biophys Res Commun. 310(4), 1124-32. Hirokawa M , Sugihara K , Sai T, Monobe Y , Kudo H , Sano N and Sano T. (2002). Secretory carcinoma of the breast: a tumour analogous to salivary gland acinic cell carcinoma? Histopathology, 40(3), 223-9. Jung F, Haendeler J, Goebel C, Zeiher A M , and Dimmeler S. (2000) Growth factor-induced phosphoinositide 3-OH kinase/Akt phosphorylation in smooth muscle cells: induction of cell proliferation and inhibition of cell death. Cardiovasc Res. 48(1), 148-57. 143 Knezevich SR, McFadden DE, Tao W, Lim JF and Sorensen PH. (1998). A novel ETV6-NTRK3 gene fusion in congenital fibrosarcoma. Nat Genet., 18(2), 184-7. Lee Y J , Streuli CH. (1999) Extracellular matrix selectively modulates the response of mammary epithelial cells to different soluble signaling ligands. J Biol Chem 274: 22401-22408. LeRoith D, Roberts CT Jr. The insulin-like growth factor system and cancer.Cancer Lett. 2003 Jun 10; 195(2): 127-37. Marshman E, Streuli CH. (2002) Insulin-like growth factors and insulin-like growth factor binding proteins in mammary gland function. Breast Cancer Res. 2002;4(6):231-9. Morrison K B , Tognon CE, Garnett MJ , Deal C and Sorensen PH. (2002). ETV6-NTRK3 transformation requires insulin-like growth factor 1 receptor signaling and is associated with constitutive IRS-1 tyrosine phosphorylation. Oncogene, 21(37), 5684-95. Muise-Helmericks RC, Grimes HL, Bellacosa A , Malstrom SE, Tsichlis P N and Rosen N . (1998) Cyclin D expression is controlled post-transcriptionally via a phosphatidylinositol 3-kinase/Akt-dependent pathway. J Biol Chem. 273(45), 29864-72. Muthuswamy SK, L i D, Lelievre S, Bissell MJ , and Brugge JS. (2001) ErbB2, but not ErbBl , reinitiates proliferation and induces luminal repopulation in epithelial acini. Nat Cell Biol., 3(9),785-92. Oberman HA. (1980) Secretory carcinoma of the breast in adults. A m J Surg Pathol. 4(5):465-70. Paeng M H , Choi H Y , Sung SH, Moon BI and Shim SS. (2003). Secretory carcinoma of the breast.'J Clin Ultrasound., 31(8), 425-9. 144 Perl A K , Wilgenbus P, Dahl U , Semb H and Christofori G. (1998) A causal role for E-cadherin in the transition from adenoma to carcinoma. Nature. 392(6672), 190-3. Petersen OW, Ronnov-Jessen L, Howlett A R and Bissell MJ . (1992) Interaction with basement membrane serves to rapidly distinguish growth and differentiation pattern of normal and malignant human breast epithelial cells. Proc Natl Acad Sci U S A . 89(19), 9064-8. Qiang Y W , Yao L, Tosato G, Rudikoff S. (2003) Insulin-like growth factor I induces migration and invasion of human multiple myeloma cells. Blood. 103(1), 301-308. Richard G, Hawk JC 3rd, Baker AS Jr and Austin R M . (1990). Multicentric adult secretory breast carcinoma: D N A flow cytometric findings, prognostic features, and review of the world literature. J Surg Oncol., 44(4), 238-44. Rosen PP and Cranor M L . (1991) Secretory carcinoma of the breast. Arch Pathol Lab Med. 115(2):141-4. Roskelley CD, Wu C and Somasiri A M . (2000) Analysis of mammary gland morphogenesis. Methods Mol Biol.,136, 27-38. Roskelley CD, Desprez PY and Bissell M J . (1994) Extracellular matrix-dependent tissue-specific gene expression in mammary epithelial cells requires both physical and biochemical signal transduction. Proc Natl Acad Sci U S A . 91(26): 12378-82. Sherr CJ and Roberts J M . (1999) C D K inhibitors: positive and negative regulators of G l -phase progression. Genes Dev. 13(12), 1501-12. Sherr CJ. (1996). Cancer cell cycles. Science. 274(5293), 1672-7. 145 Siegel JR, Karcnik TJ, Hertz M B , Gelmann H and Baker SR. (1999) Secretory Carcinoma of the Breast. Breast J. 5(3):204-207. Somasiri A , Wu C, Ellchuk T, Turley S and Roskelley CD. (2000) Phosphatidylinositol 3-kinase is required for adherens junction-dependent mammary epithelial cell spheroid formation. Differentiation. 66(2-3): 116-25. Suzuki F, Saito A , Ishi K, Okazaki T, Kina K , Koyatsu J and Sugiyama K. (1999) Secretory carcinoma of the breast: an immunohistochemical and ultrastructural study. Med Electron Microsc. 32(l):50-56. Somasiri A , Howarth A , Goswami D, Dedhar S and Roskelley CD. (2001) Overexpression of the integrin-linked kinase mesenchymally transforms mammary epithelial cells. J Cell Sci., 114(Pt 6),1125-36. Sundgren NC, Giraud GD, Schultz JM, Lasarev MR, Stork PJ and Thornburg K L . (2003) Extracellular signal-regulated kinase and phosphoinositol-3 kinase mediate IGF-1 induced proliferation of fetal sheep cardiomyocytes. A m J Physiol Regul Integr Comp Physiol., 285(6), R1481-9. Tognon C, Garnett M , Kenward E, Kay R, Morrison K , and Sorensen PH. (2001). The chimeric protein tyrosine kinase ETV6-NTRK3 requires both Ras-Erkl/2 and PI3-kinase-Akt signaling for fibroblast transformation. Cancer Res., 61(24), 8909-16. Tognon C, Knezevich SR, Huntsman D, Roskelley CD, Melnyk N , Mathers JA, Becker L, Carneiro F, MacPherson N , Horsman D, Poremba C and Sorensen PH. (2002). Expression of the ETV6-NTRK3 gene fusion as a primary event in human secretory breast carcinoma. Cancer Cell., 2(5), 367-76. Van Obberghen E, Baron V , Delahaye L, Emanuelli B, Filippa N , Giorgetti-Peraldi S, Lebrun P, Mothe-Satney I, Peraldi P, Rocchi S, Sawka-Verhelle D, Tartare-Deckert S 146 and Giudicelli J. (2001) Surfing the insulin signaling web. Eur J Clin Invest., 31(11), 966-77. Wai DH, Knezevich SR, Lucas T, Jansen B, Kay RJ and Sorensen PH. (2000). The ETV6-NTRK3 gene fusion encodes a chimeric protein tyrosine kinase that transforms NIH3T3 cells. Oncogene, 19(7), 906-15. Weaver V M , Petersen OW, Wang F, Larabell CA, Briand P, Damsky C and Bissell M J . (1997) Reversion of the malignant phenotype of human breast cells in three-dimensional culture and in vivo by integrin blocking antibodies. J Cell Biol. 137(1): 231-45. Zeng H, Datta K , Neid M , L i J, Parangi S and Mukhopadhyay D. (2003). Requirement of different signaling pathways mediated by insulin-like growth factor-I receptor for proliferation, invasion, and VPF/VEGF expression in a pancreatic carcinoma cell line. Biochem Biophys Res Commun. 302(1), 46-55. Zhang D and Brodt P. (2003). Type 1 insulin-like growth factor regulates MT1-MMP synthesis and tumor invasion via PI 3-kinase/Akt signaling. Oncogene. 22(7), 974-82. 147 CHAPTER 4: The Anti-Adhesion Protein Podocalyxin Disrupts Breast Carcinoma Cell Junctions and its Overexpression Independently Predicts Breast Cancer Progression1 4.1 Summary Podocalyxin is a CD34-related cell surface molecule that functions as an anti-adhesion molecule in normal development. In the developing kidney, expression of podocalyxin on podocytes leads to a breakdown of cell junctions such that the semipermeable urinary filtration barrier can form. Thus, I hypothesized that abnormal Podocalyxin expression might be involved in the generalized perturbation of cell-junction complexes during breast carcinoma progression. Using a tissue microarray of 272 invasive human breast tumors tied to long-term outcome we found that podocalyxin was highly overexpressed in a distinct subset of invasive breast carcinomas. Strikingly, Kaplan-Meier survival and Cox regression analyses indicated that this overexpression was a more significant indicator of poor outcome than either regional lymph node involvement or HER2 overexpression amongst these patients. Furthermore, this subset of patients had a mean 6-year shorter life span compared to rest of the patients. When Podocalyxin was overexpressed in non-metastatic MCF-7 breast carcinoma cells, the transepithelial resistance decreased more than 50%, indicating a breakdown in the TJ integrity. Cell junctions were perturbed and cell shedding was induced from confluent monolayers. Furthermore, polarized spheroid formation in basement membrane gel culture was disrupted giving a phenotype similar to IDC. Therefore, Podocalyxin ' A version of this chapter has been accepted for publication. Somasiri A M , Nielsen J, Makretsov M , Gilks CB, Huntsman D, Kershaw DB, McNagny K M and Roskelley CD. The anti-adhesion podocalyxin disrupts breast carcinoma cell junctions and its overexpression independently predicts breast cancer progression. (In Press, Cancer Research) 148 overexpression is a novel, independent predictor of breast cancer metastasis and it may actively contribute to this process by initiating the dispersion of cells from the primary tumor. 4.2 Introduction Cell adhesion is often dysregulated during metastatic breast cancer progression. In lobular breast tumors, the suppression of E-cadherin is a very common event that allows single cells to break away from the primary tumor and infiltrate the surrounding stroma (Berx and van Roy, 2001). However, in ductal breast carcinoma, which is by far the most prevalent form of the disease, the situation is less straightforward; E-cadherin may or may not be present, apical-basal polarity is often disrupted, cell junctions are generally perturbed and infiltration is characterized by the movement of disorganized clusters of cells into the stroma (Cleton-Jansen et al., 2002). While the mechanisms responsible for E-cadherin suppression and apical-basal disruption are now being elucidated, the means by which cell junctions are generally perturbed in the majority of breast tumors remain largely unknown. Podocalyxin is a membrane-associated mucin that was originally identified as the major sialoprotein on the apical surface of kidney glomerular podocytes (Kerjaschki et al., 1984). This heavily sialyated and sulfated integral membrane glycoprotein belongs to the CD34 family of sialomucin proteins. In addition, Podocalyxin is highly expressed in vascular endothelial cells and high endothelial venules (Kerjaschki et al., 1984; Sassetti et al., 1998). More recently, it was found that Podocalyxin is also expressed on the surface of hematopoietic stem cells, thrombocytes and megakeryocytes, where it is known as Myb-Ets-transformed progenitor (MEP-21) or thrombomucin (McNagny et al., 1997). Podocalyxin is a 150-165 kDa transmembrane protein that consists of a mucin domain, globular domain, transmembrane domain and a cytoplasmic tail. The extracellular domain is highly negatively charged and contains numerous potential O-linked glycosylation sites. Removal of sialic acid from transfected Podocalyxin by sialidase treatment abrogates the anit-adhesion effect showing the role of the extracellular domain. In functional studies, puromycin aminonucleoside-treatment led to reduction in sialic acid 149 content of podocalyxin and alteration of glomerular slits (Kerjaschki et al., 1985; Kurihara et al., 1992). The extracellular domain shows very little sequence similarity among species except the similar mucin-like structural features. However, the cytoplasmic and the transmembrane domain sequences are highly conserved among species. The alternatively spliced Podocalyxin is an intracellular domain truncated isoform that is expressed in a tissue specific pattern in parallel with the full-length form (Li et al., 2001). The cytoplasmic domain of podocalyxin interacts with the actin cytoskeleton through ezrin, a member of ezrin-radixin-moesin (ERJVI) family of actin-binding proteins. Furthermore, the cytoplasmic domain contains a DTHL (Asp-Thr-His-Leu) amino acid sequence that can interact with PDZ domain containing proteins. Na(+)/H(+) exchange regulatory factor-2 (NHERP-2) is one such molecule that interact with the cytoplasmic domain of podocalyxin and maintain it at the apical membranes of podocytes (Li et al., 2002). The fully developed kidney maintains its filtration properties by the slit diaphragms that form between the foot processes of the podocytes. As podocytes develop, podocalyxin is first expressed just prior to the dramatic morphological shift from a 'closed' epithelial barrier with complete TJ complexes to a selectively 'open' filter consisting of cell bodies and interdigitated foot processes that are connected to each other by modified A J that ultimately become the slit diaphragms. Initially, podocalyxin is expressed in the apical surfaces of the podocytes and it then migrates laterally displacing the junction complexes to form the slit diaphragms (Schnabel et al., 1989; Reiser et al., 2000). The high negative charge on the extracellular domain of Podocalyxin is believed to be responsible for maintaining the open slits. When the surface charge of the foot processors are neutralized, the slits collapse and foot processors under go reorganization to localize TJ protein ZO-1 to the apical regions (Seiler et al., 1977; Kurihara et al., 1992). Furthermore, deletion of the Podocalyxin gene in mice results in the persistence of barrier-forming tight junctions between the developing podocytes, a lack of foot process formation, and perinatal death associated with anuria (Doyonnas et al., 2001). Conversely, when Podocalyxin is ectopically expressed in cultured kidney epithelial (MDCK) cells both adherens AJs and TJs are subtly perturbed (Tekda et al., 2000). Thus, 150 loss-of-function and gain-of-function experiments both suggest that Podocalyxin acts as an anti-adhesin during normal kidney development. Due to this anti-adhesive property of this molecule I reasoned that inappropriate Podocalyxin expression might perturb junctional complexes in breast tumors. Using a tissue microrray (TMA) I assessed Podocalyxin expression and localization in a previously characterized series of 272 invasive human breast carcinomas. Furthermore, I tested the consequences of forced Podocalyxin overexpression in the non-aggressive MCF-7 human breast carcinoma cell line. In this study I chose to use the MCF-7 line instead of the previously used scp2 (Chapter 2) and EpH4 (Chapter 3) normal mammary epithelial lines due to the difficulties resulted from inefficient transfections. MCF-7 monolayers form A J and TJ at the basolateral membrane. In 3D culture, TJs are polarized and apically localized. This MCF-7 line was easily transfected and a stable pool of Podocalyxin expressing cells were obtained to analyze the effects of Podocalyxin in cell polarity. The results from this study indicate that podocalyxin overexpression is tightly correlated with poor outcome in a distinct subset of tumors and that ectopic expression of the molecule causes a general disruption of cell junctions without suppressing E-cadherin. In addition, ectopic Podocalyxin expression perturbed spheroidal morphogenesis and apical/basal polarization in B M gel culture. Hence, Podocalyxin overexpression is very likely to be both an independent marker of, and a functional contributor for invasive breast cancer progression. 4.3 Materials and Methods 4.3.1 TMA Construction A total of 272 formalin-fixed, paraffin-embedded primary invasive breast cancer tissue blocks (outcome-linked archival cases from 1974-1995) graded according to the Nottingham method (Elston and Ellis, 1991) were used to construct a tissue microarray (TMA) as described previously (Parker et al., 2002; This tissue array was constructed by the Genetic Pathology Unit at Jack Bell Research center, Vancouver B.C., Canada) with 151 institutional review board approval (Vancouver General Hospital, Vancouver B.C., Canada). 4.3.2 TMA Immunohistochemistry, Scoring and Correlation Analysis Paraffin embedded T M A , normal breast, and positive control normal kidney sections were deparaffinized by treating with xyaline (Fisher Scientific) 3 times, 5 min each. Then they were rehydrated by placing in 100%, 95%, 70% EtOH and finally in water. Rehydrated sections were treated with citrate buffer (pH 6.00) at 100°C for 20 min to allow antigen retrieval. Then they were blocked with 3% hydrogen peroxide for 30 min, and blocked with 4% B S A in PBS for 30min. Mouse monoclonal anti-human Podocalyxin antibody 3D3 (1:80 dilution; Kershaw et al., 1997a) added to the sections and incubated overnight at 4°C. Sections were rinsed 3 times with PBS and antibody binding was detected using the Envision detection, system (Dako, Carpinteria, CA) followed by hematoxylin counterstaining, dehydration and mounting. Podocalyxin expression was determined by staining intensity and the proportion of cells stained as described in the results without knowledge of patient outcome. Mouse IgG was used instead of the primary antibody as a control. A l l scores were processed using the T M A -Deconvoluter 1.06, Cluster, and TreeView programs as previously described (Liu et al., 2002). Paired correlation analysis to nodal status, grade, size, p53 mutation, ER positivity, C A I X positivity, and HER2 overexpression was performed using the bivariate two-tailed Pearson test. Multivariate analysis was performed using the Cox proportional hazard regression model and survival analysis was performed using the Kaplan-Meier method as previously described (Makretsov et al., 2003). 4.3.3 Cell Culture, Transfection and Immunostaining T47D, MCF-7 and MDA-231 human breast carcinoma cell lines were routinely maintained in DMEM/F12 medium supplemented with 5% FBS and insulin (5 pg/ml). Endogenous podocalyxin levels were determined by Western blotting of whole cell lysates with the same 3D3 antibody used on the T M A . MCF-7 is a well- differentiated 152 breast carcinom line, which was obtained originally by pleural effusin from a breast cancer patient (Brooke et al., 1973). They form epithelial monolayers when cultured as flat monolayers and in 3D culture, they form solid spheroids without lumen formation and TJ proteins are apically localized (Wang et al., 2002). These MCF-7 spheroids are phenotypically similar to DCIS and can be used as a model to examine the progression from in situ stage to invasive stage during cancer progression. MCF-7 cells, which expressed low levels of endogenous human podocalyxin (see Fig 4.1 A), were transfected with a control pIRES-EGFP expression vector (BD Biosciences, Mississauga Canada) Or with the same vector containing a full length mouse Podocalyxin cDNA (Doyonnas et al., 2001) using DMRIE-C reagent (Invitrogen, Carlsbad CA). Pooled, stable transfectant populations were generated by continuous selection under G418 (400 pg/ml). Monolayer culture on glass coverslips and 3D-BM membrane gel culture were carried out essentially as previously described (Chapter 2, page 82; Somasiri et al., 2001). Enhanced green fluoresent protein (EGFP) and coincident mouse Podocalyxin transgene expression was confirmed by dual fluorescence excitation for EGFP and immunofluorescence for ectopic mouse Podocalyxin using a rabbit polyclonal antibody specific for mouse Podocalyxin (PCLP-1; Doyonnas et al., 2001). For the podocalyxin staining, live cells were incubated with 1:100 PCLP-1 antibody for 1 hour at room temprature and rinsed with media 3 times before fixing with MeOH at -20C°. Then the cells were blocked with 10% NGS and immunostaining performed as previously described. The precise localization of the ectopically expressed mouse Podocalyxin in monolayers was determined by confocal microscopy in conjunction with dual staining for the A J proteins E-cadherin or P-catenin (mouse monoclonals, Pharmingen, San Diego, CA) or for the TJ proteins occludin (mouse monoclonal, Pharmingen) or ZO-1 (rat polyclonal, Zymed, San Francisco C A , respectively). Dual staining for occludin and ZO-1 (rabbit polyclonal, Zymed) was carried out to specifically focus on changes in TJ localization in 3-D culture. Since the cells were fixed in MeOH, which destroys the GFP fluorescence I was able to dual stain the cells using FITC tagged secondary antibody. Steady-state occludin and E-cadherin levels were determined by Western blotting (see chapter 2) of whole cell lysates using the same antibodies. 153 4.3.4 Transepithelial Resistance Transepithelial resistance (TER) was assessed on confluent monolayers maintained on Transwell filters with a 3 pm pore size (Costar) using a Millicell . ERS electrical resistance system (Millipore, Bedford, M A ) according to manufacturer's instructions. Each transfectant was assessed in duplicate cultures; variation was less than 10% of the mean and the data presented are representative of two independent experiments. 4.4 Results 4.4.1 Podocalyxin is Highly Expressed in a Subset of Invasive Breast Carcinomas As a prelude to testing whether Podocalyxin is augmented in aggressive breast tumors I first analyzed its expression in normal breast tissue by immunoperoxidase staining using normal human kidney tissue as a positive control (Kershaw et al., 1997a). As expected, Podocalyxin was highly expressed on glomerular podocytes in the kidney while expression was low to negative on cells of the proximal and distal tubules (Fig 4.1 A). This confirmed the specificity of immunohistochemistry under the conditions used. Although Podocalyxin was also present in normal breast epithelium, its expression was limited and spatially restricted. Specifically, Podocalyxin was localized to the apical region of luminal epithelial cells (Fig 4.IB; arrowheads). In addition, Podocalyxin was present on vascular endothelial cells as has been described previously (Fig 4.1 A , 4.1C; arrows, Kershaw et al., 1995, 1997a). I next examined endogenous Podocalyxin expression in 272 invasive breast carcinoma cases using a previously characterized T M A that featured a wide range of tumors in terms of stage, grade and nodal status (Table 4.1; Makretsov et al., 2003). Sixty percent (163/272) of the cases exhibited no Podocalyxin staining in the tumor cells and were assigned a score of '0' (Fig 4.1C); 23% (62/272) exhibited staining in less than 10% of the tumor cells and were assigned a score of'1' (Fig 4.ID); 12%) (32/272) exhibited diffuse staining in more than 10% of the cells and/or 154 intense staining in less than 50% of the cells and were assigned a score of'2' (Fig 4.IE). Kaplan-Meier analysis suggested a slight decrease in longterm cumulative disease-free survival in group 2 which suggested that a moderate upregulation of Podocalyxin may be associated with poor outcome (Fig 4.2A). However, it is important to point out that there was no statistical difference in cumulative survival between groups 0, 1 and 2. The remaining 5.51% (15/272) of the cases on the T M A exhibited intense Podocalyxin staining in the majority of the tumor cells and they were originally assigned a score of'3'. These cases often exhibited a significant architectural disruption and loss of glandular morphology (Fig 4.IF). Clearly, these group 3 tumors had a significantly poorer disease-free survival than any of the other individual Podocalyxin groups (Fig 4.2A). This difference was readily observable, and statistically significant, when the 0, 1, and 2 groups were pooled (ie. low or no Podocalyxin expression) and compared to group 3 alone (ie. high Podocalyxin), both in terms of disease-free (Fig 4.2B; p<0.01) and overall (p=0.025, see supplementary data) survival. As a result, amongst those patients that had succumbed at the time of analysis, the high Podocalyxin group had a much lower mean survival time of 9.0 +/- 1.8 years compared to the 15 +/-0.5 years for the combined low or no Podocalyxin pool and the 14.9 years for the entire T M A population (see Table 4.1). I conclude that high Podocalyxin expression occurs in a subset of primary invasive breast tumors that have already become, or will become, highly metastatic. 4.4.2 High Podocalyxin Expression is an Independent Marker of Poor Outcome The T M A used here has already been characterized by assessing a number of other markers with prognostic significance (Makretsov et al., 2003). This offered me the opportunity to perform a multi-variant Cox regression analysis using these markers and high Podocalyxin expression as parameters (Table 4.2). As expected, regional lymph node involvement was an independent indicator of poor outcome (p=0.012), which confirms the validity of the analysis of this T M A . Strikingly, high Podocalyxin expression, on its own, was also a highly significant independent predictor of poor outcome (p=0.0005). What's more, the increased relative risk associated with this high 155 Podocalyxin expression (8.4 fold), while broad in terms of confidence interval, was greater than either regional lymph node involvement (3.7 fold) or HER2 overexpression (1.9 fold). Thus, this analysis indicates that Podocalyxin overexpression identifies a unique and novel subpopulation of metastatic breast tumors. The uniqueness of the metastatic high Podocalyxin cohort was further illustrated by a Pearson marker correlation analysis (Table 4.3), which indicated that there was no correlation with HER2 overexpression or nodal involvement. The latter finding suggests that Podocalyxin upregulation may occur prior to lymphagenic spread or, alternatively, that it is a marker of non-lymphagenic hematogenous metastasis. There were statistically significant positive correlations between high podocalyxin and p53 immunoreactivity, focal upregulation of the hypoxia marker carbonic anhydrase IX (CAIX; Chia et al., 2001) and increased tumor grade which was not unexpected given the high degree of histoarchitectural disruption in the Podocalyxin overexpressing tumors. Finally, there was a significant negative correlation between high Podocalyxin and the presence of estrogen receptors in the tumor. 4.4.3 Ectopic Podocalyxin Overexpression Initiates MCF-7 Breast Carcinoma Cell Delamination in Monolayer Culture Ectopic expression of Podocalyxin in polarized kidney epithelial cells disrupts cell junctions (Takeda et al., 2000). To determine i f the same occurs in breast carcinoma cells I first examined endogenous Podocalyxin levels in human breast tumor lines. T47-D and MCF-7 breast carcinoma cells, which are well-behaved, form cell junctions and are estrogen receptor positive, expressed little or no Podocalyxin. In sharp, contrast M D A -231 cells, which are highly metastatic, cell junction negative and estrogen receptor negative, expressed podocalyxin at high levels (Fig 4.3A). MCF-7 cells form cohesive monolayers in 2-dimensional culture on tissue culture plastic and distinct spheroids in 3D B M gel culture (Wang et al., 2002). Thus, they were chosen for ectopic overexpression studies. MCF-7 cells stably transfected with a control EGFP-expressing vector formed flat epithelial cobblestone monolayers that were indistinguishable from those observed in the parent line (Fig 4.3B). In contrast, cells stably transfected with the same vector 156 encoding EGFP and a full-length mouse Podocalyxin cDNA formed monolayers that contained areas of cells that bulged apically. When these Podocalyxin-transfected cultures reached confluence the bulging cells delaminated from the monolayer and were shed into the medium (Fig 4.3B). Based on triple-labelling for the EGFP marker, ectopic mouse Podocalyxin, and nuclear DNA, it was clear that the ectopically expressed Podocalyxin was appropriately targeted to the cell surface and that it was these cells that were being extruded from the monolayer (Fig 4.3B lower panel). 4.4.4 Ectopic Podocalyxin Expression Disrupts Breast Carcinoma Cell Junctions Attempts to sub-clone homogeneous populations of high Podocalyxin expressors from the pooled MCF-7 transfectants have so far failed, even after drug selection and fluorescent cell sorting for the co-expressed GFP marker, possibly because these cells are constantly being released into the medium. I therefore examined the effects of ectopic Podocalyxin expression on cell junctions in heterogeneous primary transfectants by dual immunostaining. Unlike the normal mammary epithelial cell lines scp2 and EpH4, M C F -7 cell line is capable of forming both A J and TJ in monolayers (Macek et al., 2003). In those cells that expressed low-to-negligible ectopic mouse Podocalyxin, which served as internal controls, the transmembrane AJ protein E-cadherin and the TJ protein occludin were localized to the basolateral and apical cell-cell contact domains respectively (Fig 4.4A, 4.4B, 4.6A left side of the panels). These data indicate that adherens and TJ complexes formed and localized appropriately in these control cells. In contrast, E-cadherin, P-catenin and occludin staining indicated widely distributed along the apical and lateral surfaces of cells that expressed significant amounts of ectopic mouse Podocalyxin (Fig 4.4A, 4.4B, 4.6A right side of the panels) Since I had a heterogenous population of cells I was able to image control cells and Podocalyxin expressing cells side-by-side. To detect the cells that were expressing Podocalyxin, cells had to be labeled with rat-PCLP-1 antibody prior to cell fixation. However, the control staining indicated that the anti-mouse secondary antibodies that had to be used to detect E-cadherin, P-catenin and occludin cross reacted with the rat-PCLP-1 antibody that was bound to 157 Podocalyxin (Fig 4.4C). Hence, non-specific surface staining and there may be no increase in junction protein expression observed with E-cadherin, P-catenin and occludin. This was further confirmed by the Western blot data, which showed no significant change in the steady state levels of either transmembrane junctional protein (Fig 4.5). Interestingly, although the TJ-associated protein ZO-1 remained distinctly localized it was found at the basal attachment sites of Podocalyxin-expressing cells that were being extruded from the monolayer (Fig 4.6B, right side of the panels). Taken together, these data suggest that Podocalyxin overexpression has a subtle effect on AJs and disrupts TJs in MCF-7 cells, which allows cells to extrude off the monolayers. This finding is further supported by transepithelial resistance (TER), which is a functional measure of TJ integrity, was significantly reduced in the EGFP/Podocalyxin-transfected cultures (210 2 2 ohms cm ) compared to the control-transfected cells (497 ohms cm ). 4.4.5 Ectopic Podocalyxin Expression Perturbs Basement Membrane-Dependent Polarization and Spheroidal Morphogenesis The disruption of polarized morphogenesis is a histopathologic hallmark of increased breast tumor grade (Elston and Ellis, 1991). This was also a characteristic of high Podocalyxin overexpressing tumors on the T M A (see Table 3 above). Thus, we next determined i f ectopic Podocalyxin was able to initiate such a disruption in 3D B M membrane gel culture. Vector control-transfected MCF-7 cells formed spheroids in which E-cadherin and P-catenin were membrane localized throughout (Fig 4.7). The same was true of Podocalyxin-transfected cells, however the ectoptic Podocalyxin protein was localized at the outer edge of the spheroids (Fig 4.7B). The latter suggested to us that, while Podocalyxin spheroids were capable of forming adherens junctions, they did not polarize properly as the outer edge of MCF-7 spheroids is normally the basal surface (Wang et al., 2002). This perturbation of apical/basal polarization was confirmed by an examination of TJ localization. Specifically, instead of being centrally (ie. apically) localized, as was the case in control spheroids, both ZO-1 and occludin were relocalized to the outer (ie. basal) surface of Podocalyxin spheroids (Fig 4.8A). In addition, Podocalyxin spheroids had ragged edges with a 'cluster of grapes morphology' which 158 further indicated a disruption of polarized morphogenesis that normally generates spheroids with smooth, well-defined edges like those formed by the vector-transfected control cells (Fig 4.8B). Therefore, experimental ectopic Podocalyxin overexpression induced a morphogenic perturbation that is consistent with the increased histopathological grade observed in tumors with endogenous Podocalyxin overexpression (see Fig 4.1 above). 4.5 D i s c u s s i o n Tissue microarrays afford investigators the opportunity to carry out a rapid and relatively thorough screening of molecules that may be important in specific tissues or pathologies (Kononen et al., 1998). The power of this technology, particularly when it is linked to longterm outcome data, is exemplified here. Specifically, while only 15 of the 272 cases on the T M A used here had consistently high Podocalyxin overexpression, this group was clearly different with respect to marker linkage, disease-free survival and overall survival. Therefore, Podocalyxin overexpression can be a novel independent indicator of metastatic progression in a distinct subset of invasive breast cancers. Locally invasive breast cancers can have markedly different treatment responses and outcomes. Thus, it is extremely difficult to predict which patients will most benefit, or not benefit, from adjuvant therapy (Eifel et al., 2001). Genome-wide searches and large scale expression profiling followed by cluster analysis have had some impact on this problem, particularly in terms of identifying those tumors that do not progress (van't Veer et al., 2002; Sorlie et al., 2003). Despite these advances, the identification of novel independent indicators of poor outcome continues to be extremely useful because it facilitates the development of new classification parameters that increase the resolving power of all prognostic strategies. This has proven to be the case with HER2/neu where only a consistently high level of overexpression clearly correlates with poor outcome (Pauletti et al., 2000), which is similar to our finding with Podocalyxin. Given its demonstrated role in experimental breast tumorgenesis and its cell surface localization, HER2/neu has proven to be a clinically relevant therapeutic target (Nabholtz and Slamon, 2001). As a member of the CD34 family of sialomucins, 159 Podocalyxin is also a cell surface molecule. The function of this family of molecules, which includes CD34, Podocalyxin and Endoglycan has, until recently, been controversial. The most clear-cut experiments suggest that CD34-type proteins can act as either pro-adhesive or anti-adhesive molecules depending on their site of expression and their glycosylation status (Baumhueter et al., 1993; Sassetti et al., 1998; Bistrup, et al 1999). Thus, CD34 and Podocalyxin molecules that are located on specialized high endothelial venules in lymphoid organs are decorated with the appropriate glycosylations to make them adhesive ligands for L-selectin on circulating lymphocytes. This posttranslational modification is tissue-specific and the vast majority of endothelial cells and CD34 family proteins lack it (reviewed in Krause et al., 1996). While Podocalyxin's role in lymphogenous breast cancer metastasis is not yet clear, the negative correlation between Podocalyxin overexpression and regional lymph node involvement in the clinical cases on the T M A suggests that it may not be involved. Alternatively, podocalyxin overexpression could play a role in hematogenous metastasis, which is a major contributor to post-operative relapse due to bone metastasis (Braun et al., 2000; Solakoglu et al., 2002). The latter possibility is particularly attractive as Podocalyxin is normally found on the surface of hematopoietic precursors (McNagny et al., 1997), and because there is a strong positive correlation between the expression of CD34-type proteins and bone marrow homing capacity of these progenitors (Krause et al., 2001). Regardless its role in the route of metastatic spread and/or homing, the data generated by the functional experiments performed here clearly indicate that Podocalyxin overexpression can also act as an anti-adhesive molecule that is capable of disrupting junctional interactions between breast carcinoma cells within the primary tumor. The A J protein E-cadherin is very offen downregulated in lobular breast carcinoma but it is less often repressed in more prevalent ductal tumors (Cleton-Jansen, 2002). Although ectopic Podocalxyin overexpression did not induce a significant loss of E-cadherin in MCF-7 breast carcinoma cells, it was thought that it did cause E-cadherin and P-catenin to become slightly diffused along the cell periphery in monolayer culture. However, staining control studies show that podocalyxin and E-cadherin (or p-catenin) colocalized staining that is seen in the apical regions is due the mouse secondary antibodies binding to the anti-Podocalyxin antibody (PCLP-1). Thus there is only a subtle 160 effect on E-cadherin and P-catenin localization. Regardless, the decrease in transepithelial resistance and the additional mislocalization of TJ protein ZO-1 in delaminating Podocalyxin monolayers suggested to me that apical-basal polarity was also being disrupted. The fact that the establishment of apical/basal polarity is a prerequisite for the establishment of normal breast tubule morphogenesis (Bissell and Bilder, 2003) led us to carry out 3D B M morphogenesis assays. MCF-7 cells form spheroids in 3D culture that approximate low grade, non-metastatic tumors in morphogenetic terms. Specifically, like carcinomas in situ, MCF-7 spheroids do not form central lumina, presumably because the tumor cells generate a combination of inappropriate proliferative and anti-apoptotic signaling that prevents lumen formation in normal breast epithelial cells (Debnath et al., 2002). MCF-7 spheroids do, however, polarize (Wang et al., 2002). Thus, the TJ-associated proteins ZO-1 and occludin were apically located in a very tight, non-luminal ring near the center of vector-transfected spheroids. In contrast, both TJ proteins were inappropriately localized at the outer, basal surface of podocalyxin spheroids. This indicates that polarity was severely perturbed by Podocalyxin overexpression. A similar perturbation occurs when HER-2 signaling is constitutively activated (Muthuswamy et al., 2001) or i f integrin-mediated interactions with the B M are interrupted (Weaver et al., 2002; see chapter 2). Such a perturbation of apical/basal polarity is a histopathological hallmark of increased breast tumor grade (Elston and Ellis, 1991) and increased grade was positively correlated with high Podocalyxin overexpression in the clinical cases on the T M A . Therefore, Podocalyxin's ability to disrupt cell junctions and perturb apical/basal polarity indicates that it is capable of functionally contributing to changes in tissue architecture that help drive metastatic progression (Roskelley and Bissell, 2002). In addition to the data presented here for invasive breast cancers, Podocalyxin is dysregulated in human embryonal carcinomas (Schopperle et al., 2002). Under anemic conditions Podocalyxin levels increase significantly in mouse erythroid progenitors (Kelly McNagny, Biomedical research center, U B C . unpublished obs.), which suggests that expression may be upregulated under the same hypoxic conditions that help to drive primary breast tumor progression (Knowles and Harris, 2001). Upregulation of the hypoxia-regulated marker C A I X predominates in estrogen 161 receptor-negative invasive breast carcinomas (Chia et al., 2001) and we observed a significant positive correlation between the intensity of podocalyxin and C A I X immunostaining in our TMAs. This supports the notion that Podocalyxin expression is upregulated in the hypoxic state. Although a detailed dissection of the Podocalyxin promoter has not yet been performed, it is a direct transcriptional target of the Wilm's Tumor suppressor protein (Palmer et al., 2001) and the gene's expression is altered, either directly or indirectly, by estrogen signaling during normal mammary gland development (Ginger etal., 2001). A specific perturbation of TJs has been implicated in breast tumor progression (Hoover et al., 1997; Kramer et al., 2000; Kominsky et al., 2003). In addition to mislocalization of TJ proteins, I also found that Podocalyxin overexpression functionally disrupted TJs as evidenced by decreased transepithelial resistance. The structure and function of TJs are regulated, at least in part, by a number of PDZ-domain containing proteins, including ZO-1, that influence multiprotein scaffolding and interactions with the actin cytoskeleton (Tsukita et al., 2001). Recently, the sodium-hydrogen exchanger regulatory factorNHERF (also called EBP50) and its close relative NHERF2 (also called E3KARP and TKA1), have been shown to bind the cytoterminal D T H L sequence of Podocalyxin via tandem PDZ domains. NHERF proteins also contain an E R M motif that links Podocalyxin and other CD34 family members to the cytoskeleton (Takeda et al., 2001; Weinman, 2001; L i et al., 2002). Interestingly, NHERF is transcriptionally upregulated by estrogen in breast carcinoma cell lines (Ediger et al., 1999) and its expression is greatly diminished in estrogen receptor negative breast tumors (Stemmer-Rachamimov et al., 2001). In contrast, the opposite appears to be true of Podocalyxin; Podocalyxin overexpression was highest in the estrogen receptor negative MDA231 breast carcinoma cell line, which has very low levels of NHERF (Ediger et al., 1999) and both podocalyxin overexpression and mislocalization were positively correlated with estrogen receptor loss on the T M A . Thus, one intriguing possibility is that the normal function of NHERF is to bind Podocalyxin and aid in its localization to discrete apical membrane microdomains distinct from the basolateral domains containing AJs. In this scenario, loss of NHERF expression could lead to inappropriate targeting of Podocalyxin to all membrane domains such that it would perturb TJ-dependent polarity. This notion is 162 supported by the findings that, loss of polarity in kidney glomerular epithelial cells is closely correlated with the uncoupling of NHERF from Podocalyxin (Takeda et al., 2001). Further testing of this model in breast cells will require establishing whether both NHERF-loss and Podocalyxin-upregulation are strict prerequisites for tumor progression and whether ectopic NHERF expression is capable of reverting metastatic behavior. Regardless the outcome of these future studies, our current findings suggest that the regulation of TJ dynamics, like AJs (Fujita et al., 2003; Osterreich et al., 2003), may be dysregulated by changes in hormonal status during breast tumor progression. This ectopic expression data indicate that Podocalyxin expression alone was not sufficient for these cells to become invasive, however it could be an early event in cancer progression from DCIS to IDC. 163 Table 4.1 Characteristics of the Invasive Breast Carcinoma Tissue Microarray (n=272) Age Mean 61.7 years Range 28-86 years Follow Up Time Median 14.9 years Mean 14.5 years Range 6.3-26.7 years Tumor Size <20mm 133 (48.8%) >20mm 103 (38.1%) Unknown 36 (13.1%) Lymph Node Involvement Negative 160 (58.8%) Positive 80 (29.4%) Unknown 32 (11.8%) ER Status Negative 48 (17.6%) Positive 194 (71.3%) Unknown 30 (11.1%) Tumor Grade 1 54 (19.9%) 2 147 (54%) 3 71 (26.1%) HER2 Immunoreactivity Present 34 (12.5%) Absent 224 (82.4%) Unknown 14 (5.1%) p53 Immunoreactivity Positive 34 (12.4%) Negative 202 (74.3%) Unknown 36 (13.3%) Mean Survival Disease Specific 14.7 ± 0.5 years Overall 11.9 ± 1.3 years 164 Table 4.2 Cox Regression Multi-Variant Analysis of Disease Specific Survival Marker Degrees Significance Relative 95% Confidence of (P)* Risk Interval for RR Freedom (RR)** Lower Upper High Podocalyxin 1 0.0001 8.446 2.982 23.917 P53 Immunoreactivity 1 0.581 1.329 0.485 3.643 ER Status 1 0.498 0.716 0.273 1.881 HER2 Overexpression 1 0.136 1.913 0.814 4.494 Lymph Node Involvement 1 0.012 3.688 1.581 8.601 Tumor Grade, High 2 0.663 1.253 0.454 3.545 Tumor Size, >2cm 1 0.369 1.364 0.692 2.689 *Considered a significant independent indicator of poor outcome at p<0.05. **Mean fold increase in relative risk of the group compared to the entire population used to generate the tissue microarray ***Upper and lower margins of relative risk using 2 std. deviations of variation about the mean 165 Table 4.3 Marker Correlation with High Podocalyxin Expression Marker Pearson Correlation* Significance** p53 Immunoreactivity +0.165 0.011 ER Status -0.263 0.003 HER2 Overexpression +0.074 0.233 Lymph Node Involvement -0.041 0.532 Tumor Grade, High +0.192 0.001 Hypoxia (CAIX Expression) +0.241 0.0007 *Positive correlation between zero and +1; negative correlation between zero and **Considered to be a significant positive or negative correlation at p<0.05. 166 Figure 4.1 Podocalyxin is highly expressed in a subset of invasive breast tumors Normal tissue sections (A, B) and sections from an invasive breast carcinoma tissue microarray (C-F) were immunostained for Podocalyxin. In positive control kidney tissue (A), podocytes within the glomerulus stained intensely (brown) while the tubular epithelium was negative. The vascular endothelium within the kidney cortex was also positive (see inset; arrows). In normal breast tissue (B), positive staining was observed in the vascular endothelium and in the apical regions of luminal breast epithelial cells (see inset; arrowheads). On the tissue microarray 272 invasive breast carcinomas were scored as: '0' (ie. C) i f there was no discernible staining on the carcinoma cells (see inset; positive staining is on endothelial cells, arrow); '1' (ie. D) i f less than 10% of the cells stained positively; '2' (ie. E) i f there was a mixture of diffuse staining in more than 10% of the cells and/or intense staining in less than 50% of the cells; or '3' (ie. F) i f there was intense staining in more than 50% of the cells. The significant disruption of epithelial tissue architecture was a consistent feature of the latter tumors (bar= 60pm in low power views, 30pm in insets). 167 168 Figure 4.2 High Podocalyxin expression is associated with poor outcome In A , disease-specific survival analysis of tumors scored on the tissue microarray those with greater than 50% of the cells staining intensely for Podocalyxin (ie. high Podocalyxin overexpressors, category '3', closed circles) had a poorer outcome than those of the other three Podocalyxin staining categories ('0' open squares; '1' hatched marks; '2' closed triangles). Therefore, in B, categories 0 to 2 were combined as "no or low Podocalyxin" (inverted closed triangles) and compared to category 3 alone which was designated as "high Podocalyxin" (closed circles). 169 0 10 20 Total Follow-up (years) Low/No Podocalyxin High Podocalyxin 10 20 Total Follow-up (years) 170 Figure 4.3 Expression of Podocalyxin in human breast carcinoma cell lines A - Endogenous Podocalyxin levels in three human breast carcinoma lines, T47D, M C F -7 and MDA-231, were assessed by Western blotting (lower panel, E R K 1/2 loading control). B - Human MCF-7 cells were stably transfected with vectors expressing EGFP alone or co-expressing EGFP and full length mouse Podocalyxin. EGFP/Vector-control transfected cells formed classical MCF-7 cobblestone epithelial monolayers (top panel) while bulging cells were shed from the surface of the EGFP/Podocalyxin transfected cells (middle panel). EGFP (green) and mouse Podocalyxin (red) were coordinately expressed in cells transfected with the EGFP/Podocalyxin vector (lower panel). Note that the ectopic mouse Podocalyxin protein was only expressed by a subset of the transfected cells but that the protein was appropriately targetted to the cell surface and it was consistently expressed by cells that bulged apically as demonstrated by the upward migration of the DAPI-stained nuclei (blue). Upper two panels, live phase microscopy, bar = 50pm; lower panel, Z-series confocal microscopy, bar = 15pm. 171 165kD— ••—Podocalyxin 42kD — £=ERK 1/2 172 Figure 4.4 Podocalyxin overexpression has a subtle effect on E-cadherin and occludin localization in MCF-7 cell monolayers A, B - EGFP/Podocalyxin transfectants in monolayer culture were triple-stained for mouse Podocalyxin (red), DAPI/Nuclei (blue) and either the transmembrane adherens junction protein E-cadherin (A, green) or the transmemembrane tight junction protein occludin (B, green) and subjected to Z-series confocal microscopy. Note that where ectopic Podocalyxin was not expressed (ie. left side of panels) E-cadherin was localized basolaterally at sites of cell-cell contact and occludin was discretely localized at apical terminal bars. In contrast, in the cells expressing the mouse Podocalyxin (ie. right side of panels) both E-cadherin and occludin localization were disrupted relocalized to the entire outer surface of apically bulging cells where they co-localized with Podocalyxin (bar=15um). C- The control cells (top panel) and Podocalyxin overexpressing cells (lower panel) where were first exposed to rat anti PCPL-1 (for Podocalyxin) antibody and then fixed and exposed to FITC-anti mouse second antibodies. These were the same second antibodies used to detect E-cadherin, occludin and P-catenin (see Figure 4.6). Thus, the apical staining observed with E-cadherin and occludin was due to the non-specific second antibody binding to PCLP-1 antibody. 173 Podocalyxin E-Cadherin +DAPI • _ odocalyxin Occludin +DAPI Rat PCLP-1 Anti Ms secondary Merged +DAPI Rat PCLP-1 Anti Ms secondary Merged +DAPI 174 F i g u r e 4.5 P o d o c a l y x i n overexpress ion has no effect on steady state levels of E-c a d h e r i n a n d O c c l u d i n . Steady state levels of E-cadherin and occludin in EGFP vector control-Podocalyxin/EGFP- transfectants were determined by Western blotting. There was detectable change in the expression levels of E-cadherin and occludin. 175 176 Figure 4.6 Podocalyxin overexpression alters the localization of peripheral membrane cell junction proteins in MCF-7 cell monolayers. EGFP/Podocalyxin transfectants in monolayer culture were double-stained for mouse Podocalyxin (red) and either the adherens junction P-catenin (A, green) or the tight junction protein ZO-1 (B, green). Note that in cells expressing the mouse Podocalyxin transgene (right side of the panel) the normally basolateral P-catenin (arrows) became circumferentially localized while the normally apical ZO-1 was found at the base of apically bulging cells (arrowheads; confocal Z series microscopy, bar=15pm). 177 V „ Podocalyxin G-Catenin B Podocalyxin ZO-1 178 Figure 4.7 Podocalyxin overexpression does not disrupt adherens junctions in basement membrane gel culture MCF-7 cells stably transfected with the control EGFP- or the EGFP/Podocalyxin-expression vectors were cultured as spheroids on basement membrane gels for four days followed by dual staining for ectopic mouse Podocalyxin (red) with either E-cadherin (A) or B-catenin (B, green; EGFP signal was lost when cells were fixed in MeOH). Note that mouse Podocalyxin was localized around the outer edge of the spheres, which is the basal surface under these conditions. Both E-cadherin and B-catenin were found at sites of cell-cell contact in both control and Podocalyxin expressing spheroids indicating that adherens junctions were likely intact under both conditions. Localization of the immunostained proteins was then assessed by confocal microscopy (bar = 15pm). 179 180 Figure 4.8 Podocalyxin overexpression disrupts apical/basal polarity and spheroid architecture. A - Control- and EGFP/Podocalyxin- transfected cells cultured on basement membrane gels were dual stained for the tight junction proteins ZO-1 (red) and occludin (green; EGFP signal was lost when in MeOH fixation). Protein localization was then assessed by confocal microscopy (bar=15pm). Note that in the control spheroids the tight junction proteins were located at central sites of cell-cell contact (ie. at apical cell surfaces within the spheroid) while in the Podocalyxin-transfected spheroids both proteins were located on the outer edge of the cells (ie. the basal surface where ectopic Podocalyxin was also localized, see Figure 4.6 above). B- The morphology of the spheroids was assessed by live phase microscopy (bar=50pm). Notice that the control transfectants formed spheroids with smooth, well-defined edges while the Podocalyxin transfectants formed ragged-edged, multi-lobulated spheroids. 181 182 4.6 References Aubele, M . , Mattis, A. , Zitzelsberger, H. , Walch, A. , Kremer, M . , Welzl, G., Hofler, H . , and Werner M . (2000). Extensive ductal carcinoma in situ with small foci of invasive ductal carcinoma: evidence of genetic resemblance by C G H . Int. J. Cancer. 55, 82-86. Baumhueter, S., Singer, M.S., Henzel, W., Hemmerich, S., Renz, M . , Rosen, S.D. and Lasky, L .A . (1993). Binding of L-selectin to the vascular sialomucin CD34. Science 262, 436-8. Berx, G., and Van Roy, F. (2001). The E-cadherin/catenin complex: an important gatekeeper in breast cancer tumorigenesis and malignant progression. Breast Cancer Res. 3, 289-293. Bissell, M.J. and Bilder, D. (2003). Polarity determination in breast tissue: desmosomal adhesion, myoepithelial cells, and laminin 1. Breast Cancer Res. 5, 117-119. Bistrup, A. , Bhakta, S., Lee, J.K., Belov, Y . Y . , Gunn, M.D. , Zuo, F.R., Huang, C.C., Kannagi, R., Rosen, S.D., and Hemmerich, S. (1999). Sulfotransferases of two specificities function in the reconstitution of high endothelial cell ligands for L-selectin. J. Cell Biol. 145, 899-910. Braun, S., Pantel, K. , Muller, P., Janni, W., Hepp, F., Kentenich, C.R.M. , Gastroph, S., Wischnik, A. , Dimpfl, T., Kindermann, G., Riethmuller, G., and Schlimok, G. (2000). Cytokeratin-positive cells in the bone marrow and survival of patients with stage I, II or III breast cancer. New Eng. J. Med. 342, 525-533. Brooks, S., Locke, R., and Soule, H. (1973). Estrogen receptor in a human cell line (MCF-7) from breast carcinoma. J. Biol. Chem. 248(17), 6251-6253. 183 Chia, S.K., Wykoff, C C , Watson, P.H., Han, C , Leek, R.D., Pastorek, J., Gatter, K . C , Ratcliffe, P., Harris, A . L . Prognostic significance of a novel hypoxia-regulated marker, carbonic anhydrase IX, in invasive breast carcinoma. J. Clin. Oncol. 16, 3660-3668. Cleton-Jansen, A . M . (2002). E-cadherin and loss of heterozygosity at chromosome 16 in breast carcinogenesis: different genetic pathways in ductal and lobular breast cancer? Breast Cancer Res. 4, 5-8. Debnath, J., Mills, K.R., Collins, N.L. , Reginato, M.J. , Muthuswamy, S.K., and Brugge, J.S. (2002). The role of apoptosis in creating and maintaining luminal space within normal and oncogene-expressing mammary acini. Cell 111, 29-40. Doyonnas, R., Kershaw, D.B., Duhme, C , Merkens, H. , Chelliah, S., Graf, T. and McNagny, K . M . (2001). Anuria, omphalocele, and perinatal lethality in mice lacking the CD-34-related protein Podocalyxin. J Exp Med, 794,13-27. Ediger, T.R., Kraus, W.L., Weinman, E.J. and Katzenellenbogen, B.S. (1999). Estrogen receptor regulation of the Na+/H+ exchange regulatory factor. Endocrinology 140, 2976-82. Eifel, P., Axelson, J.A., Costa, J., Crowley, J., Curran, W.J. Jr, Deshler, A . , Fulton, S., Hendricks, C.B., Kemeny, M . , Kornblith, A .B . , Louis, T.A., Markman, M . , Mayer, R., and Roter, D. (2001). National Institutes of Health Consensus Development Conference Statement: adjuvant therapy for breast cancer. J. Nat'l. Cancer Inst. 93, 979-89. Elston, C W . and Ellis, I.O. (1991). Pathological prognostic factors in breast cancer. I. The value of histological grade in breast cancer: exprerience from a large study with long-term follow-up. Histopathology 19, 403-410. Fearon, E.R. (2003). Connecting estrogen receptor function, transcriptional repression, and E-cadherin expression in breast cancer. Cancer Cell 3, 307-309. 184 Forozan, F., Mahlamaki, E.H., Monni, O., Chen, Y . , Veldman, R., Jiang, Y . , Gooden, G.C., Ethier, S.P., Kallioniemi, A . and Kallioniemi, O.P. (2000). Comparative genomic hybridization analysis of 38 breast cancer cell lines: a basis for interpreting complementary D N A microarray data. Cancer Res. 60, 4519-25. Fujita, N . , Jaye, D.L., Kajita, M . , Geigerman C , Moreno, C.S., and Wade, P.A. (2003). M T A 3 , a Mi-2/NuRD complex subunit, regulates an invasive growth pathway in breast cancer. Cell 113,207-219. Gillett, C.E., Miles, D.W., Ryder, K. , Skilton, D., Liebman, R.D., Springall, R.J., Barnes, D .M. , Hanby, A . M . (2001). Retention of the expression of E-cadherin and catenins is associated with shorter survival in grade III ductal carcinoma of the breast. J. Pathol, 193, 433-441. Ginger, M.R., Gonzalez-Rhnbau, M.F., Gay, J.P. and Rosen, J .M. (2001). Persistent changes in gene expression induced by estrogen and progesterone in the rat mammary gland. Mol . Endocrinol 15, 1993-2009. Hoover, K .B . , Liao, S.-Y., and Bryant, P.J. (1997). Loss of tight junction M A G U K ZO-1 in breast cancer. Am. J. Pathol. 153, 1767-1773. Kerjaschki, D., Sharkey, D.J., and Farquhar, M . G . (1984) Identification and characterization of podocalyxin-the major sialoprotein of the renal glomerular epithelial cell. J. Cell Biol. 98, 1591-1596. Kerjaschki D, Vernillo AT, Farquhar M G (1995). Reduced sialylation of podocalyxin-the major sialoprotein of the rat kidney glomerulus—in aminonucleoside nephrosis. A m J Pathol. 118(3):343-9. 185 Kershaw, D.B., Thomas, P.E., Wharram, B.L. , Goyal, M . , Wiggins, J.E., Whiteside, C.I., and Wiggins, R.C. (1995). Molecular cloning, expression, and characterization of podocalyxin-like protein 1 from rabbit as a transmembrane protein of glomerular podocytes and vascular endothelium. J. Biol. Chem. 270, 29439-29446. Kershaw, D.B., Beck, S.G., Wharram, B.L. , Wiggins, J.E., Goyal, M . , Thomas, P.E., and Wiggins, R.C. (1997a). Molecular cloning and characterization of human podocalyxin-like protein. J. Biol. Chem. 272, 15708-15714. Kershaw, D.B., Wiggins, J.E., Wharram, B.L. , and Wiggins, R.C. (1997b). Assignment of the human podocalyxin-like protein (PODXL) gene to 7q32-q33. Genomics 45, 239-240. Knowles, H.J., and Harris, A . L . (2001). Hypoxia and oxidative stress in breast cancer. Hypoxia and tumorigenesis. Breast Cancer Res. 3, 318-322. Kominsky, S.L., Argani, P., Korz, D., Everon, E., Raman, V. , Garrett, E., Rein, A. , Sauter, G., Kallioniemi, O.-P., and Sukumar, S. (2003). Loss of the tight junction protein claudin-7 correlates with histological grade in both ductal carcinoma in situ and invasive ductal carcinoma of the breast. Oncogene 22, 2021-2033. Kononen, J., Bubendorf, L. , Kallioniemi, A . , Barlund, M . , Schraml, P., Leighton, S., Torhorst, J., Mihatsch, M.J. , Sauter, G., and Kallioniemi, O.P. (1998). Tissue microarrays for high-throughput molecular profiling of tumor specimens. Nat. Med. 4, 844-7. Kramer, F., White, K. , Kubbies, M . , Swisshelm, K. and Weber, B.H.F. (2000). Genomic organization of claudin-1 and its assessment in hereditary and sporadic breast cancer. Hum. Genet. 101, 249-256. Krause, D.S., Fackler, M.J. , Civin, C.I., and May, W.S. (1996). CD34: structure, biology and clinical utility. Blood 87, 1-13. 186 Krause, D.S., Theise, N.D. , Collector, M.I., Henegariu, O., Hwang, S., Gardner, R., Neutzel, S., and Sharkis, S.J. (2001). Multi-organ, multi-lineage engraftment by a single bone marrow-derived stem cell. Cell 105, 369-377. Kurihara H , Anderson JM, Kerjaschki D and Farquhar M G . (1992) The altered glomerular filtration slits seen in puromycin aminonucleoside nephrosis and protamine sulfate-treated rats contain the tight junction protein ZO-1. A m J Pathol. 141(4), 805-16. L i , Y . , L i , J., Straight, S.W., Kershaw, D.B. (2002). PDZ domain-mediated interaction of rabbit podocalyxin and Na(+)/H(+) exchange regulatory factor-2. Am. J. Physiol. Renal Physiol. 282, F l 129-39. Liu, C.L., Prapong, W., Natkuman, Y . , Alizadeh, A. , Montgomery, K., Gilks, C.B., van de Rijn, M . (2002). Software tools for high-throughput analysis and archiving of immunohistochemistry staining data obtained with tissue microarrays. Am. J. Pathol. 161, 1557-1564. Macek R, Swisshelm K , Kubbies M . (2003) Expression and function of tight junction associated molecules in human breast tumor cells is not affected by the Ras -MEKl pathway. Cell Mol Biol. 49(1):1-11 Makretsov, N . , Gilks, B., Coldman, A. , Hayes, M . , and Huntsman, D. (2003). Tissue microarray analysis of neuroendocrine differentiation and its prognostic significance in breast cancer. Human Pathol. In Press. McNagny, K . M . , Pettersson, I., Rossi, F., Flamme, I., Shevchenko, A . , Mann, M . , and Graf, T. (1997). Thrombomucin, a novel cell surface protein that defines thrombocytes and multipotent hematopoietic progenitors. J. Cell Biol. 738, 1395-1407. 187 Muthuswamy, S.K., L i , D., Lelievre, S., Bissell, M.J. , and Brugge, J.S. (2001). ErbB2, but not ErbBl , reinitiates proliferation and induces luminal repopulation in epithelial acini. Nat. Cell Biol. 3, 785-92. Nabholtz, J.M., and Slamon, D. (2001). New adjuvant strategies for breast cancer: meeting the challenge of integrating chemotherapy and trastuzumab (Herceptin). Semin. Oncol. 28, 1-12. Oesterreich, S., Deng, W., Jiang, S., Cui, X . , Ivanova, M . , Schiff, R., Kang, K. , Hadsell, D.L., Behrens, J. and Lee, A . V . (2003). Estrogen-mediated down-regulation of E-cadherin in breast cancer cells. Cancer Res. 63, 5203-8. Palmer, R.E., Kotsianti, A. , Cadman, B., Boyd, T., Gerald, W., and Haber, D.A. (2001). WT1 regulates the expression of the major glomerular podocyte membrane protein Podocalyxin. Curr Biol. 11, 1805-1809. Pauletti, G., Dandekar, S., Rong, H., Ramos, L. , Peng, H. , Seshadri, R., and Slamon D.J. (2000). Assessment of methods for tissue-based detection of the HER-2/neu alteration in human breast cancer: a direct comparison of fluorescence in situ hybridization and immunochemistry. J. Clin. Oncol. 78, 3561-3564. Parker, R.L., Huntsman, D.G., Lesack, D.W., Cupples, J.B., Grant, D.R., Akbari, M . and Gilks, C B . (2002). Assessmentof interlaboratory variation in the immunohistochemical determination of estrogen receptor status using a breast cancer tissue microarray. A m J Clin Pathol 111, 723-728. ' /• Pollack, J.R., Sorlie, T., Perou, C M . , Rees, C.A., Jeffrey, S.S., Lonning, P.E., Tibshirani, R., Botstein, D., Borresen-Dale, A . L . , Brown, P.O. (2002). Microarray analysis reveals a major direct role of D N A copy number alteration in the transcriptional program of human breast tumors. Proc. Nat'l. Acad. Sci. (USA). 99, 12963-12968. 188 Reiser, J., Kriz, W., Kretzler, M . , and Mundel, P. (2000). The glomerular slit diaphragm is a modified adherens junction. J. Am. Soc. Nephrol. 71, 1-8. Roskelley, C D . and Bissell, M.J . (2002). The dominance of the microenvironment in breast and ovarian cancer. 72, 97-104. Sassetti, C , Tangemann, K. , Singer, M.S., Kershaw, D.B., and Rosen, S.D. (1998). Identification of podocalyxin-like protein as a high endothelial venule ligand for L-selectin: parallels to CD34. J. Exp. Med. 757, 1965-1975. Schnabel, E., Dekan, G., Miettinen, A. , and Farquhar, M . G . (1989). Biogenesis of podocalyxin-the major glomerular sialoglycoprotein-in the newborn rat kidney. Eur. J. Cell Biol. 48, 313-326. Schopperle, W.M. , Kershaw, D.B., and DeWolf, W.C. (2002). Human embryonal carcinoma antigen, Gp200/GCTM-2 is podocalyxin. Biochem Biophys. Res. Comm. 300, 285-290. Seiler M W , Rennke HG, Venkatachalam M A , and Cotran RS. (1977) Pathogenesis of polycation-induced alterations ("fusion") of glomerular epithelium. Lab Invest. 36(1), 48-61. Solakoglu, O., Maierhofer, C , Lahr, G., Breit, E., Scheunemann, P., Heumos, I., Pichlmeier, U . , Schlimok, G., Oberneder, R., Kollermann, M.W., Kollermann, J., Speicher, M.R., and Pantel, K. (2002). Heterogeneous proliferative potential of occult metastatic cells in bone marrow of patients with solid epithelial tumors. Proc. Nat'l. Acad. Sci. (USA) 99, 2246-51. Somasiri, A. , Howarth A. , Goswami D., Dedhar S., and Roskelley, C D . (2001). Overexpression of the integrin-linked kinase mesenchymally transforms mammary epithelial cells. J. Cell Sci 114, 1125-1136. 189 Sorlie, T., Tibshirani, R., Parker, J., Hastie, T., Matron, J.S., Nobel, A. , Deng, S., Johnsen, H. , Pesich, R., Geisler, S., Demeter, J., Perou, C M . , Lonning, P.E., Brown, P.O., Borresen-Dale, A . L . and Botstein, D. (2003). Repeated observation of breast tumor subtypes in independent gene expression data sets. Proc. Natl. Acad. Sci. (USA) 100, 8418-23. Stemmer-Rachamimov, A.O., Wiederhold, T., Nielsen, G.P., James, M . , Pinney-Michalowski, D., Roy, J.E., Cohen, W.A., Ramesh, V . , Louis, D.N. (2001) NHE-RF, a merlin-interacting protein, is primarily expressed in luminal epithelia, proliferative endometrium, and estrogen receptor-positive breast carcinomas. Am. J. Pathol. 758, 57-62. Takeda, T., Go, W.Y., Orlando, R.A., and Farquhar, M . G . (2000). Expression of podocalyxin inhibits cell-cell adhesion and modifies junctional properties in Madin-Darby Canine Kidney Cells. Mol . Biol. Cell 71, 3219-3232. Takeda, T., McQuistran, T., Orlando, R.A., and Farquhar, M . G . (2001). Loss of glomerular foot processes is associated with uncoupling of podocalyxin from the actin cytoskeleton. J. Clin. Invest. 708, 289-301. Tsukita, S., Furuse, M . , and Itoh, M . (2001). Multifunctional strands in tight junctions. Nat. Rev. Mol . Cell Biol. 2, 285-293. van't Veer, L.J. , Dai, H. , van de Vijver, M.J. , He, Y .D . , Hart, A . A . , Mao, M . , Peterse, H.L., van der Kooy, K. , Marton, M.J. , Witteveen, A.T., Schreiber, G.J., Kerkhoven, R .M. , Roberts, C , Linsley, P.S., Bernards, R., and Friend, S.H. (2002). Gene expression profiling predicts clinical outcome of breast cancer. Nature 415, 530-536. Wang, F., Hansen, R.K., Radisky, D., Yoneda, T., Barcellos-Hoff, M.H. , Petersen, O.W., Turley, E.A., Bissell, M.J . (2002). Phenotypic reversion or death of cancer cells by 190 altering signaling pathways in three-dimensional contexts. J. Nat'l. Cancer Inst. 94, 1494-503. Weaver, V . M . , Lelievre, S., Lakins, J.N., Chrenek, M.A. , Jones, J .C , Giancotti, F., Werb, Z., Bissell, M.J . (2002). B4 integrin-dependent formation of polarized three-dimensional architecture confers resistance to apoptosis in normal and malignant mammary epithelium. Cancer Cell 2, 205-16. Weinman, E.J. (2001). New functions for the NHERF family of proteins. J Clin Invest. 10%, 185-186. 191 CHAPTER 5: Overexpression of the integrin-linked kinase leads to loss of polarity, cell junctions and mesenchymally transforms mammary epithelial cells2 5.1 Summery Cell surface integrin receptor interaction with laminin in the B M contributes to mammary epithelial cell morphogenesis and differentiation. The integrin-linked kinase (ILK) is one of the signaling moieties that associates with the cytoplasmic domain of p i integrin subunits with some specificity. Forced expression of a dominant negative, kinase-dead form of ILK subtly altered mouse mammary epithelial cell morphogenesis but it did not prevent differentiative milk protein expression. In contrast, forced overexpression of wild-type ILK strongly inhibited both morphogenesis and differentiation. Overexpression of wild-type ILK also caused the cells to lose the cell-cell adhesion molecule E-cadherin, become invasive, reorganize cortical actin into cytoplasmic stress fibers, and switch from an epithelial cytokeratin to a mesenchymal vimentin intermediate filament phenotype. Forced expression of E-cadherin in the latter mesenchymal cells rescued epithelial cytokeratin expression and it partially restored the ability of the cells to differentiate and undergo morphogenesis. These data demonstrate that ILK, which responds to interactions between cells and the extracellular matrix, induces a mesenchymal transformation in mammary epithelial cells, at least in part, by disrupting cell-cell junctions. 2 A version of this chapter has been published. Somasiri A , Howarth A , Goswami D, Dedhar S and Roskelley CD. (2001) Overexpression of the integrin-linked kinase mesenchymally transforms mammary epithelial cells. J Cell Sci. 114(Pt 6), 1125-36. 192 5.2 Introduction Integrins mediate cell-extracellular matrix (ECM) interactions, which is crucial for most cellular functions including normal growth and development (Bissell et al., 1982; Adams and Watt, 1993; Giancotti and Roushlahti, 1999). "Outside-in signals" generated by integrin-ECM interaction can induce intracellular signals that regulate gene expression that affect cell migration, adhesion as well as cell polarity. Furthermore integrin ligation can activate Rho family of small GTPases (Rho, Rac and Cdc42), which can modulate the organization of the actin cytoskeleton to influence cell shape as well as cellular junctions. The effects Outside-in signals are particularly evident in the mammary gland where interactions between mammary epithelial cells and the basement membrane E C M profoundly influence proliferation, differentiation, morphogenesis and apoptosis during the developmental cycles of pregnancy, lactation and involution (see chapter 1; Schmeichel et al., 1998). Many of these phenotypes can be modeled in vitro using 3D culture systems (Somasiri and Roskelley, 1999). For example, when mammary epithelial cells derived from mid-pregnant mice are placed on reconstituted basement membrane gels they round-up, aggregate and polarize. This morphogenic process is completed when the resulting spherical structures cavitate to form a central lumen (Aggeler et al., 1991). The cells within these structures express milk proteins and secrete them vectorially into the central lumen (Barcellos-Hoff et al., 1989). Differentiative expression of one of these milk proteins, p-casein, is transcriptionally regulated (Schmidhauser et al., 1990) and this regulation requires an interaction between p i subunit-containing integrins on the cell surface and laminin in the basement membrane (Streuli et al., 1991; Streuli et al., 1995; Muschler et al., 1999). In addition, when some these integrin-laminin interactions are functionally blocked the cells no longer respond appropriately to soluble factors in the microenvironment and they ultimately undergo apoptosis (Boudreau et al., 1995; Pullan etal., 1996). Integrins have overlapping and competing affinities for individual E C M ligands (Hynes, 1992). In response to E C M ligation, integrins cluster in the plane of the membrane and recruit large, protein plaques that associate with the cytoskeleton and initiate numerous signaling cascades (Juliano and Haskill, 1993; Yamada and Geiger, 1997). Components of the M A P kinase signaling pathway are often associated with these 193 plaques and, in some cases, this pathway is stimulated by the integrin-mediated activation of the focal adhesion kinase (FAK; Giancotti and Ruoslahti, 1999). F A K is activated when mammary epithelial cells contact laminin (Roskelley and Bissell, 1995) but the functional consequences of this activation are not clear as inhibition of the M A P kinase pathway does not prevent differentiative milk protein gene induction (Wartmann et al., 1996). Integrin ligation also activates PI3-K (Khwaja et a l , 1997; King et al., 1997) and I have shown that PI3-K is required for both BM-dependent morphogenesis and differentiative P-casein expression in mammary epithelial cells (Somasiri et al., 2000). BM-dependent integrin signaling also increase the ability of IGFs to activate PI3-K signaling and prevent apoptosis in these cells (Farrelly et al., 1999). One integrin-associated signaling molecule that has been implicated in PI3-K signaling is ILK. ILK associates with the cytoplasmic domain of p i and P3 integrin subunits and the kinase activity of the molecule, which is serine/threonine directed, is modulated by integrin ligation in a PI3-K dependent manner (Hannigan et al., 1996; Delcomenne et al., 1998). ILK also acts downstream of PI3-kinase by stimulating the phosphorylation of protein kinase B (PKB; also known as Akt) on serine 473 (Lynch et al., 1999; Persad et al., 2000). This potentiation of P K B plays a role in the ability of ILK to downregulate glycogen synthase kinase-3, which in turn influences both P-catenin and AP-1 signaling (Novak et al., 1998; Troussard et al., 1999). ILK also contains four non-catalytic ankyrin repeats which help target the molecule to focal adhesion complexes and link it to growth factor receptor tyrosine kinase signaling via the adapter protein PinchAJnc97 (Tu et al., 1998; Hobert et al., 1999; L i et al., 1999). In epithelial cells, forced overexpression of wild-type ILK suppresses suspension-mediated apoptosis and stimulates anchorage-independent growth (Radeva et al., 1997). Thus, ILK overexpression appears to constitutively activate integrin signaling pathways such that the anchored state is mimicked in the absence of cell-ECM interactions (Attwell et al., 2000). In this report I investigated the role of ILK in the BM-dependent differentiation and morphogenesis of the scp2 mouse mammary epithelial cell line. I found that ILK activity is increased by cellular interactions with either a reconstituted B M gel or by purified laminin. Forced expression of a dominant negative form of ILK did not prevent differentiative P-casein induction but it did subtly affect spheroidal morphogenesis. In 194 contrast, forced overexpression of wild-type ILK inhibited differentiation, prevented morphogenesis and it initiated a mesenchymal transformation of the cells that was associated with a complete loss of cell polarity and cell-cell junctions. Furthermore, these IKL-transformed cells invade the B M as single cells similar to that of ILC. 5.3 Materials and Methods 5.3.1 Cell culture Routine culturing of scp2 cells were done as described in chapter 2 (page 82). In addition, cell monolayers were overlaid with purified laminin (Sigma, 50 mg/ml) diluted in the presence of lactogenic hormones (Streuli et al., 1995). 5.3.2 Viral infection and transfection Wild-type (wt) ILK and kinase-dead (kd) ILK cDNAs were epitope-tagged in the pcDNA3.1 V5-His expression vector (Invitrogen, Carlsbad C A ; Persad et al., 2000) and subcloned into the MSCV-pac retroviral packaging vector. The kdILK was originally generated by site-directed mutagenesis to change the glutamic acid at position 359 to a lysine residue in the ILK kinase domain (Nonak et al., 1998). Replication-defective ecotropic retrovirus was produced by transient transfection of the latter construct into BOSC-29 packaging cells (Pear et al., 1993). The supernatant from these cells were collected and used to infect scp2 cells. Forty eight hours after retroviral infection, scp2 cells were either lysed in RIPA buffer for an initial analysis of transgene expression or they were put under selection for 5 days (puromycin, 2 mg/ml) and then pooled. Empty MSCV-pac containing virus was used as the control in these experiments. scp2 cells were also transfected using lipofectamine (Gibco-BRL; Gaithersburg MD) with either a control antisense (ILK-14) or a sense wtILK cDNA (ILK-13) under the constitutive control of the C M V promoter (Hannigan et al., 1996). Stable transfectants were produced by selection under G-418 (400 mg/ml) and clones were isolated by limiting dilution. From a 195 total of twelve ILK-13 clones isolated, three were chosen for further study based on their varying levels of ILK overexpression (Novak et al., 1998; also see Table 1). The highest overexpressing clone, ILK-13-8, was re-transfected with a mouse E-cadherin cDNA under the constituitive control of the B-actin promoter (Nose et al., 1988) and a hygromycin selection plasmid (Strategene, La Jolla CA). Stable double-transfectants were produced under hygromycin selection (100 mg/ml) and clones were isolated by limiting dilution. Hygromycin vector alone was stably transfected into ILK13-8 cells as a double-transfectant control. 5.3.3 Immunofluorescence Cells were grown on glass coverslips, fixed in cold (-20°C) MeOH for 15 min and then permeabilized in cold methanol-acetone (1:1) for all antigens except ILK and B-catenin, for which cells were fixed in paraformaldehyde (4%, 15 minutes in PBS) followed by permeabilization with Triton X-100 (0.1%, 10 minutes) at room temperature. Fixed and permeabilized cells were then incubated with antibodies against I L K (mouse monoclonal; L i et al., 1999), E-cadherin and P-catenin (mouse monoclonals, Transduction Labs, Lexington K Y ) , pan-cytokeratin (rabbit polyclonal, Dako, Carpenteria, CA) or vimentin (mouse monoclonal, Sigma). Mouse and rabbit IgGs were used as controls. Binding of primary antibodies was visualized by epifluorescence microscopy after incubation with fluorescently labelled anti-species-specific second antibodies, f-actin was visualized after the binding of rhodamine phalloidin (Molecular Probes, Eugene, OR) to paraformaldehyde-fixed, acetone-permeabilized cells. 5.3.4 Western blotting For ILK and epitope-tag immunoblotting, 5 mg of whole cell NP-40 lysates (1% NP-40, 0.5% deoxycholate) were separated by 10% SDS-PAGE, transferred to nylon membranes (PVDF, Bio-Rad, Hercules, CA) and probed with either an affinity-purified rabbit polyclonal antibody against ILK (91-3; Hannigan et al., 1996) or a mouse monoclonal antibody against the 14 amino acid V5 epitope (Invitrogen, Carlsbad CA). 196 For intermediate filament immunoblotting, 30 mg of solubilized cytoskeletal protein obtained after high salt extraction (Achtstaetter et al., 1986) was separated by 8.5% SDS-PAGE, transferred and probed with either a 1:1 mixture of AE1/AE3 monoclonal antibodies which recognize the great majority of cytokeratins (Sun et al., 1985) or a goat polyclonal antibody against vimentin (Accurate Chemical, Westbury, NY) . For E-cadherin and P-catenin expression levels, 50 mg and 20 mg of whole cell RIPA lysates were assayed as described in chapter 2 (page 84). To assess mammary-specific differentiation, cells were maintained either on tissue culture plastic in the presence or absence of a laminin overlay or on Matrigel, all in the presence of lactogenic hormones for up to five days which results in maximal differentiation of the parent scp2 cell line (Roskelley et al., 1994). On the days indicated, cells were liberated from the matrix using a neutral protease (Dispase, Collaborative Res, Bedford, M A ) , lysed in RIPA buffer and 10 mg of total protein was assayed for P-casein as described previously (Chapter 2, page 84). 5.3.5 ILK kinase activity Scp2 parent cells were maintained in lactogenic hormones as monolayers, as laminin-overlaid clusters, or as spherical structures on Matrigel for 5 days. The cells were then lysed in NP-40 buffer and 200 mg of total protein was immunoprecipitated using the anti-ILK polyclonal antibody. Protein kinase assays were performed in 50 ml kinase reaction buffer (50 m M HEPES, pH 7.0, 10 m M MgC12, 2 m M NaF, 1 m M Na3Vo4) containing 10 mCi [g-32]ATP and 5 mg myelin basic protein as substrate (Upstate Biotechnologies, Lake Placid NY) . Reactions were incubated at 30°C for 30 minutes, stopped by the addition of SDS-PAGE sample buffer and the products were resolved by 15% SDS-PAGE and visualized by fluorography (Hannigan et al., 1996). 5.3.6 RT-PCR Five pg total R N A was subjected to first strand cDNA synthesis kit (Pharmacia, Morgan, Canada) in a total volume of 33 pi. 10 pi of cDNA was then mixed with 200 m M dNTPs, 10 m M Tris-HCl (pH 8.3), 50 m M KC1, 3.0 m M M g C l 2 , 0.001% gelatin, 2.5 197 U taq polymerase (Gibco, Grand Island N Y ) in a final volume of 100 pi and subjected to 35 cycles of amplification using 0.5 pmoles of the appropriate primers. Primers used, which span introns to detect specific mRNA sequences, were 5'-GGGTGACT A C A A A A T C - A A T C - 3 ' and 5' -GGGGGC A G T A A G G G C T C T T T - 3 ' , which amplifies a 252 bp fragment of the transmembrane domain of E-cadherin (Risinger et al., 1995), or, 5" -TGATCCACAT-CTGCTGGAAGGTGG-3 ' and 5'-G G A C C T G A C T G A C T A C C T C - A T G A A - 3'which amplifies a 510 bp fragment of b-actin. Each amplification cycle consisted of denaturation at 94°C for 60 seconds, primer annealing at 58°C for 90 seconds, extension at 72°C for 120°C, and a final extension at 72°C for 15 minutes. Reactions were terminated by adding 6pl of D N A gel loading buffer to each sample and they were separated on a 1% agarose gel. 5.4 Results 5.4.1 Endogenous ILK kinase activity in mammary epithelial cells is upregulated by interactions with the basement membrane Scp2 mouse mammary epithelial cells undergo ECM-dependent morphogenesis and differentiation (Desprez et al., 1993; Roskelley et al., 1994). Laminin in the B M matrix is sufficient to induce P-casein expression and this induction requires functional p i integrins (Streuli et al., 1995; Muschler et al., 1999). In the current study I found that cellular interactions with laminin and B M gels slightly decreased the amount of immunoprecipitable ILK from NP-40 whole cell lysates but at the same time increased the kinase activity of this material (Figure 5.1). These observations led me to ask whether ILK plays a role in the BM-dependent morphogenesis and differentiation of scp2 mammary epithelial cells. 198 5.4.2 Overexpression of wtILK inhibits basement membrane-dependent morphogenesis and differentiation Forced expression of a kinase dead (kd) mutant form of ILK blocks endogenous ILK kinase activity while overexpression of wild-type (wt) ILK leads to a constitutive increase in ILK kinase activity (Persad ef al., 2000). Thus, I used a retroviral delivery system to express these two forms of ILK in scp2 cells. Retroviral infection caused an appreciable expression of the ILK transgenes two days after infection (Figure 5.2A). A 5 day genetic selection period (puromycin, 2 mg/ml) removed about half of the cells and further increased the expression of both kdILK and wtILK on a population basis (Figure 5.2A). Infected, selected and pooled cells were then used rapidly, within two passages, for further experiments. It is important to note that these populations were not cloned, and thus are heterogeneous. In 3D culture, virally-infected kdILK expressing cell populations aggregated and formed spheres. However, these structures were much smaller than those that formed in the vector-infected controls (Figure 5.2B). Despite these morphological effects, induction of p-casein expression was not affected by forced kdILK expression (Figure 5.2C). In contrast, virally-infected wtILK overexpressing cell populations did not express P-casein (Fig. 5.2B), BM-dependent sphere formation was prevented, and the cells migrated into the gel (Figure 5.2C). 5.4.3 wtILK overexpression disrupts adherens junctions Scp2 mammary epithelial cells form classical cobblestone epithelial monolayers in 2D culture (see chapter 2; Desprez et al., 1993). In the current study, the same was true of the virally-infected vector controls and kdILK expressing cell populations, although a subtle cell rounding was observed in the latter population (Figure 5.3A). In contrast, virally-infected wtILK overexpressing cell populations were morphologically heterogeneous; some cells remained close-packed and epithelial but many others detached from each other and became bipolar and fibroblastic (Figure 5.3A). Scp2 mammary epithelial cells form E-cadherin-mediated AJs in both 2D monolayers and in 3D B M gel culture (see chapter 2; Lochter et al., 1997; Somasiri et al., 2000). The localization of E-cadherin and P-catenin to areas of cell-cell interaction in 2D monolayers 199 indicated that AJs junctions formed in the virally-infected vector controls and kdILK expressing cell populations (Figure 5.3A). In contrast, E-cadherin localization was discontinuous or the protein was completely absent from the majority of the cells in the virally-infected wtILK overexpressing population (Figure 5.3A). The latter observation was reflected by an overall decrease in steady state E-cadherin levels (Figure 5.3B), an observation that has been made previously in wtILK overexpressing intestinal epithelial cells (Wu et al., 1998). Junctional B-catenin localization was also disrupted in a significant number of the cells of the wtILK overexpressing population. In some of the latter cells B-catenin was observed in the nucleus (Figure 5.3A). Steady-state B-catenin levels were similar in all of the virally-infected populations (Figure 5.3B). 5.4.4 wtILK overexpression initiates an epithelial to mesenchymal transformation The disruption of AJs in the virally-infected wtILK overexpressing cell population was associated with changes to the actin cytoskeleton. In vector control and kdlLK-infected cells most of the filamentous (f-) actin was localized to cell-cell junctions. As such, the f-actin formed honeycomb patterns in 2D monolayers in both of these pooled cell populations (see chapter 2; Figure 5.4). In contrast, many of the fibroblastic cells in the wtILK overexpressing population contained prominent cytoplasmic actin stress fibers. The combination of a fibroblastic morphology, disruption of AJs and the acquisition of stress fibers suggested to me that some of the cells in this heterogeneous wtILK overexpressing population may have undergone an EMT. To address this possibility I examined the intermediate filament cytoskeleton. As expected (Desprez et al., 1993), the great majority of the virally-infected control cells contained epithelial cytokeratins and very few cells contained mesenchymal vimentin intermediate filaments (Figure 5.4). The same was true of virally-infected kdILK expressing population. In contrast, approximately half of the wtlLK-infected cells had lost cytokeratin filaments and a considerable number of the cells had acquired vimentin filaments. To determine i f wtILK overexpression truly leads to intermediate filament switching and mesenchymal transformation I also examined homogeneous populations of 200 wtILK overexpressing cells. Previously, I stably transfected scp2 cells with a wtILK expression vector and isolated clones in which the transgene is overexpressed to varying degrees (Novak et al., 1998). In this earlier study I demonstrated that increasing stable wtILK overexpression decreased steady state E-cadherin levels, caused nuclear translocation of P-catenin and stimulated invasion into basement membrane gels (see Table 5.1 for summary; Figure 5.5 and Figure 5.10 below). I now demonstrate that increasing wtILK overexpression in these homogeneous clones caused the cells to become fibroblastic in monolayer culture and it decreased cytokeratin intermediate filaments and increased vimentin intermediate filaments in each of these clones (Figure 5.6). The transfection control, ILK 14-1 is the reverse orientation of the full length ILK and it has not effect on the endogenous ILK expression (Novak et al., 1998). These control cells show similar phenotype and a similar cytokeratin expression pattern to the parent cells. In the highest wtILK overexpressing clone, ILK13-8, all of the cells contained only mesenchymal vimentin filaments. This wtlLK-induced intermediate filament protein switching was confirmed by western blotting (Figure 5.7A,B). In agreement with the findings described above for the heterogeneous viral wtILK overexpressing populations, increasing wtILK overexpression in the homogeneous transfected clones also inhibited P-casein induction (Figure 5.7C). Therefore, high levels of stable wtILK overexpression initiated a mesenchymal transformation of scp2 cells that was incompatible with BM-dependent morphogenesis and differentiation. 5.4.5 Subcellular ILK localization is altered in mesenchymally transformed cells At first glance it was somewhat surprising that wtILK overexpression prevented BM-dependent P-casein expression given the fact that differentiative contact with laminin upregulated endogenous ILK activity in the parental scp2 cells. However, subtle, but potentially significant, alterations in ILK subcellular localization may provide an explanation. Specifically, the ILK protein was often localized to elongated streaks at the cell-substratum interface in the mesenchymal (i.e. keratin negative) wtILK overexpressing cells (Figure 5.8). Morphologically these streaks, which resemble focal 201 adhesions, were very different from the discrete punctate spots of endogenous ILK staining observed in control populations. Therefore, ILK function may differ in the mesenchymal overexpressors and the epithelial parental cell line. These observations as well as the phenotype of the kdILK expressing cells suggest that, in addition to the kinase activity, the localization of ILK may also important for the transforming activity. Furthermore, it is possible that the kinase activity is required for this transformed phenotype. 5.4.6 Forced E-cadherin expression causes an epithelial reversion in mesenchymal wtlLK-overexpressing cells Steady state E-cadherin levels were reduced in the heterogeneous virally-infected wtILK overexpressing cell population (see Figure 5.2 above). In addition, the homogenous ILK13-8 clone did not contain any observable E-cadherin protein or mRNA (Figure 5.9A). Given the importance of E-cadherin in regulating mammary epithelial cell function (Gilles and Thompson, 1996), I next asked i f forced expression of this cell-cell adhesion molecule would rescue the epithelial phenotype in the mesenchymal ILK13-8 clone. As such, I stably transfected ILK13-8 cells with a mouse E-cadherin cDNA under the control of the constitutively active P-actin promoter (Nose et al., 1988) and selected a second set of doubly-transfected clones by limiting dilution. ILK13-8 cells sham-transfected with a second selection vector only were designated ILK13-8Vect. Like the original ILK13-8 cells these controls did not express E-cadherin and they did not undergo BM-dependent differentiation (Figure 5.9). Five doubly-transfected clones expressing increasing amounts of E-cadherin were designated ILK13-8EC-a to ILK13-8EC-e. Two of these clones (EC-a and EC-c) differentiated and expressed p-casein when cultured on B M gels (Figure 5.9). However, there was no correlation between the steady-state levels of residual wtILK overexpression or E-cadherin and the restoration of the differentiated phenotype. ILK13-8Vect control cells did not contain epithelial cytokeratin filaments and they remained invasive in 3D B M culture (Figure 5.10). In contrast, the majority of the cells in all of the ILK13-8EC clonal lines were cytokeratin positive and non-invasive. Interestingly, however, only 2 of the 5 clones underwent complete spherical 202 morphogenesis in 3D culture and these were the same two clones that expressed (3-casein under the same conditions (EC-a and EC-c). Therefore, while forced E-cadherin expression re-epithelialized the cells it only weakly restored differentiative and morphogenic potential. 5.8 Discussion The products of many oncogenes and tumor suppressor genes function by impinging upon the cell cycle (Hunter, 1997). However, changes in tissue architecture are often used to classify solid tumors in vivo and it is becoming increasingly clear that such changes can profoundly influence the function of oncogenic signaling molecules by altering their sub-cellular positioning and/or their interacting partners (Huang and Ingber, 1999). In large part, tissue architecture is regulated by adhesion molecules that mediate the interactions between cells and the interactions between cells and the E C M (also see Chapter 1; Gumbiner, 1996). Mutation changes in expression or protein localization, and functional alteration of these molecules has been observed in many cancers. Thus, adhesion molecules are being increasingly viewed as architectural oncogenes and tumor suppressors (Boudreau and Bissell, 1998). In carcinomas, which are the most prevalent of all human solid tumors, an inappropriate epithelial to mesenchymal transformation (EMT) is a key architectural event that contributes to invasion and metastasis (Birchmeier et al., 1996). In breast carcinomas this inappropriate E M T is associated with alterations in both cell-cell and cell-ECM interactions (Gilles and Thompson, 1996). Experimental evidence suggests that these alterations are functionally interrelated. For example, forced expression of an activated form of the metalloprotease stromelysin-1 in normal mammary epithelial cells degrades the E C M as well as the extracellular domain of E-cadherin, and both changes contribute to mesenchymal transformation (Lochter et al., 1997). In addition, a phenotypic switching from the anchoring a6(34 integrin to migratory (31 integrins disrupts junctional E-cadherin localization and epithelial architecture during breast tumor cell progression (Weaver et al., 1997). The latter also known to constitutively upregulate endogenous ILK kinase activity (Wang et al., 1998). In this report, I demonstrate that the 203 overexpression of wtILK caused functional mouse mammary epithelial scp2 cells to become non-differentiative, mesenchymal and invasive giving a phenotype quite similar to that of ILC. When they are placed upon a B M gel, scp2 cells undergo cell-cell junction-dependent spheroidal morphogenesis (chapter2; Lochter et al., 1997; Somasiri et al., 2000). Under these conditions the cells also express the milk protein B-casein, a differentiative event that is regulated, at least in part, by interactions between laminin in the B M and BI integrins on the cell surface (Streuli et al., 1991; Streuli et al., 1995; Muschler et al., 1999). ILK interacts with the cytoplasmic tail of BI integrin subunits and the kinase activity of the molecule is activated when intestinal epithelial cells adhere to E C M in a BI integrin-dependent manner (Hannigan et al., 1996). Thus, it was not unexpected that adhesion of mammary epithelial cells to a B M also increased I L K activity. However, this adhesion also decreased the amount of ILK that could be immunoprecipitated from NP-40 cell lysates. The latter result may have occurred because of a steady-state decrease in the ILK protein, or it may indicate BM-mediated changes in cytoplasmic ILK solubility, perhaps due to changes in association with the cytoskeleton. If the latter were the case it would suggest that the ILK not associated with the cytoskeleton (i.e. the NP-40 soluble pool) has a significantly elevated kinase activity. Regardless, the finding that both the B M and purified laminin altered ILK kinase activity led me to examine the role of ILK in scp2 cell morphogenesis and differentiation. Forced expression of a kdILK mutant did not prevent BM-dependent B-casein expression or the formation of spheres on B M gels. However, the latter structures were considerably smaller than those formed by the control cells. While it is possible that cell-cell aggregation, which is a pre-requisite for spheroidal morphogenesis, was decreased in the kdILK expressing cells. The parental scp2 cells do not form functional TJ in monolayer culture (Woo et al., 2000; see chapter 2) but these junctions do form in an apical location in fully polarized spheres on B M gels. ILK plays a role in anchorage-dependent growth and survival (Radeva et al., 1997) and wtILK overexpression in scp2 cells prevents anoikic apoptosis (Attwell et al., 2000). Therefore, it is also possible that kdILK expression reduced sphere size by impinging upon on the B M ' s ability to protect mammary epithelial cells from undergoing apoptosis (Boudreau et al., 1995). If this is 204 indeed the case, it will be important to determine the relative contribution of both integrins and growth factor receptors as ILK also augments insulin-mediated signaling (Delcomenne et al., 1998; Tu et al., 1998). Interestingly, the latter pathway has been shown to inhibit mammary epithelial cell apoptosis in a laminin- and p i integrin-dependent manner (Farrelly et al., 1999). Forced overexpression of wtILK did not cause scp2 cells to express P-casein in the absence of B M matrix. On the contrary, stable wt ILK overexpression caused scp2 cells to become completely non-differentiative, even in the presence of B M . These findings do not completely rule out the possibility that some degree of ILK-mediated signaling may be compatible with mammary morphogenesis and differentiation. For example, differences in the kinetics of activation between matrix-stimulated parental cells and chronically wtILK overexpressing cells could be responsible for the differences in the phenotypic endpoints observed. This is certainly the case for the erb-B2 receptor where a precisely controlled developmental activation of this tyrosine kinase contributes to late-stage mammary alveolar morphogenesis and differentiation (Yang et al., 1995) while chronic stimulation causes a loss of differentiation and mammary tumor formation (Muller et al., 1988). In a similar manner, I L K could act to either promote or inhibit the differentiation of the mammary epithelium in a developmental stage-specfic manner, as is the case during myogenic differentiation (Huang et al., 2000) and mouse epiblast polarization (Sakai et al. 2003). Non-differentiative wtILK overexpressing scp2 cells lost cytokeratins, became fibroblastic and they invaded B M gels. These fibroblastic cells also acquired vimentin, cytoplasmic stress fibers and they no longer expressed E-cadherin. Thus, stable overexpression of wtILK caused a complete EMT in scp2 cells. When E-cadherin is force-expressed in true fibroblasts, cell-cell junctions partially form and the production of fibronectin, a mesenchymal E C M protein, is downregulated (Nagafuchi et al., 1987; Finneman et al., 1995; Yonemura et al., 1995). Forced E-cadherin expression also causes an epithelialization of metastatic breast carcinoma cells (Meiners et al., 1998) and of invasive, mesenchyme-like ovarian surface epithelial cells (Auersperg et al., 1999). Taken together, these data led me to tentatively conclude that the loss of E-cadherin might be a critical component of the wtlLK-induced E M T in scp2 cells. I directly tested this hypothesis by force-expressing E-cadherin in the completely mesenchymal wt lLK-205 overexpressing clone ILK13-8. This caused the reappearance of cytokeratin and it prevented cellular invasion of the B M in all of the clones examined. Therefore, E-cadherin appeared to restore a modicum of the epithelial phenotype. In contrast, only two of five of these clones underwent BM-dependent differentiation. There was no correlation between the levels of wtILK overexpression or E-cadherin expression and this differentiative rescue. However, those clones that did differentiate underwent the most complete spheroidal morphogenesis. Therefore, a combined morphological and differentiative rescue by E-cadherin may rely upon the complete restoration of adhesive and TJs between the cells. wtILK overexpression promotes the phosphorylation and activation of protein kinase B (PKB) which then phosphorylates glycogen synthase kinase-3 (GSK-3) and decreases its activity (Lynch et al., 1999; Persad et al., 2000). During development, wnt/wingless signaling also inhibits GSK-3 activity, which prevents it from phosphorylating B-catenin. As a result, free p-catenin escapes degradation and can be transported to the nucleus where it interacts with the Tcf/LEF-1 family of architectural transcription factors to regulate the expression of genes involved in regulating the cell cycle and developmental mesenchymal transformation (McCartney and Peifer, 2000). While it has been suggested that PKB-mediated decreases in GSK-3 activity alone, in the absence of wnt/wingless signaling, are not be sufficient to translocate P-catenin to the nucleus (Yuan et al., 1999), I did observe nuclear P-catenin in wtILK overexpressing scp2 mammary epithelial cells and the same occurs in wtILK overexpressing intestinal epithelial cells (Novak et al., 1998). Therefore, one-way that ILK overexpression could induce an E M T in mammary epithelial is by mediating its ability to upregulate nuclear P-catenin signaling. P-Catenin/LEF-1-mediated transactivation can be reversed by the forced expression of cadherins or cadherin cytoplasmic domains (Fagotto et al., 1996; Miller and Moon, 1997; Sadot et al., 1998, Orsulic et al., 1999). Presumably this effect is mediated by the binding and sequestering of P-catenin. One of the candidate targets of ILK-mediated P-catenin signaling that could initiate the mesenchymal transformation candidate is Snail, a transcription factor that acts to repress E-cadherin expression by binding to the E-box in the promoter of the gene (Batlle et al., 2000; Cano et al., 2000). Furthermore, when ILK is overexpressed in colon carcinoma cells, Snail expression is upergulated (Tan et al, 2001) and thereby downregulating E-cadherin expression. Thus, it 206 is possible that the forced expression of E-cadherin in the mesenchymal wtILK overexpressing scp2 cells caused a re-epithelialization by reversing p-catenin signaling as well as overriding the effects of Snail repression. The role of ILK in human cancer has yet to be definitively established. However, endogenous ILK activity is increased in breast tumor cells (Wang et al., 1998) and ILK is ovexpressed in 100% of Ewing's sarcomas so far examined (Chung et al., 1998). Furthermore, increased levels of ILK are observed in prostate tumorgenesis (Graff et al., 2001) as well as in highly invasive gastric carcinomas (Ito et al., 2002). Experimentally, targeted wtILK overexpression in the mammary glands of transgenic mice leads to initial hyperplasia which progresses into invasive tumors (White et al., 2001) and steady-state ILK levels are upregulated by the overexpression of the oncogene erbB2 in the suprabasal layers of mouse skin (Xie et al., 1998). A l l these observations and the data presented in this thesis suggest that, overexpression of ILK can cause the complete loss of cell junctions and cell polarity leading towards an EMT. Hence, ILK-transformed phenotype is similar to the transition from LCIS to ILC where carcinoma cells break away from the primary tumor and invade the surrounding stroma as single cells. 207 Table 5.1 ILK overexpression Cell line wtILK Overexpression P-Catenin localization Scp2 (parent) Endogenous Plasma membrane ILK14-1 (Control) None Plasma membrane ILK13-4 Low Plasma membrane ILK13-6 Moderate Discontinuous at plasmamembrane and cytoplasm ILK13-8 High Cytoplasmic and nuclear scp2 mouse mammary epithelial cells were stably transfected with a mammalian expression vector containing either a control antisense ILK cDNA (ILK 14) or a wild-type ILK sense cDNA (ILK13) and clones were isolated by limited dilution. The levels of ILK overexpression were assessed by western blotting and P-catenin localization was determined by immunofluorescence (see Figure 5.5). 208 Figure 5.1 Endogenous ILK activity is upregulated by cellular interactions with the basement membrane. Scp2 mammary epithelial cells were maintained on tissue culture plastic (PL) or they were overlaid with purified laminin (LN) or placed on basement membrane gels (BM) for 5 days (see Methods for details). Cells were then lysed, ILK was immunoprecipitated and kinase activity was determined using myelin basic protein as the substrate. ILK protein levels in the immunoprecipitates were determined by western blotting as was differentiative P-casein milk protein expression in whole cell lysates. 209 P L L N B M 210 Figure 5.2 wtILK overexpression in viral infectants prevents basement membrane-dependent differentiation. Scp2 cells were infected with vector control, kdILK or wtlLK-containing retroviruses. In A , expression of the transgene was assessed by western blotting for the V5 epitope-tagged ILK fusion protein (-60 kDa), either before (-) or after (+) genetic selection with puromycin. In B, genetically-selected pooled heterogeneous cell populations were maintained on basement membrane gels for 5 days and morphology was assessed by phase microscopy. Bar, 50 mm. In C, pooled cell populations were maintained on either tissue plastic (-) or a basement membrane gel (+) for 5 days and differentiative induction of P-casein (P-Cas) expression was assessed by western blotting. 211 212 Figure 5.3 wtILK overexpression in viral infectants disrupts cell-cell junctions. Pooled heterogeneous cell populations were maintained as 2-dimensional monolayers on tissue culture plastic. In A , morphology was assessed by phase microscopy and E-cadherin and P-catenin localization was visualized by immunofluorescence microscopy; arrows indicate the position of E-cadherin negative wtlLK-infected cells (bar, 25 mm phase; 15 mm immunofluorescence). In B, steady state levels of E-cadherin (E-Cad) and p-catenin (P-Cat) in whole cell lysates were determined by western blotting. 213 B r - 1 E-Cad Vect kdILK wtILK B-Cat 214 Figure 5.4 wtILK overexpression in viral infectants alters the cytoskeleton. Pooled heterogeneous cell populations were maintained as 2D monolayers on tissue culture plastic, f-actin was then visualized after rhodamine phalloidin staining while epithelial cytokeratin and mesenchymal vimentin intermediate filaments were visualized by immunofluorescence microscopy (different fields for each staining; bar, 15 um). 215 ^ Vector kd I L K ^ ^ ^ wtlU< f-actin m Keratin > " / o ; X Vimentin 216 Figure 5.5 ILK overexpression in scp2 cells leads to loss of E-cadherin and increased nuclear P-catenin activity. (A) ILK, E-cadherin, P-catenin, and LEF-1 expression in scp2. Nonidet P-40 cell lysates (10 pg for ILK and LEF-1; 20 pg for E-cadherin and P-catenin) were analyzed by Western blotting. Clone 14-1: control transfectants, transfected with antisense ILK cDNA; Clones 13-4, 13-6, and 13-8: transfected with ILK-sense cDNA. 13-8 clone had the highest level of ILK expression. 217 I L K E - C A D l i - C A T L E F - 1 2I8 Figure 5.6 Increasing overexpression of wtILK in stable transfectants induces an epithelial to mesenchymal transformation. Scp2 parental mammary epithelial cells, ILK14-1 control transfectants and ILK13-4, ILK13-6, and ILK13-8 overexpressing transfectant clones were maintained as 2D monolayers. wtILK overexpression in the ILK13 transfectants increases as indicated (see Table 1 and Novak et al., 1998). Morphology was assessed by phase microscopy while epithelial cytokeratin and mesenchymal vimentin intermediate intermediate filaments were visualized by immunofluorescence microscopy (different fields for phase and each staining; bar, 20 mm). 219 scp2 ILK14-1 ILK13-4 ILK13-6 ILK13-8 220 Figure 5.7 Increasing overexpression of wtILK in stable transfectants causes a loss of cytokeratin proteins, an increase in vimentin protein, and a decreased ability to undergo basement membrane-dependent differentiation. Cytoskeletal preparations of stable transfectants, obtained by high salt extraction, were probed by western blotting. In A , cytokeratins were assessed using a mixture of AE1 and AE3 monoclonal antibodies, which recognize all cytokeratins except cytokeratin 18. Three cytokeratins were identified which, given their molecular masses, can be tentatively characterized as: #1=CK5; #2=CK8, #3=CK14. In B, vinientin was assessed using an anti-vimentin monoclonal antibody. In C, the cells were maintained for the indicated time on basement membrane gels, lysed and differentiative P-casein induction was assessed (lanes: 1, scp2 parental; 2, ILK14-1 control; 3, ILK13-4 low overexpressor; 4, ILK13-6 medium overexpressor; 5, ILK13-8 high overexpressor). 221 1 2 3 4 5 —CK #1 CK#2 - C K # 3 - V i m C DayO Day1 Day2 Day3 Day4 Day5 1 2 3 4 5 wtLK Overexpressor 222 Figure 5.8 ILK localization is altered in wtILK overexpressing, mesenchymally transformed cells. Virally-infected, heterogeneous cell populations (a, vector control, b, wt ILK) and homogenous stable transfectants (c, ILK-14-1 control; d, ILK13-8 overexpressor) were maintained as 2D monolayers. Virally-infected cells were co-immunostained for epithelial cytokeratin (red) and ILK (green) and epifluorescent photomicrographs were taken at the plane of the cell/substratum interface. ILK staining (yellow) was confined to small, discrete circular dots in the vector control cells (a) and in those wtILK infected cells that remained epithelial (i.e. contained cytokeratin, b). In contrast, a significant amount of the ILK protein (green) localized to elongated streaks in the keratin-negative (i.e. mesenchymal) cells (b). Stable transfectants were stained for ILK only (white). ILK was localized in small punctate spots in the epithelial ILK14-1 control cells (c) and in larger elongated streaks in the mesenchymal ILK13-8 overexpressing cells (d), none of which contain cytokeratin (see Figs 5 and 6 above). Bar, 8 mm (a,b); 5 mm (c,d). 223 IK- ' l 5 r - • ^VflB ^^^^^^^^^ b 7-^  1 c 224 Figure 5.9 w t I L K overexpression downregulates E-cadherin. In A , the steady state levels of ILK and E-cadherin protein as well as E-cadherin (E-Cad) mRNA in epithelial scp2 cells (lane 1) and mesenchymal wtILK13-8 stable transfectants (lane 2) were assessed by western blotting and RT-PCR, respectively. For the PCR, quality of the cDNA was assessed by amplifying f-actin. In B, mesenchymal ILK13-8 cells were transfected with a either a second selection vector alone (Vect) or with an E-cadherin c-D N A (13-8EC). After genetic selection, the vector control (Vect) and five E-cadherin transfected clones (13-8ECa-e) were assessed for steady state ILK and E-Cad protein and the ability to express B-casein (B-cas) in basement membrane culture by western blotting of whole cell lysates. 225 226 Figure 5.10 Forced E-cadherin expression rescues the epithelial phenotype in wtILK overexpressing cells. Mesenchymal ILK13-8 cells transfected with vector alone (Vect) and the five E-cadherin expressing clones (ECa-e) were maintained as 2-D monolayers and E-cadherin and cytokeratin localization were assessed by immunofluorescence microscopy. The cells were also maintained in 3D B M culture and the ability of the cells to undergo spheroidal morphogenesis was assessed by phase microscopy. Bar, 15 mm immunofluorescence; 50 mm phase. 227 E-Cadherin Keratin Morphogenesis 228 5.9 References Achtstaetter, T., Hatzfeld, M . , Quinlan, R. A. , Parmell, D. C. and Franke, W. W. (1986). Separation of cytokeratin polypeptides by gel electrophoretic and chromatographic techniques and their identification by immunoblotting. Meth. Enzymol. 134, 355-371. Adams, J. C. and Watt, F. C. (1993). Regulation of development and differentiation by the extracellular matrix. Development 117, 1183-1198. Aggeler, J., Ward J., Blackie L. M . , Barcellos-Hoff, M . H. , Streuli, C. H. and Bissell, M . J. (1991). Cytodifferentiation of mouse mammary epithelial cells cultured on a reconstituted basement membrane reveals striking similarities to development in vivo. J. Cell Sci. 99, 407-417. Attwell, S., Roskelley, C. D. and Dedhar, S. (2000). The integrin-linked kinase (ILK) inhibits anoikis. Oncogene 19, 3811-3815. Auersperg, N . , Pan, J., Grove, B. D., Peterson, T., Fisher, J., Maines-Bandiera, S., Somasiri, A . and Roskelley, C. D. (1999). E-cadherin induces mesenchymal to epithelial transition in human ovarian surface epithelial cells. Proc. Nat. Acad. Sci. U S A 96, 6249-6254. Barcellos-Hoff, M . H. , Aggeler J., Ram, T. G. and Bissell, M . J. (1989). Functional differentiation and alveolar morphogenesis of primary mammary cultures on reconstituted basement membrane. Development 105, 223-242. Batlle E., Sancho, E., Franci, C , Dominguez, D., Monfar, M . , Baulida, J. and Garcia De Herreros, A . (2000). The transcription factor snail is a repressor of E-cadherin gene expression in epithelial tumour cells. Nat. Cell Biol. 2, 84-89. 229 Birchmeier, C , Bierchmeier, W. and Brand-Saberi, B. (1996). Epithelial-mesenchymal transitions in cancer progression. Acta Anatomica 156, 217-226. Bissell, M . J., Hall, H. G. and Parry, G. (1982). How does extracellular matrix direct gene expression? J. Theor. Biol. 99, 31-68. Boudreau, N . , Sympson, C. J., Werb, Z. and Bissell, M . J. (1995). Suppression of ICE and apoptosis in mammary epithelial cells by extracellular matrix. Science 267, 891-893. Boudreau, N . and Bissell M . J. (1998). Extracellular matrix signaling: integration of form and function in normal and malignant cells. Curr. Opin. Cell Biol. 10, 640-646. Cano, A. , Perez-Moreno, M A , Rodrigo, I., Locascio, A. , Blanco, M . J., del Barrio, M . G., Portillo, F and Nieto, M . A . (2000). The transcription factor snail controls epithelial-mesenchymal transitions by repressing E-cadherin expression. Nat. Cell Biol. 2, 76-83. Chung, D. H. , Lee, J. I., Kook, M . C , Kim, J. R., Kim, S. H. , Choi, E. Y . , Park, S. H. and Song, H . G. (1998). ILK: a novel immunohistochemical marker for Ewing's sarcoma and primitive neuroectodermal tumour. Virchows Arch. 433, 113-117. Delcommenne, M . , Tan, C , Gray, V. , Rue, L. , Woodget, J. and Dedhar, S. (1998). Phosphoinositide-3-OH kinase-dependent regulation of glycogen synthase kinase 3 and protein kinase B / A K T by the integrin-linked kinase. Proc. Nat. Acad. Sci. USA 95, 11211-11216. Desprez, P.-Y., Roskelley, C , Campisi, J. and Bissell, M . J. (1993). Isolation of functional cell lines from a mouse mammary epithelial cell strain: the importance of basement membrane and cell-cell interactions. Mol . Cell Differ. 1, 99-110. 230 Fagotto, F., Funayama, N . , Gluck, U . and Gumbiner, B. M . (1996). Binding to cadherins antagonizes the signaling activity of b-catenin during axis formation in Xenopus. J. Cell Biol. 132, 1105-1114. Farrelly, N . , Lee, Y . J., Oliver, J., Dive, C. and Streuli, C. H. (1999). Extracellullar matrix regulates apoptosis in mammary epithelium through control of insulin signaling. J. Cell Biol. 144, 1337-1348. Finnemann, S., Kuhl, M . , Otto, G. and Wedlich, D. (1995). Cadherin transfection of Xenopus X T C cells downregulates expression of substrate adhesion molecules. Mol . Cell. Biol. 15, 5082-5091. Giancotti, F. G. and Rouslahti, E. (1999). Integrin signaling. Science 285, 1028-1032. Gilles, C. and Thompson, E. (1996). The epithelial to mesenchymal transition and metastatic progression in breast carcinoma. Breast J. 2, 83-96. Graff JR, Deddens JA, Konicek BW, Colligan B M , Hurst B M , Carter H W and Carter JH. (2001) Integrin-linked kinase expression increases with prostate tumor grade. Clin Cancer Res. 7(7), 1987-91. Gumbiner, B. M . (1996). Cell adhesion: the molecular basis of tissue architecture and morphogenesis. Cell 84, 345-357. Hannigan, G. E., Leung-Hagesteijn, C , Fitz-Gibbon, L., Coppolino, M . G., Radeva, G., Filmus, J., Bell, J. C. and Dedhar, S. (1996). Regulation of cell adhesion and anchorage-dependent growth by a new bl-integrin-linked protein kinase. Nature 379, 91-96. Hobert, O., Moerman, D. G., Clark, K. A. , Beckerle, M . C. and Ruvkun, G. (1999). A conserved L I M protein that affects muscular adherens junction integrity and mechanosensory function in C. elegans. J. Cell Biol. 144, 45-57. 231 Huang, S. and Ingber, D. E. (1999). The structural and mechanical complexity of cell-growth control. Nat. Cell Biol. 1, E131-138. Huang, Y . , L i , J., Zhang, Y . , Wu, C. (2000). The roles of integrin-linked kinase in the regulation of myogenic differentiation. J. Cell Biol. 150, 861-872. Hunter, T. (1997). Oncoprotein networks. Cell 88, 333-346. Hynes, R. O. (1992). Integrins: versatility, modulation, and signaling in cell adhesion. Cell 69, 11-25. Ito R, Oue N , Zhu X , Yoshida K , Nakayama H, Yokozaki H and Yasui W. (2003) Expression of integrin-linked kinase is closely correlated with invasion and metastasis of gastric carcinoma. Virchows Arch. 442(2), 118-23. Juliano, R. L. and Haskill, S. (1993). Signal transduction from the extracellular matrix. J. Cell Biol. 120,577-585. Kaetzel, C. S. and Ray, D. B. (1984). Immunochemical characterization with monoclonal antibodies of three major caseins and a-lactalbumin from rat milk. J. Dairy Sci. 67, 64-75. Khwaja, A. , Rodriquez-Viciana, P., Wennstrom, S., Warne, P. H . and Downward, J. (1997). Matrix adhesion and ras transformation both activate a phosphoinositide 3-OH kinase and protein kinase B/Akt cellular survival pathway. E M B O J. 16, 2783-2793. King, W. G., Mattaliano, M . D., Chan, T. O., Tsichlis, P. N . and Brugge, J. S. (1997). PI3-kinase is required for integrin-stimulated A K T and Raf-1/mitogen-activated protein kinase pathway activation. Mol . Cell. Biol. 17, 4406-4416. 232 L i , F., Zhang, Y . and Wu, C. (1999). Integrin-linked kinase is localized to cell-matrix focal adhesions but not cell-cell adhesion sites and focal adhesion localization of integrin-linked kinase is regulated by the PINCH-binding A N K repeats. J. Cell Sci. 112, 4589-4599. Lochter, A. , Galosy, S., Muschler, J., Freedman, N . , Werb, Z. and Bissell, M . J. (1997). Matrix metalloproteinase stromelysin-1 triggers a cascade of molecular aterations that leads to a stable epithelial-to-mesenchymal conversion and a premalignant phenotype in mammary epithelial cells. J. Cell Biol. 139, 1861-1872. Lynch, D. K. , Ellis, C. A. , Edwards, P. A . W. and Hiles, I. D. (1999). Integrin-linked kinase regulates phosphorylation of serine 473 of protein kinase B by an indirect mechanism. Oncogene 18, 8024-8032. McCartney, B. M . and Peifer, M . (2000). Teaching tumour suppressors new tricks. Nat. Cell Biol. 2, E-58-E60. Meiners, S., Brinkmann, V. , Naundorf, H. and W. Birchmeier, W. (1998). Role of morphogenetic factors in metastasis of mammary carcinoma cells. Oncogene 16, 9-20. Miller, J. R. and Moon R. T. (1997). Analysis of the signaling activities of localization mutants of b-catenin during axis specification in Xenopus. J. Cell Biol. 139, 229-243. Muller, W. J., Sinn, E., pattengale, P. K. , Wallace, R. and Leder, P. (1988). Single-step induction of mammary adenocarcinoma in transgenic mice bearing the activated c-neu oncogene. Cell 54, 105-115. Muschler, J., Lochter, A . , Roskelley, C. D., Yurchenco P. and Bissell, M . J. (1999). Division of labour among integrins and an E3 laminin receptor signal morphogenesis and b-casein expression in mammary epithelial cells. Mol . Biol. Cell 10, 2817-2828. 233 Nagafuchi, A. , Sirayoshi, Y . , Okazaki, K. , Yasuda, K . and Takeichi, M . (1987). Transformation of cell adhesion properties by exogenously introduced E-cadherin. Nature 329, 340-343. Nguyen, D. D. and Neville, M . C. (1998). Tight junction regulation in the mammary gland J. Mammary Gland Biology and Neoplasia 3, 233-246. Nose, A. , Nagafuchi, A . and M . Takeichi, M . (1988). Expressed recombinant cadherins mediate cell sorting in model systems. Cell 54, 993-1001. Novak, A. , Hsu, S., Leung-Hagesteijn, C , Radeva, G., Papkoff, J., Montessano, R., Roskelley, C , Grosschedl, R. and Dedhar, S. (1998). Cell adhesion and the integrin-linked kinase regulate the LEF-1 and b-catenin signaling pathways. Proc. Nat. Acad. Sci. USA 95, 4374-4379. Orsulic, S., Huber, O., Aberle, H. , Arnold, S. and Kemler, R. (1999). E-cadherin binding prevents b-catenin nuclear localization and b-catenin/LEF-1-mediated transactivation. J. Cell Sci. 112, 1237-1245. Pear, W. S., Nolan, G. P., Scott, M . L. and Baltimore, D. (1993). Production of high-titer helper-free retroviruses by transient transfection. Proc. Nat. Acad. Sci. USA 90, 8392-8396. Persad, S., Attwell, S., Gray, V. , Delcommenne, M . , Troussard, A. , Sanghera, J. and Dedhar, S. (2000). Inhibition of ILK suppresses activation of protein kinase B/Akt and induces cell cycle arrest and apoptosis of PTEN-mutant prostate cancer cells. Proc. Nat. Acad. Sci USA 97, 3207-3212. Pullan, S., Wilson, J., Metcalfe, A. , Edwards, G. M . , Goberdhan, N . , Tilly, J., Hickman, J. A , Dive, C. and Streuli, C. H . (1996). Requirement of basement membrane for the suppression of programmed cell death in mammary epithelium. J. Cell Sci. 109, 631-642. 234 Radeva, G., Petrocelli, T., Behtrend, E., Leung-Hagesteijn, C , Filmus, J., Slingerland, J. and Dedhar, S. (1997). Overexpression of the integrin-linked kinase promotes anchorage-independent cell cycle progression. J. Biol. Chem. 272, 13937-13944. Risinger, J. I., Berchuck, A. , Kohler, M . F. and Boyd, J. (1995). Mutation of the E-cadherin gene in human gynecologic cancers. Nat. Gen. 7, 98-102. Roskelley, C. D., Desprez, P.-Y. and Bissell, M . J. (1994). Extracellular matrix-dependent tissue-specific gene expression in mammary epithelial cells requires both physical and biochemical signal transduction. Proc. Nat. Acad. Sci. USA 91, 12378-12382. Roskelley, C. D. and Bissell, M . J. (1995). Dynamic reciprocity revisited: A continuous, bidirectional flow of information between cells and the extracellular matrix regulates mammary epithelial cell function. Biochem. Cell Biol. 73, 391-397. Sadot, E., Simcha, I., Shutman, M . , Ben-Ze'ev, A . and Geiger, B. (1998). Inhibition of b-catenin-mediated transactivation by cadherin derivatives. Proc. Nat. Acad. Sci. U S A 95, 15339-15344. Sakai T, L i S, Docheva D, Grashoff C, Sakai K, Kostka G, Braun A , Pfeifer A , Yurchenco PD and Fassler R. (2003) Integrin-linked kinase (ILK) is required for polarizing the epiblast, cell adhesion, and controlling actin accumulation. Genes Dev. 17(7), 926-40. Schmeichel, K . L. , Weaver, V . M . and Bissell M . J. (1998). Structural cues from the tissue microenvironment are the major determinants of the human mammary epithelial cell phenotype. J. Mammary Gland Biol. 3, 201-213. 235 Schmidhauser, C , Bissell, M . J., Myers, C. A . and Casperson, G. F. (1990). Extracellular matrix and hormones transcriptionally regulate bovine b-casein 5' sequences in stably transfected mouse mammary cells. Proc. Nat. Acad. Sci. USA 87, 9118-9122. Somasiri, A . and Roskelley, C. D. (1999). Cell shape and integrin signaling regulate the differentiation state of mammary epithelial cells. Meth. Mol . Biol. 129, 271-284. Somasiri, A . M . , Wu, C , Ellchuk, T., Turley, S. and Roskelley, C. D. (2000). Phosphoinositide 3-OH kinase is required for adherens junction-dependent mammary epithelial cell spheroid formation. Differentiation 66, 116-125. Streuli, C. H. , Bailey, N . and Bissell, M . J. (1991). Control of mammary epithelial differentiation: the separate roles of cell-substratum and cell-cell interactions. J. Cell Biol. 115, 1383-1395. Streuli, C. H. , Schmidhauser, C , Bailey, N . , Yurchenco, P., Skubitz, A . N . , Roskelley, C. D. and Bissell M . J. (1995). Laminin mediates tissue-specific gene expression in mammary epithelia. J. Cell Biol. 129, 591-603. Sun, T. T., Tseng, S. C , Huang, A . J., Cooper, D., Schermer, A. , Lynch, M . H. , Weiss, R. and Eichner, R. (1985). Monoclonal antibody studies of mammalian epithelial keratins: a review. Ann. N Y Acad. Sci. USA 455, 307-329. Tan, C , Costello, P., Sanghera, J., Domiguez, D., Baulida, J., Garcia de Herreros, A . and Dedhar, S. (2000). Inhibition of integrin linked kinase (ILK), suppresses P-catenin-Lef/Tcf-dependent transcription and expression of the E-cadherin repressor, Snail, in A P C -/-human colon carcinoma cells. Oncogene 20(1), 133-140. Troussard, A . A , Tan, C , Youganathan, T. N . and Dedhar, S. (1999). Cell-extracellular matrix interactions stimulate the AP-1 transcription factor in an integrin-linked kinase-and glycogen synthase kinase 3-dependent manner. Mol . Cell. Biol. 19, 7420-7427. 236 Tu, Y . , L i , F. and Wu, C. (1998). Nck-2 a novel src homology 2/3-containing adaptor protein that interacts with the LIM-only protein PINCH and components of growth factor receptor kinase-signaling pathways. Mol . Biol. Cell 9, 3367-3382. Wang, F., Weaver, V . M . , Petersen, O. W., Larabell, C. A. , Dedhar, S., Briand, P., Lupu, R. and Bissell, M . J. (1998). Reciprocal interactions between b l integrin and epidermal growth factor receptor in 3-dimensional basement membrane cultures: A perspective in normal and malignant breast epithelial biology. Proc. Nat. Acad. Sci. U S A 95, 14821-14826. Wartmann, M . , Cella, N . , Hofer, P., Groner, B., Liu, X . , Henninghausen, L. and Hynes, N . E. (1996). Lactogenic hormone activation of Stat5 and transcription of the b-casein gene in mammary epithelial cells is independent of p42 ERK2 mitogen-activated protein kinase activity. J. Biol. Chem. 271, 31863-31868. Weaver, V . M . , Petersen, O. W., Wang, F., Larabell, C. A. , Briand, P., Damsky, C. and Bissell, M . J. (1997). Reversion of the malignant phenotype of human breast cancer cells in three-dimensional culture and in vivo by integrin blocking antibodies. J. Cell Biol. 137, 231-245. White DE, Cardiff RD, Dedhar S and Muller WJ. (2001) Mammary epithelial-specific expression of the integrin-linked kinase (ILK) results in the induction of mammary gland hyperplasias and tumors in transgenic mice. Oncogene. 20(48), 7064-72. Woo, P. L., Cercek, A. , Desprez, P.-Y. and Firestone, G. L. (2000). Involvement of the helix-loop-helix protein Id-1 in the glucocorticoid regulation of tight junctions in mammary epithelial cells. J. Biol. Chem. 275, 28649-28658. Wu, C , Keightley, S. Y . , Leung-Hagesteijn, C , Radewa, G., Coppolino, M . , Goicoechea, S, McDonald, J. A and Dedhar, S. (1998). Integrin-linked protein kinase 237 regulates fibronectin matrix assembly, E-cadherin expression, and tumorigenicity. J. Biol. Chem. 273, 528-536. Xie, W., L i , F., Kudlow, J. E. and Wu, C. (1998). Expression of ILK in mouse skin: loss of expression in suprabasal layers of the epidermis and up-regulation by erbB-2. Am. J. Pathol. 153, 367-372. Yamada, K. M . and Geiger, B. (1997). Molecular interactions in adhesion complexes. Curr. Opin. Cell Biol. 9, 76-85. Yang, Y . , Spitzer, E., Meyer, D., Sachs, M . , Niemann, C , Hartmann, G., Weidner, K. M . , Birchmeier, C. and Birchmeier, W. (1995). Sequential requirement of hepatocyte growth factor and neuregulin in the morphogenesis and differentiation of the mammary gland. J. Cell Biol. 131,215-226. Yonemura, S., Itoh, M . , Nagafuchi, A . and Tsukita, S. (1995). Cell-to-cell adherens junction formation and actin filament organization: similarities and differences between non-polarized fibroblasts and polarized epithelial cells. J. Cell Sci. 108, 127-142. Yuan, FL, Mao, J., L i , L and Wu, D. (1999). Suppression of G S K activity is not sufficient for LEF-1 activation. J. Biol. Chem. 274, 30419-30423. 238 CHAPTER 6: General Discussion and Future Directions. Disruption of the epithelial architecture is always observed during histopathological examination of early stages of breast cancer progression. Thus, loss of, or changes in, the proteins that are important in maintaining this normal tissue architecture can play a prominent role during cancer progression. It is also clear that maintenance of epithelial apical/basal polarity is an integral part of the normal architecture. Therefore, a better understanding of how epithelial cells are apically/basally polarized during mammary gland development will give us clues on how the polarity is disrupted during disease progression. From a cell biological point of view, one of the fundamental questions is how the epithelium is polarized. The work presented in this thesis contributes to our understanding on how TJ are polarized in mammary epithelium and how disruption of this polarity may contribute to the progression of invasive breast carcinoma (Figure 6.1). Cell-cell junctions and cell-BM interactions provide structural and spatial cues to establish and maintain apical/basal polarity in the normal epithelium. In the work presented here, I have investigated possible mechanisms of mammary epithelial TJ formation and polarization using a mouse mammary epithelial culture model. The significance of this model is the ability to examine both TJ formation and polarization, independent of each other, in a developmental context. Initial studies have shown that when differentiated mammary alveolar epithelial cells from mid pregnant mice are enzymatically dissociated and cultured on tissue culture plastic, they lose their alveolar morphology and dedifferentiate to form confluent monolayers of polygonal shaped cells (Emerman and Pitelka, 1977). When these cells are placed on attached, thin collagen gels, they continue to form continuous, flat monolayers that are similar to monolayers that form on rigid tissue culture plastic substrata. However, when the same collagen gels are rendered flexible by floating them in media, the gel contract and cells begin to roundup. With time, these cells on the floating gels become columnar in shape and their nuclei become basally located. At the ultrastructural level, apical microvilli as well as apical cell-cell junction complexes are formed (Emerman and Pitelka, 1977). 239 Furthermore, only the cells on floating gels are able to correctly polarize and deposit an intact basement membrane (Parry et al., 1985; Streuli and Bissell 1990). To determine i f B M deposition observed in floating collagen gel culture is responsible for mammary epithelial morphogenesis, these cells have been placed directly on reconstituted 'Engelbreth-Holm-Swarm' tumor derived matrices known as Matrigel. When cells, are initially placed on matrigel, they form small aggregates and with time they 'pull ' the flexible matrix around them, cavitate and cells become apically/basally polarized as seen in the alveoli of a midpregnant or lactating mammary gland (Barcellos-Hoff et al., 1989; Aggeler et al., 1991). These early observations show that the flexible B M is a critical microenvironmental effector of architectural changes that allows changes to the cell shape during mammary epithelial morphogenesis. Futhermore, due to the morphological similarities, it is not always easy to distinguish the difference between normal epithelial cells and malignant cells in a 2D culture. However, in 3D environment these cells behave differently, showing diverse architectural patterns. These architectural organizations can represent the normal glandular structure to different stages of breast carcinoma progression (Briand et al., 1996; Weaver et al., 1995). Specifically, the 3D mammary epithelial culture model closely mimic the structure and the function of the in vivo mammary epithelium by forming acini like structures and responding to the lactogenic hormones by inducing milk protein expression. Therefore, as I have shown in this thesis, a 3D culture model is a better system to study how the normal mammary epithelial architecture is build and also to investigate factors that can lead to the loss of this architecture. Because of the hierarchical nature of the culture model used, I was able to examine TJ formation in a progressive manner. At the first level of hierarchy, epithelial monolayers have localized A J proteins E-cadherin and P-catenin at cell-cell interaction sites. In addition localization of f-actin in a similar manner indicated proper AJ formation. Surprisingly, western blotting shows expression of ZO-1 and occludin in these monolayers, though I was unable to immunolocalize much of the proteins to the cytoplasm or the membrane. One possibility is that because of diffused cytoplasmic distribution of ZO-1 I was not able to detect by immunolocalization. 240 In the second level of the hierarchy, cells were allowed to round up and cluster, which was similar to the flexible matrices. This change of cell shape and clustering was sufficient to localize ZO-1 and occludin to the cell membranes indicating TJ formation. The colocalization of ZO-1 with P-catenin further agrees with other reports showing possible interactions of AJ and TJ proteins in the early stages of TJ formation (Yokoyama et al., 2001; Rajasekaran et al., 1996). The colocalization I have observed and the protein interactions others have reported suggest that A J may function as a scaffolding site for TJ assembly (Asakura et al., 1999; Ando-Akatsuka et al., 1996). Similar junctional 'scaffolding' has been observed transiently in M D C K cells after calcium switch, but unlike in our mammary system this 'scaffolding' cannot be regulated since M D C K cells are able to spontaneously polarize (Nelson, 2003). Therefore, this provides us with a unique opportunity to study molecular scaffolding during morphogenesis. For instance, P-catenin contains a " D T D L " amino acid sequence that can interact with PDZ domain containing proteins (unpublished observation; McNagny K M ) , possibly with ZO-1. ZO-1 is also able to interact with all three transmembrane TJ proteins occludin, claudin and J A M (Matter and Balda, 2003). Therefore, it is possible that an initial membrane localization of ZO-1 can promote further recruitment of TJ proteins to the membrane. This idea needs further investigation using direct interaction studies. One way to directly test for these interactions in real time would be the use of fluorescence resonance energy transfer (FRET) technique. This technique would allow the examination of a possible interaction between ZO-1 and P-catenin as well as other junction proteins. Furthermore, the ability to carry out live cell experiments would give the opportunity to identify if, and when these interactions are taking place. One of the most commonly used methods to examine protein-protein interactions during cell-cell junction formation is co-immunoprecipitation (co-IP). In this method, cells are lysed in a mild lysis buffer (0.5-1% TritonX-100) and the soluble fractions are frequently used to co-IP while cytoskeletal pellets are discarded. One problem with these studies is that the reported interactions may not take place at the junction, but instead in the cytoplasm of the cell. In addition, the much harsher lysis conditions needed for the cytoskeletal pellet may disrupt any protein-protein interactions. These problems further suggest the need of FRET studies to confirm these possible AJ and TJ protein interactions. 241 In the second level of the hierarchy, cells were allowed to round up and cluster, which was similar to the flexible matrices. This change of cell shape and clustering was sufficient to localize ZO-1 and occludin to the cell membranes indicating TJ formation. The colocalization of ZO-1 with P-catenin further agrees with other reports showing possible interactions of A J and TJ proteins in the early stages of TJ formation (Yokoyama et al., 2001; Rajasekaran et al., 1996). The colocalization I have observed and the protein interactions others have reported suggest that AJ may function as a scaffolding site for TJ assembly (Asakura et al., 1999; Ando-Akatsuka et al., 1996). Similar junctional 'scaffolding' has been observed transiently in M D C K cells after calcium switch, but unlike in our mammary system this 'scaffolding' cannot be regulated since M D C K cells are able to spontaneously polarize (Nelson, 2003). Therefore, this provides us with a unique opportunity to study molecular scaffolding during morphogenesis. For instance, P-catenin contains a " D T D L " amino acid sequence that can interact with PDZ domain containing proteins (unpublished observation; McNagny K M ) , possibly with ZO-1. ZO-1 is also able to interact with all three transmembrane TJ proteins occludin, claudin and J A M (Matter and Balda, 2003). Therefore, it is possible that an initial membrane localization of ZO-1 can promote further recruitment of TJ proteins to the membrane. This idea needs further investigation using direct interaction studies. One way to directly test for these interactions in real time would be the use of fluorescence resonance energy transfer (FRET) technique. This technique would allow the examination of a possible interaction between ZO-1 and p-catenin as well as other junction proteins. Furthermore, the ability to carry out live cell experiments would give the opportunity to identify if, and when these interactions are taking place. One of the most commonly used methods to examine protein-protein interactions during cell-cell junction formation is co-immunoprecipitation (co-IP). In this method, cells are lysed in a mild lysis buffer (0.5-1% TritonX-100) and the soluble fractions are frequently used to co-IP while cytoskeletal pellets are discarded. One problem with these studies is that the reported interactions may not take place at the junction, but instead in the cytoplasm of the cell. In addition, the much harsher lysis conditions needed for the cytoskeletal pellet may disrupt any protein-protein interactions. These problems further suggest the need of FRET studies to confirm these possible A J and TJ protein interactions. 241 In the third level of hierarchy, when a soluble B M was added on to the naked clusters they formed spheroids with a central lumen. In response to the soluble B M , these cells undergo secondary differentiative events (Streuli et al., 1995). Now the TJs were localized at the apical domains that surround the lumen while the basal surfaces of the cells interact with the B M . Thus, the cell-BM interactions were sufficient to correctly polarize and localize the TJ apically. Furthermore, there was little or no co-localization of ZO-1 and P-catenin seen in these spheroids, indicating a complete separation of the junctions into two independent complexes. Future FRET studies could provide a better understanding of when the junctions begin to separate. The fractionation study and insoluble/soluble ratio of junction proteins further demonstrated that there was no change in the A J proteins, while there was a steady increase in membrane localization of TJ protein during the development of spheroids. The a6-integrin blocking studies show partial disruption of TJ polarization while not affecting TJ formation, further suggesting that the role of integrin-BM interaction in induction of apical/basal polarity of the epithelial cells. The observed partial disruption could be due to inefficient blocking of the integrin. I was unable to block P4 integrin due to the lack of a mouse blocking P4-antibody. I also tried the use of a P4 truncated mutant (Weaver et al., 2002), but these experiments were unsuccessful because the truncated form failed to suppress the endogenous P4 integrin. Future studies need to be carried out to further confirm the effects of a6p4 integrin in TJ polarization. One of the techniques that could be used is siRNA, which can be used to block both endogenous a6 and P4 to further study their role in TJ polarity. Furthermore, downstream signaling pathways leading from cell-BM interaction to TJ polarity are yet to be identified. The significance of the laminin in the B M is further exemplified by the co-culture studies of luminal and my epithelial cells. Luminal cells cultured on collagen gells form reverse polarized spheroids similar to the naked clusters I observed in absence of a B M . However, when these luminal cells are co-cultured with myoepithelial cells on collegen gels, they form double-layered acini with correct polarity. Furthermore, the polarizing component of the myoepithelial cells was found to be laminin (Gudjonsson et al., 2002). Thus, luminal and myoepithelial co-cultures may also provide a good in vitro model that closely mimic the in vivo mammary 242 gland. However, in our system adding myoepithelial cells will make it difficult to identify B M independent events. Dystroglycan is another laminin binding receptor that links the B M to the actin cytoskeleton. In the absence of Dystroglycan, mammary tumor cells fail to undergo polarization (Muschler et al., 2002). It is also required for apical/basal polarity establishment in the Drosophila epithelium (Deng et al., 2003). However, downstream mechanisms that may be responsible for dystroglycan dependent polarization are still unknown. One possibility is that the dystroglycan interaction with the actin cytoskeleton may also be involved in the establishment of the polarized junctions. Furthermore, actin modulating GTPases, Rac and Cdc42 may also be involved in the organization of actin at the apical junction. Interestingly, Cdc42 can interact with cell polarity associated molecules, specifically the PAR family of proteins. Future studies examining the expression and localization of PAR3, PAR6 and aPKC in our culture system will provide more information on how and when TJ are polarized. Based on our current understanding, I predict that PAR6 will be localized at cell-cell interaction sites in monolayers and naked clusters, while both PAP6 and PAR3 will be localized at the apical TJ sites in the polarized spheroids. The data from the current study demonstrate that TJ formation is influenced by changes in the cell shape, which can be facilitate by the cellular microenvironment. In the mammary gland, surrounding connective tissue stroma provides a flexible medium for epithelial cells to manipulate cell shape changes and cell-cell interactions. Then, signals generated by epithelial cell interactions with the B M , specifically laminin, would allow the proper localization of the TJ to the apical domains of the lateral membrane, polarizing the junctions. The outcome is an apically/basally polarized functional mammary epithelium. In this study, I further examined the effects of three genes E N , podocalyxin and ILK to determine their influences on cell junctions, polarity and morphogenesis. E N is a fusion oncoprotein that is highly expressed in SBC. Although SBC is a subtype of IDC, they often do not become metastatic. This phenotype is possibly correlated by my observation of E N expressing mammary epithelial cells in 3D culture. Ectopic E N expression in mammary epithelial cells resulted in an increase in cell proliferation leading to multilayering of 3D spheroids while maintaining the overall normal 3D spheroid 243 architecture. Cell attachment to the B M allows cross-talk between B1 integrin and IGF signaling pathways (Lee and streuli, 1999). Unlike the controls, E N expressing cells did not stop proliferation in response to interactions with the B M . Hence, one possibility could be that a change in integrin signalling from (34 to (31 allowed enhanced IGF signaling and continued proliferation. These spheroids were able to form AJ junctions at cell-cell interaction sites and TJ surrounding the lumen. Thus, the non-invasive nature of these cells could be due to the inability of E N to influence cell junction and polarity regulators. The similarities between the actual SBC phenotype and E N expressing spheroids suggest a strong functional importance of E N in generating the SBC. The normal mammary epithelium can respond to an abnormal increase in cell growth by apoptotic removal of cells that are not in contact with the B M (Debnath et al., 2002). For example, Bim, a pro-apoptotic protein is highly expressed upon cell detachment and function as a sensor for integrin and growth factor signalling (Reginato et al., 2003). Loss of such protein can deregulate both integrin and growth factor pathways leading towards to tumor progression. Interestingly, E N expressing spheroids do not seem to respond to these apoptotic signals. Therefore, it will be interesting to examine i f E N expressing cells are protected against apoptosis, thus allowing multilayering. It is quite possible that pro-apoptotic proteins such as Bim are lost in these cells or downstream signaling is disrupted. This should be examined with future studies. Furthermore, we can also use these EN-spheroids to identify additional oncoproteins or microenvironmental changes that may be necessary to disrupt the spheroid architecture and cause them to become invasive. The phenotype observed in ectopic Podocalyxin expressing MCF-7 cells can be compared to the phenotype of IDC. The tight junctions of MCF-7 monolayers were perturbed when Podocalyxin was expressed and cells were extruded off the monolayers into the media. TJ localization was greatly disrupted in Podocalyxin expressing spheroids indicating a breakdown of apical/basal polarity. However, cell-cell adhesions were maintained by AJ . The TJ localization in these spheroids was similar to that of the naked clusters of normal mammary epithelial cells. These observations further suggest that loss of cell polarity is a more prominent event in Podocalyxin overexpressing cells. Given the anti-adhesive function of Podocalyxin it is not surprising that its overexpression can 244 disrupt epithelial architecture (Takeda et al., 2000). In this study, I examined the effects of Podocalyxin expression in a well-differentiated cancer cell line MCF-7. Therefore, it will be important to examine i f Podocalyxin alone can induce similar effects on normal mammary epithelial cells or whether additional signals are required. Based on the observations of the current study, I would predict that transgene expressing normal cells might lose cell polarity as well as cell-cell adhesion. To avoid the difficulties I had, a retroviral infection system can be developed to introduce the Podocalyxin gene into the normal mammary epithelial cells, which might give better success in generating a pool of cells that express the trans-gene. Furthermore, an inducible system will be valuable to determine whether Podocalyxin can disrupt already polarized spheroids. Thus, with the use of the 3D-culture system, now we can begin to examine the anti-adhesive effects of Podocalyxin during morphogenesis. For example, since Podocalyxin is linked to the actin cytoskeleton via NHERF-2, we can examine mutants that may disrupt this cytoskeletal interaction leading to redistribution of Podocalyxin on the cell membrane making the cell less adhesive. Furthermore, in vivo targeted overexpression of Podocalyxin in the mammary gland as well as mammary fat pad implantation of Podocalyxin expressing MCF-7 cells will provide a better understanding of its role during cancer progression. By examining a patient tumor tissue array, I was also able to determine that Podocalyxin overexpression in a subset of tumors that became metastatic. Thus, increased levels of Podocalyxin may be used as a prognostic indicator to identify a subpopulation of breast cancer patients. This information may allow the design of tailored therapies to this high-risk subpopulation of patients. The Genetic Pathology Unit at the Vancouver General Hospital is currently constructing a T M A with patient-outcome-linked 3000 breast tumor samples, which will allow us to increase the statistical significant of the present study. The functional data presented in this thesis indicates that Podocalyxin overexpression is associated with architectural changes to the mammary epithelium. In ILC, cells invade into the surrounding stroma as spindle shaped single cells that have mesenchymal characteristics. Similarly, when ILK was overexpressed in mammary epithelial cells, an E M T occurred and the cells invaded the B M . These ILK-transformed mesenchymal cells have not only lost their apical/basal polarity, they have also lost cell-cell adhesion. This was further evident by loss of E-cadherin expression in the presence 245 of high ILK levels. These observations are further confirmed by the data showing transgenic mice that are overexpressing ILK in the mammary glands initially show hyperplasia, which then progresses into invasive tumors undergoing an E M T (White et al., 2001). The tumor suppressor gene PTEN inhibits the ILK activity and the loss of PTEN expression correlates with breast cancer progression and lymph node metastasis (Chung et al., 2003; Morimoto et al., 2000). Therefore, it is valid to suggest that ILK could be constitutively activated in PTEN mutated or suppressed cells promoting cancer progression. ILK overexpression also results in suppression of apoptosis and promotion of anchorage independent growth (Atwell et al., 2000). Thus, the cells with high levels of ILK may also have high metastatic potential. This is the case with gastric carcinoma where epithelia polarity is disrupted and cells become highly invasive and metastatic (Ito et al., 2003). I L K is able to suppress glycogen synthase kinase-3 (GSK-3) and enhance Wnt signaling (Delcommenne et al., 1998; Novak et al., 1998; Troussard et al., 1999), which can be responsible for the observed inhibition of E-cadherin and induction of an EMT. Hence, loss of E-cadherin at the cell junctions may be the critical event in inducing invasive tumor progression. The E-cadherin re-expression study further demonstrates the power as architectural gene in preventing invasive cancer progression. The work presented in this thesis highlights the importance of genes that are responsible for regulating the normal mammary epithelial architecture. Disruption of epithelial tight junctions and apical/basal polarity seems to, at least in part, promote breast cancer progression into metastatic disease. Hence, tight junction proteins and cell polarity proteins can be good therapeutic targets in the fight to block cancer cell invasion, by restoring the normal epithelial architecture. Furthermore, loss of these architectural protein expression or gene mutations can be useful early prognostic indicators, as is the case with podocalyxin. 246 Figure 6.1 Hypothetical linear model of breast cancer progression showing the findings of this thesis. A. During normal mammary epithelial morphogenesis, A J may function as scaffolds to build the TJs. TJ polarization is at least in part dependent on the integrin-MB interaction. B. E N is functionally significant in generating the SBC phenotype. C. Podocalyxin is an independent prognostic indicator for invasive breast cancer progression and may be functionally important to generating the IDC phenotype. D. ILK is functionally important in E M T generating the ILC phenotype. 247 to oo •Scaffolding •Integrin-BM dependency - Polarized TJ Normal Glandular Epithelium Hyperplasia t Proliferation ^Apoptosis Atypical Hyperplasia Slight Architectural Disruption Lobular Carcinoma in situ Ductal Carcinoma in situ f Architectural Disruption ETV6-NTRK3 ILK Invasive Lobular Carcinoma Single Cel l Invasion Podocalyxin Invasive Ductal Carcinoma Disorganized Cel l Clusters Secretory Breast Carcinoma t Multilayering Differentiated Clusters 6 . 1 R e f e r e n c e s Aggeler J, Ward J, Blackie L M , Barcellos-Hoff M H , Streuli C H and Bissell MJ . (1991). Cytodifferentiation of mouse mammary epithelial cells cultured on a reconstituted basement membrane reveals striking similarities to development in vivo. J Cell Sci. 99 (Pt 2): 407-17. Asakura, T., Nakanishi, H. , Sakisaka, T., Takahashi, K. , Mandai, K. , Nishimura, M . , Sasaki, T., and Takai, Y . (1999). Similar and differential behavior between the nectin-afadin-ponsin and cadherin-catenin systems during the formation and disruption of the polarized junctional alignment in epithelial cells. Genes Cells 4:573-581 Ando-Akatsuka, Y . , Yonemura, S., Itoh, M . , Furuse, M . , and Tsukita, S. (1996). Differential behavior of E-cadherin and occludin in their colocalization with ZO-1 during the establishment of epithelial cell polarity. J. Cell. Physiol. 179:115-125. Attwell, S., Roskelley, C. D. and Dedhar, S. (2000). The integrin-linked kinase (ILK) inhibits anoikis. Oncogene 19, 3811-3815. Debnath, J., K. Mills, N . Collins, M . Reginato, S. Muthuswamy, and J. Brugge. 2002. The role of apoptosis in creating and maintaining luminal space within normal and oncogene-expressing mammary acini. Cell. I l l :29-40. Barcellos-Hoff M H , Aggeler J, Ram TG, and M J Bissell. (1989) Functional differentiation and alveolar morphogenesis of primary mammary cultures on reconstituted basement membrane. Development. 105(2):223-35. Briand P, Nielsen K V , Madsen M W and OP Petersen. (1996) Trisomy 7p and malignant transformation of human breast epithelial cells following epidermal growth factor withdrawal. Cancer Res 56:2039-2044. 249 Chung MJ , Jung SH, Lee BJ, Kang MJ , Lee DG. (2004) Inactivation of the PTEN gene protein product is associated with the invasiveness and metastasis, but not angiogenesis, of breast cancer. Pathol Int. 54(1): 10-5. Delcommenne, M . , Tan, C , Gray, V. , Rue, L., Woodget, J. and Dedhar, S. (1998). Phosphoinositide-3-OH kinase-dependent regulation of glycogen synthase kinase 3 and protein kinase B / A K T by the integrin-linked kinase. Proc. Nat. Acad. Sci. USA 95, 11211-11216. Deng W M , Schneider M , Frock R, Castillejo-Lopez C, Gaman E A , Baumgartner S and Ruohola-Baker H. (2003) Dystroglycan is required for polarizing the epithelial cells and the oocyte in Drosophila. Development. 130(1): 173-84. Emerman, J. T., and D. R. Pitelka. (1977) Maintenance and induction of morphological differentiation in dissociated mammary epithelium on floating collagen membranes. In Vitro, 13(5):316-328. Ito R, Oue N , Zhu X , Yoshida K , Nakayama H , Yokozaki H and Yasui W. (2003) Expression of integrin-linked kinase is closely correlated with invasion and metastasis of gastric carcinoma. Virchows Arch. 442(2), 118-23. Lee Y J and Streuli CH. (1999) Extracellular matrix selectively modulates the response of mammary epithelial cells to different soluble signaling ligands. J Biol Chem. 274(32):22401-8. Matter K and Balda MS. (2003) Signaling through tight junctions. Nature Reviews Molecular Cell Biology 4:225 -237 Morimoto A M , Tomlinson M G , Nakatani K, Bolen JB, Roth R A , Herbst R. (2000) The M M A C 1 tumor suppressor phosphatase inhibits phospholipase C and integrin-linked kinase activity. Oncogene. 19(2): 200-9. 250 Muschler J, Levy D, Boudreau R, Henry M , Campbell K and Bissell M J . (2002) A role for dystroglycan in epithelial polarization: loss of function in breast tumor cells. Cancer Res. 62(23):7102-9. Nelson WJ. (2003) Adaptation of core mechanisms to generate cell polarity. Nature. 422(6933):766-74. Novak, A. , Hsu, S.-S., Leung-Hagesteijn, C , Radeva, G., Papkoff, J., Montessano, R., Roskelley, C , Grosschedl, R. and Dedhar, S. (1998). Cell adhesion and the integrin-linked kinase regulate the LEF-1 and b-catenin signaling pathways. Proc. Nat. Acad. Sci. USA 95, 4374-4379. Reginato MJ, Mills KR, Paulus JK, Lynch DK, Sgroi DC, Debnath J, Muthuswamy SK and Brugge JS. (2003) Integrins and EGFR coordinately regulate the pro-apoptotic protein Bim to prevent anoikis. Nat Cell Biol. 5(8):733-40. Streuli C H , Schmidhauser C, Bailey N , Yurchenco P, Skubitz AP , Roskelley C, and Bissell MJ . (1995) Laminin mediates tissue-specific gene expression in mammary epithelia. J Cell Biol. 129(3):591-603. Takeda T, Go W Y , Orlando R A and Farquhar M G . (2000) Expression of podocalyxin inhibits cell-cell adhesion and modifies junctional properties in Madin-Darby canine kidney cells. Mol Biol Cell. ll(9):3219-32. Troussard, A . A , Tan, C , Youganathan, T. N . and Dedhar, S. (1999). Cell-extracellular matrix interactions stimulate the AP-1 transcription factor in an integrin-linked kinase-and glycogen synthase kinase 3-dependent manner. Mol. Cell. Biol. 19, 7420-7427. Weaver V M , Lelievre S, Lakins JN, Chrenek M A , Jones JC, Giancotti F, Werb Z and Bissell MJ . (2002) beta4 integrin-dependent formation of polarized three-dimensional 251 architecture confers resistance to apoptosis in normal and malignant mammary epithelium. Cancer Cell. 2(3):205-16. Weaver V M , Howlett AR, Langton-Webster B, Petersen OW and M.J . Bissell. (1995) The development of a functionally relevant cell-culture model of progressive human breast cancer. Semin Cancer Biol 6:175-184. White DE, Cardiff RD, Dedhar S and Muller WJ. (2001) Mammary epithelial-specific expression of the integrin-linked kinase (ILK) results in the induction of mammary gland hyperplasias and tumors in transgenic mice. Oncogene. 20(48), 7064-72. Yokoyama S, Tachibana K , Nakanishi H , Yamamoto Y , Irie K , Mandai K , Nagafuchi A , Monden M , Takai Y . (2001) alpha-catenin-independent recruitment of ZO-1 to nectin-based cell-cell adhesion sites through afadin. Mol Biol Cell. 12(6): 1595-609. 252 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0091859/manifest

Comment

Related Items