Effects and regulation of dystroglycan glycosylation in cancer

Transcription

1 University of Iowa Iowa Research Online Theses and Dissertations Spring 2015 Effects and regulation of dystroglycan glycosylation in cancer Michael Raymond Miller University of Iowa Copyright 2015 Michael Miller This dissertation is available at Iowa Research Online: Recommended Citation Miller, Michael Raymond. "Effects and regulation of dystroglycan glycosylation in cancer." PhD (Doctor of Philosophy) thesis, University of Iowa, Follow this and additional works at: Part of the Biophysics Commons

2 EFFECTS AND REGULATION OF DYSTROGLYCAN GLYCOSYLATION IN CANCER by Michael Raymond Miller A thesis submitted in partial fulfillment of the requirements for the Doctor of Philosophy degree in Molecular Physiology and Biophysics in the Graduate College of The University of Iowa May 2015 Thesis Supervisor: Associate Professor Michael D. Henry

3 Graduate College The University of Iowa Iowa City, Iowa CERTIFICATE OF APPROVAL PH.D. THESIS This is to certify that the Ph.D. thesis of Michael Raymond Miller has been approved by the Examining Committee for the thesis requirement for the Doctor of Philosophy degree in Molecular Physiology and Biophysics at the May 2015 graduation. Thesis Committee: Michael D. Henry, Thesis Supervisor Kevin Campbell Charles Brenner Michael Wright Christopher Stipp

4 To my wife and daughter ii

5 ACKNOWLEDGMENTS I would like to thank all members of the Henry Lab, past and present for scientific guidance, enlightening discussion, and their friendship. Thank you to my thesis committee for the support and feedback. Thank you to the Medical Scientist Training Program and all of the support it has provided. A special thank you to Dr. Michael Henry for his guidance, support, continued motivation, enthusiasm and mentorship. iii

6 ABSTRACT The interplay between cancer cells and the extracellular matrix (ECM) remains a critical regulator of both normal tissue organization and cancer cell invasion. Proteins that function as ECM receptors function to link the cell with the ECM. Abberations in either the structure of the ECM or the expression of ECM receptors leads to disrupted interaction and downstream signaling effects. Dystroglyan (DG) is an ECM receptor that is expressed in a variety of tissue types and functions to mediate sarcolemma stability, epithelial polarity, and is critical in the early formation of basement membranes. However, DG has primarily been studied in muscle where loss of its function is linked to a host of muscular dystrophies. In the epithelium, the role of DG remains enigmatic. While DG has repeatedly been shown to lose function during cancer development and progression, the mechanism and functional consequence of its loss are currently unknown. In order to increase our understanding of DG in cancer development, we analyzed its expression and glycosylation, a functional requirement for DG, in a range of prostate cancer cell lines. Previous work has shown DG to be downregulated in prostate cancer, but the mechanism by which this occurs has remained largely unclear. We found that DG expression is maintained while its glycosylation was heterogeneous in the cell lines. Further investigation revealed that lines with hypoglycosylated DG strongly associated with the loss of expression of the glycosyltransferase LARGE2. Further this enzyme is frequently downregulated in human cancers and appears to serve as a required enzyme in DG glycosylation within prostate epithelium. This is the first work to demonstrate the functional requirement of LARGE2 for DG, and the only work to implicate loss-offunction of LARGE2 in cancer progression. To determine whether loss of LARGE2 is found in other tumor types, we analyzed human clear cell renal cell carcinoma (ccrcc) samples by iv

7 immunohistochemistry and via in silico analysis with the Cancer Genome Atlas (TCGA). Our work demonstrated a frequent and significant downregulation of LARGE2 expression and its association with DG hypoglycosylation. Additionally, we found the loss of LARGE2 strongly associated with increased mortality. Thus, we again demonstrated a functional requirement of LARGE2 but also found a clinical correlate with increased mortality. Finally, we examined the functional outcome of DG hypoglycosylation or loss of expression in both a mouse model of prostate cancer and a variety of cell lines models. We found that while loss of DG expression does not increase prostate cancer growth or metastasis in one model of cancer, loss of its glycosylation does seem to mediate downstream metabolic changes within cells. The mechanism for this change remains unclear. In summary, these studies have contributed to our understanding of DG glycosylation and function in both prostate and renal carcinoma. Additionally, we have shown a novel mechanism by which DG glycosylation is lost with downregulation of LARGE2 expression. Finally, while we were unable to demonstrate a clear mechanism by which signaling changes arose, we were able to demonstrate a strong correlation between DG hypoglycosylation and increased mortality in ccrcc. These insights could be used to improve treatment of multiple cancer types as our understanding of DG function continues to improve. v

8 PUBLIC ABSTRACT Cancer remains a major cause of mortality even while healthcare quality continues to improve. While there are many reasons that tumors result in mortality, one of the most common is metastatic spread of the tumor and disruption of normal bodily processes. There are a number of ways by which tumors metastasize, but one of the earliest steps is interruption of the normal connections that exist between a cell and the surrounding acellular material, the extracellular matrix (ECM). One of the proteins that mediates this connection is dystroglycan (DG). DG sits on the cell surface and interacts with the ECM to keep the cell attached and in its designated space. When cancer causes cells to grow abnormally, the production and processing of DG is decreased such that it no longer functions to bind the cells to the ECM. Our work has focused on understanding the changes that lead to decreased DG function and the effects that has on tumor progression and patient prognosis. We found that one of thirteen proteins known to modify DG is frequently downregulated in advanced disease, and the loss of this protein, LARGE2, leads to a loss of DG function in prostate and renal cell carcinomas. Additionally, loss of LARGE2 leads to a significant reduction in overall survival of patients with renal cell carcinoma. While we were unable to identify the mechanism by which DG disruption leads to decreased survival, our work lays the groundwork for future experiments and possible therapeutics in cancer patients. vi

9 TABLE OF CONTENTS LIST OF TABLES... xi LIST OF FIGURES... xii CHAPTER I. INTRODUCTION... 1 The Extracellular Matrix... 1 Dystroglycan... 2 The Dystroglycan Glycoprotein... 4 Muscular Dystrophies and Dystroglycan Glycosylation... 5 Protein O-Mannosyl Modification... 7 β1,2 GlcNac Modification of O-Mannose... 8 O-Mannose Phosphorylation... 9 Xylosyl-glucoronyl modification of the phospho-o-mannose Factors of unknown function Dystroglycan Interactions in Muscle Dystroglycan Interactions in Epithelium Dystroglycan Expression and Glycosylation in Cancer Relevance and Focus II. LOSS OF LARGE2 DISRUPTS FUNCTIONAL GLYCOSYLATION OF α-dystroglycan IN PROSTATE CANCER Introduction Materials and Methods Immunohistochemistry Cell Culture shrna Knockdown of LARGE Overexpression of LARGE qpcr Analysis Western Blot Flow Cytometry Analysis Laminin-111 Binding Immunofluorescence Transwell Migration Asssay Matrigel Invasion Assay Growth Assay Results αdg Glycosylation and βdg Expression In Both Primary And Metastatic Prostate Cancer Functional Glycosylation Of αdg Across Prostate Cancer Cell Lines Is Heterogeneous Functional Glycosylation of αdg Correlates with Expression of the LARGE2 mrna LARGE2 Functionally Glycosylates αdg in the Prostate LARGE2 Overexpression Restores DG Function And Diminishes Invasion And Cell Proliferation Potential vii

10 LARGE2 Expression Diminishes During Prostate Cancer Progression Discussion Contributions III. IV. DOWNREGULATION OF MULTIPLE ENZYMES LEADS TO LOSS OF DYSTROGLYCAN GLYCOSYLATION IN CLEAR CELL RENAL CELL CARCINOMA Introduction Materials and Methods Human Tissue Samples Immunohistochemistry The Cancer Genome Atlas Cell Lines Flow Cytometry Statistical Analysis Results Dystroglycan Expression and Glycosylation are Diminished in Clear Cell Renal Cell Carcinoma The DG Glycosylation Pathway is Perturbed During Clear Cell Renal Cell Carcinoma Development DG-Associated Glycosyltransferases Associate with Grade and Stage Pathway-Corrected Analysis Identifies Genes Associated with Mortality Regulation of GYLTL1B Links DG Glycosylation with the Epithelial-to-Mesenchymal Transition Pathway Discussion Contributions DYSTROGLYCAN MODULATES METABOLIC ACTIVITY AND AUTOPHAGY INDUCTION Introduction Materials and Methods Cell Lines Antibodies Used Cell Proliferation Assays Seahorse Bioanalyzer Metabolic Analysis Western Blot Flow Cytometry ATP Determination Cell Immunofluoresence Clonogenic Survival Autophagic Flux Assay Statistics Results Disruption of DG does not lead to changes in cellular proliferation Inhibiting Dystroglycan Decreases Metabolic Rate Disruption of DG Does Not Affect Intracellular ATP AMPK and Akt Signaling are Unaffected by DG Function Autophagic Capacity is Dependent Upon Dystroglycan Function DG Function Does Not Mediate Clonogenic Survival Following Nutrient Deprivation DG Mediates MRLCII Phosphorylation viii

11 Conclusions V. PROSTATE-SPECIFIC DELETION OF DYSTROGLYCAN DOES NOT EXACERBATE DISEASE IN A PTEN-DFIEICIENT MOUSE MODEL OF PROSTATE CANCER Introduction Materials and Methods Animals Tiisue Collection and Processing Immunofluorescent Staining Statistics Results Dystroglycan Hypoglycosylation Occurs in a Mouse Model of Early Prostate Cancer Prostate Histology at 3 Months Prostate Histology at 6 Months Gross Pathology and Prostate Histology at 12 Months Discussion VI. SUMMARY AND FUTURE DIRECTIONS REFERENCES ix

12 LIST OF TABLES Table 2-1. Demographics of the cohort used for retrospective analysis Table 3-1. Cohort characteristics of patients utilized within this study x

13 LIST OF FIGURES Figure 2-1. αdg glycosylation and βdg expression are reduced in prostate adenocarcinoma αdg functional glycosylation is nearly absent in prostate adenocarcinoma metastases Association of biochemical recurrence with glycosylation status of αdg in a retrospective analysis αdg functional glycosylation is heterogeneous in prostate cancer LARGE2 mrna expression correlates with hypoglycosylation of αdg LARGE2 alters IIH6 immunoreactivity in multiple prostate cancer cell lines LARGE2 functionally glycosylates αdg reducing invasive and proliferative potential Levels of LARGE and LARGE2 expression differ in cancer cell lines derived from different tissues LARGE2 expression is diminished during prostate cancer progression Dystroglycan expression and glycosylation is frequently lost in clear cell renal cell carcinoma Neither loss of DG glycosylation nor expression associates with disease recurrence Multiple proteins in the DG glycosylation pathway are downregulated during tumor progression Multiple genes in the DG-glycosylation pathway show a significant association with both disease grade and stage Loss of GYLTL1B and ISPD independently predict increase mortality for clear cell carcinoma patients Methylation of the GYLTL1B CpG island negatively correlates with expression of the gene A genome-wide correlation analysis shows ZEB1 as a potential regulator of GYLTL1B expression IIH6 immunoreactivity is lost in a murine cell line of renal carcinoma Inhibition of DG is achieved through shrna-mediated knockdown and Cremediated knockout of either DG or LARGE xi

14 4-2 Inhibition of DG function either by disruption of DG expression or its glycosylation causes no discernible growth phenotype Loss of DG function causes a significant reduction in both glycolytic activity and oxidative phosphorylation Relative ATP is unaffected by DG status in PC-3E+, 22Rv1, and MEF cells mtorc1 signaling is unaffected by DG function in PC-3E+ cells Loss of DG reduces autophagy-associated LC3 accumulation DG-associated disruption of autophagic flux is observed in PC-3E+, but not 22Rv1 cells DG does not affect clonogenic potential following glucose deprivation Loss of DG function prevents serum-starvation induced phosphorylation of MRLCII PTEN deletion but not acute inflammation associate with loss of IIH6 immunoreactivity in the mouse prostate Three month histology shows both desmoplasia and nuclear atypia in PTENFL/fl prostates At 6 months, the desmoplastic response and cellular atypia are exaggerated in PTENfl/fl prostates Gross histopathology of PTENfl/fl animals demonstrate significant anterior prostate swelling with associated seminal vesicle necrosis Disease progression 12 months is driven entirely by PTEN expression Laminin organization is maintained in the murine prostate regardless of DG expression Lymph nodes lack indications of metastasis but do exhibit hypercellularity and macrophage infiltration xii

15 1 CHAPTER I INTRODUCTION The Extracellular Matrix The extracellular matrix (ECM) is an abundant framework of secreted molecules, glycoproteins, collagens, glycosaminoglycans and proteoglycans. It is responsible for the strength and support of the majority of tissues in vivo including bone and cartilage. In combination with its structural role, the ECM also serves as a readily accessible pool of growth factors bound to numerous proteins within the matrix, such as perlecan. This not only serves as an organizing center for tissue, but also provides the means for growth and development. Furthermore, it is within this matrix that the cell-surface mediated signaling interactions occur. Cell adhesion to this structure requires a number of ECM receptors that include integrins, discoidin domain receptors, syndecans, and dystroglycan (Frantz, Stewart et al. 2010). The ECM thus plays a critical role in organismal homeostasis, structural support, and development. Basement membranes (BM) are highly specialized, thin, sheet-like extracellular matrices common to epithelial tissues. They serve as an organizing scaffold as well as a primary barrier between cells and the surrounding stroma. A primary component of BM is type IV collagen (Kalluri 2003), a target of histological stains that is often used to determine basement membrane continuity in histology. Cellular attachment to the basement membrane also prevents anoikis, a specialized form of apoptosis triggered in anchorage-dependent cells (Ruoslahti and Reed 1994). Therefore, the continuity of the basement membrane and the integrity of cell-ecm interactions must be maintained for tissue homeostasis. It s well established that the basement membrane is often compromised during the development of cancer and correlates with metastatic potential (Liotta, Tryggvason et al.

16 2 1980, Barsky, Siegal et al. 1983). Loss of the basement membrane results in a microenvironment where cells are no longer anchored nor organized, and for tumor cells, this allows for escape from the primary site of tumor growth. The cell-ecm interaction is therefore a primary point of study in understanding the development of metastatic potential in human cancers. While a number of proteins function to regulate this interaction, the work presented herein will focus primarily on one, dystroglycan. Dystroglycan Dystroglycan (DG) is an extracellular matrix interacting protein that was originally discovered in skeletal muscle as a component of the dystrophin glycoprotein complex (DGC) (Ervasti, Ohlendieck et al. 1990, Ervasti and Campbell 1991). This complex is required for mechanical stability of the sarcolemma in skeletal muscle. Mutation of any component of the DGC leads to the development of muscular dystrophies characterized by muscle wasting and associated weakness. These dystrophies vary in severity but involve a combination of musculoskeletal, neuronal, and optical pathologies. Due to its pathologic association with these muscular dystrophies, dystroglycan has primarily been studied in the context of muscle pathophysiology; however DG s expression is not limited to muscle. Dystroglycan is also ubiquitously expressed in neuronal and epithelial tissues where its function is still being elucidated (Durbeej, Henry et al. 1998). Dystroglycan is encoded by the DAG1 gene located on chromosome 3p21 and is composed of two exons and a single intron. DG is expressed as a propeptide which undergoes autoproteolysis into a 72 kda alpha and 43 kda beta subunit (Ibraghimov- Beskrovnaya, Milatovich et al. 1993, Akhavan, Crivelli et al. 2008). This proteolytic processing is conserved through all vertebrates but does not occur in D. melanogaster or C. elegans (Deng, Schneider et al. 2003, Johnson, Kang et al. 2006). While the function of this cleavage is unknown, if DG is artificially expressed as a non-cleavable mutant in a

17 3 mouse model, the animal develops a muscular dystrophy phenotype (Jayasinha, Nguyen et al. 2003). Native DG is observed at a much higher molecular weight by gel electrophoresis due to extensive glycosylation, primarily of alpha-dg (Ibraghimov- Beskrovnaya, Ervasti et al. 1992, Ervasti and Campbell 1993, Ervasti, Burwell et al. 1997). Alpha-DG resides entirely on the cell surface with its C-terminus non-covalently linked to beta-dg. The alpha subunit exists as a general barbell shape with a handle composed of a serine/threonine-rich mucin domain flanked on the N-terminus by an IGlike domain fused to a ribosomal protein S6-like domain (Akhavan, Crivelli et al. 2008). The C-terminus of the barbell is composed of a cadherin-like domain (Brancaccio, Schulthess et al. 1995, Brancaccio, Schulthess et al. 1997, Akhavan, Crivelli et al. 2008). It has also been shown that alpha-dg may be cleaved at the extreme N-terminus by a furin protease, however the functional significance of this cleavage is unclear (Singh, Itahana et al. 2004). Beta-DG is a single-pass transmembrane protein that serves as the functional link between alpha-dg and intercellular DG binding partners. The ectodomain of beta-dg is known to exist as a natively unfolded protein which is thought to allow for more fluid interactions with binding partners (Bozzi, Bianchi et al. 2003). The cytoplasmic portion of beta-dg contains a nuclear localization sequence located in the juxtamembrane region that is responsible for translocation of the full-length protein into the nucleus, in vitro (Fuentes-Mera, Rodriguez-Munoz et al. 2006, Oppizzi, Akhavan et al. 2008, Lara- Chacon, de Leon et al. 2010), however this finding has not yet been confirmed in vivo and its importance to DG physiology is still debated. DG also contains a WW-domainbinding PPxY motif and an SH2 binding domain near its C-terminus (Moore and Winder 2010). In muscle, the DGC is found at the sarcolemmal membrane and in neuromuscular junctions, while in epithelium, its expression is limited to the basolateral surface of cells, with most tissues exhibiting greatest expression along the basement membrane (Durbeej, Henry et al. 1998).

18 4 The Dystroglycan Glycoprotein Both alpha- and beta-dg are post-translationally modified by a variety of enzymes. Glycosylation of the alpha subunit is the best-characterized modification due primarily to its necessity for ligand binding. Protein glycosylation itself occurs through a small number of primary mechanisms. N-linked glycosylation attaches an N-acetylglucosamine (GlcNAc) to an asparagine residue of the target protein, and a glycan chain can then be built from this foundation. Another form of glycosylation which attaches to serine and threonine residues is O-linked glycosylation. This type of modification includes linkage of a variety of sugar moieties such as fucose, glucose, GlcNAc, and N-galactosamine (GalNAc). While it is not frequently observed in higher eukaryotes, O-linked mannosylation can also occur. Dystroglycan is known to be modified by both N- and O- linked glycan moieties, but it is the O-linked mannosylation that is critical for dystroglycan function (Combs and Ervasti 2005). This modification is rather unusual and has only been characterized in a small number of proteins including RPTPβ/ζ (Dwyer, Baker et al. 2012), CD24 (Bleckmann, Geyer et al. 2009), neurofascin (Pacharra, Hanisch et al. 2012) and the cadherin family (Vester-Christensen, Halim et al. 2013, Winterhalter, Lommel et al. 2013). Multiple forms of the core glycan have been identified including the more common Galβ1,3GalNAcα-O-S/T and two less common O-mannose glycans, NeuAcα2,3Galβ1,4GlcNAcβ1,2Manα-O-S/T and the more recently identified NeuAcα2,3Galβ1,4GlcNAcβ1,4Manα-O-S/T (Nilsson, Nilsson et al. 2010, Yoshida- Moriguchi, Yu et al. 2010, Stalnaker, Aoki et al. 2011). The functional glycan required for ligand binding is now known to be the latter of the two O-mannose glycans. A highly unusual phosphorylation of the O-mannose has been shown to be necessary for ligand binding (Yoshida-Moriguchi, Yu et al. 2010) and recognition by the glycosyltransferase LARGE for further extension (Hara, Kanagawa et al. 2011). The glycosyltransferases are

19 5 a group of enzymes that enzymatically attach glycan moieties to their respective protein targets. LARGE acts to extend the glycan chain on alpha-dg from this phospho-mannose via its recently identified xylosyl-and glucoronyltransferase activity, a function it shares with its homolog, LARGE2 (Inamori, Yoshida-Moriguchi et al. 2012, Inamori, Hara et al. 2013). This highly modified glycan chain is required for dystroglycan to bind DG s ligands, but as is discussed below, the β1,2 O-mannose glycan clearly plays a functional role as its disruption also causes a muscular dystrophy phenotype. As previously mentioned, the alpha-dg subunit has a predicted molecular weight of ~70 kda, but its observed molecular weight by electrophoresis ranges from 120 kda in brain to 156 kda in muscle (Ibraghimov-Beskrovnaya, Ervasti et al. 1992). The N-linked glycans account for only ~ 4kDa of total molecular weight, indicating that the remainder of the glycosylation is achieved via O-linked modifications (Ervasti and Campbell 1991). These N-glycans do not appear to play a functional role in the ability of alpha-dg to interact with its ligands (Ervasti and Campbell 1993). Muscular Dystrophies and Dystroglycan Glycosylation The alpha-dg protein is known to interact with a variety of extracellular matrix proteins containing laminin G domains including laminin (Ervasti and Campbell 1991), perlecan (Peng, Ali et al. 1998), agrin (Gee, Montanaro et al. 1994), pikachurin (Sato, Omori et al. 2008) and neurexin (Sugita, Saito et al. 2001). Additionally, the binding epitope is co-opted by the Lymphocytic Choriomeningitis Virus and Lassa Fever Virus for cell binding and entry(cao, Henry et al. 1998, Smelt, Borrow et al. 2001). The interaction between dystroglycan and its ECM ligands is critical for appropriate formation and organization of the basement membrane as evidenced by the fact that knockout of dystroglycan is embryonic lethal in mice due to the lack of Reichert s

20 6 membrane development, one of the first basement membranes to develop during embryogenesis (Williamson, Henry et al. 1997). In skeletal muscle, dystroglycan is an integral part of the DGC which confers structural stability to the sarcolemma during muscle contraction (Petrof, Shrager et al. 1993). This structural integrity is critical to long-term function of muscle, and alterations to the DGC protein complex leads to a variety of muscular dystrophies including Duchenne s, Becker s, and a host of congenital muscular dystrophies (CMDs) referred to collectively as the dystroglycanopathies (Straub and Campbell 1997). The dystroglycanopathies are so named because they all share a characteristic finding of decreased or absent dystroglycan function. Thus far, only a single case report of a primary dystroglycanopathy has been described (Hara, Balci-Hayta et al. 2011), while the majority of patients present with secondary dystroglycanopathies that arise from improper DG post-translational modification due to mutational loss of the enzymes responsible for DG function. These diseases include, from most to least severe, Walker Warburg Syndrome (WWS), Muscle-Eye-Brain (MEB) disease, Fukuyama-type congenital muscular dystrophy, congenital muscular dystrophy 1C and 1D, and limbgirdle muscular dystrophy 2I. The 12 enzymes described in the following section are all known or putative glycosyltransferases that affect DG glycosylation and subsequent function. Mutation of any of these enzymes has been linked to muscular dystrophies in both patients and mouse models. Other proteins that do not specifically interact with DG have also been implicated in dystroglycanopathies, highlighting the functional significance of dystroglycan glycosylation and O-mannosylation on human physiology.

21 7 Protein O-Mannosyl Modification The functional glycosylation of dystroglycan requires O-mannose modification of the alpha-dg mucin domain in order for ligand binding to occur. This O-mannosyl moiety is added by the paired action of protein O-mannosyltransferase 1 and 2 (POMT1 and POMT2). The activity of these enzymes is dependent on the expression of both proteins for proper function (Manya, Chiba et al. 2004). Together, POMT1 and POMT2 link a mannose moiety to alpha-dystroglycan via an O-linkage from a dolichol-phosphate mannose (DPM) donor. DPM synthesis is performed by the DPM synthase complex located on the endoplasmic reticulum (ER). The loss of DPM synthase function via mutation of any of its subunits (DPM1, DPM2, or DPM3) leads to dystroglycanopathy with associated neuronal involvement (Lefeber, Schonberger et al. 2009, Barone, Aiello et al. 2012, Yang, Ng et al. 2013). Likewise, disruption of either member of the transferase complex (POMT1 or POMT2) also leads to the development of WWS (Beltran-Valero de Bernabe, Currier et al. 2002, Akasaka-Manya, Manya et al. 2004, van Reeuwijk, Janssen et al. 2005). The presence of O-mannosyl moieties is relatively rare in higher order eukaryotes, and dystroglycan remains one of the few identified targets. More recent work has pointed to a novel protein required for the functional O- mannosylation activity of the POMT1/2 complex (Willer, Lee et al. 2012). The isoprenoid-synthase domain containing protein (ISPD) is now known to be a frequent mutation leading to WWS in a number of patients with previously unknown genetic cause for their disease. ISPD does not have clear glycosyltransferase activity, but its absence significantly inhibits activity of the POMT1/2 complex. While the bacterial orthologue of ISPD is known to act in the non-mevalonate pyrenoid synthesis pathway, this pathway is not utilized by higher eukaryotes, which leaves the functional role of ISPD in question. However, multiple papers in the past year indicate that ISPD is a frequent cause of WWS in disease not previously attributed to one of the canonical DG

22 8 glycosyltransferase mutations, suggesting that the relative prevalence of this pathway to disease is quite common (Roscioli, Kamsteeg et al. 2012, Willer, Lee et al. 2012, Cirak, Foley et al. 2013, Tasca, Moro et al. 2013). β1,2 GlcNac Modification of O-Mannose The second sugar attached on dystroglycan glycan structures is N-acetylglucosamine (GlcNac) through a β1,2 linkage. This action is accomplished by the protein-omannosyl-β1,2-n-acetylglucosamine transferase (POMGnT1) located in the cis-golgi (Chiba, Matsumura et al. 1997). Mutation of POMGnT1 leads to a range of CMDs (Yoshida, Kobayashi et al. 2001, Biancheri, Bertini et al. 2006, Hehr, Uyanik et al. 2007, Voglmeir, Kaloo et al. 2011). POMGnT1 β1,2 transferase activity was recently confirmed (Stalnaker, Aoki et al. 2011), but this is somewhat unusual as this glycan is not necessary for extension of the LARGE-mediated glycan, which is thought to facilitate ligand binding from a β1,4 GlcNac modification (Yoshida-Moriguchi, Yu et al. 2010). Laminin binding deficiency due to POMGnT1 mutation may be overcome by overexpression of LARGE (Barresi, Michele et al. 2004) indicating that the glycan moiety is not necessary for LARGE-mediated function, however basal expression of POMGnT1 does improve ligand binding efficiency. POMGnT1 is one of the only glycosyltransferases identified that is also capable of causing a dystroglycanopathy via mutation of its promoter region, which leads to a decrease in its expression level (Raducu, Baets et al. 2012). This is especially intriguing since this pathway is differentially regulated, but not commonly mutated in cancer cells. Fukutin and fukutin related protein (FKRP) are associated with the function of POMGnT1. Fukutin and FKRP are a pair of highly homologous proteins that contain DXD motifs characteristic to glycosyltransferases, but the exact activity of either protein has yet to be elucidated. Mutation of either of these proteins results in hypoglycosylation

23 9 of alpha-dg and subsequent CMD development (Kobayashi, Nakahori et al. 1998, Brockington, Blake et al. 2001, Beltran-Valero de Bernabe, Voit et al. 2004). Recent work has shown that fukutin is capable of directly interacting with POMGnT1 and modulating its ability to glycosylate DG (Xiong, Kobayashi et al. 2006). Of note, O- mannose phosphorylation of DG still occurs in a conditional knockout of fukutin (Beedle, Turner et al. 2012), Taken together, these data indicate that fukutin (and possibly FKRP) act in concert with POMGnT1 to modify the O-mannose moiety, however these proteins are not necessary for the O-mannose phosphorylation required for LARGE/LARGE2 activity. O-Mannose Phosphorylation In addition to modification with an unusual O-mannose moiety, dystroglycan also requires phosphorylation of this mannose for recognition by LARGE and subsequent glycan chain extension (Yoshida-Moriguchi, Yu et al. 2010). This phosphorylation depends upon a number of more recently identified proteins whose mutation causes one of the alpha-dystroglycanopathies. These proteins include the glycosyltransferase-like domain containing 2 (GTDC2), β 1,3-N-acetylgalactosaminyltransferase2 (B3GALNT2), and SGK196 (Yoshida-Moriguchi, Willer et al. 2013). It was found that GTDC2 exhibits the GlcNac β1,4 linkage activity that POMGnT1 does not have. This modification then receives N-acetylgalactosamine (GalNac) modification via a β3 linkage by B3GALNT2 activity. The product of these two enzymes is necessary for phosphorylation of the O- mannose residue. A previously uncharacterized kinase, SGK196 phosphorylates the O- mannose residue following the activity of GTDC2 and B3GALNT2 (Yoshida-Moriguchi, Willer et al. 2013). Like other enzymes, each of these proteins have been linked to alphadystroglycanopathies when mutated (Manzini, Tambunan et al. 2012, Jae, Raaben et al. 2013, Stevens, Carss et al. 2013).

24 10 Xylosyl-glucoronyl modification of the phospho-o-mannose Discovery of the functional importance of O-mannosyl phosphorylation on DG was soon followed by work showing requirement of the LARGE-mediated glycan built upon this structure (Yoshida-Moriguchi, Yu et al. 2010, Hara, Kanagawa et al. 2011). Both LARGE and LARGE2 share significant homology as bifunctional glycosyltransferases and they function in a similar manner. These enzymes both share xylosyltransferase and glucoronyltransferase activity with the only documented difference being ph optimum for optimal enzymatic activity (Inamori, Yoshida-Moriguchi et al. 2012, Inamori, Hara et al. 2013). This repeating disaccharide is the critical component for the ability of DG to bind its laminin-g domain containing extracellular ligands, specifically laminin and perlecan (Inamori, Yoshida-Moriguchi et al. 2012). LARGE and LARGE2 have both shown the capacity to rescue DG function in overexpression systems regardless of underlying mutations in other DG glycosyltransferase enzymes assuming a certain level of functional activity is maintained within the pathway(barresi, Michele et al. 2004, Fujimura, Sawaki et al. 2005). Though LARGE and LARGE2 differ in their respective contributions to disease. LARGE, like the other enzymes in this pathway, has been linked to alpha-dystroglycanopathies via its functional loss (Grewal, Holzfeind et al. 2001, Grewal and Hewitt 2002, Longman, Brockington et al. 2003), however this association has not been documented for LARGE2. This difference is likely explained by tissuespecific expression of LARGE with the highest levels in muscle, eye, and brain, consistent with many of the dystroglycanopathies phenotype (Grewal, McLaughlan et al. 2005). The lack of a dystrophy phenotype associated with LARGE2 has left it relatively under-investigated, however the more recent revelation that it serves as a critical enzyme for DG glycosylation in prostate epithelium points to its functional importance (Esser, Miller et al. 2013).

25 11 LARGE and potentially LARGE2 interact with a novel mediator of dystroglycan glycosylation, β3-n-acetylglucosaminyltranserase (B3GNT1). B3GNT1 was shown to coimmunoprecipitate with LARGE in CHO cells overexpressing tagged LARGE and B3GNT1, but this interaction was not seen in purified protein studies suggesting that the interaction may be mediated by an as-of-yet unidentified scaffold protein (Bao, Kobayashi et al. 2009). Overexpression of B3GNT1 is capable of rescuing DG hypoglycosylation in a prostate cancer cell line, but the function of B3GNT1 requires the LARGE or LARGE2 activity indicating a strong reliance of B3GNT1 on LARGE. Furthermore, B3GNT1 was recently identified as a causative mutation in a subset of WWS patients. While many questions remain regarding the function of B3GNT1, it is clear that its activity is required for dystroglycan glycosylation. Factors of unknown function An elegant set of experiments utilizing a haploid genetic approach revealed a number of genes affecting dystroglycan glycosylation (as measured by Lassa Virus entry) (Jae, Raaben et al. 2013). A number of novel genes were identified that could facilitate DGmediated Lassa Virus recognition, but only two of the previously unidentified genes have been implicated in families with MEB and WWS. The first, SGK196, has since been shown to mediate O-mannose phosphorylation. However, the second protein, TMEM5, currently has no attributed function and no predicted glycosyltransferase activity (Jae, Raaben et al. 2013). Many questions still exist concerning the functional role of a small handful of dystroglycanopathy-associated proteins, and since causative mutations still have not been identified in many WWS, MEB, and LGMD, this work highlights the need for better understanding of the entire pathway by which dystroglycan glycosylation is performed.

26 12 While all of these proteins have been shown to cause dystroglycan hypoglycosylation and subsequent muscular dystrophy, one of the biggest questions that still remains is the enzyme specificity and whether other protein targets are affected by their activity. Approximately 30% of total O-glycosylation in the brain is thought to be due to O- mannose (Chai, Yuen et al. 1999), but the relative prevalence of O-mannosyl glycans is unchanged in a conditional neural knockout of DG (Stalnaker, Aoki et al. 2011). This finding strongly suggests alternative protein targets for the O-mannosyl glycosylation pathway, but currently DG remains the only well-studied target. Dystroglycan Interactions in Muscle As part of the DGC, dystroglycan interacts with a number of proteins in skeletal muscle. The interactions with alpha-dg are primarily determined by the specific ECM components, and skeletal muscle is composed of a significant amount of laminin-211 making it the most common interactor with alpha-dg in muscle (Leivo and Engvall 1988). Mutational loss of laminin alpha-2 results in a muscular dystrophy phenotype highlighting the functional importance of the DGC interaction network for sarcolemma stability(helbling-leclerc, Zhang et al. 1995). Beta-DG interacts directly and indirectly with a host of proteins that makeup the rest of the DGC including dystrophin, the sarcoglycans (Ervasti and Campbell 1991), sarcospan (Crosbie, Heighway et al. 1997), and the syntrophins (Yang, Ibraghimov-Beskrovnaya et al. 1994). The interaction with dystrophin is mediated through the PPXY domain in the C-terminus of beta-dg and the WW domain of dystrophin (Ervasti and Campbell 1991). The stability of this interaction is increased due to the presence of two Ca 2+ -binding EF-hand domains located on dystrophin(chung and Campanelli 1999). Dystrophin binds to the actin cytoskeleton providing the final functional piece for dystroglycan to link the ECM to the cytoskeleton.

27 13 Beyond the canonical DGC, beta-dg interacts with a number of other proteins in muscle as well as other tissue. Beta-DG interacts with caveolin-3, a muscle-specific member of a family of proteins known to localize to the caveolae. This interaction is mediated by a WW-like domain identified on caveolin-3 and competes with dystrophin by binding at the same PPXY motif responsible for the dystrophin-dg interaction (Sotgia, Lee et al. 2000). Mutation of caveolin-3 leads to limb-girdle muscular dystrophy 1C indicating that its role in muscle, potentially through the DGC is required for homeostasis of the muscle (Minetti, Sotgia et al. 1998). Beta-Dg also interacts with growth factor receptor-bound protein 2 (Grb2) via Grb2 s SH3 domain (Yang, Jung et al. 1995). In bovine brain samples, DG was shown to complex with Grb2 and focal adhesion kinase (Cavaldesi, Macchia et al. 1999), though the implications of this interaction are still unclear. Other signaling and adaptor proteins can interact with beta-dg directly including the post-synaptic protein rapsyn (Cartaud, Coutant et al. 1998), dynamin (Zhan, Tremblay et al. 2005), and ezrin (Spence, Chen et al. 2004). Beta-DG mediated interaction with ezrin was shown to localize to microvilli and induce cytoskeletal rearrangement and filopodia formation (Spence, Chen et al. 2004, Zhan, Tremblay et al. 2005). The association between DG and filopodia formation has also been observed in differentiating oligodendroglia (Eyermann, Czaplinski et al. 2012). Additionally, beta-dg has been shown to interact directly with both the cytoskeletal interacting protein, plectin 1 (Rezniczek, Konieczny et al. 2007) and the adaptor protein, ankyrin-g which functions to localize dystroglycan to the costamere (Ayalon, Davis et al. 2008). Dystroglycan interaction with laminin is known to play a critical role in sarcolemmal stability during muscle contraction, but there is also indication that this interaction leads to pro-survival signaling through the PI3K/AKT pathway (Langenbach and Rando 2002). While not apparently interacting with beta-dg directly, similar signaling mechanisms exist that utilize the DGC syntrophin binding to Rac1 to promote pro-survival signaling

28 14 through JNK in a laminin dependent fashion. Therefore, multiple pathways to promote survival exist via a ligand-dependent outside-in signaling mechanism. It s been shown that binding of alpha-dg to extracellular ligands causes a phosphorylation of beta-dg at Tyr892, one of the residues involved in interaction with dystrophin (James, Nuttall et al. 2000, Ilsley, Sudol et al. 2001). This phosphorylation is mediated by the c-src kinase and is thought to facilitate a switch of dystroglycan interaction from WW domain containing proteins to the SH2/SH3 containing proteins (Sotgia, Lee et al. 2001). Therefore, this phosphorylation represents a potential mechanism whereby outside-in signaling is mediated through dystroglycan directly, though the phosphorylation site is not readily accessible by c-src when beta-dg is bound to dystrophin, thus this signaling may only be active in a small subset of biological circumstances. Dystroglycan Interactions in Epithelium The binding partners and associated function of dystroglycan in the epithelial compartment remain poorly understood. However, a small number of studies have investigated the signaling impact and potential interactions within the epithelium. DGC composition in epithelium is distinct from muscle organization, but DG still maintains interaction with syntrophin and dystrobrevin mediated by utrophin binding in Madin- Darby kidney cells (MDCK) (Kachinsky, Froehner et al. 1999). Utrophin is a homologue of dystrophin that is more widely expressed across a variety of cell types. Sarcospan and sarcoglycan are expressed in a more limited subgroup of epithelia, but they do not appear to associate with DG here (Durbeej and Campbell 1999). A key point, therefore, is that the signaling complex is not fully maintained between muscle and epithelium, and the individual pathways investigated in muscle cannot assume to be active in epithelium.

29 15 Following are a number of studies using both in vitro and in vivo analyses of DG function in a variety of cell types. One of the earliest characterizations of DG outside of the muscle arose following the discovery of DG s requirement in formation of Reichert s membrane (Williamson, Henry et al. 1997). Embryonic stem cell clones generated during creation of the DG-null mouse show significant aberrations in their ability to bind and organize laminin on the extracellular surface (Henry and Campbell 1998). Additionally, embryoid bodies devoid of dystroglycan are incapable of forming an intact basement membrane. DG expression is thus required to organize its extracellular ligands in an early developmental state supporting DG s critical role in basement membrane formation. Beyond organizing its ligands, DG is also required for appropriate polarization of cells. DG knockout in mammary epithelial cells reduces laminin binding capacity, and in 3-D culture, DG-null cells are incapable of appropriately polarizing when induced by addition of laminin-111 (Weir, Oppizzi et al. 2006). The requirement for DG in polarization is also seen in studies in drosophila where loss of DG disrupts apical-basal polarity in epithelium and anteroposterior polarity of the developing oocyte (Deng, Schneider et al. 2003). Furthermore, this polarization requires binding between dystroglycan and drosophila dystrophin as well as the expression of perlecan (Schneider, Khalil et al. 2006, Yatsenko, Kucherenko et al. 2009). The polarization of the cell may partially depend on the inside-out signaling of PAR-1 to regulate laminin assembly of DG (Masuda-Hirata, Suzuki et al. 2009). Therefore, DG functions to not only organize the extracellular matrix but also provide the foundation for the signaling necessary for induction of cell polarity. Extending beyond cell polarity, dystroglycan also signals by inhibiting pro-growth signaling pathways. DG is known to negatively modulate laminin-induced activation of ERK signaling through the α6β1 integrin likely by sequestering ERK (and MEK) preventing the activation of the downstream pathways (Ferletta, Kikkawa et al. 2003,

30 16 Spence, Dhillon et al. 2004). It is proposed that this mechanism could rely on differential localization of the two components, ERK to focal adhesion and MEK to membrane ruffles, to prevent activation of the pathway. Associated with DG s role regulating pro-growth signaling, studies in murine alveolar epithelial cells have shown that DG can modulate stretch-induced signaling (Budinger, Urich et al. 2008). Upon stretching the alveolar cells signal to phosphorylate acetyl-coa carboxylase through AMP-activated kinase (AMPK) activity. This requires DG and its interaction with plectin, a protein that is known to interact directly with both DG and AMPK (Gregor, Zeold et al. 2006, Rezniczek, Konieczny et al. 2007, Takawira, Budinger et al. 2011). A functional link pathway of DG-mediated AMPK activity is still unclear, but it is intriguing that DG is placed in a functional role within the epithelium that allows it to inhibit pro-growth MEK/ERK signaling and promote anti-proliferative AMPK signaling. In addition to mediating signals through direct protein binding, beta-dg is cleaved and produces a 31 kda fragment that is seen by gel electrophoresis thought to be a byproduct of MMP activity (Yamada, Saito et al. 2001, Singh, Itahana et al. 2004). This cleavage can modulate beta-dg availability for binding to alpha-dg. Furthermore, with the identified NLS on beta-dg and the studies showing its nuclear localization (Oppizzi, Akhavan et al. 2008, Mathew, Mitchell et al. 2013), its possible that this cleavage could function to liberate DG for nuclear activity, though the functional implication of nuclear localization is still muddled. A DG function at the organismal level was identified utilizing a temperature preference assay in Drosophila (Takeuchi, Nakano et al. 2009). Loss of DG in Drosophila causes the insects to prefer colder temperature secondary to increased basal metabolic rate, cellular ATP concentration, and pyruvate dehydrogenase activity. The researchers proposed a model whereby cellular stability is disrupted when DG is absent leading to increased permeability of the membrane and increased intracellular Ca 2+ and

31 17 subsequent activation of mitochondrial ATP output. These findings again propose a potential link between DG expression and potential implications for cell growth and stability. In determining the functional role of a protein, targeted knockouts of genes in animal model remain a gold standard. While the whole body, neural specific, and muscle specific knockouts produced the expected muscular dystrophy-associated phenotypes, there are now tissue specific knockouts of dystroglycan using the Cre-Lox expression system in the kidney, prostate, and mammary epithelium. These models highlight the most direct analysis of DG-associated signaling in epithelial compartments to date. Prostate specific knockout of dystroglycan using a probasin driven Cre to disrupt DAG1 in the luminal cells of the murine prostate yielded no discernible phenotype (Esser, Cohen et al. 2010). Glandular architecture, organization of the DG ligands, the ultrastructure of the basement membrane, cell polarity, and even regenerative capacity following castration were all maintained in the knockout mouse. The lack of pathology in this model was unexpected, but the finding that DG was upregulated in regenerated prostates suggested a potential role of the basal cell expression of DG to compensate for the luminal cell knockout. A follow up to these experiments was published as part of AK Esser s thesis, but knockout of DG in the basal cell population yielded a similar finding suggesting that DG is not required for the maintenance and homeostasis of the murine prostate. DG knockout in the kidney was accomplished with an array of Cre expression systems to differentiate DG-required functions in the different renal compartments. Regardless of the promoter driving Cre, these animals exhibited only mild to absent phenotypic changes. The glomerular basement membrane is slightly thickened during podocyte specific deletion, but urinary symptoms were limited to mild proteinuria without detectable levels of albumin indicating a mostly intact glomerular filter. Additionally, the mouse was subjected to renal injury with LPS and found that DG was

32 18 not required for appropriate repair of the tissue (Jarad, Pippin et al. 2011). Both the renal and prostate findings suggest that DG does not play a critical role in tissue homeostasis or repair. These findings suggest that other ECM-interacting proteins such as the integrins can compensate for the lack of dystroglycan in this setting. The only epithelium specific knockout to show a more marked phenotype is the mammary knockout which utilized a keratin14 driven Cre expression system. DG knockout inhibits epithelial outgrowth and lactation post-pregnancy, but does not alter normal tissue architecture or basement membrane formation. The lack of ECM dysregulation is similar to the findings from the prostate and renal DG knockouts. The authors recognized the similarity to a STAT5 knockout, and they showed that DG is required for STAT5 signaling, but this requirement is dependent on the extracellular alpha-dg interactions (Leonoudakis, Singh et al. 2010). These findings suggest STAT5- associated signaling functions in a form of extracellular cross-talk between DG and the hormone receptors for prolactin and growth hormone. While this phenotype was robust, the specific setting required to elicit the findings suggest that DG s function is only critical given the appropriate condition and/or genetic background. Dystroglycan Expression and Glycosylation in Cancer Disruption of DG affects growth-signaling pathways, cellular membrane stability, metabolic activity, cell polarity, and stability of the cell-ecm interaction. Therefore, DG is positioned to critically regulate components that are commonly disrupted in the development of cancer. While cancer studies have traditionally focused more on the integrin family, it is now clear the DG expression and glycosylation are frequently perturbed in a wide variety of tumor types. DG was originally shown to be significantly downregulated in breast and prostate cancer with the greatest dysregulation found in the higher-grade tumors (Henry, Cohen et

33 19 al. 2001). These findings have since been recapitulated in other studies in breast, colon, pancreatic, gastric, and renal adenocarcinoma (Sgambato, Migaldi et al. 2003, Sgambato, Camerini et al. 2007, Moon, Rha et al. 2009, Jiang, Rieder et al. 2011, Shen, Xu et al. 2012). These findings have also been seen in squamous cell carcinomas originating from the oral cavity, cervix, and vulva (Sgambato, Tarquini et al. 2006, Sgambato, Caredda et al. 2010). Even cancers as divers as pediatric rhabdomyosarcoma and glioma (Martin, Glass et al. 2007) as well as adult gliomas (Calogero, Pavoni et al. 2006) demonstrate dysregulatd DG function. Most of the studies investigating DG function utilize either the IIH6 or VIA41 antialpha-dg antibodies that recognize the extensively glycosylated DG, therefore it is difficult to accurately determine whether the underlying cause for loss of staining is due to downregulated DG expression, decreased alpha-dg glycosylation, or proteolytic cleavage of the alpha-dg subunit. A comprehensive analysis of beta-dg expression showed that DG expression is commonly downregulated in breast, esophageal, and colorectal adenocarcinomas (Cross, Lippitt et al. 2008). These findings are balanced by studies that have shown little to no downregulation of the core DG but show significant loss of the functional DG glycan (Martin, Glass et al. 2007, de Bernabe, Inamori et al. 2009, Sgambato, Camerini et al. 2010), therefore both changes in expression of the core protein as well as dysregulation of the glycan moiety may produce a loss of immunoreactivity of cancer samples. The loss of DG immunostaining is associated with not only disease grade, but it also demonstrates predictive prognostic value in a number of studies. Overall survival is inversely associated with DG staining in renal adenocarcinoma, gastric tumors, oral cancers, colon, and breast adenocarcinomas (Sgambato, Migaldi et al. 2003, Sgambato, Camerini et al. 2007, Moon, Rha et al. 2009, Sgambato, Caredda et al. 2010). One study has demonstrated that specific loss of alpha-dg immunoreactivity, but not beta-dg, inversely associates with survival and may serve as a novel biomarker for gastric

34 20 carcinoma (Shen, Xu et al. 2012). Correspondingly, investigation of DG as a prognostic biomarker is still being pursued by a number of laboratories. Dysregulation of DG not only associates with increased disease, but its disruption in multiple in vitro studies causes increased motility, proliferation, and metastatic potential. Loss of DG glycosylation in breast cancer cell lines is a frequent event that associates with more aggressive disease. Cell proliferation and tumor formation in vivo are diminished with restoration of the laminin binding glycan (Akhavan, Griffith et al. 2012). Similar findings have shown restoration of this glycan also affects laminin anchoring and migration potential in a transwell assay in both breast and prostate cancer cells (Bao, Kobayashi et al. 2009, de Bernabe, Inamori et al. 2009, Esser, Miller et al. 2013). Additionally, overexpression of DG in prostate cancer cells decreases cell growth in a three-dimensional soft agar growth assay, but this was associated with an increase in cell motility in an invasion assay (Mitchell, Mathew et al. 2013) arguing that the role of dystroglycan and its glycosylation affect classical metastatic phenotypes, though not in a clearly defined manner. In determining what causes the functional loss of DG, a number of studies have revealed that only a small handful of glycosyltransferases appear to mediate hypoglycosylation. Several studies have identified LARGE as the most likely candidate mediating changes in the functional DG glycan in breast cancer cells (de Bernabe, Inamori et al. 2009, Akhavan, Griffith et al. 2012). Similar, studies in prostate cells illustrate that B3GNT1, with concomitant expression of LARGE, is required for appropriate expression of the ligand-binding DG ligand (Bao, Kobayashi et al. 2009). Therefore, more studies are needed to continue to investigate the causative changes in DG-glycosylation and/or expression that generate the observed changes in tumorassociated changes in DG function.

35 21 Relevance and Focus The cell-ecm interaction is frequently perturbed in cancer. Loss of the interaction is associated with decreased cellular polarity, increased cancer invasiveness, and poorer patient outcomes. DG is a critical component that mediates the appropriate cell-ecm interaction, and DG is frequently perturbed in cancer. Its dysfunction is frequently associated with decreased overall survival and more aggressive disease characteristics. Intriguingly, functional disruption of DG is seen in a wide variety of tumor types from multiple organ sources. The breadth of studies that show DG-associated changes in tumors suggest that loss of DG function is a near-universal event adenocarcinoma tumor progression. Only a small number of studies have identified causative changes in the glycosylation pathway that may explain the changes observed in tumor specimens. There is still a dearth of data indicating the signaling impact that DG disruption causes. Therefore, the focus of my thesis work was to determine what underlying changes in the glycosyltransferases mediate functional disruption of DG. Additionally, my studies have interrogated the signaling pathways affected by DG expression and glycosylation. The long-term goal of this project is to better characterize the implications of reduced DG immunoreactivity in tumor specimens. The results of my work indicate the variety of underlying molecular changes that occur in the DG-glycosylation pathway, and could be used to guide therapeutic intervention for some cancers.

36 22 CHAPTER II LOSS OF LARGE2 DISRUPTS FUNCTIONAL GLYCOSYLATION OF α- DYSTROGLYCAN IN PROSTATE CANCER Introduction Dystroglycan (DG) is an extracellular matrix receptor discovered as a component of the dystrophin glycoprotein complex in muscle (Ervasti, Ohlendieck et al. 1990, Ervasti and Campbell 1991). DG is expressed in many tissue types including muscle, neuronal, adipose and epithelial tissues. Transcribed from a single gene, DG is post-translationally cleaved into two non-covalently attached subunits (Ibraghimov-Beskrovnaya, Ervasti et al. 1992, Ibraghimov-Beskrovnaya, Milatovich et al. 1993). The transmembrane βdg subunit binds to dystrophin or utrophin, whereas the extracellular αdg subunit interacts with laminin-g domain-containing matrix proteins including agrin, perlecan, and members of the laminin family (Ibraghimov-Beskrovnaya, Ervasti et al. 1992, Ervasti and Campbell 1993, Bowe, Deyst et al. 1994, Campanelli, Roberds et al. 1994, Gee, Montanaro et al. 1994). DG thus serves as a link between the cytoskeleton and extracellular matrix. αdg is heavily glycosylated with both O- and N-linked carbohydrates, and much attention has focused on an O-mannosyl glycan that is required for ligand binding. Production of this laminin-binding glycan involves at least six known and putative glycosyltransferases (POMT1, POMT2, POMGnT1, Fukutin, FKRP and LARGE) (Aravind and Koonin 1999, Takahashi, Honda et al. 2000, Brockington, Blake et al. 2001, Grewal, Holzfeind et al. 2001, Hayashi, Ogawa et al. 2001, Manya, Chiba et al. 2004). Mutations in these enzymes result in hypoglycosylation of αdg and prevent it from binding its ligands (Michele, Barresi et al. 2002, Kanagawa, Saito et al. 2004, Kanagawa, Michele et al. 2005, Hara, Kanagawa et al. 2011). Many of these

37 23 glycosyltransferases have been implicated in diseases of muscle and neural tissue, in mice and in humans (Kobayashi, Nakahori et al. 1998, Brockington, Blake et al. 2001, Grewal, Holzfeind et al. 2001, Yoshida, Kobayashi et al. 2001, Beltran-Valero de Bernabe, Voit et al. 2004). Recently, the laminin-binding moiety of αdg was shown to be a unique phosphorylated O-mannosyl glycan (Yoshida-Moriguchi, Yu et al. 2010). This modification is mediated by the xylosyl-glucuronyltransferase LARGE (Hara, Kanagawa et al. 2011, Inamori, Yoshida-Moriguchi et al. 2012). Notably, the overexpression of LARGE has been shown to functionally rescue αdg hypoglycosylation in cells from patients harboring mutations in other αdg glycosyltransferases (Barresi, Michele et al. 2004). In addition, a mutation in LARGE is responsible for the myodystrophy mouse (Large myd, MDC1D) phenotype, features of which include αdg hypoglycosylation (Grewal and Hewitt 2002). Although LARGE is expressed in many tissues the myodystrophy mouse phenotype is primarily evident in muscle and brain. A paralog of LARGE, LARGE2, has a narrower expression profile and is not expressed in muscle and brain. LARGE2 overexpression results in αdg hyperglycosylation and enhanced laminin binding, suggesting that LARGE and LARGE2 may be functionally redundant (Fujimura, Sawaki et al. 2005, Grewal, McLaughlan et al. 2005). DG is expressed in many epithelial tissues where it is thought to mediate cell-matrix interactions important for cell polarity and tissue morphogenesis, but recent studies have shown that DG is not required for the normal development and/or maintenance of some epithelial tissues, including prostate (Durbeej, Larsson et al. 1995, Deng, Schneider et al. 2003, Weir, Oppizzi et al. 2006, Esser, Cohen et al. 2010, Jarad, Pippin et al. 2011). Loss of DG function has been implicated in various cancer cell phenotypes including increased growth in soft agar, invasive behavior in 2- and 3-dimensional culture systems, and xenograft tumor growth (Muschler, Levy et al. 2002, Sgambato, Camerini et al. 2004, Bao, Kobayashi et al. 2009). It is proposed that DG is involved in these processes by

38 24 participating in adhesive interactions with the extracellular matrix and/or by affecting intracellular signaling (Higginson and Winder 2005, Bao, Kobayashi et al. 2009). In a variety of adenocarcinomas, including prostate, expression of DG protein, αdg and/or βdg, is reduced and this is associated with increased tumor aggressiveness and loss of extracellular matrix integrity, e.g. (Henry, Cohen et al. 2001, Sgambato, Migaldi et al. 2003). So far, all evidence points toward post-transcriptional mechanisms, including proteolysis and hypoglycosylation to account for the observed loss of DG expression in cancer (Sgambato, Migaldi et al. 2003, Singh, Itahana et al. 2004, Shang, Ethunandan et al. 2008, Bao, Kobayashi et al. 2009, de Bernabe, Inamori et al. 2009). Herein we describe a novel mechanism accounting for the loss of functional αdg glycosylation in prostate cancer cell lines and prostate cancer specimens reduced expression of LARGE2, a protein not previously implicated in cancer. Furthermore, we show that endogenous LARGE2, and not LARGE, is responsible for mediating the functional glycosylation of αdg in prostate cancer cells and normal prostatic epithelium. Materials and Methods Immunohistochemistry Prostate tissue embedded in paraffin was sectioned at 7 µm, deparaffinized, and rehydrated. For the metastatic tissue microarray, samples were first baked at 70 C for 64 hours prior to antigen retrieval. Antigen retrieval for all samples was performed by a 20- min exposure to either citrate buffer (βdg) or proteinase K (αdg). Endogenous peroxidases were quenched with 3% hydrogen peroxide for 10 min. Sections were blocked in 10% horse serum and then incubated overnight with primary antibody, either IIH6 (1:100, Santa Cruz) or 8D5 (1:100, Novocastra). Samples were washed and incubated for 1 hour at room temp with a biotinylated secondary antibody which was then

39 25 detected using the ABC reagent (Vector) and DAB plus (Dako) followed by counterstain with hematoxylin. Hematoxylin and eosin (H and E) staining was performed using standard procedures. Immunohistochemistry was performed by the UI Department of Pathology Core Lab. For evaluation of immunohistochemistry, pathologists (MMM and MBC) were blinded to sample identity. For metastases analysis, samples were scored as either positive or negative for membranous staining with the antibody. Cell Culture 22Rv1 (American Type Culture Collection, ATCC), PC-3 (ATCC), LNCaP (M Cohen), PC-3E+ and TEM4-18 (PC-3 derivatives, Henry Lab) and GP2-293 packaging cells (BD Biosciences) were grown with 10% FBS (HyClone, Logan, Utah) and 1% nonessential amino acids. Cultures were maintained at 37 o C in 95% air and 5% CO 2 and subcultured as needed. shrna Knockdown of LARGE2 In order to knockdown LARGE2, plko1 lentiviral vectors targeting human LARGE2 were purchased from Open Biosystems (TRC collection, RHS4533- NM_152312). Five different constructs were transfected into GP2-293 cells along with the lentiviral second generation plasmid VSV-G (kind gift from the Trono Lab) and culture supernatant from these cells was used to infect PC-3E+, DU145, and 22Rv1 cells for 8 hours followed by puromycin selection (1.0 µg/ml). We were able to achieve 70%- 80% knockdown of LARGE2 using hairpin construct sequence TRCN and TRCN

40 26 Overexpression of LARGE2 Murine LARGE2 was excised from a mlarge2 containing plex vector using NotI and cloned into the pqcxip NotI site. Cells were selected and maintained with standard media containing 1.0 ug/ml puromycin. qpcr Analysis qpcr analysis was performed as described previously (Svensson, Barnes et al. 2007). For the analysis of bulk tumor samples, the TissueScan cdna array (HRT103, Origene) was used. Relative expression values were calculated using the comparative Ct method (Pfaffl 2001). Western Blot Protein lysates were prepared from sub-confluent cultures following scrape harvesting in RIPA buffer (150 mm NaCl 2, 1% NP40, 10 mm deoxycholate-na, 0.1% SDS, 50 mm Tris, ph 8.0; rocked at 4 C for 1 h, boiled 5 min) with protease inhibitors. Protein samples (50 µg, Lowery protein analysis) were electrophoresed (4-20% Tris-HCl gels) and transferred (PVDF) prior to antibody incubation. Antibodies were used as follows: 1 -IIH6, glycosylation sensitive αdg, 1:500 overnight at 4 C in low salt buffer (50mM Tris, 100mM NaCl2,0.1% Tween 20), 2 -donkeyαmouse IgM HRP, 1:2000 for 1h at room temperature (RT); 1 -sheep 5 core αdg, 1:100 overnight at 4 C, 2 donkeyαmouse IgG HRP, 1:2000 for 1h at RT; 1 -βdg specific 8D5 (Novocastra/Leica) for 1:200 overnight at 4 C, 2 donkey α-mouse IgG, 1:2000 for 1hr at RT; 1 -β-actin (Sigma), 1:5000 for 1h at RT, 2 -goatαrabbit IgG HRP, 1:10,000 for 1h at RT, was

41 27 utilized as a marker for protein loading. All secondary antibodies were obtained from Jackson Immunoresearch. Flow Cytometry Analysis Cells grown to 60-80% confluency were detached (Versene, Invitrogen) and washed prior to antibody incubation. The cells were then resuspended at 5 x 10 5 cells in 50ul FACS buffer (PBS % sodium azide + 5% BSA) + IIH6 antibody (1:100) for 45 min on ice. Next, cells were washed and incubated with secondary antibody (100µl FACS buffer plus 1:100, donkey anti-mouse IgM Dylight 488, Jackson Immunoresearch) for 30 min on ice in the dark. Cells were washed again and resuspended in 400µl of FACs buffer, and transferred to polystyrene FACs tubes (BD Biosciences). Samples were analyzed using the Becton Dickinson LSR II flow cytometer at The University of Iowa Flow Cytometry Core Facility. Laminin-111 Binding The methods were used as described above but cells were incubated with 50 µl 100 nm laminin-111 (BD Biosciences) prior to incubation with anti-laminin-111 primary antibody (1:100, Sigma) and the subsequent detection with the DyLight488 secondary antibody. To inhibit laminin-111 binding, a subset of samples was first incubated with αdg IIH6 antibody at 1:100 in 100µl FACS+ buffer. All steps were done on ice. Immunofluorescence Tissues were processed and antigen retrieval performed as described above. IIH6 was incubated overnight at 1:00. Tissues were washed in blocking buffer and incubated with

42 28 donkey anti-mouse IgG Cy3 (Jackson Immunoresearch) at 1:200 and counterstained with DAPI (1:5000, Sigma) for 1hr at RT. Transwell Migration Asssay 24 well, 8.0 µm transwell permeable inserts (Corning Life Sciences) were prepared as previously described (Drake, Barnes et al. 2010). Briefly, transwells were coated with laminin-511 for 1 hour at room temperature prior to washing with PBS. Luciferase expressing cells were plated at 8.0 x 10 5 cells/well in serum-free DMEM/F12 and allowed to migrate down a chemoattractant gradient (DMEM/F12 supplemented with 10% FBS in lower chamber) for 18 hours prior to trypsinization followed by quantification via bioluminescence. Data is presented as migration relative to vector control cells across an uncoated membrane. Matrigel Invasion Assay 24 well, 8.0 µm BD Biocoat matrigel inserts (BD Biosciences) were thawed and equilibrated according to the manufacturer. Cells were plated in the top chamber at 8.0 x 10 5 cells/well in serum-free DMEM/F12. The lower chamber was filled with DMEM/F12 containing 10% serum to induce a chemoattractive response. Following 18 hours of incubation, cells were trypsinized and quantified as described above. Data is presented as invasion relative to vector control cells. Growth Assay 30,000 GS689.Li cells containing either empty vector or mlarge2were plated in triplicate into 6-well dishes. On the indicated day, cells were trypsinized and counted

43 29 using a Z2 series Coulter Counter (Beckman Coulter) with a cell diameter range of 8-24 µm. Three counts were averaged per well. Results αdg Glycosylation and βdg Expression In Both Primary And Metastatic Prostate Cancer We previously demonstrated that βdg expression is reduced in prostate cancer, a finding which has been extended to loss of αdg immunostaining by other groups (Henry, Cohen et al. 2001, Sgambato, De Paola et al. 2007, Shimojo, Kobayashi et al. 2011). These studies utilized glycosylation-sensitive antibodies IIH6 and VIA41, which detect the laminin-binding glycan on αdg (Ervasti and Campbell 1993). However, it remains unclear to what extent the loss of DG immunoreactivity reflects a reduction of the core protein expression or a loss of the epitopes on αdg recognized by IIH6 and VIA41. This remains a difficult question to answer due to the lack of an antiserum that reacts specifically with the core αdg protein in histochemical preparations. Therefore, we utilized a prostate cancer tissue microarray panel to assess both DG subunits on adjacent sections. Fig. 2-1A shows representative staining of sections from Gleason grade 3 and 4 patterns observed in tissue microarrays. Interestingly, we noted examples of some grade 3 glands (Figure 2-1A) in which IIH6 staining (glycosylated DG) was reduced at the same time that 8D5 staining (βdg) was normal, i.e. levels were similar to the adjacent, healthy glands. This strongly suggested that the glycosylation of αdg is affected independently of core DG protein expression. To quantify DG staining intensity in these sections, we developed a quartile scoring system (Figure 2-1B) grading each specimen with both a primary and secondary score, similar to the Gleason scoring system where a Gleason score (range: 2-10) is generated by summing the most prevalent and

44 30 second-most prevalent Gleason grades (range: 1-5). The Gleason grades are generated based upon the histological architecture of the tumor as originally defined by Donald Gleason in 1974 and refined in 2005 by the International Society of Urological Pathology (Gleason and Mellinger 1974, Epstein, Allsbrook et al. 2005). For both IIH6 and 8D5, staining intensity was reduced across cases with Gleason scores 6-8, with IIH6 staining more dramatically reduced than 8D5 staining (Figure 2-1C). The observed reduction of both IIH6 (p<0.0022) and 8D5 (p<0.0003) staining relative to that in benign glands was significant, and both markers are independent predictors of Gleason score. We next assessed DG status in a panel of prostate cancer metastases to determine if the expression and glycosylation patterns differ from those observed in primary tumors. Because prostate cancer metastases do not retain the glandular architecture seen in the prostate, samples were graded according to a positive/negative system; a signal was evaluated as positive if membranous staining was present on the tumor cell(s). Strikingly, IIH6 staining was present in less than 1% (1/127) of cases, whereas 8D5 reactivity was present in 47.6% (59/124) of cases (Figure 2 2A-D). This indicates that during metastatic progression, αdg glycosylation is more frequently perturbed than is expression of the core protein. Staining for both IIH6 and 8D5 was clearly evident in the positive control (healthy kidney glomeruli) (Figure 2 2E-F). Having shown that DG glycosylation is more frequently lost in metastatic disease, we attempted to determine if IIH6 staining is an independent prognostic indicator of disease recurrence in a cohort of 135 prostate cancer patients seen at University of Iowa Hospitals and Clinics during the years Patient characteristics are listed in Table 2-1. Sections from prostatectomy specimens were stained with IIH6 and scored by the method described above. We first demonstrated that DG glycosylation was inversely correlated with Gleason score (p=0.0367) consistent with our finding from the tissue microarray. We then assessed whether DG glycosylation status correlates with either mortality or biochemical recurrence. With only 9 patients in this cohort having died from

45 31 prostate cancer, despite having a mean follow-up time of 12.3 years, we were unable to draw any conclusion regarding DG status and mortality. However, we did detect a modest, though not statistically significant, association between DG hypoglycosylation and biochemical recurrence, using both univariate and multivariate regression analysis (Figure 2-3). Functional Glycosylation Of αdg Across Prostate Cancer Cell Lines Is Heterogeneous. To determine the status of DG expression in prostate cancer cell lines, we examined DG expression by western blot analysis using an antiserum that detects the αdg core protein. αdg protein was expressed in all prostate cancer cell lines examined (Figure 2-4A). The observed molecular weight of the bands was within the range of ~80 to 150kDa, with 22Rv1 and PC-3 expressing higher molecular weight species than LNCaP and TEM4-18. This heterogeneous staining pattern is consistent with the fact that DG has a number of carbohydrate modifications, only some of which are involved in ligand binding and recognized by IIH6 (Ervasti and Campbell 1991, Ervasti and Campbell 1993, Yoshida-Moriguchi, Yu et al. 2010, Hara, Kanagawa et al. 2011, Inamori, Yoshida- Moriguchi et al. 2012). The expression of βdg protein (8D5 immunoreactivity) was equivalent across the cell lines (Figure 2-4A). These data indicate that both αdg and βdg are expressed in the prostate cancer cell lines examined. Because ligand binding requires proper glycosylation of cell-surface localized αdg, we carried out flow cytometry analysis with the IIH6 antibody; this revealed that αdg glycosylation is heterogeneous across the prostate cancer cell lines (Figure 2-4B). Among the cell lines examined, 22Rv1 and PC-3E+ (E-cadherin-positive (Drake, Strohbehn et al. 2009)) were positive for IIH6 staining, whereas PC-3 (obtained from the ATCC) was composed of distinct IIH6-positive (IIH6+) and IIH6-negative (IIH6-) subpopulations. The TEM4-18 and LNCaP cell lines were IIH6-, consistent with the

46 32 western blot data analyzing glycosylated DG. Interestingly, cells of the TEM4-18 line, an aggressive subpopulation of the PC-3 line that were isolated following migration through an endothelial cell monolayer, are IIH6- (Drake, Strohbehn et al. 2009). In contrast, cells of the PC-3E+ line, a less aggressive subpopulation also derived from the PC-3 line, are IIH6+. Thus, in PC-3 cells, loss of DG glycosylation is associated with a more aggressive phenotype. We next assessed whether the glycosylation status of αdg correlates with laminin binding ability using a flow cytometry-based ligand binding assay. IIH6+ 22Rv1, PC-3, and PC-3E+ cells were able to bind exogenous laminin-111 at the cell surface whereas the IIH6- LNCaP and TEM4-18 cells lines were not (Figure 2-4C). This laminin binding was specific to DG as it was inhibited by prior incubation with IIH6. Some laminin immunoreactivity was also detected in cells in the absence of exogenous laminin-111, likely representing endogenous laminin bound during tissue culture. Overall, these results are consistent with those of IIH6 immunoreactivity, indicating that functional αdg glycosylation in metastatic prostate cancer cell lines is variable. Functional Glycosylation of αdg Correlates with Expression of the LARGE2 mrna To elucidate the molecular mechanism responsible for the loss of functional αdg glycosylation in prostate cancer cells, we compared mrna expression levels of the six known and putative αdg glycosyltransferases in the PC-3 derived cell lines PC-3E+ (IIH6+) and TEM4-18 (IIH6-). Given that the glycosyltransferase β3gnt1 has also been implicated in LARGE function, it was also included in the glycosyltransferase panel (Bao, Kobayashi et al. 2009). Among these enzymes, only LARGE2 exhibited a dramatic reduction in expression: mrna levels were 100-fold lower in TEM4-18 relative to PC- 3E+ cells (Figure 2-5A). We next compared levels of the LARGE2 mrna across the prostate cancer cell line panel to those in PC-3 cells (Figure 2-5B). Notably, they

47 33 generally correlated with the IIH6 immunoreactivity of each cell line. The exception was the LNCAP cell line, suggesting that an alternate mechanism is responsible for the reduction of αdg glycosylation in LNCaP cells. LARGE2 Functionally Glycosylates αdg in the Prostate LARGE is required for the proper glycosylation of αdg in skeletal muscle and neural tissue. However, prior studies indicated that in human prostate tissue, the expression of LARGE2 mrna was significantly greater than that of LARGE (Grewal, McLaughlan et al. 2005). Therefore, we next evaluated αdg glycosylation status of prostate tissue from the myodystrophy (myd) mouse (Large myd, MDC1D), which harbors a loss-of-function mutation in LARGE. Immunofluorescence staining with the IIH6 antibody in a wild-type mouse revealed positive staining in the muscle surrounding the urethra and on the basal side of the prostate epithelium (Figure 2-5C). As expected, functional αdg glycosylation was reduced in the muscle tissue of the myd mouse.. However it was detectable in the prostate epithelium indicating that LARGE is not required for αdg glycosylation in the prostate. Tissue from a prostate-specific DG knockout mouse (Esser, Cohen et al. 2010) is shown as a control; as expected, IIH6 staining is maintained in the muscle but lost in the prostate gland. To assess whether endogenous LARGE2 functionally glycosylates αdg, we generated PC-3E+ (IIH6+) cells that express LARGE2 targeted shrnas. qrt-pcr analysis revealed that two separate shrnas were capable of producing >70% reduction of LARGE2 mrna expression (Figure 2-6A) with no significant effect on LARGE mrna expression (Figure 2-6B). Staining with IIH6 revealed that αdg glycosylation was reduced by 77% with expression of shrna #147 and 83% with expression of shrna #148 (Figure 2-7A, comparison is to a vector-transfect control). To confirm that ligand binding was reduced in the knockdown cells, we assessed laminin-111 binding by

48 34 flow cytometry. As expected, laminin binding in the shlarge2 #147 was significantly reduced relative to the control line (Figure 2-7B); moreover, it was nearly absent in the shlarge2 #148 cell line. Prior incubation with the IIH6 antibody inhibited laminin-1 binding in all cases demonstrating that binding was specific to DG. We further validated the requirement for LARGE2 by knocking-down the protein in two additional IIH6+ cell lines, 22Rv1 and DU145. Flow cytometry revealed that IIH6 immunoreactivity was significantly reduced (Figure 2-6C-D). Thus, LARGE2 is required for the functional glycosylation of αdg in these cell lines. To examine whether reduced LARGE2 expression accounts for αdg hypoglycosylation in other epithelial cancers, we assessed the glycosylation status of αdg and the expression levels of glycosyltransferase mrna in breast, colon and pancreatic cancer cell lines. αdg glycosylation was heterogeneous across the cell lines in the panel (Figure 2-8A), and levels of the LARGE2 mrna correlated with αdg glycosylation status. Of these cancer lines, the breast cancer line ZR75-1 and the colon cancer line HT-29, in which αdg was highly IIH6+, expressed the highest levels of LARGE2, though less than the reference PC-3 cell line; the pancreatic cancer lines BxPC3 and PANC-1 expressed LARGE2 at very low levels. Similarly, the IIH6- MDA.MB231 breast cancer cells expressed very low levels of LARGE2 mrna (Figure 2-8B). Notably, expression of the LARGE mrna expression was also inversely correlated with αdg glycosylation status in the HT-29 and ZR75-1 lines, consistent with prior studies in breast cancer cells (de Bernabe, Inamori et al. 2009); expression was very low in the BxPC3, PANC-1, and MDA-MB-231 (Figure 2-8C). Therefore, the utilization of LARGE or LARGE2 for αdg functional glycosylation is likely to depend on tissue type.

49 35 LARGE2 Overexpression Restores DG Function And Diminishes Invasion And Cell Proliferation Potential Overexpression of LARGE2 has been shown to functionally glycosylate αdg (Fujimura, Sawaki et al. 2005). To assess whether we could revert the hypoglycosylation phenotype in the IIH6- cell lines TEM4-18 and LNCaP, we stably expressed murine LARGE2 in each. Flow cytometry revealed that LARGE2 overexpression indeed led to a prominent, homogenous increase in IIH6 immunoreactivity within the LNCaP cells (Figure 2-7C). The TEM4-18 cells exhibited two distinct subpopulations of IIH6 high and low staining cells with a reduction of IIH6+ population during serial passage suggesting that rescue of LARGE2 may not be stable in these cells (Figure 2-6E). Therefore we overexpressed mlarge2 in a cell line similar to TEM4-18, GS689.Li, that was derived from two in vivo passages in mice as previously described (Drake, Strohbehn et al. 2009). Overexpression of mlarge2 in these cells produced a significant increase in IIH6 reactivity (Figure 2-7C). Furthermore, expression of exogenous mlarge2 produced a significant increase in the laminin binding capacity in both the LNCaP and GS689 lines (p<0.0001, Figure 2-7D) showing that the rescue of IIH6 reactivity coincides with functional DG glycosylation. Earlier studies have demonstrated that increasing DG glycosylation reduces migration, orthotopic growth, and invasion of cancer cells (Bao, Kobayashi et al. 2009, Yoneyama, Angata et al. 2012). We first assessed the effect of LARGE2 knockdown in our less metastatic PC-3E+ cells, but we were unable to show a significant difference in transwell migration, gap closure rate, or single-cell motility (data not shown); therefore, we sought to assess the functional consequences of rescuing DG glycosylation in the GS689.Li cell line. We performed the transwell migration assay and found that DG glycosylation inhibited directional migration through both uncoated and laminin-10 (laminin-511) coated wells (Figure 2-7E). Furthermore, utilizing a Matrigel invasion

50 36 assay, we found a >60% reduction in invasive potential with re-expression of mlarge2 (Figure 2-7F). Having defined a clear effect on migratory and invasive potential, we next performed a growth assay under standard tissue culture conditions and showed a significant difference in the growth characteristics after 6d in culture of the mlarge2 expressing cells when compared to vector control (Figure 2-7G). These results illustrate a clear effect on multiple characteristics that could impact on metastatic tumor growth in vivo. LARGE2 Expression Diminishes During Prostate Cancer Progression We next assessed whether LARGE2 expression status is associated with hypoglycosylation of αdg in prostate cancer patients. We first used qrt-pcr on samples of laser-capture microdissected tumors and patient-matched benign tissue. Normalization of values for all samples to a random benign control revealed that LARGE2 expression is significantly decreased (benign=3.999 ± a.u., tumor=1.271 ± a.u., p=0.0079) in tumors compared to healthy prostate epithelium (Figure 2-9A). In order to mitigate patient-to-patient variation, we next compared LARGE2 expression in each tumor to its patient-matched, benign control. LARGE2 expression was reduced in both Gleason grade 3 and 4 samples relative to benign tissue (p= and p=0.0020, respectively). Moreover, although comparison of the data for Gleason grade 3 and Gleason grade 4 samples did not reveal a statistically significant difference between the two (p=0.0809), it did suggest a trend toward reduction of LARGE2 levels with advancing disease grade (Figure 2-9B). Finally, in order to confirm that LARGE2 expression inversely correlates with tumor progression, we analyzed LARGE2 on a tissue cdna microarray organized according to disease stage. Pathological prostate cancer staging is determined by the extent of disease involvement in the patient with stage III being significant as it is the lowest pathological stage where there is evidence of

51 37 extraprostatic extension of the tumor (Edge and American Joint Committee on Cancer. 2010). Analysis of the values normalized to Stage II disease suggested that as the disease progresses from a localized (stage II; ± a.u.) to an invasive (stage III; ± a.u.) phenotype, LARGE2 is significantly reduced (p=0.0112, Figure 2-9C). This is consistent with reduced LARGE2 being an underlying cause of the loss of lamininbinding glycans on αdg in prostate cancer. Discussion As documented by multiple groups, loss of DG expression is a remarkably consistent feature in many cancer types including breast, prostate, colon, cervical, renal adenocarcinomas, squamous cell carcinomas, and neural tumors, e.g. (Henry, Cohen et al. 2001, Sgambato, Migaldi et al. 2003, Calogero, Pavoni et al. 2006, Sgambato, Tarquini et al. 2006, Martin, Glass et al. 2007, Sgambato, Camerini et al. 2007, Shang, Ethunandan et al. 2008). These studies have implicated proteolysis of α- and βdg, and/or hypoglycosylation of αdg detected via IIH6 or VIA41 monoclonal antibodies. Whether there is a relationship between hypoglycosylation of αdg and proteolysis of α- and/or βdg is unclear at present. Here we identify a new mechanism by which the functional properties of αdg become abrogated in prostate cancer-loss of expression of LARGE2, a putative glycosyltransferase that mediates the formation of laminin-binding glycans on αdg. LARGE2 has not been previously implicated in cancer, and very little is known about its function. Based on its sequence similarity to LARGE, it is inferred that it has similar function. As mentioned above, LARGE has recently been demonstrated to have a unique dual xylosyl- glucuronyl-transferase activity that extends this disaccharide chain that is essential for extracellular matrix ligand binding from phospho-mannose residues in the mucin-like domain of αdg. Our studies suggest that endogenous LARGE2 has this activity in prostate epithelial cells, and prostate cancer cell lines; where LARGE

52 38 expression is markedly absent in these cells. This outcome could be predicted from consideration of the tissue-specific expression of LARGE and LARGE2 which shows relatively high expression of LARGE2 in the prostate compared to muscle and neural tissue that show much higher expression of LARGE (Grewal, McLaughlan et al. 2005). Consistent with the tissue-specific expression pattern of LARGE2, we also find a correlation between low LARGE2 expression and αdg hypoglycosylation in pancreatic cancer cell lines. It remains to be determined whether LARGE2 elaborates glycan structures identical to LARGE and whether client proteins other than αdg exist for LARGE2. The complexity of αdg functional glycosylation raises a number of other possible causes underlying the hypoglycosylation observed in cancer. Previous studies showed that epigenetic silencing of LARGE results in hypoglycosylation of αdg in breast cancer (Takahashi, Honda et al. 2000). We do not yet know whether LARGE2 is epigenetically silenced in prostate cancer, but this represents one of several possible mechanisms by which its expression may be reduced. In our studies, we found that the LNCaP cell line, LARGE2 mrna expression is equivalent to other prostate cancer cell lines that are IIH6+ and we did not detect mutations in the coding sequence of LARGE2 in this cell line. Nonetheless, overexpression of LARGE2 in LNCaP cells restored IIH6-reactivity. Overexpression of LARGE has been shown to override deficiencies in earlier steps of αdg glycosylation, which suggests that one of these may be defective in LNCaP cells (Barresi, Michele et al. 2004). A previous study in prostate cancer showed that β3-nacetylglucosaminyltransferase (β3gnt1) cooperated with LARGE to indirectly promote glycosylation of αdg (Bao, Kobayashi et al. 2009). In this study, reduced expression of β3gnt1 was noted in IIH6- PC-3 cell derivatives and was suggested as an underlying cause of αdg hypoglycosylation. However, here we did not detect a conspicuous loss of β3gnt1 in IIH6- PC-3 cells, whereas LARGE2 was dramatically reduced. In the previous study, LARGE2 mrna was not detected in prostate cancer cell lines and the

53 39 basis for this discrepancy with our results is unclear. We cannot rule out a role for β3gnt1 in glycosylation of αdg, however our data indicates that it is unlikely to cooperate with LARGE in prostate epithelial tissue and cancer based on multiple lines of evidence: 1) As mentioned, prior work indicates that LARGE2 is preferentially expressed in the normal prostate compared to LARGE (Grewal, McLaughlan et al. 2005); 2) we show here that in a mouse that lacks a functional LARGE gene, that αdg is still glycosylated in normal prostate tissue; 3) using multiple probe sets, we find very little expression of LARGE in IIH6+ prostate cancer cell lines compared to much higher levels of LARGE2; 4) knockdown of LARGE2 in IIH6+ prostate cancer cells results in hypoglycosylation of αdg indicating a role for endogenous LARGE2 in this process. It remains possible that β3gnt1 cooperates with LARGE2 to promote glycosylation of αdg, but this awaits further investigation. As mentioned above, loss of DG function is associated with cancer cell phenotypes that may promote disease progression such as loss of cell polarity, decreased adhesion to extracellular matrix, increased invasive potential and altered intracellular signaling. Prostate cancer cell lines exhibit heterogeneous functional glycosylation of αdg. Our previous work identified a subpopulation of PC-3 prostate cancer cells, designated TEM4-18, which was isolated on the basis of its enhanced ability to invade an endothelial monolayer (Drake, Strohbehn et al. 2009). These cells also show a Zeb1-dependent epithelial-to-mesenchymal transition and behave more aggressively in metastatic colonization models. Interestingly, in TEM4-18 cells, IIH6 staining was absent, while these cells did still express the core αdg protein and βdg. Using another metastatic derivative of PC-3 cells (GS689.Li), which lack LARGE2 expression and demonstrate hypoglycosylation of αdg, we showed that restoration of LARGE2 expression rescues functional glycosylation of αdg and results in diminished ability of these cells to migrate in response to chemotactic stimulus, invade through Matrigel and reduces cell proliferation. However, we were unable to show that knockdown of LARGE2 promotes

54 40 invasive phenotypes. These data are consistent with the interpretation that loss of LARGE2 is is necessary, but not sufficient, to promote cellular phenotypes associated with prostate cancer progression and metastasis. Consistent with the idea that αdg hypoglycosylation is associated with a more aggressive disease phenotype, hypoglycosylation of αdg has, in some small retrospective studies, demonstrated potential prognostic value in predicting disease recurrence or survival (Sgambato, Camerini et al. 2007, Moon, Rha et al. 2009, Sgambato, Caredda et al. 2010, Jiang, Rieder et al. 2011, Shen, Xu et al. 2012). Prior work in prostate cancer demonstrated that αdg hypoglycosylation is associated with increasing Gleason score (Sgambato, De Paola et al. 2007, Shimojo, Kobayashi et al. 2011). Here we have independently confirmed these findings in two separate patient cohorts using prostatectomy specimens. Expanding upon this, we report here for the first time that αdg glycosylation is virtually undetectable in prostate cancer metastases, while βdg immunostaining is more frequently retained. Because of the extensive loss of IIH6 in metastatic disease, we attempted to determine if loss of αdg glycosylation could have independent prognostic value in prostatectomy specimens in a retrospective patient cohort. Unfortunately, due to lower than expected mortality in our cohort, we were unable to determine if IIH6 status predicted survival. However, there was a trend toward an association between loss of αdg glycosylation and biochemical recurrence, but, this did not reach statistical significance. Moving forward, analysis of a much larger number of patients will be necessary to evaluate the prognostic value of αdg glycosylation in prostate cancer. In addition, our finding that LARGE2 expression is diminished during prostate cancer progression suggests that it, too, may be investigated as a potential prognostic biomarker in this disease.

55 41 Contributions This work was originally published in 2013 in the Journal of Biological Chemistry (Esser, Miller et al. 2013). My contribution to this paper includes all data from figures 2-2 and 2-3, 2-6, 2-7 and 2-9. Additionally, I performed flow cytometry from figure 2-4 and 2-8.

56 Figure 2-1 αdg glycosylation and βdg expression are reduced in prostate adenocarcinoma. A. Representative images of αdg and βdg immunostained human prostatic tissue showing loss of DG correlating with cancer progression. Scale bar = 100 µm B. Grading was determined using 0-3 grading scale where 0 = basal staining absent; 1 = discontinuous thin basal staining; 2 = thin basal staining with extension between cells or thin dark basal stain without basal-lateral staining; 3 = thick, dark basal staining with basal-lateral staining, characteristic of benign glands. C. Mean DG staining of human prostate cancer tissue samples shows that DG expression is reduced across Gleason scores 6-8 (αdg p< and βdg p<0.0003), while the magnitude of loss of αdg is greater than loss of βdg immunoreactivity. 42

57 Figure 2-2 αdg functional glycosylation is nearly absent in prostate adenocarcinoma metastases. Representative images of IIH6 staining (top row) and serially sectioned analysis of 8D5 (bottom row) showing retention of glycosylation (A), loss of glycosylation only (B), or loss of protein expression (C). Analysis of DG in metastatic tumor shows a significant difference between expression and glycosylation (p<0.0001) (D). 8D5 staining reveals strong cell surface staining with βdg (E). IIH6 staining reveals specific, though less robust, staining of αdg at the cell surface (F). Bar =100 µm 43

58 Table 2-1 Demographics of the cohort used for retrospective analysis 44

59 .Figure 2-3 Association of biochemical recurrence with glycosylation status of αdg in a retrospective analysis. A. Frequency of biochemical recurrence in the cohort over 15 years of follow-up (solid line) with the 95% CI shown by the dashed lines. B. Kaplan-Meier recurrence-free survival curve comparing high (5-6) vs. medium (3-4) vs. low (0-2) glycosylation of DG. C. Hazard ratio analysis of DG staining using univariable and multivariable (adjusted for Gleason score and stage) regression. D. Univariable and multivariable regression analysis ( 1 adjusted for Gleason score and cancer stage) utilizing grouped glycosylation scores to compare aggregated data. Major = most common pattern. Minor = second most common pattern. Combined = major + minor. 45

60 Figure 2-4 αdg functional glycosylation is heterogeneous in prostate cancer. A. Both αdg core protein (80-100kDa, Sheep 5) and the transmembrane βdg (43kDa, 8D5) protein were expressed in all prostate cancer cell lines. Glycosylated αdg is apparent in PC-3, PC-3E+, and 22Rv1 cell lysates. β- actin was used as a loading control. B. Analysis of cell surface αdg shows that 22Rv1, PC-3 and PC-3E+ cell lines are positive for IIH6 staining while LNCaP and TEM4-18 cell lines are negative by flow cytometry analysis. C. 22Rv1, PC-3 and PC-3E+ cells are positive for DG-dependent laminin-111 staining by flow cytometry (signal relative to cells not incubated with laminin) following incubation with exogenous laminin-111 (100nM). LNCaP and TEM4-18 cells are negative. Prior Incubation with the antibody IIH6 inhibits laminin-111 binding. An isotype control was used for all FACS experiments (grey). (n=2) 46

61 Figure 2-5 LARGE2 mrna expression correlates with hypoglycosylation of αdg. A. mrna analysis by qrt-pcr shows dramatically reduced LARGE2 expression, compared to other genes known to influence αdg glycosylation, in TEM4-18 cells compared to PC-3E+ cells. B. LARGE2 mrna is detectable in the prostate cancer cell line panel and is greatly reduced in TEM4-18 and GS689.Li cells. All LARGE2 samples validated with two probe sets on two separate RNA isolations. Numbers on top of bars are Ct values: GAPDH/gene of interest. C. Immunofluorescence staining shows αdg glycosylation (IIH6) is present in a WT mouse prostate and the myodystrophy mouse (Large myd, MDC1D) prostate (arrows). Staining is not present in a prostate specific DG knockout mouse. Muscle staining (asterisks) indicates tissue specific deletion. Bar= 50µm. 47

62 Figure 2-6 LARGE2 alters IIH6 immunoreactivity in multiple prostate cancer cell lines. A. qrt-pcr analysis demonstrates a significant reduction in LARGE2 transcript with two independent shrna constructs. B. qrt-pcr analysis of LARGE mrna expression in PC-3E+ LARGE2 knockdown cells. IIH6 flow cytometry analysis of 22Rv1(C) and DU145(D). Both cell lines were stably transfected with either of two shrnas against LARGE2 or empty vector (control). An IgM isotype control for the vector only cell lines is shown in grey.e. Overexpression of a murine LARGE2 in TEM4-18 cells induces IIH6 immunoreactivity in an unstable sub-population of cells. An isotype control for the cells is shown in grey with the vector only IIH6 shown in red. 48

63 Figure 2-7 LARGE2 functionally glycosylates αdg reducing invasive and proliferative potential. A. PC-3E+ cells transfected with a LARGE2 targeted shrna #147 or #148 reduced IIH6 staining by 86% and 77% respectively by FACS analysis when compared to vector (plko) control. B. PC-3E+ plko control cells bind exogenous laminin-111. Incubation with IIH6 prior to laminin-111 inhibited binding. Laminin antibody only was used as a background control. PC-3E+ shlarge2 #147 cells have reduced laminin- 111 binding when compared to vector control (*, p<0.001). PC-3E+ shlarge2 #148 cells were further inhibited and significantly different from both vector (*, p<0.001) and shlarge2 #147 (, p<0.001). C. LNCaP and GS689.Li cells transfected with mlarge2 or vector control show a significant shift in IIH6 immunoreactivity via flow cytometry. D. LARGE2 expression in IIH6- cell lines induces a significant increase in laminin binding capacity that is specific to DG as shown via inhibition with IIH6 antibody. E. GS689.Li cells show retarded migration through a Boyden chamber both in the absence and presence of a laminin-511 substrate. F. Rescue of αdg glycosylation by mlarge2 also inhibits migration through a Matrigel substrate by >60%. G. mlarge2 expression causes a significant reduction in growth of GS689.Li cells under standard growth conditions (*, p<0.05; **, p<0.01; ***, p<0.001) 49

64 Figure 2-8 Levels of LARGE and LARGE2 expression differ in cancer cell lines derived from different tissues. A. Flow cytometry analysis reveals αdg glycosylation status (IIH6 staining) is heterogeneous across pancreatic (PANC-1, BxPC-3), colon (HT-29) and breast (ZR75-1, MDA-MB-231) cancer cell lines. IgM isotype control is shown in grey. B. qrt-pcr of LARGE2 mrna levels in pancreatic, colon and breast cancer cell lines. C.qRT-PCR of LARGE mrna expression in pancreatic, colon and breast cancer cell lines. Ct values are displayed as GAPDH/target. Reactions were performed in triplicate. 50

65 Figure 2-9 LARGE2 expression is diminished during prostate cancer progression. A. Microdissected samples of prostate tumor and benign tissue cdna were obtained and analyzed via qrt-pcr. All samples were normalized to a random, benign control. Carcinoma samples show a markedly reduced level of LARGE2 expression when compared to benign epithelium (benign=3.999 ± a.u., tumor=1.271 ± a.u., p=0.0079). B. Microdissected samples were all normalized to the patient-matched benign sample with the normalized value of the benign, 1.0, marked by the dashed line. Both Gleason grade 3 and Gleason grade 4 samples show a significant decrease in expression levels (p= and p=0.0020) when compared to benign. C. Analysis of the tissue cdna array normalized to benign samples. There is a significant decrease of LARGE2 (p=0.0112) when stage II (disease is completely confined to the prostate) is compared to stage III (extension beyond prostatic capsule or seminal vesicle involvement). 51

66 52 CHAPTER III DOWNREGULATION OF MULTIPLE ENZYMES LEADS TO LOSS OF DYSTROGLYCAN GLYCOSYLATION IN CLEAR CELL RENAL CELL CARCINOMA Introduction Renal cell carcinoma is a highly prevalent disease that will newly affect approximately 62,000 people in 2013 (Siegel, Naishadham et al. 2013). There are multiple subtypes of clear cell carcinoma, and the most prevalent is clear cell renal cell carcinoma (ccrcc). ccrcc is a highly aggressive, often fatal disease that exhibits chemotherapeutic resistance with only small increases in survival length in patients with disease resistant to normal therapy (Escudier, Eisen et al. 2007). Currently, primary prognostic information is derived from the Fuhrman nuclear grade, a grade based on nuclear size and morphology, and disease staging, a measure of tumor progression at the time of resection (Sorbellini, Kattan et al. 2005). However, a molecular understanding of this disease has begun to emerge in recent years with two critical papers defining the molecular subtype clear cell renal cell carcinoma, a common, aggressive, chemo-resistant subtype of renal cell carcinoma (Brannon, Reddy et al. 2010, Cancer Genome Atlas Research 2013). While these studies shed light on global genetic patterns developing during disease progression, utilization of smaller pathway analyses could begin to extrapolate the individual roles these pathways serve in modulating disease aggression. Dystroglycan (DG) is an extracellular matrix interacting protein that is frequently downregulated in a variety of tumor types (Henry, Cohen et al. 2001, Sgambato, Migaldi et al. 2003, Sgambato, Tarquini et al. 2006), and loss of DG glycosylation associates with increased mortality in clear cell renal cell carcinoma patients (Sgambato, Camerini et al. 2007, Sgambato, Camerini et al. 2010). Unfortunately, DG s regulatory pathway is

67 53 complex because it relies upon the concerted action of a number of glycosyltransferases which function to heavily glycosylate alpha-dg. This glycosylation is required for DG s functional activity, and only a handful of studies have attempted to identify the underlying changes in the glycosyltransferase pathway that work to mediate DG s carcinoma-associated hypoglycosylation. With at least 13 identified proteins required for glycosylation of DG, it is difficult to analyze the entire pathway in smaller studies of cancer-associated hypoglycosylation. The development of the Cancer Genome Atlas has created a wealth of data that can be utilized to investigate smaller pathways, like DG glycosylation, to determine which components most strongly correlate with disease markers (Cancer Genome Atlas Research 2013). TCGA data also allows for pathologic changes within pathways to be identified in an unbiased fashion. Herein, we utilize the TCGA dataset to analyze the DG glycosyltransferase pathway and identify a number of genes that strongly correlate with classical tumor aggression markers. We further demonstrate that downregulation of these genes associates with increased overall mortality. Finally, we demonstrate a method to utilize the TCGA in a pathway-specific fashion to identify novel regulatory components of GYLTL1B, the gene controlling expression of LARGE2. Materials and Methods Human Tissue Samples All samples were obtained and handled according to the IRB approved project # Tissue samples were obtained from the University of Iowa Hospitals and Clinics pathology department. All patients received partial or radical nephrectomy to treat a documented case of clear cell renal cell carcinoma and had uninvolved surgical margins

68 54 subsequent to removal. Formalin fixed, paraffin embedded samples were obtained and blocks were selected based on amount of tumor present. Immunohistochemistry Paraffin embedded renal tissue was sectioned at 5 µm, deparaffinized and rehydrated using a xylene to ethanol gradient. Antigen retrieval for all samples was performed by a 20-min exposure to either citrate buffer (βdg) or proteinase K (αdg). Endogenous peroxidases were quenched with 3% hydrogen peroxide for 10 min. Sections were blocked in 10% horse serum and then incubated overnight with primary antibody, either IIH6 (1:100, Santa Cruz Biotechnology) or 8D5 (1:100, Novocastra). Samples were washed and incubated for 1 h at room temp with a biotinylated secondary antibody, which was then detected using the ABC reagent (Vector) and DAB plus (Dako) followed by counterstain with hematoxylin. Hematoxylin and eosin (H and E) staining was performed using standard procedures. Immunohistochemistry was performed by the UI Department of Pathology Core Lab. Pathologists (DM and MG) were blinded to progression status at the time of analysis. Samples were scored according to a quartile system whereby: 3: positive ( 90% of cells intensely positive); 2: heterogeneous (regional positivity with >10% of cells negative); 1: reduced (>10% of cells negative and decreased intensity of staining); and 0: loss ( 1% of cells positive). Discrepancies between scores were assessed individually and a consensus was agreed upon. The Cancer Genome Atlas All TCGA data was obtained using the University of Iowa Institute for Clinical and Translational Science s (ICTS) data portal. ICTS created a custom database system for storing the large volumes of data required by the TCGA Dataset. This database utilizes a

69 55 distributed open source platform, Cassandra from the Apache Foundation. Data was extracted from each of the data files downloaded from TCGA Website, then loaded into a representative Cassandra table. Once the data was loaded for each type, we were then able to query and combine the data based on the barcode values for each sample. This combination work has been done in several ways. The first attempt was completed using Perl Scripts and direct access to the files. The current system uses a JAVA Web Application, connecting directly to the Cassandra database via a JDBC Driver ( Clinical data was obtained from the clinical_kirc.tar.gz (06/14/2012), transcript data from the IlluminaHiSeq_RNASeqV2.Level_ (01/08/2012), and the methylation data from KIRC.HumanMethylation450.Level_ (05/13/2013) databases. Only normalized, gene specific transcript data was obtained and integrated (rsem.genes.normalized_results). All gene RNA-Seq by Expectation Maximization (RSEM) values were collected and log-transformed to correct for non-normal distribution. The cbioportal was utilized for all copy number alterations and mutational analysis (Cerami, Gao et al. 2012, Gao, Aksoy et al. 2013). Cell Lines Renca cells expressing luciferase (Norian, Kresowik et al. 2012) were maintained in RPMI (Gibco) supplemented with 10% fetal bovine serum (Atlanta Biologicals), Penicillin-Streptomycin (Gibco), Sodium pyruvate (Gibco), non-essential amino acids (NEAA, Gibco), and glutamine (Gibco). Immortalized kidney tubular epithelial cells (kind gift from Dr. Roy Zent, Vanderbilt University) were maintained in DMEM supplemented with 10% FBS and 1% NEAA.

70 56 Flow Cytometry Cells grown to 60 80% confluency were detached (Versene, Invitrogen) and washed prior to antibody incubation. The cells were then resuspended at 1 x 10 6 cells in 100 µl FACS buffer (PBS % sodium azide + 5% BSA) + IIH6 antibody (1:100), OR E- cadherin antibody for 45 min on ice. Next, cells were washed and incubated with secondary antibody (100 µl FACS buffer plus 1:100, secondary antibody, Jackson Immunoresearch) for 45 min on ice in the dark. Cells were washed again and resuspended in 400 µl of FACs buffer containing Hoechst for live/dead differentiation, and transferred to polystyrene FACs tubes (BD Biosciences). Samples were analyzed using the Becton Dickinson LSR II flow cytometer at The University of Iowa Flow Cytometry Core Facility. Statistical Analysis To compare expression in tumor-normal matched samples, we carried out paired t- tests of differences in expression on the log scale. Associations between expression and stage/grade were calculated using a proportional odds regression model, adjusting for age and sex. Here, stage and grade were treated as ordinal outcomes. The effects of differential expression on mortality were assessed using a proportional hazards model, again adjusting for age and sex. For Figure 4, separate models were fit for each gene to assess the marginal associations between each gene and disease progression. For Figure 5, a single model including expression levels for all genes was fit to isolate the effects of individual genes within the context of the entire DG glycosylation pathway. Kaplan- Meier curves were also fit to illustrate the effects of differential expression on overall mortality.

71 57 Results Dystroglycan Expression and Glycosylation are Diminished in Clear Cell Renal Cell Carcinoma To determine the status of dystroglycan expression and glycosylation, we assessed 59 available patient samples utilizing the glycosylation sensitive αdg antibody, IIH6, and the βdg antibody, 8D5. Patients were selected such that 28 experienced disease recurrence following radical nephrectomy while 37 remained disease-free (TABLE 3-1). Samples were scored based upon intensity and frequency of staining observed throughout the tumor, and scores were assigned 0-3(FIGURE 3-1 A-H). All adjacent normal tissue exhibited a score of 3 in order to be utilized for the study (FIGURE 3-1 I, J). Both DG glycosylation and DG expression are frequently decreased in clear cell renal cell carcinoma % of alpha and 86.44% of beta exhibited staining levels below that of adjacent benign. Additionally, αdg exhibited a slightly lower average score of (SEM = ) compared to the average score of βdg, (SEM= ). Additionally, grade does not associate with staining of either alpha- or beta-dg (FIGURE 3-1 K, L). We next assessed whether dystroglycan staining could be used as a clinically useful marker to guide treatment decisions. The cohort described (TABLE 3-1) represent a well-controlled patient set to assess whether loss of DG, expression or glycosylation, associates with disease recurrence following radical nephrectomy. The two patient populations have largely identical clinical features with the disease progression group exhibiting a slightly higher Fuhrman grade, on average. In order to determine if this association is present, we compared those samples that exhibited no staining with all other samples (0 vs. 1-3) for both alpha- and beta-dg. We found that neither loss of expression (p=1.000) nor glycosylation (p=0.1859) associate strongly with disease

72 58 progression as assessed by Fisher s exact test (FIGURE 3-2). However, alpha-dg shows an increased incidence of loss in patients with disease progression, indicating an increase in sample size may be capable of reaching significance. The DG Glycosylation Pathway is Perturbed During Clear Cell Renal Cell Carcinoma Development We used the Cancer Genome Atlas database in order to query the dystroglycan glycosylation pathway to determine which components are most often perturbed during tumorigenesis and disease progression. We utilized information from those samples that had matched benign tissue and compared mrna expression levels of DG-associated proteins. In order to visually represent the data, we plotted the findings using both a volcano plot and a relative expression plot to highlight both the magnitude and significance of the changes (FIGURE 3-3 A, B). GYLTL1B, the gene for LARGE2, demonstrates the greatest magnitude change with a nearly 80% mrna reduction between tumor and normal. The two most statistically significant genes were found to be DAG1, the gene for dystroglycan, and POMGNT2. Interestingly, these two genes showed nearly identical levels of loss, and when the sample set was analyzed for copy-number alterations, we found that both POMGNT2 and DAG1, resident genes of chromosome 3p, exhibit nearly 90% levels of heterozygous loss. Clear cell carcinoma is unique in that it is the only tumor type to significantly associate with loss of the von Hippel-Lindau tumor suppressor, VHL, whose gene also resides on chromosome 3p (Gnarra, Tory et al. 1994). Finally, after accounting for these chromosomal alterations, we discovered that LARGE, the homologue of LARGE2, exhibits the next highest level of downregulation with a ~56% reduction in tumor compared to normal. Importantly, neither GYLTL1B nor LARGE show a significant loss of heterozygosity. Finally, mutational data was analyzed using Memorial Sloan Kettering s cbioportal (Cerami, Gao et al. 2012, Gao, Aksoy et al.

73 ), which revealed that this pathway has a very low mutation rate with POMGNT1 exhibiting those most recorded mutations at 3/424 tumor samples. DG-Associated Glycosyltransferases Associate with Grade and Stage In order to better assess the association between the expression changes in the DG glycosylation pathway and disease progression, odds ratios were calculated for both Fuhrman nuclear grade and stage (FIGURE 3-4 A, B). Those genes to the right of the y- axis inversely associate with increasing grade or stage. For example, a two-fold down regulation of ISPD associates with a ~1.8 fold increase in the odds of a higher grade and stage for the tumor. We found that a number of the genes associated with DG glycosylation inversely associate with both grade and stage. POMT1, ISPD, FKTN, B3GNT1, and GYLTL1B all inversely associate with both factors. Interestingly, POMGNT1, seems to positively associate with increasing grade and stage which is at odds with the data clearly showing loss of glycosylation during disease development and progression (Sgambato, Camerini et al. 2007). The analysis is complicated by the fact that similar pathways may regulate certain genes within the pathway. Because of this, a correlation heat map was constructed with an associated dendrogram to determine which of these genes demonstrate similar expression patterns (FIGURE 4-4 C). There are three demonstrated groupings revealed by the analysis. As an example, POMGNT2, DAG1, and LARGE represent one of these families. This implies a potential shared regulatory pathway between some of these genes.

74 60 Pathway-Corrected Analysis Identifies Genes Associated with Mortality Despite the lack of predictive power with alpha- and beta-dg immunohistochemistry, we next assessed the possibility that specific genes within this pathway associate with patient survival. We performed a hazard ratio analysis with these genes and the overall survival data available within the cancer genome atlas. Initial analysis, corrected for age and sex, revealed an inverse prognostic association with several genes in the DG glycosylation pathway. Downregulation of ISPD, FKTN, B3GNT1, GYLTL1B, all also associated with grade and stage, significantly associate with increased mortality (FIGURE 4-5 A). Because of the earlier findings showing some genes expression tend to strongly correlate, we performed an analysis to correct for the expression of other genes in the pathway. After correcting for pathway expression, only GYLTL1B and ISPD associate with increased mortality (FIGURE 4-5 B). This indicates that each of these genes associates independently with mortality and suggests a potential role in disease progression. A Kaplan-Meier survival curve highlights the strong association between a loss of expression (1 standard deviation below the mean) of both GYLTL1B and ISPD and fatal patient outcomes (log-rank p < ) (FIGURE 4-5 C). We also see a single gene that associates with mortality in the opposite direction: FKRP. It is difficult to determine the functional implication of this finding with the data available. Regulation of GYLTL1B Links DG Glycosylation with the Epithelial-to-Mesenchymal Transition Pathway As LARGE2, the product of GYLTL1B, has recently been identified as a critical enzyme for DG glycosylation in prostate cancer, we sought to determine whether mechanisms regulating its expression could be identified utilizing the cancer genome atlas (Esser, Miller et al. 2013). Inappropriate methylation patterns are frequently

75 61 observed in carcinoma, and hypermethylation of genes often, though not always, function to downregulate the expression of the given gene (Eden and Cedar 1994, Esteller, Tortola et al. 2000). Therefore, we first utilized the methylation data available through the database and looked at the relative methylation analyzed at each of the sites targeted by the Infinium HumanMethylation450 bead array. We found that the majority of sites demonstrated higher levels of methylation relative to normal (FIGURE 3-6A). Analysis of the promoter region for GYLTL1B using the UCSC genome browser, revealed a CpG island containing 130 CpG sites. We calculated spearman correlation coefficients for each of the sites, and found that throughout the CpG island, increased DNA methylation was associated with decreased mrna expression (FIGURE 3-6B). Finally, we analyzed the CpG island in aggregate by averaging the relative methylation of the island for each patient and found that tumor samples demonstrated a statistically significant increase in methylation that exhibited a negative association with expression (FIGURE 3-6C, D) While these data suggest DNA methylation plays a role in regulating GYLTL1B, the relative increase in individual sites methylation is relatively small. Therefore, we sought to identify transcription factors that could potentially regulate the expression of this gene. Since there is nothing known about the regulation of LARGE2 transcription, we used data available from the TCGA dataset to investigate. To do this, spearman correlation coefficients were generated for every gene in the database against GYLTL1B from the prostate cancer, breast cancer, lung cancer, and clear cell renal cell carcinoma datasets. Those genes demonstrating significance (define by p < 1.25 x 10-5 ) were isolated for both positive and negative correlations. We then identified those genes that shared positive or negative correlation across all four tumor types (FIGURE 3-7A, B). One of the discovered negative correlates, ZEB1, was pursued because of its identified role in epithelial-to-mesenchymal transition, a process that helps induce a metastatic phenotype in cancers (Nauseef and Henry 2011). We then over-expressed ZEB1 in PC-3E + prostate cancer cells, a derivative of PC-3, which we previously characterized as being both E-

76 62 cadherin positive and IIH6 positive. ZEB1 expression diminishes E-cadherin levels, as expected, and substantially reduces DG glycosylation in this cell line (FIGURE 3-7 C, D). Discussion For years, physicians have relied upon the most basic information gleaned from tumor pathology to elucidate patient prognoses. In order to improve prognostic accuracy and better direct clinical management of patients, a wealth of biomarkers have arisen in recent years for a number of different tumor types. Dystroglycan has frequently been identified as a potential prognostic biomarker in a number of different disease states (Moon, Rha et al. 2009, Jiang, Rieder et al. 2011, Coco, Zannoni et al. 2012, Shen, Xu et al. 2012). Herein we attempted to determine whether dystroglycan could be used as a predictor of disease recurrence following surgical resection with curative intent. Unfortunately, while loss of dystroglycan staining has previously been linked to grade and increased mortality in patients with ccrcc, we did not observe a significant association between DG expression or glycosylation, and disease recurrence. There is a noticeable trend in loss of alpha-dg staining in patients that recur when compared to the non-recurrent comparison suggesting that an increased patient cohort could provide the necessary power to observe a statistically significant effect. We did not perform a power estimate based on earlier observations because the original description of DG in ccrcc did not provide a sufficient explanation of staining patterns (Sgambato, Camerini et al. 2007), which explains why our dataset may be too small to detect a significant difference. ccrcc, in comparison to many other cancer types, represents a unique opportunity to dissect the functional necessity of DG. This arises from the fact that DAG1 and POMGNT2 both reside on the same chromosomal arm as VHL, the most commonly lost gene in ccrcc. DG does not require both copies of the gene for functional expression in

77 63 murine stem cells or developing epithelial tissue (Henry and Campbell 1998, Esser, Cohen et al. 2010), but its heterozygosity coupled with a concurrent reduction in one of the glycosyltransferases critical for the functional glycosylation of DG provides an ideal system that may exhibit increased sensitivity to smaller changes in the other components of the glycosyltransferase pathway. Consequently, the frequency of DG protein disruption is very high with nearly 90% of alpha- and beta-dg stained samples exhibited reduction compared to the normal tubules. While this work shows that a number of DG glycosyltransferases are downregulated in tumor development, we still do not know whether downregulation of any of these lead to loss of alpha-dg staining and function. We are currently working to extract RNA from stained samples in order to identify which of the DG glycosyltransferases genes appears to be playing a causative role in the reduction of alpha-dg staining. Signaling pathways and their regulation are complex networks with input from a number of differential transcription factors. The identification of clades within the dataset suggests that multiple components responsible for DG glycosylation may share upstream regulatory control. Thus, it is possible that classical tumor suppressor pathways may mediate the regulation of larger groups of genes that result in the frequently observed loss of DG glycosylation in various tumor types. Since the majority of studies focusing on this glycosylation pathway have traditionally concentrated on dystroglycanopathies and associated mutations within this pathway, little is known about the transcriptional regulation of its components. As more work is done to elucidate their regulation, a better understanding of the implications of DG glycosylation in cancer will be available. The discovery that FKRP positively correlated with overall mortality was unexpected in light of DG s hypoglycosylation phenotype. This finding highlights one of the shortcomings of this type of study. The findings presented represent an association between the observed changes and a clinical outcome. Causation is not implied, and this

78 64 requires that we demonstrate a functional requirement for DG glycosylation. In order to address this problem, we are currently working with a murine renal carcinoma cell line that exhibits DG hypoglycosylation relative to normal murine tubular epithelium (FIGURE 3-8). This model metastasizes in an orthotopic model of disease progression. Therefore, we are working to restore DG glycosylation in this line to determine whether functional restoration of DG in this model serves to inhibit metastatic dissemination. Due to the pathway style analysis we employed, we did not attempt correcting for grade and stage in our analyses. Having identified a clear correlation with grade and stage in a number of genes analyzed, the further fragmentation of the data by these factors would have significantly reduced the power of these studies and made significant findings all but impossible. However, the pathway correction employed in our statistical analysis allowed for the identification of internal driver genes associated with clinical mortality. As the number of samples available for analysis grows, both within the TCGA and abroad, increasingly complex analyses will provide a much more definitive description of causative pathway changes. Large-scale datasets allow a unique opportunity to utilize in silico statistical approaches to identify novel mechanisms of regulation. We utilized genome-wide correlation analyses across multiple tumor types to probe associated changes with GYLTL1B. Utilizing this unbiased approach, we were able to generate a small list of genes that either positively or negatively correlates with GYLTL1B s expression. In doing so, we discovered a transcription factor traditionally associated with EMT, ZEB1. We were then able to demonstrate that overexpression of this transcription factor results in decreased DG glycosylation, and other work in our lab has confirmed that ZEB1 is capable of directly blocking LARGE2 expression. This workflow can be expanded to other member of the glycosyltransferase pathway in order to continue to define the networks associated with DG-related signaling, both upstream and down. Additionally, these findings suggest that loss of DG glycosylation in the context of multiple cancer

79 65 types could be a relevant biomarker as an early identifier of EMT activation, but more work will need to be performed to demonstrate the functional necessity of LARGE2 in other tissue types. Collectively, this work shows that DG glycosylation is frequently downregulated in ccrcc, and while its prognostic implications by immunohistochemistry are somewhat limited, the transcript level analysis illustrates a number of genes that have significant clinical correlations. As we build upon these findings, we will be able to determine whether these changes have a direct functional impact on tumor progression or if these are merely passengers to other, more functionally relevant, drivers. Contributions All work performed in this chapter was performed by the primary author with the statistical analyses performed in conjunction with the University of Iowa Biostatistics Core.

80 Tumor'Stage 1 T Number'of'Patients Disease'Recurrence Disease3Free P'Value Age'(Range) 57.19&( ) 60.61&( ) & Median'Time'to'Relapse'(years) Median'Follow3up'(years) & Sex Smoking'Status 1.59&( ) 6.47&( ) Male Female Yes No & Fuhrman'Grade T3 & Grade&1+2 Grade& * Table 3-1 Cohort characteristics of patients utilized within this study. Patients treated by radical nephrectomy either remained disease free or experienced recurrence. Chi-square contingency analysis was performed to determine differences between groups. Statistical significance is indicated by *. 1 Staging is according to TNM classification.

81 Figure 3-1 Dystroglycan expression and glycosylation is frequently lost in clear cell renal cell carcinoma. Samples stained for either alpha-dg (IIH6) or beta-dg (8D5) were scored according to a quartile scoring system whereby 3 = positive ( 90% of cells intensely positive); 2 = heterogeneous (regional positivity with >10% of cells negative); 1 = reduced (>10% of cells negative and decreased intensity of staining); and 0 = loss ( 1% of cells positive). A-D. IIH6 staining and associated scores. E-H. Beta-DG staining and associated scores. I,J. Normal tissue demonstrating positive staining. K. Alpha- and betastaining score distribution. L. Scores were grouped by Fuhrman nuclear grade and demonstrate no significant association between staining and grade. 67

82 Figure 3-2 Neither loss of DG glycosylation nor expression associates with disease recurrence. Following scoring, data was compiled and analyzed comparing patients experiencing disease recurrence and those that remained disease-free. Neither alpha-dg (A) not beta-dg (B) demonstrated a significant association with disease recurrence when analyzed by Fisher s exact test. 68

83 Figure 3-3 Multiple proteins in the DG glycosylation pathway are downregulated during tumor progression. Analysis of the DG glycosyltransferase pathway describes expression, somatic copy number alterations, and mutational frequency. A. A volcano plot demonstrating significance and magnitude change for the analyzed gene set. B. A bar chart demonstrates differences in log-transformed expression of the various glycosyltransferases. C. Utilization of the GISTIC copy number alteration analysis tool through the cbioportal shows frequency of heterozygosity in the analyzed gene set. D. Mutational analysis from the MSKCC cbioportal demonstrates the low frequency of mutation within the pathway. Each vertical grouping represents one patient. Mutations are marked in green. Tumor samples with no detectable mutations are not shown. Total number of tumors analyzed is

84 Figure 3-4 Multiple genes in the DG-glycosylation pathway show a significant association with both disease grade and stage. Odds ratios were calculated for the DG glycosyltransferase pathway. A-B. Stage and grade analysis, adjusted for sex and age, plotted such that a two-fold downregulation of the indicated genes associates with an increase of odds (right of the axis). 95% confidence intervals are shown by the horizontal bar. P values are listed to the right of the respective genes. C. A heat map and associated dendrogram illustrating frequency of correlation between the various genes. Color intensity indicates the strength of the correlation. Blue is negatively correlated and red is positively correlated. 70

85 Figure 3-5 Loss of GYLTL1B and ISPD independently predict increase mortality for clear cell carcinoma patients. A-B. Stage and grade analysis, adjusted for sex, age and other genes in the pathway, plotted such that a two-fold downregulation of the indicated genes associates with an increase of odds (right of the axis). 95% confidence intervals are shown by the horizontal bar. P values are listed to the right of the respective genes. C. Kaplain-Meier survival curve generated from patients with downregulation of the stated gene greater than 1 standard deviation from the mean. Number of affected cases: normal, 342; Low GYLTL1B, 55; Low ISPD, 55; Both Low,

86 Figure 3-6 Methylation of the GYLTL1B CpG island negatively correlates with expression of the gene. Analysis of the methylation data from the TCGA database. A. Relative methylation pooled for all samples, normal and tumor, and plotted based upon chromosome position. The CpG island is defined above. B. A table showing the correlation values between methylation and GYLTL1B expression. The CpG island is highlighted in green. C. All CpG island sites were averaged per tumor sample and compared to normal controls. D. Combined methylation values were plotted against log-transformed GYLTL1B values demonstrating correlation. 72

87 Figure 3-7 A genome-wide correlation analysis shows ZEB1 as a potential regulator of GYLTL1B expression. A combination of all significant spearman correlates discovered during the genome-wide analysis. A. A Venn diagram demonstrating all the significant findings from ccrcc, prostate adenocarcinoma, breast adenocarcinoma, and lung adenocarcinoma. The adjacent image depicts the shared genes and their pathway associations. B. Same analysis as A but for negative correlates. C, D. Flow cytometry analyzing either E-cadherin(C) or IIH6 (D) in PC-3E + cells expressing either a vector control (red) or Zeb1 (blue). Isotype controls are shown in grey. 73

88 Figure 3-8 IIH6 immunoreactivity is lost in a murine cell line of renal carcinoma. Flow cytometry for IIH6 in either immortalized kidney tubular epithelial cells (A) or the murine renal cell carcinoma cell line, Renca (B). Isotype is shown in grey. 74

89 75 CHAPTER IV DYSTROGLYCAN MODULATES METABOLIC ACTIVITY AND AUTOPHAGY INDUCTION & Introduction& Dystroglycan (DG) is a transmembrane protein expressed in multiple tissue types, including epithelial, neuronal, adipose, and skeletal muscle tissue (Ervasti and Campbell 1991, Durbeej, Henry et al. 1998). As a part of the dystrophin glycoprotein complex (DGC), DG is active as a heterodimer where the alpha-dg subunit functions to bind extracellular laminin-g domain proteins through its extensively glycosylated mucin domain (Ibraghimov-Beskrovnaya, Ervasti et al. 1992). Alpha-DG non-covalently associates with beta-dg at the cell surface, and the transmembrane beta-dg then interacts with a number of identified proteins on the cytoplasmic surface including dystrophin in skeletal muscle and utrophin in epithelial cells (Ibraghimov-Beskrovnaya, Ervasti et al. 1992, Durbeej and Campbell 1999). Within skeletal muscle, DG is required for mechanical stability of the sarcolemma, and loss of DG function results in membrane fragility and eventual cell necrosis (Petrof, Shrager et al. 1993). Loss of dystroglycan function, as well as loss of other members of the DGC, results in a muscular dystrophy phenotype whereby muscle wasting leads to skeletal muscle weakness. However, the role of DG in the epithelium is very poorly understood, and the extensive mechanical stresses imparted within the musculature do not correlate well with forces experienced by epithelial tissues, thus determining the function of DG in epithelium is a pressing question within the field.& Recent work has begun to unveil potential roles for DG in epithelial signaling. Of the reported findings, there appear to be a number of studies that link DG signaling to metabolism and energy sensing pathways. Loss of DG in Drosophila has recently been

90 76 identified to confer a cryophilic phenotype whereby DG knockout animals prefer colder temperatures due to a gross increase in the insect s metabolic rate(takeuchi, Nakano et al. 2009). These findings were ascribed to changes in membrane fragility leading to increasing Ca 2+ and subsequent activation of mitochondrial activity. DG has also been suggested to be required for appropriate AMP activated kinase (AMPK)-mediated phosphorylation of acetyl CoA carboxylase in rat alveolar epithelial cells exposed to stretch-induced signaling(budinger, Urich et al. 2008, Takawira, Budinger et al. 2011). Furthermore, dystroglycan (and its ligand, perlecan) have been recently identified to be necessary during energetic stress signaling through AMPK-mediated phosphorylation of Sqh, the Drosophila homologue of myosin regulatory light chain II (MRLCII) (Mirouse, Christoforou et al. 2009). [Please note, during the preparation of this document, the findings from the Mirouse publication were retracted (Oct 28, 2013). This will be considered in the discussion section of this chapter.] & AMP kinase is a critical metabolic switch that is activated in response to an increased AMP to ATP ratio by its upstream kinase, LKB1(Woods, Johnstone et al. 2003). Phosphorylation of AMPK affects multiple pathways including the proliferationassociated mtorc1 signaling pathway via AMPK-mediated phosphorylation of the mtorc1 inhibitor, tuberous sclerosis complex 2 (TSC2) (Inoki, Corradetti et al. 2005). Additionally, AMPK directly phosphorylates and activates Ulk1, a critical regulator of macroautophagy (herein referred to as autophagy) (Egan, Kim et al. 2011). Furthermore, Ulk1 has been shown to phosphorylate and activate ZIP kinase (ZIPK), an upstream kinase that acts upon MRLCII in periods of energetic stress (Tang, Wang et al. 2011). Consequently, the finding that Ulk1 regulates MRLCII activation though regulation of ZIP kinase (ZIPK) means that multiple points along this signaling pathway have been directly linked to a requirement for DG. & Experiments to determine whether DG is important in mediating cellular metabolism and the metabolic stress response are critical. Based on recent discoveries, it is possible

91 77 that loss of DG function could interrupt appropriate signaling by AMPK and its downstream targets, thus resulting in an improper response to energetic stress. Loss of DG is a well-established characteristic of advancing tumors, and correlating loss with metabolic signaling would allow for a more complete understanding of the implications of DG disruption. Understanding the signaling implications of DG loss might provide a potential avenue to describe DG expression in a more complete prognostic, and possibly treatment-driven paradigm.& Materials and Methods Cell Lines Human prostate cancer cell lines 22Rv1 and PC-3 E-cadherin positive cells (PC-3E + ) were maintained in either RPMI or DMEM/F12, respectively, supplemented with 10% fetal bovine serum (FBS), 1% non-essential amino acids, and 400 µg/ml genetecin. Both lines were stably transduced with shrna targeting either dystroglycan or LARGE2 and maintained in 1 µg/ml puromycin. Murine embryonic stem cells either heterozygous or homozygous null for DG (described previously (Henry and Campbell 1998)) were maintained in DMEM supplemented with 20% heat-inactivated FBS, 2 mm L-Glutamine, 1% non-essential amino acids, 1000 U/ml leukemia inhibitory factor (LIF), and 0.001% β-mercaptoethanol (BME). Murine embryonic fibroblasts (MEFs) were derived from the Pb-Cre4-;DG fl/fl mouse (Esser, Cohen et al. 2010) using standard protocols. Briefly, E14.5 embryos were dissected from the uterine horn followed by mechanical and chemical dissociation before plating into DMEM media supplemented with 10% FBS. These cells were spontaneously immortalized following a 3T3-like passaging process. Following derivation and immortalization, the MEFs were infected with either Cre.eGFP adenovirus or egfp adenovirus at a multiplicity of infection (MOI) 500, which results in

92 78 greater than 80% infections rates. All cell lines were maintained in a humidified incubator at 37 C under 5% CO 2. Antibodies Used The following antibodies from Cell Signaling were used in the studies phosphoribosomal protein S6 S235/236 (2211), ribosomal S6 (2217), phospho-ulk1 S757 (6888), phosphor-ulk1 S555 (5869), total Ulk1 (8054), pampk T172 (2535), total AMPK (5831), Akt S473 (4060), Akt T308 (4056), LC3 (3868), myosin regulatory light chain (MRLC, 8505), pmrlc S19 (3675). Additionally, IIH6, an antibody that recognizes the functional glycoepitope of α-dg (kind gift from Kevin Campbell), and 8D5, which recognizes the cytoplasmic domain of β-dg (Santa Cruz) were also used. Cell Proliferation Assays Cellular proliferation was assessed by two methods. Cells for both were plated into either 96 well plates or 6 well plates at a density such that cells would be near confluence by the completion of the assay. For method one, cells were trypsinised off of the plates using 0.25% trypsin-edta followed by neutralization with serum containing media. The solution was then counted using a Z2 Coulter Counter. Method two utilized the WST-1 reagent (Roche) which was added to wells at a 100x dilution and allowed to incubate for 2 hours at 37 C prior to imaging with a plate spectrophotometer set to measure absorbance at 450 nm.

93 79 Seahorse Bioanalyzer Metabolic Analysis Cells are plated into a specialized 96-well plate specific for the Seahorse at a density such that cells are 80-90% confluent 48 hours after plating. The plate is transferred to the associated liquid handling instrument and all media is exchanged with DMEM supplemented with 25 mm glucose and 1 mm pyruvate. Cells are equilibrated for 30 minutes prior to initiation of the assay then loaded into the Seahorse XF96 Bioanalyzer (Seahorse). As previously described (Dranka, Benavides et al. 2011), when indicated 300 nm carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) and 1 µm oligomycin are added at separate points during data acquisition to determine maximal mitochondrial respiration and oxygen consumption due to proton leak, respectively. Following data acquisition, cells are trypsinized from the plate, neutralized with serum containing media and counted using the Z2 Coulter Counter to normalize all data to a per cell value. Western Blot Lysis buffer containing 50 mm Tris, ph 7.5, 100 mm NaCl, 1 mm EDTA, 2% SDS, 1% Triton X-100, and 1 mm NaVO 4 supplemented with complete protease inhibitor tablets (Roche) and Phostop (Roche) phosphatase inhibitor tablets was applied to cells at 150 µl/well in 60 mm dishes or 500 µl/well in 10 cm dishes. Lysates were scrape harvested and mechanically sheared using a 27.5 Ga needle multiple times until viscosity was akin to water. Protein concentrations were determined using the Biorad DC protein reagent assay and a plate spectrophotometer set to the 750 nm wavelength. Lysates were loaded at ~20 ug protein/well in either 10, 15, or 17-well 4-12% bis-tris gels (Life Technologies). After gel electrophoresis, proteins were transferred in NuPage Transfer Buffer supplemented with 20% methanol to immobilon-fl PVDF (Millipore). Following

94 80 transfer, membranes were blocked with Odyssey blocking buffer (OBB) mixed 1:1 with PBS for one hour at room temperature. Primary antibody, diluted 1:1000, unless otherwise stated, was diluted into 1:1 OBB:PBS supplemented with 0.1% Tween-20 and incubated overnight, with shaking at 4 C. Following 3x5 minute rinses at room temperature secondary antibody diluted at either 10,000X (800 nm) or 20,000X (680 nm) in 1:1 OBB:PBS supplemented with 0.1% tween-20 and 0.03% SDS. Following incubation for 1 hour at room temperature, membranes were visualized using the LiCor Odyssey infrared imaging system and quantified with the accompanying software to acquire integrated intensity counts. Flow Cytometry Adherent cells grown to 60-70% confluency were removed from tissue culture dishes with 2 ml of Versene (Gibco) to preserve surface protein integrity and analyzed as described in chapter 2. Briefly, the cells are mixed with antibody at 1:100 dilution in a 5% BSA PBS solution (FACS buffer). Cells are incubated for 45 minutes on ice with frequent mixing. Primary antibody is diluted out and cells are pelleted at 1 rcf. Secondary antibody is similarly diluted 1:100 in FACS buffer and incubated for 30 minutes on ice, in the dark with frequent mixing. The antibody is again diluted and cells are pelleted. Finally, cells are resuspended in 400 µl of FACS buffer containing Hoechst at 4 µg/ml and analyzed with the Becton Dickinson LSR II. Digital gates are used to isolate single cell, appropriately sized, live cells prior to analysis. ATP Determination Cellular ATP was determined using the ATP Determination Kit (Life Technologies) according to manufacturer s protocol. Briefly, cells are lysed with 1% NP-40 in PBS and

95 81 cellular debris pelleted at 13g for 10 minutes µg of protein lysate is then combined with a reaction mix containing D-luciferin and luciferase and incubated for 10 minutes prior to imaging using the AMI1000 bioluminescent imager. ATP is calculated using an ATP standard curve provided in the kit, and normalized to the appropriate DG control. Cell Immunofluorescence Coverslips were coated with rat tail collagen I (BD Biosciences) for 1 hour at room temperature then washed 2 times with PBS before plating cells. Following adherence for >48 hours, when cells reached 60-70% confluence, the coverslips were placed in room temperature 4% paraformaldehyde for 20 minutes at room temperature. Cells were immediately washed with PBS and permeabilized for 15 minutes in ice cold methanol for 15 minutes at -20 C. Following another PBS wash, cells were blocked in 5% bovine serum albumin diluted in PBS for 1 hour at room temperature. Anitbody diluted in blocking buffer at 1:200 (unless otherwise stated) was incubated on cells overnight at 4 C. Following washes and incubation with the appropriate secondary antibody diluted 1:200 for 1 hour at room temperature, coverslips were mounted with fluorogel and allowed to cure for 24 hours at 4 C in the dark before imaging with the Zeiss 710 confocal microscope. Clonogenic Survival Clonogenic assays were performed by plating cells in a 6-well plate at 50,000 cells/well. After 24 hours, wells were washed and put into DMEM containined either 5 mm or 1 mm glucose. Following the indicated times, cells were detached with 0.25% trypsin-edta and live cells were counted based on trypan blue exclusion. 6-well plates were then used to plate 300 cells/well in DMEM 25 mm glucose media supplemented

96 82 with 10% FBS. 14 days after plating, each well was washed with PBS and fixed with a 6% gluteraldehyde, 0.05% bromophenol blue solution overnight at room temperature. The plates were washed and imaged using a Nikon backlit with a standard lightbox. Counts were performed on the digital images in a blinded fashion. Autophagic Flux Assay Cells were plated in quadruplicate in a 6-well plate on day 0 at a density of 30-50,000 cells/well. On day 2 the media was exchanged, and on day 3, the cells were used for the starvation assay. Cells were either maintained in standard growth media (cell line dependent) or exchanged for minimal essential media (MEM) following a PBS wash. For each nutrient condition, the cells were exposed to either 10 µm chloroquine or vehicle control. Each well was incubated for four hours prior to a PBS rinse and lysis as described above. Statistics For simple comparisons, statistical significance was assessed using the student s T- test unless otherwise specified in the text. For grouped analyses, a one-way ANOVA was employed. Statistical significance was assumed if the p value Results Disruption of DG does not lead to changes in cellular proliferation In order to directly assess whether modulating DG either by inhibiting its expression or its glycosylation affects cellular growth phenotypes, we first generated three different

97 83 cell types: an E-cadherin positive cell line dervived from PC-3 (Drake, Strohbehn et al. 2009), the 22Rv1 human prostate cancer cell line (ATCC), or MEFs derived from a mouse with a floxed DAG1 allele (Williamson, Henry et al. 1997, Esser, Cohen et al. 2010). The prostate cell lines were transduced with one of three shrnas targeting DG or LARGE2 to generate DG-deficient cells, and the MEFs were infected with Crerecombinase expressing adenovirus to generate DG-null cells for analysis. Knockdown efficiency of DG was assessed by immunoblot using the cy8d5 antibody, and the effect on surface glycosylation of DG was assessed by flow cytometry for IIH6 (FIGURE 4-1). Highly efficient known down was achieved with two different shrnas targeting DG (FIGURE 4-1 A-B), and Cre-mediated knockout of DAG1 produced a relatively homogenous population of DG-deficient MEFs (FIGURE 4-1 C,F). LARGE2 knockdown successfully reduced the amount of surface glycosylated DG with two different shrnas in both the PC-3E + and 22Rv1 cell lines (FIGURE 4-1 D,E). Disruption of DG glycosylation by the shrna #147 helps to illustrate the functional necessity of LARGE2 in these cell lines, but this shrna exhibits evidence of off-target effects, therefore it is not included in any of the coming analyses. Having confirmed the relative efficacy of our system, we next assessed the overall growth characteristics of these various lines under standard growth conditions (FIGURE 4-2 A-C). Under these growth conditions, both DG knockdown and LARGE2 knockdown was unassociated with any measurable difference in overall growth in any of the cell lines examined. While a single condition in both PC-3E + (shlg2) and 22Rv1 (shdg1) appear to grow more slowly by day 5, this was attributed to a slightly increased plating density at the initiation of the experiment (data not shown). Because of work showing a specific function of DG in mediating signaling under energetic stress (Mirouse, Christoforou et al. 2009), we next assessed growth under restricted serum and/or glucose conditions in the PC-3E + and 22Rv1 lines (Figure 4-2 D- G). Serum restriction removes both growth factors and nutrients whereas glucose

98 84 restriction removes a main carbon source for the cells. Regardless of whether glucose or serum was withheld from the media, loss of DG itself or its ligand binding glycan did not cause an observable change in the growth characteristics of either cell line, and any apparent differences were not found to be statistically significant when analyzed by twoway ANOVA. Inhibiting Dystroglycan Decreases Metabolic Rate While growth is not impaired in prostate cancer cell lines or MEFs by loss of DG function, we still wanted to assess the overall metabolic profile of the cells to determine if other changes unrelated to proliferation were occurring. In order to do this, we analyzed the cells using a Seahorse Bioanalyzer. This analysis provides a surrogate measure for oxidative phosphorylation (oxygen consumption rate, OCR) and active glycolysis (extracellular acidification rate, ECAR, measure of lactate output). Measurement of the all of the cell lines, including a murine embryonic stem cell line that is deficient for DG, shows that OCR is reduced in basal conditions, and this difference is exacerbated by the addition of FCCP, a proton ionophore that uncouples oxidative phosphorylation, which acts to demonstrate maximal mitochondrial respiration (FIGURE 4-3 A, C, E). Therefore the PC-3E + s exhibit both decreased basal and maximal oxidative phosphorylation. Contrary to what is predicted by the Warburg Effect (Vander Heiden, Cantley et al. 2009), the cells also exhibit decreased glycolytic activity as measured by the ECAR (FIGURE 4-3 C, D). This data collectively suggests that these cells are metabolically less active due to a deficiency in DG. The highly cohesive 22Rv1 cells were unable to be accurately assessed in this analysis due to the necessity of a highly accurate cell count post-analysis.

99 85 Disruption of DG Does Not Affect Intracellular ATP Having demonstrated a significant difference in mitochondrial activity with no discernible difference in cellular proliferation characteristics, we next investigated whether disruption of DG alters intracellular levels of ATP. To do this, we utilized an ATP determination kit on cells grown to ~70% confluence and lysed in the dish. ATP levels were determined for a LARGE2 knockdown of PC-3E + s, a DG knockdown for 22Rv1s, and the DG deficient MEFs. In all three analyses performed (n=3 per experiment), there was no significant difference of total ATP between any of the compared lines (FIGURE 4-4). Therefore, the observed decrease in the ECAR and OCAR of these cells does not associate with either a growth or cellular ATP phenotype. AMPK and Akt Signaling are Unaffected by DG Function With ATP levels so similar between the two cells, we next examined the energy sensing Akt/mTOR and LKB1/AMPK pathways. These two pathways signal in opposition of one another as Akt activity positively modulates proliferation signaling (Lawlor and Alessi 2001) while the AMPK pathway inhibits proliferation and promotes the catabolic process of autophagy (Liang, Shao et al. 2007). Each pathway senses nutrient status within the cell and responds to promote either proliferation or energy storage under the appropriate nutrient conditions. We first assessed mtorc1-mediated signaling in our PC-3E + DG panel. We blotted for pulk1 S757, an inactivating phosphorylation downstream of mtorc1 (Kim, Kundu et al. 2011), as well as canonical markers of mtorc1 activity, both upstream and down (Akt S473 and ps6, respectively). Under basal conditions, there is no discernible difference in any of the markers of mtorc1 signaling between the DG and LARGE2 disrupted cells. Additionally, after removing serum and glucose from the cells for four hours, we observe the expected loss

100 86 of both pakt S473 and S6 phosphorylation with no grossly observable difference in the phosphorylation status of Ulk1 S757 (FIGURE 4-5 A-D). Because of the robust nature of the PC-3E + cells, standard starvation models did not acutely activate the AMPK pathway very efficiently, so we utilized 2-deoxy-d-glucose (2-DG) to induce AMPK and determine whether DG expression modulated the capacity of AMPK to be activated by LKB1. Both DG knockdowns exhibit rapid and efficient phosphorylation of AMPK following 20 minutes of treatment with 20 µm 2-DG (FIGURE 4-5 E). Thus the capacity of both mtorc1 signaling and AMPK activation appears to be intact within the PC-3E + knockdown lines. Autophagic Capacity is Dependent Upon Dystroglycan Function In order to examine both AMPK-associated signaling and mtorc signaling, we introduced a new stress scheme that would allow the analysis of more downstream effects of activated AMPK. The central role of AMPK in metabolic signaling has been widely reported, but one pathway that is significantly associated with its activity is the induction of autophagy which can be measured by immunoblot analysis for the autophagosomeassociated protein LC3. Upon autophagosome formation, LC3 is lipidated and becomes more mobile during gel electrophoresis allowing for its specific quantification. By combining chloroquine, an inhibitor of lysosomal acidification with serum and nutrient starvation, we can effectively analyze a cell s autophagic flux including its basal activity as well as its capacity (Klionsky, Abdalla et al. 2012). Additionally, utilizing of rapamycin, an inhibitor of mtorc1 activity, further dissects the signaling pathway. Using this schema, we analyzed mtorc1 and AMPK signaling in the DG deficient MEFs. A representative blot of the results (FIGURE 4-6 A-B) demonstrates the relative changes induced by serum starvation and chloroquine treatment. ps6 levels are significantly inhibited by serum withdrawal, as are Akt phosphorylation levels (FIGURE

101 B, G, H). Furthermore, rapamycin clearly inhibits mtorc1 signaling while preserving Akt phosphorylation. ULK1 phosphorylation at each serine site demonstrates the expected patterning, and the observable phosphorylation of AMPK is grossly increased by serum starvation. LC3B-II accumulation is significantly decreased in the DG-deficient MEFs compared to WT controls (FIGURE 4-6 F). This difference is statistically significant when comparing the different cell lines by a paried t-test analysis (p=0.0015) but the only individual treatment that is significant is the serum condition without chloroquine (+/-) condition as assessed by multiple t-test analysis (p= , FDR=1.0%). The pathways are otherwise completely preserved between the two cell lines with no other notable differences observed (FIGURE 4-6 C-F). We then assessed the two cancer cell lines to determine if the autophagy phenotype is correlated in those lines. The disruption of DG function by either direct shrna downregulation or through disruption of its glycosylation causes a decrease in LC3-II accumulation in PC-3E +, but not 22Rv1, cells (FIGURE 4-7 A, B). Except for the serum starved, chloroquine containing (-/+) condition, these findings are not-significant as measured across multiple blots, but the observed trend is replicated across multiple experiments (n=3, FIGURE 4-7 C, D). Therefore, mtorc1 signaling is unaffected by DG function while AMPK activity is intact through Ulk1 phoshporylation, but DG disrupted cells exhibit decreased autophagic flux. DG Function Does Not Mediate Clonogenic Survival Following Nutrient Deprivation Because AMPK is efficiently induced by energetic stress, we next assessed whether clonogenic potential of the DG modified PC-3E + cells was affected by their reduced capacity to properly initiate autophagy. We tested cells treated between 1 and 7 days under either 5 mm glucose or 1 mm glucose conditions and first show the, as reported earlier, disruption of DG does not affect cell growth characteristics under nutrient

102 88 restriction (FIGURE 4-8 A, B). Colony number (FIGURE 4-8 C, D) and clonogenic fraction (FIGURE 4-8 E, F) are both reported for the analyzed wells (representative: FIGURE 4-8 G). Clonogenic fractions are largely unaffected with the exception of the shdg10 cells in both 5mM and 1mM glucose treatments. Collectively, there is a trend indicating that once starvation conditions are initiated, there is no difference in the colonies observed. Interestingly, Loss of DG or LARGE2 under control conditions may inhibit clonogenic potential, though the mechanism behind this is unclear, as AMPK activity and autophagy are relatively similar under non-stressed conditions. DG Mediates MRLCII Phosphorylation To determine whether Ulk1-mediated phosphorylation of MRLCII, a protein previously identified to require DG under energetic stress, requires DG in PC-3E + cells, we assessed its phosphorylation status in a serum deprivation kinetics experiment (FIGURE 4-9 A, B). In vector control PC-3E + cells, MRLCII is efficiently phosphorylated over 20 hours during serum deprivation. However, both DG and LARGE2 knockdown results in a significant reduction in overall phosphorylation. The time-course, as assessed by a one-way ANOVA shows a statistically significant difference between plko and shdg10 (p=0.0374) and a near-significant difference compared to shlarge2 (p=0.0595). The experiment was repeated focusing on the four hour time point as the observed kinetics appear to change at this time point (FIGURE 4-9 B). In the repeated experiments (n=4), there is no detectable difference in relative pmrlcii under basal conditions, but serum-starvation produces a ~3 fold increase in the vector controls, while knockdown of DG and LARGE2 both prevent this phosphorylation (FIGURE 4-9 C). Attempts at measuring pmrlcii in both MEFs and 22Rv1s were unable to detect the protein by immunoblot. Regardless, these findings suggest that the

103 89 requirement proposed by the drosophila findings demonstrating a requirement of DG for phosphorylation of MRLCII are recapitulated in the prostate cancer cell line, PC-3E. Conclusions Since discovering DG is downregulated in advanced tumors, there has been significant effort by the field to determine the prognostic, mechanical, and signaling consequence of this tumor-associated alteration (Henry, Cohen et al. 2001, Sgambato, Camerini et al. 2007, Mitchell, Mathew et al. 2013). These studies have been able to continuously demonstrate a prognostic association between DG expression and disease progression/mortality (Moon, Rha et al. 2009, Sgambato, Camerini et al. 2010, Shen, Xu et al. 2012). Unfortunately, attempts to demonstrate a functional role for DG in vitro has been limited to studies that provide little insight regarding signaling (Thompson, Moore et al. 2010, Mitchell, Mathew et al. 2013), or studies that rely on overexpression of DG or its glycosyltransgerases to produce a phenotype (Bao, Kobayashi et al. 2009, Akhavan, Griffith et al. 2012, Esser, Miller et al. 2013). One study did demonstrate an interesting functional role related to STAT5 signaling required for mammary outgrowth in preparation for lactation, but the lack of a follow-up study and lack of a mechanistic description leave the findings difficult to interpret (Akhavan, Griffith et al. 2012). Herein, we attempt to demonstrate the functional results of directly disrupting dystroglycan via targeted shrna knockdown of DG or its required glycosyltransferase, LARGE2. These studies demonstrate that standard in vitro characteristics of cells are not grossly perturbed by disrupting DG, but reveal a potential role for DG in modulating phosphorylation of MRLCII that provides a mechanism by which DG is required for nutrient deprivationinduced signaling. While stressed conditions provide an interesting insight into the potential microenvironments that cancer cells encounter during primary tumor growth and metastatic

104 90 dissemination, we primarily wanted to determine if modulation of DG function has a measurable effect on cell growth and survival characteristics. We show that DG is not required for normal growth in standard tissue culture conditions, but there is an apparent reduction in overall metabolic rate within these cells as assessed by the Seahorse bioanalyzer. The lack of a difference in ATP, AMPK phosphorylation, or mtor signaling indicates that these cells, while exhibiting an apparent metabolic deceleration, are not responding as if the deficit is stressing the cellular machinery. These metabolic findings remain an enigma as there is still no clear mechanistic hypothesis to explain the observed changes. Additionally the findings are in direct opposition to those present in the DG-null drosophila model that showed a significant increase in its basal metabolic rate (Takeuchi, Nakano et al. 2009). Preliminary work has been performed to look at mitochondrial health via a number of experimental measures, but thus far these findings have demonstrated no change in the mitochondria (data not shown), and the lack of a mechanistic hypothesis makes continued investigation difficult to direct. The AMPK related signaling pathways associated with DG in previous studies remain some of the most intriguing possibilities to study. AMPK has been demonstrated to signal through an Ulk1-ZIPK-mRLCII pathway to mediate matg9 localization which presents the most appealing hypothesis for describing the mechanism by which DG could modulate autophagic capacity (Tang, Wang et al. 2011). Unfortunately, the critical paper in maintaining this hypothesis was recently retracted, as its findings regarding nutrient deprivation were attributed to an artifact generated by the starvation condition (Mirouse, Christoforou et al. 2009). Without this data, a clear pathway by which DG could function to modulate MRLCII phosphorylation is significantly muddled. The established connection between DG and myosin within skeletal muscle is clear, but the utrophin-dg connection in epithelium is more tenuous, and there is little data to link MRLCII to this complex. Further work must be performed to clarify this interaction. It is possible that loss of DG results in decreased stability or altered organization of MRLCII which would

105 91 then result in decreased availability of MRLCII for ZIPK (or myosin light chain kinase)- mediated phosphorylation. One of the looming questions posed by multiple biomarker analyses is whether DG expression or glycosylation are more frequently affected during tumor progression. Evidence exists for both possibilities, but there remains a distinct possibility that loss of DG could allow for a different phenotype than loss of glycosylation. β-dg contains a PPxY motif capable of WW domain interactions (Moore and Winder 2010), therefore its expression, regardless of glycosylation could function as a molecular scaffold for other proteins. However, glycosylation of α-dg is implicated in mediating a phosphorylation of β-dg that can mediate its interactions with proteins and function as a switch between WW domain and SH2 domain containing proteins. Therefore, the discovery that both impaired glycosylation and DG expression exhibit a similar autophagy and MRLCII phenotype indicates that DG requires laminin binding capacity for its role in mediating MRLCII phosphorylation and that expression of a DG incapable of binding ligand is unable to rescue this phenotype. Collectively this data demonstrates that DG function, interrupted either by knockdown of DG LARGE2, results in no discernible growth phenotype in prostate cancer cells or MEFs. However, there appears to be a significant effect of DG disruption on cellular metabolism, autophagic induction, and MRLCII phosphorylation. Currently, no clear mechanistic pathway is described, and a more complete understanding of epithelial DG physiology will be required before this may be effectively investigated.

106 Figure 4-1 Inhibition of DG is achieved through shrna-mediated knockdown and Cre-mediated knockout of either DG or LARGE2. Immunoblot and flow cytometry analysis to examine the knowckdown efficiency of shrnas targeting either DG or LARGE2. A-C. Immunoblots performed using Beta- DG (8D5) antibody in PC-3E +, 22Rv1, and MEFs. Actin is is shown as a loading control. D-E. DG surface glycosylation analyzed by flow cytometry demonstrates effect of LARGE2 knockdown in PC-3E + and 22Rv1. F. Ad.GFP or Ad.Cre infected DEF fl/fl MEFs demonstrate the effective loss of surface localized glycosylated DG following Cre excision. Isotype is shown in grey. 92

107 Figure 4-2 Inhibition of DG function either by disruption of DG expression or its glycosylation causes no discernible growth phenotype. Cellular proliferation analyzed by cell counts on the indicated day in PC-3E +, 22Rv1, and MEFs. A, D, F. With the PC-3E + DG panel, cells were grown in either standard media (A), glucose deprived media (B), or serum deprived media (C) and fold growth was plotted relative to day 1 measurements. B, E, G. The same analysis was performed in the 22Rv1 panel. C. Proliferation analysis of MEFs in standard media. No statistical significance was identified in any of the comparisons. 93

108 Figure 4-3 Loss of DG function causes a significant reduction in both glycolytic activity and oxidative phosphorylation. Both extracellular acidification and oxygen consumption rates were determined using the Seahorse XF96 bioanalyzer. A-B. PC-3E + cells were analyzed over 8 measurements for metabolic utilization. C-D. The same analysis was performed with murine embryonic stem cells either excpressing (C3) or deficient for (B11) DG. E. Oxygen consumption was determine in MEFs. Injection of 300 nm FCCP is indicated by the F and arrow. Injection of 1 µm oligomycin is indicated by the O and arrow. 94

109 Figure 4-4 Relative ATP is unaffected by DG status in PC-3E +, 22Rv1, and MEF cells. ATP was assessed using the ATP determination kit and plotted as ATP relative to vector controls. A. Relative ATP comparing shlarge2.148 PC- 3E + cells and vector control. B. Relative ATP comparing shdg1 cells and vector control in 22Rv1s. C. Relative ATP comparing DGWT and DGKO MEFs. All statistical comparisons were performed using an unpaired, t-test. 95

110 Figure 4-5 mtorc1 signaling is unaffected by DG function in PC-3E + cells. Immunoblot analyses showing mtorc1 and AMPK associated signaling in PC-3E + cells. A, E. Representative images of immunoblot analyses for ULK1 phoshphorylation at S757 and S555, AKT phosphorylation at S172, AMPK phosphorylation at S172, and S235/236 phosphorylation of ribosomal protein S6. Actin was probed as a loading control. B-D. Quantification of 3 replicates from blot A. 96

111 Figure 4-6 Loss of DG reduces autophagy-associated LC3 accumulation. Immunoblot analysis and quantification of AMPK and AKT signaling in MEFs. Cells were treated with the indicated combination of serum, rapamycin, and choloroquine for 4 hours prior to lysis. A. Representative images are shown of immunoblots analyzing the indicated proteins. B. AKT phosphorylation was analyzed via immunoblot in the indicated conditions in MEFs. Actin was used as a loading control.c-f. Quantification of 3 experiments of blot A. G-H. Quantification of blot B. 97

112 98 * Figure 4-7 DG-associated disruption of autophagic flux is observed in PC-3E + but not 22Rv1 cells. Immunoblot analysis probing for autophagic marker LC3B in PC-3E + and 22Rv1 cells. A, C. Representative images are shown of immunoblots against LC3B following a 4 hour treatment of cells in the indicated combination of serum and chloroquine. B, D. 3 experimental replicates were quantified, normalized to actin, and made relative to vector control +/- conditions. * = p <0.05. The absence of statistical indication implies lack of significance.

113 Figure 4-8 DG does not affect clonogenic potential following glucose deprivation. Clonogenic survival was performed following glucose deprivation in either 5 mm or 1 mm glucose containing media for the indicated time period. A, B. Cell growth was analyzed by cell counts and made relative to day 0 counts during the course of setting up the clonogenic assay. C-D, Following plating of 300 cells and growth for 14 days, stained colonies were counted and absolute colony number is shown. E-F. Relative clonogenic potential for each of the cell lines was calculated based upon #colonies in control/#colonies in stressed condition. G. Representative images of single wells from the indicated time following growth in 1 mm glucose containing DMEM. 99

114 Figure 4-9 Loss of DG function prevents serum-starvation induced phosphorylation of MRLCII. Immunoblot analysis for T18/S19 phoshproylation of myosin regulatory light chain II in PC-3E + cells. A. Serum starvation for the indicated times was performed and analyzed for the phosphorylation of MRLCII. B. Quantification of A. C. Quantification of 6 experiments analyzing MRLCII phosphorylation following 4 hours of serum starvation. * = p <

115 101 CHAPTER V PROSTATE-SPECIFIC DELETION OF DYSTROGLYCAN DOES NOT EXACERBATE DISEASE IN A PTEN-DFIEICIENT MOUSE MODEL OF PROSTATE CANCER Introduction Murine models of human disease allow investigation into novel mechanisms of tumor development and progression while also providing more complex and functionally relevant methods than those afforded by in vitro experimentation. Utilizing tissuespecific Cre recombinase expression coupled with homologous recombination for placement of LoxP sites has become a common and focused technique to analyze the effects of localized gene deletion (Sauer and Henderson 1988, Nagy 2000). Engineering of these animals not only allows a better model of human disease development and progression, but also facilitates highly targeted investigation into the roles of a specific protein or proteins in the mechanism of disease progression. Dystroglycan (DG) is a protein that been firmly established to be dysregulated either by improper hypoglycosylation or by loss of DG expression in a variety of tumor types (Henry, Cohen et al. 2001, Sgambato, Migaldi et al. 2003, Sgambato, Tarquini et al. 2006). In order to determine the functional necessity of DG in the developing prostate, a prostate-specific deletion of DG using probasin promoter-driven Cre recombinase was established. Surprisingly, the mouse model exhibited largely preserved basement membrane integrity, apical/basal cell polarity, and regenerative capacity following castration (Esser, Cohen et al. 2010). The authors noted that the Pb-Cre promoter used to drive Cre recombinase expression was only active in the luminal epithelial cells of the prostate, while the basal cells maintained appropriate DG expression. Therefore, to

116 102 determine whether the basal cell population was able to maintain the functional capacity of DG in this setting, unpublished follow-up work was performed demonstrating that deletion of DG in the basal cell population via keratin 5 promoter-driven Cre expression results in a similar phenotype. Therefore, DG, while frequently disrupted in cancers, is not necessary for development of the murine prostate. In order to more directly determine the role of dystroglycan in prostate tumor progression, we sought to combine tissue-specific DG knockout with a previously described murine prostate cancer model. The phosphatase and tensin homolog deleted on chromosome 10 (PTEN) is a protein that is frequently mutated in human prostate cancer (Cairns, Okami et al. 1997, Li, Yen et al. 1997). PTEN removes the phosphate on phosphatidylinositol (3,4,5)-triphosphate and prevents downstream activation of the progrowth Akt pathway (Maehama and Dixon 1998). Prostate-specific deletion of PTEN on the C57BL/6 background provides an ideal backdrop for combinatorial DG deletion. The PTEN model exhibits very slow disease progression with micro invasive cancer noted only in the 12 month-old animals (Svensson, Haverkamp et al. 2011). Furthermore, while PTEN knockout animals on different genetic backgrounds exhibit metastases and highly aggressive disease (Wang, Gao et al. 2003), on the C57BL/6 background no metastases were noted and disease morbidity and mortality was largely limited to urogenital obstruction due to fluid-filled enlargement of anterior prostate glands. Therefore, this model demonstrates a much lower baseline disease severity which allows us to better determine the contribution of DG loss to disease progression. No mouse model has yet examined the effects of DG loss on cancer progression, and herein we describe the combination of a Pb-Cre transgenic mouse that deletes both the tumor suppressor PTEN and the extracellular matrix interacting protein dystroglycan. We show that loss of PTEN, as previously described, leads to extensive cellular atypia, prostatic hyperplasia, and a dramatic fibrotic response. Unfortunately, combinatorial loss of DG in these animals shows no significant effect on the progression of prostate cancer

117 103 at either 3, 6, or 12 months. Microinvasive disease was not readily detectable in any of the observed animals, and no lymph node metastases were observed in any of the 12 month old animals. Materials and Methods Animals All procedures involving animals were performed according to The University of Iowa Animal Care and Use Committee policies. Female C57BL/6 mice containing a floxed DG allele (Esser, Cohen et al. 2010) were crossed with C57BL/6, PB-Cre4+, floxed PTEN male mice (Svensson, Haverkamp et al. 2011) to generate Cre+ animals of each of the following four genotypes: DG fl/wt;pten fl/wt, DG fl/fl;pten fl/wt, DG fl/wt;pten fl/fl, and DG fl/fl;pten fl/fl. For genotyping, tail DNA was extracted (REDExtract-N-Amp Tissue PCR Kit, Sigma) and PCR was performed. Eight animals for each genotype were assessed at 3, 6, and 12 months of age. Analysis of inflammation was carried out using POET-3 mice housed and treated by Tim Ratliff at Purdue University (Haverkamp, Charbonneau et al. 2011). POET-3 mice were either naïve or treated with 1 x 10 7 ovalbumin-specific CD8 + T cells and allowed to incubate for either 6 or 30 days prior to analysis. Tissue Collection and Processing All animals were euthanized by CO 2 inhalation prior to manual cervical dislocation. The urogenital sinus and the inguinal and lumbar lymph nodes were removed and placed in 4% paraformaldehyde for 24 hours then transferred to 70% ethanol and kept at 4 C. Following fixation, tissues were processed to isolate the prostate and remove surrounding

118 104 adipose tissue prior to paraffin embedding using a Ventana Automated Tissue Processor. Sections were cut at 5 µm and rehydrated using a xylene to ethanol gradient. Hematoxylin and eosin staining was carried out using standard protocols. Analysis of the samples was performed by the primary author and confirmed by veterinary pathology. Immunofluorescent Staining Sections were stained using either a IIH6 alpha-dg antibody (Santa Cruz) or a Panlaminin antibody. Briefly, following rehydration samples were treated with proteinase K at 37 C for 20 minutes for antigen retrieval. Tissues were then blocked for 1 hour at room temperature with a blocking solution containing 2% donkey serum, 1% BSA, 0.1% Triton X-100, and 0.05% Tween-20 in PBS. Primary antibody was diluted 1:100 (IIH6) or 1:200 (LMN) in the blocking solution and allowed to incubate overnight at room temperature in a humidity chamber. Appropriate secondary antibody with DAPI was applied in blocking solution to the tissue for one hour at room temperature and coverslips were mounted using Fluorogel (Electron Microscopy Sciences). Images were acquired using either an Olympus BX-61 or a Leica DM2500 microscope. Statistics Statistical significance was computed using a multiple comparison ANOVA with a Bonferroni s multiple comparison correction with the Graphpad Prism software. All p- values <0.05 were considered significant.

119 105 Results Dystroglycan Hypoglycosylation Occurs in a Mouse Model of Early Prostate Cancer Prostate cancer that developed in the C57BL/6 PTEN-deficient mice exhibits a slow disease progression that eventually includes microinvasive carcinoma. This animal accurately mimics the early stages of human prostate carcinoma, thus we sought to determine whether dystroglycan glycosylation is lost similar to the pattern seen in human samples. To analyze DG glycosylation in the PTEN-deficient mice we performed immunohistochemistry with the IIH6 antibody and scored samples based on a quartile scoring system where 3 = membranous staining seen throughout the gland, 2 = basal staining continuous around the gland, 1 = basal staining that is discontinuous, and 0 = absent staining (Figure 5-1A). We looked at a small number of animals from either the PTEN wild type or PTEN knockout cohort (Figure 5-1B). We found that animals at the earliest ages (9-12 weeks) have very high levels of DG glycosylation throughout the glands. At 25 weeks, there is a small, insignificant difference in IIH6 staining between PTEN WT and PTEN KO prostates. Between 25 and 52 weeks of age, the animals lose a significant portion of DG glycosylation in the prostate, and from 52 weeks through weeks, the animals exhibit a significant reduction in DG glycosylation relative to the PTEN WT controls (Figure 5-1C). Thus loss of DG glycosylation occurs in a manner similar to human disease. The PTEN deficient mouse model of prostate cancer exhibits extensive immunoreactive stromal changes that increase with age (Svensson, Haverkamp et al. 2011). In order to determine whether DG glycosylation was directly affected by the increased immune response, we gathered tissues from a mouse model of prostatitis that relies on injection of CD8 + T cells which specifically react with the prostate epithelium (Haverkamp, Charbonneau et al. 2011). We compared naïve animals to those injected

120 106 with T cells and found that regardless of whether prostatitis was present for 6 or 30 days, both groups were statistically inseparable from the naïve group (Figure 5-1D). Therefore, short-term induction of an inflammatory infiltrate is unable to reduce DG glycosylation in a mouse model of prostatitis. Prostate Histology at 3 Months Using the standards put forth by the most recent consortium publication describing murine models of prostate cancer (Ittmann, Huang et al. 2013), animals were first analyzed for gross histopathologic abnormalities. Previous reports have shown that heterozygous loss of DG has little effect on protein distribution and function (Henry and Campbell 1998, Esser, Cohen et al. 2010). Additionally, heterozygous loss of PTEN yields a phenotype only observed in mice older than 6 months (Svensson, Haverkamp et al. 2011). Predictably, DG fl/wt ;PTEN fl/wt animals appear to have normal glandular architecture of the prostate with no evidence of cellular hyperplasia, atypical cellular architecture, or increase in the interleaved fibrous stroma of the prostate. The DG fl/fl ;PTEN fl/wt mice also exhibit benign prostate histology that was indistinguishable from DG fl/wt ;PTEN fl/wt animals (Figure 5-2A-F). In contrast to the PTEN fl/wt animals, both PTEN fl/fl genotypes exhibit striking phenotypes (Figure 5-2G-L). Both sets of animals show significant cellular hyperplasia most prevalent in the dorsal and lateral lobes, as previously reported (Svensson, Haverkamp et al. 2011). Furthermore, at high magnification, cellular and nuclear atypia are apparent in a high proportion of the glands (Figure 5-2I,L). Based upon these findings, the majority of these glands exhibit characteristics of mpin grade 3 and 4 (Park, Walls et al. 2002). Finally, both of these genotypes exhibit an extensive fibroblastic response termed desmoplasia. While this feature is descriptive of adenocarcinoma according to pathology guidelines, the lack of an infiltrative glandular architecture

121 107 implies that these glands remain defined as high grade mpin. In conclusion, at 3 months, the phenotype observed is dominated by the PTEN status of the mouse, and DG expression does not result in worsening disease progression at this time point. Prostate Histology at 6 Months Similar to the 3 month animals, no detectable prostate pathology is seen in either the DG fl/wt ;PTEN fl/wt or the DG fl/fl ;PTEN fl/wt mice (Figure 5-3A-F). In comparing the relative sizes of prostates between the two DG variants on the PTEN fl/wt background, there is no discernible difference. Of the 16 animals examined that are heterozygous for PTEN, only 1 animal from each DG genotype shows any sign of pathology. The cellular architecture is unchanged, but each of these animals exhibit a single small focus of inflammation. There are no associated architectural changes associated with the inflammatory response, and the equivalent incidence between DG fl/wt and DG fl/fl mice indicates that this finding is unlikely to be due to differential expression of DG. In animals with homozygous PTEN deletion, prostates continue to show changes associated with the development of adenocarcinoma. Cellular hyperplasia was apparent on all prostatic lobes with nearly all glands showing mild nuclear atypia consistent with mpin grades 3 and 4. The desmoplastic response increased in overall severity with large pockets of inflammatory cells observed in many of the glands. By 6 months, all PTEN fl/fl animals also exhibit destruction of prostate glands by fibrous tissue. The relative occurrence of this event appears to be independent of DG expression. Furthermore, inflammatory cells and cellular debris are commonly seen within glandular lumens. Similar to the findings in the 3 month animals, PTEN status is the primary driver of disease progression in these animals, and DG status appears not to contribute to any observed histopathologic features.

122 108 Gross Pathology and Prostate Histology at 12 Months Grossly, PTEN fl/wt animals are indistinguishable from 12 month wild-type C57BL/6 animals. There is no observable pathology in either the prostates or the nearby lumbar and inguinal lymph nodes. In contrast, the 12 month PTEN fl/fl animals, regardless of DG status, exhibit significant fibrotic reactions surrounding the prostate causing adhesion between the various lobes of the prostate. The anterior prostates are grossly enlarged with 100% penetrance, and the anterior lobes are significantly enlarged due to fluid retention in the absence of any solid tumor growth. Furthermore, many of the animals exhibit enlarged and fused seminal vesicles with necrotic foci (Figure 5-4). By 12 months all of the animals exhibit histologic evidence of cellular hyperplasia and focal areas of thickened stroma, however, the PTEN fl/wt animals exhibit this at a much lower rate and severity when compared to the PTEN fl/fl animals. There is now evidence of focal cellular and nuclear atypia consistent with mpin in the PTEN fl/wt animals, but the frequency appears to be very low. These findings are consistent with the previous report that PTEN heterozygotes developing mpin lesions by one year of age (Svensson, Haverkamp et al. 2011). At this time point, all of the animals exhibit pockets of inflammation scattered throughout the prostate that include lymphocytes, macrophages, and scattered neutrophils. Loss of DG expression does not exacerbate any of these findings, and the two PTEN fl/wt genotypes are indistinguishable by gross histology. Similar to previous age groups, PTEN fl/fl animals all share the same pathologic characteristics regardless of DG expression. At this time, extensive desmoplastic reaction is visible in all lobes of the animals. The normal morphology of the glands is completely obliterated with some glands showing marked dilation with small, atrophied epithelia. Inflammatory infiltrates now include lymphocytes, plasma cells, neutrophils and macrophages. Cellular hyperplasia is hallmarked by cribriform and papillary patterning

123 109 extending into the lumen of the gland. There is considerable nuclear and cellular atypia with enlarged nuclei and the occasional prominent nucleoli. In order to determine whether loss of DG has any functional consequence on basement membrane organization, we utilized a pan-laminin antibody to assess the integrity of the basement membrane. DG fl/fl animals exhibit incomplete knockout of the protein in prostatic tissue at 12 months of age (Figure 5-6D,J; arrows). This finding has been previously reported when using the probasin promoter ARR 2 PB and is thought to be caused by maintained basal cell DG expression (Wu, Wu et al. 2001, Esser, Cohen et al. 2010). The majority of glands in all samples exhibited intact basement membranes as assessed by pan-laminin staining (Figure 5-6B,E,H,K; arrowheads). There is potentially a modest reduction in laminin staining in the DG fl/fl ;PTENfl /fl prostate, but the reduction is very mild and the surrounding signal from the desmoplastic reaction makes quantification very difficult. Staining of the muscle surrounding the urethra demonstrates specificity of the Pb-Cre mediated DG (DAG1) recombination. Due to the extensive desmoplastic response observed in both PTEN fl/fl genotypes, it is difficult to determine whether epithelial cells truly demonstrate a microinvasive phenotype. In order to evaluate whether any of the neoplastic cells metastasized, local lymph nodes including the lumbar and inguinal nodes were removed at necropsy and examined histologically. Grossly, PTEN fl/fl animals exhibited larger lymph nodes at both sites, and this difference is exaggerated in the 12 month age group (data not shown). Histologically, these nodes exhibit increased cellularity with an abundance of macrophages, lymphocytes, and some scattered mast cells. These cells are found throughout the medullary sinus and subcapsular sinus. Unfortunately, prostatic epithelial cells were not observed in any of the nodes. No architectural changes were present and no evidence of disruption of the subcapsular sinus was observed. Androgen receptor staining was attempted, but reactive cells from the lymph nodes exhibited very high background staining making the elimination of false positives all but impossible.

124 110 Discussion Dystoglycan function is frequently lost in the epithelial compartment of carcinomas from a variety of tissues (Henry, Cohen et al. 2001, Sgambato, Migaldi et al. 2003, Sgambato, Tarquini et al. 2006, Sgambato, Camerini et al. 2007). Interestingly, its expression is not required for appropriate development or homeostasis of the murine prostate (Esser, Cohen et al. 2010). We and others have shown through in vitro experiments that rescue of DG function results in decreased cellular proliferation and migratory capacity (Bao, Kobayashi et al. 2009, Esser, Miller et al. 2013). Unfortunately, DG knockdown experiments are exceedingly limited and the effects of directly mitigating dystroglycan function in a tumor background have been shown to produce changes in cellular adhesion (Thompson, Moore et al. 2010) and anchorage-independent growth (Mitchell, Mathew et al. 2013). We therefore generated a prostate epithelium-specific DG knockout in the background of a PTEN-deficient mouse model of prostate cancer. We show that both gross prostate anatomy and histopathology are unchanged with loss of DG. Furthermore, we show that basement membrane organization is largely maintained and glandular architecture is entirely dependent on PTEN expression status. Therefore, while DG may play a role in tumor progression in human disease, in a PTEN-deficient model of prostate cancer, DG knockout has no effect on overall disease progression through 12 months of age. We furthermore show a previously unreported finding that PTEN-deficient mouse prostate carcinomas demonstrate an age-dependent loss of IIH6 immunoreactivity. This finding mimics the observed inverse relationship between DG glycosylation and Gleason grade, a common histopathologic scoring system used for prostate cancer (Shimojo, Kobayashi et al. 2011, Esser, Miller et al. 2013). We had previously observed a subjective relationship between DG glycosylation and immune infiltrates in human prostate samples, where hypoglycosylation was frequently observed in the presence of a

125 111 prominent immune reaction. We tested this phenomenon directly using the POET-3 mouse model of acute prostatitis and found no changes in DG glycosylation upon induction of prostatitis. These findings collectively point to a change in either DG expression or glyscosyltransferase-mediated modification that occurs over time due to the cellular proliferation induced by PTEN deletion. Therefore, multiple possibilities arise that could explain this observation. Because of the delay in IIH6 loss, it is unlikely that DG function is abrogated due to direct effects from activated Akt signaling. Therefore, a prime candidate for upstream control of DG-associated gene changes may be the previously described p53-acitvated senescence pathway this model (Chen, Trotman et al. 2005). This theory could be directly assessed by examining the prostate of PTEN -/- ;p53 -/- animals at corresponding time points to determine if DG glycosylation is similarly affected. Another possibility is that increased cellular proliferation leads to dysregulated DNA repair allowing for any of the identified DG glycosylation-associated genes to be mutated in these animals. This is less likely as the DG staining within these animals glands is more uniform than this theory would likely show. Finally, an as of yet unidentified consequence of PTEN deficiency could be the cause of this change. PTEN knockout in the context of prostate cancer has been reported in a number of genetic backgrounds with disease severity ranging from extensive metastatic dissemination to highly penetrant mpin with little evidence of invasion (Wang, Gao et al. 2003, Svensson, Haverkamp et al. 2011). The C57BL/6 background is most often associated with a less severe disease phenotype, which provides an ideal model system to assess whether disease severity may be increased via disruption of other genes including overexpression of ERG (Carver, Tran et al. 2009) or mutant K-ras (Mulholland, Kobayashi et al. 2012) and knockout of p53 (Chen, Trotman et al. 2005) or Smad4 (Ding, Wu et al. 2011), among others. Ultimately, progressing beyond the localized mpin phenotype requires modifications to major signaling pathways classically associated with tumor regulation. Therefore, it is perhaps unsurprising that knockout of DG produces no

126 112 such phenotype. While DG is a critical mediator of cell-extracellular matrix (ECM) interactions, with only PTEN disruption in the epithelial compartment there remains a sufficient number of molecular inhibitors to metastasis that DG disruption is incapable of overcoming. It is therefore intriguing to examine the effects of dystroglycan disruption on the background of a more aggressive disease model to determine if DG function is critical in preventing dissemination in a different context. In addition to the multitude of signaling networks that prevent metastatic disease in the PTEN-deficient murine model, there are likely a number of other ECM-interacting proteins such as the integrin family that maintain expression in the prostatic epithelium. While knockout of DAG1 in mice leads to embryonic lethality due to disruption of Reichert s Membrane (Williamson, Henry et al. 1997), an epiblast-specific deletion of DG shows that animals are capable of surviving to birth (Satz, Barresi et al. 2008). This bolsters the idea that while DG may specifically be necessary for Reichert s development, other basement membranes found throughout the animal are able to maintain integrity via alternative ECM-interacting proteins. This functional redundancy implies that loss of dystroglycan on a homogenous genetic background is not sufficient to disrupt the basement membrane integrity in a manner capable of promoting metastatic dissemination. Therefore, it is likely that other ECM-interacting proteins such as integrins are capable of maintaining basement membrane adhesion and stability in the murine prostate. In this model metastatic dissemination was primarily assessed by analysis of the primary site and local lymph nodes. At necropsy, all organs were analyzed for signs of gross abnormalities, but there still exists the possibility that small areas of micrometastases might be present in these distant sites, or even within the unanalyzed portions of the procured tissue. Utilization of a bioluminescent reporter similar to that used in the original PTEN study from our lab could provide an opportunity for both noninvasive monitoring of disease to identify extra-prostatic growth as well as a more

127 113 thorough analysis of tissues ex vivo to address the possibility of micrometastases. In closing, there were no observable effects of DG deletion on a PTEN-defieicent mouse model of prostate cancer. Therefore, either dystroglycan does not contribute to disease progression at this point in the disease process, or its contributions are sufficiently limited that no difference could be observed by the methods employed in this study. Future work using more aggressive models, longer incubation times, and reporter genes may allow for a better analysis of the role of dystroglycan in prostate cancer progression.

128 114 *** *** Figure 5-1 PTEN deletion but not acute inflammation associate with loss of IIH6 immunoreactivity in the mouse prostate. A. IIH6 staining divided into quartiles with 3 = membranous staining seen throughout the gland, 2 = basal staining continuous around the gland, 1 = basal staining that is discontinuous, and 0 = absent staining. B. Number of animals examined to determine IIH6 retention in the PTEN fl/fl animal. C. Mean glandular score of the examined animals with a significant difference observed at 52 and 70 weeks (***, p < ). D. Analysis of IIH6 staining in POET mice treated with ovalbumin sensitized CD8 + T cells (OT-1) or naïve. No significant differences exist between any of the groups.

129 Figure 5-2 Three month histology shows both desmoplasia and nuclear atypia in PTEN FL/fl prostates. Representative images taken at various magnifications of each of the four examined genotypes. Both DG fl/wt ;PTEN fl/wt (A-C) and DG fl/fl ;PTEN fl/wt (D-F) show highly similar benign histology with no cellular atypia and the normal murine scant stroma. Similarly, both DG fl/wt ;PTEN fl/fl (G-I) and DG fl/fl ;PTEN fl/fl (J-L) demonstrate a marked desmoplastic response (marked with *). At high power, both PTEN fl/fl genotypes also show cellular hyperplasia and nuclear atypia (marked by arrow) that is characteristic of mpin lesions. Black scale bar = 100 µm. 115

130 Figure 5-3At 6 months, the desmoplastic response and cellular atypia are exaggerated in PTEN fl/fl prostates. Representative images taken at 4x-20x magnification of each of the four genotypes. The lack of pathology in both DG fl/wt ;PTEN fl/wt (A-C) and DG fl/fl ;PTEN fl/wt (D-F) is again seen at 6 months, similar to the findings at 3 months. Both DG fl/wt ;PTEN fl/fl (G-I) and DG fl/fl ;PTEN fl/fl (J-L) exhibit extensive desmoplastic changes (marked by *). An example of immune infiltrate is marked by #. Black scale bar = 50 µm. 116

131 Figure 5-4 Gross histopathology of PTEN fl/fl animals demonstrate significant anterior prostate swelling with associated seminal vesicle necrosis. A., E. Gross palpation of the anterior prostate is easily performed in live mice and outgrowth is clearly visible. B., F. the outgrowth is contained within the abdominal cavity with no associated adhesions to the abdominal pleura. C., G. Exposure of the genitourinal tract shows swelling of both the anterior prostate lobes (arrows) and the closely associated seminal vesicles (*). D., H. Ex vivo imaging of the genitourinal tract with anterior prostate, seminal vesicles, and bladder (+) representing the grossly identifiable components. 117

132 Figure 5-5 Disease progression 12 months is driven entirely by loss of PTEN expression. Representative images of 10X and 40X of the murine prostate across the 4 genotypes. A-E. At 12 months, both PTEN fl/wt genotypes begin to exhibit hypercellularity and nuclear atypia (arrows) consistent with mpin. G- K. PTEN fl/fl animals exhibit a marked desmoplastic response with overall destruction of the prostate architecture and increased cellular hyperplasia. An example of cribriform patterning is shown (*, K). Scale bar = 50 µm 118

133 Figure 5-6 Laminin organization is maintained in the murine prostate regardless of DG expression. Representative images of IIH6, laminin, and DAPI staining of prostate from all four examined genotype. A-C. DG fl/wt ;PTEN fl/wt prostates show strong DG and laminin staining. D-E DG fl/fl ;PTEN fl/wt prostates show markedly reduced DG expression in the basement membrane, but laminin organization is unaffected. F-G DG fl/wt ;PTEN fl/fl prostates exhibit the characteristic cellular proliferation with maintained DG expression and laminin organization. H-I DG fl/fl ;PTEN fl/fl exhibit significantly reduced DG expression while laminin organization remains largely intact. M, Para-urethral muscle IH6 staining highlights PB-Cre4 specificity. Arrows indicate maintained DG expression in DG fl/fl animals. Arrowheads indicate basement membrane staining of laminin. 119