EXPLOITING THE USE OF PLASMINOGEN KRINGLE DOMAINS FOR CANCER THERAPY

Sabrina R. Perri Division of Experimental Medicine, Deparlment of Medicine Faculty of Medicine, McGiII University Montreal, Quebec, Canada February 2007

A thesis submitled to the Faculty of Graduate Studies and Research in partial fulfillment of the requirements of the degree of Doctor of Philosophy (Ph. O.).

©2007 Sabrina R. Perri Library and Bibliothèque et 1+1 Archives Canada Archives Canada Published Heritage Direction du Branch Patrimoine de l'édition

395 Wellington Street 395, rue Wellington Ottawa ON K1A ON4 Ottawa ON K1A ON4 Canada Canada

Your file Votre référence ISBN: 978-0-494-32376-2 Our file Notre référence ISBN: 978-0-494-32376-2

NOTICE: AVIS: The author has granted a non­ L'auteur a accordé une licence non exclusive exclusive license allowing Library permettant à la Bibliothèque et Archives and Archives Canada to reproduce, Canada de reproduire, publier, archiver, publish, archive, preserve, conserve, sauvegarder, conserver, transmettre au public communicate to the public by par télécommunication ou par l'Internet, prêter, telecommunication or on the Internet, distribuer et vendre des thèses partout dans loan, distribute and sell th es es le monde, à des fins commerciales ou autres, worldwide, for commercial or non­ sur support microforme, papier, électronique commercial purposes, in microform, et/ou autres formats. paper, electronic and/or any other formats.

The author retains copyright L'auteur conserve la propriété du droit d'auteur ownership and moral rights in et des droits moraux qui protège cette thèse. this thesis. Neither the thesis Ni la thèse ni des extraits substantiels de nor substantial extracts from it celle-ci ne doivent être imprimés ou autrement may be printed or otherwise reproduits sans son autorisation. reproduced without the author's permission.

ln compliance with the Canadian Conformément à la loi canadienne Privacy Act some supporting sur la protection de la vie privée, forms may have been removed quelques formulaires secondaires from this thesis. ont été enlevés de cette thèse.

While these forms may be included Bien que ces formulaires in the document page count, aient inclus dans la pagination, their removal does not represent il n'y aura aucun contenu manquant. any loss of content from the thesis. ••• Canada ABSTRACT

It is well characterized that tumors depend upon in order to progress and form metastases. In this regard, strategies to inhibit neovascularization have been developed either aimed to target endothelial cells directly or indirectly via suppression of pro-angiogenic growth factors produced by the tumor or surrounding stromal cells.

Several anti-angiogenic agents are present endogenously in our body as cleavage products of larger precursor . Angiostatin is one of the most widely studied inhibitors of angiogenesis and encompasses the first 3 or 4 kringle domains of human plasminogen (PIg). Of particular interest, is the fifth kringle of human Plg (K5), which displays higher anti-angiogenic potency than angiostatin. In fact, kringles 1 to 5 exhibit greater inhibitory activity against endothelial cells than angiostatin, and this finding suggests that K5 domain acts synergistically to enhance the anti-angiogenic effect of angiostatin. Employing a gene-based approach for the delivery of an anti-angiogenic agent provides sustained, biologically relevant levels of the therapeutic of interest and circumvents the limitations associated with direct recombinant protein use, such as the necessity for prolonged repeated administration, high dosages, high expense and risk of toxicity. Therefore, we proposed that the K5 domain --Qn its OWll- could act as a potent anti-angiogenic and anti-tumor protein within the context of a cancer gene therapy strategy.

In our first study, we assessed the angiostatic properties of the K5 peptide domain

III an orthotopic brain cancer model. We demonstrated that the disulfide bridging conformation of K5, necessary to maintain its functionality, is conserved upon secretion by gene-modified mammalian cells. Kringle 5 retrovirally gene-engineered human U87

11 glioma celIs produced functional soluble KS protein capable of suppressing -induced endothelial celI migration in vitro and inhibiting glioma-induced angiogenesis in vivo. Interestingly, secreted KS prote in blocked the recruitment of tumor­ associated CD4S+Mac3+0rl" macrophages in vivo and inhibited the migration of CD206+ human monocyte-derived macrophages in vitro. Moreover, in a clinicalIy relevant orthotopic glioma model, soluble KS induced long-term survival in a majority of test animaIs. Thus, these findings validate the use of a gene therapy approach to deliver Plg

KS protein and suggest that KS acts as a novel 2-pronged anti-tumor agent, mediating its inhibitory effect via its action on host-derived endothelial ceUs and tumor-associated macrophages.

To determine if KS mediated its anti-tumor effect by modulating other immune effector cells, we tested the use of soluble KS in a murine DAl3 mammary adenocarcinoma mode!. Soluble KS produced by retrovirally gene-modified DAl3 ceUs led to long-term survival (over 1 year) of immunocompetent BALB/c mice however, we failed to observe tumor rejection in immunodeficient NOD-SCID and BALB/c nude mice. Further analysis revealed that KS enhances the recruitment of tumor-infiltrating

CD3+ lymphoid cells, in particular the natural killer T (NKT)-lymphocyte phenotype.

Consistent with our previous findings, we demonstrated that KS led to a significant decrease in tumor-associated microvessel length and density. Interestingly, KS tumors were characterized by a robust neutrophilic infiltrate. This may be explained by the ability of KS to act as a strong chemotactic agent for human neutrophils in vitro as weU as its ability to promote CD64+ activation within the CD 11 b+ neutrophil phenotype.

These findings confirm that KS acts as a potent angiostatic agent and possesses a novel

111 pro-inflammatory role via its ability to recruit tumor-associated neutrophils and NKT­ lymphocytes, leading to a strong anti-tumor response.

Tumor-associated macrophages (TAM) are key immune effector cells implicated in promoting tumor progression and metastasis. It would thus be desirable to explore strategies to reduce T AM infiltration within the tumor microenvironment. In our first study, we demonstrated that soluble K5 protein blocks macrophage recruitment. In addition, the recent observation that angiostatin reduces macrophage infiltration in an atherosc1erosis model prompted our laboratory to further explore the use of angiostatin as an anti-macrophage agent. We demonstrated that angiostatin suppresses the in vitro migration of both murine peritoneal macrophages and human monocyte-derived CD206+ macrophages. Furthermore, we showed that angiostatin led to a decrease in the gelatinolytic activity of macrophage-produced matrix -9, which may explain, in part, the observed angiostatin-mediated inhibition of migration. Additionally, we detected the presence of the /3-subunit of A TP synthase on the cell-surface of macrophages. A TP synthase was previously found to be a receptor for angiostatin on the cell-surface endothelial cells. We propose that the presence of A TP synthase on the surface of macrophages may promote interaction with angiostatin and prevent migration, similar to what has been reported with endothelial cells. Our findings suggest that angiostatin holds promise as an inhibitory agent against macrophages.

Overall, our data supports the use of Plg K5 protein as a potent anti-angiogenic agent and reveals that K5 as weIl as angiostatin possess novel immune modulatory properties that potentially enhance their utility as therapeutic agents against cancer.

IV RÉSUMÉ

Il a été largement cerné que les tumeurs dépendent de l'angiogenèse afin d'accomplir leur progression et former des métastases. À cet effet, des méthodes visant à inhiber la néovascularisation ont été élaborées afin de cibler les cellules endothéliales de façon directe ou indirecte via la suppression des facteurs de croissance pro-angiogéniques produits par la tumeur ou les cellules stromales environnantes. Plusieurs agents anti­ angiogéniques demeurent présents endogéniquement dans notre organisme en tant que produits afférents à de plus vastes protéines et ce en tant que précurseurs. L' angiostatine s'avère être un des inhibiteurs les plus étudiés et englobe les trois ou quatre premiers domaines kringle du plasminogène humain. II importe de noter qu'il s'agit du cinquième kringle du Plg humain (K5) qui démontre une capacité anti-angiogénique plus élevée que l'angiostatine. En effet, les kringles 1 à 5 présentent une activité inhibitrice supérieure face aux cellules endothéliales que ne le démontre l' angiostatine. De plus, cette découverte suggère que la région K5 agit synergiquement afin d'augmenter l'effet anti­ angiogénique de l'angiostatine. L'utilisation d'une approche de thérapie génique dans le transfert d'un agent anti-angiogénique fournit un niveau biologique significatif et soutenu de la protéine concernée et circonvient les limites associées à l'usage direct du recombinant de protéines telle que par exemple la nécessité d'administrer le produit de façon prolongée et répétée, en dosage élevé et en tenant compte du risque élevé de toxicité. Nous proposons donc que le domaine K5 agit de lui-même comme puissant agent anti-angiogénique et comme protéine anti-tumeur dans le cadre d'une thérapie génique contre le cancer.

v Lors de notre première étude, nous avons constaté les propriétés angiostatiques de la peptide du domaine K5 à même un modèle orthotopique de cancer du cerveau. Nous avons démontré que les ponts disulfides participant à la conformation du K5, nécessaire au maintien de la fonctionnalité de ce dernier, sont conservés lors de la sécrétion provenant de cellules mammaliennes génétiquement modifiées. Les cellules gliomiques humaines U87 génétiquement modifiées de façon rétrovirale produisent de la protéine KS soluble fonctionnelle capable de réduire les facteurs induits de la croissance de cellules endothéliales migrant in vitro et inhibant l'angiogénèse gliomiquement induite in vivo. Il importe de noter que la protéine K5 secrétée bloque le recrutement des macrophages in vivo CD4S+Mac3+Grr associé à la tumeur et inhibe la migration in vitro des macrophages CD206+ dérivé des monocytes. De plus, au sein d'un modèle orthotopique significatif de cellules gliomique, du K5 soluble entraîne la survie à long terme de la majorité des animaux testés. Ainsi, ces résultats valident l'approche préconisée par l'usage de la thérapie génique afin de transférer la protéine Plg K5 et suggèrent que le K5 agit en soi comme agent anti-tumeur à double action qui s'interpose via son effet d'hôte dérivé sur les cellules endothéliales et les macrophages associés à la tumeur.

Afin de déterminer si le KS a interposé son effet anti-tumeur en modulant d'autres cellules immunitaires effectrices, nous avons testé l'usage du K5 soluble dans un DAl3 murin au sein d'un modèle adénocarcinomique mammaire. Le K5 soluble, produit de façon rétrovirale par les cellules DAl3 génétiquement modifiées, mène à la survie à long terme (au-delà d'un an) des souris BALB/c immuno-compétententes. Par contre, nous n'avons pas observé de rejet de la tumeur chez les souris NOD-SCID et chez les souris

BALB/c nues immunodéficitaires. Une analyse a de surcroît révélé que le KS augmente

VI le recrutement de cellules lymphatiques CD3+ qui infiltrent la tumeur et particulièrement le phénotype du Natural Killer T du lymphocyte (NKT). De pair avec nos résultats précédents, nous avons démontré que le K5 mène à la réduction significative de la longueur et de la densité des micro-vaisseaux afférents à la tumeur. Il importe de souligner que les tumeurs K5 se caractérisent par un infiltrat neutrophilique robuste. Ceci qui s'explique par la capacité du K5 à agir comme puissant agent chimiotactique pour les neutrophiles humains in vitro ainsi que par son aptitude à promouvoir l'activation du

CD64+ au sein du phénotype neutrophilique CDllb+. Ces résultats confirment que le K5 agit comme puissant agent angiostatique et qu'il possède en soi un rôle pro-inflammatoire

à travers sa capacité à recruter des neutrophiles associés au tumeurs et des lymphocytes

NKT, menant ainsi à une forte réponse anti-tumeur.

Les macrophages associés à la tumeur (TAM) demeurent des effecteurs immunitaires clés impliqués dans la promotion et la progression des métastases. Il serait donc désirable d'explorer des stratégies visant à réduire l'infiltration de TAM au sein du micro-environnement de la tumeur. Dans notre première étude, nous avons démontré que la protéine K5 soluble bloque le recrutement des macrophages. De plus, de récentes observations révèlent que l'angiostatine réduit l'infiltration de macrophages dans un modèle d'athérosclérose incitant notre laboratoire à davantage explorer l'utilisation de l'angiostatine comme agent anti-macrophage. Nous avons démontré que l'angiostatine réduit la migration in vitro à la fois des macrophages péritonéaux murins et des macrophages CD206+, dérivés des monocytes. De surcroît, nous avons démontré que l'angiostatine mène à la réduction de l'activité gélatinolytique de la matrice

.métalloprotéinase-9 produite par les macrophages, ce qui expliquerait, en partie, la

vu réduction observée de la migration au moyen de l'angiostatine. De plus, nous avons détecté la présence de la sous-unité ~ de la synthase ATP sur la surface cellulaire du macrophage. 11 a été précédemment démontré que la synthase A TP est un récepteur d'angiostatine sur la surface cellulaire des cellules endothéliales. Nous proposons que la présence de la synthase ATP sur la surface des macrophages risquerait de promouvoir l'interaction avec l'angiostatine et éviter la migration, similairement à ce qui a été démontré chez les cellules endothéliales. Nos résultats suggèrent que l'angiostatine démontre un potentiel comme agent inhibiteur face aux macrophages.

Tout compte fait, nos données appuient l'utilisation de la protéine Plg K5 comme puissant agent anti-angiogénique et nos données révèlent que le K5 tout comme l'angiostatine possède en soi des propriétés immunitaires modulatrices qui augmentent potentiellement leur emploi en tant qu'agent thérapeutique anti-cancer.

Vlll ACKNOWLEDGEMENTS

1 would like to express my sincere gratitude to several individuals who have contributed greatly in helping me to achieve this endeavor. First and foremost, 1 would like to thank my supervisor, Dr. Jacques Galipeau for having so graciously accepted me as part of "The Galipeau Lab". Jacques, 1 thank you for your constant support and encouragement, for your smart ideas, for your ability to turn a negative result into a new question, for your optimism, for sharing your passion and for believing in me. Thank you to aIl the members of the Galipeau team, past and present, for creating a chaIlenging and motivating leaming environment. It has been a pleasure to work alongside the foIlowing individuals: Diana (Jaalouk), Nicoletta (Eliopoulos), John (Stagg), Milena (Crosato), Laurence (Lejeune), Abdul-Aziz (AI-Khaldi), François (Mercier), Sébastien (Landry), Hilal (AIsabti), Daniel (Martineau), Nathaniel (Bouganim), Richard (Marcotte), Kapil (Sharma), Jessica (Cuerquis), Moïra (François), Patricia (pelletier), Moutih (Rafei), Claudia (Penafuerte), Shala (Yuan), Sandra (Pommey), Daniel (Legault), lan (Copland), Terry (Kucic) and RaphaeIle (Romieu). Thank you for your humor and positive spirit, for your company especially during the evenings and weekends and for yOUf encouragement. 1 would also like to thank aIl the representatives of the 2001-2004 Experimental Medicine Graduate Students' Society. Thank you to Randy (Levitt) for editing parts of this thesis, for your friendship and your creative projects. Special thanks to David (Martone) for translating the thesis abstract, for your positive energy, encouragement and friendship when 1 needed it most. Many thanks to Liam (Gore) for your encouragement from the very beginning, for motivating me more than you will ever know and for your love. 1 am indebted to my parents, Dominic and Rosetta, for always being there, for believing in me, for encouraging me to strive for the very best, for your support during my failures and successes and most of aIl for your unconditionallove. Very special thanks to my younger sister Romina, for inspiring me to push myself even more, for your determination and ambitious spirit, for your smiles and your love.

IX Success is the abi/ity ta ga tram tai/ure ta tai/ure withaut /asing yaur enthusiasm.

- Sir Winston Churchill PREFACE

The use of recombinant angiostatic proteins for cancer treatment has encountered sorne drawbacks primarily due to short protein half-life in circulation, difficulty in protein production and maintenance of protein functional activity in transport and storage. Therefore, the use of gene therapy techniques to provide sustained high-Ievel anti-angiogenic protein expression at the tumor site would overcome the limitations associated with recombinant protein use. In this regard, the aim of the research performed as part of this thesis was to assess the utility of human plasminogen kringle 5 protein as an anti-tumor agent in brain and breast models within the context of a cancer gene therapy strategy. In addition, given the pivotaI role that tumor-associated macrophages play in promoting tumor growth and progression, we investigated the ability of angiostatin, a weIl characterized anti-angiogenic agent, to affect macrophage migration.

During the course of my doctoral studies, 1 have developed nov el anti-angiogenic gene therapy strategies for the treatment of cancer. 1 have elected to present my research in a manuscript-based format. Three manuscripts are presented in their integrality as distinct chapt ers of this thesis in accordance with the strict guidelines for thesis preparation from the Faculty of Graduate Studies and Research at McGill University

("Guidelines for Thesis Preparation ''):

.:. Chapter 2: Perri SR, Nalbantoglu J, Annabi B, Koty Z, Lejeune L, François M, Di Falco MR, Béliveau R, Galipeau J. Plasminogen Kringle 5-Engineered Glioma Cells Block Migration of Tumor-Associated Macrophages and Suppress Tumor Vascularization and Progression. Cancer Research. 2005 Sept 15;65(18): 8359- 65.

x .:. Chapter 3: Perri SR, Martineau D, Francois M, Lejeune L, Durocher Y, Galipeau J. Plasminogen Kringle 5 Blocks Tumor Progression by Anti-Angiogenic and Pro-Inflammatory Pathways. Molecular Cancer Therapeutics. 2007;6(2): 1-9. (Figure seZected as Cover Art) . •:. Chapter 4: Perri SR, Annabi B, Galipeau J. Angiostatin Blocks Macrophage Migration. 2007. (manuscript submitted to The FASEB Journal).

Xl CONTRIBUTION OF AUTHORS

The data presented in Chapters 2, 3 and 4 of this thesis has been performed by the doctoral candidate under the supervision of Dr. J. Galipeau. Other individuals who have contributed to the work are listed below:

Chapter 2

~ Dr. Josephine Nalbantoglu is affiliated with the Department of Neurology and

Neurosurgery at McGill University. The orthotopic implantation of gene-modified

glioma cells was performed by an experienced animal technician in her

laboratory. Dr. Nalbantoglu provided her expertise and also assumed the costs

associated with the entire implantation experiment.

~ Dr. Borhane Annabi is a Canada Research Chair in Molecular Oncology at

Université du Québec à Montréal. Dr. Annabi taught the doctoral candidate how

to perform the endothelial cell migration assay.

~ Zafiro Koty is an experienced animal technician and works in Dr. Nalbantoglu's

laboratory. She performed the orthotopic implantation of gene-modified human

glioma cells in mice and the hematoxylin and eosin staining ofbrain sections.

~ Laurence Lejeune is a research assistant in Dr. Galipeau's laboratory and

provided her expertise in flow cytometry.

~ Moira François is a research assistant in Dr. Galipeau's laboratory and provided

her expertise in animal-related experiments.

Xll ~ Dr. Marco R. Di Fa/co is a Proteomics Scientist at the McGill University and

Génome Québec Innovation Center. He performed mass spectrometry and

MALDI-QTof on kringle 5 protein.

y Dr. Richard Béliveau is a Professor at Université du Québec à Montréal. He

invited the doctoral candidate to his laboratory to leam how to perform in vitro

angiogenesis endothelial cell assays.

Chapter 3

~ Dr. Daniel Martineau is an Associate Professor at Université de Montréal. He is

a trained veterinary pathologist and analyzed the hematoxylin and eosin sections

of Matrigel explants.

y Moira François is a research assistant in Dr. Galipeau's laboratory and provided

her expertise in animal-related experiments.

y Laurence Lejeune is a research assistant III Dr. Galipeau's laboratory and

provided her expertise in flow cytometry.

~ Dr. Yves Durocher is a Scientist at the Biotechnology Research Institute. A

research assistant under his supervision performed the large-scale transient

transfection in HEK 293 cells and provided our laboratory with concentrated

conditioned media, which was used to purifY kringle 5 protein.

Chapter4

y Dr. Borhane Annabi is a Canada Research Chair in Molecular Oncology at

Université du Québec à Montréal. Dr. Annabi performed the gelatin zymography

assay using concentrated conditioned media from macrophages obtained from our

laboratory.

Xlll TABLE OF CONTENTS

Title page ...... i Abstract...... ii Résumé ...... v Acknowledgements ...... ix Preface ...... x Contribution of Authors ...... xii Table of Contents ...... xiv List of Abbreviations ...... xix LIS· t 0 fF·19ures ...... XXll..

CHAPTER 1: GENERAL INTRODUCTION...... 1 1.1 Basic Principles of Vasculogenesis and Angiogenesis ...... 2 1.1.1 Vasculogenesis ...... 2 1.1.2 Angiogenesis ...... 2 1.1.2.1 Vasodilation and Endothelial Cell Permeability ...... 3 1.1.2.2 Basement Membrane Degradation ...... 3 1.1.2.3 Endothelial Cell Proliferation and Migration ...... 4 1.1.2.4 Endothelial Cell Differentiation and Vessel Stabilization ...... 5 1.1.2.5 Endothelial Cell Survival...... 6 1.1.3 Physiological and Pathological Angiogenesis ...... 7 1.1.4 Tumor Vasculature ...... 7 1.1.5 Vasculogenic Mimicry ...... 8 1.1.6 Angiogenic Switch ...... 9 1.1.7 Angiogenic-Independent Tumors ...... 10 1.1.8 Transgenic Mouse Models of Angiogenesis ...... 11

1.2 Exploiting Anti-Angiogenic Agents Therapeutically...... 12 1.2.1 Tumor Dormancy Induced by Angiogenic Inhibitors ...... 12 1.2.2 Angiogenesis Inhibitors ...... 14 1.2.2.1 Direct Angiogenic Inhibitors ...... 14 1.2.2.2 Indirect Angiogenic Inhibitors ...... 15 1.2.2.3 Endogenous Inhibitors of Angiogenesis ...... 17 1.2.2.4 Discovery of Angiostatin ...... 18

1.3 Plasminogen and its Kringle Domains ...... 19 1.3.1 Proteo1ytic Cleavage of Plasminogen ...... 22 1.3.2 Structural Features of Angiostatin Kringles ...... 26 1.3.3 Anti-Endothelial Cell Properties of Angiostatin ...... 27 1.3.4 Limitations Associated with Recombinant Angiostatin Use ...... 27 1.3.5 Putative Endothelial Cell-Surface Receptors for Angiostatin ...... 28 1.3.5.1 Discovery of Cell-Surface Expressed ATP Synthase ...... 29 1.3.5.2 Angiostatin Binds Endothelial Cell-Surface UvP3- ...... 31 /-~ 1.3.5.3 Angiostatin Binds Endothelial Cel1-Surface Angiomotin ...... 32 1.3.5.4 Angiostatin Possesses Multiple Binding Targets ...... 33

XIV 1.3.6 Plasminogen Kringle 5 Protein ...... 34 1.3.6.1 Anti-Endothelial Cell Potency of Kringle 5 ...... 34 1.3.6.2 Structural Features of Kringle 5 ...... 42 1.3.6.3 Preclinical Testing: Direct Use of Recombinant Kringle 5 Protein ...... 43 1.3.6.4 Putative Endothelial Cell-Surface Receptors for Kringle 5 ...... 44 1.3.6.4.1 Voltage-Dependent Anion Channel...... 44 1.3.6.4.2 Glucose-Regulated Protein 78 Receptor ...... 46 1.3.6.5 Other Kringle-Containing Proteins ...... 48

1.4 Cancer Gene Therapy ...... 49 1.4.1 Anti-Angiogenic Gene Therapy Strategies ...... 52 1.4.2 Cancer Gene Delivery Vehicles ...... 52 , 1.4.2.1 Retroviral Vectors ...... 53 1.4.2.1.1 Inhibition of Tumor Angiogenesis Using Retroviral Vectors ...... 56 1.4.2.2 Lentiviral Vectors ...... 57 1.4.2.3 Adenoviral Vectors ...... 57 1.4.2.4 Adeno-Associated Vectors (AAVs) ...... 59 1.4.2.5 Vaccinia Vectors ...... 60 1.4.2.6 Herpes Simplex Virus Type 1 Vectors ...... 60 1.4.2. 7 Non-Viral Vectors ...... 62 1.4.3 Future of Cancer Gene Therapies ...... 62

1.5 Specifie Research Aims ...... 63

CHAPTER 2: PLASMINOGEN KRINGLE-5 ENGINEERED GLIOMA CELLS BLOCK MIGRATION OF TUMOR-ASSOCIATED MACROPHAGES AND SUPPRESS TUMOR V ASCULARIZATION AND PROGRESSION ...... 65

2.1 Abstract...... 66 2.2 Introduction ...... 67 2.3 Materia1s & Methods ...... 69

2.3.1 Cell Culture Reagents ...... 69 2.3.2 Soluble Human Kring1e 5-HisTag Expression Vector, Protein Purification & Detection ...... 69 2.3.3 Proteomic Analysis ofPurified Soluble hK5His ...... 70 2.3.4 VSV-G Pseudotyped Retroviral Vector Design, Synthesis and Titer Assessment...... 71 2.3.5 Transduction of Human U87MG Glioma Cells ...... 71 2.3.6 Western Blot Analysis ...... 72 2.3.7 Migration Assays ...... 72 2.3.8 In Vivo Matrigel™ Assay and Analysis of Cellular Infiltrate by Cytometry ...... 72 2.3.9 MatrigeITM ExpIant Histochemistry ...... 73 2.3.10 Stereotactic Intracerebral Surgery ...... 74

xv 2.3.11 Brain Tissue Analysis ...... 74

2.4 Results ...... 72

2.4.1 MALDI-QToF Mass Spectrometry Analysis of Expressed Soluble hK5His Peptide Confirms Predicted Disulfide Bridging Conformation ...... 75

2.4.2 hK5His Protein Secreted by Retrovirally Gene-Modified Human Glioma Cells Inhibits In Vitro Endothelial Cell Migration ...... 80

2.4.3 hK5His Blocks Glioma Associated Angiogenesis and Possesses Novel Anti- Macrophage Properties ...... 83

2.4.4Glioma Targeted hK5His Expression Suppresses Tumor Growth and Pro longs Survival in an Experimental Orthotopic Brain Cancer Model...... 91

2.5 Discussion ...... 94 2.6 Financial Support ...... 97

CHAPTER 3: PLASMINOGEN KRINGLE 5 BLOCKS TUMOR PROGRESSION BY ANTI-ANGIOGENIC AND PRO-INFLAMMATORY PATHWAYS ...... 98

3.1 Abstract ...... 100 3.2 Introduction ...... 101 3.3 Materials & Methods ...... 102

3.3.1 Cell Culture Reagents ...... 102 3.3.2VSV-G Pseudotyped Retroviral Vector Design, Synthesis and Titer Assessment ...... 102 3.3.3 Transduction of Murine Mammary DAl3 Cells ...... 103 3.3.4 Western Blot Analysis ...... 103 3.3.5 In Vivo Matrigel™ Assay and Analysis of Cellular Infiltrate by Cytometry ...... " .. 103 3.3.6 MatrigeFM ExpIant Histochemistry ...... 104 3.3.7 Production ofPurified Kringle 5 Protein ...... 105 3.3.8 Neutrophil Isolation ...... 105 3.3.9 Fluorescence-Based PMN Migration Assay ...... 106 3.3.10 Cell Staining ...... 107

3.4 Results ...... 107

3.4.1 DAl3 Mouse Mammary Tumor Cell Line Retrovirally Engineered to Express Plasminogen K5 Domain ...... 107

xvi 3.4.2 Anti-Tumor Property of hK5His Protein is Dependent upon an Intact Immune System ...... 11 0

3.4.3 hK5His Protein Expression by DA/3 Cells Leads to Recruitment of CD3+ Lymphocytes In Vivo ...... 114

3.4.4 hK5His Protein Expression by DA/3 cells Influences the Quantity and Size ofTumor-Associated Microvasculature ...... '" ...... , ...... 114

3.4.5 hK5His Induces Neutrophilic Tumor Infiltration ...... 117

3.4.6 hK5His Protein is Chemotactic for Neutrophils and Promotes Their Activation ...... 122

3.4.7MHC 1 Expression in hK5His-Expressing DA/3 Cells ...... 125

3.5 Discussion ...... 125 3.6 Financial Support ...... 128

CHAPTER 4: ANGIOSTATIN BLOCKS MACROPHAGE MIGRATION...... 130

4.1 Abstract ...... 132 4.2 Introduction ...... 133 4.3 Materials and Methods ...... 134

4.3.1 Macrophage Isolation ...... 134 4.3.2 Macrophage Migration Assay ...... 13 5 4.3.3 Immunoblot Analysis ...... 136 4.3.4 Assay ...... '" ...... 136 4.3.5 Gelatin Zymography ...... 136

4.4 Results ...... 137

4.4.1 Characterization ofIsolated Macrophages ...... 137

4.4.2 Angiostatin Inhibits Macrophage Migration ...... 140

4.4.3 Angiostatin Does Not Induce Caspase Activation in Macrophages ...... 143

4.4.4 Angiostatin Decreases the Gelatinolytic Activity of Macrophage-Derived MMP-9 ...... 146

4.5 Discussion ...... 146

CHAPTER 5: DISCUSSION AND CONCLUSION ...... 152

XVll 5.1 Anti-Angiogenic Agents: From Bench to Bedside ...... 153 5.2 Adverse Effects Associated with Anti-Angiogenic Agents ...... 156 5.3 Avastin: A Clinically Successful Anti-Angiogenic Agent...... 157 5.4 Exploiting Gene Therapy Based on Lessons Leamed from Anti-Angiogenic Clinical Trials ...... 157 5.5 Development of a Novel Anti-Angiogenic Gene Therapy Strategy for Cancer ...... 160 5.5.1 Hypothesis 1 ...... 161 5.5.2 Hypothesis 2 ...... 165 5.5.3 Hypothesis 3 ...... 167 5.6 Summary ofFindings ...... 168 5.7 Future Directions ...... 169 5.8 Concluding Remarks ...... 170

CONTRIBUTION TO ORIGINAL KNOWLEDGE...... 172

REFERENCES ...... 175

APPEND IX ...... 202

Reprint of Published Manuscript Research Compliance Letter

XV111 LIST OF ABBREVIATIONS

bp base pairs g grams kDa kilodaltons Jlg mIcrogram JlI microliter JlM micromolar Jlm mIcron mg milligram ml milliliter mm millimeters mM millimolar ng nanograms U units

AAVs adeno-associated viruses ACN acetonitrile Ang-l -l

BAE bovine arterial endothelial cens BCEs bovine endothelial cens bFGF basic fibroblast growth factor

CAR Coxsackie/adenovirus receptor CEPs circulating endothelial progenitors CHO Chinese hamster ovary

DAB 3,3'-diaminobenzidine DNA deoxyribonucleic acid dsDNA double-stranded DNA

EGFP enhanced green fluorescent protein EPCs endothelial progenitor cens ER endoplasmic reticulum

FAK focal adhesion kinase FBS fetal bovine serum FUCA fluorochrome-labeled inhibitors of caspases FSD free sulfhydryl donor

GFP green fluorescent protein GM-CSF granulocyte-macrophage colony stimulating factor glucose regulated protein 78 /-, GRP78

XIX hK5His human kringle 5 His-tagged H&E hematoxylin and eosin HBSS Hank's balanced salt solution HIF hypoxia-inducible transcription factor HPF high power field HUVEC human umbilical endothelial cell

IGF insulin-like growth factor IL interleukin IFN interferon IRES internaI ribosomal entry site

K kringle Kl-3/4 angiostatin Kl-4.5 angiostatiI14.s

LLC-LM Lewis lung carcinoma-Iow metastatic LTR long terminal repeats

MALDI-QTof matrix assisted laser desorption ionization- quadruple time offlight MCP-1 monocyte chemoattractant protein MDR-1 multi drug resistant-1 METH -1 metallospondin-I MHC major histocompatibility class MMP matrix MoMuLV Moloney murine leukemia virus mrK5 mouse recombinant kringle 5 MTD maximum tolerated dose

NOD-SCID non-obese diabetic severe combined immunodeficient

PAI-l plasminogen activator inhibitor type-l PBMC peripheral blood mononuclear cell PBS phosphate-buffered saline PDGF platelet-derived growth factor PDGF-R platelet-derived growth factor receptor PD-EGF platelet-derived endothelial growth factor PFU plaque-forming units PGK pHe extracellular pH PI propidium iodide PMN polymorphonuclear neutrophil Plg plasminogen PLGF PPT polypurine tract

xx RAG recombinase activating gene RANTES Regulated on Activation, Normal T -cell Expressed and Secreted Rb retinoblastoma RBCs red blood cells RIP rat insulin II gene promoter RGD arginine-glycine-aspartic acid

SDS-PAGE sodium dodecyl sulfate/polyacrylamide gel electrophoresis SEM standard error of the mean SK streptokinase SV40 simian virus 40

Tag SV 40 large T antigen TAM tumor-associated macrophage TFA trifluoroacetic acid TGF transforming growth factor TNF tumor necrosis factor tRNA transfer ribodeoxynuc1eic acid TSP-l -l t-AMCHA trans-4-aminomethy1cyc1ohexane-l-carboxylic acid t-PA tissue-type plasminogen activator u-PA urinary/-type plasminogen activator

VDACI voltage-dependent anion channel VEGF vascular endothelial growth factor VEGFR vascular endothelial growth factor receptor VEGI vascular endothelial growth inhibitor VPF vascular permeability factor VSV-G vesicular stomatitis virus-G protein vWF von Willebrand factor

XXI LIST OF FIGURES

Chapter 1 Figure l.I: Plasminogen and its kringle domains ...... 20 Figure 1.2: Angiostatin fonnation by tumors ...... 23 Figure l.3: Schematic of the amino acid sequence ofhuman plasminogen kringle 5 ...... 35 Figure 1.4: Comparison of the anti-endothelial cell properties of angiostatin and kringle 5 ...... '" ...... 38 Figure l.5: Synergistic inhibition induced by KI-5 ...... 40

Chapter 2 Figure 2.1: Schematic illustration of human Plg and its kringle domains ...... 76 Figure 2.2: Proteomic analysis ofpurified soluble hK5His ...... 78 Figure 2.3: Development and characterization of hK5His-GFP gene-modified cells ...... 81 Figure 2.4: hK5His acts as a potent anti-angiogenic agent...... 84 Figure 2.5: hK5His possesses novel anti-macrophage property and decreases migration potential ofhuman macrophages ...... '" ...... 87 Figure 2.6: U87-GFP and U87-hK5His-GFP cell populations possess equivalent proliferation rates in vitro ...... 89 Figure 2.7: Suppression ofintracerebral U87MG glioma growth by soluble hK5His ..... 92

Chapter 3 Figure 3.1: Development and characterization of hK5His-GFP gene-modified DA/3 cells ...... '" ...... 108 Figure 3.2: hK5His-GFP-implanted immunocompetent mice survive long-tenn and modulate the immune system to suppress tumor growth ...... 111 Figure 3.3: hK5His possesses novel immunostimulatory property ...... 115 Figure 3.4: hK5His acts as a potent anti-angiogenic agent...... 118 Figure 3.5: hK5His protein induces dense pericapillary neutrophilic infiItrate ...... 120 Figure 3.6: hK5His prote in acts as a neutrophil chemoattractant and promotes activation ...... 123

Chapter 4 Figure 4.1: Macrophage characterization ...... 138 Figure 4.2: Angiostatin inhibits macrophage migration ...... 141 Figure 4.3: Angiostatin does not induce macrophage apoptosis ...... 144 Figure 4.4: Angiostatin mediates decrease in MMP-9 gelatinolytic activity ...... 147

XXIl CHAPTER 1

GENERAL INTRODUCTION

1 CHAPTERl:GENERALINTRODUCTION

1.1 BASIC PRINCIPLES OF V ASCULOGENESIS & ANGIOGENESIS

1.1.1 Vasculogenesis

Vasculogenesis refers to the de nova fonnation of blood vessels from undifferentiated endothelial precursor cells called angioblasts that migrate and assemble to fonn a primary capillary plexus. Growth factors, such as vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF), and VEGF receptor 2 induce the primitive capillary plexus to differentiate, fonn new blood vessels and branch from pre-existing , a process known as angiogenesis (1). or matrix receptors which mediate interactions between endothelial cells and matrix macromolecules also affect vasculogenesis. Vasculogenesis leads to the fonnation of immature, poody functional vasculature and it was initially believed to occur only during embryonic development. However, it was recently reported that vasculogenesis also occurs in the adult organism. Circulating endothelial cells (stem cell derivatives) as well as bone marrow-derived endothelial cells were identified in adults and it was described that VEGF, granulocyte-monocyte colony-stimulating factor (GM-CSF), bFGF and insulin-like growth factor (IGF)-I promote these endothelial precursor cells to differentiate and mobilize to sites of interest (2, 3) such as angiogenic tumor microenvironments or to areas necessitating revascularization post trauma (ex: following cardiac ischemia) (1).

1.1.2 Angiogenesis

Angiogenesis is derived from 2 Greek words: angio meaning blood vessel and genesis meaning beginning. Angiogenesis is the fonnation ofblood vessels from pre-existing ones and consists of 2 distinct processes: endothelial cell sprouting and intussusceptive microvascular

2 growth (IMO), which refers to the splitting of vessel lumens (4). Sprouting angiogenesis involves endothelial cell migration, proliferation and tube formation (4). In contrast, IMO separates existing vessel lumens by forminglinserting tissue folds and columns of interstitial tissue into vessellumen (4).

1.1.2.1 Vasodilation & Endothelial Cell Permeability

The angiogenic process is initiated with vasodilation, which involves nitric oxide.

VEOF increases vascular permeability of pre-existing capillaries or post-capillary venules and allows extravasation of plasma proteins, which serve to create a provisional scaffold for migrating activated endothelial cells. Permeability is required for angiogenesis however, excessive vascular leakage is associated with circulatory collapse, intracranial hypertension, formation of adhesion, metastasis, premenstrual discomfort and blindness (1). Endothelial ceIls present within capillaries or postcapillary venules receive pro-angiogenic stimuli from paracrine growth factors such as VEOF and bFOF, which are produced by surrounding tumor or stromal cells as weIl as the extracellular matrix (5).

1.1.2.2 Basement Membrane Degradation

Emigration of endothelial ceIls from their initial resident site involves the loosening of interendothelial ceIl contacts and periendothelial (pericyte) ceIl support, destabilizing mature vessels. Angiopoietin (Ang) 2, an endogenous inhibitor of Tie2 (6) receptor (tyrosine kinase receptor exclusively expressed by endothelial cells) signaling, may be implicated in detaching smooth muscle cells and loosening the matrix (7). Loosening of the pericyte covering is foIlowed by vascular basement membrane and extracellular matrix degradation to allow

3 endothelial cells to migrate toward a chemotactic angiogenic stimuli (8). Plasminogen activator proteinases, matrix metalloproteinases (MMP), chymase or heparanase families affect angiogenesis by degrading matrix molecules and by activating or releasing growth factors sequestered in the extracellular matrix such as bFGF, VEGF and IGF-I (9). For ex ample,

MMP-3, MMP-7 and MMP-9 influence angiogenesis in bones (10) and tumors (11). Tissue inhibitors of MMPs 1 and 3 or PEX, which is the endogenous MMP-2 fragment that prevents binding ofMMP-2 to U vP3, inhibit tumor angiogenesis (12).

1.1.2.3 Endothelial Cell Proliferation & Migration

Following matrix degradation, dividing endothelial cells can migrate to distant sites forming a migration column. Generally, normal quiescent endothelial cells trailing behind the migrating ceUs quickly proliferate and replace ceUs that have been triggered to move toward developing capillary sprouts (5). Sprouting endothelial ceUs form tubes and connect to other blood vessels to aUow for blood flow. The endothelial ceU migration pro cess is influenced by specific growth factors such as VEGF (13), placental growth factor (PLGF), VEGF-B, VEGF­

C, VEGF-D and their respective receptors VEGFR2, VEGFR3 (14) and neuropilin-l (a co­ receptor of VEGFR2) (15). Unlike VEGF, Angl alone do es not induce endothelial network organization. Angl phosphorylates Tie2 and acts as a chemotactic agent for endothelial ceUs.

Angl induces sprouting and enhances the effect of VEGF by stabilizing endothelial ceU networks initiated by VEGF via stimulating endothelial - periendothelial ceU interactions, but do es not induce endothelial ceU proliferation (16, 17). This implies that Angl plays a role in the later stages of angiogenesis. Endothelial cell spreading is also mediated by molecules involved in ceU-ceU or ceU-matrix interactions, such as U vP3 integrin, which is responsible for

4 localizing MMP-2 at the endothe1ial cell surface (18). As a consequence, antagonists of av~3 integrin inhibit angiogenesis.

1.1.2.4 Endothelial Cell Differentiation & Vessel Stabilization

In normal angiogenesis, a migration column leads to a differentiation zone, where endothelial cells alter shape and adhere to one another to create a lumen. Endothelial cell proliferation within the vascular wall induces an increase in the blood vessel diameter. The final step in the maturation of the endothelial network involves remodeling and "pruning" of the capillary-like vessels to a uniform size and conversion of their irregular organization into a structured network of branching vessels (1). This maturation process involves intussusception, a pro cess whereby large vessels split to form smaller vessels. Assembly of endothelial cells into solid cords precedes lumen formation. Endothelial cell thinning, referred to as intercalation, followed by fusing of preexisting vesse1s allows vessels to increase their diameter and length.

VEGF 189 decreases luminal diameter, in contrast to VEGF 121 and VEGF 165 and their associated receptors that increase lumen formation as well as increase vessellength. Angl in combination with VEGF also increases luminal diameter (17), whereas thrombospondin (TSP)-l is an endogenous inhibitor of lumen formation (1). Recruitment of perivascular cells leads to the production of a vascular basal lamina around the newly formed blood vessels. A ligand for the endothelial Tie2 receptor, Angl, is an endogenous inhibitor of vascular permeability and tightens pre-existing vesse1s. Administration of Angl to adult vesse1s prevents plasma leakage without affecting the morphology of vascular structures (19). Activation of AngllTie2

(tyrosine kinase 2) receptor initiates endothelial cells to become quiescent, matures vessels with

5 the completion of the basement membrane and recruits mesenchymal cells which eventually become pericytes (5).

1.1.2.5 Endothelial Cell Survival

In the adult, subsequent to new vessel assembly, the vasculature usually remains quiescent with endothelial cells being among the longest-lived cells outside of the nervous system. Endothelial cell survival is essential since it has been reported that reduced survival results in vascular regression in the embryo (20). Tie2 is involved in vascular maintenance since low levels of phosphorylated Tie2 have been detected in the adult quiescent vasculature.

However, endothelial cell apoptosis is also important and represents a natural mechanism of vessel regression in the ovary following birth and in the retina. Several angiogenic inhibitors utilized for therapeutic purposes also exert their effects by inducing endothelial cell apoptosis.

Lumen obstruction caused by spasms, thrombi, the shedding of dead endothelial cells or a change in angiogenic gene profile as well as nutrient deprivation and lack of survival signaIs induces endothelial cell apoptosis (21-23). For instance, premature babies exposed to hyperoxia possess decreased retinal VEGF levels and vessel regression is induced (1, 24). Ang1 promotes, in contrast to Ang2, which suppresses endothelial cell survival in the absence of angiogenic stimuli, and has been described to be implicated in regression of co-opted tumor

vessels (1, 17). As mentioned previously, antagonists of U v P3 integrin disrupt the interaction of proliferating endothelial cells with matrix macromolecules and lead to endothelial cell apoptosis (1, 18). Other factors inducing endothelial cell apoptosis include: nitric oxide, reactive oxygen species, angiostatin, TSP-1, the metallospondin METH-1, interferon (IFN)-y, tissue factor pathway inhibitor and vascular endothelial growth inhibitor (VEGI) (1).

6 1.1.3 Physiological & Pathological Angiogenesis

Angiogenesis is required for a few adult tissues necessitating regeneration such as the female reproductive organs, organs undergoing growth or injured tissue, as weIl as in chronic inflammatory conditions. Neovascularization is associated with several pathological pro cesses besides tumor growth such as arthritis, endometriosis, diabetic retinopathy and macular degeneration (5). In order to adapt to the local physiological requirements of host-associated vasculature, endothelial cells acquire specific characteristics. For example, the development of the blood-brain barrier requires normal angiotensinogen levels as weIl as an interaction between astroglial cells expressing glial fibrillary acidic protein and pericytes (25).

1.1.4 Tumor Vasculature

Normal physiological angiogenesis IS regulated by a balance of pro- and anti­ angiogenic signaIs and results in the formation of mature and stable blood vessels. This is in stark contrast to the pathological angiogenesis associated with tumors, which is non-quiescent and leads to continuous tumor blood vessel growth. The heterogeneous nature of tumors leads to heterogeneity in tumor vasculature. Tumor-associated vasculature is abnormal in several ways: it is multilayered, irregularly shaped, enlarged, dilated, tortuous and possesses a slow and inefficient blood flow incapable of meeting metabolic demands (26). Furthermore, it contains protruded extensions bridging and splitting of vessels, contains intracellular and transcellular holes, displays uncontrolled permeability and undergoes continuous remodeling (1). Tumor vessels are often leaky and haemorrhagic due to excess production of VEGF, which is also referred to as vascular permeability factor (VPF). Tumor vessel blood flow is irregular, moves slowly and tends to oscillate, leading to dysfunctional capillaries with heterogeneous vascular

7 patterns. In contrast to physiological angiogenesis, tumor vessels do not undergo a maturation processs and thus, remain immature and leaky. This is thought to be due to the expression of

Ang2 (antagonist of Angl), which promotes destabilization of blood vessels and vascular remodeling in the presence of angiogenic stimulators (5). Perivascular cens are mural cens

(cens of the blood vessel wall) related to the vascular smooth muscle lineage that usually remain in close contact with endothelial cells, however these perivascular cells become more loosely associated with tumor-associated endothelium as well as less abundant (27, 28). In normal vessels, the association of pericytes with endothelium leads to a decrease in the proliferation of endothelial cells as well as reduces the dependence on host-tissue VEGF-A production (29). The molecular mechanisms involved in tumor blood vessel abnormalities are under much investigation, however it is widely accepted that an imbalance between the expression of positive and negative regulators of angiogenesis results in drastic vascular differences.

1.1.5 Vasculogenic Mimicry

In light of these distinguishing features, tumor blood vessels are referred to as "mosaic" in the literature, meaning that the tumor vessels are lined with both endothelial cells and malignant "vasculogenic" tumor cells, commonly referred to as "vasculogenic mimicry" (l,

30-32). This concept was tirst observed in melanoma and describes the ability of aggressive melanoma cells to express endothelium-associated and form extracellular matrix-rich vasculogenic-like networks in three-dimensional culture conditions similar to an endothelial cell feature (31,33, 34). In fact, patients with a melanoma that has undergone vasculogenic mimicry have a poor prognosis (35). Vasculogenic tumor cells have been reported for other

8 malignancies as well. Given this salient feature of sorne tumors, epitopes specific for tumor

endothelial cells have been targeted for anti-cancer therapeutic purposes. For instance, Ellerby

et al. (36) exploited this characteristic by designing short peptides composed of 2 functional

domains: a tumor vessel "homing" motif and a programmed cell death inducing sequence. The

synthetic peptides demonstrated selective toxicity for angiogenic endothelial cells and resulted

in an anti-cancer response in mice.

1.1.6 Angiogenic Switch

Angiogenesis associated with growing tumors has been described as undergoing 2 phases separated by the "angiogenic switch" (37). The first phase is the avascular phase

corresponding to small and occult lesions with a diameter of no more than 1-2 mm. Lesions of this size remain dormant by maintaining a steady state between proliferation and apoptosis. In

fact, it has been reported by Black et al., that dormant tumors have been discovered during

autopsies ofindividuals who died of non-cancer related causes (38). These findings support the view that only a small fraction of dormant tumors enter the second phase, referred to as the

vascular phase whereby the tumor grows exponentially. Similarly, dormant metastases may be

activated by the removal of the primary tumor mass (39). The angiogenic switch can be tumed

"on" by oncogene-driven tumor expression of pro-angiogenic factors (40) such as VEGF, bFGF, interleukin (IL)-8, PLGF, TGF-p, platelet-derived endothelial growth factor (PD-EGF), pleiotrophin and several other angiogenic proteins (41).

Tumor angiogenesis can also be modulated by hypoxia. Low oxygen tumor conditions activate hypoxia-inducible transcription factors (HIF-1a, HIF, 1p and HIF-2a), which induce upregulation of several angiogenic factors such as VEGF, VEGFR1, VEGFR2, neuropilin-1,

9 Ang-2, nitric oxide synthase, TGF~-l, PDGF-BB, endothelin-l, IL-8, IGF-II, Tiel, cyclooxygenase-2 and others (1, 42). During nonnoxia (nonnal oxygen conditions), the expression of hypoxia-inducible target genes is suppressed by the von Hippel Lindau tumor suppressor gene product (1). It has been shown that tumors that lack HIF-1a or HIF-1~ fail to develop a vascular system and hypoxia does not lead to VEGF expression (43). Pro-angiogenic factors can also be released by fibroblasts within the tumor microenvironment (44) and by progenitor endothelial cells that are recruited from the bone marrow by the tumor (45).

1.1. 7 Angiogenic-Independent Tumors

Although angiogenesis is a prerequisite for the majority of tumors to progress, sorne tumors such as astrocytomas are called "non-angiogenic" tumors as they do not require neovascularization. Instead, they co-opt with existing nonnal brain blood vessels and acquire a blood supply without initiating angiogenesis. Astrocytomas are very oxygen-sensitive and exploit the vas cul ar-ri ch brain parenchyma by growing along blood vessels, without possessing a tumor capsule, invade and can enlarge to a size similar to that of an angiogenic tumor (46).

The progression of grade III astrocytomas to grade IV astrocytomas/glioblastomas

(glioblastoma multifonne), causes them to become hypoxic and necrotic, partially due to the regression of vessels, as a result of Ang2 upregulation (47) and increased tumor cell proliferation. Hypoxic conditions induce the fonnation of new blood vessels and the most common hypoxia-induced angiogenic factors are VEGF and VEGFR-2, which along with

Ang2, promote vascular remodeling and sprouting (46, 48). A distinguishing feature of glioblastomas is the appearance of proliferating endothelial cells and a high blood vessel density, which differentiates them from lower-grade astrocytomas (49).

10 1.1.8 Transgenic Mouse Models ofAngiogenesis

The development of oncogene-driven mouse tumor models has facilitated the in vivo characterization of tumor angiogenesis. The RIPTag model, which employs the rat insulin II gene promoter (RIP) to target expression of the simian virus (SV)-40 large T tumor antigen (T­

Ag) oncoprotein to cells of pancreatic islets, results in the formation of pancreatic-islet carcinomas (50). Multistage progression to invasive squamous cell carcinoma of the epidermis is reproduced in transgenic mice expressing the human papillomavirus type (HPV)-16 under the control of the human keratin 14 promoter/enhancer (E6 and E7 oncogenes) (50). These animal models require that neovascularization begin early in the dysplastic phase in order for tumor formation to occur. The SV 40 Tag, E6 and E7 oncogenes bind and inactivate p53 and the retinoblastoma (RB) tumor suppressor gene products however, this is not sufficient to induce neovascularization. In order to initiate neovascularization, the angiogenic switch must be turned on and it has been described in the pancreatic-islet cancer model that VEGF is required for the switch to occur (51). Matrix metalloproteinase (MMP)-9 has been reported to release sequestered VEGF and allows VEGF to interact with its receptors (52). Higher levels of

MMP-9 production also resulted in more vascularized tumors in mammary-tumor-prone mi ce

(MMTV-neu) that lack TSP-l (angiogenic inhibitor), since higher MMP-9 production causes higher levels ofVEGF available for interaction VEGFR-2. In contrast, overexpression of TSP-

1 in MMTV-neu mice leads to delayed tumor growth and can even result in up to 20% inhibition of tumor development (53). It has been demonstrated that TSP-l expression is inversely correlated with mammary, lung and melanoma tumor progression (54). Watnick et al.

(55) demonstrated that the ability to grow primary epithelial cells in vivo depended upon the level of HRAS expression. Epithelial cells that expressed low levels of RAS remained dormant

11 and non-angiogenic, whereas those that expressed high levels of RAS developed tumors and

expressed moderate levels ofVEGF-A as weIl as high levels ofTSP-l. Overexpressing VEGF-

A in those cells that expressed low levels of RAS induced tumor formation. The sum of these

findings supports the notion that angiogenesis is a prerequisite for tumor formation, that a

balance of pro-angiogenic and anti-angiogenic factors influences the angiogenic switch and that

oncogene expression can affect the balance (8). However, there is also evidence that indicates

that this multi-step model, whereby the angiogenic switch is necessary for tumor development,

yet is not always a requirement for the growth of angiogenic tumors. It has been reported that

expression of deregulated MYe, BeL2 or BeL-XI led to the formation of angiogenic and

invasive pancreatic-islet tumors that did not require additional mutations to proceed to the

premalignant stages (56). There have been however no published reports of human tumors

bypassing the premalignant stages.

1. 2 EXPLOITING ANTI-ANGIOGENIC AGENTS THERAPEUTICALLY

1.2.1 Tumor Dormancy lnduced by Angiogenic lnhibitors

An extension of the angiogenic switch model is that the switch may be tumed "on" at p 1 the site of a primary tumor and "off" at sites of distant metastases. This has been observed by

Folkman and others who have noted an increase in the growth of metastatic tumors following

the removal of a large primary tumor. In experimental models, it was described that metastatic

tumor resurgence was not due to seeding of tumor cells at the time of primary tumor resection,

but due to the fact that the pre-existing micrometastases were being held "in check" by the

presence of the primary tumor and are induced to grow after removal of the primary tumor (47,

57, 58). There is overproduction of pro-angiogenic molecules within the primary tumor as weIl

12 as production of inhibitors of angiogenesis, albeit at much lower levels than the angiogenic stimulators and thus angiogenesis is sustained and tumor growth ensues. The angiostatic molecules produced at the primary tumor site enter the bloodstream and possess a longer half­ life in circulation and therefore angiogenesis at distant sites is tumed off, preventing the growth of micrometastases. Primary tumor resection results in an overall decrease in the systemic levels of angiostatic agents and thus pro-angiogenic factors secreted by metastatic tumors become dominant and the shift in angiogenic balance favors metastatic growth. O'Reilly et al.

(59) utilized a variant of the Lewis lung carcinoma cell line (LLe-LM) possessing low metastatic potential and subcutaneously implanted the cells in mice. Primary tumors were resected 14 days post-implantation and mice were compared to those whose primary tumors

were left intact after a mock surgery (60). The presence of an intact primary tumor resulted in a

10-fold suppression oflung metastases in contrast to mice with resected primary tumors (60).

Metastatic lesions in mice with intact primary tumors formed perivascular cuffs devoid of neovascularization (60). Thus, this demonstrates that the primary tumor maintains metastatic lesions dormant, possibly by secreting anti-angiogenic proteins that inhibit angiogenesis.

In light of the importance of angiogenesis for tumor growth and progression of dormant lesions reviewed above, it would be highly desirable to exploit the utility of angiogenic

inhibitors to prevent the angiogenic switch from being tumed on, resulting in inhibition of

tumor progression and metastasis. This strategy could be used therapeutically especially for

those patients who are at high risk for developing cancer, recurrence of cancer or metastases.

Anti-angiogenesis therapy targets cells that support tumor growth instead oftargeting the tumor itself like conventional therapeutic modalities such as chemotherapy and radiation therapy, and therefore presents a new treatment option for patients with refractory malignancies. Vascular

13 endothelial cells are genetically stable nonnal diploid cells, homogeneous and have a low mutation rate, therefore they are less likely to develop mutations as compared to unstable mutating heterogenous tumor cells. This in turn minimizes the risk for the development of drug resistance towards the anti-angiogenesis therapy.

1.2.2 Angiogenesis Inhibitors

A variety of angiogenesis inhibitors have been discovered and are classified in two distinct groups: direct and indirect inhibitors.

1.2.2.1 Direct Angiogenic Inhibitors

Direct angiogenesis inhibitors target microvascular endothelial cells by preventing them from responding to endothelial mitogens, and thus inhibit endothelial cell proliferation and migration as well as overcome the avoidance of cell death in response to an array of pro­ angiogenic stimuli (61). Angiostatin, an endogenous inhibitor of angiogenesis, is an internaI proteolytic fragment of plasminogen (a circulating glycoprotein acting as the inactive precursor of ) and has been shown to bind to ATP synthase, angiomotin and II on endothelial cells, preventing endothelial cell proliferation and migration (62-64). Another endogenous , , is an internaI fragment C-tenninal collagen type

XVIII. Endostatin has been reported to target integrin-usPI to inhibit endothelial cell proliferation and migration, and induces apoptosis of dividing endothelial cells (65, 66).

Canstatin, an endogenous human basement membrane-derived fragment of the alpha2 chain of

collagen type IV, has been described to bind integrin U v P3 to inhibit endothelial cell proliferation, migration and capillary tube fonnation (67). Avastin () is a

14 recombinant humanized monoclonal directed against VEGF, developed by

Genentech. It has shown great promise in treating colorectal, lung and metastatic breast cancer and was approved in 2004 as first-line treatment for patients with advanced colorectal cancer

(68). In June 2006, Avastin was approved as a treatment for second-line advanced colorectal cancer (patients previously treated for their cancer). Vitaxin is a humanized monoclonal antibody developed by Medlmmune, which targets ŒvP3 integrin and is another direct inhibitor of angiogenesis (69).

Direct inhibitors of angiogenesis have a lower risk of inducing acquired drug resistance since they target stable endothelial cells and Boehm et al., report that Lewis lung carcinoma, fibrosarcoma as well as melanoma tumors treated with a direct anti-angiogenic agent

(endostatin) did not develop drug resistance in mice (70). O'Reilly et al., describe the successful use of another direct angiogenic inhibitor, angiostatin, in the reduction of Lewis lung carcinoma metastases formation (63).

1.2.2.2 Indirect Angiogenic Inhibitors

Indirect inhibitors of angiogenesis target tumor cells by preventing the expression or blocking the activity of a tumor protein (oncogene product that drives the angiogenic switch) that activates angiogenesis, or by blocking the expression of its receptor on endothelial cells

(39, 40, 71). Unlike direct inhibitors of angiogenesis, indirect inhibitors are prone to develop resistance. Studies on the activity of oncogene and tumor suppressor gene products were initially conducted in vitro and assessed cancer cell proliferation, resistance to apoptosis, immortalization and anchorage independence (40, 71). Oncogene activation demonstrated increased cancer-cell proliferation and decreased apoptosis in vitro and this correlated very

15 weIl with in vivo tumor growth. Activating mutations in oncogenes as weIl as in the anti­ apoptotic factor BCL-2 have been demonstrated to upregulate angiogenic pro teins and to downregulate anti-angiogenic proteins. For instance, the HER-2/neu blocking antibody

Herceptin (trastuzumab), which inhibits the tyrosine kinase activity of the HER-2/neu receptor also inhibited the angiogenic potential of a tumor by decreasing the expression of VEGF, Ang-

1, TGF-/3, plasminogen activator inhibitor type-1 (PAI-1) and by upregulating the expression of

TSP-1 (39). Likewise, Arbiser et al. (72) demonstrated that when normal cells in mice were immortalized with SV40 tumor antigen, the production of microscopic avascular dormant tumors was observed. However, when the immortalized cells were transfected with the Ras oncogene, tumor angiogenesis and growth ensued (73). Other studies supporting this concept are as follows: activating mutations in Kras and Hras have been shown to upregulate VEGF expression and downregulate TSP-1 (angiogenic inhibitor) expression (74, 75) and transfection of the Bcl2 oncogene into tumor cells resulted in enhanced VEGF expression (76).

Furthermore, transfection of human osteosarcoma cells with the Ras oncogene and subsequent injection in mice led to increased VEGF expression, decreased thrombospondin expression and resulted in large neovascularized tumors as compared to untransfected control mice that only formed microscopic avascular dormant tumors that possessed a balanced apoptotic rate (75). In a similar fashion, re-activation of tumor suppressors such as p53 can also lead to the inhibition of angiogenesis (77, 78). These findings imply that anti-cancer drugs that were initially developed to block an oncogene product could also be utilized to target angiogenesis indirectly.

For example, EGF receptor tyrosine kinase inhibitors such as Iressa (ZD1839), Tarceva

(OSI1774) and Erbitux (IMC225) block the production of EGF receptor products including

VEGF, bFGF and TGF-a (79). Other examples of indirect angiogenesis inhibitors target the

16 VEGF receptor on endothelial cells and these include: PTK787/ZK222584 (80), synthesized and developed by Novartis Pharma AG and Schering AG, which exhibits strong inhibition of

VEGFR-2, slightly weaker inhibition of VEGFR-l and also active against other tyrosine kinases specifically belonging to the same family as the VEGF receptors such as the PDGF receptor tyrosine kinase; SU6668 (81) and SU11248 (80), developed by SUGEN (acquired by

Pfizer) are capable of inhibiting the tyrosine kinase activity of the VEGF, FGF and PDGF receptors (VEGFR-1I2, PDGFR, KIT, FLT3); and ZD6474 (80), developed by AstraZeneca, blocks VEGF and EGF receptor tyrosine kinase activity. IFN-a is an exarnple ofboth a direct and indirect angiogenesis inhibitor since it inhibits endothelial cell migration (82) and suppresses bFGF synthesis at the mRNA and protein level by tumor cells (83). It is interesting to note that sorne tumors such as giant-cell tumors of the bone and angioblastornas produce mainly bFGF and thus when patients were treated with low daily doses of IFN-a, drug resistance was not observed during the course of 1-3.5 years oftherapy (84,85).

1.2.2.3 Endogenous Inhibitors ofAngiogenesis

The majority of anti-angiogenic agents exist endogenously in the body and have either been isolated frorn the body or tissue-derived cell lines. Among the reported endogenous angiogenesis inhibitors, at least 50% are proteolytic fragments oflarger precursor proteins (61).

Endogenous angiostatic agents possess the added benefit of being non-toxic and non­ immunogenic (61). To date, there have been no reports describing toxicities or adverse effects for endogenous anti-angiogenic agents. Therapeutically, the use of endogenous inhibitors is less likely to cause suppression of hernatopoiesis and gastrointestinal symptoms (86). In fact, naturally-occurring angiostatic proteins tend to specifically target newly formed tumor

17 vasculature rather than quiescent nonnal blood vessels. There are several characteristics that make endogenous angiogenic inhibitors of particular interest. First, they are typically proteolytic fragments of larger precursor proteins that are biologically inactive at suppressing angiogenesis. However, these cryptic fragments are processed by nonnal host tissues or tumors and exert inhibitory angiogenic effects. Second, the majority of endogenous angiostatic agents can directly inhibit endothelial cell proliferation even in the presence of angiogenic stimuli.

Third, they generally act synergistically with other anti-cancer therapeutic modalities such as chemotherapy and radiation therapy. Finally, since endogenous angiogenic inhibitors target genetically-stable endothelial ceIls, the risk for acquired drug resistance is minimized, although it can occur (87).

1.2.2.4 Discovery ofAngiostatin

A better understanding of the angiogenic switch led to the possibility that exogenous administration of angiogenic inhibitors following removal of the primary tumor could suppress the growth of metastases. In order to identify the angiogenic inhibitor responsible for inducing donnancy in metastatic lesions, ü'Reilly and colleagues (59) collected over 100 L of urine from LLC-LM tumor-bearing mice and purified a protein that was capable of inhibiting angiogenesis both in vitro and in vivo. It was also observed that this protein could prevent the neovascularization and growth ofmetastatic LLC-LM tumors, even in those mice with resected primary tumor. The 38kDa protein possessed an amino tenninus composed of VYLSEXKT, which corresponded to an internaI fragment of plasminogen and its molecular weight indicated that it encompassed the first 4 of the 5 plasminogen (PIg) kringles (60). This novel angiostatic

18 protein was named "angiostatin." This discovery prompted further investigation to elucidate the mechanism of angiostatin generation.

1.3 PLASMINOGEN & ITS KRINGLE DOMAINS

Ruman Plg is a circulating glycoprotein that acts as the inactive precursor of plasmin, which is an active that is responsible for dissolving fibrin in blood clots (Figure

1.1). Plasmin is also implicated in embryonic development, tissue remodeling, inflammation and tumor invasion. Plg is produced and secreted by the kidneys as a zymogen and is found in plasma and extracellular fluid (61). Full-Iength Plg (glu-PIg) is a tightly compacted structure that is only activated by Plg activators upon epsilon-amino caprioic acid (EACA) binding, which induces the structure to become less compact and more susceptible to activation (88). It is activated by urinary/urokinase-type or tissue-type plasminogen activator (u-PA and t-PA respectively), which cleave Plg to form plasmin (88). Plg is composed of 5 internaI kringle domains and serine protease domain at its C-terminus (88). A kringle (kringla in Swedish) is a type of Scandinavian cookie folded into 2 rings (61). The term was coined to describe a triple looped structure linked by 3 pairs of disulfide bonds and was originally used to describe the triple disulfide bonds in prothrombin. The kringles are numbered starting from the amino terminus. Several proteins contain either 1 or more kringle structures. In addition to the triple disulfide linkages, kringle structures possess a sequence of approximately 80 amino acids.

Typically, amino acids flanking the third and fourth of the 6 conserved cysteine residues that form the disulfide bridges are conserved as well, in contrast to other amino acids within the kringle structure that are not similar among different

19 Figure 1.1. Plasminogen and its kringle domains. The full-length plasminogen amino acid sequence is depicted. Disulfide bonds are displayed as black bars connecting pairs of cysteine residues. Glycosylation sites are displayed as red wavy lines in the third kringle domain. The plasmin catalytic domain is shown in gray, kringle 5 in purple, kringle 4 in red and kringles 1-3 in orange. This schematic was developed by Wahl et al. (90). r FIGURE 1.1 kringle domains. Interactions between K5 and the N-terminal ''jinger domain" of plasminogen influence the compact conformation of plasminogen (88). Constructs devoid of the finger domain exist only in an extended conformation and are not activated by EACA (88). Lysine­ binding sites in KI and K4 also affect maintenance of the closed conformation (88).

1.3.1 Proteolytic Cleavage ofPlasminogen

Plg possesses various cleavage sites, which leads to the production of several smaller cryptic fragments that can be produced in a context-dependent manner. Hogg and co-workers

(89) proposed a model to elucidate the proteolytic processing steps that occur in the generation of angiostatin and K5 (Figure 1.2). Plasmin reduction is the first step in the formation of angiostatin fragments KI-3, KI-4 and KI-4.5. A plasmin reductase reduces the disulfide bonds in K5 in the presence of certain cofactors such as reduced glutathione or cysteine. A serine protease then cleaves peptide bonds in K5 to yield KI-4.5, which then undergoes further cleavage by MMPs to pro duce KI-3 or KI-4. These MMPs may be activated by pro-angiogenic signaling pathways such as induction of HIF-l. Phosphoglycerate kinase (PGK), a glycolytic outer mitochondrial membrane enzyme that is usually associated with the voltage-dependent anion channel (VDACI) (90), has been reported to be the plasmin reductase involved in the reduction of K5 and has been observed to be present in the plasma at high levels in tumor­ bearing animaIs (91). Both Stathakis et al. (89, 92) and Soff et al. (93) described that the carboxy-terminus of cleaved angiostatin was found within the fifth Plg kringle yielding a 52-

55kDa protein and thus designated the name "angiostatin4.5" (KI-4.5) to refer to KI-4 plus approximately 85% ofK5. KI-4.5 was found to be produced via plasmin autoproteolysis,

22 Figure 1.2. Angiostatin formation by tumors. Plasminogen is converted to plasmin by the u-PA enzyme followed by plasmin K5 reduction by phosphoglycerate kinase (PGK).

A serine proteinase and matrix metalloproteinases (MMPs) participate in the formation of

angiostatin. This schematic was obtained from Geiger et al. (88).

------FIGURE 1.2

Plasminogen

).lPA Serine Proteinase A~;~~atin PGK Reduced • Kringle Plasmin --..... Plasmin K5 1-41/2 MMPs< K1·3 mediated by a free sulfhydryl donor (FSD), or by a plasmin reductase. This form of angiostatin has been found in ascites and in human plasma (93). It has also been reported that human prostate cancer cells convert Plg to angiostatin via the following mechanism: u-PA or t-PA converts Plg to plasmin and in the presence of a FSD, plasmin undergoes autoproteolysis to generate angiostatin, where plasmin acts as both the enzyme and the substrate (60, 94, 95). The plasmin receptor was shown to be expressed on the cell-surface of activated murine macrophages and it induces plasmin autoproteolysis, which generates angiostatin-like fragments (96).

There are other molecules that can induce cleavage of Plg to angiostatin such as in Lewis lung carcinoma (86) and in human prostate carcinoma (95) as well as prostate specifie antigen (97), which has been reported to also act as an inhibitor of angiogenesis (98). In addition, human matrilysin (MMP-7), gelatinase B/type IV (MMP-9) and a macrophage­ derived metalloproteinase (99, 100) have also been implicated in the generation of angiostatin.

K5 however, is not produced via this pathway. Instead, Plg is cleaved by neutrophil elastase to generate angiostatin KI-3, free K4 and what is referred to as "miniplasminogen ", comprising the serine protease domain plus K5 (90). MMPs are then responsible for releasing

K5. Although KI-3 has been found in the circulation of tumor-bearing animaIs as a consequence of activation of pro-angiogenic signaling cascades (90), the in vivo generation of

K5 is still unknown (90).

It has been found that tumor cells do not produce angiostatin or PIg, but rather secrete proteases that cleave circulating Plg into angiostatin. These tumor-derived proteases cleave Plg at several sites and thus angiostatin is composed of a different number of kringles that independently possess varying degrees of endothelial cell inhibitory properties (61). For

25 example, MMP-2, -3, -7, -9, -12, as weIl as elastase plus pepsm or chymotrypsin or

Staphylococcus Aureas V8 protease or streptokinase and plasmin autodigestion have aIl been

reported to release KI, K2-K3, angiostatin (KI-3 or KI-4), or K5 from the parent Plg molecule

(61). These findings highlight the implications ofPlgiangiostatin cleavage and associated post-

translational modifications as weIl as the importance of proteolysis in angiogenesis

suppreSSIOn.

1.3.2 Structural Features ofAngiostatin Kringles

KI, K2, K3 and K4 possess high amino acid sequence similarity (about 50% homology)

(61). Nuclear magnetic resonance and X-ray crystallography ofKI-4 demonstrate that the high

sequence homology translates into a similar conformation (61). It was reported that intact Plg

do es not possess any angiostatic properties and is incapable of affecting endothelial cell

growth, angiogenesis or tumor growth. However, the cryptic kringle fragments vary

considerably in their ability to inhibit endothelial cell growth. Of the first 4 kringles, KI

displays the highest potency at impairing endothelial cell growth, followed by K3, K2 and K4

(61). In fact, K4 is almost incapable at suppressing endothelial cell proliferation and

demonstrates high affinity for lysine binding (61). However, since KI has also been shown to

bind lysine with high affinity (101), it suggests that lysine binding does not play a role in

manifesting an endothelial cell inhibitory effect (61). Further investigation into the 3-

dimensional structure of K4 revealed that it contains 2 clusters of positively charged lysine

residues adjacent to 2 cysteines, which interact with lysine 59 to create an exposed positively

charged area in K4 (61). This cationic cluster is not present in other kringle structures and it

may be speculated that the lysine-enriched K4 domain interacts with a negatively charged

)

26 domain in another protein, resulting in a sequestering of K4 to induce a loss-of-function effect

(61).

1.3.3 Anti-Endothelial Cell Properties ofAngiostatin

Angiostatin has been characterized as an endothelial cell inhibitor via its ability to suppress endothelial cell proliferation, endothelial cell capillary tube formation on a Matrigel assay as weIl as FGF- and VEGF-induced endothelial cell migration in a modified Boyden chamber through a gelatinized membrane (59, 94, 95). Moreover, KI-3, KI-4 and KI-4.5 have been shown to induce apoptosis of endothelial cells in vitro and this has been associated with activation of focal adhesion kinase (102). Kl-4.5 was also shown to inhibit extracellular­ matrix-enhanced, t-PA-catalyzed Plg activation, which prevents plasmin generation and may explain the mechanism employed by angiostatin to suppress endothelial cell migration (103).

Angiostatin possesses a longer half-life than angiogenic factors such as VEGF and FGF and as a result, primary tumors that produce angiostatin can inhibit their metastases but fail to inhibit their own growth (61). Angiostatin has been shown to induce dormancy in metastatic and primary tumors such as breast, prostate, colon and lung carcinoma (59, 104) as weIl as reduce the growth of a murine hemangioendothelioma in vivo (105). The anti-tumor effect of angiostatin has been attributed to its anti-angiogenic property.

1.3.4 Limitations Associated with Recombinant Angiostatin Use

Angiostatin requires intact kringle structures to maintain its anti-endothelial cell effect and it has been demonstrated that disruption of intrachain and interchain disulfide bonds by reducing agents entirely abrogates its inhibitory property. Unfolded or misfolded angiostatin or

27 kringles also causes a loss-of-function effect. For instance, bacterially-expressed angiostatin must be properly folded to inhibit endothelial cell growth and angiogenesis.

The shorter version of angiostatin containing only the first 3 kringles (Kl-3) has been demonstrated to be more potent than the Kl-4 proteolytic fragment (86). Thus, Kl-3 was selected for anti-cancer therapeutic purposes and is currently undergoing Phase II clinical trials.

One of the limitations associated with Kl-3 due to its shorter amino acid sequence is that it possesses a shorter half-life in vivo and cleared within minutes, thus continuous infusion is necessary for effective anti-cancer therapy. This was observed in a Phase 1 clinical trial, whereby daily administration of angiostatin for 2 week intervals with a l-week break period to monitor toxicity, caused VEGF and bFGF to reSurge in the circulation as early as 1 day into the drug-free week (106). Therefore, these results suggested that anti-angiogenic agents would require continuous infusion to suppress tumor growth and those agents possessing longer half­ lives would be more efficacious for clinical use. Other shortcomings associated with angiostatin use include protein production difficulties and maintenance of functional activity during storage and transport (90). These issues have spurred interest in attempting to identify angiostatin receptors involved in mediating its anti-angiogenic effects to develop angiostatin mimetics, which are more stable, easier to pro duce and possess enhanced pharmacokinetic properties (90).

1.3.5 Putative Endothelial Cell-Surface Receptorsfor Angiostatin

Although angiostatin is derived from PIg, Plg itself and plasmin do not display any angiostatic properties, implying that angiostatin interacts with a different binding site. To date,

28 there are 3 main cell-surface receptor binding targets for angiostatin on endothelial ceIls: the mitochondrial F IFo ATP synthase, ŒvP3-integrin and angiomotin.

1.3.5.1 Discovery of Cell-Surface Expressed ATP Synthase

Moser et al. (62) sought to determine the endothelial cell binding sites for angiostatin in order to identify a putative receptor. Cell binding assays using radiolabeled angiostatin revealed that Kl-3 bound to the surface of human umbilical endothe1ial cells (HUVECs) at a different site and with a different binding kinetic than Plg (62). Scatchard plot analysis revealed a ~ of

245 nM and 38 000 binding sites per cell in contrast to Plg with a ~ of 158 nM and 870 000 sites per cell (62). In addition, it was shown that Plg did not compete with angiostatin for

HUVEC binding, confirming different binding sites (62).

In order to identify the endothelial cell binding site for angiostatin, Moser et al. prepared an affinity column of angiostatin coupled to Sepharose as weIl as a Pig-coupled column as a control. HUVECs were cell-surface biotin-Iabe1ed and plasma membrane extracts were subsequently passed over each column and washed to remove unbound material. Bound proteins were eluted from each column and subjected to denaturing gel electrophoresis followed by immunoblot analysis. Specifie bands were se1ected to undergo further proteomic analysis. The Plg affinity column yie1ded a 44 kDa protein band, which was confirmed to be annexin II by immunoassays. In contrast, the angiostatin affinity column yielded several protein bands ranging from 20-65 kDa, with a dominant band at 55 kDa. Gel extraction, tryptic digestion and mass fingerprinting revealed that the 55 kDa band was composed of the Œ- and p­ subunits of the FIFo ATP synthase.

29 This finding was quite controversial since FJFo ATP synthase was believed to be strictly a component of the mitochondrial inner membrane and this discovery implied that a form of

F jFO ATP synthase was also present on the cell-surface of HUVECs. At the time, only 1 other report by Das et al. (107) described the detection ATP synthase on the cell-surface of A549 human lung carcinoma cells and since this finding was unexpected, it was attributed to aberrant trafficking in an unstable tumor cell line or a cell culture artifact. Although the concept of ectopic localization of mitochondrial pro teins was initially met with much skepticism, other research groups also reported identifying plasma membrane proteins that were in fact mitochondrial matrix proteins, but that possessed extramitochondrial functions (108, 109).

Additional reports confirmed angiostatin binding to cell-surface expressed FJ-Fo ATP synthase on sorne tumor cells such as A549 human lung carcinoma cells and K562 human erythroleukemia cells (107, 110), as well as on the cell-surface of other normal cell types such as keratinocytes and adipocytes (111, 112). It has also been recentlY demonstrated that the

V gamma9Vdelta2 T cell receptor on gammadelta T cells interacts with cell-surface expressed

Fj-ATP synthase in association with apolipoprotein A-Ion tumor cells, promoting an innate anti-tumor response (113). Further investigation revealed that several of the enzymes and components of the mitochondrial electron transport chain as well as ATP synthesis generating mechanisms are located on the plasma membrane of endothelial cells.

It was hypothesized that endothelial cell-surface ATP synthase is responsible for maintaining intracellular pH in the acidic tumor microenvironment. Binding of angiostatin to endothelial cell-surface ATP synthase inhibits the proton pump and causes the intracellular pH to decrease leading to endothelial cell apoptosis (114, 115).

30 Magnetic resonance imaging (MRI) measurements of the in vivo pH of the tumor extracellular environment reveals that it is between pH 5.6 to 7.6 on average, whereas the extracellular pH of normal tissue is between 7.2 to 7.6 (116). These MRI measurements reflect the average pH value of various cell types including stromal and endothelial cells. Tumor cells however, are able to maintain a normal intracellu1ar pH (117). It was hypothesized that endothelial cell-surface ATP synthase is responsible for maintaining intracellular pH in the acidic tumor microenvironment. Although angiostatin can bind and inhibit ATP synthase over a pH range of 6.5 to 7.5, it has been reported to be more potent at low extracellular pH, and therefore the tumor microenvironment is ideal to maximize its anti-tumor effect (116). Binding of angiostatin to endothelial cell-surface ATP synthase inhibits the proton pump and causes the intracellular pH to decrease, which leads to the inhibition of metabolic enzymes that require a pH of 7.0 and results in endothelial cell apoptosis (114, 115). In addition, it was reported that angiostatin does not inhibit proliferation of endothelial cells cultured at a normal pH (116).

This implies that angiostatin's target is implicated in pH maintenance and suggests that angiostatin-induced inhibition of the proton flux via A TP synthase causes endothelial cells to become more sensitive to pH stress. Further research is being pursued to investigate whether other pH-lowering compounds possess anti-angiogenic properties (118).

1.3.5.2 Angiostatin Binds Endothelial Cell-Surface avP3Jntegrin

Takada et al. (119) have shown that angiostatin binds to the uvP3-integrin on the surface ofChinese hamster ovary (CHO) cells and bovine arterial endothelial (BAE) cells in an EACA­ dependent manner. They demonstrated that the binding was dose-dependent, saturable and blocked by integrin-specific as well as the RGD (arginine-glycine-aspartic acid)

31 peptide. Integrin-expressing cells bound to KI-5, KI-4 and KI-3, but not to full-Iength Pig.

They also demonstrated that plasmin, unlike PIg, is a ligand for av~3-integrin and induces endothelial cell migration. Plasmin-induced cell migration was strongly inhibited by angiostatin. Specific localization of the protease domain of plasmin the integrin on the cell surface is necessary to maintain its function. Angiostatin, and possibly other Plg kringles, block this interaction. Further investigation is required to determine the molecular basis for angiostatin-integrin interaction, as there are no RGD motifs found in either human or mouse

Plg or angiostatin sequences. Cell binding assays have not yet revealed whether the interaction is direct or indirect, for example through a binding site that is different from but influenced by an integrin. In addition, there have been no reports to date describing a human integrin-human angiostatin interaction, and since av~3-integrin knockout cells still display angiogenic features, av~3-integrin is not absolutely required for angiogenesis.

1.3.5.3 Angiostatin Binds Endothelial Cell-Surface Angiomotin

In 2001, an additional novel receptor for angiostatin was discovered on the endothelial cell surface and was called "angiomotin" (64) since it plays a role in endothelial cell migration and capillary tube formation. Angiomotin was identified as an angiostatin-binding receptor in a yeast 2-hybrid system using a placental cDNA library, which is rich in vascular endothelium mRNAs. It is important to note that employing a yeast 2-hybrid approach to screen membrane proteins presents 2 limitations: post-translational protein modifications essential for angiostatin signaling cannot be precisely recapitulated in the yeast nucleus and functional cell-surface domains required for protein-protein interactions may bind with reduced or altered specificity

(90). Angiomotin is composed of a conserved coiled-coil and PDZ binding domains (120) and

32 IS expressed in a variety of endothelial cell lines with particular high expression on microvascular and umbilical vein endothelial cells (64). Angiomotin was also found to be highly expressed on both capillaries and large vessels and its expression was observed to be restricted to endothelial cells. Angiomotin-transfected cells were capable of binding and intemalizing angiostatin (in a RGD-dependent fashion), leading to activation of focal adhesion kinase (F AK) activity. Since FAK is implicated in cell motility and adhesion-dependent cell

survival, angiostatin binding was found to suppress migration and lead to apoptosis via a RGD­ independent FAK activation pathway (102). Endothelial cells expressing truncated angiomotin

(deletion of 3 amino acids in the carboxy-terminal PDZ domain) were incapable of migrating and forming tubes in vitro (121). Furthermore, transgenic mice expressing truncated angiomotin died at embryonic day 9.5 due to non-migration of endothelial cells into the neuroectoderm and intersomitic regions (121). Endothelial cells exposed to exogenous angiomotin caused stabilization of established capillary tubes and increased invasion of endothelial cells derived from the formed tubules (122). Angiomotin remains to be confirmed as an angiostatin receptor using dose response, saturable binding kinetic and competitive antibody studies. AdditionaIly, angiomotin is not implicated in other angiostatin-induced effects such as inhibition of endothelial cell proliferation and promotion of endothelial cell death (90).

1.3.5.4 Angiostatin Possesses Multiple Binding Targets

In summary, the overall findings suggest that angiostatin binds several partners and cell-surface targets. Plg also binds multiple cell-surface, soluble and membrane-associated targets. Since Plg is activated during fibrinolysis, it participates in many binding events.

33 However, the kringle domains change their binding specificity once proteolytically cleaved to smaller fragments, since Plg does not seem to bind many of the targets that have been found to interact with its cleavage products (123-127). It is important to note that in light of the combinatorial complexity associated with different angiostatin fragments, it is highly likely that different investigators may be working with various angiostatin forms that may alter its binding specificity and therefore yield different conclusions regarding the key mediators of angiostatin action. It is also worthy to note that none of the angiostatin binding targets aforementioned are capable of affecting endothelial cell survival, proliferation, migration, mvaSlOn, morphogenesis, phenotype and gene expression (90). Thus, with the data available to date, angiostatin appears to induce its effects via interaction with multiple binding sites and receptors among which include FIFo ATP synthase (responsible for endothelial cell survival maintenance under pH stress), (role in endothelial cell attachment, survival and migration) and angiomotin (implicated in endothelial cell migration) (90).

1.3.6 Plasminogen Kringle 5 Protein

Human Plg K5 (Figure 1.3) contains 80 amino acid residues and it possesses highest sequence homology (about 60%) with KI and lower, but still significant, homology with K2,

K3 and K4 (61). Consistent with the lysine-binding properties of KI and K4, K5 also binds lysine residues, however in contrast to K4, K5 does not possess a lysine-ri ch cluster (61).

1.3.6.1 Anti-Endothelial Cell Potency ofKringle 5

The rationale of utilizing K5 as the anti-angiogenic therapeutic transgene for the purpose ofthis thesis is due to the fact that both proteolytic and appropriately folded

34 Figure 1.3. Schematic of the amino acid sequence of human plasminogen kringle 5.

K5 displays the typical features of kringle domains, possessing 3 disulfide bonds, which form a double looped conformation. In order to form angiostatiI14.s, the 2 circled disulfide bonds must be reduced to allow for separation of the carboxyl terminal fragment from the kringle domains that form angiostatiI14.s. The black bar indicates LySS3h which is at the carboxy-terminus of angiostatiI14.s. This figure was obtained from Stathakis et al. (92). FIGURE 1.3 recombinant K5 proteins exhibit substantially higher potency than any of the other individual kringle fragments. Of particular interest, it was shown by Cao et al. (128) that proteolytically­ derived as well as recombinant human K5 --on its own- displayed several orders of magnitude greater inhibitory effect on endothe1ial cell proliferation than Kl-4 (Figure 1.4). Moreover, Ji et al. (129) demonstrated that K5 also inhibits endothelial cell migration, which is an important process in angiogenesis. In addition, several studies report that K5 inhibits growth-factor stimulated angiogenesis by inducing cell-cycle arrest in Gl and apoptosis specifically in proliferating endothe1ial cells (128-130). It has been reported that the K5 domain enables Plg to bind to HUVEC with high affinity (131). In vitro studies indicate that the overall order of kringle fragmentes) endothe1ial cell inhibition is as follows: K5>Kl-3>Kl-4>K1>K3>K2>K4.

It is important to note that this ranking order may not be translatable in vivo, since K5 has a shorter half-life than either form of angiostatin.

Further studies performed by Cao et al. (132) demonstrated that Kl-5, that is angiostatin plus K5, acts as a more potent anti-angiogenic agent than angiostatin (KI-4) alone in preventing blood vesse1 formation in the mouse comea (Figure 1.5). It was observed that upon implantation of a pellet of FGF (an angiogenic stimulator) in the comea, the growth of blood vessels was reduced in mice receiving injections of angiostatin as compared to the PBS­ treated control mice. However, blood vessel formation was most significantly reduced in those mice receiving injections of Kl-5. These results suggest that angiostatin and K5 work synergistically to markedly improve the angiostatic effect. Moreover, this same group also investigated whether Kl-5 could induce an anti-tumor response (132). Syngeneic C57BL6/J mi ce were subcutaneously implanted with murine fibrosarcoma tumor cells, followed by daily systemic subcutaneous injections ofKl-5. Consistent with theirprevious findings, Kl-5

37 Figure 1.4. Comparison of the anti-endothelial cell properties of angiostatin and kringle 5. (A) Bovine capiIlary endothelial ceIls (BCE) were stimulated to proliferate with bFGF and exposed to either purified human proteolytic kringle 5 (hK5), mouse recombinant K5 (mrK5) or human proteolytic angiostatin KI-4. Although aIl 3 agents inhibited BCE proliferation in a dose-dependent manner, the human and mouse K5 prote in was significantly more potent than angiostatin KI-4. This figure was obtained from Cao et al. (128). (B) Human dermal microvascular endothelial ceIls were bFGF­ stimulated to migrate through gelatin-coated polycarbonate filters. Although a concentration-dependent inhibition of bFGF-stimulated endothelial ceIl migration was observed with angiostatin KI-4 and angiostatin KI-4.5, K5 induced the most potent inhibition of migration (50% inhibition of migration < 0.1 nM). Angiostatin KI-4.5 possessed an ED50 of approximately 2.8 nM compared with 100 nM for angiostatin K 1-4.

This figure was obtained from Soff et al. (60). FIGURE 1.4

A B

r-·-·-···------....., Endothelial cell migration

! bFGF (1 nglml) alone 100,. o T 80 + ii K1-4 + ~ -25 ~ 60 ~ hK5 ~.L 40 .S -50 (1) en s::: 20 ..c'" (.) -75 'o~ 0

-100 f 20 40 80 160 320 o 0.10 1.0 10 100 1000 Protein Cone. (!lM) Concentration (nM) Figure 1.5. Synergistic inhibition induced by Kl-5. (A) KI-5 inhibits FGF-induced

mouse corneal neovascularization more significantly than angiostatin (KI-4) as compared

to PBS-treated mice. (B) Daily systemic subcutaneous KI-5 injections is more effective

at reducing the tumor volume of subcutaneous fibrosarcoma tumors than KI-4 at the

same dose as compared to PBS-treated control mice. This figure was modified from Cao

et al. (132).

------FIGURE 1.5

A Inhibition of Mouse Comeal Neovascularization PBS K1-4 K1-5

B ln Vivo Tumor Suppression 4000 ,------, Dose = 2 mglkglday

~ 3000 63%tumor ---- K1-4 ! reduction j 2000 ~ )o{l·S ()"- e 1000 {!. a 0 5 la 15 20 Tlme(Days) induced a 63% reduction in tumor volume as compared to the saline-treated mice. It is

interesting to note that angiostatin at the same dose as KI-5 did not block tumor growth. This

suggests that K5 may mediate its endothelial cell suppressive effects via an alternative

mechanistic pathway as compared to angiostatin or that K5 is a more potent inducer of

inhibitory regulators on endothelial cells. Therefore, in light of the anti-angiogenic potency of

KI-5, it would be desirable to explore the use ofK5 alone as an angiostatic agent within a gene

therapy setting for the treatment of cancer.

1.3.6.2 Structural Features ofKringle 5

The high efficacy, cell-type specificity and short amino acid sequence of K5 suggest

that it possesses tremendous potential as an anti-angiogenic therapeutic agent. Due to its great

promise, the majority of groups to date have focused their research efforts on assessing the

anti-tumor ability of K5 and as a result, there is little evidence describing the structural and

mechanistic features ofK5. It was reported that the lysine-binding site ofK5 is not involved in

mediating its endothelial cell inhibitory properties, smce addition of trans-4-

aminomethylcyclohexane-I-carboxylic acid (t-AMCHA) had no effect in a migratory assay

(133). In addition, alkylation as well as reduction of the disulfide bonds within K5 enhances its

anti-migratory activity, suggesting that the kringle conformation may shield the functional

elements of K5 from effectively interacting with endothelial cells (129). Reduction of

angiostatin however, induces inactivation indicating that the kringle domains employ varying mechanisms to exert endothelial cell inhibition (88).

Recently, Cai et al. (134) reported that a deletion mutant ofPlg K5 possesses enhanced

anti-angiogenic properties when compared to the intact K5 protein in an oxygen-induced

42 retinopathy rat model. More specifically, PCR was used to delete 10 amino acids from the N­ terminus and 1 amino acid from the C-terminus, without disrupting the kringle domains. These

findings suggest that the flanking amino acid arms ofK5, which reside outside of the disu1fide bond region, may not be essential for the angiostatic activity of K5 and may in fact prevent the

functional elements of K5 from optimally interacting with endothelial cells. It is also possible

that the deletion mutant may possess a different conformational structure than intact K5 due to

the absence of these flanking amino acid arms.

1.3.6.3 Preclinical Testing: Direct Use ofRecombinant Kringle 5 Protein

Zhang et al. (135, 136) employed a rat model of oxygen-induced retinopathy, which

displays an overexpression of VEGF, pre-retinal neovascularization and haemorrhage. It was

demonstrated that intravitreal injection of K5 prior to the development of neovascularization

resulted in decreased neovascular tufts and pre-retinal vascular cells compared to control rats

receiving PBS injections (135). Moreover, injection of K5 after the development of

neovascularization inhibited the increase in the pre-retinal vascular cells (135). Therefore, K5

possesses the ability to prevent the development as well as haIt the progression of retinal

neovascularization (135). Injection of K5 did not induce any detectable inflammatory response

in the retina or histological toxicity to retina neurons and pre-existing vessels (135). A

subsequent report by Zhang et al. (136) utilizing the same rat model demonstrated that

intravitreal injection of K5 reduced vascular permeability, without affecting permeability in

normal rats. It was shown that K5 reduced vascular permeability at doses substantially lower

than that required for inhibition of retinal neovascularization and that the K5-induced decrease

in vascular permeability correlated with its ability to downregulate VEGF expression in the

43 retina (136). Moreover, K5 inhibited IGF-I-induced hyperpermeability, which is known to arise through upregulation of endogenous VEGF expression (136). Further studies by Zhang et al. (136) demonstrated that topical K5 administration delayed the onset of alkali-burn-induced comeal neovascularization and decreased areas of neovascularization in a dose-dependent manner. In addition, topical application of K5 post formation of comeal neovascularization induced regression in newly formed vessels and downregulated VEGF expression.

Yang et al. (137) recently confirmed that recombinant human K5 protein inhibited proliferation and induced apoptosis of primary endothelial cells in a dose-dependent manner.

They also confirmed that the in vitro inhibitory properties of K5 are specifie for endothe1ial cells, since K5 was unable to affect the proliferation of hepatoma cells. Ventral K5 injection suppressed the growth of grafted hepatoma tumor cells as well as prevented tumor growth by

68% in an athymie hepatocarcinoma xenograft mouse model as compared to control-treated mice. K5 exerted a potent anti-angiogenic effect in both mouse mode1s as evidenced by a substantial reduction in microvesse1 density. Interestingly, it was shown that K5 induced apoptosis of tumor tissues in vivo as observed by increased cleavage of pro-caspase 3 in tumor tisssues.

1.3.6.5 Putative Endothelial Cell-Surface Receptorsfor Kringle 5

1.3.6.5.1 Voltage-Dependent Anion Channel

In 2003, Pizzo and colleagues identified the first possible receptor for K5 as the voltage-dependent anion channel (VDAC1) present on the cell-surface of endothe1ial cells

(138). VDACs are a group of proteins that form channels through the outer mitochondrial membrane and have also been observed in human ske1etal muscle, B lymphocytes and

44 hematopoietic cell plasma membranes (138-140). It is believed that VDACI is concentrated in caveolae and caveolae-related domains of the plasma membrane (l38, 141). Caveolae are lmown to cluster groups of fibrinolytic proteinases (l38). Caveolin-l and -2 are highly expressed in normal human endothelial cells (l38, 142). VEGF-induced endothelial cell proliferation leads to a decrease in the expression of caveolin-l (l38). Angiostatin has been shown to block VEGF-induced caveolin-l downregulation (143).

It was reported that K5 mediates binding of Plg to the Plg activator streptokinase (SK)

(144, 145). It was also found that sequence similarities between SK and the mitochondrial human VDACI exist, although no functional relationship between these proteins has been reported to date (146). The region of homology is between SK residues T~52_Lys283 and between VDACI residues Tyr224_Lys255 (146). Pizzo et al. (l38) generated antibodies against peptides within these regions and used them to identifY VDACI on the cell-surface of

HUVECs by flow cytometry analysis. Purified K5 protein was found to bind with high affinity

(KI = 28 nM) to endothelial cells and binding was inhibited by the generated antibodies (l38).

Via its interaction with VDAC, K5 was shown to inhibit VEGF-stimulated HUVEC proliferation and interferes with signaling mechanisms regulating cytosolic intracellular free 2 Ca + (l38). In addition, K5 binding to HUVEC induced a decrease in intracellular pH and hyperpolarization of the mitochondrial membrane (l38), an early feature of apoptosis. Highly purified VDACI was shown to bind K5 after reconstitution of the receptor into liposomes

(l38). These findings suggest that VDACI is a receptor for K5. It was reported by Komatsu et al. (147) that cytosolic pH is increased in endothelial cells prior to angiogenesis and blood vessel repair. This group also demonstrated that cell proliferation is dependent on modulation

45 2 of intracellular Ca + concentration. Thus, K5 may act as an anti-angiogenic agent by interfering with both of these mechanisms in HUVEC.

1.3.6.5.2 Glucose-Regulated Protein 78 Receptor

More recently, it was reported by Davidson et al. (148) from Abbott Laboratories that

K5 binds with high affinity to the glucose regulated protein 78 (GRP78), which was shown to be expressed on the surface of hypoxic or cytotoxic stressed tumor cells and on the surface of stimulated endothelial cells. GRP78, also referred to as BiP, is involved in regulating the function of the endoplasmic reticulum (ER) due to its roI es in protein folding and assembly, targeting misfolded proteins for degradation, ER Ca(2+)-binding and controlling the activation of trans-membrane ER stress sensors (149). GRP78 possesses an anti-apoptotic property, thus stress induction of GRP78 represents an important pro-survival component of the unfolded protein response (149). Most notably, the discovery of cell-surface expressed GRP78 on a wide variety of stressed cancer cells may be further exploited to develop anti-cancer therapeutics. It has been reported that when tumor cells are exposed to stressful conditions such as hypoxia or cytotoxic stress, the emergence of an apoptosis-resistant phenotype emerges, which has been linked to GRP78 cell-surface expression.

Davidson et al. demonstrated that the peptide sequence PRKLYDY around the lysine­ binding pocket of K5 mediates apoptosis of proliferating endothelial cells and competes with

K5 for binding to endothelial cells. An immobilized NH2-terminally biotinylated PRKLYDY peptide was used to isolate K5 binding proteins from endothelial cell surfaces. Tryptic peptides of the major eluted protein underwent mass spectrometric sequencing and revealed sequences corresponding to GRP78. Confirmation that GRP78 acts as a cell-surface receptor for K5 was performed using coprecipitation of biotinylated endothelial cell-surface proteins with S-tagged

46 K5. Equilibrium dialysis was used to demonstrate direct binding of recombinant GRP78 protein

(yeast-derived) to 3HK5 (yeast-derived) in a concentration-dependent manner with a Kt = 0.25 nM. It is important to note that the GRP78:K5 binding possesses a significantly lower dissociation constant compared to the VDAC:K5 binding, which suggests that GRP78 is more likely to be the main high affinity endothelial cell-surface binding receptor for K5. In addition,

Pizzo et al. utilized confluent, static endothelial cell monolayers to examine K5 binding to

VDAC, whereas Davidson et al. used subconfluent endothelial cell culture conditions since

GRP78 was minimally exposed on confluent endothelial cells. Therefore, under static, confluent culture conditions other proteins maya more significant role for K5 binding. It has also been reported that VDAC pro teins get chaperoned to the cell-surface by heat shock or glucose-regulated proteins. Thus, it is possible that K5 binds to cell-surface expressed GRP78, which is associated with VDAC. Furthermore, Davidson et al. demonstrated that K5 induces apoptosis of several hypoxic tumor cells that upregulate GRP78 and that caspase-7 is implicated in mediating the apoptotic effect (148). It was also recently reported that the ATPase domain of GRP78 binds to procaspase-7, preventing activation of procaspase-7 and as a result, decreases the stress-induced cellular apoptotic response (150). Nuclear magnetic resonance studies performed by Davidson et al. using a recombinant GRP78 ATPase domain indicate that

K5 binds directly to the ATPase domain of GRP78 and immunohistochemical analysis demonstrated that the ATPase domain of GRP78 is present on the extracellular surface of stressed tumor cells (148). It was thus proposed that K5 binding to the cell-surface expressed

ATPase domain of GRP78 on stressed tumor cells leads to endocytosis of the K5:GRP78 complex, whereby K5 competes with procaspase-7 for GRP78 binding resulting in caspase-7 activation and promotion of apoptosis (148). To date, Davidson et al. provide the first

47 evidence that an anti-angiogenic agent possesses the ability to induce apoptosis of stressed tumor cells (148). It was postulated that K5 would induce apoptosis of stimulated tumor­ associated endothe1ial cells and the subsequent hypoxia would induce tumor cells to upregulate cell-surface GRP78, which would lead to tumor cell apoptosis via K5 binding (148).

Future studies may reveal other novel binding receptors/partners for K5, which can be exploited for therapeutic anti-cancer purposes. Moreover, further investigation on the biology of K5 may divulge that K5 possesses a broader cell specificity than initially be1ieved, targeting other cell types in addition to stressed tumor cells and stimulated tumor-associated endothelial cells.

1.3.6.6 Other Kringle-Containing Proteins

Besides PIg, several other pro teins have been identified as containing internaI kringle domains and inc1ude the following: prothrombin, the unactivated form of , t-PA, u­

PA, apolipoprotein A, TrK-related tyrosine kinases and hepatocyte growth factor (88).

Typically, kringle fragments have been characterized as protein recognition domains and usually display affinity for C-terminal lysine residues, which can be mimicked by EACA (88).

Kringle domains have also been reported to interact with proteins that do not possess a C­ terminal lysine residue such as tetranectin, which binds to Plg K4, the pneumo coccus surface protein PAM, which binds to K2, factor Va, which binds to KI and K2 of prothrombin, and the

NG2 proteoglycan, which binds to angiostatin (88). It is also common for kringle domains to be found in multiple tandem repeats as is the case for plasminogen, which contains 5 kringle domains and apolipoprotein, which contains at least 41 kringle repeats (151). Sorne of these

48 kringle-containing proteins possess angiostatic activity such as prothrombin kringle 2 (152) and

NK4, which encompasses kringles 1 through 4 ofhepatocyte growth factor (153).

1.4 CANCER GENE THERAPY

Gene therapy involves the transfer of foreign genetic material into a patient in order to

treat a particular disease. In the past decade, there has been much effort to develop cancer gene

therapy as a novel strategy to treat cancer. Since cancer is a multi-stage disease, various gene therapy strategies have been examined such as: (i) transfer of toxic transgene products such as

tumor suppressor genes or suicide genes/enzymes (pro-drug approach) or expression of

antisense, ribozymes or smaIl interfering RNAs (siRNAs) for dominant oncogenes; (ii)

expression of immunomodulatory cytokines, co-stimulatory molecules or tumor specifie

antigens; (iii) expression ofmolecules affecting angiogenesis, ceIl adhesion and metastasis; (iv)

chemosensitization and radiosensitization approaches; (v) chemoprotection of stem ceIls; (vi)

expression of ceIl-surface receptors/ligands that target tumor ceIls; (vii) expression of

transgenes with tumor-specific promoters and lastly, (viii) the use of replicating (oncolytic) vectors (154).

Many malignancies are associated with loss oftumor suppressor genes such as p53, the retinoblastoma gene pRE and p16/NK4A (155). In order to induce apoptosis or ceIl cycle arrest, tumor suppressor genes have been introduced into tumor ceIls using a strategy known as "gene

replacement" (156). It has been reported that p53 can also suppress angiogenesis and tumor invasion, which further enhances its anti-tumor effect (157, 158). Transfer oftumor suppressor genes has either been performed by directly injecting gene-expressing vectors into the tumor

49 mass, by using tumor-targeted vectors or in an ex vivo approach whereby tumor cells are killed as is the case for bone marrow purging prior to autologous bone marrow transplantation (154).

The suicide gene therapy approach utilizes vectors expressing suicide genes whose protein product exposed to small molecular weight drugs results in cancer cell death (referred to as the suicide gene/enzyme pro-drug approach (154). For instance, a vector harboring bacterial cytosine deaminase in combination with systemic delivery of 5' -fluoro-cytosine can convert the pro-drug 5' -fluoro-cytosine to the cytotoxic drug, 5' -fluoro-uracil. The herpes simplex thymidine kinase gene has also been used in combination with ganciclovir, which pro duces phosphorylated ganciclovir as the toxic species. This approach is further enhanced by the "bystander effect" that ensues following the generation of the toxic gene product, which kills adjacent non-gene-modified tumor cells as weIl (154). Vector delivery can be performed via direct injection in the tumor, ex vivo or via tumor-targeted systemic administration.

An alternative strategy involves inhibition of dominant oncogenes (154). Vectors containing antisense oligodeoxynucleotides, ribozymes or siRNAs can inhibit the expression of oncogenes such as ras and HER-2/neu. Both in vivo tumor targeting as weIl as ex vivo tumor ceIl killing can be utilized. These approaches can also be used to target specific growth factors and cytokines implicated in angiogenesis, such as VEGF. There are also other genes that play key roles in tumor invasion and metastasis that are either deficient in tumor cells, such as and TIMP-l, or overexpressed such as metalloproteinase enzymes, which can be specifically targeted and either enhanced or suppressed respectively using gene-expressing vectors that can be directly injected intratumorally or targeted to tumor-associated vasculature

(154)

50 Strategies aimed at modulating the immune system to enable it to recognize cancerous cells as "foreign" cells have also been explored extensively. Immune escape is one of the mechanisms employed by tumor cells to avoid immune recognition and killing, thus developing methods to circumvent this escape is crucial. Tumor cells evade recognition by the immune system via the following mechanisms: lowering major histocompatibility class 1 (MHC-I) and

MHC-II expression, decreased growth and differentiation of immune effector cells and via defects in the expression of costimulatory molecules (154). Immune modulatory approaches attempt to activate the immune system or alter the local immune microenvironment by introducing genes such as IL-2, IL-4, IL-12, tumor necrosis factor (TNF), IFN-y, GM-CSF, immunostimulatory genes (allogenic MHC antigens) or T-cell costimulatory molecules such as

87.1 and 87.2 (154). Cell therapy approaches have been tested whereby autologous tumor cells are engineered with cytokines, immunostimulatory or costimulatory molecules and subsequently used as a vaccine to treat the patient. Other strategies involve engineering T­ lymphocytes or dendritic cells with cytokines to ameliorate their anti-cancer fighting properties as well as expressing variable regions of tumor-specific monoclonal antibodies to recognize and and kill tumor cells (154).

Alternative approaches involve sensitizing cancer cells to chemotherapy and radiotherapy by transferring for example wild-type p53, which has been shown to have chemosensitizing effects (156), or the liver cytochrome P450 gene, which has demonstrated chemosensitization to oxazaphosphorines, like cyclophosphamide and ifosfamide, that require activation by cytochrome enzymes (154). Another strategy involves engineering hematopoietic stem cells with drug resistant genes such as the multi drug resistant-l gene (MDR-l) to

51 increase their resistance to cytotoxic drug therapy (154). This approach allows patients to receive higher doses of chemotherapeutic agents.

1.4.1 Anti-Angiogenic Gene Therapy Strategies

Delivery of anti-angiogenic proteins can be performed using 2 main strategies. The tirst approach utilizes systemic gene therapy, whereby gene-engineered vectors are injected into the patient in order to provide high levels of circulating angiostatic agents (159). Preclinical animal models and early Phase l clinical trials using this approach to test the efticacy of angiostatin or endostatin did not demonstrate toxicity (66, 160). A caveat is the possibility that systemic vector administration could result in potential adverse effects, especially when employing viral­ based vectors (5). The second approach utilizes tumor-targeted anti-angiogenic gene therapy, which aims to increase the levels of angiostatic agents within the local tumor microenvironment (161). High gene expression is maintained in the tumor region and thus possible systemic toxicity due to the biological properties of the anti-angiogenic agent or the vector is avoided. One major shortcoming associated with this approach is that the tumor may be inaccessible for vector delivery of the protein of interest, either due to the tumor location or multiple metastases (5).

1.4.2 Cancer Gene Delivery Vectors

A gene delivery vector enables the transfer of genetic material into target cells. There are two main types of vectors currently employed to deliver the therapeutic gene of interest for cancer gene therapy strategies: viral-based and non-viral based. Virus es evolved natural mechanisms to deliver their genomes into cells therefore they are ideal vectors to deliver

52 foreign DNA. The most frequently used viral vectors are derived from retroviruses, adenoviruses and adeno-associated viruses (AAVs). Non-viral vectors are comprised of either plasmid DNA or chemically synthesized compounds. The use of viral vectors is advantageous for use in anti-angiogenic cancer gene therapy since: (1) the majority of gene therapy vectors can be easily manufactured and purified to high titers; (2) they are stable and can be stored for a long duration; (3) they can be engineered to specifically bind tumor-associated endothelial cell markers as well as selectively express genes long-term in tumor vasculature and (4) they can be engineered to directly kill tumor-derived endothelial cells upon binding. Selection of the optimal vector type is based on the following criteria: (1) target cell type- tumor cells or host cells as well as developmental nature of the target cells (epithelial origin, hematopoietic stem ceIls, neuronal cells or skin cells etc.); (2) cell-cycle status oftarget ceIl- active mitotic division or quiescent; (3) gene transfer efficiency requirement; (4) vector targeting and vector tropism required to be able to administer the vector locally or systemicaIly; (5) duration of transgene expression- short-term or long-term; (6) development of adverse effects due to vector-mediated gene therapy- (for example: the presence of pre-existing antibodies against the vector) and ability to cease therapy in the event of a harmful side-effect; (7) vector should not be capable of integrating into reproductive cells and; (8) size of the gene to be delivered (154).

1.4.2.1 Retroviral Vectors

Retroviruses are composed of 2 positive single RNA strands, which upon infection into the target cells are copied by the viral reverse transcriptase into double-stranded DNA

(dsDNA). The dsDNA then migrates to the nucleus where it stably integrates into the genome of dividing target cells (162). The most commonly used retro virus is the Moloney Murine

53 Leukemia Virus (MoMuLV), which is composed of a double stranded RNA genome tlanked by

5' and 3' long terminal repeats (LTR). The transfer ribodeoxynuc1eic acid (tRNA) primer binds

adjacent to the 5'LTR and reverse transcription is initiated. This region also contains splice

donor and encapsulation sequences, gag, pol and env genes, which encode for the viral

structural proteins ribonuc1eoprotein core (gag); protease (pro), reverse transcriptase/RNaseH

and integrase enzymes (pol); and the envelope glycoproteins (env). In order to generate

recombinant retroviral vectors, coding sequences from gag, pol and env genes are replaced by

the cDNA of the therapeutic gene of interest (up to 8 kb). Deletion of required viral sequences

necessitates that these genes be expressed in trans in what are referred to as "packaging" cell

lines. Retroviral packaging cell lines constitutively express the gag, pol, pro and env genes,

which provide the necessary helper sequences for viral propagation, however these genes are

expressed without their packaging sequence (154). A plasmid encoding the therapeutic gene of

interest with minimal viral sequences is subsequently transferred into packaging cells, resulting

in the generation of replication incompetent retro viral partic1es, which are able to transfer the

therapeutic gene into target cells. The minimal viral sequences comprise the following: (1) the long terminal repeats (LTRs), (2) the primer binding site, (3) the polypurine tract (PPT) and (4) the packaging sequence (163). Retroviral infection into the target cell genome requires that the viral envelope glycoprotein attach to the cell-surface receptors of the target cell (164).

MoMuLV-derived vectors employ 2 different envelope glycoproteins to induce gene transfer

into target cells: the ecotropic glycoprotein, which binds an amino acid transporter present only on murine cells (165), and the amphotropic glycoprotein, which binds a phosphate transporter that is present on most cell types inc1uding rodents and humans (166). Subsequent to viral entry

54 into the cytoplasm, the viral RNA is reverse-transcribed into DNA and undergoes random integration into the host cell genome (164).

The use of retro viral vectors is advantageous since they preferentially gene-modify dividing cells, except for lentiviral vectors (a subc1ass of retroviruses), and they induce stable integration and expression of the therapeutic transgene in the genome of the target cell.

Typically, transgenes up to 8-10kb in size could be introduced within retro viral vectors. Viral titers range from 105 to 106 infectious partic1es per mL and can be concentrated to more than

1012 infectious partic1es per mL when the viral partic1es incorporate the G-glycoprotein from the vesicular stomatitis virus enve10pe (167).

There are sorne limitations associated with the use of retroviral vectors. Although

retro viral vectors can stably integrate the transgene expression cassette into the genome of the

target cell, accidentaI random integration into the host has occurred and caused

de1eterious effects such as activation of specific protooncogenes by insertional mutagenesis or

suppression of tumor suppressor genes (154). Random integration has been shown to cause

leukemia even when using replication incompetent retro viral vectors. This phenomenon has been observed in mice mode1s and in humans with severe combined immunodeficiency disease

who received bone marrow transplants transduced with a therapeutic retroviral vector (164).

Transplantation of the retrovirally gene-modified autologous bone marrow cells resulted in

development of leukemia only after a long delay. This suggests that mutagenesis due to random

integration may involve other factors such as the characteristics of the particular retro viral

preparation (retroviral vector may carry "passenger" noncoding regulatory RNAs possessing

the ability to upregulate genes affecting cell proliferation) (164, 168). Another major

disadvantage is the possible generation of replication competent retroviruses (ReR). There has

55 been no report to date describing such an event however, it is possible that a replication­ incompetent retro viral vector administered in vivo could recombine with endogenous virus es to create a new RCR virus.

1.4.2.1.1 Inhibition of Tumor Angiogenesis Using Retroviral Vectors

Several independent groups have utilized retroviral gene therapy to deliver anti­ angiogenic proteins long-term in preclinical tumor models. Most studies have used the

MoMuLV -based retroviral vectors. Since these viral vectors can only infect dividing cells, replication-deficient, third-generation lentiviral vectors based on the human immunodeficiency virus-l have been developed to overcome this limitation (5).

Our laboratory is the first to test the efficiacy of K5 protein alone in the context of a gene therapy strategy for brain (169) and breast cancer (170). Our approach exploits the use of ex vivo retroviral gene transfer to gene-modify tumor cells prior to implantation and further assessment in vivo. Other groups have also employed ex vivo retroviral gene transfer to test the efficacy of various angiostatic agents. Scappaticci et al. (171) showed that a combination treatment of retrovirally delivered angiostatin and endostatin induced a synergistic effect in animal models of melanoma and leukemia by prolonging the survival of animaIs receiving the combination treatment when compared to animaIs receiving either protein alone. Tanaka et al.

(172) demonstrated that implantation of ex vivo retrovirally-engineered platelet factor-4- expressing glioma cells gave rise to smaller and less vascularized tumors when compared to control non-gene-modified tumors. Consistent with these results, Machein et al. (173) also observed suppression of tumor growth in a rat model when orthotopically implanting glioma cells that were retrovirally-modified ex vivo with the dominant negative Flkl (lacks

56 intracellular signaling domain). It is important to highlight that retroviral vectors are efficient at stably transducing tumor cells ex vivo, however their efficiency decreases substantially when they are used for direct in vivo gene transfer in pre-established tumors.

1.4.2.2 Lentiviral Vectors

Recombinant lentiviral vectors have also been used for gene transfer. Similar to retroviral vectors, lentiviral vectors integrate in host-cell (174). They can be exploited for applications that require long-term transgene expression. Elimination of viral open reading frames from recombinant vectors allows for a nonimmunogenic gene delivery approach, however immunogenicity could develop against the transgene itself and from the virion-associated antigens that enter cell at the time of infection (174). The use of lentiviral vectors is attractive since they can transduce non-dividing cells (both ex vivo and in vivo) and possess stable and relatively large genomic capacity. Stable transmission of large genomic fragments allows for improved transgene expression. The development of third-generation packaging systems derived from HIV -1 will allow for a broader investigation of lentiviral­ mediated gene transfer. Risk of insertional mutagenesis and low viral titer limit the use of lentiviral vectors. The development of improved lentiviral vector targeting is currently underway using virions pseudotyped with alternative envelopes or genetically engineered envelope substitutes (174). Lentiviral vectors are therapeutically useful for strategies that require efficient gene transfer in stem cells, T lymphocytes and tumor cells.

1.4.2.3 Adenoviral Vectors

57 Adenoviruses are double-stranded DNA non-enveloped viruses that replicate in the nucleus outside the chromosome and are capable of infecting a wide variety of cells, including dividing and non-dividing cells (154). Adenoviral vectors can be prepared at extremely high titers in the range of 1012 plaque-forming units (PFU) per mL, which is sufficient for in vivo use. Adenoviruses attach to target cells using a fiber protein that binds to the

Coxsackie/adenovirus receptor (CAR) and enter cells by receptor-mediated endocytosis (154).

Adenoviral vectors have a larger packaging capacity compared to retroviruses and thus can accommodate large transgene sequences (30-35kb) (154). In contrast to retroviruses, adenoviruses do not integrate into the genome of the infected cell and this feature minimizes the development of malignancy. The non-integrating nature of adenoviral vectors leads to high but transient transgene expression. In addition, adenoviruses elicit a strong host immune response and this also contributes to the short-lived transgene expression. There are 3 main concems associated with the use of adenoviral vectors: (1) organ inflammation due to immune reactions against the vector, (2) the development of tolerance to the vector that could result from disease upon infection with wild-type adenoviruses (cause respiratory infections) and (3) the generation ofreplication competent adenoviruses (175).

Several replication-defective adenoviral vectors for cancer therapy are in late-stage clinical trials, including Advexin (lntrogen Therapeutics), which expresses wild-type p53, and

GV AX (Cell Genesys), which expresses GM-CSF and is used for the ex vivo generation of autologous cancer vaccines (176). Other adenovirus-based gene therapies in early-stage trials include TNFerade (Gen Vec), which expresses TNF-a and EG009 (Ark Therapeutics), which expresses HSV-TK (176).

58 Another approach involves the use of replication selective adenoviruses. Deletion of

viral genes necessary for efficient replication in nonnal cells but expendable in tumor cells has been used to engineer replication selectivity (176). ONYX-15 was the first adenovirus designed

using this strategy. McConnick hypothesized that an adenovirus with deletion of a gene

encoding a p53-inhibitory protein, EIB-55-kDa, would be selective for tumors that had already

inhibited or lost p53 function (176). More than 230 patients have been treated with ONYX-15

via different administration routes and ONYX-15 was reported to be well tolerated (176).

1.4.2.4 Adeno-Associated Vectors (AA Vs)

AA Vs are single-stranded DNA viruses that are derived from a non-pathogenic and

defective human parvovirus that often coexists with adenoviruses. AAVs replicate and

integrate in the nucleus in the presence of a helper virus (177). Thus, AA Vs require co­

infection with another virus, typically an adenovirus or a herpes simplex virus. Although 90%

of humans show evidence of prior infection with AA V s, it has not been associated with human

disease (177, 178). Wild-type AA Vs integrate at a specifie region on chromosome 19 however,

AAVs integrate randomly since the Rep protein necessary for site-directed integration is

lacking (178). Similar to retroviral vectors, all coding sequences within AA Vs are deleted and

thus the development of an immune response toward viral proteins is avoided. AAVs are

relatively safe and nonimmunogenic, provide high titers and possess the ability to infect many

cell types among which include quiescent cells, hematopoietic cells as well as tenninally

differentiated cells. AAV s are integrating vectors and thus long-tenn gene expression has been

achieved, lasting up to 1 year post-infection. However, due to the small size of AA Vs, only

small inserts of up to 5kb can be packaged. AAV s also possess potential risks such as (J) the

59 presence of contaminating adenovirus which may cause adverse effects, (2) the potential for

insertional mutagenesis and (3) the generation of wild-type AAVs due to recombination

between the vector and the packaging plasmid (179).

1.4.2.5 J7accinia J7ectors

The poxvirus vaccinia has been extensively used clinically for smallpox vaccination.

Vaccinia virus is a large virus with a double envelope surrounding an 186-kb DNA genome.

Vaccinia vectors have been utilized in human clinical trials to deliver genes encoding tumor­

associated antigens and cytokines. The vectors have been administered via different routes

including intratumoral, subcutaneous and intradermal. One major disadvantage of vaccinia

vectors is that they are replication-competent viruses, which poses a safety concem in

immunosuppressed cancer patients. This problem has been addressed by developing vectors

based on other poxviruses, such as fowl pox and canary pox, which do not replicate in humans.

Despite this effort, vaccinia and other poxvirus vectors express a large number of viral proteins

in addition to the cytokine or tumor-associated antigen of interest. As a consequence, the

antiviral immune response tends to dominate the antitumor immune response and this effect has

limited the use of these vectors.

1.4.2.6 Herpes Simplex J7irus Type 1 J7ectors

The Herpes simplex virus type 1 (HSV-1) is another large virus used in cancer gene therapy. The genome is a linear, double-stranded DNA molecule (180). Replication-defective

HSV -1 vectors, which cannot pro duce infectious progeny and thus have reduced toxicity to infected cells, can be generated by deleting one or more of the immediate early genes essential

60 for virus repli cation (180). The main advantage of replication-defective HSV-1 vectors for

cancer gene therapy is their large cloning capacity, which has been exp10ited to deliver distinct

therapeutic genes to the heterogeneous population of cells within a solid tumor (180).

Oncolytic viral therapy represents a promising strategy for treating cancer that involves

replication-competent viral vectors that can replicate in situ in tumor cells, induce oncolytic

activity by direct cytocidal effects and then spread throughout the tumor (181). In addition,

these vectors are capable of transferring and expressing foreign genes in host cells (181). HSV-

1 possesses many features that make it attractive for cancer therapy: it infects most tumor cell

types, its life cycle is well studies, its genome has been sequenced, the function of the majority

of genes has been identified, the genes can be manipulated and the large size of the genome

(181). HSV -1 is a neurotropic virus and many of the genes necessary for neuropathogenicity

are nonessential and can be mutated (181). Therefore, HSV-1 is particularly useful for brain

tumor therapy. The therapeutic benefits of oncolytic HSV-1 vectors depend on the extent of

both intratumoral viral repli cation and induction of host antitumor immune responses (181).

Third-generation oncolytic HSV -1 vectors possess enhanced properties and retain the safety

features of second-generation vectors (pathogenicity substantially attenuated by genetically

engineering mutations in two different genes). This approach has features that make it very

attractive for clinical application: tumor cells are targeted irrespective of their genetic makeup,

it can be combined with conventional therapies such as surgery, radiation therapy and

chemotherapy, combination with immunotherapy has potential synergistic effects and it can act

as a vehicle for gene delivery in vivo (181). It has been demonstrated in several clinical trials

that oncolytic HSV -1 vectors can be administered safely in humans and CUITent efforts are underway to improve antitumor efficacy without compromising safety.

61 1.4.2.7 Non-Viral Vectors

Non-viral vectors transfer genetic material without utilizing a viral particle such as naked DNA plasmids amplified in bacteria or eukaryotic ceIls, as weIl as chemically synthesized oligonucleotides (182). Plasmid DNA can harbor up to 15kb of genetic material and requires cationic liposomes or receptor-mediated targeting in order to efficiently enter a target cell. Synthesized oligodeoxynucleotides contain between lOto 25 bases and are generally used to alter processmg, translation or stability of targeted RNA.

Oligodeoxynucleotides can also be used to inhibit DNA transcription or as decoys to transcription factors. The main advantage of using non-viral vectors is the absence of risk for generating replication competent viruses. However, one major drawback is the typically transient expression of plasmid-encoded genes. Potential risks associated with the use of non­ viral vectors include: (1) the possibility of insertional mutagenesis if the plasmid integrates and

(2) toxicities induced by the compounds used to facilitate cell entry (183).

1.4.3 Future of Cancer Gene Therapies

Although several cancer gene therapies have entered clinical trials and have demonstrated safe and well-tolerated results, evidence of significant clinical efficacy has not been forthcoming. The main reason for the lack of successful treatment involves the inefficiency of targeting tumor cells using systemic delivery methods. Preclinical data indicates that tumor-targeted gene delivery approaches using tumor-specific promoters holds greater promise and remains to be tested in clinical trials. It is believed that cancer gene therapies will be most efficacious if used as adjuvants and in conjunction with CUITent treatment modalities such as chemotherapy and radiation therapy.

62 1.5 SPECIFIC RESEARCH AIMS

The main objective of the research presented in this thesis was to develop novel anti­ anglOgemc gene therapy strategies for cancer. The specific research aims were as follows:

1. To determine whether retrovirally gene-modified human U87 glioma cells

producing soluble human plasminogen kringle 5 protein could suppress tumor

progression and angiogenesis in an orthotopic brain cancer model;

2. To determine whether human plasminogen kringle 5 protein, when secreted by

retrovirally gene-engineered murine DN3 mammary adenocarcinoma cells, IS

dependent upon innate immune effector cells to mediate its anti-cancer effect;

3. To determine whether recombinant human angiostatin protein, in addition to its

strong anti-angiogenic activity, can also exert inhibitory action on macrophages.

63 CHAPTER2

64 CHAPTER2

Plasminogen Kringle 5-Engineered Glioma Cells Block Migration of Tumor­ Associated Macrophages and Suppress Tumor Vascularization and Progression.

Reference: Perri SR, Nalbantoglu J, Annabi B, Koty Z, Lejeune L, François M, Di Falco MR, Béliveau R, Galipeau J. Plasminogen Kringle 5-Engineered Glioma Cells Block Migration of Tumor-Associated Macrophages and Suppress Tumor Vascularization and Progression. Cancer Research. 2005 Sept 15;65(18): 8359-65.

65 2.1 ABSTRACT

Angiostatin, a weIl characterized angiostatic agent, is a proteolytic cleavage product of human plasminogen encompassing the first four kringle structures. The fifth kringle domain (K5) of human plasminogen is distinct from angiostatin, and has been shown -on its own- to act as a potent endothelial cell inhibitor. We propose that tumor­ targeted K5 cDNA expression may act as an effective therapeutic intervention as part of a cancer gene therapy strategy. In this study, we provide evidence that eukaryotically­ expressed His-tagged human K5 cDNA (hK5His) is exported extracellularly and maintains predicted disulfide bridging conformation in solution. Functionally, hK5His protein produced by retrovirally-engineered human U87MG glioma cells suppresses in vitro migration of both human umbilical vein endothelial cells and human macrophages.

Subcutaneous implantation of Matrigel™ -embedded hK5His-producing glioma cells in

NOD-SCID mice reveals that hK5His induces a marked reduction in blood vessel formation and significantly suppresses the recruitment of tumor-infiltrating

CD45+Mac3+Grl- macrophages. TherapeuticaIly, we demonstrate in a nude mouse orthotopic brain cancer model that tumor-targeted K5 expression is capable of effectively suppressing glioma growth and promotes significant long-term survival (>120 days) of test animaIs. These data suggest that plasminogen K5 acts as a novel two-pronged anti­ cancer agent, mediating its inhibitory effect via its action on host-derived endothelial cells and tumor-associated macrophages, resulting in a potent, clinically-relevant anti­ tumor effect.

66 2.2 INTRODUCTION

Glioblastomas are very aggresslve solid tumors characterized by hypervascularization and extensive tumor cell invasion into normal brain parenchyma

(49) . CUITent therapeutic modalities primarily target rapidly dividing malignant cells and include combinations of surgery, radiotherapy and chemotherapy (49, 184). However, these therapies remain ineffective with more than 90% of patients experiencing local reCUITence and a 5-year survival rate of only 9% (185).

The urgent need for the development of potent anti-glioma therapeutics as weIl as the importance of angiogenesis in glioma growth (186, 187) has fueled the identification and characterization of numerous anti-angiogenic agents, aimed at interrupting new vessel formation and ultimately arresting tumor growth. The use of anti-angiogenic agents as therapeutics is an appealing approach targeting nonmalignant endothelial cells that form the tumor vasculature and indirectly affects tumor ceIls, thus minimizing the risk of toxicity. In addition, the problem of drug resistance associated with conventional chemotherapy agents, is avoided since normal endothelial cells are genetically stable unlike tumor cells (188). Moreover, the blood-brain barrier is another obstacle often encountered with therapies targeting malignant cells within the brain and may be exploited with anti-angiogenic strategies by maintaining the angiogenic inhibitor within the brain vasculature for a prolonged period. Although angiogenesis inhibition offers several advantages over traditional therapeutic approaches, it is expected to induce a cytostatic effect resulting in tumor stabilization not eradication. Furthermore, single agent anti-angiogenic therapy may lead to a compensatory increase in the production of other angiogenic factors, which may then sustain angiogenesis (1). Despite these limitations, it

67 is now weIl recognized that angiostatic agents either singly or in combination with other treatment modalities could offer superior therapeutic benefits than currently attainable

(189-195), converting the tumor into a controIled, quiescent chronic disease. Direct delivery of purified recombinant anti-angiogenic proteins however necessitates large quantities and repeated administration of the therapeutic gene product for a prolonged period of time (66, 104). To circumvent the pitfalls associated with conventional treatments, we have developed a potent anti-angiogenic gene therapy strategy for brain cancer. Moreover, since malignant gliomas localize within the CNS and do not form distant metastases (196), they represent an attractive target for local gene therapy, which provides high and sustained local production of the desired therapeutic agent and offers a viable alternative to existing therapeutic interventions.

Several inhibitors of angiogenesis exist endogenously as proteolytic cleavage products of larger precursor molecules. Angiostatin, a well characterized anti-angiogenic agent discovered by O'Reilly and colleagues in Folkman's laboratory, was initially identified as an endothelial cell growth inhibitor present in urine and plasma of animaIs harboring solid tumors (59, 197), is an internaI cryptic fragment of human plasminogen

(PIg) encompassing the first four kringle (KI-4) domains (Figure. 2.1A). The K5 domain of Plg has been expressed as a recombinant protein in bacteria and been found to be more potent than KI-4 or any of the Plg kringles expressed individually in inhibiting growth factor stimulated proliferation of endothelial cells in vitro (86, 128). It has also been demonstrated that kringle domains 1-5 (KI-5) act as more potent endothelial cell inhibitors in vitro and are more effective in suppressing fibrosarcoma tumor growth in vivo as compared to KI-4 alone (132). We propose that the K5 domain could serve as a

68 potent angiostatic agent on its own and that it may act as a useful therapeutic transgene

within a cancer gene therapy strategy. To test this hypothesis, we have developed a human K5-expressing retroviral vector and tested its efficacy in vivo using an orthotopic brain cancer model.

2.3 MATERIALS AND METHODS

2.3.1 Cell Culture Reagents.

293GPG pantropic retrovirus-packaging cell line (198) was a gift from Dr.

Richard C. Mulligan (Children's Hospital, Boston, MA). Human glioma U87MG cellline

(199) was a gift from Dr. Stéphane Richard (Lady Davis Institute, Montreal, Quebec,

Canada). Freshly isolated primary human umbilical endothelial cells (HUVEC) were

generously provided by Dr. Mark Blostein (Lady Davis Institute) and maintained in

endothelial-basal media (EBM-2; Cambrex, Walkersville, MD) supplemented with EBM-

2 SingleQuots. Fresh human peripheral blood mononuc1ear cells were obtained from

normal volunteers and maintained in DMEM-F12 (Wisent Technologies) supplemented with 2% human serum and 500U/ml granulocyte/macrophage colony stimulating factor

(GM-CSF; Immunex, Thousand Oaks, CA).

2.3.2 Soluble Human Kringle 5-HisTag Expression Vector, Protein Purification &

Detection.

pBLAST -hK5 cDNA (InvivoGen, San Diego, CA) encodes for hK5 and contains

a 5' interleukin-2 signal peptide (IL-2sp) coding sequence. A 3' 6-histidine tag was PCR

c10ned to generate IL-2sp/hK5/HisTag (hereafter hK5His). Conditioned media was

69 collected from hK5His stably transfected human kidney 293T cells, concentrated using a

Centricon® Plus-20 column (Millipore, Billerica, MA) and purified using a Ni-NTA­ based resin affinity chromatography purification system (Novagen, San Diego, CA).

Immunoblot analysis was performed on eluted fractions using a polyclonal anti-His

antibody (Santa Cruz Biotechnology, Santa Cruz, CA).

2.3.3 Proteomic Analysis of Purified Soluble hK5His.

Lyophilized hK5His samples were resuspended ln water:5% acetonitrile

(ACN):O.l % trifluoroacetic acid (TFA), desalted using reverse phase C18 ZipTips

(Millipore, Billerica, MA), eluted with water:70% ACN :0.1 % TF A. The eluate was

spotted onto a matrix assisted laser desorption ionization (MALDI) sample plate and

analyzed using a MALDI-Time of Flight (ToF) Voyager-DE (Applied Biosystems, Foster

City, CA). The eluted fraction was also spotted onto polybrene coated TF A filter disks

for Edman degradation-based N-terminal sequence analysis using a Pro cise 492

automated microsequencer (Applied Biosystems). One eluate aliquot was treated with

10mM dithiothreitol for lhr, diluted with 80lll of 50mM Ambic containing lOOng of

Trypsin gold (Promega, Madison, WI) and digested ovemight. For dual enzyme digest,

the eluted fraction was first digested ovemight with immobilized trypsin (Pierce,

Rockford, IL) followed by an ovemight digestion with lOOng of Endoproteinase-Asp-N

(Roche Diagnostics, Laval, Quebec, Canada). Digested solutions were ziptipped and

eluted onto a MALDI sample plate. Peptide fragment analysis was performed with an

Ultima MALDI-Quadrupole ToF mass spectrometer (Waters Limitée, Dorval, Quebec,

70 Canada). Spectra analysis and peptide identity assignment was perfonned using Biolynx, a software tool ofMasslynx v4.0 (Waters).

2.3.4 VSV-G Pseudotyped Retroviral Vector Design, Synthesis and Titer

Assessment.

The bicistronic murine retrovector pIRES-EGFP was previously generated in our laboratory (200) and contains an enhanced green fluorescent protein (EGFP). The hK5His cDNA was ligated into pIRES-EGFP to generate phK5His-IRES-EGFP. VSV-G pseudotyped retroparticles encoding hK5His-IRES-EGFP were generated by tetracycline­ withdrawal as previously published (200, 201). Engineered retroparticles were devoid of replication-competent retro virus as detennined by GFP marker rescue assay utilizing conditioned media from target cens. The titer of the control GFP and hK5His 293GPG single clone population was assessed as previously described (201).

2.3.5 Transduction of Human U87MG Glioma Cells.

Glioma cens were transduced with hK5His-expressing retroviral particles as previously described (200). Stably transduced glioma cens were culture-expanded and sorted to obtain both polyclonal and single clone populations based on GFP expression using a Becton Dickinson FACST AR sorter.

71 2.3.6 Western Blot Analysis.

Conditioned media was collected from confluent control OFP and hK5His transduced U87 glioma ceIls, as weIl as stably transfected hK5His 293T cells (positive control), concentrated and detected by anti-His immunoblot analysis as previously mentioned.

2.3.7 Migration Assays.

4 4 HUVEC (lxl0 ) or human macrophages (5x10 ) were plated onto 0.15% gelatinlPBS-coated 8-flm pore chemotaxis membranes (Coming, Acton, MA) within

Boyden chamber inserts. Migration assays were performed as previously described (202).

HUVEC and macrophages were exposed to conditioned media from control OFP and hK5His transduced U87 glioma cells for 4 hours and 24 hours respectively under growth factor-stimulated conditions (EBM-2 SingleQuots) and serum-stimulation respectively.

Each sample was tested in triplicate and the average number of migrating cells per field was assessed by counting 3 random high power fields per filter.

2.3.8 In Vivo Matrigel™ Assay and Analysis of Cellular InnItrate by Cytometry.

Culture-expanded U87 -OFP and U87 -hK5His-OFP cells were aliquoted to create

50fll cell suspensions containing 2xl04 ceIls, mixed with 500fll Matrigel™ (BD

Biosciences, San Jose, CA) at 4°C and implanted subcutaneously in the right lateral flank of 6-week old immunodeficient NOD-SCID mice (Charles River, Laprairie, Quebec,

Canada). Four weeks post-implantation, mi ce were sacrificed, implants excised and

72 processed as previously described (203). To detennine the expIant cellular infiltrate, cell suspensions were stained with the following antibodies: purified rat anti-mouse

CD16/CD32 (mouse Fc block) followed by APC-conjugated rat anti-mouse CD45, PE­ conjugated rat anti-mouse Mac-3, biotin-conjugated rat anti-mouse Ly-6G (Gr-l) and their corresponding isotypic controls. Biotinylated antibodies were revealed using PE­

Cy7-streptavidin (BD Phanningen, San Diego, CA). Cells were fixed with 1% parafonnaldehyde and events were acquired using a FACSCalibur flow cytometer

(Beckman Coulter, Fullerton, CA) and analyzed using the CellQuest software (BD).

2.3.9 Matrigel™ Expiant Histochemistry.

Explants were fixed in fonnalin, embedded in paraffin and 4 J.Lm sections were prepared in order to generate representative sections of the border and central regions of the explants. Sections were treated with xylene, rehydrated and antigen retrieval was perfonned using two microwave boils in a 10 mM sodium citrate buffer (pH 6.0) solution. Endogenous biotin activity was blocked using a kit (Zymed Laboratories,

Markham, Ontario), sections were subsequently blocked with 2.5% bovine serum in PBS and incubated with a rabbit polyc1onal raised against murine von

Willebrand factor (vWF; N eomarkers, Fremont, CA dilution 1: 100) ovemight. After three washes, the sections were incubated with biotinylated goat anti-rabbit IgG antibody

(BD Phanningen, dilution 1:200) for 2 hours, washed and incubated with streptavidin­ peroxidase (VectorLabs, Burlingame, CA) for 1 hour prior to the addition of 3,3'­ diaminobenzidine (DAB) chromogenic substrate (VectorLabs). Meyer's hematoxylin was used for counterstaining. The blood vessel surface area of all vWF-positive vessels was

73 calculated using a Leica light microscope ocular micrometer (Lei ca Microsystems Inc.,

Richmond Hill, Ontario, Canada) at 400X magnification. Two observers counted 3 random sections per expIant in each experimental group. The blood vessel surface area to total section surface area was calculated and expressed as a percent age.

2.3.10 Stereotactic Intracerebral Surgery.

CDI nu/nu female 6-week old athymic nude mlce (Charles River) were anesthetised usmg a ketamine-xylezine-saline-acepromazine cocktail (1 OOmglml,

20mglml, 0.9%, IOmglml respectively) dosed at 1OOIll/I OOg by i.p. injection. Mice were secured in stereotactic apparatus (KopfInstruments, Tujunga, CA) and incisor bar set to -

1.5. Midline incision made on scalp to expose Bregma & Lambda. Coordinates from

Bregma were: AP: +0.5, LM: -2.0, DV: -4.4. A bUIT hole was made and a Hamilton syringe (IOIlI, 26 gauge needle, Hamilton Co., Reno, NV) was gently lowered. Cell suspension (IxIOs in 3111 ofHBSS) was injected at a rate ofO.251l1 per 3 minutes for 36 minutes.

2.3.11 Brain Tissue Analysis.

After euthanasia, brains were processed for H&E staining as previously described

(204). Digital images were retrieved using an Olympus microscope (Olympus America,

Melville, NY). Tumor volumes were calculated using the formula length x width2 x 0.4.

74 2.4RESULTS

2.4.1 MALDI-QToF Mass Spectrometry Analysis of Expressed Soluble hK5His

Peptide Confirms Predicted Disulfide Bridging Conformation.

The 381bp human Plg K5 domain cDNA, corresponding to amino acid residues

447 to 546 of PIg, includes a 20 amino acid IL-2 signal sequence and a His-tag at the C terminus (Figure 2.lB), encodes for the hK5His protein with a kringle structure as predicted by previously published NMR structure study (205). Affinity chromatography was performed on pre-cleared conditioned media from stably transfected hK5His­ expressing 293T cells. Anti-His immunoblot, following SDS-PAGE separation (Figure

2.2A, upper), demonstrates a major band of 15kDa (Figure 2.2A, lower). N-terminal amino acid sequencing indicated proper cleavage of the IL-2 signal sequence directly upstream of Ser20 and analysis by MALDI-QToF mass spectrometry revealed a major peak at 12108Da consistent with the predicted molecular weight of hK5His (Figure

2.2B). To fully characterize cystinyl bridge structure and confirm kringle domain tertiary structure, an orthogonal digestion was carried out either with agarose-immobilized trypsin alone (Figure 2.2C) or with trypsin followed by incubation with Asp-N endopeptidase (Figure 2.2D) under reducing and non-reducing conditions. Peptide mapping of non-reduced single and dual protease hK5His digests by MALDI-QToF mass spectrometry confirmed the presence of disulfide bridges between Cys 17 -Cys 96, Cys38_

Cys79, and Cys67_Cys91 (by deduction). These findings demonstrate that Plg hK5His domain spontaneously adopts a tertiary structure consistent with native conformation as displayed within Plg when expressed by genetically-engineered eukaryotic cells.

75 Figure 2.1. Schematic illustration of human Plg and its kringle domains. (A) Plg is composed of 791 amino acids with a single N-linked glycosylation at Asn289 and a single

346 O-linked glycosylation at Thr . Angiostatin, a weIl known inhibitor of angiogenesis, encompasses the first three to four kringle structures of Plg. This is a modified schematic courtesy of Dr. M. R. Llimis and co-workers, Carnegie Mellon University, Pittsburgh, PA

(www.chem.cmu.edu/groups/Llinas/res/structure/hpk.html). (B) Human kring1e 5 (hK5) consists of the last cryptic fragment of Plg (Cys462 _Cys541) and is composed of 80 amino acid residues with 3 distinct disulfide bonds. FIGURE 2.1

A

B Human Knng• 1e 5 (h K5His) -~t@:]~5A&~'\E(H~

IL-2 Signal Sequence

COOH Figure 2.2. Proteomic analysis of purified soluble hK5His. (A, upper) Silver stain perfonned on pooled eluted fractions detects purified protein at the expected molecular weight of 15kDa. (A, lower) Anti-His immunoblot analysis demonstrates the presence of soluble hK5His peptide in pooled eluted fractions. (B) Analysis of the purified hK5His peptide by MALDI-QToF mass spectrometry revealed a major peak at 12108Da consistent with the theoretical molecular weight of monomeric K5 at 12109.6Da. The doubly charged fonn and dimer are detectable at 6059Da and 24195Da respectively. (C, lelt) Expected peptide fragment of interest generated for hK5His protein digested with trypsin alone under oxidizing conditions. (C, right) MALDI-QToF mass spectra confinning Cys38:79 disulfide bond within hK5His. (D, left) Expected peptide fragment of interest generated for hK5His protein digested with trypsin followed by Asp-N endopeptidase treatment under oxidizing conditions. (D, right) Spectra confinning

Cys17:96 and Cys38:79 disulfide bonds. FIGURE 2.2

A B ,~ Monomer

30

25

15 10 Doubly Charged J

30 Dimer 25 15 10

C Trypsin Digestion Residue , Sequence lM. Hl Disulfide Bridge 100. OXIDlZEO ",.!li", 30-48 A~GTPCaDWAAQEPHR 3914.13 (M.H) 391473 ~-C79 69-85 NPDGDVGG~C~

% . 'l'OO=,u o 3800 3900• mil

100 OXlOIZED 100 2441.09 (M+H]

16-23 DTMFGNGK 1902.78 92-101 DVPQCAAPSF % ,Jill< 1902.78IM+H]

30-39 A~GTPCQ 24.41.09 o 0 73-85 100 'llREDUCED 11l % ,.. % 'f o 1900 ml. 02400 ml. 2.4.2 hK5His Protein Secreted by RetroviraUy Gene-Modified Human Glioma CeUs

Inhibits In Vitro Endothelial CeU Migration.

The hK5His cDNA was cloned into our previously published (200) bicistronic

retro viral vector (Figure 2.3A). The hK5His retrovector plasmid was stably transfected

into 293GPG retro viral packaging cells and retro virus producer cells were selected as

described in Materials and Methods. Tetracycline-withdrawal from the culture media led

to the production of VSVG-typed hK5His-GFP retroviral particles which were

subsequently concentrated 100-fold by ultracentrifugation and a viral titer of -2.5xI06

infectious particles/mL was obtained (data not shown). Concentrated VSVG-typed

hK5His-GFP retroviral particles were utilized to transduce human glioma cells. The

U87MG human glioblastoma cell line was selected for this study because it

overexpresses vascular endothelial growth factor (VEGF) producing tumors in vivo that

are characterized by increased vascularity, lack of necrosis and absence of peripheral

invasion in early tumor development (49). Following retro viral transduction, single

U87MG clones were selected and assessed for GFP expression by flow cytometry. The

U87 single clone used for the study was sorted based on GFP expression and was 99%

GFP positive (Figure 2.3B). To ensure hK5His transgene expression and proper secretion,

anti-His immunoblot analysis was performed on conditioned supernatant collected from hK5His transduced human U87 glioma cells and detects a major 15kDa protein consistent

with the predicted molecular weight of soluble hK5His (Figure 2.3C). Using a semi­ quantitative western blot titration curve, it was estimated that hK5His-GFP-expressing

U87 cells secrete 0.5pM (or 7.5ng) of soluble hK5His per 106 cells per 24 hours (data not shown). To test the functionality of secreted hK5His from hK5His-GFP-expressing U87

80 Figure 2.3. Development and characterization of hK5His-GFP gene-modified cells.

(A) Schematic illustration of the human hK5His-lRES-EGFP retroviral plasmid construct. The retrovector plRES-EGFP construct served as the control plasmid. (B)

Flow cytometry analysis of untransduced (grey) and hK5His transduced (black) hum an

U87 glioma cells. (C) Anti-His immunoblot analysis reveals functional secretion of hK5His migrating as a doublet at ~ 15kDa from retrovirally gene-modified U87 glioma cells and stably transfected 293T cells (positive control). (D) Conditioned media from hK5His transduced glioma cells inhibits HUVEC migration up to 75 ± 3%; bars, S.E.M.

Statistical analysis was performed using the student (-test. FIGURE 2.3

B'

.!!l c ::> o U K5 Retrovector 7076bps

c US7 U87-GFP ..-15kOa o 100 --f "P< 0.0004 oC 80'; 1;&0... CIl i .. 0

fi. 20

------cens, a HUVEC migration assay was perfonned as it has been previously demonstrated that recombinant K5 protein is capable of inhibiting the migration of bovine capillary endothelial (BCE) cens (129). Soluble hK5His present in the conditioned media of hK5His-GFP-expressing U87 cens was capable of inhibiting growth factor-induced migration of HUVEC up to 75 ± 3% (P < 0.0004) as compared to conditioned media from GFP-expressing U87 cens (Figure 2.3D). This confinns the in vitro functionality of hK5His secreted by hK5His-GFP-expressing U87 cens.

2.4.3 hK5His Blocks Glioma Associated Angiogenesis and Possesses Novel Anti­

Macrophage Properties.

To further characterize the mechanism of action of soluble hK5His in vivo, we investigated whether hK5His could interfere with the tumor microenvironment. More specificany, we assessed whether soluble hK5His could affect the recruitment of host­ derived tumor-infiltrating endothelial as wen as inflammatory cens implicated in tumor

4 angiogenesis. To test this hypothesis, 2xl0 U87-GFP (n = 10) or U87-hK5His-GFP (n =

10) cens were Matrigel™ -embedded and implanted subcutaneously in NOD-SCID mice.

To verify if tumor-secreted hK5His induced an anti-angiogenic effect in vivo, explants (n

= 4) in each experimental group were retrieved 4 weeks post-implantation (Figure 2.4A) and stained with von Willebrand factor (vWF) antibody (Figure 2.4B). The section surface area occupied by vWF-positive blood vessels to total section surface area ratio was ca1culated and expressed as a percentage (Figure 2.4C). Our results demonstrate that soluble hK5His domain is capable of exerting a potent anti-angiogenic effect (P = 0.04)

83 Figure 2.4. hK5His acts as a potent anti-angiogenic agent. (A) Implant (arrow) retrieval 4 weeks post-implantation of 2 x 104 Matrigel™ -embedded U87-GFP or U87- hK5His-GFP cells (representative images are shown at the same magnification). (B)

Representative images of vWF-positive immunostained blood vessels (red arrows) from two different U87-GFP-containing (a, b) and U87-hK5His-GFP-containing (c, d) implants; magnification 200X. (C) The vWF-positive blood vessel surface area to total section surface area ratio is plotted as a percentage; bars, S.E.M. (S.E.M. for Matrigel™• only group was based on 3 random section counts from 1 animal since remaining implants resorbed completely). Statistical analysis was performed using the student t-test. FIGURE 2.4

A

U87-GFP U87-K5-GFP B .« < , • .:;:J<,~,4 }~:\.: .» .,: ::~~jl~~Q~i

0.9

C 0.8 o Matrigel™ Only GJ U87-GFP 0.7 • U87-K5-GFP * p= 0.04 ca 0.6 CU '- 0.5 ct > 0.4 al 0.3 ~0 0.2

0.1

0 in vivo when secreted by retrovirally-engineered glioma cells. The remaining explants (n

= 6) in each experimental group were dissolved in a collagenase solution and single-cell

suspensions were generated and counted. (Figure 2.5A, left). The hK5His-GFP explants

contained about two thirds less cells than the control GFP explants (P = 0.01). The

cellular infiltrate in the explants was analyzed for GFP expression as well as for cell

surface marker expression to distinguish the various cell types present within the explants

(Figure 2.5A, right). The GFP positive cell fraction representing glioma cells, is

significantly reduced in the hK5His-GFP explants as compared to the control GFP

explants (P = 0.03). It is important to note that the difference in absolute tumor cell

number is not due to differences in the proliferation rates of the control GFP and hK5His­

GFP glioma cells, since both possess equivalent proliferation rates in vitro (Figure 2.6).

Although immunocompromised NOD-SCID mice possess functional macrophages and

granulocytes, they lack functional T and B lymphocytes. Flow cytometry analysis after

three-color staining indicated that a substantial part of the cellular infiltrate consisted of

CD45-positive hematopoietic cells (Figure 2.5B, left). More specifically, there was a

reduced absolute number ofrecruited CD45+Mac+Grl+ (P = 0.04) cells, which represents

the monocyte/macrophage fraction and CD45+Mac3+Grl- (P = 0.03) cells, which

represents the macrophages present in the U87-hK5His-GFP-containing explants as

compared to the control U87-GFP-containing explants (Figure 2.5B). There was no

significant difference in the number of CD45+Mac3-Grl + (P = 0.43) cells, representing

the infiltrated granulocytes present in the explants.

In order to confirm our in vivo observations, we tested whether hK5His could

,~ inhibit macrophage migration in vitro. A migration assay was performed using human

86 Figure 2.5. hK5His possesses novel anti-macrophage property and decreases migration potential of human macrophages. (A) Lejt, Absolute total cell count performed post-collagenase digestion of the explants. Right, Absolute number of GFP positive cells was enumerated using flow cytometry and represents the GFP-marked glioma ceIls; bars, S.E.M. (B) Absolute number of infiltrated CD45+ hematopoietic ceIls,

CD4S+Mac3+Grl + monocytes/macrophages, CD4S+Mac3+Grr macrophages and

CD4S+Mac3-Grl + granulocytes in the explants; bars, S.E.M. Statistical analysis was performed using the student (-test. (C) Macrophages were exposed to various conditioned media for 24 hoUTs under serum stimulation. Basal invasion represents migrating macrophages under serum-free conditions; bars, S.E.M. Insets, Representative stains of migrated cells from each experimental group; magnification 200X. Statistical analysis was performed using the student t-test. FIGURE 2.5

14 A **p= 0.01 w- [[] U87-GFP .....012 *P= 0.03 X • U87 -K5-GFP -=10 Q) ,.Q 8 E ::J Z 6

Q) 4 0 S 2 0 t- 0 * Total Cell Count 4 B *** p= 0.03 ~3.5 ~ ..... u P=O.04 U87-GFP • U87-K5-GFP ~ 3· * P= 0.43 '- .82.5 E z::J 2 =ai 1.5 0 1 cu ;§0.5 0 Mac3+Gr1+ Mac3+Gr1" Mac3"Gr1+ 1 C CD45+

350 .==;======

11. 300 a. 250 ~ * P= 0.007 Ci) 200 o "C 150 .! E 100 Cl) !Ë 50 o Supernatant Source Figure 2.6. U87-GFP and U87-hKSHis-GFP cell populations possess equivalent proliferation rates in vitro. FIGURE 2.6

10000 ,------,

M -0 -x 1000 -c .2 -O-U87-GFP ta 100 '- ___ U87 -K5-GFP ~ 0 '- a.. 10 C) 0 ...J

o 24 48 72 96 144 1 68 Time (hrs) monocytes isolated from peripheral blood mononuclear cells that were stimulated with

GM-CSF to differentiate into CD206+ macrophages (data not shown). The ability of

macrophages to migrate through a gelatin matrix was assessed under serum-stimulated

conditions, where macrophages were directly exposed to conditioned media collected

from GFP-expressing glioma cells and hK5His-GFP-expressing glioma cells. The results

indicate that soluble hK5His decreases the migration potential of human macrophages by

approximately 60% as compared to the GFP control (P = 0.007, Figure 2.5C).

2.4.4 Glioma Targeted hK5His Expression Suppresses Tumor Growth and Prolongs

Survival in an Experimental Orthotopic Brain Cancer Model.

In order to test the efficacy of retrovirally-engineered glioma cells secreting

5 hK5His in a clinically relevant proof-of-principle experiment, 10 U87-GFP (n = 5) and

U87-hK5His-GFP cells derived from either a single clone (n = 5) and an independently

generated polyclonal population (n = 5) were implanted intracerebrally in athymic nu/nu

(nude) mice. Mice were sacrificed 32 days post-implantation, brains were removed and

processed for H&E staining (Figure 2.7A). Tumor volumes from each experimental group

were estimated from H&E-stained cryostat sections and calculated as described in

Materials & Methods. Consistent with our previous study (204), brain tumors in the GFP

control group had extensive growth. In contrast, the U87-hK5His-GFP-implanted mice

possessed significantly reduced mean tumor volumes (0.35 ± 0.18mm3 for the single

clone population and 0.07 ± 0.03mm3 for the polyclonal population) as compared to the

3 ~- control U87-GFP-implanted (12.23 ± 6.07mm ) mice (P = 0.04) (Figure 2.7B).

91 Figure 2.7. Suppression of intracerebral U87MG glioma growth by soluble hKSHis.

(A) H&E-stained brain tissue sections from control U87-GFP implanted mice (n = 5) and

U87-hK5His-GFP implanted mice (n = 5) 32 days post-implantation. Representative stains are demonstrated at the same magnification, 40X. (D) Tumor volumes were estimated from post-mortem H&E-stained brain sections obtained from mice implanted with U87-GFP, U87-hK5His-GFP single clone (sc) and U87-hK5His-GFP polyclonal

(pc) cells. The experiment was performed twice with similar results; bars, S.E.M.,

Student (-test performed. (C) Kaplan-Meier long-term survival curve of mice implanted with control U87 -GFP (n = 10) and U87 -hK5His-GFP (n = 15) cells. Log-rank statistical test performed. FIGURE 2.7

A

~ - C _...... ~·_·· __ ·_.. _-_·_-_·-~--~l 100 1 --U81-GP P <: 0.0001 1 --U81~P l'1 ~so d 'S: 5&0 (f) 1:40 ~ ;200 1 Q.. 0 t 0 125 175 225 275 Days We perfonned a long-tenn in vivo study, where we evaluated the survival of the therapeutic cohort implanted with U87-hK5His-GFP cells (n = 15) and the control U87-

GFP-implanted (n = la) mice. Although all control mi ce succumbed to excessive tumor burden by ~8 weeks, the U87-hK5His-GFP-implanted mice possessed a clear survival advantage with 53% ofthe mice surviving over 120 days (P <0.0001) (Figure 2.7C).

2.5 DISCUSSION

This present study demonstrates that eukaryotically-expressed secreted hK5His protein preserves the prototypical native kringle disulfide bridging confonnation deemed essential for the angiostatic activity of plasminogen-derived peptide domains. lndeed, our cysteine bridging pattern data confinns the results of the only other study which characterizes the structure of recombinant K5 domain using NMR technology (205). We have shown that VSV -G pseudotyped retro viral vectors engineered to express human Plg hK5His domain can efficiently gene-modify human U87 glioma cells ex vivo. We also demonstrate that hK5His-engineered U87 cells can secrete biologically functional soluble hK5His protein capable of inhibiting growth factor-stimulated endothelial cell migration in vitro.

Moreover, our in vivo data in immunodeficient NOD-SCID mice suggests that hK5His secreted by Matrigel™-embedded U87 engineered cells acts as a potent anti­ angiogenic agent by significantly decreasing blood vessel fonnation as well as a novel anti-macrophage agent by significantly preventing the recruitment ofhost-derived tumor-

94 associated CD45+Mac3+Grl +/- monocytes/macrophages within the local tumor microenvironment. The involvement of tumor-associated macrophages in angiogenesis and tumor progression is a well established concept (206-208). Activated macrophages pro duce tumor necrosis factor-u and IL-l, inducing the production of VEGF and IL-8 in glioma cells, which further attracts and activates macrophages (209). Macrophage infiltration has been closely correlated with neovascularization and observed more frequently in grade IV glioblastomas than in grade II or III gliomas (209).

It may be speculated that the potent anti-angiogenic properties of soluble K5 mediating a reduction in tumor-associated endothelial cells, as evidenced by the lower surface area occupied by blood vesse1s in the U87-hK5His-GFP-containing explants, is partially responsible for the decreased infiltration ofhost-derived macrophages within the tumor area. We also demonstrate that soluble hK5His independently reduces macrophage migration in vitro, suggesting that the K5 domain can impart a direct inhibitory effect on macrophage motility. To date, the K5 peptide has been consistently characterized as an endothelial cell-specific inhibitor (128, 129) and Gonzalez-Gronow et al. (138) suggest that K5 mediates its effects by binding to the human voltage-dependent anion channel, which is a receptor present on human endothelial cells. Unlike K5, the mechanism of action of angiostatin has been studied more extensively and its endothelial cell binding sites include the cell surface-associated ATP synthase receptor (62), angiomotin (64) and integrin u v J33 (119). Although angiostatin mediates its inhibitory action primarily on endothe1ial cells, it has been reported by Moulton et al. (210) that it reduces plaque angiogenesis and atherosclerosis by reducing plaque-associated macrophages. However,

Moulton et al. suggest that angiostatin, at doses that reduce endothelial cell sprouting,

95 does not directly inhibit monocytes since it do es not inhibit VEGF-induced monocyte migration in vitro. Thus, our findings provide novel mechanistic insight and suggest that soluble hK5His protein inhibits two unique cell types most critical for tumor angiogenesis, mediating a dual role as an anti-angiogenic agent via its anti-endothelial cell properties as weIl as an anti-tumor agent by reducing the infiltration of host-derived macrophages within the tumor microenvironment.

We utilized a clinically relevant orthotopic implantation approach to test whether soluble hK5His could suppress glioma progression post-implantation of hK5His-GFP- expressing U87 cells in athymic nu/nu mice. Stereotactic implantation procedures are particularly suitable for glioma gene therapy approaches allowing for constant long- lasting transgene expression within the local tumor microenvironment and minimize the risk of toxicity as is often encountered with systemic therapies. Our data successfully demonstrates that soluble hK5His significantly decreases tumor growth by approximately

97% as compared to controls. In doing so, soluble hK5His also promotes long-term survival, with 53% of U87-hK5His-GFP-implanted mice surviving past 120 days. We postulate that upon stereotactic implantation of our ex vivo retrovirally-engineered hK5His glioma ceIls, stable integration and sustained expression of the hK5His transgene

~ in target U87 cells provided continuous release of soluble hK5His within the loco- regional tumor microenvironment, resulting in potent suppression of tumor growth. We speculate that tumor-targeted delivery of hK5His via viral vectors such as pseudotyped retrovectors (200) and other gene delivery platforms would lead to similar results.

96 In summary, we pro vide compelling evidence that secreted hK5His protein expressed by engineered glioma cells possesses potent anti-angiogenic as well as novel anti-macrophage properties which block glioma progression and promote survival in an orthotopic experimental brain cancer model. Further studies on the pleiotropic effects of

K5 peptide on host-derived angiogenic and macrophage response may offer new avenues of exploration for the development of therapeutic strategies for the treatment of glioma and other loco-regional refractory malignancies such as prostate and ovarian cancer.

2.6 FINANCIAL SUPPORT

S. R. Perri was a recipient of a Canadian Prostate Cancer Research Initiative­

Canadian Institutes of Health Research (CIHR) Doctoral Award and is currently a recipient of the D.S. Army Medical Research and Materiel Command Breast Cancer

Research Predoctoral Traineeship, Award, DAMDI7-03-1-0545. J. Galipeau is a recipient of the CIHR Clinician-Scientist Award (MOP-15017). J. Nalbantoglu is a

FRSQ-FCAR and Killam Scholar as well as a recipient of a National Cancer Institute of

Canada grant and a Valorisation-Recherche Québec grant.

97 CHAPTER3

98 CHAPTER3

Plasminogen Kringle 5 Blocks Tumor Progression by Anti-Angiogenic and

Pro-Inflammatory Pathways.

Reference: Perri SR, Martineau D, Francois M, Lejeune L, Bisson L, Durocher Y, Galipeau J. Plasminogen Kringle 5 Blocks Tumor Progression by Anti-Angiogenic and Pro-Inflammatory Pathways. Molecular Cancer Therapeutics. 2007;6(2): 1-9. (Figure se!ected as Caver Art).

Preface: In Chapter 2, we demonstrated that soluble human plasminogen kringle 5 (K5) protein produced by gene-modified human glioma cells possesses potent anti-angiogenic as well as novel macrophage inhibitory properties when tested in immunodeficient mice. In order to determine whether soluble K5 protein modulates other immune cellular effectors, we tested the efficacy of soluble K5 protein in a munne mammary adenocarcinoma model using mice with intact immune systems.

99 3.1 ABSTRACT

Proteolytic processing of human plasminogen generates potent anti-angiogenic peptides such as angiostatin. The plasminogen kringle 5 (K5) domain - which is distinct from angiostatin - possesses potent anti-angiogenic properties on its own which can be exploited in cancer therapy. It has been recently observed that anti-angiogenic agents promote leukocyte-vessel wall interaction as part of their anti-tumor effect. Though we have previously shown that K5 suppresses cancer growth in tumor xenograft models, its modulation of inflammation in experimental mice with intact immune systems is unknown. To determine whether K5 possesses immune pro-inflammatory properties, we investigated the effects of K5 in an immune competent model of breast cancer and observed that tumor rejection is substantially reduced in NOD-SCID and BALB/c nude when compared to wild-type BALB/c mice, suggesting an important role for T -lymphoid cells in the anti-tumor effect of K5. Tumor expIant analysis demonstrates that K5 enhances tumor recruitment of CD3+ lymphoid cells, in particular the NKT phenotype.

We also observed a significant decrease in tumor-associated microvessel length and density consistent with anti-angiogenic activity. Histological analysis of K5 tumors also revealed a robust neutrophilic infiltrate, which may be explained by the neutrophil chemotactic activity of K5 as well as its ability to promote CD64 upregulation within the

CD Il b + adhesive neutrophil population. In sum, our findings confirm that the K5 protein acts as a potent angiostatic agent and possesses a novel pro-inflammatory role via its ability to recruit tumor-associated neutrophils and NKT -lymphocytes, leading to a potent anti-tumor response.

100 3.2 INTRODUCTION

Proteolytic cleavage products of plasminogen - as weIl as individual plasminogen kringle domains such as kringle 5 - possess anti-angiogenic properties whose use in cancer therapy is under much scrutiny (60, 86, 88, 128, 205, 211). It has been recently proposed that an array of angiostatic agents - including plasminogen derivatives like angiostatin - can significantly stimulate leukocyte-vessel wall interactions in vivo by the up-regulation of endothelial adhesion molecules in tumor vessels (212) and may enhance anticancer immune response and allow the immune system to overcome tumor immune resistance (213). However, in the setting of experimental peritonitis, angiostatin has also been shown to behave as an antiadhesive/anti-inflammatory substance (214). The sum of these observations readily supports the anti-angiogenic properties of angiostatin in the setting of pathological neovascularization, but its immune modulatory effects remain controversial. The fifth kringle domain of plasminogen (hereafter K5), when utilized on its own displays robust anti-angiogenic properties (86, 128, 211) and will also directly lead to apoptosis of anoxie tumor cells in vitro (148). These features make K5 an appealing anti-tumor biopharmaceutical with combined anti-tumoral as weIl as anti­ angiogenic properties. Indeed, we have previously demonstrated that genetic engineering of tumor cells for expression of K5 abolishes tumor growth in vivo, potently suppresses cancer-associated angiogenesis and will inhibit recruitment of tumor-associated macrophages in an immune deficient human tumor xenograft model (169). The latter observation suggests that K5 may possess immune modulatory properties on blood­ derived immune competent cells - in keeping with the observations made by others with angiostatin. To address the question of K5 influence on immunity, we compared the

101 biology of K5-engineered breast tumor cells in immune defective mice as well as in tumor MHC-matched mice with normal immune systems. As expected, we find that K5 is a potent inhibitor of cancer-associated angiogenesis. However, we also observed that the anti-tumor effect of K5 is utterly dependent on an intact immune system, in particular on the T -lymphocyte subset, and that recruitment of polymorphonuclear cells is a dominant feature of K5 tumors. These observations provide new evidence that the K5 plasminogen derivative possesses anti-angiogenic as well immune stimulatory anti-tumor properties.

3.3 MATERIALS AND METHODS

3.3.1 CeU Culture Reagents.

293GPG pantropic retrovirus-packaging cell line (198) was a gift from Dr.

Richard C. Mulligan (Children's Hospital, Boston, MA). Murine mammary adenocarcinoma DN3 cell line (215) was a gift from Dr. Moulay Alaoui-Jamali (Lady

Davis Institute, Montreal, Quebec, Canada).

3.3.2 VSV -G Pseudotyped Retroviral Vector Design, Synthesis and Titer

Assessment.

The bicistronic murine retrovector pIRES-EGFP was previously generated in our laboratory (200) and contains an enhanced green fluorescent protein (EGFP). The hK5His cDNA was ligated into pIRES-EGFP to generate phK5His-IRES-EGFP. The hK5 cDNA was histidine-tagged for protein detection and purification purposes, since there is no commercially available anti-human K5 antibody. VSV-G pseudotyped

102 retroparticles encoding hK5His-IRES-EGFP were generated by tetracycline-withdrawal as previously pub li shed (200, 201). Engineered retroparticles were devoid of replication­ competent retrovirus as determined by GFP marker rescue assay utilizing conditioned media from target cens. The titer of the control GFP and hK5His 293GPG single clone population was assessed as previously described (201).

3.3.3 Transduction of Murine Mammary DAl3 Cells.

DAl3 cens were transduced with hK5His-expressing retroviral particles as previously described (200). Stably transduced DAl3 cens were culture-expanded and sorted to obtain polyclonal populations based on GFP expression using a Becton

Dickinson F ACST AR sorter.

3.3.4 Western Blot Analysis.

Conditioned media was conected from confluent DAl3-GFP control and hK5His­

GFP transduced DAl3 cens, concentrated and detected by anti-His immunoblot analysis as previously described (169).

3.3.5 In Vivo Matrigel™ Assay and Analysis of Cellular InÏlltrate by Cytometry.

Culture-expanded DAl3-GFP and DAl3-hK5His-GFP cens were aliquoted to create 50).11 ceIl suspensions containing 106 ceIls, mixed with 500).11 Matrigel™ (a semi­ solid matrix derived from a murine sarcoma ceIlline, BD Biosciences, San Jose, CA) at

4°C and implanted subcutaneously in the right lateral flank of 6-week old immunocompetent BALB/c mi ce (Jackson Laboratories, Quebec, Canada). Three days

103 post-implantation mice were sacrificed, implants excised and processed as previously described (203). To determine the expIant cellular infiltrate, cell suspensions were stained with the following antibodies: purified rat anti-mouse CD16/CD32 (mouse Fc block) followed by APC-conjugated rat anti-mouse CD45, APC-conjugated rat anti-mouse CD3,

PE-conjugated rat anti-mouse CD4, PE-Cy7-conjugated rat anti-mouse CD25, FITC­ conjugated rat anti-mouse CD8, PE-conjugated rat anti-mouse NK-T/NK and their corresponding isotypic controls. AlI antibodies were purchased from BD Pharmingen,

San Diego, CA. Cells were fixed with 1% paraformaldehyde and events were acquired using a FACSCalibur flow cytometer (Beckman Coulter, Fullerton, CA) and analyzed using the CellQuest software (BD).

3.3.6 Matrigel™ Expiant Histochemistry.

Mice were sacrificed, implants excised, fixed in formalin, embedded in paraffin and 41lm sections were prepared in order to generate representative sections of the border and central regions of the explants. Sections were either stained with hematoxylin and eosin or underwent an antigen retrieval pro cess using two microwave boils in a 1OmM sodium citrate buffer (pH 6.0) solution. Endogenous biotin activity was blocked using a kit (Zymed Laboratories, Markham, Ontario), sections were subsequently blocked with

2.5% bovine in PBS and incubated with a rabbit polyclonal raised against murine von Willebrand factor (vWF; Neomarkers, Fremont, CA dilution 1: 100) ovemight. After three washes, the sections were incubated with biotinylated goat anti­ rab bit IgG antibody (BD Pharmingen, dilution 1:200) for 2 hours, washed and incubated with streptavidin-peroxidase (VectorLabs, Burlingame, CA) for 1 hour prior to the

104 addition of 3,3'-diaminobenzidine (DAB) chromogenic substrate (VectorLabs). Meyer's hematoxylin was used for counterstaining. The microvessel length was measured on thirty randomly selected vWF-positive vessels from each experimental group using an ocular and stage micrometer and the microvessel density was calculated by counting aIl vWF-positive vessels on 3 randomly selected sections from each group using a Leica light microscope (Leica Microsystems Inc., Richmond Hill, Ontario, Canada) at 400X magnification and dividing by the total section surface area. The mean number of vWF­ positive blood vessels per mm2 was plotted.

3.3.7 Production of Purified Kringle 5 Protein.

Human embryonic kidney 293 cells were transiently transfected in a large-scale system with a K5-containing expression vector, conditioned media was collected 72 hours post-transfection, concentrated and purified using an immobilized metal affinity chromatography system (Novagen, La Jolla, CA). Eluted protein fractions were analyzed by anti-His immunoblot. Semi-quantitative anti-His immunoblot was performed to quantify the amount of K5 protein (data not shown).

3.3.8 Neutrophil Isolation.

Heparinized whole blood (50mL) was collected from a healthy human donor, layered over a Ficoll-density gradient and centrifuged at 1600rpm for 30 minutes. The polymorphonuclear neutrophil (PMN)-rich fraction was collected and resuspended in

0.9% NaCl. Red blood cells (RBCs) were lysed by exposure to cold 0.2% NaCI for 30 seconds, followed by neutralization with cold 1.6% NaCI and spun at 1000rpm for 5 minutes. Lysis steps were repeated until RBC fraction was removed. Calcein AM

105 (Molecular Probes, Eugene, OR) (lOmg/mL) was added to a 5mL PMN suspension in

Hank's Balanced Salt Solution (HBSS) and cells were incubated for 30 minutes at 37°C.

PMN were then washed twice with PBS, counted and resuspended in OPTI-MEM to the

desired concentration.

3.3.9 Fluorescence-Based PMN Migration Assay.

A 96-well 3f..lm pore size chemotaxis chamber (ChemoTx, Neuro Probe,

Gaithersburg, MD) was utilized and the migration assay was performed as described by

Frevert et al. (216). Briefly, the bottom chambers were filled with 29f..l1 of either

increasing concentrations of purified hK5His protein (80 to 160ng) diluted in OP TI­

MEM or OPTI-MEM alone as a negative control. In order to determine the total

fluorescence of PMN added to the top si de of the filter, 25f..l1 of each cell suspension (0,

6 0.25, 0.5, 1.0, 2.0 3.0, 5.0 X 10 PMN/mL) was placed directly in duplicate wells in the

bottom chamber. The polycarbonate filters were positioned on the loaded microplate and

4 secured in place with corner pins. PMNs (7.5 x 10 , 50f..l1) were placed directly onto the

filter sites and the chamber was incubated for 7 hours at 37°C. The non-migrating cells

on the top si de of the filter were removed by gently wiping the filter with a tissue. The

chemotaxis filter was placed in a FLUOstar OPTIMA fluorescent plate reader (Fisher

Scientific, Ottawa, Ontario, Canada) and the cells that migrated into the bottom chamber

were measured by using the calcein fluorescence signal. The fluorescent plate reader was

configured in order for the probe to be in a bottom-read position, which allows for

detection of fluorescence in each well of the chemotaxis chamber (excitation, 485nm;

/~ emission, 530nm).

106 3.3.10 Cell Staining.

PMNs were exposed to increasing doses of purified hK5His protein (80, 120 or

160ng) diluted in OPTI-MEM or OPTI-MEM alone as a negative control or phorbal myristate acetate as a positive control for 2 hours at 37°C. PMNs were then co-stained with PE-Iabeled anti-human CDllb antibody and biotin-Iabeled anti-human CD64 antibody. Gene-modified DA/3 cells were stained with biotin-conjugated anti-H-2Kd

(MHC-I) antibody. The biotinylated antibody was revealed using APC-streptavidin.

Antibodies were purchased from BD Pharmingen, San Diego, CA. Cells were fixed with

1% paraformaldehyde and events were acquired using a F ACSCalibur flow cytometer

(Beckman Coulter, Fullerton, CA) and analyzed using the CellQuest software (BD).

3.4 RESULTS

3.4.1 DA/3 mouse mammary tumor cellline retrovirally engineered to express plasminogen K5 domaine

The human K5 histidine-tagged cDNA (hK5His) was cloned as previously described (169) into a bicistronic retroviral vector construct (200) (Figure 3.1A). The hK5His retrovector plasmid was stably transfected into 293GPG retroviral packaging cells and retrovirus producer cells were selected as described in Materials and Methods.

Tetracycline-withdrawal from the culture media led to the production of VSVG-typed hK5His-GFP retro viral particles which were subsequently concentrated 100-fold by ultracentrifugation and a viral titer of ~2.5x106 infectious particles/mL was obtained.

Concentrated VSVG-typed hK5His-GFP retroviral particles were utilized to transduce

107 Figure 3.1. Development and characterization of hKSHis-GFP gene-modified DAi3 cells. (A) Human kringle 5 (hK5His) retrovector when integrated in DAl3 cells. hK5His consists of the last cryptic fragment of plasminogen (Cys462_Cys54I) and is composed of

80 amino acid residues with 3 distinct disulfide bonds. (B) Anti-His immunoblot analysis reveals functional secretion of hK5His migrating at ~ 15kDa from retrovirally gene­ modified DAl3 cells. FIGURE 3.1

A B

r'l' ___~ 5'1 LTR H hK5His-IRES-GFPI-----l LTR 13' kDa Ctl hK5His 20-

1 15- .., the BALB/c-compatible DAl3 murine mammary cancer cell line. The DAl3 munne mammary adenocarcinoma cell line is estrogen independent and serves as a munne model of locally advanced breast cancer (215, 217-219). Following retroviral transduction, polyclonal gene-modified DAl3 cells were assessed for GFP expression by flow cytometry and sorted to obtain a 100% GFP-positive population. To ensure hK5His transgene expression and proper secretion, anti-His immunoblot analysis was performed on conditioned supernatant collected from hK5His transduced murine DAl3 mammary tumor cells and detects a major 15kDa protein consistent with the predicted molecular weight of soluble hK5His (Figure 3.1B). Using a semi-quantitative western blot titration curve, it was estimated that hK5His-GFP-expressing DAl3 cells secrete 0.02pmol/L (or

0.25ng) of soluble hK5His protein per 106 cells per 24 hours (data not shown).

3.4.2 Anti-tumor property of hK5His protein is dependent upon an intact immune system.

We tested the efficacy of hK5His against breast cancer in mice with functional immune systems. One million DAl3 (n = 10), genetically engineered DAl3-GFP (n =

10), or DAl3-hK5His-GFP (n = 10) polyclonal cells were implanted subcutaneously in immune competent BALB/c mi ce and animal survival was monitored over time. One year post-implantation, 65% of hK5His-GFP-implanted mice survived tumor-free (P <

0.0001) as compared to 20% in both DAl3- and DAl3-GFP-implanted control mice

(Figure 3.2A). The anti-tumor effect was severely diminished when 106 genetically

110 Figure 3.2. hK5His-GFP-implanted immunocompetent mice survive long-term and modulate the immune system to suppress tumor growth. Kaplan-Meier long-tenn survival of (A) immunocompetent Balb/c mice implanted subcutaneously with 106 non­ gene-modified DA/3 (n = 10), GFP-expressing DAl3 (n = 10) or hK5His-GFP-expressing

DAl3 cells (n = 10). The experiment was perfonned twice with similar results. Kaplan­

Meier survival of (B) immunodeficient NOD-SCID mice and (C) athymic BALB/c nude

6 mice implanted subcutaneously with 10 GFP-expressing DAl3 (n = 10, n = 6 respectively) and hK5His-GFP-expressing DAl3 cells (n = 10, n = 6 respectively). (D)

Kaplan-Meier survival of immunocompetent BALB/c mice implanted subcutaneously with irradiated 5 x 105 GFP-expressing DAl3 ("GFP", n = 10) or hK5His-GFP­ expressing DAl3 cells ("K5", n = 10) and challenged 14 days later (day 0 on graph) on the opposite flank with 5 x 105 non-irradiated GFP-expressing DAl3 cells. BALB/c mice

("CTL", n = 10) were independently implanted subcutaneously with 5 x 105 non­ irradiated GFP-expressing DAl3 cells on the same day as the challenge as a technical control to demonstrate the tumor growth potential. Log-rank statistical test perfonned for aIl Kaplan-Meier graphs. FIGURE 3.2

A c BALBle BAL Ble Nude 100 100 ~-..---,_--...... - 80 80 CFP K5 60 P

20 Jmtb:rr:c:I:::JI::l::lIIOI:arr:r:r:oc-tXl-I~ Nu Il 20 ni CFP .~ 0 O~~----~~--~~~~ > ~ 0 100 200 300 400 o 20 40 60 80 100 120 ::::3 (1) B o '#- NOD"SCID BAL Ble Challenge 100 100 ~ ___ ~t-r.I 80 CFP 80 60 60 K5 40 P

(RAG)-2 gene, which impairs rearrangement of separate gene elements of the immunoglobulin and T -cell antigen receptor genes and thus disrupts differentiation of both B and T -lymphocyte progenitor cells (220). Although hK5His-GFP-implanted mice succumbed to excessive tumor burden by 2 months post-implantation, their survival was significantly prolonged as compared to control mice (P < 0.0001), possibly due to the angiostatic property of hK5His protein. This observation suggests that the immune system is playing a role in the tumoricidal effects of hK5His protein. We proceeded to test whether the immune modulatory effect of hK5His protein was mediated via T lymphocytes by subcutaneously implanting 106 DA/3-GFP (n = 6) or DA/3-hK5His-GFP

(n = 6) polyclonal cells in athymic BALB/c nude (T -lymphocyte deficient) mice

(Figure 3.2C). Our results demonstrate that T lymphocytes are required for hK5His protein to exert its anti-tumor action, since all hK5His-GFP-implanted mice had to be sacrificed due to tumor development by 3 months post-implantation, albeit delayed survival compared to GFP-implanted control mice (P < 0.0005). In order to assess if hK5His protein elicited an adaptive systemic protective immune response, we

5 subcutaneously implanted wildtype BALB/c mice with irradiated 5 x 10 DA/3-GFP (n =

10) or DA/3-hK5His-GFP (n = 10) cells and 14 days later challenged the mi ce on the opposite flank with 5 x 105 non-irradiated DA/3-GFP cells (Figure 3.2D). Protective

113 immunity was not observed since there was no significant difference between the survival rate of challenged GFP-implanted and hKSHis-GFP-implanted mi ce (P = 0.37).

3.4.3 hK5His protein expression by DAl3 ceUs leads to recruitment of CD3+ lymphocytes in vivo.

To further characterize the immune cellular mediators involved in hKSHis protein

6 anti-tumor effect, we embedded 10 DA/3-GFP (n = 4) or DA/3-hKSHis-GFP (n = 4) polyclonal cells in Matrigel™ and subcutaneously implanted the cells in immunocompetent BALB/c mice. Implants were retrieved 3 days post-implantation and collagenase-digested to obtain a single cell suspension. Cells in each experimental group were counted using a hemacytometer and the cellular infiltrate was analyzed by staining the single cell suspension with 4-color antibodies enabling the identification of different immune subsets by flow cytometry analysis. Our infiltrate analysis confirms that hKSHis protein secreted by gene-engineered DA/3 cells induces a potent host-derived cellular infiltrate including CD4S+ hematopoietic cells (Figure 3.3A). T-Iymphoid subset analysis

(Figure 3.3B), demonstrates a substantial increase in the absolute number of infiltrated

CD3+ lymphocytes (P < O.OOS), in particular CD3+NKT+ cells (P < O.OS).

3.4.4 hK5His protein expression by DAl3 ceUs influences the quantity and sÎZe of tumor-associated microvasculature.

We have previously demonstrated that hKSHis secretion by human glioma cells leads to a profound anti-angiogenic effect in NOD-SCID mice (169). To test whether

114 Figure 3.3. hKSHis possesses novel immunostimulatory property. (A) lefi Absolute total celI count performed post-colIagenase digestion of the explants 3 days post­

6 implantation of 10 GFP-expressing DA/3 (n = 4) or hKSHis-GFP-expressing DAl3 celIs

(n = 4) embedded in Matrigel™; right Absolute number of infiltrated CD4S+ hematopoietic cells; bars, S.E.M. (B) Absolute number of infiltrated lymphoid cells:

CD3+, CD3+CD4+, CD4+CD2S+, CD3+CD8+, CD8+CD2S+, CD3~KT+; bars, S.E.M.

Statistical analysis was performed using the student t-test. FIGURE 3.3

A 10 o DA3-GFP TOTAL * • DA3-K5-GFP 8

6 * P< 0.05 ...-.. (!) 0 4 .,.... * >< 2 ---1-

~ 0 J .. o«oo_.. _««_«««« __ L. __ EB ::l 1.6 o DA3-GFP NS P> 0.05 Z 1.4 • DA3-K5-GFP * P< 0.05 * "* P< 0.005 - 1.2 -Cl) + + 1.0 - Il) + Il) + o ** N CO N t- o o 0 ~ 0.8 U U U Z + + + + 0.6 ...... M CO M o o 0 0 0.4 u U U U

0.2 ...* NS o -00 •• • ~ __ .-==- this effect was replicated in immune competent BALB/c mice, we embedded 106 DAl3 (n

= 4), DAl3-GFP (n = 4) or DAl3-hK5His-GFP (n = 4) polyclonal cens in Matrigel™ and subcutaneously implanted the cens in BALB/c mice. One week post-implantation, the

Matrigel™ plugs were retrieved (Figure 3.4A), sectioned and stained with von

Willebrand factor (vWF) antibody (Figure 3.4B, left panel). The number (Figure 3.4C) as wen as the length (Figure 3.4B, right panel) ofblood vessels was significantly reduced in the hK5His plugs consistent with a potent anti-angiogenic effect in vivo.

3.4.5 hK5His induces neutrophilic tumor infiltration.

In an effort to understand the effect of hK5His on host-derived inflammation, we

6 embedded 10 DAl3 (n = 4), DAl3-GFP (n = 4) or DAl3-hK5His-GFP (n = 4) polyclonal cens in Matrigel™ and subcutaneously implanted the cens in immunocompetent BALB/c mice. Implants were retrieved 3 days post-implantation, sectioned, stained with hematoxylin and eosin (H&E) and analyzed by a veterinary pathologist (co-Author: DM).

The histological analysis revealed that there were almost no intact islet (acini) of tumor cells in the hK5His-containing implants as compared to the GFP-containing implants

(Figure 3.5E, P = 0.0001), where there was a moderate number ofislets, sometimes acini, of intact tumor cens present at the periphery of the implants as seen in Figure 3.5A & c.

The DAl3-hK5His-GFP tumor cells occurred as single cells, rarely in pairs, in contrast with the DAl3-GFP tumor cens which occurred in islets of at least 6 cens. Most of the

DAl3-hK5His-GFP tumor cens observed were either necrotic or degenerated (Figure

3.5D & F) and were often surrounded by inflammatory cells as compared to the DAl3-

117 Figure 3.4. hK5His acts as a potent anti-angiogenic agent. (A) Implant (arrow) retrieval 1 week post-implantation of 106 Matrigel™ -embedded DAl3, DA/3-GFP or

DAl3-hK5His-GFP cells (representative images are shown at the same magnification).

(B) left panel Representative images of vWF-positive immunostained blood vessels

(black arrows) from two different DAl3-GFP-containing (a, b) and DAl3-hK5His-GFP­ containing (c, d) implants; C, center of implant; B, border of implant; magnification

160X; right panel The mean vWF-positive vessel length (Jlm) is plotted for both experimental groups; bars, S.E.M. (C) The microvessel density is plotted as the mean

2 number of vWF-positive blood vessels divided by the mean section surface area (mm ); bars, S.E.M. Statistical analysis was perforrned using the student t-test. FIGURE 3.4

A

P

c -8 "-'noAiGFP'~~~"i ~ 7 .OA3-K5-GFP $6 ~5 P

GFP-expressing DAl3 cells (A, 2.5X & C, 20X) or hK5His-GFP-expressing DAl3 cells

(B, 2.5X, D, 40X & E, 40X). (B) inflammatory infiltrate (red arrows) demonstrated, (C) eosinophils present along the border region (red arrows) and tumor islets composed of 6 cells (black arrows) are depicted, (D & F) neutrophils within the center region as well as within the border capsule region (red arrows) and representative examples of necrotic tumors (black arrows) are shown. (E) Quantitative histological analysis representing the number of intact tumor cells per high power field (left) and the number of neutrophils per border region (right) composed of concentric layers of edematous immature connective tissue; bars. S.E.M. Statistical analysis was performed using the student t-test. FIGURE 3.5

DA/3-GFP DA/3-K5-GFP B Center

Denselnflammatory infiltrate - o - ,II' Necrotic Tumor Neulrophils

E;s P / o DA3-GFP F .DA3-K5-GFP .. ~Necrotic ~ ... Tumor. teJ~ ~ ..:Ir.. • 1'= 0.0001 ;/Mlf ". ':!> # E Cl ::J r z 1- = 1) ... u.. Neutr0J!!1Ils 5 -.

0 ceUslHPF GFP tumor cells that appeared healthy (no vacuolation, no swelling, no loss of tinctorial affinity, acinar or islet architecture preserved). Both GFP- and hK5His-GFP-containing implants were incompletely surrounded by a capsule at the border regions (Figure 3.5C &

F respectively) composed of concentric layers of edematous immature connective tissue

(fibroblasts with variable numbers ofinflammatory cells). We observed that the hK5His­

GFP-containing implants were markedly thickened by a dense population of inflammatory cells (Figure 3.5B & D) composed of at least 50% neutrophils, qualified as suppurative inflammation. Enumeration confirmed a substantial increase in the number of infiltrated neutrophils in the hK5His-GFP-containing implants as compared to the control implants (Figure 3.5E, P = 0.0001). There were several eosinophils in the border zone of both groups. However, there was no significant difference in the number of infiltrated eosinophils between test and control (data not shown).

3.4.6 hK5His protein is chemotactic for neutrophils and promotes their activation.

In order to explain the enhanced recruitment of host-derived neutrophils within the hK5His-GFP-containing implants, we assessed the ability of soluble hK5His protein to act as a neutrophil chemoattractant. PMN isolated from heparinized human peripheral blood as described in the Materials and Methods were assayed for cell-surface expression of the adhesion marker CD Il b (Mac-1) by flow cytometry analysis, and were found to be strongly positive (Figure 3.6A). PMN were plated on the top surface of the migration filter and exposed to either increasing doses of purified hK5His protein (80, 120 or

160ng) diluted in OPTI-MEM or OPTI-MEM a10ne as a control. We observed that

122 Figure 3.6. hK5His protein acts as a neutrophil chemoatfractant and promotes activation. (A) Expression of CD11b cell-surface adhesion marker on isolated PMN; isotype antibody (leJt), test antibody (right). (B) Relative fluorescence units (RFU) of migrated PMN toward either increasing doses of purified hK5His protein (80, 120,

160ng) diluted in OPTI-MEM or OPTI-MEM alone. (C) Percent expression of

CD11b+CD64+ PMN post-exposure to either increasing doses ofpurified hK5His protein

(80, 120, 160ng) diluted in OPTI-MEM or OPTI-MEM alone or phorbal myristate acetate

(PMA, lOf..lg/mL) as a positive control; bars, S.E.M. Statistical analysis was performed using the student (-test. FIGURE 3.6

0 8 0 A e.I Isotype (\1 CD11b ...fi> ....(1) 98% c 40/0 C :J :l 0 0 0 0

0 0 1 100 10 1G2 103 104 ,ÎP 101 1if. .03 ,04 B ...... ~ 10 ( ~ ** ->< s ::::)6 LL * El:: 4

2

o ""_~"_"L ____~~~======~ c 20 "'"-"DCC--1-"'C-1'7-~b"-+-~Cc--D=-c-"C6:""4'-~:--*""~d"_' * • p< 0_05 ...... P< W CTL \.. 80 120 160..) PMA 10pglmL hK5 (ng) hK5His protein acts as a potent neutrophil chemoattractant and displays a dose-response effect (Figure 3.6B). Additionally, we demonstrate that neutrophil exposure to soluble hK5His protein induces increased cell-surface expression of the granulocyte activation marker CD64 (Fey receptor 1) within the CDllb+ (Mac-1) adhesive neutrophil subset

(Figure 3.6C). These findings are consistent with our histological analysis and suggest that soluble hK5His protein produced by gene-modified DN3 cells acts as a strong neutrophil chemoattractant as weIl as promotes activation of neutrophils within the tumor microenvironment.

3.4.7 MHC 1 expression in hK5His-expressing DA/3 cells.

MHC 1 is a key molecule involved in immune surveillance and its level of expression may influence interaction with innate effectors (221-224). Therefore, we ascertained whether hK5His altered this phenotype. We assessed the cell-surface expression of major histocompatibility antigen (MHC) class 1 on GFP- and hK5His-GFP­ expressing DN3 ceIls by flow cytometry. Our results indicate that hK5His-GFP­ expressing DN3 ceIls express similar levels (54.4%) of MHC-I (H-2Kd) molecule as

GFP-expressing DN3 ceIls (64.1 %) (P> 0.05) (data not shown).

3.5 DISCUSSION

We have previously demonstrated that tumor-expressed hK5His leads to a substantial anti-angiogenic effect and cure of human glioma orthotopic xenografts in a majority of athymic nu/nu nude mi ce (169). In the CUITent study, we observed that

125 NOD-SCID and athymic nude BALB/c mice with impaired T -lymphoid immune systems do not mount as effective an anti-tumor response to DA/3 cells expressing hK5His, when compared to immunologically normal rodents. Furthermore, despite apparent long term cures lasting more than 1 year in normal BALB/c mice, we were unable to demonstrate an effective adaptive immune response to tumor challenge, suggesting that components of the innate immune system play an important role in the effectiveness of K5 anti-tumor properties (Figure 3.2). These observations are congruent with the hypothesis that anti­ angiogenic compounds derived from plasminogen and other sources have pleiotropic effects, which likely involve a synergistic recruitment of an inflammatory anti-tumor response (212, 213). We have observed that tumor expression of hK5His leads to a robust CD3+ lymphoid tumor infiltration, in particular NKT cells, suggesting a role for these immune effector cells in K5-mediated tumor rejection (Figure 3.3). Histological examination of tumor implants confirmed a robust suppreSSIOn of cancer neovascularization (Figure 3.4) as well as a marked neutrophilic suppurative reaction

(Figure 3.5). This neutrophilic infiltration may be mediated by soluble K5 protein, which we demonstrate acts as a chemoattractant for neutrophils and promotes upregulation of the CD64 granulocyte activation marker (Figure 3.6). This observation follows in the stead of the recently discovered property of an array of anti-angiogenic pharmaceuticals - including plasminogen derivatives such as angiostatin - to alter the phenotype of cancer­ associated neovasculature in a manner which leads to enhanced tumor recruitment of leukocytes. Dirkx et al., recently reported that a synthetic angiogenesis inhibitor, anginex, enhances leukocyte-vessel wall interactions in tumor vessels by upregulating tumor endothelial VCAM-l and E-selectin expression, and subsequently increasing infiltration

126 of CD45+ leukocytes and cytotoxic CD8+ lymphocytes into the tumor to suppress tumor growth (212). Although Dirkx et al., report that angiostatin can also increase the leukocyte-vessel wall interaction in vivo, Chavakis et al., daim that angiostatin acts as an anti-adhesive and anti-inflammatory agent since it inhibits peritonitis-induced neutrophil emigration in vivo via its interaction with a4~ 1-integrin and Mac-1 (aM~2-integrin)

(214). Our observation that K5 enhances recruitment ofNKT cells and neutrophils is in support of a pro-inflammatory mode of action and buttresses the theory that inflammation and microvascular suppression act synergistically in the observed anti-tumor effects of hK5. Our survival data in both immunodeficient NOD-SCID and BALB/c nude mice demonstrates that although host-derived neutrophils may be implicated in suppressing tumor progression, their presence is not sufficient in T -cell deficient mi ce to eradicate the tumor, suggesting a cooperative anti-tumor effect between these immune effector cells.

There is precedence for combined lymphocytic and neutrophilic involvement in tumor rejection. Cairns et al., (225) reported that gene-engineered myeloma cells expressing lymphotactin induced infiltration of CD4+, CD8+ and neutrophils leading to effective tumor regression in vivo. Lee et al., (226) described that in vivo injection of IL-8- transfected human ovarian cancer cells induced dramatic neutrophilic infiltration and resulted in decreased tumor growth. In the context of anti-angiogenic therapy, Pike et al.,

(227) also observed in vivo neutrophilic and lymphocytic infiltration in vasostatin

(angiogenic inhibitor)-treated Burkitt lymphoma tumors, which led to tumor suppression.

Work from Abbott Laboratories demonstrates that cancer cells rendered anoxic in vitro translocate GRP78 to the cell membrane which then serves as a ligand for K5. K5- bound GRP78 thereafter initiates an apoptotic cascade (148). The significance of this

127 phenomenon was not tested in experimental animaIs; however it dearly buttresses the daim that K5 can directly affect the phenotype and cellular function of tumor cells. We did not observe an increased apoptosis index or reduced cell growth of K5-expressing

DAl3 cells in standard tissue culture conditions or in a 3% oxygen hypoxic environment, nor were we able to detect membrane-bound GRP78 despite its intracellular abundance in

DAl3 cells (data not shown). These observations suggest that K5 likely interacts with tumor cells via a plurality of pathways of which GRP78 may be one in certain circumstances. Thus, it may be speculated that hK5His protein breaks immune tolerance by inducing a cross-talk to occur between host-derived neutrophils and other innate infiltrating immune cells such as NKT -lymphocytes. Furthermore, K5 from DAl3 may lead to an altered adhesive phenotype of tumor-associated vessels and promote local recruitment of leukocytes from blood stream.

In aggregate, our data confirms the anti-angiogenic potency of hK5His and suggests that hK5His protein also relies upon multiple innate cellular effectors to induce its in vivo anti-neoplastic effect. More specifically, our findings demonstrate that hK5His protein requires functional neutrophils and T-lymphocytes in order to induce optimal tumor rejection. Further studies on the pleiotropic effects of K5 protein on vasculature and inflammatory effector cells may provide new insights allowing K5 to be therapeutically exploited to treat cancer.

3.6 FINANCIAL SUPPORT

S. R. Perri is a recipient of the V.S. Army Medical Research and Materiel

Command Breast Cancer Research Predoctoral Traineeship, Award, DAMDI7-03-1-

128 /~ 0545. J. Galipeau is a recipient of the CIHR Clinician-Scientist Award and this work was

supported by CIHR operating grant MOP-15017.

129 CHAPTER4

130 CHAPTER4

Angiostatin Blocks Macrophage Migration.

Reference: Perri SR, Annabi B, Galipeau J. Angiostatin Blocks Macrophage Migration. 2007. (manuscript submitted to The FASEB Journal).

Preface: Tumor-associated macrophages (T AM) play a pivotaI role in promoting tumor growth and metastasis. Thus, strategies to prevent the recruitment of T AM are appealing as potentiai anti-cancer therapeutics. In Chapter 2, we demonstrated that hum an plasminogen kringle 5 acts as an anti-macrophage agent by suppressing T AM recruitment in vivo and by blocking in vitro migration of human monocyte-derived macrophages. In Chapter 4, we investigated if other internaI plasminogen fragments possessing anti­ angiogenic activity such as angiostatin aiso induced macrophage inhibitory properties.

131 4.1 ABSTRACT

In light of the involvement of tumor-associated macrophages (T AM) in the promotion of tumor growth and inetastasis, strategies to prevent TAM infiltration within the tumor microenvironment are currently under investigation. Our laboratory previously demonstrated that human plasminogen (PIg) kringle 5 (K5) protein blocks recruitment of host-derived tumor-infiltrating macrophages in vivo and suppresses the migration of human monocyte-derived macrophages in vitro. In addition, the recent observation that angiostatin reduces macrophage infiltration in an atherosclerosis model prompted our laboratory to further explore the use of angiostatin as a macrophage inhibitory agent. We demonstrate that angiostatin blocks migration ofboth murine peritoneal macrophages and human monocyte-derived CD206+ macrophages in vitro. Furthermore, we show that angiostatin leads to a decrease in the gelatinolytic activity of macrophage-produced matrix metalloproteinase-9, which may explain, in part, the observed angiostatin­ mediated inhibition of migration. Additionally, we detect the presence of the p-subunit of

ATP synthase on the cell-surface of macrophages by immunoblot analysis. ATP synthase was previously found to be a receptor for angiostatin on the cell-surface endothelial cells.

We propose that the presence of A TP synthase on the surface of macrophages may promote interaction with angiostatin and prevent migration, similar to what has been reported with endothelial cells. Our findings suggest that angiostatin holds promise as an inhibitory agent against macrophages.

132 4.2 INTRODUCTION

It is weIl established that tumor-associated macrophages (TAMs) are implicated

ln tumor progression and metastasis. T AMs are skewed toward the M2 phenotype

(polarized type II macrophages) (228), which is oriented toward promotion of

angiogenesis, tissue remodeling and repair (228). T AMs are typified by low cytotoxic

potential against tumor cells due to reduced production of nitric oxide (NO) and pro­

inflammatory cytokines (229). T AMs may possess decreased inflammatory cytokine

production as a consequence of their exposure to tumor-derived anti-inflammatory

molecules such as interleukin (lL)-4, IL-10, transforming growth factor-~l and

prostaglandin E2 (230, 231). Type II macrophages promote tumor cell proliferation and

metastasis by producing an array of pro-angiogenic factors and metalloproteinases

(MMP) as weIl as their involvement in signaling pathways regulating the function of

fibroblasts in the tumor stroma (228,229,232,233). Several reports have demonstrated a

correlation between infiltration of T AMs and prognosis in cancer patients, whereby

increased macrophage infiltration resulted in worse prognosis (232, 234-236). T AMs

have also been associated with immune suppression and tolerance (237) as weIl as

inhibition of T cell responses by upregulating NO, prostaglandins, tumor necrosis factor­

a and arginase activity leading to apoptosis of activated T cells (238, 239).

T AMs are derived from circulating monocytes and are recruited to the tumor

microenvironment by tumor-produced monocyte chemotactic factors (240) sorne of

which include: CCL2/monocyte chemoattractant protein (MCP)-l (produced by gliomas,

sarcomas, lung, breast, cervix, ovary, melanoma and pancreatic cancer) (241-244),

~' CXCL8/IL-8 (produced by breast and melanoma) (245, 246) and CCL5/RANTES

133 (Regulated on Activation, Normal T -cell Expressed and Secreted, produced by breast and melanoma) (247, 248). High expression of monocyte chemotactic factors is associated with increased macrophage infiltration (249). Upon T AM tumor infiltration, cancer cells sustain T AM survival, which subsequently induces T AMs to produce growth factors and matrix degradation enzymes leading to tumor proliferation, angiogenesis and invasion

(230, 250-252).

It would thus be desirable to further explore strategies to reduce T AM infiltration in the tumor microenvironment. Our laboratory demonstrated that human plasminogen kringle 5 (K5) domain acts as a potent anti-angiogenic agent as weIl as blocks migration of human monocyte-derived macrophages in vitro and substantially reduces CD45+Mac-

3 +Gr- r macrophage recruitment to the tumor microenvironment in vivo (169).

Interestingly, it was reported by Moulton et al. (210) that angiostatin, another internaI proteolytic fragment of plasminogen encompassing the first 4 kringle domains, also reduced in vivo infiltration of Mac3+ macrophages in the plaque and surrounding vasa vasorum in an atherosc1erosis model. This finding prompted our laboratory to further investigate the use of angiostatin as an inhibitory agent against macrophages in order to develop a strategy to target T AMs to suppress tumor promotion and angiogenesis.

4.3 MATERIALS & METHODS

4.3.1 Macrophage Isolation.

Murine peritoneal macrophages were isolated from C57B1/6 mi ce (Charles River,

Laprairie, Quebec, Canada) by lavage of the abdominal cavity with RPMI solution.

Harvested peritoneal macrophages were resuspended in media (RPMI containing 10%

134 fetal bovine serum, FBS and 1% penicillin-streptomycin, pen/strep, Wisent Technologies,

St. Bruno, Quebec, Canada), plated on non-adherent tissue culture dishes for 4 hours, non-adherent cens were removed, adherent fraction was trypsinized and used for further experiments. Fresh human peripheral blood mononuc1ear cens (PBMCs) were obtained from normal volunteers and maintained in DMEM-F12 (Wisent Technologies) supplemented with 2% human serum and 500U/ml granulocyte/macrophage colony stimulating factor (GM-CSF; Immunex, Thousand Oaks, CA) to induce differentiation of monocytes into macrophages.

4.3.2 Macrophage Migration Assay.

5 Murine or human macrophages (1.5x10 ) were plated onto 0.15% gelatin/PBS­ coated 5-J-lm pore chemotaxis membranes (Coming, Acton, MA) within Boyden chamber inserts. Murine macrophages were exposed to recombinant human Kl-3 protein (1.0 and

5.0 J-lM) for 24 hrs. Human macrophages were exposed to Kl-3 (10 J-lM) for 4 hrs.

Species-specific recombinant GM-CSF was used as a chemoattractant and diluted in serum-containing RPMI media. Each sample was tested in triplicate and non-migrating cens were removed by gently wiping the upper surface of the filter. The average number of migrating murine macrophages per field was assessed by counting 3 random high power fields per filter and plotted. The total number of migrating human macrophages per filter was counted and plotted.

135 4.3.3 Immunoblot Analysis.

Cell-surface membrane proteins were isolated from munne macrophages and human umbilical vein endothelial cells (as a positive control) using the Cell-Surface

Protein Isolation kit (Pierce, Rockford, IL), which biotinylates and purifies cell-surface membrane proteins. Immunoblot analysis was performed on isolated protein lysates using anti-a-subunit and anti-p-subunit ATP synthase antibodies (Molecular Probes,

Burlington, Ontario, Canada).

4.3.4 Apoptosis Assay.

Murine peritoneal macrophages were plated on 60-mm dishes for 3 hours in

RPMI media containing 10% FBS and 1% penlstrep. Non-adherent cells were removed by washing 3 times in PBS. Adherent macrophages were either exposed to 2.0 !lM or increasing concentrations of recombinant human angiostatin (KI-3) protein (0.1, 0.5, 1.0 and 2.0 !lM) diluted in media for 16 ho urs in normal culture conditions. Whole-cell staining was performed using the poly-caspase FUCA-PI (fluorochrome-Iabeled inhibitors of caspases-propidium iodide) apoptosis kit (lmmunochemistry Technologies,

Bloomington, MN) and staining was quantified by flow cytometry analysis.

4.3.5 Gelatin Zymography.

Murine macrophages were exposed to either 1 !lM recombinant human KI-3 protein diluted in media or media alone for 24 hrs, conditioned media was collected and concentrated 4-fold. Gelatinolytic activity in conditioned media samples was assessed by gelatin zymography. An aliquot (20 ,ul) of the culture medium was subjected to sodium

136 dodecyl sulfate/polyacrylamide gel electrophoresis (SDSIPAGE) with a 7.5% (w/v) polyacrylamide gel containing 0.1 mg/ml gelatin. The gels were then incubated for 30 minutes at room temperature twice in 2.5% (v/v) Triton X-lOO to remove SDS, then rinsed five times in doubly distilled water. The gels were incubated at 37°C for 20 hrs in

20 mM NaCI /5 mM CaCh /0.02% (v/v) Brij-35/50 mM Tris / HCI buffer (pH 7.6), then stained with 0.1% Coomassie Brilliant Blue R-250, followed by destaining in 10% (v/v) acetic acid / 30% (v/v) methanol in water. Gelatinolytic activity was detected as unstained bands on a blue background and was quantified by densitometric measurement.

4.4 RESULTS

4.4.1 Characterization of Isolated Macrophages.

We isolated both murine and human macrophages in order to test the effect of recombinant human angiostatin (Kl-3) exposure. Murine macrophages were harvested by peritoneal lavage of thioglycollate-stimulated C57BI/6 mice. A Giemsa stain was performed on a cytospin sample prepared immediately post-isolation and a representative stain is depicted in Figure 4.IA, whereby vacuole-containing macrophages and macrophages with engulfed particIes is demonstrated. Human monocyte preparations were isolated from peripheral blood mononucIear cells (PBMCs) of healthy donors by a standard leukaphoresis protocol. Monocytes were induced to differentiate by exposure to human recombinant granulocyte-macrophage-colony stimulating factor (GM-CSF) for 7 days in normal culture conditions and a cell aliquot was subsequently stained with human anti-CD206 antibody, which detects a mannose receptor that is expressed on the plasma

137 Figure 4.1: Macrophage characterization. (A) Giemsa stain perfonned on a cytospin preparation of murine peritoneal macrophages. Representative stain of macrophages with ingested particles is demonstrated, 100X magnification. (B) Human monocyte-derived macrophages stained with human anti-CD206 antibody and analyzed by flow cytometry.

Monocytes induced to differentiate into macrophages using human GM-CSF were 88%­ positive for the macrophage marker CD206. (C) Cell-surface membrane proteins from murine peritoneal macrophages "M" and human umbilical vein endothelial cells ("EC" as a positive control) were subjected to anti-a and anti-13 A TP synthase immunoblot analysis. Murine peritoneal macrophages express the l3-subunit of ATP synthase (52 kDa) on the cell-surface. FIGURE 4.1 A

1

v v Q e B - Isotype hCD206+ (") M Q- 0.4% ..-Q 88.60/0

N N 0 0 ~ r-

-s -e

0 Cl Ct 0 4 Hl 10

C ATP-Synthase EC M a-subunit ---+

(3-subunit ---+ membrane of macrophages. Flow cytometry analysis demonstrates that monocyte-derived macrophages are 88% CD206+ (Figure 4.1B).

It was reported by Moser et al. (62) that cell-surface expressed ATP synthase (a­ and j3-subunits) acts as a putative receptor for angiostatin on primary endothelial cells and that angiostatin exerts its anti-endothelial cell properties via binding to this receptor.

We sought to determine whether either subunit of ATP synthase was expressed on the cell-surface of macrophages. The cell-surface membrane pro teins from murine peritoneal macrophages and human umbilical vein endothelial cells (HUVECs) (as a positive control) were isolated using a commercially available kit and subjected to either anti-a­ or anti-j3-subunit ATP synthase immunoblot analysis (Figure 4.1 C). The presence of the j3-subunit of ATP synthase was detectable on the cell-surface ofmurine macrophages as a band migrating at 52 kDa. Our results suggest that murine macrophages do not express a-subunit of ATP synthase since we were unable to detect a band at 55 kDa. Both the a­ and j3-subunit ATP synthase are detectable on the cell-surface of HUVECs. These findings suggest that angiostatin may affect macrophages, in addition to endothelial cells, by interacting with cell-surface expressed j3-subunit of A TP synthase.

4.4.2 Angiostatin Inhibits Macrophage Migration.

In light of the presence of the j3-subunit of ATP synthase on the cell-surface of murine macrophages, we tested the ability of recombinant human angiostatin (KI-3)

4 protein to affect macrophage migration. Murine peritoneal macrophages (1.5 x 10 ) previously exposed to either KI-3 or media alone were plated on the upper chamber of gelatin-coated filters within modified Boyden chambers (Figure 4.2A). KI-3 was capable

140 Figure 4.2: Angiostatin inhibits macrophage migration. (A) Depiction of a modified­

Boyden chamber, whereby macrophages are plated on a gelatin-coated polycarbonate 5

Jlm-pore filter in the upper chamber "A" and subjected to chemoattractants in the lower

chamber "B". (B) Murine peritoneal macrophages were exposed to either RPMI or 1 JlM

or 5 JlM recombinant angiostatin (Kl-3) protein for 45 minutes at 37°C culture

conditions and subsequently plated in the upper chamber "A". Angiostatin inhibited the

migration of macrophages chemoattracted by murine GM-CSF in a dose-dependent

manner. The number of migrated macrophages per high power field (HPF) is shown;

S.E.M. (C) Angiostatin (Kl-3) inhibits the migration of human macrophages

chemoattracted by human GM-CSF. The total number of migrated macrophages lS

shown; bars, S.E.M. Statistical analysis was performed using the student {-test.

------FIGURE 4.2 A

B 1()() -r---~::::;;::=:;::;;:;:;::;;;::==:::::::---;r P < 0.00005 '\ r P< 0.005 .,

O-f--­ A RPMI RPMI 5pM K1-3 B RPMI Serum + Serum + mGM .. CSF mGM-CSF c

tJ) - (' P < 0.05 ' o0360 ." .!40 ! tJ) i20

o A DMEM/F12 10pM K1-3 B Serum + Serum + hGM-CSF hGM-CSF of blocking the migration of macrophages chemoattracted by media containing 10% serum and 500 U/mL of recombinant murine GM-CSF in a dose-dependent fashion (1

!JM, P < 0.005 and 5 !JM, P < 0.00005; Figure 4.2B). Consistent with these results, we also observed a Kl-3-induced suppression of migration when human monocyte-derived macrophages were utilized in a similar migration assay (P < 0.05, Figure 4.2C). These observations suggest that direct exposure of macrophages to Kl-3 protein may lead to interaction with cell-surface expressed p-subunit of ATP synthase, which may decrease the migratory capacity of macrophages either by inducing macrophage apoptosis or by affecting downstream signa1ing mediators implicated in macrophage chemotaxis.

4.4.3 Angiostatin Does Not Induce Caspase Activation in Macrophages.

We investigated whether macrophages exposed to Kl-3 led to apoptotic cell death. Murine peritoneal macrophages were exposed to either increasing doses of Kl-3

(0.1, 0.5, 1.0 and 2.0 !JM) diluted in serum-containing media or serum-containing media alone for 16 hrs in normal culture conditions. Cells were harvested and subsequently stained with a commercially available fluorescence-based FUCA/PI kit, which selectively stains cells containing active caspases. Results reveal that KI-3 was unable to induce cell death at the doses tested, as evidenced by negligible FUCA+ staining as weIl as similar FUCA+PI+ staining as control-treated cells (Figure 4.3). The data suggests that

KI-3 exposure to macrophages does not lead to cell death, even though angiostatin at the same dose has been reported to cause endothelial cell death. It may be postulated that Kl-

3 requires interaction with both cell-surface expressed Œ- and p-ATP synthase subunits to induce apoptosis, as was observed for endothelial cells.

143 Figure 4.3: Angiostatin does not indu ce macrophage apoptosis. Murine peritoneal macrophages were exposed to either increasing doses (0.1, 0.5, 1.0, 2.0 /lM) of recombinant Kl-3 protein or 10% serum-containing RPMI media as a control for 16 hours in nonnal culture conditions. Cells were then stained with FLICAIPI and analyzed by flow cytometry. Kl-3 does not induce FLICA+ staining (ie: apoptosis) at the doses tested. The percentage of PI+ cells is not affected by KI-3 exposure. The upper most panel depicts typical FLICA + staining dot plot and was obtained from Immunochemistry

Technologies. Representative flow cytometry dot plots with appropriate staining percentages are shown on the right panel. FIGURE 4.3

(;::t<:: Cv::., ~': P')!U~,.,;

.-:;' .. '.

J~:;()O~\'~:~: Ç",'I$ !/... rt: ~. ,':. M P._·.').~;: n

% expression ""= CTL <'> :;::: 22

""~ = 77 0.3

=<=> 2 3 4 ~100 '0 10 10 10 ..,. ' 9 O.1JlM ""=- K1-3 24 N <:>

<=> 75 0.2

<::> -= 2 3 4 - ~1Oü 10 10 10 10 c.. =- ' <-> O.5JlM C':t - 19 N K1-3 <= = 80 0.3

103 104 """=

>C"''':. 1JlM = 20 N K1-3 '2

c::'> 78 0.3 <=> S'10° 10 102 10~ 104 '

MMP-9.

In attempt to explain the mechanism by which KI-3 induces inhibition of macrophage migration, we decided to test whether the activity of key proteases implicated in cell chemotaxis/motility such as matrix metalloproteinase (MMP)-2 and -9 were affected by KI-3 exposure. Murine peritoneal macrophages were exposed to 1 f-tM

KI-3 for 24 hrs in normal culture conditions. Conditioned media was collected, concentrated 4-fold and subjected to gelatin zymography analysis. We observed that exposure of macrophages to K 1-3 led to a decrease in the ability of macrophage-derived soluble MMP-9 to lyse ge1atin as compared to control-treated cells (Figure 4.4). No change in the gelatinolytic activity of MMP-2 was detectable. We speculate that KI-3- mediated decrease in the activity of macrophage-produced MMP-9 may be in part responsible for the observed a KI-3-induced migration blockade.

4.5 DISCUSSION

Our findings demonstrate that angiostatin (KI-3), which has been touted as a promising anti-angiogenic agent, possesses additional anti-tumoral properties by exerting inhibitory action on macrophages. More specifically, we show that recombinant human

KI-3 protein blocks in vitro migration of murine peritoneal macrophages (Figure 4.2B) and human monocyte-derived CD206+ macrophages (Figure 4.2C). This result is consistent with that observed by Benelli et al. (253) who report that angiostatin is capable of inhibiting MCP-I-induced monocyte migration. Furthermore, we detect expression of the 13-subunit of ATP synthase on the cell-surface of macrophages by immunoblot

146 Figure 4.4: Angiostatin mediates decrease in MMP-9 gelatinolytic activity. Murine peritoneal macrophages were exposed to either 1 /lM KI-3 or media al one for 24 hrs.

Conditioned media was concentrated 4-fold and run on a discontinuous polyacrylamide gel containing gelatin (enzyme substrate for the MMPs being tested). Macrophages exposed to Kl-3 possess decreased MMP-9 (95 kDa) activity. No change was observed in MMP-2 (72 kDa) activity. FIGURE 4.4

4X MMP-2 stds - + K1-3

« MMP-9

« MMP-2 analysis (Figure 4.1C). ATP synthase (both the u- and ~-subunits) was the first reported receptor for angiostatin and was found to be expressed on the cell-surface of endothe1ial cells (62). It was also recently reported that tumor cells express extracellular surface ATP synthase, where increased activity was observed under low extracellular pH (pHe) conditions typical of the tumor microenvironment (110). Angiostatin exposure under low pHe led to intracellular acidification of tumor cells and direct cytotoxicity (110).

Consistent with these findings, we speculate that direct contact of macrophages with K1-

3 may promote interaction with cell-surface expressed ~-subunit of ATP synthase, causing inhibition of migration, similar to that observed by Moser et al. (62) with primary endothelial cells. To specifically determine whether angiostatin interacts via the macrophage cell-surface expressed ~-subunit of ATP synthase to suppress migration, we plan to assess the effect of adding an A TP synthase blocking antibody on migration.

Further experiments need to be performed to confirm whether angiostatin induces macrophage apoptosis. Interestingly, we also observed a Kl-3-mediated decrease in the gelatinolytic activity of macrophage-produced MMP-9 (Figure 4.4). MMP-9 is linked with leukocyte transmigration as was recently observed by Kolaczkowska et al. (254) in a murine mode1 of induced peritonitis, whereby ge1atinase B-deficient mi ce (MMP-9 -/-) mice possessed impaired leukocyte infiltration compared to controls. MMP-9 has also been reported to be participate in lung-specific metastasis (255). Downregulation of

MMP-9 activity may be associated with the observed angiostatin-mediated inhibition of macrophage migration.

It is worthy to note that other well-characterized endogenous anti-angiogenic agents have also been reported to possess broadened cell specificity than initially

149 believed. For instance, Moulton et al. (210) reported that soluble VEGF receptor 1 (256) is capable of inhibiting plaque-associated endothelial cell sprouting as well as suppressing VEGF-induced migration of human blood monocytes through fibronectin­ coated transwell membranes. In an experimental model of bone metastasis induced by breast cancer cells that mimic metastasis observed in naturally occurring human breast cancer (257), it was shown that angiostatin acts as a very effective inhibitor of bone resorption by directly inhibiting cancer-induced bone destruction via direct inhibition of osteoc1ast activity and generation. Furthermore, our laboratory confirmed that human K5 protein, another internaI c1eavage product of plasminogen, acts as a potent endothelial cell inhibitor (169) as has been observed by others (128). Moreover, we demonstrate that

K5 directly inhibits in vitro migration of human monocyte-derived CD206+ macrophages and leads to a significant reduction in CD45+Mac-3+Gr-r macrophage infiltration within tumor implants (169). Additionally, we show that K5 protein is chemotactic for CDllb+ neutrophils in an in vitro chemotaxis assay and that K5-containing tumor implants possess a cellular infiltrate predominantly composed of neutrophils (170). Similarly, angiostatin was reported to inhibit IL-8- and macrophage inflammatory protein 2-induced neutrophil migration (253).

Overall, we provide evidence supporting the use of angiostatin as a promising therapeutic agent against macrophages. We speculate that angiostatin's ability to suppress tumor-infiltrating leukocytes, in addition to its previously reported endothelial cell inhibitory properties as well as its direct cytotoxicity on tumor cells, could markedly enhance its effectiveness as an anti-cancer agent. Moreover, in light of angiostatin's

150 novel macrophage suppressive property, its utility could also be extended to target M2- phenotypic macrophages implicated in immune regulation or tissue remodeling (258).

151 CHAPTER 5

DISCUSSION & CONCLUSION

152 CHAPTER 5: DISCUSSION AND CONCLUSION

5.1 Anti-Angiogenic Agents: From Bench to Bedside

In 1971, published a landmark reVlew ln the New England

Journal of Medicine (186) proposing that tumors produced what he referred to as a

"tumor angiogenesis factor" that was responsible for recruiting new blood vessels toward the tumor, allowing it the ability to grow and persist. Folkman also speculated that inhibition of this diffusible factor could prevent the tumor from progressing. Although his hypothesis was initially met with much skepticism, further research revealed that his notion was indeed correct. In the last 30 years, angiogenesis research has gained tremendous interest and has generated over 25 000 pub li shed reports. It is now well established that cancers require blood vessels to progress from premalignant, noninvasive growths into aggressive tumors that can metastasize to distant sites in the body.

In light of the fact that anti-angiogenic therapy targets genetically stable vascular endothelial cells and not the tumor itself, researchers in the field explored various avenues to develop a pan-tumor therapy focused on eliminating tumor blood vessels with the goal to induce tumor regression. The initial anti-angiogenesis hype was dampened by animal and preliminary human clinical trials that revealed that different tumors possess different response rates toward anti-angiogenic therapy. There are several factors that could affect the treatment response such as the level of angiogenic factors produced by the tumor, expression of oncogenes and resistance to hypoxia (8). In a rat model of glioblastoma, anti-VEGF therapy induced, as expected, decreased tumor vessel density and increased apoptosis. An additional unexpected effect of the treatment was increased co-option of existing blood vessels, thus overcoming the requirement for the tumor to

153 initiate angiogenesis (259). Resistance to anti-angiogenesis therapy has been reported in several studies: TSP 1 therapy for glioblastomas (260), therapy for renal cell carcinoma (261) and p53-negative tumors become less responsive to combinations of low-dose cytotoxic and anti-angiogenic agents (262). Anti-angiogenic therapy also has the ability to contribute to hypoxia due to the regression of oxygen-transporting blood vessels. As a result, hypoxia as weIl as VEGF-A production leads to the expression of anti-apoptotic genes (263, 264) and multidrug-resistance-associated proteins (265), and thus allows the tumor to become protected from the effects of cytotoxic drugs.

Over 50% of human cancers possess mutations in p53 which can affect the angiogenic balance. For instance, tumor cells with p53 mutations are selected for based on their ability to survive in hypoxia (262) and have been reported to downregulate TSPI expression as weIl as upregulate VEGF-A expression (266). Consistent with these findings, wild-type p53 tumors inhibit VEGF transcription (78) and promote MDM2- mediated ubiquitylation and proteasomal degradation of HIF-la, which is the main transcriptional activator of VEGF under hypoxia (77). In normal ceIls, hypoxia induces upregulation of VEGF-A and downregulation of TSP1, independent of p53 status.

However, when normal cells are transformed with ElA and RAS, TSPI expression is decreased, independent of hypoxia, whereas VEGF-A expression remains normal (267).

Further studies have implicated p53 in the regulation of tumor angiogenesis. Both

Holmgren et al., (268) and Gautam et al., (269) demonstrate that introduction of wild­ type p53 induced dormancy and inhibited metastasis by affecting the TSPI :VEGF-A ratio, implying that mutations in p53 allow tumors the ability to circumvent the effects of

154 anti-angiogenic therapy. P53-mediated drug resistance should be overcome by utilizing a combination of anti-angiogenic regimens.

Studies using the mouse model of pancreatic-islet carcinogenesis have also demonstrated that the success of anti-angiogenic treatment is dependent on the stage of disease (270). For instance, endostatin (39), MMP (8) and VEGF inhibitors (271, 272) are most effective in treating early-stage disease, by preventing the angiogenic switch from being tumed on in dysplastic lesions as weIl as inhibiting the progression of small solid tumors. These same anti-angiogenic therapies cannot stabilize or induce regression of bulky end-stage tumors. Conversely, other angiogenic inhibitors such as TNP-

470/AGM-1470 (8), which prevent endothelial cell proliferation and smooth muscle cell migration, are capable of reducing the size of bulky end-stage tumors, but have no effect on early-stage tumors. These observations suggest that there may be specific differences in the angiogenic vasculature of early versus late-stage tumors or in the regulation of angiogenesis induction during the growth of tumor vessels. It was demonstrated in a pancreatic-islet model (273) that combination of SU5416 (blocks VEGFR) and SU6668

(blocks PDGFR and to a lesser extent FGFR and VEGFR) prevents the angiogenic switch and induces the regression of both small and large tumors with improved synergistic effects at aIl stages, as compared to single agent treatment. IdeaIly, for therapeutic purposes, it would be advantageous to be able to predict which combination of anti­ angiogenic agents synergisticaIly. Cline et al., (274) exposed endothelial cells to various anti-angiogenic agents and subsequently performed gene-expression profiling studies to predict which combination of anti-angiogenic therapies would most effectively inhibit tumor growth in animal models. Further studies in this area will help to more accurately

155 predict which combination of angiogenic inhibitors act synergistically in the clinical setting.

Despite much success with the use of anti-angiogenic agents in pre-clinical studies, it is worthy to note that the tirst anti-angiogenic clinical trials failed to reduce tumor burden and did not prolong patient survival. However, clinical trials that failed attempted to treat aggressive refractory tumors of end-stage patients using endostatin or agents targeting VEGF signaling. Aggressive tumors develop mechanisms to overcome the effects of single anti-angiogenic therapy by activating alternative pathways, so in retrospect it is not surprising that these trials were unsuccessful.

5.2 Adverse Effects Associated with Anti-Angiogenic Agents

The reported toxicities associated with anti-angiogenic agents are not severe and consist of the following: hypertension, blood clotting, proteinurea and more rarely gastrointestinal performations of the bowel, which can be severe (275). Blood clot formation may be worsened by combining anti-angiogenic agents with certain regimens of chemotherapy, such as SU5416, a VEGFR2 antagonist, combined with gemcitabine and cisplatin. However, many grade 3 and grade 4 adverse effects have been manageable with medical treatment during ongoing therapy (276). Another concern with the use of anti-angiogenics therapeutically is the development of acquired resistance as was observed by Casanovas et al. (277) in a mouse model of pancreatic islet carcinogenesis.

Following an initial period oftumor growth suppression, resistance to VEGFR2 blockade emerged as tumors re-grew during treatment. It was reported that resistance involved reactivation of tumor angiogenesis, independent of VEGF, and associated with hypoxia-

156 rnediated induction of other pro-angtogemc factors, including rnernbers of the FGF farnily (277). It has also been recently reported that anti-VEGF agents under clinical deve10prnent can "prune away" 75% of turnor vesse1s, as well as possess the ability to affect Il of 17 organs analyzed by inducing substantial but reversible capillary regression

(278).

5.3 Avastin: A Clinically Successful Anti-Angiogenic Agent

To date, the rnost clinically successful anti-angiogenic agent is A vastin

(bevacizurnab), a hurnanized monoclonal antibody directed against VEGF, developed by

Genentech. Avastin was tested in cornbination with conventional chemotherapy in a large-scale Phase III clinical trial and the cornbination treatment prolonged survival of patients with metastatic colorectal cancer when compared to chemotherapy alone (68). As a result, standard chemotherapy cornbined with Avastin was approved in 2004 by the

Food and Drug Administration and other countries as first-line therapy for metastatic colorectal cancer. Moreover, Phase III clinical trials assessing A vastin cornbined with chemotherapy for the treatment of breast and non-small cell lung cancer have also generated successful results (279).

5.4 Exploiting Gene Therapy Based on Lessons Learned from Anti-Angiogenic

Clinical Trials

Within the last decade, several angiogenic inhibitors have been discovered, with a rnajority consisting of endogenous proteins or fragments of larger precursor proteins normally present in the body. Clinical trials assessing the efficacy of anti-angiogenic

157 therapies have provided valuable insight (34) and emphasized the potential benefits of employing gene therapy strategies.

1. Genetie stability of endothelial eells: Since endothelial cens are more geneticany

stable than tumor cens, they possess a lower chance of accumulating mutations

that lead to the development of early drug resistance (39). Thus, gene therapy

approaches targeting endothelial cens via delivery of anti-angiogenic proteins are

preferred over those aimed at tumor cens, which may cause the tumor to develop

an escape mechanism.

2. Use of metronomie dosing: Delivering anti-angiogenic agents continuously at

low doses has proven to be more effective than the intermittent peaks associated

with repeated administration of recombinant proteins (280). Gene therapy

strategies can provide sustained expression of the therapeutic protein over a long

duration.

3. Harnessing the power of the angiogenie switeh: It has become evident that the

angiogenic switch must be tumed on for tumor growth and progression (8). In situ

tumors present in the body will only progress if there is an imbalance between the

tumor-produced pro-angiogenic signaIs compared to the body' s ability to

counteract these signaIs (281). The development of gene therapy techniques to

boost the body's levels of endogenous angiogenic inhibitors can be exploited to

balance the levels in order to promote tumor suppression.

4. lnvolvement of multiple angiogenie pathways. The contribution of several

angiogenic factors promotes tumor proliferation and progression. Conventional

anti-angiogenic therapies attempt to target a number of pro-angiogenic mediators

158 by administering repeated, systemic boluses of multiple anglOgemc inhibitors

(282). The use of gene therapy could facilitate the targeting of numerous

angiogenic pathways since biologically relevant levels of several angiogenesis

inhibitors could be delivered.

Although recombinant angiostatic proteins have been found to possess unique anti-cancer properties, their utility as therapeutics poses several obstacles that need to be overcome to better elucidate mechanisms of action, dosing schedule and the overall biological effectiveness in the clinic. In addition, the structural complexity possessed by many of these inhibitors as well as the difficulty in cost-effectively manufacturing large quantities for clinical use also raises a concem. Optimized gene therapy strategies can be exploited to bypass sorne of the limitations associated with recombinant anti-angiogenic protein use. An ideal anti-angiogenic gene therapy approach involves the use of autologous cells that can be genetically-engineered with a vector in order to deliver sustained, biologically relevant quantities of the therapeutic protein for an extended period of time.

Therapeutically, it is necessary that the cellular biopharmaceutical be gene-modified with a vector that is safe and po orly immunogenic, since repeated administration may be required over a period of several months to achieve disease stabilization and complete cancer regresSlOn.

Overall, gene therapy approaches can bypass sorne of the limitations associated with conventional therapies. This led to the first gene therapy clinical trial, which aimed to treat X-linked SCID (also known as the gamma chain deficiency) in children. This is an inherited disease characterized by an early block in T and natural killer lymphocyte differentiation caused by mutations of the gene encoding the gamma chain cytokine

159 receptor subunit of IL-2, -4, -7, -9, and -15 receptors (283). Cavazzana-Calvo and colleagues employed the use of autologous CD34+ bone marrow stem cells that were ex vivo retrovirally gene-modified with the gamma chain gene (283). In 2000 and 2002, it was reported that the majority of patients were successfully treated (283, 284). However, the initial enthusiasm was quickly dampened 3 years post-therapy when 2 of the 10 children developed leukemia-like conditions. The development of leukemia was linked to retroviral vector integration proximal to the LM02 proto-oncogene promoter, causing aberrant transcription and LM02 expression (285). This setback prompted investigators to develop better vectors, yet the possibility of oncogene activation by retroviral integrating vectors still remains a valid concem. Despite the results of the SCID clinical trial, gene therapy strategies still possess attractive features that can be exploited for the treatment of cancer. Most likely, anti-angiogenic gene therapies will provide the greatest benefit ifutilized as adjuvants and in combination with current treatment modalities.

5.5 Development ofa Novel A nti-Angiogenic Gene Therapy Strategy for Cancer

In light of the importance of angiogenesis in tumor progression as well as the impediments associated with recombinant angiostatic protein use and the benefits of gene therapy, the main objective of this thesis was to develop a novel anti-angiogenic gene therapy strategy for cancer. The specific research aims were as follows:

1. To determine whether retrovirally gene-modified human U87 glioma cells

producing soluble human plasminogen kringle 5 protein could suppress tumor

progression and angiogenesis in an orthotopic brain cancer model;

160 2. To determine whether human plasminogen kringle 5 protein, when secreted by

retrovirally gene-engineered murine DA/3 mammary adenocarcinoma cells, lS

dependent upon innate immune effector cells to mediate its anti-cancer effect;

3. To determine whether recombinant human angiostatin protein, in addition to its

strong anti-angiogenic activity, can also exert inhibitory action on macrophages.

The rationale for use of human Plg K5 protein as the anti-angiogenic agent of choice for this thesis was based upon studies comparing the potency of internaI proteolytic fragments of human Pig. It was shown that proteolytically-derived purified human K5 acts as a stronger inhibitor of in vitro bFGF-stimulated endothe1ial cell proliferation (128) and migration (60) than human proteolytic angiostatin (Kl-4). Additionally, it was demonstrated that Kl-5 is more potent than Kl-4 in suppressing in vitro bFGF-stimulated endothelial cell proliferation (132) and migration (60). Kl-5 also displayed strong inhibition ofbFGF-stimulated corneal neovascularization and tumor growth, whereas Kl-

4 at the same dose was complete1y ineffective (132). Therefore, these findings suggest that the cryptic K5 domain acts synergistically to markedly enhance the potency of Kl-4.

5.5.1 Hypothesis 1

In light of these reported observations, we decided to test the ability of K5 domain produced by retrovirally gene-engineered human U87 glioma cells to act as an anti­ angiogenic agent and assessed its efficacy in suppressing tumor progression in an orthotopic brain cancer model. Our results demonstrate that:

161 i) Soluble K5 protein spontaneously adopts a tertiary structure consistent

with native conformation as displayed within Plg when expressed by

genetically-engineered eukaryotic ceIls;

ii) K5 protein produced by retrovirally gene-modified human U87 glioma

cells is appropriately secreted and inhibits in vitro growth factor­

stimulated endothelial cell migration;

iii) Soluble K5 protein blocks glioma-associated angiogenesis in vivo;

iv) Soluble K5 protein possesses novel anti-macrophage properties,

capable of suppressing in vitro migration of human monocyte-derived

macrophages and prevents the recruitment of host-derived tumor­

infiltrating macrophages in vivo and;

v) Glioma-targeted K5 expression suppresses tumor growth and pro longs

survival in an experimental orthotopic brain cancer model.

Since it has been reported that the tertiary conformation adopted by plasminogen kringles is necessary for maintenance of functional angiostatic activity, we confirm that eukaryotically-expressed Plg K5 protein preserves the prototypical native kringle disulfide bridging conformation. Indeed, our cysteine bridging pattern data is consistent with the only other report that characterized the structure of recombinant K5 domain using NMR technology (286). Functionally, we show evidence that soluble K5 protein produced by gene-engineered glioma cells is capable of suppressing growth factor­ induced endothelial cell migration in vitro, as has been previously reported using proteolytically-derived K5 protein (60). Our demonstration that soluble K5 protein acts as a potent angiostatic agent by reducing blood vessel formation in vivo validates the

162 concept that tumor cells can be ex vivo retrovirally gene-modified to secrete functional

K5 protein capable of providing sustained expression in order to exert an anti-endothelial cell effect in vivo. Additionally, our observation that soluble K5 protein can substantially reduce the recruitment of in vivo host-derived tumor-infiltrating macrophages by imparting a direct anti-migratory effect, as shown in its ability to inhibit the migration of human monocyte-derived CD206+ macrophages in vitro, provides novel mechanistic insight. Until this point, K5 protein had been consistently characterized as an endothelial cell specific inhibitor. We provide first evidence that K5 possesses strong inhibitory properties against both tumor-infiltrating endothelial cells as well as tumor-associated macrophages, 2 of the key cellular mediators implicated in tumor angiogenesis.

Leukocyte infiltration into tumors is analogous to the wound healing process, since recruited inflammatory cells are programmed by the tumor to divert to a wound­ healing phenotype (231). Thus, cancer is often referred to as "the wound that will not heal." Among leukocytes that are attracted toward the tumor, macrophages are the most multifunctional and can influence several processes such as angiogenesis, matrix degradation and cytotoxicity. Tumor-associated macrophages (TAMs) enter the tumor vasculature as blood monocytes and migrate to avascular areas of the tumor, differentiate and c1uster in hotspots (231). It has been reported that increased macrophage density is associated with increased angiogenesis and decreased survival rates (206). Recruitment of T AM has been inhibited by drugs such as linomide and thalidomide known to also possess anti-angiogenic properties (287). In fact, Moulton et al. (210) observed that angiostatin reduces plaque angiogenesis and atherosc1erosis by reducing plaque­ associated macrophages. Thus, our finding that soluble K5 protein exhibits inhibitory

163 action on two distinct cell types is consistent with what has been reported for other angiostatic agents. Furthennore, the ability of soluble K5 to directly inhibit macrophage migration suggests that K5 may interact with a cell-surface receptor expressed on macrophages that leads to a decrease in macrophage motility.

Moreover, we demonstrate that soluble K5 protein mediates a potent anti-tumor response in an orthotopic glioma mode1 by significantly prolonging long-tenn survival in test animaIs. This validates the concept that tumor cells can be retrovirally gene- engineered ex vivo to provide sustained expression of K5 protein within the loco-regional tumor microenvironment in order to induce tumor regression. Our findings also reveal that K5 protein mediates its anti-tumoral activity via 2 unique cell types implicated in tumor angiogenesis.

Our study was the first to test the efficacy of plasminogen K5 protein in a gene therapy setting for the treatment of cancer. Angiostatin, the most commonly studied internaI plasminogen kringle fragment, has been tested within the context of gene therapy strategies. Joseph et al. (288) assessed the ability of intravenously injected recombinant adenovirus encoding human KI-3 protein (coupled to to increase the in vivo half-life) and failed to demonstrate an anti-tumor effect in human neuroblastoma xenografts, despite high circulating leve1s of KI-3 protein. Similar to the human U87 glioma modei empIoyed by our laboratory, the neuroblastoma mode1 utilized by Joseph et al. also secretes high levels of VEGF and their findings demonstrate that angiostatin is unable to reverse the pro-angiogenic phenotype of the tumor. Interestingly, it was recently shown that delivery of full-iength plasminogen kringles KI-5 by a single adenoviral intratumorai injection resuited in 40% tumor regression, arrested tumor

164 growth by 90% and induced substantial reduction in neovascularization in a human breast carcinoma model, whereas KI-3 did not lead to an anti-tumor effect (289). Overall, these findings support the notion that the K5 domain enhances the potency of angiostatin and that it can be utilized on its own as an effective anti-tumor protein using a retroviral gene therapy approach.

5.5.2 Hypothesis 2

Our finding that K5 protein acts as a 2-pronged anti-tumor agent inhibiting both endothelial cells and macrophages, prompted our laboratory to investigate whether K5 mediated its anti-tumor effects by modulating other immune cell types. Angiostatin has been shown to possess immune modulatory properties by enhancing leukocyte-vessel interactions via upregulation of endothelial cell adhesion molecules (212). Angiostatin has also been shown to act as an anti-adhesive/anti-inflammatory agent in the setting of experimental peritonitis (214). Therefore, we assessed the immune modulatory activity of soluble K5 protein employing a murine mammary adenocarcinoma model in the context of a gene therapy strategy. We demonstrate that:

i) Retroviral gene-engineered DN3 mammary adenocarcinoma cells can

secrete soluble K5 protein;

ii) Soluble K5 protein relies upon the innate immune system to exert an anti­

tumor effect in vivo;

iii) Expression of soluble K5 protein by gene-modified DN3 cells within the

tumor microenvironment leads to the recruitment of CD3+ lymphoid cells,

more specifically natural killer T -lymphocytes;

165 iv) Expression of soluble K5 protein by gene-modified DAl3 cells within the

tumor microenvironment reduces the quantity and size of tumor-associated

vasculature;

v) Soluble K5 protein induces massive neutrophilic infiltration within the in

vivo tumor microenvironment and;

vi) Purified K5 protein acts as a chemoattractant as well as promotes

neutrophil activation in vitro.

Our findings confirm that retroviral gene modification of murine mammary tumor cells leads to functional production of K5 protein in order to induce an anti-cancer response in vivo. Consistent with our previous study, we confirm that K5 acts as a potent anti­ angiogenic agent capable of reducing the quantity and size of blood vessels within the tumor microenvironment. Interestingly, we observed that soluble K5 protein is dependent upon an intact innate immune system to exert an effective long-term anti-tumor response.

Despite the significant delay in the survival rate of K5-containing immunodeficient

NOD-SCID and BALB/c nude mi ce as compared to control mi ce, aU animaIs eventually developed large tumors that exceeded acceptable endpoints and required euthanasia.

Soluble K5 induced long-term survival only in immunocompetent BALB/c mice, suggesting the requirement of functional innate immune effector cells. More specifically, preferential recruitment of CD3+ lymphoid cells, with predominance for the natural killer

T-Iymphocyte subset, was detected in K5-containing tumor explants. Further analysis revealed that soluble K5 also induced massive neutrophilic infiltration within the tumor microenvironment. We provide evidence that K5 protein acts as a strong neutrophil

166 chemoattractant and promotes neutrophil activation VIa upregulation of CD64 granulocyte cell-surface marker.

Overall, we demonstrate that in addition to the ability of K5 protein to act as a potent angiostatic agent, K5 also recruits CD3+ lymphocytes as well as neutrophils to cause in vivo tumor regression. Thus, K5 mediates its anti-tumor effect via anti­ angiogenic and pro-inflammatory modes of action. It has been previously reported that combined lymphocytic and neutrophilic infiltration leads to tumor rejection (225-227).

5.5.3 Hypothesis 3

In our first study, we observed that soluble K5 protein suppressed the recruitment of tumor-associated macrophages in vivo as well as prevented the migration of human monocyte-derived macrophages in vitro. Given this novel finding, we sought to determine whether angiostatin, another internaI plasminogen kringle fragment possessing angiostatic activity, also affects macrophages. The development of strategies to inhibit

T AMs is warranted since T AMs play a key role in tumor progression and metastasis. Our results demonstrate that:

i) Human recombinant angiostatin (KI-3) protein inhibits GM-CSF-induced

migration of murine peritoneal macrophages;

ii) Kl-3 protein inhibits migration of human monocyte-derived CD206+

macrophages and;

iii) KI-3 protein decreases the gelatinolytic activity of macrophage-derived

MMP-9.

Our demonstration that angiostatin blocks macrophage migration is consistent with what has been observed by Benelli et al. (253), who report that angiostatin inhibits MCP-l-

167 induced monocyte migration. We speculated that macrophages express a cell-surface receptor that angiostatin interacts with in order to mediate its anti-migratory effect. In fact, we found that macrophages express the ~-subunit of ATP synthase on the cell­ surface. ATP synthase was identified as the first possible receptor for angiostatin expressed on the cell-surface of endothelial cells (210), and more recently detected on the cell-surface oftumors in a low extracellular pH environment (110). Previous reports have also shown that angiostatin suppresses neutrophil migration in vitro (253) and inhibits integrin-mediated leukocyte adhesion (214). Therefore, similar to our observations with

K5 protein, angiostatin also appears to possess multiple cell targets. Additionally, we observed a KI-3-mediated decrease in the ge1atinolytic activity of macrophage-produced

MMP-9, which has been associated with leukocyte transmigration (254) and thus, may be linked with the ability of Kl-3 to block macrophage migration. These results suggest that angiostatin could be exploited to inhibit tumor-associated macrophages and its utility could also be extended to target M2-phenotypic macrophages implicated in tissue remode1ing or immune regulation (258).

5.6 Summary of Findings

In conclusion, the data presented in this thesis demonstrates that human plasminogen K5 protein can be efficiently secreted by retrovirally gene-engineered tumor cells to exert an anti-tumor effect in vivo in animal models of brain and breast cancer.

Functionally, we have shown that soluble K5 protein acts as a potent anti-angiogenic agent by its inhibition of growth factor-induced endothe1ial cell migration in vitro and its ability to suppress tumor-induced vascularization in vivo. Furthermore, we have

168 discovered additional novel properties for soluble K5 protein that can be exploited therapeutically, including its ability to block tumor-associated macrophages, preferentially recroit CD3+ lymphoid cells (natural killer T -cells in particular) and act as a strong chemoattractant and activator for tumor-infiltrating neutrophils. Moreover, we have demonstrated that human plasminogen Kl-3 protein blocks the migration of macrophages and leads to a decrease in the gelatinolytic activity of macrophage-derived

MMP-9.

5. 7 Future Directions

In view of these findings, it would be interesting to further identify and characterize binding receptors and downstream signaling mediators for both K5 and K 1-

3. A broader understanding of their modes of action on leukocytes might expand the therapeutic utility of these proteins, for instance in the setting of atherosclerosis. It would also be worthy to design combinatorial or fusion anti-angiogenic transgenes with differing targets in order to maximize therapeutic benefit such as K5 fused to soluble

FLTI (which is an endogenous inhibitor ofVEGFRl). Phase III clinical trials using anti­

VEGF agents have demonstrated that tumors can escape the anti-VEGF effect by producing other pro-angiogenic agents to recroit blood vessels (279). It has also been shown that multi-targeted tyrosine kinase inhibitors such as sunitinib (which targets

VEGFR2, PDGFR-~, FLT3 and c-Kit) that target several pathways implicated in both tumor growth and angiogenesis can be utilized as monotherapies to increase patient survival (279). Finally, anti-angiogenic gene therapies could be utilized as adjuvants with conventional treatment or in combination with CUITent treatment modalities such as

169 chemotherapy and radiation therapy. In fact, it was proposed by Teicher et al. (290) that combining anti-angiogenic therapy with chemotherapy would induce a synergistic effect, since both tumor and endothe1ial cells would be targeted. Although a number of preclinical studies have shown that angiostatic agents reduce the number of tumor­ associated blood vesse1s and antagonize the effect of chemotherapy and radiation therapy by impeding delivery (291), it has also been demonstrated that anti-VEGF therapies

"normalize" tumor vasculature (291, 292). Vesse1 normalization induced by anti-VEGF agents like bevacizumab reverses sorne of the abnormalities possessed by tumor blood vessels by suppressing vesse1 leakiness and pruning immature growing blood vesse1s

(291). As a result, normalization of tumor vessels creates mostly mature functional vesse1s and thus, the delivery of drugs and oxygen can be increased and ultimate1y lead to improved outcome (279).

The studies presented III this thesis served to test the proof-of-concept that retroviral gene modification of tumor cells can lead to the production of biologically relevant levels of an anti-angiogenic agent in order to induce in vivo tumor regression.

However, for clinical purposes, an autologous cellular de1ivery vehic1e, such as bone marrow stroma, could be employed. Our laboratory has validated the process of retrovirally gene-modifying marrow stromal cells for the in vivo delivery of therapeutic proteins in preclinical models ofmelanoma (203) and anemia (293).

5.8 Concluding Remarks

Overall, our findings extend the existing body of knowledge and propose novel avenues for the development of anti-angiogenic strategies against cancer. Our data also

170 highlights the importance of further investigating the immune modulatory properties of anti-angiogenic agents in order to enhance their anti-tumor effect.

171 CONTRIBUTION TO ORIGINAL

KNOWLEDGE

172 CONTRIBUTION TO ORIGINAL KNOWLEDGE

The data presented in this thesis has provided several original contributions to the existing body of scientific knowledge: vi) Soluble human plasminogen (PIg) kringle 5 (K5) protein spontaneously adopts a

tertiary structure consistent with native conformation as displayed within Plg

when expressed by genetically-engineered eukaryotic cells. vii) Human U87 glioma cells and murine DN3 mammary adenocarcinoma cells can

be retrovirally gene-modified to secrete functional K5 protein. viii) Soluble K5 protein inhibits growth factor-stimulated endothelial cell migration in

vitro. lx) Soluble K5 protein blocks glioma and mammary adenocarcinoma-associated

angiogenesis in vivo by reducing the quantity and Slze of tumor-associated

vasculature. x) Soluble K5 protein possesses novel anti-macrophage properties, capable of

suppressmg in vitro migration of human monocyte-derived macrophages and

prevents the recruitment ofhost-derived tumor-infiltrating macrophages in vivo. xi) Glioma-targeted K5 expression suppresses tumor growth and prolongs survival in

an experimental orthotopic brain cancer model. xii) Soluble K5 protein is dependent upon the innate immune effector cells to exert an

anti-tumor effect in vivo.

173 xiii) Expression of soluble K5 protein by gene-modified DAl3 cells within the tumor

microenvironment leads to the recruitment of CD3+ lymphoid cells and more

specifically natural killer T-Iymphocytes. xiv) Soluble K5 protein induces robust neutrophilic infiltration within the tumor

microenvironment. xv) Purified K5 protein acts as a chemoattractant as well as promotes neutrophil

activation in vitro. xvi) Human recombinant angiostatin (KI-3) protein inhibits GM-CSF-induced

migration of murine peritoneal macrophages and blocks migration of hum an

monocyte-derived CD206+ macrophages. xvii) KI-3 protein decreases the ge1atinolytic activity of macrophage-derived MMP-9.

174 REFERENCES

175 References

1 Carmeliet,P. Mechanisms ofangiogenesis and arteriogenesis, Nat.Med., 6: 389- 395,2000.

2 Peichev,M., Naiyer,AJ., Pereira,D., Zhu,Z., Lane,W.J., Williams,M., Oz,M.C., Hicklin,D.J., Witte,L., Moore,M.A. and Rafii,S. Expression ofVEGFR-2 and AC133 by circulating human CD34(+) cells identifies a population offunctional endothelial precursors, Blood, 95: 952-958, 2000.

3 Takahashi,T., Kalka,C., Masuda,H., Chen,D., Silver,M., Keamey,M., Magner,M., Isner,J.M. and Asahara,T. Ischemia- and cytokine-induced mobilization ofbone marrow-derived endothelial progenitor cells for neovascularization, Nat.Med.,5: 434-438, 1999.

4 Patan,S. Vasculogenesis and Angiogenesis. In: M. Kirsch and P. McL.Black (eds.), Angiogenesis in Brain Tumors, pp. 3-32. Kluwer Academic Publishers: Boston, 2004.

5 Gyorffy,S., Gauldie,J., Stewart,A. and Wen,x.-Y. Antiangiogenic gene therapy of cancer. In: D. Curiel and J. Douglas (eds.), Cancer Gene Therapy, pp. 261-271. Humana Press Inc.: Totowa, NJ, 2005.

6 Yancopoulos,G.D., Davis,S., Gale,N.W., Rudge,J.S., Wiegand,S.J. and Holash,J. Vascular-specific growth factors and blood vessel formation, Nature, 407: 242- 248,2000.

7 Maisonpierre,P.c., Suri,C., Jones,P.F., Bartunkova,S., Wiegand,S.J., Radziejewski,C., Compton,D., McClain,J., Aldrich,T.H., Papadopoulos,N., Daly,T.J., Davis,S., Sato,T.N. and Yancopoulos,G.D. Angiopoietin-2, a natural antagonist for Tie2 that disrupts in vivo angiogenesis, Science, 277: 55-60, 1997.

8 Bergers,G. and Benjamin,L.E. Tumorigenesis and the angiogenic switch, Nat.Rev.Cancer, 3: 401-410, 2003.

9 Coussens,L.M., Raymond,W.W., Bergers,G., Laig-Webster,M., Behrendtsen,O., Werb,Z., Caughey,G.H. and Hanahan,D. Inflammatory mast cells up-regulate angiogenesis during squamous epithelial carcinogenesis, Genes Dev., 13: 1382- 1397, 1999.

10 Vu,T.H., Shipley,J.M., Bergers,G., Berger,J.E., Helms,J.A., Hanahan,D., Shapiro,S.D., Senior,R.M. and Werb,Z. MMP-9/gelatinase B is a key regulator of growth plate angiogenesis and apoptosis ofhypertrophic chondrocytes, Cell, 93: 411-422, 1998.

Il Bajou,K., Noel,A., Gerard,R.D., Masson,V., Brunner,N., Holst-Hansen,C., Skobe,M., Fusenig,N.E., Carmeliet,P., Collen,D. and Foidart,J.M. Absence of

176 host plasminogen activator inhibitor 1 prevents cancer invasion and vascularization, Nat.Med., 4: 923-928, 1998.

12 Brooks,P.C., Silletti,S., von Schalscha,T.L., Friedlander,M. and Cheresh,D.A. Disruption of angiogenesis by PEX, a noncatalytic metalloproteinase fragment with integrin binding activity, Cell, 92: 391-400, 1998.

13 Ferrara,N. Role ofvascular endothelial growth factor in the regulation of angiogenesis, Kidney lnt., 56: 794-814, 1999.

14 Taipale,J., Makinen,T., Arighi,E., Kukk,E., Karkkainen,M. and Alitalo,K. Vascular endothelial growth factor receptor-3, Curr.Top.Microbiol.Immunol., 237: 85-96, 1999.

15 Soker,S., Takashima,S., Miao,H.Q., Neufeld,G. and Klagsbrun,M. Neuropilin-l is expressed by endothelial and tumor cells as an isoform-specific receptor for vascular endothelial growth factor, Cell, 92: 735-745, 1998.

16 Gale,N.W. and Yancopoulos,G.D. Growth factors acting via endothelial cell­ specific receptor tyrosine kinases: VEGFs, , and ephrins in vascular development, Genes Dev., 13: 1055-1066, 1999.

17 Suri,C., McClain,J., Thurston,G., McDonald,D.M., Zhou,H., Oldmixon,E.H., Sato,T.N. and Yancopoulos,G.D. Increased vascularization in mice overexpressing angiopoietin-l, Science, 282: 468-471, 1998.

18 Vamer,J.A., Brooks,P.C. and Cheresh,D.A. REVIEW: the integrin alpha V beta 3: angiogenesis and apoptosis, Cell Adhes.Commun., 3: 367-374, 1995.

19 Thurston,G., Rudge,J.S., Ioffe,E., Zhou,H., Ross,L., Croll,S.D., Glazer,N., Holash,J., McDonald,D.M. and Yancopoulos,G.D. Angiopoietin-l protects the adult vasculature against plasma leakage, Nat.Med., 6: 460-463, 2000.

20 Carmeliet,P., Lampugnani,M.G., Moons,L., Breviario,F., Compemolle,V., Bono,F., Ba1coni,G., Spagnuolo,R., Oostuyse,B., Dewerchin,M., Zanetti,A., Angellilo,A., Mattot,V., Nuyens,D., Lutgens,E., Clotman,F., de Ruiter,M.C., Gittenberger-de,G.A., Poelmann,R., Lupu,F., Herbert,J.M., Collen,D. and Dejana,E. Targeted deficiency or cytosolic truncation ofthe VE-cadherin gene in mice impairs VEGF-mediated endothelial survival and angiogenesis, Cell,98: 147-157, 1999.

21 Carmeliet,P., Ng,Y.S., Nuyens,D., Theilmeier,G., Brusselmans,K., Comelissen,I., Ehler,E., Kakkar,V.V., Stalmans,I., Mattot,V., Perriard,J.C., Dewerchin,M., Flameng,W., Nagy,A., Lupu,F., Moons,L., Collen,D., D'Amore,P.A. and Shima,D.T. Impaired myocardial angiogenesis and ischemic cardiomyopathy in mice lacking the vascular endothelial growth factor isoforms VEGF164 and VEGFI88, Nat.Med., 5: 495-502, 1999.

177 22 Gerber,H.P., Hillan,K.J., Ryan,A.M., Kowalski,J., Keller,G.A., Rangell,L., Wright,B.D., Radtke,F., Aguet,M. and Ferrara,N. VEGF is required for growth and survival in neonatal mice, Development, 126: 1149-1159, 1999.

23 Jain,R.K., Safabakhsh,N., Sckell,A., Chen,Y., Jiang,P., Benjamin,L., Yuan,F. and Keshet,E. Endothelial cell death, angiogenesis, and microvascular function after castration in an androgen-dependent tumor: role of vascular endothelial growth factor, Proc.Natl.Acad.Sci.U.S.A, 95: 10820-10825, 1998.

24 Alon,T., Hemo,l., Itin,A., Pe'er,J., Stone,J. and Keshet,E. Vascular endothelial growth factor acts as a survival factor for newly formed retinal vessels and has implications for retinopathy of prematurity, Nat.Med., 1: 1024-1028, 1995.

25 Lindahl,P., Hellstrom,M., Kalen,M. and Betsholtz,C. Endothelial-perivascular cell signaling in vascular development: lessons from knockout mice, Curr.Opin.Lipidol., 9: 407-411, 1998.

26 Helmlinger,G., Yuan,F., Dellian,M. and Jain,R.K. Interstitial pH and p02 gradients in solid tumors in vivo: high-resolution measurements reveal a lack of correlation, N at.Med., 3: 177-182, 1997.

27 Benjamin,L.E., Golijanin,D., Itin,A., Pode,D. and Keshet,E. Selective ablation of immature blood vessels in established human tumors follows vascular endothelial growth factor withdrawal, J.Clin.Invest, 103: 159-165, 1999.

28 Morikawa,S., Baluk,P., Kaidoh,T., Haskell,A., Jain,R.K. and McDonald,D.M. Abnormalities in pericytes on blood vessels and endothelial sprouts in tumors, Am.J.Pathol., 160: 985-1000,2002.

29 Benjamin,L.E., Hemo,1. and Keshet,E. A plasticity window for blood vessel remodelling is defined by pericyte coverage of the preformed endothelial network and is regulated by PDGF-B and VEGF, Development, 125: 1591-1598, 1998.

30 Folberg,R., Hendrix,M.J. and Maniotis,A.J. Vasculogenic mimicry and tumor angiogenesis, Am.J.Pathol., 156: 361-381, 2000.

31 Maniotis,A.J., Folberg,R., Hess,A., Seftor,E.A., Gardner,L.M., Pe'er,J., Trent,J.M., Meltzer,P.S. and Hendrix,M.J. Vascular channel formation by human melanoma cells in vivo and in vitro: vasculogenic mimi cry, Am.J.Pathol.,155: 739-752, 1999.

32 McDonald,D.M., Munn,L. and Jain,R.K. Vasculogenic mimi cry: how convincing, how novel, and how significant?, Am.J.Pathol., 156: 383-388,2000.

33 Hendrix,M.J., Seftor,E.A., Hess,A.R. and Seftor,R.E. Vasculogenic mimicry and tumour-cell pl asti city: les sons from melanoma, Nat.Rev.Cancer, 3: 411-421, 2003.

178 34 Tandle,A., Blazer,D.G., III and Libutti,S.K. Antiangiogenic gene therapy of cancer: recent developments, J.Transl.Med., 2: 22, 2004.

35 Makitie,T., Summanen,P., Tarkkanen,A. and Kivela,T. Microvascular loops and networks as prognostic indicators in choroidal and ciliary body melanomas, J.Natl.Cancer Inst., 91: 359-367, 1999.

36 EIlerby,H.M., Arap,W., EIlerby,L.M., Kain,R., Andrusiak,R., Rio,G.D., Krajewski,S., Lombardo,C.R., Rao,R., Ruoslahti,E., Bredesen,D.E. and Pasqualini,R. Anti-cancer activity oftargeted pro-apoptotic peptides, Nat.Med., 5: 1032-1038, 1999.

37 Hanahan,D. and Folkman,J. Patterns and emerging mechanisms of the angiogenic switch during tumorigenesis, CeIl, 86: 353-364, 1996.

38 Black,W.C. and Welch,H.G. Advances in diagnostic imaging and overestimations of disease prevalence and the benefits oftherapy, N.Engl.J.Med.,328: 1237- 1243, 1993.

39 Kerbel,R. and Folkman,J. Clinical translation of angiogenesis inhibitors, Nat.Rev.Cancer, 2: 727-739,2002.

40 Rak,J., Mitsuhashi,Y., Sheehan,C., Tamir,A., Viloria-Petit,A., Filmus,J., Mansour,SJ., Ahn,N.G. and Kerbel,R.S. Oncogenes and tumor angiogenesis: differential modes of vascular endothelial growth factor up-regulation in ras­ transformed epithelial cells and fibroblasts, Cancer Res., 60: 490-498, 2000.

41 Relf,M., LeJeune,S., Scott,P.A., Fox,S., Smith,K., Leek,R., Moghaddam,A., Whitehouse,R., BickneIl,R. and Harris,A.L. Expression of the angiogenic factors vascular endothelial cell growth factor, acidic and basic fibroblast growth factor, tumor growth factor beta-l, platelet-derived endothelial cell growth factor, placenta growth factor, and pleiotrophin in human primary breast cancer and its relation to angiogenesis, Cancer Res., 57: 963-969, 1997.

42 Semenza,G.L. Hypoxia-inducible factor 1: master regulator of 02 homeostasis, Curr.Opin.Genet.Dev., 8: 588-594, 1998.

43 Carmeliet,P., Dor,Y., Herbert,J.M., Fukumura,D., Brusselmans,K., Dewerchin,M., Neeman,M., Bono,F., Abramovitch,R., MaxweIl,P., Koch,C.J., Ratc1iffe,P., Moons,L., Jain,R.K., Collen,D. and Keshert,E. Role ofHIF-lalpha in hypoxia-mediated apoptosis, cell proliferation and tumour angiogenesis, Nature, 394: 485-490, 1998.

44 Fukumura,D., Xavier,R., Sugiura,T., Chen,Y., Park,E.C., Lu,N., Selig,M., Nielsen,G., Taksir,T., Jain,R.K. and Seed,B. Tumor induction ofVEGF promoter activity in stromal ceIls, CeIl, 94: 715-725, 1998.

179 45 Shi,Q., Rafii,S., Wu,M.H., Wijelath,E.S., Yu,C., Ishida,A., Fujita,Y., Kothari,S., Mohle,R., Sauvage,L.R., Moore,M.A., Storb,R.F. and Hammond,W.P. Evidence for circulating bone marrow-derived endothelial cells, Blood, 92: 362-367, 1998.

46 Vajkoczy,P., Farhadi,M., Gaumann,A., Heidenreich,R., Erber,R., Wunder,A., Tonn,J.C., Menger,M.D. and Breier,G. Microtumor growth initiates angiogenic sprouting with simultaneous expression ofVEGF, VEGF receptor-2, and angiopoietin-2, J.Clin.Invest, 109: 777-785,2002.

47 Holash,J., Wiegand,S.J. and Yancopoulos,G.D. New model oftumor angiogenesis: dynamic balance between vessel regression and growth mediated by angiopoietins and VEGF, Oncogene, 18: 5356-5362, 1999.

48 Holash,J., Maisonpierre,P.C., Compton,D., Boland,P., Alexander,C.R., Zagzag,D., Yancopoulos,G.D. and Wiegand,S.J. Vessel cooption, regression, and growth in tumors mediated by angiopoietins and VEGF, Science, 284: 1994- 1998, 1999.

49 Bello,L., Giussani,C., Carrabba,G., Pluderi,M., Costa,F. and & Bikfalvi,A. Angiogenesis and Invasion in Gliomas. In: M. Kirsch and P. M. Black (eds.), pp. 263-284. Kluwer: Norwell, 2004.

50 Radovanovic,I., D'Angelo,M. and Aguzzi,A. Angiogenesis in Transgenic Models of Multistep Angiogenesis. In: M. Kirsch and P. McL.Black (eds.), pp. 97-114. Kluwer Academic Publishers: Boston, 2004.

51 Inoue,M., Hager,J.H., Ferrara,N., Gerber,H.P. and Hanahan,D. VEGF-A has a critical, nonredundant role in angiogenic switching and pancreatic beta cell carcinogenesis, Cancer Cell, 1: 193-202, 2002.

52 Bergers,G., Brekken,R., McMahon,G., Vu,T.H., Itoh,T., Tamaki,K., Tanzawa,K., Thorpe,P., Itohara,S., Werb,Z. and Hanahan,D. Matrix metalloproteinase-9 triggers the angiogenic switch during carcinogenesis, Nat.Cell Biol., 2: 737-744, 2000.

53 Rodriguez-Manzaneque,J.C., Lane,T.F., Ortega,M.A., Hynes,R.O., Lawler,J. and lruela-Arispe,M.L. Thrombospondin-l suppresses spontaneous tumor growth and inhibits activation of matrix metalloproteinase-9 and mobilization of vascular endothelial growth factor, Proc.Natl.Acad.Sci.U.S.A,98: 12485-12490,2001.

54 Zabrenetzky,V., Harris,C.C., Steeg,P.S. and Roberts,D.D. Expression ofthe extracellular matrix molecule thrombospondin inversely correlates with malignant progression in melanoma, lung and breast carcinoma celllines, InU.Cancer,59: 191-195, 1994.

55 Watnick,R.S., Cheng,Y.N., Rangarajan,A., Ince,T.A. and Weinberg,R.A. Ras modulates Myc activity to repress thrombospondin-l expression and increase tumor angiogenesis, Cancer Cell, 3: 219-231,2003.

180 56 Pelengaris,S., Khan,M. and Evan,G.1. Suppression of Myc-induced apoptosis in beta cells exposes multiple oncogenic properties ofMyc and triggers carcinogenic progression, Cell, 109: 321-334, 2002.

57 Funatsu,H., Yamashita,H., Noma,H., Mochizuki,H., Mimura,T., Ikeda,T. and Hori,S. Outcome ofvitreous surgery and the balance between vascular endothelial growth factor and endostatin, Invest Ophthalmol.Vis.Sci., 44: 1042-1047,2003.

58 Holmgren,L., O'Reilly,M.S. and Folkman,J. Dormancy ofmicrometastases: balanced proliferation and apoptosis in the presence of angiogenesis suppression, Nat.Med., 1: 149-153, 1995.

59 O'Reilly,M.S., Holmgren,L., Shing,Y., Chen,C., Rosenthal,R.A., Moses,M., Lane,W.S., Cao,Y., Sage,E.H. and Folkman,J. Angiostatin: a novel angiogenesis inhibitor that mediates the suppression of metastases by a Lewis lung carcinoma, Cell, 79: 315-328, 1994.

60 Soff,G.A. Angiostatin and angiostatin-related proteins, Cancer Metastasis Rev., 19: 97-107, 2000.

61 Cao,Y. and Xue,L. Angiostatin, Semin.Thromb.Hemost., 30: 83-93,2004.

62 Moser,T.L., Stack,M.S., Asplin,I., Enghild,J.J., Hojrup,P., Everitt,L., Hubchak,S., Schnaper,H.W. and Pizzo,S.V. Angiostatin binds ATP synthase on the surface of human endothelial cells, Proc.Natl.Acad.Sci.U.S.A, 96: 2811-2816, 1999.

63 O'Reilly,M.S. Angiostatin: an endogenous inhibitor ofangiogenesis and oftumor growth, EXS, 79: 273-294, 1997.

64 Troyanovsky,B., Levchenko,T., Mansson,G., Matvijenko,O. and Holmgren,L. Angiomotin: an angiostatin binding protein that regulates endothelial cell migration and tube formation, J.Cell Biol., 152: 1247-1254,2001.

65 Dixelius,J., Larsson,H., Sasaki,T., Holmqvist,K., Lu,L., Engstrom,A., Timpl,R., Welsh,M. and Claesson-Welsh,L. Endostatin-induced tyrosine kinase signaling through the Shb adaptor protein regulates endothelial cell apoptosis, Blood,95: 3403-3411, 2000.

66 O'Reilly,M.S., Boehm,T., Shing,Y., Fukai,N., Vasios,G., Lane,W.S., Flynn,E., Birkhead,J.R., Olsen,B.R. and Folkman,J. Endostatin: an endogenous inhibitor of angiogenesis and tumor growth, Cell, 88: 277-285, 1997.

67 Kamphaus,G.D., Colorado,P.C., Panka,D.J., Hopfer,H., Ramchandran,R., Torre,A., Maeshima,Y., Mier,J.W., Sukhatme,V.P. and Kalluri,R. Canstatin, a novel matrix-derived inhibitor of angiogenesis and tumor growth, J.Biol.Chem., 275: 1209-1215,2000.

181 68 Hurwitz,H., Fehrenbacher,L., Novotny,W., Cartwright,T., Hainsworth,J., Heim,W., Berlin,J., Baron,A., Griffing,S., Holmgren,E., Ferrara,N., Fyfe,G., Rogers,B., Ross,R. and Kabbinavar,F. Bevacizumab plus irinotecan, fluorouracil, and leucovorin for metastatic colorectal cancer, N.Engl.J.Med., 350: 2335-2342, 2004.

69 Gutheil,J.C., Campbell,T.N., Pierce,P.R., Watkins,J.D., Huse,W.D., Bodkin,DJ. and Cheresh,D.A. Targeted antiangiogenic therapy for cancer using Vitaxin: a humanized monoclonal antibody to the integrin alphavbeta3, Clin.Cancer Res., 6: 3056-3061,2000.

70 Boehm,T., Folkman,J., Browder,T. and O'Reilly,M.S. Antiangiogenic therapy of experimental cancer does not induce acquired drug resistance, Nature, 390: 404- 407, 1997.

71 Kerbel,R.S., Viloria-Petit,A., Okada,F. and Rak,J. Establishing a link between oncogenes and tumor angiogenesis, Mol.Med., 4: 286-295, 1998.

72 Arbiser,J.L., Yeung,R., Weiss,S.W., Arbiser,Z.K., Amin,M.B., Cohen,C., Frank,D., Mahajan,S., Herron,G.S., Yang,J., Onda,H., Zhang,H.B., Bai,X., Uhlmann,E., Loehr,A., Northrup,H., Au,P., Davis,!., Fisher,D.E. and Gutmann,D.H. The generation and characterization of a cellline derived from a sporadic renal angiomyolipoma: use of telomerase to obtain stable populations of cells from benign neoplasms, AmJ.Pathol., 159: 483-491, 2001.

73 Arbiser,J.L., Moses,M.A., Femandez,C.A., Ghiso,N., Cao,Y., Klauber,N., Frank,D., Brownlee,M., Flynn,E., Parangi,S., Byers,H.R. and Folkman,J. Oncogenic H-ras stimulates tumor angiogenesis by two distinct pathways, Proc.Natl.Acad.Sci.U.S.A, 94: 861-866,1997.

74 Rak,J., Yu,J.L., Kerbel,R.S. and Coomber,B.L. What do oncogenic mutations have to do with angiogenesis/vascular dependence oftumors?, Cancer Res., 62: 1931-1934, 2002.

75 Udagawa,T., Femandez,A., Achilles,E.G., Folkman,J. and D'Amato,R.J. Persistence of microscopic human cancers in mice: alterations in the angiogenic balance accompanies loss oftumor dormancy, FASEB J, 16: 1361-1370,2002.

76 Femandez,A., Udagawa,T., Schwesinger,C., Beecken,W., chilles-Gerte,E., McDonnell,T. and D'Amato,R. Angiogenic potential of prostate carcinoma cells overexpressing bcl-2, J Natl.Cancer Inst., 93: 208-213, 2001.

77 Ravi,R., Mookerjee,B., Bhujwalla,Z.M., Sutter,C.H., Artemov,D., Zeng,Q., Dillehay,L.E., Madan,A., Semenza,G.L. and Bedi,A. Regulation oftumor angiogenesis by p53-induced degradation ofhypoxia-inducible factor lalpha, Genes Dev., 14: 34-44, 2000.

182 78 Zhang,L., Yu,D., Hu,M., Xiong,S., Lang,A., Ellis,L.M. and Pollock,R.E. Wild­ type p53 suppresses angiogenesis in human leiomyosarcoma and synovial sarcoma by transcriptional suppression of vascular endothelial growth factor expression, Cancer Res., 60: 3655-3661, 2000.

79 Ciardiello,F., Caputo,R., Bianco,R., Damiano,V., Fontanini,G., Cuccato,S., De,P .S., Bianco,A.R. and Tortora,G. Inhibition of growth factor production and angiogenesis in human cancer cells by ZD1839 (Iressa), a selective epidermal growth factor receptor tyrosine kinase inhibitor, Clin. Cancer Res., 7: 1459-1465, 2001.

80 Tille,J.C., Wood,J., Mandriota,S.J., Schnell,C., Ferrari,S., Mestan,J., Zhu,Z., Witte,L. and Pepper,M.S. Vascular endothelial growth factor (VEGF) receptor-2 antagonists inhibit VEGF and basic fibroblast growth factor-induced angiogenesis in vivo and in vitro., J Pharmaco1.Exp.Ther., 299: 1073-1085,2001.

81 Hoekman K SU6668, a multitargeted angiogenesis inhibitor., Cancer J., 7: S134- S138,2001.

82 Brouty-Boye,D. and Zetter,B.R. Inhibition of cell motility by interferon, Science, 208: 516-518, 1980.

83 Singh,R.K., Gutman,M., Bucana,C.D., Sanchez,R., Llansa,N. and Fidler,l.J. Interferons alpha and beta down-regulate the expression ofbasic fibroblast growth factor in human carcinomas, Proc.Nat1.Acad.Sci.U.S.A, 92: 4562-4566, 1995.

84 Kaban,L.B., Mulliken,J.B., Ezekowitz,R.A., Ebb,D., Smith,P.S. and Folkman,J. Antiangiogenic therapy of a recurrent giant cell tumor of the mandible with interferon alfa-2a, Pediatrics, 103: 1145-1149, 1999.

85 MarIer,J.J., Rubin,J.B., Trede,N.S., Connors,S., Grier,H., Upton,J., Mulliken,J.B. and Folkman,J. Successful antiangiogenic therapy of giant cell angioblastoma with interferon alfa 2b: report of2 cases, Pediatries, 109: E37, 2002.

86 Cao,Y., Ji,R.W., Davidson,D., Schaller,J., Marti,D., Sohndel,S., McCance,S.G., O'Reilly,M.S., Llinas,M. and Folkman,J. Kringle domains ofhuman angiostatin. Characterization of the anti-proliferative activityon endothelial cells, J.Biol.Chem., 271: 29461-29467, 1996.

87 Folkman,J. Endogenous angiogenesis inhibitors, APMIS, 112: 496-507, 2004.

88 Geiger,J .H. and Cnudde,S.E. What the structure of angiostatin may tell us about its mechanism of action, J.Thromb.Haemost., 2: 23-34, 2004.

89 Stathakis,P., Fitzgerald,M., Matthias,L.J., Chesterman,C.N. and Hogg,P.J. Generation of angiostatin by reduction and proteolysis of plasmin. Catalysis by a plasmin reductase secreted by cultured cells, J.Bio1.Chem., 272: 20641-20645, 1997.

183 90 Wahl,M.L., Kenan,D.J., Gonzalez-Gronow,M. and Pizzo,S.V. Angiostatin's molecular mechanism: aspects of specificity and regulation elucidated, J .Cell Biochem., 96: 242-261, 2005.

91 Lay,A.J., Jiang,X.M., Kisker,O., Flynn,E., Underwood,A., Condron,R. and Hogg,P.J. Phosphoglycerate kinase acts in tumour angiogenesis as a disulphide reductase, Nature, 408: 869-873,2000.

92 Stathakis,P., Lay,AJ., Fitzgerald,M., Schlieker,C., Matthias,L.J. and Hogg,P.J. Angiostatin formation involves disulfide bond reduction and proteolysis in kringle 50fplasmin, J.Biol.Chem., 274: 8910-8916, 1999.

93 Soff,G.A., Hong,J., Fishman,D., Schultz,R., Cundiff,D., Park,S., Enghild,J., Stack,M.S. and Gately,S. Angiostatin4.5: a naturally occurring human angiogenesis inhibitor, Proceedings ofthe American Association for Cancer Research, 40: 4088, 1999.

94 Gately,S., Twardowski,P., Stack,M.S., Patrick,M., Boggio,L., Cundiff,D.L., Schnaper,H.W., Madison,L., Volpert,O., Bouck,N., Enghild,J., Kwaan,H.C. and Soff,G.A. Human prostate carcinoma cells express enzymatic activity that converts hum an plasminogen to the angiogenesis inhibitor, angiostatin, Cancer Res., 56: 4887-4890, 1996.

95 Gately,S., Twardowski,P., Stack,M.S., Cundiff,D.L., Grella,D., Castellino,F.J., Enghild,J., Kwaan,H.C., Lee,F., Kramer,R.A., Volpert,O., Bouck,N. and Soff,G.A. The mechanism of cancer-mediated conversion ofplasminogen to the angiogenesis inhibitor angiostatin, Proc.Natl.Acad.Sci.U.S.A, 94: 10868-10872, 1997.

96 Falcone,D.J., Khan,K.M., Layne,T. and Femandes,L. Macrophage formation of angiostatin during inflammation. A byproduct of the activation of plasminogen, J.Biol.Chem., 273: 31480-31485, 1998.

97 Heidtmann,H.H., Nettelbeck,D.M., Mingels,A., Jager,R., Welker,H.G. and Kontermann,R.E. Generation of angiostatin-like fragments from plasminogen by prostate-specific antigen, Br.J.Cancer, 81: 1269-1273, 1999.

98 Fortier,A.H., Holaday,J.W., Liang,H., Dey,C., Grella,D.K., Holland-Linn,J., Vu,H., Plum,S.M. and Nelson,B.J. Recombinant prostate specific antigen inhibits angiogenesis in vitro and in vivo, Prostate, 56: 212-219,2003.

99 Dong,Z., Kumar,R., Yang,x. and Fidler,I.J. Macrophage-derived metalloelastase is responsible for the generation of angiostatin in Lewis lung carcinoma, Cell, 88: 801-810, 1997.

100 Patterson,B.c. and Sang,Q.A. Angiostatin-converting enzyme activities ofhuman matrilysin (MMP-7) and gelatinase B/type IV collagenase (MMP-9), J.Biol.Chem., 272: 28823-28825, 1997.

184 101 Rejante,M.R., Byeon,I.J. and Llinas,M. Ligand specificity of hum an plasminogen kringle 4, Biochemistry, 30: 11081-11092, 1991.

102 Claesson-Welsh,L., Welsh,M., Ito,N., nand-Apte,B., Soker,S., Zetter,B., O'Reilly,M. and Folkman,J. Angiostatin induces endothelial cell apoptosis and activation of focal adhesion kinase independently of the integrin-binding motif RGD, Proc.Natl.Acad.Sci.U.S.A, 95: 5579-5583, 1998.

103 Stack,M.S., Gately,S., Bafetti,L.M., Enghild,J.J. and Soff,G.A. Angiostatin inhibits endothelial and melanoma cellular invasion by blocking matrix-enhanced plasminogen activation, Biochem.J., 340 (Pt 1): 77-84, 1999.

104 O'Reilly,M.S., Holmgren,L., Chen,C. and Folkman,J. Angiostatin induces and sustains dormancy ofhuman primary tumors in mice, Nat.Med., 2: 689-692, 1996.

105 Lannutti,BJ., Gately,S.T., Quevedo,M.E., Soff,G.A. and Paller,A.S. Human angiostatin inhibits murine hemangioendothelioma tumor growth in vivo, Cancer Res., 57: 5277-5280,1997.

106 DeMoraes,E.D., Fogler,W.E., Grant,D.S., Wahl,M.L., Leeper,D.B., Zrada,S., Malin,A., Connors,S., Fortier,A.H., Dabrow,M., Sidor,C. and Capizzi,R.L.Recombinant human angiostatin (rhA): A Phase 1 clinical trial assessing safety, pharmacokinetics (PK) and pharmacodynamcs (PD)., American Society of Clinical Oncology, abstract number 10 2001 (Abstract)

107 Das,B., Mondragon,M.O., Sadeghian,M., Hatcher,V.B. and Norin,AJ. A novel ligand in lymphocyte-mediated cytotoxicity: expression of the beta subunit ofH+ transporting ATP synthase on the surface oftumor celllines, J.Exp.Med., 180: 273-281, 1994.

108 Soltys,BJ. and Gupta,R.S. Mitochondrial proteins at unexpected cellular locations: export of proteins from mitochondria from an evolutionary perspective, Int.Rev.Cytol., 194: 133-196,2000.

109 Soltys,B.J. and Gupta,R.S. Mitochondrial-matrix proteins at unexpected locations: are they exported?, Trends Biochem.Sci., 24: 174-177, 1999.

110 Chi,S.L. and Pizzo,S.V. Angiostatin is directly cytotoxic to tumor cells at low extracellular pH: a mechanism dependent on cell surface-associated ATP synthase, Cancer Res., 66: 875-882,2006.

111 Burrell,H.E., Wlodarski,B., Foster,B.J., Buckley,K.A., Sharpe,G.R., Quayle,J.M., Simpson,A.W. and Gallagher,J.A. Human keratinocytes release ATP and utilize three mechanisms for nucleotide interconversion at the cell surface, J.Biol.Chem., 280: 29667-29676,2005.

185 112 Wilson-Fritch,L., Burkart,A., Bell,G., Mendelson,K., Leszyk,J., Nicoloro,S., Czech,M. and Corvera,S. Mitochondrial biogenesis and remode1ing during adipogenesis and in response to the insulin sensitizer rosiglitazone, Mol.Cell Biol., 23: 1085-1094,2003.

113 Scotet,E., Martinez,L.O., Grant,E., Barbaras,R., Jeno,P., Guiraud,M., Monsarrat,B., Saulquin,X., Maillet,S., Esteve,J.P., Lopez,F., Perret,B., Collet,X., Bonneville,M. and Champagne,E. Tumor recognition following V gamma9V delta2 T cell receptor interactions with a surface F1-ATPase-related structure and apolipoprotein A-l, Immunity., 22: 71-80, 2005.

114 Moser,T.L., Kenan,DJ., Ashley,T.A., Roy,J.A., Goodman,M.D., Misra,U.K., Cheek,D.J. and Pizzo,S.V. Endothelial cell surface F1-FO ATP synthase is active in ATP synthesis and is inhibited by angiostatin, Proc.Natl.Acad.Sci.U.S.A,98: 6656-6661,2001.

115 Moser,T.L., Stack,M.S., Wahl,M.L. and Pizzo,S.V. The mechanism of action of angiostatin: can you teach an old dog new tricks?, Thromb.Haemost., 87: 394- 401,2002.

116 Wahl,M.L., Owen,C.S. and Grant,D.S. Angiostatin induces intracellular acidosis and anoikis in endothelial cells at a tumor-like low pH, Endothelium, 9: 205-216, 2002.

117 Yamagata,M. and Tannock,LF. The chronic administration of drugs that inhibit the regulation ofintracellular pH: in vitro and anti-tumour effects, Br.J.Cancer, 73: 1328-1334, 1996.

118 Contarino,M.R., Fang,J., Pizzo,S.V. and Wahl,M.L.The anti-angiogenic potential of pH regulation inhibitors., Proceedings of the American Association for Cancer Research, 04-AB-1548-AACR 2004 (Abstract)

119 Tarui,T., Miles,L.A. and Takada,Y. Specifie interaction of angiostatin with integrin alpha(v)beta(3) in endothelial cells, J.Biol.Chem., 276: 39562-39568, 2001.

120 Bratt,A., Wilson,W.J., Troyanovsky,B., Aase,K., Kessler,R., Van Meir,E.G. and Holmgren,L. Angiomotin belongs to a novel protein family with conserved coiled-coil and PDZ binding domains, Gene, 298: 69-77, 2002.

121 Levchenko,T., Aase,K., Troyanovsky,B., Bratt,A. and Holmgren,L. Loss of responsiveness to chemotactic factors by deletion of the C-terminal protein interaction site ofangiomotin, J.Cell Sei., 116: 3803-3810,2003.

122 Levchenko,T., Bratt,A., Arbiser,J.L. and Holmgren,L. Angiomotin expression promotes hemangioendothelioma invasion, Oncogene,23: 1469-1473,2004.

186 123 Gonzalez-Gronow,M., Edelberg,J.M. and Pizzo,S.V. Further characterization of the cellular plasminogen binding site: evidence that plasminogen 2 and lipoprotein a compete for the same site, Biochemistry, 28: 2374-2377, 1989.

124 Gonzalez-Gronow,M., Grenett,H.E., Fuller,G.M. and Pizzo,S.V. The role of carbohydrate in the function of hum an plasminogen: comparison ofthe protein obtained from molecular cloning and expression in Escherichia coli and COS cells, Biochim.Biophys.Acta, 1039: 269-276, 1990.

125 Gonzalez-Gronow,M., Gawdi,G. and Pizzo,S.V. Characterization of the plasminogen receptors of normal and rheumatoid arthritis human synovial fibroblasts, J.Biol.Chem., 269: 4360-4366, 1994.

126 Stack,M.S., Moser,T.L. and Pizzo,S.V. Binding ofhuman plasminogen to basement-membrane (type IV) collagen, Biochem.J., 284 (Pt 1): 103-108, 1992.

127 Stack,M.S., Rinehart,A.R. and Pizzo,S.V. Comparison ofplasminogen binding and activation on extracellular matrices produced by vascular smooth muscle and endothelial ceIls, Eur.J.Biochem., 226: 937-943, 1994.

128 Cao,Y., Chen,A., An,S.S., Ji,R.W., Davidson,D. and Llinas,M. Kringle 5 of plasminogen is a novel inhibitor of endothelial cell growth, J.Biol.Chem., 272: 22924-22928, 1997.

129 Ji,W.R., Barrientos,L.G., Llinas,M., Gray,H., Villarreal,X., Deford,M.E., Castellino,F.J., Kramer,R.A. and Trail,P.A. Selective inhibition by kringle 5 of human plasminogen on endothelial cell migration, an important process in angiogenesis, Biochem.Biophys.Res.Commun., 247: 414-419, 1998.

130 LU,H., Dhanabal,M., Volk,R., Waterman,M.J., Ramchandran,R., Knebelmann,B., Segal,M. and Sukhatme,V.P. Kringle 5 causes cell cycle arrest and apoptosis of endothelial ceIls, Biochem.Biophys.Res.Commun., 258: 668-673, 1999.

131 Wu,H.L., Wu,I.S., Fang,R.Y., Hau,J.S., Wu,D.H., Chang,B.I., Lin,T.M. and Shi,G.Y. The binding ofplasminogen fragments to cultured human umbilical vein endothelial ceIls, Biochem.Biophys.Res.Commun., 188: 703-711, 1992.

132 Cao,R., Wu,H.L., Veitonmaki,N., Linden,P., Famebo,J., Shi,G.Y. and Cao,Y. Suppression of angiogenesis and tumor growth by the inhibitor KI-5 generated by plasmin-mediated proteolysis, Proc.Natl.Acad.Sci.U.S.A, 96: 5728-5733, 1999.

133 Ji,W.R., CasteIlino,FJ., Chang,Y., Deford,M.E., Gray,H., Villarreal,X., Kondri,M.E., Marti,D.N., Llinas,M., SchaIler,J., Kramer,R.A. and Trail,P.A. Characterization ofkringle domains of angiostatin as antagonists of endothelial cell migration, an important pro cess in angiogenesis, FASEB J., 12: 1731-1738, 1998.

187 134 Cai,W., Ma,J., Li,C., Yang,Z., Yang,x., Liu,W., Liu,Z., Li,M. and Gao,G. Enhanced anti-angiogenic effect ofa deletion mutant ofplasminogen kringle 5 on neovascularization, J.Cell Biochem., 96: 1254-1261,2005.

135 Zhang,D., Kaufman,P.L., Gao,G., Saunders,R.A. and Ma,J.X. Intravitreal injection of plasminogen kringle 5, an endogenous angiogenic inhibitor, arrests retinal neovascularizatÏon in rats, Diabetologia, 44: 757-765,2001.

136 Zhang,S.X., Sima,J., Shao,C., Fant,J., Chen,Y., Rohrer,B., Gao,G. and Ma,J.X. Plasminogen kringle 5 reduces vascular leakage in the retina in rat models of oxygen-induced retinopathy and diabetes, Diabetologia, 47: 124-131,2004.

137 Yang,x., Cheng,R., Li,C., Cai,W., Ma,J.X., Liu,Q., Yang,Z., Song,Z., Liu,Z. and Gao,G. Kringle 5 of hum an plasminogen suppresses hepatocellular carcinoma growth both in grafted and xenografted mice by anti-angiogenic activity, Cancer Biol.Ther., 5: 399-405, 2006.

138 Gonzalez-Gronow,M., Kalfa,T., Johnson,C.E., Gawdi,G. and Pizzo,S.V. The voltage-dependent anion channel is a receptor for plasminogen kringle 5 on human endothelial cells, J.Biol.Chem., 278: 27312-27318, 2003.

139 Jan,L.Y. and Jan,Y.N. Voltage-sensitive ion channels, Cell, 56: 13-25, 1989.

140 Jurgens,L., Ilsemann,P., Kratzin,H.D., Hesse,D., Eckart,K., Thinnes,F.P. and Hilschmann,N. Studies on human porin. IV. The primary structures of "Porin 31 HM" purified from human skeletal muscle membranes and of "Porin 31 HL" derived from human B lymphocyte membranes are identical, Biol.Chem.Hoppe Seyler, 372: 455-463, 1991.

141 Bathori,G., Parolini,l., Tombola,F., Szabo,l., Messina,A., Oliva,M., De,P., V, Lisanti,M., Sargiacomo,M. and Zoratti,M. Porin is present in the plasma membrane where it is concentrated in caveolae and caveolae-related domains, J.Biol.Chem., 274: 29607-29612, 1999.

142 Okarnoto,T., Schlegel,A., Scherer,P.E. and Lisanti,M.P. Caveolins, a family of scaffolding proteins for organizing "preassembled signaling complexes" at the plasma membrane, J.Biol.Chem., 273: 5419-5422, 1998.

143 Liu,J., Razani,B., Tang,S., Terrnan,B.I., Ware,J.A. and Lisanti,M.P. Angiogenesis activators and inhibitors differentially regulate caveolin-l expression and caveolae formation in vascular endothelial cells. Angiogenesis inhibitors block vascular endothelial growth factor-induced down-regulation of caveolin-l, J.Biol.Chem., 274: 15781-15785, 1999.

144 Lin,L.F., Houng,A. and Reed,G.L. Epsilon amino caproic acid inhibits streptokinase-plasminogen activator complex formation and substrate binding through kringle-dependent mechanisms, Biochemistry, 39: 4740-4745, 2000.

188 145 Nihalani,D. and Sahni,G. Streptokinase contains two independent plasminogen­ binding sites, Biochem.Biophys.Res.Commun., 217: 1245-1254, 1995.

146 McCabe,K.M., Wheeler,D.A, Adams,V. and McCabe,E.R. Comparison ofhuman VDACI with streptococcal streptokinase and bovine bactericidal permeability increasing protein: role of structural information in identifying functionally significant domains, Biochem.Mol.Med., 56: 176-179, 1995.

147 Komatsu,S., Sawada,S., Tamagaki,T., Tsuda,Y., Kono,Y., Higaki,T., Imamura,H., Tada,Y., Yamasaki,S., Toratani,A, Sato,T., Akamatsu,N., Tsuji,H. and Nakagawa,M. Effect ofprobucol on intracellular pH and proliferation ofhuman vascular endothelial cells, J.Pharmacol.Toxicol.Methods, 41: 33-41, 1999.

148 Davidson,D.J., Haskell,C., Majest,S., Kherzai,A., Egan,D.A, Walter,K.A, Schneider,A, Gubbins,E.F., Solomon,L., Chen,Z., Lesniewski,R. and Henkin,J. Kringle 5 ofhuman plasminogen induces apoptosis of endothelial and tumor cells through surface-expressed glucose-regulated protein 78, Cancer Res., 65: 4663- 4672,2005.

149 Li,J. and Lee,AS. Stress induction ofGRP78/BiP and its role in cancer, Curr.Mol.Med., 6: 45-54, 2006.

150 Reddy,R.K., Mao,C., Baumeister,P., Austin,R.C., Kaufman,RJ. and Lee,AS. Endoplasmic reticulum chaperone protein GRP78 protects cells from apoptosis induced by topoisomerase inhibitors: role of ATP binding site in suppression of caspase-7 activation, J.Biol.Chem., 278: 20915-20924, 2003.

151 McLean,J.W., Tomlinson,J.E., Kuang,WJ., Eaton,D.L., Chen,E.Y., Fless,G.M., Scanu,AM. and Lawn,R.M. cDNA sequence ofhuman apolipoprotein(a) is homologous to plasminogen, Nature, 330: 132-137, 1987.

152 Lee,T.H., Rhim,T. and Kim,S.S. Prothrombin kringle-2 domain has a growth inhibitory activity against basic fibroblast growth factor-stimulated capillary endothelial cells, J.Biol.Chem., 273: 28805-28812, 1998.

153 Tomioka,D., Maehara,N., Kuba,K., Mizumoto,K., Tanaka,M., Matsumoto,K. and Nakamura,T. Inhibition of growth, invasion, and metastasis ofhuman pancreatic carcinoma cells by NK4 in an orthotopic mouse model, Cancer Res., 61: 7518- 7524,2001.

154 Seth,P. Vector-mediated cancer gene therapy: an overview, Cancer Biol.Ther., 4: 512-517,2005.

155 Weir,B., Zhao,X. and Meyerson,M. Somatic alterations in the human cancer genome, Cancer Cell, 6: 433-438, 2004.

189 156 Roth,J.A. and Grammer,S.F. Tumor Suppressor Gene Replacement for Cancer. In: D. Curiel and J. Douglas (eds.), Cancer Gene Therapy, pp. 19-33. Humana Press: Totowa, New Jersey, 2005.

157 Nishizaki,M., Fujiwara,T., Tanida,T., Hizuta,A., Nishimori,H., Tokino,T., Nakamura,Y., Bouvet,M., Roth,J.A. and Tanaka,N. Recombinant adenovirus expressing wild-type p53 is antiangiogenic: a proposed mechanism for bystander effect, Clin. Cancer Res., 5: 1015-1023, 1999.

158 Seth,P. Pre-clinical studies with tumor suppressor genes, Adv.Exp.Med.Biol., 465: 183-192,2000.

159 Folkman,J. Antiangiogenic gene therapy, Proc.Natl.Acad.Sci.U.S.A, 95: 9064- 9066, 1998.

160 Herbst,R.S., Hess,K.R., Tran,H.T., Tseng,J.E., MuIlani,N.A., Charnsangavej,C., Madden,T., Davis,D.W., McConkey,D.J., O'Reilly,M.S., Ellis,L.M., Pluda,J., Hong,W.K. and Abbruzzese,J.L. Phase I study of recombinant human endostatin in patients with advanced solid tumors, J.Clin.Oncol., 20: 3792-3803,2002.

161 Kong,H.L. and Crystal,R.G. Gene therapy strategies for tumor antiangiogenesis, J.Natl.Cancer Inst., 90: 273-286, 1998.

162 Vannus H and Brown P Retroviruses, Washington, DC, 1989.

163 Pages,J.C. and Bru,T. Toolbox for retrovectorologists, J.Gene Med., 6 Suppl 1: S67-S82,2004.

164 Liu,Y. and Deisseroth,A. Tumor vascular targeting therapy with viral vectors, Blood, 107: 3027-3033,2006.

165 Wang,H., Kavanaugh,M.P., North,R.A. and Kabat,D. Cell-surface receptor for ecotropic murine retroviruses is a basic amino-acid transporter, Nature, 352: 729- 731,1991.

166 Miller,D.G. and Miller,A.D. A family ofretroviruses that utilize related phosphate transporters for cell entry, J.Virol., 68: 8270-8276, 1994.

167 Stagg J and Galipeau J Pseudotyped retrovectors for tumor-specific delivery of toxic suicide genes, IDrugs, 4: 928-934, 2001.

168 Deisseroth,A. Nonnal and pathological functions of mammalian retroelements, Proc.Natl.Acad.Sci.U.S.A, 102: 12292-12293,2005.

169 Perri,S.R., Nalbantoglu,J., Annabi,B., Koty,Z., Lejeune,L., Francois,M., Di Falco,M.R., Beliveau,R. and Galipeau,J. Plasminogen kringle 5-engineered glioma cells block migration of tumor-associated macrophages and suppress tumor vascularization and progression, Cancer Res., 65: 8359-8365,2005.

190 170 Perri SR, Martineau D, Francois,M., Lejeune L, Durocher Y and Galipeau,J. Plasminogen kringle 5 blocks tumor progression by anti-angiogenic and pro­ inflammatory pathways, submitted to Molecular Cancer Therapeutics, 2006.

171 Scappaticci,F.A., Smith,R., Pathak,A., Schloss,D., Lum,B., Cao,Y., Johnson,F., Engleman,E.G. and Nolan,G.P. Combination angiostatin and endostatin gene transfer induces synergistic antiangiogenic activity in vitro and antitumor efficacy in leukemia and solid tumors in mice, Mol.Ther.,3: 186-196,2001.

172 Tanaka,T., Manome,Y., Wen,P., Kufe,D.W. and Fine,H.A. Viral vector-mediated transduction of a modified cDNA inhibits angiogenesis and tumor growth, Nat.Med., 3: 437-442, 1997.

173 Machein,M.R., Risau,W. and Plate,K.H. Antiangiogenic gene therapy in a rat glioma model using a dominant-negative vascular endothelial growth factor receptor 2, Hum.Gene Ther., 10: 1117-1128, 1999.

174 Sadelain,M. and Riviere,!. The Advent of Lentiviral Vectors: Prospects for Cancer Therapy. In: E. C. Lattime and S. L. Gerson (eds.), Gene Therapy of Cancer, Academic Press: 2002.

175 McConnell,M.J. and Imperiale,M.J. Biology of adenovirus and its use as a vector for gene therapy, Hum.Gene Ther., 15: 1022-1033,2004.

176 Howe,J.A., Ralston,R. and Ramachandra,M. Development ofOncolytic Adenoviruses. In: D. Curiel and J. Douglas (eds.), Cancer Gene Therapy, pp. 211- 233. Humana Press Inc.: 2005.

177 Daly,T.M. Overview of adeno-associated viral vectors, Methods Mol.Biol., 246: 157-165,2004.

178 McCarty,D.M., Young,S.M., Jr. and Samulski,R.J. Integration ofadeno­ associated virus (AAV) and recombinant AAV vectors, Annu.Rev.Genet., 38: 819-845,2004.

179 Tenenbaum,L., Lehtonen,E. and Monahan,P.E. Evaluation ofrisks related to the use of adeno-associated virus-based vectors, CUIT.Gene Ther., 3: 545-565, 2003.

180 Douglas,J. and Curiel,D. Viral Vectors in Cancer Gene Therapy. In: D. Curiel and J. Douglas (eds.), Cancer Gene Therapy, pp. 379-391. Humana Press In.: 2005.

181 Todo,T. and Rabkin,S.D. Development ofOncolytic Replication-Competent Herpes Simplex Virus Vectors. In: D. Curiel and J. Douglas (eds.), Cancer Gene Therapy, pp. 199-210. Humana Press Inc.: 2005.

182 Mehier-Humbert,S. and Guy,R.H. Physical methods for gene transfer: improving the kinetics of gene delivery into cells, Adv.Drug Deliv.Rev., 57: 733-753,2005.

191 183 Glover,DJ., Lipps,HJ. and Jans,D.A. Towards safe, non-viral therapeutic gene expression in humans, Nat.Rev.Genet., 6: 299-310, 2005.

184 Holland,E.C. Glioblastoma multiforme: the terminator, Proc.Natl.Acad.Sci.U.S.A, 97: 6242-6244,2000.

185 Black,P.M. Brain tumor. Part 2, N.Engl.J.Med.,324: 1555-1564, 1991.

186 Folkman,J. Tumor angiogenesis: therapeutic implications, N.Engl.J.Med., 285: 1182-1186, 1971.

187 Maher,E.A., Fumari,F.B., Bachoo,R.M., Rowitch,D.H., Louis,D.N., Cavenee,W.K. and DePinho,R.A. Malignant glioma: genetics and biology of a grave matter, Genes Dev., 15: 1311-1333,2001.

188 Puduvalli,V.K. Inhibition of Angiogenesis as a Therapeutic Strategy against Brain Tumors. In: M. Kirsch and P. M. Black (eds.), pp. 307-336. Kluwer: Norwell, 2004.

189 Bello,L., Carrabba,G., Giussani,C., Lucini,V., Cerutti,F., Scaglione,F., Landre,J., Pluderi,M., Tomei,G., Villani,R., Carroll,R.S., Black,P.M. and Bikfalvi,A. Low­ dose chemotherapy combined with an antiangiogenic drug reduces human glioma growth in vivo, Cancer Res., 61: 7501-7506,2001.

190 Bello,L., Giussani,C., Carrabba,G., Pluderi,M., Lucini,V., Pannacci,M., Caronzolo,D., Tomei,G., Villani,R., Scaglione,F., Carroll,R.S. and Bikfa1vi,A. Suppression of malignant glioma recurrence in a newly developed animal model by endogenous inhibitors, Clin. Cancer Res., 8: 3539-3548, 2002.

191 Gorski,D.H., Beckett,M.A., Jaskowiak,N.T., Calvin,D.P., Mauceri,HJ., Salloum,R.M., Seetharam,S., Koons,A., Hari,D.M., Kufe,D.W. and Weichselbaum,R.R. Blockage of the vascular endothelial growth factor stress response increases the antitumor effects of ionizing radiation, Cancer Res., 59: 3374-3378, 1999.

192 Griscelli,F., Li,H., Cheong,C., Opolon,P., Bennaceur-Griscelli,A., Vassal,G., Soria,J., Soria,C., LU,H., Perricaudet,M. and Yeh,P. Combined effects of radiotherapy and angiostatin gene therapy in glioma tumor model, Proc.Natl.Acad.Sci.U.S.A, 97: 6698-6703,2000.

193 Hwu,W.J., Raizer,J., Panageas,K.S. and Lis,E. Treatment ofmetastatic melanoma in the brain with temozolomide and thalidomide, Lancet Oncol., 2: 634-635, 2001.

194 Li,L., Rojiani,A. and Siemann,D.W. Targeting the tumor vasculature with combretastatin A-4 disodium phosphate: effects on radiation therapy, InU.Radiat.Oncol.Biol.Phys., 42: 899-903, 1998.

192 195 Mauceri,H.J., Hanna,N.N., Beckett,M.A., Gorski,D.H., Staba,M.J., Stellato,K.A., Bigelow,K., Heimann,R., Gately,S., Dhanabal,M., Soff,G.A., Sukhatme,V.P., Kufe,D.W. and Weichselbaum,R.R. Combined effects of angiostatin and ionizing radiation in antitumour therapy, Nature, 394: 287-291, 1998.

196 Shir,A. and Levitzki,A. Gene therapy for glioblastoma: future perspective for delivery systems and molecular targets, Cell Mol.Neurobiol., 21: 645-656, 2001.

197 Kirsch,M., Strasser,J., Allende,R., Bello,L., Zhang,J. and Black,P.M. Angiostatin suppresses malignant glioma growth in vivo, Cancer Res., 58: 4654-4659, 1998.

198 Ory,D.S., Neugeboren,B.A. and Mulligan,R.C. A stable human-derived packaging cellline for production ofhigh titer retrovirus/vesicular stomatitis virus G pseudotypes, Proc.Natl.Acad.Sci.U.S.A, 93: 11400-11406, 1996.

199 Ponten,J. and Macintyre,E.H. Long term culture of normal and neoplastic human glia, Acta Pathol.Microbiol.Scand., 74: 465-486, 1968.

200 Galipeau,J., Li,H., Paquin,A., Sicilia,F., Karpati,G. and Nalbantoglu,J. Vesicular stomatitis virus G pseudotyped retrovector mediates effective in vivo suicide gene delivery in experimental brain cancer, Cancer Res., 59: 2384-2394, 1999.

201 Eliopoulos,N. and Galipeau,J. Green Fluorescent Protein. In: B.W.Hicks (ed.), pp. 353-371. Humana Press Inc.: Totowa, 2002.

202 Annabi,B., Lachambre,M.P., Bousquet-Gagnon,N., Page,M., Gingras,D. and Beliveau,R. Green tea polyphenol (-)-epigallocatechin 3-gallate inhibits MMP-2 secretion and MTI-MMP-driven migration in glioblastoma cells, Biochim.Biophys.Acta, 1542: 209-220, 2002.

203 Stagg,J., Lejeune,L., Paquin,A. and Galipeau,J. Marrow stromal cells for interleukin-2 delivery in cancer immunotherapy, Hum.Gene Ther., 15: 597-608, 2004.

204 Li,H., onso-Vanegas,M., Colicos,M.A., Jung,S.S., Lochmuller,H., Sadikot,A.F., Snipes,G.J., Seth,P., Karpati,G. and Nalbantoglu,J. Intracerebral adenovirus­ mediated p53 tumor suppressor gene therapy for experimental human glioma, Clin. Cancer Res., 5: 637-642, 1999.

205 Chang,Y., Mochalkin,I., McCance,S.G., Cheng,B., Tulinsky,A. and Castellino,F.J. Structure and ligand binding determinants of the recombinant kringle 5 domain ofhuman plasminogen, Biochemistry, 37: 3258-3271, 1998.

206 Leek,R.D., Lewis,C.E., Whitehouse,R., Greenall,M., Clarke,J. and Harris,A.L. Association of macrophage infiltration with angiogenesis and prognosis in invasive breast carcinoma, Cancer Res., 56: 4625-4629, 1996.

193 207 Richter,G., Kruger-Krasagakes,S., Hein,G., Huls,C., Schmitt,E., Diamantstein,T. and Blankenstein,T. Interleukin 10 transfected into Chinese hamster ovary cells prevents tumor growth and macrophage infiltration, Cancer Res., 53: 4134-4137, 1993.

208 Sunderkotter,C., Steinbrink,K., Goebeler,M., Bhardwaj,R. and Sorg,C. Macrophages and angiogenesis, J.Leukoc.Biol., 55: 410-422, 1994.

209 Nishie,A., Ono,M., Shono,T., Fukushi,J., Otsubo,M., Onoue,H., Ito,Y., Inamura,T., Ikezaki,K., Fukui,M., Iwaki,T. and Kuwano,M. Macrophage infiltration and heme oxygenase-l expression correlate with angiogenesis in human gliomas, Clin.Cancer Res., 5: 1107-1113, 1999.

210 Moulton,K.S., Vakili,K., Zurakowski,D., Soliman,M., Butterfield,C., Sylvin,E., Lo,K.M., Gillies,S., Javaherian,K. and Folkman,J. Inhibition ofplaque neovascularization reduces macrophage accumulation and progression of advanced atherosclerosis, Proc.Natl.Acad.Sci.U.S.A, 100: 4736-4741, 2003.

211 Cao,Y., Cao,R. and Veitonmaki,N. Kringle structures and antiangiogenesis, Curr.Med.Chem.Anti.-Canc.Agents, 2: 667-681,2002.

212 Dirkx,A.E., oude Egbrink,M.G., Castermans,K., van der Schaft,D.W., Thijssen,V.L., Dings,R.P., Kwee,L., Mayo,K.H., Wagstaff,J., Bouma-ter Steege,J.c. and Griffioen,A.W. Anti-angiogenesis therapy can overcome endothelial cell anergy and promo te leukocyte-endothelium interactions and infiltration in tumors, F ASEB J., 20: 621-630, 2006.

213 Sun,x', Kanwar,J.R., Leung,E., Lehnert,K., Wang,D. and Krissansen,G.W. Angiostatin enhances B7.1-mediated cancer immunotherapy independently of effects on vascular endothelial growth factor expression, Cancer Gene Ther., 8: 719-727,2001.

214 Chavakis,T., Athanasopoulos,A., Rhee,J.S., Orlova,V., Schmidt-Woll,T., Bierhaus,A., May,A.E., Celik,l., Nawroth,P.P. and Preissner,K.T. Angiostatin is a novel anti-inflammatory factor by inhibiting leukocyte recruitment, Blood, 105: 1036-1043,2005.

215 Medina,D. Mammary tumorigenesis in chemical carcinogen-treated mice. VI. Tumor-producing capabilities of mammary dysplasias in BALB/cCrgl mice, J.Natl.Cancer Inst., 57: 1185-1189, 1976.

216 Frevert,C.W., Wong,V.A., Goodman,R.B., Goodwin,R. and Martin,T.R. Rapid fluorescence-based measurement of neutrophil migration in vitro, J.Immunol.Methods, 213: 41-52, 1998.

217 Matory,Y.L., Chen,M., Goedegebuure,P.S. and Eberlein,T.J. Anti-tumor effects and tumor immunogenicity following IL2 or IL4 cytokine gene transfection of threemousemammarytumors, Ann.Surg.Oncol., 2: 502-511,1995.

194 218 Matory,Y.L., Chen,M., Dorfman,D.M., Williams,A., Goedegebuure,P.S. and Eberlein,T.J. Antitumor activity ofthree mouse mammary cancer celllines after interferon-gamma gene transfection, Surgery, 118: 251-255, 1995.

219 Baruch,A., Hartmann,M., Zrihan-Licht,S., Greenstein,S., Burstein,M., Keydar,l., Weiss,M., Smorodinsky,N. and Wreschner,D.H. Preferential expression ofnovel MUCI tumor antigen isoforms in human epithelial tumors and their tumor­ potentiating function, InU.Cancer, 71: 741-749, 1997.

220 Prochazka,M., Gaskins,H.R., Shultz,L.D. and Leiter,E.H. The nonobese diabetic scid mouse: model for spontaneous thymomagenesis associated with immunodeficiency, Proc.Natl.Acad.Sci.U.S.A, 89: 3290-3294, 1992.

221 Karre,K. NK cells, MHC class 1 molecules and the missing self, Scand.J.Immunol., 55: 221-228,2002.

222 Levitsky,H.I., Lazenby,A., Hayashi,R.J. and Pardoll,D.M. In vivo priming oftwo distinct antitumor effector populations: the role ofMHC class 1 expression, J.Exp.Med., 179: 1215-1224, 1994.

223 Moron,G., Dadaglio,G. and Leclerc,C. New tools for antigen delivery to the MHC class 1 pathway, Trends Immunol., 25: 92-97,2004.

224 Trombetta,E.S. and Mellman,1. Cell biology of antigen processing in vitro and in vivo, Annu.Rev.Immunol., 23: 975-1028,2005.

225 Caims,C.M., Gordon,J.R., Li,F., Baca-Estrada,M.E., Moyana,T. and Xiang,J. Lymphotactin expression by engineered myeloma cells drives tumor regression: mediation by CD4+ and CD8+ T cells and neutrophils expressing XCRI receptor, J.Immunol., 167: 57-65,2001.

226 Lee,L.F., Hellendall,R.P., Wang,Y., Haskill,J.S., Mukaida,N., Matsushima,K. and Ting,J.P. IL-8 reduced tumorigenicity ofhuman ovarian cancer in vivo due to neutrophil infiltration, J.Immunol., 164: 2769-2775, 2000.

227 Pike,S.E., Yao,L., Jones,K.D., Chemey,B., Appella,E., Sakaguchi,K., Nakhasi,H., Teruya-Feldstein,J., Wirth,P., Gupta,G. and Tosato,G. Vasostatin, a fragment, inhibits angiogenesis and suppresses tumor growth, J .Exp.Med., 188: 2349-2356, 1998.

228 Mantovani,A., Allavena,P. and Sica,A. Tumour-associated macrophages as a prototypic type II polarised phagocyte population: role in tumour progression, Eur.J.Cancer, 40: 1660-1667,2004.

229 Mills,C.D., Kincaid,K., Alt,J.M., Heilman,M.J. and Hill,A.M. M-IIM-2 macrophages and the ThllTh2 paradigm, J.Immunol., 164: 6166-6173, 2000.

195 230 Bingle,L., Brown,NJ. and Lewis,C.E. The role oftumour-associated macrophages in tumour progression: implications for new anticancer therapies, J.Pathol., 196: 254-265, 2002.

231 Leek,R.D. and Harris,A.L. Tumor-associated macrophages in breast cancer, J.Mammary.Gland.Biol.Neoplasia., 7: 177-189,2002.

232 Oosterling,S.J., van der Bij,G.J., Meijer,G.A., Tuk,C.W., van,G.E., van,R.N., Meijer,S., van,d.S., Jr., Beelen,R.H. and van,E.M. Macrophages direct tumour histology and clinical outcome in a colon cancer model, J.Pathol., 207: 147-155, 2005.

233 Sinha,P., Clements,V.K. and Ostrand-Rosenberg,S. Reduction ofmyeloid-derived suppressor ceIls and induction of Ml macrophages facilitate the rejection of established metastatic disease, J.Immunol., 174: 636-645,2005.

234 Barbera-Guillem,E., Nyhus,J.K., Wolford,C.C., Friece,C.R. and Sampsel,J.W. Vascular endothelial growth factor secretion by tumor-infiltrating macrophages essentiaIly supports tumor angiogenesis, and IgG immune complexes potentiate the process, Cancer Res., 62: 7042-7049, 2002.

235 Lewis,C. and Murdoch,C. Macrophage responses to hypoxia: implications for tumor progression and anti-cancer therapies, Am.J.Pathol., 167: 627-635, 2005.

236 Shimura,S., Yang,G., Ebara,S., Wheeler,T.M., Frolov,A. and Thompson,T.C. Reduced infiltration of tumor-associated macrophages in human prostate cancer: association with cancer progression, Cancer Res., 60: 5857-5861,2000.

237 Condeelis,J. and PoIlard,J.W. Macrophages: obligate partners for tumor ceIl migration, invasion, and metastasis, CeIl, 124: 263-266, 2006.

238 Kusmartsev,S. and Gabrilovich,D.1. STATI signaling regulates tumor-associated macrophage-mediated T ceIl deletion, J.Immunol., 174: 4880-4891, 2005.

239 Saio,M., Radoja,S., Marino,M. and Frey,A.B. Tumor-infiltrating macrophages induce apoptosis in activated CD8(+) T ceIls by a mechanism requiring ceIl contact and mediated by both the ceIl-associated form ofTNF and nitric oxide, J.Immunol., 167: 5583-5593,2001.

240 Bottazzi,B., Polentarutti,N., Acero,R., Balsari,A., Boraschi,D., Ghezzi,P., Salmona,M. and Mantovani,A. Regulation of the macrophage content of neoplasms by chemoattractants, Science, 220: 210-212, 1983.

241 Mantovani,A. The chemokine system: redundancy for robust outputs, Immunol.Today, 20: 254-257, 1999.

196 242 Matsushima K, Larsen CG, DuBois GC and Oppenheim JJ Purification and characterization of a nove1 monocyte chemotactic and activating factor produced by a human myelomonocytic cellline., J.Exp.Med., 169: 1485-1490, 1999.

243 Saji,H., Koike,M., Yamori,T., Saji,S., Seiki,M., Matsushima,K. and Toi,M. Significant correlation of monocyte chemoattractant protein-l expression with neovascularization and progression ofbreast carcinoma, Cancer, 92: 1085-1091, 2001.

244 Yoshimura,T., Robinson,E.A., Tanaka,S., Appella,E., Kuratsu,J. and Leonard,E.J. Purification and amine acid analysis of two human glioma-derived monocyte chemoattractants, J.Exp.Med., 169: 1449-1459, 1989.

245 Azenshtein E, Meshel T, Shina S, Barak N, Keydar l and Ben-Baruch A The angiogenic factors CXCL8 and VEGF in breast cancer: regu1ation by an array of pro-malignancy factors., Cancer Lett, 217: 73-86, 2005.

246 Haghnegahdar,H., Du,J., Wang,D., Strieter,R.M., Burdick,M.D., Nanney,L.B., Cardwell,N., Luan,J., Shattuck-Brandt,R. and Richmond,A. The tumorigenic and angiogenic effects ofMGSAIGRO proteins in melanoma, J.Leukoc.Biol., 67: 53- 62,2000.

247 Payne,A.S. and Come1ius,L.A. The ro1e of chemokines in melanoma tumor growth and metastasis, J.Invest Dermatol., 118: 915-922, 2002.

248 Ueno,T., Toi,M., Saji,H., Muta,M., Bando,H., Kuroi,K., Koike,M., Inadera,H. and Matsushima,K. Significance of macrophage chemoattractant protein-l in macrophage recruitment, angiogenesis, and survival in human breast cancer, Clin. Cancer Res., 6: 3282-3289, 2000.

249 Sica,A., Schioppa,T., Mantovani,A. and Allavena,P. Tumour-associated macrophages are a distinct M2 polarised population promoting tumour progression: potentia1 targets of anti-cancer therapy, EurJ.Cancer, 42: 717-727, 2006.

250 Knowles,H., Leek,R. and Harris,A.L. Macrophage infiltration and angiogenesis in human malignancy, Novartis.Found.Symp.,256: 189-200,2004.

251 Pollard,J. W. Tumour-educated macrophages promote tumour progression and metastasis, Nat.Rev.Cancer, 4: 71-78,2004.

252 Wyckoff,J., Wang,W., Lin,E.Y., Wang,Y., Pixley,F., Stanley,E.R., Graf,T., Pollard,J.W., Segall,J. and Condeelis,J. A paracrine loop between tumor cells and macrophages is required for tumor cell migration in mammary tumors, Cancer Res., 64: 7022-7029, 2004.

253 Benelli,R., Morini,M., Carrozzino,F., Ferrari,N., Minghelli,S., Santi,L., Cassatella,M., Noonan,D.M. and Albini,A. Neutrophi1s as a key cellular target for

197 angiostatin: implications for regulation of angiogenesis and inflammation, FASEB J., 16: 267-269, 2002.

254 Kolaczkowska,E., Chadzinska,M., Scislowska-Czamecka,A., Plytycz,B., Opdenakker,G. and Amold,B. Gelatinase B/matrix metalloproteinase-9 contributes to cellular infiltration in a murine model of zymosan peritonitis, Immunobiology, 211: 137-148, 2006.

255 Hiratsuka,S., Nakamura,K., Iwai,S., Murakami,M., Itoh,T., Kijima,H., Shipley,J.M., Senior,R.M. and Shibuya,M. MMP9 induction by vascular endothelial growth factor receptor-1 is involved in lung-specific metastasis, Cancer Cell, 2: 289-300, 2002.

256 Kendall,R.L. and Thomas,K.A. Inhibition ofvascular endothelial cell growth factor activity by an endogenously encoded soluble receptor, Proc.Natl.Acad.Sci.U.S.A,90: 10705-10709, 1993.

257 Peyruchaud,O., Serre,C.M., NicAmhlaoibh,R., Foumier,P. and Clezardin,P. Angiostatin inhibits bone metastasis formation in nude mice through a direct anti­ osteoclastic activity, J.Biol.Chem., 278: 45826-45832, 2003.

258 Mantovani A Macrophage diversity and polarization: in vivo veritas, Blood, 108: 408-409,2006.

259 Rubenstein,J.L., Kim,J., Ozawa,T., Zhang,M., Westphal,M., Deen,D.F. and Shuman,M.A. Anti-VEGF antibody treatment of glioblastoma prolongs survival but results in increased vascular cooption, Neoplasia., 2: 306-314,2000.

260 Filleur,S., Volpert,O.V., Degeorges,A., Voland,C., Reiher,F., Clezardin,P., Bouck,N. and Cabon,F. In vivo mechanisms by which tumors producing bypass its inhibitoryeffects, Genes Dev., 15: 1373-1382, 2001.

261 Douglas,M.L., Reid,J.L., Hii,S.I., Jonsson,J.R. and Nicol,D.L. Renal cell carcinoma may adapt to and overcome anti-angiogenic intervention with thalidomide, BJU.Int., 89: 591-595,2002.

262 Yu,J.L., Rak,J.W., Coomber,B.L., Hicklin,D.J. and Kerbel,R.S. Effect ofp53 status on tumor response to antiangiogenic therapy, Science, 295: 1526-1528, 2002.

263 Tran,J., Master,Z., Yu,J.L., Rak,J., Dumont,D.J. and Kerbel,R.S. A role for survivin in chemoresistance of endothelial cells mediated by VEGF, Proc.Natl.Acad.Sci.U.S.A, 99: 4349-4354, 2002.

264 Wang,J.H., Wu,Q.D., Bouchier-Hayes,D. and Redmond,H.P. Hypoxia /-, upregulates Bcl-2 expression and suppresses interferon-gamma induced

198 antiangiogenic activity in human tumor derived endothe1ial cells, Cancer, 94: 2745-2755, 2002.

265 Zhang,X.S., Zhu,X.F., Gao,J.S., Qian,C.N., Kuang,Z.J., Liu,Z.C. and Zeng,Y.X. Multiple drug resistance phenotype ofhuman endothelial cells induced by vascular endothelial growth factor 165, Acta Pharmacol.Sin., 22: 731-735, 2001.

266 Kwak,C., Jin,R.J., Lee,C., Park,M.S. and Lee,S.E. Thrombospondin-l, vascular endothelial growth factor expression and their relationship with p53 status in prostate cancer and benign prostatic hyperplasia, BJU.Int., 89: 303-309,2002.

267 Laderoute,K.R., Alarcon,R.M., Brody,M.D., Calaoagan,J.M., Chen,E.Y., Knapp,A.M., Yun,Z., Denko,N.C. and Giaccia,A.J. Opposing effects ofhypoxia on expression of the angiogenic inhibitor thrombospondin 1 and the angiogenic inducer vascular endothelial growth factor, Clin. Cancer Res., 6: 2941-2950, 2000.

268 Holmgren,L., Jackson,G. and Arbiser,J. p53 induces angiogenesis-restricted dormancy in a mouse fibrosarcoma, Oncogene, 17: 819-824, 1998.

269 Gautam,A., Densmore,C.L., Melton,S., Golunski,E. and Waldrep,J.C. Aerosol delivery ofPEI-p53 complexes inhibits BI6-FlO lung metastases through regulation of angiogenesis, Cancer Gene Ther., 9: 28-36, 2002.

270 Bergers,G., Javaherian,K., Lo,K.M., Folkman,J. and Hanahan,D. Effects of angiogenesis inhibitors on multistage carcinogenesis in mi ce, Science, 284: 808- 812, 1999.

271 Fong,T.A., Shawver,L.K., Sun,L., Tang,C., App,H., Powell,T.J., Kim,Y.H., Schreck,R., Wang,x., Risau,W., Ullrich,A., Hirth,K.P. and McMahon,G. SU5416 is a potent and selective inhibitor of the vascular endothelial growth factor receptor (Flk-IIKDR) that inhibits tyrosine kinase catalysis, tumor vascularization, and growth of multiple tumor types, Cancer Res., 59: 99-106, 1999.

272 Laird,A.D., Vajkoczy,P., Shawver,L.K., Thurnher,A., Liang,C., Mohammadi,M., Schlessinger,J., Ullrich,A., Hubbard,S.R., Blake,R.A., Fong,T.A., Strawn,L.M., Sun,L., Tang,C., Hawtin,R., Tang,F., Shenoy,N., Hirth,K.P., McMahon,G. and Cherrington SU6668 is a potent antiangiogenic and antitumor agent that induces regression of established tumors, Cancer Res., 60: 4152-4160, 2000.

273 Bergers,G., Song,S., Meyer-Morse,N., Bergsland,E. and Hanahan,D. Benefits of targeting both pericytes and endothelial cells in the tumor vasculature with kinase inhibitors, J.Clin.Invest, 111: 1287-1295,2003.

274 Cline,E.I., Bicciato,S., DiBello,C. and Lingen,M.W. Prediction ofin vivo synergistic activity of antiangiogenic compounds by gene expression profiling, Cancer Res., 62: 7143-7148,2002.

199 275 Kerbel,R.S. Antiangiogenic therapy: a univers al chemosensitization strategy for cancer?, Science, 312: 1171-1175,2006.

276 Duda,D.G. Antiangiogenesis and drug delivery to tumors: bench to bedside and back, Cancer Res., 66: 3967-3970,2006.

277 Casanovas,O., Hicklin,D.J., Bergers,G. and Hanahan,D. Drug resistance by evasion of antiangiogenic targeting ofVEGF signaling in late-stage pancreatic islet tumors, Cancer Cell, 8: 299-309, 2005.

278 Kamba T, Tarn B, Hashizume H, Haskell A, Sennino B, Mancuso MR, Norberg SM, O'Brien SM, Davis RB, Gowen LC, Anderson KD, Thurston G, Joho S, Springer ML, Kuo CJ and McDonald DM VEGF-dependent pl asti city of fenestrated capillaries in the normal adult microvasculature, Am J Physiol Heart Circ Physiol, 290: H560-H576, 2006.

279 Jain,R.K., Duda,D.G., Clark,J.W. and Loeffler,J.S. Lessons from phase III clinical trials on anti-VEGF therapy for cancer, Nat.Clin.Pract.Oncol., 3: 24-40, 2006.

280 Kisker,O., Becker,C.M., Prox,D., Fannon,M., D'Amato,R., Flynn,E., Fogler,W.E., Sim,B.K., Allred,E.N., Pirie-Shepherd,S.R. and Folkman,J. Continuous administration of endostatin by intraperitoneally implanted osmotic pump improves the efticacy and potency of therapy in a mouse xenograft tumor model, Cancer Res., 61: 7669-7674, 2001.

281 Folkman,J. and Kalluri,R. Cancer without disease, Nature, 427: 787,2004.

282 Liau,G., Su,E.J. and Dixon,K.D. Clinical efforts to modulate angiogenesis in the adult: gene therapy versus conventional approaches, Drug Discov.Today, 6: 689- 697,2001.

283 Cavazzana-Calvo,M., Hacein-Bey,S., de Saint,B.G., Gross,F., Yvon,E., Nusbaum,P., Selz,F., Hue,C., Certain,S., Casanova,J.L., Bousso,P., Deist,F.L. and Fischer,A. Gene therapy ofhuman severe combined immunodeticiency (SCID)­ Xl disease, Science, 288: 669-672,2000.

284 Hacein-Bey-Abina,S., Le,D.F., Carlier,F., Bouneaud,C., Hue,C., de Villartay,J.P., Thrasher,A.J., Wulffraat,N., Sorensen,R., Dupuis-Girod,S., Fischer,A., Davies,E.G., Kuis,W., Leiva,L. and Cavazzana-Calvo,M. Sustained correction of X-linked severe combined immunodeticiency by ex vivo gene therapy, N.Engl.J.Med., 346: 1185-1193,2002.

285 Hacein-Bey-Abina,S., Von,K.C., Schmidt,M., McCormack,M.P., Wulffraat,N., Leboulch,P., Lim,A., Osborne,C.S., Pawliuk,R., Morillon,E., Sorensen,R., Forster,A., Fraser,P., Cohen,J.!., de Saint,B.G., Alexander,!., Wintergerst,U., Frebourg,T., Aurias,A., Stoppa-Lyonnet,D., Romana,S., Radford-Weiss,!., Gross,F., Valensi,F., Delabesse,E., Macintyre,E., Sigaux,F., Soulier,J., Leiva,L.E., Wissler,M., Prinz,C., Rabbitts,T.H., Le,D.F., Fischer,A. and Cavazzana-Calvo,M.

200 LM02-associated clonaI T ceU proliferation in two patients after gene therapy for SCID-Xl, Science, 302: 415-419, 2003.

286 Chang,Y., Mochalkin,I., McCance,S.G., Cheng,B., Tulinsky,A. and Castellino,F.J. Structure and ligand binding determinants of the recombinant kringle 5 domain ofhuman plasminogen, Biochemistry, 37: 3258-3271, 1998.

287 Joseph,I.B. and Isaacs,J.T. Macrophage role in the anti-prostate cancer response to one class ofantiangiogenic agents, J.Natl.Cancer Inst., 90: 1648-1653, 1998.

288 Joseph,J.M., Bouquet,C., Opolon,P., Morizet,J., Aubert,G., Rossler,J., Gross,N., Griscelli,F., Perricaudet,M. and Vassal,G. High Ievel of stabilized angiostatin mediated by adenovirus delivery do es not impair the growth ofhuman neuroblastoma xenografts, Cancer Gene Ther., 10: 859-866,2003.

289 GaIaup,A., Magnon,C., Rouffiac,V., Opolon,P., Opolon,D., Lassau,N., Tursz,T., Perricaudet,M. and GrisceUi,F. Full kringles ofplasminogen (aa 1-566) mediate complete regression ofhuman MDA-MB-231 breast tumor xenografted in nude mi ce, Gene Ther., 12: 831-842, 2005.

290 Teicher,B.A. A systems approach to cancer therapy. (Antioncogenics + standard cytotoxics-->mechanism(s) of interaction), Cancer Metastasis Rev., 15: 247-272, 1996.

291 Jain,R.K. Normalization oftumor vasculature: an emerging concept in antiangiogenic therapy, Science, 307: 58-62,2005.

292 Jain,R.K. Normalizing tumor vasculature with anti-angiogenic therapy: a new paradigm for combination therapy, Nat.Med., 7: 987-989,2001.

293 Eliopoulos,N., Gagnon,R.F., Francois,M. and GaIipeau,J. Erythropoietin delivery by genetically engineered bone marrow stromal cells for correction of anemia in mice with chronic renal failure, J.Am.Soc.Nephrol., 17: 1576-1584,2006.

201 APPENDIX

202

Plasrninogen Kringle 5-Engineered Gliorna Cells Block Migration of Turnor-Associated Macrophages and Suppress Turnor Vascularization and Progression

Sabrina R. Perri,J2 Josephine NaJbantoglu,3 Borhane Annabi,6 Zafiro Koty: Laurence Lejeune: J2 Moïra François,' Marcos R. Di FaJco: Richard Béliveau,7 and Jacques Galipeau .8

'Division of Experimental Medicine, 'Lady Davis Institute for Medical Research: 'Depart ment of Neurology and Neuroscience, "Montreal Neurologieal Institute, Mc Gill University: "McGiII University and Genome Quebec Innovation Centre; 'Department of Biochemistry, 'Laboratoire de Médicine Moléculaire. Centre de Cancérologie Charles, Bruneau. Hôpital Sainte'Justine. Université de Québec à Montréal: and 'Di\ision of Hematology/Oncology. Department of Medicine, Jewish General Hospital. Montreal. Quebec. Canada

Abstract patients experiencing local recurrence and a 5-year survival rate of only 9% (3). Angiostatin, a well-characterized angiostatic agent, is a The urgent need for the development of potent antiglioma proteolytic c1eavage product of human plasminogen encom­ therapeutics as weIl as the importance of angiogenesis in glioma passing the first four kringle structures. The fifth kringle growth (4, 5) has fueled the identification and characterization of domain (K5) of human plasminogen is distinct from angios­ numerous antiangiogenic agents, aimed at interrupting new vessel tatin and has been shown, on ils own, to act as a potent formation and ultimately arresting tumor growth. The use of endothelial cell inhibitor. We propose that tumor-targeted antiangiogenic agents as therapeutics is an appealing approach K5 cDNA expression may act as an effective therapeutic targeting nonmalignant endothelial cens that form the tumor intervention as part of a cancer gene therapy strategy. In this vasculature and indirectly affects tumor cens thus minimizing the study, we provide evidence that eukaryotically expressed risk of toxicity, In addition, the problem of drug resistance His-tagged human K5 cDNA (hK5His) is exported extracellu­ associated with conventional chemotherapy agents is avoided larly and main tains predicted disulfide bridging conformation because normal endothelial cells are genetically stable unlike in solution. Functionally, hK5His protein produced by retro­ tumor cells (6)_ Moreover, the blood-brain barrier is another virally engineered human U87MG glioma cells suppresses obstacle olten encountered with therapies targeting malignant cells in vitro migration of both human umbilical vein endothelial within the brain and may be exploited with antiangiogenic cells and human macrophages. Subcutaneous implantation of strategies by maintaining the angiogenic inhibitor within the brain Matrigel-embedded hK5His-producing glioma cells in non­ vasculature for a prolonged period_ Although angiogenesis obese diabetic/severe combined immunodeficient mice inhibition otIers severa! advantages over traditional therapeutic reveals that hK5His induces a marked reduction in blood approaches, it is expected to induce a cytostatic effect resulting in vessel formation and significantly suppresses the recruitment tumor stabilization not eradication. Furthermore, single-agent of tumor-infiltrating CD45+Mac3+Grl- macrophages. Thera­ antiangiogenic therapy may lead to a compensatory increase in peutically, we show in a nude mouse orthotopic brain cancer the production of other angiogenic factors, which may then sustain model that tumor-targeted K5 expression is capable of angiogenesis (1)_ Despite these limitations, it is now weil effectively suppressing glioma growth and promotes signifi­ recognized that angiostatic agents either singly or in combination cant long-term survival (> 120 days) of test animais. These data with other treatment modalities could offer superior therapeutic suggest that plasminogen K5 acts as a novel two-pronged benefits than currently atlainable (7-13), converting the tumor into anticancer agent, mediating ils inhibitory effect via its action a controlled, quiescent chronic disease, Direct delivery of purified on host-derived endotheJial cells and tumor-associated recombinant antiangiogenic proteins however necessitates large macrophages, resulting in a potent. c1inically relevant anti­ quantities and repeated administration of the therapeutic gene tumor effect. (Cancer Res 2005; 65(18): 8359-65) product for a prolonged period of time (14, 15), To circumvent the pitfalls associated with conventional treatments, we have devel­ Introduction oped a potent antiangiogenic gene therapy strategy for brain Glioblastomas are very aggressive solid tumors characterized cancer. Moreover, because malignant gIiomas localize within the by hypervascularization and extensive tumor cell invasion into central nervous system and do not form distant metastases (16), normal brain parenchyma (1). Current therapeutic modalities they represent an attractive target for local gene therapy, which primarily target rapidly dividing malignant cens and include provides high and sustained local production of the desired thera­ combinations of surgery, radiotherapy, and chemotherapy (1, 2). peutie agent and offers a viable alternative to existing therapeutic However, these therapies remain ineffective with >90% of interventions_ Severa! inhibitors of angiogenesis exist endogenously as proteo­ Iytic c1eavage products oflarger precursor molecules_ Angiostatin, a well-characterized antiangiogenic agent discovered by O'Reilly et al, Nole: Supplementary data for this article are available at Cancer Research Online in Folkman's laboratory, was initially identified as an endothelial (http://cancerres,aacrjournals,orgl). Requests for reprints: Jacques Galipeau. Lady Davis Institute for Medical cell growth inhibitor present in urine and plasma of animais Research. 3755 Cote-Ste-Catherine Road. Montreal. Quebec. Canada H3T 1E2 harboring solid tumors (17, 18), is an internai cryptic fragment of Phone: 514-340.8214; Fax: 514-340.8Z81; E-mail: j,galipea@lab,jgh,mcgill,ca ©2005 American Association for Cancer Research. human' pIasminogen encompassing the first four kringle (KI-4) doi:IO,ll58/0008-5472,CAN-05-0508 domains (Fig. lA). The K5 domain of plasminogen has been www.aacrjournals_org 8359 Cancer Res 2005; 65: (18). September 15, 2005 Cancer Research cxpressed as a recombinant protein in bacteria and been found to human serum and 500 units/mL granulocyte/macrophage colony stimulat­ be more potent than KI-4 or any of the plasminogen kringles ing factor (GM-CSF; Immunex, Thousand Oaks, CA). expressed individually in inhibiting growth factor stimulated Soluble hm",m kringIe 5-hisTag expression vedor, protein purifi­ cation, and detec:tion. pBLAST-hK5 cDNA (InvivoGen, San Diego, CA) proliferation of endothelial cells in vitro (I9, 20). It has a1so been encodes for hK5 and contains a 5' interleukin-2 signal peptide (IL-2sp) shown that kringle domains 1-5 (KI-5) act as more potent coding sequence. A 3' 6-histidine tag was PCR c10ned to generate IL-2sp/ endothelial cell inhibitors in vitro and are more effective in hK5/HisTag (hereafter hK5His). Conditioned medium was collected from suppressing fibrosarcoma tumor growth in vivo compared with hK5His stably transfected human kidney 293 T cells, concentrated using KI-4 alone (21). We propose that the K5 domain could serve as a a Centricon Plus-20 column (Millipore, Billerica, MA) and purified using a potent angiostatic agent on its own and that it may act as a useful Ni-NTA-based resin affinity chromatography purification system (Novagen, therapeutic transgene within a cancer gene therapy strategy. To test San Diego, CA). Immunoblot analysis was done on eluted fractions using this hypothesis, we have developed a hum an K5-expressing a polyclonal anti-His antibody (Santa Cruz Biotechnology, Santa Cruz, CA). retroviral vector and tested its efficacy in vivo using an orthotopic Proteornic aDaIysis of puri6ed soluble hK5His. Lyophilized hK5His brain cancer mode!. samples were resuspended in water/5% acetonitrile/O.l % trifluoroacetic acid (TFA), desalted using reverse-phase CI8 ZipTips (Millipore), eluted with water/7O% acetonitrile/O.l % TFA. The eluate was spotted onto a Materials and Methods matrix-assisted laser desorption ionization (MALDI) sample plate and analyzed using a MALDI time-of·flight (TOF) Voyager-DE (Applied Cell culture reagents. 293GPG pantropic retrovirus-packaging cell line Biosystems, Foster City, CA). The eluted fraction was also spotted onto (22) was a gift from Dr. Richard C. Mulligan (Children's Hospital, Boston, polybrene-coated TFA filter dises for Edman degradation-based NH2 - MA). Human gIioma U87MG cell line (23) was a gift from Dr. Stéphane terminal sequence analysis using a Procise 492 automated microsequencer Richard (Lady Davis Institute, Montreal, Quebec, Canada). Freshly isolated (Applied Biosystems). One eluate aliquot was treated with 10 mmol/L DIT primary human umbilical endothelial cells (HUVEC) were generously for 1 hour, diluted with 80 fiL of 50 mmol/L Ambic containing 100 ng of provided by Dr. Mark B10stein (Lady Davis Institute) and maintained in Trypsin gold (Promega, Madison, Wl). and digested ovemight. For dual­ endothelial-basal media (EBM-2; Cambrex, Walkersville, MD) supplemented enzyme digest. the eluted fraction was first digested ovemight with with EBM-2 SingleQuots. Fresh human peripheral blood mononuclear cells immobilized trypsin (Pierce, Rockford, IL) followed by an ovemight were obtained from normal volunteers and maintained in DMEM-FI2 digestion with 100 ng of Endoproteinase-Asp-N (Roche Diagnostics, Laval, (Wisent Technologies, St-Bruno, QC, Canada) supplemented with 2% Quebec, Canada~ Digested solutions were ziptipped and eluted onto a MALDI sample plate. Peptide fragment analysis was done with an Ultima MALDI-Quadrupole TOF mass spectrometer {Waters Limitée. Dorval, Quebec, Canada~ Spectra analysis and peptide identity assignment was done using Biolynx, a software tool of Masslynx v4.0 (Waters). VSV-G-pseudotyped retroviraJ vector design, synthesis, and titer assessrnent. 111e bicistronic murine retrovector pIRES-EGFP was previ­ ously generated in our laboratory (24) and contains an enhanced green fluoescent protein (EGFP). The hK5His cDNA was ligated into pIRES-EGFP to generate phK5His-IRES-EGFP. VSV-G-pseudotyped retroparticles encod­ ing hK5His-lRES-EGFP were generated by tetracycline withdrawal as previously publisbed (24, 25). Engineered retroparticles were devoid of replication-competent retrovirus as detennined by GFP marker rescue assay using conditioned medium from target cells. The titer of the control GFP and hK5His 293GPG single clone population was assessed as previously described (25). Transduction of human US7MG gIioma celIs. Glioma cells were transduced with hK5His-expressing retroviral particles as previously described (24). Stahly transduced gIioma cells were culture expanded and sorted to obtain both polyclonal and single clone populations based on GFP expression using a Becton Dickinson FACSTAR sorter. Western blot analysis. Conditioned medium was collected from confluent control GFP and hK5His-transduced U87 g1ioma cells, as weil as stably translëcted hK5His 293 T cells (positive control), concentrated, and detected by anti-His immunoblot analysis as previously mentioned. 4 4 Migration auays. HUVEC (I X 10 ) or human macrophages (5 x 10 ) were plated onto 0.15% gelatin/PBS-coated 8-flm pore chemotaxis membranes (Corning. Acton, MA) within Boyden chamber inserts. Migration assays were done as previously described (26). HUVEC and macrophages were exposed to conditioned medium from control GFP and hK5His transduced US7 glioma cells for 4 and 24 hours, respectively, under growth factor-slimulated conditions (EBM-2 SingleQuots) and serum stimulation, respectively. Each sample was tested in triplicate and the Figure 1. Schematic illustration 01 human plasminogen (Plg) and its kringle average number of migrating cells per field was assessed by counting three domains. A, plasminogen is composed 01 791 amino acids with a single N-linked glycosylation at Asn289 and a single O-linked glycosylation at Thf"4S. random high-power fields per mter. Angiostatin, a well-known inhibitor 01 angiogenesis, encompasses the lirst ln vivo Malrigel assay and analysis of cellular infIltrate by three to lour kringle structures 01 plasminogen. This is a modilied schematic cytornetry. Culture-expanded U87-GFP and U87-hK5His-GFP celIs were courtesy 01 Dr. M.R. Uinas et al., Carnegie Mellon University, Pillsburgh, 4 aliquoted to create 5O-fiL cell suspensions containing 2 x 10 celIs, rnixed PA (hllp://www.chem.cmu.edu/groupslUinaslreslstructurelhpk.html). 8, hK5 consists 01 the last cryptic Iragment 01 plasminogen (CYS'52 _Cys541) with 500 fiL Matrigel (BD Biosciences, Sanjose;CA) at 4°C, and implanted and is composed 01 8O-amino-acid residues with three distinct disuHide bonds. s.c. in the right lateral flank of 6-week-old nonobese diabetic/severe

Cancer Res 2005; 65: (18). September 15,2005 8360 www.aacrjournals.org K5 Possesses Angiostatic and Antimacrophage Properties eombined immunodefieient (NOD-SClD) miee (Charles River, Laprairie, predicted molecular weight of hK5His (Fig. 2B). To fully Quebee, Canada). Four weeks after implantation, miee \Vere sacrificed and characterize cystinyl bridge structure and confirm kringIe domain implants exeised and processed as pre\~ously deseribed (27). To determine tertiary structure, an orthogonal digestion was carried out either the expIant cellular infiltrate, cell suspensions were stained with the with agarose-immobilized trypsin alone (Fig. 2C) or with trypsin follo,,~ng antibodies: purified rat anti-mouse CDI6/CD32 (mouse Fe block) followed by incubation with Asp-N endopeptidase (Fig. W) under followed by APC-eonjugated rat anti-mouse CD45, PE-conjugated rat anti­ reducing and nonreducing conditions. Peptide mapping of non­ mouse Mac-3, biotin-conjugated rat anti-mouse Ly-6G (Gr-I), and their corresponding isotypie controls. Biotinylated antibodies \Vere revealed using reduced single and dual protease hK5His digests by MALDI-Q TOF PE-Cy7-strepta~din (BD PharMingen, San Diego, CA). Cells were fIXed with mass spectrometry confirmed the presence of disulfide bridges 1% paraformaldehyde and events were acquired using a FACSCalibur fiow among CYSJ7_Cys96, Cys38_Cys79, and Cys67_Cys9J (by deduction). cytometer (Beckman Coulter, Fullerton, CA) and analyzed using the These findings show that plasminogen hK5His domain spontane­ CelIQuest software (BD PharMingen). ously adopts a tertiary structure consistent with native conforma­ Matrigel expIant histochemistry. Explants were fixed in formalin, tion as displayed within plasminogen when expressed by embedded in paraffin, and 4-f1m sections were prepared to generate genetically engineered eukaryotic cells. representative sections of the border and central regions of the explants. hK5His protein secreted by retrovirally gene-modified Sections \Vere treated with xylene, rehydrated, and antigen retrieval was done human g1ioma cells inhibits in vitro endothelial cell migra­ using Iwo microwave boils in a 10 mmol/L sodium citrate buffer (pH 6.0) tion. The hK5His cDNA was c10ned into our previously published solution. Endogenous biotin acti~ty was blocked using a kit (Zymed Laboratories, M arkh am, Ontario, Canada), sections were subsequently (24) bicistronic retroviral vector (Fig. 3A). The hK5His retrovector blocked with 2.5% bo~ne serum a1bumin in PBS and incubated with a rabbit plasmid was stably transfected into 293GPG retroviral packaging polyclonal raised against murine von Willebrand factor (vWF; Neomarkers, Fremont, CA dilution 1:100) overnight. After three washes, the sections were incubated with biotinylated goat anti-rabbit IgG antibody (BD PharMingen, B dilution 1:200) for 2 hours, washed, and incubated with strepta~din­ peroxidase (Vector Labs, Burlingame, CA) for 1 hour before the addition of 3,3' -diaminobenzidine chromogenic substrate (Vector Labs). Meyer's hema­ toxylin was used for counterstaining. The blood vessel surface area of ail vWF­ positive vessels was calculated using a Leica light microscope ocular micrometer (Leica Microsystems, Inc., Richmond Hill, Ontario, Canada) at 400x magnification. Two observers counted three random sections per expIant in each experimental group. The blood vessel surface area to total section surface area was calculated and expressed as a percentage. Stereotactic intracerebral surgery. COI nu/nu female 6-week-oId athymic nude mice (Charles River) were anesthetized using a ketamine/ xylezine/saline/acepromazine cocktail (100 mg/mL. 20 mg/mL. 0.9%, 10 mg/mL. respectively) dosed at 100 f1L/lOO g by i.p. injection. Mice were secured in stereotactic apparatus (Kopf Instruments, Tujunga, CA) and incisor bar set to -1.5. Midline incision made on scalp to expose Bregma c and Lambda Coordinates from Bregma were AI', +0.5; LM, -2.0; and DY, -4.4. A burr hole \Vas made and a Hamilton syringe (IO-f1L 26-gauge needle, ·0 Hamilton Co., Reno, NV) was gently lowered. Cell suspension (1 x Hl" in 3 f1L of HBSS) was injected at a rate of 0.25 f1L per 3 minutes for 36 minutes. Brain tissue analysis. After euthanasia, brains were processed for H&E ~[~- staining as pre~ously described (28). Digital images \Vere retrieved using an . ,.. .. ------_.~-_._~--- -=----- OIympus microscope (Olympus America, Mel~lIe, NY). Tumor volumes were caIculated using the formula, length x width2 x 0.4. D r,.,,.,,,. ~.,4f btf ...... _. ~ .... ~ ..... Results Matrix-assisted laser desorption ionization-quadrupole time-of-flight mass spectrometry analysis of expressed soluble hK5His peptide confirms predicted disulfide bridging confor­ mation. The 381-bp human plasminogen K5 domain cDNA, corresponding to amino acid residues 447 to 546 of plasminogen, includes a 20-amino-acid IL-2 signal sequence and a His-tag at the Figure 2. Proteomic analysis of purified soluble hK5His. A, silver stain done on COOH terminus (Fig. lB) and encodes for the hK5His protein with pooled eluted fractions detects purified protein at the expected molecular weight a kringle structure as predicted by previously published nuclear of 15 kDa (top). Anti-His immunoblot analysis demonstrates the presence of soluble hK5His peptide in pooled eluted fractions (boNom). B, analysis of the magnetic resonance (NMR) structure study (29). Mfinity chroma­ purified hK5His peptide by MAlDl-Q TOF mass spectrometry revealed a major tography was done on precIeared conditioned medium from stably peak at 12,108 Da consistent with the theoretical molecular weight of monomeric transfected hK5His-expressing 293T cells. Anti-His immunoblot, K5 at 12.109.6 Da. The doubly charged form and dimer are detectable at 6,059 and 24,195 Da, respectively. C, expected peptide fragment of interest following SDS-PAGE separation (Fig. 2A. top), shows a major band generated for hK5His protein digested with trypsin alone under oxidizing of 15 kDa (Fig. 2A. bottom). NH2-terminal amino acid sequencing conditions (/eft). C, MALDI-Q TOF mass spectra confirming Cys38:79 disulfide indicated proper c1eavage of the IL-2 signal sequence directIy bond within hKSHis (right). D, expected peptide fragment of interest generated 20 for hK5His protein digested with trypsin followed by Asp-N endopep,tidase upstream of Ser and analysis by MALDI-Q TOF mass spectrom­ treatment under oxidizing conditions (/eft). Spectra confirming Cys 7:96 and etry revealed a major peak at 12,108 Da consistent with the Cys38:79 disulfide bonds (right). www.aacrjournals.org 8361 Cancer Res 2005; 65: (18). September 15, 2005 Cancer Research

whether hK5His could interfere with the tumor microenvironment. More specifically, we assessed whether soluble hK5His could affect the recruitment of host-derived tumor-infiltrating endothelial as weIl as inllammatory ceUs implicated in tumor angiogenesis. To 4 test this hypothesis. 2 x 10 U87-GFP (n = 10) or U87-hK5His-GFP (n = 10) ceUs were Matrigel embedded and implanted s.e. in NOD­ C UST U81.(iFP SCID mi ce. To verify if tumor-secreted hK5His induced an antiangiogenic effect in vivo, explants (n = 4) in each experimental group were retrieved 4 weeks after implantation (Fig. 4A) and o stained with vWF antibody (Fig. 4B). The section surface area 1 "P<'J)9Ç4 ! occupied by vWF-positive blood vessels to total section surface 1:: ! ~ -0' 1 #" 1 •

Figure 3. Development and characterization 01 hK5His·GFP gene·modified cens. A. schematic illustration 01 the human hK5His·IRES·EGFP retroviral plasmid construct. The retrovector pIRES·EGFP construct served as the control plasmid. B, Ilow cytometry analysis 01 untransduced (gray) and hK5His· transduced (black) human U87 glioma cens. C, anti·His immunoblot analysis reveals lunctional secretion 01 hK5His migrating as a doublet at -15 kDa Irom retrovirany gene-modilied U87 glioma cens and stably translected 293T cens (positive control). D, conditioned medium Irom hK5His-transduced glioma cens inhibits HUVEC migration up to 75 ± 3%. Bars, SE. Statistical analysis was done using Student's t test. B cells and retrovirus producer cells were selected as described in Materials and Methods. Tetracycline withdrawal from the culture medium led to the production of VSV-G-typed hK5His-GFP retroviral particles that were subsequently concentrated lOO-fold by ultracentrifugation and a viral titer of - 2.5 x 106 infectious particles/mL was obtained (data not shown), Concentrated VSV-G­ typed hK5His-GFP retroviral particles were used to transduce human g1ioma cells. The U87MG human g1ioblastoma cell line was selected for this study because it overexpresses vascular endothelial growth factor (VEGF)-producing tumors in vivo that are charac­ terized by increased vascularity, lack of necrosis, and absence of peripheral invasion in early tumor development (1). Following retroviral transduction, single U87MG clones were selected and assessed for GFP expression by flow cytometry. The U87 single clone used for the study was sorted based on GFP expression and was 99% GFP positive (Fig. 3B). To ensure hK5His transgene expression and proper secretion, anti-His immunoblot analysis was done on conditioned supernatant collected from hK5His transduced human U87 g1ioma cells and detects a major 15-kDa protein consistent with 0.6 1\1 the predicted molecular weight of soluble hK5His (Fig, 3C), Using ...QI 0.5 a semiquantitative Western blot titration curve, it was estimated ~ > 004 that hK5His-GFP-expressing U87 cells secrete 0.5 pmol/L (or 7.5 ng) ID 6 0.3 of soluble hK5His per 10 cells per 24 hours (data not shown). To 0~ test the functionality of secreted hK5His from hK5His-GFP­ 0.2 expressing U87 cells, a HUVEC migration assay was done as it has * 0.1 been previously shown that recombinant K5 protein is capable of inhibiting the migration of bovine capillary endothelial cells (30). 0 Soluble hK5His present in the conditioned medium of hK5His-GFP­ expressing U87 cells was capable of inhibiting growth factor­ Figure 4. hK5His acls as a potent antiangiogenic agent. A, implant (alTOW) induced migration ofHUVEC up to 75 ± 3% (P < 0.0004) compared retrieval4 weeks alter implantation 01 2 x 10' Matrigel-embedded U87-GFP with conditioned medium from GFP-expressing U87 ceUs (Fig. 3D). or U87-hK5His-GFP celfs (representative images are shown at Ihe same magnificalion). 8, representalive images of vWF-posilive immunoslained blood This confirms the in vitro functionality of hK5His secreted by vessels (recf arrows) lrom two differenl U87-GFP-conlaining (a and b) and hK5His-GFP-expressing U87 ceUs. U87-hK5His-GFP-containing (c and cf) implants; magnification, 2OOx. hK5His blocks glioma-associated angiogenesis and pos­ C, vWF-positive blood vessel surface area to lotal seclion surface area ratio is plotted as a percentage. Bars, SE (SE for Matrigel-only group was based on sesses nove) antimacrophage properties. To further characterize three random section counts Irom one animal because remaining implants the mechanism of action of soluble hK5His in vivo, we investigated resorbed completely). Stalistical analysis was done using Studenl's 1 test

Cancer Res 2005; 65: (18). September 15, 2005 8362 www.aacrjournals.org K5 Possesses Angiostatic and Antimacrophage Properties area ratio was calculated and expressed as a percentage (Fig. 4C). (P = 0.03). It is important to note that the difference in absolute Our results show that soluble hK5His domain is capable of exerting tumor cell number is not due to differences in the proliferation a potent antiangiogenic effect (P = 0.04) in vivo when secreted by rates of the control GFP and hKSHis-GFP glioma cells because both retrovirally engineered g1ioma cells. The remaining explants (n = 6) possess equivalent proliferation rates in vitro (Supplementary in each experimental group were dissolved in a collagenase Fig. 1). Although immunocompromised NOD-SCm mice possess solution and single-cell suspensions were generated and counted. functional macrophages and granulocytes, they lack functional (Fig. SA, left). The hK5His-GFP explants contained about two thirds T and B lymphocytes. Flow c)'tometry anal)'sis after three-color less cells than the control GFP explants (P = 0.01). The cellular staining indicated that a substantial part of the cellular infiltrate infiltrate in the explants was analyzed for GFP expression as weil as consisted of CD45-positive hematopoietic cells (Fig. 5B, left). More for cell surface marker expression to distinguish the various cell specificall)'. there was a reduced absolute number of recruited types present within the explants (Fig. SA, right). The GFP-positive CD4S+Mac+Gr1+ (P = 0.04) cells, which represents the monocyte/ cell fraction representing g1ioma cells is significantly reduced in the macrophage fraction and CD45'Mac3+GrC (P = 0.03) cells, which hK5His-GFP explants compared with the control GFP explants represents the macrophages present in the U87-hK5His-GFP­ containingexplants compared with the control U87-GFP-containing explants (Firr 5B). There was no significant difference in the number A 14 of CD4S+Mac3-Gr1+ (P = 0.43) cells representing the infiltrated ~ 012 D U87-GFP granulocytes present in the explants. X • U87-K5-GFP ~10 To confirm our in vivo observations, we tested whether hK5His ~ could inhibil macrophage migration in vitro. A migration assay was ~ 8 :1 done using human monocytes isolated from peripheral blood Z 6 mononuclear cells that were stimulated with GM-CSF to differen­ ~ 4 tiate into CD206+ macrophages (data not shown). The ability of ! 2 macrophages to migrate through a gelatin matrix was assessed ~ o * under serum-stimulated conditions, where macrophages were directl)' exposed to conditioned medium collected from GFP­ B expressing g1ioma cells and hK5His-GFP-expressing g1ioma cells. ••• P= 0.03 ~3.5 o U87-GFP The results indicate that soluble hK5His decreases the migration E 3 • U87-K5-GFP potential oC human macrophages b)' - 60% compared with the ~ 2.5 GFP control (P = 0.007, Fig. 5C). E :1 2 Glioma-largeted hK5His expression suppresses tumor Z 'i 1.5 growth and prolongs survival in an experimental orthotopic ~1 brain cancer modeJ. To test the efficacy of retrovirally engineered S g1ioma cells secreting hK5His in a c1inieally relevant proof-of­ ~05 5 o principle experiment, 10 U87-GFP (n =5) and U87-hK5His-GFP cells derived from either a single clone (n = 5) and an independentl)' generated polyclonal population (n = 5) were implanted intracere­ c brally in athymie nu/nu (nude) mice. Mice were sacrificed 32 days after implantation and brains were removed and processed for H&E staining (Firr 6A). Tumor volumes from each experimental group were estimated from H&E-stained cryostat sections and calcu­ 350 lated as described in Materials and Methods. Consistent with our u. 300 previous study (28), brain tumors in the GFP control group had Q. :I: 250 extensive growth. In contrast, the U87-hK5His-GFP-implanted ] CD 200 miee possessed significantly reduced mean tumor volumes t) (0.35 ± 0.18 mm3 for the single clone population and 0.07 ± "C 150 .. 0.03 mm3 for the polyclonal population) compared with the control ~ 100 3 CI U87-GFP-implanted (12.23 ± 6.07 mm ) mice (P = 0.04; Fig. 6B). :ë 50 We did a long-tenn in vivo study, where we evaluated the o~--~----~----~--~ survival of the therapeutic cohort implanted with U87-hK5His-GFP Supernatant Source cells (n = 15) and the control U87-GFP-implanted (n = 10) miee. Although ail control miee succumbed to excessive tumor burden by Figure 5. hK5His possesses nover antimacrophage property and decreases - 8 weeks, the U87-hKSHis-GFP-implanted miee possessed a c1ear migration potential of human macrophages. A, absolule lolal cell counl survival advantage with 53% of the mice surviving over 120 days performed alter collagenase digestion 01 Ihe explanls (Ieft). Absolute number < 01 GFP positive ceUs was enumerated using Ilow cytometry and represents the (P 0.0001; Fig. 6C). GFP-marked glioma ceUs (righ/). Bars, SE. B, absolute number 01 inliltrated CD45+ hematopoietic ceUs, CD45+Mac3+Grl + monocytes/macrophages, CD45+Mac3+Grl- macrophages and CD45+Mac3-Grl+ granulocytes in the Discussion explants. Bars, SE. Statistical analysis was done using Student's t test. C, macrophages were exposed to various conditioned medium lor 24 heurs This present study shows that eukaryotieally expressed secreted under serum stimulation. Basal invasion represents migrating macrophages hK5His protein preserves the prototypieal native kringle disulfide under serum·lree conditions. Bars, SE. Inse/s, representative stains 01 migrated ceUs lrom each experimental group; magnification, 200 x . Statistical analysis bridging conCormation deemed essential for the angiostatie activity was done using Student's / test. of plasminogen-derived peptide domains. Indeed, our cysteine www.aacrjournals.org 8363 Cancer Res 2005; 65: (18). September 15, 2005 Cancer Research

Gonzalez-Gronow et al. (3S) suggest that K5 mediates its effects by

B 20 binding to the human voltage-dependent anion channel, which is a ·P"'O.64 ;:;- 18 receptor present on hum an endothelial cells. Vnlike K5, the E " É. 14 mechanism of action of angiostatin has been studied more Ë 12 extensively and its endothelial cell binding sites include the cell "0" " > surface-associated ATP synthase receptor (36), angiomotin (37), ~ 6 o E • and integrin IXvr'>3 (38). Although angiostatin mediates its inhibitory sc pc ?!. 2 __...l...... - action primarily on endothelial cells, it has been reported by GFP K5-GFP Moulton et al. (39) that it reduces plaque angiogenesis and ...... -U87-GFP atherosclerosis by reducing plaque-associaled macrophages. How­ __ Ue7-K5-GFP ever, Moulton et al. suggest that angiostatin, at doses that reduce endothelial cell sprouting, does not directly inhibit monocytes because it does not inhibit VEGF-induced monocyte migration in vitro. Thus, our findings provide novel mechanistic insight and suggest that soluble hKSHis protein inhibits two unique cell types 75 175 225 375 Days ." most critical for tumor angiogenesis, mediating a dual role as an antiangiogenic agent via its antiendothelial cell properties as weil as an antitumor agent by reducing the infiltration of host-derived Figure 6. Suppression 01 intracerebral U87MG glioma groWlh by soluble hK5His. A, H&E·stained brain tissue sections Irom control U87·GFP implanted macrophages within the tumor mieroenvironment. mice (n = 5) and U87·hK5His·GFP implanted mice (n = 5) 32 days alter We used a c1inically relevant orthotopic implantation approach implantation. Representative stains are shown at the same magnificat ion, 40x. to test wh ether soluble hKSHis could suppress g1ioma progression B, tumor volumes were estimated Irom postmortem H&E·stained brain sections obtained Irom mice implanted with U87·GFP, U87-hK5His-GFP single clone after implantation of hK5His-GFP-expressing V87 cens in athymie (sc), and U87-hK5His-GFP polyclonal (pc) ceUs. The experiment was done twice nu/nu mice. Stereotactic implantation procedures are particularly with similar results. Bars, SE. Student's t test. C, Kaplan-Meier long-term suitable for g1ioma gene therapy approaches allowing for constant survival curve 01 mice implanted with control U87-GFP (n = 10) and U87-hK5His-GFP (n = 15) cells. Log-rank statistical test. long-lasting transgene expression within the local tumor microen­ vironment and minimize the risk of toxicity as is often encountered with systemic therapies. Our data successfully shows that soluble bridging pattern data confirms the results of the only other study hKSHis significantly decreases tumor growth by - 97% compared that characterizes the structure of recombinant KS domain using with controls. In doing so, soluble hK5His also promotes long-term NMR technology (29). We have shown that VSV-G-pseudotyped survival, with S3% of U87-hKSHis-GFP-implanted mice surviving retroviral vectors engineered to express human plasminogen past 120 days. We postulate that upon stereotactic implantation of hKSHis domain can efficiently gene modify human U87 g1ioma our ex vivo retrovirally engineered hK5His g1ioma cens, stable cens ex vivo. We also show that hKSHis-engineered U87 cells can integration and sustained expression of the hKSHis transgene in secrete biologically functional soluble hKSHis protein capable target U87 cens provided continuous release of soluble hK5His of inhibiting growth factor-stimulated endothelial cell migration within the loco-regional tumor microenvironment, resuIting in in vitro. potent suppression of tumor growth. We speculate that tumor­ Moreover, our in vivo data in immunodeficient NOD-SCm mice targeted delivery of hKSHis via viral vectors such as pseudotyped suggests that hKSHis secreted by Matrigel-embedded U87 engi­ retrovectors (24) and other gene delivery platforms would lead to neered cells acts as a potent antiangiogenic agent by significantly similar results. decreasing blood vessel formation as weil as a novel antimacro­ ln summary, we provide compelling evidence that secreted phage agent by significantly preventing the recruitment of hKSHis prolein expressed by engineered g1ioma cens possesses host-derived tumor-associated CD4S+Mac3+Grl +/- monocytes! potent antiangiogenic as weil as novel antimacrophage properties macrophages within the local tumor microenvironment. The that block g1ioma progression and promote survival in an involvement of tumor-associated macrophages in angiogenesis orthotopic experimental brain cancer model. Further studies on and tumor progression is a well-established concept (31-33). the pleiotropic effects of KS peptide on host-derived angiogenic Activated macrophages produce tumor necrosis factor-a and IL-l, and macrophage response may offer new avenues of exploration inducing the production of VEGF and IL-8 in g1ioma cells, which for the development of therapeutic strategies for the treatment of further attracts and activates macrophages (34). Macrophage g1ioma and other loco-regional refractory malignancies snch as infiltration has been c10sely correlated with neovascularization prostate and ovarian cancer. and observed more frequently in grade 4 g1ioblastomas than in grade 2 or 3 g1iomas (34). Acknowledgments It may be speculated that the potent antiangiogenic properties of Received 2/15/2005; revised 6/27/2005; aceepted 7/7/2005. soluble K5 mediating a reduction in tumor-associated endothelial Grant supperi: Canaruan Prostate Cancer Resean:h lnitiative-Canadian Institutes cens, as evidenced by the lower surface area occupied by blood of Health Research doctoral award (S.R. Perri), V.S. Army Medical Research and Materiel Co~ Breast Cancer Research predoctoral traineeship award DAMDI7- vessels in the U87-hK5His-GFP-containing explants, is partially 03-1-0545 (S.R. Perri). Canadian Institutes of Health Research Clinician-Scientist responsible for the decreased infiltration of host-derived macro­ award MOP-I5017 (J. Galipeau). National Scholar of Fonds de la Recherche en Santé phages within the tumor area. We also show that soluble hK5His du Québec-Foods pour la fonnation de Chercheurs et rAide à la Recherche (J. Nalbantoglu). Killam scholar (J. Nalbantoglu), National Cancer Institute of Canada independently reduces macrophage migration in vitro, suggesting (J. Nalbantoglu). and Valorisation-Recherche Québec (J. Nalbantoglu). B. Annabi holds that the K5 domain can impart a direct inhibitory effect on a Canada Resmn:h Chair in Molecular Oncology from the CIHR. The costs of publication of this article were deCrayed in part by the payment of page macrophage motiIity. To date, the K5 peptide has been consistently charges. This lllticle must therefore be hereby marked advertisement in accordance characterized as an endothelial cell-specific inhibitor (19, 30) and with 18 V.S.c. Section 1734 solely to indicate this fact.

Cancer Res 2005; 65: (18). September 15, 2005 8364 www.aacrjournals.org K5 Possesses Angiostatic and Antimacrophage Properties

References 15. O'Reilly MS. Boehm T. Shing Y. et al. Endostatin: an 28. Li H, onso-Vanegas M, Colicos MA. et al. Intracere­ endogenous inhibitor of angiogenesis and tumor bral adeno\'Îrus-mediated p53 tumor suppressor gene l. Bello L. Giussani C. Carrabba G. Pluderi M. Costa F. growth. Cell 1997:88:277-85. therapy for experimental human g1ioma Clin Cancer Bikfalvi A. Angiogenesis and invasion in Gliomas. In: 16. Shir A. Levitzki A. Gene lherapy for glioblastoma: Res 1999-,5:637 - 42. Kirsch M, Black PM. editors. Norwell: Klm"er: 2004. future perspective for delivery systems and molecular 29. Chang Y. Moehalkin 1. MeCance SG, Cheng B, p.263-84. targets. Cell Mol Neurobiol 2001;21:645-56. Tulinsky A Castelli no FI. Structure and ligand binding 2, Holland Ee. Glioblastoma multiforme: the terminator. 17. Kirsch M. Strasser J. Allende R, Bello L, Zhang J. detenninants of the recombinant kringle 5 domain of Proc Natl Acad Sci USA 2000:97:6242-4. Black PM. Angiostatin suppresses malignant glioma human plasminogen. Biochemistry 1998:37:3258-71. 3. Black PM. BTain tumOT. Part 2. N Engl J Med 1991; growth in vivo. Cancer Res 1998;58:4654-9. 30, Ji \NR, Barrientos LG, L1inas M, et al. Selective 324:1555-64. 18. O'Reilly MS, Holmgren L, Shing Y. et al. Angiostatin: a inhibition by kringle 5 of human plasminogen on 4. Folkman J. Tumor angiogenesis: therapeutÎc implica­ nove} angiogenesis Înhibitor that mediates the suppres­ endothelial cell migration, an important pTocess in tions. N Engl J Med 1971;285:]]82-6. sion of metastases by a Lewis lung carcinoma Cell angiogenesis. Biochem Biophys Res Commun 1998;247: 5, Maher EA, Furnari FB, Bachoo RM, et al. Malignant 1994;79-.315-28. 414-9. g1ioma: genetics and biology of a grave matter. Genes 19. Cao Y, Chen A, An 55, Ji RW. Davidson D, L1inas M. 31. Leek RD, Lewis CE, "''hitehouse R, Greenall M, Dev 2001;15:13]]-33. Kringle 5 of plasminogen is a nove) Înhibitor of Clarke J, Harris AL Association of macrophage infiltra­ 6, Puduvalli VK. Inhibition of angiogenesis as a thera­ endothelial cell growth. J Biol Chem 1997:272:22924-8. tion with angiogenesis and prognosis in invasive breast peu tic strategy against brain tumors. In: KiTsch M, 20. Cao Y, Ji RW. Davidson D. et al. Kringle domains carcinoma Cancer Res 1996:56:4625-9. Black PM, editors. Norwell: Kluwer; 2004. p. 307-36. of hum an angiostatin. Char3cterization of the anti­ 32. Richter G. Kruger·Krasagakes S, Hein G, et al. 7. Bello L, Carrabba G. Giussani C, et al. Low-dose proliferative activity on endothelial cells. J Biol Chem Interleukin 10 tr.nsfected into Chinese hamster ovary chemotherapy combined with an antiangiogenic drug 1996:271 :29461-7. cells prevents tumor growth and macrophage infiltra­ reduces human gliorna growth in yivo. Cancer Res 21. Cao R, Wu HL. Veitonmaki N, et al. Suppression of tion. Cancer Res 1993:53:4134-7. 2001;61:7501-6. angiogenesis and tumOT growth by the inhibitor KI-5 33. Sunderkotter C. Steinbrink K, Goebeler M, Bhardwaj 8, Bello L, Giussani C, Carrabba G, et al. Suppression of generated by plasmin-mediated proteolysis. Proc Nat! R, Sorg e. Macrophages and angiogenesis. J Leukoc Biol malignant glioma recurrence in a newly developed Acad Sci USA 1999:96:5728-33. 1994;55:410-22. animal model by endogenous inhibitors. Clin Cancer 22. Ory DS, Neugeboren BA. Mulligan Re. A stable 34. Nishie A, Ono M, Shono T, et al. Macrophage Res 2002;8:3539-48. human-derived packaging cell line for production of infiltration and heme oxygenas.. l expression correlate 9. Gorski DH, Beckett MA, Jaskowiak NT, et al. Blockage high titer retrovirus/vesJcuJar stomatitis virus G pseu­ with angiogenesis in hum an gliomas. Clin Cancer Res of the vascuJar endothelial growth factor stress response dotypes. Proc Nat! Acad Sei USA 1996;93:11400-6. 1999',5:1107-13. increases the antitumor effects of ionizing radiation. 23. Ponten J, Macintyre EH. Long term culture of normal 35. Gonzalez-Gronow M, Kalfa T, Johnson CE, Gawdi G, CanceT Res 1999;59:3374-8. and neoplastic human g1ia. Acta Pathol MicTobiol Seand Pizzo SV. The voltage-dependent anion channel is a 10, Griscelli F, Li H, Cheong C, et al. Combined effects of 1968;74:465-86. receptor for plasminogen kringle 5 on human endothe­ radiotherapy and angiostatin gene therapy in g1ioma 24. Galipeau J, Li H, Paquin A, Sicilia F, Karpati G, liai cens. J Biol Chem 2003;278:27312-8. tumoT mode!. Proc Nat! Acad Sei USA 2000;97: Nalbantoglu}. Vesicular stomatitis virus G pseudotyped 36. Moser TL, Stack MS, Asplin l, et al. Angiostalin binds 6698-703. relroveclor mediates effective in vivo suicide gene ATP synthase on the surface of human endothelial cells. Il, Hwu W], Raizer J, Panageas KS, Lis E. TTeatment of delivery in experimental brain cancer. Cancer Res Proc Nat! Acad Sci USA 1999:%:28]]-6. metastatic melanoma in the brain with temozolomide 1999',59:2384-94. 37. Troyanovsky B, Levchenko T, Mansson G, Matvijenko and thalidomide. Lancet Oncol 2001:2:634-5. 25. Eliopoulos N, Galipeau J. Green fluorescent pTotein. 0, Holmgren L. Angiomotin: an angiostatin binding 12, Li L, Rojiani A, Siemann DW. Targeting the tumor ln: Hicks BW, editor. Totowa: Humana Press Inc.; 2002. protein that regulates endothelial cell migration and vaseulaLure with combretastatin A-4 disodium phosphate: p.353-71. tube formation. J Cell Biol 2001:152:1247-54. effects on radiation therapy. Int J Radiat Oncol Biol 26. Annabi B, Lachambre MP, Bousquet-Gagnon N, 38. Tarui T, Miles LA Takada Y. Specific interaction of Phys 1998;42:899-903. Page M, Gingras D, Beliveau R Green tea polyphenol angiostatin with integrin a(v)J>(3) in endothelial cells. 13. Mauceri H), Hanna NN, Beckett MA, et al. Combined (-)-epigallocatechin 3-gallate inhibils MMP-2 secretion J Biol Chem 2001;276:39562-8. effects of angiostatin and ionizing radiation in anti­ and MTl-MMP-driven migration in g1ioblastoma cells. 39. Moulton KS, Vakili K, Zurakowski D, et al. tumour therapy. Nature 1998;394:287-91. Biochim Biophys Acta 2002;1542:209-20. Inhibition of plaque neovascularization reduces mac­ )4. O'Reilly MS. Holmgren L, Chen C, Folkman J. 27. Stagg J, Lejeune L, Paquin A, Galipeau J. Marrow rophage accumulation and progression of advanced Angiostatin induces and sustains dormancy of human stromal cens for interleukin-2 delivery in cancer atheroselerosis. PTOC Nat! Acad Sei USA 2003;100: primary tumors in miee. Nat Med 1996;2:689-92. immunotherapy. Hum Gene Ther 2004;15:597-608. 4736-41.

www.aacrjournals.org 8365 Cancer Res 2005; 65: (18). September 15, 2005 HÔPITAL GÉNÉRAL JUIF UNIVERSITÉ SIR MORTIMER B. DAVIS MCGlll JEWISH GENERAL HOSPITAL UNIVERSITY

INSTITUT LADY DAVIS DE RECHERCHES MÉDICALES LADY DAVIS INSTITUTE FOR MEDICAL RESEARCH

MARK A. WAINBERG, O.C., Ph.D., F.R.S.C. PROFESSEUR TITULAIRE - FACULTË DE MËDECINE DIRECTEUR DE LA RECHERCHE DIRECTEUR, CENTRE SIDA DE L'UNIVERSITË McGILL August 8, 2006

RE: Sabrina Perri Thesis submission to McGill University

To Whom It May Concern:

The foHowing will confinn the safety and ethical conditions with regard to aH laboratories at the Lady Davis Institute for Medical Research of the Sir Mortimer B. Davis Jewish General Hospital.

The Lady Davis Institute for Medical Research is a free-standing biomedical research institute attached to the Sir Mortimer B. Davis Jewish General Hospital, a major adult teaching hospital affiliated with McGill University. There is approximately 110,000 sq. ft. of laboratory space occupied by over 65 independent investigators aH of whom hold McGill University appointments in appropriate departments. At the present time we have approximately 100 graduate students and post-doctoral feHows registered with McGill University. Total external funding is in the range of $18 million (exclusive of overhead), which is large by Canadian standards.

Professor Andrew Mouland functions as Chair of our Health and Safety Committee, while Professor John Hiscott, a virologist, heads the subcommittee of the Health and Safety Committee on Biohazards.

Dr. Mouland affinns the safety program which addresses the question of radioactive material licence and Dr. Hiscott affinns the question of the use of recombinant DNA in the experiments. Dr. Lorraine Chalifour affinns the ethical use of animaIs.

As a constituent part of McGill University the Institute foHows and meets aH of the requirements of the McGill Laboratory Biosafety Manual. For those rare circumstances not covered by the McGill Laboratory Biosafety Manual we use the following as reference guidelines:

3755, CHEMIN DE LA CÔTE-STE-CATHERINE. MONTRÉAl, QUÉBEC H3T 1 E2 1 TÉL.: (514) 340-8260 FAX: (514) 340-7502 Continuation .. ./2 Letter-August 8, 2006 Sabrina Perri Thesis submission to McGill University

- Biosafety in Microbiological and Biomedical Laboratories, 3rd edition (1993). U.S. Department of Health and Human Services (Public Health Service), Centers for Disease Control and Prevention, and the National Institutes of Health. U.S. Government Printing Office, Washington, D.C.

- Laboratory Biosafety Manual, 2nd edition (1993). World Health Organization, Geneva, Switzerland.

- Laboratory Biosafety Guidelines (1990). Medical Research Council of Canada and Laboratory Centre for Disease Control, Health Protection Branch, Health Canada, Ottawa, Canada.

- Laboratory Safety, 2nd edition (1995). D.O. Fleming, J.H. Richardson, J.1. Tullis and D. Vesley, editors. ASM Press, Washington, D.C.

- Guidelines for Research Involving Recombinant DNA Molecules (1996). U.S. Department of Health and Human Services, National Institutes of Health, Bethesda, M.D.

Regular inspection of safety procedures are conducted by the Canadian Nuclear Safety Commission (CNSC) (annual), Contex Environment (laboratory health and safety consultants - several times per year), the Sir Mortimer B. Davis Jewish General Hospital Chairman of Infection Control - Dr. Mark Miller (monthly), the Canadian Council on Animal Cafe (bi-annually), the McGill animal Care Committee (monthly), as weIl as by the Hospital Fire Marshal - Mr. Thomas Prokos, and the City of Montreal Fire Department (semi-annually).

The Lady Davis Institute (LDI) laboratories in which the research will be conducted are specially designed and equipped for scientific research and are used solely for that purpose. Specialized facilities are provided for the safe conduct of scientific research, including but not limited to, waste disposaI facilities for biohazardous, radioactive and chemical wastes, a level 3 containment facility and state-of-the-art animal facilities. An automatic sprinkler system (water) is installed throughout the building. Fire extinguishers are located in several locations on each of six floors, as are fire alarms. Fire doors and alarm pulls are marked. In general, the laboratories and animal rooms are designed to have pressure differentials negative to adjacent areas. Fume hoods are monitored on a semi-annual basis. Fume hoods exhaust to the outside. Biological safety cabinets are certified semi-annually.

2