Progression, Angiogenesis, and Response to Chemotherapy. Allison Shayna Betof. Department of Pathology Duke University.

Transcription

1 Therapeutic Aerobic Exercise in a Mouse Model of Breast Cancer: Effects on Tumor Progression, Angiogenesis, and Response to Chemotherapy by Allison Shayna Betof Department of Pathology Duke University Date: Approved: Mark W. Dewhirst, Supervisor Lee W. Jones Salvatore V. Pizzo Christopher D. Kontos Zeljko Vujaskovic Dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Pathology in the Graduate School of Duke University 2012

2 ABSTRACT Therapeutic Aerobic Exercise in a Mouse Model of Breast Cancer: Effects on Tumor Progression, Angiogenesis, and Response to Chemotherapy by Allison Shayna Betof Department of Pathology Duke University Date: Approved: Mark W. Dewhirst, Supervisor Lee W. Jones Salvatore V. Pizzo Christopher D. Kontos Zeljko Vujaskovic An abstract of a dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Pathology in the Graduate School of Duke University 2012

3 Copyright by Allison Shayna Betof 2012

4 Abstract Over the past decade, exercise has gained increasing attention from both clinicians and patients as a beneficial adjunct therapy to maintain or enhance quality of life in breast cancer patients. Recent epidemiological studies indicate that aerobic exercise following a diagnosis of breast cancer may be associated with reductions in cancer- specific and all- cause mortality. However, the mechanisms by which physical activity affects tumor physiology are poorly understood. Accordingly, clinicians lack critical information to properly advise breast cancer patients on the use of exercise as an adjunct to more conventional therapies. The beneficial effects of exercise on systemic vasculature are well known, including improved endothelial function and increased perfusion. In this work, we evaluate the hypothesis that exercise slows tumor growth and improves the structure and function of tumor blood vessels, resulting in decreased hypoxia and increased effectiveness of cyclophosphamide chemotherapy. To investigate the effects of exercise on tumor microenvironment, we injected syngeneic 4T1 breast tumor cells into the mammary fat pad of immunocompetent BALB/c mice. The exercise intervention (voluntary wheel running) was designed to mimic four clinically relevant scenarios: (1) patients who are sedentary before and after diagnosis, (2) previously sedentary patients who begin exercising after diagnosis, (3) iv

5 previously active patients who stop exercising after diagnosis, and (4) previously active patients who continue to exercise after diagnosis. Animals in Groups 3 and 4 exercised prior to tumor cell transplant, whereas Groups 1 and 2 were sedentary during that time. Immediately after transplant, Groups 2 and 4 were running, and Groups 1 and 3 were sedentary. Tumor growth was monitored for 18 days, and then perfusion was mapped using MRI and tumors were removed for analysis. In a follow- up experiment, BALB/c mice were immediately implanted with tumor cells and then randomized to running or sedentary conditions with or without cyclophosphamide chemotherapy given one week after tumor cell transplant (three 100 mg/kg doses, every other day). Tumors were again allowed to progress for three weeks, and MRI was performed prior to tumor removal. Animals voluntarily ran 5 to 6 km per day prior to transplant and 4 to 5 km per day after transplant. Body weight was unaffected by exercise. Voluntary wheel running reduced tumor growth rate nearly twofold and significantly increased apoptosis. Additionally, running after tumor implantation caused significant increases in microvessel density (CD31) and vessel maturity (colocalization of CD31 with NG2 and desmin). Hypoxia (EF5) was significantly reduced in the exercising animals, and MRI showed that tumors were more uniformly perfused in the running groups. Furthermore, exercise significantly enhanced the effectiveness of cyclophosphamide in slowing tumor growth. Taken together, these results suggest that aerobic exercise slows breast tumor v

6 growth, improves tumor vessel structure and function, and augments the effectiveness of cyclophosphamide chemotherapy. These findings have important implications for the use of exercise in cancer treatment. Using a clinically relevant animal model, we provide the first conclusive evidence that exercise may do more than decrease symptoms and improve quality of life for cancer patients. Our data suggest that exercise may, in fact, be an effective antitumor intervention both alone and in combination with other cytotoxic therapies. vi

7 Dedication This dissertation is dedicated to: My parents. You paved the path, then walked with me hand- in- hand. Ari and Shauna. You have supported and guided me through the lowest of lows to the highest of highs. Rachel Brems. Not many people are lucky enough to have more than one family. You made me part of yours, and I am so grateful that you are forever part of mine. Anya and Kayla. You have filled my life with light and joy. I work with the hope that you will grow up to live in a world that no longer fears the word cancer. vii

8 Contents Abstract... iv List of Tables... xii List of Figures... xiii List of Abbreviations... xv Acknowledgments... xviii 1. Introduction Epidemiology of Breast Cancer Pathogenesis of Breast Tumors The Hallmarks of Cancer The Tumor Microenvironment Components of the Physical Microenvironment The Physiological Microenvironment: Hypoxia and Angiogenesis Antiangiogenesis and Vascular Normalization Introduction to Exercise Oncology Exercise and Cancer Symptom Control Exercise and Breast Cancer Survivorship: Epidemiological Evidence Biological Mechanisms of Exercise Effects on Breast Cancer: Clinical Evidence Exercise and Cancer Progression: Preclinical Studies Biological Mechanisms of Exercise s Effects on Breast Cancer: Preclinical Evidence Conclusions viii

9 2. Materials and Methods T1 Breast Cancer Model Exercise Intervention Magnetic Resonance (MR) Tumor Perfusion Imaging In Vivo and Ex Vivo Bioluminescence Imaging For Metastasis Detection Exercise Plus Chemotherapy Model Immunohistochemistry Ki67 and Cleaved Caspase CD31, Desmin, and NG oxo- 2 - deoxyguanosine EF Microscopy Image Analysis Optical Spectroscopy Measurement of Total Hemoglobin and Hemoglobin Saturation Angiogenesis- Related Gene Expression Statistics Effect of Voluntary Aerobic Exercise on Primary Breast Tumor Growth and Progression Introduction Results Voluntary Exercise Performance Effect of Aerobic Exercise on Body Weight ix

10 3.2.3 Effects of Exercise on Primary Tumor Growth Effects of Exercise on Breast Tumor Metastasis Discussion Effects of Voluntary Aerobic Exercise on the Tumor Microenvironment Introduction Results Effects of Exercise on Tumor Vascularity and Angiogenic Signaling Exercise- Mediated Changes in Tumor Vessel Maturity Effects of Exercise on Tumor Perfusion Effects of Exercise on Tumor Hypoxia Exercise- Induced Decreases in Oxidative Stress Effects of Exercise on the Physical Microenvironment: Innate Immunity Discussion Efficacy of Combining Aerobic Exercise with Cyclophosphamide Chemotherapy for Breast Cancer Treatment Introduction Results Effects of Chemotherapy on Voluntary Exercise Performance Effects of Aerobic Exercise and Cyclophosphamide on Body Weight Effects of Aerobic Exercise and Cyclophosphamide on Primary Tumor Growth Exercise- Mediated Changes in Tumor Blood Flow Discussion x

11 6. Discussion and Future Directions Confirmation and Optimization of the Exercise Model Exploring the Effects of Aerobic Exercise on Tumor Metastasis Investigating Mechanisms of Exercise- Induced Apoptosis Exploring the Role of the Immune System in Exercise Oncology Understanding Exercise- Mediated Changes in Angiogenesis and Vascular Maturity Investigating the Efficacy of Aerobic Exercise in Combination with Other Therapies Identifying Biomarkers for Exercise Efficacy in Breast Cancer Exploring the Role of Exercise in Tumor Dormancy Conclusions Appendix A: Experimental Protocols References Biography xi

12 List of Tables Table 1: Clinical Studies of the Association Between Exercise and Breast Cancer Outcomes Table 2: Biological Mechanisms of Exercise Effects in Cancer Patients from Clinical Studies Table 3: Description of Preclinical Exercise Oncology Studies and Outcomes Table 4: Description of Exercise Interventions in Preclinical Exercise Oncology Studies 35 Table 5: Summary of Immunohistochemistry Protocols Table 6: Summary of PCR Protocols xii

13 List of Figures Figure 1: Effects of hypoxia on the hallmarks of cancer Figure 2: Overview of experimental design to determine the effects of voluntary wheel running on primary breast tumor growth Figure 3: Experimental design to determine the effects of voluntary aerobic exercise combined with cyclophosphamide chemotherapy on primary breast tumor growth Figure 4: Mean voluntary wheel running distance before and after 4T1 tumor cell transplantation Figure 5: Body weights over time of BALB/c mice randomized to voluntary wheel running or sedentary control Figure 6: Tumor response to voluntary aerobic exercise Figure 7: Effect of exercise after tumor transplantation on proliferation of 4T1 breast tumor cells Figure 8: Apoptosis in primary 4T1 breast tumors in response to exercise after tumor transplantation Figure 9: Representative bioluminescence scans for 4T1 tumor metastasis Figure 10: Mean lung and liver weights from animals bearing 4T1 breast tumors Figure 11: Exercise- mediated increases in tumor microvessel density Figure 12: Effects of voluntary wheel running on expression of VEGF, VEGFR- 1, and VEGFR- 2 mrna in tumor tissue Figure 13: Effects of aerobic exercise on vascular maturity in primary 4T1 breast tumors Figure 14: Effects of exercise on mrna expression levels of components of the angiopoietin signaling axis in 4T1- luc tumor tissue Figure 15: Effects of exercise on tumor perfusion xiii

14 Figure 16: Exercise- mediated decreases in tumor hypoxia Figure 17: Exercise- induced changes in tumor oxidative stress Figure 18: Effects of voluntary wheel running on innate immune cell infiltration Figure 19: Mean daily voluntary wheel running distance in the exercise and exercise plus cyclophosphamide groups Figure 20: Body weights of BALB/c mice bearing 4T1- luc tumors randomized to no treatment, cyclophosphamide, exercise, or exercise plus cyclophosphamide Figure 21: Tumor response to voluntary aerobic exercise combined with cyclophosphamide chemotherapy Figure 22: Effects of cyclophosphamide and/or exercise after tumor transplantation on proliferation of 4T1 breast tumor cells Figure 23: Apoptosis in primary 4T1 breast tumors in response to cyclophosphamide and/or exercise after tumor transplantation Figure 24: Effects of exercise on tumor total hemoglobin xiv

15 List of Abbreviations 8- oxo- dg, 8- oxo- 2 - deoxyguanosine ADCperf, apparent diffusion coefficient AMPK, AMP- activated protein kinase Ang, angiopoietin bfgf, basic fibroblast growth factor BMI, body mass index CA IX, carbonic anhydrase IX CI, confidence interval DMBA, 7, 12- dimethylbenz[α]anthracene DMEM, Dulbecco s Modified Eagle Medium EC, endothelial cell ECM, extracellular matrix ELISA, enzyme- linked immunosorbent assay ER, estrogen receptor FDA, (U. S.) Food and Drug Administration FPLSD, Fisher s Protected Least Significant Difference HER2, human epidermal receptor 2 HIF- 1, hypoxia- inducible factor- 1 IL, interleukin xv

16 LAK, lymphokine- activated killer (cell) MMP, matrix metalloproteinases MR, magnetic resonance MVD, microvessel density NK, natural killer NMU, N- methyl- N- nitrosourea ODD, oxygen- dependent degradation PBS, phosphate- buffered saline PBST, phosphate- buffered saline with 0.3% Tween 20 PDGF- B, platelet- derived growth factor PDGFR- β, platelet- derived growth factor receptor- β PECAM, platelet- endothelial cell- adhesion molecule PFA, paraformaldehyde PVC, perivascular cell qpcr, quantitative real- time polymerase chain reaction ROI, region of interest SEM, standard error of the mean SOD, superoxide dismutase TNF, tumor necrosis factor TGF, tumor growth factor xvi

17 VEGF, vascular endothelial growth factor VEGFR, vascular endothelial growth factor receptor VHL, von Hippel- Lindau (complex) WMES, weighted mean effect size xvii

18 Acknowledgments If I have seen further it is by standing on the shoulders of giants. Isaac Newton, letter to Robert Hooke, 1676 An endeavor like this does not come to fruition without the support and guidance of many caring and generous people. First, I must thank my thesis advisor, Dr. Mark Dewhirst. I entered the Dewhirst lab as a medical student who loved science, and I emerged as a scientist. The transformation would not have been possible without him. He taught me the thinking and discipline necessary to pursue worthwhile research while giving me the confidence and freedom to follow my interests and intuition. For all of the help he has given me in the past and the support he will undoubtedly provide in the future, I will be forever grateful. I must also thank the members of my committee. Dr. Lee Jones introduced me to the field of exercise oncology and gave me the opportunity to combine my lifelong love of fitness with my passion for cancer research. His enthusiasm and boundless energy are incredibly inspiring. Lee helped to build my confidence as a scientist (while simultaneously humbling me as an athlete!), and he went above and beyond what I had the right to hope for in a mentor. Dr. Zeljko Vujaskovic helped me to write the first grant I ever got funded (hopefully the first of many), and has served as a wonderful example of how to combine a successful career in research with exemplary patient care. I owe a large debt of gratitude to Drs. Salvatore Pizzo and Christopher Kontos, past and present xviii

19 Directors of the Medical Scientist Training Program at Duke. Sal recruited me to the Duke MSTP, an opportunity that shaped my professional and personal lives in ways I am only beginning to understand. Throughout my time in the program, he has been my biggest cheerleader, and (in classic Philadelphia style) a voice of reason when I needed one most. Chris became the Director in my third year of the MSTP and has been a source of constant guidance and support ever since. Together, we have traveled to conferences, hiked a few mountains (both literally and figuratively), talked about training (medical, scientific, and athletic), and even managed to discuss a little vascular biology. Both Sal and Chris have embodied the terms physician- scientist and mentor. I hope someday to live up to their examples. To the members of the Dewhirst Lab, I can never thank you enough. Chelsea Landon assisted with some of the experiments in this dissertation, and was a source of constant support and encouragement. We shared desk space, dog stories, chocolate, and a lot of laughs. My fellow grad students Kelly Kennedy, Andrew Fontanella, and Keara Boss have provided ideas, feedback, support, and a lot of fun along the way. I must thank our wonderful post- docs Kate Ashcraft, Peter Scarbrough, Jennifer Ayers (an honorary Dewhirst Lab member), and Diane Fels, who have become friends as well as colleagues and mentors. Lastly, I owe a huge thank you to Ken Young, our lab manager, who simultaneously kept the research train on the tracks, cheered me on, and provided photographic documentation of the entire journey. xix

20 I must thank several people who have not directly contributed to this work but who deserve considerable credit for helping me to reach this milestone. My fellow students in the Duke MSTP, Department of Pathology, and School of Medicine have made this a time of personal as well as professional growth. I must especially thank Giselle Lopez, Jessica Amenta Hennessey, Dawn Kernagis, Lauren Jackson, Caroline Hadley, Amy Treece, Tracy Wester, Sara Jiang, Erin Wilfong, and Kim Cocce. Additionally, I owe a huge thank you to Dr. Robert Drucker for being a wonderful medical school advisor. Though his judgment in collegiate hockey teams is somewhat suspect (GO BIG RED!), his faith in me has never wavered. My undergraduate research advisor, Dr. John Hermanson, taught me a love of science and encouraged me to pursue it as a career. I hope that this is the first of many scientific works of which he can be proud. Last and most important, I must thank my family. My parents were the first in their respective families to get doctorate degrees. They taught me to prioritize education, and they have supported me as I have pursued every one of my dreams. My brother Ari, sister- in- law Shauna, and two amazing nieces Anya and Kayla are a constant source of love, support, encouragement, and laughter. Without them, none of this would be worthwhile. Rachel Brems, a member of our family by choice if not by blood, has been a source of inspiration, comfort, and love since before I can remember. Mere words cannot express my gratitude to each and every one of you. xx

21 1. Introduction The most recent estimates from the American Cancer Society suggest that nearly one in eight women will be diagnosed with invasive breast cancer in her lifetime [1]. Accordingly, improving both the quality of life of breast cancer patients and the efficacy of treatment is of the utmost importance. To date, surgical, pharmacological, and radiation- based approaches have been the mainstays of breast cancer therapy. However, over the past decade, exercise has gained legitimacy as a potentially beneficial intervention, and the field of exercise oncology has received increasing attention [2]. Specifically, recent epidemiological studies indicate that regular moderate- intensity exercise is associated with significant reductions in the risk of cancer- specific and all- cause mortality in patients diagnosed with early- stage breast cancer [3-11]. However, little is known about the mechanisms responsible for exercise- mediated inhibition of tumor progression. Clinical studies of exercise in breast cancer patients have revealed some pathways of interest, but variability in tumor stage, exercise interventions, and patients characteristics are fraught with confounding factors that have complicated efforts to identify biological and molecular mechanisms of exercise effects on tumor progression. Additionally, animal models of exercise in cancer have yielded widely varying results, and exploration of underlying mechanisms has been complicated by study heterogeneity and limitations of the models used. 1

22 The work in this dissertation focuses on developing an improved animal model of exercise to recapitulate the effects of exercise in women with breast cancer as accurately as possible. To do so, we studied tumors in their native microenvironment, which has profound effects on tumor behavior. We sought to understand whether exercise affects tumor progression in this model. Additionally, we explored biological mechanisms underlying this relationship. Importantly, use of a clinically relevant model of the microenvironment allowed us to explore the effects of exercise on components of the tumor s environment that greatly influence aggressiveness of the disease and efficacy of treatment. This work focuses mainly on the structure and function of blood vessels and associated changes in tumor perfusion and oxygenation, features that are known to be abnormal in breast tumors and to strongly influence tumor behavior and therapeutic outcomes. Understanding these effects on tumor progression and tumor physiology is a critical first step toward optimizing the safety and efficacy of aerobic exercise as an adjunct therapy for breast cancer. 1.1 Epidemiology of Breast Cancer Malignancy of the breast trails only skin cancer as the most common tumor type in women worldwide, and it is second to lung cancer as the leading cause of cancer death in American women [12]. According to the most recent estimates by the American Cancer Society, approximately 226,870 women will be diagnosed with invasive breast 2

23 cancer in 2012, and 39,510 women will die of the disease [1]. Additionally, over the next 15 years, the number of women with breast cancer is expected to grow by about one- third as the population ages [13]. Given the large number of women affected by breast cancer and the expectation of further increases, considerable effort has been concentrated on understanding its causes and finding new ways to treat it. 1.2 Pathogenesis of Breast Tumors The etiology of breast cancer is often divided into hereditary causes, which are associated with a family history or germ line mutation, and sporadic causes, which arise as a result of exposure to other risk factors. A collaborative analysis of 52,809 women with breast cancer reported that approximately 13% of female breast cancer patients have a family history of breast cancer in a first- degree relative [14]. The most common hereditary causes of breast cancer are mutations in BRCA1 and BRCA2, two autosomal- dominant genes responsible for approximately one- quarter of inherited breast cancers. Several other genes have been implicated in the pathogenesis of hereditary breast cancer, but much is left to be learned about familial risk. Though hereditary causes of breast cancer have received considerable attention, eight of nine women who develop breast cancer do not have a known history in a first- degree relative [14]. The most common risk factors for sporadic breast cancer are exposure to hormones, including ages at menarche and menopause; pregnancy and 3

24 breastfeeding; and exogenous estrogen intake. These tumors most commonly occur in postmenopausal women, and they often overexpress the estrogen receptor (ER). Estrogen can affect breast carcinogenesis in a variety of ways, including increased proliferation in premalignant and malignant lesions and generation of mutations and free radicals that damage DNA as a result of estrogen metabolites [15]. However, many breast tumors occur in women who were not exposed to high levels of estrogen or do not express ER, so tumorigenesis is clearly driven by more than estrogen. 1.3 The Hallmarks of Cancer The development of a tumor entails progression of normal cells through a series of steps that drive malignant transformation and eventually neoplastic growth [16]. In their seminal paper in 2000, Hanahan and Weinberg distilled the complexities of tumorigenesis into six hallmarks of cancer that are essential to malignant growth: self- sufficiency in growth signals, insensitivity to growth- inhibitory signals, evasion of programmed cell death, limitless replicative potential, sustained angiogenesis, and tissue invasion and metastasis [17]. Each of these hallmarks is an acquired capability, gained during tumor development and reflecting the evolving tumor s ability to evade antitumor mechanisms in the host. Individual tumor types and subtypes use different mechanisms and require different amounts of time to acquire these capabilities. These variables play important roles in determining tumor phenotype. However, the six 4

25 hallmarks are nearly ubiquitous during tumor formation and progression, and because these capabilities are critical to tumor growth, scientists have attempted to inhibit these pathways to prevent and treat cancer. Although the hallmarks of cancer focus on malignant progression of tumor cells, development of a tumor involves countless other cells and signaling molecules that coexist in a complicated milieu, the tumor microenvironment. Just as the development of children is highly dependent on their surroundings, tumor development and acquisition of the hallmark capabilities are greatly influenced by the microenvironment, and these interactions have profound effects on tumor physiology and efficacy of treatment. Of particular interest to this dissertation, many tumors develop imbalances between the supply of and demand for oxygen, resulting in hypoxic regions of a tumor that are particularly aggressive and resistant to treatment. The work in this dissertation explores how exercise affects the tumor microenvironment and focuses specifically on exercise- induced changes in tumor blood vessel structure and function, oxygenation, and the consequences of these changes on efficacy of treatment. 1.4 The Tumor Microenvironment In the native tumor microenvironment, extracellular matrix, blood vessels, immune cells, and other supporting structures surround tumor cells. Though early research focused on targeting and killing tumor cells, the role of the microenvironment 5

26 is now understood to be central to tumor development, progression, metastasis, and treatment. Microenvironmental research falls into two main categories: physical and physiological. The physical microenvironment consists of cells, cellular interactions, and the matrix that constitutes the structure of a tumor. In contrast, the physiological microenvironment mediates the ability of the tumor vasculature to exchange oxygen, nutrients, and waste products. We have described the development and interactions within the tumor microenvironment in more detail in previous work [18], but key points will be discussed in the following section. This dissertation focuses on ways in which changes in the physical and physiological microenvironments affect tumor progression in a murine model of breast cancer Components of the Physical Microenvironment Transformation and progression of cancer cells occur within the volume and environment of a host organ. Activation of the host tissue precedes or coincides with malignant cell transformation. These processes trigger proliferation of fibroblasts, recruitment of inflammatory mediators, and development of new blood vessels [19-21]. Cancer- associated fibroblasts from the host stroma respond by producing collagen, which contributes to the growing tumor parenchyma [21]. These fibroblasts are critical for transformation, proliferation, and invasion, three of the six hallmarks of cancer [22]. 6

27 Within the heterogeneous structure of the growing tumor, these cellular components are surrounded by the extracellular matrix (ECM), a three- dimensional supporting structure consisting of proteins produced by fibroblasts. The ECM contains proteins, secreted by tumor cells, that participate in key processes of tumor progression, such as cellular adhesion, motility, communication, and invasion [23]. It is constantly undergoing remodeling, mediated by matrix metalloproteinases (MMPs), adamalysin- related membrane proteases, endoglucosidases, and tissue serine proteases [24, 25]. To offset the effects of the degradative enzymes, activated fibroblasts produce collagen type I and fibronectin, supported by IL- 1β and nitric oxide synthase released from macrophages [26, 27]. The constant remodeling of the ECM yields a loosely woven, disorganized stroma. Immune cells are early and essential components of the tumor microenvironment that reside in and interact with the tumor stroma as well as affect tumor cells. Mast cells are the first to invade a growing tumor. Upon arriving, they degranulate and secrete key mediators, such as vascular endothelial growth factor (VEGF), serine proteases, and MMP- 9, which enhance the angiogenic and invasive capabilities of the growing tumor [28-30]. The arrival of mast cells is followed by macrophage infiltration and activation, resulting in secretion of molecules that are critical to the development of blood vessels necessary for sustained tumor growth [27]. 7

28 1.4.2 The Physiological Microenvironment: Hypoxia and Angiogenesis Within the tumor microenvironment, proliferation of tumor cells and recruitment and activity of supporting cells create a greater need for oxygen than in normal tissue. Whereas the normal oxygen tension in tissues is at least 20 mm Hg, hypoxia is defined as po2 10 mm Hg. Many solid tumors have regions of hypoxia caused by insufficient blood supply and increased demand due to proliferation [31, 32]. Specifically, 25% to 98% of invasive breast tumors from various clinical cohorts contain regions of hypoxia [33-39]. Our group showed that in a series of 209 patients with invasive breast cancer treated at Duke University Medical Center, 88% of tumors stained positively for carbonic anhydrase (CA) IX, an endogenous marker of tumor hypoxia [40]. In a variety of cancer types, identification of hypoxia is associated with poor outcomes, and it is thought to confer on the tumor more aggressive properties, including increased invasiveness, decreased host immune resistance to cancer, and metastasis [41-45]. Additionally, hypoxia has been reported to mediate resistance to chemotherapy in breast cancer [40, 46]. Lastly, poor blood supply and increased intratumoral interstitial fluid pressure, both associated with hypoxia, limit the delivery of systemic therapies such as chemotherapy and targeted cytotoxic agents [47]. For these reasons, many researchers have sought methods to prevent or combat hypoxia as a means of sensitizing cells to other antitumor treatments. 8

29 Fundamentally, tumor hypoxia is caused by an inability of the vasculature to meet the demand for oxygen. Virchow first reported that tumor blood vessels have an abnormal architecture [48]. The vessels themselves are highly heterogeneous, and they are arranged in networks that vary in terms of vessel number, length, spatial distribution, and flow, among other factors [49-52]. Though the networks constantly undergo remodeling and new vessels mature, disorganized vascular systems are characteristic of tumors, and hypoxia results. Our laboratory previously described eight key features of tumor hypoxia [32]. First, limited arteriolar supply results in low oxygenation in the small downstream vessels that branch off the supplying arteriole [53-55]. The heterogeneous orientation of vessels also affects oxygenation, as some regions are well perfused, whereas others are not adequately vascularized [56]. On a similar note, the third characteristic of tumor hypoxia is that the periphery of the tumor tends to be much more vascular than the center. Within the vessels themselves, red blood cell content varies widely, resulting in considerable variation in red blood cell flux, the fourth characteristic of tumor hypoxia [57]. The fifth is the resulting imbalance between oxygen supply and demand [58]. Patterns of erythrocyte flow account for the last three features of tumor hypoxia. Increased blood viscosity due to stiffening of hypoxic red blood cells results in slower 9

30 flow rates and abnormal distribution of flow at vascular bifurcations [59]. Furthermore, tumor vascular networks are marred by shunts that carry arteriolar blood to veins, diverting flow away from regions downstream of the arteriole [60]. Lastly, tumor hypoxia exhibits periodicity as a result of variations in erythrocyte flux in microvessels, a phenomenon known as cycling hypoxia [32, 61, 62]. Hypoxia is an environmental stressor that triggers an array of cellular responses. These responses enable cells to adapt to the harsh conditions while simultaneously triggering vascularization to increase tissue oxygenation. Goldman was the first to hypothesize that angiogenesis, the growth of new blood vessels from previously existing vessels, is important for tumor growth [63]. In 1971, Judah Folkman proposed that angiogenesis is necessary for tumor growth, and he promoted antiangiogenesis as a strategy to treat cancer [64]. Since Folkman s initial work in the 1970s, studies have demonstrated that angiogenesis is essential both for the growth of tumors beyond a very small size and for tumor metastasis. Beginning with Weidner and colleagues in 1991, many studies, several reviews, and one meta- analysis have all confirmed that high density of microvessels in the tumor and angiogenesis are associated with worse prognosis for patients with invasive breast cancer [65-67]. Adaptation to hypoxia is mediated in large part by a family of hypoxia- inducible transcription factors, the most common of which is hypoxia- inducible factor- 1 (HIF- 1). 10

31 First characterized by Semenza, HIF- 1 is often referred to as the master regulator of oxygen homeostasis [68]. The molecule is a heterodimer of α and β subunits. The β subunit is constitutively expressed, whereas the concentration of the α subunit is tightly regulated by tissue oxygenation. The predominant mechanism for regulation of HIF- 1α protein levels is degradation by a family of prolyl hydroxylases. Under normoxic conditions, these enzymes hydroxylate proline residues in the oxygen- dependent degradation (ODD) domain of HIF- 1α, targeting the protein for proteosomal degradation via the von Hippel- Lindau (VHL) complex. Hydroxylation requires molecular oxygen, so under hypoxic conditions the prolyl hydroxylases cease to function, and HIF- 1α is spared from degradation. As HIF- 1α levels rise, the protein complexes with HIF- 1β and binds to hypoxia response elements (HREs) in the promoter and enhancer regions of target genes. Accumulation of HIF- 1 increases the expression of genes involved in all six of the hallmarks of cancer, as shown in figure 1. Of these hallmarks, the effects of HIF- 1 on tumor angiogenesis has received the most attention, as the transcription factor increases expression of many proangiogenic genes, including directly activating transcription of VEGF [69]. In early breast tumors, HIF- 1 levels correlate with the expression of VEGF and amount of angiogenesis [70]. High expression levels of VEGF and other 11

32 proangiogenic molecules, mediated in large part by HIF- 1, trigger the formation of aberrant and poorly functional neovasculature in tumors [71-74]. Figure 1: Effects of hypoxia on the hallmarks of cancer. Hypoxia and accumulation of HIF- 1 greatly affect the six hallmarks of cancer (red) through a variety of signaling pathways (blue). Of the hallmarks, the interaction between hypoxia and sustained angiogenesis has received the most attention. (Figure reproduced with permission from Betof, A.S. and M.W. Dewhirst Establishing the tumor microenvironment. In: D.W. Siemann, (ed). Tumor Microenvironment. West Sussex, UK: John Wiley & Sons, Ltd.) 12

33 As with many biological phenomena, the occurrence or absence of angiogenesis is balanced like a seesaw, with proangiogenic factors, such as VEGF, vascular endothelial growth factor receptor (VEGFR), angiopoietin- 1 (Ang- 1), and Tie2, balanced against antiangiogenic molecules, such as VEGFR- 1, soluble VEGFR- 1, and Ang- 2, the antagonist of Ang- 1. The intricate balance between these promoters and inhibitors of angiogenesis is known as the angiogenic switch, and the net balance of these factors determines whether angiogenesis occurs in a particular time and location [75]. When the angiogenic switch is on, tumor vessels grow by sprouting from preexisting vessels, coopting nearby vessels, incorporating endothelial precursor cells from bone marrow, or intussusception, the splitting of previously existing vessels [76]. Signaling of VEGF through VEGFR2 causes weakening of the endothelial cell (EC) cytoskeleton and junctions between cells, resulting in dilation and leakiness of the developing vasculature [77]. This is one of many signaling molecules that result in abnormal vascular structure in new tumor vessels, and gene expression profiles differ greatly between tumor and normal endothelium [78]. Additionally, the perivascular cells (PVC) that surround and support blood vessels are highly aberrant, loosely connected, or even absent in tumor vessels [79-81]. Perivascular cells include smooth muscle cells and pericytes, the latter of which are recruited to the basement membrane around vessels to interact with EC and limit 13

34 permeability. Several signaling pathways are involved in the recruitment of pericytes, which enhance vessel stability: platelet- derived growth factor (PDGF- B) is released from EC and binds to platelet- derived growth factor receptor- β (PDGFR- β) on pericytes [73, 82, 83]. Ang- 1 contributes by enhancing the connections between pericytes and EC. Sphingosine- 1- phosphate- 1, endothelial differentiation sphingolipid G protein- coupled receptor- 1, and transforming growth factor- β are also involved in pericyte stabilization of vasculature [73, 82, 83]. Detachment of PVC from mature vessels is critical to angiogenesis, causing EC to migrate and form new vessels, but attachment to the newly formed vessels is key to stabilizing the resulting structures and enhancing their functionality [79]. This process is highly irregular in tumors. PVC can be observed in a haphazard pattern around tumor vessels [84]. Detachment of PVC is promoted by VEGF and Ang- 2, and then the detached cells become activated and release additional VEGF and basic fibroblast growth factor (bfgf), contributing to a self- perpetuating cycle of abnormal angiogenesis and hypoxia [85-88]. Given that hypoxia and aberrant vasculature increase tumor aggressiveness, promote metastasis, and interfere with delivery and efficacy of antitumor therapeutics, as discussed earlier, considerable effort has been put forth to inhibit abnormal angiogenesis and promote circulation and oxygenation to tumors. 14

35 1.5 Antiangiogenesis and Vascular Normalization In 1971, Folkman pioneered the notion of administering systemic antiangiogenic therapy to cut off a tumor s supply of oxygen and nutrients, resulting in tumor death or dormancy [64]. Though early preclinical studies using xenografts of sarcoma, glioblastoma, and colorectal carcinoma suggested that inhibition of VEGF could delay tumor growth and reduce metastasis [89, 90], subsequent clinical trials using this approach have yielded more modest results. For example, only a 6.7% response rate was observed in women with metastatic breast cancer treated with single- agent bevacizumab, a Food and Drug Administration (FDA) approved monoclonal antibody against human VEGF [91]. Additionally, phase III clinical trials have failed to demonstrate a clinically meaningful survival benefit using bevacizumab as monotherapy, raising considerable doubt about the ability of antiangiogenic therapy alone to significantly inhibit tumor growth [92]. In contrast to the failures of antiangiogenic monotherapy, combinations of bevacizumab with other systemic chemotherapeutic agents have improved clinical outcomes in several tumor types, including metastatic colorectal and non- small cell lung cancer [93-97]. However, combining antiangiogenic therapy with cytotoxic therapy has been less effective in breast cancer. In fact, in 2011, the FDA repealed its approval of bevacizumab to treat 15

36 metastatic breast cancer, having concluded that the risks did not outweigh the benefits to patients [98]. The observed improvement in the efficacy of systemic chemotherapy when combined with an antiangiogenic agent is paradoxical, because reduction of the tumor blood supply should decrease the delivery of cytotoxic chemotherapy and oxygen, thereby reducing therapeutic effectiveness. In an effort to explain this inconsistency, Jain put forth the vascular normalization hypothesis, which suggests that using limited doses of antiangiogenic agents may correct phenotypic and functional abnormalities in tumor vasculature [79, 99, 100]. Normalization entails pruning of immature, leaky, dilated, poorly functional vessels and remodeling of the remaining blood vessels so that they are less permeable, more uniform in diameter and direction, and more extensively and evenly surrounded by pericytes and basement membrane [100]. Tumor vessel normalization could have multiple beneficial effects on the tumor microenvironment. The resultant vasculature is better organized and more uniformly distributed, which could homogenize patterns of perfusion in the tumor [79]. Additionally, the improved coverage by perivascular cells could reduce permeability, thereby lowering tumor interstitial fluid pressure [79]. It is unlikely that antiangiogenic therapy will ever result in truly normal vessels, but the improvements in structure and function could drastically improve perfusion and decrease hypoxia and the associated 16

37 complications. Additionally, more functional and uniform vessels could prevent cancer cell intravasation and extravasation during the metastatic process and increase the delivery and efficacy of systemically administered chemotherapy [79]. Achieving normalization requires modulation of the pathological proangiogenic phenotype of tumors. Factors that promote angiogenesis, such as VEGF, bfgf, and Ang- 2, must be balanced by those that promote vascular maturity and stability, such as thrombospondin- 1 and Ang- 1 [73, 79]. Pharmacological inhibition of the VEGF signaling axis has been demonstrated to cause normalization in preclinical models and clinical studies, but other angiogenesis pathways have also been implicated, including the angiopoietin- Tie- 2 axis [73, 101, 102]. Notably, despite the evidence that combining bevacizumab with systemic chemotherapy does not provide significant benefit to patients [98], several groups have demonstrated that tumors, including breast cancer, are more sensitive to chemotherapy when tumor vessels are normalized [93, 94, 96, 97, ]. The main limitation to tumor vessel normalization as a widespread therapeutic technique is that effects are transient. The term normalization window has been used to describe the period following antiangiogenic therapy during which tumor vessels exhibit characteristics of normalization [100]. Vascular normalization generally begins within 2 days of the onset of the therapy, but the effects are reportedly lost within 3 days 17

38 to 6 weeks [79, 106, 107]. The normalization window may close as a result of vascular pruning caused by extended or high- dose antiangiogenic therapy. Alternatively, the tumor may develop resistance to the blockade of angiogenesis, resulting in a new wave of vascularization. However, when the window closes, the benefits of normalization are lost. For reasons that will be described in the next section, we hypothesize that aerobic exercise will cause tumor vessel normalization similar to that associated with antiangiogenic therapy. Furthermore, because this is a sustainable, nonpharmacological approach, exercise may result in a prolonged normalization window that sensitizes cancer cells to endogenous and exogenous mediators of tumor cell death. 1.6 Introduction to Exercise Oncology The health benefits of regular physical activity have been recognized for centuries, but it was not until the early 1900s that scientists first explored a relationship between exercise and malignancy [108]. Since that time, numerous clinical and preclinical studies have been conducted to investigate the relation between exercise and cancer prevention [109, 110]. The role of exercise in patients who already have cancer has received comparatively little attention. However, growing numbers of cancer patients and survivors have sought novel adjunct therapies to improve survivorship and quality of life. As such, exercise oncology has garnered increasing attention and 18

39 legitimacy as a field of research [111]. Based on this work, Australia and the United States have both formally recommended exercise for cancer patients during and after antitumor therapy [112, 113]. 1.7 Exercise and Cancer Symptom Control To date, approximately 80 studies have explored the effects of exercise in patients with a confirmed diagnosis of cancer. Most focused on maintaining fitness and improving quality of life by controlling symptoms. Cancer and the various modalities used to treat it are associated with a variety of psychosocial and pathophysiological signs and symptoms including loss of cardiorespiratory fitness, weight gain or loss, cardiac and pulmonary dysfunction, nausea, fatigue, depression, and anxiety, all of which can severely diminish patients quality of life [114]. Over the past three decades, researchers have attempted to address these concerns by designing trials to explore whether physical activity can offset chemotherapy- and radiotherapy- induced toxicity. The first studies, conducted in the late 1980s, investigated whether structured exercise training could mitigate therapy- associated fatigue and loss of cardiorespiratory fitness in women with early- stage breast cancer [ ]. The field of exercise oncology blossomed from this early work, and various groups have performed numerous studies to explore the safety and efficacy of exercise as an adjunct therapy before, during, and/or 19

40 after cancer therapy. In fact, to date more than half of the studies on exercise in cancer patients have investigated the role of physical activity during cytotoxic therapy [121]. The available literature on symptom control has been extensively reviewed elsewhere, and at least one meta- analysis of the data has been performed [114, ]. The findings are summarized here. Most studied breast cancer patients, while others focused on colorectal cancer, non- Hodgkin s lymphoma, or patients with mixed tumor types. Physical activity modalities included 2 to 24 weeks of aerobic exercise or aerobic exercise combined with resistance training, and all regimens entailed moderate to vigorous activity (50% to 75% of baseline capacity), three or more sessions per week, for 10 to 60 minutes per session. While the outcome measures differed, most studies reported on cardiorespiratory fitness, muscular strength, quality of life, pain, and depression. A meta- analysis by Speck et al. analyzed the effects of exercise on 60 physical and psychosocial outcomes in cancer patients [126]. The authors obtained data from 82 studies and calculated weighted mean effect sizes (WMES) and confidence intervals (CI) based on 66 high quality studies. They found significant improvements in patient- reported outcomes such as quality of life (WMES: 0.29; 95% CI: ), fatigue (WMES: ; 95% CI: ), and anxiety (WMES: ; 95% CI: ), as well as physiological outcomes like muscular strength (Upper Body- WMES: 0.99; 95% 20

41 CI: ; Lower Body- WMES: 0.90; 95% CI: ) and aerobic fitness (WMES: 0.32; 95% CI: ). Importantly, the authors observed a low incidence of adverse events. Taken together, these data suggest that exercise is well tolerated by and beneficial to cancer patients during and after cancer therapy. Key improvements in both psychosocial and physiological measures have been repeatedly observed in patients with a variety of cancer types who underwent different treatment regimens and modalities. 1.8 Exercise and Breast Cancer Survivorship: Epidemiological Evidence The success of exercise in controlling cancer symptoms has led to important questions about the effects of physical activity on tumor- related outcomes. Of the 20 clinical studies to date that investigated the effects of exercise on cancer- specific and/or all- cause mortality, 10 focused on breast cancer [3-11, 128]. These studies are summarized in table 1. Six (60%) studies reported a significant positive correlation between physical activity and improved outcomes. The reported risk reductions were 15% to 50% for cancer- specific mortality and 18% to 67% for all- cause mortality. 21

42 Table 1: Clinical Studies of the Association Between Exercise and Breast Cancer Outcomes. Study Author, Year Borugian et al., 2004 Holmes et al., 2005 Pierce et al., 2007 Holick et al., 2008 Irwin et al., 2008 Dal Maso et al., 2008 Sternfeld et al., 2009 Bertram et al., 2011 Chen et al., 2011 Irwin et al., 2011 N Cohort Breast Cancer- Specific Mortality All- Cause Mortality Risk Reduction Amount of Physical Activity Risk Reduction Amount of Physical Activity 603 Breast cancer patients after surgery before adjuvant treatment 2987 Stage I- III breast cancer; Nurses Health Study 1490 Stage I- IIIa breast cancer; Women s Healthy Eating and Living Study 4482 Invasive breast cancer free of recurrence >2 years after diagnosis 933 Breast cancer survivors; Health, Eating, Activity, and Lifestyle Study 1453 Invasive breast cancer survivors 1970 Stage I- IIIa breast cancer; Life After Cancer Epidemiology 2361 Stage I- III breast cancer survivors within 4 years, no current chemotherapy; WHEL study 4826 Stage I- III breast cancer, 6 months after diagnosis; Shanghai Breast Cancer Survival Study 4643 Invasive breast cancer survivors; Women s Health Initiative *multi- variate adjusted relative risk 1.0* Exercise was not associated with breast cancer mortality. 0.5* MET- hr/week MET- min/week 0.51* 21 MET- hr/week (p <0.05) 0.65* 9 MET- hr/week >1 time/week n/a n/a 0.56* MET- hr/week (p <0.05) n/a n/a 0.44* 21 MET- hr/week (p <0.05) 0.33* 9 MET- hr/week (p <0.05) 0.85* 2 hr/week 0.82* 2 hr/week 0.69* 3- <6 hr/week moderate activity 0.66* 3- <6 hr/week moderate activity n/a n/a 0.47* MET- hr/week 0.60* 2 times/ week (p <0.05) 0.60* 9 MET- hr/week (p <0.05) 0.70* 2 times/ week (p <0.05) 0.58* 9 MET- hr/week (p <0.05) The earliest evidence of a relationship between physical activity and cancer mortality came from Holmes et al. [11]. This study analyzed the association between self- reported physical activity and cancer- specific outcomes in the Nurses Health Study, consisting of a cohort of 2987 female nurses who had been diagnosed with stage I- III 22

43 breast cancer. They found that women who reported 9 or more metabolic equivalent time (MET)- hours per week of exercise (equivalent to walking briskly for 1 hour, 5 days per week), had a 6% reduction in unadjusted absolute mortality risk at 10 years compared with women who reported fewer than 3 MET- hours per week (equivalent to walking 2 to 2.9 mph for 1 hour), and exercise had particularly beneficial effects on outcome in women with hormone- responsive tumors [11]. Several subsequent studies conducted in other populations have confirmed that physical activity is associated with a risk reduction for cancer- specific and all- cause mortality for breast cancer patients [3-10]. A study by Borugian et al. involving 603 breast cancer patients enrolled after surgery but before the start of adjuvant therapy found that self- reported exercise was not associated with breast cancer mortality 10 years after enrollment [128]. The reasons for this discrepancy are unclear, but significant reductions in exercise during primary adjuvant therapy could play a role [129]. The groundbreaking studies on breast cancer paved the way for more recent investigations into the effects of exercise on prognosis in patients with other malignancies. Though other malignancies are not the focus of this dissertation, studies have reported significant inverse relationships between exercise behavior and mortality in colorectal, prostate, lung, ovarian, and brain cancers [ ]. Clearly, the epidemiological evidence suggests that physical activity can affect cancer- specific and 23

44 all- cause mortality in patients with breast cancer and other malignancies. However, all of the studies to date have analyzed physical activity based on patients retrospective self- reports, which may not be reliable. This is a key weakness of these data, and controlled trials of supervised exercise with long- term follow- up are needed to confirm the results. One such trial in breast cancer survivors is currently under way [140]. 1.9 Biological Mechanisms of Exercise Effects on Breast Cancer: Clinical Evidence Equally important to understanding the effects of physical activity on cancer outcomes is elucidating the mechanisms by which exercise exerts these effects. Many pathways have been suggested as possible biological mechanisms, including changes in metabolism, sex hormone levels, immune system function, systemic inflammation, and oxidative stress [141]. To date, 16 clinical studies have investigated associations between exercise and changes in these pathways in breast cancer patients [ ]. An overview of these studies is provided in table 2. 24

45 Table 2: Biological Mechanisms of Exercise Effects in Cancer Patients from Clinical Studies Study Author, Year Fairey et al., 2003 Irwin et al., 2005 Fairey et al., 2005, Brain Behavior and Immunity Fairey et al., 2005, Journal of Applied Physiology Schmitz et al, 2005 Irwin et al., 2006 Irwin et al., 2007 Payne et al., 2008 Ligibel et al., 2008 Pierce et al., 2009 N Cohort Biological Mechanisms 53 Postmenopausal breast cancer patients randomized to a supervised 15- week aerobic training intervention or sedentary control. 710 Stage 0- IIIA breast cancer survivors; Health, Eating, Activity, and Lifestyle Study 53 Randomized controlled trial of exercise training in postmenopausal survivors of stage I- IIIB breast cancer who had completed surgery, RT, and/or chemotherapy; Rehabilitation Exercise for Health after Breast Cancer Trial 53 Randomized controlled trial of exercise training in postmenopausal survivors of stage I- IIIB breast cancer who had completed surgery, RT, and/or chemotherapy; Rehabilitation Exercise for Health after Breast Cancer Trial 85 Breast cancer patients 4-36 months after adjuvant therapy 474 Stage 0- IIIA breast cancer survivors; Health, Eating, Activity, and Lifestyle Study 522 Stage 0- IIIA breast cancer survivors; Health, Eating, Activity, and Lifestyle Study 20 Postmenopausal breast cancer patients receiving hormonal therapy 101 Sedentary, overweight, early- stage breast cancer patients in home- based aerobic and resistance training program or sedentary control for 16 weeks 741 Stage 0- IIIA breast cancer survivors; Health, Eating, Activity, and Lifestyle Study - decreases in IGF- 1 and IGF- 1:IGFBP- 3 molar ratio - increase in IGFBP- 3 - no significant effect on fasting insulin, glucose, or insulin resistance. - significant decreases in C- peptide and leptin - increases in IGF- 1 and IGFBP- 3 - trend toward decreased CRP levels - increases in NK cell cytotoxic activity and unstimulated mononuclear cell function - no significant effect on mononuclear cell production of pro- inflammatory (IL- 1, TNF- α, IL- 6) or antiinflammatory cytokines (IL- 4, IL- 10, TGF- 1) - significant decreases in IGF- II and IGFBP- 3 levels - No significant effect on glucose, insulin, IGF- 1, IGFBP- 1, or IGFBP- 2 - significant decrease in mammographic dense area and percent density in postmenopausal women with BMI 30 kg/m 2 - significant increase in percent density in premenopausal women with BMI < 30 kg/m 2 - significant decrease in mammographic dense area in postmenopausal women with BMI 30 kg/m 2 - increase in percent dense area in women with BMI <25 kg/m 2 - no significant association with cortisol or IL- 6 levels - 28% decrease in fasting insulin - improvement in insulin sensitivity - significant reductions in CRP - decreases in serum amyloid A concentration 25

46 Evans et al., 2009 Irwin et al., 2009 George et al., Breast cancer patients within 6 months of completion of all major cancer therapy and healthy controls matched for age and physical activity level; Get REAL & HEEL Breast Cancer Program 75 Sedentary postmenopausal breast cancer survivors diagnosed 1-10 years earlier with stage 0- IIIA breast cancer who completed adjuvant treatment at least 6 months before enrollment 746 Stage 0- IIIA breast cancer survivors; Health, Eating, Activity, and Lifestyle Study Tosti et al., Stage I- III invasive breast cancer within 6 months of completion of all major cancer treatments and no evidence of metastatic disease, matched to healthy controls Janelsins et al., 2011 Zheng et al., Sedentary patients with stage 0- IIIb breast cancer who had completed treatment 1-30 months prior to enrollment 75 Inactive postmenopausal women diagnosed with stage 0- IIIA breast cancer who had completed adjuvant treatment at least 6 months prior to enrollment - breast cancer patients had significantly lower post- exercise blood lactate levels following high- intensity exercise (70% of VO2max) - no significant differences at low and moderate intensity - decrease in insulin, IGF- 1, and IGFBP- 3 levels - physical activity offset the negative effects of poor diet quality on CRP levels - patients with breast cancer had significantly lower blood lactate response to exercise across a variety of intensities - Trend toward more elevated glucose responses in women with cancer before and after exercise - insulin levels did not rise in exercising patients compared to increase seen in sedentary controls - IGF- 1 increased more in exercising patients than controls - IGFBP- 3 increased in exercising group - significant reduction in methylation of L3MBTL1 tumor suppressor gene Abbreviations: BMI, body mass index; IGF, insulin- like growth factor; IGFBP, insulin- like growth factor binding protein; CRP, C- reactive protein; RT, radiotherapy; NK, natural killer; TNF, tumor necrosis factor. The most commonly studied physiological effects of exercise in cancer patients are changes in metabolism, particularly with respect to the insulin- glucose axis. Eight clinical studies have examined metabolic changes as a result of exercise in breast cancer patients [143, 145, 148, 150, 151, 153, 154, 156]. The first trial, published in 2003, involved 53 postmenopausal early- stage breast cancer patients randomized to either a supervised 15- week aerobic training protocol or sedentary control [143]. Exercise was associated with significant improvements in insulin- like growth factor- 1 (IGF- 1) concentrations (- 26

47 7.4 ng/ml), insulin- like growth factor binding protein- 3 (IGFBP- 3) concentrations ( ng/ml), and IGF- 1:IGFBP- 3 molar ratio ( ). In this cohort, no significant differences were seen between the exercise and sedentary groups with respect to fasting glucose, insulin, or insulin resistance. Similar results were observed in larger studies of both aerobic and resistance exercise in breast cancer patients [145, 154, 156]. Ligibel et al. randomly assigned 101 overweight, sedentary early- stage breast cancer patients to a 16- week home aerobic and resistance exercise program or sedentary control [148]. Interestingly, in this population, patients experienced a borderline significant decrease in fasting insulin (28%, p =0.07), a result that was duplicated by the randomized Yale Exercise and Survivorship Study [151]. Two studies in breast cancer have also evaluated energy utilization during exercise and reported differing results with respect to blood lactate response to varying intensities of physical activity [150, 153]. In addition to metabolism, immune function and systemic inflammation have received considerable attention as potential mediators of exercise effects on tumor progression. Fairey et al. observed increases in natural killer (NK) cell and mononuclear cell function from early- stage breast cancer patients who underwent 15 weeks of aerobic training [144]. The authors did not observe significant differences in either proinflammatory (e.g., IL- 1, TNF- α, IL- 6) or antiinflammatory cytokines (e.g., IL- 4, IL- 10, transforming growth factor- 1). A 2008 study by Payne et al. also reported no significant 27

48 effects of exercise on IL- 6 and cortisol levels in 20 postmenopausal breast cancer patients receiving hormonal therapy [157]. One common measure of systemic inflammation is C- reactive protein (CRP), an acute- phase biomarker that has been widely shown to rise in the blood in accordance with systemic inflammation. Elevated CRP levels are associated with a worse prognosis for breast cancer patients [158]. Three studies in breast cancer have shown that exercise is associated with decreases in CRP levels [142, 149, 152, 159]. A recent report suggests that epigenetic changes in important tumor- related genes may result from physical activity. Tissue from 75 postmenopausal women with stage 0- IIIA breast cancer was analyzed, and the authors found an association between physical activity and decreased methylation of the L3MBTL1 tumor suppressor gene [155]. Methylation is often associated with decreased expression of the associated gene, so exercise- induced changes in methylation status could have important implications for tumor progression and treatment. Two mechanisms unique to breast cancer have been evaluated. Hormonal status, particularly with respect to ER expression, has been widely discussed as an important determinant of the effectiveness of physical activity in preventing breast cancer; this work has been reviewed extensively elsewhere [160]. Additionally, mammographic density is an independent risk factor for breast malignancy [161]. Two reports based on the Health, Eating, Activity, and Lifestyle study suggest that exercise may affect 28

49 mammographic density. In the first study of 474 women with stage 0- IIIA breast cancer, the authors report an inverse relationship between physical activity and mammographic dense area and percent density in postmenopausal women with a body mass index (BMI) 30 kg/m 2, but in women with BMI < 30 kg/m 2, exercise was associated with increased percent density [147]. In a subsequent report on 522 women, exercise was again associated with increased density in obese women, but mammographic density was not significantly increased in women with BMI < 25 kg/m 2 [146]. The results of these studies hint at some biological mechanisms that could mediate the effects of exercise on cancer progression. However, to date, no clinical studies have evaluated the effects of exercise on tumor perfusion and oxygenation, key components of the microenvironment with a large impact on treatment and outcomes. Related to hypoxia and oxygenation, oxidative stress has also been implicated in the effects of exercise on cancer progression. Though the effects of exercise on oxidative stress have not been investigated in breast cancer, two studies of colorectal and lung cancer suggest that physical activity may affect it. Allgayer et al. reported that short- term moderate- intensity exercise is associated with significantly decreased urinary excretion of 8- oxo- dg, a marker of oxidative damage to DNA, in colorectal cancer patients after the completion of primary therapy [162]. In contrast, Jones et al. recently found that physical activity is associated with increased levels of F2- isoprostanes, 29

50 eicosanoid markers of oxidative stress, in 16 patients with stage I- IIIB non- small cell lung cancer [163]. Given that perfusion and oxygenation play a key role in tumor progression as described in the first section, these components of the tumor microenvironment warrant investigation. It is well accepted that aerobic exercise leads to vascular adaptations [ ]. Improved endothelial function has been reported following aerobic training in several clinical populations with abnormal blood vessels, including patients with coronary artery disease, diabetes, hypercholesterolemia, and congestive heart failure [164, 168]. However, to date, the effects of exercise on aberrant tumor vasculature have not been investigated in clinical studies. The work in this dissertation explores this pathway in preclinical models of exercise in tumor- bearing animals Exercise and Cancer Progression: Preclinical Studies Preclinical investigations can provide key information on exercise- mediated changes in tumor progression. To date, 14 studies have been conducted in animal models to determine whether exercise slows tumor progression, as suggested by the epidemiological data, and which intratumoral and systemic pathways may be involved [ ]. Tables 3 and 4 describe the results and exercise methods of these studies, respectively. In summary, rodent models were used for all studies (n = 8, mice; n = 6, rats). Five (36%) studies were performed in breast cancer; the others involved sarcomas 30

51 (n = 3), hepatomas (n = 1), prostate adenocarcinoma (n = 1), pancreatic adenocarcinoma (n = 1), transformed salivary gland cells (n = 1), and transformed fibroblasts (n = 1). In 11 studies, tumor cells were implanted into host animals (n = 8 allografts; n = 3 xenografts), whereas carcinogens were administered to induce tumor growth in three studies. Only one of the cell injection studies investigated orthotopic implantation to closely mimic the natural tumor microenvironment. The exercise regimens in these studies ranged from 2 to 20 weeks and involved forced treadmill running (n = 6), voluntary wheel running (n = 6), and forced swimming (n = 4) 5 to 7 days per week. Forced exercise sessions ranged from 10 minutes to 12 hours per session, whereas voluntary wheel running regimens provided free access to the running wheel at all times throughout the study. Table 3: Description of Preclinical Exercise Oncology Studies and Outcomes. Tumor Type Study Author, Year Rodent Model (Strain) Method of Tumor Initiation Effect of Exercise on Primary Tumor Growth (% of control) Description of Biological Mechanisms of Primary Tumor Exercise Effects Growth Intratumoral Systemic Breast cancer Cohen et al. Rat (F344) Tail vein 1991 injection of 37.5 mg/kg N- methyl- N- nitrosourea Inhibition (not evaluated) 31 Exercise Not evaluated significantly delayed tumor appearance and increased tumor latency. Tumor incidence was significantly lower in exercised animals than in controls. No significant difference in Exercise did not significantly affect prolactin levels.

52 tumor burden or mean tumor volume. Breast cancer Breast cancer Breast cancer Breast cancer Sarcoma Hoffman- Mouse Goetz et al., (BALB/c) 1994 Jones et al., 2005 Sáez et al., 2007 Jones et al., 2010 Hoffman et al., 1962 Mouse (Athymic nu/nu) Rat (Sprague- Dawley) Lateral tail vein injection of 1x10 4 MMT 66 cells Not evaluated Evaluated pulmonary metastases from a tail vein injection of tumor cells. Exercise beginning after tumor injection did not significantly affect multiplicity of lung metastases. Not evaluated Subcutaneous injection of 5x10 6 MDA- MB- 231 cells Gastric intubation of DMBA. 5 mg wkly for 4 wk. No change (Not evaluated) Augmentation (+200%) Mouse Orthotopic No change (Athymic) (dorsal (+21%) mammary fat pad) injection of 1x10 6 MDA- MB- 231 cells Rat (Wistar) Subcutaneous Inhibition injection of 2 ml (- 97%) of Walker 256 cell suspension No significant Not evaluated difference in tumor growth delay Exercise group Not evaluated had significantly higher tumor growth rate. No significant differences in survival time or tumor multiplicity. No significant difference in survival time based on tumor growth to 1500 mm 3. Tumor weight significantly lower in the exercise group. Mice that began running on a voluntary wheel only after tumor injection had significantly higher LAK cell activity than those that ran before and after tumor injection. Not evaluated Not evaluated Exercised group Not evaluated had significantly more perfused vessels. HIF- 1 protein levels significantly higher in viable tumor in exercised group. No significant differences in the levels of CD31, VEGF, ATP, PGC- 1α, or AMPK. Not evaluated Not evaluated 32

53 Sarcoma Sarcoma Hepatoma Pancreatic cancer Uhlenbruck Mouse and Order, (BALB/c) 1991 Foley et al., 2004 Baracos, 1989 Roebuck et al., 1990 (concentration not specified) Subcutaneous injection of 2.5x10 4 L- 1 cells Inconclusive (- 44% to +86%) Rat (F344) Subcutaneous No change injection of 1x10 7 (- 31%) C10 cells Rat (Sprague- Dawley) Rat (F344, Lewis) Subcutaneous injection of 20 µl Morris hepatoma 777 finely chopped tumors Inhibition (- 25%) F344 rats: 3 Inconclusive doses of 30 (- 1 to - 11%) mg/kg azaserine 4-5 d apart. Lewis rats: 1 dose of 30mg/kg azaserine The group that Not evaluated ran 200 m daily had significantly lower tumor weights, but the other running distances did not significantly affect tumor weight. No significant Not evaluated difference in tumor weight as a result of exercise. Tumor weight Not evaluated was significantly lower in the exercise groups. No significant differences in tumor weight between groups exercised for different durations. Exercised F344 rats had significantly fewer foci and smaller volume percentage of foci. No difference in focal diameter. No significant differences in Lewis rats. Not evaluated Significant increase in insulin- stimulated glucose transport. No significant differences in blood glucose levels or lipid peroxidation in skeletal muscle based on exercise treatment. Not evaluated No significant Not evaluated difference in the amount of DNA synthesis based on treatment condition. 33

54 Transformed Japel et al., salivary 1992 gland cells Mouse (NMRI) Transformed MacNeil C3H/He fibroblasts and Hoffman- Goetz, 1993 Neoplastic lymphoid cells Prostate cancer Zielinski et al., 2004 Zheng et al., 2011 BALB/c Mouse (SCID) Subcutaneous injection of 1.5x10 6 S- 180 cells Not evaluated Not evaluated Not evaluated Moderate- intensity physical exercise begun after injection of tumor cells did not significantly change macrophage phagocytosis. Lateral tail vein injection of 3x10 5 CIRAS1 cells Not evaluated Evaluated pulmonary metastases from a tail vein injection of tumor cells. Exercise beginning after tumor injection did not alter lung tumor density relative to sedentary controls. Not evaluated Subcutaneous Inconclusive injection of 2x10 7 (+5%) EL- 4 cells Subcutaneous Inhibition injection of 2.5x10 6 LNCaP cells in media + matrigel. Tumors grew for 4-6 wk; then animals were surgically castrated to mimic androgen deprivation Exercise significantly delayed tumor appearance. No significant difference in peak tumor volume. Non- syngeneic tumors rejected by immune system of exercised mice. Voluntary wheel running moderately inhibited androgen- independent tumor growth. Not evaluated No significant Not evaluated difference in the fluid content of tumors. Significantly less vessel density in exercised animals on days 6, 8, 10, and 14 than controls. Not evaluated Not evaluated 34

55 Table 4: Description of Exercise Interventions in Preclinical Exercise Oncology Studies Tumor Type Study Author, Year Breast cancer Cohen et al., 1991 Breast cancer Hoffman- Goetz et al., 1994 Breast cancer Jones et al., 2005 Breast cancer Sáez et al., 2007 Breast cancer Jones et al., 2010 Sarcoma Sarcoma Sarcoma Hoffman et al., 1962 Uhlenbruck and Order, 1991 Foley et al., 2004 Exercise Modality Voluntary wheel running Treadmill running or voluntary wheel running Treadmill running Forced swimming Voluntary wheel running Continual running on a 20 foot runway + swimming + run on revolving drum Treadmill running Voluntary wheel running Daily Exercise Time Treadmill: 30 min/day min Frequency of Exercise Intensity Duration of Exercise of Exercise 35 Time From Tumor Initiation to Exercise Onset Control Characteristics Daily 20 wks 2 days Tumor- bearing sedentary animals Daily 18 m/min, 0% slope 5 days/wk 70-75% VO2max (10 m/min for 10 min up to 18 m/min for 45 min, 0% slope) 8 wks prior to injection and/or 3 wks after injection 30 min/day 5 days/wk days Runway: duration not specified, Swimming: 20 min- 4 hr, Revolving drum: 12 hr Varied. Constant speed (0.3 m/sec) for 200 m, 400 m, or 800 m Daily 36 hours Tumor- bearing sedentary animals 8 wks 14 days Tumor- bearing sedentary animals exposed to unmoving treadmill days Exercise began 1 days after appearance of first tumor Tumor- bearing sedentary animals were immersed several times in water but did not swim 2 days Tumor- bearing sedentary animals Daily 21 days Immediate Tumor- bearing sedentary animals Daily 0.3 m/sec 2 wks Not specified Not specified Daily 17 days 3 days Tumor- bearing sedentary animals

56 Hepatoma Pancreatic cancer Transformed salivary gland cells Transformed fibroblasts Neoplastic lymphoid cells Prostate cancer Baracos, 1989 Roebuck et al., 1990 Japel et al., 1992 MacNeil and Hoffman- Goetz, 1993 Zielinski et al., 2004 Forced swimming Voluntary wheel running Treadmill running Voluntary wheel running Treadmill running Zheng et al., Voluntary 2011 wheel running Low: 5 min/day, increased by 5 min/day Medium: 10 min/day, increased by 10 min/day High: 15 min/day, increased by 15 min/day 5 days/wk 3 wk 3 days Tumor- bearing sedentary animals Daily 18 wk 4-7 days after administration of carcinogen 30 min/day Daily m/sec 3 hours or until volitional fatigue Daily Daily Gradually increasing speed, m/min, 5% slope 3 wk prior to injection and/or 3 wk after injection 9 wk prior to injection and/or 3 wk after injection Not specified Not specified 5-14 days Injected immediately after first exercise session Daily 42 days 4-6 wks after tumor cell injection, mice were surgically castrated. Exercise began after castration. Tumor- bearing sedentary animals Not specified Tumor- bearing sedentary animals Tumor- bearing sedentary animals housed near treadmill during time of exercise and exposed to noise of treadmill Tumor- bearing sedentary animals 36

57 Eleven of the 14 preclinical studies (79%) reported primary tumor growth as an end point. Of the four preclinical breast tumor studies that evaluated primary tumor growth, one showed inhibition [169], two reported no change [171, 172], and one reported augmentation of tumor growth [173]. Cohen et al. injected the carcinogen N- methyl- N- nitrosourea (NMU) into the tail vein of F344 rats to induce breast adenocarcinoma, and two days later animals were randomized to 20 weeks of voluntary wheel running or sedentary control [169]. In this model, exercise significantly delayed breast tumor appearance and increased tumor latency. Two previous studies by Jones et al. reported no effect of treadmill running or voluntary wheel running on the growth rate of primary human tumor xenografts in athymic mice [171, 172]. In contrast, one study reported augmentation of breast tumor growth as a result of exercise. In this experiment, Sprague- Dawley rats were administered 7,12- dimethylbenz[α]anthracene (DMBA) to induce breast adenocarcinoma, and after tumors appeared, they were randomized to forced swimming 30 minutes per day for 38 to 65 days or sedentary control in which tumor- bearing animals were immersed several times but did not swim [173]. In this study, tumor growth rate was double the rate in controls, but there were no significant differences in survival time. Given the limited number of preclinical exercise studies on breast cancer, it is useful to evaluate the effects of exercise on other tumor types in similar animal models in order to gain insights on mechanisms and outcomes that may translate to breast 37

58 malignancy. Seven of the nine preclinical studies conducted using other tumor types reported on primary tumor growth. Three of these studies, all in different tumor models, found that exercise inhibited primary tumor growth [169, 174, 177, 182]. In the first study, Hoffman et al. injected Walker 256 sarcoma cells subcutaneously into Wistar rats that either were assigned to running and swimming daily for 21 days or were sedentary controls; they reported that tumor weight was significantly lower in the exercise group (percent of control: - 97%) [174]. Baracos et al. followed up on this experiment by injecting finely chopped Morris hepatoma 777 tumors subcutaneously into Sprague- Dawley rats that were assigned to forced swimming or were sedentary controls for three weeks [177]. In this experiment, tumor weight was significantly lower in the exercise group (percent of control: - 25%). Lastly, in a recent publication, Zheng et al. reported that voluntary wheel running inhibited tumor growth in SCID mice injected subcutaneously with human LNCaP prostate cancer cells then subjected to surgical castration to mimic androgen deprivation [182]. In contrast, three studies reported inhibition as a result of exercise in some groups and no effect in others [175, 178, 181]. Roebuck et al. examined the effects of up to 18 weeks of voluntary wheel running in a pancreatic adenocarcinoma model induced by injection of azaserine [178]. They found that F344 rats that exercised had significantly fewer tumors and smaller volume percentage, but Lewis rats subjected to the same treatment did not exhibit significant differences between groups. In a different model, 38

59 Uhlenbruck and Order injected L- 1 sarcoma cells subcutaneously into BALB/c mice that ran 200, 400, or 800 m daily on a treadmill [175]. Tumor weight was significantly lower in the group that ran 200 m daily than in sedentary animals, but there were no significant differences in tumor weight for the other distances. Additionally, Zielinski et al. injected EL- 4 neoplastic lymphoid cells subcutaneously in BALB/c mice and found that treadmill running delayed tumor appearance, but there was no significant difference between groups with respect to maximum tumor volume [178, 181]. Similar to the breast tumor studies by Jones et al., tumor growth was comparable between exercise and control groups in one study involving 17 days of voluntary wheel running following subcutaneous injection of C10 sarcoma cells in F344 rats [176]. Taken together, these data yield interesting yet conflicting results on the effects of exercise on primary tumor growth. Inhibition of tumor growth was observed in a variety of tumor types, including breast cancer, prostate cancer, lymphosarcoma, and hepatoma. With respect to breast cancer, only one of four studies showed growth inhibition, though the epidemiological evidence that exercise delays breast tumor progression is strong. Reasons for the variability in outcomes are not yet well understood, but differences in study methods, tumor cell line or carcinogen, location of tumor growth in cases of transplanted tumor cells, and volume and intensity of exercise could all play a role. Importantly, the one study that reported exercise- related augmentation of tumor growth used forced swim training, which has been shown to 39

60 induce a stress response that promotes tumor progression [183]. Accordingly, models that utilize voluntary exercise may yield more reliable results. Additionally, most preclinical exercise studies have utilized flank tumor models and/or tumor xenografts in immunodeficient hosts; neither of these models accurately recapitulates the physiological tumor microenvironment. Notably, one study evaluated the effects of exercise combined with chemotherapy. The authors reported that treadmill running up to 45 minutes daily for 8 weeks did not significantly affect the efficacy of doxorubicin chemotherapy in nude mice bearing flank tumor xenografts of human breast cancer cells. To date, this is the only preclinical study in any tumor type that evaluated the combination of exercise and other antitumor therapies. As discussed previously, the efficacy of cytotoxic interventions, such as chemotherapy and radiotherapy, depend heavily on the tumor microenvironment and oxygenation status. As a result, it is unclear whether using a more appropriate model of the microenvironment may affect the efficacy of combining exercise with chemotherapy. As most cancer deaths result from metastasis, there is considerable interest in whether exercise mediates tumor progression and spread to distant sites. To date, two studies have investigated the effects of exercise on the incidence and density or multiplicity of pulmonary metastases resulting from intravenous injection of tumor cells, and both reported null results [170, 180]. In the first study, MacNeil and Hoffman- 40

61 Goetz injected CIRAS1 transformed fibroblasts into the lateral tail vein of C3H/He mice and then randomized animals to voluntary wheel running or sedentary controls for 3 weeks [180]. Hoffman- Goetz et al. followed up on that work by injecting MMT 66 breast cancer cells intravenously into BALB/c mice and then randomizing animals to three weeks of treadmill running, voluntary wheel running, or sedentary control [180]. Exercise after tumor cell injection did not affect the development of lung metastases in either of these experiments. However, both of these experiments involved intravenous injection of cancer cells into otherwise healthy recipients, a model that fails to evaluate the ability of tumor cells to break through the tissue s basement membrane and invade the capillary. The effects of exercise on these essential components of the metastatic cascade warrant further attention, and transgenic or spontaneously metastasizing implanted tumor models may be more appropriate to elucidate whether physical activity may affect metastasis Biological Mechanisms of Exercise s Effects on Breast Cancer: Preclinical Evidence As in the clinical experiments, researchers have investigated exercise- induced differences in intratumoral and systemic factors that might affect tumor progression. This dissertation builds upon these early groundbreaking studies. Preclinical experiments are particularly well suited to this purpose because variables such as diet, training conditions, and measurement frequency and technique can all be standardized. Four studies (29%), including two in breast cancer, sought to elucidate changes in 41

62 systemic factors resulting from exercise in tumor- bearing host animals. Cohen et al. reported no significant differences in circulating prolactin levels as a result of 20 weeks of voluntary wheel running in rats with breast adenocarcinoma [169]. With respect to the immune system, Hoffman- Goetz et al. reported that lymphokine- activated killer (LAK) cell activity was significantly higher as a result of voluntary wheel running in mice injected intravenously with MMT 66 breast cancer cells [184]. In contrast, Japel et al. found that 3 weeks of treadmill running did not significantly affect macrophage phagocytosis in mice implanted with S- 180 salivary gland cells [179]. Lastly, one study investigated changes in systemic metabolism in animals. Foley et al. found that 17 days of voluntary wheel running significantly increased insulin- stimulated glucose levels in a rat sarcoma model [176]. As with human patients, no significant differences in blood glucose levels were observed as a result of exercise. In addition to systemic changes as a result of exercise, the availability of tumor tissue from necropsy facilitates evaluation of exercise- mediated changes in the tumor itself. Despite evidence from both clinical and preclinical studies that exercise affects systemic metabolism, its effect on tumor metabolism has not been investigated. Changes in host metabolism are likely to have a large impact on cancer cell energy use and resultant tumor growth. Notably, tumors rely more heavily than normal tissue on glycolysis; in this phenomenon, known as the Warburg effect, tumor cells often use less efficient anaerobic methods to derive energy from glucose rather than rely on oxidative 42

63 metabolism, even when tissue is adequately oxygenated [185]. This requires the tumor to have high levels of glucose, glucose transporters, and glycolytic enzymes [185]. Exercise- mediated changes in insulin, insulin receptors, insulin- mediated glucose receptors, and glucose delivery could have profound effects on tumor metabolism. The effects of exercise on tumor metabolism have not been investigated. Though not the focus of this dissertation, this topic certainly warrants further investigation. Only 3 (21%) of the 14 preclinical studies evaluated effects of exercise on intratumoral markers of neoplastic phenotype, focusing on DNA synthesis and vascularity (Table 3). In their model of azaserine- induced pancreatic cancer, Roebuck et al. reported that voluntary wheel running for up to 18 weeks did not significantly affect the rate of DNA synthesis assessed by 3 H- thymidine incorporation [178]. The other two experiments assessed markers of tumor angiogenesis. Zielinski et al. injected BALB/c mice subcutaneously with EL- 4 neoplastic lymphoid cells and treated them with 5-14 days of treadmill running; this group reported that intratumoral blood vessel density was decreased as a result of exercise [181]. More recently, Jones et al. showed that voluntary wheel running caused significantly increased HIF- 1 protein levels and number of perfused blood vessels in breast tumors implanted orthotopically in nude mice [172]. However, there were no significant differences between groups with respect to protein levels of CD31, VEGF, or AMP- activated protein kinase (AMPK). 43

64 These were the first experiments to investigate the effects of aerobic exercise on tumor vasculature, suggesting that exercise may result in changes in vessel density and function. However, blood vessel structure and function were not thoroughly investigated in these experiments, nor were the consequences of these vascular changes on the physiological microenvironment. This dissertation attempts to address these gaps in knowledge Conclusions Over the past decade, exercise has garnered increasing attention as a possible adjunct therapy for the treatment of cancer. Though much of the work to date has focused on symptom control, more recent clinical and preclinical studies have focused on the role of exercise in preventing tumor progression in breast cancer and other tumor types. Epidemiological evidence strongly suggests that physical activity can decrease cancer- specific and all- cause mortality in several tumor types, including breast cancer, and several potential biological mechanisms have emerged from these studies. Additionally, preclinical studies have been conducted to further evaluate this relationship, but the diverse models have yielded varying results. Furthermore, though exercise is widely known to cause improvements in vascular structure and function in other diseases, such as diabetes and coronary artery disease, little is known about the effects of exercise on aberrant tumor blood vessels and the tumor microenvironment. Studies reported in chapter 3 of this dissertation explore the effects of voluntary exercise 44

65 on primary breast tumor growth rate in an optimized model of the tumor microenvironment. Chapter 4 addresses exercise- mediated changes in vascular structure and function along with resultant changes in the physiological microenvironment. Chapter 5 describes studies on the effects of combining exercise with systemic chemotherapy to capitalize on vascular changes and improve therapeutic efficacy. We hope that this body of work yields valuable information that will lead to optimal use of exercise as an adjunct therapy to improve treatment efficacy and outcomes for breast cancer patients. 45

66 2. Materials and Methods 2.1 4T1 Breast Cancer Model 4T1 murine breast cancer cells that stably express luciferase (4T1- luc) were cultured in Dulbecco s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum. All cells were harvested by trypsinization at approximately 80% confluence in log phase growth. Forty- eight adult female BALB/c mice were obtained from Harlan Laboratories, Inc. (Madison, WI). All mice were individually housed (21 C, 35% to 45% humidity, 12:12- hour light- dark cycle) in cages with contact bedding. The animals were fed Purina Rodent Chow 5058 (LabDiet, Richmond, IN) and water ad libitum. Mice were allowed to acclimate for 4 weeks prior to experimental procedures. Animal care and all experimental procedures were in accordance with the Institutional Animal Care and Use Guidelines at Duke University Medical Center. Following a period of preconditioning (described later), all animals were injected orthotopically into the right dorsal mammary fat pad with 0.1 ml of DMEM containing 5 x T1- luc cells. 2.2 Exercise Intervention The exercise modality was voluntary wheel running, which is characterized by intermittent periods of high- intensity, short- duration exercise against low resistance throughout the dark cycle [172]. Animals randomized to exercise groups were given continuous access to a wheel 11.5 cm in diameter (Mini Mitter, Bend, OR), and wheel 46

67 revolutions were monitored continuously by a magnetic sensor using the VitalView data acquisition program (Respironics, Inc., Murrysville, PA). Mice randomized to sedentary groups were singly housed in cages without wheels. An overview of the experimental design is shown in figure 2. We conducted a pretraining period prior to injection of tumor cells during which animals were randomized to either 9 weeks of voluntary wheel running (n = 24) or sedentary control (n = 24). This was intended to mimic the clinical scenario, in which patients may present with either high or low levels of fitness at the time of breast cancer diagnosis. All mice were weighed, and running data were obtained weekly during the pretraining period. On day 0, following tumor cell injection, animals were randomized again to either running or sedentary groups such that there were four final animal groups: sedentary both before and after transplantation (SS), sedentary before and running after transplantation (SR), running before and sedentary after transplantation (RS), and running both before and after transplantation (RR) (n = per group). One animal in the RR group was euthanized due to development of a maxillary growth, which is thought to be unrelated to the exercise intervention. Animals were monitored 3 days a week for 18 days after tumor transplantation. Body weight, tumor size (by calipers), and running data were recorded. 47

68 Figure 2: Overview of experimental design to determine the effects of voluntary wheel running on primary breast tumor growth. Female BALB/c mice were randomized to either voluntary wheel running or sedentary controls for 9 weeks (days - 63 to - 1). On day 0, all animals were injected in the dorsal mammary fat pad with 5 x T1- luc cells and then randomized again to either running or sedentary conditions (RR, running then running; RS, running then sedentary; SR, sedentary then running; SS, sedentary then sedentary; n = per group). Exercising animals were given free access to a voluntary running wheel that uses a computer system to track wheel revolutions. Tumor volume (calipers), body weight, and running distance were monitored three times weekly. On the last day of the experiment (day 18), tumor perfusion was assessed using magnetic resonance imaging, and metastases were evaluated using bioluminescence. Tissues were snap- frozen in liquid nitrogen and processed for histology. 48

69 2.3 Magnetic Resonance (MR) Tumor Perfusion Imaging On the last day of the experiment, animals in the 4T1 experiment underwent MR imaging (n = 3 per group) to determine tumor perfusion using a Bruker 7T (70/30) system (Bruker Biospin, Billerica, MA, USA). This system has a quadrature surface receiver and volume transmit coil setup with active decoupling. Animals were anesthetized with isoflurane and positioned in an MR- compatible cradle that maintains body temperature using warm water circulation. Temperature and respiratory rate were continuously monitored. First, anatomic information was acquired by T2- weighted imaging using a RARE- based fast spin echo sequence with TR=4200, TE=12, RARE factor 8, 1 mm slice thickness, FOV 2.4 cm, 256 x 256, with respiratory gating. Tumor perfusion maps were generated by a double spin- echo planar pulse sequence using pairs of bipolar gradients at specific predetermined signs in each of three orthogonal directions. The combination of gradient directions allows cancellation of all off- diagonal tensor elements, hence measurement of the diffusion tensor trace, and so provides unambiguous and rotationally invariant apparent diffusion coefficient (ADC) values. Seven b values (b = 0, 50.0, 100, 150, 200, 500, 1000) were acquired, with a matrix size of 128 x 128, slice thickness 1.0 mm. Volume images (one for each b value) were created from raw DICOM images. For voxels within the matrix with a signal value above 2000, the ADC at each voxel was calculated using an exponential moving fit by the following method: ADC = ln [S(b = b1) S(b = b2)/b2- b1. The b1 and b2 values of 100 and 49

70 200, respectively, are sensitive to blood flow apparent diffusion changes in small arteries and capillaries. ADC maps were generated using mono- exponential fitting as above, and T2 images were zero- filled to prior to analysis. Parametric images were analyzed in anatomic regions of interest using Bruker Paravision software and offline using Osirix software. 2.4 In Vivo and Ex Vivo Bioluminescence Imaging For Metastasis Detection Luciferase expression from 4T1- luc cells, which was used to detect sites of metastatic tumor growth, was quantified in vivo in relative light units with a bioluminescence imaging system (Caliper Life Sciences, Hopkinton, MA) and the associated software (Living Image, Caliper Life Sciences). Images of total animal luciferase expression and organ- specific expression from the lungs and liver were acquired for each animal immediately prior to and after euthanasia, respectively, according to the manufacturer s specified procedures. During the imaging session, the mice were anesthetized with isoflurane and imaged 19 minutes after administration of 20 mg/kg of luciferin stock solution. Immediately after imaging, animals were administered a lethal dose (250 mg/kg) of pentobarbital. Tumors, lungs, and livers were excised and weighed. Immediately thereafter, total luciferase expression was assessed in the lungs and liver of each animal. Then, all organs were snap- frozen over liquid nitrogen and stored at - 80 C. 50

71 2.5 Exercise Plus Chemotherapy Model Figure 3 describes the experimental design used to determine the effects of combining voluntary wheel running and cyclophosphamide chemotherapy on primary breast tumor growth. Female BALB/c mice (8 to 10 weeks of age) were obtained from Harlan Laboratories, Inc. (Madison, WI). They were housed under the same conditions as described earlier. After acclimation, mice were injected orthotopically in the right dorsal mammary fat pad with 0.1 ml of DMEM containing 5 x T1- luc cells on day 0. Immediately following injection, mice were randomized to one of four groups: no treatment, exercise alone, cyclophosphamide alone, or exercise plus cyclophosphamide (n = 17 per group). Animals in exercising groups were given immediate access to a voluntary running wheel, and sedentary animals were singly housed in cages without wheels. Tumors were allowed to grow for 1 week, and then animals in the cyclophosphamide groups were injected with the maximal tolerated dose (100 mg/kg) of cyclophosphamide intraperitoneally on days 7, 9, and 11. Animals in the groups not receiving cyclophosphamide were handled similarly but not administered any chemotherapeutic agent. 51

72 Figure 3: Experimental design to determine the effects of voluntary aerobic exercise combined with cyclophosphamide chemotherapy on primary breast tumor growth. Female BALB/c mice were injected orthotopically with 5 x T1- luc cells on day 0 then randomized to exercise plus chemotherapy; chemotherapy; exercise; or no treatment (n = 17 per group). Exercising animals were given free access to a voluntary running wheel. Tumors were allowed to grow for 1 week; then animals randomized to receive chemotherapy received 100 mg/kg cyclophosphamide intraperitoneally on days 7, 9, and 11. Tumor volume (calipers), body weight, running distance, and tumor blood flow (optical spectroscopy) were measured three times weekly. Animals were euthanized on day 18 (when the first animals reached a volume of 1500mm 3 ), and tissues were processed for histology. As in the first experiment, all mice were weighed, tumor volume was measured by calipers, and running data were obtained three times weekly. Additionally, total 52

73 hemoglobin and hemoglobin saturation within the tumor were measured using optical spectroscopy (described later). Prior to euthanasia, three animals were randomly selected from each group to undergo MR imaging of tumor perfusion. Tumors were excised when they reached 1500 mm 3 or ulcerated. Three hours prior to euthanasia, mice received an intraperitoneal injection of 80 mg/kg EF5 [2-2(nitro- 1H- imidazole- 1- yl)- N- (2,2,3,3,3- pentafluoropropyl acetamide)] (generously provided by Dr. Cameron Koch, University of Pennsylvania, Philadelphia, PA), a marker that selectively targets regions of hypoxia. Tumors were excised, weighed, and then snap- frozen over liquid nitrogen and stored at - 80 C. 2.6 Immunohistochemistry Frozen tissue sections (10 µμm thick) were stained using specific antibodies as previously described [186]. Briefly, unless otherwise specified, sections were fixed in ice- cold methanol for 30 minutes. Non- specific binding was blocked by a 30- minute incubation at room temperature with 10% donkey serum. Sections were incubated with primary antibodies overnight at 4 C, followed by incubation with the appropriate fluorescently- conjugated secondary antibody. Samples were washed three times for 5 minutes in phosphate- buffered saline (PBS) between each successive step, and all antibodies were diluted in PBS unless otherwise stated. All staining protocols concluded with application of Hoechst (0.2 mg/ml, Sigma Aldrich, St. Louis, MO) for 5 minutes at room temperature to counterstain for cellular nuclei. Exclusion of the 53

74 primary antibody served as a negative control. Immunostained slides were stored at 4 C in 4% paraformaldehyde until imaging. Specific staining protocols are summarized in Table 5 and described below. Table 5: Summary of Immunohistochemistry Protocols. Marker (Source) Ki67 (Dako, Inc.) Cleaved caspase- 3 (Cell Signaling Technology, Inc.) CD31 (BD Biosciences) Desmin (Dako North America, Inc.) NG2 (EMD Millipore) 8- oxo- dg (Genox Co.) EF5 (ELK3-51, Dr. Cameron Koch) Fixation Solution Permeabilizing Solution 4% PFA 0.5% saponin in PBS Antibody Dilution/Wash Solution 0.5% saponin in PBS Blocking Solution 10% donkey serum in PBS 4% PFA n/a PBS 10% donkey serum Ice- cold methanol Ice- cold mixture of acetone and methanol (1:1) Ice- cold methanol Ice- cold acetone Ice- cold methanol n/a PBS 10% donkey serum n/a PBS 10% donkey serum n/a PBS 10% donkey serum 0.3% Tween 20 in PBS 0.3% Tween 20 in PBS 10% goat serum n/a PBS 10% donkey serum Primary Antibody Dilution Second Antibody Dilution 1:100 1:1000 1:400 1:1000 1:100 1:1000 1:200 1:1000 1:400 1:1000 1:500 1: :2 n/a Ki67 and Cleaved Caspase-3 To evaluate tumor cell proliferation, sections were fixed with 4% paraformaldehyde (PFA), rendered permeable with 0.5% saponin, and incubated overnight with rat anti- Ki67 (Dako North America, Inc., Carpinteria, CA) diluted 1:100. This was followed by incubation with donkey anti- rat fluorescent secondary antibody 54

75 diluted 1:1000 (DyLight 488, Jackson). All washes and antibody dilutions were performed with 0.5% saponin in PBS. To evaluate apoptosis, slides were fixed in 4% PFA and stained with anti- cleaved caspase- 3 antibody (Cell Signaling Technology, Inc., Danvers, MA), diluted 1:400. This was followed by incubation with donkey anti- rabbit fluorescent secondary antibody (DyLight 594, Jackson), diluted 1: CD31, Desmin, and NG2 To stain for vasculature, sections were incubated for 1 hour with rat anti- mouse CD31 antibody (BD Pharmingen, San Diego, CA), diluted 1:100, followed by incubation with donkey anti- rat fluorescent secondary antibody (DyLight 488, Jackson ImmunoResearch Laboratories, Inc., West Grove, PA), diluted 1: Pericyte coverage was assessed by desmin colocalization with CD31, and this was confirmed with NG2 colocalization with CD31. For both desmin and NG2 staining, slides were fixed in a 1:1 mix of acetone and methanol. Following fixation and blocking, slides were treated with rabbit polyclonal anti- Desmin (Abcam, Cambridge, MA), diluted 1:200 or rabbit polyclonal anti- NG2 (Millipore, Billerica, MA), diluted 1:400 overnight, followed by incubation with donkey anti- rabbit fluorescent secondary antibody (DyLight 594, Jackson), diluted 1: oxo-2 -deoxyguanosine Oxidative stress was identified by 8- oxo- 2 - deoxyguanosine (8- oxo- dg), a marker of oxidative damage to DNA. After fixation, slides were rendered permeable using PBS 55

76 with 0.3% Tween 20 (PBST), blocked with 10% goat serum, and incubated with mouse monoclonal anti- 8- hydroxy- 2 - deoxyguanosine (Genox, Baltimore, MD) diluted 1:500 overnight at 4 C, followed by incubation with Alexa Fluor- 594 goat anti- mouse IgG1 secondary antibody (Invitrogen, Grand Island, NY) diluted 1: EF5 The hypoxia marker EF5 was detected using a Cy5- conjugated mouse monoclonal ELK3-51 antibody (purchased from Dr. Cameron Koch, University of Pennsylvania) applied 1:2 overnight at 4 C Microscopy Slides were imaged using a high- resolution solid- state camera mounted on a fluorescence microscope (Axioscop Zplus, Carl Zeiss, Inc., Thornwood, NY). A 4,6- diamidino- 2- phenylindole filter was used to detect Hoechst CD31, Ki67, and cleaved caspase- 3 staining were detected with a FITC filter. NG2 and desmin staining were detected with a TRITC filter. EF5 images were detected with a Cy5 filter. All images were captured at 5x magnification in 16- bit monochrome signal depth. Fixed exposure times were preselected for each fluorochrome. A computer- controlled motorized stepping stage (Metamorph Imagining System, Molecular Devices Corporation, Sunnyvale, CA) was used to tile individual fields and stitch them to generate composite images of every tissue section. 56

77 2.6.6 Image Analysis The nuclear counterstain permitted region of interest (ROI) contour lines to be drawn around each tumor, and necrotic regions were excluded from ROIs based on the absence of tumor cell nuclei. Microvessel density (MVD) was calculated for each section by overlaying a fixed grid on the CD31 image (Adobe Photoshop CS2, Adobe Systems, Inc., San Jose, CA). Individual CD31- positive vessels were manually counted in every other visible field to determine a mean value for the section. All other immunostains were analyzed using Image J software (NIH, First, a single optimal threshold value was set for each channel on every set of images. All images in a series were analyzed using the same threshold to determine positively stained pixels within the ROI. For the nuclear stains Ki67 and cleaved caspase- 3, positive cells were counted from thresholded images using the particle analysis tool. To analyze pericyte coverage of tumor vessels, thresholded desmin images were overlaid on thresholded CD31 images. This procedure was repeated for NG2 colocalization with CD31. The total tumor area fraction containing colocalized pixels and the CD31- positive area fraction containing colocalized vessels were calculated to determine the fraction of tumor area containing mature blood vessels and vessel area covered by pericytes, respectively. Hypoxic tumor fraction was assessed by determining the amount of EF5- positively stained area divided by the total tumor area. Images were colored and adjusted for brightness and contrast for presentation only. 57

78 2.7 Optical Spectroscopy Measurement of Total Hemoglobin and Hemoglobin Saturation Optical spectroscopy measurements were made as described previously [187]. Briefly, a fiberoptic probe composed of one read fiber surrounded by six illumination fibers (Ocean Optics, Inc., Dunedin, FL) was coupled to a USB4000 Miniature Fiber Optic Spectrometer (Ocean Optics, Inc.). Optical spectroscopy measurements were made by placing the fiberoptic probe in direct contact with the tissue surface; contact was aided by use of ultrasound gel. Calibration was carried out daily by normalizing the diffuse reflectance data wavelength- by- wavelength to a reflectance puck measurement (Labsphere, Inc., North Sutton, NH) made with the probe in flush contact. The diffuse reflectance spectra ( nm) were measured through the skin surface above the tumor three times weekly and then processed using a Monte Carlo- based model of diffuse reflectance to extract the physiological information related to absorption and scattering properties of the tissue [188]. The amounts of total, oxygenated, and deoxygenated hemoglobin were calculated from these properties. In order to maximize the sampling depth of the method and minimize any influence of skin on these measurements, diffuse reflectance fits were performed over the range of 480 to 600 nm to extract the absorption properties. 2.8 Angiogenesis-Related Gene Expression Total RNA was extracted from the 4T1- luc tumors using the Qiagen RNeasy Miniprep Kit (Qiagen, Inc., Valencia, CA) according to the manufacturer s instructions. 58

79 Approximately 1 µμg of RNA was reverse- transcribed using the Bio- Rad iscript cdna Synthesis Kit (Bio- Rad Laboratories, Inc., Hercules, CA), and resulting cdna was stored at - 20 C until analysis. Quantitative real- time polymerase chain reaction (qpcr) was performed using iq SYBRGreen Supermix (Bio- Rad) according to the manufacturer s instructions. Gene expression was quantified (2 - ΔΔCt ) relative to the RPL19 housekeeping gene. A summary of the primers and PCR conditions used is provided in table 6. 59

80 Table 6: Summary of PCR Protocols. Gene of Interest Forward Primer Reverse Primer Thermocycle Conditions Ang1 5 - CGGATTTCTCTTCCCAGAAAC TCCGACTTCATATTTTCCACAA- 3 Denaturation: 95 C for 10 min 40 cycles: denaturation at 95 C for 15 sec, annealing and elongation at 60 C for 1 min Ang2 5 - CACACTGACCTTCCCCAACT CCCACGTCCATGTCACAGTA- 3 Denaturation: 95 C for 10 min 40 cycles: denaturation at 95 C for 15 sec, annealing and elongation at 60 C for 1 min Flt GTCACAGATGTGCCGAATGG TGAGCGTGATCAGCTCCAGG- 3 Denaturation: 95 C for 10 min 40 cycles: denaturation at 95 C for 15 sec, annealing and elongation at 60 C for 1 min RPL ATCCGCAAGCCTGTGACTGT TCGGGCCAGGGTGTTTTT- 3 Tie TTGAAGTGACGAATGAGAT ATTTAGAGCTGTCTGGCTT- 3 Denaturation: 95 C for 10 min 40 cycles: denaturation at 95 C for 15 sec, annealing and elongation at 60 C for 1 min VEGF 5 - CTGTGCAGGCTGCTGTAACG GTTCCCGAAACCCTGAGGAG- 3 Denaturation: 95 C for 10 min 30 cycles: denaturation at 94 C for 30 sec, annealing at 56 C for 30 sec, elongation at 72 C for 30 sec Dissociation 95 C for 15 sec, 60 C for 15 sec, 95 C for 15 sec VEGFR2 5 - AGAACACCAAAAGAGAGAGGAACG Statistics 5 - GCACACAGGCAGAAACCAGTAG- 3 Denaturation: 95 C for 10 min 30 cycles: denaturation at 94 C for 30 sec, annealing at 58 C for 30 sec, elongation at 72 C for 30 sec Dissociation 95 C for 15 sec, 60 C for 15 sec, 95 C for 15 sec Statistical analysis was carried out using StatView (SAS Institute, Inc., Cary, NC). Descriptive statistics are presented as mean ± standard error of the mean (SEM). 60

81 One- way ANOVA was used to compare differences in tumor variables between groups. Repeated- measures ANOVA was used to compare differences in body weight and running distance between groups throughout the experiments. All ANOVA and repeated- measure ANOVA tests were followed by Fisher s protected least significant difference (FPLSD) to determine whether individual groups differed from the control group if the data were normally distributed or the Mann- Whitney U test for data that were not normally distributed. Tumor growth rates were determined by linear regressions, and differences in growth rate between groups were assessed using an ANCOVA. For all tests, p <0.05 was considered statistically significant. 61

82 3. Effect of Voluntary Aerobic Exercise on Primary Breast Tumor Growth and Progression 3.1 Introduction Using tumor- bearing animal models, it is possible to evaluate both the effects of physical activity on cancer growth and progression and the biological mechanisms that mediate observed effects. As described in chapter 1, previous preclinical studies reported a variety of model systems and exercise techniques [ ]. The results have varied widely and have not adequately modeled or explained the epidemiological evidence of exercise- mediated decreases in breast tumor progression. The most reasonable explanation for differences between the clinical observations and the results of preclinical experiments is that the experimental model systems do not accurately represent the effects of exercise on host and tumor physiology. Many of the previous studies used forced exercise, treadmill running [170, 171, 174, 175, 179, 181] and/or swimming [173, 174, 177]. Forced exercise models in rodents have been shown to induce a stress response that promotes cancer progression [183]. Also, most preclinical exercise studies have utilized heterotopic subcutaneous tumor models [171, , 179, 181], which are not optimal to evaluate the effectiveness of anticancer therapies because the subcutaneous microenvironment differs greatly from that of the native tumor microenvironment, and tumors grown heterotopically rarely metastasize [189, 190]. Orthotopic tumor models, which involve 62

83 growth of a tumor in the native tissue of origin, more accurately mimic site- specific tumor- host interactions, organ- specific gene expression, and clinically relevant patterns of metastasis [189, 190]. Thus, orthotopic models more accurately predict tumor response to therapy [189, 190]. At present, only three studies have evaluated the effects of exercise in orthotopic models of breast cancer; two entailed chemical induction of breast tumors [169, 173], and one entailed implantation of human breast cancer cells into the mammary fat pad of immunodeficient mice [172]. Though human tumor xenografts enable investigators to study the behavior of human cancer cells, these models are problematic for studying efficacy of treatment because they require immunodeficient animals, and the immune system is a key component of the tumor microenvironment and host defense against tumor progression [190]. To date, not one study has evaluated the effects of voluntary exercise in an immunocompetent animal model of syngeneic tumor cells implanted in the orthotopic position [191]. To address this gap, we studied voluntary aerobic exercise in immunocompetent BALB/c mice that received transplanted syngeneic 4T1- luc cells in the dorsal mammary fat pad. The exercise intervention was designed to mimic four clinically relevant scenarios: (1) patients who are sedentary before and after diagnosis, (2) previously sedentary patients who begin exercising after diagnosis, (3) physically active patients who stop exercising after diagnosis, and (4) previously active patients who continue to 63

84 exercise after diagnosis. Voluntary wheel running was selected as the most realistic, least stressful model of aerobic exercise. Accordingly, animals in Groups 3 and 4 exercised for 9 weeks before tumor cell transplant, and animals in Groups 1 and 2 were sedentary during that time. Immediately after transplant, animals in Groups 2 (SR; sedentary before transplantation, then running) and 4 (RR; running before transplantation, then running) were given access to a voluntary running wheel, and animals in Groups 1 (SS; sedentary before and after transplantation) and 3 (RS; running before transplantation, then sedentary) were sedentary. Tumor volume, running distance, and body weights were recorded three times weekly for 18 days. Then, tumor perfusion was mapped using MR, bioluminescence was used to detect metastases, and tumors were removed for analysis. 3.2 Results Voluntary Exercise Performance Average running distances are shown in figure 4. Animals ran significantly more prior to 4T1 tumor cell transplantation than after (6.1 ± 0.4 km/day prior to transplant vs. 4.7 ± 0.4 km/day after transplant; p <0.02). There were no significant differences in mean running distances between the groups that exercised prior to tumor cell transplantation (6.6 ± 0.6 km/day RR vs. 5.6 ± 0.5 km/day RS; p >0.05) or after tumor implantation (4.4 ± 0.6 km/day RR vs. 4.9 ± 0.6 km/day SR; p >0.05). 64

85 Figure 4: Mean voluntary wheel running distance before and after 4T1 tumor cell transplantation. As shown here, BALB/c mice were given access to a voluntary running wheel for 9 weeks prior to tumor cell implantation in the RR ( ) and RS ( ) groups, and average daily running distance was assessed weekly. Mice were implanted with 5.0 x T1- luc cells in the dorsal mammary fat pad on day 0. Following transplantation, animals in the RS group were no longer given access to a voluntary wheel, animals in the RR group continued running, and previously sedentary animals in the SR ( ) group began running. Animals in the SS group were sedentary throughout the experiment. After tumor cell transplantation, running distances were assessed three times weekly for 18 days. Animals ran significantly less following tumor cell transplantation (p <0.02). There were no significant differences in mean running distances between groups. Data are expressed as mean ± SEM, Student s t- test. 65

86 3.2.2 Effect of Aerobic Exercise on Body Weight Exercise increases caloric expenditure and is often associated with changes in body weight. In a study of tumor growth, it is important to evaluate changes in body weight, as differences between groups could affect patterns of tumor growth. Figure 5 shows mean animal body weights throughout the experiment. Body weight increased in all groups over the course of the experiment, but there were no significant differences in body weight between treatment groups (starting weight: 21.8 ± 0.4 g SS, 22.1 ± 0.4 g RS, 22.1 ± 0.4 g SR, 22.8 ± 0.5 g RR, p >0.05; weight at necropsy: 24.7 ± 0.5 g SS, 24.5 ± 0.4 g RS, 24.5 ± 0.5 g SR, 25.6 ± 0.6 g RR, p >0.05). 66

87 Figure 5: Body weights over time of BALB/c mice randomized to voluntary wheel running or sedentary control. Mean body weights ± SEM (error bars) are shown. Body weights increased over time in all groups, but there were no significant differences between treatment groups at any time point (n = per group), one- way ANOVA. 67

88 3.2.3 Effects of Exercise on Primary Tumor Growth We use a clinically relevant model involving orthotopic implantation of syngeneic tumor cells in immunocompetent mice to assess the effects of voluntary aerobic exercise on the growth of primary breast tumors. On day 0, following the 9- week preconditioning period, all mice were injected with 4T1- luc cells in the dorsal mammary fat pad. Then animals in the posttransplantation exercise groups were given immediate access to a voluntary running wheel. Tumor volume was monitored by caliper measurements three times weekly for 18 days. Primary tumor growth rates are shown in figure 6A. Animals that ran before, after, or before and after tumor cell transplantation had significantly lower tumor growth rates than sedentary controls (p <0.01), but there were no significant differences between groups that ran only before, only after, or both before and after tumor cells were injected. To confirm that tumor growth was slowed by exercise, tumor weights were compared at the time of necropsy (figure 6B). Tumors were significantly smaller in the animals that ran at any point during the experiment than in sedentary controls (0.74 ± 0.10 g SS, 0.34 ± 0.04 g RS, 0.46 ± 0.04 g SR, 0.34 ± 0.04 g RR; SS vs. RS p <0.002, SS vs. SR p <0.002, SS vs. RR p <0.0001). 68

89 Figure 6: Tumor response to voluntary aerobic exercise. A. 4T1 tumor volume was measured by calipers three times weekly beginning on day 8. Data points are mean volume ± SEM (error bars). Tumor growth rate was assessed by linear regression, and differences between groups were assessed by ANCOVA. Tumor growth rate was significantly greater in the SS group ( ) than in animals that ran before and/or after tumor cell transplantation [RS( ), SR ( ), RR ( )], *p <0.01. B. Tumor weight at necropsy, 18 days after tumor cell transplantation. Tumors weights were significantly lower in the RR, SR, and RS groups than the SS group. **p <0.002, *** p<0.0001, one- way ANOVA with FPLSD post- hoc test. 69

90 To investigate whether the exercise- induced decrease in primary tumor growth rate was caused by decreased tumor cell proliferation or increased apoptosis, we immunostained sections for Ki67 and cleaved caspase- 3, respectively. Ki67 is a nuclear marker expressed exclusively by proliferating cells; it is present during all active phases of the cell cycle and absent from resting cells [192]. Caspase- 3 is a key downstream effector caspase, known to play a large role in breast cancer, that is involved in both the death receptor and mitochondrial pathways of apoptosis [193]. Produced as an inactive proenzyme, the cleaved form is the active enzyme that participates in apoptosis. The antitumor effects of exercise prior to tumor development are well accepted, so we were particularly interested in the effects of running in tumor- bearing hosts. Accordingly, data were analyzed with respect to activity level after tumor cell transplantation irrespective of the pretraining condition. Animals in the SS and RS groups were considered sedentary, and animals in the SR and RR groups were considered to be exercising. Figure 7 shows representative images of Ki67 immunostaining and provides quantification. There were no significant differences between groups in the density of proliferating cells (1549 ± 244 cells/mm 2 sedentary versus 1559 ± 211 cells/mm 2 running, p >0.05). In contrast, apoptosis was approximately 1.5- fold higher in the running animals than in sedentary controls (1544 ± 154 cells/mm 2 sedentary versus 2168 ± 259 cells/mm 2 running, p <0.04, figure 8). 70

91 Figure 7: Effect of exercise after tumor transplantation on proliferation of 4T1 breast tumor cells. Representative color composites of proliferating cells (Ki67+, green) in animals that were sedentary (A) or running (B) after tumor cell transplantation. Cellular nuclei are stained with Hoechst (blue). 5x objective, magnification 100%. The density of proliferating cells was quantified (C) as Ki67 positive cells per square millimeter of tumor area. Columns, mean; bars, SEM; p >0.05, Student s t- test. 71

92 Figure 8: Apoptosis in primary 4T1 breast tumors in response to exercise after tumor transplantation. Representative color composites of immunostaining for apoptotic cells (cleaved caspase- 3, green) in tumors from animals that were sedentary (A) or running (B) after tumor cell transplantation. Cellular nuclei are stained with Hoechst (blue). 5x objective, magnification 100%. The density of proliferating cells was quantified (C) as cleaved caspase- 3 positive cells per square millimeter of tumor area. Columns, mean; bars, SEM; *p <0.04, Student s t- test. 72

93 3.2.4 Effects of Exercise on Breast Tumor Metastasis Given that voluntary wheel running inhibited growth of primary breast tumors and increased tumor cell apoptosis, we sought to determine whether this intervention affected progression to metastatic disease. In BALB/c mice, 4T1 tumors spontaneously metastasize to the lung, liver, lymph nodes, and brain, while primary tumors grow in situ [ ]. We used 4T1 cells that constitutively express luciferase, enabling us to detect metastases using bioluminescence imaging. Mice were injected with 20 mg/kg of luciferin intraperitoneally, and live animals were imaged 19 minutes later. Animals were then euthanized, and the liver and lungs from each animal were imaged ex vivo to ensure that a weak signal from small metastatic sites was not missed. Representative images are shown in figure 9. In all animals, luciferase was clearly expressed by primary tumors, but there was no evidence of metastasis in any of the animals on imaging either in vivo or ex vivo. 73

94 Figure 9: Representative bioluminescence scans for 4T1 tumor metastasis. Animals bearing 4T1- luc tumors were injected with luciferin and scanned for bioluminescence. A. Animals were anesthetized with isoflurane, and in vivo whole animal bioluminescence was measured. In this representative image, the primary tumor is clearly visible, and there is no evidence of distant metastasis. B. Lungs (left) and liver (right) from each animal were removed and imaged ex vivo to ensure that weak signals were not missed from small sites of metastasis. There was no evidence of metastatic disease in any of these organs. To confirm that there was no evidence of tumor metastasis, lungs and livers were weighed, and organ weights were compared across groups. There were no significant differences in organ weights between treatment groups (Lung: 0.21 ± 0.01 g SS, 0.21 ± 0.01 g RS, 0.22 ± 0.01 g SR, 0.23 ± 0.01 g RR, p >0.05; Liver: 1.41 ± 0.03 g SS, 1.42 ± 0.04 g RS, 1.44 ± 0.08 g SR, 1.36 ± 0.04 g RR, p >0.05), further suggesting that there was no difference in tumor metastasis between groups after 18 days of tumor growth. 74

95 Figure 10: Mean lung and liver weights from animals bearing 4T1 breast tumors. Lung and liver weights at necropsy were compared between treatment groups for evidence of metastatic disease. Organ weights were comparable between groups, suggesting that there were no significant differences in tumor metastasis. 75

96 3.3 Discussion Epidemiological evidence strongly suggests that exercise is associated with improved survival rates following a diagnosis of breast cancer [4-6, 11]. However, previous preclinical studies conducted in mice and rats have yielded conflicting results regarding the effects of exercise on primary tumor growth. These studies have been plagued by inadequacies of the model system to mimic the effects of exercise on a tumor growing in the native microenvironment in an otherwise healthy host. Shortcomings of previous model systems include heterotopic locations for tumor growth and the use of xenografts in immunodeficient animals, both of which limit key tumor- host interactions and microenvironmental influences that are known to affect tumor growth [189, 190]. Additionally, interpretation of previous experiments is complicated by the common use of forced exercise techniques, which stress the animals and have been shown to promote cancer progression [183]. To address these deficiencies with a more clinically relevant model, we implanted immunocompetent female mice orthotopically with syngeneic murine breast tumor cells and selected voluntary wheel running to minimize stress to the animal and mimic exercise behaviors in patients. Using this model, we demonstrate that voluntary aerobic exercise inhibits growth of primary breast tumors. Animals that exercised did not have significantly different body weight from those that were sedentary, so it is unlikely that exercise- 76

97 mediated changes in body size account for the differences in tumor growth. In general, tumor growth rate is determined by a balance of cellular proliferation and death. In an effort to understand the biological mechanisms underlying the exercise- induced reduction in tumor growth rate, we evaluated both of these processes using immunohistochemistry. Immunostaining for the nuclear proliferation marker Ki67 suggests that exercise did not change the rate of cellular proliferation. This is supported by a previous study conducted in rats with pancreatic cancer, which reported that voluntary wheel running did not affect the rate of DNA synthesis in tumor cells [178]. In contrast, exercise caused a significant increase in tumor cell apoptosis as assessed by the downstream apoptotic marker cleaved caspase- 3. This is in accordance with a recent report showing that anaerobic exercise caused a twofold increase in Walker 256 tumor cell apoptosis in rats [197]. The authors of that study attributed the observed increase in apoptosis and decrease in tumor growth rate to exercise- mediated increases in oxidative stress, increased expression of the pro- apoptotic protein Bax, and decreased expression of the apoptosis inhibitor Bcl2. The effects of aerobic exercise on oxidative stress and other microenvironmental factors will be discussed in more detail in the next chapter, but voluntary wheel running was associated with decreased oxidative stress in our model. In contrast to acute, severe oxidative stress, chronic exposure to reactive oxygen and nitrogen species can cause resistance to apoptosis through p53-77

98 mediated mechanisms and by stabilizing HIF- 1, leading to up- regulation of prosurvival signals [198, 199]. Therefore, exercise- induced reductions in tumor oxidative stress could increase the sensitivity of cancer cells to apoptotic signaling. However, given that similar increases in apoptosis were observed in two studies that reported opposite effects of exercise on oxidative stress, it seems unlikely that reactive oxygen species are responsible for exercise- induced increases in tumor cell apoptosis. Further investigation of the causes of apoptosis are clearly needed to understand and capitalize on these effects for tumor control. As a starting point, it would be helpful to determine whether expression of Bax and Bcl2 are affected by exercise in our model, but other key mediators of apoptosis, such as p53, Fas ligand, inhibitor of apoptosis proteins, and tumor necrosis factor should be examined as well. It is noteworthy that exercise- induced inhibition of tumor growth occurred whether exercise was performed before or after tumor cell transplantation. Our data indicate that exercise conducted after tumor development slows tumor progression, supporting the epidemiological evidence that exercising after a diagnosis of breast cancer improves outcomes. Furthermore, these results suggest that in addition to the known effects of preventing cancer, exercise performed before the development of a tumor may affect the behavior of cancer cells that appear later. Interestingly, exercise- induced increases in tumor cell apoptosis were observed only in animals that were given 78

99 access to a running wheel after tumor transplantation; apoptosis rates were similar in animals in the RS and SS groups and significantly different from those in the SR and RR groups, which suggests that the effects of physical activity before tumor development are different from those that occur with exercise after the tumor appears. Proliferation rates were comparable in all groups, so it is unlikely that the pretraining period inhibited tumor growth by direct effects on cellular proliferation. It is unclear how exercise conducted before tumor cell transplantation inhibited tumor growth. It is possible that systemic changes in metabolism, oxidative stress, hormone levels, or other growth factors or similar changes in the tissue microenvironment where the tumor would later develop could account for these results. Both of these possibilities warrant further exploration, and potential approaches are discussed in chapter 6. Though inhibition of tumor growth and local tumor control are important outcomes of any cancer therapy, metastatic disease is the most common cause of cancer death. The survival benefit observed in epidemiological exercise oncology studies suggests that exercise may inhibit progression to metastasis. We selected the 4T1 model of breast cancer because these cells can spontaneously metastasize while the primary tumor grows in situ [ ]. We assessed metastasis in two ways, using both bioluminescence to image distant tumor cells and weighing common target organs to screen for differences between groups. Neither method revealed any evidence of 79

100 metastatic disease in any of the animals. It is likely that the primary tumors grew so quickly that there was not time to observe significant metastatic growth before the primary tumors were large enough to necessitate euthanasia. As described in chapter 6, future studies should further explore the effects of aerobic exercise on metastasis. In conclusion, we demonstrated in this study that voluntary aerobic exercise slows tumor growth rate and increases apoptosis in immunocompetent mice implanted orthotopically with syngeneic breast cancer cells. Similar growth inhibitory effects were observed whether exercise was performed only before, only after, or both before and after tumor cell transplantation, though exercise- induced increases in apoptosis were limited to animals that ran after injection of 4T1 cells. This is the first study to use this clinically relevant model, and it confirms that aerobic exercise inhibits growth of primary breast tumors. 80

101 4. Effects of Voluntary Aerobic Exercise on the Tumor Microenvironment 4.1 Introduction Tumors are heterogeneous structures consisting of much more than just cancer cells. As tumors grow, malignant cells are surrounded by a variety of supporting structures, including blood vessels, fibroblasts, extracellular matrix, and immune cells, all of which interact physically and physiologically to influence the composition and behavior of a tumor [18]. Together, these structures and the physiological consequences of their interactions make up the tumor microenvironment, which plays a key role in development, progression, and metastasis of the tumor, and in efficacy of treatment. Accordingly, understanding and therapeutically modifying the microenvironment have been a major research focus in oncology. Abnormal tumor vasculature and resultant hypoxia are key aspects of the microenvironment that profoundly affect progression of the tumor and efficacy of treatment. This is a particularly common problem in breast cancer. Between 25% and 98% of invasive breast tumors from different clinical cohorts have been reported to contain regions of hypoxia [33-40]. The implications of this are important, because hypoxia increases aggressiveness, invasiveness, and metastasis of the tumor and decreases the host s immune response, resulting in poor outcomes [41-45]. Furthermore, hypoxia is associated with increased resistance to chemotherapy in breast cancer [40, 46]. 81

102 As a result, many groups sought ways to prevent or treat hypoxia, but efforts to increase intratumoral oxygenation or target hypoxia have yielded only modest results to date [200]. Finding effective and lasting ways to improve tumor oxygenation continues to be an important therapeutic goal in breast cancer. The studies described in this chapter were designed to investigate whether aerobic exercise can combat tumor hypoxia and improve tissue oxygenation. The causes of hypoxia in tumors are well characterized. Inadequate oxygenation results from an inability of the vasculature to deliver enough oxygen to fulfill the metabolic requirements of the tumor. Tumor vessels have highly abnormal structure. They tend to be dilated, tortuous, and leaky, with large fenestrations between endothelial cells and very little coverage by basement membrane and perivascular cells [201]. This creates a vicious circle. Structurally abnormal vessels function poorly, resulting in hypoxia, stabilization of HIF- 1, and increased expression of VEGF and other angiogenic signaling molecules, which in turn trigger the formation of new poorly functional vessels [70-74]. To reduce tumor hypoxia, groups have sought to interrupt this cycle of hypoxia and unproductive angiogenesis by restoring balance between proangiogenic and antiangiogenic signaling. Jain popularized the term vascular normalization to describe the change from aberrant and poorly functional tumor blood vessels to remodeled, mature, 82

103 functional vasculature that results from blockade of VEGF signaling [79, 99, 100]. The proposed benefits of normalization include improved patterns of perfusion, reduced vessel permeability, decreased interstitial fluid pressure, and increased delivery of oxygen and systemic cytotoxic agents to the tumor [79]. Furthermore, normalization has been shown to improve the efficacy of chemotherapy in several types of cancer, including breast cancer [93, 94, 96, 97, ]. However, the clinical utility of vascular normalization is limited by the brevity of the normalization window, which is the period during which tumor vessels are normalized by antiangiogenic therapy. The benefits of normalization are reportedly lost within 6 weeks of beginning therapy, likely because tumors become resistant to the blockade of angiogenesis [79, 106, 107]. We think that aerobic exercise may be an alternative way to create long- lasting normalization of tumor vasculature. The benefits of aerobic exercise on systemic vascular function are well accepted. Exercise- mediated improvements in endothelium and vascular function have been reported in several clinical populations, including patients with coronary artery disease, diabetes, and congestive heart failure [164, 168, 202, 203]. Accordingly, it is plausible that exercise could cause similar improvements in tumor vasculature. Two previous preclinical experiments provided preliminary evidence that exercise affects tumor vasculature. Zielinski et al. showed that treadmill running decreased blood vessel 83

104 density in subcutaneous lymphoid tumors in BALB/c mice [181]. Additionally, Jones et al. reported that voluntary wheel running caused an increase in the number of perfused tumor blood vessels in nude mice bearing orthotopic human breast tumor xenografts [172]. These early studies hinted at exercise- induced changes in the number of tumor blood vessels and patterns of flow through them, but they did not fully explore the effects on vessel structure and function. This chapter describes experiments designed to determine whether exercise normalizes tumor vasculature and to explore the implications of exercise- induced changes on the tumor microenvironment. To do this, we analyzed tumor tissue from the experiments described in chapter 3. An overview of the experimental design is provided in figure 2. Briefly, BALB/c mice were assigned to a pretraining period of 9 weeks of voluntary wheel running or sedentary control and then injected orthotopically with syngeneic 4T1 breast tumor cells. Immediately following tumor cell transplantation, animals were randomized again to either exercise or sedentary conditions so that there were four total treatment groups [sedentary then sedentary (SS), running then sedentary (RS), sedentary then running (SR), and running then running (RR); n = per group). Tumors were allowed to progress for 18 days, and then perfusion was mapped using MR, followed by removal of the tumor for analysis. We used immunohistochemistry and qpcr to examine the effects of exercise on tumor angiogenesis, vessel normalization, and resultant effects on the 84

105 microenvironment. Our primary interest was the microenvironmental effects of aerobic exercise after the development of a tumor, so we analyzed data with respect to exercise after tumor cell transplantation. To mimic the clinical scenario in which patients present at the time of diagnosis with varying histories of physical activity, all animals that were sedentary after injection of 4T1 cells (SS and RS) were considered to be sedentary. Similarly, animals in the SR and RR groups were analyzed together in the exercise group. 4.2 Results Effects of Exercise on Tumor Vascularity and Angiogenic Signaling To determine whether exercise affects the number of tumor blood vessels, frozen sections were stained for the endothelial cell marker CD31. Representative images and quantification are shown in figure 11. Voluntary wheel running resulted in a significant increase in microvessel density (23 ± 1 vessels/mm 2 sedentary versus 32 ± 3 vessels/mm 2 running; p <0.003). 85

106 Figure 11: Exercise- mediated increases in tumor microvessel density. Representative color images of immunostaining for endothelial cells (CD31, green) in animals that were animals that were sedentary (A) or running (B) after tumor cell transplantation. 5x objective, magnification 100%. Microvessel density was quantified (C) as CD31+ structures per square millimeter of viable tumor area. Columns, mean; bars, SEM; *p <0.003, Student s t- test. Given the observed increase in tumor vascularity caused by exercise, we evaluated changes in expression of key genes involved angiogenesis. As shown in figure 12, we evaluated expression of VEGF and its two principle receptors VEGFR- 1 (Flt1) and VEGFR- 2, using qpcr to determine whether changes in expression of these key angiogenic signaling proteins could explain the exercise- mediated increase in vascular density. Expression of VEGF was nearly 1.5- fold higher in the running group than in sedentary controls (0.81 ± 0.06 relative expression level sedentary versus 1.14 ± 0.10 relative expression level running; p <0.01). Expression of VEGFR- 2, the key mediator of VEGF- induced angiogenesis [204], was not significantly different between groups (1.87 ± 0.52 relative expression level sedentary versus 1.40 ± 0.21 relative expression level 86

107 running; p >0.05). Although its function is not fully understood, VEGF has been also shown to bind to VEGFR- 1. So we evaluated expression of this receptor, but there were not significant differences in expression between groups (0.84 ± 0.10 relative expression level sedentary versus 0.84 ± 0.11 relative expression level running; p >0.05). Figure 12: Effects of voluntary wheel running on expression of VEGF, VEGFR- 1, and VEGFR- 2 mrna in tumor tissue. Total RNA was extracted from 4T1- luc tumors and reverse transcribed. Gene expression was quantified relative to the RPL19 housekeeping gene. Each sample was run in triplicate. Columns, mean; bars, SEM; *p <0.01, Student s t- test. 87

108 4.2.2 Exercise-Mediated Changes in Tumor Vessel Maturity Given the increase in tumor angiogenesis described in the previous section, we were interested in whether the maturity and functionality of the vessels were affected by aerobic exercise. The maturity of tumor vessels was assessed by staining for pericytes (desmin or NG2) surrounding tumor vessels (CD31). Pericyte- covered vessels were identified by colocalization of CD31 with desmin. Representative images and quantification are shown in figure 13. Running significantly increased the amount of viable tumor area containing pericyte- covered vessels compared to sedentary controls (0.4 ± 0.1% of tumor area sedentary versus 1.9 ± 0.5% running; p <0.0003); exercise also significantly increased the vessel area covered by pericytes (22.1 ± 3.7% of vessel area sedentary versus 39.6 ± 6.4% of vessel area running; p <0.02). 88

109 Figure 13: Effects of aerobic exercise on vascular maturity in primary 4T1 breast tumors. Representative two- color composite images of tumor vessel pericyte coverage determined by colocalization of tumor vessels (CD31, green) and pericytes (desmin, red) in animals that were sedentary (A) or running (B) after tumor cell transplantation. 5x objective, magnification 100%. The percentage of tumor area containing pericyte- covered blood vessels was quantified (C) as CD31- desmin colocalized pixels/viable tumor area pixels x100%. The percentage of vessel area covered by pericytes was quantified (D) as CD31- desmin colocalized pixels/cd31+ pixels x100%. Columns, mean; bars, SEM; *p <0.003, **p <0.02, Student s t- test. 89

110 The observed exercise- mediated increase in vessel pericyte coverage was confirmed by immunostaining for the pericyte marker NG2, and similar results were observed (0.8 ± 0.3% of viable tumor area sedentary versus 3.3 ± 0.6% running; p <0.0004; 16.5 ± 3.8% of vessel area sedentary versus 46.6 ± 7.3% of vessel area running; p <0.0006; data not shown). In an effort to explain how exercise could influence pericyte coverage of tumor vessels, we examined the angiopoietin signaling axis using qpcr. Ang1 and Ang2 are ligands that bind to the Tie- 2 tyrosine kinase receptor on endothelial cells [205]. This ligand- receptor signaling axis is central in the regulation of vascular maturity. Expression of Ang1 causes phosphorylation of Tie- 2, maintaining mature vessels with normal pericyte coverage and deterring remodeling [206]. In contrast, Ang2 prevents Tie- 2 autophosphorylation, causing pericytes to dissociate from endothelial cells and the basement membrane to degrade, preparing endothelial cells to initiate vascular remodeling and angiogenesis in response to VEGF [207]. Results of qpcr are shown in figure 14. There were no significant differences between groups in mrna levels of Ang1, Ang2, or Tie- 2 (Ang1: 7.70 ± 1.55 relative expression level sedentary versus 6.19 ± 1.84 relative expression level running; p >0.05; Ang2: 1.70 ± 0.13 relative expression level sedentary versus 1.55 ± 0.16 relative expression level running; p >0.05; Tie- 2: ±

111 relative expression level sedentary versus ± 3.62 relative expression level running, p >0.05). Figure 14: Effects of exercise on mrna expression levels of components of the angiopoietin signaling axis in 4T1- luc tumor tissue. Total RNA was extracted from 4T1- luc tumors and reverse- transcribed. Gene expression was quantified relative to the RPL19 housekeeping gene. Each sample was run in triplicate. Columns, mean; bars, SEM; Student s t- test Effects of Exercise on Tumor Perfusion Changes in vessel number and maturity are often associated with improved perfusion, a fundamental tenet of the vessel normalization hypothesis. To determine whether blood flow was affected by voluntary wheel running, we used MR to generate 91

112 perfusion maps in six randomly selected animals per group. Perfusion maps and mean ADCperf for each tumor are shown in figure 15. Perfusion of the tumors was highly heterogeneous. There were no significant differences in mean ADCperf between groups (1.2x10-6 ± 0.2 sedentary versus 1.6x10-6 ± 0.2 running; p >0.05). However, qualitative inspection of the perfusion maps indicates that exercise affects patterns of blood flow. Tumors from animals in the control group tended to show a ring- enhancement pattern of perfusion, with most of the blood flow around the outer edges of the tumor and little in the center. Comparatively, tumors from running animals showed much more uniform distribution of flow throughout. Figure 15, C and D shows the distribution and variability of perfusion values, respectively, in running versus sedentary animals. Clearly, tumor perfusion is much more variable in sedentary animals than in those that ran after tumor cell transplantation. 92

113 Figure 15: Effects of exercise on tumor perfusion. Animals from both running and sedentary treatment groups (n = 6/group) bearing 4T1- luc tumors in the mammary fat pad underwent in vivo MR perfusion mapping. Mean ADCperf was calculated for each tumor, and patterns of perfusion were analyzed qualitatively. Tumors from sedentary animals (A) showed highly heterogeneous patterns of flow, with more perfusion around the edges of the tumor and less in the center. In comparison, tumors from animals that ran (B) were more uniformly perfused. (C) Overlaid shadow histogram showing the distribution of perfusion values in exercising and sedentary animals. (D) Bar chart showing the variability of perfusion values in the tumor (ROI) in exercising versus sedentary animals Effects of Exercise on Tumor Hypoxia Tumor hypoxia results from an inability of the tumor blood vessels to meet the demand for oxygen. Improvements in vessel maturity and patterns of tumor perfusion 93

114 would be expected to improve delivery of oxygen to the tissue, thereby reducing hypoxia. To assess whether aerobic exercise affected hypoxia in 4T1- luc tumors, BALB/c mice were injected orthotopically with 4T1 breast tumor cells then randomized to running or sedentary controls for 18 days. Three hours prior to euthanasia, animals were administered 80 mg/kg of the hypoxia marker EF5 [2- (2- nitro- 1H- imidazole- 1- yl)- N- (2,2,3,3,3- pentafluoropropyl acetamide)], a nitroimidazole that binds selectively to hypoxic cells [208]. Frozen sections were immunostained with an antibody specific to EF5. As shown in figure 16, the percent of tumor that was EF5 positive was approximately 2.5- fold higher in the sedentary animals than those that ran (46.3 ± 7.8% of viable tumor area sedentary versus 19.2 ± 6.4% of viable tumor area running; p <0.002). 94

115 Figure 16: Exercise- mediated decreases in tumor hypoxia. BALB/c mice with 4T1- luc breast tumors in the mammary fat pad were treated with voluntary wheel running or sedentary control for 18 days. Three hours prior to euthanasia, animals were injected intraperitoneally with the hypoxia marker EF5. Cellular nuclei are stained with Hoechst 33342, and EF5 (red) was detected by immunohistochemistry. Representative whole tumor images are shown for sedentary (A) and running (B) animals. Quantification of hypoxic tumor percentage (C) was calculated as EF5+ pixels/total viable tumor pixels x100%. Columns, mean; bars, SEM; *p <0.002, Student s t- test Exercise-Induced Decreases in Oxidative Stress An important consequence of aberrant and dysfunctional tumor vasculature is the generation of free radical species, which tip the oxidative balance and result in oxidative stress in tumors [32]. To determine whether exercise affected oxidative stress in 4T1- luc tumors, we immunostained sections with an antibody against 8- oxo- 2 - deoxyguanosine (8- oxo- dg), an oxidized derivative of deoxyguanosine that forms under conditions of oxidative stress [209]. As shown in figure 17, tumors from animals in the 95

116 running group had significantly less 8- oxo- dg than tumors from sedentary control animals (3.6 ± 0.9% of viable tumor area sedentary versus 1.4 ± 0.5% of viable tumor area running; p <0.05). Figure 17: Exercise- induced changes in tumor oxidative stress. Tumor sections were immunostained for 8- oxo- dg (red), a marker of oxidative damage to DNA. Cellular nuclei are stained with Hoechst x objective, magnification 100%. A and B. Representative two- color composite images from tumor sections from sedentary or running animals. C. Quantification of 8- oxo- dg+ pixels per square millimeter of viable tumor area. Columns, mean; bars, SEM; *p <0.05, Mann- Whitney U test Effects of Exercise on the Physical Microenvironment: Innate Immunity As described in chapter 1, there are many important interactions between the physical and physiological microenvironments. Among the most important for tumor progression are the effects on immune function, as this is a key host defense against the tumor. Hypoxia and oxidative stress are known to affect the function of the innate immune system, particularly macrophages [210, 211]. To investigate whether voluntary 96

117 wheel running affects infiltration of innate immune cells, we immunostained tumor sections for CD11b, a marker on the surface of monocytes, granulocytes, macrophages, and NK cells [212]. Representative images and quantification are shown in figure 18. CD11b staining showed no significant differences between groups (3.8 ± 0.4% of tumor area sedentary versus 4.0 ± 0.4% of tumor area running; p >0.05) Figure 18: Effects of voluntary wheel running on innate immune cell infiltration. Representative two- color composite images of immunostaining for innate immune cells (CD11b, green) in animals that were animals that were sedentary (A) or running (B) after tumor cell transplantation. 5x objective, magnification 100%. Quantification (C) was performed as CD11b+ pixels/total tumor pixels x100%. Columns, mean; bars, SEM; p>0.05, Student s t- test. 4.3 Discussion The microenvironment of a tumor plays a key role in determining the behavior of a developing tumor and its responsiveness to therapy. The abnormal structure of tumor vessels causes them to function poorly, leading to regions of hypoxia. In breast cancer, this is a common problem that is known to affect the tumor s aggressiveness, 97

118 invasiveness, metastasis, and response to chemotherapy [33-46]. Normalizing tumor vasculature using moderate amounts of antiangiogenic agents to improve vessel structure and function has been proposed as way to reduce tumor hypoxia and its consequences [79, 99, 100]. This approach is reported to improve patterns of perfusion, reduce hypoxia, increase delivery of systemic chemotherapy to the tumor, and improve the efficacy of chemotherapy in several tumor types, including breast cancer [79, 93, 94, 96, 97, ]. The major limitation to using vascular normalization as a therapeutic approach is that the effects of antiangiogenic therapy are transient [79, 106, 107]. Given that aerobic exercise is associated with improvements in vessel structure and function in healthy people as well as patients with other diseases involving abnormal vasculature [164, 168, 202, 203], we conducted these experiments to determine whether voluntary wheel running could cause tumor vessel normalization in mice. BALB/c mice bearing orthotopic 4T1- luc tumors either were treated with voluntary wheel running or served as sedentary controls for 18 days after tumor cell transplantation. At this time, tumor tissue was analyzed for exercise- induced changes in the microenvironment. Immunofluorescence staining of tumor sections for the endothelial marker CD31 suggests that tumor blood vessel density is increased as a result of aerobic exercise. This contradicts an earlier study by Zielinski et al. that reported decreases in the number of 98

119 tumor blood vessels, identified by staining for von Willebrand factor VIII, in mice with subcutaneous flank tumors of lymphoma cells treated with 3 hours of daily treadmill running or sedentary control [181]. This discrepancy could be explained by differences in the model systems. The previous study used a subcutaneous model of lymphoma cells. As discussed in chapter 3, heterotopic models are suboptimal for evaluating microenvironmental effects because they fail to recapitulate tissue- specific tumor- host interactions [189, 190]. In our experiment, syngeneic tumor cells were injected orthotopically into the mammary fat pad of mice. Differences in exercise intervention could also have affected tumor angiogenesis. Zielinski et al. reported up to 3 hours of continuous treadmill running. In our experiment, the exercise intervention consisted of voluntary wheel running, which generally consists of intermittent high- intensity, short- duration exercise against low resistance throughout the 12- hour dark cycle [172]. The authors of the lymphoma cell study hypothesized that the observed decrease in tumor vessel density could have been caused by a decrease in VEGF levels resulting from exercise, but this was not evaluated. We used qpcr to determine the effects of voluntary wheel running on the VEGF signaling axis. We found that exercise resulted in increases in VEGF; expression levels of the main VEGF receptors were not significantly different between groups. Increased expression of VEGF likely plays a role in the observed increases in angiogenesis with exercise, but other proangiogenic signaling molecules, 99

120 such as fibroblast growth factor and PDGF, warrant further exploration as well. As these are the only two studies that have evaluated changes in tumor microvessel density in response to exercise, general conclusions are difficult to draw. Future studies using different model systems, exercise interventions, and breast tumor lines are necessary to clearly understand the effects of exercise on breast tumor angiogenesis. The increased number of tumor blood vessels and expression of VEGF suggest that exercise causes the formation of new blood vessels, but it is unclear whether the newly formed vessels are mature and functional. To investigate tumor vessel maturity, a key feature of vascular normalization, we used immunohistochemistry to detect pericytes surrounding blood vessels. Using two markers of pericytes (NG2 and desmin), we demonstrated that voluntary wheel running increased the viable tumor area containing pericyte- covered vessels as well as the vessel area covered by pericytes. In other words, tumors from running animals contained a greater number of mature blood vessels than those of sedentary controls, and more of the vessels that were present were mature. Increased vessel maturity is generally associated with improved functionality, so we used MR to generate perfusion maps of 4T1- luc tumors. Our data indicate that tumors are more uniformly perfused in running animals than in sedentary controls. In addition to suggesting that running improves vascular function, the uniformity of tumor 100

121 flow has important implications for response to therapy. Previous work from our group indicates that heterogeneous ring- enhancement patterns of perfusion similar to those seen in the sedentary animals in our study are associated with increased resistance to chemotherapy and hyperthermia [213]. Therefore, by homogenizing patterns of blood flow throughout the tumor, exercise may increase the sensitivity of tumor cells to radiation. Future experiments designed to explore this relationship are described in chapter 6. Increased vascular maturity and improved perfusion are key outcomes of tumor vessel normalization. To determine whether these changes affect tumor hypoxia, animals were given the hypoxia marker EF5 prior to euthanasia. Immunohistochemical detection of EF5 revealed that hypoxia was significantly decreased in tumors from running animals. This is a notable finding, because tumor hypoxia perpetuates abnormal angiogenesis, increases aggressiveness, increases metastasis, and decreases sensitivity to chemotherapy [32]. Many groups have tried to prevent or treat tumor hypoxia, but efforts to pharmacologically increase intratumoral oxygen or target hypoxia have yielded only modest results [214]. Therefore, exercise- mediated improvements in tumor oxygenation are an exciting new approach to targeting hypoxia and the tumor microenvironment. 101

122 Taken together, the findings of exercise- related increases in tumor vessel density and maturity, more uniform perfusion, and decreases in hypoxia provide strong evidence that aerobic exercise normalizes tumor vasculature. The mechanisms underlying this effect are unclear at present. We used qpcr to examine changes in expression of the angiopoietins and their receptor Tie- 2, key mediators of vascular maturity, but we did not observe any significant changes in mrna levels as a result of exercise. However, Ang1- induced phosphorylation of Tie- 2 is critical to promoting vascular maturity, and qpcr is not capable of detecting changes in phosphorylation. In future work, protein phosphorylation should be assessed using Western blotting or immunohistochemistry to determine whether changes in phosphorylation status could account for the improvements in vessel structure. Several other pathways are important for vascular maturity, and these warrant further exploration as well. For example, PDGF- B is involved in the recruitment of pericytes, and sphingosine- 1- phosphate- 1, endothelial differentiation sphingolipid G protein- coupled receptor- 1, and transforming growth factor- β are involved in pericyte stabilization of vasculature [73, 82, 83]. Expression of these signaling molecules in response to exercise should be investigated in future work. In addition to changes in angiogenic signaling, exercise- induced vascular normalization may also be explained by changes in vessel shear stress as a result of 102

123 exercise. In other diseases, including diabetes and coronary artery disease, exercise causes changes in vascular flow [165, 168, 215]. Modulations of flow cause changes in vascular tone, and sustained changes in vascular tone lead to changes in tumor vessel architecture [216, 217]. These dynamic changes in flow and structure are difficult to study ex vivo. Possible ways to study vessel flow and remodeling dynamics in vivo in the future are discussed in chapter 6. The experiments discussed in this chapter also revealed exercise- mediated changes in oxidative stress. Previous experiments in rodents have demonstrated that levels of reactive oxygen species are higher in tumors than in normal tissues [32, 218]. Sources of oxidative stress include mutations in mitochondrial DNA, which may result in inefficiencies in the electron transport chain; hypoxia- induced inefficiencies in electron transport leading to excess mitochondrial superoxide; up- regulation of NADPH oxidase; and overexpression of nitric oxide synthases [ ]. Immunohistochemical detection of 8- oxo- dg, a marker of oxidative damage to DNA, showed that exercise results in a 60% reduction in oxidative stress in the tumor. This finding aligns with a previous clinical study that reported decreases in systemic oxidative stress as a result of aerobic exercise in colorectal cancer patients [223]. In contrast, Jones et al. reported that aerobic exercise increased urinary F(2) isoprostanes, a marker of systemic oxidative stress, in patients with non- small cell lung cancer [163]. As discussed in chapter 3, 103

124 anaerobic exercise in rats caused an increase in oxidative stress in Walker 256 tumors [197]. This discrepancy could be explained by differences in exercise type and intensity, as higher- intensity exercise is known to increase oxidative stress [224]. Chronic oxidative stress can increase tumor growth and metastasis, as well as stabilize HIF- 1, causing increased tumor angiogenesis, so exercise- mediated reductions in oxidative stress may have profound effects on breast tumor growth and progression [198]. Experimental approaches to exploration of the role of oxidative stress in exercise- mediated effects on breast cancer are detailed in the future directions section of chapter 6. In summary, the experiments in this chapter demonstrate that voluntary aerobic exercise improves the structure and function of tumor blood vessels. Using immunohistochemistry, we showed that exercise increased the density and maturity of tumor blood vessels. Immunostaining for the hypoxia marker EF5 indicated that exercise resulted in improved tumor oxygenation. Furthermore, MR perfusion mapping demonstrated that tumors received more uniform blood flow in exercising animals than in sedentary controls. Taken together, the data in this chapter provide the first evidence that aberrant tumor blood vessels can be normalized non- pharmacologically using voluntary aerobic exercise. 104

125 5. Efficacy of Combining Aerobic Exercise with Cyclophosphamide Chemotherapy for Breast Cancer Treatment 5.1 Introduction The previous two chapters have described evidence that voluntary wheel running inhibits primary breast tumor growth and normalizes tumor vasculature in an orthotopic model of murine breast cancer. Though exercise shows promise for inhibiting tumor growth as a monotherapy, the vessel- normalizing effect presents the opportunity for improved antitumor efficacy when combined with chemotherapy. As described in chapter 1, abnormalities in tumor vessel structure result in inefficient delivery of oxygen and systemically administered therapies to the tumor [225]. The normalization hypothesis put forth by Jain suggests that phenotypic improvements in tumor blood vessels seen with judicious use of antiangiogenic therapy should improve the efficacy of systemically delivered cytotoxic therapy by decreasing interstitial fluid pressure, improving vascular function, and decreasing tumor hypoxia [100]. Several clinical studies that combined anti- VEGF therapy with systemic chemotherapy, including one in breast cancer, have supported this hypothesis [93, 94, 96, 97, 105]. Given that exercise caused improvements in tumor vessel structure and function as described in chapter 2, we hypothesized that voluntary wheel running would augment the efficacy of cyclophosphamide chemotherapy. To date, only one group has 105

126 examined the combination of exercise with chemotherapy. In this study, nude mice bearing subcutaneous human breast tumor xenografts were treated with doxorubicin chemotherapy and moderate- intensity treadmill running [171]. Exercise did not influence the antitumor effect of doxorubicin chemotherapy. However, there are several noteworthy limitations to interpreting this study. First, the exercise modality was treadmill running, a forced exercise intervention that may cause a tumor- promoting stress response as discussed in chapter 2 [183]. Additionally, treatment efficacy was investigated in immunodeficient mice bearing heterotopic human tumor xenografts. Heterotopic models are not ideal for investigating the efficacy of anticancer agents because the model tumor environment differs greatly from the native tumor microenvironment, and this difference has important effects on tumor response to therapy [189, 190]. Furthermore, the use of immunodeficient mice neglects the inhibitory effects of the host immune system on tumor progression [190]. As exercise is known to have important effects on the immune system [226], it is difficult to evaluate the antitumor effects of exercise or other therapies in an immunodeficient model system. We elected to conduct this study in a more clinically relevant model, which involved transplantation of syngeneic 4T1- luc tumor cells into the mammary fat pad of immunocompetent BALB/c mice. The experimental design is shown in figure 3. Immediately after tumor cell transplantation, mice were randomized to one of four 106

127 groups: no treatment, exercise alone, chemotherapy alone, or exercise plus chemotherapy (n = 17 per group). Mice in the exercise groups were given free access to a voluntary running wheel, a model of aerobic exercise that minimizes stress to the animal. Tumors were allowed to grow for 1 week, and then animals in the chemotherapy groups were treated with the maximal tolerated dose (100 mg/kg) of cyclophosphamide, injected intraperitoneally, on days 7, 9, and 11. Cyclophosphamide is an alkylating agent that crosslinks DNA strands to prevent cell division; this is an irreversible process that causes cell death. It was selected because it is a commonly used first- line chemotherapy agent for breast cancer [ ]. Tumor volume, running distance, and body weights were monitored three times weekly. Additionally, optical spectroscopy was used to measure total hemoglobin and hemoglobin saturation in the tumor three times weekly. Tumors were allowed to grow for a total of 18 days, and then perfusion was mapped using MR and tumors were removed for analysis. 5.2 Results Effects of Chemotherapy on Voluntary Exercise Performance Average running distances are shown in figure 19. Treatment with cyclophosphamide did not deter mice from voluntary exercise. There were no significant differences between groups in mean daily running distance (6.8 ± 0.5 km/day running versus 7.4 ± 0.5 km/day running + cyclophosphamide; p >0.05). 107

128 Figure 19: Mean daily voluntary wheel running distance in the exercise and exercise plus cyclophosphamide groups. BALB/c mice were injected with 4T1- luc tumor cells and given immediate access to a voluntary running wheel on Day 0. Animals exercise + cyclophosphamide group ( ) were administered 100 mg/kg cyclophosphamide intraperitoneally on days 7, 9, and 11, whereas animals in the exercise group ( ) were handled similarly but not injected with chemotherapy. Running distances were assessed three times weekly for 18 days. There were no significant differences in mean running distances between groups at any of the time points. Data are expressed as mean ± SEM, Student s t- test. 108

129 5.2.2 Effects of Aerobic Exercise and Cyclophosphamide on Body Weight Both aerobic exercise and chemotherapy treatment are commonly associated with decreases in body weight, which has the potential to confound observed effects on tumor growth. Mean body weights over the course of the experiment are shown in figure 20. Body weights increased in all groups during the experiment, but there were no significant differences between groups (starting weight: 22.4 ± 0.6 g no treatment, 22.5 ± 0.7 g cyclophosphamide, 23.0 ± 0.6 g exercise, 22.8 ± 0.7 g exercise + cyclophosphamide; p >0.05; weight at necropsy: 24.1 ± 0.7 g no treatment, 24.1 ± 0.7 g cyclophosphamide, 24.3 ± 0.6 g exercise, 23.9 ± 0.7 g exercise + cyclophosphamide; p >0.05). 109

130 Figure 20: Body weights of BALB/c mice bearing 4T1- luc tumors randomized to no treatment, cyclophosphamide, exercise, or exercise plus cyclophosphamide. Mean body weights ± SEM (error bars) are shown. Body weights increased over time in all groups, but there were no significant differences between treatment groups at any time point (n = 17 per group), one- way ANOVA Effects of Aerobic Exercise and Cyclophosphamide on Primary Tumor Growth Using a clinically relevant model of breast cancer involving orthotopic implantation of syngeneic tumor cells in immunocompetent mice, we assessed the effects of combining voluntary aerobic exercise with cyclophosphamide chemotherapy on the growth of primary breast tumors. On day 0, mice were injected with 4T1- luc cells 110

131 in the dorsal mammary fat pad. Immediately after injection, animals in the exercise groups were given access to a voluntary running wheel. Tumors were allowed to grow for 1 week. On days 7, 9, and 11, animals in the chemotherapy groups were injected IP with the maximal tolerated dose (100 mg/kg) of cyclophosphamide [230]. Tumor volume was monitored by caliper measurements three times weekly for 18 days. Primary tumor growth rates are shown in figure 21. Both exercise and cyclophosphamide monotherapies significantly slowed tumor growth relative to animals that received no treatment (p <0.01). Notably, the tumor growth rate of animals treated with exercise was comparable to that of animals treated with cyclophosphamide (p >0.05). Additionally, animals in the exercise plus cyclophosphamide group had significantly slower tumor growth rates than animals treated with exercise alone or cyclophosphamide alone (p <0.01). 111

132 Figure 21: Tumor response to voluntary aerobic exercise combined with cyclophosphamide chemotherapy. BALB/c mice were injected with 4T1- luc cells in the dorsal mammary fat pad and randomized to voluntary wheel running or sedentary controls (n = 17 per group) starting on day 0. Animals in chemotherapy groups received 100 mg/kg cyclophosphamide IP on days 7, 9, and 11. Data are presented as mean ± SEM (error bars). Tumor growth rate is significantly slower in groups receiving exercise and cyclophosphamide monotherapies than in sedentary controls. The combination of exercise and cyclophosphamide results in a significant reduction in tumor growth rate relative to the groups that received no treatment, exercise alone, or cyclophosphamide alone. *p <0.01, ANCOVA. In chapter 3, we demonstrated that voluntary wheel running caused increased rates of tumor cell apoptosis in BALB/c mice bearing orthotopic 4T1- luc breast tumors. Cellular proliferation, assessed by immunostaining for Ki67, was unchanged by exercise. 112

133 Here we demonstrate that the addition of exercise to cyclophosphamide chemotherapy slows tumor growth rate more than either monotherapy alone. To determine whether the observed decrease in growth rate was caused by changes in proliferation or apoptosis, we euthanized five animals per group immediately after the last dose of chemotherapy. This enabled us to determine the response of the tumor to the chemotherapy. We immunostained sections for Ki67 and cleaved caspase- 3 to measure proliferation and apoptosis, respectively. Representative images of Ki67 immunostaining are shown, and quantification is provided, in figure 22. There were no significant differences between groups in the density of proliferating cells (621 ± 132 cells/mm 2 no treatment, 412 ± 286 cells/mm 2 cyclophosphamide, 511 ± 142 cells/mm 2 exercise, 306 ± 104 cells/mm 2 exercise + cyclophosphamide; p >0.05). Figure 23 shows representative images of cleaved caspase- 3 immunostaining and quantification. Though there is a trend toward increased apoptosis in animals treated with cyclophosphamide and/or exercise, there are no significant differences between groups (1605 ± 451 cells/mm 2 no treatment, 2178 ± 396 cells/mm 2 cyclophosphamide, 2881 ± 297 cells/mm 2 exercise, 2448 ± 696 cells/mm 2 exercise + cyclophosphamide; p >0.05). 113

134 Figure 22: Effects of cyclophosphamide and/or exercise after tumor transplantation on proliferation of 4T1 breast tumor cells. A. Representative color composites of proliferating cells (Ki67+, green) in 4T1 tumors from animals that were treated with cyclophosphamide and/or voluntary wheel running after tumor cell transplantation. Cellular nuclei are stained with Hoechst (blue). 5x objective, magnification 100%. The density of proliferating cells was quantified (B) as Ki67 positive cells per square millimeter of tumor area. Columns, mean; bars, SEM; p >0.05, one- way ANOVA. 114