Cardiac Atrophy Due to Cancer: Characterization, Mechanisms, and Sex Differences

Transcription

1 University of Colorado, Boulder CU Scholar Molecular, Cellular, and Developmental Biology Graduate Theses & Dissertations Molecular, Cellular, and Developmental Biology Spring Cardiac Atrophy Due to Cancer: Characterization, Mechanisms, and Sex Differences Pippa Froukje Cosper University of Colorado Boulder, Follow this and additional works at: Part of the Cell Biology Commons, and the Molecular Biology Commons Recommended Citation Cosper, Pippa Froukje, "Cardiac Atrophy Due to Cancer: Characterization, Mechanisms, and Sex Differences" (2011). Molecular, Cellular, and Developmental Biology Graduate Theses & Dissertations This Dissertation is brought to you for free and open access by Molecular, Cellular, and Developmental Biology at CU Scholar. It has been accepted for inclusion in Molecular, Cellular, and Developmental Biology Graduate Theses & Dissertations by an authorized administrator of CU Scholar. For more information, please contact

2 CARDIAC ATROPHY DUE TO CANCER: Characterization, Mechanisms and Sex Differences by Pippa F. Cosper B.S., University of Texas, Austin, 2002 A thesis submitted to the Faculty of the Graduate School of the University of Colorado In partial fulfillment of the requirement for the degree of Doctor of Philosophy Department of Molecular, Cellular, Developmental Biology 2011

3 This thesis entitled: Cardiac Atrophy due to Cancer: Characterization, Mechanisms, and Sex Differences written by Pippa Froukje Cosper has been approved for the Department of Molecular, Cellular, and Developmental Biology Dr. Greg Odorizzi Dr. Leslie Leinwand Date: The final copy of this thesis has been examined by the signatories, and we find that both the content and the form meet acceptable presentation standards of scholarly work in the above mentioned discipline.

4 Abstract Cosper, Pippa Froukje (Ph.D., Molecular, Cellular, Developmental Biology) Cardiac Atrophy due to Cancer: Characterization, Mechanisms, and Sex Differences Thesis directed by Professor Leslie A. Leinwand Approximately one-third of cancer deaths are caused by cachexia, a severe form of skeletal muscle and adipose tissue wasting that affects men more than women. The heart also undergoes atrophy in cancer patients but the extent, functional consequences, mechanisms and sex differences have not been elucidated. In a mouse colonadenocarcinoma model, cancer causes a loss of cardiac mass due to a decrease in cardiac myocyte size that is associated with reduced levels of all sarcomeric proteins. I provide evidence that published reports showing a selective decrease in myosin heavy chain (MyHC) during cancer cachexia are likely an artifact resulting from muscle lysis methods which do not solubilize myosin out of myofibrils. I show that MyHC decreases in parallel with other myofibrillar proteins in cachectic cardiac and skeletal muscle. Unlike skeletal muscle, atrophic hearts do not upregulate the ubiquitin-proteasome system (UPS) or its activity but increase autophagy. Thus, cancer causes cardiac atrophy by a mechanism distinct from that in skeletal muscle. This murine model recapitulates the sexual dimorphism associated with cachexia in human patients. I demonstrate that male tumorbearing mice have a more severe phenotype than females, including greater cardiac mass loss and mortality. In females, estrogen protects against cancer-induced cardiac atrophy and body weight loss by signaling through its receptor. I also examined the effect of iii

5 dietary phytoestrogens on cancer cachexia. A soy diet worsens cachexia in both sexes, while a casein-based diet decreases the extent of skeletal and cardiac muscle mass loss in males yet increases cancer mortality in both sexes. Finally, I established an in vitro model of cardiac atrophy. Interferon-γ (IFN), a pro-inflammatory cytokine commonly increased in the serum of cancer patients and in the myocardium of patients suffering from Chagas disease, causes cardiac myocyte atrophy, but unlike the mechanism in vivo, IFN activates the UPS and causes the specific degradation of MyHC in a proteasome-dependent manner. Together, these studies provide mechanistic insight into cardiac atrophy due to cancer and inflammatory cytokines, and demonstrate that cardiac and skeletal muscle cachexia are modulated by both sex and diet. iv

6 Acknowledgments Many people have contributed to this body of work either directly or indirectly and I would like to express my gratitude to all of them. I would first like to thank my mentor, Leslie Leinwand, for her continual support and allowing me to take full control of this project as an independent scientist. She set a positive example of a successful scientist involved in both academia and the biotechnology industry, and is very inspiring to me. Leslie was also a good mentor in matters beyond science and I am so grateful to have her influence in my life, which will hopefully be maintained throughout my career. I would also really like to thank Arthur Gutierrez-Hartmann, who saw my potential and has supported me in this MD/PhD journey. I have had the opportunity to work with wonderful people and very talented scientists. I would like to thank Massimo Buvoli for his continual willingness to talk about science with me, providing brilliant ideas, and of course for making fun of me nearly every day. Steve Langer, Bob Thompson, and Brooke Harrison have also been very helpful and full of knowledge. I would really like to thank Ann Robinson for all of her support and friendship, for doing most of the lab duties, and especially for preparing the neonatal rat ventricular myocytes. Thank you to Tom Giddings and Christina Clarissa for their help in acquiring beautiful electron microscopy images of the heart. I also want to thank Janice Jones and Sandra Duff, who s organizational and computer skills have been instrumental in submitting grants and manuscripts. I owe a big thanks to Margaret Isenhart for taking such wonderful care of my mice and always looking out for their health. I would also like to thank all the mice that have sacrificed their lives for these studies. v

7 I would like to thank one of my best friends, Kristen Barthel, for teaching me so many valuable science and life lessons and for always being there for me. Thank you to Pam Harvey, who has helped me immensely with advice on publishing and writing this dissertation, and for being such a great friend. I would also like to thank Betsy Luczak, Cecilia Riquelme, and Dave Busha for their friendship and camaraderie. I could not have done this without my wonderful family, who have been nothing but supportive throughout my 20,000 years of education. My mother and father truly are the best parents and I thank them for all of their love, encouragement, and support throughout these trying years. My sister has also been wonderful and supportive and I am proud to be her sister. Last but not least, I would like to thank my husband. Oh wait, I don't have one because I have been stuck in lab too much I have had a wonderful, fulfilling experience in this lab and have learned so much. I could not have done it without all the help of the students, staff, and faculty in the MCDB department. I express my gratitude to all of you. vi

8 CONTENTS CHAPTER I: Introduction 1 Cancer Cachexia 1 Causes of Cancer Cachexia: pro-inflammatory cytokines 3 Skeletal Muscle and Atrophy 7 The Ubiquitin Proteasome System 10 The Lysosome 16 Autophagy 17 Ca 2+ -dependent Calpains and Caspases 19 Cardiac Muscle 22 Cardiac Atrophy 27 The Ubiquitin Proteasome System 33 The Lysosome and Autophagy 38 Ca 2+ -dependent Calpains 41 Conclusions and Research Aims 43 CHAPTER II: Cancer causes cardiac atrophy and autophagy in a sexually 46 dimorphic manner Introduction 46 Materials and Methods 48 Results 54 Discussion 73 CHAPTER III: Quantification of MyHC in atrophic muscle 79 Introduction 79 Materials and Methods 81 Results 84 Discussion 95 CHAPTER IV: Interferon-γ causes cardiomyocyte atrophy and the specific 102 degradation of myosin heavy chain in vitro. Introduction 102 Materials and Methods 105 Results 109 Discussion 125 vii

9 CHAPTER V: The effects of sex and diet on cancer survival and cachexia 131 Introduction 131 Materials and Methods 136 Results 138 Discussion 158 CHAPTER VI: Conclusions and Future Directions 166 CHAPTER VII: References 179 Appendix I. Real-time PCR primers 201 viii

10 LIST OF TABLES Table 1. Serum cytokine levels in control and tumor-bearing male 58 and female mice. Table 2. Transthoracic echocardiography measurements 62 Table 3. The effects of sex and diet on weight loss and cardiac gene 149 expression. ix

11 LIST OF FIGURES 1. Skeletal muscle architecture 8 2. Myofibrillar proteins in the cardiac sarcomere Body weight and cardiac mass loss Pro-inflammatory cytokines in tumor-bearing mice Estrogen receptor signaling is required for the maintenance of 59 cardiac and body mass in female tumor-bearing mice. 6. Fibrosis and echocardiography Cardiomyocyte size and apoptosis MyHC and myofibrillar protein levels Changes in cardiac MyHC isoform expression Atrogin-1 and MuRF-1 expression Quantification of ubiquitinated proteins and UPS activity Effects of bortezomib on cancer cachexia Autophagy in atrophic hearts MyHC solubility in high and low salt MyHC in supernatant vs. pellet MyHC levels in cachectic muscle in different lysis buffers Quantification of MyHC in the same muscle lysed with different buffers Sarcomeric α-actin levels Western blot analysis of MyHC in cachectic muscle Quantification of MyHC and actin in cachectic muscle Neonatal rat ventricular myocyte size upon IFN treatment 111 x

12 22. Cardiac MyHC levels in atrophic NRVMs Myofibrillar protein levels in atrophic NRVMs Protein synthesis and degradation rates in IFN-treated NRVMs Degradation of S 35 labeled proteins The proteasome and MyHC loss and myocyte atrophy UPS and immunoproteasome activity in atrophic NRVMs Autophagy and MyHC degradation in NRVMs Physiological levels of IFN induce MyHC degradation Sex differences in skeletal muscle mass loss due to cancer Sex differences in atrogin-1 and MuRF-1 expression in skeletal muscle Effects of estrogen on male cardiac atrophy Effects of sex and diet on colon-26 adenocarcinoma survival Effects of sex, diet and cancer on body weight Effects of sex and diet on body weight loss at days 15 and Effects of sex, diet and cancer on skeletal muscle mass Effects of sex and diet on skeletal muscle loss at days 15 and Effects of sex, diet and cancer on heart weight Effects of sex and diet on cardiac muscle loss at days 15 and Effects of sex and diet on cardiac MyHC gene expression Effects of sex and diet on cardiac autophagy gene expression 159 xi

13 CHAPTER I. Introduction Cachexia is a form of severe adipose tissue and skeletal muscle wasting associated with several chronic diseases, including acquired immunodeficiency syndrome (AIDS), chronic obstructive pulmonary disease (COPD), renal disease and cancer (Morley, Thomas et al. 2006). Cachexia affects more than 5 million people in the United States (Morley, Thomas et al. 2006). Approximately 50% of cancer patients suffer from cachexia and it contributes to greater than 20% of cancer deaths (Tisdale 1997; Tijerina 2004). Not all cancer patients suffer from cachexia; the incidence of cachexia is low in leukemia and breast cancer, but very high in gastrointestinal and lung cancers (Acharyya and Guttridge 2007). In fact, cachexia is a causative factor in 30-50% of deaths in patients with gastrointestinal cancer, and up to 80% of deaths in patients with pancreatic cancer (Bachmann, Heiligensetzer et al. 2008). Weight loss associated with cachexia differs from that due to starvation as it is mostly attributed to loss of skeletal muscle over adipose tissue. For an equivalent amount of weight loss, there is a greater degree of skeletal muscle mass lost in cancer cachexia (Heymsfield and McManus 1985). This preferential loss of lean body mass contributes to mortality by causing generalized weakness, increased susceptibility to infection and decreased responsiveness to both chemotherapy and radiation. The presence of cachexia therefore implies a poor prognosis as the degree of weight loss is inversely correlated with survival (Argiles, Busquets et al. 2005). The heart is also known to atrophy in humans and rodents with cancer (Burch, Phillips et al. 1968; Zhou, Wang et al. 2010). 1

14 However, cardiac atrophy is under-studied compared to skeletal muscle cachexia, thus the main goal of this dissertation is to describe the mechanisms responsible for cancerinduced cardiac myocyte atrophy. Carbohydrate, Lipid and Protein Metabolism in Cancer Patients Cachexia is caused by abnormal metabolism, which is initiated by an inflammatory state. Increases in basal metabolic rate can occur before the onset of weight loss, which indicates that metabolic changes are causative rather than a consequence of cachexia (Tisdale 1997). Solid tumors mainly rely on anaerobic metabolism and therefore consume large amounts of glucose, which can rival the amount consumed by the brain. Malignant tumors obtain over half of their energy requirements from glycolysis, which keeps the host in a constant state of gluconeogenesis (Tijerina 2004). This increased demand for glucose results in increased Cori cycle activity, which produces glucose from lactate in the liver, and is also an energy wasting process. Normal glucose homeostasis is therefore significantly disturbed as glucose is continually produced and consumed (De Blaauw, Deutz et al. 1997). This increased glucose turnover contributes to the increased energy expenditure of the tumor-bearing host (Tisdale 1997). A large proportion of body weight loss in cancer patients is fat loss due to increased lipolysis. Cancer patients commonly have increased plasma glycerol levels and free fatty acids and in some cases, tissues utilize these as a preferred energy source over glucose (Tisdale 1997). Decreased lipoprotein lipase levels and activity also contribute to increased triglycerides in blood. Cancer patients with depleted fat stores therefore commonly have hypertriglyceridemia, which contributes to immunosupression (De 2

15 Blaauw, Deutz et al. 1997). Tumors may also produce lipid-mobilizing factors that increase release of fatty acids for use as energy. Increased fatty acid oxidation decreases fat stores, and combined with increased gluconeogenesis and increased metabolic rate, results in a net loss of body weight. One of the most characteristic features of cancer cachexia is loss of lean body mass. Muscle is the largest protein pool in the body and normally accounts for nearly 50% of whole body protein turnover (De Blaauw, Deutz et al. 1997). The skeletal muscles are therefore the major sites of protein loss (Tisdale 1997). Increased muscle protein turnover can occur in patients with very small tumor burdens and is therefore independent of tumor size. Muscle mass loss is often due to decreased protein synthesis coupled with increased protein degradation, resulting in a negative nitrogen balance. Hepatic protein synthesis actually increases in order to increase production of acute phase proteins, which are produced in order to respond to inflammation and tumors (De Blaauw, Deutz et al. 1997). Thus, net protein loss occurs in the musculature, while the weights of other organs, such as the kidneys and lungs, are unaffected by cancer. Causes of Cancer Cachexia The etiology of cancer cachexia is multifactorial, but appears to be humorally mediated (Norton, Moley et al. 1985). Pro-inflammatory cytokines are known contributors to body mass loss. Cytokines may be produced by a tumor itself or by the host when mounting an immune response to the tumor. The main pro-inflammatory cytokines that play a direct role in cancer cachexia are tumor necrosis factor α (TNF-α), interleukin-6 (IL-6), interleukin-1β (IL-1β) and interferon-γ (IFN-γ). Evidence for the 3

16 importance of tumor-derived cytokines lies in one study that showed that host derived cytokines are quantitatively less important than cytokines derived from the tumor as genetic deletion of host cytokines does not decrease tumor growth or improve cachexia (Cahlin, Korner et al. 2000). TNF-α, also known as cachectin, has been implicated in cachexia in numerous studies. Mice inoculated with CHO cells that constitutively express TNF-α develop a syndrome that resembles cancer cachexia, namely anorexia, muscle atrophy, and early death (Oliff, Defeo-Jones et al. 1987). The same symptoms occur when mice are injected with recombinant human TNF-α for 10 days (Tracey, Wei et al. 1988). Administration of anti-tnf antibodies to rats with hepatomas decreases protein degradation rates in skeletal and cardiac muscle but does not prevent body weight loss (Costelli, Carbo et al. 1993). TNF-α can also activate muscle proteolysis directly (Mahony, Beck et al. 1988). Despite these established roles of TNF-α in cachexia, it is not always increased in the serum of cancer patients or tumor-bearing rodents, suggesting that other factors are also involved in causing cachexia (De Blaauw, Deutz et al. 1997; Tisdale 1997). TNF-α also plays a role in the energetic inefficiency of the tumor-bearing host. Uncoupling proteins (UCP) dissipate the proton gradient across the inner mitochondrial membrane, which uncouples respiration from ATP synthesis and contributes to energy wasting. Tumor growth and TNF-α increase the expression of UCPs and TNF-α is able to uncouple respiration in isolated mitochondria (Argiles, Moore-Carrasco et al. 2003; Argiles, Busquets et al. 2005). Thus TNF-α appears to be involved in both muscle proteolysis and hypermetabolic changes. 4

17 IL-6 has been implicated in causing cancer cachexia mainly through studies in rodent tumor models. In a murine colon-26 adenocarcinoma model, IL-6 is a cachectic factor as monoclonal antibody to IL-6, but not to TNF-α, was able to decrease the development of cachexia (Strassmann, Fong et al. 1992). Additionally, mice with a mutation in the tumor suppressor gene Apc develop intestinal tumors and have increased serum IL-6 levels. These mice develop severe cachexia and a decrease in gastrocnemius weight, while IL-6 null mice in the same background did not lose any skeletal muscle mass. Interestingly, the presence of a tumor in addition to IL-6 was necessary for the induction of cachexia in this model (Baltgalvis, Berger et al. 2008). Most importantly, cancer patients with weight loss have statistically significant increased IL-6 levels compared to those that do not lose weight (Scott, McMillan et al. 1996). In vitro studies with skeletal muscle myotubes have shown that IL-6 increases the activity of proteolytic pathways and thereby decreases the half-life of long lived proteins (Ebisui, Tsujinaka et al. 1995). IL-1β is another cytokine implicated in cancer cachexia and is involved in enhancing lipolysis (Tisdale 1997). IL-1β alone cannot cause cachexia and it appears to require the presence of other cytokines, namely IL-6, to exert cachectic effects. It does so by inducing IL-6 production by the tumor, which increases muscle wasting. Thus inhibiting IL-1β signaling can reduce cachexia, albeit indirectly (Strassmann, Masui et al. 1993). However, administration of an IL-1β receptor antagonist does not reverse cachexia in a rat cancer model (Costelli, Llovera et al. 1995). IL-1β therefore likely acts synergistically with other cytokines in order to cause cachexia but can contribute to weight loss directly by inducing anorexia (Argiles, Busquets et al. 2005). 5

18 IFN-γ is a cytokine produced by activated lymphocytes and has biological activities that overlap with TNF-α (Argiles, Busquets et al. 2005). Administration of IFN-γ to non-tumor bearing rats causes a decrease in food intake and body weight. Conversely, treatment of tumor-bearing rats with antisera to rat IFN-γ results in increased food intake, decreased body weight loss, and longer survival, suggesting that IFN-γ is an important mediator of cachexia (Langstein, Doherty et al. 1991). Other studies have also shown a direct link between IFN-γ and cachexia as injection of CHO cells that constitutively express IFN-γ causes profound cachexia (Matthys, Dijkmans et al. 1991), but the presence of a tumor was required for this effect, implying that IFN-γ cannot, by itself, cause cachexia. Additionally, administration of anti-ifn-γ antibody reverses muscle wasting in a Lewis lung carcinoma model (Matthys, Heremans et al. 1991). Patients with colorectal or pancreatic cancer, two cancers that cause severe cachexia have increased serum concentrations of IFN-γ that at least in colorectal cancer patients, is associated with higher resting energy expenditure and weight loss (Ravasco, Monteiro- Grillo et al. 2007; R, A et al. 2009). Proteolysis inducing factor (PIF) is another humoral factor implicated in cancer cachexia. It is a glycoprotein found in the blood and urine of tumor-bearing animals and humans with weight loss, but not in cancer patients without weight loss. PIF has been purified from a cachexia inducing tumor and is thought to be secreted by the tumor, rather than produced by the host s immune system (Todorov, Cariuk et al. 1996). Injection of PIF into non-tumor-bearing mice causes rapid lean body mass loss and treatment of isolated skeletal muscles with PIF induces protein degradation (Todorov, Cariuk et al. 1996; Todorov, McDevitt et al. 1996). 6

19 Although other factors such as anorexia and immobilization also contribute to cancer cachexia, there is strong evidence that pro-inflammatory cytokines play the main role in the etiology of cancer cachexia. Cytokines can increase metabolism and induce muscle proteolysis, though there is not a consensus as to which cytokine plays the most important role. It is very likely that many pro-inflammatory cytokines and other tumorderived proteins act together to cause cachexia, which may be a reason that targeted treatments against individual cytokines have failed (Boddaert, Gerritsen et al. 2006; Lynch, Schertzer et al. 2007). Skeletal Muscle Skeletal muscle makes up 40% of total body mass and is required for support, stability, and movement of the body. Muscle tissue is arranged in a highly organized, hierarchical manner. Muscle is composed of muscle fibers, which in turn are composed of many myofibrils. These myofibrils are composed of myofilaments, which form sarcomeres, the contractile units of muscle (Figure 1). Sarcomeres are composed of thick and thin filaments. Myosin is the main component of the thick filament and comprises 40-50% of total muscle protein. The thin filament is composed of filamentous actin (Factin). Myosin is a heterohexameric molecule that is composed of two heavy chains and four light chains. Each heavy chain has a globular head region that contains ATPase activity, and binds actin, and a tail region that dimerizes with another heavy chain to form an α-helical coiled-coil. Multiple myosin molecules polymerize to form the thick filaments. The myosin heads bind actin in an ATP-dependent manner and couple ATP hydrolysis with force production, which results in muscle contraction. 7

20 Figure 1. Skeletal muscle architecture. Skeletal muscle is composed of bundles of muscle fibers, which are composed of myofibrils. Myofibrils are composed of myofilaments, which form sarcomeres, the basic contractile unit of skeletal muscle. The main myofilaments are myosin, which form the thick filaments, and actin, which form the thin filaments. Adapted from edcenter.sdsu.edu 8

21 There are eight different myosin heavy chain (MyHC) isoforms. Each isoform has a different contractile velocity due to distinct ATPase activity in the myosin head. Four isoforms are expressed in adult skeletal muscle: type I/β, or slow MyHC and three fast isoforms, IIa, IId/x and IIb. Thus skeletal muscle fibers may be composed of primarily slow-twitch, oxidative fibers such as the soleus, or composed of fast-twitch, more glycolytic fibers like the gastrocnemius. The expression of these MyHC genes is highly regulated (Allen, Sartorius et al. 2001) and certain conditions, such as muscle atrophy, can induce MyHC isoform shifts. Changes in MyHC isoform expression affect muscle function since the ATPase activity in the myosin head confers distinct contractile velocities. Skeletal Muscle Atrophy Skeletal muscle mass is preserved by maintaining a delicate balance between protein synthesis and degradation. Muscle growth (hypertrophy) occurs when protein synthesis exceeds degradation. Conversely, a reduction in muscle mass (atrophy) occurs when protein degradation outweighs synthesis. Several different conditions and diseases are associated with skeletal muscle atrophy: denervation, aging, prolonged bed rest, sepsis, cancer, AIDS, COPD, and chronic kidney disease (Lynch, Schertzer et al. 2007). Muscle is the body s major protein reservoir. Under acute stress, muscle proteins can be mobilized to supply the body with free amino acids that can be used to maintain protein synthesis in vital organs. These amino acids can also provide energy by serving as substrates for gluconeogenesis or by direct oxidation. In certain disease states, free amino acids derived from skeletal muscle serve as precursors for acute phase protein synthesis 9

22 in the liver, which aids in the immune response. Although acute muscle catabolism may be beneficial to the host, chronic muscle catabolism is detrimental and can decrease the host s ability to fight infection, recover from stress and can impair respiratory function, which culminates in increased morbidity and mortality (Ventadour and Attaix 2006). Muscle mass loss in cancer patients is associated with a decreased prognosis and is inversely correlated with survival (Bachmann, Ketterer et al. 2009). Cancer induced skeletal muscle atrophy is largely due to increases in protein degradation. Decreases in muscle protein synthesis may also contribute to decreased muscle mass (Smith and Tisdale 1993), but these data are disputable (Plumb, Fearon et al. 1991). In cancer cachexia, the fast, glycolytic fibers lose more protein than the slower, oxidative fibers (Mitch and Goldberg 1996) and MyHC appears to be specifically decreased compared to other myofibrillar proteins (Acharyya, Ladner et al. 2004). Like most tissues, skeletal muscle contains four main proteolytic pathways that are responsible for the turnover of different proteins in the cell. These pathways include the ubiquitin-proteasome pathway (UPS), the lysosome, the Ca 2+ -dependent calpains, and the caspases (Ventadour and Attaix 2006). Each pathway has been implicated in cancer cachexia to some degree. The Ubiquitin-Proteasome System Most cellular proteins are degraded by the UPS (Lecker, Solomon et al. 1999), which is the main proteolytic pathway implicated in skeletal muscle cancer cachexia. The UPS is responsible for degrading several types of proteins, including those involved in signaling, cell cycle progression, regulation of transcription, proteins that have been misfolded or damaged from oxidative stress and myofibrillar proteins (Kisselev and 10

23 Goldberg 2001). Proteins are tagged for degradation by the addition of a poly-ubiquitin chain, which is recognized by the proteasome. The conjugation of ubiquitin molecules to proteins is a highly regulated process. First, the E1 enzyme activates ubiquitin in an ATPdependent manner. A ubiquitin carrier protein, E2, then accepts this activated ubiquitin and transfers it to an E3 ligase, which confers specificity for a particular protein substrate (Melstrom, Melstrom et al. 2007). Multiple rounds of ubiquitin ligation create a poylubiquitin chain on the substrate, marking it for degradation. There is only one E1 enzyme in eukaryotic cells and its expression is not regulated in catabolic states, yet it has high intrinsic activity and is capable of activating excess amounts of E2 (Wing 2005; Ventadour and Attaix 2006). There are approximately 40 E2s in mammalian cells, and several (E2 14k, E2 20k ) are overexpressed during muscle wasting, implying some form of regulation at the E2 level. The E3 ligases confer specificity for the ubiquitin conjugation cascade and there are thousands of E3s, each of which recognizes specific sequences in a substrate. Atrogin-1/muscle atrophy F-box (Mafbx) and muscle ring finger protein-1 (MuRF-1) are muscle-specific E3 ligases. They were first identified because they are both upregulated in many forms of skeletal muscle atrophy, including atrophy induced by denervation, immobilization and unloading (Bodine, Latres et al. 2001). Expression of these ligases also increases in muscles undergoing atrophy due to starvation, diabetes, uremia, and cancer (Lecker, Jagoe et al. 2004). The insulin-like growth factor-1 (IGF- 1)/insulin/PI3K/Akt pathway and the Forkhead box-containing protein, O subfamily (FoxO) family of transcription factors control expression of these ligases (Stitt, Drujan et al. 2004). Under growth conditions, the Akt pathway is activated, leading to 11

24 phosphorylation of FoxO3 and its nuclear exclusion, rendering it unable to activate transcription of atrogin-1, MuRF-1, and other genes involved in atrophy. Conversely, nutrient deprivation and atrophic conditions decrease Akt activation resulting in increased atrogin transcription and atrophy. Inhibition of FoxO3 prevents atrogin-1 induction in murine muscle after fasting, while constitutively active FoxO3 increases atrogin-1 transcription and causes significant muscle atrophy (Sandri, Sandri et al. 2004). In skeletal muscle, atrogin-1 targets the initiation factor eif3-f and MyoD (Tintignac, Lagirand et al. 2005; Lagirand-Cantaloube, Offner et al. 2008), which contributes to atrophy by decreasing general translation and the transcription of skeletal muscle-specific genes. MuRF-1 is known to bind to the sarcomeric proteins titin, troponin-i, and troponin-t and most recently has been found to bind and ubiquitinate β- MyHC and MyHC-IIa, myosin light chains 1 and 2 and myosin binding protein C in skeletal muscle (Witt, Granzier et al. 2005; Clarke, Drujan et al. 2007; Fielitz, Kim et al. 2007; Cohen, Brault et al. 2009). Overexpression of MuRF-1 causes skeletal muscle atrophy, while its deletion decreases atrophy (Bodine, Latres et al. 2001). Thus increases in atrogin-1 and MuRF-1 during muscle atrophy are likely partially responsible for the decreases in myofibrillar proteins, based upon their known targets. However, overexpression of atrogin-1 alone is not sufficient to induce myotube or muscle atrophy (Sandri, Sandri et al. 2004). Although smooth muscle does not atrophy in response to the same stimuli as skeletal muscle, both atrogin-1 and MuRF-1 are also associated with uterine involution after parturition (Bdolah, Segal et al. 2007). After the E3 ligases have poly-ubiquitinated the appropriate substrates, they are shuttled to the proteasome, where proteolysis occurs. 12

25 The proteasome is composed of a 20S core and two, 19S regulatory particles at each end, which form lid-like structures. The 19S particles regulate the entry of substrates into the proteasome by binding the ubiquitinated substrates and enabling their entry into the proteolytic core. These particles have ATPase activity, which allows them to unfold protein substrates for subsequent degradation by the 20S proteasome. Additionally, the N-termini of many of the α-subunits project into the opening of the cylinder, thereby preventing access of substrates until the proteasome has been appropriately activated. Activation of the proteasome occurs when the base of the 19S associates with the 20S particle, causing rearrangement of the α-subunits, which allows entry and unfolding of the substrate into the proteolytic core (Powell 2006). The core is composed of four stacked rings that enclose a central cavity. The outer two rings are composed of α-subunits, which form a gate and prevent unwarranted access to the proteolytic core. The inner two rings are composed of β-subunits, which contain the proteolytic activity, and contain at least three active sites each that are unique for certain types of bonds (Lecker, Solomon et al. 1999). The β5 subunit contains chymotryptic activity, the β2 subunit tryptic activity, and the β1 subunit caspase-like activity (Powell 2006). The six catalytic β-subunits can be replaced by corresponding immunoforms upon IFN-γ stimulation, which forms the so-called immunoproteasome. Immunoproteasome activity is correlated with the early phases of an antiviral immune response, favors cleavage behind hydrophobic residues, and has an increased substrate degradation rate than the constitutive subunits (Strehl, Textoris-Taube et al. 2008). These changes in the proteasome favor the formation of antigenic peptides for presentation on major histocompatibility antigens during an immune response. Proteasome substrates are 13

26 cleaved multiple times in a processive fashion in order to ensure that digested proteins do not accumulate. Digested peptides released by the proteasome range from 3 to 25 residues and are further hydrolyzed to individual amino acids in the cytosol. Proteasome inhibitors have enabled the discovery of multiple roles of the UPS. Not all sites need to be inhibited to decrease protein breakdown rates. Inhibiting only the chymotrypsin-like site causes a significant decrease in protein breakdown, while inhibition of the trypsin or caspase-like sites has little effect on total proteolytic activity (Kisselev and Goldberg 2001), implying a major role for the chymotrypsin-like site. Accordingly, most inhibitors of the UPS target the chymotrypsin-like site. The most common proteasome inhibitors used in the laboratory are MG-132, lactacystin, and bortezomib. MG-132 is a peptide aldehyde, which is a potent, reversible inhibitor and forms an adduct with the active site. MG-132 is usually the first choice to study proteasome involvement due to its cost and effectiveness (Kisselev and Goldberg 2001). However, MG-132 also inhibits calpains and cathepsins but at 10-fold higher concentrations than is required to inhibit the proteasome. Lactacystin is a non-peptide inhibitor that gets transformed into a β-lactone that acylates the catalytic residue resulting in irreversible inhibition of the proteasome. Lactacystin is the least stable of all proteasome inhibitors, but does not have off target effects like the aldehydes (Kisselev and Goldberg 2001). Bortezomib (PS-341) is a peptide boronate, which are 100-fold more potent inhibitors and are more selective than the aldehydes (Kisselev and Goldberg 2001). Because the proteasome is involved in cell signaling, cell cycle and general protein turnover, inhibition may have toxic effects on cells. Toxicity occurs more in 14

27 rapidly proliferating cells than in non-proliferating cells, thus short-term proteasome inhibition in muscle cells is likely not toxic. The UPS is the main proteolytic pathway implicated in nearly all models of skeletal muscle atrophy including denervation, starvation, and cancer (Lecker, Solomon et al. 1999). The muscles of tumor-bearing rodents have increased ubiquitin expression, increased proteasome subunit expression, and increased ubiquitin-protein conjugates indicating increased activation of the UPS (Temparis, Asensi et al. 1994; Baracos, DeVivo et al. 1995). Inhibition of the lysosomal or Ca 2+ -dependent proteolytic pathways does not attenuate increased muscle proteolysis of tumor bearing rats, while depletion of ATP almost completely suppresses the increased proteolysis associated with cancer (Temparis, Asensi et al. 1994). Skeletal muscles of tumor-bearing rats also have increased ubiquitin conjugating activity (Baracos, DeVivo et al. 1995), which is another indication of increased activity of the UPS. Increased expression of proteasome subunits has also been found in the skeletal muscle of cancer patients (Khal, Hine et al. 2005). The proteasome can degrade purified myosin, actin, troponin, and tropomyosin, but is unable to degrade these myofibrillar proteins when they are in an intact myofibril (Solomon and Goldberg 1996). This suggests that the rate-limiting step in their degradation is their cleavage and dissociation from the sarcomere. How myofibrillar proteins are initially dissociated from the sarcomere is an area of debate. There is evidence, which is discussed below, suggesting that caspases or calpains may be involved in the cleavage of proteins from the sarcomere, freeing them for subsequent degradation by the UPS. Despite the overwhelming evidence supporting the role of the UPS in muscle atrophy, there is evidence that the lysosome also plays a role in this catabolic process. 15

28 The lysosome The lysosome is an evolutionarily conserved degradative compartment present in all mammalian cells. Lysosomes are membrane-bound vesicles that contain a high concentration of acid hydrolases. The lumen of the lysosome is very acidic, which is maintained by a vacuolar-type ATP-proton pump in the membrane. The hydrolases in this acidic lumen include proteases, nucleases, lipases and phosphatases, allowing the lysosome to degrade a variety of molecules. Lysosomes can degrade extracellular components after endocytosis or phagocytosis, but can also degrade soluble cytoplasmic components and are implicated in the turnover of organelles, including mitochondria, peroxisomes and even components of the nucleus (Bechet, Tassa et al. 2005). The acidic ph of the lysosome promotes unfolding of substrates, which increases their susceptibility to proteolysis by the cathepsins, the major lysosomal proteases. Cathepsins B, L, H, and D are widely expressed in many tissues, including muscle. Cathepsins B, H, and L are cysteine proteases, and cathepsin D is an aspartic protease, but all are endopeptidases and cleave at the inner peptide bonds of proteins. Cathepsin D can degrade myofibrillar proteins in vitro and binds and hydrolyzes myosin faster than actin and tropomyosin (Jones, Ogunro et al. 1983). Cathepsins B and L can also hydrolyze myosin. Cathepsin L digests myosin more extensively, and interestingly, hydrolyzes fast MyHC isoforms 4-fold faster than the slow isoforms (Dufour, Ouali et al. 1989). Skeletal muscle contains low concentrations of cathepsins in relation to other tissues with high protein turnover, such as the liver. Immunohistochemical studies have found that cathepsins B and L are distributed in discrete vesicles around the nucleus and 16

29 beneath the sarcolemma, but are also in the sarcoplasm (Stauber and Ong 1981; Bechet, Tassa et al. 2005). Cathepsin L has been recognized as a marker of muscle atrophy, and its expression increases during muscle atrophy induced by sepsis, glucocorticoids, disuse, and cancer (Deval, Mordier et al. 2001). Cathepsin B is also implicated in muscle atrophy as leupeptin, a cathepsin B inhibitor, decreases protein breakdown in denervated rat muscles but to the same extent as in normal muscles (Libby and Goldberg 1978). Cathepsin B and L activities also increase in the muscles of tumor-bearing rodents (Fujita, Tsujinaka et al. 1996). Despite these findings, the role of lysosomes in myofibrillar protein degradation is very controversial. Myofibrillar proteins have been found in lysosomes, though whether they entered lysosomes as intact myofibrils or were already partially digested is unknown (Gerard and Schneider 1979). Lysosomes may therefore be involved in myofibrillar protein degradation at some point, but there is overwhelming evidence that the UPS plays the main role in the degradation of myofibrillar proteins in skeletal muscle (Solomon and Goldberg 1996). Autophagy Because lysosomal hydrolases are within the lysosome and are physically separated from their cytosolic substrates, various mechanisms exist to join these two components together. One such mechanism is autophagy, which literally means to eat oneself. Macroautophagy (hereafter referred to as autophagy) is the main mechanism that cells use to degrade long-lived proteins and organelles (Levine and Kroemer 2008). 17

30 Autophagy is upregulated during starvation, periods of high energy demand, or during cellular stress that involves accumulation of protein aggregates, misfolded proteins or damaged organelles (Levine and Kroemer 2008). Mammalian target of rapamycin (mtor) regulates autophagy and inhibits it under nutrient rich conditions. Upon nutrient deprivation, substrates in the cytoplasm are sequestered inside double-membraned vesicles and then delivered to the lysosome. Vesicle formation is a highly organized process and involves activation of class III PI3K, lipid kinase signaling, and ubiquitin-like conjugation systems mediated by the Atg proteins. There are more than 20 Atg genes, and all are evolutionarily conserved and were originally described in yeast. The products of these genes are involved in vesicle nucleation, elongation, and recycling of autophagosome components (Levine and Kroemer 2008). Several in vivo models exist to examine the roles of Atg genes and to monitor the progression of autophagy in different pathological settings. Autophagy can be detected biochemically by following the formation of phosphatidylethanolamine (PE)-conjugated Atg8 (Kabeya, Mizushima et al. 2000). Upon induction of autophagy, cytosolic microtubule-associated light chain 3 (LC3), the mammalian homologue of Atg8, is conjugated to PE to form LC3-II, which incorporates into the autophagosomal membrane. LC3-II is the only well-characterized protein that is specifically localized to autophagic vacuoles, and is therefore used as a marker for autophagy (Barth, Glick et al. 2010). Transgenic mice expressing a GFP-LC3 fusion protein provide an excellent method for visualizing autophagic vesicle formation. For example, starvation induces the formation of distinct, GFP-positive puncta in skeletal muscle (Mizushima, Yamamoto et 18

31 al. 2004). The fully formed autophagosome containing cytoplasmic constituents must ultimately fuse with a lysosome in order for degradation to occur. The lysosome associated membrane protein, LAMP-2, is required for proper fusion with autophagosomes (Levine and Kroemer 2008). Interestingly, some human myopathies, such as Danon disease, are due to deficiencies in proteins required for fusion, which result in the accumulation of autophagic vacuoles, causing severe myofiber degeneration (Nishino, Fu et al. 2000). Autophagy is implicated in skeletal muscle atrophy due to starvation, oxidative stress, denervation and aging (O'Leary and Hood 2009; Sandri 2010). As is the case with the UPS, the FoxO family of transcription factors activates autophagy in skeletal muscle and overexpression of FoxO3 induces skeletal myocyte atrophy in vitro and in vivo (Zhao, Brault et al. 2007; Sandri 2010). The role of autophagy in muscle homeostasis is perplexing because both induction and inhibition of autophagy induce muscle atrophy and the formation of inclusion bodies, which are implicated in certain myopathies (Mammucari, Milan et al. 2007; Masiero, Agatea et al. 2009). It therefore seems that a delicate balance is required to prevent dysregulation of autophagy and maintain muscle homeostasis. Cancer likely disrupts this balance leading to activation of autophagy and muscle proteolysis. Ca 2+ -dependent calpains Ca 2+ -activated proteolytic enzymes, such as calpains, are found in a variety of tissues. The role of calpains in cancer cachexia is under-studied compared to the other proteolytic pathways, but there is some evidence that they are involved in the initial 19

32 phases of muscle protein breakdown. These enzymes are cysteine proteases and require micro- or millimolar levels of calcium for activation and are therefore known as micro- or milli-calpains. Cells containing calpains also express calpastatin, which is a natural inhibitor of calpain protease activity (Croall and DeMartino 1991). Calpain 1 (µ-calpain), calpain 2 (m-calpain), and calpain 3, a muscle specific isoform, are all highly expressed in skeletal muscle (Kramerova, Kudryashova et al.). Rates of proteolysis significantly increase when Ca 2+ is added to the media of isolated muscles, and this can be decreased by inhibiting Ca 2+ release from the sarcoplasmic reticulum (Baracos, Greenberg et al. 1986). Localization of calpains at the Z-line in skeletal muscle support a potential role in the initial cleavage of sarcomeric proteins during muscle atrophy (Croall and DeMartino 1991). Mice deficient for calpain 3 have decreased rates of muscle atrophy after unloading and decreased rates of growth upon reloading. This study shows that calpain 3 is important for sarcomeric remodeling and that it acts upstream of the UPS and is necessary for the ubiquitination of muscle proteins (Kramerova, Kudryashova et al.). Calpains may also be involved in cancer cachexia as muscles of tumor-bearing rats have decreased calpastatin activity and decreased calpain substrates (Costelli, Tullio et al. 2001). Additionally, calpain inhibitors prevent sepsis-induced muscle protein degradation in rats (Fareed, Evenson et al.) and reduce unloading-induced muscle atrophy by 30% (Tidball and Spencer). Calpains are thought to mainly perform regulatory functions, as opposed to degradative, as they are known to cleave at only a few, specific sites (Croall and DeMartino). This correlates with the theory that activation of calpains initiates cleavage of myofibrillar proteins from the sarcomere, which then allows for their 20

33 ubiquitination and proteasomal degradation (Tidball and Spencer 2002). Although this potential role for calpains in muscle atrophy is intriguing, this dissertation focused on the proteolytic pathways involved in the bulk degradation of proteins, such as the UPS and the lysosome. Caspases Caspases are aspartic proteases that mediate apoptosis. It has been proposed that caspases may also serve other functions, such as cleavage of myofibrillar proteins upon cellular stress. The sarcomere is a highly organized contractile unit is likely disassembled by an unknown mechanism prior to degradation by the UPS. Myofibrillar cleavage by caspases may be an initial step in myofibrillar proteolysis since the UPS cannot degrade intact myofibrils (Solomon and Goldberg 1996). Recombinant caspase-3 cleaves actomyosin complexes in rat muscle lysates and produces a characteristic 14 kd cleaved actin fragment that is subsequently degraded by the UPS (Du, Wang et al. 2004). Additionally, the proteolytic activity of caspases 1, 3, 6, 8 and 9 significantly increases in the muscles of tumor-bearing mice without an increase in apoptosis (Belizario, Lorite et al. 2001). This suggests that caspases play a role in cancer-induced muscle atrophy that is independent of apoptosis. Caspase-3 is also implicated in cleaving certain myofibrillar proteins in cardiac muscle. Caspase-3 can cleave ventricular myosin light chain, which alters MyHC and actin interactions and results in decreased contractility (Moretti, Weig et al. 2002). Another group found that caspase-3 can also cleave sarcomeric α-actin and tropomyosin in vitro but not MyHC, or the troponins (Communal, Sumandea et al. 2002). Despite this 21

34 evidence that caspases can cleave myofibrillar proteins, it remains to be determined if they are involved in the initial cleavage of myofibrillar proteins during muscle atrophy in vivo. The mechanisms of skeletal muscle atrophy and the proteolytic pathways involved have therefore been well studied. It has been established that the UPS plays the main proteolytic role in skeletal muscle atrophy due to cancer, while the lysosome and autophagy play secondary or tertiary roles. Very little is known about the proteolytic pathways induced in atrophic cardiac muscle in humans or rodents with cancer. Because cardiac muscle is also a striated muscle, it is logical to assume that the proteolytic pathways activated in the hearts of animals with cancer are the same as those in skeletal muscle. However, cardiac and skeletal muscle differ in many ways, such as in the relative contributions of the different proteolytic pathways in normal and diseased states. The following section discusses cardiac muscle physiology and the proteolytic pathways implicated in different types of cardiac disease and atrophy. Cardiac Muscle The heart is a striated muscle, and is therefore organized similarly to skeletal muscle. Cardiac muscle is composed of thick and thin filaments that together, form the sarcomere. Elements of the thick filaments include myosin, myosin-binding protein C (MyBP-C), and the myosin light chains (MyLC) 1 and 2. Actin, tropomyosin (Tm), and troponins T, I and C (TnT, TnI, TnC) make up the thin filament (Figure 2). Z-disks at each end of the sarcomere serve to anchor the myofilaments and link them from one sarcomere to the next via the large protein titin. 22

35 Figure 2. Myofibrillar proteins in the cardiac sarcomere. Components of the thin filament (troponin (Tn) T, I and C, tropomyosin (Tm), and actin) and thick filament (MyHC, myosin light chain, and myosin binding-protein C) are shown. Calcium binding to Tn C causes a decrease in the interaction between TnI and actin, allowing Tm to move away from the MyHC binding site on actin. The myosin head then binds actin and converts energy from ATP hydrolysis into a force that moves the head, resulting in contraction. Adapted from The New England Journal of Medicine. Mar 13, 1997; 336 (11). 23

36 Contraction in the heart is regulated by calcium (Ca 2+ ). At rest, TnI binds actin, which changes the conformation of tropomyosin so that it prevents myosin and actin interactions. Upon initiation of contraction, there is an influx of Ca 2+, which binds to TnC and strengthens its interaction with TnI. This releases TnI from actin, causes movement of Tm away from the actin-binding site on MyHC, and allows for myosin-actin crossbridge formation and contraction. Like skeletal muscle, the 220 kd MyHC protein has two main regions; the head, which contains the ATPase activity and actin binding site, and the rod, which forms helical coiled-coil dimers, tetramers and multimeric complexes to form the thick filament. Cardiac muscle expresses two MyHC isoforms, α and β, which are 93% homologous but functionally distinct (McNally, Kraft et al. 1989). α-myhc has higher ATPase activity and actin filament sliding velocity than β (Hoh, McGrath et al. 1978; VanBuren, Harris et al. 1995). β-myhc is the same gene expressed in skeletal muscle and although it has less ATPase activity, it generates more force per unit ATP and is therefore more energy efficient than α-myhc (Holubarsch, Goulette et al. 1985). Addition of small amounts of β-myhc to α-myhc in in vitro preparations can decrease contractile speeds to levels that occur in pure β preparations (Harris, Work et al. 1994), indicating that minor changes in MyHC isoform content in the heart can have profound effects on contractile properties. Numerous pathological stimuli affect cardiac MyHC isoform expression, which ultimately affects contractile function of the heart. Hypertension, diabetes, hypothyroidism, caloric restriction, hypertrophy and cancer all cause an increase in β- MyHC expression in the heart (Gupta 2007; Tian, Nishijima et al. 2010). A small 24

37 increase in β-myhc in the murine heart can cause decreased Ca 2+- activated ATPase activity and systolic function and augments the decline in cardiac function in stressed hearts (Tardiff, Hewett et al. 2000; Krenz and Robbins 2004). In vitro studies have shown that power generation and loaded shortening velocities are less at all loads in cardiomyocytes expressing only β-myhc (Herron, Korte et al. 2001). It is important to note that these studies are all performed in rodent cardiomyocytes, which express mostly α-myhc, hence a small increase in β will have adverse effects. Human hearts are composed of approximately 90% β-myhc, which is not by itself, pathologic. β-myhc expression can still increase in failing human hearts, yet there is debate as to whether this causes a decline in function or is pathologic (Noguchi, Camp et al. 2003). Clinical improvement in patients with dilated cardiomyopathy was correlated with increases in α- MyHC and decreases in β-myhc mrna, and human hearts in failure have no detectable α protein, while normal hearts express approximately 7% α-myhc protein (Miyata, Minobe et al. 2000). This indicates that increased β-myhc expression also contributes to pathology in the human heart (Abraham, Gilbert et al. 2002). Expression of the two MyHC isoforms is developmentally regulated and differs between species. In rodents, α- and β-myhc are detected at embryonic day 8 and β- MyHC expression increases in the ventricles as fetal development continues (Lyons, Schiaffino et al. 1990). Immediately before birth, α-myhc expression increases and becomes the dominant isoform. Adult mouse hearts express >99% α-myhc (Gupta 2007), while humans express approximately 10% (Miyata, Minobe et al. 2000). Thus faster-contracting hearts express more α-myhc and slower contracting hearts express more β, which correlates with the ATPase and actin sliding velocities of these isoforms. 25

38 MyHC expression is controlled both transcriptionally and post-transcriptionally. Transcriptional regulation of cardiac MyHC is complex and incompletely understood. Four groups of transcription factors are implicated in MyHC gene expression: the GATA family, Nkx2.5, serum response factor (SRF) and MEF2 (Gupta 2007). These factors are necessary for expression of several cardiac genes but deletion of GATA and Nkx2.5 does not affect MyHC gene expression, implying the existence of other factors. Deletion of SRF, however, does decrease the expression of both α- and β-myhc transcripts (Balza and Misra 2006). Thyroid hormone is the most potent regulator of MyHC gene transcription. Upon activation of the thyroid hormone receptor, the α-myhc promoter is activated while the β-myhc promoter is repressed. Adrenergic stimulation also regulates MyHC expression as β-adrenergic agonists increase α while antagonists or cardiac denervation causes a shift in favor of β (Gupta 2007). As reviewed in Gupta, this regulation also occurs in humans. Patients with chronic heart failure receiving catecholamines have 40% more α- MyHC expression compared to untreated patients. Healthy, physiologic stimuli such as exercise, IGF-1 and growth hormone induce cardiac hypertrophy and α-myhc mrna, while pathological hypertrophy induced by pressure or volume overload is associated with an induction of β-myhc at the expense of α (Gupta 2007). Thus a switch in MyHC isoforms from α to β is the hallmark of pathology in the heart. Post-transcriptional mechanisms of MyHC isoform regulation are less clear but may involve alternative splicing (Sindhwani, Ismail-Beigi et al. 1994). Mechanical parameters such as contraction, have also been shown to regulate both α and β-myhc mrna levels (Gupta 2007). Additionally, contractile arrest in cardiomyocytes increases 26

39 MyHC protein degradation (Samarel, Spragia et al. 1992), which could be another mechanism for the replacement of one isoform over the other. Cardiac Atrophy Like skeletal muscle, the heart can remodel and hypertrophy or atrophy in a stimulus-specific manner. Cardiac hypertophy can be physiological in the case of exercise or pregnancy, or can be pathological under conditions of volume or pressure overload. Cardiac hypertrophy continues to be an area of intense investigation, while cardiac atrophy is largely overlooked likely due to its low prevalence in the clinic. Atrophy of the heart was first described as a pathological entity in 1649, but was first fully described clinically by Senac in his Treatise on Heart Disease in 1749 (Senac 1749). Senac recognized that cardiac atrophy occurred in people with chronic systemic diseases. Clinical features associated with atrophy of the heart were distant heart sounds, small and narrow pulse, and diminished precordial activity. Studies began to associate atrophy with illnesses in which inadequate nutrition is involved, such as gastrointestinal disease, tuberculosis, typhoid fever and cancer. Autopsies of patients with tuberculosis discovered hearts one-half the size of normal that appeared pale with increased tortuosity of coronary vessels. Hellerstein et al. performed a comprehensive study of 85 cases of cardiac atrophy in humans in the late 1940s (Hellerstein and Santiago-Stevenson 1950). The most common cause of cardiac atrophy in this group was cancer, which accounted for 73% of cases. The criteria for diagnosis was decreased cardiac weight, increased pigmentation of muscle fibers, tortuosity of coronary vessels, increased nuclei to muscle fiber ratio, and 27

40 decreased muscle fiber size. Additionally, the heart was found to lose mass to a greater degree than the body. Histological analysis of the myofibers of these hearts revealed many narrow and short fibers, increased spaces between fibers, and fibers that were twisted. Of note, sarcomeres were intact as both longitudinal and transverse striations were preserved. The valves of the heart did not atrophy as much as the cardiac muscle itself, but atrophy of other organs such as the liver, skeletal muscle, and thyroid did occur in some patients. The authors of this study noted that the majority of patients had decreased physical activity, some were bed-ridden, many were malnourished or experienced anorexia secondary to their disease or had prolonged low grade fever. These factors undoubtedly contributed to cardiac atrophy and the authors state that this was a form of disuse atrophy of the heart due to the decreased work placed on it. Most patients with cardiac atrophy also had decreased blood pressure, slow pulse rate, decreased cardiac output, and decreased voltage of the QRS and T waves on electrocardiogram but normal sinus rhythm. Despite these significant changes in the cardiac musculature, only three out of the 85 patients died with heart failure, implying that the atrophic heart is usually adequate to meet the demands of the diseased individual. Cardiac muscle can also undergo atrophy due to HIV, diabetes, prolonged bed rest, spaceflight, quadriplegia, starvation, and left ventricular assist device placement (Gottdiener, Gross et al. 1978; Kessler, Pina et al. 1986; Rakusan, Heron et al. 1997; Perhonen, Franco et al. 2001; Hu, Klein et al. 2008; Pruznak, Hong-Brown et al. 2008). The extent of cardiac mass loss and mechanisms are similar in all of these cases. Transgenic rats expressing the HIV-1 protein have a 12% decrease in cardiac mass that is not due to decreased protein synthesis, and a decrease in stroke volume (Pruznak, Hong- 28

41 Brown et al. 2008). Mice with streptozotocin-induced diabetes lose approximately 20% of their cardiac mass, which is associated with decreased fractional shortening, thinning of the ventricular walls, increased cardiac protein degradation and decreased MyHC protein in a proteasome-dependent manner (Bilim, Takeishi et al. 2008; Hu, Klein et al. 2008; Zu, Bedja et al. 2010). 12-week bed rest results in a 16% decrease in left ventricular mass, and 10 days of spaceflight causes a 12% decrease (Perhonen, Franco et al. 2001). Quadriplegic and paraplegic patients have a 26% lower left ventricular mass that is associated with reduced cardiac output (Kessler, Pina et al. 1986). Thus in the pathological settings of immune dysfunction, metabolic disease, and prolonged rest, cardiac muscle loss is observed with concomitant decreases in function. A form of cardiac atrophy that is functionally adaptive occurs with decreased load on the heart such as after the placement of a left ventricular assist device (LVAD) into patients with heart failure. Unloading of failing hearts is clinically used to partially reverse hypertrophy, which improves function and increases survival (Bugger, Leippert et al. 2006). Unloading of the heart is achieved in laboratory animals by heterotopic cardiac transplantation. This involves placing a donor heart in the abdomen of the recipient and anastomosing the ascending aorta of the donor with the abdominal aorta of the recipient. This allows perfusion and contraction of the heart without a load (Razeghi, Baskin et al. 2006). Ventricular mass decreases 50% three weeks after transplantation (Rakusan, Heron et al. 1997). This decrease in heart size is due to decreased cardiomyocyte size and increased protein degradation, rather than increased cell death (Razeghi and Taegtmeyer 2006). Unloaded rat hearts have normal fractional shortening and relaxation times, 29

42 implying that function is maintained despite a 40% decrease in myocyte size (Welsh, Dipla et al. 2001). The mechanisms associated with this form of cardiac atrophy are unique because this is a form of disuse atrophy and humoral factors are usually not involved. Autophagy is also not involved in this form of cardiac atrophy as markers of autophagy decrease in patients with LVAD support (Kassiotis, Ballal et al. 2009). This indicates that autophagy is a maladaptive mechanism in the failing heart since it decreases upon improvement of cardiac function. Anorexia and starvation are other well-established causes of cardiac atrophy, which are associated with severe cardiac complications in 80% of patients, and even sudden death (Olivares, Vazquez et al. 2005). Anorexic patients lose approximately 17% of their cardiac mass and have increased QT intervals and decreased cardiac output. Increased QT intervals may be explained by the histological abnormalities of cardiac myocytes that include changes in the orientation and structure of myofibers that are associated with myofibrillar atrophy (Olivares, Vazquez et al. 2005). This decrease in cardiac mass is also associated with left ventricular systolic dysfunction, though most studies do not report a decrease in fractional shortening or ejection fraction (St John Sutton, Plappert et al. 1985; Galetta, Franzoni et al. 2005). It is possible that this type of atrophy is also due to decreased load on the heart as hypotension induced by fasting causes a decrease in afterload, which stimulates the reduction in cardiac mass. Cardiac atrophy associated with starvation is reversible, and cardiac mass increases toward normal upon adequate weight gain (Gottdiener, Gross et al. 1978). 30

43 A proposed mechanism for starvation-induced cardiac atrophy involves insulin deficiency. Insulin signaling promotes cardiac and skeletal muscle growth, and decreases in insulin like growth factor I (IGF-I) contribute to muscle atrophy. Mice with a cardiacspecific deletion of the insulin receptor have a 20-30% smaller heart implying an important role for insulin in regulating cardiac size (Belke, Betuing et al. 2002). Insulin deficiency, which occurs in starvation, can also increase cardiac proteolysis and decrease protein synthesis. It is well accepted that starvation-induced cardiac atrophy is mainly due to decreased protein synthesis of myofibrillar proteins and that this is likely mediated by decreases in insulin (Crie, Sanford et al. 1980; Preedy, Smith et al. 1984; Samarel, Parmacek et al. 1987). The heart also atrophies in humans and rodents with cancer. In the late 1960s, Burch et al. characterized cardiac atrophy in patients with cancer at autopsy and correlated these pathologic features with electrocardiographic abnormalities (Burch, Phillips et al. 1968). Burch coined the phrase cachectic heart because it represents the cardiac pathology and describes the diseases with which it is associated, namely diseases that cause whole body cachexia. This term, however, should not be confused with cardiac cachexia which refers to skeletal muscle atrophy secondary to heart failure. Cachectic hearts were mostly found in patients who died of malignancy. Like Hellerstein, Burch observed that the main features of cachectic hearts were decreased size and mass, a soft, flabby appearance, minimal or no atherosclerosis, and tortuosity of the coronary arteries. Microscopic examination revealed attenuated cardiac muscle fibers with minimal or absent longitudinal striations, and wide, empty spaces between muscle fibers, consistent with a myofiber atrophy phenotype. Thus, in contrast to Hellerstein, 31

44 Burch found disrupted sarcomeres and marked myofiber atrophy perhaps because most of his specimens were from cancer patients as opposed to patients with other chronic diseases as in Hellerstein s report. Electrocardiograms from patients with lung cancer showed a decreased QRS voltage that worsened with time. Atrophic hearts could also be seen on X-ray as a reduction in size of the cardiac silhouette, though this was not always an accurate way of diagnosing cachectic hearts and the diagnosis must be confirmed postmortem. The studies by Hellerstein and Burch provide early insight into cardiac atrophy and its association with malignant disease. Although these studies described the gross pathology, they did not offer mechanistic insight. Rodent models of cancer cachexia were therefore employed to further study the molecular mechanisms and functional consequences of cardiac atrophy. Early studies aimed to determine if cardiac physiology in tumor-bearing rats contributed to the increased whole-body energy expenditure and skeletal muscle wasting observed in these animals (Drott, Ekman et al. 1986). Cardiac function was studied in perfused hearts from tumor-bearing rats and this cardiac muscle had decreased mass, normal to increased left ventricular work, normal contractile function, and higher oxygen consumption. This study established that atrophic cardiac muscle maintains its function but its increased oxygen consumption may contribute to the elevated energy expenditure of the host. Analysis of hearts from mice with a sarcoma revealed reduced amounts of myofibrillar, soluble and collagen proteins with no change in interstitial volume (Sjostrom, Wretling et al. 1987). Atrophic hearts also contained less mitochondria, myofibrils and myosin filaments per fiber. Despite this finding, activity of enzymes required for energy production was not affected by the presence of a tumor, indicating 32

45 that the heart is able to maintain functional capacity better than structural components. Another study involving mice with a sarcoma found that myofibrillar proteins are not selectively decreased but that cardiac protein synthesis and total RNA content are decreased (Drott, Lonnroth et al. 1989). Characterization of changes in myofibrillar proteins revealed no change in MyHC isoform expression but did show a decrease in the ATPase activity of cardiac MyHC purified from tumor-bearing rats and mice. Cardiac function decreases in some models of cancer cachexia (Broussard, Lang et al. 1994; Tian, Nishijima et al. 2010) but is not affected in others or in functional studies using isolated hearts (Heymsfield, Bethel et al. 1978; Drott, Ekman et al. 1986). Interestingly, hearts from tumor-bearing rats have increased sensitivity and reactivity to β-adrenergic agonists (Drott, Waldenstrom et al. 1987). Because this was also seen in hearts from starved rats, it may be a reaction to decreased contractile mass rather than a tumor-specific reaction. The UPS in cardiac muscle Reductions in cardiac mass observed in several disease states that are also associated with decreased skeletal muscle mass implicates mechanisms of proteolysis in skeletal muscle. Indeed, cardiac muscle contains the same proteolytic pathways as skeletal muscle: the UPS, lysosome, and the calpains. The UPS is responsible for majority of protein turnover and is necessary for the regulation of cell signaling, apoptosis, and sarcomere quality control in the heart (Willis and Patterson 2006). Expression of UPS components and proteasome activity is 50% greater in cardiac than skeletal muscle, likely because of the high metabolic demands of the heart (Liu, Miers et 33

46 al. 2000). The proteasome is primarily associated with the cytoskeleton and ER in the cytosol, but is also associated with the plasma membrane and nucleus (Wojcik and DeMartino 2003). The proteasome has the same basic composition in all tissues, but subunit composition does differ between tissues and organs in order to confer specificity and selectivity in proteolytic activity (Willis and Patterson 2006). Gomes et al. found that cardiac muscle contains at least 2 species of 26S proteasomes, which suggests the existence of subpopulation-dependent regulation and function. This study also found that 20S proteasomes in cardiac tissue are more complex and heterogeneous in terms of subunit composition than other cell types (Gomes, Zong et al. 2006). Cardiac muscle also contains all three inducible immunoforms of the β-subunits, which indicates that the heart can be immunoreactive. The variety of proteasome components implies diverse functions of the UPS in the heart. As such, the UPS regulates membrane channels and receptors such as the voltage-gated Na + and K + channels, connexin 43, and the β2-adrenergic receptor, implying an important role in regulating conduction and contractility (Willis and Patterson 2006). The UPS also regulates important transcription factors and signal transduction pathways in the heart. Calcineurin, a phosphatase that regulates NFAT and promotes pathological hypertrophic growth, is degraded by the UPS, as are the proapoptotic factors MDM2, Bax, caspase-3 and p53 (Li, Kedar et al. 2004). Given these important substrates of the UPS in the heart, the UPS is implicated in many cardiac diseases such as ischemia-reperfusion injury, hypertrophy, heart failure, cardiomyopathies and chemotherapy-induced cardiotoxicity (Willis and Patterson 2006). 34

47 Atrogin-1 and MuRF-1, the muscle specific E3 ubiquitin ligases, also play important roles in the heart but their roles in cardiac atrophy are less clear than in skeletal muscle. As in skeletal muscle, the FoxO transcription factors, which lie downstream of Akt, regulate these ligases and transfection of cardiomyocytes with FoxO3a causes increased atrogin-1 and MuRF-1 expression and cardiomyocyte atrophy in vitro and in vivo (Skurk, Izumiya et al. 2005). Roles for atrogin-1 in decreasing cardiomyoycte size are strongly supported in the literature. The known substrates of atrogin-1 in the heart are calcineurin, and α-actinin-2, which correlates with its localization to the Z-disc of cardiomyocytes (Li, Kedar et al. 2004). Atrogin-1 inhibits calcineurin-dependent cardiomyocyte hypertrophy in vitro and pressure overload-induced hypertophy in vivo. Atrogin-1 overexpression does not affect heart weight or cardiac function, implying that it alone, cannot induce cardiac atrophy (Li, Kedar et al. 2004). Subsequent studies by the same group provided evidence that atrogin-1 can also prevent Akt-dependent cardiac hypertrophy by regulating the ubiquitination of FoxO, which affects their transcriptional activities (Li, Willis et al. 2007). This study also reported that atrogin-1 null mice had more physiological hypertrophy after running than WT, which implies that atrogin-1 is involved in preventing both physiological and pathological cardiac hypertrophy. MuRF-1 targets cardiac troponin I (Li, Kedar et al.) and titin (Li, Kedar et al. 2004). MuRF-1 has also been suggested to regulate the transcription of α- and β-myhc (Mearini, Gedicke et al. 2009). There are three closely related MuRF proteins (MuRF-1, MuRF-2, and MuRF-3) in heart and skeletal muscle, and all share the RING finger domain, which confers ubiquitin ligase activity. MuRF-1 and MuRF-2 are functionally 35

48 redundant as they target many of the same proteins (Witt, Granzier et al. 2005). Mice lacking either MuRF-1 or MuRF-2 lacked phenotypes but a deletion in both genes caused mice to develop cardiac and skeletal muscle hypertrophy (Witt, Witt et al. 2008). Like atrogin-1, MuRF-1 is implicated in decreasing cardiac myocyte hypertrophy (Li, Kedar et al.). Mice lacking MuRF-1 were unable to fully reverse thoracic aortic constriction (Schena, Kurimoto et al.)-induced hypertrophy (Willis, Rojas et al. 2009). However, cardiac muscle still atrophied without MuRF-1, indicating that other factors are involved in causing atrophy. MuRF-1 was also shown to mediate dexamethasone-induced cardiac atrophy in mice (Willis, Rojas et al. 2009), which is considered a primary form of atrophy. Mice with a deletion of MuRF-1 and -3 develop cardiac hypertrophy and accumulate β-myhc and its degradation products. MuRF-1 and -3 were therefore implicated in cardiac β-myhc turnover in vivo (Fielitz, Kim et al. 2007). MuRF-1 knock-out mice have normal cardiac size as do young MuRF-1 transgenic mice (Mearini, Gedicke et al. 2009). Adult MuRF-1 transgenic mice, however, exhibit progressive thinning of the left ventricular wall (without a change in total cardiac mass), decreased cardiac function, and increased progression to heart failure following thoracic aortic banding (Schena, Kurimoto et al. 2004; Willis, Schisler et al. 2009). Interestingly, increased cardiac MuRF-1 expression is not associated with increased ubiquitination and degradation of sarcomeric proteins, which implies that MuRF-1 s ubiquitin ligase activity may not be the mechanism by which it affects cardiac size and function (Willis, Schisler et al. 2009). 36

49 Both atrogin and MuRF-1 expression increase in the hearts of rodents with chronic heart failure (Adams, Linke et al. 2007), TAC-induced hypertrophy (Willis, Rojas et al. 2009), and those undergoing hypoxia (Razeghi, Baskin et al. 2006). An increase in these ligases during heart failure is therefore likely an attempt to restrict hypertrophic growth in order to preserve cardiac function. Interestingly, doxorubicin, a chemotherapeutic agent known to cause cardiotoxicity, also causes increased expression of atrogin-1 in cardiac myocytes in vitro and in vivo (Yamamoto, Hoshino et al. 2008). Beneficial cardiac atrophy however, as occurs with LVAD placement during heart failure, is associated with decreased atrogin-1 and MuRF-1 transcription in the rat (Razeghi and Taegtmeyer 2006), but not in humans (Willis, Rojas et al. 2009). This decrease likely represents an attempt at reversing the high levels that were induced upon cardiac hypertrophy, which necessitated LVAD placement in the first place. Other components of the UPS are also upregulated in cardiac muscle in pathological settings. The hearts of diabetic rats have increased ubiquitin mrna expression (Liu, Miers et al. 2000). Hyperglycemia also causes an increase in cardiac expression of the 11S-activator, which is a component of the immunoproteasome and activates the 20S proteasome (Powell, Samuel et al. 2008). The unloaded, atrophic rat heart expresses higher transcripts of ubiquitin B, and the ubiquitin conjugating enzyme UbcH2 (Razeghi, Baskin et al. 2006). Additionally, doxorubicin stimulates UPS activity in cardiac myocytes and promotes the degradation of ubiqutinated proteins (Liu, Zheng et al. 2008). Paradoxically, UPS activity and components also increase during cardiac hypertrophy, which is unlike what occurs in skeletal muscle. Patients with dilated 37

50 cardiomyopathy exhibit increased ubiquitinated proteins, and E1 and E2 enzyme expression in the myocardium (Weekes, Morrison et al. 2003). Ubiquitination of cytoskeletal proteins near the intercalated disk is also increased in pressure-overloaded feline hearts (Balasubramanian, Mani et al. 2006). Because this occurs in the first hours of pressure-overload, this may represent active protein turnover in order to favor hypertrophic growth. Thus increased ubiquitination of cardiac proteins in the setting of cardiac hypertrophy may be a maladaptive mechanism to cope with the rapid growth. Accordingly, inhibition of the UPS prevents cardiomyocyte hypertrophy and its associated markers in vitro, and pressure-overload hypertrophy in vivo (Meiners, Dreger et al. 2008). The lysosome and autophagy in cardiac muscle Lysosomes play the same role in healthy cardiac muscle as they do in skeletal muscle: they are involved in the turnover of membrane and cytoplasmic proteins and organelles. However, the autophagosome-lysosome pathway is more active in cardiac than skeletal muscle. Cardiac muscle has greater functional demands than skeletal muscle, and therefore has a higher metabolic rate. Cardiac muscle consumes nearly 5 times the amount of oxygen as pale skeletal muscle and has higher protein turnover rates (Hansen-Smith, Maksud et al. 1977; Earl, Laurent et al. 1978). Accordingly, protein degradation by the lysosome is greater in cardiac than skeletal muscle (Wing, Chiang et al. 1991). Cardiac muscle also stains more heavily than skeletal muscle for active cathepsin B (Stauber and Ong 1981) suggesting a more active role in protein degradation. 38

51 The lysosome is responsible for approximately 25% of protein degradation in the heart (Wildenthal, Wakeland et al. 1980). Like skeletal muscle, there is controversy surrounding its role in myofibrillar degradation. While inhibiting the lysosome in vitro decreases general protein degradation, it does not affect the degradation of myosin or other myofibrillar proteins (Wildenthal, Wakeland et al. 1980; Wildenthal and Wakeland 1985). However, cathepsin D can hydrolyze purified cardiac myosin and actin (Ogunro, Lanman et al. 1979) and cardiac proteins with a sequence that targets them to the lysosome are depleted in the myofibrillar fraction upon starvation (Wing, Chiang et al. 1991). There is also evidence that cathepsins B and L are involved in myofibrillar protein degradation in a model of myocardial infarction (Tsuchida, Aihara et al. 1986). It has been suggested that myofibrillar proteins require an initial cleavage from the sarcomere and partial degradation before they can undergo lysosomal proteolysis. Myofibrillar proteins have not been found in autophagic vacuoles by electron microscopy (Pfeifer and Strauss 1981), but this does not exclude the possibility that they are degraded in this compartment because it is rare to be able to identify digested materials inside the lysosome. Like skeletal muscle, FoxO transcription factors also regulate autophagy in the heart. FoxO factors bind the promoters of many genes that induce autophagy, including Atg8 (LC3), Gabarap1, Atg12, and beclin (Ferdous, Battiprolu et al. 2010). FoxO3 plays the main role as its overexpression can decrease cardiomyocyte size and induce autophagy genes, while cardiomyocytes expressing dominant negative FoxO do not decrease in size upon starvation (Sengupta, Molkentin et al. 2009). 39

52 The roles of autophagy in the heart have been determined using transgenic mice. Cardiac specific reduction of Atg5, a protein required for autophagy, causes contractile dysfunction, hypertrophy, and heart failure (Nakai, Yamaguchi et al. 2007). These hearts have increased levels of ubiquitinated proteins and UPS activity, increased protein synthesis, disorganized sarcomeres, and increased apoptosis. Therefore, autophagy plays a beneficial role in the heart by maintaining protein turnover, sarcomere integrity, and overall cardiac structure and function. Autophagy is also involved in pathologic responses to stimuli that stress the heart. Autophagy is increased in the heart during ischemia, heart failure, pressure overload, doxorubicin treatment, and starvation (Shimomura, Terasaki et al. 2001; Yan, Vatner et al. 2005; Zhu, Tannous et al. 2007; Takemura, Kanamori et al. 2009; Kobayashi, Volden et al. 2010). Like most other tissues, autophagy in the heart is upregulated during nutrient deprivation. As little as 12 hours of starvation causes an increase in cardiac autophagy while inhibition of autophagy during starvation decreases cardiac function (Takemura, Kanamori et al. 2009). Studies that have inhibited cardiac autophagy concluded that it is protective in the heart, leading to the hypothesis that increased autophagy during cardiac disease, such as heart failure, is also beneficial. This, however, is not the case. Mice with pressure overload-induced heart failure have increased cardiac autophagy that contributes to the pathologic phenotype. Genetically disrupting autophagy by decreasing beclin 1 in these mice decreases pathological remodeling of the heart suggesting that autophagy is detrimental during hemodynamic stress (Zhu, Tannous et al. 2007). Conversely, mice lacking Atg5, a protein necessary for autophagy, have more severe cardiac dysfunction after pressure overload (Nakai, Yamaguchi et al. 2007). It therefore seems that there is a 40

53 fine line between under- and over-activation of autophagy and a change from baseline in either direction can have severe consequences for cardiac function. Chronic cardiac stress causes cardiac myocytes to remodel their cellular architecture and elongate and hypertrophy to cope with an increased workload. Autophagy is upregulated in these conditions, potentially in order to provide energy substrates for growth but also to clear aggregated proteins. The UPS can become overwhelmed during pressure overload induced cardiac hypertrophy, and cannot process the increased flux of damaged or misfolded proteins (Tannous, Zhu et al. 2008). A biopsy from a patient with cardiomyopathy and early-stage heart failure showed ubiquitin positive inclusions, loss of myofibrils, and autophagic vacuoles containing mitochondria and tubulofilamentous inclusions, which may represent aggregated myofibrillar proteins (Fidzianska, Bilinska et al. 2010). It has been proposed that accumulation of aggregated proteins in the heart triggers the activation of autophagy in order to clear the increased flux of damaged proteins. In fact, protein aggregation is sufficient to induce autophagy in cardiomyocytes in vitro, while inhibition of autophagy causes in increase in UPS activity (Tannous, Zhu et al. 2008). This suggests that autophagy and the UPS function in parallel in order to maintain clearance of aggregated proteins during cardiac stress. Calpains in the heart Although the UPS and the lysosomal-autophagy pathways are the major proteolytic pathways in the heart, the calpains are also involved in maintaining cardiac homeostasis. Cardiac muscle contains calpains 1 and 2, and they have mostly been studied as mediators of post-ischemic myocardial injury (Sandmann, Prenzel et al. 2002). 41

54 Overexpression of calpain 1 in cardiomyocytes causes increased protein degradation and accumulation of ubiquitinated proteins, suggesting that calpain lies upstream of the UPS (Galvez, Diwan et al. 2007). Calpain overexpression is lethal, but inducible transgenic mice rapidly develop heart failure, accumulate ubiquitinated proteins, and have increased UPS activity in the heart. Conversely, overexpression of calpastatin, the natural calpain inhibitor, caused a dilated cardiomyopathy that involved sarcomeric disruption, protein aggregates, and the presence of autophagic vacuoles (Galvez, Diwan et al. 2007). These results suggest that calpains are necessary for normal protein ubiquitination and their degradation by the UPS. Thus, it appears that, like skeletal muscle, the calpains also regulate protein turnover upstream of the UPS in the heart. Human unloaded failing hearts express higher calpain 2 transcript levels and unloaded rat hearts have increased calpain activity and higher levels of a calpain-specific breakdown product (Razeghi, Volpini et al. 2007). Other proteolytic pathways must be involved, however, because transgenic overexpression of calpastatin, an endogenous calpain inhibitor, in the same study failed to attenuate unloading-induced atrophy. The hearts of tumor-bearing rats have decreased levels of calpastatin, which implies increased calpain activation (Costelli, Tullio et al. 2001). Despite this evidence, it is not known if calpains initiate myofibrillar protein cleavage and ubiquitination during cardiac atrophy due to cancer. 42

55 Conclusions and Research Aims Cancer cachexia is a significant clinical problem as it contributes to 30% of cancer deaths. Cachexia has been heavily researched in both humans and animals and the basic causes and mechanisms of skeletal muscle atrophy have been determined. Despite this, there is no treatment for cachexia and most clinical trials have failed to prevent muscle mass loss. Research on cachexia has primarily focused on skeletal muscle mass loss, and the mechanisms mediating this loss have been well studied. The heart is also a muscle, but relatively little attention has been paid to cardiac muscle atrophy in cancer patients. Cardiac muscle atrophy may have been ignored because doctors and patients cannot visualize a smaller heart, unlike skeletal muscle where atrophy and mass loss is obvious. Additionally, most oncologists are unaware of cardiac atrophy in cancer patients, likely because it has not been well studied or described. The purpose of this project was therefore to determine the extent, mechanisms, and consequences of cardiac atrophy in males and females. Elucidation of the mechanisms behind this pathology will illuminate the need for clinical consideration of cardiac atrophy during the care of cancer patients, particularly in those with pre-existing heart conditions. The extent, mechanisms and functional consequences of cancer-induced cardiac atrophy in a mouse model are discussed in Chapter 2. The pathological characteristics of cardiac atrophy are described in detail as are the changes in myofibrillar proteins and the proteolytic pathways that likely mediate these changes. Additionally, a sex difference in cardiac mass loss and the mechanism behind it is also described. It has been previously reported that MyHC is specifically decreased in cachectic skeletal muscle (Acharyya, Ladner et al. 2004), which has implications for the 43

56 mechanisms of muscle atrophy. Accurate quantification of sarcomeric myosin is essential in order to determine the components of the sarcomere that are being degraded during atrophy. However, myosin quantification is difficult due to its low solubility in traditional low-salt lysis buffers and due to its high abundance in muscle. Many studies on muscle atrophy lyse muscle in low-salt in order to quantify MyHC, but this provides inaccurate results as it artificially decreases the levels of MyHC in the soluble portion of the lysate. The solubility of MyHC in different lysis buffers and the most accurate method for determining MyHC levels in atrophic muscle is discussed in Chapter 3. In order to study the mechanisms involved in cardiac myocyte atrophy in more detail, an in vitro model using neonatal rat ventricular myocytes was established. Similar to studies in skeletal muscle, the pro-inflammatory cytokine interferon-γ was used to mimic an inflammatory response to a tumor. The mechanisms of interferon-γ induced cardiac myocyte atrophy and the associated changes in MyHC levels are discussed in Chapter 4. The contribution of each proteolytic pathway in total protein turnover and MyHC degradation is also determined. Diet has a profound effect on several diseases, including cancer. A conventional soy-based mouse diet contains high amounts of phytoestrogens, which can have estrogenic effects. Because cancer cachexia exhibits a sexual dimorphism that is estrogen dependent, the effects of dietary phytoestrogens on cancer cachexia were investigated in Chapter 5. The differences in body weight, skeletal and cardiac muscle mass loss, and survival in male and female mice on soy, casein, and phytoestrogen-supplemented diets are discussed. The effect of diet on cardiac MyHC and autophagy gene expression in tumor-bearing mice is also determined. 44

57 Collectively, this body of work contributes to the cachexia field by describing a previously unappreciated aspect of cancer cachexia: atrophy of the heart. Herein, we describe the gross and molecular pathologies associated with cardiac atrophy and the sex differences therein. The proteolytic pathways involved in cardiac atrophy in vivo and those involved in the specific degradation of MyHC in vitro are also described. Importantly, we find that the proteolytic pathway upregulated in cardiac atrophy is distinct from that in skeletal muscle. Together, the results of these studies should not only highlight the mechanistic differences between cardiac and skeletal muscle atrophy, but will hopefully create awareness in the oncology field that cardiac atrophy, and the sex differences therein, should be considered in the care of cancer patients. 45

58 CHAPTER II Cancer causes cardiac atrophy and autophagy in a sexually dimorphic manner Introduction Cachexia is a severe form of skeletal muscle and adipose tissue wasting often associated with diseases such as cancer, sepsis, and AIDS. Cachexia affects approximately one-half of cancer patients and causes nearly one-third of cancer deaths (Tisdale 2002). Weight loss is due to complex alterations in carbohydrate, lipid, and protein metabolism (De Blaauw, Deutz et al. 1997). The consequences of these metabolic changes include anemia, insulin resistance, production of acute phase proteins, and a negative nitrogen balance, which cannot be reversed with nutritional supplementation. These pathologic perturbations result in a significant loss of muscle protein, leading to pronounced muscle weakness and fatigue, increased sensitivity to infections, decreased responsiveness to both chemotherapy and radiation treatment and can ultimately lead to cardiac or respiratory failure (Tisdale 2002). There is a significant sexual dimorphism in muscle mass loss and survival in cancer patients. Multiple studies have found that male cancer patients lose more body weight and muscle mass than females and have shorter overall survival (Hendifar, Yang et al. 2009; Baracos, Reiman et al.). Since cachexia increases mortality, it is likely that sex hormones are mediating these differences because post-menopausal women lose their survival advantage, and estrogen therapy decreases colon cancer mortality (al-azzawi and Wahab 2002; Koo, Jalaludin et al. 2008). Humoral factors secreted from or induced by the tumor are responsible for initiating skeletal muscle and body mass loss. Tumors induce a host immune response 46

59 resulting in increased serum levels of pro-inflammatory cytokines, which cause skeletal muscle protein loss and cachexia in vitro and in vivo (Tisdale 1997; Acharyya, Ladner et al. 2004). Muscle mass loss can be due to increased protein degradation, decreased protein synthesis, or both (Smith and Tisdale 1993), but cancer cachexia is primarily due to increased proteolysis (Temparis, Asensi et al. 1994). Cardiac and skeletal muscles utilize three major proteolytic pathways: the lysosome, Ca 2+ -dependent calpains, and the ubiquitin-proteasome system (UPS). The UPS is responsible for the bulk breakdown of long-lived proteins and plays a major role in skeletal muscle protein degradation due to cancer (Lecker, Solomon et al. 1999). Components of the UPS are upregulated in the muscles of cancer patients and tumor-bearing rodents and inhibition of the UPS, but not the other proteolytic pathways, suppresses tumor-induced muscle proteolysis in vitro (Baracos, DeVivo et al. 1995; Khal, Hine et al. 2005). Cardiac muscle is labile and can undergo atrophy due to anorexia, prolonged bed rest, left ventricular assist device placement, and HIV (Gottdiener, Gross et al. 1978; Hill and Olson 2008; Pruznak, Hong-Brown et al. 2008). Given the high prevalence and mortality rate of cachexia and the fact that the heart is a striated muscle like skeletal muscle, it is surprising that relatively little attention has been paid to cardiac muscle atrophy in cancer patients. In 1968, Burch et al. observed that cancer patients had smaller hearts and decreased amplitude and duration of the QRS complex, implying functional defects (Burch, Phillips et al. 1968). Cardiac atrophy in tumor-bearing rodents has also been observed (Tessitore, Costelli et al. 1993), but the extent of cardiac muscle atrophy, functional implications, biochemical mechanisms and sex differences have never been fully characterized in any tumor model. 47

60 We have established a murine model of cancer-induced cardiac atrophy. Here we report that cardiac mass decreases rapidly during the course of tumor progression and there are multiple, significant differences in the disease phenotype between males and females. Males lose more body weight, skeletal and cardiac muscle than females and have a worse phenotype in all cardiac parameters we studied. We also show that cardiac atrophy is due to a decrease in all myofibrillar proteins, as opposed to myosin heavy chain (MyHC) specifically as reported in skeletal muscle cachexia (Acharyya, Ladner et al. 2004). Most importantly, we provide data implicating autophagy as the main proteolytic pathway involved. To our knowledge, this report provides the first insight into this previously unappreciated aspect of cancer cachexia in both sexes and shows that the mechanisms of cardiac muscle atrophy are distinct from those in skeletal muscle. Materials and methods Animals CD2F1 (Balb/c X DBA2) is the mouse strain used in all cachexia studies. Colon- 26 adenocarcinoma (C-26) cells were grown in RPMI media supplemented with 10% fetal bovine serum, 2% L-Glutamine, and 1% penicillin-streptomycin. Cells were trypsinized, washed in PBS, pelleted and resuspended in PBS at a concentration of 5 x 10 6 cells/ml. 100 µl of this (5 x 10 5 cells) was injected subcutaneously into the right flank of 8-week old male and female mice. Mice were monitored and weighed daily after 7 days and were anesthetized, weighed, and cervically dislocated at the indicated time points. Tissues were rapidly excised, weighed, flash-frozen in liquid nitrogen and stored at -80 C for further analysis. Serum from control and tumor-bearing animals was 48

61 analyzed with a cytokine multiplex panel (Milliplex TM, Millipore). All animal studies were reviewed and approved by the Institutional Animal Care and Use Committee at the University of Colorado at Boulder. Drug treatments Fulvestrant (ICI 182,780; Sigma) was dissolved in 70% ethanol and Cremaphor EL. Female mice were injected subcutaneously with 5 mg of ICI once weekly for 4 weeks starting the day of tumor cell inoculation. Bortezomib (Velcade ) was kindly provided by Millennium Pharmaceuticals. Male mice were injected with 0.8 mg/kg every other day for 11 days. Treatment started 10 days after tumor-cell inoculation and continued until sacrifice at 21 days. A known side effect of Bortezomib is hypothermia. The mouse cages were therefore kept on a heating pad throughout the study. Echocardiography Echocardiography was performed on male and female mice 27 days post-tumor cell inoculum. The mice were anesthetized with isofluorane, positioned supine and M- mode recordings were taken and analyzed as previously described (Stauffer, Konhilas et al. 2006). Animals with heart rate at > 500 beats per minute (b.p.m.) were included in this study. Heart rates were not significantly different between the two study groups. Immunohistochemistry and Histology Hearts from 27-day tumor bearing mice were excised, rinsed in ice-cold PBS and stored in 10% buffered formalin for 24 hours. The tissue was then transferred to 70% 49

62 EtOH and kept at 4 C until further processing by Premier Laboratory (Longmont, CO). Paraffin embedded tissues were cut transversally at three locations: 1 mm from the apex, the middle of the heart, and 1 mm from the base. These segments were each sectioned at 4 microns and adhered on slides. Slides were de-paraffinized, soaked in PBS, and stained with Texas red-x conjugated wheat germ agglutinin (1:50) or anti-cathepsin D (Zhu, Tannous et al. 2007) in 2.5% goat serum overnight at 4 C. At least 5 fields per heart section were imaged and analyzed with a Nikon Eclipse E800 microscope. At least 150 myocytes were analyzed per heart (n = 3 hearts per group) using imagej software. For fibrosis analysis, slides were stained with picrosirius red and analyzed with polarized light microscopy (Zeiss Universal Microscope). Collagen fibers show bright yellow/orange birefringence under polarized light, while ordered myofibrils appear green and show only slight birefringence. Non-uniform myocytes do not exhibit birefringence, which indicates the presence of myocellular disarray. Fibrosis/collagen content was determined by digitizing five random images per slide and quantifying the total area of collagen per image with imagej. Five images each from 4 animals of each group were analyzed and collagen content was expressed as a percent of total image area. Western blots For MyHC analysis, left ventricles were homogenized in high salt myosin extraction buffer (Butler-Browne and Whalen 1984). Lysates were spun at 12,000 rpm for 15 min at 4 C. The supernatant was removed, and protein concentration determined (Biorad DC protein assay). 0.5 µg of protein per sample were run on 7.5% SDS-PAGE gels and transferred to polyvinylidene difluoride (PVDF) membranes. Membranes were blocked in 5% non-fat dry milk for 30 minutes followed by primary antibody (hybridoma 50

63 supernatant 1:5) for two hours at room temperature and goat-anti-mouse secondary antibody 1:10,000 for 1 hour. MyHC antibodies were: MF20 (Developmental Studies Hybridoma Bank), anti-α-myhc (BA-G5 hybridoma, ATCC) and anti-β-myhc (VP- M667, Vector Laboratories). For caspase-9 and ubiquitin blots, frozen left ventricles were homogenized in lysis buffer containing 50 mm tris, 150 mm NaCl, 50 mm NaF, 1 mm EDTA, 0.5% Triton X- 100, ph 7.5 and complete protease inhibitor cocktail (Roche). For LC3-II analysis, tissue was homogenized in 30 mm HEPES, 20 mm KCl, 2 mm NaF, 5 mm EDTA, 1% Triton- X100 and complete protease inhibitor cocktail (Roche). SDS-PAGE and blotting protocols were similar to those above except µg of protein was analyzed per sample and 4-20% gradient gels (Pierce) were used for LC3 analysis. Antibodies were: Caspase-9, LC3b, and GAPDH antibodies were obtained from Cell Signaling Technology (Beverly, MA) and used at 1:1000 concentration. Anti-ubiquitin antibody (Santa Cruz P4D1) was used at 1:500. Anti-sarcomeric α-actin (Sigma clone 5C5) was used at 1:1000. Immunoreactivity was visualized using a Western Lighting chemiluminescence detection system (Perkin Elmer) and quantified using ImageJ. Myofibrillar gels Cardiac myofibrils were purified as previously described (Cohen, Brault et al. 2009). Gels were loaded with the same percent of final heart weight (0.005%). The resolving gel was composed of 15% acrylamide (124:1 acrylamide: bis), 9.79% glycerol, 0.73M Tris ph 9.3, and 0.1% SDS. The stacking gel was composed of 2.95% acrylamide (5.7:1), 0.13M Tris ph 6.8, 10% glycerol, and 0.1% SDS. Gels were run at 16 ma for six 51

64 hours at 4 C and stained with colloidal Coomassie Blue (ProtoBlue Safe, National Diagnostics). Gels were scanned with a LI-COR Odyssey and quantified using ImageJ. RNA analysis Total RNA was purified using TRI Reagent (Molecular Research Center, Inc.) according to the manufacturer s protocol. cdna was synthesized with Superscript II reverse transcriptase (Invitrogen) and random hexamer primers. Gene expression was determined by quantitative real-time reverse transcription PCR (qrt-pcr) using SYBR Green dye with gene specific primer sets and an Applied Biosystems 7500 Real-Time PCR system. Primer sequences are listed in Appendix I. Caspase-3 and proteasome activity assays Caspase-3 activity in cardiac lysates was determined by monitoring the rate of cleavage of a fluorogenic caspase-3 specific substrate (Acetyl-AspGluValAsp-AMC; Calbiochem) as previously described (Stauffer, Konhilas et al. 2006). Proteasome activity assays were performed as described (Powell, Davies et al. 2007) with Suc-LLVY-AMC (Boston Biochem) as the substrate. Cleavage data was obtained over one hour and activity was determined by calculating the slope of the linear portion of the graph. Ubiquitin conjugation assay Briefly, left ventricular tissue homogenate (1 mg/ml) was incubated with 2 µg myc-ub, 2 mm Energy Regeneration Solution, 50 µm MG-132, and 1 µm ubiquitin aldehyde (all Boston Biochem) at 37 C for 30 minutes. The reaction was quenched with 52

65 SDS-PAGE sample buffer. Ubiquitin-protein conjugates were detected by an anti-myc antibody (Cell Signaling). Electron Microscopy Electron microscopy was performed as described (Zhu, Tannous et al. 2007). Mice were anesthetized with 400 ul of 2.5% Avertin. Hearts were retrograde perfused with 2% gluteraldehyde in 0.1 M Na-cacodylate buffer (ph 7.4). The heart was removed and the atria and right ventricle were removed and discarded. The left ventricle and interventricular septum were cut into 1 mm chunks, and left to fix overnight in 2% gluteraldehyde/0.1 M cacodylate buffer. All the following steps require that the tissue be rotating. Tissue was then washed (3 x 30 min) in 0.1 M cacodylate buffer. Tissue was post-fixed in 2% osmium tetroxide in 0.1 M Na-cacodylate buffer for one hour, rinsed with water (2 x 10 min), and post-fixed again with 1% aqueous uranyl acetate for one hour at room temperature. After a brief rinse in water, tissue was dehydrated using an ascending series of ethanol washes (50%, 70%, 90%, 100% ethanol, one hour each). Tissue was then rinsed in propylene oxide (2 x 10 min) and then transferred to a 1:1 mixture of propylene oxide: Epon-Araldite epoxy resin for 2 hours. The tissue was then transferred to pure Epon resin and incubated overnight at room temperature. The tissue was transferred to fresh Epon with accelerator (DMP-30) and incubated for 5 hours before placing it into molds and then polymerized at 60 C for 48 hours. Sections were cut with a Diatome Ultra diamond knife, post-stained for 6 min in 2% uranyl acetate in 70% methanol, 30% water, rinsed in methanol/water, then stained 4 min in Reynold's lead citrate. Sections were viewed using Philips CM100 electron microscope (FEI, Inc, 53

66 Hillsboro, OR). Data and statistical analysis Data are presented as mean ± SEM. Differences between groups were evaluated for statistical significance using Student s t-test. P values less than 0.05 were considered significant. Results Cardiac atrophy caused by colon-26 adenocarcinoma is more pronounced in males than females. In order to define cardiac atrophy due to cancer and to determine whether there are sex differences, male and female mice were inoculated subcutaneously with colon-26 adenocarcinoma (C-26) cells, which cause a well-characterized cancer cachexia, resulting in rapid and severe skeletal muscle atrophy that is independent of anorexia (Tanaka, Eda et al. 1990). Male mice lost significantly more body weight than females at all time points studied (Figure 3A). Both male and female mice lost significant skeletal muscle mass at 15 days, and continued to lose muscle mass along the course of disease. The rate and extent of muscle mass loss was significantly greater in males than females (Figure 3B). We measured cardiac mass at 15, 21, and 27 days post-tumor cell inoculum. Male mice lost 8.3% of their cardiac mass at 15 days, which progressed to 21.8% at 27 days (Figure 3C). Interestingly, female mice initially gained cardiac mass and lost only 9.8% by day 27. Males lost significantly more cardiac mass than females at days 15 and

67 Figure 3. Colon-26 adenocarcinoma causes cardiac muscle mass loss and body weight loss in a sexually dimorphic manner. (A) Percent body weight loss in male and female tumor-bearing mice. Body weight loss was calculated as the final body weight minus the tumor weight and was compared to the body weight of age-matched control mice. (B) Percent skeletal muscle mass (gastrocnemius) loss in male and female tumor-bearing mice. (C) Percent cardiac mass loss in male and female tumor-bearing mice. H&E stain of male hearts 27 days post-tumor cell inoculum (bottom). (D) Tumor mass in male and female mice. n = 6-8 per group. Graphs are mean ± SEM. *p < 0.05, **p < 0.01, ***p < vs. female. 55

68 The rate at which cardiac mass was lost also differed between the sexes (Figure 3C). This sexual dimorphism cannot be attributed to differences in tumor mass (Figure 3D). In addition to having less cardiac atrophy, female tumor-bearing mice survived about one week longer than males (data not shown). Since pro-inflammatory cytokines contribute to skeletal muscle cancer cachexia (Tisdale 1997), we performed a Mulitplex ELISA on male and female mouse serum at days 15 and 27. We found that males had a trend toward higher levels of pro- and antiinflammatory cytokines than females (Figure 4 and Table 1), which correlates with findings in colorectal cancer patients (Sharma, Greenman et al. 2010). The enhanced inflammatory profile in male tumor-bearing mice may therefore contribute to the increased cardiac and skeletal muscle mass loss and decreased survival we observed. Estrogen signaling is required for the maintenance of cardiac muscle mass in female tumor-bearing mice. Female cancer patients lose less skeletal muscle and body mass than men (Baracos, Reiman et al.). Because we observed this in our mouse model, we determined if estrogen plays a role in the prevention of cardiac muscle mass loss in females by injecting female mice with Fulvestrant (ICI 182,780), a potent and specific estrogen receptor (ER) antagonist (Wakeling, Dukes et al. 1991). Fulvestrant did not affect either cardiac mass or body weight in non-tumor-bearing mice (Figures 5A and C). Fulvestrant did, however, have a dramatic effect on tumor-bearing females. Inhibiting ER signaling caused female tumor-bearing mice to lose as much cardiac and body mass as tumorbearing males (Figures 5B and D). The extent of skeletal muscle mass loss, however, 56

69 Figure 4. Tumor-bearing males have increased levels of pro- and anti-inflammatory cytokines compared to females. A Multiplex ELISA was used to quantify the levels of pro-inflammatory cytokines (A) and anti-inflammatory cytokines (B) in male and female mouse serum at days 15 and 27 post-tumor cell inoculum. n = 3 per group. Male and female differences were not significant due to the variability between mice. Mean ± SEM. ** p < 0.01 and *p < 0.05 versus control. Cytokine abbreviations are; interleukin (IL), granulocyte macrophage colony-stimulating factor (GM-CSF), interferon-γ (IFN-γ), tumor necrosis factor-α (TNF-α). 57

70 Table 1. Serum cytokine levels in control and tumor-bearing male and female mice at day 15. Male Female Cytokine Control C-26 Control C-26 IL-1β 2.0 ± ± 1.5* 1.8 ± ± 1.4* IL ± ± ± ± 0.1 IL ± ± ± 0.04 ND IL ± ± ± ± 2.4** IL ± ± 53 * 3.4 ± ± 85 IL ± ± ± ± 2.7 IL ± ± ± 0.6 IL ± ± ± ± 3.3 IL ± ± ± ± 1.3 GM-CSF 3.3 ± ± ± ± 1.4 IFN-γ 1.3 ± ± ± ± 0.07 TNF-α 2.1 ± ± ± ± 0.4 MCP ± ± 30* 9.6 ± ± 8.9 A Multiplex ELISA was performed on serum from male and female control and tumorbearing mice 15 days post-tumor cell inoculum. Values are in picograms/milliliter (pg/ml) and results are represented as mean ± SEM. n = 3 per group. ** p < 0.01 and *p < 0.05 versus control. Cytokine abbreviations are; interleukin (IL), granulocyte macrophage colony-stimulating factor (GM-CSF), interferon-γ (IFN-γ), tumor necrosis factor-α (TNF-α), monocyte chemotactic protein-1 (MCP-1). 58

71 Figure 5. Estrogen receptor signaling is required for the maintenance of cardiac and body mass in female tumor-bearing mice. Fulvestrant treatment does not affect cardiac mass (A) or body weight (C) in control mice. Inhibition of estrogen signaling in tumor-bearing females results in equivalent cardiac mass (B) and body weight (D) loss as tumor-bearing males. % loss in ICI group was calculated relative to the non-ici controls. Heart weight (HW) is normalized to tibia length (TL) to normalize for mouse age and size. n = 6 per group. Graphs are mean ± SEM. **p < 0.01, ***p <

72 was not significantly affected with Fulvestrant treatment nor was tumor mass. Hence, in this cancer model, estrogen signaling prevents cardiac muscle and total body mass loss. Atrophic hearts have increased fibrosis and decreased aortic pressure and velocity. Cardiac fibrosis is often implicated in cardiac pathology and contributes to decreased function (Jalil, Doering et al. 1989). In order to determine the extent of fibrosis in atrophic hearts, transverse sections were stained with Picrosirius red, which reveals myocellular disarray and collagen deposition. We found that both male and female atrophic hearts had significant increases in fibrosis (50% and 65% respectively) and significant myocellular disarray (Figure 6A). Hearts from both control and tumor-bearing females had less (22% and 17% respectively) fibrosis than males but the fold change in fibrosis due to cancer was not significantly different between the sexes. Because collagen deposition and myocellular disarray cause myocardial stiffness and a decrease in cardiac function (Thiedemann, Holubarsch et al. 1983), we performed M-mode echocardiography on male and female mice at day 27. Male atrophic hearts had a significant 30% decrease in aortic pressure and a 16% decrease in aortic velocity (Figure 6B). Female tumorbearing mice, however, did not have a decrease in any of the functional parameters studied (Figure 6B and Table 2). Surprisingly, neither ejection fraction nor fractional shortening changed in either sex despite the extensive cardiac muscle loss. Cardiac atrophy is due to a decrease in myocyte size, not an increase in cell death. Cancer-induced skeletal muscle atrophy is primarily due to increased protein degradation resulting in decreased myofiber size (Temparis, Asensi et al. 1994). Though 60

73 Figure 6. Atrophic hearts have increased fibrosis and decreased function. (A) Picrosirius red staining of male hearts 27 days post-tumor cell inoculum. Collagen fibers appear as bright spots. Both male and female atrophic hearts have increased fibrosis (lower panel). (B) Echocardiography of male mice at day 27 showed significantly decreased aortic velocity and pressure, while female mice did not have any functional deficits. n = 6 per group. Mean ± SEM. *p < 0.05, ***p < vs. control. indicates significance compared to males in each respective treatment group. 61

74 Table 2. Transthoracic echocardiography on male and female control and tumor-bearing mice 27 days post-tumor cell inoculum. Male Female Control C-26 Control C-26 Heart rate (bpm) ± ± ± ± 4 Anterior wall, diastole (mm) 0.73 ± ± ± ± 0.05 Anterior wall, systole (mm) 1.19 ± ± ± ± 0.07 Posterior wall, diastole (mm) 0.88 ± ± ± ± 0.01 Posterior wall, systole (mm) 1.25 ± ± ± ± 0.07 LVID, diastole (mm) 4.05 ± ± ± ± 0.07 LVID, systole (mm) 2.50 ± ± ± ± 0.13 Fractional shortening (%) ± ± ± ± 3 Ejection fraction (%) ± ± ± ± 3.3 Aortic velocity (mm/sec) ± ± 29.2 * ± ± 46 Aortic pressure (mm Hg) 3.50 ± ± 0.19 * 2.56 ± ± 0.31 Cardiac output (ml/min) 19.4 ± ± ± ± 1.69 Stroke Volume (µl) ± ± ± ± 3.1 Male and female mice were subcutaneously injected with saline (control) or C-26 adenocarcinoma cells in the right flank. M-mode echocardiography was performed on each group 27 days post-tumor cell or saline inoculum. Values are represented as mean ± SEM. N = 9 per group *p < 0.05 versus control. Abbreviation is; Left ventricular internal diameter (LVID). 62

75 apoptosis does occur in cachectic skeletal muscle, it does not significantly contribute to muscle mass loss (Belizario, Lorite et al. 2001). In order to determine how cardiac muscle atrophies in our mouse model, we first quantified myocyte size in male and female control and atrophic hearts (day 27). Cardiac myocyte cross-sectional area from male tumor-bearing mice was 31% smaller than male controls, while atrophic female myocytes were only 16% smaller than female controls (Figure 7A). The sexually dimorphic decrease in myocyte area correlates with the sex difference in cardiac mass loss. To determine if increased cell death also plays a role in cardiac atrophy in males, we measured caspase-3 activity and the levels of caspase-9 cleavage products in cardiac muscle extracts and did not find an increase in either of these apoptotic markers at day 15 or 27 (Figures 7B and 7C). The amount of DNA per mg of tissue also did not change in atrophic cardiac muscle, while the amount of protein per mg tissue significantly decreased, as expected (Figure 7D). Together, these results indicate that cardiac muscle atrophy is due to a decrease in cell size, rather than an increase in apoptosis. All sarcomeric proteins are equally decreased in atrophic cardiac muscle. Myofibrillar proteins make up 40% of left ventricular dry weight (Samarel 1993). Therefore, a decrease in myocyte size must be accompanied by a decrease in myofibrillar proteins. When the same percent of final heart weight was analyzed, we found that male atrophic hearts contained 22% less MyHC than controls (Figure 8A). Contrary to previous studies in skeletal muscle (Acharyya, Ladner et al. 2004), we also found a significant decrease in sarcomeric actin that paralleled the decrease in MyHC (Figure 8A 63

76 Figure 7. Cancer-induced cardiac atrophy is due to a decrease in myocyte size and not an increase in apoptosis. (A) Immunohistochemistry of transverse sections of control and atrophic (day 27) male hearts with WGA. Scale bar: 300 µm. At least 150 myocytes per heart (n = 3 per group) were quantified. (B) Caspase-3 activity in cardiac myocytes from male tumor-bearing mice. n = 5 per group. (C) Pro-caspase-9 and caspase-9 cleavage products did not change in atrophic hearts. (D) Protein and DNA content per mg of cardiac tissue. n = 8 per group. Mean ± SEM. *p < 0.05, ***p <

77 Figure 8. All sarcomeric proteins are decreased in atrophic hearts. (A) Coomassiestained gel (0.005% final cardiac mass) shows MyHC and actin levels decrease in parallel in atrophic hearts. (B) Western blots of MyHC and actin extracted in high salt show MyHC and actin levels decrease in parallel in atrophic hearts. (C) SDS-PAGE of purified myofibrils revealed that all sarcomeric proteins decrease in parallel during cardiac atrophy. Samples were run on the same gel but were noncontiguous. Mean ± SEM. *p < 0.05, **p < 0.01, ***p < vs. control. n = 4-5 per group. 65

78 and 8B). MyHC/actin ratios therefore did not change, which suggests that entire cardiac sarcomeres are degraded during cancer-induced atrophy. To determine if all sarcomeric components decrease in parallel, we purified myofibrils from male atrophic and control hearts. When the same percent of final heart weight was loaded, we found a parallel decrease in all myofibrillar components (Figure 8C). Therefore, the absolute amounts of myofibrillar proteins decreased but the ratio of myofibrillar proteins in the sarcomere was maintained, presumably to preserve cardiac function. Cardiac muscle contains two MyHC isoforms: α and β. The murine heart is primarily composed of α-myhc, which has faster ATPase kinetics than β (VanBuren, Harris et al. 1995). A small increase in β-myhc can cause decreased Ca 2+ -activated ATPase activity and systolic function (Tardiff, Hewett et al. 2000) and is a marker of pathology. qrt-pcr revealed that β-myhc mrna expression increased in both sexes at day 15, and significantly increased 16-fold in females and 22-fold in males at day 27 (Figure 9A). α-myhc mrna levels did not change in males but significantly increased 3-fold in females at day 15, which correlates with the sex difference in cardiac mass at that time. In contrast to mrna data, western blots revealed that α-myhc protein decreased starting at day 15 and continued to decrease to 70% of control levels at day 27 in male hearts (Figure 9B). These results indicate that cardiac MyHCs are posttranscriptionally regulated, and suggest that increased protein degradation is responsible for the observed atrophy. Importantly, there was a significant increase in β-myhc protein (Figure 9B and 9C), which is considered pathologic in the rodent heart. 66

79 Figure 9. (A) Fold change of MyHC mrna (normalized to 18S expression) in each sex compared to controls. (B) α- and β-myhc protein expression in males at day 27. MyHC was solubilized in a high salt buffer and 0.5 µg was loaded. It was therefore not possible to re-probe the blot for a loading control. (C) High-resolution PAGE gels revealed a significant increase in β-myhc protein in the hearts of male and female tumor-bearing mice. Mean ± SEM. *p < 0.05, **p < 0.01, ***p < vs. control. n = 4-5 per group. 67

80 The UPS is not upregulated in cancer-induced cardiac atrophy. Skeletal muscle atrophy is accompanied by increases in transcription of a common, specific set of genes that are involved in protein degradation by the UPS (Lecker, Jagoe et al. 2004). Atrogin-1 and MuRF-1, muscle-specific E3 ubiquitin ligases, did not increase in male or female atrophic hearts, while expression increased 8-and 11- fold respectively, in the gastrocnemius of the same male animals (Figures 10A and B). Interestingly, the heart had 1.5- and 3-fold higher levels of atrogin-1 and MuRF-1 transcripts, respectively, than skeletal muscle and males had higher levels of these transcripts than females (Figures 10C and D). Because the UPS is primarily responsible for skeletal muscle protein degradation (Solomon and Goldberg 1996), we quantified the levels of poly-ubiquitinated proteins in both the soluble and insoluble fractions of male cardiac muscle lysates and did not find significant differences at any time point (Figure 11A). Cachectic skeletal muscle has increased ubiquitination activity (Solomon, Baracos et al. 1998), but we did not find a change in ubiquitin conjugation activity in atrophic cardiac muscle at day 15 or 27 (Figure 11B). Additionally, proteasome activity in atrophic cardiac muscle lysates did not change at day 15 or 27, while it increased 2-fold in the gastrocnemius (Figure 11C). Together, these results indicate that unlike skeletal muscle, UPS activity is not upregulated in the hearts of tumor-bearing mice. Inhibiting the proteasome in vivo does not prevent cardiac muscle mass loss The UPS may be contributing to cardiac atrophy even though it is not upregulated. We wanted to determine if inhibiting the proteasome in vivo would decrease cardiac muscle mass loss in male tumor-bearing mice. Bortezomib is a potent and 68

81 Figure 10. Atrogin-1 and MuRF-1 expression levels do not change in atrophic hearts, but are significantly upregulated in atrophic skeletal muscle. (A) Atrogin-1 and MuRF-1 mrna fold changes in male and female control and atrophic hearts. (B) Relative atrogin- 1 and MuRF-1 gene expression at day 27 in male heart and skeletal muscle (gastrocnemius). (C) Relative atrogin-1 and MuRF-1 gene expression in control male versus female cardiac muscle. (D) Relative atrogin-1 and MuRF-1 gene expression in cardiac (left ventricle) versus skeletal muscle. All genes were normalized to 18S expression. n = 4 per group. Mean ± SEM. *p < 0.05, **p < 0.01, ***p <

82 Figure 11. The UPS is not upregulated in the hearts of tumor-bearing mice. (A) Western blot of day 15 cardiac muscle lysates for total ubiquitin. (B) Ubiquitin conjugation assays do not show a difference in ubiquitination activity between groups. (C) Proteasome activity in atrophic cardiac and skeletal muscle. Bortezomib is a potent and specific proteasome inhibitor. n = 4 per group. Graphs are mean ± SEM. *p < 0.05 vs. control. 70

83 specific proteasome inhibitor used clinically in the treatment of multiple myeloma. Mice treated with bortezomib suffered from serious side effects including dehydration and hypothermia and appeared ill at 27 days. In order to avoid potential confounding effects from drug toxicity, we chose to sacrifice mice at the 21-day time point rather than the usual 27. Surprisingly, bortezomib decreased cardiac mass in non-tumor-bearing mice, which may be evidence of some toxicity (Figure 12A). 21 days of bortezomib treatment resulted in increased cardiac mass loss compared to untreated tumor-bearing mice (Figure 12B). Bortezomib treatment did not affect body weight or skeletal muscle mass loss (Figure 12B), but did decrease tumor mass (Figure 12C). Although we cannot verify that the proteasome was effectively inhibited in all tissues, these results imply that long-term proteasome inhibition does not prevent cancer cachexia and exacerbates cardiac mass loss. Autophagy is upregulated in atrophic hearts. Autophagy, a mechanism by which cells degrade large quantities of intracellular protein during periods of cellular stress, has been shown to play a larger role in the heart than in skeletal muscle (Wing, Chiang et al. 1991). Cathepsin L, beclin, and LC3 (microtubule-associated protein 1 light chain 3) are well-characterized markers of increased lysosomal activation during myocyte atrophy and autophagy (Mammucari, Milan et al. 2007). Cathepsin L mrna significantly increased 2-fold in both male and female atrophic hearts, while LC3 mrna increased approximately 1.5-fold at day 27 (Figure 13A). Cytosolic LC3-I is lipidated to form LC3-II upon activation of autophagy. LC3-II is the only well-characterized protein that is specifically localized to autophagic 71

84 Figure 12. Inhibiting the proteasome in vivo does not decrease body weight, skeletal or cardiac mass loss. (A) Cardiac mass of treated and untreated male mice 21 days posttumor cell inoculum. Bortezomib causes a decrease in cardiac mass in males without cancer. (B) Percent body weight (BW), heart weight (HW) and gastrocnemius (gastroc) mass loss in treated and untreated male tumor-bearing mice. (C) Bortezomib decreases tumor mass. Graphs are mean ± SEM. *p < 0.05, **p < 0.01, ***p < vs. untreated controls. n = 7-8 per group. 72

85 vacuoles and serves as an accurate marker for autophagy (Barth, Glick et al.). LC3-II protein levels were 7-fold higher in male atrophic hearts and only 3-fold higher in females at days 15 and 27 (Figure 13B). LC3-II was significantly higher in male than female hearts at day 27, which could explain the increased cardiac mass loss in tumorbearing males. Direct evidence of autophagy was obtained by electron microscopy, which revealed the presence of numerous double-membraned autophagic vacuoles that contained portions of cytoplasm, mitochondria, and myelin-like structures (Figure 13C). Autophagic vacuoles were rarely detected at day 15, but were abundant by day 27. Additionally, atrophic hearts stained more heavily for cathepsin D, a lysosomal protease (Figure 13D). These results show that autophagy increases along the course of disease in cardiac muscle of tumor-bearing mice and is likely playing a major role in the enhanced protein degradation responsible for cardiac muscle atrophy. Discussion More than one-half of cancer patients suffer from cachexia, a severe musclewasting syndrome that results in decreased prognosis and survival (Tisdale 2002). Cancer also causes cardiac muscle atrophy, a phenomenon that has been understudied by the scientific and oncology communities. A recent study found a direct correlation between muscle mass loss and death, and postulated that cardiac muscle atrophy contributed to the decreased survival in tumor-bearing mice (Zhou, Wang et al. 2010). Our studies reveal the phenotype, mechanisms and sex differences of cardiac atrophy and show that it is distinct from skeletal muscle. Our results demonstrate that cardiac atrophy is an important feature of cancer cachexia that should be considered during the treatment and 73

86 Figure 13. Autophagy is upregulated in male and female atrophic hearts and is nearly three-fold higher in males. (A) Fold change of autophagy marker transcripts (normalized to 18S) in male and female atrophic hearts at days 15 and 27. (B) Western blots revealed increases in LC3-II levels in the hearts of both male and female tumor-bearing mice, though it was only significant in males. Quantification represents the band density of blots containing males and females on the same gel (not shown). n = 4 per group. (C) Electron micrographs from the left ventricle of tumorbearing males (day 27). Atrophic hearts contain numerous autolysosomes and double-membraned autophagosomes (arrows) containing cellular components. (D) Immunohistochemistry for cathepsin D revealed more intense staining in atrophic hearts. Mean ± SEM. *p < 0.05, **p < 0.01, ***p < vs. control. indicates significance compared to males. 74

87 management of cancer patients, particularly in the setting of chemotherapy-induced cardiotoxicity or pre-existing heart disease. We demonstrated that atrophic hearts had significantly decreased levels of all myofibrillar proteins. The issue whether sarcomeric proteins are selectively depleted is controversial; some have shown a specific decrease in MyHC (Acharyya, Ladner et al. 2004), while others report that MyHC is not selectively lost in muscle atrophy (Cohen, Brault et al. 2009). Although we show a parallel decrease in all sarcomeric proteins in the later stages of atrophy, it is possible that MyHC is specifically decreased early in cardiac atrophy, before any measurable mass loss. Presumably, absolute decreases in the number of sarcomeres would result in decreased cardiac function, which is why we were surprised to find that neither ejection fraction nor fractional shortening changed in either sex. However, other models of cardiac atrophy have also shown that atrophic hearts maintain their ejection fraction (Welsh, Dipla et al. 2001; Artaza, Reisz-Porszasz et al. 2007), implying that cardiac muscle is able to adapt functionally to muscle protein loss up to a certain point. We did find that male tumor-bearing mice had a significant decrease in aortic pressure and velocity, which correlates with clinical findings of hypotension in cancer patients. The increases in β-myhc we observed in the atrophic heart could have significant functional implications that may not be detected by echocardiography. β- MyHC induction in the rodent heart is pathologic and can contribute to myocardial contractile dysfunction since cardiac pressure development and power output decrease as β-myhc increases (Tardiff, Hewett et al. 2000; Korte, Herron et al. 2005). A more gradual model of cardiac atrophy might reveal higher levels of β-myhc than we 75

88 observed, which could affect cardiac function. Since cachexia in our mouse model is very rapid and leads to death in approximately four weeks, there may have been decreased cardiac function if the mice had lived longer. Because most cancer patients live with a tumor burden for many years, it is possible that cardiac abnormalities occur but likely remain undiagnosed due to insufficient monitoring. The proteolytic pathways involved in skeletal muscle cachexia have been wellcharacterized and we were not expecting to find differences in the pathways upregulated in cardiac muscle. The UPS mediates skeletal muscle atrophy, but we did not find an increase in total ubiquitinated proteins, ubiquitin conjugating activity, or proteasome activity in atrophying cardiac muscle. The heart may not upregulate the UPS because it already has a high basal activity: cardiac muscle has a higher metabolic rate than skeletal muscle and has a higher protein turnover rate (Earl, Laurent et al. 1978). Accordingly, expression of UPS components and proteasome activity are greater in cardiac than skeletal muscle (Liu, Miers et al. 2000). The baseline levels and activity of UPS components in the heart may therefore be sufficient to process the increased supply of substrates during cardiac atrophy. Interestingly, inhibiting the proteasome in vivo with bortezomib resulted in increased cardiac mass loss when compared to non-treated controls. This may have been due to drug toxicity in our study because of the long treatment period. Alternatively, bortezomib is known to negatively affect the heart as many myeloma patients receiving bortezomib develop cardiac abnormalities that progress into heart failure (Hacihanefioglu, Tarkun et al. 2008). Other studies, however, have found that bortezomib can prevent decreases in muscle protein and myosin in models of diaphragm weakness 76

89 due to heart failure and denervation-induced atrophy, but the treatment time in these studies was only 2-10 days (Beehler, Sleph et al. 2006; van Hees, Li et al. 2008). The effects of bortezomib on cancer cachexia has been studied in humans and bortezomib was unable to reduce weight loss in patients with pancreatic cancer (Jatoi, Alberts et al. 2005). We confirmed this, as we did not observe a change in body weight or skeletal muscle mass loss in treated mice. These results could have been due to the dose used or that the extent of proteasome inhibition in certain tissues was insufficient. Autophagy plays only a minor role in skeletal muscle atrophy (Baracos, DeVivo et al. 1995), but we found a significant increase in autophagy in the hearts of tumorbearing mice. Autophagy is essential for cardiac homeostasis and cardiac-specific reduction of autophagy results in contractile dysfunction, increased levels of polyubiquitinated proteins and increased apoptosis (Nakai, Yamaguchi et al. 2007), indicating that autophagy is important for protein turnover and clearance of misfolded or aggregated proteins. In fact, cardiac autophagy is induced in response to intracellular protein aggregates (Tannous, Zhu et al. 2008) and it is thought to be beneficial by removing aggregates that are unable to be cleared by the UPS. In addition to degrading mitochondria, it is possible that autophagosomes also degrade myofibrillar proteins cleaved from the sarcomere during cardiac atrophy in order to prevent protein aggregation and to preserve cardiac function. Although there is evidence for lysosomal degradation of myofibrillar proteins (Gerard and Schneider 1979), it is a very controversial issue, and will be an important area of future investigation. Cachexia increases mortality in both humans and rodents (Bachmann, Ketterer et al. 2009; Zhou, Wang et al. 2010). Interestingly, male cancer patients have more severe 77

90 cachexia and increased mortality than females (Palomares, Sayre et al. 1996; Baracos, Reiman et al. 2010). Accordingly, we found striking sex differences in cardiac and body mass loss and in all of our molecular analyses. We showed that ER signaling protects females against body weight and cardiac mass loss, indicating that estrogen could also be involved in the increased survival we observed in females. The increase in body mass loss we observed with Fulvestrant treatment must be due to increased fat loss since skeletal muscle mass was not affected. Female ERα knockout mice have increased skeletal muscle mass (Brown, Ning et al. 2009), which indicates that muscle mass is not positively regulated by ER signaling. Although decreases in estrogen are typically known to increase fat mass, the role of estrogen in regulating fat mass in a cachexia model is unknown, and will be an interesting area of future study. Collectively, these studies offer much-needed insight into the effects of cancer on the heart, and suggest that cardiac function in cancer patients, particularly in males, requires closer monitoring. 78

91 CHAPTER III. Quantification of MyHC in atrophic muscle Introduction Muscle comprises 40% of the human body and is an important amino acid reservoir during times of starvation and disease. Muscle protein breakdown is initially beneficial to the organism because it provides the liver with amino acids for gluconeogenesis and acute phase protein synthesis, and serves as an energy source for other cell types such as immune cells (Hasselgren and Fischer 2001). However, chronic muscle catabolism is detrimental and causes profound muscle weakness and fatigue, decreased respiratory function, decreased overall prognosis, and contributes to increased mortality. Accelerated skeletal muscle mass loss in the context of disease is known as cachexia, and it occurs in patients with chronic heart failure, AIDS, and cancer. In fact, approximately 50% of patients with cancer suffer from cachexia (Tisdale 2002). Tumors induce a pro-inflammatory response, leading to increased serum levels of proinflammatory cytokines, which cause muscle protein degradation by activating the ubiquitin proteasome system (UPS) (Morley, Thomas et al. 2006). Myofibrillar proteins comprise at least 60% of all muscle proteins, and they decrease significantly in both humans and rodents with cachexia. Several groups have shown that myosin heavy chain (MyHC) is decreased by up to 80% and is selectively targeted for degradation in skeletal muscle during catabolic conditions, including cancer (Acharyya, Ladner et al. 2004; Moore-Carrasco, Garcia-Martinez et al. 2006; Schmitt, Martignoni et al. 2007; Eley, Skipworth et al. 2008; Paul, Gupta et al. 2010). 79

92 Myosin is the motor component of muscle and causes contraction upon actin binding and ATP hydrolysis. Each myosin molecule is composed of two heavy chains and four light chains. Each heavy chain is 220 kd and contains three protein domains: the head, neck, and rod. The N-terminal head domain is globular, has ATPase activity, and associates with actin. There is a short α-helical neck region that extends from the head and binds the myosin light chains. This neck extends into the rod domain, which is the C- terminal 1100 amino acid portion of the molecule. The rod is responsible for myosin assembly and dimerizes to form an α helical coiled-coil (Weiss and Leinwand 1996). The amino acid sequence of the rod domain is highly conserved, leading to very stable hydrophobic and electrostatic interactions. The rod is composed of 40 repeats of 28 amino acids each (Parry 1981). Each 28-residue repeat is arranged into alternating bands of positive and negative charges leading to differentially charged zones along the outer surface of the rod (McLachlan and Karn 1982). Electrostatic interactions between oppositely charged portions of two rods likely provide the driving force for assembly of the α helical coiled-coil and higher order oligomers and filaments. Only high ionic strength buffers ( 300 mm) can disrupt these electrostatic interactions. The rod domain therefore confers the solubility and aggregation properties of MyHC. Limited tryptic digest of myosin yields two distinct fragments: heavy meromyosin (HMM) and light meromyosin (LMM). HMM contains the head and motor domain, while LMM is the C-terminal two-thirds of the rod domain. LMM is responsible for the solubility and assembly of myosin. In low ionic strength buffers, LMM spontaneously aggregates and forms highly ordered paracrystalline structures in vitro (Chowrashi and Pepe 1977). MyHC solubility is therefore dependent on salt concentration and MyHC 80

93 assembles into filaments in low ionic strength solutions (less than 200 mm) and disassembles as ionic strength increases (greater than 300 mm) (Szent-Gyorgyi 1953; Trinick and Cooper 1980). Studies on cachectic rodents and humans have found that MyHC is specifically and significantly decreased in atrophic muscles compared to other myofibrillar proteins (Acharyya, Ladner et al. 2004; Schmitt, Martignoni et al. 2007; Eley, Skipworth et al. 2008). Interestingly, all of these studies used very low ionic strength lysis buffers for their MyHC anaylsis. This is surprising given that the properties of MyHC insolubility in the context of low ionic strength have been known for decades. As discussed above, MyHC is insoluble in low salt, and greater than 90% of MyHC is found in the insoluble, pellet fraction and is virtually undetectable in the supernatant (Cote and McCrea 1987). It is therefore not possible to accurately quantify sarcomeric MyHC levels in the soluble fraction of these low salt lysates. Here, we show the extent of MyHC solubility and extraction from muscle between conventional, low salt, lysis buffers (LB) and high salt, myosin extraction buffer (MEB). We propose that accurate quantification of MyHC can only be achieved when muscles are homogenized in a high salt (300 mm) buffer. Finally, using a colon-26 adenocarcinoma mouse model, we show that MyHC levels do not specifically decrease in cachectic muscles and that all myofibrillar proteins decrease in parallel. Materials and Methods Animal studies 81

94 Female Balb/c mice and male DBA/2 mice (The Jackson Laboratory) were crossed to produce CD2F1 mice, which is the strain used in all colon-26 adenocarcinoma cachexia studies. Animals were housed under standard conditions and fed a standard diet. All animal studies were reviewed and approved by the Institutional Animal Care and Use Committee at the University of Colorado at Boulder. Colon-26 adenocarcinoma (C-26) cells were a generous gift from Dr. Denis Guttridge (The Ohio State University) and were cultured as described (Acharyya, Ladner et al. 2004). C-26 cells were trypsinized, washed in PBS, pelleted and resuspended in PBS at a concentration of 5 x 10 6 cells/ml. 100 µl of this was injected subcutaneously into the right flank of 8-week old male mice. Mice were weighed and cervically dislocated 27 days post-tumor cell inoculum. Tissues were rapidly excised, weighed, flash-frozen in liquid nitrogen and stored at -80 C for further analysis. Muscle homogenization and lysis buffers For MyHC analysis, muscles were homogenized in myosin extraction buffer (Butler- Browne and Whalen 1984) (with modification from McKoy et al 1998) which is composed of 300 mm NaCl, 0.1 M NaH 2 PO 4, 0.05 M Na 2 HPO 4, 0.01 M Na 4 P 2 O 7, 1mM MgCl 2, 10 mm EDTA, and 1 mm DTT, ph 6.5 and complete protease inhibitor cocktail (Roche). LB is composed of 50 mm Tris, 150 mm NaCl, 50 mm NaF, 1 mm EDTA, 0.5% Triton X-100, ph 7.5 and complete protease inhibitor cocktail. Muscles were homogenized in 300 µl of lysis buffer using a drill press. Lysates were centrifuged at 12,000 rpm for 15 min at 4 C. The supernatant was removed, and protein concentration determined (Bio-Rad DC protein assay). The insoluble material in the pellet was 82

95 resuspended in 0.5 M NaOH, rotated for 30 minutes at 4 C, homogenized by hand using an eppendorf tube homogenizer, and then centrifuged again at 700 rpm for 1 minute, and protein concentration determined. To determine the volume required to load a certain percent of final muscle weight, muscle weight (in µg) was divided by the protein concentration of the lysate. The range of protein for optimal detection of MyHC in MEB lysates is µg for immunoblots and 0.5 µg 1 µg for Coomassie-stained SDS- PAGE gels. The protein range for MyHC detection in LB is 5-10 µg for immunoblots and µg for Coomassie-stained SDS-PAGE gels, but this may vary due to variable extraction of MyHC in LB. Myofibrillar gels and western blots The myofibrillar resolving gel was composed of 10% acrylamide (37.5:1), 1.5 M Tris ph 8.8, and 0.1% SDS. The stacking gel was composed of 4% acrylamide, 0.5 M Tris ph 6.8, and 0.1% SDS. Gels were run for two hours at 100V, stained with Imperial Protein Stain (Thermo Scientific) and scanned and analyzed with ImageJ. Gels for western blots were 7.5% acrylamide (37.5:1), 1.5 M Tris ph 8.8, and 0.1% SDS and a stacking gel as above. These gels were allowed to stack at 80 V for 30 minutes, then run at 100 V for two hours. Gels were transferred to polyvinylidene difluoride (PVDF) membranes overnight at 4 C, blocked in 5% non-fat dry milk (NFDM), exposed to primary antibody for 2 hours at room temperature (MyHC antibody: MF20 hybridoma supernatant 1:5) or overnight at 4 C (α-actin antibody: Sigma clone 5C5). Secondary HRP-conjugated antibody was used at 1:8000 dilution for 1 hour at room temperature in 83

96 2.5% NFDM. Immunoreactivity was visualized using a Western Lighting chemiluminescence detection system (Perkin Elmer) and quantified using ImageJ. Data and statistical analysis Data are presented as mean ± SEM. Differences between groups were evaluated for statistical significance using Student s t-test. P values less than 0.05 were considered significant. Results We first determined the relative solubilities of MyHC in LB and MEB in control tibialis anterior muscle of 12-week old male mice. The conventional LB contains 150 mm salt, while MEB contains 300 mm salt, which fully solubilizes sarcomeric MyHC (Szent-Gyorgyi 1953). When equal amounts of the soluble fraction from each lysis procedure were loaded, SDS-PAGE revealed that lysis in LB results in 94% less soluble MyHC than lysis in MEB (Figure 14A). Because MyHC is insoluble in low salt, we reasoned that most of the MyHC would be found in the insoluble pellet fraction in the LB lysate. Pellets from both lysis conditions were resuspended in 0.5 M NaOH and analyzed by SDS-PAGE. Insoluble pellets from the lysis in LB contain 3-fold more MyHC than pellets from the lysis in MEB (Figure 14B). This does not account for all of the MyHC in the LB lysate, thus some MyHC was likely lost in the extraction procedure or remained insoluble after extraction of the pellet. Thus, muscle lysis in high salt solubilizes MyHC nearly 20-fold more than lysis in low salt. 84

97 Figure 14. MyHC solubility in high salt myosin extraction buffer (MEB) and low salt lysis buffer (LB). (A) Coomassie-stained myofibrillar gel of the supernatant fraction of tibialis anterior muscle lysed in MEB and LB. 10 µg of protein was loaded for each sample. (B) Coomassie-stained myofibrillar gel of the pellet fraction of tibialis anterior muscle lysed in MEB and LB. 5 µg of protein was loaded for each sample. Histograms to the right of each figure represent MyHC and actin quantities in each sample. * p < 0.05, *** p < Graphs are mean ± SEM. 85

98 To determine the ratio of MyHC in the soluble vs. insoluble fractions of each buffer, equal amounts of protein were loaded from the supernatant and pellet on the same SDS-PAGE gel. Muscle lysis in MEB results in 10-fold higher levels of MyHC in the supernatant compared to the pellet (Figure 15A), while lysis in LB results in greater than 5-fold levels of MyHC in the pellet versus the supernatant (Figure 15B). Conventional low-salt buffers therefore solubilize less than 20% of sarcomeric MyHC, which agrees with solubility studies in vitro (Szent-Gyorgyi 1953). Thus, lysis in LB does not represent the actual, physiological state of MyHC expression. Additionally, only µg of MEB lysate is required for analysis by SDS-PAGE or Western blot because the lysates contain a very high concentration of MyHC. Conversely, the concentration of MyHC in LB lysates is very low, requiring at least 20 µg of protein for visualization by SDS-PAGE. To address the extent of MyHC loss in cachectic muscles, we used a murine colon-26 adenocarcinoma model, which induces a severe and rapid cachexia (Tanaka, Eda et al. 1990). We have found that male tumor-bearing mice lose approximately 50% of their tibialis anterior (TA) muscle mass in 27 days. In order to accurately determine MyHC levels in cachectic versus control muscle, we first solubilized TA muscle of agematched control and 27-day tumor-bearing mice in MEB. When 1 µg of total protein was analyzed, we did not observe a difference in MyHC levels between the two groups (Figure 16A). This is likely because myosin comprises approximately 30% of myocyte protein and loading the same amount of protein results in artificially overloading MyHC in cachectic samples. Other groups using low-salt lysis buffers have found that MyHC is significantly decreased (~80%) in cachectic muscle (Acharyya, Ladner et al. 2004). We first studied 86

99 Figure 15. Relative amounts of MyHC in the soluble and insoluble fractions of MEB and LB muscle lysate. The supernatant is the soluble fraction of each lysate after centrifugation, and the pellet is the insoluble fraction. (A) Coomassie-stained myofibrillar gel of tibialis anterior muscle lysed in MEB shows that most of the MyHC is in the soluble fraction. (B) Coomassie-stained myofibrillar gel of tibialis anterior muscle lysed in LB shows that most of the sarcomeric MyHC is in the insoluble, pellet fraction of the lysate. Histograms to the right of each figure represent MyHC quantities in each condition. *** p < Graphs are mean ± SEM. 87

100 Figure 16. MyHC levels in control and cachetic (C-26) tibialis anterior muscles in MEB and LB. (A) MyHC levels in control and cachectic muscle lysed in MEB are the same. 1 µg of protein was loaded per sample. (B) A representative coomassie-stained gel showing increasing amounts of protein of a normal (N) muscle and a cachectic (C) muscle lysed in LB. The cachectic muscle has less MyHC than the control. Greater than 20-fold more protein is required to visualize MyHC in LB. (C) MyHC levels in control and cachectic muscle lysed in LB are variable. 20 µg of protein was loaded per sample. Graphs are mean ± SEM. 88

101 only one muscle in each treatment group. Lysis with LB revealed a decrease in MyHC in cachectic muscle, as others have observed (Figure 16B). Of note, up to 75 µg of protein was loaded, once again illustrating the inefficiency of MyHC extraction with LB. However, when we lysed three control and cachectic TA muscles in LB and quantified MyHC levels, we found variable levels of MyHC in both control and cachectic samples (Figure 16C). We did not observe a consistent decrease in MyHC in the cachectic muscles although some samples compared side by side would reveal a decrease, as seen in Figure 16B. The variability in MyHC is likely due to the inefficient solubilization of myosin in low salt conditions. To more precisely examine the difference in MyHC levels in control and cachectic muscles in each lysis buffer, we homogenized one-half of one muscle (gastrocnemius) in MEB and the other half in LB of both control and cachectic muscles. We have shown elsewhere that the gastrocnemius also decreases in mass by 50% in mice with tumors (Cosper and Leinwand 2011). All samples were homogenized at the same time using identical protocols with exception of the lysis buffer. Analysis of MyHC in the same muscle in each buffer provides an internal control for the different lysis buffer conditions. As expected, we did not observe a difference in MyHC levels when 1 µg of MEB lysate from control and cachectic muscles was analyzed by SDS-PAGE (Figure 17). We observed even more variability in MyHC levels than with the TA when we lysed the same muscles in LB and loaded 20 µg of total protein (Figure 17). Both the control and the cachectic groups contained significant variability in MyHC levels such that when all levels were compared, there was not a statistically significant difference between the two groups. 89

102 Figure 17. Comparison of MyHC levels between control and cachectic (C-26) gastrocnemius muscle using MEB and LB in the same muscle. Each letter (above a sample) corresponds to a muscle. Thus each muscle was lysed in both MEB and LB for direct comparison. Due to the inefficiency of MyHC extraction in LB, 20 µg of LB lysate was analyzed, as opposed to 1 µg of MEB lysate. The histogram represents MyHC and actin levels in each disease and buffer condition. Lysis in MEB provides consistent MyHC and actin levels, while lysis in LB results in variable MyHC, but not actin, levels. Graphs are mean ± SEM. 90

103 Many published studies also show a large variation in MyHC levels in the cachexia samples (Acharyya, Ladner et al. 2004; Schmitt, Martignoni et al. 2007; Eley, Skipworth et al. 2008) but still report that there is an overall decrease. Our results seem to indicate that the extent of MyHC extraction in LB is not only inefficient and incomplete, but also highly variable, which makes quantification difficult and prone to error. For example in Figure 17, comparing lane C with lane E in the LB group would lead to the conclusion that there is significantly less MyHC in cachectic muscle. This experiment was repeated with TA muscles and the same results were obtained (data not shown). The statement that MyHC is specifically degraded implies that other myofibrillar proteins, such as sarcomeric actin, do not decrease in cachectic muscles. The myosin to actin ratio is physiologically relevant because it determines the extent of contraction. We therefore wanted to ensure that actin was also solubilized in both the high and low salt conditions so that we could accurately compare MyHC and actin levels. As seen in Figure 14, actin levels are not different between the two conditions. However, the actin band seen on SDS-PAGE represents all forms of actin in the myocyte. Because only sarcomeric α-actin is physiologically relevant to muscle contraction, we immunoblotted control muscle lysates in each buffer for both sarcomeric α-actin and total MyHC. When 10 µg of protein was loaded, we found 70% higher levels of sarcomeric α-actin in the MEB lysate, implying that this buffer is also better for solubilizing and quantifying sarcomeric actin (Figure 18). We also analyzed the insoluble pellet of each lysate and did not find a difference in sarcomeric actin levels (data not shown). We have determined the appropriate conditions for optimal MyHC and actin solubilization and analysis, but an accurate method for comparing sarcomeric protein 91

104 Figure 18. α-sarcomeric actin levels in MEB and LB tibialis anterior muscle lysates. A western blot for MyHC and α-sarcomeric actin revealed a significant increase in both proteins in MEB lysate compared to LB lysate. 10 µg of protein was loaded for each sample. ** p < 0.01, *** p < Graphs are mean ± SEM. 92

105 levels between control and atrophic muscles remains a controversial issue. As discussed above, muscle lysate in MEB contains a very high concentration of MyHC, requiring less than 1 µg of total protein for analysis. Thus loading the same amount of protein on a gel results in artificially high levels of MyHC in the cachectic samples, erroneously suggesting that MyHC levels do not change in atrophic conditions (refer to Figure 17). We reasoned that loading gels based upon percent of final muscle weight would allow us to better quantify changes in MyHC and other myofibrillar proteins since the number of myocytes is the same in both control and atrophic muscle (Temparis, Asensi et al. 1994). Comparison of the same percent of muscle weight, then, allows us to quantify the relative amounts of MyHC and other myofibrillar proteins. When 0.001% of final TA weight in MEB was analyzed by Western blot, we observed a 20% decrease in MyHC levels in the cachectic muscles and a parallel decrease in actin (Figure 19A). When 0.1% of final TA weight was loaded from LB lysate, we did not detect a difference in MyHC in cachectic samples, as MyHC levels were variable across all groups (Figure 19A). Actin also did not decrease in the cachectic samples in LB when we analyzed the same percent of final muscle mass. Similar analysis of the insoluble pellet fraction from lysis in LB revealed a decrease in both MyHC and α-actin (Figure 19B). This suggests that the insoluble fraction in LB is more accurate than the soluble fraction for quantifying MyHC, likely because that fraction contains the majority of sarcomeric MyHC. Although Western blots allow for evaluation of specific proteins, they are not quantitative. We therefore analyzed 0.002% of final gastrocnemius muscle weight in MEB by SDS-PAGE and stained the gel with colloidal Coomassie, which is technically quantitative. This revealed a 27% decrease in MyHC and a 22% decrease in actin in the 93

106 Figure 19. MyHC levels in control and cachectic (C-26) skeletal muscle when analyzed based on final muscle weight. (A) Immunoblot of MyHC shows a 20% decrease in the cachectic samples when 0.001% of final tibialis anterior (TA) weight was analyzed from MEB (top). Analysis of 0.1% of TA weight in LB did not show a decrease in MyHC in the cachectic samples (bottom). (B) Immunoblot of the insoluble, pellet fraction of LB lysates shows a significant decrease in MyHC in the cachectic samples when 0.005% of TA weight was analyzed. Graphs are mean ± SEM. * p < 0.05 vs. control. 94

107 cachectic muscles (Figure 20). The MyHC/actin ratio did not change. Therefore both MyHC and actin decrease in parallel during cancer cachexia. We performed the same experiment in atrophic cardiac muscle and also found that MyHC decreased in parallel with actin (Cosper and Leinwand 2011). One of the main goals of this study was to determine if MyHC is specifically decreased with respect to other myofibrillar proteins such as actin. This is an important question because it has physiological implications and may provide insight into the mechanism of muscle atrophy. Our results indicate that the previously published findings are an artifact from MyHC extraction in low salt and that both MyHC and actin decrease in parallel. In conclusion, we have described the optimal lysis conditions and analysis method to compare MyHC levels in control and atrophic muscle. A high salt lysis buffer is required to fully solubilize MyHC from the sarcomere and allows for accurate quantification of its levels. Low salt lysis buffers do not solubilize MyHC, resulting in variable amounts in the soluble fraction, and an inaccurate representation of sarcomeric MyHC quantities. Additionally, comparison of MyHC levels in different sized muscles should be done according to a certain percent of final muscle weight. Together, these methods provide an accurate way to quantify MyHC in the context of catabolic disease. Discussion Cancer causes significant muscle mass loss that negatively impacts the patient. Cachexia was first described centuries ago, but the specific myofibrillar proteins that are degraded have just recently been described. A pivotal study by Acharyya et al. demonstrated that MyHC is specifically decreased over all other myofibrillar proteins 95

108 Figure 20. MyHC levels in control and cachectic (C-26) gastrocnemius muscle. Muscle was homogenized in MEB, and analyzed based on final muscle weight (0.002%). Gel was stained with Coomassie and protein levels were quantified in the histogram (right). Graphs are mean ± SEM. * p < 0.05 vs. control. 96

109 (Acharyya, Ladner et al. 2004). Because we routinely lyse muscle in high salt for MyHC analysis, we were unable to reproduce those results, likely because we were analyzing the same amount of total protein per sample. Additionally, we analyzed only µg of protein while Acharyya et al. and Paul et al. analyzed µg of protein when quantifying MyHC (Acharyya, Ladner et al. 2004; Paul, Gupta et al. 2010). That amount of protein would fall well above the linear range of detection in our assays. It became apparent that large amounts of protein were necessary because those groups used a conventional, low-salt lysis buffer and analyzed the soluble fraction of the lysate in their MyHC studies. We therefore wanted to determine the efficiency of MyHC extraction in both high and low salt lysis buffers so that we could accurately establish the relative changes of MyHC in cachectic muscle. As Szent-Gyorgyi originally published in 1953, we found that MyHC is insoluble in low salt (150 mm) and more than 95% soluble in high salt (300 mm). We also demonstrated that the majority of MyHC is found in the insoluble pellet of low salt lysate. Most groups analyze the soluble supernatant fraction of cell or tissue lysates where MyHC is not accurately represented, which leads to artificially low levels upon analysis. When we compared MyHC in each lysis buffer, we found it necessary to load 20-fold more protein in the LB group than in the MEB group because MyHC extraction in LB is so inefficient. Because most other cellular proteins are soluble in this low salt buffer, MyHC appears to be decreased in relation to other proteins, leading to the conclusion that MyHC is specifically decreased. Our results show that this conclusion is based upon an artifact from the tissue lysis procedure. 97

110 One of the most important findings of this study is that MyHC decreases in parallel with other myofibrillar proteins in cachectic skeletal muscle. We reached this conclusion by analyzing fully solubilized sarcomeric MyHC in equivalent percentages of muscle mass. We found that MyHC and actin decrease in parallel, resulting in consistent MyHC/actin ratios. This result provides insight into the mechanisms of muscle atrophy as it implies that whole sarcomeres are degraded during atrophy, as opposed to MyHC alone. The ubiquitin proteasome pathway mediates myofibrillar protein degradation (Solomon and Goldberg 1996) but how sarcomeric proteins, including MyHC, are removed from the sarcomere and ubiquitinated is unknown. MyHC is embedded in the sarcomere and it has been proposed that during atrophy, components of the thick filament (myosin light chains, myosin-binding protein C) are lost from the myofibril first, which increases the susceptibility of MyHC to ubiquitination and dissociation from the sarcomere (Cohen, Brault et al. 2009). This group also found a decrease in all myofibrillar proteins in a denervation atrophy model, which agrees with our results. Therefore, the claim that MyHC is specifically decreased in atrophic muscles is likely incorrect given that MyHC cannot be targeted for degradation while associated with the actomyosin complex (Solomon and Goldberg 1996; Cohen, Brault et al. 2009). If MyHC is insoluble in low salt regardless of disease state, why do some studies show a consistent decrease in MyHC levels in atrophic muscles? There is evidence that the MyHC present in the soluble fraction of low salt lysates is newly synthesized protein that has not yet been incorporated into sarcomeres, or that it is an intermediate in the normal turnover of myofibrils. Easily releasable myofilaments are elements of the 98

111 sarcomere that easily dissociate in a low-salt, ATP-containing relaxing solution (van der Westhuyzen, Matsumoto et al. 1981). Less than 3% of total myofibrillar protein is easily releasable but analysis of this population revealed that MyHC and actin are the two predominant proteins. Moreover, this MyHC was shown to be newly synthesized, which indicates that it had not yet been incorporated into a sarcomere (van der Westhuyzen, Matsumoto et al. 1981). This correlates with the finding that newly synthesized myosin is located at the periphery of myofibrils (Morkin 1970). Additionally, a low concentration of MyHC remains unassembled at equilibrium with filaments (Saad, Pardee et al. 1986), suggesting the existence of a small, unincorporated and soluble pool of myosin. Thus the soluble MyHC in low salt that we, and others, have observed may represent newly synthesized myosin. Because protein synthesis often decreases in cachectic muscle (Smith and Tisdale 1993), it would be logical to find less of this soluble MyHC in the muscles of tumor-bearing mice. There is also evidence that this easily releasable pool represents an intermediate in the breakdown of myofibrils. The yield of easily releasable filaments increases during muscle atrophy induced by fasting and cancer suggesting that they are an intermediate in myofibrillar turnover (Dahlmann, Rutschmann et al. 1986; Neti, Novak et al. 2009). Inhibiting proteolysis decreases the size of the releasable filament fraction, while addition of Ca 2+ increases it (van der Westhuyzen, Matsumoto et al. 1981). Calpain, a Ca 2+ - activated protease implicated in myofibril cleavage, can also cause the release of these filaments from myofibrils (Neti, Novak et al. 2009). This implies that there is an initial cleavage step that releases MyHC and actin from the surface of the sarcomere during normal protein turnover and in catabolic states. If the soluble pool of MyHC was just 99

112 cleaved from the sarcomere, it would be counterintuitive to find less of this in cachectic muscle where rates of myofibrillar proteolysis are increased. However, depending on the time point analyzed, muscle may be protecting itself from further mass loss by decreasing myofilament cleavage, which would lead to detection of decreased levels of free MyHC in cachectic muscle. Regardless of the identity of the low salt-soluble MyHC pool, it does not represent sarcomeric MyHC because the extent of MyHC loss reported by others cannot be physiologically correct. Some groups have shown greater than an 80% reduction in MyHC, which is not realistic (Acharyya, Ladner et al. 2004; Eley, Skipworth et al. 2008; Zu, Bedja et al. 2010). If muscle mass decreased by 50% and myosin comprises 30% of total muscle protein, then myosin should decrease only by 15%, which is 5-fold less than what has been reported. The results we obtained with high-salt MyHC extraction showed a ~20% decrease in cachectic muscle, which is near the expected, calculated value. We have only addressed the effect of salt concentration on the efficiency of MyHC extraction in different buffers. Other variables such as types and amounts of detergent, ph, or EDTA may also affect the soluble pool of MyHC. However, the effects of those buffer components on MyHC extraction are most likely very minor compared to the ionic strength of the buffer, given the known relationship between salt concentration and myosin solubility. One downfall to lysing muscle in MEB is that it is not possible to reprobe an immunoblot for a loading control, such as GAPDH. This is because the proportion of MyHC in the lysate is so high that less than 1 µg is required for detection, which is not enough protein to detect other proteins. In order to accurately quantify both MyHC and 100

113 other cytoplasmic proteins in the same muscle, it is necessary to lyse portions of the same muscle in different buffers and use the MEB lysate for MyHC and actin analysis and the LB lysate for all other proteins. The solubility properties of myosin have been known for decades but have been largely ignored in many studies of muscle atrophy, likely because one tissue lysate was used to quantify all proteins in the cell. The method described here allows for accurate comparison of MyHC levels between different sized muscles and provides physiologically relevant results. Most importantly, we have shown that MyHC is not specifically decreased in atrophic muscle. We believe that interpreting results on MyHC levels in low-salt lysates should be done with extreme caution, as they could be artificially low. 101

114 CHAPTER IV. Interferon-γ causes cardiomyocyte atrophy and the specific degradation of myosin heavy chain in vitro. Introduction Cachexia is a syndrome characterized by excessive muscle mass loss and occurs in patients suffering from inflammatory diseases such as heart failure, chronic obstructive pulmonary disease, chronic kidney disease, and cancer (Morley, Thomas et al. 2006). This wasting state is largely mediated by pro-inflammatory cytokines such as tumor necrosis factor α (TNF-α), interleukin-6 (IL-6), and interferon-γ (IFN-γ), which initiate complex metabolic alterations leading to increased tissue catabolism (Tisdale 1997). These cytokines are implicated in muscle atrophy as they alone or in combination, can cause muscle mass and protein loss in vivo (Tisdale 1997). Decreased muscle mass increases susceptibility to infections and can lead to cardiac or respiratory failure resulting in decreased prognosis and survival (Tisdale 1997). Cachexia is a causative factor in 30-50% of deaths in patients with gastrointestinal cancer, and up to 80% of deaths in patients with pancreatic cancer (Bachmann, Heiligensetzer et al. 2008). Cardiac muscle also atrophies in humans and rodents with cancer (Burch, Phillips et al. 1968; Zhou, Wang et al. 2010). We, and others, have observed that tumor-bearing male mice lose approximately 20% of their cardiac mass in one month (Zhou, Wang et al. 2010; Cosper and Leinwand 2011). While the mechanisms of skeletal muscle atrophy and cachexia have been extensively studied in vivo and in vitro (Guttridge, Mayo et al. 2000; Acharyya, Ladner et al. 2004; Clarke, Drujan et al. 2007), equivalent studies on cardiac myocytes are lacking. Because elevated serum levels of pro-inflammatory cytokines are 102

115 the hallmark of cachexia and play a key role in muscle wasting (Argiles, Busquets et al. 2003), most studies employ cytokines to mimic the inflammatory response to a tumor or infection in vitro. For example, TNF-α and IFN-γ synergistically cause a specific decrease in myosin heavy chain (MyHC) mrna and protein in skeletal myocytes in vitro, implicating them in the pathogenesis of cancer cachexia (Acharyya, Ladner et al. 2004). Pro-inflammatory cytokines induce muscle atrophy and protein loss by activating proteolytic pathways that increase degradation of muscle proteins (Lecker, Solomon et al. 1999; Morley, Thomas et al. 2006; Ventadour and Attaix 2006). Skeletal and cardiac muscle utilize three major proteolytic pathways: the lysosome, the Ca 2+ -dependent calpains, and the ubiquitin proteasome system (UPS). The UPS degrades approximately 80-90% of all intracellular proteins (Lee and Goldberg 1998), including myofibrillar proteins, and it is the main proteolytic pathway involved in skeletal muscle atrophy (Lecker, Solomon et al. 1999). The proteolysis of skeletal muscle proteins has been wellstudied but equivalent models in cardiac myocytes have not been reported, which is surprising given that cardiac atrophy has been recognized as an important clinical feature of inflammatory diseases such as HIV and cancer (Pruznak, Hong-Brown et al. 2008; Zhou, Wang et al. 2010). IFN-γ (hereafter referred to as IFN) is directly implicated in cancer cachexia as mice inoculated with CHO cells engineered to produce IFN develop a severe cachexia, while administration of anti-ifn antibodies to mice with Lewis lung carcinoma prevents body weight loss (Matthys, Dijkmans et al. 1991; Matthys, Heremans et al. 1991). IFN is produced by lymphocytes and natural killer cells as part of the adaptive immune 103

116 response. One of IFN s main functions is to activate macrophages and increase the expression of major histocompatibility complex (MHC) I and II and to increase the diversity of peptides displayed on these molecules (Schroder, Hertzog et al. 2004). Although IFN s primary function is in the acute immune response, chronically high IFN levels, that occur in patients with cancer or infections, can have deleterious effects on many tissues, including skeletal and cardiac muscle. Patients with pancreatic and colon cancer, two cancers that cause severe muscle atrophy, have significantly elevated IFN levels, which were associated with higher resting energy expenditure and weight loss (Ravasco, Monteiro-Grillo et al. 2007; R, A et al. 2009). There have been very few studies on the direct effects of IFN on cardiac myocytes. IFN is induced in cardiac muscle during sepsis and slightly increases cardiac MyHC mrna expression in adult rat cardiac myocytes (Patten, Kramer et al. 2001). IFN is also cholinergic and decreases cardiac contractility (Borda, Leiros et al. 1991; Sun, Delbridge et al. 1998). Unlike skeletal muscle, there are no studies regarding IFN and cardiac atrophy with respect to cancer cachexia, but there is evidence that IFN plays a role in regulating cardiac muscle mass. Transgenic mice with high serum levels of IFN undergo severe cardiomyopathy with associated cardiac myofiber atrophy (Reifenberg, Lehr et al. 2007). IFN also partially reverses cardiac hypertrophy in vivo and in vitro but short-term treatment with IFN alone does not affect heart weight (Jin, Li et al. 2005). Because the mechanisms of pro-inflammatory cytokine induced cardiac atrophy are unknown, we established an in vitro model using neonatal rat ventricular myocytes (NRVM). Our results show that IFN causes significant myocyte atrophy and the selective degradation of MyHC in a proteasome-dependent manner. This model may be applicable 104

117 not only to cancer-induced cardiac atrophy, but also to inflammatory diseases in which the heart is exposed to high levels of IFN, such as autoimmune myocarditis and Chagas disease. Materials and Methods Cell culture Neonatal rat ventricular myocytes (NRVM) were isolated and cultured as previously described (Waspe, Ordahl et al. 1990). Briefly, freshly isolated cardiomyocytes from 1- day old rats were placed on gelatin-coated plates in MEM medium with 5% calf serum and BrdU to prevent non-cardiomyocyte proliferation. 24 hours later, the media was replaced with serum-free MEM supplemented with 20 mm Hepes, vitamin B 12, BrdU, and 0.01 mg/ml BSA, insulin and transferrin. After an 18-hour recovery period, cells were treated with PBS or 200 U/mL IFNγ (Calbiochem). Proteasome inhibitors used were: 1 µm MG-132 (Boston Biochem), 5 nm Bortezomib (Velcade) for 48 hours. Bortezomib was kindly provided by Millenium Pharmaceuticals, Inc. Lysosomal inhibitors were: 10 mm NH 4 Cl (Sigma), and 2.5 mm 3-methyladenine (3-MA). This concentration was chosen as in (Kobayashi, Volden et al. 2010) to prevent potential secondary effects that occur with higher concentrations. Myofibrillar gels and Western blots For MyHC analysis, NRVMs were lysed in high salt myosin extraction buffer (MEB)(Butler-Browne and Whalen 1984). Lysates were spun at 12,000 rpm for 15 min at 4 C. The supernatant was removed, and protein concentration determined (Biorad DC 105

118 protein assay). 2.5 µg of protein was loaded per sample onto myofibrillar gels, which are composed of: 15% acrylamide (124:1 acrylamide: bis), 9.79% glycerol, 0.73 M Tris ph 9.3, and 0.1% SDS. The stacking gel was composed of 2.95% acrylamide (5.7:1), 0.13 M Tris ph 6.8, 10% glycerol, and 0.1% SDS. Gels were run at 16 ma for six hours at 4 C and silver stained (Biorad Silver Stain Plus). Gels were scanned and band densities were quantified using ImageJ. For MyHC western blots, antibodies were: anti-α-myhc (BA- G5 hybridoma from ATCC, 1:5) and anti-β-myhc (VP-M667 Vector Laboratories 1:400). For LC3-II analysis, NRVMs were lysed in 30 mm HEPES, 20 mm KCl, 2 mm NaF, 5 mm EDTA and 1% Triton-X100. Lysates were set to rotate at 4 C, then spun at 12,000 rpm for 10 minutes. The supernatant was removed, and protein concentration determined (Biorad DC protein assay). 15 µg of protein was loaded onto a 4-20% gradient gel (Pierce), then transferred to a PVDF membrane, blocked for 30 minutes in 5% NFDM, and primary antibody (LC3b, 1:1000 Cell Signaling) overnight. Immunoreactivity was visualized using a Western Lighting chemiluminescence detection system (Perkin Elmer) and quantified using ImageJ. Immunocytochemistry NRVMs grown on coverslips were fixed in 4% para-formaldehyde for 20 minutes, then rinsed and stored in 1 mm Ca ++ in PBS. Cells were permeabilized with 0.1% Triton X- 100, blocked with 10% horse serum, and stained with anti-α-actinin (1:200) antibody overnight at 4 C. Goat-anti-mouse-FITC secondary antibody was used at 1:100 for one hour. Digitized images of cells (3 fields per condition in triplicate) were analyzed using ImageJ software. 106

119 Cell volume determinations After 48 hours of IFN treatment, NRVMs were trypsinized, and resuspended in PBS containing 1% bovine-calf serum and 10 mm EDTA. Cell volume was measured in a coulter counter (Beckman Coulter) that provides the median cell volume per sample. Each sample was in triplicate and the experiment was repeated twice. RNA analysis Total RNA was purified from NRVMs using TRIzol reagent (Molecular Research Center, Inc). First strand cdna was synthesized using Superscript II reverse transcriptase (Invitrogen) and random hexamer primers. Gene expression was then determined by quantitative real-time reverse transcription polymerase chain reaction (qrt-pcr) using SYBR Green dye with gene specific primer sets and Applied Biosystems 7500 Real-Time PCR system. Primer sequences are listed in Appendix I. Caspase-3 and Proteasome activity assays Caspase-3 activity in cardiac lysates was determined by monitoring the rate of cleavage of a fluorogenic caspase-3 specific substrate (Acetyl-Asp-Glu-Val-Asp-AMC; Calbiochem). NRVMs were scraped in PBS, pelleted, and lysed in 70 µl of caspase lysis buffer (20 mm Tris, ph 7.0, 5 mm EDTA, 1 mm EGTA, 150 mm NaCl, 1% Triton X- 100), rotated for 20 minutes at 4 C, and spun at 14,000 rpm for 10 minutes. 50 µl of each sample was added to 50 µl of ICE buffer (50 mm Tris, ph 7.0, 0.5 mm EDTA, 20% glycerol) in an opaque 96-well plate. DTT was added to a final concentration of 4 mm and final substrate concentration was 20 µm. Data was obtained every 2 minutes for 107

120 2 hours at 37 C. Caspase activity was calculated as the slope of the linear portion of the line divided by the sample s protein concentration. Proteasome activity was determined as described (Powell, Davies et al. 2007) except ATP was not added to the reaction mixture because we found that it significantly decreased activity. Activity was determined using 15 µg of protein and 20 µm substrate. The substrate for proteasome activity was Suc-LLVY-AMC (Boston Biochem) and the substrate for immunoproteasome activity was Ac-PAL-AMC (Millenium Pharmaceuticals). 5 nm bortezomib was used to inhibit the proteasome to ensure specificity of the assay. Data was obtained every 2 minutes for one hour at 37 C. Proteasome activity was calculated as the slope of the linear portion of the line for each sample. Pulse and pulse-chase assays To study protein synthesis, NRVMs were treated with 200 U/mL IFNγ for 46 hours. Cells were washed and incubated with 3 H-Tyrosine (4 µci/ml) for 2 hours. Cells were collected in 10% TCA and the incorporated radioactivity was determined using a scintillation counter. Protein synthesis is represented as the number of counts normalized to protein concentration per well. Protein degradation rates in NRVMs were determined as previously described (Zhao, Brault et al. 2007), with the exception that after the initial 24-hour labeling period, cells were treated with IFNγ for 24 hours in medium containing 2 mm unlabeled tyrosine. Cells were then treated with either proteasome (1 µm MG-132) or lysosomal inhibitors (10 mm NH 4 Cl or 100 µm chloroquine) for 2 hours prior to collecting media samples. In other experiments, cells were labeled with S 35 for 24 hours 108

121 and then treated with IFN for 24 hours in the presence of 2 mm unlabeled methionine. Cellular proteins were extracted in MEB, and run on myofibrillar gels as above. Gels were dried, autoradiography was performed, and scanned. All experiments were performed three times and measurements were performed in triplicate. Data and statistical analysis Data are presented as mean ± SEM. Differences between groups were evaluated for statistical significance using Student s t-test. P values less than 0.05 were considered significant. Results IFN-γ causes myocyte atrophy and a decrease in cardiac MyHC in vitro. Previous studies on skeletal muscle cancer cachexia have shown that in vitro, the pro-inflammatory cytokines TNF-α and IFN mimic the inflammatory response to a tumor and recapitulate the atrophic phenotype observed in vivo (Acharyya, Ladner et al. 2004). We found that IFN alone could cause the degradation of fast-twitch MyHC in myotubes (unpublished observations). We therefore chose to work with IFN alone in order to avoid the complex effects TNF-α on cardiomyocytes. To determine if IFN treatment could cause cardiac myocyte atrophy and induce MyHC loss, we treated NRVMs with IFN for 48 hours. We chose to use 200 U/mL of IFN, which is a similar concentration used in studies on skeletal muscle myotubes and rat cardiomyocytes (Stephanou, Brar et al. 2000; Patten, Kramer et al. 2001; Acharyya, Ladner et al. 2004). 109

122 In order to determine if IFN causes cardiac myocyte atrophy, we quantified cell size using two different methods; immunocytochemistry to measure cell area and a coulter counter to determine cell volume. Both methods revealed that 48-hour IFN treatment causes a significant decrease in myocyte size. Immunocytochemistry for α- actinin revealed a 26% decrease in myocyte surface area (Figure 21A), while cell volume measurements on live cells revealed a significant 9% decrease (Figure 21B). IFN-induced cardiac myocyte atrophy is associated with a 58% decrease in total MyHC (Figure 22A). Cardiac muscle expresses two MyHC isoforms, α- and β-myhc. β-myhc is expressed during fetal development and an increase in its expression is considered pathologic in adult rodent hearts. Unlike adult rodents, NRVMs express α- and β-myhc equally. We therefore wanted to determine if IFN caused a preferential decrease in one MyHC isoform over the other. Western blots for the two cardiac MyHC isoforms revealed an equivalent decrease in both α- and β-myhc protein (Figure 22A). This decrease in MyHC is post-transcriptionally mediated since α-myhc and β-myhc mrna levels do not significantly decrease (Figure 22B). Interestingly, β-myhc mrna increases 1.3-fold after 24 hours of treatment. To ensure that this decrease in MyHC was specific to IFN, we treated NRVMs with TNF-α or dexamethasone, a glucocorticoid that induces both cardiac and skeletal muscle atrophy (Clarke, Drujan et al. 2007; Willis, Rojas et al. 2009). Neither treatment causes a decrease in MyHC at high or low concentrations, implying that IFN is unique in inducing a decrease in MyHC (Figure 22C). In order to determine if IFN causes a decrease in other myofibrillar proteins, we analyzed the myofibrillar fractions of treated NRVMs. SDS-PAGE gels revealed that IFN treatment causes a 58% decrease in MyHC 110

123 Figure 21. IFN causes cardiac myocyte atrophy. (A) Immunocytochemistry of NRVMs for α-actinin. At least 50 cells were quantified per group as represented in the bottom histogram. (B) Myocyte volume, as determined by a coulter counter, also decreased in the IFN treated group. (C) IFN does not cause an increase in caspase-3 activity, and is therefore not affecting myocyte viability. Graphs are mean ± SEM. * p < 0.05, ***p <

124 Figure 22. IFN causes a specific decrease in MyHC protein. (A) A silver-stained SDS- PAGE gel and immunoblots reveal a significant decrease in both isoforms of MyHC in IFN treated NRVMs. (B) α- and β-myhc gene expression after 24 and 48 hours of treatment as determined by qrt-pcr. Gene expression was normalized to histone 3B. n = 5 per group. (C) Western blot for MyHC shows that only IFN causes a decrease in MyHC protein. (D) SDS-PAGE of NRVM myofibrils revealed that only MyHC is decreased in the IFN treated group, whereas all other myofibrillar proteins are not changed. n = 3 per group. Each experiment was performed in triplicate. Graphs are mean ± SEM. **p < 0.01, ***p <

125 Figure 23. SDS-PAGE of NRVM myofibrils revealed that only MyHC is decreased in the IFN-treated group, whereas all other myofibrillar proteins are not changed. 113

126 levels, while all non-myosin myofibrillar proteins; actin, troponin, tropomyosin, and myosin light chains 1 and 2 are unaffected (Figure 23). Myosin/actin ratios therefore consistently decrease while all other myofibrillar protein ratios do not significantly change. IFN treatment does not affect myocyte viability (Jin, Li et al. 2005). We have confirmed this by measuring caspase-3 activity and surprisingly, found a significant decrease upon IFN treatment (Figure 21C). Additionally, there was no increase in expression of the pathological markers ANF or BNP (data not shown). Thus, IFN treatment causes both NRVM atrophy and a specific decrease in MyHC protein without causing cell death. Cardiomyocyte atrophy is not due to changes in total protein synthesis or degradation. Cancer-induced skeletal muscle atrophy is primarily due to cytokine-mediated muscle proteolysis (Temparis, Asensi et al. 1994), although decreased protein synthesis can also occur (Smith and Tisdale 1993). Because IFN treatment causes a decrease in MyHC protein but does not decrease MyHC mrna levels, we determined whether this was due to a decrease in protein synthesis, an increase in protein degradation, or both. We performed a pulse assay with 3 H-tyrosine in order to study changes in total protein synthesis upon IFN treatment. Interestingly, 48-hour IFN treatment does not affect total protein synthesis rates (Figure 24A). Since IFN does not affect protein synthesis, we reasoned that the decrease in MyHC must be due to enhanced protein degradation. In order to determine total protein degradation rates, we performed pulse-chase assays in which myocyte proteins were 114

127 labeled with 3 H tyrosine for 24 hours, then treated with IFN in media containing unlabeled tyrosine. 3 H tyrosine release was measured in the medium 24 hours after IFN treatment and surprisingly, there was no change in total protein degradation rates compared to PBS treated cells (Figure 24B). To ensure that 24-hour IFN treatment is sufficient to cause a decrease in MyHC, we performed myofibrillar gels as in Figure 1 and found an equivalent decrease in MyHC as in 48-hour treatment (data not shown). The fact that total protein degradation rates do not change is further evidence that IFN causes the specific degradation of myosin while other cellular proteins are spared. We also co-treated myocytes with either a proteasome inhibitor (MG-132) or lysosomal inhibitor (NH 4 Cl) (Figure 24B) in order to determine the role that each pathway plays in normal and IFN-induced protein turnover. Consistent with the observation that IFN alone does not affect protein degradation rates, co-treatment with MG-132 or NH 4 Cl also did not affect protein degradation when compared to control cells treated with either proteolysis inhibitor. Additionally, the results of these pulse-chase assays revealed that the UPS is the major degradative pathway active in NRVMs; inhibiting the proteasome reduces total protein degradation by 72%, while inhibiting the lysosome only decreases it by 31% (Figure 24C). We also repeated these experiments with chloroquine, another lysosomal inhibitor that raises intralysosomal ph (Seglen 1983), and did not find any difference in protein degradation rates in the IFN group (data not shown). In order to verify that MyHC is specifically targeted for degradation, we labeled NRVM proteins with S 35 methionine for 24 hours, then treated cells for 24 hours with IFN in media containing unlabeled methionine. Cell extracts were run on myofibrillar 115

128 Figure 3. IFN treatment does not change total protein synthesis or degradation rates in NRVMs. (A) Control or IFN-treated NRVMs were pulsed with 3 H-tyrosine for two hours prior to collection of cells. Protein synthesis is represented as the amount of radioactivity incorporated into cells and normalized to the total protein concentration per well. (B) Pulse-chase analysis of NRVMs treated with IFN with or without the proteasome inhibitor, MG132, or a lysosomal inhibitor, NH 4 Cl. Cells were labeled for 24 hours with 3 H-tyrosine, then treated with IFN for 24 hours in media with an excess of cold tyrosine. Protein degradation is represented as the percent of radioactivity released into the media from the total 3 H-tyrosine incorporated into cellular proteins. (C) The contribution of the UPS and the lysosomal pathway in basal NRVM protein degradation. n = 3 per group, and each experiment was performed in triplicate. Graphs are mean ± SEM. ***p<

129 Figure 25. S 35 labeling experiments in atrophic NRVMs. NRVMs were allowed to incorporate S 35 -methionine for 24 hours and were then treated with IFN or PBS for 24 hours in media with an excess of unlabeled methionine. SDS-PAGE of lysates and autoradiography revealed a specific decrease in MyHC in IFN-treated NRVMs. n = 3 per group, and each experiment was performed in triplicate. Graphs are mean ± SEM. ***p<

130 gels and autoradiography was performed in order to visualize the specific proteins that are degraded upon treatment. As hypothesized, MyHC is selectively decreased in the IFN-treated myocytes (Figure 25). Thus, IFN causes the specific degradation of MyHC and does not cause a general increase in protein turnover. Cardiac MyHC is degraded by the ubiquitin proteasome system in vitro. The UPS is the main proteolytic pathway implicated in skeletal muscle atrophy and inhibition of the UPS, but not other proteolytic pathways, decreases tumor-induced proteolysis (Baracos, DeVivo et al. 1995). In order to determine if the UPS is involved in IFN-induced cardiac MyHC degradation, we co-treated NRVMs with IFN and MG-132 and quantified MyHC levels. MG-132 treatment completely prevents MyHC loss in IFN treated NRVMs and increases MyHC levels slightly above that of control myocytes (Figure 26A). We also co-treated NRVMs with Bortezomib, which is a more specific and potent inhibitor of the chymotryptic activity of the proteasome. Bortezomib treatment also prevents the degradation of MyHC in IFN-treated cells (Figure 26A, bottom panel). In order to determine if proteasome inhibition could also prevent IFN-induced myocyte atrophy, we measured cell size after 48 hours of IFN treatment with or without Bortezomib. Bortezomib treatment prevents myocyte atrophy and surprisingly, caused a significant increase in myocyte size over control values (Figure 26B). Because we found that the UPS degrades MyHC, and that the UPS is the main proteolytic pathway active in NRVMs, we wanted to determine if IFN affects UPS activity. Proteasome activity assays revealed 1.6-fold higher activity after 24 hours of treatment and 2.4-fold higher activity after 48 hours (Figure 27A). The inner proteolytic 118

131 Figure 26. IFN causes MyHC degradation in a proteasome-dependent manner. (A) NRVMs were treated with IFN or PBS in the presence of either 0.1% DMSO or 1 µm MG-132 for 48 hours. SDS-PAGE of lysates reveals that proteasome inhibition restored MyHC levels without affecting the levels of other myofibrillar proteins. Bortezomib treatment also fully restored MyHC levels in IFN treated cells (bottom panel). (B) Bortezomib treatment prevented IFN-induced myocyte atrophy and caused an increase in cell size in both control and IFN-treated groups. * significant compared to control in each group, significant compared to the non-bortezomib group. For protein data, n = 3 per group with each experiment performed in triplicate. Graphs are mean ± SEM. **p<0.01, ***p< vs. control. 119

132 core of the proteasome is composed of two rings of β-subunits which contain chymotryptic, tryptic, and caspase-like activity (Powell 2006). IFN causes the replacement of three of these β-subunits with corresponding immunoforms; LMP2, LMP7 and MECL-1 (Schroder, Hertzog et al.). The incorporation of these inducible subunits collectively forms the immunoproteasome, which favors formation of peptide fragments for display on major histocompatibility antigens during an immune response. Using a substrate that has 10-fold higher specificity for the immunoproteasome over the proteasome, we found a striking 8-fold increase in immunoproteasome activity after 48 hours of IFN treatment, while it increased only 2.7-fold after 24 hours (Figure 27A). The fold induction of immunoproteasome activity is significantly higher than proteasome activity induction at each time point. Immunoproteasomes are upregulated in cardiac muscle of mice with diabetes and coxsackievirus myocarditis and have been shown to play a role in decreasing cardiac mass (Szalay, Meiners et al. 2006; Zu, Bedja et al. 2010). The immunoproteasome is therefore likely involved in IFN-induced NRVM atrophy and MyHC degradation. Atrogin-1 and MuRF-1 are muscle specific E3-ubiquitin ligases, which are transcriptionally upregulated during skeletal muscle atrophy and mark specific substrates for degradation by the proteasome (Lecker, Jagoe et al. 2004). Atrogin-1 has been shown to prevent cardiac hypertrophy (Li, Kedar et al. 2004) and cause cardiomyocyte atrophy (Yamamoto, Hoshino et al. 2008) and MuRF-1 has been implicated in cardiac atrophy in vivo (Willis, Rojas et al. 2009). We hypothesized that these ligases were involved in IFNinduced cardiomyocyte atrophy because they are known to target the degradation of myofibrillar proteins, including β-myhc, in a proteasome dependent manner. We 120

133 Figure 27. IFN causes an increase in proteasome activity. (A) Both proteasome and immunoproteasome activity increase in cardiomyocytes treated with IFN for 48 hours. Bortezomib is a proteasome inhibitor and decreased activity 98% (B) Atrogin-1 and MuRF-1 expression did not change in atrophic NRVMs at either time point, as determined by qrt-pcr. Gene expression data was normalized to histone 3B. n = 4 per group. Experiments were repeated at least twice. 121

134 measured atrogin-1 and MuRF-1 mrna levels in cardiac myocytes after both 24 and 48 hours of IFN treatment and did not find an induction of either mrna at either time point (Figure 27B). Although atrogin-1 and MuRF-1 expression increase in atrophying skeletal muscle (Lecker, Jagoe et al. 2004), they do not increase in atrophying cardiac muscle in vivo (Sharma, Ying et al. 2006; Zhou, Wang et al. 2010; Cosper and Leinwand 2011), which is consistent with these in vitro results. IFN does not cause increased autophagy or lysosome mediated MyHC degradation. The lysosome is responsible for approximately 25% of cardiac protein turnover (Wildenthal and Wakeland 1985). We have shown that the autophagosomal-lysosomal pathway is upregulated in the atrophic hearts of tumor-bearing mice (Cosper and Leinwand 2011) and wanted to determine if this proteolytic pathway is also involved in IFN induced MyHC degradation and myocyte atrophy. We co-treated NRVMs with NH 4 Cl and IFN and analyzed the levels of MyHC and other sarcomeric proteins. Inhibition of the lysosome in cardiac myocytes does not prevent MyHC loss to any degree (Figure 28A), which indicates that the lysosome is not involved in MyHC degradation in this model. To determine if autophagy is involved in MyHC degradation, we inhibited autophagy initiation with 3-methyladenine (3-MA), which blocks autophagosome formation (Seglen and Gordon 1982). 3-MA treatment also does not prevent MyHC loss (Figure 28B) implying that autophagy is not responsible for IFNinduced MyHC degradation in vitro. To determine if IFN causes an increase in autophagic vacuole formation, we quantified the levels of lipidated microtubule associated light chain 3 (LC3-II), which is a 122

135 Figure 28. MyHC degradation is not mediated by the lysosome/autophagosome pathway. (A) SDS-PAGE of NRVMs treated with IFN and a lysosomal inhibitor, NH 4 Cl, revealed an equivalent decrease in MyHC as IFN treatment alone. (B) Inhibiting autophagosome formation with 3-MA also does not prevent MyHC loss. (C) Western blot for LC3-II does not reveal an increase with IFN treatment. Rapamycin (Rap) is a known inducer of autophagy and was used as a control to show increased LC3-II expression. Graphs are mean ± SEM. ***p< vs. control. 123

136 protein specifically localized to autophagic vacuoles and is a marker of increased autophagic activity (Barth, Glick et al. 2010). IFN alone does not increase LC3-II levels, while rapamycin, a known inducer of autophagy, causes a 2-fold increase in LC3-II (Figure 28C). Together, these results show that the autophagolysosome is not involved in IFN-induced cardiac MyHC degradation. We previously reported that tumor-bearing mice have 8.8-fold higher levels of IFN than control mice 15 days after tumor-cell inoculation (Cosper and Leinwand 2011). The serum concentrations of IFN in these mice are comparable to those in human pancreatic cancer patients (R, A et al. 2009), but are in the low pg/ml range, which is 1000-fold less than the concentrations used in our studies. We therefore determined whether physiological levels of IFN would also cause MyHC degradation to ensure the validity of our model. NRVMs treated with 17 pg/ml of IFN lose MyHC to the same extent as myocytes treated with the higher dose (Figure 29A). These results indicate that the IFN levels present in the serum of cancer patients is sufficient to cause MyHC loss in vitro. Many cell types regulate expression of the IFN receptor in order to adapt their response to this cytokine. To determine if regulation of the IFN receptor occurs in cardiac myocytes of tumor-bearing mice, we measured the transcript levels during the early and late stages of atrophy. IFN receptor mrna significantly decreases early in cardiac atrophy when serum IFN levels are high (Figure 29B). Accordingly, as IFN levels decrease over the course of disease, IFN receptor expression in atrophic cardiac muscle returns to baseline. IFN receptor expression is also regulated in treated NRVMs. The IFN receptor decreases in NRVMs after 48 hours of treatment but its expression does not 124

137 change after 24 hours (Figure 29C). Thus, exposure to IFN causes a decrease in IFN receptor expression both in vivo and in vitro, perhaps to protect against further cellular damage. Discussion IFN is directly implicated in cancer-induced muscle atrophy (Matthys, Dijkmans et al. 1991), though whether IFN causes similar effects on cardiac muscle is unknown. We developed a novel, in vitro model for IFN-induced cardiomyocyte atrophy and cardiac MyHC degradation. Here, we have show that IFN has significant adverse effects on cardiac myocytes, namely the proteasome-dependent specific degradation of MyHC and consequential myoycte atrophy. This model is applicable to many diseases in which there are high circulating IFN levels, such as cancer, sepsis and myocarditis. IFN is initially beneficial and protective as it stimulates antitumoral and antimicrobial activities in macrophages as well as leukocyte attraction, growth and differentiation (Schroder, Hertzog et al. 2004). However, chronically high levels of IFN that occur in cancer, myocarditis and Chagas disease can adversely affect the heart. Numerous studies have shown that IFN decreases contractility in vivo and in vitro, and has been shown to be an important pathological mediator in autoimmune myocarditis (Borda, Leiros et al. 1991; Perez Leiros, Goren et al. 1997). IFN also plays important protective roles in cancer by stimulating tumoricidal pathways (Schroder, Hertzog et al. 2004). However, chronic IFN also contributes to immunodeficiency, weight loss and an overall poor prognosis in cancer patients (Brandacher, Winkler et al. 2006). Not only is IFN increased in cancer patients, but it has 125

138 Figure 29. IFN concentrations present in the serum of tumor-bearing mice are sufficient to cause MyHC loss. (A) SDS-PAGE of NRVM proteins treated with physiological concentrations of IFN, which causes a decrease in MyHC. (B) IFN receptor (IFN-R) mrna expression is initially downregulated in the hearts of tumor-bearing mice as determined by qrt-pcr. Gene expression is normalized to 18S. n = 4 per group. (C) IFN receptor (IFN-R) mrna expression after 24 hours and 48 hours of IFN treatment in NRVMs. Gene expression is normalized to histone 3B. * p < 0.05, ** p < Graphs are mean ± SEM. 126

139 also been used to treat certain types of cancer. Studies on humans receiving IFN therapy have observed significant cardiovascular adverse events including cardiotoxicity and abnormalities on electrocardiogram (Sriskandan, Garner et al. 1986; Mattson, Niiranen et al. 1991). The findings in our study could explain these adverse events and provides additional relevance to human disease. We have shown that IFN stimulates the selective degradation of cardiac MyHC protein without affecting mrna levels. These results correlate with findings in atrophic cardiac muscle (Cosper and Leinwand 2011), and suggest that changes in cardiac MyHC expression due to increased circulating cytokines occur post-transcriptionally. Since MyHC comprises a large proportion of total NRVM proteins by mass, we expected to find an increase in protein degradation rates, but pulse-chase analysis did not reveal any differences in total protein degradation rates between control and IFN treated myocytes. NRVMs were only treated for 24 hours in the pulse-chase analysis and we showed that proteasome and immunoproteasome activity increased only slightly at 24 hours but increased several fold more after 48 hours of treatment. Thus, total protein degradation may increase at later time points or it is possible that our assay was not sensitive enough to detect a small, MyHC specific, increase in protein turnover. Pulse-chase experiments confirmed the primary role of the UPS in normal cardiac myocyte protein turnover and led us to believe that the proteasome was a likely candidate for MyHC degradation. IFN is also known to upregulate components of the UPS in skeletal muscle leading to increased protein degradation (Llovera, Carbo et al. 1998) and components of the UPS are elevated in the skeletal muscles of cancer patients, and in rodent models of sepsis and cancer (Tiao, Fagan et al. 1994; Baracos, DeVivo et al. 1995; 127

140 Khal, Hine et al. 2005). As expected, we found increased proteasome activity in IFN treated myocytes and determined that inhibition of the proteasome completely prevented MyHC loss. These findings correlate with the mechanisms established in skeletal muscle atrophy but not in cardiac muscle atrophy in vivo. We reported elsewhere that the atrophic hearts of tumor-bearing mice have increased autophagy, rather than increased UPS activity, and that MyHC decreases in parallel with other myofibrillar proteins (Cosper and Leinwand 2011). The proteolytic pathways upregulated and the myofibrillar proteins lost during cardiac atrophy in vivo are therefore different than those in vitro. There are many possible explanations for this. First, tumor-bearing mice have increased circulating levels of many different pro-inflammatory cytokines, while in vitro, cardiac myocytes are only exposed to IFN. Pro-inflammatory cytokines often signal synergistically and different combinations are likely to have different effects on the heart. Second, cancer causes an increase in resting energy expenditure (Tisdale 1997) and the hearts of tumor-bearing mice could be in an energy-deprived state, leading to upregulation of autophagy. Third, the heart is likely exposed to numerous, as yet unknown, factors either secreted by the tumor directly or by the immune system that may play a role in activating autophagy rather than the UPS. IFN caused a striking increase in immunoproteasome activity in cardiac myocytes. Immunoproteasomes have altered cleavage site preferences and protein cleavage rates in order to increase antigenic peptide presentation. Immunoproteasomes may therefore have increased affinity for MyHC and may be responsible for its degradation in this model. In fact, mice immunized with cardiac α-myhc peptide 128

141 fragments develop severe myocarditis (Caforio, Mahon et al. 2002), implying a potential positive feedback loop where inflammation and high IFN levels increase MyHC peptide fragment generation resulting in a continued immune response and further inflammation and IFN secretion. There is evidence that immunoproteasomes regulate cardiac muscle mass (Zu, Bedja et al. 2010), thus it is possible that increased immunoproteasome activity in our model is also responsible for the myocyte atrophy we observed. Additionally, transgenic mice that have constitutively high levels of circulating IFN experience chronic myocarditis and severe cardiomyopathy that is characterized by significant cardiac myofiber atrophy, fibrosis, and decreased cardiac function (Reifenberg, Lehr et al. 2007). Because IFN induces myocarditis (Reifenberg, Lehr et al. 2007) and myocarditis is associated with increased cardiac immunoproteasome activity (Szalay, Meiners et al. 2006), immunoproteasome-mediated MyHC degradation may be the causative factor in the transition from myocarditis to cardiomyopathy. Our model therefore provides new insights into the molecular mechanisms responsible for the cardiac phenotypes observed in many inflammatory diseases. Atrogin-1 and MuRF-1 are well-characterized markers of skeletal muscle atrophy and have also been implicated in regulating cardiac myocyte size. Neither atrogin-1 nor MuRF-1 mrnas were upregulated in atrophic NRVMs, which correlates with cardiac atrophy in vivo (Zhou, Wang et al. 2010; Cosper and Leinwand 2011). Because cardiac muscle expresses approximately 3-fold higher levels of atrogin-1 and MuRF-1 than skeletal muscle (Cosper and Leinwand 2011), it is possible that the baseline levels of 129

142 these ligases are sufficient to forgo the need for upregulation, or that cardiac muscle utilizes ubiquitin ligases distinct from skeletal muscle during atrophy. Together, these results provide evidence that IFN causes cardiac pathology in vitro. It remains to be determined if IFN also causes the specific degradation of MyHC in vivo. Patients suffering from myocarditis and sepsis who have high levels of proinflammatory cytokines, including IFN, often develop cardiomyopathy and decreased function (Granton, Goddard et al. 1997). IFN-induced cardiac MyHC loss could be occurring in these patients, as this would likely cause the decreased contractility and cardiac function observed clinically. Our in vitro model of cardiac atrophy and MyHC degradation could therefore provide insight into the molecular mechanisms underlying many human cardiac inflammatory pathologies, including cardiac atrophy due to cancer, chronic myocarditis, sepsis, and Chagas disease. 130

143 CHAPTER V. The effects of sex and diet on cancer survival and cachexia Introduction Greater than 1.5 million people were estimated to be diagnosed with cancer in 2010 and over 500,000 people were expected to succumb to it (Jemal, Siegel et al. 2010). Additionally, this study reported that men have a higher incidence of all types of cancer (except breast) and greater mortality than women. Analysis of 1.6 million cases of 26 forms of cancer revealed that overall, women have a significantly lower risk of death but that this female advantage decreases with age (Micheli, Ciampichini et al. 2009). The basis for sex-dependent differences in cancer incidence and mortality has been well studied and it appears that the female hormone, estrogen, plays an important role in preventing women from developing certain kinds of cancer and in increasing their overall survival. Colorectal cancer (CRC) is one of the most common forms of cancer. The agestandardized incidence rates of CRC are higher in men than women, especially in developed countries such as the USA, Australia, and the UK (Koo and Leong 2010). Several studies have found that women develop CRC approximately 10 years later than men (Brenner, Hoffmeister et al. 2007). The age-standardized mortality rate is also consistently higher in men than women. However, many prospective studies have not found associations between sex and survival in CRC, but this is likely due to confounding variables, such as co-morbidities (Deans, Patterson et al. 1994). After correcting for 131

144 confounders, six out of fourteen studies did detect a survival advantage in women (Koo and Leong 2010). Additionally, when survival data are analyzed in a cancer-specific (i.e. site, stage, grade) manner, women have improved survival (Wichmann, Muller et al. 2001; McArdle, McMillan et al. 2003) and also survive longer after surgical resection of the tumor (Paulson, Wirtalla et al. 2009). Interestingly, women less than 50 years of age have significantly improved survival compared to men in the same age group, but this survival advantage is reversed in older women (Koo, Jalaludin et al. 2008; Hendifar, Yang et al. 2009). The authors of these studies postulated that estrogen may be responsible for the survival advantage in younger women and that the decreases in estrogen that occur during menopause may contribute to the loss of this survival advantage. Further evidence that estrogen is involved in preventing CRC is that postmenopausal women taking hormone replacement therapy have a 30-40% reduction in CRC incidence (al-azzawi and Wahab 2002; Chlebowski, Wactawski-Wende et al. 2004). Hormone replacement therapy also decreases a woman s probability of dying from CRC by approximately 40% (Slattery, Anderson et al. 1999). It has been suggested that estrogen protects against CRC by inducing apoptosis and inhibiting cell proliferation, and some studies have suggested that the estrogen receptor has tumor-suppressor activity (Koo and Leong 2010). Estrogen receptor β (ERβ) likely mediates the protective effects of estrogen in CRC and ERβ is downregulated as estrogen levels decrease postmenopause (Wong, Malcomson et al. 2005). Loss of ERβ is linked with tumor progression because it is lost in malignant tissue. Further, mice lacking ERβ have 132

145 increased proliferation and decreased differentiation of colonic mucosal cells (Hendifar, Yang et al. 2009). There is also a sex difference in survival in patients with lung cancer, which is the leading cause of cancer-related death for both sexes. As is the case with CRC, women have decreased development and progression of non-small cell lung cancer and increased survival (McGovern, Liao et al. 2009). This improved survival is independent of tobacco use. However, this study found that both pre- and post-menopausal women survive longer than men. A similar study also found that women of all ages with lung cancer survive the same amount of time. However, they still survive longer than their male counterparts (Moore, Mery et al. 2003), suggesting that factors other than estrogen are involved in increased female survival, at least in non-small cell lung cancer. Colorectal, pancreatic and lung cancers are the main cancers that cause cachexia, a severe tissue-wasting syndrome that contributes to mortality in patients with advanced cancer. Cachexia involves a preferential loss of skeletal muscle, which can approach 75% (Melstrom, Melstrom et al. 2007). Many studies have shown that males lose more muscle mass and body weight than females, which could contribute to their overall decrease in survival. Body composition analysis in men and women with lung cancer revealed that 61% of men met the criteria for severe muscle wasting, as opposed to 31% of women (Baracos, Reiman et al. 2010). The women in this study also survived 23% longer. Another study on patients with lung cancer also found that men lost significantly more body weight than women, and that this occurred at an 8-fold faster rate (Palomares, Sayre et al. 1996). Two of the strongest predictors of survival in this patient population were 133

146 initial weight loss rate and gender, which suggests that the sex differences in weight loss may mediate the differences in overall survival. A significant correlation exists between severity of weight loss and mortality in patients with advanced, metastatic cancer. Weight loss in males is more frequent and severe than females and men have higher resting energy expenditure than women at all degrees of weight loss (Sarhill, Mahmoud et al. 2003). Weight loss can be due to fat or muscle loss, but muscle mass loss has more severe consequences and implications. Interestingly, men lose more muscle mass than fat, while women lose both muscle and fat equally (Sarhill, Mahmoud et al. 2003). Pro-inflammatory cytokines are implicated in the etiology of cancer cachexia as they can directly cause muscle protein loss in mice (Argiles, Busquets et al. 2005). Increases in circulating pro-inflammatory cytokines are correlated with weight loss and decreased prognosis. Accordingly, male CRC patients have significantly higher cytokine levels than females (Sharma, Greenman et al. 2010), which may contribute to their increased muscle mass loss and decreased survival. We have shown elsewhere that male tumor-bearing mice have a trend toward higher levels of pro-inflammatory cytokines than females (Cosper and Leinwand 2011). There are two pieces of important evidence that estrogen levels in women regulate cytokine levels. Healthy post-menopausal women receiving oral estradiol have decreased serum IL-6 levels, implying that estrogen is able to affect whole body cytokine levels (Rachon, Suchecka-Rachon et al. 2006). Premenopausal women with oophorectomies and thus significantly reduced estrogen levels, have increases in multiple proinflammatory cytokines as soon as 1 week after surgery (Pacifici, Brown et al. 1991). 134

147 Estrogen replacement therapy in these women decreased the secretion of these cytokines in as little as two weeks. Together, these studies indicate that estrogen prevents the secretion of pro-inflammatory cytokines from multiple cell types, which could contribute to the increased prognosis and survival observed in female cancer patients. The beneficial effects of estrogen on cachexia and survival in female cancer patients and tumor-bearing mice have been documented, but the effects of plant-based estrogens, or phytoestrogens, on these parameters in both sexes are unknown. The protein content of conventional mouse chow is derived from soy, which contains very high levels of phytoestrogens. Serum levels of phytoestrogens in mice eating a soy-based diet can reach 8.5 µm (Brown and Setchell 2001), which is 30,000-60,000 fold greater than the concentration of endogenous estrogens (Thigpen, Setchell et al. 1999). The main phytoestrogens in soy are the isoflavones daidzein and genistein, which are structurally similar to 17β-estradiol and therefore have estrogenic properties. They can bind and activate both estrogen receptors (ERα and ERβ) but have a higher affinity for ERβ (Setchell, Brown et al. 2002). Phytoestrogens can activate transcription of estrogenresponsive genes, but do so more than a thousand-fold less effectively than 17β-estradiol (Safford, Dickens et al. 2003). Genistein and daidzein can also function independently of the estrogen receptor. For example, genistein is a well-known tyrosine kinase inhibitor (Kim, Peterson et al. 1998). Diet, particularly soy, can significantly affect disease phenotype. A soy diet worsens hypertrophic cardiomyopathy in male mice by decreasing function and increasing apoptosis compared to mice on a soy-free diet (Stauffer, Konhilas et al. 2006). Phytoestrogens can also have beneficial effects on certain diseases, such as cancer. 135

148 Genistein can inhibit the growth of cancer cells by regulating genes involved in cell cycle and apoptosis, has anti-oxidant properties and can inhibit angiogenesis and metastasis (Sarkar and Li 2002). We have previously shown in a colon cancer model that blocking the estrogen receptor in females increases body weight and cardiac mass loss to male levels (Cosper and Leinwand 2011). These results begged the question of whether dietary estrogens also play a role in muscle mass loss due to cancer. The aims of the present study were to i) fully characterize the sex differences in muscle mass loss and the related changes in gene expression, ii) determine if estrogen can prevent cardiac and body mass loss in males, and iii) determine the effects of dietary phytoestrogens on muscle mass loss and survival in both sexes. Materials and Methods Experimental animals Female Balb/c mice and male DBA/2 mice (The Jackson Laboratory) were crossed to produce CD2F1 mice, which is the strain used in all C-26 colon cancer studies. Animals were housed under standard conditions and fed a standard diet. All animal studies were reviewed and approved by the Institutional Animal Care and Use Committee at the University of Colorado at Boulder. Colon-26 adenocarcinoma (C-26) cells were a generous gift from Dr. Denis Guttridge, (The Ohio State University) and were cultured as described (Acharyya, Ladner et al. 2004). C-26 cells were trypsinized, washed in PBS, pelleted and resuspended in PBS at a concentration of 5 x 10 6 cells/ml. 100 µl of this was injected subcutaneously into the right flank of 8-week old male and female mice. A 136

149 subset of male mice were treated with 17β-estradiol pellets (90-day release, 0.25 mg; Innovative Research of America), which were subcutaneously implanted in the interscapular region. Mice were monitored and weighed daily after 7 days and were anesthetized, weighed, and cervically dislocated at the indicated time points. Tissues were rapidly excised, weighed, flash-frozen in liquid nitrogen and stored at -80 C for further analysis. Diets Mice were fed ad libitum either a soy diet (Harlan Teklad 8640), casein diet (Research Diets D10001), or a phytoestrogen supplemented casein diet (Research Diets). These diets are described in detail in (Stauffer, Konhilas et al. 2006). All three diets contain approximately the same percentage of protein, fat, carbohydrate, and fiber. The main differences between the diets are the source of protein (casein in D10001 versus alfalfa and soybean in 8640), carbohydrate (sucrose and corn starch in D10001 versus corn and soybean meal in 8640), and fat (corn oil in D10001 versus soybean oil in 8640). The phytoestrogen-supplemented diet was created by adding genistein (229.5 mg/g dry food weight) and daidzein (206 mg/g dry food weight) to the casein diet. These amounts of isoflavones are comparable to those in the soy diet. RNA analysis Total RNA was purified from frozen control and atrophic left ventricles and skeletal muscles using TRIzol reagent (Molecular Research Center, Inc.). First strand cdna was synthesized using Superscript II reverse transcriptase (Invitrogen) and random hexamer 137

150 primers. Gene expression was then determined by quantitative real-time reverse transcription polymerase chain reaction (qrt-pcr) using SYBR Green dye with gene specific primer sets and Applied Biosystems 7500 Real-Time PCR system. Primer sequences are listed in Appendix I. Data and statistical analysis Data are presented as mean ± SEM. Differences between groups were evaluated for statistical significance using Student s t-test. P values less than 0.05 were considered significant. Percent loss in each group was calculated by comparing the tumor-bearing mice to the control mice within each respective diet and sex. Males treated with estrogen were compared to untreated, non-tumor-bearing controls. Results Sex differences in muscle mass loss and E3 ubiquitin ligase expression We have previously shown that male tumor-bearing mice lose more body weight, gastrocnemius muscle mass and cardiac muscle mass than females (Cosper and Leinwand 2011). To determine if this is true for all muscles, we compared tibialis anterior (TA) and soleus muscle mass loss between males and females. Males also lose more TA weight than females throughout the course of disease, though this was only significant 21 days post-tumor cell inoculum (Figure 30A). Attempts at accurately quantifying soleus mass proved difficult due to its small size. Soleus muscle mass decreased significantly in both sexes by day 27, but due to the variability in mass, male soleus mass loss was not significantly different than females at any time point (Figure 30B). 138

151 Figure 30. Sex differences in skeletal muscle mass loss. (A) Percent tibialis anterior (TA) muscle mass loss is greater in males than females, but is only significant 21 days posttumor cell inoculum. (B) Percent soleus mass loss is not significantly different between males and females though males trend toward more loss at the end of disease. Graphs are mean ± SEM. * p < 0.05 versus females. 139

152 Figure 31. Sex differences in skeletal muscle E3 ubiquitin ligase gene expression. Atrogin-1 and MuRF-1 gene expression in the gastrocnemius (A) and the soleus (B) of tumor-bearing mice as determined by qrt-pcr. Males trended toward increased expression of both ligases in both muscles, but this was not significant. Gene expression was normalized to 18S. Graphs are mean ± SEM. 140

153 Skeletal muscle atrophy due to cancer is known to occur through the ubiquitin proteasome system (UPS) (Baracos, DeVivo et al. 1995). Proteins destined for degradation by the proteasome are tagged with ubiquitin molecules by ubiquitin E3 ligases. Atrogin-1 and MuRF-1 are muscle-specific E3 ligases that have been implicated in many models of muscle atrophy, and are transcriptionally upregulated during atrophy (Lecker, Jagoe et al. 2004). In order to determine if sex differences in muscle mass loss are correlated with atrogin-1 or MuRF-1 gene expression, we performed qrt-pcr in both fast-twitch (gastrocnemius) and slow-twitch (soleus) atrophic muscle 27 days posttumor cell inoculum. There was a trend toward increased atrogin-1 and MuRF-1 mrna in male versus female gastrocnemius, though this was not significant (Figure 31A). The same trend occurred in the soleus but with much more variability (Figure 31B). Effects of estrogen treatment on cancer cachexia in males We have previously shown that female tumor-bearing mice lose less body weight and cardiac mass than males and that this is dependent on estrogen signaling through its receptor (Cosper and Leinwand 2011). We therefore wanted to determine if estrogen would also prevent body and cardiac mass loss in males. 17β-estradiol pellets were subcutaneously implanted into male mice 1 week prior to tumor cell inoculation and tissues were weighed at the 27-day time point. Estrogen treatment did not affect the extent of body weight loss or gastrocnemius loss in tumor-bearing males (Figure 32A). Surprisingly, estrogen treatment in males increased cardiac mass loss by 30% (Figure 32A). Estrogen also significantly decreased tumor weight (Figure 32B), which is suggestive of anti-tumor effects and correlates with clinical findings that estrogen 141

154 Figure 32. Estrogen treatment exacerbated cardiac atrophy in males. (A) Percent body weight (BW), heart weight (HW), and gastrocnemius (Gastroc) mass loss in tumorbearing males treated with 17β-estradiol compared to untreated males. Tissue weights were compared with age-matched, untreated, cancer-free males. (B) Estrogen treatment significantly decreases tumor mass in males. Graphs are mean ± SEM. ** p < 0.01 versus untreated controls. 142

155 treatment decreases CRC incidence (al-azzawi and Wahab 2002). These results indicate that unlike females, estrogen worsens certain parameters of cachexia in males, but is beneficial in terms of reducing tumor growth. Effects of sex and diet on survival of mice with colon-26 adenocarcinoma We observed that tumor-bearing females have significantly increased survival compared to males (Figure 33A). Tumor-bearing males appear moribund approximately 23 to 25 days post-tumor cell inoculum, and generally do not survive past 27 days. Tumor-bearing females, however, appear relatively healthy, remain active, and many survive up to 40 days post tumor-cell inoculum. Our mouse model therefore exhibits the same female survival advantage that is observed in CRC patients. Because estrogen seems to protect females from body and cardiac mass loss but worsens cardiac mass loss in males, we wanted to determine how dietary phytoestrogens affect survival and the severity of cancer cachexia in both sexes. To study this, we compared male and female mice fed a conventional soy-based diet with those fed a phytoestrogen-free, casein-based diet. In order to determine the role of genistein and daidzein alone, we also fed male and female mice a phytoestrogen-supplemented casein diet (hereafter referred to as the PE diet). A striking, initial observation was that both male and female tumor-bearing mice on the casein diet died much sooner than those on the soy diet (Figures 33B and 33C). Tumor-bearing males on the soy diet survived an average of 27 days, while those on the casein diet only survived a median of 22 days. Males on the PE diet had increased survival compared to those on the casein diet, but survival was decreased compared to those on soy. At day 27, only 38% of males on the 143

156 Figure 4. The effects of diet on survival in male and female mice with colon-26 adenocarcinoma. (A) Male versus female survival on the soy diet. Most males had to be sacrificed by day 27 because they were very ill. Survival curve analysis requires that this be entered as the equivalent of survival, while in fact they would have died. This interfered with the statistical analysis. Males on the soy diet never survived beyond 28 days, while females survived up to 38 days. (B) Survival of males on the soy, casein or phytoestrogen-supplemented (PE) diet. Males on soy had significantly increased survival compared to males on the casein or PE diets. (C) Survival of females on the soy, casein or PE diet. Females on soy had significantly increased survival compared to females on the casein or PE diets. * p < 0.05, ** p < 0.01, *** p <

157 casein diet and 66% of males on the PE diet were alive whereas 97% of males on the soy diet were alive (Figure 33B). Females on the casein and PE diets had significantly increased mortality compared to females on soy and unlike males, percent survival was similar on the casein and PE diets (Figure 33C). The casein diet increased male mortality more than females and female tumor-bearing mice still outlived males, though this was not significant. Together, these results imply that females have a survival advantage that is independent of diet and that elements in soy, other than genistein and daidzein, increase survival in both male and female tumor-bearing mice. Effects of sex and diet on body weight To establish the effects of diet on cancer cachexia in each sex, we obtained body and muscle weights for male and female mice on each diet 15 and 27 days post-tumor cell inoculum. Figure 34 shows how body weight changed with respect to cancer, sex, and diet 27 days post-tumor cell inoculum. As previously described in Stauffer et al., male mice on the casein diet weigh more than those on soy. All tumor-bearing male mice lose a significant amount of body weight, regardless of diet. Interestingly, female mice eating soy do not lose a significant amount of weight, while females on the casein and PE diets do. Unlike males, diet does not affect weight in control female mice. At 15 days, male mice lose nearly 2-fold more body mass on the casein diet than soy or PE, while diet does not affect female body weight loss at this time point (Figure 35A). Females on the soy and casein diets lose significantly less weight than their respective male counterparts. By day 27, male mice on the casein diet lose 30% more body weight than those eating soy, while males on the PE diet lose more weight than 145

158 Figure 34. Effects of sex, diet, and cancer on body weight. Colon-26 adenocarcinoma (C-26) causes a significant decrease in body weight in both sexes on all diets except females on the soy diet. A casein-based diet also increases body weight in males. Tumor weight was subtracted from final body weight in the C-26 groups. PE is the phytoestrogen-supplemented diet. Graphs are mean ± SEM. * p < 0.05, *** p < versus control. 146

159 Figure 35. Effects of sex and diet on body weight loss. (A) Percent body weight loss 15 days post-tumor cell inoculum in males and females on the soy, casein or phytoestrogensupplemented (PE) diet. (B) Percent body weight loss at day 27, the end-point of the study. For each sex and diet, tumor weight was subtracted from final body weight and compared to the body weight of age-matched control mice. Ŧ represents a significant difference between diets within the same sex, represents a significant difference between the sexes on each diet. Graphs are mean ± SEM. 147

160 males on soy, but this was not significant (Figure 35B). The same pattern between diets also occurs in females. Female mice eating casein or PE lose approximately 2-fold more body weight than females eating soy. Once again, female mice on all three diets lose significantly less weight than the respective males. Thus the sex difference in body weight loss occurs regardless of diet. Mice on a casein diet have higher body fat than mice on a soy diet (Stauffer, Konhilas et al. 2006), thus it is likely that the increase in body mass loss we observed in the casein group is due to increases in fat loss rather than muscle. All body weight and muscle mass loss in the different groups is summarized in Table 3. Effects of sex and diet on skeletal muscle mass To determine if the differences in body weight loss are due to skeletal muscle mass loss, we analyzed the mass of a fast-twitch muscle, the gastrocnemius, and a slowtwitch muscle, the soleus. Control gastrocnemius muscle weight does not differ between the diets in males, but is greater in females on the casein diet than in females on soy (Figure 36). Both male and female tumor-bearing mice lose a significant amount of gastrocnemius mass on all diets (Figure 36). When we compared the extent of muscle mass loss between the sexes and diets, we did not find differences in gastrocnemius loss between the diets or the sexes early in atrophy at day 15 (Figure 37A). By day 27, however, male mice on the soy diet lose significantly more (26%) gastrocnemius weight than those on the casein or PE diets (Figure 37B). Once again, diet does not affect muscle mass loss in female mice. There also were not any significant sex differences on any diet, 148

161 Table 3. The effects of sex and diet on cancer cachexia and cardiac gene expression MALE FEMALE Soy Casein PE Soy Casein PE BW loss (%) Gastroc loss (%) Soleus loss (%) HW loss (%) β-myhc LC Cathepsin L Table 3. The effects of sex and diet on body weight (BW), skeletal muscle (Gastrocnemius and soleus), and heart weight (HW) loss in mice with colon-26 adenocarcinoma. Values are percent loss for tissue weights, and fold-change increase for gene expression (β-myhc, LC3, and cathepsin L). Gray boxes indicate the diet associated with the largest change in each variable, and sex. 149

162 Figure 36. Effects of sex, diet, and cancer on gastrocnemius muscle weight. Colon-26 adenocarcinoma (C-26) causes a significant decrease in gastrocnemius mass in both sexes on all diets. Females on the casein diet have increased gastrocnemius mass compared to those on soy. PE is the phytoestrogen-supplemented diet. Graphs are mean ± SEM. ** p < 0.01, *** p < versus control. 150

163 Figure 37. Effects of sex and diet on skeletal muscle mass loss. Percent gastrocnemius mass loss (A) and soleus mass loss (C) 15 days post-tumor cell inoculum in males and females on the soy, casein or phytoestrogen-supplemented (PE) diet. Percent gastrocnemius mass loss (B) and soleus mass loss (D) 27 days post-tumor cell inoculum in both sexes and all diets. To calculate percent loss, muscle mass in the C-26 group was compared to the muscle mass of age-matched control mice in each respective sex and diet. Ŧ represents a significant difference between diets within the same sex, represents a significant difference between the sexes on each diet. Graphs are mean ± SEM. 151

164 with the exception of soy, which has been previously reported (Cosper and Leinwand 2011). The soleus muscle is composed of primarily slow-twitch muscle fibers, which are known to atrophy less than fast-twitch (Mitch and Goldberg 1996). We wanted to determine if slow-twitch muscle loss is affected differently by sex and diet than muscle composed of fast-twitch fibers. At day 15, there are no differences in soleus mass between the sexes or the diets (Figure 37C). However, by day 27, male mice on soy lose greater than 7-fold more soleus mass than males on the casein or PE diets (Figure 37D). Soleus mass loss follows a similar pattern in females, though this was not significant. This suggests that sex and diet affect slow and fast-twitch muscle atrophy similarly. These results are summarized in Table 3. Together, these results indicate that a soy diet worsens muscle cachexia but the dietary phytoestrogens (genistein and daidzein) themselves are not responsible for this effect. Effects of sex and diet on cardiac muscle mass loss We have previously shown that cancer also causes cardiac muscle atrophy that is more severe in males than females (Cosper and Leinwand 2011). We compared cardiac mass loss on all diets and in both sexes. Control heart weights do not differ between the diets in males, but like the gastrocnemius, females on the casein diet have a greater heart weight than females on soy (Figure 38). Interestingly, tumor-bearing males on all diets lose a significant amount of heart weight, while this occurs only in tumor-bearing females on the soy diet. Females eating a casein or PE-based diet therefore do not lose a significant amount of heart weight. 152

165 Figure 38. Effects of sex, diet, and cancer on cardiac muscle weight. Colon-26 adenocarcinoma (C-26) causes a significant decrease in cardiac mass in males on all diets, but only in females on the soy diet. Females on the casein diet have increased cardiac mass compared to those on soy. PE is the phytoestrogen-supplemented diet. Heart weight (HW) is normalized to tibia length (TL) to account for mouse age and size. Graphs are mean ± SEM. * p < 0.05, ** p < 0.01, *** p < versus control. 153