Calorie Restriction and the Aging Muscle: a Multispecies Approach

CALORIE RESTRICTION AND THE AGING MUSCLE: A MULTISPECIES APPROACH

Melissa C. Orenduff

A dissertation submitted to the faculty at the University of North Carolina at Chapel Hill in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Nutrition in the Gillings School of Global Public Health.

Chapel Hill 2020

Approved by:

Stephen D. Hursting

Melinda A. Beck

Saame Raza Shaikh

Kim M. Huffman

Jane F. Pendergast

ABSTRACT

Melissa C. Orenduff: Calorie Restriction and the Aging Muscle: A Multispecies Approach (Under the direction of Stephen D. Hursting)

Aging is associated with a progressive decline in muscle mass, strength, and physical function termed sarcopenia. Sarcopenia is an important risk factor for loss of independence, reduced quality of life, and increased mortality. Calorie restriction (CR) is an effective dietary treatment for several age-related conditions, including sarcopenia. Recent work conducted in our laboratory suggests insulin-like growth factor-1 (IGF-1) is involved in the anti-aging effects of

CR. The aim of this work is to understand the role of circulating IGF-1 in CR-induced effects on aging skeletal muscle. Using a multispecies approach, we sought to determine if IGF-1- mediated effects of CR are conserved among mice, nonhuman primates, and humans. As an extension into our investigation of CR and IGF-1 on aging muscle, we conducted an in vivo experiment to examine the role of a CR mimetic, rapamycin, on age-related loss of muscle strength and function. Our results suggest that the effects of CR on aging skeletal muscle are realized through biological changes consistent with regenerative and repair-related mechanisms and a shift in transcriptional profiles towards improved metabolic and inflammatory signaling pathways that promote enhanced mitochondrial function and biogenesis. Collectively, this work demonstrates CR-induced effects on aging muscle are, in part, supported by reduced circulating

IGF-1. Further, the CR-mimetic, rapamycin may hold promise as a pharmacological alternative to CR to attenuate loss of muscle strength and function associated with age.

iii

To my fiancé, Carl, my Children, Morgan and Garrett, and extended family, thank you for the encouragement, support, and love throughout this journey.

TABLE OF CONTENTS

LIST OF TABLES ...... viii

LIST OF FIGURES ...... ix

LIST OF ABBREVIATIONS...... x

THESIS PRELUDE ...... 1

CHAPTER I: OVERVIEW AND SPECIFIC AIMS ...... 3

Overview ...... 3

Specific Aims ...... 6

CHAPTER II: BACKGROUND AND SIGNFICANCE...... 9

Impact of age-related changes on skeletal muscle architecture and metabolism ...... 9

The calorie restriction paradigm: Dietary modulation of aging ...... 10

Rodent CR studies...... 11

Nonhuman primate CR studies ...... 11

Human CR study ...... 12

Calorie restriction (CR) mitigates sarcopenia in various mammalian species...... 13

CR-induced effects on age-related loss in mice ...... 14

CR-induced effects on age-related loss in nonhuman primates ...... 15

CR-induced effects on age-related muscle loss in humans ...... 15

Summary of CR-induced effects on age-related loss in multiple species...... 16

Role of IGF signaling in skeletal muscle growth, maintenance, and repair, a process that involves stimulating anabolic pathways and suppressing catabolic activity ...... 17

Supporting roles of IGFBPs in CR-induced effects to mitigate Sarcopenia ...... 18

Effect of the CR mimetic, Rapamycin on biological aging ...... 19

CHAPTER III: THE IMPACT OF CALORIE RESTRICTION OR RAPAMYCIN ON AGE- AND SARCOPENIA-ASSOCIATED TRANSCRIPTIONAL SIGNITURES IN SKELETAL MUSCLE ...... 23

Introduction ...... 23

Materials and Methods ...... 26

Calorie restrction on transcriptional signatures in skeletal muscle...... 26

Effects of CR versus Rapamycin on Transcriptional signatures in skeletal muscle ...... 28

Results ...... 34

CR induces changes in skeletal muscle mass of mice and humans ...... 34

Gene expression analysis reveals metabolic networks are engaged by CR in skeletal muscle of mice and humans ...... 35

Influence of CR and Rapamycin on age-related loss of muscle mass, strength, and function ...... 39

Effects of rapamycin and CR on gene expression in skeletal muscle ...... 41

Discussion ...... 44

CHAPTER IV: MECHANISTIC ROLE OF RECUDED CIRCULATING LEVELS OF FREE IGF-1 INDUCED BY CALORIE RESTRICTION ON AGE-RELATED MUSCLE LOSS ...... 84

Introduction ...... 84

Materials and Methods ...... 87

Results ...... 90

Discussion ...... 92

CHAPTER V: SYNTHESIS ...... 100

Overview of research findings ...... 100

Public health mplications and future perspectives ...... 102

APPENDIX 1: EFFECT OF CALORIE RESTRICTION ON SYSTEMIC BIOMARKERS OF THE IGF SYSTEM IN MICE, NONHUMAN PRIMATES, AND HUMANS...... 106

APPENDIX 2: ASSOCIATIONS BETWEEN INSULIN-LIKE GROWTH FACTOR (IGF) BINDING PROTEIN-7 AND SOCIO-DEMOGRAPHIC VARIABLES, BODY MASS INDEX, AND ALL-CAUSE MORTALITY IN POSTMENOPAUSAL WOMEN FROM THE WOMEN'S HEALTH INITIATIVE OBSERVATION STUDY...... 116

REFERENCES ...... 136

vii

LIST OF TABLES

Table 2.1: Hallmark and BioCarta gene set enrichment in quadriceps muscle from CR versus control mice ...... 22

Table 3.1: Gene set enrichment analysis at 12 months ...... 58

Table 3.2: Gene set enrichment analysis at 24 months ...... 59

Table 3.3: Gene list for Reactome and KEGG gene sets commonly enriched in mice and humans ...... 60

Table 3.4: Gene Set Enrichment (Hallmark) for male mice ...... 72

Table 3.5: Gene Set Enrichment (Hallmark) for female mice ...... 72

Table 3.6: Gene Set Enrichment (Hallmark) for male and female mice in rapamycin group ...... 72

Supplemental Table 3.1S. Gene set enrichment analysis for mouse samples ...... 73

Supplemental Table 3.2S. Gene set enrichment analysis for human samples ...... 77

Table 4.1. Gene set enrichment analysis ...... 92

Appendix 2 Table 1. Characteristics of participants enrolled in the ancillary study of the Women’s Health Initiative Observational Study at Baylor College of Medicine and Wake Forest University School of Medicine between February 1995 and July 1998 ...... 126

Appendix 2 Table 2. Spearman Correlation Coefficients ...... 127

Appendix 2 Table 3. Regression Coefficients ...... 128

Appendix 2 Table 4. Survival Analysis ...... 129

viii

LIST OF FIGURES

Figure 1.1: Conceptual Model ...... 8

Figure 2.1: Schematic of age-related changes in muscle cell signaling that contribute to loss of muscle mass, strength, and function ...... 21

Figure 3.1: Mice: Body Weight and Relative Lean Mass ...... 51

Figure 3.2: Humans: Body weight and Relative Lean Mass ...... 52

Figure 3.3: Mice: Consensus Clustering ...... 53

Figure 3.4: Humans: Consensus Clustering ...... 55

Figure 3.5: Venn diagram of overlapping enriched gene sets between humans and mice 12 months ...... 58

Figure 3.6: Venn diagram of overlapping enriched gene sets between humans and mice at 24 months ...... 69

Figure 3.7: Average Daily Food Intake for Control and Rapamycin mice ...... 63

Figure 3.8: Body weight and fasting blood glucose...... 63

Figure 3.9: Absolute and relative muscle mass for male and female mice ...... 65

Figure 3.10: Functional tests ...... 67

Figure 3.11: Gene expression analysis for female mice...... 70

Figure 3.12: Gene expression analysis for male mice ...... 71

Figure 3.13: Differentially expressed genes in females versus males ...... 71

Figure 4.1. Body weight and fasting blood glucose ...... 90

Figure 4.2. Exogenous PEG-IGF-1 increases IGF-1 serum levels in CR mice ...... 91

Figure 4.3. qPCR analysis of mRNAs derived from quadriceps muscle ...... 91

Figure 4.4. Gene expression analysis ...... 92

LIST OF ABBREVIATIONS

Akt Protein Kinase B (serine/threonine protein kinase)

BMI Body Mass Index

CR Caloric Restriction

EGF Epidermal growth factor eRapa Enteric rapamycin

ERK Extracellular signal-regulated kinases

FoxO Forkhead family of transcription factors

IGF Insulin-like Growth Factor

IGF-1 Insulin-like Growth Factor - 1

IGF-1R Insulin-like Growth Factor – 1 Receptor

IGFBP Insulin-like Growth Factor Binding Protein

MAPK Mitogen-activated protein kinases mTOR Mammalian Target of Rapamycin

PEG Polyethylene glycol

PI3K Phosphoinositide 3-kinases qPCR Quantitative polymerase chain reaction

TNFα Tumor Necrosis Factor α

TGF-β Transforming growth factor beta

ROS Reactive Oxygen Species

Wnt Wingless-related integration site

THESIS PRELUDE

This body of work investigates the effect of calorie restriction (CR) on age-related loss of muscle mass, strength, and function (referred to as sarcopenia). Evidence has shown that CR delays the onset and progression of sarcopenia by acting at several critical control points and signaling pathways that regulate muscle fiber number, size, and composition – primary determinants of muscle mass, strength, and function. Previous work conducted by our lab and others support a role for insulin-like growth factor (IGF) signaling in the anti-aging effects of

CR. Thus, we sought to investigate the role of IGF-1 mediated effects of CR on age-related muscle loss. As a unique feature of this work, we have incorporated a multispecies approach utilizing samples from mice, nonhuman primate, and human CR studies to develop a comprehensive assessment of CR-induced molecular adaptations on skeletal muscle and circulating IGF-1 across mammalian species. Further, as an extension to our investigation of these CR-induced effects on skeletal muscle, we examined the potential role of rapamycin, a pharmacological agent and established CR-mimetic, to attenuate age-related loss of muscle mass, strength, and function.

This thesis is divided into the following six Chapters: Chapter I provides a general overview to introduce our overarching research questions: 1. Does CR induce changes in circulating bioactive IGF-1 levels and is this change conserved across mice, nonhuman primates, and humans? 2. Does the reduction of circulating levels of total IGF-1 observed in mice mechanistically contribute to attenuation of muscle mass, strength, and function related to age?

3. Is there a role for rapamycin, as a therapeutic intervention, to prevent loss of muscle strength

and function associated with age?

Chapter II includes a background on the central themes of this work including: a) the impact of age-related loss of muscle mass, strength, and function (sarcopenia); b) proposed mechanisms associated with the CR paradigm on these anti-aging effects in mice, nonhuman primates, and humans; c) role of IGF-1 signaling in skeletal muscle, in the context of CR; d) supporting roles for IGF binding proteins (IGFBPs) in CR-induced effects on age-related muscle loss; and e) evidence to support the use of rapamycin as a CR mimetic. Chapter III presents our work investigating CR-induced gene expression changes in skeletal muscle. By employing both mice and human samples, we were able to conduct comparative analysis to identify differences and similarities in gene expression between species (mice and human) and diet treatments (CR and ad libitum control). Additionally, as a potential alternative to CR, we examined the effects of rapamycin on age-related changes in skeletal muscle of mice. Chapter IV reports on a mouse

CR study designed to investigate the mechanistic role of free, bioavailable IGF-1 on CR-induced effects on age-related muscle loss. In Chapter V, we summarize our major findings and conclusions and include how understanding the mechanisms by which CR interrupts age-related muscle loss can provide insight and potentially contribute to the development of novel therapies, such as rapamycin to improve the health and active life span of aging populations.

Lastly, in Appendix 1, we present our ongoing work to incorporate a collaborative multispecies approach to investigate the CR-induced effects on both biologically active and biologically inert forms of circulating levels of IGF-1 in mice, nonhuman primates, and humans.

CHAPTER I: OVERVIEW AND SPECIFIC AIMS

Overview

Aging is associated with numerous physiological changes that affect health and lifespan.

One of the major deleterious effects of biological aging is sarcopenia, or the progressive loss of muscle mass, strength, and function related to aging1. Sarcopenia contributes to functional decline, disability, frailty, increased risk of falls, and mortality2,3. The prevalence of sarcopenia in the US is estimated to be 14% of adults aged 60-70 years and 53% of adults aged >80 years3–5.

With an unprecedented rise in the number of older adults, the public health implications of untreated sarcopenia are far-reaching and will place significant burdens on the provision of health care and age-related services6. Understanding the etiology and pathology that contribute to sarcopenia and developing interventions that can intervene with the disease process is of great importance to the health of aging populations2.

Calorie restriction (CR), defined as a reduction in caloric intake without malnutrition, has been shown to attenuate age-related muscle loss in various mammalian species including mice, nonhuman primates, and humans7–9. CR is purported to promote transcriptional reprogramming associated with increased muscle growth, maintenance, and repair and reduced muscle breakdown10,11. During CR, while an absolute amount of muscle (or lean mass) is lost, the proportion of lean mass as a percent of total body mass increases, a phenotypic signature associated with reduced risk of metabolic disease, including sarcopenia12–15. One proposed mechanism by which CR elicits its effects on skeletal muscle is through the insulin-like growth

factor (IGF) signaling axis16. Specifically, the research suggests that CR increases IGF/mTOR signaling, to effectively regenerate muscle while decreasing tumor necrosis factor α (TNFα) signaling and inducing elements of antioxidant response to reduce muscle degradation16–18.

IGF-1 is produced primarily in the liver; however, various other tissues including skeletal muscle are responsible for the local synthesis of IGF-119,20. Liver-derived IGF-1 is stimulated by pituitary growth hormone and is responsible for >75% of circulating IGF-1 levels21; whereas, the production of locally produced and utilized IGF-1 in skeletal muscle is, in part, regulated by systemic IGF-120,22. Systemic and muscle-derived IGF-1 regulates various stages of myogenesis, including proliferation, differentiation, and fusion of muscle precursor cells23. These processes are stimulated via activation of PI3K/Akt/mTOR and MAPK/ERK signaling pathways and result in increased muscle fiber number and size20. Conversely, reduced circulating levels of IGF-1 and inhibition of downstream myogenic signaling through PI3K/Akt/mTOR and MAPK/ERK in skeletal muscle are associated with aging and linked to several hallmarks of sarcopenia

(including muscle atrophy, accompanied by loss of strength, mobility, and function)24–26.

In circulation, IGF-1 exists in either an unbound (free) or bound state when sequestered by carrier IGF binding proteins (IGFBPs). Nearly all circulating IGF-1 is bound to IGF binding proteins (IGFBP 1-6), leaving less than 1% IGF-1 in a free form that can bind to the IGF-1 receptor (IGF-1R) and/or insulin receptor (IR)27. IGFBPs 1-6 regulate IGF-1 bioavailability and facilitate tissue-specific activity19,28,29, including growth factor signaling and metabolic regulation, in an IGF-dependent fashion. IGFBPs can also function independently of IGF-128. In particular, IGFBP7, also called IGFBP-related protein-1 (IGFBP rp1) or Mac25, has a 100-fold lower affinity for IGF-1 relative to the other IGFBPs, but it binds strongly to insulin and regulates IGF-1R and IR receptor activity. IGFBP7 is explicitly produced in skeletal muscle30,

and along with IGFBP 3 (which is the most abundantly expressed BP in circulation), has been shown to interfere with TNFα signalingan IGF-1 independent pathway known to be involved in age-related muscle degradation31–34. Although the role(s) of IGFBPs in the context of CR have yet to be identified, IGFBP 3 putatively plays an especially critical role in regulating systemic

IGF-1 levels and activity28,34, while IGFBP7 acts primarily as an IGF-1R and IR receptor modulator to regulate signals downstream of IGF-1R and IR.

To date, nearly all assays used for assessing serum or plasma levels of IGF-1 in humans or animals have been limited to measuring total IGF-1, which cannot distinguish the free versus bound form of IGF-135. In some reports, the molar ratio of total IGF-1 to IGFBP-3, the most abundant IGFBP in blood, has been used as a proxy of bioavailable IGF-1, although the concordance of this marker with actual bioavailable IGF-1 is unclear. Total IGF-1 and the ratio of IGF-1:BP3 have been informative markers for assessing associations between the IGF system, aging, and some age-related chronic diseases. In general, high total circulating IGF-1 or IGF-

1:IGFBP3 is associated with increased risk of many cancers36, but decreased risk of cardiovascular disease37, Moreover, the impact of diet on total IGF-1 and IGF-1:IGFBP-3 has been inconsistent, with apparent differences across species. For example, CR in mice decreases serum total IGF-1 concentrations by ~40%38, but the effects of CR on these markers in nonhuman primates39 are less pronounced and in humans are equivocal40. This apparent disparity between mice and primates represents an important knowledge gap in CR research, with a particular need for better assays of bioavailable IGF-1 and direct comparisons between species on multiple markers of the IGF system.

As an extension to our investigation of CR-induced effects on age-related muscle loss, we

also examined an alternative pharmaceutical strategy using the CR mimetic, rapamycin. The objective of this strategy was to identify a safe drug or nutraceutical agent that could potentially emulate the anti-aging effects of CR on skeletal muscle by targeting metabolic and stress response pathways, without restricting dietary intake. To date, CR and the CR mimetic rapamycin are the only two interventions able to extend median and maximal lifespan in rodent aging studies41. However, the effects of rapamycin on age-related sarcopenia are unclear.

Mechanistically, rapamycin mimics CR through its inhibition of the mechanistic target of rapamycin complex 1 (mTORC1), which is downstream of IR and IGF-1R and is a key regulator of protein and nucleotide synthesis, survival signals, and macronutrient metabolism.

To address the gap of between species discrepancy in circulating IGF-1 and examine molecular changes induced by CR in skeletal muscle, we have employed a translational approach to investigate CR-induced effects across mammalian species including mice, nonhuman primates, and humans. Additionally, we conducted in vivo experiments to first: test the mechanistic role of reduced circulating IGF-1 levels observed in mice on aging skeletal muscle; and second: to investigate the pharmacological agent, rapamycin as a potential drug to mitigate loss of muscle mass, strength, and function associated with age (Figure 1.1). The objectives of this work were addressed by completing the following specific aims:

Specific Aims

Specific Aim 1a: Identify CR-induced effects on gene expression (including myogenic and IGF regulated targets) in skeletal muscle of mice and humans. 1b: Determine if the CR mimetic, rapamycin confers similar molecular changes induced by CR.

Hypothesis: 1a: Gene expression of myogenic and IGF-associated genes are upregulated in

skeletal muscle in CR groups, relative to ad libitum. 1b: Effects of rapamycin on gene expression in skeletal muscle align with changes observed in CR with respect to transcriptional adaptations associated with growth, maintenance, and repair.

Specific Aim 2: Delineate the effect of CR on free IGF-1 and major constituents of the IGF-1 system in circulation that are purported to be involved in CR-induced effects including IGFBP 3 and IGFBP 7 in mice, nonhuman primates, and humans.

Hypothesis: CR reduces levels of circulating, free (biologically active) IGF-1, and this reduction is conserved across mammalian species.

Specific Aim 3: Determine if CR-induced effects to mitigate age-related muscle loss in mice are dependent on decreased circulating free IGF-1.

Hypothesis: Reduced circulating free IGF-1 is causally related to the protective effect of CR on age-related muscle loss in mice.

Chapter I: Figure

Figure 1.1. Conceptual model. We present a conceptual model serving as the basis for the experimental approach underlying our study design. We conceive a causal path from implementing a CR diet to attenuation of age related muscle loss. This occurs by means of mediating factors (IGF-1 signaling) that affect aging skeletal muscle. The specific aims are directed at understanding the role of CR-induced reductions in circulating levels of free IGF-1 on aging muscle and discerning how CR-induced effects translate to higher mammalian species. Moreover, we will investigate the ability of the CR mimetic, rapamycin to delay or prevent age-related decline in muscle strength and function.

CHAPTER II: BACKGROUND AND SIGNIFICANCE

Impact of age-related changes on skeletal muscle architecture and metabolism

Skeletal muscle is the largest organ in the body and generally contributes ~50% of total body mass in young adults42. In the normal course of aging, as early as the 3rd decade, the proportion of muscle (or lean muscle mass) decreases on average about 5% every 10 years43. In a study on human aging and muscle mass, Lexell et al. reported the cross-sectional area of the vastus lateralis (quadriceps) muscle decreased by 40% between the ages of 20 and 80 years44.

This gradual loss of skeletal muscle is defined as sarcopenia and is characterized by muscle atrophy and reduced metabolic performance (a component of muscle quality). These features of sarcopenia are accompanied by a progressive loss of muscle strength and function and are directly linked to poor quality of life, increased falls and factures, disability, loss of independence, and mortality45.

At a physiological level, skeletal muscle is composed of bundles of muscle fascicles that are made up of individual muscle fibers. Muscle fibers contribute to the functionality and metabolic integrity of the whole muscle and are classified by characteristic movement rates

(‘fast’ or ‘slow’ twitched) and metabolic styles (oxidative and glycolytic). Although a constellation of fiber types is vital for optimal muscle function, an abundance of ‘fast’ and ‘slow’ twitched oxidative fibers are associated with younger phenotypes and contribute to increased energy-generation capacity compared with glycolytic fibers46,47. The architectural composition

(number and size) of each muscle fiber is dependent upon a tightly regulated balance between

protein synthesis and protein degradation48. In youth, this equilibrium is maintained (synthesis is stimulated and degradation is inhibited), in large part by insulin-like growth factor 1 (IGF-1) and downstream signaling through PI3K/Akt/mTOR and MAPK/ERK pathways25. Aging, however, disrupts this system resulting in a net loss of skeletal muscle mass, primarily by decreased anabolic activity (reduced IGF-1 signaling) and increased muscle degradation via inflammatory factors such as tumor necrosis factor α (TNFα)18. Additionally, oxidative damage, due to an accumulation of reactive oxygen species (ROS), has been proposed as a major contributor of age-related muscle atrophy48. As the largest consumer of oxygen in the body, skeletal muscle generates high ROS flows as by-products of the mitochondrial electron transport chain.

Although ROS are produced as a normal product of cellular metabolism, aged muscle exhibit reduced antioxidant response and subsequent oxidative damage49 (Figure 2.1). Collectively, these age-induced effects on skeletal muscle occur in parallel with increased fat infiltration and altered cellular regulation, resulting in a metabolic shift from oxidative to a more glycolytic landscape46. Overall, these age-related changes in skeletal muscle contribute to increased risk of adverse outcomes associated with sarcopenia50,51. It is estimated that over the next 4 decades,

~200 million people worldwide will experience poor health-related effects due to sarcopenia52.

Thus, the development of treatments and strategies to counteract the negative impact of sarcopenia will likely improve active life expectancy in older people and lead to substantial health-care savings and improved quality of life.

The calorie restriction paradigm: Dietary modulation of aging

Calorie restriction (CR), or the daily restriction of caloric intake without a reduction in essential nutrients, is one of the most robust interventions shown to delay the onset and

progression of age-related diseases53 and extend lifespan54. Whether CR is prescribed as an intervention or practiced as a lifestyle dietary regimen, the main objective is to improve health and delay the effects of aging55. Since McCay et al. first observed the striking effects of CR to reduce age-related disease risk and extend lifespan in rats56, similar findings have been subsequently demonstrated across a wide and diverse range of species from unicellular organisms (yeast) to nematodes, invertebrates, and mammals54,57. The following sections review the findings observed from CR studies conducted in a.) rodents; b.) nonhuman primates; and c.) humans to support the role of CR to attenuate age-related diseases and extend life- and health- span in multiple mammalian species.

a.) Rodent CR studies

The CR paradigm has provided one of the most useful research tools for investigating mechanisms of aging and longevity, and CR studies conducted in rodents have paved the way for these insights39. CR studies conducted in small laboratory animals indicate that the benefits of CR on disease risk and lifespan extension are attained through complex interactions of biological adaptations55,58. CR has been shown to delay, and even prevent several chronic diseases and age- related disorders in rodents including cancer, diabetes, autoimmune and neurodegenerative diseases, and sarcopenia59–62. These results suggest CR impinges on multiple signaling pathways that regulate metabolism, oxidative stress, inflammation, and autophagy to promote longevity and healthy aging39,55.

b.) Nonhuman primate CR studies

The relevance and value of CR as a research tool to understand human aging hinges on

conservation of the CR-response on aging in primates. Nonhuman primates are similar to humans in their genetic, anatomical, and physiological profiles and susceptibility to infectious and metabolic diseases to humans, and thus serve as an excellent model to investigate the translatability of CR’s benefits to humans63,64. Additionally, nonhuman primate models also offer the ability to tightly control experimental conditions and allow for the implementation of a rigorous, yet consistent research protocol, such as a CR diet7. The first two prospective CR studies in nonhuman primates (rhesus monkeys) were conducted in 1987 at the National Institute on Aging (NIA)65, and in 1989 at the University of Wisconsin's National Primate Research

Center (WNPRC)66. These long-term studies (18 years) in nonhuman primates evaluated the effects of CR on a variety of biological measures and represents an important step towards bridging the gap between animal and human CR research. A complete description of the NIA and WNPRC studies, including major findings and key comparisons between the two studies are detailed elsewhere14,67–69. In brief, the most compelling evidence of the benefits of CR in nonhuman primate studies includes decreased cardiovascular risk; improved insulin sensitivity and glucose tolerance; and lower incidence of type 2 diabetes, cancer, and cardiovascular disease. These changes coincide with favorable changes in body composition (i.e., lower body fat and attenuation of age-related muscle loss, or sarcopenia)14,15,67,70–84. In summary, studies conducted in nonhuman primates not only validate the findings from rodent models, but also provide rigorous preclinical support for assessing CR effects on aging and the prevention of age- related conditions in human studies76.

c.) Human CR study

Due to high scientific interest to extend the CR paradigm to humans, in 2002 the NIA

sponsored the first randomized clinical trial (RCT) of CR, the CALERIE (Comprehensive

Assessment of the Long-term Effects of Reducing Intake of Energy) Study85. CALERIE was implemented as a sustained (2-year), multicenter (Pennington Biomedical Research Center ,

Tufts University, and Washington University), RCT investigating the effects of CR (25%) in nonobese men (21-50 years of age) and women (21-47 years of age, to avoid perimenopausal confounding effects)86. The primary goals of CALERIE were to test the effect of long-term CR in humans and examine whether CR would result in sustained metabolic adaptations, including reductions in core body temperature and reduced resting metabolic rate (RMR). Secondary outcome measures included energy metabolism, cardiovascular risk factors, immune function, glucose and insulin parameters, endocrine response, qualify of life, psychological and cognitive functioning, physical activity, body composition (height, weight, and percent lean and fat mass), and nutrient intake. Following baseline testing, enrolled participants were randomly assigned in a 2:1 ratio to either a CR or ad libitum diet group85,86. A summary of results suggest CR participants exhibited beneficial effects of CR with respect to glucose control, including reductions in: a) systemic insulin and glucose levels, as well as insulin resistance as determined by the homeostatic model assessment of insulin resistance (HOMA-IR); cardiometabolic risk factors, including reductions in blood lipids, total cholesterol, and triglycerides); systemic biomarkers of inflammation (including reduced levels of TNF-a); and an increase in relative lean mass, a measure associated with attenuation of age-related muscle loss and impaired function (i.e. sarcopenia)12,87,88. These results from CALERIE provide evidence for the benefit of CR on health and longevity in humans.

Calorie restriction (CR) mitigates sarcopenia in various mammalian species

Possibly the most compelling evidence of the effects of CR to attenuate age-related disease

and promote markers of longevity is demonstrated by the impact of CR on age-related muscle mass. Despite an absolute loss of fat and lean mass, studies conducted in various species

(including mice, nonhuman primates, and humans) have demonstrated CR effectively increases the proportion of lean muscle mass, as a percent of total body mass (hereafter referred to as relative lean or muscle mass)12,13,17,61,89–93 (Figure 2.2A-C). Increased relative lean mass is a body composition phenotype associated with reduced risk of metabolic disease, including sarcopenia, and an independent predictor of longevity in older adults94,95. The following sections summarize the results from CR studies conducted in a.) mice; b.) nonhuman primates; and c.) humans to demonstrate the conserved anti-aging effects of CR on skeletal muscle.

a.) CR-induced effects on age-related muscle loss in mice

In a CR study, previously conducted by our collaborators at Wageningen University, mice prescribed a CR diet experienced a significant increase in relative lean mass compared with ad libitum control mice13. Additionally, when assessing the absolute amount of muscle loss during aging (a diagnostic feature of sarcopenia), CR mice exhibited a protective effect against age-associated muscle loss compared with ad libitum controls. Functional measures of sarcopenia, including grip strength, balance, agility, maximum/relative force, and activity level were increased in CR or found to be analogous, relative to age-matched ad libitum controls, despite CR mice having a significantly lower absolute muscle mass and lower total body weight.

These results suggest CR mice, relative to ad libitum-fed control mice, display a higher degree of muscle strength (in relation to muscle mass ratio) and functional capacity13. These data appear to corroborate findings from higher mammalian CR studies conducted in nonhuman primates and humans.

b.) CR-induced effects on age-related muscle loss in nonhuman primates

CR studies conducted in nonhuman primates (rhesus monkeys) reinforce the rodent findings that CR mitigates age-related muscle loss. Investigators on our collaborative research team at NIA and University of Wisconsin found age-related muscle loss was significantly reduced in the CR group compared with ad libitum controls15,96. Moreover, CR animals experienced a marked increase in relative lean mass compared with controls15,81,83,97. To assess physical function, activity was measured using an accelerometer, and results indicate CR animals were significantly more active than ad libitum controls97. In addition, measures of poor endurance assessed by the reduction in energy efficiency calculated from metabolic cost of movement for each animal were conducted. CR animals demonstrated a significantly higher endurance index compared with age-matched control counterparts97. Attenuation of age-related muscle mass and physical activity was further confirmed with biometric and metabolic rate data.

Preserved metabolic efficiency in CR animals correlated with the retention of muscle mass in aged monkeys96.

c.) CR-induced effects on age-related muscle loss in humans

In the human CALERIE study, conducted by researchers at the NIA and an extension of our collaborative research team, an increase in relative lean mass in CR versus ad libitum controls was observed. Specifically, at the end of the 2-year CR period, in nonobese, middle aged adults (~38 years), there was a mean increase in lean mass from 67% at baseline to 72% at endpoint compared with no change in ad libitum controls (67% lean mass at baseline and endpoint)12. For functional assessments, CR resulted in a higher relative aerobic capacity

(measured by maximum treadmill exercise) compared with ad libitum controls. Measures of absolute leg strength revealed no significant differences at 2 years between groups, despite the

CR group having significantly lower absolute muscle mass and total body weight98. This finding aligns with our mouse study showing increased muscle strength to muscle mass ratio in the CR group13.

d.) Summary of CR-induced effects on age-related muscle loss in multiple species

Taken together, the observations from mouse, nonhuman primate, and human studies support the role of CR to prevent muscle loss, in the context of weight loss, and mitigate age- related sarcopenia in aging mice and nonhuman primate models. Further, these results demonstrate CR-induced effects on skeletal muscle are conserved across mammalian species12,13,67. Although much remains to be discovered about the mechanisms by which CR elicits its protective, promyogenic effects in skeletal muscle, it has been shown CR results in molecular adaptations that promote a shift in cellular regulations that support regeneration and inhibit degradation in skeletal muscle46,99. Evidence suggest these CR-induced benefits are realized through a complex interplay of biological changes involving anabolic effects of IGF-1 signaling (supporting growth and repair) and complimentary effects involving anti-oxidant, anti- inflammatory, and cellular energy metabolic pathways to support survival16. Collectively, these data support the potential role of CR to serve as a relevant experimental intervention delay and possibly prevent age-related muscle loss in humans. While CR is challenging for most people to adopt and sustain as a lifestyle modification, our goal is elucidate the mechanisms underlying the

CR-associated protection of the aging muscle to identify new targets and strategies for decreasing the burden of age-related sarcopenia.

Role of IGF signaling in skeletal muscle growth, maintenance, and repair, a process that involves stimulating anabolic pathways and suppressing catabolic activity.

The IGF axis is an evolutionary conserved pathway and essential for virtually every biological process associated with growth and proliferation, particularly in skeletal muscle100,101.

Major components of the IGF system include the ligand, IGF-1, the receptor, IGF-1R, and IGF binding proteins 1-7 (IGFBPs 1-7) and IGF related binding proteins. In skeletal muscle, IGF-I stimulates proliferation and differentiation of myoblasts, promotes myotube hypertrophy, while also inhibiting protein degradation102–105. These mitogenic, anabolic, and anticatabolic effects of

IGF-I on muscle cells are mediated through specific binding to IGF-IR106. This ligand-receptor interaction activates major intracellular signaling pathways through the mitogen-activated protein kinases (MAPKs), particularly extracellular signal-regulated kinase (ERK), and phosphatidylinositol 3 kinase (PI3K) targets. In downstream succession, PI3K activates protein kinase B (Akt) to trigger anabolic signaling via mTOR, while also inhibiting protein degradation by repressing FoxO family transcription factors107,108. These key signaling pathways in skeletal muscle are indispensable in order to maintain equilibrium, throughout the lifespan, for muscle cell proliferation, differentiation, and inhibition of protein degradation107,109. However, in the normal course of biological aging, the equilibrium that protects skeletal muscle mass in youth becomes unbalanced110. This disruption is caused by decreased levels of IGF-1 and increased age-associated inflammatory signaling via TNFα pathways16. Collectively, these age-induced effects on reduced IGF signaling and increased inflammation contribute to sarcopenia, which is characterized by a progressive loss of muscle mass, strength, and function111. As described in the previous section, multiple research studies (conducted in mice, nonhuman primates, and humans) have demonstrated CR’s ability to attenuate muscle loss and mediate these age-related effects on skeletal muscle. In support of these findings, data generated from our pilot study

indicate IGF-1 gene expression and downstream anabolic signaling pathways (IGF, mTOR, and

MYC) are upregulated, while proteins involved in catabolic processes, inflammatory response, and apoptosis are downregulated in skeletal muscle of CR mice, relative to ad libitum controls

(Table 2.1). We posit these anti-aging effects on skeletal muscle are a result of CR-induced transcriptional adaptations that effectively preserve muscle mass.

Supporting roles of IGFBPs in CR-induced effects to mitigate sarcopenia

IGFBPs 1-7 are ubiquitously expressed and function in a complex fashion to mediate

IGF-1 bioactivity. As carrier proteins, IGFBPs 1-6 regulate the bioavailability of IGF-1 in circulation112,113. Whereas at target tissues, IGFBPs 1-7 act as autocrine/paracrine regulators and can either reduce or facilitate IGF-1 action in local microenvironments29,114,115. Further, extensive evidence has recently elucidated that IGFBPs 1-7 have many IGF-independent actions29. By associating directly with several extracellular receptors, IGFBPs are able to cause a variety of unique effects to induce growth and differentiation and block protein degradation

(independent of IGF-1)34. For example, IGFBPs 3 and 7 are purported to inhibit TNFα-induced catabolic activity by binding to its receptor (TNFR1), thereby preventing protein degradation and apoptosis32. Moreover, IGFBP7 has been shown to directly mediate TNFα activity by downregulating its expression and suppressing its proinflammatory signaling116. Despite the major regulatory functions of IGFBPs 1-7 involving IGF-1 bioavailability and tissue-specific activity, and exerting potent IGF-1–independent effects on cells, it is surprising that IGFBPs 1-7, in the context of CR, remain poorly characterized. To our knowledge, the only published data reporting on CR-induced effects on circulating IGFBPs is from the CALERIE trial. Fontana et al. report that, while long-term CR was not associated with changes in circulating total IGF-1,

there was a substantial (21%) and persistent (2 years) increase in circulating IGFBP1 levels in the CR group versus ad libitum40. Similarly, our lab has shown IGFBP7 is significantly increased (p<0.001) in CR mice, relative to ad libitum controls after 10 weeks of diet onset

(unpublished data). In sum, these observations provide evidence that CR may impact levels of

IGFBPs in circulation that favor a reduction in IGF-1 bioactivity, by either binding directly to

IGF-1 (IGFBP3) or through receptor modulation (IGFBP7).

Effect of the CR mimetic, rapamycin on biological aging

The physiological effects of CR to slow the rate of aging and increase both median and maximal life span are important from the standpoint of gaining new insights to improve human health. However, much research effort has also been devoted to pharmacological alternatives that mimic the beneficial effects of CR without requiring dietary limitations. These drug compounds are referred to as CR-mimetics and include resveratrol, metformin and rapamycin117.

Of these compounds, overwhelming evidence suggests rapamycin is most CR-comparable drug associated with universal anti-aging effects – that is, it extends lifespan in all models tested

(yeast to mammals), delays the onset and progression of age-related disease, and protects several organs and tissues against age-related functional decline118–120. For example, rapamycin treatment is associated with cardioprotection, antineurodegenerative effects, cancer prevention and anti-aging effects on skeletal muscle121–125.

Aging muscle is subject to chronic activation of mTOR signaling (a pathway responsible for protein synthesis and regenerative processes)126. This overstimulation of mTOR is purported to increase mitochondrial oxidative stress and inhibition of mitophagy – pivotal mediators of mTOR-induced catabolism26. These processes in turn lead to muscle fiber loss, atrophy, cellular

damage, and deterioration of muscle strength and function (features of muscle quality)26.

Rapamycin acts by inhibiting mTOR signaling, and thus is hypothesized to protect aging muscle from atrophy and loss of muscle quality by reducing mitochondrial oxidative stress and cellular damage related to advanced age in skeletal muscle126,127. Our investigation is designed to assess the relative role of rapamycin, compared with CR, on age-related loss of muscle mass, strength, and function in aged mice.

Chapter II: Table and Figure

Figure 2.1: Schematic of age-related changes in muscle cell signaling that contribute to loss of muscle mass, strength, and function. In youth, IGF-1R activation is increased and downstream signaling PI3K/Akt/mTOR and MAPK is upregulated. Apoptosis and MuRF signaling are decreased. Age-induced effects increase TNFα signaling, apoptosis, and protein degradation. ROS accumulation results in increased oxidative damage and mitochondrial dysfunction. Reduced IGF-1 signaling inhibits protein synthesis, cell survival, differentiation, and proliferation. Signaling associated with youth is green arrow and aged is red arrow.

Table 2.1: Hallmark and BioCarta gene set enrichment in quadriceps muscle from CR versus control mice (6 months post diet onset)

Gene Set Enrichment in Quadriceps Muscle CR vs. control (Hallmark) NES FDR q-val

HALLMARK_MTORC1_SIGNALING 1.304 0.017 HALLMARK_MYC_TARGETS_V1 1.699 0.020 HALLMARK_MYC_TARGETS_V2 1.314 0.020 HALLMARK_TNFA_SIGNALING_VIA_NFKB -1.855 0.001 HALLMARK_APOPTOSIS -1.872 0.001

Gene Set Enrichment in Quadriceps Muscle CR vs. control (BioCarta) NES FDR q-val BIOCARTA_IGF1_PATHWAY 1.41 0.044

CHAPTER III: THE IMPACT OF CALORIE RESTRiCTION OR RAPAMYCIN ON AGE- AND SARCOPENIA-ASSOCIATED TRANSCRIPTIONAL SIGNATURES IN SKELETAL MUSCLE

Introduction

Sarcopenia, defined as a geriatric syndrome and characterized by reduced muscle mass, strength and/or physical function, is a significant cause of physical disability, poor quality of life, and death128. In the US, sarcopenia affects approximately 5-13% of adults 60-79 years and 11-

50% of older adults 80 years and older4. With current older adult population estimates doubling by 2050 in the US, sarcopenia represents a major public health concern, from both a societal and economic front129.

Although the precise molecular mechanisms associated with sarcopenia are unknown, emerging evidence suggests multiple pathophysiological processes are involved, including a reduction in growth factor signaling (IGF-1), infiltration of pro-inflammatory cytokines (TNFα), and oxidative damage due to the accumulation of reactive oxygen species (ROS)20,130–132. These molecular changes, in turn, lead to reduced muscle quality and muscle quantity24,133. Muscle quality is defined as muscle strength (or power) per unit of muscle mass and its loss is intimately linked to metabolic dysregulation, including reduction in insulin sensitivity, impaired oxidative defense, and decreased mitochondrial function134. Age-related loss of muscle quantity is an overall reduction of muscle mass to body weight ratio, or relative muscle mass. Reduced relative muscle mass is widely used as an index of sarcopenia and independent risk factor for loss of muscle function, adverse health outcomes, and increased mortality. At a more granular level,

loss of muscle quantity associated with age is characterized by the loss of muscle fiber number and size, especially type II (fast-twitch glycolytic) muscle fibers responsible for high-intensity/or highly fatiguing activities47. Compared with type I (slow-twitch oxidative) muscle fibers, utilized for endurance-type of activities, loss of type II muscle fibers result in reduced muscle strength and decreased ability to perform activities of daily living such as rising from a chair, climbing stairs, getting dressed, and maintaining personal hygiene47. The detrimental consequences for an individual with sarcopenia strongly impact societal goals to promote healthy aging, improve functionality and quality of life, and reduce disability and dependency135.

Research efforts focused on understanding biological factors and molecular changes associated with age-related loss of muscle quality and quantity may facilitate the development of novel strategies for the prevention and/or treatment of sarcopenia.

One particularly interesting question under investigation considers if sarcopenia is amenable to intervention136. As such, one nonpharmacological intervention that seems promising is calorie restriction (CR)137. CR is a dietary intervention that reduces daily caloric intake without malnutrition and is the only intervention to date that consistently decreases the rate of biological aging and increases both mean and maximum lifespan53. The effects of CR are realized through multiple biological adaptations that are characterized by both synergistic and mutually non-exclusive properties55. The net effect of these physiological changes is a reduction in oxidative damage, inflammation, autophagy and improved energy metabolism, insulin sensitivity, and mitochondrial metabolism138. With respect to skeletal muscle, this multiplex perspective speaks to the efficacy of CR on a range of physiological systems, which ultimately amount to an inhibitory effect on age-related muscle loss138. CR enhances mitochondrial and stem cell function and repair mechanisms to preserve muscle quality, while maintaining muscle

quantity of both muscle fiber types (I and II)139. While CR results in an initial loss of whole body mass (muscle or lean mass and fat mass), this reduction is acute and remains stable in advanced age140. In studies conducted in preclinical models (mice and nonhuman primates) to human clinical trials, the observed effect of CR on muscle mass demonstrates that after an initial loss of absolute muscle mass, the proportion of muscle mass as a percentage of body mass

(referred to as relative muscle mass) increases12,13,15. Increased relative lean mass is a potent preventative parameter for multiple metabolic diseases and a more meaningful marker of muscle preservation and function than absolute mass141.

Understanding the multidimensional effects of CR on age-related muscle loss is the subject of much research. Notably, CR-induced changes on skeletal muscle occur at a molecular, cellular, and systemic level and are likely associated with gene expression changes142.

For example, gene expression analysis from CR studies conducted in nonhuman primates have provided clues to the metabolic connections between CR and age-related muscle loss. Rhoads et al. reported profound CR-induced changes in gene expression resulting in metabolic shifts within the cellular environment96. These molecular adaptations were then linked to pathways involved in muscle protein synthesis and breakdown, RNA processing, and lipid synthesis96. To extend the reports in nonhuman primates to mice and humans, we conducted an analysis of gene expression data using both targeted and untargeted approaches to derive the common genetic impact of CR-induced changes in skeletal muscle in mice and humans. We hypothesize that gene expression of metabolic, myogenic, and IGF-associated gene targest are upregulated in skeletal muscle in CR groups, relative to ad libitum, in both mice and humans. Further, we posit gene regulatory factors linked to mitochondrial function will be enhanced in CR groups. Taken together, these gene expression profiles of CR-induced effects on skeletal muscle of multiple

species may serve as a source by which cross-species comparisons can be made and provide a valuable tool for aging research. Additional anti-aging interventions that are receiving scientific attention are CR mimetics. These pharmacological compounds including resveratrol, metformin, and rapamycin have shown similar CR-related longevity benefits without the need for diet restrictions117. Of these pharmacological agents, rapamycin appears to be the most promising with respect to anti-aging effects on skeletal muscle127. To investigate the potential role of rapamycin as an anti-aging alternative to CR, we conducted a study in C57BL/6 J mice to assess the effects of rapamycin on age-related changes in skeletal muscle associated with loss of muscle mass, strength, and function, relative to control and CR. We hypothesize that the effects of rapamycin will closely align with CR with respect to physical function and genetic indices associated with muscle quality.

Materials and Methods

A. Calorie Restriction Effects On Transcriptional Signatures in Skeletal Muscle

Mouse Study Design

Male C57BL/6 J mice (7 weeks of age; purchased from Janvier [Cedex, France]) were housed in pairs (12-h light/dark cycle and light on at 4 a.m.). Mice were provided with ad libitum access to water and received a standard American Institute of Nutrition (AIN)-93 G diet

(Research Diet Services, Wijk bij Duurstede, The Netherlands). After a 14 day acclimation period (at 9 weeks of age), mice were individually housed and randomized to one of two diet intervention groups: (1) Control diet (AIN-93 W), provided ad libitum (n = 18) or (2) CR diet

(AIN-93W-CR), providing 70% of daily food intake (n=18). Daily food portion size of the CR group was adjusted at the age of 6, 12, 18, and 24 months and based on measurements consumed

in the control group on the week prior to the switch. The CR daily food portions were provided once a day, just before the start of the dark cycle, at 3:30 p.m. CR diets were supplemented with

100% vitamins, essential fatty acids, and minerals to match the control diet. Body weight was recorded bi-monthly. Six animals per group were euthanized at the age of 6, 12, and 24 months.

Skeletal muscle (quadriceps) were harvested and processed (flash frozen in liquid nitrogen) at time of sacrifice and stored at -80°C.

The mouse CR study was conducted by our collaborators at Wageningen University and

Research as part of a collaboration with our laboratory; tissues were sent to UNC for processing and analysis.

Human Study Design

CALERIE was a multisite single-protocol study conducted at three study sites: Pennington

Biomedical Research Center (Baton Rouge, LA, USA), Washington University School of

Medicine (St Louis, MO, USA) and Tufts University (Boston, MA, USA). Duke Clinical

Research Institute (Durham, NC USA) served as the central coordinating center for the study.

All participants provided written informed consent and received financial compensation. A data and safety monitoring board provided oversight of the study, and institutional review board approval was obtained by each of the individual study sites. A sample size of 218 male (21-50 years) and female (21-47 years) nonobese participants were enrolled, and assigned to either the

CR intervention or an ad libitum control group. A 2:1 allocation ratio in favor of the CR intervention was applied to maximize the number of subjects receiving the intervention of greater scientific interest and in consideration of the potential nonadherence rate. Participants in both treatment arms were followed over a period of 24 months. The active intervention targeted a

sustained 25% restriction in calorie intake vis-à-vis ad libitum energy intake measured by doubly labeled water (DLW) at baseline. The CR intervention was implemented by a multidisciplinary team including dietitians, psychologists, and physicians. No specific diet composition was mandated. Rather, the CR intervention was tailored to the needs of the individual participant, with specific nutritional and behavioral strategies employed. A comprehensive set of evaluations were performed prior to initiating the intervention, with follow-up evaluations at Months 1, 3, 6,

9, 12, 18 and 24 after randomization. Body weight was measured in duplicate in the morning following a 12-hour fast, with the participant wearing a hospital gown and no shoes. Fat mass and fat-free mass was measured by dual-energy x-ray absorptiometry (DXA; Hologic 4500A,

Delphi W or Discovery A scanners; Marlborough, MA, USA) with all scans analyzed at

University of California, San Francisco (San Francisco, CA, USA). Scanner performance was monitored with baseline and longitudinal phantom cross calibrations. Tissue extractions (muscle biopsies) of the quadriceps muscle were conducted at Baseline, 12 and 24 months. Muscle tissue was flash frozen in liquid nitrogen and stored at -80°C.

B. Effects of CR Versus Rapamycin On Transcriptional Signatures in Skeletal Muscle

Mouse Study Design

Eight-month-old male (n=45) and female (n=45) C57BL/6J mice were purchased from the Jackson Laboratory (Sacramento, CA). Two female mice were euthanized before the study began due to neurological symptoms. The remaining 43 female mice were included in the data analysis. Mice were housed in standard laboratory cages with alpha-dry bedding and enrichment and maintained on a 12:12 light-dark cycle. During acclimation and baseline testing, mice were provided with ad libitum access to water and a purified control diet (D12450J) formulated by

Research Diets, Inc. Mice were allowed to acclimate for one week prior to the commencement of balance beam testing. Body composition at baseline was conducted using quantitative magnetic resonance imaging (qMR) (UNC Nutrition Obesity Research Center: Animal

Metabolism Phenotyping Core, Chapel Hill, NC). All procedures concerning the use and care of animals were approved and monitored by the Institutional Animal Care and Use Committee at

University of North Carolina at Chapel Hill.

Diet Intervention

After the acclimation period, baseline functional testing, and qMRs were conducted, mice were randomized to one of three diet groups: 1) control diet (D20011502), provided ad libitum;

2) Calorie restricted (CR) diet (D20011503) containing 70% of daily caloric intake and provided once a day between 9-11 am; or 3) Rapamycin diet (D20011501) containing an enteric, encapsulated form of rapamycin (eRAPA) at a concentration of 14 mg/kg, provided ad libitum.

Both the control and CR diets contained the vehicle compound, Eudragit. The micronutrient content of the CR diet was matched to the control diet providing 100 % of all vitamins, minerals, fatty acids, and amino acids. Mice were given ad libitum access to water. All diets were formulated by Research Diets, Inc.

Anthropometric and Physical Function Procedures

Body weight was recorded weekly. Body composition (lean mass and fat mass) was measured at baseline using qMR. Physical function was assessed using grip strength, inverted screen, and balance beam performance tests. On the day of functional testing sessions, mice were transferred in their home cages from the mouse housing room to the experimental room 30

minutes prior testing. This allowed mice to fully awaken and acclimate to the experimental environment. Functional testing was conducted between the hours of 10:00 am and 2:00 pm at baseline and endpoint (time of euthanization).

Grip strength test is used to measure forelimb grip strength. The grip-strength meter is a calibrated grip strength apparatus (Panlab), consisting of a grasping trapeze connected to a force transducer. Strength measurements included a set of three maximum effort repetitions. Mean and maximum grip strength were determined from the three trials.

Inverted screen test is an assessment of coordination and whole body strength. Mice were placed individually on top of a square (7.5 cm x 7.5 cm) wire screen that was secured within a wooden frame. The screen was inverted and suspended 1 meter from the floor.

Bedding was placed under the screen to prevent injury when the mice fell from the screen. The test concluded when the mouse fell from the screen and the maximum hang time was recorded.

Balance beam test assesses agility, balance, and coordination. The balance beam apparatus was composed of two smooth wooden beams, each with a 1 meter in length and 6 mm or 12 mm in width. The beams were securely suspended 50 cm from the floor using sturdy trestles. Enclosed safe houses were placed at the escape ends of each of the two beams and bedding was added to encourage the mice to enter. To prevent injury to animals should they fall off the beam, a net was tethered to the trestles 25 cm below the beam. A distance of 80 cm was marked on the beam, indicating the length at which the animal is tested, with an extra distance of

10 cm behind the starting line to allow a space for the mouse to be placed on the beam and 10 cm

after the finish line in case the mouse paused or hesitated right before entering the safe house.

Mice that paused or hesitated were retested up to 3 times on testing session 1 and up to 5 times on testing session 2 per beam. Mice were placed on the beam and a stopwatch was used to time the mice from the start line to the finish line and a video camera recorded foot slips. The video camera was placed on a tripod 50 cm from the starting end of the beam and positioned to view the mouse from behind. This position allowed the cameras’ point of view to capture the entire length of the beam and enabled the investigator to observe right and left foot slips.

mRNA Extraction

Sections of frozen skeletal muscle (quadriceps) (~50 mg) were homogenized using bead- based Qiagen TissueLyser II in l mL TRIzol™ Reagent (Invitrogen™ ThermoFisher

#15596026). Following homogenization, mRNA was extracted using Qiagen RNeasy Mini Kit

(Qiagen #74104) and stored at -80°C.

Affymetrix Gene Expression Microarray Analysis

The Functional Genomics Core at The University of North Carolina at Chapel Hill performed the expression profiling and genotyping using an Affymetrix Clariom™ S microarray, specifically for mouse or human species. The Affymetrix assay plates were read on a Beckman

4.0 software, Gene Set Enrichment Analysis (GSEA), SAS version 9.4, RStudio computing software, and Gene Cluster 3.0, Coulter’s Biomek® FXP Target Prep Express robot and the

GeneTitan Instrument from Affymetrix. Gene Expression Analysis was conducted using

Transcriptome Analysis Console (TAC).

Consensus clustering (or aggregated clustering) is a robust approach that relies on multiple iterations of the chosen clustering method on subsamples of the dataset. By inducing sampling variability with subsampling, this provides a metric to assess parameter decisions (i.e.

K and linkage) that determine the number of clusters in the data and to test the stability of the discovered clusters. Additional information about consensus clustering is provided in figure legend.

Statistical Analysis

General consideration: The study labelled Calorie Restriction Effects On

Transcriptional Signatures in Skeletal Muscle assesses similarities and differences in the transcriptional response to ad libitum control vs. CR diet regimens in muscle RNA across species

(human / mouse) and time (6 / 12 month). Six mice were included in the analysis at each time (6

&12) and CR (CR & control) group, for a total of 24 mice. Using the CALERIE subjects, muscle RNA was available for 14 control and 25 CR participants for analysis at both the 6 and

12 time points. Gene expression analysis included a next generation transcriptome-wide gene- level expression profiling assessment of 22,206 annotated gene. As a first step, gene expression was assessed in the mouse and human models using Transcriptome Analysis Console (TAC) software. The differentially altered genes between species (mice and humans) and study groups

(CR and control) were identified. Application of false discovery rate (FDR) adjustment was employed to control for potential Type-I error rate inherent in the testing of multiple outcomes. Gene Cluster 3.0 analysis was also employed to group genes that were differentially expressed into unique clusters. We then utilized Gene Set Enrichment Analysis (GSEA) software to examine biological pathways enriched in CR mouse and human samples compared

with control. Genes identified in these pathways were retained for subsequent analysis to develop a meta-analytic model, which calculated a summary of homogeneity and heterogeneity effects to both species.

The study labelled Effects of CR Versus Rapamycin On Transcriptional Signatures in Skeletal Muscle involves mice only using a 2 (sex) by 3 (diets: control, CR, and rapamycin) block design with two repeated measures for some measures (e.g., performance and weight), and single measures for others (including post-mortem gene and muscle measures). For the repeated measures, the pre- and post-outcome measures were transformed to change scores (post-baseline) and analyzed by OLS regression. Following good statistical clinical trial practice143, baseline was incorporated into the analytic structure as a covariate. With the exception of the baseline covariate in repeated measures, the analytic structure was be constant across each of the primary and secondary outcomes. By ANOVA, for each outcome, main effects and the diet group by sex interaction was assessed. If statistically significant, follow-up contrasts were computed to assess where the gender and diet differences existed. Since the test of the interaction was nonsignificant, only main effects of sex and diet will be tested on (1 and 2 df, respectively) with pairwise comparisons of the difference in change for the 3 diets. Additional statistical analysis performed using GraphPad Prism 7 for Windows, version 7.0c (GraphPad Software Inc.).

Significance was determined by p<0.05.

Results

A. CR induces changes in skeletal muscle mass of mice and humans

Mouse CR Study

The original study, included a cohort of male C57BL/6 J mice fed control AIN-93G diet

ad libitum (n=89 control) and or the 30% CR regimen (n=117 CR). From the original study, our analysis includes a subset of n=18 control and n=18 CR mice. At the onset of the study, 9-week- old mice were randomized into control or CR (30% restricted) diet groups. At 6, 12, and 24 months, a random sample of mice were euthanized [n=12 per time point and n=6 per diet group

(control / CR)]. Body weight (bi-weekly) and body composition (DEXA at baseline and endpoint) were obtained throughout the study and are reported elsewhere13. In brief, CR animals weighed less and had lower absolute lean muscle mass. However, relative lean mass (lean mass as a percentage of body mass) was significantly increased in CR mice in comparison to ad libitum-fed controls. Figure 3.1A-C reports the weights recorded at baseline and respective euthanasia time point (6, 12, or 24 months). While body weight was not different between control and CR groups at initiation of CR, at all subsequent time points, control mice were significantly heavier than CR mice. This indicated that CR abrogated weight increase as mice aged. Indeed, CR mice weight was not significantly different to that at initiation.

Human CR study

CALERIE was a multicenter randomized, controlled trial aimed at evaluating the time- course effects of 25% calorie restriction. Muscle biopsies were collected at Baseline, 12, and 24 months. Of the 218 participants (n=75 control; n=143 CR), this substudy includes samples from

39 participants (n=14 control; n=25 CR). Anthropometric outcomes for the complete cohort

(n=218) have been previously reported12,87. For the subset of participants included in this analysis, body weight and body composition (DEXA) collected at baseline, 12, and 24 months are presented in Figure 3.2A-D. In line with previous studies indicating that low compliance and high heterogeneous response in human subjects to CR limit weight loss effects of CR, we

did not observe a significant reduction in body weight or increase in relative lean body mass in this subset of men (n=16) and women (n=23) subjects. However, in the complete CALERIE data set, Das et al. reported weight change in the CR was significantly decreased (-7.6 +/- 0.3 kg)12.

B. Gene expression analysis reveals metabolic networks are engaged by CR in skeletal muscle of mice and humans

CR has been widely reported to preserve muscle function in humans and mice, yet much more dramatic body compositional changes are observed in mice than humans as evident in our data. Hence, we sought to determine whether CR induced, transcriptional reprogramming of muscle metabolism is conserved across mice and humans. Following transcriptomic

(microarray) analysis, we first performed consensus clustering to identify distinct, non- overlapping subpopulations within each data set (mouse and human) based on differentially expressed genes.

Consensus clustering is an unsupervised analysis method that relies on multiple iterations on subsamples of the dataset to aggregate or cluster data based on similarities in the data set.

Use of random sampling with subsampling derives a metric to assess the stability of the clusters and the parameter chosen (i.e. K). The heat map provides a visual component of this analysis.

This analysis was conducted to delineate whether global muscle gene expression changed uniformly following CR at each time point (i.e. resulting in the generation of clusters overwhelming populated by individuals from either the CR or control groups). The mouse transcriptomic data revealed two (i.e. K=2) unique subpopulations (Figure 3.3A-E). In Figure

3.3B, consensus matrix heat maps indicate for each possible number of clusters (2-10), K=2 was determined to be optimal as it maximize the consensus score. Figure 3.3C-D confirm K=2 as

optimal though demonstration of maximal stability of Cumulative distribution function. Figure

3.3E indicates that at assignment of individuals at K=2 remains robust through addition of additional clusters. Thus, we conclude that K=2 generated the most robust segregation of individuals based on these data. Clustering strength was determined using a silhouette plot and

SigClust, demonstrating K1 and K2 to be significantly separated (Figure 3.3F-G), indicating that each individual was robustly identified as a member of their respective cluster. Hence, we concluded that these individuals were well described by their clustering. The heat map (Figure

3.3H) shows the majority of samples identified in K2 were derived from control mice and all

CR-derived samples were uniquely clustered in K1. Each time point (6, 12, and 24 months) further segregated within each cluster. Indicating that while CR induced changes in muscle transcriptomics over time, the most robust segregation of individuals was achieved between CR and control.

Analysis of the human data revealed three (K=3) unique clusters (Figure 3.4A-E); however, as clearly shown in Figure 3.4E, cluster K3 is smaller in relative density (marginally less stable) than clusters K1 and K2. Figure 3.4B indicates the consensus matrix heat maps for each possible number of clusters (2-10), K=3 was determined to be optimal as it maximized the consensus score. Figure 3.4C-D indicate K=3 as optimal though maximization of stability of cumulative distribution function. Figure 3.4E indicates that at assignment of individuals at K1-2 remains robust through addition of additional clusters. However, the additional clusters indicate the potential for a small forth cluster in the data, arising from marginally low stability of K3.

Thus we conclude that K=3 generated the most robust segregation of individuals. A silhouette plot and SigClust demonstrate K1 and K3 are significantly separated from K2, with K3 not significantly separated from K1 (Figure 3.4F-G). Hierarchical clustering (Figure 3.4H) shows

that cluster K2 mostly contains samples from the control group and K1 and K3 predominately contain samples from the CR group. In contrast to our mouse analysis, times points (baseline,

12, and 24 months) assessed in the human samples did not cluster within diet treatment.

We conclude that CR in both humans and mice induces robust transcriptomic changes that readily segregate control and CR treated individuals via consensus clustering. Our mouse analysis demonstrated stronger overall clustering and robust subclustering of time points indicating a lower degree of heterogeneity than that seen in human samples where clustering was marginally less robust, and time points were not further subclustered. Genetic, environmental, dietary, and social differences in human subjects may contribute to high degrees of heterogeneity in addition to modest compliance with CR protocols.

Having established that CR altered muscle transcriptomes of both mice and humans, we next sought to determine which pathways and processes may be activated/repressed by CR. To investigate the effect of CR on biological pathways within skeletal muscle of mice and humans, we performed gene set enrichment analysis using subcollections of the Molecular Signatures

Database (MSigDB), including Kyoto Encyclopedia of Genes and Genomes (KEGG), Reactome,

Gene Ontology (GO), and BioCarta. These library tools provide interpretation, analysis, and translation of Omics data into functional components and biological pathways. The KEGG collection represents molecular interactions and high-level function of biological system and the

Reactome model network of biological reactions and group into pathways. GO and BioCarta feature common metabolic, biochemical, and signal transduction pathways for gene set enrichment144. GSEA analysis of mouse samples indicated a profound metabolic reprogramming of muscle tissue in CR with numerous metabolic gene sets significantly enriched between control and various CR time points. Fatty acid and lipid metabolism processes were prominent, which

may be particularly relevant to the CR phenotype given CR’s preferential reduction in fat mass.

Several other CR related pathways were enriched following CR including IGF-1 and mTOR signaling, development, and reproduction (Supplemental Table 3.1S (A-D)). For the human data, due to the small sample size (n=39), great variability in human genetics, diet, and lifestyle, and possible confounding due to compliance to diet, the false discovery rate (FDR) corrected q value threshold for the human data was set to <0.25 for GSEA. Using this threshold, enriched pathways in the human data were associated with metabolic processes including histidine, phenylalanine, and bile acid metabolism. In addition, antioxidant responsive pathways including xenobiotic and drug metabolism were enriched in CR which supports previous reports that purport antioxidant defense is a mechanism by which caloric restriction (CR) increases longevity145 (Supplemental Table 3.2S (A-D)).

Next, to determine shared phenotypic responses to CR in mouse and human samples, we collated molecular pathways that were coherently regulated at 12 and 24 months for both species

(Figures 3.5 and 3.6; Tables 3.1A-B and 3.2A-B). We found remarkably coherent upregulation of energy metabolism, specifically central carbon metabolism and mitochondrial bioenergetics.

Pyruvate metabolism and TCA cycle was significantly enriched in the CR group of both species at 12 month time point, concurrent with histidine and phenylalanine metabolism. Fatty acid and bile acid metabolism were enriched in the CR groups from both species at the 24 month time point in addition to pentose phosphate metabolism, tyrosine, and retinol metabolism. Several pathways involved in the detoxification of drugs were similarly enriched in the CR groups.

Electron transport and uncoupling heat production was negatively associated with both species at

12 months, indicating reallocation of the energetic payload of central carbon and fatty acid metabolism. Thus, we conclude that metabolic reprogramming of muscle by CR is conserved

across mice and humans. Much of this metabolic reprogramming was temporally aligned between species indicating the potential for dynamic regulation of metabolism by CR. This is of particular interest given the overall modest overlap in conservation of phenotype between humans and mice. Table 3.3A-D provides the gene list for each of the Reactome and KEGG pathways enriched for mice and humans. As indicated by the blue highlighted cells, numerous genes overlap in >2 pathways.

C. Influence of CR and rapamycin on age-related loss of muscle mass, strength, and function

Data analysis included male (n=45) and female (n=43) C57BL/6 J mice. At ~8 months of age, mice were randomized to one of 3 diet groups: control (n=15 males, n=14 females), rapamycin (eRapa) (n=15 males, n=14 females) or CR (n=15 males, n=15 females). Weekly food intake and body weights are reported in Figures 3.7 and 3.8A-B, respectively. As expected, CR mice consistently exhibited lower body mass compared with mice fed control or rapamycin-supplemented diets. There were no differences observed in body mass or food intake

(Figure 3.8) between Control and Rapamycin animalsFasting blood glucose (FBG) was significantly lower in CR mice, relative to control and rapamycin groups in both sexes (Figure

3.7C-D). These results align with expected results for these interventions, and indicate that the low dose of rapamycin provided was not sufficient to elevate blood glucose or alter body weight in these animals. For CR male mice, absolute mass of the quadriceps (quad) and gastrocnemius

(gastroc) muscle were significantly reduced compared with control-fed and rapamycin-treated male mice. Tibialis anterior (TA) was reduced in CR males, relative to rapamycin-treated male mice. When muscle mass was normalized to whole body mass, CR males had increased relative muscle mass for each muscle type, relative to control and rapamycin. For female CR mice, there

was a significant reduction in all muscle types compared with control and rapamycin. Female

CR mice did not display altered relative lean mass for any muscle tested. (Figure 3.8A-D). CR has been reported to promote loss of both lean and fat mass, but preferentially induces loss of fat mass. Our findings from male mice concord with these findings (primarily derived from male mice); however, the female mice in our study appeared to lose both lean and fat mass at similar rates.

Muscle force generation is derived through both muscle mass and muscle function.

Hence, we conducted a series of assays of muscle function in these mice to disentangle the CR effects on muscle function and muscle mass. Functional measures indicate that CR and rapamycin have a protective effect on muscle strength and function. Despite the differences in absolute muscle mass (quads, gastroc, and TA muscles) between CR compared with control and rapamycin, mean grip strength did not differ between CR compared with rapamycin and control groups. However, for females in the rapamycin group, strength was significantly increased relative to control (Figure 3.9A). The literature suggest that CR increases strength capacity [unit of force (newtons) per body mass (g)] compared with control13, hence we normalized grip strength to animal mass and found that CR increased strength capacity in males and females

(Figure 3.9A). To investigate agility, balance, and coordination, a balance beam test was performed (Figure 3.9B). This test determines the speed with which a mouse will cross a narrow beam (6/12 mm), with faster times indicating higher agility. Overall, across both beams

(6 mm and 12 mm), we found mice in the intervention groups (CR and rapamycin) were significantly faster than control mice. These results suggest CR and rapamycin treatment prevented age-related decline in function, with respect to balance and agility. This finding supports a previous study that demonstrated improved motor coordination and balance in CR

rodents, relative to control13. Whole body strength and coordination were assessed with the inverted screen test (Figure 3.9C). Maximum hang time for both females and males was significantly increased in CR compared with control and rapamycin. To determine if the increased hang time observed in the CR mice could be explained by weight differences between groups (CR mice weighed ~30% less than control and rapamycin mice), we computed holding impulse to correct for the effects of body mass on hang time. Results indicate hang time for CR females remained significantly greater than control and rapamycin females. For males, hang time was not different to control or rapamycin. Taken together, these data indicate that in male mice CR promotes preferential retention of muscle, and improved muscle function. In female mice, CR did not preferentially protect lean mass from weight loss, but did promote increased measures of muscle function, thus implicating a role for CR beyond a simple increase of lean mass per unit body weight. In contrast, Rapamycin only promoted improvements in agility (as measured by balance beam and otherwise did not demonstrate the breadth of such coherent enhanced muscle function observed with CR.

D. Effects of rapamycin and CR gene expression in skeletal muscle

To explore the effects of diet/treatment on the transcriptional changes in skeletal muscle we conducted microarray analysis using mRNA from the quadriceps muscle. By applying principle component analysis (PCA) we found the gene expression patterns of female and male mice exposed to CR formed distinct clusters, whereas control and rapamycin displayed a more similar transcriptional profile (Figure 3.10A and Figure 3.11A). Hierarchical clustering displayed in the heat map further shows a clear and distinct separation of groups based on dietary/treatment manipulation. As represented in the dendrogram at the top of the heat map, the

greatest separation between groups is between CR and control or rapamycin for both males and female mice (Figure 3.10B and Figure 3.11B). Analysis of sex-specific differences between differentially expressed genes indicated a substantial increase in the number of genes up and down regulated in males versus females in the rapamycin group compared with control or CR

(Figure 3.12).

Next, to evaluate the effects of diet/treatment on biological signaling pathways, we conducted GSEA analysis. Using MSigDB Hallmark curated gene set list, results from GSEA indicated there were striking differences between male and female mice (Figure 3.13), through comparisons of diet groups. Differences in transcriptional profiles between male and female mice in the rapamycin group were most surprising for mTOR signaling. Given the role of rapamycin to inhibit mTOR signaling, we expected the hallmark pathway, mTORC1 to be downregulated in mice (males and females) exposed to rapamycin. For males in the rapamycin group, as expected, mTORC1 signaling was negatively enriched compared with control

(FDR=0.01) and CR (FDR=0.01) (Table 3.6). However, for females in the rapamycin group, mTOR signaling was not enriched, relative to control (FDR=0.83) or CR (FDR=0.94) (Table

3.7). This may be influenced by sex differences in the pharmacokinetics or pharmacodynamics of rapamycin, differences in food consumption, or sex differences in regulation of mTOR signaling independent of rapamycin.

GSEA revealed negative associations of EMT, cholesterol homeostasis, and inflammatory related gene sets in male CR mice compared with control. EMT was similarly negatively associated with CR in female mice when compared with control. By comparison, hallmark pathways found to be commonly enriched in rapamycin males compared with control and CR males included xenobiotic metabolism, cholesterol homeostasis, mTOR signaling, and

adipogenesis (Table 3.6). The enrichment for xenobiotic metabolism in all rapamycin treated samples compared to all others indicates that at least part of the transcriptomic response to rapamycin may direct its elimination from the muscle and thus may add complexity to interpretation. Taken together, these data indicate that while CR induced relatively few gene set enrichments relative to control, EMT was coherently regulated between CR and control in both males and females. These mice also demonstrated the most robust improvements in muscle function on testing, thus the coherent expression of EMT like makers may indicate a novel mechanism through which CR may mediate its pro-muscle effects. Regulation of inflammatory signaling may also play and important role in mediating these effects, at least in male mice, as various inflammatory mediators have been shown to promote muscle breakdown and limit myogenesis146. While rapamycin induced more gene set enrichments the functional significance of these is less clear as rapamycin induced limited protective effects on muscle. Additionally, in gene set enrichment analysis between male and female mice for each diet/treatment group, results revealed nutrient-sensitive metabolic pathways and inflammatory/antioxidant focused signaling were enriched for female mice exposed to rapamycin, relative to males (Table 3.8).

Notably, PCA and hierarchical clustering revealed a clear distinction between diet/treatment groups. While pathway analysis using GSEA did not reveal the same strong distinction between diet/treatment groups, there were intriguing sex-dependent transcriptional changes in skeletal muscle for all groups. Overall, we observed a sex-dependent effect on gene expression analysis and a shift (downregulation) in gene set enrichment of metabolic signaling pathways in male rapamycin mice compared with control and an enrichment of inflammatory response adaptations in female mice exposed to rapamycin, relative to CR and control.

Discussion

Sarcopenia is a major adverse health condition associated with the process of biological aging and contributes to poor outcomes at both individual and population levels. The effects of

CR to oppose the progression of aging likely result from diverse biological adaptations acting at molecular, cellular, and systemic levels, rather than a single mechanism or causal pathway53,62.

In skeletal muscle, CR is purported to reduce or delay many of the deleterious effects related to age, including loss of muscle quantity (mass) and muscle quality147. Based on the results presented here in mice, gene expression analysis suggests that CR intervenes with age-related loss of muscle mass via pathways associated with regeneration, maintenance, and repair.

Specifically, we found IGF-1 and mTOR signaling pathways were enriched in skeletal muscle of

CR mice compared with control. These findings both contradict and support previous reports.

For instance, in contrast to our results, a study conducted by Masternak et al., found IGF-1 gene expression was downregulated in skeletal muscle of CR rats148. However, similar to our observations, Kim et al. reported an increase in IGF-1 gene expression in skeletal muscle of CR rats and van Norren et al. reported a significant increase in mTOR expression in CR mice13,149.

Systemic reduction of IGF1 is well established in rodent models of CR, yet the role for muscle derived IGF-1 transcription is more ambigious. Through the use of GSEA we have implicated a coordinated program of IGF-1 and mTOR signaling in muscle following CR.

Although gene expression analysis from human samples did not replicate the same strong associations between CR and increased IGF-1/mTOR signaling in skeletal muscle (possibly due to confounding issues inherent to human clinical trials such as protocol adherence), we found processes related to muscle contraction were downregulated in both species. The process of muscle contraction is generated within muscle tissue, and results in alterations of muscle

geometry, such that increased force is exerted on the tendons150. CR may alter changes in muscles to improve efficiency, rather than increasing its size, hence these differences may reflect force production arising from muscles that are structurally different16. Indeed in our animal data we demonstrate increase muscle performance from lower muscle mass, indicating that for at least some assays of muscle function muscles from CR mice generate more force per unit mass than muscle from control mice. Additionally, biological pathways associated with metabolic regulation and mitochondrial function (i.e. features associated with muscle quality)147 were enriched in skeletal muscle of both mice and humans receiving the CR regimen, relative to control diets. Specifically, energy metabolism, central carbon metabolism, and mitochondrial bioenergetics were upregulated in both species. These findings support a role for enhanced capacity of central carbon/TCA metabolism and may reflect a central node of CR response in muscle of mice and humans. These results are in agreement with previous research that demonstrates adaptations associated with CR promote enhanced bioenergetic metabolism and mitochondrial function151,152. Such findings are of critical importance in the context of aging as numerous studies have implicated decline in metabolic/mitochondrial fitness as key mediators of muscle decline with age

A reduction of oxidative stress is linked to the anti-aging effects of CR and intriguingly, xenobiotic metabolism was enriched in CR groups in both species. Xenobiotic metabolism protects tissues from the damaging effects of ingested toxins and harmful endogenous molecules produced through normal metabolism153. Research suggest that increased expression of xenobiotic processes decreases the rate of age-dependent decline in tissue homeostasis154. Our results align with previous reports that suggest CR increases xenobiotic enzymatic activity154, however; this is the first study, to our knowledge, that has observed an increase in xenobiotic

metabolism in skeletal muscle of CR mice and humans. Gene enrichment analysis also revealed electron transport and uncoupling heat production was downregulated in both species, resulting in reduced thermogenesis. This finding is in line with previous research that has shown CR alterations include a reduction of heat energy and core body temperature53. Taken together, these results suggest IGF-1 and mTOR have a role in CR effects to regenerate and repair muscle in mice. Although we did not detect a role for IGF-1 and mTOR signaling in human samples, our findings do suggests that CR induces a shift in transcriptional reprogramming that supports enhanced metabolism and improved mitochondrial function in both mice and human species.

Admittedly, gene expression analyses does not confer mechanistic associations or reveal casual relationships between differentially expressed genes and the effects of CR on skeletal muscle. Yet, genome-wide investigations and pathway-based analyses are powerful tools for identifying novel, and possibly unexpected, associations that warrant further experimental exploration. Hence, employing a globally centered approach to assess gene expression changes induced by CR on age-related changes and metabolic features linked to muscle mass and quality provides insight to the molecular underpinnings associated with the anti-aging effects of CR on skeletal muscle. Identifying gene expression patterns that are uniquely as well as commonly altered by CR in skeletal muscle of mice and humans marks an important milestone in laying the foundation for understanding the functional contribution of CR-induced transcriptional changes associated with improved health and longevity. On-going analysis and the development of quantitative models or bio-signatures based on whole-genome wide microarray datasets of CR- induced changes in skeletal muscle will provide a valuable tool for aging research and potentially shed light on the mechanistic link between aging muscle and CR. Our work makes important contributions to this field through analysis of mouse and human samples at identical timepoints

we can allow for refinement of preclinical work to remove rodent specific CR effects where appropriate. Further by conducting detailed functional testing in addition to transcriptomic analysis on male and female mice we generate important associations between functional CR effects and transcriptional effects.

Given physiological effects of CR are of great scientific interest from the standpoint of human health, much effort has been devoted to the development of CR mimetics, or drugs that offer beneficial aspects of CR diets, without reducing caloric intake. Of these compounds, rapamycin has been shown to protect several tissues from age-related function decline155,156.

Only recently, have the effects of rapamycin been explored in the context of aging muscle. In a study conducted by Tang et al., chronic activation of mTOR signaling (a common characteristic of aging muscle) was shown to stimulate progressive muscle damage, while inhibition of mTOR via rapamycin was associated with reduced fiber loss.126 In an effort to establish the relative importance of mTOR, in the loss of muscle strength and function associated with age, we treated older (8 month) mice with an enteric form of rapamycin.

For measures of body composition, we observed no difference in body weight or relative muscle mass between rapamycin and control. As expected, CR mice exhibited a loss of both weight and absolute muscle mass. Relative lean mass significantly increased for male mice, but did not differ between female CR mice and control or rapamycin. Although unexpected, this disparity between relative lean mass in CR male versus female mice is supported by previous reports. For example, Boldrin et al. found clear differences in quantitative measures of muscle mass between CR male and CR female mice157. Although the researchers attributed the differences in body composition to sex hormones, the association is still inconclusive. For functional tests, our findings suggest rapamycin and CR may have some protective effects on

loss of balance and coordination, relative to control. Male and female mice exposed to CR or rapamycin significantly improved their time to cross the balance beams (6 mm and 12 mm), relative to control. The measure of normalized grip strength also improved for CR males and females as well as rapamycin females. As a measure of force per unit of muscle mass, increased normalized grip strength suggest a greater strength and functional capacity relative to muscle mass. In support of this finding, van Norren et al. found improved measures of relative muscle strength in CR male mice13. For whole body strength, CR male and female mice had the greatest improvement in maximum hang time, whereas only female mice retained this advantage when body weight was included in the statistical model. With respect to functional tests, our data suggest CR may be partially protective against age-related functional decline and the effects appear to be sex-specific. Rapamycin produced modest if any improvements of muscle function.

Collectively these functional assessments corroborate with previous reports that demonstrate CR attenuates age-related functional decline. To date, studies investigating the effect of rapamycin on functional capacity have been associated with the pathogenesis of disease and not aging per se. Thus, these reports offer novel findings on rapamycin-induced effects on age-related changes in skeletal muscle with respect to strength and function.

At the level of gene expression, the CR-induced transcriptional changes, relative to control are consistent with evidence from previously conducted in vivo studies158. Several studies have reported on the effects of CR driving an altered transcriptional profile compared with control fed mice13,159. For gene expression analysis between rapamycin versus control, although both rapamycin and control groups were fed ad-libitum, we anticipated (a priori), the groups would form clear, distinct clusters. We found that although the two groups clustered, the transcriptional profiles of rapamycin and control were less distinct than we had hypothesized.

A major unexpected finding was revealed in gene set enrichment analysis. Based on previous studies that demonstrate a range of transcriptional adaptations associated with CR in skeletal muscle, we expected to find clear differences in gene set enrichment analysis of metabolic signaling pathways, particularly nutrient-sensitive processing between CR and control13,160. We did, however detect metabolic signaling pathways enriched in CR and control male mice versus rapamycin male mice. These results align with the role of rapamycin to inhibit many (but not all) of mTOR’s downstream effects involving metabolism. For female mice, although changes in gene expression of metabolic transcription was relatively absent, rapamycin treatment appeared to enrich biological pathways involved in inflammation. This finding is in line with previous research that has identified rapamycin as a regulator of immune-related functions161.

In this study, extraordinary diversity was found due to biological variability between males and females, especially in mice exposed to rapamycin. Even though sex-specific differences were predicted, we found that rapamycin had particularly significant effects on gene expression in males versus females compared with CR or control. This theme continued in gene set enrichment analysis of biological pathways altered in males versus females. Although genetic variably due to sex differences is commonly recognized, we are the first study, to our knowledge, to report on the disparate findings between males and female mice exposed to rapamycin.

In conclusion, this work represents the first attempt to identify a CR-induced molecular signature in skeletal muscle common to both mice and humans. Additionally, we revealed several biological pathways responsive to CR that are conserved among mice and human species.

We identify several CR induced metabolic adaptations which are conserved across time points

and species. Further, through paired functional testing and transcriptomic analysis we determine that suppression of an EMT signature is associated with CR muscle protective effects. Our investigation of the therapeutic potential of the CR-mimetic, rapamycin may potentially extend our current knowledge of the mechanisms associated with CR and possibly provide a step towards the discovery of effective strategies for the prevention and management of sarcopenia.

Chapter III: Tables and Figures

Figure 3.1: Mice: Body Weight and Relative Lean Mass

Body Weight Relative Lean Mass A ✱✱✱✱ ✱✱✱✱ B 50 90 ✱✱✱✱ ✱✱✱✱

) 40

% 85

)

(

g s

(

t 30 a

h 80

M i

n a W 75

10 e L

0 70 Baseline 12 24 Baseline 12 24 Months Months

Figure 3.1: Body weight and Relative Lean Mass for Mice (n=6 per time point). (A) Body weight and (B) relative lean mass were assessed at Baseline and time of euthanization at 12 and 24 months. Weight is presented in grams (g) and relative lean mass is presented as percent (%) and calculated by dividing lean mass by total body mass. Regression model was used for analysis to detect differences between control and CR at 12 and 24 months and controlling for baseline weight, ****p<0.0001.

Figure 3.2: Humans: Body Weight and Relative Lean Mass

Men Women

✱ ✱✱✱

100 ✱✱✱ 90 ✱✱✱✱

90 80

)

( (

80 70

g i

i 70 60

W W 60 50

40 50 Baseline 12 24 Baseline 12 24 Months Months

Men Women

✱ ) ) 90 ✱✱✱

90 % (

% ✱✱✱

( ✱✱✱✱

s s

s 85

a 80

80 n

a 70

e e

75 L

e 60

i i

70 t

a a

l e

e 65 50 R R Baseline 12 24 Baseline 12 24 Months Months

Figure 3.2: CALERIE (human): Body Weight and Relative Lean Mass for men (n=16) and women (n=23). Body weight was assessed at Baseline (0), 12 months, and 24 months. Body weight for (A) men and (B) women participants. Relative lean mass for (C) men and (D) women. Weight is presented in kilograms (kg) and relative lean mass is presented as percent (%) and calculated by dividing lean mass by total body mass. Regression model was used for analysis to detect differences between control and CR at 12 and 24 months and controlling for baseline weight, *p<0.05, ***p<0.005, ****p<0.0001.

Figure 3.3 Mice: Consensus clustering

F H

…

Figure 3.3: Consensus Clustering: Mouse data set. (A) Consensus matrix color legend. Consensus values range from 0 (white) and indicates never cluster together to 1 (dark blue) and indicates always cluster together. (B) Heat maps were generated by running multiple iterations on random subsamples of data for every ‘K’ to be tested. For this data set, nine iterations were subsampled (K=2…10). Although there are >2 clusters forming in heat maps K=7…10, the most clearly defined clusters are K=2. The matrices are ordered by the consensus clustering, which is depicted as a dendrogram at the top of the heat map. The colored rectangles placed between the dendrogram and heat map are the cluster memberships and compare a clusters’ member count in the context of their consensus, or cluster boundaries. (C) Consensus Cumulative Distribution Function (CDF) illustrates the cumulative distribution functions of theconsensus matrix for each K (indicated by colors). The plot indicates K=2 reaches an approximate maximum (or maximum stability). (D) Delta area plot shows the relative change in

area under the CDF curve comparing K and K-1. Although 3 clusters formed, the strongest stability was contained in 2 clusters. (E) Sample tracking plot shows the consensus cluster assignment of samples (columns) for each K (rows) by color. The colors correspond to the colors of the consensus matrix class assignments. Hatch marks below the plot indicate samples. This plot provides a view of item cluster membership across different ‘K’s and tracks the history of clusters relative to earlier clusters. Samples with multiple colors (changing colors within a column) indicate an unstable membership and clusters with an abundance of unstable members suggest an unstable cluster. Data indicate there are 2 distinct members, shown in light and dark blue. (F) Silhouette width plot visually represents how well the sample fits into each cluster (closest fit =1). Average of clusters 1 and 2 =0.61. (G) SigClust analysis estimates the ‘uniqueness’ of each cluster (p-value=0). (H) Heat map was generated using hierarchical clustering with K2. Genes identified in control are mostly included in K2 (light blue bar) and genes identified in CR are all in K1 (purple bar). Time points (6. 12, and 24 months) also uniquely cluster for both control and CR groups (grey, green, and dark blue bars, respectively).

Figure 3.4 Humans: Consensus clustering

H F

Figure 3.4: Consensus Clustering: Human data set. (A) Consensus matrix color legend. (B) Heat maps were created using nine iterations (K=2…10) and indicate there are 2 dominate clusters with a smaller, but consistent 3rd cluster. (C) CDF plot indicates K=3 has greatest stability. (D) Delta area plot indicates a 4th potential cluster. (E) Sample tracking plot indicates the human data clusters into 3 unique groups, with K3 (green) of smaller density than K1 (dark blue) or K2 (light blue). (F) Silhouette width plot illustrates the average of 3 clusters = 0.12. (G) SigClust analysis estimates clusters 1-3 are relatively distinct: 1,2 (p-value=0), 1.3 (p- value=0.062* nonsignificant), and 2,3 (p-value =0.025). (H) Heat map generated using hierarchical clustering with K=3. Genes identified in control samples are mostly in K2 (light blue bars) and identified genes in CR are contained predominately in K1 and K3 (purple bars). Clustering based on time points does not appear to have an association with groups of genes formed from diet intervention (grey, green, and dark blue bars, respectively).

Figure 3.5. Venn diagram of overlapping enriched gene sets between humans and mice at 12 months. (A) Reactome and (B) KEGG biological pathway collections.

Table 3.1: Gene set enrichment analysis: Commonly upregulated gene sets in humans and mice at 12 months. (A) Reactome and (B) KEGG, (FDR<0.05).

Figure 3.5. Venn diagram (Reactome)

Table 3.1. Reactome Gene Sets Gene Set Enrichment Common to mice and humans (12 months) Human CR vs. control Mouse CR vs. control (Reactome) NES FDR q-val NES FDR q-val REACTOME_PYRUVATE_METABOLISM_AND_CITRIC_ACID_TCA_CYCLE 1.628 0.014 2.407 0.000

Figure 3.5. Venn diagram (KEGG)

Table 3.1. KEGG Gene Sets

Gene Set Enrichment Common to mice and humans (12 months) Human CR vs. control Mouse CR vs. control (KEGG) NES FDR q-val NES FDR q-val KEGG_HISTIDINE_METABOLISM 1.812 0.004 3.017 0.000 KEGG_PHENYLALANINE_METABOLISM 1.620 0.027 2.969 0.000

Figure 3.6. Venn diagram of overlapping enriched gene sets between humans and mice at 24 months. (A) Reactome and (B) KEGG biological pathway collections.

Table 3.2 Gene set enrichment analysis: Commonly upregulated gene sets in humans and mice at 24 months. (A) Reactome and (B) KEGG, (FDR<0.05).

Figure 3.6. Venn diagram (Reactome)

Table 3.2. Reactome Gene Sets Gene Set Enrichment Common to mice and humans (24 months) Human CR vs. control Mouse CR vs. control (Reactome) NES FDR q-val NES FDR q-val REACTOME_FATTY_ACIDS METABOLISM 1.504 0.018 1.777 0.002 REACTOME_BILE_ACID_AND_BILE_SALT_METABOLISM 1.594 0.009 1.801 0.001

Figure 3.6. Venn diagram (KEGG)

Table 3.2. KEGG Gene Sets

Gene Set Enrichment Common to mice and humans (24 months) Human CR vs. control Mouse CR vs. control (KEGG) NES FDR q-val NES FDR q-val KEGG_PENTOSE_AND_PHOSPHATE PATHWAY 1.683 0.003 1.504 0.018 KEGG_METABOLISM_OF_XENOBIOTICS_BY_CYTOCHROME_P450 1.591 0.010 2.149 0.000 KEGG_TYROSINE_METABOLISM 1.638 0.006 1.919 0.000 KEGG_RETINOL_METABOLISM 1.619 0.004 1.926 0.000 KEGG_DRUG_METABOLISM_OTHER_ENZYMES 1.520 0.024 1.872 0.000 KEGG_DRUG_METABOLISM_CYCTOCHROME_P450 1.500 0.041 1.839 0.001

Table 3.3 Gene list for Reactome and KEGG gene sets commonly enriched in mice and humans. (A) Reactome and (B) KEGG at 12 months. (C) Reactome and (D) KEGG at 24 months. Blue highlight indicates the genes overlap >2 gene sets.

12 Month Upregulated CR / control Enriched Gene Set and Gene List REACTOME_PYRUVATE_METABOLISM_AND_CITRIC_ACID_TCA_CYCLE ACO2 IDH3G LDHA MPC2 PDK3 SDHC ADHFE1 L2HGDH LDHAL6A NNT PDK4 SDHD BSG FH LDHAL6B OGDH PDP1 SLC16A1 CS GLO1 LDHB PDHA1 PDP2 D2HGDH GSTZ1 LDHC PDHA2 PDPR DLAT HAGH ME1 PDHB PPARD DLD IDH2 ME2 PDHX RXRA DLST IDH3A ME3 PDK1 SDHA FAHD1 IDH3B MPC1 PDK2 SDHB

B 12 Month Upregulated CR / control Enriched Gene Set and Gene List KEGG_HISTIDINE_METABOLISM ACY3 ALDH3B1 ASPA HDC MAOB ALDH1A3 ALDH3B2 BUD23 HEMK1 METTL2B ALDH1B1 ALDH7A1 CNDP1 HNMT METTL6 ALDH2 ALDH9A1 DDC LCMT1 TRMT11 ALDH3A1 AMDHD1 FTCD LCMT2 UROC1 ALDH3A2 AOC1 HAL MAOA KEGG_PHENYLALANINE_METABOLISM ALDH1A3 AOC2 GOT2 MAOB PAH ALDH3A1 AOC3 HPD MIF PRDX6 ALDH3B1 DDC IL4I1 NAA80 TAT ALDH3B2 GOT1 MAOA

24 Month Upregulated CR / control Enriched Gene Set and Gene List REACTOME_FATTY_ACIDS METABOLISM REACTOME_Bile_acid and bile salt metabolism AADAT BLVRA ENO2 HSD17B4 PDHB ABCB11 BAAT NCOA1 RXRA ACAA1 BMPR1B ENO3 HSD17B7 PPARA ABCC3 CH25H NCOA2 SCP2 ACAA2 BPHL EPHX1 HSDL2 PRDX6 ACOT8 CYP27A1 NR1H4 SLC10A1 ACADL CA2 ERP29 HSP90AA1 PSME1 ACOX2 CYP39A1 OSBP SLC10A2 ACADM CA4 ETFDH HSPH1 PTPRG AKR1C1 CYP46A1 OSBPL1A SLC27A2 ACADS CA6 FABP1 IDH1 PTS AKR1C2 CYP7A1 OSBPL2 SLC27A5 ACADVL CBR1 FABP2 IDH3B RAP1GDS1 AKR1C3 CYP7B1 OSBPL3 SLCO1A2 ACAT2 CBR3 FASN IDH3G RDH11 AKR1C4 CYP8B1 OSBPL6 SLCO1B1 ACO2 CCDC58 FH IDI1 RDH16 AKR1D1 FABP6 OSBPL7 SLCO1B3 ACOT2 CD1D FMO1 IL4I1 REEP6 ALB HSD17B4 OSBPL9 STARD5 ACOT8 CD36 G0S2 INMT RETSAT AMACR HSD3B7 PTGIS ACOX1 CEL GABARAPL1 KMT5A S100A10 ACSL1 CIDEA GAD2 LDHA SDHA ACSL4 CPOX GAPDHS LGALS1 SDHC

62 ACSL5 CPT1A GCDH LTC4S SDHD

ACSM3 CPT2 GLUL MAOA SERINC1 ACSS1 CRAT GPD1 MCEE SLC22A5 ADH1C CRYZ GPD2 MDH1 SMS ADH7 CYP1A1 GRHPR MDH2 SUCLA2 ADIPOR2 CYP4A11 GSTZ1 ME1 SUCLG1 ADSL CYP4A22 H2AZ1 METAP1 SUCLG2 ALAD D2HGDH HADH MGLL TDO2 ALDH1A1 DECR1 HADHB MIF TP53INP2 ALDH3A1 DHCR24 HAO2 MLYCD UBE2L6 ALDH3A2 DLD HCCS NBN UGDH ALDH9A1 DLST HIBCH NCAPH2 UROD ALDOA ECH1 HMGCL NSDHL UROS AOC3 ECHS1 HMGCS1 NTHL1 VNN1 APEX1 ECI1 HMGCS2 ODC1 XIST AQP7 ECI2 HPGD OSTC YWHAH AUH EHHADH HSD17B10 PCBD1 BCKDHB ELOVL5 HSD17B11 PDHA1

24 Month Upregulated CR / control D Enriched Gene Set and Gene List KEGG_PENTOSE_AND_PHOSPHATE PATHWAY KEGG_METABOLISM_OF_XENOBIOTICS_BY_CYTOCHROME_P450 ALDOA GPI PGM1 RPEL1 ADH1A ALDH3B2 CYP3A5 GSTM5 UGT1A5 ALDOB H6PD PGM2 RPIA ADH1B CYP1A1 CYP3A7 GSTO1 UGT1A6 ALDOC PFKL PRPS1 TALDO1 ADH1C CYP1A2 DHDH GSTO2 UGT1A7 DERA PFKM PRPS1L1 TKT ADH4 CYP1B1 EPHX1 GSTP1 UGT1A8 FBP1 PFKP PRPS2 TKTL1 ADH5 CYP2B6 GSTA1 GSTT1 UGT1A9 FBP2 PGD RBKS TKTL2 ADH6 CYP2C18 GSTA2 GSTT2 UGT2A1 G6PD PGLS RPE ADH7 CYP2C19 GSTA3 GSTZ1 UGT2A3 AKR1C1 CYP2C8 GSTA4 MGST1 UGT2B10 AKR1C2 CYP2C9 GSTA5 MGST2 UGT2B11 AKR1C3 CYP2E1 GSTK1 MGST3 UGT2B15 AKR1C4 CYP2F1 GSTM1 UGT1A1 UGT2B17 ALDH1A3 CYP2S1 GSTM2 UGT1A10 UGT2B28 ALDH3A1 CYP3A4 GSTM3 UGT1A3 UGT2B4 ALDH3B1 CYP3A43 GSTM4 UGT1A4 UGT2B7 KEGG_TYROSINE_METABOLISM KEGG_RETINOL_METABOLISM ADH1A ALDH3B2 GOT1 METTL2B ADH1A CYP26A1 CYP3A5 RDH11 UGT1A7 ADH1B AOC2 GOT2 METTL6 ADH1B CYP26B1 CYP3A7 RDH12 UGT1A8 ADH1C AOC3 GSTZ1 MIF ADH1C CYP26C1 CYP4A11 RDH16 UGT1A9 ADH4 AOX1 HEMK1 NAA80 ADH4 CYP2A13 CYP4A22 RDH5 UGT2A1 ADH5 BUD23 HGD PNMT ADH5 CYP2A6 DGAT1 RDH8 UGT2A3 63 ADH6 COMT HPD TAT ADH6 CYP2A7 DGAT2 RETSAT UGT2B10

ADH7 DBH IL4I1 TH ADH7 CYP2B6 DHRS3 RPE65 UGT2B11 ALDH1A3 DCT LCMT1 TPO ALDH1A1 CYP2C18 DHRS4 UGT1A1 UGT2B15 ALDH3A1 DDC LCMT2 TRMT11 ALDH1A2 CYP2C19 DHRS4L2 UGT1A10 UGT2B17 ALDH3B1 FAH MAOA TYR AWAT2 CYP2C8 DHRS9 UGT1A3 UGT2B28 MAOB TYRP1 BCO1 CYP2C9 LRAT UGT1A4 UGT2B4 CYP1A1 CYP3A4 PNPLA4 UGT1A5 UGT2B7 CYP1A2 CYP3A43 RDH10 UGT1A6 KEGG_DRUG_METABOLISM_OTHER_ENZYMES KEGG_DRUG_METABOLISM_CYTOCHROME_P450 CDA GMPS UCK2 UGT2B10 CYP3A5 GSTK1 MAOA UGT2B15 MGST2 CES1 GUSB UCKL1 UGT2B11 ADH1A GSTM1 MAOB UGT2B17 MGST3 CES2 HPRT1 UGT1A1 UGT2B15 ADH1B CYP2A13 MGST1 UGT2B28 UGT1A1 CES5A IMPDH1 UGT1A10 UGT2B17 ADH1C CYP2A6 CYP3A7 UGT2B4 UGT1A10 CYP2A13 IMPDH2 UGT1A3 UGT2B28 ADH4 CYP2A7 FMO1 UGT2B7 UGT1A3 CYP2A6 ITPA UGT1A4 UGT2B4 ADH5 CYP2B6 FMO2 GSTM2 UGT1A4 CYP2A7 NAT1 UGT1A5 UGT2B7 ADH6 CYP2C18 FMO3 GSTM3 UGT1A5 CYP3A4 NAT2 UGT1A6 UMPS ADH7 CYP2C19 FMO4 GSTM4 UGT1A6 CYP3A43 TK1 UGT1A7 UPB1 ALDH1A3 CYP2C8 FMO5 GSTM5 UGT1A7 CYP3A5 TK2 UGT1A8 UPP1 ALDH3A1 CYP2C9 GSTA1 GSTO1 UGT1A8 CYP3A7 TPMT UGT1A9 UPP2 ALDH3B1 CYP2D6 GSTA2 GSTO2 UGT1A9 DPYD TYMP UGT2A1 XDH ALDH3B2 CYP2E1 GSTA3 GSTP1 UGT2A1 DPYS UCK1 UGT2A3 AOX1 CYP3A4 GSTA4 GSTT1 UGT2A3 CYP1A2 CYP3A43 GSTA5 GSTT2 UGT2B10 GSTZ1 UGT2B11

Figure 3.7. Average Daily Food Intake

Average Daily Food Intake 3.4

3.2

s 3.0

a r

G 2.8

2.6

2.4 s s s s le le le le a a a a M m M m e e F F Control Rapamycin 000000000 Figure 3.7: Average daily food intake for Control and000 Rapamycin mice. Weekly food weights were obtained throughout the 14-week intervention. Data is presented in grams.

Males: Body Weight Females: Body Weight A B 45 35

40 )

) 30

g (

(

t 35 h

h 25

g i

i 30

e W W 20 25

20 15 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 Weeks on Diet Weeks on Diet

C Males D Females Fasting Blood Glucose Fasting Blood Glucose

✱✱✱✱

✱✱✱✱ e

200 ✱✱✱ e 200 s

s ✱✱✱✱ o

u l

150 l 150

)

) G

/ o

100 o 100

(

( g

50 g 50

t s

s a 0 a

F 0 F l l o in R o in R tr c C tr c C n y n y o m o m C a C a p p a a R R

Figure 3.8: Body weight and fasting blood glucose data. Body weight was recorded weekly throughout the 14 weeks for (A) males and (B) females. Fasting blood glucose (6 hours) was obtained at time of euthanization for (C) males and (D) females. Weight is presented in grams (g) and fasting blood glucose is in mg/dL. Sample size: control (n=15 males) (n=14 females); Rapamycin (n=15 males) (n=14 females); CR (n=15 males) (n=15 females), ***p<0.005, ****p<0.0001. Values are means SD. Statistical tests: (C) and (D) one-way ANOVA and Tukey’s multiple comparison post hoc test.

Figure 3.9

(A) Muscle mass: quadriceps, gastrocnemius, and Tibialis Anterior

Quadriceps mass Gastrocnemius mass Tibialis Anterior mass ✱✱✱ ✱✱✱✱ ✱✱✱✱ ✱ 0.6 0.6 ✱✱✱✱ 0.25

) 0.20

)

(

)

(

0.4

0.4 s s

s 0.15 a

o 0.10

Males

r A

a t 0.2

0.2 T

s a Q 0.05 G

0.0 0.0 0.00 l l n l o in R o i R o in R r c C tr c C tr c C t y n y n y n o m o m o m C a C a C a p p p a a a R R R

(B) Relative muscle mass: quadriceps, gastrocnemius, and Tibialis Anterior

Quadriceps Gastrocnemius Tibialis Anterior (relative mass) (relative mass) (relative mass) ✱✱✱ ✱ ✱✱ ✱ ✱✱

2.0 ) 1.6 1.0

)

g /

g ✱✱

) /

(

g 0.8 (

s 1.5 s

1.4

s a

M 0.6

y d

1.0 1.2

Males

d o

o d

B 0.4

0.5 / d 1.0

a 0.2

Q a

0.0 G 0.8 0.0 l o in R l n l r c C o i R o in R t y tr c C tr c C n n y n y o m o m o m C a C a C a p p p a a a R R R

Quadriceps mass Gastrocnemius mass Tibialis Anterior mass

✱✱✱✱ ✱✱✱✱ ✱✱

0.5 ✱✱✱✱ 0.5 ✱✱✱✱ 0.20 ✱

) ) 0.4 g 0.4

(

ales (

) 0.15

(

s 0.3 0.3 s

a m

Fem 0.10

o d

0.2 0.2

a t

T a Q 0.05

0.1 G 0.1

0.0 0.0 0.00 l l l o in R o in R o in R tr c C tr c C tr c C n y n y n y o m o m o m C a C a C a p p p a a a R R R

(D) Relative muscle mass: quadriceps, gastrocnemius, and tibialis anterior

Quadriceps Gastrocnemius Tibialis Anterior (relative mass) (relative mass) (relative mass)

2.0 ) )

2.0

g 0.8

)

(

1.8 /

ales s

s 1.5

( s

0.6

s M

1.6 s

Fem

y d

1.0 M

o 0.4 o

1.4 y

B d

0.5 o

1.2 / 0.2

Q a

1.0 G 0.0 0.0 l o in R l r c C o in R l t y tr c C o in R n n y tr c C o m o m n y C a C a o m p p C a a a p R R a R Figure 3.9. Absolute and relative muscle mass for male and female mice. (A) Absolute muscle mass for males (n=15) per diet group. (B) Relative muscle mass for males. (C) Absolute muscle mass for females (n=14) for control and rapamycin diet groups and (n=15) for the CR diet group. Relative muscle mass is derived by dividing muscle mass by whole body mass, *p<0.05, **p<0.01, ***p<0.005, ****p<0.0001. Values are means SD. Statistical tests: one-way ANOVA and Tukey’s multiple comparison post hoc test.

(A) Grip Strength Test

Males Females Grip strength Grip strength

1.5 ✱✱✱ 1.5

s 1.0 n

s 1.0

o n

w e N 0.5 N 0.5

0.0 0.0 e l n n o i R li tr c C e l n R e n y n ro i s o li t c C a m e n y C a s o m B p a C a a B p R a R

Normalized Grip Strength Normalized Grip Strength

✱✱ ✱ ✱ 0.05 0.05

s 0.04 0.04 m

r g

g 0.03 0.03

s n

o t

t 0.02 0.02

w e

e N N 0.01 0.01

0.00 0.00

e l n e l n n o i R n o i R li tr c C li tr c C e n y e n y s o m s o m a C a a C a B p B p a a R R

(B) Balance beam Test

Males Females Balance Beam 12 mm Balance Beam 12 mm ✱✱✱ ✱✱✱ ✱

10 8 ✱✱✱✱

)

8 ) s

s 6 d

n o

6 o

c e

e 4

(

e m

m i

i 2 T 2 T

0 0

e l n e l n n o i R n o i R li tr c C li tr c C e n y e n y s o m s o m a C a a C a B p B p a a R R

Balance Beam 6 mm Balance Beam 6 mm

✱✱✱✱ ✱✱

✱ ✱ 15 10

✱✱ )

) 8

s s

d d n

n 10 o

o 6

c c

e e

s s

( (

4 e e

m m

i i T T 2

0 0 e l n e l n n o i R n o i R li tr c C li tr c C e n y e n y s o m s o m a C a a C a B p B p a a R R

Males Females Max Hang Time Max Hang Time

✱✱✱ ✱✱✱✱ 25 50 ✱✱✱

✱✱✱ )

) 20 40

s s

t u u 30

15 n

n i

(

10 e

e m

i T T 5 10

0 0

e l n R e l n R n ro i n o i li t c C li tr c C e n y e n y s o m s o m a C a a C a B p B p a a R R

Holding Impulse Holding Impulse ✱ 300 ✱✱

500

s s

d 400 d

n n

o 200

e e

s 300

n n

o 200

t o

100 t

e e

N 100 N 0 0 l e n R e l n n ro i n o i R li t c C li tr c C e n y e n y s o m s o m a C a a C a B p B p a a R R

Figure 3.10. Functional tests: Measures of muscle strength, function, balance, and coordination were assessed. (A) Schematic of grip strength: Mean grip strength (derived from the three replicated measurements) and relative grip strength [(derived from dividing unit of force (newtons) by body mass (g)]. (B) Schematic of balance beam: Latency to cross the balance beam presented as the mean of three trials. (C) Schematic of inverted screen test: Data is presented as maximum hang time and holding impulse (body mass multiplied by maximum hang time and converted to newtons), *p<0.05, **p<0.01, ***p<0.005, ****p<0.0001. Values are means SEM. Statistical tests: two-way ANOVA and Tukey’s multiple comparison post hoc test. In the graph for holding impulse for females, the highest score was truncated to reduce the influence of an extreme outlier.

Figure 3.11

(A) Females PCA plot (B) Females heat map

Figure 3.11. Gene expression analysis for female mice: (A) Principle component analysis (PCA) plot and (B) Heat Map. For PCA plots unit variance scaling is applied to rows; singular value decomposition (SVD) with imputation is used to calculate principal components. X and Y axis show principal component 1 and principal component 2 that explain 39.7% and 11.8% of the total variance, respectively. Prediction ellipses are such that with probability 0.95, a new observation from the same group will fall inside the ellipse. PCA and heat map were generated using ClustVis.

Figure 3.12

(A) Males PCA plot (B) Males heat map

Figure 3.12. Gene expression analysis for male mice: (A) Principle component analysis (PCA) plot and (B) Heat Map. For PCA plots unit variance scaling is applied to rows; singular value decomposition (SVD) with imputation is used to calculate principal components. X and Y axis show principal component 1 and principal component 2 that explain 39.7% and 11.8% of the total variance, respectively. Prediction ellipses are such that with probability 0.95, a new observation from the same group will fall inside the ellipse. PCA and heat map were generated using ClustVis.

Figure 3.13

Figure 3.13. Differentially expressed genes in females versus males. Analysis conducted using Transcriptomic analysis console software (TAC) and Graph generated with Prism software.

Table 3.4. Gene Set Enrichment (Hallmark) for male mice. (FDR<0.05)

Gene Set Enrichment in Quadriceps muscle CR vs Control Rapamycin vs Control Rapamycin vs CR (Hallmark) NES FDR q-value NES FDR q-value NES FDR q-value HALLMARK_COAGULATION -1.79 0.00 -1.92 0.00 -1.14 0.36 HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION -1.65 0.00 -0.99 0.66 1.52 0.12 HALLMARK_XENOBIOTIC_METABOLISM -1.41 0.13 -1.83 0.00 -1.33 0.11 HALLMARK_ADIPOGENESIS -0.62 1.00 -1.77 0.00 -1.64 0.02 HALLMARK_CHOLESTEROL_HOMEOSTASIS 1.01 0.90 -1.76 0.00 -1.89 0.00 HALLMARK_FATTY_ACID_METABOLISM -0.63 1.00 -1.71 0.00 -1.50 0.05 HALLMARK_MTORC1_SIGNALING 0.84 1.00 -1.64 0.01 -1.72 0.01

Table 3.5. Gene Set Enrichment (Hallmark) for female mice. (FDR<0.05)

Gene Set Enrichment in Quadriceps Muscle CR vs Control Rapamycin vs Control Rapamycin vs CR (Hallmark) NES FDR q-value NES FDR q-value NES FDR q-value HALLMARK_COAGULATION -0.93 0.93 2.33 0.00 2.47 0.00 HALLMARK_XENOBIOTIC_METABOLISM 1.29 0.86 2.32 0.00 2.07 0.00 HALLMARK_BILE_ACID_METABOLISM 1.23 0.67 2.24 0.00 1.89 0.00 HALLMARK_PEROXISOME 0.89 0.97 1.53 0.03 1.22 0.22 HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION -1.81 0.01 -0.98 1.00 1.64 0.01 HALLMARK_COMPLEMENT -1.23 0.25 1.41 0.07 1.58 0.01 HALLMARK_INTERFERON_ALPHA_RESPONSE -1.28 0.22 0.82 1.00 1.49 0.03 HALLMARK_INTERFERON_GAMMA_RESPONSE -1.58 0.03 -0.91 1.00 1.40 0.07 HALLMARK_ALLOGRAFT_REJECTION -1.57 0.02 -1.00 1.00 1.27 0.16 HALLMARK_CHOLESTEROL_HOMEOSTASIS -1.54 0.02 -1.32 0.49 0.97 0.77

Gene Set Enrichment in Quadriceps Muscle CR vs Control Rapamycin vs Control Rapamycin vs CR (Hallmark) NES FDR q-value NES FDR q-value NES FDR q-value HALLMARK_MTORC1_SIGNALING -1.20 0.25 -1.19 0.83 -0.94 0.94

Table 3.6. Gene Set Enrichment (Hallmark) for male and female mice in rapamycin group. (FDR<0.05)

Gene Set Enrichment in Quadriceps Muscle Females vs Males (Hallmark) NES FDR q-value HALLMARK_COAGULATION 2.444 0.000 HALLMARK_XENOBIOTIC_METABOLISM 2.314 0.000 HALLMARK_BILE_ACID_METABOLISM 1.843 0.001 HALLMARK_FATTY_ACID_METABOLISM 1.728 0.003 HALLMARK_CHOLESTEROL_HOMEOSTASIS 1.630 0.011 ALLMARK_COMPLEMENT 1.577 0.016 HALLMARK_KRAS_SIGNALING_UP 1.533 0.024 HALLMARK_MTORC1_SIGNALING 1.498 0.032 HALLMARK_ADIPOGENESIS 1.475 0.037

Supplemental Table 3.1. Gene set enrichment analysis for Mouse samples: Reactome at (A) 12 and (B) 24 months and KEGG at (C) 12 and (D) 24 months

A). Reactome 12 months

Gene Set Enrichment in Quadriceps Muscle (12 months) CR vs. control (Reactome) NES FDR q-val REACTOME_CHOLESTEROL_BIOSYNTHESIS 2.456 0.000 REACTOME_CHYLOMICRON_MEDIATED_LIPID_TRANSPORT 2.235 0.002 REACTOME_BILE_ACID_AND_BILE_SALT_METABOLISM 2.213 0.002 REACTOME_RORA_ACTIVATES_CIRCADIAN_EXPRESSION 2.201 0.002 REACTOME_LIPID_DIGESTION_MOBILIZATION_AND_TRANSPORT 2.176 0.002 REACTOME_CIRCADIAN_REPRESSION_OF_EXPRESSION_BY_REV_ERBA 2.158 0.003 REACTOME_SYNTHESIS_OF_BILE_ACIDS_AND_BILE_SALTS 2.124 0.003 REACTOME_FORMATION_OF_FIBRIN_CLOT_CLOTTING_CASCADE 2.115 0.003 REACTOME_FATTY_ACYL_COA_BIOSYNTHESIS 2.042 0.006 REACTOME_METABOLISM_OF_AMINO_ACIDS_AND_DERIVATIVES 2.029 0.006 REACTOME_PHASE1_FUNCTIONALIZATION_OF_COMPOUNDS 2.009 0.007 REACTOME_TRIGLYCERIDE_BIOSYNTHESIS 1.975 0.009 REACTOME_BIOLOGICAL_OXIDATIONS 1.957 0.010 REACTOME_NUCLEAR_RECEPTOR_TRANSCRIPTION_PATHWAY 1.948 0.011 REACTOME_LIPOPROTEIN_METABOLISM 1.929 0.012 REACTOME_REGULATION_OF_GENE_EXPRESSION_IN_BETA_CELLS 1.913 0.013 REACTOME_INTERACTION_BETWEEN_L1_AND_ANKYRINS 1.892 0.017 REACTOME_BMAL1_CLOCK_NPAS2_ACTIVATES_CIRCADIAN_EXPRESSION 1.852 0.024 REACTOME_METABOLISM_OF_STEROID_HORMONES_AND_VITAMINS_A_AND_D 1.824 0.029 REACTOME_NUCLEAR_SIGNALING_BY_ERBB4 1.818 0.029 REACTOME_CYTOCHROME_P450_ARRANGED_BY_SUBSTRATE_TYPE 1.816 0.028 REACTOME_CIRCADIAN_CLOCK 1.809 0.028 REACTOME_METABOLISM_OF_LIPIDS_AND_LIPOPROTEINS 1.806 0.028 REACTOME_GLUCONEOGENESIS 1.786 0.033 REACTOME_INTRINSIC_PATHWAY 1.782 0.032 REACTOME_PI3K_CASCADE 1.741 0.043

Gene Set Enrichment in Quadriceps Muscle (24 months) CR vs. control B) Reactome (Reactome) NES FDR q-val REACTOME_PEROXISOMAL_LIPID_METABOLISM 2.416 0.000 24 months REACTOME_METABOLISM_OF_VITAMINS_AND_COFACTORS 2.331 0.000 REACTOME_METABOLISM_OF_LIPIDS_AND_LIPOPROTEINS 2.279 0.001 REACTOME_TRIGLYCERIDE_BIOSYNTHESIS 2.269 0.000 REACTOME_CHOLESTEROL_BIOSYNTHESIS 2.233 0.000 REACTOME_GLYCEROPHOSPHOLIPID_BIOSYNTHESIS 2.229 0.001 REACTOME_BIOSYNTHESIS_OF_THE_N_GLYCAN_PRECURSOR_DOLICHOL_LIPID_LINKED_OLIGOSACCHARIDE_LLO_AND_TRANSFER_TO_A_NASCENT_PROTEIN2.224 0.000 REACTOME_PEPTIDE_CHAIN_ELONGATION 2.172 0.001 REACTOME_TCA_CYCLE_AND_RESPIRATORY_ELECTRON_TRANSPORT 2.144 0.002 REACTOME_E2F_MEDIATED_REGULATION_OF_DNA_REPLICATION 2.142 0.001 REACTOME_SYNTHESIS_OF_DNA 2.119 0.002 REACTOME_FATTY_ACYL_COA_BIOSYNTHESIS 2.104 0.002 REACTOME_PHOSPHOLIPID_METABOLISM 2.082 0.003 REACTOME_GLUCONEOGENESIS 2.067 0.003 REACTOME_G1_S_TRANSITION 2.063 0.003 REACTOME_M_G1_TRANSITION 2.058 0.003 REACTOME_CDT1_ASSOCIATION_WITH_THE_CDC6_ORC_ORIGIN_COMPLEX 2.052 0.003 REACTOME_GLUCOSE_METABOLISM 2.052 0.003 REACTOME_RESPIRATORY_ELECTRON_TRANSPORT 2.037 0.003 REACTOME_DNA_STRAND_ELONGATION 2.030 0.004 REACTOME_FATTY_ACID_TRIACYLGLYCEROL_AND_KETONE_BODY_METABOLISM 2.030 0.003 REACTOME_MITOTIC_G1_G1_S_PHASES 1.995 0.006 REACTOME_ER_PHAGOSOME_PATHWAY 1.976 0.007 REACTOME_RESPIRATORY_ELECTRON_TRANSPORT_ATP_SYNTHESIS_BY_CHEMIOSMOTIC_COUPLING_AND_HEAT_PRODUCTION_BY_UNCOUPLING_PROTEINS_1.969 0.007 REACTOME_S_PHASE 1.965 0.007 REACTOME_BRANCHED_CHAIN_AMINO_ACID_CATABOLISM 1.955 0.007 REACTOME_CDK_MEDIATED_PHOSPHORYLATION_AND_REMOVAL_OF_CDC6 1.952 0.007 REACTOME_DNA_REPLICATION 1.951 0.007 REACTOME_GLYCOLYSIS 1.935 0.008 REACTOME_MITOTIC_M_M_G1_PHASES 1.934 0.008 REACTOME_INTERFERON_ALPHA_BETA_SIGNALING 1.926 0.009 REACTOME_PYRUVATE_METABOLISM_AND_CITRIC_ACID_TCA_CYCLE 1.925 0.009 REACTOME_TELOMERE_MAINTENANCE 1.921 0.009 REACTOME_3_UTR_MEDIATED_TRANSLATIONAL_REGULATION 1.919 0.009 REACTOME_ACTIVATION_OF_THE_PRE_REPLICATIVE_COMPLEX 1.914 0.009 REACTOME_ASSEMBLY_OF_THE_PRE_REPLICATIVE_COMPLEX 1.909 0.009 REACTOME_REGULATION_OF_ORNITHINE_DECARBOXYLASE_ODC 1.902 0.009 REACTOME_MITOCHONDRIAL_PROTEIN_IMPORT 1.877 0.011 REACTOME_SCFSKP2_MEDIATED_DEGRADATION_OF_P27_P21 1.875 0.011 REACTOME_TRANSFERRIN_ENDOCYTOSIS_AND_RECYCLING 1.864 0.013 REACTOME_ORC1_REMOVAL_FROM_CHROMATIN 1.848 0.014 REACTOME_REGULATION_OF_APOPTOSIS 1.846 0.014 REACTOME_SRP_DEPENDENT_COTRANSLATIONAL_PROTEIN_TARGETING_TO_MEMBRANE 1.821 0.018 REACTOME_EXTENSION_OF_TELOMERES 1.813 0.019 REACTOME_LATENT_INFECTION_OF_HOMO_SAPIENS_WITH_MYCOBACTERIUM_TUBERCULOSIS 1.804 0.020 REACTOME_CITRIC_ACID_CYCLE_TCA_CYCLE 1.797 0.021 REACTOME_CELL_CYCLE_MITOTIC 1.792 0.022 REACTOME_INSULIN_RECEPTOR_RECYCLING 1.790 0.022 REACTOME_SCF_BETA_TRCP_MEDIATED_DEGRADATION_OF_EMI1 1.778 0.024 REACTOME_APC_C_CDC20_MEDIATED_DEGRADATION_OF_MITOTIC_PROTEINS 1.775 0.024 REACTOME_PROCESSIVE_SYNTHESIS_ON_THE_LAGGING_STRAND 1.769 0.025 REACTOME_P53_INDEPENDENT_G1_S_DNA_DAMAGE_CHECKPOINT 1.768 0.025 REACTOME_CELL_CYCLE_CHECKPOINTS 1.749 0.028 REACTOME_G1_S_SPECIFIC_TRANSCRIPTION 1.748 0.028 REACTOME_SYNTHESIS_OF_PIPS_AT_THE_GOLGI_MEMBRANE 1.748 0.028 REACTOME_AUTODEGRADATION_OF_CDH1_BY_CDH1_APC_C 1.736 0.030 REACTOME_REGULATION_OF_MITOTIC_CELL_CYCLE 1.726 0.032 REACTOME_ACTIVATION_OF_NF_KAPPAB_IN_B_CELLS 1.722 0.032 REACTOME_ANTIGEN_PROCESSING_CROSS_PRESENTATION 1.718 0.033 REACTOME_ACTIVATION_OF_THE_MRNA_UPON_BINDING_OF_THE_CAP_BINDING_COMPLEX_AND_EIFS_AND_SUBSEQUENT_BINDING_TO_43S1.716 0.033 REACTOME_AUTODEGRADATION_OF_THE_E3_UBIQUITIN_LIGASE_COP1 1.711 0.034 REACTOME_SPHINGOLIPID_METABOLISM 1.704 0.035 REACTOME_TRANSLATION 1.701 0.036 REACTOME_VIF_MEDIATED_DEGRADATION_OF_APOBEC3G 1.698 0.036 REACTOME_LAGGING_STRAND_SYNTHESIS 1.685 0.039 REACTOME_GLYCOSPHINGOLIPID_METABOLISM 1.680 0.040 REACTOME_DESTABILIZATION_OF_MRNA_BY_AUF1_HNRNP_D0 1.679 0.040 REACTOME_TRAF6_MEDIATED_NFKB_ACTIVATION 1.676 0.040 REACTOME_CROSS_PRESENTATION_OF_SOLUBLE_EXOGENOUS_ANTIGENS_ENDOSOMES 1.675 0.039 REACTOME_CELL_CYCLE 1.674 0.039 REACTOME_IRON_UPTAKE_AND_TRANSPORT 1.654 0.040 REACTOME_MEIOTIC_RECOMBINATION 1.653 0.041 REACTOME_MITOCHONDRIAL_TRNA_AMINOACYLATION 1.648 0.042

C). KEGG 12 months

Gene Set Enrichment in Quadriceps Muscle CR vs. control (KEGG) NES FDR q-val KEGG_NITROGEN_METABOLISM 2.386 0.000 KEGG_DRUG_METABOLISM_CYTOCHROME_P450 2.333 0.000 KEGG_GLYCINE_SERINE_AND_THREONINE_METABOLISM 2.325 0.000 KEGG_ALANINE_ASPARTATE_AND_GLUTAMATE_METABOLISM 2.129 0.001 KEGG_STEROID_BIOSYNTHESIS 2.108 0.001 KEGG_TRYPTOPHAN_METABOLISM 2.050 0.004

KEGG_RETINOL_METABOLISM 1.978 0.008 KEGG_PRIMARY_BILE_ACID_BIOSYNTHESIS 1.977 0.007

KEGG_METABOLISM_OF_XENOBIOTICS_BY_CYTOCHROME_P450 1.907 0.009 KEGG_COMPLEMENT_AND_COAGULATION_CASCADES 1.847 0.016 KEGG_PENTOSE_PHOSPHATE_PATHWAY 1.822 0.018 KEGG_ARGININE_AND_PROLINE_METABOLISM 1.816 0.018 KEGG_HISTIDINE_METABOLISM 1.815 0.017 KEGG_WNT_SIGNALING_PATHWAY 1.793 0.018 KEGG_INSULIN_SIGNALING_PATHWAY 1.789 0.018 KEGG_STEROID_HORMONE_BIOSYNTHESIS 1.745 0.025 KEGG_TYROSINE_METABOLISM 1.719 0.029

D). KEGG 24 months

Gene Set Enrichment in Quadriceps Muscle CR vs. control (KEGG) NES FDR q-val KEGG_PEROXISOME 2.757 0.000 KEGG_VALINE_LEUCINE_AND_ISOLEUCINE_DEGRADATION 2.416 0.000 KEGG_PENTOSE_PHOSPHATE_PATHWAY 2.413 0.000 KEGG_CITRATE_CYCLE_TCA_CYCLE 2.364 0.000 KEGG_PYRUVATE_METABOLISM 2.348 0.000 KEGG_PROPANOATE_METABOLISM 2.279 0.000

KEGG_PPAR_SIGNALING_PATHWAY 2.259 0.000 KEGG_FATTY_ACID_METABOLISM 2.202 0.000

KEGG_GLYCOLYSIS_GLUCONEOGENESIS 2.186 0.000 KEGG_LYSOSOME 2.125 0.000 KEGG_STEROID_BIOSYNTHESIS 2.113 0.001 KEGG_BIOSYNTHESIS_OF_UNSATURATED_FATTY_ACIDS 1.992 0.002 KEGG_AMINO_SUGAR_AND_NUCLEOTIDE_SUGAR_METABOLISM 1.958 0.002 KEGG_FRUCTOSE_AND_MANNOSE_METABOLISM 1.957 0.002 KEGG_BUTANOATE_METABOLISM 1.940 0.003 KEGG_HISTIDINE_METABOLISM 1.935 0.003 KEGG_CYTOSOLIC_DNA_SENSING_PATHWAY 1.932 0.003 KEGG_RIBOSOME 1.915 0.003 KEGG_DRUG_METABOLISM_OTHER_ENZYMES 1.873 0.006 KEGG_DRUG_METABOLISM_CYTOCHROME_P450 1.858 0.007 KEGG_PROTEASOME 1.840 0.008 KEGG_TERPENOID_BACKBONE_BIOSYNTHESIS 1.765 0.015 KEGG_PYRIMIDINE_METABOLISM 1.765 0.014 KEGG_INSULIN_SIGNALING_PATHWAY 1.754 0.015 KEGG_TRYPTOPHAN_METABOLISM 1.726 0.017 KEGG_PHENYLALANINE_METABOLISM 1.696 0.021 KEGG_ADIPOCYTOKINE_SIGNALING_PATHWAY 1.671 0.025

KEGG_GALACTOSE_METABOLISM 1.648 0.029 KEGG_OXIDATIVE_PHOSPHORYLATION 1.642 0.030

KEGG_BETA_ALANINE_METABOLISM 1.637 0.030 KEGG_ACUTE_MYELOID_LEUKEMIA 1.609 0.037 KEGG_ANTIGEN_PROCESSING_AND_PRESENTATION 1.608 0.036 KEGG_TYROSINE_METABOLISM 1.600 0.036 KEGG_DNA_REPLICATION 1.567 0.043

Supplemental Table 3.2. Gene set enrichment analysis for Human samples: Reactome at (A) 12 and (B) 24 months and KEGG at (C) 12 and (D) 24 months

A). Reactome 12 months

Gene Set Enrichment in Quadriceps Muscle CR vs. control (Reactome) NES FDR q-val REACTOME_OLFACTORY_SIGNALING_PATHWAY 1.966 0.024 REACTOME_RESPIRATORY_ELECTRON_TRANSPORT_UNCOUPLING_PROTEINS 1.930 0.025 REACTOME_TERMINATION_OF_O_GLYCAN_BIOSYNTHESIS 1.853 0.057 REACTOME_RESPIRATORY_ELECTRON_TRANSPORT 1.833 0.055 REACTOME_AMYLOIDS 1.733 0.161

B.) Reactome 24 months

Gene Set Enrichment in Quadriceps Muscle CR vs. control (Reactome) NES FDR q-val REACTOME_OLFACTORY_SIGNALING_PATHWAY 2.319 0.000 REACTOME_GAP_JUNCTION_ASSEMBLY 1.589 0.158 REACTOME_CLASS_A1_RHODOPSIN_LIKE_RECEPTORS 1.579 0.162 REACTOME_BILE_ACID_AND_BILE_SALT_METABOLISM 1.594 0.167 REACTOME_NA_CL_DEPENDENT_NEUROTRANSMITTER_TRANSPORTERS 1.602 0.170 REACTOME_PEPTIDE_LIGAND_BINDING_RECEPTORS 1.615 0.188 REACTOME_INCRETIN_SYNTHESIS_SECRETION_AND_INACTIVATION 1.602 0.190 REACTOME_TRANSPORT_OF_GLUCOSE_AND_OTHER_SUGARS_BILE_SALTS_AND_ORGANIC_ACIDS_METAL_IONS_AND_AMINE_COMPOUNDS1.553 0.198 REACTOME_LIGAND_GATED_ION_CHANNEL_TRANSPORT 1.658 0.201 REACTOME_SYNTHESIS_OF_BILE_ACIDS_AND_BILE_SALTS 1.543 0.203 REACTOME_SYNTHESIS_SECRETION_AND_INACTIVATION_OF_GLP1 1.638 0.203 REACTOME_AMINE_COMPOUND_SLC_TRANSPORTERS 1.619 0.209 REACTOME_DEFENSINS 1.520 0.238 REACTOME_AMINE_DERIVED_HORMONES 1.500 0.241 REACTOME_PHASE1_FUNCTIONALIZATION_OF_COMPOUNDS 1.504 0.244

C). KEGG 12 months

Gene Set Enrichment in Quadriceps Muscle CR vs. control (KEGG) NES FDR q-val KEGG_HISTIDINE_METABOLISM 2.750 0.059 KEGG_PHENYLALANINE_METABOLISM 1.950 0.230

D). KEGG 24 months

Gene Set Enrichment in Quadriceps Muscle CR vs. control (KEGG) NES FDR q-val KEGG_OLFACTORY_TRANSDUCTION 2.273 0.000 KEGG_PENTOSE_AND_GLUCURONATE_INTERCONVERSIONS 1.683 0.060 KEGG_TASTE_TRANSDUCTION 1.686 0.087 KEGG_METABOLISM_OF_XENOBIOTICS_BY_CYTOCHROME_P450 1.591 0.138 KEGG_TYROSINE_METABOLISM 1.494 0.198 KEGG_RETINOL_METABOLISM 1.537 0.201 KEGG_DRUG_METABOLISM_OTHER_ENZYMES 1.505 0.203 KEGG_NEUROACTIVE_LIGAND_RECEPTOR_INTERACTION 1.516 0.212 KEGG_DRUG_METABOLISM_CYTOCHROME_P450 1.469 0.222

CHAPTER IV: MECHANISTIC ROLE OF REDUCED CIRCULATING LEVELS OF FREE IGF-1 INDUCED BY CALORIE RESTRICTION ON AGE-RELATED MUSCLE LOSS

Introduction

In recent decades, the study of aging has expanded exponentially. In large part, these research efforts have been motivated by emerging trends in global aging. Today, people are living longer129. Although this represents a triumph in the last century of medical advances, economic development, and more effective public health interventions and polices, it is not without a new set of issues135. Life expectancy has more than doubled during the 20th century and the number of people 100 years and over is projected to increase by more than five times by

203016. Increased life expectancy is met with an overall increase in population aging. Latest estimates predict that by 2030, one billion people (1 in 8 individuals) will be 65 years and older162. These trends present a host of challenges at both an individual and societal level and will likely place a great burden on national health care resources135. Meeting these challenges with strategies that combine an increased life expectancy with healthy aging may be a viable window of opportunity to address these issues of population aging.

Aging is a complex process, defined as a time-dependent progressive decline in physical health and wellness, and ultimately leads to a loss of physical function and an increase in age- specific mortality16. One of the greatest factors contributing to the deleterious effects of aging is the decline of functional ability due to loss of muscle mass, strength, and function, a condition termed sarcopenia134. Skeletal muscle, which comprises ~50% of total body mass in young individuals, is a fundamental organ in maintaining physical health and function in older age163.

Peak muscle mass is reached by~30-40 years of age, after which there is a progressive loss of mass and deterioration of muscle quality (determinant of muscle strength and function)134,163. On average, an individual can expect to lose ~30% of muscle mass over a lifetime, and by 80-90 years of age, as much as 50% of muscle mass, strength, and function may be lost163. At a physiological level, skeletal muscle metabolism is guided by a dynamic process characterized by a balance between synthesis and breakdown of muscle proteins164,165. Maintaining equilibrium in this balance is fundamental to the structure and function of skeletal muscle, while a disruption in the balance is a driving etiological factor of sarcopenia134. Evidence suggest, insulin-like growth factor 1 (IGF-1) is, in part, responsible for supporting a careful balance between the forces that regulate anabolic and catabolic processes in skeletal muscle20,166. IGF-1 stimulates protein synthesis while inhibiting protein degradation; promotes differentiation of satellite cells

(or muscle stem cells); and plays an important role in the maintenance of muscle mass and function with aging167,168. Understanding mechanisms of sarcopenia and the potential therapeutic role of IGF-1 signaling in aging muscle may lead to the development of treatments and interventions to attenuate and/or prevent sarcopenia.

Intervention strategy for aging muscle

Calorie restriction (CR), or the reduction of calorie intake without malnutrition, is an intervention that has consistently been shown to extend lifespan and delay the onset and progression of various age-related diseases, including sarcopenia169. In skeletal muscle, CR reduces myopathies, oxidative stress, and mitochondrial dysfunction169. Perhaps most intriguing,

CR attenuates age-related muscle loss (a seemingly anabolic process), despite reducing systemic levels of IGF-1, at least in rodents170. For example, CR studies in rodents conducted by our lab

(unpublished) and others149 have shown an increase in IGF/PI3K/AKT/mTOR gene expression in skeletal muscle, despite a significant decrease in systemic IGF-1 (>40%).

One explanation for this paradoxical phenomenon may involve the role of muscle- derived IGF-1 in muscle development and regenerative processes. IGF-1 is produced primarily by the liver in response to growth hormone. However, in addition to the liver, many other tissues including skeletal muscle, produce a local source of IGF-120. Muscle-derived IGF-1 is controlled by a paracrine/autocrine system that appears to be regulated, in part, by systemic IGF-

120. A considerable body of research has shown skeletal muscle-derived IGF-1 has an essential role in local and global muscle growth, maintenance, and repair. For instance, mice harboring a muscle-specific IGF-1 knock out allele had significantly smaller muscles and impaired whole body growth compared with control mice171–173. Conversely, near ablation of circulating IGF-1 in mice with a liver-specific IGF-1 knock out allele had only marginal effects on muscle protein synthesis or postnatal growth29,174,175. These data strongly suggest that IGF-1, produced specifically in skeletal muscle, has a quintessential role in supporting and maintaining muscle mass29,175; however, the influence or regulatory role of systemic IGF-1 on normal muscle growth and development is still a developing field. Emerging evidence suggest there is a reciprocal interplay between systemic and locally expressed IGF-1 and this relationship is crucial for appropriate cell and tissue growth and function29. To investigate the regulatory role of systemic

IGF-1, in the context of CR, on skeletal muscle , the primary objective of this murine study was to experimentally manipulate levels of systemic IGF-1 in vivo to test our hypothesis that reduced circulating IGF-1 is causally related to the protective effect of CR on age-related muscle loss.

However, administration of recombinant IGF-1 for this in vivo purpose has technical challenges due to its short systemic half-life (<4 hours in rodent models) and side effects including

hypoglycemia176 Thus, the secondary objective was to test the efficacy of an IGF-1 derivative, a polyethylene glycol (PEG)-coupled variant of IGF1 (PEG-IGF1), on CR-induced effects in skeletal muscle. A major advantage of PEG-IGF-1 is improved in vivo pharmacokinetic properties. PEG-IGF-1 provides sustained steady-state levels of circulating IGF-I with minimal dosing regimens due to its decreased renal clearance. Additionally, although its affinity to the

IGF-1R is only marginally reduced compared with endogenous IGF-1, PEG-IGF-1 substantially reduces hypoglycemic side effects177. The use of PEG-IGF-1 may offer a therapeutic option to age-related effects on skeletal muscle and is an effective way to increase systemic IGF-1 levels in our study.

Materials and Methods

Mice and Diets

Eight-month-old male (n=20) C57BL/6J mice were purchased from the Jackson

Laboratory (Sacramento, CA). One mouse was euthanized during the injection procedure due to an adverse event. The remaining 19 mice were included in the data analysis. Mice were housed in standard laboratory cages with alpha-dry bedding and enrichment and maintained on a 12:12 light-dark cycle. During a seven-day acclimation period, mice were provided with ad libitum access to water and a purified control diet (D12450J). After acclimation, mice were randomized to continue on the purified control diet (n=5), provided ad libitum or to a calorically restricted diet (30%) (CR, #D0302702) (n=14), provided daily between 9-11am. The micronutrient content of the CR diet was matched to the control diet providing 100 % of all vitamins, minerals, fatty acids, and amino acids. Diets were formulated by Research Diets, Inc. Mice remained on diet assignments for 5 weeks

PEGylated IGF-1

PEGylated IGF-1 is an optimized IGF-1 variant achieved by attaching a 40 kDa polyethylene glycol (PEG) chain to the N-terminal α-amino group of lysine residues of IGF-1 to form PEG-IGF-I. PEG-IGF-I has a slower renal clearance which allows for less dosing than

IGF-1 to maintain steady-state serum levels in mice. Additional advantages of PEGylation are an increase in solubility and a decrease in protein immunogenicity. PEGylated IGF-1 was generously donated by Dr. Metzger and the Roche Innovative Center in Basel, Switzerland.

PEGylated IGF-1 Treatment

After 5 weeks on diet, mice in the CR diet group were randomized to receive injections

(100ul) of PEGylated IGF-1 at concentrations of 100 ug/kg (n=5) or 500 ug/kg (n=5), or vehicle

(saline) (n=4). Mice in the control diet group (n=5) received a vehicle (saline) injection. Mice were weighed weekly and euthanized 3 days post injection.

mRNA Extraction

Muscle tissue was prepared as previously described (Chapter II). Briefly, sections of frozen skeletal muscle (quadriceps) (~50 mg) were homogenized using bead-based Qiagen

TissueLyser II in l mL TRIzol™ Reagent (Invitrogen™ ThermoFisher #15596026). Following homogenization, mRNA was extracted using Qiagen RNeasy Mini Kit (Qiagen #74104) and stored at -80°C.

Affymetrix Gene Expression Microarray Analysis

The Functional Genomics Core at The University of North Carolina at Chapel Hill

performed the expression profiling and genotyping using an Affymetrix Clariom™ S microarray, specifically for mouse or human species. The Affymetrix assay plates were read on a Beckman

Coulter’s Biomek® FXP Target Prep Express robot and the GeneTitan Instrument from

Affymetrix.

Affymetrix Gene Expression Analysis

Gene expression analysis was conducted using Transcriptome Analysis Console (TAC)

4.0 software, ClustVis, and Gene Set Enrichment Analysis (GSEA).

Biomarker Assays

Serum Total IGF-1 was measured using Quantikine ® ELISA IGF-1 Immunoassay (Cat #

MG100) and PEGylated IGF-1 was measured using PEGylated Protein ELISA quantitative detection kit (ab133065). ELISAs were analyzed using Bio-Plex® MAGPIXTM Multiplex

Reader.

Statistical Analysis

Gene expression Analysis: This is a pilot study assessing the impact of PEGylated IGF1 on genetic response. Twenty, 8 month old mice, were randomized into 4 groups of 5 mice each and followed for 5 weeks: Group 1 was control; Group 2 injected with vehicle (placebo) and

30% CR; Group 3 was injected with 100ug/kg and 30% CR; and Group 4 was injected with

500ug/kg and 30% CR. The injections occurred 3 days prior to sacrifice. As a first step, gene expression profiling analysis of the Affymetrix microarrays were assessed using (TAC)

Software. The differentially altered genes between study groups [control and CR (100ug/kg,

500ug/kg, and vehicle)] were identified. Application of false discovery rate (FDR) adjustment was employed to control for potential Type-I error rate inherent in the testing of multiple outcomes. Gene Cluster 3.0 analysis was also employed to group genes that were differentially expressed into unique clusters. We then utilized Gene Set Enrichment Analysis (GSEA) software to examine biological pathways enriched in these samples.

Results

As expected, CR mice had a sustained reduction in body weight compared with control mice. Fasting blood glucose assessed at time of euthanization was significantly reduced in all

CR groups. PEG-IGF-1 treatment appeared to have no effect on fasting blood glucose 3 days post injection (Figure 4.1A-B). Analysis of serum PEGylated protein and total IGF-1 levels were conducted using ELISA based assays. The presence of PEGylated proteins was detected in both CR groups (CR_100 and CR_500) that received PEG-IGF-1 at concentrations of 100µg/kg and 500µg/kg, respectively (Figure 4.2A). When endogenous IGF-1 levels were added to

PEGylated protein levels (a method previously validated177), we observed a significant increase in IGF-1 for both CR_100 and CR_500 groups compared with CR_Vehicle (Figure 4.2B).

To explore the molecular adaptations in skeletal muscle of mice after exposure to PEG-

IGF-1, mRNA was extracted from the quadriceps muscles collected 3 days post injection. Using a preliminary approach to a genetic inquiry about the transcriptional impact of PEG-IGF-1 exposure, we first performed qPCR assays on select IGF-1-responsive genes to identify differential gene expression between groups. Our results indicate the gene expression of myoblast determination protein 1 (MyoD) was significantly downregulated in CR-Vehicle compared with CR-100 (Figure 4.3A-C). While there were no significant differences in gene

expression of the remaining targets (including MyoG), a trend was observed for the gene expression of fatty acid synthase that suggest there were differences between CR_Vehicle and

CR_100. With these findings, we proceeded with a genome-wide transcriptomic investigation.

Principle Component Analysis (PCA) of the top 1200 most variable genes (based on the

F-statistic<0.05), showed higher similarity between CR groups (Vehicle, 100, and 500) compared with control (Figure 4.4A). Similarly, the heat map generated using hierarchical clustering of the 1200 genes presented in Figure 4.4B shows two clusters with distinct expression profiles between the samples from mice exposed to CR versus control. However, (as shown by the dendrogram and sample color scale at the top of the figure) additional clusters were formed within the CR groups, with CR-Vehicle and CR_500 having more similarity to one another than CR_100. To determine if biological pathways were effected by diet and/or PEG-

IGF-1 treatment, we conducted gene set enrichment analysis (GSEA). As shown in Table 4.1, there were 11 hallmark gene sets enriched in CR_Vehicle compared with all other groups

(control, CR_100, and CR_500). These gene sets of various pathways are commonly acknowledged to be affected by CR including G2M checkpoint, E2F Targets, MTORC1 signaling, DNA repair, and apoptosis. In conclusion, although there was only a modest difference in gene expression within CR groups, as shown by hierarchical clustering, gene set enrichment analysis detected significant differences in CR-responsive pathways that were diminished by PEG-IGF-1 treatment.

Discussion

The world’s population of older adults is increasing rapidly129. Whether this increase in lifespan is perceived as an opportunity or a burden for individuals or societies depends on the

extend to which older adults are experiencing the negative health effects of advanced age178. The progressive loss of muscle mass combined with a loss of muscle strength and function lead to impaired physical capacity and increased vulnerability to morbidity and mortality3. IGF-1 plays an active role in muscle regeneration and repair by maintaining a homeostatic relationship between muscle synthesis and degradation132. Previous research conducted in preclinical models suggests the effects of IGF-1 on skeletal muscle are orchestrated via elaborate crosstalk between circulating and locally synthesized (muscle derived) IGF-120.

CR is one of the most effective interventions to attenuate age-related loss of muscle mass, strength, and function111. CR reduces circulating levels of IGF-1 in rodent models, but the connection between the anti-aging effects of CR on muscle and reduced circulating levels of

IGF-1 has not been well studied. Our observations in rodents led to our hypothesis that reduced circulating IGF-1 is causally related to the protective effect of CR on age-related muscle loss in mice.

In this study, we investigated the mechanistic role of circulating IGF-1 in CR-induced effects on skeletal muscle by utilizing a PEGylated form of IGF-1. Due to the novel use of PEG-

IGF-1 in the context of CR-induced effects on skeletal muscle, we also evaluated potential dosing concentration curves. Although we expected circulating levels of PEGylated proteins to reflect each of the respective dosing concentrations (100µg/kg versus 500 µg/kg), we found only a marginal difference between doses in CR_100 compared with CR_500. We posit this observation may be due to an ‘anabolic threshold-effect’ that has been previously reported176,177,179. Under this theory, the dosing concentration of 500 µg/kg may have exceeded the saturation levels for IGF-1 and these mice may have compensated with an increased excretion rate176. For circulating total IGF-1 levels, mice exposed to CR had reduced circulating

levels compared with control. However, an important finding was the increase in adjusted total

IGF-1 levels, observed in PEG-IGF-1 treated mice (i.e. CR_100 and CR_500 groups). Adjusted total IGF-1 levels is derived by combining the concentrations of PEGylated proteins with total

IGF-1 levels. This method is employed due to the inability of the total IGF-1 assay to detect the

PEGylated form of IGF-1176,177.

In our preliminary targeted gene expression analysis, we selected a sample of muscle– specific IGF-1 responsive genes in order to assess potential molecular underpinnings induced by

PEG-IGF-1 treatment. From the genes selected, we found MyoD gene expression was significantly downregulated in CR_Vehicle versus CR_100. MyoD is highly expressed during myoblast proliferation and since regeneration and proliferation in skeletal muscle are tightly coordinated, a reduction in MyoD may imply CR is enhancing regenerative events and limiting proliferation. In global gene expression analysis, PCA results indicate the transcriptional response to PEG-IGF-1 treatment was less pronounced among the CR groups than we anticipated. One possibility for the lack of separation between CR groups may be due to the robust transcriptional impact of CR compared with control. It may be that transcriptional differences between CR groups are obscured, in comparison, to the differences between CR and control.

An exciting and potentially impactful result was found with gene set enrichment analysis.

Biological pathways associated with cell cycle regulation, inflammation, metabolism (including mTOR) are signature pathways that have been previously shown to be upregulated in CR mice versus control.

In conclusion, these results indicate that a 5-week exposure to CR in aged mice resulted in significant molecular changes in skeletal muscle of mice, relative to control, and importantly,

these molecular changes were partially blunted by PEG-IGF-1 treatment. Thus, reduced circulating levels of IGF-1 observed in mice may drive a compensatory mechanism in skeletal muscle, such that IGF-1 signaling is increased, and, hence, the impact of aging on muscle mass is ameliorated. Taken together, this work provides a stepping stone for future research investigating the mechanistic relationship between attenuation of age-related loss of muscle mass, strength, and function and reduced circulating levels of IGF-1 induced by CR.

Chapter IV: Tables and Figures Body Weight A Body Weight

4040

) 35 )

g Control

(

t 30

h CR-Vehicle

g i

g 25 i e CR-PEG 100

25 e W CR-PEG 500 W 20 20 15 150 1 2 3 4 5 0 1 2 3 4 5 Weeks on Diet Weeks on Diet

B Fasting Blood Glucose

e 150

s o

l ) G 100

g o

( 50

t s

0 F

l e 0 0 o l 0 0 tr ic 1 5 n h o e G G C V E E _ -P -P R R R C C C

Figure 4.1. Body weight and fasting blood glucose. (A) Body weight was recorded weekly throughout the 5 weeks. (B) Fasting blood glucose (6 hours) was obtained at time of sacrifice. Weight is presented in grams (g) and fasting blood glucose is in mg/dL. Sample size: Control (n=5); CR_Vehicle (n=4); CR_100 (n=5); CR_500 (n=5), *p<0.05, **p<0.01. Values are means SD. Statistical tests: (B) one-way ANOVA and Tukey’s multiple comparison post hoc test.

PEGylated IGF-1 A B ✱✱✱✱

) 500

L ✱✱✱✱

m /

g 400

n (

- 300

F G

- 200 G

l 100

a t

o T 0

l e 0 0 o l 0 0 tr ic 1 5 n h _ _ o e R R C V _ C C R C Figure 4.2. Exogenous PEG-IGF-1 increases IGF-1 serum levels in CR mice. (A) Serum concentration levels of PEGylated proteins were detected PEG-IGF-1 treated mice. (B) Total endogenous IGF-1 was reduced in CR mice compared to control (checkered bars for CR_100 and CR_500), however when PEGylated protein levels were added to endogenous IGF-1 levels, there was a significant increase in IGF-1 levelsC in CR_100 and CR_500 groups compared to CR_Vehicle. *p<0.05, **p<0.01, ****p<0.0001. Values are means SEM. Statistical tests: one-way ANOVAwith (A) Dunnett’s multiple comparison post hoc test, and (B) Tukey’s multiple comparison post hoc test

A B C MVy oblast Determination Protein 1 Myogenin Fatty Acid Synthase (MyoD) (MyoG) (FASN) 2.0 2.0 10 8

✱ 1.5 e

1.5 e

g g

g 6 n

h h

h 1.0 4 C

1.0 C

l o

o 2

F F F 0.5 0.5 0

0.0 0.0 -2 e l 0 0 l e 0 0 l e 0 0 l o 0 0 o l 0 0 ro l 0 0 ic tr 1 5 tr ic 1 5 t ic 1 5 h n _ _ n _ _ n h _ _ o h o e R R e R R o e R R C V V C C C C V C C _ C C _ _ R R R C C C

Figure 4.3. qPCR analysis of mRNAs derived from quadriceps muscle. Data were normalized by the amount of Actin and RPL4. (A) Myoblast determination protein 1 (MyoD). (B) Myogenin (MyoG). (C) Fatty acid synthase (FASN).

A PCA plot B Heat map

Figure 4.4. Gene expression analysis. (A) Principle component analysis (PCA) plot and (B) Heat Map. For PCA plots unit variance scaling is applied to rows; singular value decomposition (SVD) with imputation is used to calculate principal components. X and Y axis show principal component 1 and principal component 2 that explain 39.7% and 11.8% of the total variance, respectively. Prediction ellipses are such that with probability 0.95, a new observation from the same group will fall inside the ellipse. PCA and heat map were generated using ClustVis.

Table 4.1. Gene set enrichment analysis: Hallmark gene sets enriched in CR_Vehicle versus Control or CR+IGF-1 groups

Gene Set Enrichment Analysis CR_Vehicle versus Control CR_Vehicle versus CR_100 CR_Vehicle versus CR_500 (Hallmark) NES FDR q NES FDR q NES FDR q G2M_CHECKPOINT 2.82 0.00 2.73 0.00 2.80 0.00 E2F_TARGETS 2.74 0.00 2.69 0.00 2.81 0.00 MITOTIC_SPINDLE 1.79 0.00 2.07 0.00 2.26 0.00 DNA_REPAIR 1.64 0.00 1.46 0.02 1.71 0.00 HEME_METABOLISM 2.45 0.00 2.32 0.00 2.45 0.00 MTORC1_SIGNALING 1.85 0.00 1.61 0.01 1.70 0.00 MYC_TARGETS_V1 1.78 0.00 1.40 0.04 1.74 0.00 TNFA_SIGNALING_VIA_NFKB 1.79 0.00 1.94 0.00 1.62 0.00 REACTIVE_OXIGEN_SPECIES 1.68 0.00 1.55 0.01 1.67 0.00 COMPLEMENT 1.49 0.02 1.82 0.00 1.66 0.00 APOPTOSIS 1.58 0.01 1.74 0.00 1.44 0.02

CHAPTER V: SYNTHESIS

Overview of Research Findings

In the US, both as a percentage and absolute number, of the population elders is growing.

For example, the number of people in the oldest- old age group, aged 85 and over is projected to grow from 5.9 million in 2012 to 8.9 million in 203016. A major challenge accompanying the demographic shift towards an aging population in the US and many other countries is the maintenance of independence, physical health, and optimal function178. The aging process is accompanied by an increase in chronic health conditions and a high prevalence of functional limitations and disability associated with sarcopenia (defined as a decline in lean mass) and strength178.

Calorie restriction (CR) offers many beneficial anti-aging and health-promoting effects by acting at various levels of function and modulating a number of molecular, cellular, and systemic pathways to promote health and delay age-related decline55. However, the effects of CR on muscle health, particularly age-related sarcopenia, are incompletely understood.

This study adds to our understanding about the role of IGF-1-mediated effects of CR on phenotype-genotype relationships in skeletal muscle. These results provide biological insights into mechanism-based strategies to delay muscle aging. We present, for the first time to our knowledge, a comprehensive study of CR-induced processes in skeletal muscle across multispecies. We present results at both the muscle transcriptional level and systemic biomarker level (ongoing). We also investigated the efficacy of rapamycin, a CR mimetic drug, on age-

related changes in muscle. Finally, our results speak to mechanism testing of circulating IGF-1 on CR-induced effects on aging muscle.

This research largely focused on the functional and molecular basis underlying CR- induced changes in skeletal muscle in the context of aging. Our major findings are: (1)

Phenotypic changes in body composition, particularly an increase in relative lean mass, paralleled changes in gene expression that were associated with metabolic networks, antioxidant defense, and mechanical contractile force, in both mice and humans (Chapter III). (2) The pharmacological agent, rapamycin is a potential CR-mimetic treatment to counteract the effects of aging on skeletal muscle. Specifically, our novel preclinical findings revealed that critical functional, metabolic, and immune-related features were associated, in a sex-dependent fashion, with sarcopenia and were successfully improved with rapamycin treatment. This result provides evidence that even in advanced middle age, skeletal muscle is a highly flexible organ that adapts its transcriptional program to different dietary/treatment challenges (Chapter III). (3) We demonstrated that the reduced systemic levels of IGF-1 in response to the CR diet regimen drives age-related transcriptional changes in muscle towards an anti-aging gene signature and away from a sarcopenic signature (Chapter IV).

This dissertation work has many strengths. First, in the design, we incorporated advanced age and CR as variables in order to more closely model human populations. We developed and improved on existing protocols for balance beam testing in mice (submitted manuscript ; Orenduff et al.). Here, we developed a formalized testing protocol, which provided precise and reliable measures. We were able to show a decrement in performance with aging in mice and this change was modified by CR and rapamycin, albeit modestly in rapamycin mice.

Through our collaborations, we were able to: (1) Incorporate samples from multiple species and

test the reproducibility of major study findings across mice, nonhuman primates, and humans.

(2) Acquire a PEGylated form of IGF-1 that allowed us to manipulate circulating levels of IGF-

1 in mice. (3) Conduct analysis of CR-induced changes on IGF biomarkers (including free IGF-

1, a technically challenging analyt to measure) among multiple species. Additionally, our study involving the Women’s Health Initiative data set provided us with an opportunity to test the impact circulating IGF-1 and IGFBP7 concentrations on 20yr survival, and cancer in the largest study sample to date (Appendices Section 2 - submitted manuscript; Orenduff et al.).

This work also has several limitations. First, we are utilizing preclinical systems to model human age-related sarcopenia, and inherent differences exists across the human and mouse.

Another limitation, was the brief treatment exposure to PEG-IGF-1, which did not allow us to fully assess the effects of increased circulating IGF-1 over-time. Additionally we were also limited by not being able to obtain an endpoint body composition (MRI) assessment of mice in the CR/rapamycin study (Chapter III) due to COVID 19-related closures. The body composition measure at endpoint would have provided us with a more comprehensive measure of whole body changes in muscle and fat mass.

Public Health Implications and Future Perspectives

In animal model studies, many variants of CR differing in degree of restriction, length of

CR regimen, and timing (e.g., intermittent or time-restricted CR), have been explored, and, in general, demonstrate protection against age-related changes in skeletal muscle180 (ref). A major challenge in this area of research is establishing that the beneficial outcomes of CR on skeletal muscle observed in short-lived rodent species translate to humans. Unfortunately, there is a research gap from preclinical to clinical studies, which has been largely bridged by studies in

nonhuman primates sharing a high degree of similarity to humans. Establishing the link between

CR and muscle quality in humans has proven to be a challenge for many reasons, including the difficulty of maintaining a CR regimen in human subjects for an extended period of time55.

The results obtained from this work show that CR attenuates loss of muscle mass, strength, and function. At the tissue level, CR results in maintained contractile content and attenuated age-related metabolic shifts associated with mitochondrial function. However, despite these benefits, CR is a highly intensive dietary regime that is not easily achieved successfully due to the requirement of strict adherence to the diet plan involving undernutrition without malnutrition (inadequate micronutrient intake). Thus, the lack of medical supervision in older adults undergoing a CR diet might result in severe adverse effects including excessive loss of bone and muscle mass and increased risk of cardiometabolic disease and immune deficiencies9.

In fact, I would not recommend the implementation of a chronic, severe CR regimen in older adults as a way to intervene on age-related changes in muscle, out of concern for adverse effects.

However, lessons learned from our studies about druggable molecular targets will hopefully contribute to the development of new intervention strategies (such as mTOR inhibitors like rapamycin) to prevent age-associated sarcopenia.

The goal of our work was to identify the molecular underpinnings and mechanisms associated with the role of circulating IGF-1 on CR-induced effects in skeletal muscle and to test the efficacy of rapamycin, a CR-mimetic, on anti-aging effects in muscle. By demonstrating a potential causal link between reduced circulating levels of IGF-1, observed in mice with the anti- aging effects of CR, we identified new targets for mitigating age-related decline in muscle mass and/or function.. Further, we have shown rapamycin may have a potential role to attenuate loss of muscle strength and function associated with age. Following from this work, it will be

important to determine if reduced circulating levels of IGF-1 are responsible for triggering an autocrine signaling loop in skeletal muscle such that muscle regeneration and metabolic efficiency are stimulated due to nutrient stress. Future experiments to further unravel the role of circulating IGF-1 in age-related sarcopenia could include: (1) In vitro and ex vivo muscle experiments that could provide insight into the precise mechanism associated with IGF-1 signaling, both systemically and locally within skeletal muscle. (2) An in vivo experiment with an extended treatment duration of PEG-IGF-1 and incorporating functional tests to assess potential changes in muscle strength and function, may allow for a more definitive conclusion to the CR-induced role of circulating IGF-1 on aging muscle. This experiment was planned but could not be completed due to COVID-19. An important aspect to consider in response to dietary intervention and aging muscle is sexual dimorphism. Given the importance of sex as a biological variable, inclusion of both male and female mice is vitally important for all future studies addressing age-related sarcopenia.

An area of research that has not been explored is the effect of CR and the role of circulating IGF-1 in the context of cancer-associated cachexia. Cachexia is a systemic syndrome characterized by pathological weight loss due to excessive wasting of skeletal muscle and adipose tissue mass. This state of metabolic perturbation frequently accompanies many cancers, and results in anorexia, inflammation, insulin resistance, and increased muscle protein breakdown. Our findings suggesting systemic IGF-1 is part of a central pathway regulating muscle quantity and quality in disease states may provide new therapeutic targets for the prevention and treatment of several muscle wasting diseases, such as cachexia. (5) Moreover, our findings that CR and rapamycin both prevented age-associated decline in muscle function and mass via distinct mechanism suggest that a combination approach involving CR and rapamycin

might have greater anti-sarcopenic effects than either intervention alone. An experiment to assess the effects of a moderate CR regimen combined with rapamycin treatment may reveal a synergistic effect on aging muscle that has not yet been explored.

In conclusion, we found the phenotype-genotype relationship associated with aging skeletal muscle is malleable to dietary and pharmacological intervention. Further, our data suggest circulating IGF-1 may play a vital role in CR-induced effects on aging muscle, and our observed interactions between CR, IGF-1, and sarcopenia appear to be part of a conserved process across multiple mammalian species. These findings provide an impetus for further exploring the role of systemic IGF-1 in the etiology and mitigation of age-related sarcopenia.

APPENDIX 1

The following section contributes to this work by providing a follow-up study on IGF biomarkers thought to be involved in CR-induced effects on aging skeletal muscle. Specifically, the biomarkers included in this analysis are free IGF-1, IGFBP3, and IGFBP7. We hypothesize that changes in circulating IGF biomarkers result in a net effect to reduce the concentration of free, bioavailable IGF-1 and this effect is conserved across multiple species, including mice, nonhuman primates, and humans.

EFFECT OF CALORIE RESTRICTION ON SYSTEMIC BIOMARKERS OF THE INSULIN-LIKE GROWTH FACTOR (IGF) SYSTEM, IN MICE, NONHUMAN PRIMATES, AND HUMANS

Introduction

Insulin-like growth factor 1 (IGF-1) is a natural growth hormone and plays an essential role in normal muscle and tissue growth and development20. IGF-1 is the most potent activator of bioactivities including growth, proliferation, and differentiation of target cells, cell survival, and maintenance of cell function181. In addition to a role in regulating growth, survival, and maintenance, IGF-1 is also involved in diverse processes including regulation of macronutrient

(carbohydrate, protein, and lipid) metabolism and a major regulator of biological aging182,183.

These diverse functions of IGF-1 and the central signaling pathways at which it regulates require highly efficient regulatory mechanisms be in place to prevent a disruption in crosstalk signals between over and under stimulation and consequentially, an imbalance of IGF-signaling28.

In circulation, IGF-1 exists in two forms - bound IGF-1 and unbound or free IGF-128.

Cumulatively, these two forms of IGF-1 represent total systemic IGF-1 levels. To maintain a

100

high level of growth factor regulation in circulation, the most common form is bound IGF-1, of which is estimated to be ~99% of total IGF-1 under normal physiological conditions in adults

(~25-48 years)184. Bound IGF-1 exist in a stable tertiary complex comprised of acid-labile subunit (ALS) and one of six primary IGF binding proteins (IGFBPs) 1-6. The ratio of IGF-1 and IGFBPs (IGF:IGFBP), predominately IGFBP3, exist to maintain a delicate balance between biologically active and inactive IGF-1 and is considered to be the determinant of bioavailable

IGF-1 regulation in circulation and at the tissue level181. In addition to IGFBPs 1-6, circulating

IGF-1 activity is also regulated at the receptor level. IGFBP7, also referred to as IGFBP-related protein-1 (IGFBP rp1) or Mac25, has a 100-fold lower affinity for IGF-1, relative to IGFBPs 1-

6; however, it binds strongly to insulin and regulates IGF-1R and IR receptor activity and thereby modulates receptor activation by free IGF-1185. Of the two forms of circulating IGF-1, bound IGF-1 is biologically inactive and unable to interact with its receptor, IGF-1R, until released from the complex via proteolytic cleavage. Conversely, unbound or free IGF-1

(estimated to be ~1% of total IGF-1 in adults) is the bioavailable, more biologically relevant form, able to bind to and activate IGF-1R at target cells and trigger a series of signal transduction events181,184. Downstream of IGF-1R, intracellular substrates, insulin receptor substrate (IRS) and SRC homology containing (Shc) protein are phosphorylated and in turn, relay signals through either the MAPK or the PI3K cascade channels – both of which mediate growth, survival, and metabolic signaling pathways.

Circulating IGF-1 regulates growth, proliferation, and metabolism in all cells and thus, serves as a useful and viable biomarker for both pre-diagnostic and disease management settings181. Evidence to date suggest an increase or decrease in systemic total IGF-I levels, relative to ‘normal values’ (loosely based on age and gender) are associated with risk of multiple

101

metabolic diseases and disorders. For example, an increase in circulating total IGF-1 is associated with several types of cancers, including colon, breast, and prostate170. In contrast, reduced circulating levels of total IGF-1 have been linked to cardiovascular disease, type 2 diabetes, and age-related conditions such as sarcopenia, or the loss of muscle mass, strength, and function associated with age186.

Along with IGF-1, circulating IGFBPs (including IGFBP7), also serve as meaningful biomarkers. In addition to the classical roles of IGFBPs to modulate circulating IGF-1 bioavailability (IGFPBs 1-6) and interfere with IGF-1R activation (IGFBP7), IGFBPs also have unique and distinct roles that are independent of IGF-128,185. In particular, IGFBP7 is explicitly produced in skeletal muscle30, and along with IGFBP3 (which is the most abundantly expressed

BP in circulation), has been shown to interfere with TNFα signalingan IGF-1 independent pathway known to be involved in age-related muscle degradation31–34. Although the role(s) of

IGFBPs in the context of CR have yet to be identified, IGFBP 3 putatively plays an especially critical role in regulating systemic IGF-1 levels and activity28,34, while IGFBP7 acts primarily as an IGF-1R and IR receptor modulator to regulate signals downstream of IGF-1R and IR.

In addition to disease states, circulating total IGF-1 levels are responsive to changes in diet. For example, in studies on calorie restriction (CR) in mice, exposure to energy restrictive diets resulted in a 40% reduction in circulating total IGF-1, relative to ad libitum control mice187.

This significant decrease in circulating total IGF-1 observed in mice has been associated with

CR-induced effects to prevent the onset and progression of cancer, while also, and paradoxically, mitigates age-related diseases such as sarcopenia13,170,187. Interestingly, in CR studies conducted in nonhuman primates and humans, while many of the biological benefits observed in mice are conserved in higher mammalian species, circulating levels of IGF-1 are anomalous40,82. This

102

disparity in between-species differences in circulating levels of total IGF-1 presents an important gap in CR research and introduces several important issues about IGF-1 measurements employed in mainstream techniques to date.

Although free IGF-1 is the active and more potent form of IGF-1, total IGF-1 measures are the most broadly accepted clinical method for evaluating IGF-1 status. This is due to the relatively small concentration levels of free IGF-1, which make it technically complex and difficult to measure accurately35. As a result, total IGF-1 serves as a surrogate measure to estimate free IGF-1 levels. Most commonly, free IGF-1 is estimated to be ~1% of total IGF-1 across a variety of phenotype metrics, thus represents an inaccurate approximation at best35.

Based on between-species inconsistencies of CR-induced changes on total IGF-1 levels and questioning whether changes in free IGF-1 are reflected in total IGF-1 measures, we sought to pursue a more sensitive method to evaluate CR-induced changes on free IGF-1. One of the few laboratories, to our knowledge, capable of performing these measures of free IGF-1 with a high level of sensitivity is at McGill University in Canada. By employing the expertise at

McGill and deriving measures of free and total IGF-1, in addition to IGFBP 3 (the most abundant BP), and IGFBP7 (role in IGF-1R modulation) in mice, nonhuman primate, and human species, this work seeks to answer the following questions. Do current methods of measuring total IGF-1 and use of the 1% approximation metric for free IGF-1 accurately assess free versus bound IGF-1 levels. Further, does this estimated metric of 1% remain static across different nutrient intake levels (e.g., CR) and species? Finally, in consideration of IGFBPs and especially the prominent roles of IGFBP3 as the primary IGFBP in circulation and IGFBP7 as an important mediator of IGF-1R/IR activity, how do variables such as nutrient intake (CR) and species impact changes in circulating ratios between IGFBP3 and IGFBP7 with IGF-1?

103

This study includes plasma samples from mouse, nonhuman primates, and humans to test our hypothesis that CR reduces levels of circulating, free IGF-1 by increasing the ratio of IGF-

1:IGFBP3, and modulates receptor activity via IGFBP7. Additionally, these CR-induced effects on IGF- biomarkers are conserved across mammalian species. Further, this study provides a set of normative values, of which can be used in various clinical and research settings to compare and assess CR-induced changes on circulating free and total IGF-1 levels among mammalian species.

Materials and Methods

Mouse Study Design

The mouse CR study was conducted at Wageningen University in the Netherlands in close collaboration with the Hursting laboratory.

Male C57BL/6 J mice (7 weeks of age; purchased from Janvier (Cedex, France)) were housed in pairs (12-h light/dark cycle and light on at 4 a.m.). Mice were provided with ad libitum access to water and received a standard American Institute of Nutrition (AIN)-93 G diet

(Research Diet Services, Wijk bij Duurstede, The Netherlands). After a 14 day acclimation period (at 9 weeks of age) mice were individually housed and randomized to one of two diet intervention groups: (1) Control diet (AIN-93 W), provided ad libitum (n = 18) or (2) CR diet

(AIN-93W-CR), providing 70% of daily food intake. Portion size of the CR group was adjusted at the age of 6, 12, 18, and 24 months and based on measurements consumed in the control group on the week prior to the switch. The CR daily food portions were provided once a day, just before the start of the dark cycle, at 3:30 p.m. CR diets were supplemented with 100% vitamins, essential fatty acids, and minerals to provide a diet matched in micronutrient content as control.

104

Body weight was recorded every week during acclimation, then every two weeks for the remainder of the study.

Mouse Study Sample Collection

Six animals per group were euthanized at the age of 6, 12, and 24 months. Whole blood was collected via cardiac puncture and transferred to an EDTA-anticoagulant treated tube.

Plasma was then centrifuged (2000 x g) at 4°C for 15 minutes. The resulting supernatant was transferred into a clean 1.7 ml polypropylene tube and stored at -80°C.

Nonhuman primate Study Design

Investigators at the at the National Institute on Aging (NIA) in Baltimore, and the

University of Wisconsin's National Primate Research Center have contributed plasma samples from their CR studies conducted in nonhuman primates. These studies were by far the most elaborate and intensive CR studies conducted in any animal to date. Male and female rhesus monkeys (~9 years of age) were maintained on a 30% CR diet or ad libitum control diet for 18 years. Body composition, physical activity, and indices related to metabolic health (blood pressure, insulin and glucose parameters, and oxygen consumption) were closely monitored.

Additionally, blood samples were collected every 6 months and stored for future analyses. For purposes of this study, we obtained plasma samples from the ~18 year time point, at which the monkeys had been on diet treatments for ~9 years

Nonhuman Primate Study Sample Collection

Blood samples were collected every 6 months from rhesus monkeys over the course of 18

105

years. For the purposes of this study, the samples collected at the 18-year time-point were utilized. Animals were fasted (18 hours) and then anesthetized with ketamine hydrochloride

(10–15 mg/kg). Blood was drawn from the femoral vein into tubes containing 0.1% EDTA (pH

7.4, final concentration). Plasma was centrifuged at 1000×g, 30 min, 4°C, and stored at −80 C until assayed.

Human Study Design

CALERIE was a multisite single-protocol study conducted at three study sites:

Pennington Biomedical Research Center (Baton Rouge, LA, USA), Washington University

School of Medicine (St Louis, MO, USA) and Tufts University (Boston, MA, USA). Duke

Clinical Research Institute (Durham, NC USA) served as the central coordinating center for the study. All participants provided written informed consent and received financial compensation.

A data and safety monitoring board provided oversight of the study and institutional review board approval was obtained by each of the individual study sites.

As detailed elsewhere (REF), a sample size of 250 male (21-50 years) and female (21-47 years) non-obese participants were enrolled, and assigned to either the CR intervention or an ab libitum control group. A 2:1 allocation ratio in favor of the CR intervention was applied to maximize the number of subjects receiving the intervention of greater scientific interest. Participants in both treatment arms were followed over a period of 24 months. The active intervention targeted a sustained 25% restriction in calorie intake vis-à-vis ad libitum energy intake measured by doubly labeled water (DLW) at baseline. The CR intervention was implemented by a multi-disciplinary team including dietitians, psychologists, and physicians. No specific diet composition was mandated. Rather, the CR intervention was tailored to the needs of the individual participant,

106

with specific nutritional and behavioral strategies employed. A comprehensive set of evaluations were performed prior to initiating the intervention, with follow-up evaluations at Months 1, 3, 6,

9, 12, 18 and 24 after randomization.

Human Study (CALERIE) Sample Collection

Fasting (12 hours) blood samples were collected by a trained phlebotomist at baseline, 3,

12, and 24-month follow-up visits. Blood samples were collected and transferred to EDTA tubes. Samples were then centrifuged to separate the plasma, aliquoted, and stored in a −80°C freezer.

Biomarker Assays

Commercially available ELISA kits were used to measure total IGF-1 (Item # AL-121), free IGF-1 (Item # AL-122), IGFBP3 (Item # 120) (Ansh Labs, Webster, TX) and IGFBP7 (Item

# DV009) (Sino Biological Inc. CedarLane, Ontario, Canada). ELISA kits were species-specific for mice, nonhuman primates, humans.

Statistical Analysis

The goal of the analysis is to identify and compare a comprehensive profile of circulating

IGF proteins, including free IGF-1, total IGF-1, IGFBP3, and IGFBP7 with plasma from human, nonhuman primate, and mouse study samples and to compare the CR groups over time.

Specifically, we will examine between group differences in IGF-1 (free and total) IGFBP 3, and

IGFBP7. To control for Type-I error rate, inherent in the testing of multiple outcomes in the mouse, nonhuman primate, and human models, we will analyze the change over time using

107

multivariate repeated measures mixed models, simultaneously modeling the effect of IGF-1 (free and total), IGFBP3, and IGFBP7. Use of longitudinal mixed models has the advantage of allowing for missing values in the analysis, while employing a multivariate testing structure allowed us to assess both a generalized effect across all IGF proteins, the individual effect for each IGF protein, and differences between these individual effects. In all analyses, we tested for both group, and group by time interactions. Use of multivariate mixed models has additional power advantages relative to alpha splitting methods for individual outcomes, like Bonferroni or

Hochberg modelling several outcomes individually. Rather, if there is an omnibus effect, the test of individual outcomes can proceed without Type-I adjustment for multiple tests. Since the mouse, nonhuman primate, and human panels of growth factors will be the same (free and total

IGF-1 and IGFBP3), at the end of the analysis, the individual species results can be compared and contrasted using meta-analytic techniques, to assess the homogeneity of effect as well as heterogeneity between the mouse/nonhuman primate/human models.

108

APPENDIX 2

This work includes an additional project, involving the Women’s Health Initiative study

(WHI), conducted as a part of my doctoral training. WHI was a multisite (40 research centers), long term national health study that focused on identifying risks and developing strategies for the prevention of heart disease, breast and colorectal cancer, and osteoporosis in postmenopausal women1,2. WHI was comprised of three studies: clinical, observational, and community prevention. Our investigation includes an analysis of data from an ancillary study conducted within the observational arm2. In addition to the wealth of information collected as part of the primary observation study, we collected additional biochemical measures from serum samples.

Our goal was to assess the relationship between serum concentrations of insulin-like growth factor (IGF) biomarkers (IGF-1, IGFBP7, and IGFBP7-IGF-1 ratio) with subsequent breast and all cancer incidence and all-cause mortality in 793 postmenopausal women. The following manuscript is submitted for internal (WHI) review

Associations between Insulin-Like Growth Factor (IGF) Binding Protein-7 and Socio-Demographic Variables, Body Mass Index, and All-Cause Mortality in Postmenopausal Women from the Women's Health Initiative Observation Study

Authors: Melissa Orenduff1, Emma H. Allott2, Carl Pieper3, Christopher Ford4, Su Yon Jung5,

Mara Z. Vitolins6, Jenifer I. Fenton7, Chu Chen8, Oleg Zaslavsky9, Candyce Kroenke10, Fred K.

Tarang11, Cynthia A. Thompson12, Ana Barac13, Electra D. Paskett11,14, Shine Chang15, Michael

Pollak16, Stephen D. Hursting1, 17

Affiliations: 1Department of Nutrition, University of North Carolina, Chapel Hill, NC

2Centre for Cancer Research and Cell Biology, Queen’s University, Belfast, UK

109

3Department of Biostatistics and Bioinformatics, Duke University School of Medicine, Durham,

4Emory Global Diabetes Research Center, Emory University, Atlanta, GA

5Translational Sciences Section, Jonsson Comprehensive Cancer Center, School of Nursing,

University of California Los Angeles, Los Angeles, CA

6Department of Epidemiology and Prevention, Division of Public Health Sciences, Wake Forest

School of Medicine, Winston-Salem, NC

7Department of Food Science and Human Nutrition, Michigan State University, East Lansing,

8Program in Epidemiology, Division of Public Health Sciences, Fred Hutchinson Cancer

Research Center, Seattle, WA

9Biobehavioral Nursing and Health Systems, University of Washington School of Nursing,

Seattle, WA

10Division of Research, Kaiser Permanente Northern California, Oakland, CA

11Comprehensive Cancer Center, The Ohio State University, Columbus, OH

12Canyon Ranch Center for Prevention and Health Promotion, The University of Arizona,

Tucson, AZ

13MedStar Health Research Institute, Georgetown University, Washington, D.C.

14Division of Cancer Prevention and Control, Department of Medicine, College of Medicine, The

Ohio State University, Columbus, OH

110

15Department of Epidemiology, The University of Texas MD Anderson Cancer Center, Houston,

16Department of Oncology, McGill University, Montreal, Quebec, Canada

17Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, NC

Running Title: IGFBP7 and mortality in the Women’s Health Initiative

Key words: (7) insulin-like growth factor-1, insulin-like growth factor binding protein 7, postmenopausal, breast cancer, obesity, cancer, all-cause mortality.

Financial support: This study was supported by R35 CA197627 from the NIH/National Cancer

Institute (S.D. Hursting)

Correspondence:

Stephen D. Hursting, PhD, MPH

University of North Carolina at Chapel Hill

235 Dauer Drive, MJHRC Room 2107

Chapel Hill, NC 27599-7400

Phone: 919-966-7346

Email: [email protected]

Conflicts of Interest Disclosures: The authors have no conflicts of interest to report.

111

Abstract

Background: Reduced systemic levels of the hormone insulin-like growth factor (IGF)-1 are associated with reduced risk of several age-related chronic diseases; however, heterogeneity exists in the strength of these associations. IGF-1 bioactivity is regulated by multiple IGF binding proteins (IGFBPs). The ratio of systemic IGFBP7 to IGF-1 has recently been associated with increased cardiovascular disease risk and cardiac failure, although associations with cancer remain unclear. Herein, we describe the serum levels of IGFBP7, IGF-1, and their ratios in a cross-sectional study of older white and African American women, and the relationships between these biomarkers and the women’s baseline characteristics, cancer incidence, and all-cause mortality.

Methods: Serum total IGF-1 and IGFBP7 levels were measured at the third annual visit in 793 postmenopausal women participating in an ancillary study of the Women’s Health

Initiative Observational Study between February 1995 and July 1998, and longitudinal follow-up data were collected through 2019. We conducted regression and survival analyses to examine associations between the women’s IGF pathway biomarkers (IGFBP7, IGF-1, and their ratio), socio-demographic and phenotypic features (e.g., race, age, BMI and physical activity level), cancer incidence (total and breast-specific), and all-cause mortality, controlling for potential confounders.

Results: Serum IGFBP7 levels and the IGFBP7: IGF-1 ratio, but not IGF-1 levels, were significantly higher in obese women than nonobese women. Regression analysis showed significant associations between (a) IGF-1 and race (white, relative to African American, women had lower levels); (b) IGFBP7 and BMI, age, MET hour equivalents of physical activity per week, marital status, household income and years of education; and (c) IGFBP7:IGF-1 ratio and

112

BMI, age, race, and BMI, MET hours, household income, and years of education. Proportional hazards modeling showed significant associations between all-cause mortality and both IGFBP7 levels (hazard ratio [HR] = 2.72; 95% CI, 1.54-4.80; p<0.001) and IGFBP7:IGF-1 ratio (HR

=1.60; 95% CI, 1.21-2.11, p<0.001). Associations that approached but did not achieve significance included IGFBP7 and cancer incidence (HR = 2.24; 95% CI, 0.95-5.35; p=0.066), and IGF-1 levels and breast cancer incidence (HR=1.94; 95% CI, 0.90-4.20, p=0.093). Other associations were nonsignificant.

Conclusions: Circulating IGFBP7 levels and the IGFBP7:IGF-1 ratio are significantly associated with obesity, race, and all-cause mortality in postmenopausal women. Findings from these hypothesis-generating prospective analyses suggest that serum IGFBP-7 and the IGFBP7:

IGF-1 ratio are promising biomarkers for assessing IGF–disease relationships.

113

Introduction Insulin-like growth factor-1 (IGF-1), a key regulator of cellular growth and proliferation, is involved in the pathogenesis of breast cancer and multiple other malignancies188. IGF-1 binds to the IGF-1 receptor (IGF-1R) and triggers a signal transduction cascade that leads to increased mitogenic activity and enhanced survival – processes quintessential to cellular transformations induced during tumorigenesis188-190. These effects are in part dependent upon IGF-1 bioavailability, which are tightly regulated by an extensive family of secreted IGF binding proteins (IGFBPs). IGFBPs 1-6 inhibit IGF pathway activation by sequestering IGF-1 into an inactive complex (IGF-1:IGFBP), thus reducing ligand availability191. In contrast, IGFBP7, also known as IGFBP-related protein-1 (IGFBP-rp1) or Mac25, modulates IGF-1 activity by binding directly to IGF-1R, thereby blocking its activation and inhibiting growth-signaling pathways192.

Additionally, there is a growing consensus that IGFBP7 functions predominately as a tumor suppressor in many types of cancer, including breast, colorectal, liver, and ovarian cancers192-196.

IGFPB7 participates in several physiological processes, with the net effect of inhibiting protein synthesis, suppressing cell growth and proliferation, and activating apoptosis, independent of the

IGF signaling pathway192. However, the effects of IGFBP7 appear to be tissue-specific, as it contributes to tumor progression in some types of cancer. For example, in glioblastomas, a positive correlation was observed between IGFBP7 expression and cell growth and migration197.

Similar findings have also been reported for esophageal, prostate, gastric, and colon cancers, especially when IGFBP7 is highly expressed at the invasive front of the progressing tumor198-201.

Collectively, these results speak to the complexity and, as yet undetermined, mode of action for

IGFBP7 in cancer and other chronic disease development. IGFBP7 is a secreted protein and is expressed by virtually all cells, including tumor cells. Biological functions of IGFBP7, both anti- and protumorigenic roles, have been largely studied at the cellular level via gene expression

114

patterns of the tumor or host cell and the surrounding environment. Circulating concentrations of IGFBP7 and cancer incidence have not been widely investigated, and to our knowledge, no studies have reported on the association of circulating IGFBP7 and cancer in a large longitudinal cohort study such as the Women’s Health Initiative. There have been recent reports that increased IGFBP7:IGF-1 ratio may predict heart failure and associate with cardiovascular disease-associated mortality202-203. Thus, the objective of this study is to examine the associations between circulating serum levels of IGFBP7 or the ratio of IGFBP7: total IGF-1 and phenotypic and socio-demographic variables, total cancer incidence, breast cancer incidence and all-cause mortality in postmenopausal women participating in an ancillary study of the Women’s

Health Initiative Observational Study.

Methods

Study population

This prospective study included 793 postmenopausal women enrolled in an Ancillary

Study of the Women’s Health Initiative Observation Study (WHI-OS). Participants were recruited at the WHI clinical centers located at the Baylor College of Medicine (Houston, TX) and the Wake Forest University School of Medicine (Winston-Salem and Greensboro, NC) between February 1995 and July 1998. The Ancillary Study included baseline measures and blood collected at enrollment and a 3-year follow-up at annual visit 3 (AV3). Study participants were postmenopausal women of European-American and African-American descent, aged 50-79 years, and able to provide written consent. Subjects were excluded if they had medical conditions predictive of survival less than 3 years, had complicating conditions such as alcoholism, drug dependency, or dementia, or did not plan to reside in the area for at least 3

115

years (due to AV3). Details of the scientific rationale, eligibility requirements, and other aspects of the design of the WHI-OS have been published elsewhere204. We excluded participants with inadequate sample volume for IGFBP7 and IGF-1 measurement (n=42 of 835 total participants), resulting in a final analytic sample of 793 participants. The study was approved by the

Institutional Review Boards at the University of Texas MD Anderson Cancer Center, the Baylor

College of Medicine, the Wake Forest University School of Medicine, and the University of

North Carolina at Chapel Hill.

Data collection

Participants completed self-administered questionnaires providing demographic and behavioral information in addition to medical and reproductive history. Behavioral factors, including smoking status (never, former, current smoking), and physical activity were evaluated at baseline and AV3 (three years after baseline). Anthropometric data, including height and weight were collected at each clinic visit by trained staff. Obesity was defined as BMI ≥30 kg/m2, according to recommendations from the World Health Organization. Physical activity was converted to metabolic equivalent of task (MET) hours per week, and categorized as <10 versus ≥10 hours per week, as described205.

Laboratory methods

Fasting blood samples were collected from each participant at AV3 by trained phlebotomists. The samples were sent to clinics’ laboratories for processing, and serum aliquots were stored at -80°C. The assay for IGF-1 has been described previously19. IGFBP-7 was measured using an enzyme-linked immunosorbent assay (IGFBP-7 DuoSet ELISA, R&D

Systems, Minneapolis, MN). The lower detection limit of the IGFBP7 assay was 39.1 pg/ml. The ratio of IGFBP7:IGF-1 was also calculated from the measured analyte data for each subject.

116

Statistical analysis

This is a prospective study with the goal of assessing the relationship of IGFBP7 to demographic and health predictors, and cancer outcomes. Initially, we assessed the distribution of demographic, health, and biomarker data (IGFBP7, total IGF-1, and a calculated ratio of

IGFBP7:IGF-1). Second, we assessed the correlation of IGF-1, IGFBP7 and IGFBP7: IGF-1 with the demographics and health indicators. Subjects with missing values on the IGF biomarkers (IGF-1 and IGFBP7; n=42 due insufficient sample) were deleted, while subjects with missing values on the predictors were given imputed values – the overall mean for continuous variables and the modal value for categorical values. For the final, multivariable models we checked that these imputations did not affect the results substantively. Initially, we assessed the bivariate association of the predictors with the IGF biomarkers by Spearman’s Rho to allow for the skewed distribution in IGF markers. In a sensitivity analyses, we assessed the impact of the

3 biomarkers of interest with, race, BMI group, and age, comparing all pairwise interactions of the demographics with the biomarkers, adding each of the 2 pairwise interactions separately to the main effects model. All tests of significance were based on a p-value of 0.05 (two-tailed) significance level.

Our primary analyses focused on the associations of serum IGFBP7, IGF-1 and the

IGFBP7:IGF-1 ratio, and 3 outcomes – time to death, all cancer diagnosis, and breast cancer. All cancer was defined as breast, colon, colorectal, endometrial, ovarian, and pancreas cancers.

Initially, separately for each outcome, Kaplan-Meier analysis was used related to each of the IGF predictor groups split at the median. Time to last interview was used as a censoring event.

Further, time to death was a censoring event for time to all cancer and breast cancer diagnosis.

The metric of ‘time’ was initially defined as time to event, relative to measurement of IGF. We

117

also analyzed age at event, with and without left censoring at the age of IGF measurement to control for the ‘immortal time’ prior to the point of measurement. Finally, we ran a single ‘fully controlled’ model for each outcome and IGF combination, using the set of covariates listed above. The combination of the time metric (age/time), control (Y/N), left censoring (Y/N), outcome (death, all cancer diagnosis, breast cancer), survival model (PHreg, Kaplan-Meier), and

IGF measure (IGF1, IGFBP7 and their ratio) led to a variety of models. In each model, the impact of the IGF marker is the predictor of interest. We did not control for the underlying

Type-I error rate inherent in the estimation of multiple correlated models by, for example,

Bonferonni correction. Rather, since this analysis was exploratory and not confirmatory, we did not adjust for the Type-I error and, for any significant effect observed, we warn of the potential for Type-I error.

Results

Characteristics of participants

A total of 793 postmenopausal women 50 to 79 years old were included in this analysis

(Appendix 2 Table 1). Among the participants, 30.6% (n=243) were obese (BMI ≥30 kg/m2),

30.9% (n=245) were overweight (BMI 25.0-29.9 kg/m2), and 38.5% (n=305) were normal weight (BMI (<24.9 kg/m2). European white women were more likely to be normal weight whereas African-American women were more likely to be in either the overweight or obese categories. Obese women had the lowest level of physical activity, education, and annual household income. Marital status was similar across groups. A significantly higher percent of

African-American women were prescribed medications for diabetes (pills: 20.3% vs. 7.2%)

118

(shots: 5.6% vs. 1.7%) and hypertension (pills: 62.2% vs. 42.4%) at the 3 year follow up visit

(AV3) compared with white women, respectively.

Correlations of serum IGFBP7, IGF-1 and the IGFBP7:IGF-1 ratio with BMI and socio- demographic factors

We observed that the IGF biomarkers were distributed log-normally. So, for multivariable regression models predicting each outcome, we log-transformed each of the IGF biomarkers, which allowed the statistical testing to assume approximate normality. Appendix 2

Table 2 shows the correlation between IGF-1, IGFBP7 or their ratio with the demographic and health predictors by spearman correlation. For IGF-1, significant relationships were observed for race (r=-0.12, p<0.001) and marital status (r=-0.09, p=0.01). For IGFBP7, there was a significant positive correlation with BMI (r=0.10, p=0.0051) and age (r=0.32, p<.0001) and negative correlation with marital status (r=-0.129, p<0.0001), MET (r=-0.080, p=0.026), education (r=-0.0928, p=0.009), and annual income (r=-0.22, p<0.001). The ratio of IGFBP7:

IGF-1 had a significant positive association with BMI (r=0.071, p=0.045), age (r=0.22, p<0.0001), race/ethnicity (r=0.118, p<0.001), and inverse associations with MET (r=-0.0706, p=0.047), years of education (r=-0.0954, p=0.007), and annual household income (r=-0.136, p<0.001).

Prediction of IGFBP7, IGF-1 and the IGFBP7:IGF1 ratio:

As shown in Appendix 2 Table 3, in controlled analyses of IGF-1, in a main effects only model, race and marital status were statistically significant variables. White women had 13% lower IGF-1 levels compared with African American women and married women had 7% lower levels of IGF-1. For IGFBP-7, increasing age, higher BMI, and education were significantly

119

associated with higher circulating levels. With each year of increasing age and education,

IGFBP7 increased by 1% and .82%, respectively. Circulating IGFBP7 levels decreased by 6%

in both normal and overweight groups compared with obese. For the IGFBP-7: IGF-1: ratio,

increasing age, race, higher BMI, and education were significantly associated with higher ratios.

Survival Analysis

Control variables included: BMI, Years of Education, Income (in dollars), METs of

activity/day (hours per week), Alcohol in the last 3 months, current smoking at intake, Diabetes,

Hysterectomy, Estrogen use, Estrogen w/ progesterone use (Appendix 2 Table 4)

IGFBP7: The Kaplan-Meyer (K-M) estimates for IGFBP7, split at the median, were predictive

of time to death (p=0.0001). The K-M estimates for 18-year survival were 54.7% for above the

median and 78.4% for below the median. K-M estimates were also predictive of age at death

(p=0.0010). The K-M survival estimate at 80 years was 76.8% for above the median and 88.5%

for below the median. Using the Proportional Hazards (PH) regression model to incorporate left

censoring for age at entry into the study, we found the hazards ratio (HR) = 1.30 (95% CI = 1.04,

1.622, p=0.0226). For the estimated HR for the log of IGFBP7 (continuous), we found HR =

3.72 (95% CI = 2.17, 6.37, p<0.001). Controlling for a priori chosen variables, the estimated HR

was 2.856 (95% CI 1.60, 5.08, p=0.0004).

IGF-1: The K-M estimates for IGF-1, split at the median, were predictive of time to death

(p=0.026). The K-M estimates for 18-year survival were 69.5% for above the median and 60.8% for below the median. The K-M estimates for age at death did not differ significantly (p=0.205)

120

and the survival curves at 80 years were nearly equivalent for above and below the median at

~80%. The estimated HR for age at entry into the study was 0.95 (95% CI = 0.77, 1.18, p=0.643).

Using the log of IGF-1 (continuous), the estimated HR was 0.83 (95% CI = 0.62, 1.12, p=0.230).

In controlled models, the HR was 0.72 (95% CI = 0.52, 1.00, p=0.0509).

All cancers (breast, colon, colorectal, endometrial, ovarian, and pancreas cancers)

IGFBP7: Groups were predictive of time to first all cancer diagnosis over the 18-year follow-up

(p=0.0441) and age at first cancer (p=0.0230). K-M estimates with IGFBP7 grouped at the median were 80.2% above and 85.0% below the median for 18-year survival and 68.5% above and 77.3% below the median at 80 years of age. Using proportional hazards models with the log of IGFBP7 (continuous), we found HR = 2.60 (95% CI = 1.15, 5.90, p=0.0219) and in controlled analyses, HR = 2.32 (95% CI = 0.96, 5.57, p=0.060).

IGF-1: Grouped at the median, K-M survival estimates for time to first cancer diagnosis over the

18-year follow up were nearly identical at 85.2% versus 85.4% (NS) for above and below the median, respectively. For age at cancer free survival, the K-M estimates indicated that the curves differed (p=0.035). At age 80, the cancer free survival estimates were 82.5% for above the median versus 85.0% for below the median. The risk of age of all cancer diagnosis for IGF-1 levels and was HR = 1.19 (95% CI 0.75, 1.90, NS) and HR = 1.03 (95% CI 0.63, 1.68, NS) in controlled analyses.

121

Breast cancer

IGFBP7: The K-M risk, with IGFBP7 grouped at the median, was only at a slightly higher risk of breast cancer diagnosis (p=0.3420). The survival estimates at the 18-year follow-up were

90.9% above the median and 93.0% below the median. For age at breast cancer diagnosis, the high IGFBP7 group was associated with earlier diagnosis of breast cancer (p=0.006). The survival estimates at 80 years was (71.0% above and 82.8% below the median). When left censoring was implemented, we found the HR for IGFBP7 was 1.79 (95% CI = 0.52, 6.21, p=0.021) and when the covariates were added to the model the HR was 1.83 (95% CI = 0.47,

7.13, p=0.384).

IGF-1: For IGF-1, grouped at the median, the K-M survival curves of breast cancer diagnosis did not differ significantly (p=0.634) and the survival estimates at the 18-year follow-up were at

92.8% above and 93.1% below the median. The survival curves for age at breast cancer diagnosis were 91.6% above and 93.1% below the median and did not differ significantly

(p=0.458). When left censoring was employed, HR for IGF-1 was 1.12 (95% CI = 0.75, 1.90, p=0.460) and 1.03 (95% CI = 0.63, 1.69, p=0.914) in controlled analyses.

Discussion

This cross-sectional study of serum IGFBP7 and its ratio with total IGF-1 in cancer-free, postmenopausal women participating in the observational arm of the WHI yielded 3 important findings that should drive future research on this biomarker. First, serum IGFBP7 and the

IGFBP7:IGF-1 ratio were both significantly higher in obese relative to normoweight or overweight women. Second, IGFBP7 was significantly correlated with BMI, age, MET hour equivalents of physical activity per week, marital status, household income and years of

122

education. Similarly, IGFBP7:IGF-1 ratio also correlated with BMI, age, MET hours per week, years of education and household income, and unlike IGFBP7 alone, was also correlated with race. Total IGF-1 only correlated with race (higher levels in African American than white women) and marital status. The third key finding, from our Kaplan-Meier estimates and partial hazards regression analyses that controlled for potential confounders, was that systemic IGFBP7 levels and the ratio of IGFBP7:IGF-1 were significantly associated with all-cause mortality.

Serum IGFBP7 levels, but not the ratio, also showed a trend towards an association with total cancer incidence. In contrast, total IGF-1 showed no significant associations with any of the outcomes, although a trend towards an association between serum IGF-1 and breast cancer incidence was observed.

IGFBPs 1-6 regulate IGF-1 bioavailability and facilitate tissue-specific activity208-210, including growth factor signaling and metabolic regulation, in an IGF-dependent fashion. IGFBPs can also function independently of IGF-1209. In particular, IGFBP7, also called IGFBP rp1 or Mac25, has a 100-fold lower affinity for IGF-1 relative to the other IGFBPs, but it regulates IGF-1R and IR receptor activity.

To date, nearly all assays used for assessing serum or plasma levels of IGF-1 in humans or animals have been limited to measuring total IGF-1, which cannot distinguish the free versus bound form of IGF-1. Despite clear evidence that bioavailable IGF-1 plays an important role in obesity-associated comorbidities from animal studies and epidemiologic studies employing

123

Mendelian Randomization approaches, the majority of human studies show either no association or an inverse association between obesity and total serum IGF-1211,212. In our study, IGF-1 showed no association with BMI, socio-demographic variables, cancer incidence or all-cause mortality. New biomarkers reflecting IGF bioactivity are needed for studying mechanistic links between risk factors, such as obesity, and chronic diseases, such as cancer and cardiovascular disease. Given the role of IGFBP7 as an IGF-1 receptor and insulin receptor modulator, and its emergence as a biomarker in cardiovascular research, our findings that serum IGFBP7 is associated with BMI, age, physical activity, all-cause mortality and possibly cancer incidence suggests it should be further assessed as a biomarker of IGF bioactivity.

The IGFBP7:IGF-1 ratio has been shown to be a better predictor of the risk of dying from heart failure than either IGFBP7 or IGF-1 alone15. To our knowledge, this ratio has not been previously assessed in studies of cancer or all-cause mortality. Our findings show that the serum

IGFBP7:IGF-1 ratio, like IGFBP7 alone, was associated with BMI, age, MET hours of physical activity, and all-cause mortality, and (unlike IGFBP7) was also associated with race. However, the Spearman correlation coefficients and p values describing the relationships between the

IGFBP7:IGF-1 ratio and BMI, age, MET, suggest weaker associations for the ratio than for

IGFBP7 alone. Similarly, the HR describing the association between IGFBP7:IGF-1 and all- cause mortality (1.60) was lower than for IGFBP7 (2.72). Moreover, the trend towards an association between cancer incidence and IGFBP7 (HR=2.72; p=0.066) was not observed with the ratio (HR=1.34; p=0.24). Thus, our data suggests IGFBP7 alone, and not the IGFBP7:IGF-1 ratio, may be the better biomarker to include in future studies.

This study has several strengths, including the prospective design of the Women's Health

Initiative from which our blood samples were collected, processed and stored as part of an

124

ancillary study between 1995 and 1998. Thus, we are capitalizing on the high quality of baseline socio-demographic and anthropometric data collected on these women, and the 25 years of careful, longitudinal follow-up inherent with this study. In addition, since our serum was collected prior to the development of disease in our participants, this establishes temporality and minimizes selection bias relative to retrospective designs. Moreover, our serum IGF-1 and

IGFBP7 measures were performed in the laboratory of our collaborator, Dr. Michael Pollak, the world leader in the measurement of biomarkers of the IGF pathway. This was particularly important or the IGFBP7 assay, which is not yet broadly available commercially. There were also several weaknesses of this study, including a relatively small sample size (n=793), and thus relatively small number of cancer cases, and even fewer cancer-related deaths, over the ~25 year follow up period. This limited our assessment of cancer outcome variables to total cancer incidence and breast cancer incidence, since the rates of other female cancers were too low to assess their associations with IGFBP7. Also, other components of the IGF pathway, particularly

IGFBP1-6, may influence the relationships between IGFBP7 and outcomes, but due to budget limitations on this hypothesis-generating pilot study, we were unable to analyze those biomarkers.

In conclusion, this prospective study of postmenopausal women demonstrated that serum

IGFBP7 levels are significantly associated with obesity, age, MET hours of physical activity, and all-cause mortality. In addition, the association between IGFBP7 and cancer incidence (all cancers combined) was borderline statistically significant. In contrast, serum IGF-1 levels were associated with race but not with all-cause mortality or cancer incidence, although a borderline association was observed with breast cancer incidence. The ratio of IGFBP7:IGF-1 showed similar, albeit weaker, associations with BMI, socio-demographic variables, and all-cause

125

mortality relative to IGFBP7 alone, with the exception of race (not associated with IGFBP7) and cancer incidence (not associated with the ratio). Findings from these prospective analyses support serum IGFBP-7, and possibly the IGFBP7: IGF-1 ratio, as promising biomarkers for assessing IGF–disease relationships.

126

Body mass index in kg/m2 Normal Overweight Obese Total N (%) 305 (38.5%) 245 (30.9%) 243 (30.6%) 793 Age, years (SD) 62.7 (7.09) 62.1 (6.76) 61.1 (6.94) 62.1 (6.97) Race/ethnicity, n (%) White 282 (92.5%) 191 (78.0%) 185 (76.1%) 658 (83.0%) African American 23 (7.5%) 54 (22.0%) 58 23.9% 135 (17.0%) Marital status, n (%) Married 205 (67.2%) 157 (64.1%) 151 (62.1%) 513 (64.7%) Other 100 (32.8%) 88 (35.9%) 92 (37.9%) 280 (35.3%) (Never married/Divorced/separated/widowed) MET hours/week, n (%) ≥10 138 (45.2%) 72 (29.4%) 61 (25.1%) 271 (34.2%) <10 167 (54.8%) 173 (70.6%) 182 (74.9%) 522 (65.8%) Education, years (SD) 14.38 (2.05) 14.14 (2.06) 13.82 (2.02) 14.13 (2.05) Annual household income, $65,757 $48,853 $43,341 $53,665 USD (SD) ($48,746) ($35,774) ($30,860) ($41,152) IGF Biomarkers (ng/mL) mean (SD) 144.90 143.13 140.75 143.08 IGF-1 (56.64) (54.98) (56.30) (55.99) 82.47 82.58 88.07 84.22 IGFBP7 (18.61) (19.11) (25.27) (21.14) Incidence of Cancer and Death, n (%) Breast Cancer Diagnosis 27 (8.8%) 14 (5.7%) 13 (5.3%) 54 (6.8%) All Cancer Diagnosis 38 (12.5%) 20 (8.2%) 22 (9.0%) 80 (10.1%) (including breast) All-cause mortality 96 (31.5%) 75 (30.6%) 98 (40.3%) 269 (34.0%)

127

Appendix 2 Table 2: Spearman Correlation Coefficients, (n = 793) (Bivariate Correlations)

Correlations

Variables IGFBP7 LogIGFBP7 (p value) IGF-1 IGFBP7 IGF-1 LogIGF-1

BMI (kg/m2) -0.0161 0.1000 0.0711 0.0455

(0.651) (0.0048) (0.0454) (0.200)

Age (years) -0.0590 0.325 0.219 0.137 (0.097) (<0.0001) (<0.0001) (0.0001) Race/ethnicity -0.123 0.00923 0.118 0.118 (0.0005) (0.795) (0.0008) (0.0008) Marital status -0.0870 -0.125 0.00535 0.0519 (0.0142) (0.0004) (0.880) (0.1440) MET 0.0308 -0.0821 -0.0706 -0.05156 (0.386) (0.0208) (0.0467) (0.147) Education, years 0.0617 -0.0936 -0.0954 -0.0816 (0.0824) (0.0084) (0.0072) (0.0215) Annual household 0.0377 -0.187 -0.136 -0.0874 Income (0.289) (<0.0001) (0.0001) (0.0138)

128

Appendix 2 Table 3: Regression coefficients

IGF-1 IGFBP7 IGF-1:IGFBP7 Race (p value) White -0.134 (0.0007) 0.0233 (0.281) 0.106 (0.0025) African American ref. ref. ref. Age (per year) -0.00344 0.0103 0.01.03 (p value) (0.105) (<.0001) <0.0001

MET (hours/week) 0.0159 -0.0222 -0.0227 (p value) (0.604) (0.190) 0.407 BMI (p value) Normal 0.0365 (0.307) -0.0627 (0.0015) -0.0565 (0.0762)

Overweight 0.0185 (0.608) -0.0655 (0.0010) -0.0589 (0.0670)

Obese ref. ref. ref. Income (per $10,000) 0.0140 -0.00250 -0.00636 (p value) (0.0826) (0.574) (0.376) Marital Status (p value) Married -0.0703 (0.0321) -0.0286 (0.113) 0.00848 (0.771) Other ref. ref. ref. Education 0.00537 -0.00824 -0.0229

(p value) (0.516) (0.0427) (0.0005)

Interactions (p values) BMI group*Age 0.0017 0.518 0.018 BMI group*Race 0.528 0.494 0.880 Race*Age 0.621 0.0127 0.0881

129

Appendix 2 Table 4. Survival Analysis

Outcome Model Censoring Predictor Form Events n (%) IGF-1 IGFBP7 IGFBP7:IGF-1 ratio Hazard Ratio Hazard Ratio Hazard Ratio P value P value P value (95% CI) (95% CI) (95% CI) All-Cause End of Followup Above/below 269/793 (33.9%) K-M estimate 0.056 <0.0001 <0.0001 Mortality Median End of Followup 269/793 (33.9%) Above/below K-M estimate 0.0031 0.00308 0.074 Median

PH Entry (left censor) Above/below 268/792 (33.8%) 0.87 1.61 1.134 0.24 <0.0001 0.025 regression End of followup Median (0.68,1.10) (1.24, 2.10) (1.037, 1.72) PH Entry (left censor) Continuous (log 1.45 3.72 1.63 268/792 (33.8%) 0.23 <0.0001 0.0003 regression End of followup value) (0.62,1.21) (0.62, 1.12) (1.04, 1.98) Entry (left censor) Continuous PH 0.77 2.72 1.60 End of followup (log value) 268/792 (33.8%) 0.10 0.0006 <0.0001 regression (0.56,1.05) (1.54, 4.80) (1.21, 2.11) CONTROLLED1 Death, End of Followup All Cancer Above/below K-M estimate 118/7742 (15.2%) 0.39 0.28 0.69 Diagnosis Median

Death, End of Followup 118/774 (15.2%) Above/below K-M estimate 0.31 0.99 0.24 Median

Age at Entry (left censor), Death, PH Above/below 1.17 1.56 0.94 End of followup 0.72 0.21 0.74

130 regression Median (0.82,1.68) (0.87,1.82) (0.65,1.35) Age at Entry (left censor), Death,

PH Continuous (log 1.19 2.60 1.29 End of followup 0.72 0.022 0.30 regression value) 118/774 (15.2%) (0.75,1.90) (1.15,5.90) (0.80, 2.07) Age at Entry (left censor), Death, Continuous PH 1.04 2.242 1.34 End of followup (log value) 0.89 0.066 0.24 regression 118/774 (15.2%) (0.64, 1.69) (0.95, 5.35) (0.83, 2.16) CONTROLLED1 Death, End of Followup 52/774 Breast Above/below Cancer K-M estimate 0.28 0.48 0.14 Median Diagnosis (6.7%) Death, End of Followup Above/below 52/774 K-M estimate 0.26 0.89 0.33 Median (6.7%) PH Age at Entry (left censor), Death, Above/below 52/774 1.33 1.32 0.69 (0.40, 0.31 0.32 0.20 regression End of followup Median (6.7%) (0.77, 2.31) (0.76,2.28) 1.22) PH Age at Entry (left censor), Death, Continuous (log 52/774 1.83 0.59 2.04 (0.98,4.23 0.055 0.02 0.30 regression End of followup value) (6.7%) (0.53,6.34 ) (0.22, 1.58) Age at Entry (left censor), Death, Continuous 52/774 PH 1.94 1.85 0.63 End of followup (log value) (6.7%) 0.093 0.37 0.37 regression (0.90,4.20) (0.48,7.12) (0.23,1.72) CONTROLLED1 1. Control variables = totp tote diabetes smoke alc3mo_3 education ethnic mvmet married income base_bmi 2. 19 participants had Cancer time event prior or equal to study entry

REFERENCES

1. Larsson L, Degens H, Li M, et al. Sarcopenia: Aging-related loss of muscle mass and function. Physiol Rev. 2019. doi:10.1152/physrev.00061.2017

2. Walston JD. Sarcopenia in older adults. Curr Opin Rheumatol. 2012. doi:10.1097/BOR.0b013e328358d59b

3. Yeung SSY, Reijnierse EM, Pham VK, et al. Sarcopenia and its association with falls and fractures in older adults: A systematic review and meta-analysis. J Cachexia Sarcopenia Muscle. 2019. doi:10.1002/jcsm.12411

4. Rong S, Wang L, Peng Z, et al. The mechanisms and treatments for sarcopenia: could exosomes be a perspective research strategy in the future? J Cachexia Sarcopenia Muscle. 2020. doi:10.1002/jcsm.12536

5. Shafiee G, Keshtkar A, Soltani A, Ahadi Z, Larijani B, Heshmat R. Prevalence of sarcopenia in the world: A systematic review and meta- analysis of general population studies. J Diabetes Metab Disord. 2017. doi:10.1186/s40200-017-0302-x

6. Beaudart C, Rizzoli R, Bruyère O, Reginster JY, Biver E. Sarcopenia: Burden and challenges for public health. Arch Public Heal. 2014. doi:10.1186/2049-3258-72-45

7. Colman RJ, Anderson RM. Nonhuman primate calorie restriction. Antioxidants Redox Signal. 2011. doi:10.1089/ars.2010.3224

8. Mercken EM, Carboneau BA, Krzysik-Walker SM, De Cabo R. Of mice and men: The benefits of caloric restriction, exercise, and mimetics. Ageing Res Rev. 2012. doi:10.1016/j.arr.2011.11.005

9. Most J, Tosti V, Redman LM, Fontana L. Calorie restriction in humans: An update. Ageing Res Rev. 2017. doi:10.1016/j.arr.2016.08.005

10. Romanello V, Sandri M. Mitochondrial quality control and muscle mass maintenance. Front Physiol. 2016. doi:10.3389/fphys.2015.00422

11. Mercken EM, Crosby SD, Lamming DW, et al. Calorie restriction in humans inhibits the PI3K/AKT pathway and induces a younger transcription profile. Aging Cell. 2013. doi:10.1111/acel.12088

12. Das SK, Roberts SB, Bhapkar M V., et al. Body-composition changes in the Comprehensive Assessment of Long-term Effects of Reducing Intake of Energy (CALERIE)-2 study: A 2-y randomized controlled trial of calorie restriction in nonobese humans. Am J Clin Nutr. 2017. doi:10.3945/ajcn.116.137232

13. van Norren K, Rusli F, van Dijk M, et al. Behavioural changes are a major contributing

131

factor in the reduction of sarcopenia in caloric-restricted ageing mice. J Cachexia Sarcopenia Muscle. 2015. doi:10.1002/jcsm.12024

14. Mattison JA, Roth GS, Mark Beasley T, et al. Impact of caloric restriction on health and survival in rhesus monkeys from the NIA study. Nature. 2012. doi:10.1038/nature11432

15. Colman RJ, Beasley TM, Allison DB, Weindruch R. Attenuation of sarcopenia by dietary restriction in rhesus monkeys. Journals Gerontol - Ser A Biol Sci Med Sci. 2008. doi:10.1093/gerona/63.6.556

16. Vitale G, Pellegrino G, Vollery M, Hofland LJ. ROLE of IGF-1 system in the modulation of longevity: Controversies and new insights from a centenarians’ perspective. Front Endocrinol (Lausanne). 2019. doi:10.3389/fendo.2019.00027

17. Sharples AP, Hughes DC, Deane CS, Saini A, Selman C, Stewart CE. Longevity and skeletal muscle mass: The role of IGF signalling, the sirtuins, dietary restriction and protein intake. Aging Cell. 2015. doi:10.1111/acel.12342

18. Dirks AJ, Leeuwenburgh C. Tumor necrosis factor α signaling in skeletal muscle: effects of age and caloric restriction. J Nutr Biochem. 2006. doi:10.1016/j.jnutbio.2005.11.002

19. Kineman RD, del Rio-Moreno M, Sarmento-Cabral A. 40 years of IGF1: Understanding the tissue-specific roles of IGF1/IGF1R in regulating metabolism using the Cre/loxP system. J Mol Endocrinol. 2018. doi:10.1530/JME-18-0076

20. Bikle DD, Tahimic C, Chang W, Wang Y, Philippou A, Barton ER. Role of IGF-I signaling in muscle bone interactions. Bone. 2015. doi:10.1016/j.bone.2015.04.036 21. Puche JE, Castilla-Cortázar I. Human conditions of insulin-like growth factor-I (IGF-I) deficiency. J Transl Med. 2012. doi:10.1186/1479-5876-10-224

22. Philippou A, Barton ER. Optimizing IGF-I for skeletal muscle therapeutics. Growth Horm IGF Res. 2014. doi:10.1016/j.ghir.2014.06.003

23. Syverud BC, VanDusen KW, Larkin LM. Growth factors for skeletal muscle tissue engineering. Cells Tissues Organs. 2016. doi:10.1159/000444671

24. Barclay RD, Burd NA, Tyler C, Tillin NA, Mackenzie RW. The Role of the IGF-1 Signaling Cascade in Muscle Protein Synthesis and Anabolic Resistance in Aging Skeletal Muscle. Front Nutr. 2019. doi:10.3389/fnut.2019.00146

25. Schiaffino S, Mammucari C. Regulation of skeletal muscle growth by the IGF1-Akt/PKB pathway: Insights from genetic models. Skelet Muscle. 2011. doi:10.1186/2044-5040-1-4

26. Yoon MS. mTOR as a key regulator in maintaining skeletal muscle mass. Front Physiol. 2017. doi:10.3389/fphys.2017.00788

132

27. Janssen JAMJL, Stolk RP, Pols HAP, Grobbee DE, De Jong FH, Lamberts SWJ. Serum free IGF-I, total IGF-I, IGFBP-1 and IGFBP-3 levels in an elderly population: Relation to age and sex steroid levels. Clin Endocrinol (Oxf). 1998. doi:10.1046/j.1365- 2265.1998.00300.x

28. Allard JB, Duan C. IGF-binding proteins: Why do they exist and why are there so many? Front Endocrinol (Lausanne). 2018. doi:10.3389/fendo.2018.00117

29. Yakar S, Adamo ML. Insulin-Like Growth Factor 1 Physiology. Lessons from Mouse Models. Endocrinol Metab Clin North Am. 2012. doi:10.1016/j.ecl.2012.04.008

30. Jin L, Shen F, Weinfeld M, Sergi C. Insulin Growth Factor Binding Protein 7 (IGFBP7)- Related Cancer and IGFBP3 and IGFBP7 Crosstalk. Front Oncol. 2020. doi:10.3389/fonc.2020.00727

31. Rajah R, Lee KW, Cohen P. Insulin-like growth factor binding protein-3 mediates tumor necrosis factor-α-induced apoptosis: Role of Bcl-2 phosphorylation. Cell Growth Differ. 2002.

32. Hwang JR, Huh JH, Lee Y, Lee S Il, Rho SB, Lee JH. Insulin-like growth factor-binding protein-5 (IGFBP-5) inhibits TNF-α-induced NF-κB activity by binding to TNFR1. Biochem Biophys Res Commun. 2011. doi:10.1016/j.bbrc.2011.01.064

33. Lee YC, Jogie-Brahim S, Lee DY, et al. Insulin-like growth factor-binding protein-3 (IGFBP-3) blocks the effects of asthma by negatively regulating NF-κB signaling through IGFBP-3R-mediated activation of caspases. J Biol Chem. 2011. doi:10.1074/jbc.M111.231035

34. Bach LA. 40 years of IGF1: IGF-binding proteins. J Mol Endocrinol. 2018. doi:10.1530/JME-17-0254

35. Janssen JAMJL, Van Der Lely AJ, Lamberts SWJ. Circulating free insulin-like growth- factor-I (IGF-I) levels should also be measured to estimate the IGF-I bioactivity. J Endocrinol Invest. 2003. doi:10.1007/BF03345225

36. Knuppel A, Fensom GK, Watts EL, et al. Circulating insulin-like growth factor-I (IGF-I) concentrations and incidence of 30 cancers: prospective analyses in UK Biobank. Cancer Res. 2020. doi:10.1158/0008-5472.can-20-1281

37. Zhang WB, Aleksic S, Gao T, et al. Insulin-like Growth Factor-1 and IGF Binding Proteins Predict All-Cause Mortality and Morbidity in Older Adults. Cells. 2020. doi:10.3390/cells9061368

38. Fontana L, Weiss EP, Villareal DT, Klein S, Holloszy JO. Long-term effects of calorie or protein restriction on serum IGF-1 and IGFBP-3 concentration in humans. Aging Cell. 2008. doi:10.1111/j.1474-9726.2008.00417.x

133

39. Balasubramanian P, Howell PR, Anderson RM. Aging and Caloric Restriction Research: A Biological Perspective With Translational Potential. EBioMedicine. 2017. doi:10.1016/j.ebiom.2017.06.015

40. Fontana L, Villareal DT, Das SK, et al. Effects of 2-year calorie restriction on circulating levels of IGF-1, IGF-binding proteins and cortisol in nonobese men and women: A randomized clinical trial. Aging Cell. 2016. doi:10.1111/acel.12400

41. Harrison DE, Strong R, Sharp ZD, et al. Rapamycin fed late in life extends lifespan in genetically heterogeneous mice. Nature. 2009. doi:10.1038/nature08221

42. Rivas DA, Fielding RA. Skeletal Muscle. Encycl Hum Nutr. 2012. doi:10.1016/B978-0- 12-375083-9.00188-4

43. Volpi E, Nazemi R, Fujita S. Muscle tissue changes with aging. Curr Opin Clin Nutr Metab Care. 2004. doi:10.1097/01.mco.0000134362.76653.b2

44. Evans WJ, Lexell J. Human Aging, Muscle Mass, and Fiber Type Composition. Journals Gerontol Ser A Biol Sci Med Sci. 1995. doi:10.1093/gerona/50a.special_issue.11

45. Dhillon RJS, Hasni S. Pathogenesis and Management of Sarcopenia. Clin Geriatr Med. 2017. doi:10.1016/j.cger.2016.08.002

46. Chen CNJ, Lin SY, Liao YH, Li ZJ, Wong AMK. Late-onset caloric restriction alters skeletal muscle metabolism by modulating pyruvate metabolism. Am J Physiol - Endocrinol Metab. 2015. doi:10.1152/ajpendo.00508.2014

47. Talbot J, Maves L. Skeletal muscle fiber type: using insights from muscle developmental biology to dissect targets for susceptibility and resistance to muscle disease. Wiley Interdiscip Rev Dev Biol. 2016. doi:10.1002/wdev.230

48. Mukund K, Subramaniam S. Skeletal muscle: A review of molecular structure and function, in health and disease. Wiley Interdiscip Rev Syst Biol Med. 2020. doi:10.1002/wsbm.1462

49. Doria E, Buonocore D, Focarelli A, Marzatico F. Relationship between human aging muscle and oxidative system pathway. Oxid Med Cell Longev. 2012. doi:10.1155/2012/830257

50. Fielding RA, Vellas B, Evans WJ, et al. Sarcopenia: An Undiagnosed Condition in Older Adults. Current Consensus Definition: Prevalence, Etiology, and Consequences. International Working Group on Sarcopenia. J Am Med Dir Assoc. 2011. doi:10.1016/j.jamda.2011.01.003

51. Goodpaster BH, Park SW, Harris TB, et al. The loss of skeletal muscle strength, mass, and

134

quality in older adults: The Health, Aging and Body Composition Study. Journals Gerontol - Ser A Biol Sci Med Sci. 2006. doi:10.1093/gerona/61.10.1059

52. Phillips SM. Nutritional Supplements in Support of Resistance Exercise to Counter Age- Related Sarcopenia. Adv Nutr. 2015. doi:10.3945/an.115.008367

53. Redman LM, Ravussin E. Caloric restriction in humans: Impact on physiological, psychological, and behavioral outcomes. Antioxidants Redox Signal. 2011. doi:10.1089/ars.2010.3253

54. Fontana L, Partridge L, Longo VD. Dietary Restriction, Growth Factors and Aging: from yeast to humans. Science (80- ). 2010. doi:10.1126/science.1172539

55. Das SK, Balasubramanian P, Weerasekara YK. Nutrition modulation of human aging: The calorie restriction paradigm. Mol Cell Endocrinol. 2017. doi:10.1016/j.mce.2017.04.011

56. McCay CM, Crowell MF, Maynard LA. The effect of retarded growth upon the length of life span and upon the ultimate body size. 1935. Nutrition. 1989.

57. Kapahi P, Kaeberlein M, Hansen M. Dietary restriction and lifespan: Lessons from invertebrate models. Ageing Res Rev. 2017. doi:10.1016/j.arr.2016.12.005

58. Picca A, Pesce V, Lezza AMS. Does eating less make you live longer and better? An update on calorie restriction. Clin Interv Aging. 2017. doi:10.2147/CIA.S126458

59. Hursting SD, Slaga TJ, Fischer SM, DiGiovanni J, Phang JM. Mechanism-based cancer prevention approaches: Targets, examples, and the use of transgenic mice. J Natl Cancer Inst. 1999. doi:10.1093/jnci/91.3.215

60. Longo VD, Fontana L. Calorie restriction and cancer prevention: metabolic and molecular mechanisms. Trends Pharmacol Sci. 2010. doi:10.1016/j.tips.2009.11.004

61. Masoro EJ, Yu BP, Bertrand HA. Action of food restriction in delaying the aging process. Proc Natl Acad Sci U S A. 1982. doi:10.1073/pnas.79.13.4239

62. Anderson RM, Weindruch R. Metabolic reprogramming, caloric restriction and aging. Trends Endocrinol Metab. 2010. doi:10.1016/j.tem.2009.11.005

63. Gibbs RA, Rogers J, Katze MG, et al. Evolutionary and biomedical insights from the rhesus macaque genome. Science (80- ). 2007. doi:10.1126/science.1139247

64. Phillips KA, Bales KL, Capitanio JP, et al. Why primate models matter. Am J Primatol. 2014. doi:10.1002/ajp.22281

65. Ingram DK, Cutler RG, Weindruch R, et al. Dietary restriction and aging: The initiation of a primate study. Journals Gerontol. 1990. doi:10.1093/geronj/45.5.B148

135

66. Kemnitz JW, Weindruch R, Roecker EB, Crawford K, Kaufman PL, Ershler WB. Dietary restriction of adult male rhesus monkeys: Design, methodology, and preliminary findings from the first year of study. Journals Gerontol. 1993. doi:10.1093/geronj/48.1.B17

67. Mattison JA, Colman RJ, Beasley TM, et al. Caloric restriction improves health and survival of rhesus monkeys. Nat Commun. 2017. doi:10.1038/ncomms14063

68. Kemnitz JW. Calorie Restriction and Aging in Nonhuman Primates. ILAR J. 2011. doi:10.1093/ilar.52.1.66

69. Cava E, Fontana L. Will calorie restriction work in humans? Aging (Albany NY). 2013. doi:10.18632/aging.100581

70. Edwards IJ, Rudel LL, Terry JG, et al. Caloric restriction lowers plasma lipoprotein (a) in male but not female rhesus monkeys. Exp Gerontol. 2001. doi:10.1016/S0531- 5565(01)00107-3

71. Gresl TA, Colman RJ, Roecker EB, et al. Dietary restriction and glucose regulation in aging rhesus monkeys: A follow-up report at 8.5 yr. Am J Physiol - Endocrinol Metab. 2001. doi:10.1152/ajpendo.2001.281.4.e757

72. LANE MA, INGRAM DK, CUTLER RG, KNAPKA JJ, BARNARD DE, ROTH GS. Dietary Restriction in Nonhuman Primates: Progress Report on the NIA Study. Ann N Y Acad Sci. 1992. doi:10.1111/j.1749-6632.1992.tb27434.x

73. Lane MA, Ingram DK, Roth GS. Calorie restriction in nonhuman primates: Effects on diabetes and cardiovascular disease risk. Toxicol Sci. 1999. doi:10.1093/toxsci/52.2.41

74. Mattison JA, Lane MA, Roth GS, Ingram DK. Calorie restriction in rhesus monkeys. Exp Gerontol. 2003. doi:10.1016/S0531-5565(02)00146-8

75. Ramsey JJ, Harper ME, Weindruch R. Restriction of energy intake, energy expenditure, and aging. Free Radic Biol Med. 2000. doi:10.1016/S0891-5849(00)00417-2

76. Roth GS, Ingram DK, Lane MA. Caloric restriction in primates and relevance to humans. Ann N Y Acad Sci. 2001. doi:10.1111/j.1749-6632.2001.tb05660.x

77. Roth GS, Lane MA, Ingram DK, et al. Biomarkers of caloric restriction may predict longevity in humans. Science (80- ). 2002. doi:10.1126/science.1071851

78. Anderson RM, Weindruch R. The caloric restriction paradigm: Implications for healthy human aging. Am J Hum Biol. 2012. doi:10.1002/ajhb.22243

79. Colman RJ, Ramsey JJ, Roecker EB, Havighurst T, Hudson JC, Kemnitz JW. Body fat distribution with long-term dietary restriction in adult male rhesus macaques. Journals

136

Gerontol - Ser A Biol Sci Med Sci. 1999. doi:10.1093/gerona/54.7.B283

80. Colman RJ, Beasley TM, Kemnitz JW, Johnson SC, Weindruch R, Anderson RM. Caloric restriction reduces age-related and all-cause mortality in rhesus monkeys. Nat Commun. 2014. doi:10.1038/ncomms4557

81. Colman RJ, Anderson RM, Johnson SC, et al. Caloric restriction delays disease onset and mortality in rhesus monkeys. Science (80- ). 2009. doi:10.1126/science.1173635

82. Colman RJ, Anderson RM. Nonhuman primate calorie restriction. Antioxidants Redox Signal. 2011. doi:10.1089/ars.2010.3224

83. Colman RJ, Roecker EB, Ramsey JJ, Kemnitz JW. The effect of dietary restriction on body composition in adult male and female rhesus macaques. Aging Clin Exp Res. 1998. doi:10.1007/bf03339642

84. Edwards IJ, Rudel LL, Terry JG, Kemnitz JW, Weindruch R, Cefalu WT. Caloric restriction in rhesus monkeys reduces low density lipoprotein interaction with arterial proteoglycans. Journals Gerontol - Ser A Biol Sci Med Sci. 1998. doi:10.1093/gerona/53A.6.B443

85. Rickman AD, Williamson DA, Martin CK, et al. The CALERIE Study: Design and methods of an innovative 25% caloric restriction intervention. Contemp Clin Trials. 2011. doi:10.1016/j.cct.2011.07.002

86. Rochon J, Bales CW, Ravussin E, et al. Design and conduct of the CALERIE study: Comprehensive Assessment of the Long-term Effects of Reducing Intake of Energy. Journals Gerontol - Ser A Biol Sci Med Sci. 2011. doi:10.1093/gerona/glq168

87. Ravussin E, Redman LM, Rochon J, et al. A 2-year randomized controlled trial of human caloric restriction: Feasibility and effects on predictors of health span and longevity. Journals Gerontol - Ser A Biol Sci Med Sci. 2015. doi:10.1093/gerona/glv057

88. Kraus WE, Bhapkar M, Huffman KM, et al. 2 years of calorie restriction and cardiometabolic risk (CALERIE): exploratory outcomes of a multicentre, phase 2, randomised controlled trial. Lancet Diabetes Endocrinol. 2019. doi:10.1016/S2213- 8587(19)30151-2

89. Barger JL, Walford RL, Weindruch R. The retardation of aging by caloric restriction: Its significance in the transgenic era. Exp Gerontol. 2003. doi:10.1016/j.exger.2003.10.017

90. Mair W, Dillin A. Aging and Survival: The Genetics of Life Span Extension by Dietary Restriction. Annu Rev Biochem. 2008. doi:10.1146/annurev.biochem.77.061206.171059

91. Taormina G, Mirisola MG. Calorie restriction in mammals and simple model organisms. Biomed Res Int. 2014. doi:10.1155/2014/308690

137

92. Speakman JR, Selman C. The free-radical damage theory: Accumulating evidence against a simple link of oxidative stress to ageing and lifespan. BioEssays. 2011. doi:10.1002/bies.201000132

93. Selman C, Phillips T, Staib JL, Duncan JS, Leeuwenburgh C, Speakman JR. Energy expenditure of calorically restricted rats is higher than predicted from their altered body composition. Mech Ageing Dev. 2005. doi:10.1016/j.mad.2005.02.004

94. Hunter GR, Singh H, Carter SJ, Bryan DR, Fisher G. Sarcopenia and Its Implications for Metabolic Health. J Obes. 2019. doi:10.1155/2019/8031705

95. Srikanthan P, Karlamangla AS. Muscle mass index as a predictor of longevity in older adults. Am J Med. 2014. doi:10.1016/j.amjmed.2014.02.007

96. Rhoads TW, Clark JP, Gustafson GE, et al. Molecular and Functional Networks Linked to Sarcopenia Prevention by Caloric Restriction in Rhesus Monkeys. Cell Syst. 2020. doi:10.1016/j.cels.2019.12.002

97. Yamada Y, Kemnitz JW, Weindruch R, Anderson RM, Schoeller DA, Colman RJ. Caloric Restriction and Healthy Life Span: Frail Phenotype of Nonhuman Primates in the Wisconsin National Primate Research Center Caloric Restriction Study. Journals Gerontol - Ser A Biol Sci Med Sci. 2018. doi:10.1093/gerona/glx059

98. Racette SB, Rochon J, Uhrich ML, et al. Effects of Two Years of Calorie Restriction on Aerobic Capacity and Muscle Strength. Med Sci Sports Exerc. 2017. doi:10.1249/MSS.0000000000001353

99. Yang L, Licastro D, Cava E, et al. Long-Term Calorie Restriction Enhances Cellular Quality-Control Processes in Human Skeletal Muscle. Cell Rep. 2016. doi:10.1016/j.celrep.2015.12.042

100. Brisson BK, Barton ER. New modulators for IGF-I activity within IGF-I processing products. Front Endocrinol (Lausanne). 2013. doi:10.3389/fendo.2013.00042 101. Frost RA, Lang CH. Multifaceted Role of Insulin-Like Growth Factors and Mammalian Target of Rapamycin in Skeletal Muscle. Endocrinol Metab Clin North Am. 2012;41(2):297-322. doi:10.1016/j.ecl.2012.04.012

102. Fuentes EN, Björnsson BT, Valdés JA, et al. IGF-i/PI3K/AKT and IGF-I/MAPK/ERK pathways in vivo in skeletal muscle are regulated by nutrition and contribute to somatic growth in the fine flounder. Am J Physiol - Regul Integr Comp Physiol. 2011. doi:10.1152/ajpregu.00535.2010.

103. Musarò A, McCullagh KJA, Naya FJ, Olson EN, Rosenthal N. IGF-1 induces skeletal myocyte hypertrophy through calcineurin in association with GATA-2 and NF-ATc1. Nature. 1999. doi:10.1038/23060

138

104. Rommel C, Bodine SC, Clarke BA, et al. Mediation of IGF-1-induced skeletal myotube hypertrophy by Pl(3)K/Alt/mTOR and Pl(3)K/Akt/GSK3 pathways. Nat Cell Biol. 2001. doi:10.1038/ncb1101-1009

105. Rommel C, Clarke BA, Zimmermann S, et al. Differentiation stage-specific inhibition of the Raf-MEK-ERK pathway by Akt. Science (80- ). 1999. doi:10.1126/science.286.5445.1738

106. Le Roith D, Bondy C, Yakar S, Liu JL, Butler A. The somatomedin hypothesis: 2001. Endocr Rev. 2001. doi:10.1210/edrv.22.1.0419

107. Coolican SA, Samuel DS, Ewton DZ, McWade FJ, Florini JR. The mitogenic and myogenic actions of insulin-like growth factors utilize distinct signaling pathways. J Biol Chem. 1997. doi:10.1074/jbc.272.10.6653

108. Glass DJ. Molecular mechanisms modulating muscle mass. Trends Mol Med. 2003. doi:10.1016/S1471-4914(03)00138-2

109. Jones NC, Fedorov Y V., Rosenthal RS, Olwin BB. ERK1/2 is required for myoblast proliferation but is dispensable for muscle gene expression and cell fusion. J Cell Physiol. 2001. doi:10.1002/1097-4652(200101)186:1<104::AID-JCP1015>3.0.CO;2-0

110. Wilkinson DJ, Piasecki M, Atherton PJ. The age-related loss of skeletal muscle mass and function: Measurement and physiology of muscle fibre atrophy and muscle fibre loss in humans. Ageing Res Rev. 2018. doi:10.1016/j.arr.2018.07.005

111. Dalle S, Rossmeislova L, Koppo K. The role of inflammation in age-related sarcopenia. Front Physiol. 2017. doi:10.3389/fphys.2017.01045

112. Brahmkhatri VP, Prasanna C, Atreya HS. Insulin-like growth factor system in cancer: Novel targeted therapies. Biomed Res Int. 2015. doi:10.1155/2015/538019

113. Ryan PD, Goss PE. The Emerging Role of the Insulin-Like Growth Factor Pathway as a Therapeutic Target in Cancer. Oncologist. 2008. doi:10.1634/theoncologist.2007-0199

114. Mohan S 1 BD. IGF-binding proteins are multifunctional and act via IGF-dependent and - independent mechanisms. Eur PMC. 2002;175(1):19-31.

115. Baxter RC. Insulin-like growth factor (IGF)-binding proteins: Interactions with IGFs and intrinsic bioactivities. Am J Physiol - Endocrinol Metab. 2000.

116. Liao Y, Lei J, Liu M, et al. Mesenchymal stromal cells mitigate experimental colitis via insulin-like growth factor binding protein 7-mediated immunosuppression. Mol Ther. 2016. doi:10.1038/mt.2016.140

139

117. Madeo F, Carmona-Gutierrez D, Hofer SJ, Kroemer G. Caloric Restriction Mimetics against Age-Associated Disease: Targets, Mechanisms, and Therapeutic Potential. Cell Metab. 2019. doi:10.1016/j.cmet.2019.01.018

118. Blagosklonny M V. Rapamycin for longevity: Opinion article. Aging (Albany NY). 2019. doi:10.18632/aging.102355

119. Lamming DW. Inhibition of the mechanistic target of rapamycin (mTOR)–Rapamycin and beyond. Cold Spring Harb Perspect Med. 2016. doi:10.1101/cshperspect.a025924

120. Partridge L, Fuentealba M, Kennedy BK. The quest to slow ageing through drug discovery. Nat Rev Drug Discov. 2020. doi:10.1038/s41573-020-0067-7

121. Bové J, Martínez-Vicente M, Vila M. Fighting neurodegeneration with rapamycin: Mechanistic insights. Nat Rev Neurosci. 2011. doi:10.1038/nrn3068

122. Gu J, Hu W, Song ZP, Chen YG, Zhang DD, Wang CQ. Rapamycin inhibits cardiac hypertrophy by promoting autophagy via the MEK/ERK/Beclin-1 pathway. Front Physiol. 2016. doi:10.3389/fphys.2016.00104

123. Guertin DA, Sabatini DM. Defining the Role of mTOR in Cancer. Cancer Cell. 2007. doi:10.1016/j.ccr.2007.05.008

124. Xie J, Wang X, Proud CG. mTOR inhibitors in cancer therapy. F1000Research. 2016. doi:10.12688/f1000research.9207.1

125. Ye L, Widlund AL, Sims CA, et al. Rapamycin doses sufficient to extend lifespan do not compromise muscle mitochondrial content or endurance. Aging (Albany NY). 2013. doi:10.18632/aging.100576

126. Tang H, Inoki K, Brooks S V., et al. mTORC1 underlies age-related muscle fiber damage and loss by inducing oxidative stress and catabolism. Aging Cell. 2019. doi:10.1111/acel.12943

127. Tang, Huibin Tang Joseph B. Shrager, 1 2 and Daniel Goldmancorresponding author. Rapamycin protects aging muscle. Aging (Albany NY). 2019;11(16):5868-5870. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6738401/pdf/aging-11-102176.pdf.

128. Santilli V, Bernetti A, Mangone M, Paoloni M. Clinical definition of sarcopenia. Clin Cases Miner Bone Metab. 2014. doi:10.11138/ccmbm/2014.11.3.177

129. Ortman JM, Velkoff V a., Hogan H. An aging nation: The older population in the United States. Econ Stat Adm US Dep Commer. 2014.

130. Ferrucci L, Fabbri E. Inflammageing: chronic inflammation in ageing, cardiovascular disease, and frailty. Nat Rev Cardiol. 2018. doi:10.1038/s41569-018-0064-2

140

131. Fan J, Kou X, Yang Y, Chen N. MicroRNA-regulated proinflammatory cytokines in sarcopenia. Mediators Inflamm. 2016. doi:10.1155/2016/1438686

132. Song YH, Song JL, Delafontaine P, Godard MP. The therapeutic potential of IGF-I in skeletal muscle repair. Trends Endocrinol Metab. 2013. doi:10.1016/j.tem.2013.03.004

133. Bian A, Ma Y, Zhou X, et al. Association between sarcopenia and levels of growth hormone and insulin-like growth factor-1 in the elderly. BMC Musculoskelet Disord. 2020. doi:10.1186/s12891-020-03236-y

134. Nair KS. Aging muscle. Am J Clin Nutr. 2005. doi:10.1093/ajcn/81.5.953

135. World Health Organisation. Why Population Ageing Matters to Health: A Global Perspective. Glob Heal Aging. 2011.

136. Marzetti E, Calvani R, Bernabei R, Leeuwenburgh C. Apoptosis in skeletal myocytes: A potential target for interventions against sarcopenia and physical frailty - A mini-review. Gerontology. 2012. doi:10.1159/000330064

137. Brook MS, Wilkinson DJ, Phillips BE, et al. Skeletal muscle homeostasis and plasticity in youth and ageing: Impact of nutrition and exercise. Acta Physiol. 2016. doi:10.1111/apha.12532

138. López-Lluch G, Navas P. Calorie restriction as an intervention in ageing. J Physiol. 2016. doi:10.1113/JP270543

139. Faitg J, Leduc-Gaudet JP, Reynaud O, Ferland G, Gaudreau P, Gouspillou G. Effects of aging and caloric restriction on fiber type composition, mitochondrial morphology and dynamics in rat oxidative and glycolytic muscles. Front Physiol. 2019. doi:10.3389/fphys.2019.00420

140. Sohal RS, Forster MJ. Caloric restriction and the aging process: A critique. Free Radic Biol Med. 2014. doi:10.1016/j.freeradbiomed.2014.05.015

141. Baumgartner RN, Koehler KM, Gallagher D, et al. Epidemiology of sarcopenia among the elderly in New Mexico. Am J Epidemiol. 1998. doi:10.1093/oxfordjournals.aje.a009520

142. Swindell WR. Genes and gene expression modules associated with caloric restriction and aging in the laboratory mouse. BMC Genomics. 2009. doi:10.1186/1471-2164-10-585

143. Committee for Proprietary Medicinal Products (CPMP): points to consider on adjustment for baseline covariates. Stat Med. 2004. doi:10.1002/sim.1647

144. Gene Set Enrichment Analysis. https://www.gsea-msigdb.org/gsea/msigdb/collections.jsp.

141

145. Liang Y, Liu C, Lu M, et al. Calorie restriction is the most reasonable anti-ageing intervention: A meta-analysis of survival curves. Sci Rep. 2018. doi:10.1038/s41598-018- 24146-z

146. Webster JM, Fenton CG, Langen R, Hardy RS. Exploring the interface between inflammatory and therapeutic glucocorticoid induced bone and muscle loss. Int J Mol Sci. 2019. doi:10.3390/ijms20225768

147. Montano M. Translational models, methods and concepts in studies of aging and longevity. Transl Biol Med. 2014. doi:10.1533/9781908818652.75

148. Masternak MM, Al-Regaiey KA, Lim MMDR, et al. Effects of caloric restriction on insulin pathway gene expression in the skeletal muscle and liver of normal and long-lived GHR-KO mice. Exp Gerontol. 2005. doi:10.1016/j.exger.2005.06.003

149. Kim JH, Kwak HB, Leeuwenburgh C, Lawler JM. Lifelong exercise and mild (8%) caloric restriction attenuate age-induced alterations in plantaris muscle morphology, oxidative stress and IGF-1 in the Fischer-344 rat. Exp Gerontol. 2008. doi:10.1016/j.exger.2007.12.012

150. Eng CM, Azizi E, Roberts TJ. Structural determinants of muscle gearing during dynamic contractions. Integr Comp Biol. 2018. doi:10.1093/icb/icy054

151. Ruetenik A, Barrientos A. Dietary restriction, mitochondrial function and aging: From yeast to humans. Biochim Biophys Acta - Bioenerg. 2015. doi:10.1016/j.bbabio.2015.05.005

152. López-Lluch G, Hunt N, Jones B, et al. Calorie restriction induces mitochondrial biogenesis and bioenergetic efficiency. Proc Natl Acad Sci U S A. 2006. doi:10.1073/pnas.0510452103

153. Omiecinski CJ, Vanden Heuvel JP, Perdew GH, Peters JM. Xenobiotic metabolism, disposition, and regulation by receptors: From biochemical phenomenon to predictors of major toxicities. Toxicol Sci. 2011. doi:10.1093/toxsci/kfq338

154. Steinbaugh MJ, Sun LY, Bartke A, Miller RA. Activation of genes involved in xenobiotic metabolism is a shared signature of mouse models with extended lifespan. Am J Physiol - Endocrinol Metab. 2012. doi:10.1152/ajpendo.00110.2012

155. Johnson SC, Rabinovitch PS, Kaeberlein M. MTOR is a key modulator of ageing and age- related disease. Nature. 2013. doi:10.1038/nature11861

156. Schinaman JM, Rana A, Ja WW, Clark RI, Walker DW. Rapamycin modulates tissue aging and lifespan independently of the gut microbiota in Drosophila. Sci Rep. 2019. doi:10.1038/s41598-019-44106-5

142

157. Boldrin L, Ross JA, Whitmore C, et al. The effect of calorie restriction on mouse skeletal muscle is sex, strain and time-dependent. Sci Rep. 2017. doi:10.1038/s41598-017-04896-y

158. Pletcher SD, Macdonald SJ, Marguerie R, et al. Genome-wide transcript profiles in aging and calorically restricted Drosophila melanogaster. Curr Biol. 2002. doi:10.1016/S0960- 9822(02)00808-4

159. Park SK, Prolla TA. Gene expression profiling studies of aging in cardiac and skeletal muscles. Cardiovasc Res. 2005. doi:10.1016/j.cardiores.2005.01.005

160. Wang BT, Ducker GS, Barczak AJ, Barbeau R, Erle DJ, Shokat KM. The mammalian target of rapamycin regulates cholesterol biosynthetic gene expression and exhibits a rapamycin-resistant transcriptional profile. Proc Natl Acad Sci U S A. 2011. doi:10.1073/pnas.1103746108

161. Powell JD, Pollizzi KN, Heikamp EB, Horton MR. Regulation of immune responses by mTOR. Annu Rev Immunol. 2012. doi:10.1146/annurev-immunol-020711-075024

162. Ortman JM, Velkoff VA, Hogan H. An Aging Nation: The Older Population in the United States Current Population Reports.; 2014.

163. Goodpaster BH, Park SW, Harris TB, et al. The loss of skeletal muscle strength, mass, and quality in older adults: The Health, Aging and Body Composition Study. Journals Gerontol - Ser A Biol Sci Med Sci. 2006. doi:10.1093/gerona/61.10.1059

164. Frontera WR, Ochala J. Skeletal Muscle: A Brief Review of Structure and Function. Behav Genet. 2015. doi:10.1007/s00223-014-9915-y

165. S. Fry C, B. Rasmussen B. Skeletal Muscle Protein Balance and Metabolism in the Elderly. Curr Aging Sci. 2012. doi:10.2174/1874609811104030260

166. McCarthy JJ, Esser KA. Anabolic and catabolic pathways regulating skeletal muscle mass. Curr Opin Clin Nutr Metab Care. 2010. doi:10.1097/MCO.0b013e32833781b5

167. Schiaffino S, Dyar KA, Ciciliot S, Blaauw B, Sandri M. Mechanisms regulating skeletal muscle growth and atrophy. FEBS J. 2013. doi:10.1111/febs.12253

168. Velloso CP. Regulation of muscle mass by growth hormone and IGF-I. Br J Pharmacol. 2008. doi:10.1038/bjp.2008.153

169. Marzetti E, Lees HA, Wohlgemuth SE, Leeuwenburgh C. Sarcopenia of aging: Underlying cellular mechanisms and protection by calorie restriction. BioFactors. 2009. doi:10.1002/biof.5

170. Bowers LW, Rossi EL, O’Flanagan CH, De Graffenried LA, Hursting SD. The role of the insulin/IGF system in cancer: Lessons learned from clinical trials and the energy balance-

143

cancer link. Front Endocrinol (Lausanne). 2015. doi:10.3389/fendo.2015.00077

171. Philippou A, Barton ER. Optimizing IGF-I for skeletal muscle therapeutics. Growth Horm IGF Res. 2014. doi:10.1016/j.ghir.2014.06.003

172. Jaynes JB, Johnson JE, Buskin JN, Gartside CL, Hauschka SD. The muscle creatine kinase gene is regulated by multiple upstream elements, including a muscle-specific enhancer. Mol Cell Biol. 1988. doi:10.1128/mcb.8.1.62

173. Ostrovsky O, Eletto D, Makarewich C, Barton ER, Argon Y. Glucose regulated protein 94 is required for muscle differentiation through its control of the autocrine production of insulin-like growth factors. Biochim Biophys Acta - Mol Cell Res. 2010. doi:10.1016/j.bbamcr.2009.11.005

174. Iresjö BM, Svensson J, Ohlsson C, Lundholm K. Liver-derived endocrine IGF-I is not critical for activation of skeletal muscle protein synthesis following oral feeding. BMC Physiol. 2013. doi:10.1186/1472-6793-13-7

175. Sjögren K, Liu JL, Blad K, et al. Liver-derived insulin-like growth factor I (IGF-I) is the principal source of IGF-I in blood but is not required for postnatal body growth in mice. Proc Natl Acad Sci U S A. 1999. doi:10.1073/pnas.96.12.7088

176. Metzger F, Sajid W, Saenger S, et al. Separation of fast from slow anabolism by site- specific PEGylation of insulin-like growth factor I (IGF-I). J Biol Chem. 2011. doi:10.1074/jbc.M110.172189

177. Krieger F, Elflein N, Saenger S, et al. Polyethylene glycol-coupled IGF1 delays motor function defects in a mouse model of spinal muscular atrophy with respiratory distress type 1. Brain. 2014. doi:10.1093/brain/awu059

178. Veron J, Council NR. Preparing for an Aging World: The Case for Cross-National Research. Popul (French Ed. 2001. doi:10.2307/1534892

179. Parker K, Berretta A, Saenger S, et al. PEGylated insulin-like growth factor-I affords protection and facilitates recovery of lost functions post-focal ischemia. Sci Rep. 2017. doi:10.1038/s41598-017-00336-z

180. Smith NJG, Caldwell JL, van der Merwe M, et al. A comparison of dietary and caloric restriction models on body composition, physical performance, and metabolic health in young mice. Nutrients. 2019. doi:10.3390/nu11020350

181. Hakuno F, Takahashi SI. 40 years of IGF1: IGF1 receptor signaling pathways. J Mol Endocrinol. 2018. doi:10.1530/JME-17-0311

182. Clemmons DR. Metabolic Actions of Insulin-Like Growth Factor-I in Normal Physiology and Diabetes. Endocrinol Metab Clin North Am. 2012. doi:10.1016/j.ecl.2012.04.017

144

183. Sjögren K, Wallenius K, Liu JL, et al. Liver-Derived IGF-I is of Importance for Normal Carbohydrate and Lipid Metabolism. Diabetes. 2001. doi:10.2337/diabetes.50.7.1539

184. Key TJ, Appleby PN, Reeves GK, et al. Insulin-like growth factor 1 (IGF1), IGF binding protein 3 (IGFBP3), and breast cancer risk: Pooled individual data analysis of 17 prospective studies. Lancet Oncol. 2010. doi:10.1016/S1470-2045(10)70095-4

185. Evdokimova V, Tognon CE, Benatar T, et al. IGFBP7 binds to the IGF-1 receptor and blocks its activation by insulin-like growth factors. Sci Signal. 2012. doi:10.1126/scisignal.2003184

186. Roubenoff R, Parise H, Payette HA, et al. Cytokines, insulin-like growth factor 1, sarcopenia, and mortality in very old community-dwelling men and women: The Framingham Heart Study. Am J Med. 2003. doi:10.1016/j.amjmed.2003.05.001

187. Nogueira LM, Lavigne JA, Chandramouli GVR, Lui H, Barrett JC, Hursting SD. Dose- dependent effects of calorie restriction on gene expression, metabolism, and tumor progression are partially mediated by insulin-like growth factor-1. Cancer Med. 2012. doi:10.1002/cam4.23

188. Laura W. Bowers, 1 Emily L. Rossi, 1 Ciara H. O’Flanagan, 1 Linda A. deGraffenried 2 and Stephen D. Hursting1. The Role of the Insulin/IGF System in Cancer: Lessons Learned from Clinical Trials and the Energy Balance-Cancer Link. Front Endocrinol (Lausanne). 2015;77(6). 189. Rosenzweig SA, Atreya HS. Defining the pathway to insulin-like growth factor system targeting in cancer. Biochem Pharmacol. 2010. doi:10.1016/j.bcp.2010.06.013 190. Weroha SJ, Haluska P. The Insulin-Like Growth Factor System in Cancer. Endocrinol Metab Clin North Am. 2012. doi:10.1016/j.ecl.2012.04.014 191. Firth SM, Baxter RC. Cellular actions of the insulin-like growth factor binding proteins. Endocr Rev. 2002. doi:10.1210/er.2001-0033 192. Evdokimova V, Tognon CE, Benatar T, et al. IGFBP7 binds to the IGF-1 receptor and blocks its activation by insulin-like growth factors. Sci Signal. 2012. doi:10.1126/scisignal.2003184 193. Ruan W, Xu E, Xu F, et al. IGFBP7 plays a potential tumor suppressor role in colorectal carcinogenesis. Cancer Biol Ther. 2007. doi:10.4161/cbt.6.3.3702 194. Akiel M, Guo C, Li X, et al. IGFBP7 deletion promotes hepatocellular carcinoma. Cancer Res. 2017. doi:10.1158/0008-5472.CAN-16-2885 195. Chen D, Yoo BK, Santhekadur PK, et al. Insulin-like growth factor-binding protein-7 functions as a potential tumor suppressor in hepatocellular carcinoma. Clin Cancer Res. 2011. doi:10.1158/1078-0432.CCR-10-2774 196. Gambaro K, Quinn MCJ, Cáceres-Gorriti KY, et al. Low levels of IGFBP7 expression in

145

high-grade serous ovarian carcinoma is associated with patient outcome. BMC Cancer. 2015. doi:10.1186/s12885-015-1138-8 197. Jiang W, Xiang C, Cazacu S, Brodie C, Mikkelsen T. Insulin-like growth factor binding protein 7 mediates glioma cell growth and migration. Neoplasia. 2008. doi:10.1593/neo.08694 198. Degeorges A, Wang F, Frierson HF, Seth A, Chung LWK, Sikes RA. Human prostate cancer expresses the low affinity insulin-like growth factor binding protein IGFBP-rP1. Cancer Res. 1999. 199. Rupp C, Scherzer M, Rudisch A, et al. IGFBP7, a novel tumor stroma marker, with growth-promoting effects in colon cancer through a paracrine tumor-stroma interaction. Oncogene. 2015. doi:10.1038/onc.2014.18 200. Sato Y, Inokuchi M, Takagi Y, et al. Relationship between expression of IGFBP7 and clinicopathological variables in gastric cancer. J Clin Pathol. 2015. doi:10.1136/jclinpath- 2015-202987 201. Smith E, Ruszkiewicz AR, Jamieson GG, Drew PA. IGFBP7 is associated with poor prognosis in oesophageal adenocarcinoma and is regulated by promoter DNA methylation. Br J Cancer. 2014. doi:10.1038/bjc.2013.783 202. Barroso MC, Kramer F, Greene SJ, et al. Serum insulin-like growth factor-1 and its binding protein-7: Potential novel biomarkers for heart failure with preserved ejection fraction. BMC Cardiovasc Disord. 2016. doi:10.1186/s12872-016-0376-2 203. Hage C, Bjerre M, Frystyk J, et al. Comparison of Prognostic Usefulness of Serum Insulin-Like Growth Factor-Binding Protein 7 in Patients With Heart Failure and Preserved Versus Reduced Left Ventricular Ejection Fraction. Am J Cardiol. 2018. doi:10.1016/j.amjcard.2018.02.041 204. Langer RD, White E, Lewis CE, Kotchen JM, Hendrix SL, Trevisan M. The women’s health initiative observational study: Baseline characteristics of participants and reliability of baseline measures. Ann Epidemiol. 2003. doi:10.1016/S1047-2797(03)00047-4 205. Jung SY, Hursting SD, Guindani M, Vitolins MZ, Paskett E, Chang S. Bioavailable insulin-like growth factor-i inversely related to weight gain in postmenopausal women regardless of exogenous estrogen. Cancer Epidemiol Biomarkers Prev. 2014. doi:10.1158/1055-9965.EPI-13-1053 206. Jung SY, Hursting SD, Guindani M, Vitolins MZ, Paskett E, Chang S. Bioavailable insulin-like growth factor-i inversely related to weight gain in postmenopausal women regardless of exogenous estrogen. Cancer Epidemiol Biomarkers Prev. 2014. doi:10.1158/1055-9965.EPI-13-1053 207. Janssen JAMJL, Stolk RP, Pols HAP, Grobbee DE, De Jong FH, Lamberts SWJ. Serum free IGF-I, total IGF-I, IGFBP-1 and IGFBP-3 levels in an elderly population: Relation to age and sex steroid levels. Clin Endocrinol (Oxf). 1998. doi:10.1046/j.1365- 2265.1998.00300.x 208. Kineman RD, del Rio-Moreno M, Sarmento-Cabral A. 40 years of IGF1: Understanding

146

the tissue-specific roles of IGF1/IGF1R in regulating metabolism using the Cre/loxP system. J Mol Endocrinol. 2018. doi:10.1530/JME-18-0076 209. Allard JB, Duan C. IGF-binding proteins: Why do they exist and why are there so many? Front Endocrinol (Lausanne). 2018. doi:10.3389/fendo.2018.00117 210. Yakar S, Adamo ML. Insulin-Like Growth Factor 1 Physiology. Lessons from Mouse Models. Endocrinol Metab Clin North Am. 2012. doi:10.1016/j.ecl.2012.04.008 211. Murphy N, Knuppel A, Papadimitriou N, et al. Insulin-like growth factor-1, insulin-like growth factor-binding protein-3, and breast cancer risk: observational and Mendelian randomization analyses with ∼430 000 women. Ann Oncol. 2020. doi:10.1016/j.annonc.2020.01.066 212. Renehan AG, Frystyk J, Flyvbjerg A. Obesity and cancer risk: the role of the insulin-IGF axis. Trends Endocrinol Metab. 2006. doi:10.1016/j.tem.2006.08.006

147