Nutrigenomics and Aging: a Transcriptomic Perspective

Iowa State University Capstones, Theses and Graduate Theses and Dissertations Dissertations

2020

Nutrigenomics and aging: A transcriptomic perspective

Joe Lawrence Webb Iowa State University

Follow this and additional works at: https://lib.dr.iastate.edu/etd

Recommended Citation Webb, Joe Lawrence, "Nutrigenomics and aging: A transcriptomic perspective" (2020). Graduate Theses and Dissertations. 18244. https://lib.dr.iastate.edu/etd/18244

This Dissertation is brought to you for free and open access by the Iowa State University Capstones, Theses and Dissertations at Iowa State University Digital Repository. It has been accepted for inclusion in Graduate Theses and Dissertations by an authorized administrator of Iowa State University Digital Repository. For more information, please contact [email protected]. Nutrigenomics and aging: A transcriptomic perspective

Joseph Lawrence Webb

A dissertation submitted to the graduate faculty

in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

Major: Nutritional Sciences

Program of Study Committee: Matthew Rowling, Co-major Professor Elizabeth McNeill, Co-major Professor Andrew Bolstad Anumantha Kanthasamy Kevin Schalinske Peter Clark

The student author, whose presentation of the scholarship herein was approved by the program of study committee, is solely responsible for the content of this dissertation. The Graduate College will ensure this dissertation is globally accessible and will not permit alterations after a degree is conferred.

Iowa State University

Ames, Iowa

2020

TABLE OF CONTENTS

Page

LIST OF FIGURES ...... v

LIST OF TABLES ...... vi

NOMENCLATURE ...... vii

ABSTRACT ...... xii

CHAPTER 1. GENERAL INTRODUCTION ...... 1 General Introduction ...... 1 Dissertation Organization ...... 3 Authors’ Roles ...... 4 References ...... 5

CHAPTER 2. LITERATURE REVIEW ...... 6 Chronic Disease Prevalence ...... 6 Nutrigenomics: The Role of Diet Modifying the Transcriptome ...... 8 Whole Eggs Impact on Health & Disease ...... 10 Composition of Whole Eggs & Bioactive Mechanisms of Action ...... 11 Diabetes, Eggs, & Gene Expression ...... 15 Relationship Between Whole Eggs & Neurodegeneration ...... 16 Impact of Aging on the Transcriptome ...... 18 Comparative Aging Across Species ...... 22 Aging Associated Diseases & Dysregulation of the Transcriptome ...... 24 Predicting Chronological Aging ...... 26 Role of MicroRNAs in Health & Disease ...... 29 References ...... 32

CHAPTER 3. LARGE AND SMALL RNA SEQUENCING REVEALS OXIDATIVE- REDUCTION PATHWAYS ARE MODIFIED BY SHORT-TERM WHOLE EGG CONSUMPTION ...... 53 Abstract ...... 54 Introduction ...... 55 Methods ...... 56 Animals and Diets ...... 56 Large and small RNA Extraction & Sequencing ...... 57 Quality Control & Adapter Trimming...... 58 Read Alignment & Quantification ...... 58 Data Filtering & Quality Control ...... 58 Differential Expression Analysis using DESeq2...... 59 Heatmaps, Principal Component Analysis, & Volcano Plots ...... 59 Functional Enrichment Annotations ...... 59 qRT-PCR Validation Analyses ...... 60

iii

Results ...... 61 RNA Seq Differential Expression ...... 61 KEGG & GO Functional Enrichment Analysis ...... 61 MicroRNA Sequencing Differential Expression Analysis ...... 62 MicroRNA Gene Target Analysis ...... 62 Serum MicroRNA Refeeding Analysis ...... 62 Principal Component Analysis (PCA) & Hierarchical Clustering ...... 63 qRT-PCR validation ...... 63 Protein-protein Interaction Networks ...... 63 Food Intake & Body Weight Gain ...... 63 Discussion ...... 64 Conclusions ...... 69 References ...... 70 Tables & Figures ...... 75

CHAPTER 4. WHOLE EGG CONSUMPTION INCREASES GENE EXPRESSION WITHIN THE GLUTATHIONE PATHWAY IN THE LIVER OF ZUCKER DIABETIC FATTY RATS ...... 98 Abstract ...... 99 Background: ...... 99 Methods: ...... 99 Results: ...... 99 Conclusion: ...... 100 Introduction ...... 100 Methods ...... 101 IACUC Approval ...... 101 Animal Housing & Experimental Design ...... 101 RNA Extraction & Analysis ...... 102 SmallRNA & TotalRNA Sequencing ...... 102 Sequencing Quality Control & Adapter Trimming ...... 103 Alignment & Read Quantification ...... 103 Data Filtering & Normalization ...... 104 Differential Expression Analysis...... 104 Heatmaps, Principal Component Analysis, & Volcano Plots ...... 104 KEGG/GO Pathway Analysis ...... 105 qPCR Validation Analyses ...... 105 MicroRNA Bioinformatic Analysis ...... 106 Results & Discussion ...... 106 Total RNASeq Differential Expression ...... 107 MicroRNA Seq Differential Expression ...... 111 qPCR Analyses ...... 111 Mapping Between MicroRNAs & Target Genes ...... 111 KEGG & GO Functional Enrichment Analysis ...... 112 Strengths & Limitations ...... 113 Conclusion ...... 114 References ...... 115 Tables & Figures ...... 120

CHAPTER 5. IDENTIFICATION OF CONSERVED GENETIC FEATURES BETWEEN HUMANS AND DROSOPHILA IN AGING ...... 161 Abstract ...... 162 Introduction ...... 163 Methods ...... 164 Literature & Dataset Search ...... 164 Data Acquisition ...... 165 Quality Control & Adapter Trimming...... 165 Alignment & Read Quantification ...... 165 Data Filtering & Quality Control ...... 166 KEGG/GO Analysis & Protein Interaction Maps ...... 166 Algorithm Selection ...... 166 Regression & Classification ...... 167 Results ...... 167 Publicly Available Data Characteristics ...... 167 Algorithm Selection ...... 168 Predicting Chronological Age ...... 168 Age Group Classification ...... 170 KEGG & Go Analyses ...... 171 Discussion ...... 171 Strengths & Limitations ...... 175 Conclusion ...... 177 References ...... 179 Tables & Figures ...... 182

CHAPTER 6. GENERAL CONCLUSIONS ...... 196 Overall Summary & Conclusions...... 196 Strengths & Limitations ...... 199 Future Research ...... 202 References ...... 204

APPENDIX . IACUC APPROVAL ...... 206

LIST OF FIGURES

Page

Figure 3-1 Research Design Diagram ...... 92

Figure 3-2 PCA & Hierarchical Clustering ...... 93

Figure 3-3 Volcano Plots ...... 94

Figure 3-4 QPCR & Venn Diagram...... 95

Figure 3-5 Protein Interaction Networks...... 96

Figure 3-6 Food Intake & Body Weight Gain...... 97

Figure 4-1 Principal Component Analysis ...... 157

Figure 4-2 - Volcano Plots...... 158

Figure 4-3 Glutathione DEGs ...... 159

Figure 4-4 qPCR Comparison with RNAseq Data ...... 160

Figure 5-1 Chronological Age Prediction using XGBoost across species...... 192

Figure 5-2 Feature Selection using Aging Correlated Genes Across Species...... 193

Figure 5-3 Homologous Aging Correlated Genes ...... 194

Figure 5-4 : PANTHER Pathway Analysis...... 195

Figure 0-1 - IACUC Approval Letter ...... 206

LIST OF TABLES

Page

Table 2-1 Whole Egg Nutrient Composition...... 12

Table 3-1 Diet Composition of the WE- and CAS-based diets ...... 75

Table 3-2 Differentially expressed genes ...... 76

Table 3-3 GO Pathway Analysis Comparing Whole Egg to Casein ...... 80

Table 3-4 MicroRNA Differential Expression Analysis ...... 87

Table 3-5 MicroRNA Gene Targets ...... 89

Table 3-6 Primers for qPCRInformation ...... 90

Table 3-7 Weight Gain, Food Intake & Organ Weights ...... 91

Table 4-1 Differentially Expressed microRNAs ...... 149

Table 5-1: Study Characteristics...... 182

Table 5-2 Regression Algorithm Performance Comparison...... 183

Table 5-3 Age Ranges for Age Group Classification in Humans and Drosophila...... 184

Table 5-4 Aging Correlated Genes...... 185

Table 5-5 Regression Tables Predicting Chronological Age ...... 186

Table 5-6 Classification Tables for Age Group Prediction ...... 188

Table 5-7 Human Raw Read Counts...... 190

Table 5-8 Drosophila Raw Read Counts...... 191

vii

NOMENCLATURE

1,25D 1,25-dihdroxycholecalciferol

25D 25-hydroxycholecalciferol

ACE Angiotensin-converting enzyme

ACE-I Angiotensin-converting enzyme inhibitor

AD Alzheimer’s Disease

ALZ Amyotrophic Lateral Sclerosis

CAS Casein-based diet

CAS+D Casein-based diet with Vitamin D cAMP Cyclin adenosine monophosphate

CNS Central Nervous System

CYP27A1 Cytochrome p450, family 24, subfamily A, polypeptide 1

CYP27B1 Cytochrome p450, family 24, subfamily B, polypeptide 1

CVD Cardiovascular Disease

DAB2 Disabled-2

DMG Dimethylglycine

DNA Deoxyribonucleic Acid

DNMT DNA Methyltransferase

FTD Frontotemporal Dementia or Frontotemporal Degeneration

GNMT Glycine N-methyltransferase

GSH Glutathione

GSSG Glutathione Disulfide

viii

HD Huntington’s Disease

HDL High Density Lipoprotein

HFD High-fat Diet

IACUC Institutional Animal Care and Use Committee

IL-1 Interleukin-1

IL-6 Interleukin-6

IOM Institute of Medicine

KO Knockout

LBD Lewy Body Dementia

LDL Low Density Lipoprotein

LPS Lipopolysaccharides

MAPK Mitogen-activated protein kinases

MIR Micro Ribonucleic Acid

MS Multiple Sclerosis

MTHFR Methylene-Tetrahydrofolate Reductase

NAFLD Non-alcoholic Fatty Liver Disease

NEFA Non-esterified fatty acid

NF-kB Nuclear Factor Kappa-Light-Chain-Enhancer of Activated B Cells

NMDA N-methyl-D-aspartate Receptor

PD Parkinson’s Disease

PFC Pre Frontal Cortex

RNA Ribonucleic Acid

SAH S-Adenosyl-Homocysteine

SAM S-Adenosyl-Methionine

SD Sprague Dawley

SEQ Sequencing

STZ Streptozotocin

T1DM Type 1 Diabetes Mellitus

T2DM Type 2 Diabetes Mellitus

TNF-a Tumor necrosis factor-alpha

TLR Toll Like Receptors

UVB Ultraviolet B

VLDL Very Low-Density Lipoprotein

VDR Vitamin D receptor

VDRE Vitamin D responsive element

WE Whole egg-based diet

ZDF Zucker Diabetic Fatty

ACKNOWLEDGMENTS

With the greatest respect, I would like to express my sincere gratitude to the professors who served on my committee: Dr. Matthew Rowling, Dr. Elizabeth McNeill, Dr. Kevin

Schalinske, Dr. Peter Clark, Dr. Anumantha Kanthasamy, and Dr. Andrew Bolstad. I am deeply grateful for their support and thoughtful insight over the last four years. Without their outstanding mentorship, I would not be where I am today. It has been remarkable to work with such an amazing group of people and learning as much as I can from these scientists. I would also like to thank Dr. Schalinske and Dr. Rowling for going out of their way during my undergraduate career to answer my questions about graduate school. Thank you both for investing in me as an undergraduate student and making time to grab coffee with me at the hub. Without the conversations with you two, I would not have pursued opportunities here at

Iowa State University. I would also like to recognize Dr. Mike Baker, Tanya Murtha, and

Kevin Cavallin who have assisted our lab with our sequencing experiments over the last few years at the DNA facility. Additionally, I would like to thank Jeanne Stewart for all her guidance and assistance troubleshooting problems, while also helping our labs running DXA scans on rats in pizza boxes. There have also been many faculty members in our department who have aided in guiding me through graduate school and I would like to extend thanks to all the Department of Food Science and Human Nutrition staff members for their incredible flexibility and willingness to support our students. Without the assistance from all these people, numerous questions would have gone unanswered.

Completing these projects would have been extremely difficult to complete without the assistance from all our undergraduate research assistants over the years. I would really like to thank all of the students who have worked as research assistants with us including but not

xi limited to: Lily Harvison, Claudia Carrillo, Maggie O’Brien, Ryan Fisher, Nataja Hill, and

Brooke Vogel. Also, thank you to the other graduate students who provided tremendous support for our research studies including Dr. Samantha Jones, Dr. Cassondra Saande, Ella

Bauer, Alicia Taylor, Simon Moe, Alyssa Hohman, and Carter Reed. Finally, thank you so much to Amanda Bries, whose guidance and support was instrumental in helping me complete my PhD. Thank you for all your assistance with running experiments, with coursework, troubleshooting problems, mentoring students and most importantly for all your unwavering support, especially when our cell culture incubators broke down. Thank you for encouraging me throughout graduate school to pursue my goals, for understanding when I had to work weekends or crazy hours to handle our research studies, and for believing in me even when I didn’t believe in myself. Finally, thank you to all my family and friends. Thank you for encouraging me to pursue this lofty goal of becoming the first person in our family to attend graduate school and graduate with a PhD. Thank you for pushing me and helping me through the tough times with your words of encouragement and understanding.

xii

ABSTRACT

The transcriptome, or all Ribonucleic Acid (RNA) molecules within a tissue or cell, is altered during metabolic diseases and aging. Understanding how specific foods and biological processes alter the transcriptome will enable identification of interventions to attenuate aging- associated or metabolic disease processes. As described throughout this dissertation, we examined how aging and specific dietary patterns affect the transcriptome across multiple tissues in rats, Drosophila, and humans to achieve the following objectives: 1) Identify if short-term dietary whole egg consumption modifies microRNAs in the blood or in tissues; 2) Examine if long-term dietary whole egg consumption alters the transcriptome during type 2 diabetes mellitus

(T2DM); and 3) Determine normal transcriptomic signatures of aging and examine if these profiles can predict age in humans and Drosophila in order to identify conserved aging genes.

In our first study, we examined the role that nutrition plays in regulating the transcriptome by examining how whole egg consumption affects the blood, liver, kidney, prefrontal cortex, and adipose tissue. We identified that short-term consumption of whole egg does not alter the circulating microRNA status in the blood, but modifies the transcriptomes across the liver, kidney, brain, and adipose tissues. In the second study, we examined if long- term consumption of whole egg alters gene expression or microRNA profiles during T2DM and determined that the gene-diet interactions from consuming whole egg may yield therapeutic benefit during diabetes by improving glutathione metabolism. Collectively, we demonstrated aging-associated transcriptomic profiles across multiple species can be used to model aging, consuming whole egg doesn’t alter microRNA status in the blood but may serve as a beneficial dietary approach to improve glutathione metabolism during T2DM.

xiii

In the final study, we processed publicly available data to identify how the transcriptome in the prefrontal cortex is altered during aging between humans and Drosophila. Then we examined if gene expression in the prefrontal cortex can be used to predict age or classify subjects into age groups. We discovered 41 conserved genes across humans and Drosophila that predict aging across species and demonstrated that using genes that are highly correlated with age in Drosophila, we can predict aging in humans using machine learning algorithms. Overall, these research studies have expanded the knowledge of how aging and diet impact gene expression across tissues, while uncovering novel genes and microRNAs that might be altered during aging or while consuming diets rich in whole egg.

CHAPTER 1. GENERAL INTRODUCTION

General Introduction

In the United States, chronic diseases such as cancer, obesity, type 2 diabetes mellitus

(T2DM), and Alzheimer’s disease affect at least 92% of adults over 65 years of age, while over

45% suffer from more than one chronic diseases [1][2]. Aging, the progressive decline of organ function over time, also strongly correlates with the development of chronic diseases [3], underscoring the need to develop new therapeutic approaches to improve human health.

Advancements in medicine have ushered in the era of personalized medicine, where individual’s clinical, genetic, and environmental factors are used to develop safe and effective treatment plans for disease instead of a one-size-fits-all approach [4]. Precision nutrition, one segment of personalized medicine, is defined as the development of personalized nutrition recommendations for individuals according to their underlying lifestyle, genetic, and clinical conditions [5]. Since dietary patterns directly affect disease onset and the development of chronic diseases, it is important to understand how nutrigenomics - the study of how dietary factors and nutrients - alters gene expression. Improving our understanding of the interactions between food, genetics and disease will pave the way for effective personalized medicine to be used for the prevention and treatment of chronic diseases.

Functional foods are defined as food items that contain bioactive ingredients and nutrients that have a positive influence on health and longevity [6]. Eggs are considered a complex functional food [7] rich in healthy protein, fats, vitamin D, choline and B vitamins while also containing bioactive compounds such as omega-3 fatty acids and antioxidant properties [8]. According to the 2015–2020 Dietary Guidelines for Americans [9], eggs are a nutrient-rich and low-cost food that should be included in a healthy diet that may be especially

2 beneficial for adults over 65 years of age. Studying the connection between aging, dietary patterns, and chronic disease may identify new opportunities to slow or reverse disease progression and ultimately develop new therapeutic treatments for combatting chronic disease

[10].

Chapter 2 of this dissertation will provide the necessary background information and supporting research studies that highlight the need for additional research at the intersection of diet, aging and disease. The research presented in Chapter 3 of the dissertation will first examine the response of short-term dietary consumption of whole egg on gene expression in multiple tissues and uncover the molecular pathways that are altered by eating eggs. Chapter 4 will provide a detailed examination of how consuming whole egg may alter expression of genes and microRNAs in a rat model of T2DM and examining the extent by which glutathione is a key metabolic pathway regulated by whole egg consumption. Chapter 5 of this dissertation will examine how the transcriptome changes in the brain across time across multiple species, while also examining the utility of RNA sequencing data as a means for training machine learning models to predict chronological age across species in order to identify novel aging-associated genes. In Chapter 6, there is a detailed summary of this research and the general conclusions of these studies are compared/contrasted to the greater body of available research literature. Finally, there is one appendix chapter that highlights the Institutional Animal Care and Use Committee’s

(IACUC) approval of the animal research conducted within this dissertation.

Dissertation Organization

This dissertation is composed of 6 main chapters including a general introduction, literature review, three manuscripts, and ending with a general conclusion. The first manuscript

“Large and small RNA sequencing reveals oxidative-reduction pathways are modified by short- term whole egg consumption,” is in preparation to be submitted to PLOS ONE. In this study, we reported that acute dietary whole egg consumption doesn’t modify circulating microRNAs in the blood, as well as showed that consuming whole egg for 2 weeks can result in upregulation of oxidative-reduction pathways such as glutathione metabolism. The second manuscript “Whole

Egg Consumption Increases Gene Expression within the Glutathione Pathway in the Liver of

Zucker Diabetic Fatty Rats” is in preparation to be submitted to PLOS ONE. In this study we examined how eight weeks of dietary whole egg consumption modifies RNA and microRNA expression in multiple tissues when compared to an isocaloric casein-based diet in a genetic rat model of T2DM. The third manuscript, “Identification of Conserved Genetic Features between

Human and Drosophila in Aging,” is in preparation to be submitted to PLOS Computational

Biology. In this paper, we investigated the extent to which aging modifies gene expression in the prefrontal cortex across Drosophila and humans, as well as identified how machine learning can be used to model aging in order to identify novel aging associated genes. The literature cited throughout this dissertation are listed at the end of each chapter. The final chapter of this dissertation, general conclusions, provides an overall summary of this body of work and highlights potential future directions based on the data generated within each study. The appendix section contains the Institutional Animal Care and Use Committee (IACUC) approval form for the animal research within this dissertation. All animal studies described within this dissertation were approved by the IACUC at Iowa State University prior to the start of the study.

Authors’ Roles

Throughout my doctorate, I primarily worked on several projects examining the influence of whole egg consumption in rats, as well as processing publicly available RNA sequencing data to identify novel aging associated genes. For each of these manuscripts, I contributed to the study design, all aspects of the animal management including daily feeding/weighing, data generation, data analysis, and writing the publications. Outside of the research contained within this dissertation, I also assisted with all aspects of animal management including daily feeding/weighing, data collection, data analysis, and conducted some of the experiments in collaboration with Dr. Cassondra Saande in examining the role of whole egg influencing insulin resistance and investigating how the effects of whole egg consumption on body weight gain using a dose-response study design to determine the smallest effective dosage of whole egg to alter weight gain. Outside of the previous work with Dr. Saande, I have also aided with two studies where I will be an author examining the role of whole egg influencing Poly Cystic

Ovarian Syndrome. In these studies, I aided in every aspect of the animal management including daily feeding/weighing, conducting surgeries, data collection, data analysis, and editing the publication for a study. Additionally, I have contributed to data analysis for an additional microRNA sequencing project in collaboration with Dr. Kendal Hirschi focused on the role of amino acids influencing microRNA expression in pig muscle.

References

1. Raghupathi W, Raghupathi V. An empirical study of chronic diseases in the united states: A visual analytics approach. International Journal of Environmental Research and Public Health. 2018;15. doi:10.3390/ijerph15030431

2. Hung WW, Ross JS, Boockvar KS, Siu AL. Recent trends in chronic disease, impairment and disability among older adults in the United States. BMC Geriatrics. 2011;11: 47. doi:10.1186/1471-2318-11-47

3. Prasad S, Sung B, Aggarwal BB. Age-associated chronic diseases require age-old medicine: Role of chronic inflammation. Preventive Medicine. NIH Public Access; 2012. p. S29. doi:10.1016/j.ypmed.2011.11.011

4. Cutter GR, Liu Y. Personalized medicine: The return of the house call? Neurology: Clinical Practice. 2012;2: 343–351. doi:10.1212/CPJ.0b013e318278c328

5. Betts JA, Gonzalez JT. Personalised nutrition: What makes you so special? Nutrition Bulletin. 2016;41: 353–359. doi:10.1111/nbu.12238

6. Williamson C. Functional foods: what are the benefits? British journal of community nursing. 2009. pp. 230–236. doi:10.12968/bjcn.2009.14.6.42588

7. MicroRNAanda JM, Anton X, Redondo-Valbuena C, Roca-Saavedra P, Rodriguez JA, Lamas A, et al. Egg and egg-derived foods: Effects on human health and use as functional foods. Nutrients. MDPI AG; 2015. pp. 706–729. doi:10.3390/nu7010706

8. Zambrowicz A, Pokora M, Setner B, Dąbrowska A, Szołtysik M, Babij K, et al. Multifunctional peptides derived from an egg yolk protein hydrolysate: Isolation and characterization. Amino Acids. 2015;47: 369–380. doi:10.1007/s00726-014-1869-x

9. 2015-2020 Dietary Guidelines | health.gov. [cited 14 Mar 2020]. Available: https://health.gov/our-work/food-nutrition/2015-2020-dietary-guidelines/guidelines/

10. Samieri C, Sun Q, Townsend MK, Chiuve SE, Okereke OI, Willett C, et al. The association between dietary patterns at midlife and health in aging an observational study. Annals of Internal Medicine. 2013;159: 584–591. doi:10.7326/0003-4819-159-9-201311050-00004

CHAPTER 2. LITERATURE REVIEW

Chronic Disease Prevalence

Over 133 million Americans suffer from a chronic disease and the number of Americans suffering from at least one chronic disease continues to increase every year [1]. Chronic diseases as defined by Hwang et al in 2001 are health conditions with a duration greater than one year causing disfunction, require monitoring by a medical professional involving essential treatment

[2]. Chronic diseases include type 2 diabetes mellitus (T2DM), cardiovascular disease, neurological disorders, and cancer along with many others that in the United States account for more than 75% of healthcare expenditures and are the leading cause of death in North America

[3]. Adults greater than 65 years of age are becoming the largest proportion of the US population

[4] and more than 92% of older adults have at least 1 chronic disease [5]. Aging is an important risk factor for chronic diseases [6] and the prevalence of aging-associated chronic diseases like

Alzheimer’s disease are projected to increase 65% by 2030 [7]. There is an urgent need to develop new therapies that can treat, prevent, or slow down the progression of chronic diseases.

Recent technological advancements such as Next Generation Sequencing (NGS) are ushering in a new era of biomedical science focused on big data in hopes to solve global problems such as creating new agricultural solutions that will feed 7 billion people and developing new therapeutics for the treatment of chronic diseases. As defined by Chris Snijders, big data consists of data sets that are too large for common software tools to analyze, process, or visualize data in a short time frame [8]. In the biomedical sciences, big data allows for combined interpretation of environmental, clinical, and genetic information from a single sample, leading to breakthroughs in disease detection. Every day more than 2.5 quintillion bytes of data are generated and in the past two years the world has created more than 90% of the available

7 information in the world [9]. This new accelerating pace of data aggregation is changing how science is conducted, shifting laboratories from hoarding their data towards sharing data and making their data as publicly available via open access,

Since the sequencing of the first human genome in 2001 [10], medicine has rapidly adopted the use of genetic information to inform clinical decisions, disease diagnosis, and altered our understanding about individual predispositions to chronic disease. NGS techniques have become mainstream because the current costs to sequence a genome has drastically decreased from $2.7 billion in 2003 to <$599 in 2020. The rapid increase in data availability with the simultaneous decreased cost to examine genetic information has created a new area of science known as transcriptomics, or the study of the sum of all ribonucleic acid (RNA) molecules in a sample at a given time. To work with these large quantities of data, computer programming is gaining a more integral role in the science of collecting and analyzing DNA. These advancements have revealed far more than the role that our DNA plays in governing height or eye color; it has also highlighted that by studying transcriptomics we can discover new insights into human health and disease.

Transcriptional regulation occurs at several different checkpoints; 1) transcription of

DNA to RNA, 2) RNA processing in the nucleus, 3) RNA transport across the nucleus, 4) RNA translation into protein in the cytosol, 5) post-translational modification of proteins. At each of these steps, the regulation of gene expression can be turned up or turned down for individual genes or many genes at the same time from environmental influences such as exercise, drugs, sunlight, sleep, nutrition, aging, and disease [11]. Beneficial environmental influences like improving sleep quality have been shown to positively affect the transcriptome by upregulating gene networks regulating the immune system that may protect against disease [12], while

8 negative exposures like smoking cigarettes can upregulate the expression of deleterious oncogenes, or genes that may lead to the development of cancer [13]. With the wide array of factors that can influence the transcriptome, by interrogating how diet, aging, and metabolic disease influence the transcriptome, we can uncover new targets for developing future therapeutics to treat diseases. By improving the understanding of how diet impacts the transcriptome, we can uncover the specific role of nutrients and dietary patterns in modifying the early molecular events that result in T2DM and lead to identification of biomarkers to aid in earlier detection of disease onset. Similarly, by contributing additional knowledge of how the transcriptome changes over time, we can create more predictive tools that use this information to model aging leading to earlier detection of aging-associated diseases. Overall, this approach of using transcriptomic data to model aging or metabolic diseases will ultimately aid in the prevention and/or delay of disease complications through identifying how dietary patterns and aging relate to the onset of disease.

Nutrigenomics: The Role of Diet Modifying the Transcriptome

Diet plays an important role in controlling expression across the transcriptome [14] and the emergence of big data is revolutionizing nutrition, creating entirely new fields of study including nutrigenomics. The field of nutrigenomics is defined as studying how common dietary components affect health on a molecular level by altering gene expression [15], arising from studying how dietary patterns influence gene expression [16]. With large quantities of data becoming increasingly available and easier to collect, big data in the field of nutrition allows for the combination of data across biochemistry, metabolomics, and transcriptomics techniques to examine interactions between genes and nutrients at a molecular level. Several studies examining

9 how low caloric diets may reduce body weight are also now demonstrating the mechanism by which these results are obtained may be due to directly affecting the transcriptome in tissues such as the adipose tissue [17]. The growing awareness that foods or dietary patterns can support health or prevent disease has moved the field of nutrition toward tailoring dietary patterns based on someone’s personal transcriptomic profile. This movement in the biomedical community towards tailoring medication according to personal genetic and transcriptomic profile to prevent or treat diseases through optimizing individual’s dietary intake.

Many foods that affect the transcriptome are known as functional foods, which do not have a universally accepted definition. Functional foods within this dissertation will follow the definition as defined by Arnal et al. [18] that are nutrient dense foods rich in vitamins, minerals, and bioactive components that play key roles in nutrition. Nutrigenomics approaches to treating disease relies on dietary modification to include functional foods that provide beneficial nutrients such as choline during Parkinson's disease [19], or avoidance of specific nutrients like avoidance of phenylalanine during phenylketonuria [20]. Personalized nutritional recommendations for disease therapies focused on affecting gene expression have shown to be beneficial during

Alzheimer’s disease, where Athanasopoulos and colleagues highlighted the role of foods high in

Vitamin B-12 has been shown to affect one-carbon metabolism which would regulate the DNA methylation process [21]. Additionally, other groups like Saande et al. in 2017 [22] demonstrated that consuming diets rich in whole eggs can significantly affect body weight, but the mechanism underlying these outcomes remain unknown. Clarifying the exact transcriptomic response to functional foods may pave the way for development for personalized dietary recommendations to treat or prevent chronic diseases.

Whole Eggs Impact on Health & Disease

Hens’ eggs are a nutritious, low cost and commonly consumed functional food that is rich in complete protein, vitamins, and minerals [23]. Even though eggs are nutritious and are a low- cost food source for animal protein, vitamin A, iron, choline [24]; egg consumption varies across the world. African countries consume one of the lowest average quantities of egg per capita (36 eggs/year) [25], compared to an average egg consumption per capita in India (62 eggs/year) while Mexico consumes one of the highest egg consumption per capita (385 eggs/year) [26].

According to the American Egg Board (AEB), in the United States the average egg consumption per capita was estimated in 2017 to be (279 eggs/year) [27] which has grown by 5.25 eggs per year to (289.5/year/capita) in 2019 [28].

The consumer demand for functional foods, including hens’ eggs, is increasing because of the opportunity to decrease disease risk by consuming nutritious functional foods [29][30].

Hens’ eggs have been shown to play a beneficial role in preventing chronic disease [31] which is thought to arise from their anti-oxidative properties [32]. Several mechanisms that have been proposed to explain how hens’ eggs exert beneficial effects for disease, including modification of gene expression, vitamin D mediated regulation of transcription, and alterations in one carbon metabolism [33]. Each one of these mechanisms can be partially explained by the nutrient contents which are described in depth in the next section.

Recently, some groups have proposed the controversial idea that foods may contribute dietary microRNAs that can be absorbed through the intestine and enter circulation [34].

Additionally, it has been suggested that microRNAs in milk-derived exosomes are conserved across species [35] and it is thought that these milk-derived microRNAs could be bioavailable

11 due to encapsulation in exosomes [36,37]. Surprisingly, microRNAs have also been identified in hens’ eggs [38] and consuming these dietary egg-derived microRNAs is thought to modulate gene expression. While this is a highly controversial topic [39] that plant and animal foods contain microRNAs and it is unknown to what extent these microRNAs can be absorbed by humans. Some groups have argued that diet-derived microRNAs cannot be absorbed due to the acidic PH from human stomach acid [40] that would degrade the microRNAs prior to reaching the intestine. Other groups argue that diet derived microRNAs are protected by exosomes [41], potentially allowing them to be absorbed. While the controversy still exists, it is informative to understand how specific dietary patterns or bioactive foods can influence circulating microRNAs in the blood or affect endogenous expression in tissues. One bioactive food suggested to influence circulating microRNAs are hens’ eggs, which are described in the next section.

Composition of Whole Eggs & Bioactive Mechanisms of Action

One large hens’ egg is a rich source of in vitamins and nutrients. Briefly, one egg contains 1.1 microgram of vitamin D3, 22 micrograms of folic acid, 125 mg of choline, and bioactive peptides in egg yolk are thought to yield anti-oxidative properties upon consumption [42].

According to the American Egg Board (AEB) nutrient composition data [31,43], hens’ eggs are a great source of complete protein, and the nutrient composition of whole egg is described in detail within TABLE 1 on a 100-gram basis. This table provides data from the USDA nutrient database describing the nutrient content of hen eggs. All nutrient data represented in this table are courtesy of the USDA FoodData Central 2019 (Formerly the National Nutrient Database for

Standard Reference). Nutrient data presented are described for 100 grams of whole egg.

Table 2-1 Whole Egg Nutrient Composition. Whole Egg Nutrient Composition Per 100 Grams Nutrients Vitamins Minerals MOISTURE - g (75.81) NIACIN - mg (0.103) CALCIUM - mg (55) FAT - g (total lipid) (10.3) RIBOFLAVIN - mg (0.523) IRON - mg (1.77) ASH - g (1.16) B12 - mcg (1) MAGNESIUM - mg (11.2) CARBOHYDRATE - g (0.91) PANTOTHENIC ACID - mg (1.57) PHOSPHORUS - mg (189) GLUCOSE - g (0.25) VITAMIN A - IU (570) POTASSIUM - mg (117) CALORIES - kcal (150) THIAMIN - mg (0.067) SODIUM - mg (121) CHOLESTEROL - mg (420) PYRIDOXINE (B6) - mg (0.188) ZINC - mg (1.2) All nutrient data represented on this page courtesy of US Department of Agriculture, FoodData Central, While eggs contain2019 (Formerly numerous National bioactive Nutrient Database ingredients, for Standard it has Reference been proposed (SR). that eggs primarily exert their beneficial effects through three primary mechanisms of action [44]: 1) regulating gene expression, 2) one carbon metabolism, and 3) reducing oxidative stress. Eggs can regulate gene expression due to their rich composition of vitamins and nutrients; vitamins A, D,

& E have receptors throughout the genome that are activated in the presence of these vitamins.

After consuming eggs, these vitamins bind to their receptors to modify gene expression among their target pathways [38]. Eggs have also been shown to regulate one carbon metabolism through their choline content [21] and their B vitamin content allowing for successful re- methylation of homocysteine to methionine. Finally, eggs have been proposed to also contain anti-oxidative properties that function to attenuate oxidative stress [29]. Each one of these mechanisms will be described in more detail within the following sections.

One important nutrient in eggs is vitamin D3, a steroid hormone primarily found within the egg yolk [45]. Vitamin D3 can be primarily found in two forms in food, the biologically active steroid hormone 1-25-OH-vitD3 (1,25D3) and the main circulating form 25-OH- vitD3

(25D3) [135]. A single large egg can contain between 50 to 500 IU of 1-25D3 and 25D3 [46], which directly affects the transcriptome by modifying gene expression across the entire human genome through interactions with the vitamin D receptor [47]. Vitamin D is known to regulate

13 calcium homeostasis, immune function and suboptimal vitamin D levels may contribute to the onset or progression of osteoporosis, sarcopenia, depression [48], neurodegenerative diseases

[49], and even specific forms of cancer [50]. Even though humans can synthesize vitamin D from UV-B exposure on their skin, as of 2018 nearly 40% of participants from the USA included in the 2012 National Health and Nutrition Examination Survey (NHANES) data were classified as vitamin D deficient [45]. The clinical definition of Vitamin D deficiency is where 25- hydroxyvitamin D concentrations are lower than 50 nanomolar per liter in the serum whereas vitamin D insufficiency is denoted by 25D3 concentrations in the serum between 50 to 75 nanomolar per liter [51]. Consuming hens’ eggs has been proposed as an effective means to increase serum vitamin D status in rats [31] and humans [38]. To date, very few studies have focused on identifying the cellular mechanisms that are altered by egg consumption or the transcriptional regulation by egg consumption and future studies should work towards identifying these underlying relationships with egg consumption.

In addition to the vitamin content of hens’ eggs, these functional foods have been used to examine one carbon metabolism in part due to their folate and choline content [43]. Choline plays key roles in numerous pathways due to its role as an essential nutrient involved in the synthesis of acetylcholine, functions as a methyl donor, and its use in the generation of phospholipids [52]. Whole egg contains roughly 1,149 mg of choline on a 100-gram basis [53], and consuming three whole eggs per day has been shown to provide a similar quantity of choline compared to consuming an over the counter supplement [54]. Controversially, eggs have been previously been classified as “unhealthy” due to their high choline content that is a precursor for trimethylamine N-oxide (TMAO), a metabolite which has been shown to predict risk of cardiovascular disease [55,56]. Other studies have shown that egg-based choline may play a

14 beneficial role in preventing hyperhomocystemia, which has also been related to increased cardiovascular disease [57]. Eggs also contribute to regulating methyl group metabolism by providing B vitamins (B6, Folate, B12) and more recent work is focused on enhancing the B vitamin content of hens’ eggs to improve folate intake across the US [42,58]. While the controversy of the choline content in egg remains inconclusive, future work examining how egg- based choline affects disease progression may aid in revealing the true nature of this nutrient.

Another controversial facet of egg consumption surrounds the cholesterol and fatty acid content in the egg yolk [59–65]. Eggs have garnered a lot of attention in the media over the last

10 years; some groups have criticized eggs after identifying that greater egg intake correlates with increased mortality and heart disease risk [44], while other groups have demonstrated that the dietary cholesterol in egg yolks is poorly absorbed and doesn’t contribute to the circulating cholesterol in the plasma [66]. The literature to date surrounding the role of egg consumption and associations with cardiovascular disease are inconsistent [66] and this remains a heated topic of discussion in the field of Nutrition.

Outside of the controversy, eggs have been proposed to play a beneficial role in regulating oxidative stress [67], in part due to the bioactive metabolites ovalbumin, ovotransferrin and lysozyme that reside primarily within egg whites [68]. The lysozyme within egg whites has been shown to suppress reactive oxygen species (ROS) and activate oxidative stress response genes in hep G2 cells and protects mice against oxidative stress [33]. Other anti- oxidative molecules within egg whites include the protein ovotransferrin [69], which comprises nearly 12% of the total protein within egg whites, and ovalbumin which composes over half of the protein in the egg white [70]. Other research groups have proposed that peptides in egg, protein sequences composed of less than 50 amino acids, may be providing the antioxidative

15 properties [7]. Since oxidative stress is hypothesized to play a key role metabolic diseases and aging, increasing consumption of foods rich in dietary antioxidants may reduce the risk of developing these diseases by improving redox homeostasis. Currently there are several groups examining how enriching eggs with added antioxidants through manipulation of poultry feed may improve antioxidant levels commonly seen with dietary supplementation [71]. Future work examining how the antioxidant properties of egg may shed light on the exact role of egg consumption in preventing the onset or delaying progression of oxidative stress related diseases.

Diabetes, Eggs, & Gene Expression

Egg consumption during T2DM has been a controversial topic where several studies suggest consuming eggs increases the risk of developing T2DM [72], while other studies have found no association [73–77] or even found egg consumption decreases risk of developing

T2DM [78]. More recently, Shin et al. identified that diabetics consuming at least one egg per day were at not at increased risk of developing cardiovascular disease when compared to subjects consuming less than one egg per week but egg consumption might increase incidence of developing T2DM [76,79,80]. The literature surrounding egg consumption during T2DM remains controversial; however, the role of diabetes impacting the transcriptome has been examined thoroughly. To date, several studies have demonstrated that T2DM strongly influences the transcriptome across multiple tissues [73], but there is very little evidence examining how consuming whole eggs will influence the transcriptome [79].

During T2DM, the impact on the blood transcriptome has been characterized in animal models and humans where genes encoding mitochondrial function are strongly downregulated during T2DM [75,79–81]. In the beta cells of the pancreas that produce insulin, in diabetic

16 patients more than 440 genes were differentially expressed when compared to healthy age matched controls [77,82–88]. These data suggested that multiple pathways in the pancreas are altered due to diabetes, where oxidative phosphorylation genes were the largest category of transcripts that were downregulated during diabetes. Other studies examining how T2DM affects gene expression in the plasma have shown that genes involved in glucose transport, insulin secretion, and apoptosis are differentially regulated during diabetes [89]. While the findings across these studies do not show the exact same direction of change in these genetic regulatory pathways, they all agree that the specific expression patterns reflect the pathogenic mechanisms of diabetes [90].

Relationship Between Whole Eggs & Neurodegeneration

The most recent estimates of dementia prevalence in the USA are from 2007, where 14% of adults over 71 years old in the USA have at least one form of dementia [91]. The prevalence of neurodegenerative disorders such as Alzheimer's disease (AD) and Parkinson's disease (PD) is increasing worldwide [92], underscoring the need to identify therapies to treat these neurological diseases. Environmental factors such as lifestyle and nutrition have been shown to strongly affect neurodegenerative disease progression [92], highlighting how lifestyle interventions may be able to counteract changing metabolism during aging-associated diseases. Current dietary intervention strategies to attenuate dementia-related symptoms include reducing blood pressure to decrease risk of vascular dementia [93][94], or employing the Mediterranean diet to delay onset of neurodegeneration [17]. More recently, there is emerging evidence in favor of the hypothesis that susceptibility to aging-associated neurodegenerative diseases is influenced by vitamin D

[95], where low serum vitamin D increases risk of abnormal brain function resulting in depression or AD-like pathology [96].

To date, no studies have examined the specific role of egg consumption with prevention of AD, but some epidemiological evidence supports the hypothesis that egg consumption may be beneficial for preventing cognitive decline. Previous epidemiological studies have suggested that increased choline consumption may be beneficial for preventing AD [97] and neither cholesterol nor egg consumption was found to increase risk of developing AD among men [98]. These results corroborate previous cross-sectional human data highlighting increased consumption of eggs and shellfish decreased the odds of developing mild-cognitive impairment [25,26], which is the stage between the normal decline in cognitive function due to healthy aging and the rapid decline due to dementia. It has been well described that serum vitamin D status is decreased as a function of age [99], and that serum vitamin D status is also marked decreased in patients with neurodegenerative diseases [100]. A single large egg can contain between 50 to 500 IU of vitamin D, and the average egg consumption per day in America is on the rise [100]. Since eggs are a great source of vitamin D and may be able to more effectively increase serum vitamin D status than consuming supplemental vitamin D [101], increasing egg consumption may serve as a viable approach to improve vitamin D status throughout the lifespan as a means to decrease the risk of developing neurodegenerative diseases.

Studies examining the specific role of egg consumption in preventing aging-associated diseases are very limited [102]. One of the first studies to examine the role of egg consumption and neurodegenerative disease risk was Fall et al. in 1999 [103]. They described that Swedish individuals consuming eggs every day significantly decreased their risk of developing PD compared with individuals who never consumed eggs. Similarly in 2006, a case-control study in

China was conducted examining dietary risk factors, where Ma et al. noted subjects consuming >

5 eggs per day was related to a non-significant trend of a 33% decreased relative risk of developing PD compared to controls [101]. More recently in 2017, a larger study of 1053 self- reporting patients with PD reported a non-significant inverse relationship between consuming 1 egg per day and PD symptom severity [104]. While these data are primarily associational in nature, they highlight that employing dietary strategies to increase egg consumption may serve as a low cost, yet effective means to decrease risk of disease. These claims should be assessed experimentally to determine if the nutrients and bioactive compounds from eggs may be able to affect cognitive decline or delay the progression of dementia related symptoms.

Impact of Aging on the Transcriptome

While diet can profoundly affect the transcriptome, aging is another complex process that is known to alter the transcriptome [105] affecting every person in the world. As the USA population continues to age, understanding how aging affects the transcriptome may shed light on the unique differences between healthy normal aging and developing chronic disease. The transcriptional landscape changes drastically during aging across numerous tissues [106], however there are distinct differences in aging according to biological sex [107]. This unique dimorphism in aging has been proposed as a reason why more women develop aging-associated diseases like AD than men [7]. Studying aging in the laboratory revolves primarily around using model organisms to examine chronological aging, or the amount of time in which organism lives.

Other types of aging are often studied including the number of times cells can divide, known as replicative aging or premature aging diseases such as progeria[106]. The aging sections within this dissertation will primarily define aging through the lens of chronological aging and how the human body is altered according to the passage of time.

Studying chronological aging often employs the use of various short-lived animal models such as Drosophila and mice due to their well-defined genetics [108,109] or studying long-lived species such as naked mole rats to understand why their aging timelines defy traditional knowledge of chronological aging [110]. Non-human primates share similar aging pathologies and neurological deficits with humans [111]; however, since some species such as chimpanzees can live more than 60 years, it can be very expensive to study aging in non-human primates

[112]. Studying the aging brain transcriptome across species has revealed conserved gene expression patterns in humans, mice and rhesus macaques where genes such as neuroprotective networks regulating apolipoprotein-D are upregulated during aging while and synaptic cAMP signaling genes tend to be downregulated [113]. While the evidence across species highlights similarities, some studies examining neuropathological hallmarks of aging across 47 vertebrate species have indicated that the aging human brain displays unique cellular senescence when compared to lower organisms [114]. Other studies have argued that the aging human brain may be more closely related to brains of other mammals based on genetic features such as insulin like growth factor expression when only comparing mammalian models of aging [115]. By studying aging across species, we may be able to identify previously uncovered aging alterations in the brain that might be overlooked when studying aging within a single species by using conservation as a lens to view important aging signals that transcend multiple species.

Studying aging across species will lead to novel insights into evolutionary mechanisms that explain the complexity of aging within and across species. Transcriptomic experiments in the aging rat brain suggest that the genetic features most strongly altered by aging are immune function genes [116], while gene expression studies in humans have highlighted decreased neurotransmission gene expression and similar increases in immune regulatory genes across

2200 samples [117]. Overall, the gene expression changes occurring in the aging brain are thought to be more similar across multiple species than different, suggesting a universal aging transcriptomic signature [118]. There is not a simple blanket answer that perfectly explains the exact transcriptomic changes in aging across every species; therefore, the complex molecular changes due to chronological aging discovered within an individual species should be broadly compared to similarities during aging across species in order to identify the underlying analogous aging mechanisms that may be important among humans and other model organisms.

The transcriptome is not the only aspect of tissues that change as a function of age.

Previous work comparing aging in mice with humans has demonstrated that lipid levels in the brain change with aging similarly between mice and men [119]. Comparing aging rat brains with humans, rats display similar age-related neuronal loss starting at the end of adolescence [120], just like humans. In non-human primates such as Macaques, there are similar age-associated declines in neuronal function in the prefrontal cortex [121], emblematical of the decline in humans [122]. In mice, diverse protein families are altered by chronological aging [123] along with similar age-associated losses of neuronal function in the cortical tissues [124]. With the rise of big data, numerous types of data have become more integrated in the analysis of aging, allowing for inter-disciplinary interrogation of these data to enhance our understanding of the underlying biological networks that are altered during aging [125]. Specifically looking into the human transcriptomic during aging has identified that aging leads to an upregulating in genes regulating cellular senescence and decreased cell proliferation [126], while other studies have focused on compiling data from every organ across multiple ages to generate an atlas of the human transcriptome [127].

While the brain is the primary tissue that is examined during aging due to its large decrease in volume, morphological changes and alterations in gene expression, comparative biology studies have indicated similarities that extend across multiple species in other tissues

[128]. Each region of the human brain shows distinct transcriptomic changes with aging [129] and the complexity of the human brain has led investigators to examine the differences among specific cellular subpopulations within a single brain region at the single cell level [130]. Some of the key pathways that have been identified to change in aging fall into the following categories: 1) immune response [131], 2) oxidative stress [132,133], 3) cellular senescence, 4) autophagy [126], and 5) inflammation [134–136]. During aging, increased inflammation and oxidative stress [137] have been described as main hallmarks of aging, suggested to be a direct result from the accumulation of damage induced by these problems [138]. Transcriptomic analyses support these hypotheses across humans [117], Drosophila [132], mice [135], and other species [133,136,139] suggesting that progression of aging and aging associated disease symptoms across species could be slowed down drugs that provide anti-oxidant [140], anti- inflammatory [141] or increased immune function properties [142]. Outside of gene expression data supporting this hypothesis, emerging evidence suggests that microRNAs that regulate these pathways might also regulate aging [143]. While numerous studies support the hypothesis that microRNAs may play key roles in regulating aging and aging associated diseases [143], other studies have argued that the role in which microRNAs play may not have nearly as significant effects as alterations on the protein or mRNA levels [117].

Specifically, the brain transcriptome is altered during neurological disorders [144,145] and may be affected in numerous chronic diseases such as T2DM [146], AD [147], PD [148] and aging [149]. In rat models of T2DM, the prefrontal cortex displays an increase in inflammatory gene expression [150], apoptosis [151], oxidative stress [150] insulin resistance [152], and DNA repair pathways [153]. Similarly, in humans, it has been shown that T2DM downregulates similar neuronal networks and upregulates gene expression patterns correlated with neurodegenerative diseases [154]. Transcriptomic studies using mice have revealed that inflammatory gene expression is strongly upregulated as a function of age and disease severity

[155]. In the brains from patients diagnosed with AD, whole transcriptome profiling studies have also noted increased microglial inflammatory expression [156] but also highlighted that microRNAs may regulate disease pathology [157]. In PD, similar transcriptomic changes occur in the striatum, where decreased gene expression supporting synaptic activity and mitochondrial function have been reported in humans [158] and in rats [159]. In a meta-analysis of transcriptomic studies, similar results were noted across 300 subjects with decreased mitochondrial gene function, synaptic transmission, and DNA repair [160]. Co-expression studies in humans have shown decreased immune function and increased inflammatory gene expression [161]. In comparison to healthy aging, these diseases all result in upregulation of genes involved in inflammation signaling, oxidative stress and DNA repair [162].

Comparative Aging Across Species

Aging, the progressive decline in function as time progresses affects every living organism but the molecular effects from aging are not universal across every organism [163].

While there are distinct molecular patterns of aging among individual species [164], there are

23 also molecular commonalities of aging across multiple species [165]. Across all eukaryotic species, similar molecular characteristics of aging occur across the lifespan including decreased mitochondrial function, immune function, growth factor signaling, DNA damage repair, and dysregulated gene expression across tissues and organ systems [166]. Common laboratory rodents such as mice and rats are frequently used to study aging, where the interconversion from rodent age to human age is estimated that 10 rat days equates to one human year [167] and 9.125 mouse days equates to one human year [168]. These rodent models are often used to study aging due to their high conservation of human genes and physiological resemblance of human physiology [169]. Other common model organisms to study aging include Drosophila, but while there is not a published guide for converting Drosophila age to humans to date, following a lifespan ratio formula similar to the rodent models [167,168] above one Drosophila day as an adult can be equated to one human year.

While aging is a complex process with unique characteristics belonging to each species

[169], conserved aging-related changes in the transcriptome exist [169]. In a comparative cell culture study examining neuronal cell populations between rats, mice, and humans; gene expression signatures for cell migration, gliogenesis, and neurogenesis increase with age [170].

When just examining comparisons between rat and human brain tissues, their transcriptomic signatures both increase gene expression of inflammatory response genes as a function of age which may play an important role in depression [171]. While to date the direct comparison of

Drosophila head transcriptomes to human brain transcriptomes has not been examined, there are shared features of aging between the two species. In Drosophila, brain transcriptome changes with aging show deficits in olfactory function and memory [172] which is also common during

PD in humans [173]. The aging Drosophila brain also shows decreased gene expression in

24 oxidative phosphorylation, suggesting decreased mitochondrial function across time [174] which is emblematical of the changes in human aging transcriptomes [175].In addition to gene expression, other molecules such as microRNAs change across species as a function of age and recent evidence suggests altering microRNA status strongly impacts aging-associated diseases by either offering neuroprotection or stimulating neurodegeneration [176].

Aging Associated Diseases & Dysregulation of the Transcriptome

Aging is an important risk factor for aging-associated diseases and chronic diseases such as cancer, metabolic disease, and neurodegeneration [177]. More than 80% of US adults older than 65 years of age have at least one chronic disease [178] but understanding what separates healthy normal aging from diseased aging remains elusive. Transcriptomics experiments have been employed to identify what genetic changes distinguish healthy aging from disordered aging

[179]. The transcriptomic landscape of aging in mice demonstrate strong upregulation of inflammatory cytokine gene expression [180]; whereas studying aging across 14 species of

Drosophila suggested that individual sets of genes do not completely explain longevity, but instead complex relationships among global transcriptomic profiles may provide more insight into aging than single gene families [181]. In human tissues, Glass et al. noted that similar expression patterns occur across the skin, adipose, blood and brain that downregulating mitochondrial genes as a function of aging [182]. Studying transcriptomes across multiple species and multiple tissues may lead to discovering novel changes in the transcriptome that underlie disease development.

Transcriptomic experiments in human tissues have suggested that healthy aging gene expression patterns that promote extended lifespan may be downregulating inflammation‐related gene expression, and gene pathways decreasing insulin resistance [137]. Similarly, Wrobel et al used transcriptome-wide association analysis in the blood across 665 older adults between 70 –

76 years of age to demonstrate that aging gene expression pathways regulating the immune system were upregulated as a function of aging [183]. Other groups using comparative biology across zebrafish, mice and human blood samples demonstrated a shift from cancer associated gene expression to chronic degenerative disorder gene expression across the lifespan [184–186].

While there might be strong differences in blood gene expression profiles, the hallmarks of aging are more often characterized by changes in the brain [187].

Aging associated neurodegenerative diseases including AD, PD, Amyotrophic Lateral

Sclerosis (ALS), and Huntington's disease (HD) all undergo specific transcriptomic changes resulting in loss of neuronal function, deteriorated neuronal connections causing an inability to respond to stimuli which culminates in degeneration [188]. Transcriptomics studies examining gene expression changes in the brain have examined specific alterations occurring in these neurodegenerative diseases [188]. In 2015, Magistri et al. used transcriptomic profiling of the hippocampus from patients with AD compared to healthy age-matched control samples to identify that several long noncoding RNAs were differentially expressed due to AD and gene pathways coding for neuronal communication, blood flow and clearance of amyloid-β were downregulated [189]. These results corroborated the findings from Annese et al. in 2018

[185,186] that indicated differential expression of microRNAs play a significant role in AD where specifically microRNA-184 was upregulated in the hippocampus in parallel with decreased expression of its target genes such as NR4A2 implicated in long-term memory [185].

The transcriptomic changes occurring in PD differ from these changes in AD described by Jiang et al. in 2019. They combined transcriptomic datasets from humans and mice to show that in PD patients gene expression networks controlling circadian rhythm are upregulated in human and mouse models of PD [186].

HD, an autosomal dominant neurological disease resulting in severe motor symptoms, shares a lot of co-expression network changes with PD [144]. Kusko et al. showed that in an animal model of HD, gene expression in brain derived neurotrophic factor, glucocorticoid, and dopamine pathways in the rat striatum were significantly upregulated when compared to genetically similar control animals [190]. While studying a single disease can yield significant insight into disease mechanisms, comparing multiple neurodegenerative diseases can uncover common features shared across these diseases. Bayraktar et al. performed a meta-analysis by combining publicly available RNA sequencing data from the SRA from subjects with AD, HD,

ALS, and PD that impaired mitochondrial function was the strongest feature of neurodegeneration shared across all of these diseases [191–193]. While many neurodegenerative diseases share changes in the brain transcriptomes, emerging evidence suggests that metabolic diseases such as T2DM may also alter the brain transcriptome [194] and these changes are altered by dietary patterns [195].

Predicting Chronological Aging

Outside of using RNAseq to identify which genes are differentially expressed based on a treatment, some groups have assessed if the results from RNAseq can be used to predict aging[196] or detect diseases such as neurodegeneration [197]. By focusing on predicting chronological age, we can identify the important features that underly how statistical models

27 classify ages, which may aid in the identification of specific genes that are important in aging

[198]. Similar approaches using methylation data have also used age prediction to identify epigenetic markers that are altered by aging [199]. Neuroimaging studies predicting age have used age prediction as a proxy to detect brain abnormalities that may provide insight into early- stage neurodegeneration [197], while other scientific fields such as forensic science apply these techniques to predict the age of suspects in crime scene investigations [200]. While there are a few studies applying machine learning to next generation sequencing data in order to predict chronological age [149,201], to our knowledge no studies have sought to try to examine if machine learning models predicting aging can be used to identify novel aging genes across species.

One of the first groups to look at the predictive utility of gene expression data was

Ferrucci et al. in 2000 who built a statistical model that can classify humans according to “old”

(>75 years of age) compared to “young” (<65 years of age) [202–204]. This statistical modelling approach primarily used transcripts from blood leukocyte samples to determine which transcripts correlate with aging, then building a statistical model using these correlated features. Other groups have used proteomics data to predict aging [205] that identified specific proteins that previously were un-associated with aging may contribute to aging in humans.

Several of these previous studies have examined numerous types of transcriptomic data from methylation patterns, RNA transcripts and biochemical data using different statistical modelling approaches. One of the more accurate data types that can be used to model aging is

DNA methylation. DNA methylation status has been used to predict biological age (Corr. > 0.96, absolute error < 3.6 years) [205] indicating that support vector regression models may be useful tools in modeling aging using transcriptomics data. According to The Elements of Statistical

Learning by Hastie et al. [206], support vector regression models are non-probabilistic binary linear classifiers that function by constructing a single hyperplane or numerous hyperplanes in order to separate data by the hyperplane with the greatest distance to the closest data point for each class, thus allowing for successful regression, classification, or outlier detection models

[207]. While these methods employing support vector machines might be useful in modeling aging, they are susceptible to multicollinearity – where two variables are very highly correlated with each other – suggesting the need for improved methods to model aging that are not susceptible to the large degree of multicollinearity common among genetic data.

Other groups that examined the transcriptome from blood [149], skin, adipose tissue, blood, and brain have demonstrated that transcriptomic data is also effective at predicting age using weighted Cox regression models [202]. In skin fibroblasts, Fleischer et al. [208] demonstrated that using all transcripts from 144 cell culture samples can be effectively modelled to predict chronological age using ensemble linear discriminant analysis (LDA) classifiers can accurately predict chronological age in humans within eight years. Similarly, Horvath et al.

[209] predicted age within four years using DNA methylation data while Putin et al. [210] utilized a neural network with blood cytology data to predict age within 5.5 years. While combining machine learning with transcriptomic data to is a nascent area of research, additional studies combining these fields will enable unique opportunities to identify gene expression patterns underlying health and disease. Additionally, advancements in statistical normalization across for examining transcriptomic data across studies will allow improve our understanding of how gene expression networks are altered by aging, chronic diseases or T2DM.

Role of MicroRNAs in Health & Disease

An emerging class of RNA molecules that affects the transcriptome are microRNAs

(microRNAs), which are 17-25 nucleotide long small RNA molecules that have been shown to play a role in regulating gene expression by decreasing gene expression on their target genes without modifying the gene sequence [211]. microRNAs are non-coding molecules whose biogenesis arises from processing in the nucleus by the protein DROSHA, followed by exportation from the nucleus by Exportin 5, and are converted into mature microRNA sequences in the cytoplasm by the protein DICER [212]. When microRNA status or biosynthesis is dysregulated, they can contribute to a variety of diseases [212] and aging [213]. MicroRNAs have been suggested to regulate aging through modification of lifespan [214], cellular senescence [215], or modifying inflammation [216]. microRNAs have recently been shown to regulate lifespan in C. elegans by altering insulin and insulin-like growth factor-1(IGF1) [217].

Similarly, in humans, circulating microRNA expression of 80 microRNAs in the blood has been associated with biological age and strongly predicts all-cause mortality [218]. In the mouse brain, most of the previously identified microRNAs decrease with age [219], while in rats this same phenomenon does not occur [220]. A lot of microRNA sequences are highly conserved across multiple species [143,221–223] and there is an emerging body of evidence suggesting that studying the transcriptomic response to aging may lead to development of new therapeutics for aging-associated chronic diseases such as Alzheimer’s disease [224].

Additional evidence is emerging that microRNAs may modify the aging process [225], and that influencing microRNA status may be a novel method of combatting aging to extend lifespan. In C. elegans, it has been shown that altering the expression of lin-4 or reducing lin-14 expression can extend lifespan through the insulin like growth factor pathway [226]. Similarly,

30 regulation of the argonaute protein family in C. elegans can extend lifespan by regulating expression of alg-1 and alg-2 which also function through the insulin signaling pathway. In mice models of premature aging, microRNA status alterations that affect IGF1 have similarly been shown to extend lifespan [227–229]. Due to their influential role in regulating numerous biological pathways [149], microRNA expression levels can be used to effectively predict aging

[230]. While microRNAs may provide useful insight into understanding health and disease, their ability to regulate health and disease has yet to be fully understood.

Outside of aging and aging-associated diseases, dietary patterns have also been thought to influence circulating microRNA status in human plasma and stool samples [231]. While some groups suggest that these microRNAs in colorectal tissues are changing in result of specific dietary patterns [232], others suggest that changes in circulating microRNAs may be due to specific bioactive nutrients such as fatty acids [233], vitamins [234], or minerals [235]. Zheng et al. demonstrated that in three-week old rats fed diets differing in poly unsaturated fatty Acid content altered expression of hsa-microRNA-183-5p, hsa-microRNA-19b-3p, and microRNA -

146b-5p [233]. In addition to dietary patterns and specific nutrients altering microRNA status, some groups have sparked controversy surrounding the idea that food derived microRNAs might be absorbed via the intestine [236]. Baier et al. examined this in 2014 with healthy older adults who consumed broccoli and did not observe any changes in broccoli-derived microRNAs [237].

While the controversy remains surrounding if food derived microRNAs can be absorbed intact

[32,35,237], or not [37,40]; the underlying literature has demonstrated an important role of diet in regulating gene and microRNA expression [238].

Recent data is suggesting that microRNAs also play a role in altering the transcriptome during diabetes [239]. Since microRNAs regulate gene expression through post-transcriptional modification, disruption of microRNA has been proposed as a common mechanism occurring across multiple diseases [87], especially in metabolic [86] and neurodegenerative diseases [84].

Briefly, the primary microRNAs that have been associated with diabetes are thought to regulate gene expression networks underlying glucose metabolism, and insulin signaling [82]. MicroRNA profiles in the serum of human subjects with diabetes display altered hsa-microRNA-365a-3p, hsa-microRNA-5190, hsa-microRNA-770-5p, and hsa-microRNA-125b-5p concentrations and the expression levels of these microRNAs strongly correlate with hemoglobin A1C, a long term marker of blood sugar levels [82]. Similarly, serum concentrations of hsa-microRNA-192 and hsa-microRNA-194 have been positively correlated with diabetes incidence [82], while other groups have demonstrated that microRNAs in plasma such as hsa-microRNA-150, hsa- microRNA-15a, or hsa-microRNA-375 can be used to identify subjects with pre-diabetes or

T2DM [82]. Overall these studies support the hypothesis that diet and aging strongly impact gene expression and that by improving our understanding of how lifestyle can affect the transcriptome, we can better tailor personalized health recommendations for the prevention or treatment of metabolic or aging associated diseases. In the subsequent chapters of this dissertation, this central idea will be explored using in vivo experiments alongside computational methods to better understand how whole egg and aging affect the transcriptome across multiple tissues in animal models.

References

1. Hajat C, Stein E. The global burden of multiple chronic conditions: A narrative review. Preventive Medicine Reports. Elsevier Inc.; 2018. pp. 284–293. doi:10.1016/j.pmedr.2018.10.008

2. Hwang W, Weller W, Ireys H, Anderson G. Out-Of-Pocket Medical Spending For Care Of Chronic Conditions. 2001.

3. Raghupathi W, Raghupathi V. An empirical study of chronic diseases in the united states: A visual analytics approach. International Journal of Environmental Research and Public Health. 2018;15. doi:10.3390/ijerph15030431

4. Dall TM, Gallo PD, Chakrabarti R, West T, Semilla AP, Storm M v. An Aging Population And Growing Disease Burden Will Require A Large And Specialized Health Care Workforce By 2025. doi:10.1377/hlthaff.2013.0714

5. Hung WW, Ross JS, Boockvar KS, Siu AL. Recent trends in chronic disease, impairment and disability among older adults in the United States. BMC Geriatrics. 2011;11: 47. doi:10.1186/1471-2318-11-47

6. MacNee W, Rabinovich RA, Choudhury G. Ageing and the border between health and Disease. European Respiratory Journal. 2014;44: 1332–1352. doi:10.1183/09031936.00134014

7. 2019 ALZHEIMER’S DISEASE FACTS AND FIGURES Includes a Special Report on Alzheimer’s Detection in the Primary Care Setting: Connecting Patients and Physicians.

8. Snijders C, Matzat U, Reips U-D. “Big Data”: Big Gaps of Knowledge in the Field of Internet Science. International Journal of Internet Science. 2012;2012: 1–5. Available: http://wikipedia- miner.cms.waikato.ac.nz/

9. Roberto Zicari. Big Data for Good. | ODBMS Industry Watch. In: OBDMS Industry Watch [Internet]. 4 Jun 2012 [cited 15 Mar 2020]. Available: http://www.odbms.org/blog/2012/06/big- data-for-good/

10. Lander ES, Linton LM, Birren B, Nusbaum C, Zody MC, Baldwin J, et al. Initial sequencing and analysis of the human genome. Nature. 2001;409: 860–921. doi:10.1038/35057062

11. Jung KC, Sang CK. Environmental effects on gene expression phenotype have regional biases in the human genome. Genetics. 2007;175: 1607–1613. doi:10.1534/genetics.106.069047

12. Foo JC, Trautmann N, Sticht C, Treutlein J, Frank J, Streit F, et al. Longitudinal transcriptome- wide gene expression analysis of sleep deprivation treatment shows involvement of circadian genes and immune pathways. Translational Psychiatry. 2019;9: 1–10. doi:10.1038/s41398-019- 0671-7

13. Brody JS. Transcriptome alterations induced by cigarette smoke. International Journal of Cancer. 2012;131: 2754–2762. doi:10.1002/ijc.27829

14. Hasan MS, Feugang JM, Liao SF. A Nutrigenomics Approach Using RNA Sequencing Technology to Study Nutrient–Gene Interactions in Agricultural Animals. Current Developments in Nutrition. 2019;3. doi:10.1093/CDN/NZZ082

15. Armenise C, Lefebvre G, Carayol J, Bonnel S, Bolton J, di Cara A, et al. Transcriptome profiling from adipose tissue during a low-calorie diet reveals predictors of weight and glycemic outcomes in obese, nondiabetic subjects. The American Journal of Clinical Nutrition. 2017;106: ajcn156216. doi:10.3945/ajcn.117.156216

16. Arnal C, Lou-Bonafonte JM, Martínez-Gracia M v., Rodríguez-Yoldi MJ, Osada J. Transcriptomics and Nutrition in Mammalians. Genomics, Proteomics and Metabolomics in Nutraceuticals and Functional Foods: Second Edition. Chichester, UK: Wiley Blackwell; 2015. pp. 581–608. doi:10.1002/9781118930458.ch46

17. Blusztajn JK, Slack BE, Mellott TJ. Neuroprotective actions of dietary choline. Nutrients. MDPI AG; 2017. doi:10.3390/nu9080815

18. Sharma P, Dwivedi S. Nutrigenomics and Nutrigenetics: New Insight in Disease Prevention and Cure. Indian Journal of Clinical Biochemistry. Springer India; 2017. pp. 371–373. doi:10.1007/s12291-017-0699-5

19. Athanasopoulos D, Karagiannis G, Tsolaki M. Recent Findings in Alzheimer Disease and Nutrition Focusing on Epigenetics. Advances in Nutrition. 2016;7: 917–927. doi:10.3945/an.116.012229

20. Saande CJ, Webb JL, Curry PE, Rowling MJ, Schalinske KL. Dietary Whole Egg Reduces Body Weight Gain in a Dose-Dependent Manner in Zucker Diabetic Fatty Rats. The Journal of Nutrition. 2019;149: 1766–1775. doi:10.1093/jn/nxz143

21. Réhault-Godbert S, Guyot N, Nys Y. The golden egg: Nutritional value, bioactivities, and emerging benefits for human health. Nutrients. MDPI AG; 2019. doi:10.3390/nu11030684

22. Drewnowski A. The nutrient rich foods index helps to identify healthy, affordable foods. American Journal of Clinical Nutrition. 2010;91. doi:10.3945/ajcn.2010.28450D

23. Tukur HM. Egg production in Africa. Improving the Safety and Quality of Eggs and Egg Products: Egg Chemistry, Production and Consumption. Elsevier Inc.; 2011. pp. 27–38. doi:10.1533/9780857093912.1.27

24. Atlas of the Global Egg Industry.

25. Elaine Watson. US per capita egg consumption continues to rise as protein craze continues, says egg board. In: Food Navigator [Internet]. 29 Mar 2018 [cited 15 Mar 2020]. Available: https://www.foodnavigator-usa.com/Article/2018/03/29/US-per-capita-egg-consumption- continues-to-rise-as-protein-craze-continues-says-egg-board

26. Per capita consumption of eggs in the U.S. 2020 | Statista. [cited 27 Feb 2020]. Available: https://www.statista.com/statistics/183678/per-capita-consumption-of-eggs-in-the-us-since-2000/

27. Park SJ, Jung JH, Choi SW, Lee HJ. Association between egg consumption and metabolic disease. Korean Journal for Food Science of Animal Resources. Korean Society for Food Science of Animal Resources; 2018. pp. 209–223. doi:10.5851/kosfa.2018.38.2.209

28. Health Canada. What are Functional Foods and Nutraceuticals? 2009 Apr. Available: http://www4.agr.gc.ca/AAFC-AAC/display-afficher.do?id=1171305207040

29. Andersen C. Bioactive Egg Components and Inflammation. Nutrients. 2015;7: 7889–7913. doi:10.3390/nu7095372

30. Yousr M, Howell N. Antioxidant and ACE inhibitory bioactive peptides purified from egg yolk proteins. International Journal of Molecular Sciences. 2015;16: 29161–29178. doi:10.3390/ijms161226155

31. Wiedeman AM, Barr SI, Green TJ, Xu Z, Innis SM, Kitts DD. Dietary choline intake: Current state of knowledge across the life cycle. Nutrients. MDPI AG; 2018. doi:10.3390/nu10101513

32. Zempleni J, Baier SR, Howard KM, Cui J. Gene regulation by dietary microRNAs. Canadian Journal of Physiology and Pharmacology. Canadian Science Publishing; 2015. pp. 1097–1102. doi:10.1139/cjpp-2014-0392

33. Yousr M, Howell N. Antioxidant and ACE inhibitory bioactive peptides purified from egg yolk proteins. International Journal of Molecular Sciences. 2015;16: 29161–29178. doi:10.3390/ijms161226155

34. Zempleni J. Milk exosomes: beyond dietary microRNAs. Genes and Nutrition. BioMed Central Ltd.; 2017. doi:10.1186/s12263-017-0562-6

35. Baier S, Howard K, Cui J, Shu J, Zempleni J. MicroRNAs in chicken eggs are bioavailable in healthy adults and can modulate mRNA expression in peripheral blood mononuclear cells. The FASEB Journal. 2015;29: LB322. doi:10.1096/fasebj.29.1_supplement.lb322

36. Wagner AE, Piegholdt S, Ferraro M, Pallauf K, Rimbach G. Food derived microRNAs. Food and Function. 2015;6: 714–718. doi:10.1039/c4fo01119h

37. Witwer KW, Zhang CY. Diet-derived microRNAs: Unicorn or silver bullet? Genes and Nutrition. BioMed Central Ltd.; 2017. doi:10.1186/s12263-017-0564-4

38. Nutrient Composition Tables - American Egg Board. [cited 15 Mar 2020]. Available: https://www.aeb.org/food-manufacturers/eggs-product-overview/nutrient-composition-tables

39. Fernandez M. Eggs and Health Special Issue. Nutrients. 2016;8: 784. doi:10.3390/nu8120784

40. Cottrill KA, Chan SY. Diet-derived MicroRNAs: Separating the dream from reality. MicroRNA Diagnostics and Therapeutics. De Gruyter Open Ltd; 2014. pp. 46–57. doi:10.2478/micrnat-2014-0001

41. Otsuka K, Yamamoto Y, Matsuoka R, Ochiya T. Maintaining good microRNAs in the body keeps the doctor away?: Perspectives on the relationship between food-derived natural products and microRNAs in relation to exosomes/extracellular vesicles. Molecular Nutrition & Food Research. 2018;62: 1700080. doi:10.1002/mnfr.201700080

42. Kuang H, Yang F, Zhang Y, Wang T, Chen G. The Impact of Egg Nutrient Composition and Its Consumption on Cholesterol Homeostasis. Cholesterol. 2018;2018. doi:10.1155/2018/6303810

43. Lemos BS, Medina-Vera I, Blesso CN, Fernandez ML. Intake of 3 eggs per day when compared to a choline bitartrate supplement, downregulates cholesterol synthesis without changing the LDL/HDL ratio. Nutrients. 2018;10. doi:10.3390/nu10020258

44. Nimalaratne C, Wu J. Hen egg as an antioxidant food commodity: A review. Nutrients. MDPI AG; 2015. pp. 8274–8293. doi:10.3390/nu7105394

45. Hayes A, Duffy S, O’Grady M, Jakobsen J, Galvin K, Teahan-Dillon J, et al. Vitamin D– enhanced eggs are protective of wintertime serum 25-hydroxyvitamin D in a randomized controlled trial of adults. The American Journal of Clinical Nutrition. 2016;104: 629–637. doi:10.3945/ajcn.116.132530

46. Knekt P, Kilkkinen A, Rissanen H, Marniemi J, Sääksjärvi K, Heliövaara M. Serum vitamin D and the risk of Parkinson disease. Archives of Neurology. 2010;67: 808–811. doi:10.1001/archneurol.2010.120

47. Carlberg C. Nutrigenomics of Vitamin D. Nutrients. 2019;11: 676. doi:10.3390/nu11030676

48. Parva NR, Tadepalli S, Singh P, Qian A, Joshi R, Kandala H, et al. Prevalence of Vitamin D Deficiency and Associated Risk Factors in the US Population (2011-2012). Cureus. 2018;10. doi:10.7759/cureus.2741

49. PAULA BORDELON D, MARIA V. GHETU M, ROBERT LANGAN M. Recognition and Management of Vitamin D Deficiency - American Family Physician. In: Am Fam Physician [Internet]. 15 Oct 2009 [cited 17 Mar 2020] pp. 841–846. Available: https://www.aafp.org/afp/2009/1015/p841.html

50. Saande CJ, Jones SK, Hahn KE, Reed CH, Rowling MJ, Schalinske KL. Dietary Whole Egg Consumption Attenuates Body Weight Gain and Is More Effective than Supplemental Cholecalciferol in Maintaining Vitamin D Balance in Type 2 Diabetic Rats. The Journal of Nutrition. 2017; jn254193. doi:10.3945/jn.117.254193

51. Saande CJ, Pritchard SK, Worrall DM, Snavely SE, Nass CA, Neuman JC, et al. Dietary egg protein prevents hyperhomocysteinemia via upregulation of hepatic betaine-homocysteine s- methyltransferase activity in folate-restricted rats. Journal of Nutrition. 2019;149: 1369–1376. doi:10.1093/jn/nxz069

52. Wang Z, Klipfell E, Bennett BJ, Koeth R, Levison BS, Dugar B, et al. Gut flora metabolism of phosphatidylcholine promotes cardiovascular disease. Nature. 2011;472: 57–65. doi:10.1038/nature09922

53. Saande CJ, Pritchard SK, Worrall DM, Snavely SE, Nass CA, Neuman JC, et al. Dietary Egg Protein Prevents Hyperhomocysteinemia via Upregulation of Hepatic Betaine-Homocysteine S- Methyltransferase Activity in Folate-Restricted Rats. The Journal of nutrition. 2019;149: 1369– 1376. doi:10.1093/jn/nxz069

54. Altic L, McNulty H, Hoey L, McAnena L, Pentieva K. Validation of folate-enriched eggs as a functional food for improving folate intake in consumers. Nutrients. 2016;8. doi:10.3390/nu8120777

55. Tran NL, Barraj LM, Heilman JM, Scrafford CG. Egg consumption and cardiovascular disease among diabetic individuals: A systematic review of the literature. Diabetes, Metabolic Syndrome and Obesity: Targets and Therapy. Dove Medical Press Ltd.; 2014. pp. 121–137. doi:10.2147/DMSO.S58668

56. Fuller NR, Sainsbury A, Caterson ID, Markovi TP. Egg consumption and human cardiometabolic health in people with and without diabetes. Nutrients. MDPI AG; 2015. pp. 7399– 7420. doi:10.3390/nu7095344

57. Zhong VW, van Horn L, Cornelis MC, Wilkins JT, Ning H, Carnethon MR, et al. Associations of Dietary Cholesterol or Egg Consumption with Incident Cardiovascular Disease and Mortality. JAMA - Journal of the American Medical Association. 2019;321: 1081–1095. doi:10.1001/jama.2019.1572

58. Kim JE, Campbell WW. Dietary cholesterol contained in whole eggs is not well absorbed and does not acutely affect plasma total cholesterol concentration in men and women: Results from 2 randomized controlled crossover studies. Nutrients. 2018;10. doi:10.3390/nu10091272

59. Khawaja O, Singh H, Luni F, Kabour A, Ali SS, Taleb M, et al. Egg Consumption and Incidence of Heart Failure: A Meta-Analysis of Prospective Cohort Studies. Frontiers in Nutrition. 2017;4: 10. doi:10.3389/fnut.2017.00010

60. Rong Y, Chen L, Zhu T, Song Y, Yu M, Shan Z, et al. Egg consumption and risk of coronary heart disease and stroke: Dose-response meta-analysis of prospective cohort studies. BMJ (Online). 2013;346. doi:10.1136/bmj.e8539

61. Li Y, Zhou C, Zhou X, Li L. Egg consumption and risk of cardiovascular diseases and diabetes: A meta-analysis. Atherosclerosis. 2013;229: 524–530. doi:10.1016/j.atherosclerosis.2013.04.003

62. Xu L, Lam TH, Jiang CQ, Zhang W sen, Zhu F, Jin YL, et al. Egg consumption and the risk of cardiovascular disease and all-cause mortality: Guangzhou Biobank Cohort Study and meta- analyses. European Journal of Nutrition. 2019;58: 785–796. doi:10.1007/s00394-018-1692-3

63. Rong Y, Chen L, Zhu T, Song Y, Yu M, Shan Z, et al. Egg consumption and risk of coronary heart disease and stroke: Dose-response meta-analysis of prospective cohort studies. BMJ (Online). 2013;346: e8539. doi:10.1136/bmj.e8539

64. Alexander DD, Miller PE, Vargas AJ, Weed DL, Cohen SS. Meta-analysis of Egg Consumption and Risk of Coronary Heart Disease and Stroke. Journal of the American College of Nutrition. Routledge; 2016. pp. 704–716. doi:10.1080/07315724.2016.1152928

65. Shin JY, Xun P, Nakamura Y, He K. Egg consumption in relation to risk of cardiovascular disease and diabetes: A systematic review and meta-analysis. American Journal of Clinical Nutrition. 2013;98: 146–159. doi:10.3945/ajcn.112.051318

66. Liu H, Zheng F, Cao Q, Ren B, Zhu L, Striker G, et al. Amelioration of oxidant stress by the defensin lysozyme. American Journal of Physiology - Endocrinology and Metabolism. 2006;290: E824-32. doi:10.1152/ajpendo.00349.2005

67. Ibrahim HR, Hoq MI, Aoki T. Ovotransferrin possesses SOD-like superoxide anion scavenging activity that is promoted by copper and manganese binding. International Journal of Biological Macromolecules. 2007;41: 631–640. doi:10.1016/j.ijbiomac.2007.08.005

68. Egg Bioscience and Biotechnology - Google Books. [cited 17 Mar 2020]. Available: https://books.google.com/books?hl=en&lr=&id=OnZNqXI8zpoC&oi=fnd&pg=PR5&ots=V6mx RRPIKu&sig=ZMx4J759l8ly6vWUcLbBMP5zsjY#v=onepage&q&f=false

69. Leeson S, Caston L. Enrichment of eggs with lutein. Poultry science. 2004;83: 1709–12. doi:10.1093/ps/83.10.1709

70. Plassman BL, Langa KM, Fisher GG, Heeringa SG, Weir DR, Ofstedal MB, et al. Prevalence of dementia in the United States: The aging, demographics, and memory study. Neuroepidemiology. 2007;29: 125–132. doi:10.1159/000109998

71. Morris MC. The role of nutrition in Alzheimer’s disease: Epidemiological evidence. European Journal of Neurology. 2009. pp. 1–7. doi:10.1111/j.1468-1331.2009.02735.x

72. Shin JY, Xun P, Nakamura Y, He K. Egg consumption in relation to risk of cardiovascular disease and diabetes: A systematic review and meta-analysis. American Journal of Clinical Nutrition. 2013;98: 146–159. doi:10.3945/ajcn.112.051318

73. CAMPBELL-THOMPSON M, BUTTERWORTH EA, LENCHIK NI, ATKINSON MA, MATHEWS CE, GERLING IC. Analysis of Transcriptome Data from 261 Individual Laser- Captured Islets from Nondiabetic, Autoantibody-Positive, and Type 1 Diabetic Organ Donors. Diabetes. 2018;67: 308-LB. doi:10.2337/db18-308-lb

74. Li JW, Lee HM, Wang Y, Tong AHY, Yip KY, Tsui SKW, et al. Interactome-transcriptome analysis discovers signatures complementary to GWAS Loci of Type 2 Diabetes. Scientific Reports. 2016;6: 1–14. doi:10.1038/srep35228

75. Jenkinson CP, Göring HHH, Arya R, Blangero J, Duggirala R, DeFronzo RA. Transcriptomics in type 2 diabetes: Bridging the gap between genotype and phenotype. Genomics Data. 2016;8: 25–36. doi:10.1016/j.gdata.2015.12.001

76. Wen L, Zhang Z, Peng R, Zhang L, Liu H, Peng H, et al. Whole transcriptome analysis of diabetic nephropathy in the db/db mouse model of type 2 diabetes. Journal of Cellular Biochemistry. 2019;120: 17520–17533. doi:10.1002/jcb.29016

77. Cui Y, Chen W, Chi J, Wang L. Comparison of transcriptome between type 2 diabetes mellitus and impaired fasting glucose. Medical Science Monitor. 2016;22: 4699–4706. doi:10.12659/MSM.896772

78. Bries A, Webb J, Curry P, Fischer S, McNeill E, Clark S, et al. Acute Effects of Whole Egg Consumption on Circulating MicroRNAs in Sprague Dawley Animals (P15-001-19). Current Developments in Nutrition. 2019;3. doi:10.1093/CDN/NZZ037.P15-001-19

79. Christodoulou MI, Avgeris M, Kokkinopoulou I, Maratou E, Mitrou P, Kontos CK, et al. Blood-based analysis of type-2 diabetes mellitus susceptibility genes identifies specific transcript variants with deregulated expression and association with disease risk. Scientific Reports. 2019;9: 1–14. doi:10.1038/s41598-018-37856-1

80. Neelankal John A, Ram R, Jiang FX. RNA-Seq Analysis of Islets to Characterise the Dedifferentiation in Type 2 Diabetes Model Mice db/db. Endocrine Pathology. 2018;29: 207– 221. doi:10.1007/s12022-018-9523-x

81. Mohr S, Liew CC. The peripheral-blood transcriptome: new insights into disease and risk assessment. Trends in Molecular Medicine. 2007;13: 422–432. doi:10.1016/j.molmed.2007.08.003

82. Jiménez-Lucena R, Camargo A, Alcalá-Diaz JF, Romero-Baldonado C, Luque RM, van Ommen B, et al. A plasma circulating microRNAs profile predicts type 2 diabetes mellitus and prediabetes: from the CORDIOPREV study. Experimental and Molecular Medicine. 2018;50: 1– 12. doi:10.1038/s12276-018-0194-y

83. Bartel DP. MicroRNAs: Genomics, Biogenesis, Mechanism, and Function. Cell. Cell Press; 2004. pp. 281–297. doi:10.1016/S0092-8674(04)00045-5

84. Jaeger A, Zollinger L, Saely CH, Muendlein A, Evangelakos I, Nasias D, et al. Circulating microRNAs -192 and -194 are associated with the presence and incidence of diabetes mellitus. Scientific Reports. 2018;8: 1–14. doi:10.1038/s41598-018-32274-9

85. Application of microRNAs in diabetes mellitus in: Journal of Endocrinology Volume 222 Issue 1 (2014). [cited 18 Mar 2020]. Available: https://joe.bioscientifica.com/view/journals/joe/222/1/R1.xml

86. Satake E, Pezzolesi MG, Dom ZIM, Smiles AM, Niewczas MA, Krolewski AS. Circulating microRNA profiles associated with hyperglycemia in patients with type 1 diabetes. Diabetes. 2018;67: 1013–1023. doi:10.2337/db17-1207

87. Pordzik J, Jakubik D, Jarosz-Popek J, Wicik Z, Eyileten C, de Rosa S, et al. Significance of circulating microRNAs in diabetes mellitus type 2 and platelet reactivity: Bioinformatic analysis and review. Cardiovascular Diabetology. BioMed Central Ltd.; 2019. p. 113. doi:10.1186/s12933-019-0918-x

88. Feng J, Xing W, Xie L. Regulatory roles of microRNAs in diabetes. International Journal of Molecular Sciences. MDPI AG; 2016. doi:10.3390/ijms17101729

89. Iacomino G, Siani A. Role of microRNAs in obesity and obesity-related diseases. Genes and Nutrition. BioMed Central Ltd.; 2017. pp. 1–16. doi:10.1186/s12263-017-0577-z

90. Fernández-Hernando C, Ramírez CM, Goedeke L, Suárez Y. MicroRNAs in metabolic disease. Arteriosclerosis, Thrombosis, and Vascular Biology. 2013;33: 178–185. doi:10.1161/ATVBAHA.112.300144

91. Williamson JD, Pajewski NM, Auchus AP, Bryan RN, Chelune G, Cheung AK, et al. Effect of Intensive vs Standard Blood Pressure Control on Probable Dementia: A Randomized Clinical Trial. JAMA - Journal of the American Medical Association. American Medical Association; 2019. pp. 553–561. doi:10.1001/jama.2018.21442

92. Gardener H, Caunca MR. Mediterranean Diet in Preventing Neurodegenerative Diseases. Current Nutrition Reports. Current Science Inc.; 2018. pp. 10–20. doi:10.1007/s13668-018- 0222-5

93. Littlejohns TJ, Henley WE, Lang IA, Annweiler C, Beauchet O, Chaves PHM, et al. Vitamin D and the risk of dementia and Alzheimer disease. Neurology. 2014;83: 920–8. doi:10.1212/WNL.0000000000000755

94. Eyles DW, Burne THJ, McGrath JJ. Vitamin D, effects on brain development, adult brain function and the links between low levels of vitamin D and neuropsychiatric disease. Frontiers in Neuroendocrinology. 2013. pp. 47–64. doi:10.1016/j.yfrne.2012.07.001

95. Abdollahi AM, Virtanen HEK, Voutilainen S, Kurl S, Tuomainen TP, Salonen JT, et al. Egg consumption, cholesterol intake, and risk of incident stroke in men: the Kuopio Ischaemic Heart Disease Risk Factor Study. American Journal of Clinical Nutrition. 2019;110: 169–176. doi:10.1093/ajcn/nqz066

96. An Y, Zhang X, Wang Y, Wang Y, Liu W, Wang T, et al. Longitudinal and nonlinear relations of dietary and Serum cholesterol in midlife with cognitive decline: Results from EMCOA study. Molecular Neurodegeneration. 2019;14: 51. doi:10.1186/s13024-019-0353-1

97. Gallagher JC. Vitamin D and Aging. Endocrinology and Metabolism Clinics of North America. NIH Public Access; 2013. pp. 319–332. doi:10.1016/j.ecl.2013.02.004

98. Koduah P, Paul F, Dörr JM. Vitamin D in the prevention, prediction and treatment of neurodegenerative and neuroinflammatory diseases. EPMA Journal. Springer International Publishing; 2017. pp. 313–325. doi:10.1007/s13167-017-0120-8

99. Saande CJ, Jones SK, Hahn KE, Reed CH, Rowling MJ, Schalinske KL. Dietary Whole Egg Consumption Attenuates Body Weight Gain and Is More Effective than Supplemental Cholecalciferol in Maintaining Vitamin D Balance in Type 2 Diabetic Rats. The Journal of Nutrition. 2017; jn254193. doi:10.3945/jn.117.254193

100. Fall PA, Fredrikson M, Axelson O, Granerus AK. Nutritional and occupational factors influencing the risk of Parkinson’s disease: a case-control study in southeastern Sweden. Movement disorders : official journal of the Movement Disorder Society. 1999;14: 28–37.

101. Ma L, Zhang L, Gao XH, Chen W, Wu YP, Wang Y, et al. Dietary factors and smoking as risk factors for PD in a rural population in China: a nested case-control study. Acta Neurologica Scandinavica. 2006;113: 278–281. doi:10.1111/j.1600-0404.2005.00571.x

102. Mana R. Ehlers, Rebecca M. Todd. Genesis and Maintenance of Attentional Biases: The Role of the Locus Coeruleus-Noradrenaline System. Neural Plasticity. 2017;1: 2–3. doi:10.1155/2017

103. Johnson SC, Dong X, Vijg J, Suh Y. Genetic evidence for common pathways in human age- related diseases. Aging Cell. 2015;14: 809–817. doi:10.1111/acel.12362

104. Mischley LK, Lau RC, Bennett RD. Role of Diet and Nutritional Supplements in Parkinson’s Disease Progression. Oxidative medicine and cellular longevity. 2017/09/10. 2017;2017: 6405278. doi:10.1155/2017/6405278

105. Zeng L, Yang J, Peng S, Zhu J, Zhang B, Suh Y, et al. Transcriptome analysis reveals the difference between “healthy” and “common” aging and their connection with age-related diseases. Aging Cell. 2020;19. doi:10.1111/acel.13121

106. Lai RW, Lu R, Danthi PS, Bravo JI, Goumba A, Sampathkumar NK, et al. Multi-level remodeling of transcriptional landscapes in aging and longevity. BMB Reports. The Biochemical Society of the Republic of Korea; 2019. pp. 86–108. doi:10.5483/BMBRep.2019.52.1.296

107. Sampathkumar NK, Bravo JI, Chen Y, Danthi PS, Donahue EK, Lai RW, et al. Widespread sex dimorphism in aging and age-related diseases. Human Genetics. Springer; 2020. pp. 333– 356. doi:10.1007/s00439-019-02082-w

108. He Y, Jasper H. Studying aging in Drosophila. Methods. 2014;68: 129–133. doi:10.1016/j.ymeth.2014.04.008

109. Piper MDW, Partridge L. Drosophila as a model for ageing. Biochimica et Biophysica Acta - Molecular Basis of Disease. Elsevier B.V.; 2018. pp. 2707–2717. doi:10.1016/j.bbadis.2017.09.016

110. Edrey YH, Hanes M, Pinto M, Mele J, Buffenstein R. Successful aging and sustained good health in the naked mole rat: A long-lived mammalian model for biogerontology and biomedical research. ILAR Journal. 2011;52: 41–53. doi:10.1093/ilar.52.1.41

111. Kohama SG, Rosene DL, Sherman LS. Age-related changes in human and non-human primate white matter: From myelination disturbances to cognitive decline. Age. Springer; 2012. pp. 1093–1110. doi:10.1007/s11357-011-9357-7

112. Hof PR, Gilissen EP, Sherwood CC, Duan H, Lee PWH, Delman BN, et al. Comparative Neuropathology of Brain Aging in Primates. Aging in Nonhuman Primates. Basel: KARGER; 2002. pp. 130–154. doi:10.1159/000061462

113. Loerch PM, Lu T, Dakin KA, Vann JM, Isaacs A, Geula C, et al. Evolution of the Aging Brain Transcriptome and Synaptic Regulation. Orban L, editor. PLoS ONE. 2008;3: e3329. doi:10.1371/journal.pone.0003329

114. Kim SK. Common aging pathways in worms, flies, mice and humans. Journal of Experimental Biology. The Company of Biologists Ltd; 2007. pp. 1607–1612. doi:10.1242/jeb.004887

115. Piriz J, Muller A, Trejo JL, Torres-Aleman I. IGF-I and the aging mammalian brain. Experimental Gerontology. 2011;46: 96–99. doi:10.1016/j.exger.2010.08.022

116. Stefanova NA, Ershov NI, Maksimova KY, Muraleva NA, Tyumentsev MA, Kolosova NG. The Rat Prefrontal-Cortex Transcriptome: Effects of Aging and Sporadic Alzheimer’s Disease- Like Pathology. The journals of gerontology Series A, Biological sciences and medical sciences. 2019;74: 33–43. doi:10.1093/gerona/gly198

117. González-Velasco O, Papy-García D, le Douaron G, Sánchez-Santos JM, de Las Rivas J. Transcriptomic landscape, gene signatures and regulatory profile of aging in the human brain. Biochimica et Biophysica Acta - Gene Regulatory Mechanisms. 2020; 194491. doi:10.1016/j.bbagrm.2020.194491

118. Kang HJ, Kawasawa YI, Cheng F, Zhu Y, Xu X, Li M, et al. Spatio-temporal transcriptome of the human brain. Nature. 2011;478: 483–489. doi:10.1038/nature10523

119. Torello LB, Yates AJ, Hart R, Leon KS. A comparative-evolutionary study of lipids in the aging brain of mice. Neurobiology of Aging. 1986;7: 337–346. doi:10.1016/0197- 4580(86)90160-0

120. Morterá P, Herculano-Houzel S. Age-related neuronal loss in the rat brain starts at the end of adolescence. Frontiers in Neuroanatomy. 2012;6. doi:10.3389/fnana.2012.00045

121. Gray DT, Barnes CA. Experiments in macaque monkeys provide critical insights into age- associated changes in cognitive and sensory function. Proceedings of the National Academy of Sciences of the United States of America. 2019;116: 26247–26254. doi:10.1073/pnas.1902279116

122. Bishop NA, Lu T, Yankner BA. Neural mechanisms of ageing and cognitive decline. Nature. NIH Public Access; 2010. pp. 529–535. doi:10.1038/nature08983

123. Komljenovic A, Li H, Sorrentino V, Kutalik Z, Auwerx J, Robinson-Rechavi M. Cross-species functional modules link proteostasis to human normal aging. PLoS Computational Biology. 2019;15: e1007162. doi:10.1371/journal.pcbi.1007162

124. Calì C, Wawrzyniak M, Becker C, Maco B, Cantoni M, Jorstad A, et al. The effects of aging on neuropil structure in mouse somatosensory cortex—A 3D electron microscopy analysis of layer 1. Ginsberg SD, editor. PLOS ONE. 2018;13: e0198131. doi:10.1371/journal.pone.0198131

125. Genome, transcriptome and proteome: the rise of omics data and their integration in biomedical sciences. [cited 19 Mar 2020]. Available: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6018996/

126. Chatsirisupachai K, Palmer D, Ferreira S, de Magalhães JP. A human tissue-specific transcriptomic analysis reveals a complex relationship between aging, cancer, and cellular senescence. Aging Cell. 2019;18: e13041. doi:10.1111/acel.13041

127. Suntsova M, Gaifullin N, Allina D, Reshetun A, Li X, Mendeleeva L, et al. Atlas of RNA sequencing profiles for normal human tissues. Scientific data. 2019;6: 36. doi:10.1038/s41597- 019-0043-4

128. Kusunose K, Penn MS, Zhang Y, Cheng Y, Thomas JD, Marwick TH, et al. How Similar Are the Mice to Men? Between-Species Comparison of Left Ventricular Mechanics Using Strain Imaging. Bauer WR, editor. PLoS ONE. 2012;7: e40061. doi:10.1371/journal.pone.0040061

129. Darmanis S, Sloan SA, Zhang Y, Enge M, Caneda C, Shuer LM, et al. A survey of human brain transcriptome diversity at the single cell level. Proceedings of the National Academy of Sciences of the United States of America. 2015;112: 7285–7290. doi:10.1073/pnas.1507125112

130. Magistri M, Velmeshev D, Makhmutova M, Faghihi MA. Transcriptomics Profiling of Alzheimer’s Disease Reveal Neurovascular Defects, Altered Amyloid-β Homeostasis, and Deregulated Expression of Long Noncoding RNAs. Journal of Alzheimer’s Disease. 2015;48: 647–665. doi:10.3233/JAD-150398

131. Hashimoto K, Kouno T, Ikawa T, Hayatsu N, Miyajima Y, Yabukami H, et al. Single-cell transcriptomics reveals expansion of cytotoxic CD4 T cells in supercentenarians. Proceedings of the National Academy of Sciences of the United States of America. 2019;116: 24242–24251. doi:10.1073/pnas.1907883116

132. Moskalev AA, Shaposhnikov M v., Zemskaya N v., Koval L, Schegoleva E v., Guvatova ZG, et al. Transcriptome Analysis of Long-lived Drosophila melanogaster E(z) Mutants Sheds Light on the Molecular Mechanisms of Longevity. Scientific Reports. 2019;9: 1–11. doi:10.1038/s41598-019-45714-x

133. Crowe EP, Tuzer F, Gregory BD, Donahue G, Gosai SJ, Cohen J, et al. Changes in the Transcriptome of Human Astrocytes Accompanying Oxidative Stress-Induced Senescence. Frontiers in Aging Neuroscience. 2016;8: 208. doi:10.3389/fnagi.2016.00208

134. Haustead DJ, Stevenson A, Saxena V, Marriage F, Firth M, Silla R, et al. Transcriptome analysis of human ageing in male skin shows mid-life period of variability and central role of NF-κB. Scientific Reports. 2016;6: 1–12. doi:10.1038/srep26846

135. Kazakevych J, Stoyanova E, Liebert A, Varga-Weisz P. Transcriptome analysis identifies a robust gene expression program in the mouse intestinal epithelium on aging. Scientific Reports. 2019;9: 1–8. doi:10.1038/s41598-019-46966-3

136. Lin H, Lunetta KL, Zhao Q, Rong J, Benjamin EJ, Mendelson MM, et al. Transcriptome-wide association study of inflammatory biologic age. Aging. 2017;9: 2288–2301. doi:10.18632/aging.101321

137. López-Otín C, Blasco MA, Partridge L, Serrano M, Kroemer G. The hallmarks of aging. Cell. Cell Press; 2013. p. 1194. doi:10.1016/j.cell.2013.05.039

138. Liguori I, Russo G, Curcio F, Bulli G, Aran L, Della-Morte D, et al. Oxidative stress, aging, and diseases. Clinical Interventions in Aging. Dove Medical Press Ltd.; 2018. pp. 757–772. doi:10.2147/CIA.S158513

139. Du L, Liu Q, Shen F, Fan Z, Hou R, Yue B, et al. Transcriptome analysis reveals immune- related gene expression changes with age in giant panda (Ailuropoda melanoleuca) blood. Aging. 2019;11: 249–262. doi:10.18632/aging.101747

140. Fusco D, Colloca G, lo Monaco MR, Cesari M. Effects of antioxidant supplementation on the aging process. Clinical interventions in aging. Dove Press; 2007. pp. 377–387.

141. Abbas H, Kamel R, El-Sayed N. Dermal anti-oxidant, anti-inflammatory and anti-aging effects of Compritol ATO-based Resveratrol colloidal carriers prepared using mixed surfactants. International Journal of Pharmaceutics. 2018;541: 37–47. doi:10.1016/j.ijpharm.2018.01.054

142. Aiello A, Farzaneh F, Candore G, Caruso C, Davinelli S, Gambino CM, et al. Immunosenescence and its hallmarks: How to oppose aging strategically? A review of potential options for therapeutic intervention. Frontiers in Immunology. Frontiers Media S.A.; 2019. doi:10.3389/fimmu.2019.02247

143. Jin Jung H, Suh Y. MicroRNA in Aging: From Discovery to Biology. Current Genomics. 2012;13: 548–557. doi:10.2174/138920212803251436

144. Abdul-Rahman O, Sasvari-Szekely M, Ver A, Rosta K, Szasz BK, Kereszturi E, et al. Altered gene expression profiles in the hippocampus and prefrontal cortex of type 2 diabetic rats. BMC Genomics. 2012;13: 81. doi:10.1186/1471-2164-13-81

145. Zhou Z, Zhu Y, Liu Y, Yin Y. Comprehensive transcriptomic analysis indicates brain regional specific alterations in type 2 diabetes. Aging. 2019;11: 6398–6421. doi:10.18632/aging.102196

146. Kelly J, Moyeed R, Carroll C, Albani D, Li X. Gene expression meta-analysis of Parkinson’s disease and its relationship with Alzheimer’s disease. Molecular Brain. 2019;12: 16. doi:10.1186/s13041-019-0436-5

147. AltimicroRNAas F, Uszczynska-Ratajczak B, Camara F, Vlasova A, Palumbo E, Newhouse S, et al. Brain transcriptome sequencing of a natural model of alzheimer’s disease. Frontiers in Aging Neuroscience. 2017;9. doi:10.3389/fnagi.2017.00064

148. French L, Ma T, Oh H, Tseng GC, Sibille E. Age-Related Gene Expression in the Frontal Cortex Suggests Synaptic Function Changes in Specific Inhibitory Neuron Subtypes. Frontiers in Aging Neuroscience. 2017;9: 162. doi:10.3389/fnagi.2017.00162

149. Fleischer JG, Schulte R, Tsai HH, Tyagi S, Ibarra A, Shokhirev MN, et al. Predicting age from the transcriptome of human dermal fibroblasts. Genome Biology. 2018;19: 221. doi:10.1186/s13059-018-1599-6

150. Liu YJ, Lian ZY, Liu G, Zhou HY, Yang HJ. RNA sequencing reveals retinal transcriptome changes in STZ-induced diabetic rats. Molecular Medicine Reports. 2016;13: 2101–2109. doi:10.3892/mmr.2016.4793

151. Meng Y, Guan Y, Zhang W, Wu YE, Jia H, Zhang Y, et al. RNA-seq analysis of the hypothalamic transcriptome reveals the networks regulating physiopathological progress in the diabetic GK rat. Scientific Reports. 2016;6. doi:10.1038/srep34138

152. Abdul-Rahman O, Sasvari-Szekely M, Ver A, Rosta K, Szasz BK, Kereszturi E, et al. Altered gene expression profiles in the hippocampus and prefrontal cortex of type 2 diabetic rats. BMC Genomics. 2012;13: 81. doi:10.1186/1471-2164-13-81

153. Zhou Z, Zhu Y, Liu Y, Yin Y. Comprehensive transcriptomic analysis indicates brain regional specific alterations in type 2 diabetes. Aging. 2019;11: 6398–6421. doi:10.18632/aging.102196

154. Kim YK, Kim B, Kim VN. Re-evaluation of the roles of DROSHA, Exportin 5, and DICER in microRNA biogenesis. Proceedings of the National Academy of Sciences of the United States of America. 2016;113: E1881–E1889. doi:10.1073/pnas.1602532113

155. Rothman SM, Tanis KQ, Gandhi P, Malkov V, Marcus J, Pearson M, et al. Human Alzheimer’s disease gene expression signatures and immune profile in APP mouse models: A discrete transcriptomic view of Aβ plaque pathology. Journal of Neuroinflammation. 2018;15: 256. doi:10.1186/s12974-018-1265-7

156. Chew G, Petretto E. Transcriptional networks of microglia in alzheimer’s disease and insights into pathogenesis. Genes. MDPI AG; 2019. doi:10.3390/genes10100798

157. Annese A, Manzari C, Lionetti C, Picardi E, Horner DS, Chiara M, et al. Whole transcriptome profiling of Late-Onset Alzheimer’s Disease patients provides insights into the molecular changes involved in the disease. Scientific Reports. 2018;8: 1–15. doi:10.1038/s41598-018- 22701-2

158. Wang Q, Zhang Y, Wang M, Song WM, Shen Q, McKenzie A, et al. The landscape of multiscale transcriptomic networks and key regulators in Parkinson’s disease. Nature Communications. 2019;10: 1–15. doi:10.1038/s41467-019-13144-y

159. Konradi C, Westin JE, Carta M, Eaton ME, Kuter K, Dekundy A, et al. Transcriptome analysis in a rat model of L-DOPA-induced dyskinesia. Neurobiology of Disease. 2004;17: 219–236. doi:10.1016/j.nbd.2004.07.005

160. Mariani E, Frabetti F, Tarozzi A, Pelleri MC, Pizzetti F, Casadei R. Meta-Analysis of Parkinson’s Disease Transcriptome Data Using TRAM Software: Whole Substantia Nigra Tissue and Single Dopamine Neuron Differential Gene Expression. Chiba-Falek O, editor. PLOS ONE. 2016;11: e0161567. doi:10.1371/journal.pone.0161567

161. Keo A, Mahfouz A, Ingrassia AMT, Meneboo J-P, Villenet C, Mutez E, et al. Transcriptomic signatures of brain regional vulnerability to Parkinson’s disease. bioRxiv. 2019; 664771. doi:10.1101/664771

162. Heneka MT, Carson MJ, Khoury J el, Landreth GE, Brosseron F, Feinstein DL, et al. Neuroinflammation in Alzheimer’s disease. The Lancet Neurology. Lancet Publishing Group; 2015. pp. 388–405. doi:10.1016/S1474-4422(15)70016-5

163. Jones OR, Scheuerlein A, Salguero-Gómez R, Camarda CG, Schaible R, Casper BB, et al. Diversity of ageing across the tree of life. Nature. 2014;505: 169–173. doi:10.1038/nature12789

164. Gerstein MB, Rozowsky J, Yan KK, Wang D, Cheng C, Brown JB, et al. Comparative analysis of the transcriptome across distant species. Nature. 2014;512: 445–448. doi:10.1038/nature13424

165. Darbellay F, Necsulea A. Comparative transcriptomics analyses across species, organs and developmental stages reveal functionally constrained lncRNAs. bioRxiv. 2019; 607200. doi:10.1101/607200

166. Frenk S, Houseley J. Gene expression hallmarks of cellular ageing. Biogerontology. 2018;19: 547–566. doi:10.1007/s10522-018-9750-z

167. Sengupta P. The laboratory rat: Relating its age with human’s. International Journal of Preventive Medicine. Wolters Kluwer -- Medknow Publications; 2013. pp. 624–630.

168. Dutta S, Sengupta P. Men and mice: Relating their ages. Life Sciences. Elsevier Inc.; 2016. pp. 244–248. doi:10.1016/j.lfs.2015.10.025

169. Cohen AA. Aging across the tree of life: The importance of a comparative perspective for the use of animal models in aging. Biochimica et Biophysica Acta - Molecular Basis of Disease. Elsevier B.V.; 2018. pp. 2680–2689. doi:10.1016/j.bbadis.2017.05.028

170. Masjosthusmann S, Becker D, Petzuch B, Klose J, Siebert C, Deenen R, et al. A transcriptome comparison of time-matched developing human, mouse and rat neural progenitor cells reveals human uniqueness. Toxicology and Applied Pharmacology. 2018;354: 40–55. doi:10.1016/j.taap.2018.05.009

171. Carboni L, Marchetti L, Lauria M, Gass P, Vollmayr B, Redfern A, et al. Cross-species evidence from human and rat brain transcriptome for growth factor signaling pathway dysregulation in major depression. Neuropsychopharmacology. 2018;43: 2134–2145. doi:10.1038/s41386-018-0117-6

172. Pacifico R, MacMullen CM, Walkinshaw E, Zhang X, Davis RL. Brain transcriptome changes in the aging Drosophila melanogaster accompany olfactory memory performance deficits. Skoulakis EMC, editor. PLOS ONE. 2018;13: e0209405. doi:10.1371/journal.pone.0209405

173. Haehner A, Hummel T, Reichmann H. Olfactory Loss in Parkinson’s Disease. Parkinson’s Disease. 2011;2011. doi:10.4061/2011/450939

174. Davie K, Janssens J, Koldere D, de Waegeneer M, Pech U, Kreft Ł, et al. A Single-Cell Transcriptome Atlas of the Aging Drosophila Brain. Cell. 2018;174: 982-998.e20. doi:10.1016/j.cell.2018.05.057

175. Grimm A, Eckert A. Brain aging and neurodegeneration: from a mitochondrial point of view. Journal of Neurochemistry. Blackwell Publishing Ltd; 2017. pp. 418–431. doi:10.1111/jnc.14037

176. Aw S, Cohen SM. Time is of the essence: MicroRNAs and age-associated neurodegeneration. Cell Research. 2012;22: 1218–1220. doi:10.1038/cr.2012.59

177. Benayoun BA, Pollina EA, Singh PP, Mahmoudi S, Harel I, Casey KM, et al. Remodeling of epigenome and transcriptome landscapes with aging in mice reveals widespread induction of inflammatory responses. Genome Research. 2019;29: 697–709. doi:10.1101/gr.240093.118

178. Ma S, Avanesov AS, Porter E, Lee BC, Mariotti M, Zemskaya N, et al. Comparative transcriptomics across 14 Drosophila species reveals signatures of longevity. Aging Cell. 2018;17: e12740. doi:10.1111/acel.12740

179. Glass D, Viñuela A, Davies MN, Ramasamy A, Parts L, Knowles D, et al. Gene expression changes with age in skin, adipose tissue, blood and brain. Genome Biology. 2013;14: R75. doi:10.1186/gb-2013-14-7-r75

180. Zeng L, Yang J, Peng S, Zhu J, Zhang B, Suh Y, et al. Transcriptome analysis reveals the difference between “healthy” and “common” aging and their connection with age‐related diseases. Aging Cell. 2020;19. doi:10.1111/acel.13121

181. Harris SE, Riggio V, Evenden L, Gilchrist T, McCafferty S, Murphy L, et al. Age-related gene expression changes and transcriptome wide association study of physical and cognitive aging traits in the Lothian Birth Cohort 1936. Aging. 2017;9: 2489–2503. doi:10.18632/aging.101333

182. Aramillo Irizar P, Schäuble S, Esser D, Groth M, Frahm C, Priebe S, et al. Transcriptomic alterations during ageing reflect the shift from cancer to degenerative diseases in the elderly. Nature Communications. 2018;9: 1–11. doi:10.1038/s41467-017-02395-2

183. Courtney E, Kornfeld S, Janitz K, Janitz M. Transcriptome profiling in neurodegenerative disease. Journal of Neuroscience Methods. Elsevier; 2010. pp. 189–202. doi:10.1016/j.jneumeth.2010.08.018

184. Hofmann-Apitius M, Ball G, Gebel S, Bagewadi S, de Bono B, Schneider R, et al. Bioinformatics mining and modeling methods for the identification of disease mechanisms in neurodegenerative disorders. International Journal of Molecular Sciences. MDPI AG; 2015. pp. 29179–29206. doi:10.3390/ijms161226148

185. Kusko R, Dreymann J, Ross J, Cha Y, Escalante-Chong R, Garcia-MicroRNAalles M, et al. Large-scale transcriptomic analysis reveals that pridopidine reverses aberrant gene expression and activates neuroprotective pathways in the YAC128 HD mouse. Molecular Neurodegeneration. 2018;13: 25. doi:10.1186/s13024-018-0259-3

186. Bayraktar A, Önal-Süzek T, Süzek BE, Baysal Ö. Meta-analysis of Gene Expression in Neurodegenerative Diseases Reveals Patterns in GABA Synthesis and Heat Stress Pathways. Available: https://github.com/davidliwei/rnaseqmut

187. Magistri M, Velmeshev D, Makhmutova M, Faghihi MA. Transcriptomics Profiling of Alzheimer’s Disease Reveal Neurovascular Defects, Altered Amyloid-β Homeostasis, and Deregulated Expression of Long Noncoding RNAs. Journal of Alzheimer’s Disease. 2015;48: 647–665. doi:10.3233/JAD-150398

188. Annese A, Manzari C, Lionetti C, Picardi E, Horner DS, Chiara M, et al. Whole transcriptome profiling of Late-Onset Alzheimer’s Disease patients provides insights into the molecular changes involved in the disease. Scientific Reports. 2018;8. doi:10.1038/s41598-018-22701-2

189. Jiang P, Scarpa JR, Gao VD, Vitaterna MH, Kasarskis A, Turek FW. Parkinson’s Disease is Associated with Dysregulations of a Dopamine-Modulated Gene Network Relevant to Sleep and Affective Neurobehaviors in the Striatum. Scientific Reports. 2019;9: 1–14. doi:10.1038/s41598- 019-41248-4

190. Nam KN, Mounier A, Wolfe CM, Fitz NF, Carter AY, Castranio EL, et al. Effect of high fat diet on phenotype, brain transcriptome and lipidome in Alzheimer’s model mice. Scientific Reports. 2017;7: 1–13. doi:10.1038/s41598-017-04412-2

191. Djoussé L, Michael Gaziano J, Buring JE, Lee IM. Egg consumption and risk of type 2 diabetes in men and women. Diabetes Care. 2009;32: 295–300. doi:10.2337/dc08-1271

192. Radzeviciene L, Ostrauskas R. Egg consumption and the risk of type 2 diabetes mellitus: A case-control study. Public Health Nutrition. 2012;15: 1437–1441. doi:10.1017/S1368980012000614

193. Radzeviciene L, Ostrauskas R. Egg consumption and the risk of type 2 diabetes mellitus: A case-control study. Public Health Nutrition. 2012;15: 1437–1441. doi:10.1017/S1368980012000614

194. Djoussé L, Kamineni A, Nelson TL, Carnethon M, Mozaffarian D, Siscovick D, et al. Egg consumption and risk of type 2 diabetes in older adults. American Journal of Clinical Nutrition. 2010;92: 422–427. doi:10.3945/ajcn.2010.29406

195. Virtanen JK, Mursu J, Tuomainen TP, Virtanen HEK, Voutilainen S. Egg consumption and risk of incident type 2 diabetes in men: The kuopio ischaemic heart disease risk factor study. American Journal of Clinical Nutrition. 2015;101: 1088–1096. doi:10.3945/ajcn.114.104109

196. Kerepesi C, Daróczy B, Sturm Á, Vellai T, Benczúr A. Prediction and characterization of human ageing-related proteins by using machine learning. Scientific Reports. 2018;8: 1–13. doi:10.1038/s41598-018-22240-w

197. Wang J, Knol MJ, Tiulpin A, Dubost F, de Bruijne M, Vernooij MW, et al. Gray matter age prediction as a biomarker for risk of dementia. Proceedings of the National Academy of Sciences of the United States of America. 2019;116: 21213–21218. doi:10.1073/pnas.1902376116

198. Arabfard M, Ohadi M, Rezaei Tabar V, Delbari A, Kavousi K. Genome-wide prediction and prioritization of human aging genes by data fusion: A machine learning approach. BMC Genomics. 2019;20: 832. doi:10.1186/s12864-019-6140-0

199. Parson W. Age Estimation with DNA: From Forensic DNA Fingerprinting to Forensic (Epi)Genomics: A Mini-Review. Gerontology. 2018;64: 326–332. doi:10.1159/000486239

200. Hamano Y, Manabe S, Morimoto C, Fujimoto S, Tamaki K. Forensic age prediction for saliva samples using methylation-sensitive high resolution melting: Exploratory application for cigarette butts. Scientific Reports. 2017;7: 1–8. doi:10.1038/s41598-017-10752-w

201. Vidaki A, Ballard D, Aliferi A, Miller TH, Barron LP, Syndercombe Court D. DNA methylation-based forensic age prediction using artificial neural networks and next generation sequencing. Forensic Science International: Genetics. 2017;28: 225–236. doi:10.1016/j.fsigen.2017.02.009

202. Horvath S. DNA methylation age of human tissues and cell types. Genome Biology. 2013;14: R115. doi:10.1186/gb-2013-14-10-r115

203. Hannum G, Guinney J, Zhao L, Zhang L, Hughes G, Sadda SV, et al. Genome-wide Methylation Profiles Reveal Quantitative Views of Human Aging Rates. Molecular Cell. 2013;49: 359–367. doi:10.1016/j.molcel.2012.10.016

204. Xu C, Qu H, Wang G, Xie B, Shi Y, Yang Y, et al. A novel strategy for forensic age prediction by DNA methylation and support vector regression model. Scientific Reports. 2015;5: 1–10. doi:10.1038/srep17788

205. Hastie T, Tibshirani R, Friedman J. Springer Series in Statistics The Elements of Statistical Learning Data Mining, Inference, and Prediction.

206. Peters MJ, Joehanes R, Pilling LC, Schurmann C, Conneely KN, Powell J, et al. The transcriptional landscape of age in human peripheral blood. Nature Communications. 2015;6. doi:10.1038/ncomms9570

207. Ferrucci L, Bandinelli S, Benvenuti E, di Iorio A, Macchi C, Harris TB, et al. Subsystems contributing to the decline in ability to walk: Bridging the gap between epidemiology and geriatric practice in the InCHIANTI study. Journal of the American Geriatrics Society. 2000;48: 1618–1625. doi:10.1111/j.1532-5415.2000.tb03873.x

208. Putin E, Mamoshina P, Aliper A, Korzinkin M, Moskalev A, Kolosov A, et al. Deep biomarkers of human aging: Application of deep neural networks to biomarker development. Aging. 2016;8: 1021–1033. doi:10.18632/aging.100968

209. Herrera-Marcos L v., Lou-Bonafonte JM, Arnal C, Navarro MA, Osada J. Transcriptomics and the mediterranean diet: A systematic review. Nutrients. MDPI AG; 2017. doi:10.3390/nu9050472

210. German JB. Genetic dietetics: Nutrigenomics and the future of dietetics practice. Journal of the American Dietetic Association. 2005;105: 530–531. doi:10.1016/j.jada.2005.02.034

211. Ardekani AM, Naeini MM. The role of microRNAs in human diseases. Avicenna Journal of Medical Biotechnology. Avicenna Research Institute; 2010. pp. 161–179.

212. Thalyana SV, Slack FJ. MicroRNAs and their roles in aging. Journal of Cell Science. 2012. pp. 7–17. doi:10.1242/jcs.099200

213. Noren Hooten N, Fitzpatrick M, Wood WH, De S, Ejiogu N, Zhang Y, et al. Age-related changes in microRNA levels in serum. Aging. 2013;5: 725–740. doi:10.18632/aging.100603

214. Analysis of Plasma MicroRNAs as Predictors and Biomarkers of Aging and Frailty in Humans. [cited 19 Mar 2020]. Available: https://www.hindawi.com/journals/omcl/2018/7671850/

215. Tahamtan A, Teymoori-Rad M, Nakstad B, Salimi V. Anti-inflammatory MicroRNAs and their potential for inflammatory diseases treatment. Frontiers in Immunology. Frontiers Media S.A.; 2018. p. 1377. doi:10.3389/fimmu.2018.01377

216. Huan T, Chen G, Liu C, Bhattacharya A, Rong J, Chen BH, et al. Age-associated microRNA expression in human peripheral blood is associated with all-cause mortality and age-related traits. Aging Cell. 2018;17. doi:10.1111/acel.12687

217. Inukai S, de Lencastre A, Turner M, Slack F. Novel MicroRNAs Differentially Expressed during Aging in the Mouse Brain. Watson M, editor. PLoS ONE. 2012;7: e40028. doi:10.1371/journal.pone.0040028

218. Rao YS, Mott NN, Wang Y, Chung WCJ, Pak TR. MicroRNAs in the Aging Female Brain: A Putative Mechanism for Age-Specific Estrogen Effects. Endocrinology. 2013;154: 2795–2806. doi:10.1210/en.2013-1230

219. Li SC, Chan WC, Hu LY, Lai CH, Hsu CN, Lin W chang. Identification of homologous microRNAs in 56 animal genomes. Genomics. 2010;96: 1–9. doi:10.1016/j.ygeno.2010.03.009

220. Raj T, Li YI, Wong G, Humphrey J, Wang M, Ramdhani S, et al. Integrative transcriptome analyses of the aging brain implicate altered splicing in Alzheimer’s disease susceptibility. Nature Genetics. 2018;50: 1584–1592. doi:10.1038/s41588-018-0238-1

221. Felekkis K. MicroRNAs and Aging: Biomarkers or Therapeutic Targets? Current Aging Science. 2017;10: 80–82. doi:10.2174/1874609808666161124130206

222. Thalyana SV, Slack FJ. MicroRNAs and their roles in aging. Journal of Cell Science. The Company of Biologists Ltd; 2012. pp. 7–17. doi:10.1242/jcs.099200

223. Kinser HE, Pincus Z. MicroRNAs as modulators of longevity and the aging process. Human Genetics. Springer; 2020. pp. 291–308. doi:10.1007/s00439-019-02046-0

224. Boehm M, Slack F. Physiology: A developmental timing microRNA and its target regulate life span in C. elegans. Science. 2005;310: 1954–1957. doi:10.1126/science.1115596

225. Mariño G, Ugalde AP, Fernández ÁF, Osorio FG, Fueyo A, Freije JMP, et al. Insulin-like growth factor 1 treatment extends longevity in a mouse model of human premature aging by restoring somatotroph axis function. Proceedings of the National Academy of Sciences of the United States of America. 2010;107: 16268–16273. doi:10.1073/pnas.1002696107

226. Jung HJ, Suh Y. Regulation of IGF -1 signaling by microRNAs. Frontiers in Genetics. Frontiers Media S.A.; 2015. doi:10.3389/fgene.2014.00472

227. Margolis LM, Lessard SJ, Ezzyat Y, Fielding RA, Rivas DA. Circulating MicroRNA Are Predictive of Aging and Acute Adaptive Response to Resistance Exercise in Men. The journals of gerontology Series A, Biological sciences and medical sciences. 2017;72: 1319–1326. doi:10.1093/gerona/glw243

228. Yao Y, Jiang C, Wang F, Yan H, Long D, Zhao J, et al. Integrative Analysis of microRNA and mRNA Expression Profiles Associated With Human Atrial Aging. Frontiers in Physiology. 2019;10: 1226. doi:10.3389/fphys.2019.01226

229. Ong J, Woldhuis RR, Boudewijn IM, van den Berg A, Kluiver J, Kok K, et al. Age-related gene and microRNA expression changes in airways of healthy individuals. Scientific Reports. 2019;9: 1–8. doi:10.1038/s41598-019-39873-0

230. Ferrucci L, Bandinelli S, Benvenuti E, di Iorio A, Macchi C, Harris TB, et al. Subsystems contributing to the decline in ability to walk: Bridging the gap between epidemiology and geriatric practice in the InCHIANTI study. Journal of the American Geriatrics Society. 2000;48: 1618–1625. doi:10.1111/j.1532-5415.2000.tb03873.x

231. Tarallo S, Pardini B, Mancuso G, Rosa F, di Gaetano C, Rosina F, et al. MicroRNA expression in relation to different dietary habits: a comparison in stool and plasma samples. Mutagenesis. 2014;29: 385–91. doi:10.1093/mutage/geu028

232. Slattery ML, Herrick JS, Mullany LE, Stevens JR, Wolff RK. Diet and lifestyle factors associated with microRNA expression in colorectal tissue. Pharmacogenomics and Personalized Medicine. 2017;10: 1–16. doi:10.2147/PGPM.S117796

233. Zheng Z, Ge Y, Zhang J, Xue M, Li Q, Lin D, et al. PUFA diets alter the microRNA expression profiles in an inflammation rat model. Molecular Medicine Reports. 2015;11: 4149– 4157. doi:10.3892/mmr.2015.3318

234. Giangreco AA, Vaishnav A, Wagner D, Finelli A, Fleshner N, van der Kwast T, et al. Tumor suppressor microRNAs, microRNA-100 and -125b, are regulated by 1,25-dihydroxyvitamin D in primary prostate cells and in patient tissue. Cancer Prevention Research. 2013;6: 483–494. doi:10.1158/1940-6207.CAPR-12-0253

235. Fong LY, Farber JL, Croce CM. Zinc intake, microRNA dysregulation, and esophageal cancer. Aging. Impact Journals LLC; 2016. pp. 1161–1162. doi:10.18632/aging.100978

236. Zhang L, Hou D, Chen X, Li D, Zhu L, Zhang Y, et al. Exogenous plant MICRORNA168a specifically targets mammalian LDLRAP1: Evidence of cross-kingdom regulation by microRNA. Cell Research. 2012;22: 107–126. doi:10.1038/cr.2011.158

237. Baier SR, Nguyen C, Xie F, Wood JR, Zempleni J. MicroRNAs Are Absorbed in Biologically Meaningful Amounts from Nutritionally Relevant Doses of Cow Milk and Affect Gene Expression in Peripheral Blood Mononuclear Cells, HEK-293 Kidney Cell Cultures, and Mouse Livers. The Journal of Nutrition. 2014;144: 1495–1500. doi:10.3945/jn.114.196436

238. Quintanilha BJ, Reis BZ, Silva Duarte GB, Cozzolino SMF, Rogero MM. NutrimicroRNAomics: Role of micrornas and nutrition in modulating inflammation and chronic diseases. Nutrients. MDPI AG; 2017. p. 1168. doi:10.3390/nu9111168

239. Quinlan S, Kenny A, Medina M, Engel T, Jimenez-Mateos EM. MicroRNAs in Neurodegenerative Diseases. International Review of Cell and Molecular Biology. Elsevier Inc.; 2018. pp. 1–26. doi:10.1016/bs.ircmb.2017.04.002

CHAPTER 3. LARGE AND SMALL RNA SEQUENCING REVEALS OXIDATIVE- REDUCTION PATHWAYS ARE MODIFIED BY SHORT-TERM WHOLE EGG CONSUMPTION

Modified from a manuscript in preparation for the Journal PLOS ONE.

Authors & Affiliations

Joe L. Webb1,2,3, Amanda E. Bries1,2, Brooke Vogel1, Claudia Carrillo1, Timothy A. Day4,

Michael J. Kimber4, Rudy J. Valentine2,5, Matthew J. Rowling1,2, Stephanie Clark1, Kevin L.

Schalinske1,2, and Elizabeth McNeill1,2,6

1 Department of Food Science and Human Nutrition, Iowa State University, Ames, IA, USA

2 Interdepartmental Graduate Program in Nutritional Sciences, Iowa State University, Ames, IA, USA

3 National Science Foundation Graduate Research Fellowship Program, Alexandria, VA, USA

4 Department of Biomedical Sciences, Iowa State University College of Veterinary Medicine, Ames, IA, USA.

5 Department of Kinesiology, Iowa State University, Ames, IA, USA.

6 Genetics and Genomics Graduate Program, Iowa State University, Ames, IA, USA

Corresponding Author

Elizabeth M. McNeill, Ph.D.

Department of Food Science and Human Nutrition

220 MacKay Hall

Iowa State University

Ames, Iowa 50010, United States

Tel: +1 515 294 7421, Fax: +1 515 294 6193

E-mail address: [email protected]

Data Availability

The data created though this study can be downloaded from the National Institutes of Health

(NIH) Sequence Read Archive (SRA). Additional python scripts used to process samples will be available on GitHub at: https://github.com/joelwebb/Sprague-Dawley-Whole-Egg-RNA-Seq

Key Words: Whole egg; differential expression analysis; RNAseq; Sprague Dawley

Abstract

Scope: Eggs are protein-rich, nutrient-dense, and contain bioactive ingredients that have been shown to modify gene expression.

Methods: In order to understand the effects that egg consumption has on tissue-specific mRNA and microRNA gene expression, we examined the role of whole egg consumption (20% protein, w/w) on differentially expressed genes (DEGs) between rat (n =12) transcriptomes in the prefrontal cortex (PFC), liver, kidney and adipose tissue. Principal component analysis with hierarchical clustering were used to examine transcriptomic profiles between treatment groups.

Finally, we performed Gene Ontology and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis to examine which metabolic pathways the DEGs altered in each tissue.

Results: Overall, these data show that whole egg consumption for 2 wk modified the expression of 52 genes in the PFC, 20 genes in the adipose, and two genes in the liver (adj P < 0.05).

Additionally, microRNA125b-5p was downregulated in the adipose, microRNA-192-5p and microRNA-10b-5p were downregulated in the PFC. The main pathways influenced by WE consumption, were glutathione metabolism in the adipose and cholesterol synthesis in the PFC.

Conclusion: These data highlight potential microRNA targets and the genes that may be modified by acute consumption of whole egg-based diets.

Introduction

Eggs are a low-cost, nutrient-dense food comprised of numerous vitamins, bioactive compounds and have been proposed to play a role in disease prevention [1,2]. Eggs and their nutrients [3] have been linked to several mechanisms of modifying gene expression, such as vitamin D-mediated transcriptional regulation and by modifying one-carbon metabolism via their folate and choline content [4]. Despite the beneficial components of eggs, they remain one of the most controversial foods [5], owing to their contribution of dietary cholesterol [6,7].

Observational studies examining the role of long-term egg intake on the risk of developing cardiovascular disease (CVD) have reported inconsistent results[8], but most recently, Dehghan and others reported no significant association between egg intake and major CVD events in a conglomerate of 50 studies [9]. Although the role of whole egg (WE) consumption has been extensively examined in population-based studies [10,11] only a few studies have thoroughly identified the biological role of the individual egg components. For instance, egg yolk peptides have been shown to display anti-oxidative properties [12] and lutein, a carotenoid that is high in egg yolk, has been demonstrated to protect dopaminergic neurons from oxidative damage in a model of Parkinson’s Disease (PD) [13]. Similar effects of lutein administration have also been shown in other animal models of aging and cognitive impairment [14,15]; however, the role of these egg components and their influence on their global gene expression remains elusive. In addition to WE are decreasing oxidative stress, our laboratory has previously reported that consuming a WE-based diet reduced body weight gain in rats with type 2 diabetes [16,17]. To date, very few studies have focused on identifying the molecular mechanisms that are altered by

WE consumption in additional to the global gene expression changes in the rats fed WE vs. casein-based protein diets. By gaining an in-depth understanding of gene-diet interactions, we can shed light on the molecular mechanisms underlying WE consumption.

In this study, male Sprague Dawley rats were fed WE-based diets to examine the influence of short-term WE consumption on gene and microRNA expression in comparison with

CAS-based diets. We examined the transcriptomic profiles of the prefrontal cortex (PFC), adipose, liver, and kidney tissues to identify metabolic pathways that may be altered by WE consumption and mapped these changes to microRNAs. The primary aim of this study was to first identify if WE consumption had a significant impact on global gene expression, and if there were potential molecular drivers, such as microRNAs that were upstream effects of these observed transcriptomic differences.

Methods

Animals and Diets

This animal study was approved by the Institutional Animal Care and Use Committee at

Iowa State University and was performed according to the Iowa State University Laboratory

Animal Resources Guidelines. Male Sprague Dawley rats (n =12) were obtained at 6 wk of age

(151-175 g) from Charles River Laboratories (Wilmington, MA). Rats were individually housed in conventional cages in a temperature-controlled environment (22°C ± 2°C) following a 12-h light-dark cycle. All rats were acclimated for one week on a standard rat chow diet, whereby they were randomly assigned to one of two dietary intervention groups. There were no significant differences in baseline weights between groups (P=0.62). Rats were either placed on the control casein (CAS)-based diet, or a whole egg (WE)-based diet (Table 1) matched for 20% protein

(w/w). All ingredients were purchased from Envigo except for dried whole egg (Rose Acre

Farms, Guthrie Center, IA), as well as L-methionine and choline bitartrate (Sigma-Aldrich).

Abbreviations used: CAS, casein-based diet, WE, whole egg-based diet. The protein and lipid quantities in the diet coming from 435 g of dried whole egg was 46 (200 g) and 42% (183 g),

57 respectively. To formulate all diets such that protein was provided at 20% (w/w).For 72 h, animals underwent a controlled fasted-refeeding protocol to train them to consume food ad libitum within a 4h window (Supplemental Figure 1). After training, animals were fasted overnight for 10 h with water provided ad libitum, followed by controlled feeding (4 g) of either the CAS- or WE-based diet. At 7 wk of age, serum was collected via tail vein at 0, 2, 4, 6, and 8 h immediately following refeeding. Following the serum time curve collection, rats were maintained on their respective diets for 2-wk with food intake and body weight gain monitored daily. Following the dietary intervention, 9-wk old rats underwent a 12 h overnight fast with water provided ad libitum; rats were anesthetized with a ketamine:xylazine cocktail (90:10 mg/kg body weight) via a single intraperitoneal injection. Whole blood was collected via cardiac puncture and stored at -80○C. The epididymal adipose, kidney, liver, and PFC tissues were procured, weighed, and stored in RNAlater.

Large and small RNA Extraction & Sequencing

Total RNA and microRNA were extracted using a SPLIT Total RNA Extraction Kit

(Lexogen, Greenland, NH). RNA quantities were measured using the Qubit Flourometer

(ThermoFisher Scientific), and integrity was assessed using a Bioanalyzer 2100 (Agilent

Technologies). Large RNA libraries were prepared using an automated protocol for the

QuantSeq 3' mRNA-Seq Library Prep Kit (Lexogen, Greenland, NH) and small RNA libraries were prepared using Small RNA Library Prep Kits (Lexogen, Greenland, NH). Total RNA samples were multiplexed across two lanes using an Illumina High-Seq 3000 resulting in an average of 7.5 million reads per sample prior to quality control. Small RNA libraries were also multiplexed and run on a separate lane on an Illumina High-Seq 3000.

Quality Control & Adapter Trimming

All large and small RNA reads were inspected using Fastqc and were trimmed to remove adapter sequences using BBDUK [18]. Read segments matching common Illumina Truseq or

Nextera adapter sequences were removed for the reverse-complement or forward sequence of the adapters during processing. Subsequently, low-quality reads with average quality < 10 were discarded.

Read Alignment & Quantification

Reference genome (fasta) and genome annotation files (gtf) were obtained from the

Ensembl genome browser. Reads were aligned to the RNO version 6 release 94 of the Ensembl genome using the STAR v2.5.2 aligner [19]. Transcripts aligning to specific genes were counted using the STAR - quantMode geneCounts function to map transcripts to each genome. Files containing microRNA counts and gene counts for all samples are presented in Supplemental

Data 1 and 2, respectively. Small RNA samples were processed using the smallrnaseq python tool [20], which aligns samples using Bowtie and quantifies small RNA read counts using reference fasta and gtf files from RNAcentral.org.

Data Filtering & Quality Control

All data analysis was conducted in Python within an IPython notebook unless otherwise specified. Genes were filtered out if they were not expressed in any samples or had fewer than 10 counts in half of the samples for each gene. Data filtering and alignment settings were adapted from Lexogen‘s QuantSeq 3‘ mRNA-Seq Kit and integrated Data Analysis Pipeline on

Bluebee® platform according to the manufacturer’s instructions.

Differential Expression Analysis using DESeq2

Read normalization was conducted using weighted mean log expression ratios (trimmed mean of M values (TMM)) method to account for variable sequencing depth between samples.

Differential expression analysis was conducted using the DESeq2 [21] package in the R programming language to identify differentially expressed genes (DEGs). When applying

DESeq2, DESeqDataSetFromMatrix was used to generate P values and adjusted p-values were calculated with Benjamini-Hochberg corrections [22] for false discovery rate (FDR) correction.

For all analyses, FDR was controlled at 1% and all adjusted p-values < 0.01 were considered significant.

Heatmaps, Principal Component Analysis, & Volcano Plots

Principal Component Analysis (PCA) was conducted for the initial clustering and characterization of RNAseq data. Hierarchical clustering was used to create a dendrogram classifying samples according to similar transcriptomic profiles, while Volcano Plots were constructed to visualize genes which surpass a log-fold change of >1.5 increase or decrease to assess biological significance. PCA, clustering, and volcano plots were all constructed using

MatplotLib in python version 3.2.0rc1.

Functional Enrichment Annotations

Gene pathway analysis was performed with the Kyoto Encyclopedia of Genes and

Genomes (KEGG) database that contains annotated biological functions for genes [23]. All

KEGG pathway analysis was performed in python using the KEGG package version bio2bel- kegg 0.2.5. All differentially expressed genes (DEGs) were additionally categorized based on cellular localization, function, and processes using Gene Ontology (GO) Analysis in the

Database for Annotation, Visualization and Integrated Discovery (DAVID) v6.8 database through the online web application [24]. MicroRNA target genes were identified through

TargetScan and target genes were downloaded and mapped through Ensembl software [25].

Genes were declared significant with multiple testing correction at 5% FDR & adjusted P <0.05.

qRT-PCR Validation Analyses

Total RNA from each tissue was aliquoted and reverse transcribed with a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Catalog # 4368813) into cDNA. The cDNA was diluted to 100ng/uL and qPCR reactions were performed using 200ng of total cDNA with primers at 300nM concentration in 10 uL FastStart Sybr Green Master (Roche) according to the manufacturer’s instructions. All qPCR reactions were conducted in a Roche LightCycler 96

Real-Time PCR System. Primers sequences for qPCR are listed in the Supplemental Table 1, and 18s RNA was used for the internal control to normalize CT counts within each tissue. Each sample was assessed in triplicate. The qPCR data was analyzed with the Livak Delta-Delta CT method [26].

Protein-Protein Interaction Network Mapping

The STRING database was used for analyzing protein-protein interaction networks between DEGs using their online web portal https://string-db.org/ . The interaction maps show both direct and indirect protein associations using previously published studies and databases.

Results

RNA Seq Differential Expression

Differential gene expression analyses of the resulted in 52 DEGs in the PFC, 20 in the adipose tissue, 2 in the liver, and 0 in the kidney. Of the 74 DEGs that surpassed multiple testing corrections (adjusted P < 0.05), 1 gene was differentially upregulated across both the PFC (2.36- fold increase) and adipose (1.78-fold increase) tissue - indolethylamine N-methyltransferase

(INMT). Table 2 describes the DEGs in each tissue, whereas Supplemental Data 2 contains adjusted P-value results for all genes.

KEGG & GO Functional Enrichment Analysis

To examine the function pathway for each DEG, mRNA was mapped to KEGG/GO pathway terms, which are described in Table 3. The primary KEGG enrichment analyses of the

DEGs indicated that in adipose tissue, the top KEGG pathway that was upregulated by consuming a WE-based diet was glutathione metabolism. In the PFC, glutathione metabolism was ranked 3rd, while steroid biosynthesis and terpenoid backbone biosynthesis were the other top biological pathways. In the liver, there were no pathway processes that were determined from the differentially expressed genes.

To further investigate the molecular function and localization of these DEGs, GO analyses revealed that all 74 genes could be classified according to the categories of: biological processes, cellular structures, and molecular functions. GO Functional Analysis of DEGs indicated that in the PFC, the top altered pathways were oxidation-reduction processes, cholesterol biosynthetic processes, metal ion binding pathways, and fatty acid biosynthetic processes. In the adipose, the main pathways were organic acid transmembrane transport, phosphate ion transport, and glutathione metabolic processes as described in Table 3.

MicroRNA Sequencing Differential Expression Analysis

Differential expression analyses of the microRNAs resulted in 6 upregulated microRNAs and 10 downregulated microRNAs across all four tissues based on non-adjusted P < 0.05. No microRNAs survived multiple testing correction with FDR correction at 5%. Table 4 describes the differentially expressed microRNAs in each tissue, whereas Supplemental Data 1 contains adjusted P values results for all microRNAs. Interestingly, microRNA-10b-5p was downregulated in both the adipose and PFC (non-adjusted; P=0.03 and P=0.02, respectively) and microRNA-192-5p was downregulated in both the liver and PFC (non-adjusted; P=0.02 and

P=0.05, respectively).

MicroRNA Gene Target Analysis

Differentially expressed microRNAs from Table 2 were mapped against their human genetic targets and cross-referenced against the DEGs from Table 3. For instance, Table 5 indicates the targets of the downregulated microRNA-10b-5p in the PFC. The DEG that was upregulated in the PFC as it relates to microRNA-10b-5p was the Arrestin Domain Containing 3 protein (Arrdc3). Nine DEGs were identified in the PFC and were correlated with the suppression of microRNA-192-5p in the PFC (Table 4).

Serum MicroRNA Refeeding Analysis

There was no statistically significant effect of dietary treatment on serum microRNA 0, 2,

4, 6, or 8 h immediately following refeeding after correcting for multiple testing using FDR adjusted P < 0.05. Additionally, using an ANNOVA with repeated measures across time, there was no significant effects of dietary treatment on serum microRNA.

Principal Component Analysis (PCA) & Volcano Plots

To examine the relatedness between transcriptomic profiles, lowly expressed genes across all tissues were filtered and performed PCA, as shown in Figure 1A. The PCA plots indicate that these rat samples cluster by diet (WE vs. CAS). Subsequently, the top DEGs were visualized using a hierarchical clustering heatmap for each tissue that displays distinct differential expression patterns according to each diet, as shown in Figure 1B-D. To view the

DEGs with large fold changes, volcano plots in Figure 2 were used to visualize the results.

qRT-PCR validation and Protein-protein Interaction Networks

To validate the RNA-seq data, 5 DEGs were selected for qRT-PCR analysis in each tissue. The PCR gene expression data were correlated with the RNA-seq data in Figure 3, suggesting that RNA-seq and PCR results were in alignment with one another. Finally, we examined protein-protein interactions based on the DEGs in each tissue. In the adipose tissue and

PFC according to the functional interaction network for Rattus norvegicus protein-protein interaction maps are located in Figure 4 A & B. Notably in the PFC, nearly half of all DEG proteins interacted with one another, whereas in the adipose tissue, almost all of the DEG proteins did not interact, corroborating the GO analysis.

Food Intake & Body Weight Gain

There was no significant effect of dietary treatment on food intake or total energy intake throughout the study (Supplemental Table 4). Additionally, there was no statistically significant effect of dietary treatment on final body weight or cumulative body weight gain at 8 wk of age.

There were also no differences in organ weights at tissue collection except for the liver, where rats on the WE-based diets had 16% increased relative liver weight (P<0.01).

Discussion

Previous studies have examined specific egg components, such as hen egg lysozyme in altering gene expression in pig intestinal tissues [27], but very little information is known about how WE diet affects endogenous gene expression across multiple tissues. Our laboratory has consistently demonstrated the beneficial role of a WE-based diet in maintaining vitamin D status and modulating phenotypic outcomes in both a type 1 diabetes (T1DM) and type 2 diabetes

(T2DM) animal model [17,28]; however, we have yet to investigate how WE consumption modulates gene expression in a standard rodent model. RNA-sequencing is a powerful tool that can examine the influence of dietary patterns on whole genome gene expression. Since nutrition was one of the first major environmental factors that was identified as an epigenetic influencer, it’s important to look at more global effects of WE consumption in a standard rodent model to better understand how consuming WE alone, influences genetic expression. Measuring these effects is especially important when it comes to better understanding the molecular drivers of disease. By assessing the genes that are influenced by WE, we can contribute to scientifically sound public health policies by designing effective intervention strategies.

In this study, next generation sequencing revealed that consuming WE for 2-wk significantly modified the expression of 74 different genes across the PFC, liver, and adipose tissue. In the PFC, the top three represented GO pathways were oxidation-reduction process, cholesterol biosynthetic process, and metal ion binding. Profiling the DEGs within the given pathways indicated that 9 out of the 12 DEGs in the oxidation-reduction process pathway were downregulated. Two cytochrome P450 genes, Cyp2c22, and Cyp4A1 were upregulated in the

PFC of the animals fed WE- vs. CAS-based diets. Moreover, squalene epoxidase (Sqle), a rate- limiting gene in the sterol biosynthesis pathway [29], was significantly downregulated (-31-fold) within the PFC of rats fed WE-based vs. CAS-based diets. Sqle is important for steroidal

65 synthesis, and previous studies have demonstrated that ablation of Sqle may disrupt tumorigenesis, owing to blunted cholesterol biosynthesis [30]. Moreover, the dysregulation of

Sqle has been observed during the onset of diabetes [31,32]. Q Ge et al. [33] indicated a significant upregulation in the expression of Sqle, as well as an abundance in the protein in a chemically induced diabetic animal model [34]. Another contrast between the diabetic animal model and our WE dietary fed animals is the observed increased expression of the Cyp genes, where in contrast, there was a global reduction in Cyp51 and other cytochrome P450 genes in the liver of diabetic animals. Similar findings have been corroborated in other diabetic animal models, where it’s been detected that Cyp2c22 was markedly reduced in a T2D mouse [35].

Interestingly, Ding et al [31] conducted a cross-sectional study using obese men, examining

DEGs based on body mass index. They reported the close association between body mass index and increased expression of Fads1, Sqle, Scd, Cyp51a1, whereas post weight-loss intervention,

Sqle expression was significantly reduced. Like these findings, we observed downregulation of

Fads1 and Sqle in animals fed the WE- vs. CAS-based diets. The results from our findings may indicate a potential benefit of WE-based diet on the increased expression of these cytochrome

P450 enzymes, which may have a beneficial role in correcting the progression of T2D.

Functional annotation of the DEGs to GO terms indicated that the two targeted genes of the glutathione S-transferase (GST) pathway were glutathione S-transferase zeta 1 (Gstz1) and glutathione S-transferase pi 1 (Gstp1), which were upregulated 1.9- and 3.7-fold, respectively in the adipose tissue of the SD animals fed WE-based diets. Previously, it has been reported that deficiency of glutathione-related pathways alters antioxidant responses [36], suggesting that our data may indicate an upregulation of glutathione with WE consumption, providing protection against oxidative stress. In a recent study, a Gstp1 polymorphism was associated with increased

66 glucose intolerance and greater androgen production in women who were not obese but had with polycystic ovary syndrome [37]. Moreover, the literature indicates that dysregulated GST production has been implicated in conditions of obesity and T2D [38]. It was previously reported that glutathione metabolism is regulated by egg yolk peptide consumption in a porcine model of oxidative stress [39], while in Zucker Diabetic Fatty rats, egg white hydrolysate consumption increased glutathione concentrations in the liver [40]. In a meta-analysis of clinical studies with long-term egg consumption, there were no observed effects of WE consumption on blood inflammatory markers [41], whereas other researchers have reported elevation in endothelial and arterial inflammation from WE consumption [42] and some report vascular inflammation to be exclusive to egg white consumption and not WE [43,44]. It’s important that we examined these differences in gene expression, as we identified upregulation not only in the adipose tissue, but also observed an increase of glutathione S-transferase mu 2 (Gstm2) expression in the brain of rats fed WE-based diet. These are important considerations as the data from clinical trials regarding, WE consumption on inflammation-mediated cardiovascular disease remains inconclusive.

To better understand how the DEGs interact with one another, we used the STRING-DB to interrogate potential protein-protein interactions using models based on gene pathways and functional categories. K-nearest-neighbor (KNN) clustering was used to define relationships more granularly between DEGs and identify sub-networks in our data. In adipose tissue, very few genes interacted and KNN clustering revealed 3 main pathways that corroborated the KEGG analysis: 1) glutathione processing, 2) carbamoyl-phosphate processing and 3) peroxisome proliferator activated genes. One of the most intriguing findings resulted from examining the

PFC, where distinct protein-protein interaction networks occurred among the DEG genes. Nearly

67 all of the fatty acid biosynthesis genes and cholesterol synthesis genes physically interacted, while very few of the genes in any other pathways were interrelated. KNN clustering also revealed 3 common clusters in the PFC according to the protein interaction maps, with two main subgroups: 1) steroid biosynthesis processing; and 2) fatty acid synthesis, with the remaining genes not interacting.

Although in this study we demonstrated multiple genes were differentially expressed across these tissues, each of these genes highlighted here should be investigated further to confirm if they are regulated through, WE consumption. After examining whether these DEGs were changing across all the sequenced tissues, we did not identify any DEGs that were globally affected by WE consumption, which was a rigorous assessment. Additionally, we wanted to interrogate the mechanism by which these 74 genes were being altered by WE consumption, by performing smallRNA sequencing. Surprisingly, there was no observed effect of acute WE consumption on postprandial serum microRNAs within 8h of testing. Previous research by

Zemplini et al. reported robust effects of egg consumption on transient plasma microRNA expression [45,46]. While it is controversial whether diet consumption directly influences circulating microRNA status [47], we did not observe any serum changes in microRNAs between the two dietary treatments. MicroRNAs are a great tool for biomarker detection [48], as well as granular mediators in the pathogenesis of various diseases [49,50]. By using next generation sequencing to observe serum microRNA changes, we concluded that diet alone, is not promising to elicit acute changes in microRNA status. The tissue microRNA profiles however, may change due to diet, thus we examined the microRNA expression changes in rats following

2-wk consumption.

Since we observed prominent pathways that were altered by WE consumption, we wanted to explore whether these gene changes could be associated by their microRNA targets.

Interestingly, of the target microRNAs detected as influencers of the glutathione-mediated pathways, five microRNAs were detected from the small RNA-Seq in the adipose tissue, but only one survived multiple testing correction. MicroRNA-125b-5p was downregulated (-1.4- fold) by the WE-based diet, with a non-adjusted P-value of (P=0.007). microRNA-125b-5p does logically link with the upregulation in Gstz1 and Gstp1 in the adipocytes.

The biological significance of microRNA-125b-5p in the adipocytes has been examined in human and animal models. Brovkina et al. reported an upregulation of microRNA-125b-5p in the subcutaneous adipose tissue in individuals T2D and obesity [51]. Additionally, in a population with T1D, Satake and others [52] reported a strong positive correlation between hemoglobinA1c and the expression of microRNA-125b-5p, and this has also been reported in a diabetic db/db mouse model [53]. Future mechanistic studies exploring the relationship between microRNA-125b-5p and glutathione metabolism in a diabetic animal model are warranted.

In this study, we examined the effects of rats consuming a WE-based diet for 2 wk on gene expression in multiple tissues, revealing 74 novel DEGs across the PFC, liver, adipose and kidney transcriptomes in Sprague Dawley rats using Illumina HiSeq 3000 platform. In the adipose, KEGG analyses highlighted that glutathione metabolism was upregulated by WE consumption accompanied by downregulated microRNA-125b-5p in the adipose tissue of rats fed a WE-based diet. GO analysis showed that multiple reduction-oxidation reactions in the PFC were also altered by consuming WE. These results indicate that WE-based diets attenuate the expression of glutathione and oxidative-reduction processes because of WE-based diets.

Conclusions

In this study, we examined the effects of rats consuming a WE-based diet for 2 wk on gene expression in multiple tissues, revealing 72 novel DEGs across the PFC, liver, adipose and kidney transcriptomes in Sprague Dawley rats using Illumina HiSeq 3000 platform. Across all tissues, KEGG analyses highlighted that glutathione metabolism was upregulated by WE consumption and GO analysis showed that multiple reduction-oxidation reactions in the PFC were also altered by consuming WE. Two gene targets, Sqle and Akr1c14, were significantly downregulated. These publicly available sequencing data may serve as a valuable resource as the scientific community continues to work toward personalized nutrition by contributing to our understanding of the nutrigenomic response of dietary whole eggs.

Authors’ Contributions

All authors contributed to the study design. JW, AB, CC, and BV performed animal maintenance, prepared the animal diets, and conducted the experiments. JW and AB performed the sequencing preparations, statistical/bioinformatics analyses, and prepared the original version of this manuscript. All authors reviewed, edited, and approved the final manuscript.

Disclosures & Competing Interests

There are no competing interests to disclose.

Funding Information

This work is partially supported by the Presidential Interdisciplinary Research Seed Grant

Program at Iowa State University and by the College of Human Sciences Intramural

Collaborative Seed Grant at Iowa State University, Ames, IA 50011. JW was supported by a

National Science Foundation GRFP. The funding agencies did not play a role in designing the study, collecting data/analysis, the decision to publish, or writing any aspects of the manuscript.

Acknowledgements

We would like to thank Dr. Peng Liu for her insightful comments during the planning stages of this project, as well as Kevin Cavallin, Tanya Murtha and Dr. Mike Baker for their assistance with the sequencing experiments. We would also like to thank all the undergraduate students who helped throughout the duration of this project.

References

1. Park, S. J., Jung, J. H., Choi, S. W. & Lee, H. J. Association between egg consumption and metabolic disease. Korean Journal for Food Science of Animal Resources 38, 209– 223 (2018).

2. Andersen, C. J. Bioactive egg components and inflammation. Nutrients 7, 7889–7913 (2015).

3. Nimalaratne, C. & Wu, J. Hen egg as an antioxidant food commodity: A review. Nutrients 7, 8274–8293 (2015).

4. Wallace, T. C. & Fulgoni, V. L. Usual choline intakes are associated with egg and protein food consumption in the United States. Nutrients 9, (2017).

5. MicroRNAanda, J. M. et al. Egg and egg-derived foods: Effects on human health and use as functional foods. Nutrients 7, 706–729 (2015).

6. Alexander, D. D., Miller, P. E., Vargas, A. J., Weed, D. L. & Cohen, S. S. Meta-analysis of Egg Consumption and Risk of Coronary Heart Disease and Stroke. Journal of the American College of Nutrition 35, 704–716 (2016).

7. Zhong, V. W. et al. Associations of Dietary Cholesterol or Egg Consumption with Incident Cardiovascular Disease and Mortality. JAMA - J. Am. Med. Assoc. 321, 1081– 1095 (2019).

8. Drouin-Chartier, J. P. et al. Egg consumption and risk of cardiovascular disease: Three large prospective US cohort studies, systematic review, and updated meta-analysis. The BMJ 368, (2020).

9. Dehghan, M. et al. Association of egg intake with blood lipids, cardiovascular disease, and mortality in 177,000 people in 50 countries. Am. J. Clin. Nutr. (2020). doi:10.1093/ajcn/nqz348

10. Fuller, N. R. et al. Effect of a high-egg diet on cardiometabolic risk factors in people with type 2 diabetes: The Diabetes and Egg (DIABEGG) Study - Randomized weight-loss and follow-up phase. Am. J. Clin. Nutr. 107, 921–931 (2018).

11. DiMarco, D. M. et al. Intake of up to 3 Eggs/Day Increases HDL Cholesterol and Plasma Choline While Plasma Trimethylamine-N-oxide is Unchanged in a Healthy Population. Lipids 52, 255–263 (2017).

12. Yousr, M. & Howell, N. Antioxidant and ACE inhibitory bioactive peptides purified from egg yolk proteins. Int. J. Mol. Sci. 16, 29161–29178 (2015).

13. Nataraj, J., Manivasagam, T., Justin Thenmozhi, A. & Essa, M. M. Lutein protects dopaminergic neurons against MPTP-induced apoptotic death and motor dysfunction by ameliorating mitochondrial disruption and oxidative stress. Nutr. Neurosci. 19, 237–246 (2016).

14. Akbaraly, N. T., Faure, H., Gourlet, V., Favier, A. & Berr, C. Plasma carotenoid levels and cognitive performance in an elderly population: results of the EVA Study. J. Gerontol. A. Biol. Sci. Med. Sci. 62, 308–16 (2007).

15. Binawade, Y. & Jagtap, A. Neuroprotective effect of lutein against 3-nitropropionic acid- induced huntington’s disease-like symptoms: Possible behavioral, biochemical, and cellular alterations. J. Med. Food 16, 934–943 (2013).

16. Saande, C. J., Webb, J. L., Curry, P. E., Rowling, M. J. & Schalinske, K. L. Dietary Whole Egg Reduces Body Weight Gain in a Dose-Dependent Manner in Zucker Diabetic Fatty Rats. J. Nutr. 149, 1766–1775 (2019).

17. Saande, C. J. et al. Dietary Whole Egg Consumption Attenuates Body Weight Gain and Is More Effective than Supplemental Cholecalciferol in Maintaining Vitamin D Balance in Type 2 Diabetic Rats. J. Nutr. jn254193 (2017). doi:10.3945/jn.117.254193

18. BBMap: A Fast, Accurate, Splice-Aware Aligner. Available at: https://escholarship.org/uc/item/1h3515gn. (Accessed: 28th January 2020)

19. Dobin, A. et al. STAR: Ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).

20. Farrell, D. smallrnaseq : short non coding RNA-seq analysis with Python. Bioarxiv 110585 (2017). doi:10.1101/110585

21. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, (2014).

22. Benjamini, Y. & Hochberg, Y. Benjamini-1995.pdf. Journal of the Royal Statistical Society B 57, 289–300 (1995).

23. Kanehisa, M. & Goto, S. KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic Acids Research 28, (2000).

24. DAVID Functional Annotation Bioinformatics Microarray Analysis. Available at: https://david.ncifcrf.gov/home.jsp. (Accessed: 23rd March 2020)

25. Ensembl genome browser 99. Available at: https://useast.ensembl.org/index.html. (Accessed: 25th March 2020)

26. Livak, K. J. & Schmittgen, T. D. Analysis of relative gene expression data using real-time quantitative PCR and the 2-ΔΔCT method. Methods 25, 402–408 (2001).

27. Lee, M. et al. Hen egg lysozyme attenuates inflammation and modulates local gene expression in a porcine model of dextran sodium sulfate (DSS)-Induced Colitis. J. Agric. Food Chem. 57, 2233–2240 (2009).

28. Jones, S. K., Koh, G. Y., Rowling, M. J. & Schalinske, K. L. Whole Egg Consumption Prevents Diminished Serum 25-Hydroxycholecalciferol Concentrations in Type 2 Diabetic Rats. J. Agric. Food Chem. 64, 120–124 (2016).

29. Boone, L. R., Brooks, P. A., Niesen, M. I. & Ness, G. C. Mechanism of Resistance to Dietary Cholesterol. J. Lipids 2011, 1–9 (2011).

30. Garcia-Bermudez, J. et al. Squalene accumulation in cholesterol auxotrophic lymphomas prevents oxidative cell death. Nature 567, 118–122 (2019).

31. Ding, J. et al. Alterations of a cellular cholesterol metabolism network are a molecular feature of obesity-related type 2 diabetes and cardiovascular disease. Diabetes 64, 3464– 3474 (2015).

32. Seiki, S. & Frishman, W. H. Pharmacologic inhibition of squalene synthase and other downstream enzymes of the cholesterol synthesis pathway: A new therapeutic approach to treatment of hypercholesterolemia. Cardiology in Review 17, 70–76 (2009).

33. Ge, Q. et al. Mulberry leaf regulates differentially expressed genes in diabetic mice liver based on RNA-seq analysis. Front. Physiol. 9, (2018).

34. Ge, Q. et al. RNA-Seq analysis of the pathogenesis of STZ-induced male diabetic mouse liver. J. Diabetes Complications 34, 107444 (2020).

35. Pass, G. J. et al. Effect of hyperinsulinemia and type 2 diabetes-like hyperglycemia on expression of hepatic cytochrome p450 and glutathione s-transferase isoforms in a New Zealand obese-derived mouse backcross population. J. Pharmacol. Exp. Ther. 302, 442– 50 (2002).

36. Blackburn, A. C. et al. Deficiency of glutathione transferase zeta causes oxidative stress and activation of antioxidant response pathways. Mol. Pharmacol. 69, 650–657 (2006).

37. Savic-Radojevic, A. et al. Glutathione S-transferase (GST) polymorphism could be an early marker in the development of PCOS: An insight from non-obese and non-insulin resistant adolescents. Endokrynol. Pol. 69, 366–374 (2018).

38. Dastidar, S. G. et al. Glutathione s-transferase p deficiency induces glucose intolerance via jnk-dependent enhancement of hepatic gluconeogenesis. Am. J. Physiol. - Endocrinol. Metab. 315, E1005–E1018 (2018).

39. Young, D., Fan, M. Z. & Mine, Y. Egg Yolk Peptides Up-regulate Glutathione Synthesis and Antioxidant Enzyme Activities in a Porcine Model of Intestinal Oxidative Stress. J. Agric. Food Chem. 58, 7624–7633 (2010).

40. Garcés-Rimón, M. et al. Pepsin Egg White Hydrolysate Ameliorates Obesity-Related Oxidative Stress, Inflammation and Steatosis in Zucker Fatty Rats. PLoS One 11, e0151193 (2016).

41. Sajadi Hezaveh, Z., Sikaroudi, M. K., Vafa, M., Clayton, Z. S. & Soltani, S. Effect of egg consumption on inflammatory markers: a systematic review and meta-analysis of randomized controlled clinical trials. Journal of the Science of Food and Agriculture 99, 6663–6670 (2019).

42. David Spence, J. Dietary cholesterol and egg yolk should be avoided by patients at risk of vascular disease. J. Transl. Intern. Med. 4, 20–24 (2016).

43. Jiayu, Y. et al. Egg white consumption increases GSH and lowers oxidative damage in 110-week-old geriatric mice hearts. J. Nutr. Biochem. 76, 108252 (2020).

44. Njike, V., Faridi, Z., Dutta, S., Gonzalez-Simon, A. L. & Katz, D. L. Daily egg consumption in hyperlipidemic adults - Effects on endothelial function and cardiovascular risk. Nutr. J. 9, 28 (2010).

45. Zempleni, J., Baier, S. R., Howard, K. M. & Cui, J. Gene regulation by dietary microRNAs. Canadian Journal of Physiology and Pharmacology 93, 1097–1102 (2015).

46. Baier, S. R., Nguyen, C., Xie, F., Wood, J. R. & Zempleni, J. MicroRNAs Are Absorbed in Biologically Meaningful Amounts from Nutritionally Relevant Doses of Cow Milk and Affect Gene Expression in Peripheral Blood Mononuclear Cells, HEK-293 Kidney Cell Cultures, and Mouse Livers. J. Nutr. 144, 1495–1500 (2014).

47. Witwer, K. W. & Zhang, C. Y. Diet-derived microRNAs: Unicorn or silver bullet? Genes and Nutrition 12, 15 (2017).

48. Mosallaei, M. et al. PBMCs: a new source of diagnostic and prognostic biomarkers. Arch. Physiol. Biochem. 1–7 (2020). doi:10.1080/13813455.2020.1752257

49. Akbar, N. et al. Endothelium-derived extracellular vesicles promote splenic monocyte mobilization in myocardial infarction. JCI insight 2, (2017).

50. Fulga, T. A. et al. A transgenic resource for conditional competitive inhibition of conserved Drosophila microRNAs. Nat. Commun. 6, 7279 (2015).

51. Brovkina, O. et al. Role of MicroRNAs in the Regulation of Subcutaneous White Adipose Tissue in Individuals With Obesity and Without Type 2 Diabetes. Front. Endocrinol. (Lausanne). 10, (2019).

52. Satake, E. et al. Circulating microRNA profiles associated with hyperglycemia in patients with type 1 diabetes. Diabetes 67, 1013–1023 (2018).

53. Villeneuve, L. M. et al. Enhanced levels of microRNA-125b in vascular smooth muscle cells of diabetic db/db mice lead to increased inflammatory gene expression by targeting the histone methyltransferase Suv39h1. Diabetes 59, 2904–2915 (2010).

Tables & Figures

Table 3-1 Diet Composition of the WE- and CAS-based diets.

CAS WE

Ingredient (g/kg)

Casein 200 0

Whole Egg - 435

Cornstarch 417 365

Glucose monohydrate 150 150

Lard - 200

Corn Oil 183 0

Mineral Mix 35 35

Vitamin Mix 10 10

Choline Bitartrate 2 2

L-methionine 3 3

Biotin (1%) - 0.4

Table 3-2 Differentially expressed genes

Gene FDR-adjusted P- Tissue Ensembl_ID Gene Name Log2Fold Symbol value Adipose Upregulated ENSRNOG00000018237 Gstp1 glutathione S-transferase pi 1 1.89 1.65E-05 ENSRNOG00000011250 Inmt indolethylamine N-methyltransferase 1.78 3.42E-05

ENSRNOG00000013484 Gsta3 glutathione S-transferase alpha-3 1.37 3.20E-03 76

ENSRNOG00000058571 N/A unclassified 1.14 3.76E-02 ectonucleoside triphosphate ENSRNOG00000033206 Entpd5 1.10 2.57E-02 diphosphohydrolase 5 ENSRNOG00000032745 Slc17a3 solute carrier family 17 member 3 1.06 5.08E-02 ENSRNOG00000011573 Csad cysteine sulfinic acid decarboxylase 1.01 2.49E-03 ENSRNOG00000008755 Acox1 acyl-CoA oxidase 1 0.98 1.07E-02 ENSRNOG00000047708 Gstz1 glutathione S-transferase zeta 1 0.92 2.08E-02 Adipose Downregulated ENSRNOG00000017672 Akr1c14 aldo-keto reductase family 1, member C14 -3.43 2.66E-16 ENSRNOG00000043451 Spp1 secreted phosphoprotein 1 -3.14 3.20E-03 ENSRNOG00000010047 Ddit4l DNA-damage-inducible transcript 4-like -2.56 5.51E-03 ENSRNOG00000013704 Cps1 carbamoyl-phosphate synthase 1 -1.90 1.72E-03 methylenetetrahydrofolate dehydrogenase ENSRNOG00000010833 Mthfd2 (NADP+ dependent) 2, -1.79 3.24E-02 methenyltetrahydrofolate cyclohydrolase ENSRNOG00000058739 Snn stannin -1.56 5.51E-03

Table 3-2 Continued

ENSRNOG00000004626 Slc34a2 solute carrier family 34 member 2 -1.54 4.45E-03 ENSRNOG00000014453 Anxa5 annexin A5 -1.52 3.81E-06 prostate androgen-regulated mucin-like ENSRNOG00000002579 Parm1 -1.50 3.81E-03 protein 1 proline and arginine rich end leucine rich ENSRNOG00000003120 Prelp -1.48 3.63E-03 repeat protein ENSRNOG00000015550 Ptgds prostaglandin D2 synthase -1.35 2.45E-02 ENSRNOG00000018351 Thap4 THAP domain containing 4 -1.22 3.81E-03 ENSRNOG00000009019 Slc6a6 solute carrier family 6 member 6 -1.19 5.08E-02 Liver Upregulated

G protein-coupled receptor, class C, group 77

ENSRNOG00000003144 Gprc5c 2.13 4.43E-02 5, member C Liver Downregulated ENSRNOG00000019422 Egr1 early growth response 1 -2.06 1.04E-02 Brain Upregulated ENSRNOG00000010262 Hdc histidine decarboxylase 2.67 2.77E-08 ENSRNOG00000011250 Inmt indolethylamine N-methyltransferase 2.36 2.76E-07 ENSRNOG00000000961 Glt1d1 glycosyltransferase 1 domain containing 1 1.95 4.70E-04 ENSRNOG00000061527 Gck glucokinase 1.94 1.29E-03 ENSRNOG00000013851 Spry4 sprouty RTK signaling antagonist 4 1.90 5.19E-03 ENSRNOG00000010337 Slc13a2 solute carrier family 13 member 2 1.79 1.70E-02 ENSRNOG00000013552 Scd stearoyl-CoA desaturase 1.78 7.50E-03 ENSRNOG00000020869 mrpl9 mitochondrial ribosomal protein L9 1.70 4.23E-02 acyl-CoA synthetase medium-chain family ENSRNOG00000032246 Acsm3 1.69 1.03E-02 member 3 cytochrome P450, family 4, subfamily a, ENSRNOG00000009597 Cyp4a1 1.62 1.34E-03 polypeptide 1

Table 3-2 Continued

cytochrome P450, family 2, subfamily c, ENSRNOG00000021924 Cyp2c22 1.55 3.56E-02 polypeptide 22 ENSRNOG00000057072 Slc12a3 solute carrier family 12 member 3 1.46 1.70E-02 ENSRNOG00000045649 Arrdc3 arrestin domain containing 3 1.36 6.69E-03 ENSRNOG00000000978 N/A unclassified 1.26 8.58E-03 protein tyrosine phosphatase, receptor type, ENSRNOG00000019587 Ptprn 1.17 1.86E-02 N ENSRNOG00000011648 Aqp1 aquaporin 1 1.17 2.70E-02 ENSRNOG00000018937 Gstm2 glutathione S-transferase mu 2 0.94 7.05E-05

ENSRNOG00000003515 Ephx1 epoxide hydrolase 1 0.93 1.27E-02 78

ENSRNOG00000009421 Ivd isovaleryl-CoA dehydrogenase 0.93 4.42E-02 ENSRNOG00000004009 Xpnpep2 X-prolyl aminopeptidase 2 0.92 9.78E-04 ENSRNOG00000013949 NADP isocitrate dehydrogenase 0.92 4.04E-02 ENSRNOG00000010017 Wee1 WEE1 G2 checkpoint kinase 0.79 3.19E-03 ENSRNOG00000000645 Reep3 receptor accessory protein 3 0.76 4.59E-02 ENSRNOG00000011747 Tmem205 transmembrane protein 205 0.71 2.70E-02 ENSRNOG00000003038 Sft2d2 SFT2 domain containing 2 0.53 2.21E-02 Brain Downregulated ENSRNOG00000009550 Sqle squalene epoxidase -4.96 3.53E-14 ENSRNOG00000016690 Idi1 isopentenyl-diphosphate delta isomerase 1 -2.71 3.53E-14 ENSRNOG00000020480 fasn fatty acid desaturase 1 -2.61 5.31E-14 ENSRNOG00000012819 Gdnf glial cell derived neurotrophic factor -2.48 1.65E-03 ENSRNOG00000007234 Cyp51 cytochrome P450, family 51 -2.06 3.53E-14 ENSRNOG00000006859 Insig1 insulin induced gene 1 -1.96 8.06E-03 proprotein convertase subtilisin/kexin type ENSRNOG00000006280 Pcsk9 -1.89 6.29E-05 9

Table 3-2 Continued

3-hydroxy-3-methylglutaryl-CoA synthase ENSRNOG00000016552 Hmgcs1 -1.77 3.11E-07 1 ENSRNOG00000036615 RGD1560242 similar to RIKEN cDNA 1700028P14 -1.76 3.56E-02 ENSRNOG00000032297 Msmo1 methylsterol monooxygenase 1 -1.74 3.76E-14 ENSRNOG00000011622 Echdc1 ethylmalonyl-CoA decarboxylase 1 -1.65 3.32E-03 ENSRNOG00000005871 Il1rn interleukin 1 receptor antagonist -1.60 6.69E-03 ENSRNOG00000043377 Fdps farnesyl diphosphate synthase -1.39 3.81E-03 ENSRNOG00000006787 Dhcr24 24-dehydrocholesterol reductase -1.36 9.88E-03 ENSRNOG00000045636 Fasn fatty acid synthase -1.34 9.14E-04

ENSRNOG00000020704 Tkfc triokinase and FMN cyclase -1.32 3.07E-02 ENSRNOG00000016924 Acly ATP citrate lyase -1.30 2.15E-04 ENSRNOG00000016122 Hmgcr 3-hydroxy-3-methylglutaryl-CoA reductase -1.25 1.43E-04 acyl-CoA synthetase short-chain family ENSRNOG00000018755 Acss2 -1.20 9.88E-04 member 2 ENSRNOG00000032508 Acot5 acyl-CoA thioesterase 5 -1.09 1.95E-02 ENSRNOG00000002212 17-beta hydroxysteroid -1.08 2.36E-02 ENSRNOG00000023348 Tbc1d2 TBC1 domain family, member 2 -1.08 4.42E-02 ENSRNOG00000000658 Acacb acetyl-CoA carboxylase beta -1.02 2.59E-02 ENSRNOG00000013387 Tpcn2 two pore segment channel 2 -0.92 4.04E-02 diazepam binding inhibitor, acyl-CoA ENSRNOG00000046889 Dbi -0.89 3.39E-02 binding protein NADP-dependent steroid dehydrogenase- ENSRNOG00000057814 Nsdhl -0.64 8.06E-04 like emopamil binding protein (sterol ENSRNOG00000004903 Ebo -0.64 2.60E-02 isomerase)

Table 3-3 GO Pathway Analysis Comparing Whole Egg to Casein

Category GO_ID Term Count P- Gene Symbols Value Adipose GO:1903825 organic acid transmembrane transport 2 4.80E- Slc17a3, Slc6a6 03 GO:0044341 sodium-dependent phosphate transport 2 7.70E- Slc17a3, Slc34a2 03 GO:0006817 phosphate ion transport 2 9.70E- Slc17a3, Slc34a2 03 GO:0035435 phosphate ion transmembrane transport 2 1.70E- Slc17a3, Slc34a2 80

02 GO:0006749 glutathione metabolic process 2 4.90E- Gstz1, Gstp1 02 GO:0043200 response to amino acid 2 5.50E- Cps1, Gstp1 02 GO:0048545 response to steroid hormone 2 5.80E- Cps1, Spp1 02 GO:0098869 cellular oxidant detoxification 2 6.50E- Gstz1, Gstp1 02 GO:0035725 sodium ion transmembrane transport 2 6.60E- Slc17a3, Slc34a2 02 GO:0005436 sodium:phosphate symporter activity 2 6.70E- Slc17a3, Slc34a2 03 GO:0015321 sodium-dependent phosphate transmembrane 2 7.70E- Slc17a3, Slc34a2 transporter activity 03 GO:0042301 phosphate ion binding 2 1.10E- Mthfd2, Slc34a2 02

Table 3-3 – Continued

GO:0004602 glutathione peroxidase activity 2 2.00E- Gstz1, Gstp1 02 GO:0005504 fatty acid binding 2 2.60E- Ptgds, Acox1 02 GO:0004364 glutathione transferase activity 2 4.10E- Gstz1, Gstp1 02 KEGG Metabolic pathways 6 8.70E- Cps1, Mthfd2, 03 Csad, Ptgds, Gstz1, Acox1 Brain GO:0006695 cholesterol biosynthetic process 8 3.90E- Hmgcr, Hmgcs1, 13 sterol isomerase,

Nsdhl, Dhcr24, 81

Idi1, Fdps GO:0006633 fatty acid biosynthetic process 6 1.10E- Scd, Acsm3, Acacb, 07 Fasn, Acly, Msmo1 GO:0055114 oxidation-reduction process 12 8.70E- 17-beta, Scd, 07 Hmgcr, Fads1, Fasn, Sqle, Cyp2c22, Cyp51, Nsdhl, Dhcr24, Cyp4a1, Msmo1

GO:0016126 sterol biosynthetic process 4 5.10E- Insig1, Sqle, sterol 06 isomerase, Msmo1

GO:0008610 lipid biosynthetic process 4 8.00E- Scd, Fasn, Acly, 06 Acss2

GO:0008203 cholesterol metabolic process 5 2.30E- Pcsk9, Insig1, Sqle, 05 Nsdhl, Dhcr24

GO:0008299 isoprenoid biosynthetic process 4 2.30E- Hmgcr, Hmgcs1, 05 Idi1, Fdps

Table 3-3 – Continued

GO:0006084 acetyl-CoA metabolic process 3 1.50E- Acacb, Fasn, Acly 04

GO:0006641 triglyceride metabolic process 4 1.70E- Pcsk9, Scd, Insig1, 04 Dbi GO:0014070 response to organic cyclic compound 6 7.80E- Hmgcs1, Fads1, 04 Acacb, Ephx1, Il1rn, Gstm2

GO:0006637 acyl-CoA metabolic process 3 2.50E- Acsm3, Acot5, Dbi 03 GO:0019932 second-messenger-mediated signaling 2 8.00E- Gck, Ptprn 03 82

GO:0043588 skin development 3 9.90E- Arrdc3, Dhcr24, 03 Dbi GO:0006085 acetyl-CoA biosynthetic process 2 1.10E- Acly, Acss2 02 GO:0031999 negative regulation of fatty acid beta-oxidation 2 1.30E- Acacb, Dbi 02 GO:0006725 cellular aromatic compound metabolic process 2 1.90E- Ephx1, Sqle 02 GO:0030157 pancreatic juice secretion 2 2.10E- Aqp1, Dbi 02 GO:0006629 lipid metabolic process 3 2.40E- Fads1, Il1rn, Acly 02

GO:0009725 response to hormone 3 2.90E- Hmgcs1, Aqp1, 02 Dhcr24

GO:0021670 lateral ventricle development 2 3.20E- Aqp1, Dbi 02 GO:0070723 response to cholesterol 2 3.40E- Hmgcs1, Fdps 02

Table 3-3 – Continued

GO:0046889 positive regulation of lipid biosynthetic process 2 3.70E- 17-beta, Dbi 02 GO:0006636 unsaturated fatty acid biosynthetic process 2 3.70E- Scd, Fads1 02 GO:0032869 cellular response to insulin stimulus 3 4.30E- Gck, Pcsk9, Insig1 02

GO:0042493 response to drug 5 5.20E- Hmgcs1, Aqp1, 02 Acacb, Il1rn, Fdps

GO:0001889 liver development 3 5.80E- Pcsk9, Hmgcs1, 02 Ephx1

GO:0008584 male gonad development 3 5.90E- Hmgcs1, Gdnf, 83 02 Fdps GO:0046835 carbohydrate phosphorylation 2 6.20E- Gck, Tkfc 02 GO:0010033 response to organic substance 3 6.30E- Hmgcs1, Fads1, 02 Sqle GO:0006694 steroid biosynthetic process 2 8.20E- Cyp51, Dbi 02 GO:0019369 arachidonic acid metabolic process 2 8.20E- Fads1, Cyp4a1 02 GO:0030073 insulin secretion 2 9.00E- Ptprn, Il1rn 02 GO:0070542 response to fatty acid 2 9.90E- Scd, Insig1 02 GO:0005506 iron ion binding 5 1.80E- Scd, Cyp2c22, 03 Cyp51, Cyp4a1, Msmo1

GO:0016491 oxidoreductase activity 4 4.30E- 17-beta, Scd, 03 Fads1, Dhcr24

Table 3-3 – Continued

GO:0050660 flavin adenine dinucleotide binding 3 1.40E- Sqle, Dhcr24, Ivd 02

GO:0000287 magnesium ion binding 4 1.70E- Gck, NADP, Wee1, 02 Idi1

GO:0016831 carboxy-lyase activity 2 2.80E- Echdc1, Hdc 02 GO:0070402 NADPH binding 2 4.10E- Hmgcr, Fasn 02

GO:0005215 transporter activity 3 4.40E- Slc13a2, Aqp1, 02 Slc12a3

GO:0042803 protein homodimerization activity 6 5.50E- Hmgcr, Hmgcs1, 84 02 Fasn, Hdc, Gdnf, Gstm2 GO:0046872 metal ion binding 8 6.10E- Scd, Acsm3, Acacb, 02 Xpnpep2, Acly, Idi1, Tkfc, Fdps

GO:0020037 heme binding 3 6.80E- Cyp2c22, Cyp51, 02 Cyp4a1

GO:0019899 enzyme binding 4 7.90E- Ephx1, Dhcr24, 02 Gstm2, Slc12a3

GO:0000062 fatty-acyl-CoA binding 2 9.40E- Ivd, Dbi 02

Table 3-3 - Continued

KEGG Metabolic pathways 23 9.50E- Gck, Hmgcs1, 10 Acsm3, Acacb, Fasn, NADP, Sqle, Hdc, Acly, Acot5, Nsdhl, Fdps Cyp51, Idi1, Acss2, Tkfc

Dhcr24, Ivd 85 Cyp4a1, Hmgcr, Cyp2c22, Msmo1 KEGG Biosynthesis of antibiotics 12 1.60E- Gck, Hmgcr, 09 Hmgcs1, NADP, Sqle, Acly, Acss2, Cyp51, Nsdhl, Idi1, Msmo1, Fdps

KEGG Steroid biosynthesis 6 2.10E- Sqle, sterol 08 isomerase, Cyp51,

Nsdhl, Dhcr24, Msmo1

KEGG Terpenoid backbone biosynthesis 4 1.40E- Hmgcr, Hmgcs1, 04 Idi1, Fdps

KEGG Propanoate metabolism 3 7.00E- Echdc1, Acacb, 03 Acss2 KEGG Biosynthesis of unsaturated fatty acids 3 7.50E- Scd, Fads1, Acot5 03

Table 3-3 – Continued

KEGG Carbon metabolism 4 1.70E- Gck, NADP, Acss2, 02 Tkfc

KEGG AMPK signaling pathway 4 2.00E- Scd, Hmgcr, Acacb, 02 Fasn

KEGG Fatty acid metabolism 3 2.40E- Scd, Fads1, Fasn 02 KEGG Bile secretion 3 4.10E- Hmgcr, Aqp1, 02 Ephx1 KEGG PPAR signaling pathway 3 4.70E- Scd, Cyp4a1, Dbi 02

KEGG Fatty acid biosynthesis 2 6.10E- Acacb, Fasn 86 02

KEGG Chemical carcinogenesis 3 6.30E- Ephx1, Cyp2c22, 02 Gstm2

Table 3-4 MicroRNA Differential Expression Analysis

Tissue MicroRNA Log2Fold Raw P-value

Adipose Upregulated microRNA-221-

3p 0.89 0.024

Adipose Downregulated microRNA-140-

3p -1.01 0.002

microRNA-

125b-5p -1.18 0.007

microRNA-

191a-5p -0.77 0.012

microRNA-10b-

5p -0.49 0.034

Brain Upregulated let-7e-5p 1.34 0.005

microRNA-30a-

3p 0.39 0.030

microRNA-98-

5p 0.96 0.035

Brain Downregulated microRNA-10a-

5p -1.43 0.025

microRNA-10b-

5p -1.35 0.028

Table 3-4 – Continued

microRNA-29a-

3p -0.90 0.036

microRNA-192-

5p -0.82 0.057

Liver Upregulated microRNA-30c-

5p 0.35 0.028

microRNA-30d-

5p 0.32 0.056

Liver Downregulated microRNA-21-

5p -0.44 0.003

microRNA-192-

5p -0.19 0.022

Table 3-5 MicroRNA Gene Targets

Rat Gene non-adj p DIOPT Weighted MicroRNA Tissue Human GeneID LFC Species ID value Score Score Rank microRNA- 2.00E-05 10b-5p Brain ARRDC3 1.85 Rat Arrdc3 12 12.41 High microRNA- 2.81E-03 192-5p Brain AMER1 1.58 Rat Amer1 14 14.27 High 1.10E-02 BLCAP -0.25 Rat Blcap 14 14.27 High 1.16E-02 MYLK -0.21 Rat Mylk 12 12.46 High 1.64E-02 FABP3 1.78 Rat Fabp3 14 14.27 High 1.74E-02 TAOK1 -0.29 Rat Taok1 13 13.34 High 2.01E-02 PCDH17 1.46 Rat Pcdh17 14 14.27 High 2.62E-02 FRMD4B 0.78 Rat Frmd4b 14 14.27 High

3.37E-02 89 KIF1B 0.21 Rat Kif1b 11 11.51 High 3.73E-02 C4orf46 -0.28 Rat RGD1560010 14 14.27 High 4.17E-02 COL5A1 2.00 Rat Col5a1 14 14.27 High 4.86E-02 ZFP36L1 0.17 Rat Zfp36l1 11 11.51 High 5.19E-02 NIPAL1 0.40 Rat Nipal1 11 11.5 High 5.49E-02 PDHB -0.20 Rat Pdhb 14 14.27 High 5.88E-02 SNX33 0.18 Rat Snx33 14 14.27 High microRNA- 0.00E+00 125b-5p Adipose PARM1 -2.24 Rat Parm1 14 14.27 High 4.80E-04 DNAJC14 -0.94 Rat Dnajc14 12 12.41 High

Table 3-6 Primers for qPCR Information.

REV FWD FWD NCBI ID Amplicon Forward Primer Reverse Primer Primer Primer G/C Name Number Length Sequence Sequence Tm Tm Content Fatty acid synthase GGCGAGTCTATGCCAC GCTGATACAGAGAACG (Fasn) NM_017332.1 117 TATTC GATGAG 62 62 52.40%

Indolethylamine N- methyltransferase NM_001109022 CTGGAGAAGGAGACG CGGGCAACCACGAAGT (Inmt) .1 128 GTAGAA ATAA 62 62 52.40% Cytochrome P450, family 2, subfamily c, polypeptide 22 AGAGAGAGAGAGAGA GAGACCCTCTGCATCTC

(Cyp2c22) NC_005100.4 113 GAGAGAGA AATAC 62 62 47.80% 90

Fatty acid desaturase AGAGAGAGAGAGAGA GAGACCCTCTGCATCTC 1 (Fads1) NC_005100.4 113 GAGAGAGA AATAC 62 62 47.80% 18S ribosomal RNA AAGACGAACCAGAGC TCGGAACTACGACGGTA (Rn18s) NR_046237.1 98 GAAAG TCT 62 62 50% ACCAGTCCCAACTCAT ACTCAACAGGGCAAGC EGR1 NC_005117.4 99 CAAAC ATAC 62 62 50 Cytochrome P450, CCTTCCAGTGGTGCTC CTAAGCCACTACCCAAA family 51 (Cyp51) NC_005103.4 93 TTATT GACTATAC 62 62 47.60%

Table 3-7 Weight Gain, Food Intake & Organ Weights

Kidney Liver Adipose Final Body Weight Cumulative Weight Gain Daily Food Intake Casein 0.8 ± 0.1 8.5 ±0.8 3.6 ± 0.4 280.7 ± 2.8 48.2 ± 1.9 15.1 ± 1.7 Whole Egg 0.8 ± 0.1 10.0 ± 1.1 3.3 ± 0.6 287.7 ± 7.2 51.8 ± 3.1 16.0 ± 1.7

Figure 3-1 Research Design Diagram

Figure 3-2 PCA & Hierarchical Clustering

Figure 3-3 Volcano Plots

Figure 3-4 QPCR & Venn Diagram.

Figure 3-5 Protein Interaction Networks.

Figure 3-6 Food Intake & Body Weight Gain.

CHAPTER 4. WHOLE EGG CONSUMPTION INCREASES GENE EXPRESSION WITHIN THE GLUTATHIONE PATHWAY IN THE LIVER OF ZUCKER DIABETIC FATTY RATS

Modified from a manuscript under review for Journal PLOS ONE.

Authors & Affiliations

Joe L. Webb1,2,¶, Amanda E. Bries1,2,¶, Brooke Vogel1, Claudia Carrillo1, Lily Harvison1,

Timothy A. Day3, Michael J. Kimber3, Matthew J. Rowling1,2, Stephanie Clark1, Elizabeth

McNeill1,2,4, and Kevin L. Schalinske1,2,*

1 Department of Food Science and Human Nutrition, Iowa State University, Ames, IA, USA

2 Interdepartmental Graduate Program in Nutritional Sciences, Iowa State University, Ames, IA,

USA

3Department of Biomedical Sciences, Iowa State University College of Veterinary Medicine,

Ames, IA, USA.

4Genetics and Genomics Graduate Program, Iowa State University, Ames, IA, USA

* Corresponding Author

E-mail address: [email protected] (KS)

¶ These authors contributed equally to this work.

Abstract

Background: Nutrigenomics, the interactions between diet and gene expression underlying health and disease, has been proposed as a viable approach toward prevention and treatment of type 2 diabetes mellitus (T2DM). Eggs are nutrient dense food containing bioactive ingredients that modify gene expression and the literature surrounding the influence of egg consumption on globally affected tissues in a T2DM has not been determined. To examine the influence of whole egg consumption on gene expression during T2DM, we examined if whole egg consumption in

T2DM rats alters microRNA and gene expression in the prefrontal cortex, liver, kidney, and adipose tissue.

Methods: Male Zucker Diabetic Fatty (ZDF; fa/fa) rats (n = 12) and their lean controls (fa/+)

(n = 12) were obtained at 6 wk of age. Rats had ad libitum access to water and were randomly assigned to a semi-purified AIN93G casein-based diet (CAS), or a whole egg-based diet (WE) containing dried whole egg powder providing 20% protein (wt:wt). TotalRNA libraries were prepared using QuantSeq 3' mRNA-Seq Library Prep Kit (Lexogen) and smallRNA Libraries were prepared using the SmallRNA-Seq Library Prep Kit (Lexogen) run on an Illumina

HighSeq3000. Differential expression was conducted using DESeq2 in R reference and

Benjamini-Hochberg adjusted p-values controlling FDR at 5% where P< 0.05 were considered significant.

Results: We identified that 7 microRNAs were differentially expressed in response to 8 wk of consuming whole egg-based diets and modified 580 genes, globally. Additionally, 4 of these microRNAs were related have been previously demonstrated to be altered during diabetes and that KEGG/GO pathway analyses highlighted glutathione metabolism was the top upregulated pathway in the liver of ZDF rats. In addition, many these genes were related to metabolic processes including the TCA cycle, glycolysis, and fatty acid metabolism.

100

Conclusion: Across all tissues, glutathione metabolism gene expression was upregulated by whole egg consumption which has been shown to be negatively affected during T2D.

Additionally, we highlighted that endogenous microRNA expression may be altered in numerous tissues following 8-wk consumption of whole egg. These publicly available data provide unique insight into the nutrigenomic response of eggs during T2DM, suggesting that egg consumption may contribute to improving glutathione metabolism during T2DM.

Introduction

Nutrigenomic evidence supports the idea that Type 2 Diabetes Mellitus (T2DM) arises due to the interactions between the transcriptome, individual genetic profiles, lifestyle, and diet.

Since eggs are a nutrient dense food containing bioactive ingredients that modify gene expression, our goal was to examine the role of whole egg consumption on the transcriptome during T2DM. We analyzed whether whole egg consumption in Zucker Diabetic Fatty (ZDF) rats alters microRNA and mRNA expression across the adipose, liver, kidney, and prefrontal cortex tissue. Male ZDF (fa/fa) rats (n = 12) and their lean controls (fa/+) (n = 12) were obtained at 6 wk of age. Rats had ad libitum access to water and were randomly assigned to a modified semi-purified AIN93G casein-based diet or a whole egg-based diet, both providing 20% protein

(w/w). TotalRNA libraries were prepared using QuantSeq 3' mRNA-Seq and Lexogen smallRNA library prep kits and were further sequenced on an Illumina HighSeq3000.

Differential gene expression was conducted using DESeq2 in R and Benjamini-Hochberg adjusted P-values controlling for false discovery rate at 5%. We identified 9 microRNAs and 583 genes that were differentially expressed in response to 8 wk of consuming whole egg-based diets. KEGG/GO pathway analyses demonstrated that 12 genes in the glutathione metabolism pathway were upregulated in the liver and kidney of ZDF rats fed whole egg. WE consumption

101 primarily altered glutathione pathways such as conjugation, methylation, glucuronidation, and detoxification of reactive oxygen species. These pathways have been shown to be negatively affected during T2DM This data provides unique insight into the nutrigenomic response of dietary whole egg consumption during the progression of T2DM.

Methods

IACUC Approval

The animal studies conducted to generate these data were all approved by the

Institutional Animal Care and Use Committee at Iowa State University. All animal care was performed according to Laboratory Animal Resources Guidelines at Iowa State University.

Animal Housing & Experimental Design

Male Zucker Diabetic Fatty (ZDF; fa/fa) rats (n = 12) and their lean control (fa/+) rats

(n = 12) were purchased from Charles River (Wilmington, MA) at 6 wk of age. Rats were dual housed in conventional cages in a temperature-controlled room (25°C) with a 12-h light-dark cycle. Rats were acclimated to the facilities for 72 h while receiving ad libitum access to semi- purified (AIN-93G) diet (Research Diets Inc., New Brunswick, NJ) and water. Rats were randomly assigned an experimental diet as outlined in Table 4 consisting of either a casein-based diet (CAS), or a WE-based diet containing dried WE powder (Rose Acre Farms). Both diets provided 20% protein (wt:wt) from either CAS or WE powder. To match the diets according to lipid content (18.3%), corn oil was added into the control diet. For the remaining of the study, rats were fed ad libitum for 8 wk while body weight and food intake measurements were recorded daily. At the end of the study, the rats were anesthetized with ketamine:xylaxine (90:10 mg/kg body weight) in an intraperitoneal injection of 1µL/g body weight. Whole blood was

102 collected via cardiac puncture and subsequently stored at −80°C for downstream analysis.

Tissues were immediately excised, weighed and frozen in liquid nitrogen for −80°C storage.

RNA Extraction & Analysis

Tissue samples (20 mg) were rapidly thawed on ice and largeRNA and smallRNA fractions were extracted from the same isolate using the RNA SPLIT Kit (Lexogen) according to the manufacturer’s instructions. Briefly, samples were homogenized in an isolation buffer and phase separated using a phenol/chloroform extraction followed by a spin column-based purification procedure. All samples were aliquoted and stored at −80°C for downstream analysis.

Following extraction, sample concentrations for the largeRNA fraction were analyzed using a

Qubit 2.0 fluorometer (Thermo Fisher) using the Qubit™ Broad Range RNA Assay Kit. RNA integrity was assessed using the Bioanalyzer 2100 (Agilent Technologies) and samples with low

RIN values <5 were discarded and extracted again. SmallRNA concentrations were measured using a Qubit 2.0 fluorometer (Thermo Fisher) using the Qubit™ microRNA Assay Kit.

SmallRNA & TotalRNA Sequencing

Libraries for totalRNA were prepared using an automated protocol according to the manufacturer’s instructions for half reactions on the QuantSeq 3' mRNA-Seq Library Prep Kit

(Lexogen) using a MANTIS® Liquid Handler pipetting robot (Formulatrix). All totalRNA samples were multiplexed together across two lanes on an Illumina High-Seq 3000. SmallRNA

Libraries were prepared manually using the SmallRNA-Seq Library Prep Kit (Lexogen). Briefly,

100ng of enriched smallRNA was used as input and 3’ & 5’ adapters were ligated followed by column purifications. Subsequently, the ligation products were reverse transcribed and double stranded cDNA libraries were generated. Finally, individual sample barcodes for multiplexing

103 were introduced via 17 cycles of PCR. All libraries were assessed on the Bioanalyzer 2100

(Agilent) to examine if adapter dimers formed during PCR. All libraries were cleaned using bead purification module (Lexogen) and pooled into a single sample at 2nM (20 µL) for sequencing.

Sequencing Quality Control & Adapter Trimming

For both totalRNA and smallRNA samples, the resulting FASTQ files were analyzed using Fast-QC [19] and sequencing adapters were trimmed using on BBDUK[20] with an example of the trimming procedure: bbduk.sh in=reads.fq out=clean.fq maq=10 ref=

/bbmap/resources/adapters.fa . For smallRNA samples, reads were additionally trimmed using the literal flag to remove the Lexogen specific sequence “5’-TGGAATTCTCGGGTGC

CAAGGAACTCCAGTCAC – 3’” following similar trimming procedures. Briefly, any read segments that matched Illumina Truseq or Nextera adapters were also trimmed out during processing, along with reads whose average quality scores were lower than 10 were also discarded.

Alignment & Read Quantification

For totalRNA, reads were mapped to the Ensembl release 94 of the Rattus Norvegicus

RNO_6.0 genome using RNA STAR [21]. TotalRNA read counts were generated during the read alignment using the --genecounts function in STAR. For smallRNA samples, reference fasta files from www.RNACentral.org were downloaded for microRNA, piwiRNA, snRNA, nRNA, rRNA, and tRNA. Indexes were generated using Bowtie [22] and alignment was conducted using the smallrnaseq python tool [23]. Read counts for all reference indices and IsomicroRNAs were generated using the smallrnaseq python tool.

104

Data Filtering & Normalization

Following read count generation, quantseq data was merged into a single data frame for analysis in R. Genes were discarded from analysis if there were less than 3 samples without a single read for that gene. TotalRNA data originally generated read counts for 32,883 genes and over 50% of the trimmed reads from each sample mapped to the RNO_6 version 94 genome. The microRNA data originally generated read counts for over 350 microRNAs and the formal analysis was conducted on 60 – 150 targets across each tissue. All samples were normalized using the Trimmed Mean of M Values (TMM) method [24] which accounts for variable depth between samples by normalizing using weighted mean log expression ratios across all samples.

Differential Expression Analysis

Differential expression was conducted using DESeq2 from Bioconductor. DESeq -

DataSetFromMatrix generated P values and Benjamini-Hochberg [25] adjusted p-values controlling FDR at 5% and all adjusted p-values < 0.05 were considered significant. All differential expression analyses were conducted using R.

Heatmaps, Principal Component Analysis & Volcano Plots

Principal Component Analysis (PCA) was used to visualize sample relatedness across treatments and tissues. Subsequent hierarchical clustering grouped samples according to transcriptomic relatedness, and volcano plots were constructed to visualize samples with absolute log-fold changes >1.5. Figures were generated with MatplotLib in Python version 3.2.0rc1.

105

KEGG/GO Pathway Analysis

Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis was conducted with the bio2bel-kegg 0.2.5 package in Python. Biological pathways for each DEG were generated, along with categorization according to protein cellular localization and function. Gene

Ontology (GO) Analysis was performed in the Database for Annotation, Visualization, and

Integrated Discovery (DAVID) v6.8 Database.

qPCR Validation Analyses

TotalRNA from each tissue was aliquoted and frozen at -80C, and 2µg of totalRNA was reverse transcribed into cDNA with a High-Capacity cDNA Reverse Transcription Kit (Applied

Biosystems, Catalog # 4368813). cDNA was diluted to 250ng/µL and qPCR reactions were performed using 250ng of total cDNA with primers at 300nM concentration in 10 µL FastStart

Sybr Green Master (Roche) according to the manufacturer’s instructions. Briefly, the thermocycling protocol followed a pre-incubation at 95°C for 600 seconds, followed by 45 cycles of 3-Step Amplification 1) Denature at 95°C for 20 seconds, 2) Anneal and extend at

60°C for 20 seconds and 3) Elongate at 72°C for 20 seconds. All qPCR reactions were conducted in a Roche LightCycler 96 Real-Time PCR System. Primers sequences for qPCR are as follows:

Fatty Acid Synthase FWD: GGCGAGTCTATGCCACTATTC, REV:

GCTGATACAGAGAACGGATGAG; Indolethylamine N-methyltransferase FWD:

CTGGAGAAGGAGACGGTAGAA, REV: CGGGCAACCACGAAGTATAA; Cytochrome

P450, family 2, subfamily c, polypeptide 22 FWD: AGAGAGAGAGAGAGAGAGAGAGA,

REV: GAGACCCTCTGCATCTCAATAC; 18S Ribosomal Subunit FWD:

AAGACGAACCAGAGCGAAAG, REV:TCGGAACTACGACGGTATCT; Cytochrome P450, family 51 FWD: CCTTCCAGTGGTGCTCTTATT, REV: CTAAGCCACTACCCAAAG

106

ACTATAC. In all qPCR experiments, 18s RNA expression was used to normalize gene expression within each tissue sample that was processed in triplicate. All data were analyzed using the Livak Delta-Delta CT method [26].

MicroRNA Bioinformatic Analysis

All microRNA fastq files were processed using the smallrnaseq [23] package in python. smallrnaseq automates standard bioinformatic processes for quantification and analysis of small non-coding RNA species such as microRNA quantification and novel microRNA prediction.

Briefly, smallrnaseq uses bowtie to align fastq files to user defined reference fasta sequences and all reference sequences were downloaded from www.RNAcentral.org version 14. Following alignment to the Rattus Norvigicus genome and reference tRNA, rRNA, microRNA, lncRNA, and snRNA files, novel microRNA predictions are conducted using microRNADeep2.

Additionally, differential expression is automated using the DEseq2 package in R.

Results & Discussion

Whole eggs have predominantly been criticized for their associated risk of developing chronic diseases [23], yet the benefits of WE consumption have also been reported [24]. For instance, several groups have suggested that WE provide antioxidant properties [12,25] , either through antioxidant peptides in the egg yolk [12] or other reactive oxygen species-reducing nutrients [26]. Other studies examining the role of quail egg consumption in rat models of T2DM have demonstrated upregulation of glutathione metabolism in alloxan-induced T2DM in Wistar rats [27] and improved oxidative stress profiles in streptozotocin-injected rats [28]. Raza and colleagues [28] identified that in diabetic rat liver glutathione content and glutathione S- transferase (GST) were decreased 65% because of the T2DM phenotype.

107

Total RNASeq Differential Expression

When comparing the WE versus casein (CAS) in ZDF rats and their lean controls, differential expression analyses of the mRNAseq data resulted in 583 differentially expressed genes (DEGs) across four tissues in both genotypes in Table 1. S1 Table contains the results from DESeq2 with the results for each gene across all four tissues with data on individual genes.

S2 Table contains raw mRNA read counts for each tissue and rat across both genotypes. Among the lean controls, 13 genes were differentially expressed in the adipose tissue, 32 in the liver, and

6 in the kidney. Notably, none of the genes were differentially expressed in the PFC between dietary treatments in the lean rats. In the ZDF rats, dietary WE consumption resulted in 532 total

DEGs across all tissues where 50 genes were differentially expressed in adipose tissue, 474 in the liver, 6 in the kidney and 2 genes in the PFC following multiple testing correction using the false discovery rate (FDR) threshold of 5%. We demonstrated that consuming WE-based diets for 8 wk resulted in significant alterations in oxidative stress pathways, as well as glutathione metabolism pathways. While there were tissue-specific changes in gene expression, glutathione metabolism was altered in the kidney and liver among ZDF rats, and in the kidney of lean controls were significantly upregulated. Overall, these data highlight how consumption of WE- based diets can provide beneficial effects through modifying gene expression of oxidative reduction targets.

We previously demonstrated that WE consumption for 8 wk is effective at improving serum vitamin D status and providing nephroprotective benefits [29,30]; however, despite our gene expression findings in this study we still have yet to elucidate the mechanism underlying how WE consumption leads to decreased weight gain. We also identified that ZDF rats fed WE upregulated 11 genes involved in glutathione metabolism in the liver and kidney. In the PFC,

108

WE consumption had differing effects whereby in the lean PFC, WE consumption did not change the transcriptome, whereas in the ZDF rats WE consumption strongly downregulated the expression of 2, AY172581 exon transcripts. These exon transcripts have yet to be characterized and future proteomic studies may reveal their biological importance. Across both genotypes, the most significantly altered genes were involved in the Kyoto Encyclopedia of Genes and

Genomes (KEGG) pathways of glutathione metabolism, metabolic pathways, steroid biosynthesis, and cholesterol metabolism. After controlling for the genetic background differences of our ZDF rats, a combined analysis indicated that 428 unique genes were differentially expressed across these tissues as a product of WE consumption (data not shown).

Moreover, 13 different glutathione metabolism genes were significantly upregulated across the liver and kidney in both genotypes suggesting that increased egg consumption can increase glutathione metabolism and attenuate the decreased glutathione metabolism during diabetes.

To visualize the global differences in the transcriptomes based on dietary treatment, we performed principal component analysis (PCA) and generated volcano plots for genes that exhibited ≥1.5-fold change, respectively. Fig 1 displays the samples in a three-dimensional principal component space, whereby samples are colored in red or black to distinguish either WE or CAS, respectively. In the mRNA samples, rats on the same dietary treatment (i.e. black or red) clustered together, while animals belonging to different dietary treatments separated, indicating distinctly different patterns across global mRNA expression. These results were further visualized using volcano plots for each tissue as presented in Fig 2. These volcano plots show that the degree to which genes were upregulated or downregulated was varied across each tissue, whereas WE consumption primarily downregulated gene expression in the liver.

109

During T2DM, genes within the oxidative stress-related pathways have been shown to be upregulated [26]. Evans et al. suggested that oxidative stress was driven by the hyperglycemic environment concomitant with increased concentrations of free fatty acids in the plasma [26].

Protective antioxidant genes such as glutathione peroxidase have been shown to be downregulated during T2DM, [34] and both glutathione s-transferases (GSTs) and glutathione- dependent enzymes are important in the regulation of pathophysiological alterations in numerous chronic diseases, especially T2DM [31]. Previous work has shown that dietary intervention with direct glutathione supplementation was protective against diabetic nephropathy in a streptozotocin-induced T2DM model [28]. This current study provides new transcriptomic evidence supporting our previous report demonstrating that WE consumption protects against diabetic nephropathy, where WE consumption leads to altered gene expression in the kidney. In this study we noted that the strongest alterations in glutathione metabolism were altered in the liver, potentially due to previous reports showing that hepatic glutathione is produced at much higher concentrations (10 mM), whereas intracellular glutathione concentrations are approximately 1-2 mM [28]. This body of previous work is important in relation to our findings that several GSTs and glutathione-dependent enzymes are significantly altered during WE consumption in lean controls and during diabetes in the kidneys and livers across both genotypes. Future mechanistic studies identifying the beneficial impact of these two enzymes in chronic diseases like T2DM are warranted.

Outside of the glutathione pathways, we also observed that there were significant differences in early growth response-1 (Egr-1) gene expression following WE consumption. Egr-

1 has been implicated in the onset of insulin-resistance, as previous studies identified that loss of

110 function in Egr-1 restores insulin sensitivity via increased phosphorylation of the insulin receptor substrate-1 tyrosine kinase [32]. Notably, in our ZDF rats fed WE, we observed a 30% decrease in hepatic Egr-1 expression. This is an interesting finding as research by Garnett et al. [33] determined that exposing beta cells to hyperglycemic conditions resulted in a temporal and dose- dependent increase in Egr-1 transcription and translation. Furthermore, Egr-1 null mice are known for their inability of displaying diabetic and obese phenotypes [34] owing to their increased energy expenditure. These data suggest that consumption of WE may lead to altered

Egr-1 expression which may play a key role in regulating energy expenditure.

We also demonstrated that WE consumption resulted in tissue-specific alterations in gene expression and that there were distinct transcriptomic differences between genotypes. WE consumption did not influence gene expression in the PFC of lean animals, while 2 genes were significantly altered in the ZDF PFC. There were more stark differences when comparing the liver tissues between the two genotypes, where more than 400 genes were altered in ZDF livers that were not altered in the liver of lean controls. It has been shown that T2DM impacts a variety of tissues [1] but previous studies have provided very little evidence of how T2DM alters the nutrigenomic responses to foods in specific tissues. It is still unknown which specific egg components lead to phenotypic differences in gene expression and future studies should focus on identifying the specific egg constituents that mediate these gene expression differences. These collective findings are likely mediated through the alteration of several genes; therefore, we aimed to further examine microRNA changes involved in the underlying progression of T2DM during WE consumption.

111

MicroRNA Seq Differential Expression

We examined if endogenously expressed microRNA profiles in the adipose, liver, kidney, and prefrontal cortex tissues would be altered following 8 wk consumption of dietary WE.

Differential expression analyses of the ZDF microRNA data resulted in 1 differentially expressed microRNA in the adipose tissue, none in the liver, none in the kidney and 2 in the PFC that surpassed multiple testing correction. Among the lean rats, there were 2 marginally differentially expressed microRNAs in the adipose tissue, 4 in the liver, none in the kidney and none in the

PFC that survived multiple testing correction. Table 2 presents the differentially expressed microRNAs in the adipose, liver, kidney, and PFC tissues across both genotypes. S3 Table contains results from DESeq2 with the results for each microRNA across all four tissues and raw microRNA read counts are contained in S4 Table.

qPCR Analyses

Confirmatory analysis with qPCR demonstrated that across the 5 genes selected for confirmatory qPCR experiments, the qPCR data highly correlates with the mRNA sequencing results (R2 = 0.72). Supplemental Figure 1 highlights the correlation plot between the log fold changes in qPCR and the Quantseq analysis which displayed a high degree of similarity in the log fold changes detected by these methods.

Mapping Between MicroRNAs & Target Genes

Next, we sought out to determine if these significantly altered microRNAs were responsible for the tissue-specific differential expression of their predicted target genes.

MicroRNA mapping analyses of the differentially expressed microRNAs and their target genes demonstrates that in each of the tissues with differentially expressed microRNAs, key target

112 genes of these microRNAs were altered. For instance, in the lean liver microRNA-181a-3p was upregulated and two of its mRNA target genes were differentially expressed, Cytochrome P450

Family 7 Subfamily A Member 1 (Cyp7a1) and stearoyl-CoA desaturase (Scd). Similarly, in the lean adipose, microRNA-125b-5p was downregulated while its target gene phosphoglycolate phosphatase (Pgp) was upregulated. The microRNAs in the PFC and kidney tissue did not map to any differentially expressed genes. Table 3 summarizes the mapping between microRNAs and their gene targets.

KEGG & GO Functional Enrichment Analysis

To further examine the molecular function of the identified DEGs, KEGG pathway analysis indicated that the most prevalent pathways influenced by dietary WE across multiple tissues in the ZDF rats were: glutathione metabolism; oxidation-reduction; metabolism of xenobiotics; steroid hormone biosynthesis; and fatty acid synthesis pathways. In the livers of lean control rats, the most significantly enriched pathways were metabolic pathways and retinol metabolism. All the differentially expressed genes that map to KEGG and gene ontology (GO) pathways analyses are presented in S5 Table. To further investigate the specific genes involved in the glutathione metabolism pathways, genes were categorized into the corresponding reactions identified by Reactome.org in Fig 3. Glutathione metabolism functions in antioxidant defense, signal transduction, cytokine production, and other cellular processes such as detoxification. The role of GST, GSTK, GSTO dimers, and GPX1 which function in glutathione conjugation, glucuronidation, methylation, and detoxification of reactive oxygen species, respectively, are detailed within Fig 3. These reactions within glutathione metabolism are essential for recycling of glutathione disulfide or the conjugation of GSH that can be utilized in redox reactions.

113

Strengths & Limitations

The strengths and limitations of this study should be addressed to better understand how these results fit into the larger context of the current literature. It has been reported that in

America, males consume on average 5.9 eggs per week, whereas females consume roughly 3.8 eggs per week [24]. The dose of egg used in this study would equate to roughly 14 eggs per day for a human. While our study demonstrated that consuming a large dose of WE may alter gene expression of various metabolic pathways, particularly during T2DM, this quantity of egg would not be a standard dietary practice in humans. However, our laboratory has previously reported in

ZDF rats that even smaller dosages, such as the human equivalent of <2 eggs/day, significantly reduced weight gain in the ZDF rat and therefore may be effective for weight management during T2DM in humans [13]. After the examination of the transcriptome following our high

WE-based diet, it is warranted to examine these specific genes in a follow-up intervention study.

Future studies will focus on titrating down the egg dosages to discern the smallest dosage to elicit similar transcriptomic responses to egg consumption that will be more translatable to human consumption patterns. Overall, our findings are significant as we are the first to report that whole hen egg consumption promotes glutathione metabolism expression during T2DM and alters the transcriptome of multiple tissues using next-generation sequencing. Additionally, we provide evidence supporting the idea that egg consumption modifies endogenous microRNA expression in a tissue-specific manner.

114

Conclusion

In summary, we examined whether feeding WE modifies expression of microRNAs or gene expression profiles across multiple tissues in a diabetic versus a lean rat model. Across all tissues examined with next generation sequencing, we identified that 9 microRNAs were differentially expressed in response to consuming WE. Additionally, we have shown that these microRNAs were related to tissue-specific changes in gene expression, and that 8 wk of consuming diets high in whole egg modified 583 genes across the PFC, kidney, liver, and adipose tissue. KEGG/GO analyses identified that glutathione metabolism was highly upregulated in response to feeding WE and qPCR results validated the sequencing results. These data suggest that high WE consumption may provide beneficial effects during T2DM by improving glutathione metabolism gene expression across multiple tissues and decreasing gene expression in oxidative stress pathways.

Author Contributions

JW, AB, CC, LH, and BV worked with the animals, prepared the diets, and ran the experiments. JW & AB performed the bioinformatics, conducted statistical analyses and wrote the original draft of the manuscript. All authors participated in the study design, as well as revised, edited and approved the final manuscript.

Competing interests

There are no competing interests to disclose.

115

Grant information

This work is partially supported by the National Institutes of Health (NIH) grant and a

National Science Foundation GRFP. The work was also supported by the egg nutrition center

and Iowa State University. The funders had no role in the study design, data collection and

analysis, decision to publish, or preparation of the manuscript.

Acknowledgements

We would like to thank Dr. Peng Liu for aiding in planning this project, and the ISU

DNA facility staff members Kevin Calvalin, Tanya Murtha and Dr. Mike Baker for their

assistance sequencing our samples. Additionally, the authors would like to thank the

undergraduate research assistants that helped conduct the experiments and work with the

animals.

References

1. Zheng Y, Ley SH, Hu FB. Global aetiology and epidemiology of type 2 diabetes mellitus and its complications. Nature Reviews Endocrinology. Nature Publishing Group; 2018. pp. 88–98. doi:10.1038/nrendo.2017.151

2. Folli F, Corradi D, Fanti P, Davalli A, Paez A, Giaccari A, et al. The Role of Oxidative Stress in the Pathogenesis of Type 2 Diabetes Mellitus Micro- and Macrovascular Complications: Avenues for a Mechanistic-Based Therapeutic Approach. Current Diabetes Reviews. 2012;7: 313–324. doi:10.2174/157339911797415585

3. Raza H, John A, Howarth FC. Increased Oxidative Stress and Mitochondrial Dysfunction in Zucker Diabetic Rat Liver and Brain. Cellular Physiology and Biochemistry. 2015;35: 1241– 1251. doi:10.1159/000373947

4. Raza H, John A, Howarth FC. Alterations in glutathione redox metabolism, oxidative stress, and mitochondrial function in the left ventricle of elderly zucker diabetic fatty rat heart. International Journal of Molecular Sciences. 2012;13: 16241–16254. doi:10.3390/ijms131216241

116

5. Kanikarla-Marie P, Micinski D, Jain SK. Hyperglycemia (high-glucose) decreases l-cysteine and glutathione levels in cultured monocytes and blood of Zucker diabetic rats. Molecular and Cellular Biochemistry. 2019;459: 151–156. doi:10.1007/s11010-019-03558-z

6. Tessari P, Coracina A, Kiwanuka E, Vedovato M, Vettore M, Valerio A, et al. Effects of insulin on methionine and homocysteine kinetics in type 2 diabetes with nephropathy. Diabetes. 2005;54: 2968–2976. doi:10.2337/diabetes.54.10.2968

7. Patel D, Rooney R, Groom S. Gene Expression Profiles for the Zucker Fatty Rat Versus Zucker Diabetic Fatty Rat are Highly Consistent with Those Observed in Human Patients. Available: www.criver.com

8. Cordero-Herrera I, Martín MÁ, Goya L, Ramos S. Cocoa intake ameliorates hepatic oxidative stress in young Zucker diabetic fatty rats. Food Research International. 2015;69: 194–201. doi:10.1016/j.foodres.2014.12.039

9. Zou XR, Zhan LR, Chen L, Long QH, Yuan J, Wang L, et al. Influence of the klotho/FGF23/egr1 signaling pathway on calciumphosphorus metabolism in diabetic nephropathy and the intervention of shenyuan granules. Journal of Biological Regulators and Homeostatic Agents. 2019;33: 1695–1702. doi:10.23812/19-207-A

10. Fuller NR, Caterson ID, Sainsbury A, Denyer G, Fong M, Gerofi J, et al. The effect of a high- egg diet on cardiovascular risk factors in people with type 2 diabetes: The Diabetes and Egg (DIABEGG) study- A 3-mo randomized controlled trial. American Journal of Clinical Nutrition. 2015;101: 705–713. doi:10.3945/ajcn.114.096925

11. Djoussé L, Khawaja OA, Gaziano JM. Egg consumption and risk of type 2 diabetes: A meta- analysis of prospective studies. American Journal of Clinical Nutrition. 2016;103: 474–480. doi:10.3945/ajcn.115.119933

12. Yousr M, Howell N. Antioxidant and ACE inhibitory bioactive peptides purified from egg yolk proteins. International Journal of Molecular Sciences. 2015;16: 29161–29178. doi:10.3390/ijms161226155

13. Saande CJ, Webb JL, Curry PE, Rowling MJ, Schalinske KL. Dietary Whole Egg Reduces Body Weight Gain in a Dose-Dependent Manner in Zucker Diabetic Fatty Rats. The Journal of Nutrition. 2019;149: 1766–1775. doi:10.1093/jn/nxz143

14. Virtanen JK, Mursu J, Tuomainen TP, Virtanen HEK, Voutilainen S. Egg consumption and risk of incident type 2 diabetes in men: The kuopio ischaemic heart disease risk factor study. American Journal of Clinical Nutrition. 2015;101: 1088–1096. doi:10.3945/ajcn.114.104109

15. Rong Y, Chen L, Zhu T, Song Y, Yu M, Shan Z, et al. Egg consumption and risk of coronary heart disease and stroke: Dose-response meta-analysis of prospective cohort studies. BMJ (Online). 2013;346. doi:10.1136/bmj.e8539

117

16. Garcés-Rimón M, González C, Uranga JA, López-Miranda V, López-Fandiño R, Miguel M. Pepsin Egg White Hydrolysate Ameliorates Obesity-Related Oxidative Stress, Inflammation and Steatosis in Zucker Fatty Rats. Peterson J, editor. PLOS ONE. 2016;11: e0151193. doi:10.1371/journal.pone.0151193

17. Saande CJ, Steffes MA, Webb JL, Valentine RJ, Rowling MJ, Schalinske KL. Whole Egg Consumption Impairs Insulin Sensitivity in a Rat Model of Obesity and Type 2 Diabetes. Current Developments in Nutrition. 2019;3. doi:10.1093/CDN/NZZ015

18. Pourafshar S, Akhavan NS, George KS, Foley EM, Johnson SA, Keshavarz B, et al. Egg consumption may improve factors associated with glycemic control and insulin sensitivity in adults with pre- and type II diabetes. Food and Function. 2018;9: 4469–4479. doi:10.1039/c8fo00194d

19. Saande CJ, Jones SK, Rowling MJ, Schalinske KL. Whole Egg Consumption Exerts a Nephroprotective Effect in an Acute Rodent Model of Type 1 Diabetes. Journal of Agricultural and Food Chemistry. 2018;66: 866–870. doi:10.1021/acs.jafc.7b04774

20. Dhas Y, Mishra N, Banerjee J. Vitamin D Deficiency and Oxidative Stress in Type 2 Diabetic Population of India. Cardiovascular & Hematological Agents in Medicinal Chemistry. 2017;14: 82–89. doi:10.2174/1871525714666160426150233

21. Wimalawansa SJ. Vitamin D deficiency: Effects on oxidative stress, epigenetics, gene regulation, and aging. Biology. MDPI AG; 2019. doi:10.3390/biology8020030

22. Andrews S. FASTQC. A quality control tool for high throughput sequence data. 2010 [cited 5 Apr 2020]. Available: https://www.bibsonomy.org/person/1f230a919c34360709aa298734d63dca3/author/0

23. Tran NL, Barraj LM, Heilman JM, Scrafford CG. Egg consumption and cardiovascular disease among diabetic individuals: A systematic review of the literature. Diabetes, Metabolic Syndrome and Obesity: Targets and Therapy. Dove Medical Press Ltd.; 2014. pp. 121–137. doi:10.2147/DMSO.S58668

24. Fuller NR, Sainsbury A, Caterson ID, Markovi TP. Egg consumption and human cardiometabolic health in people with and without diabetes. Nutrients. MDPI AG; 2015. pp. 7399– 7420. doi:10.3390/nu7095344

25. Patrick Emeka Aba, Domnic Chibuike Igwebuike, Jonas Anayo Onah. Effects of Various Concentrations of Quail Egg Solution on Glycemia and Antioxidant Parameters of Alloxan- induced Diabetic Rats. Journal of Advances in Medical and Pharmaceutical Sciences. 2015.

26. Evans JL, Goldfine ID, Maddux BA, Grodsky GM. Are oxidative stress - Activated signaling pathways mediators of insulin resistance and β-cell dysfunction? Diabetes. American Diabetes Association; 2003. pp. 1–8. doi:10.2337/diabetes.52.1.1

118

27. Suchocka Z, Kobylińska K, Pachecka J. Activity of glutathione-dependent enzymes in long term diabetes. II. Glutathione contents and activity of glutathione-dependent enzymes: S- transferase and peroxidase in the kidney cytosol of alloxan induced diabetic rats. Acta Pol Pharm. 1995;52.

28. Raza H, Ahmed I, John A, Sharma AK. Modulation of xenobiotic metabolism and oxidative stress in chronic streptozotocin-induced diabetic rats fed with Momordica charantia fruit extract. In: J Biochem Mol Toxicol [Internet]. 2000 [cited 26 Apr 2020]. Available: https://www.ncbi.nlm.nih.gov/pubmed/10711628

29. Quigley JD. Effects of spray-dried whole egg and biotin in calf milk replacer. Journal of Dairy Science. 2002;85: 198–203. doi:10.3168/jds.S0022-0302(02)74068-X

30. Chen X, Du Y, Boni GF, Liu X, Kuang J, Geng Z. Consuming egg yolk decreases body weight and increases serum HDL and brain expression of TrkB in male SD rats. Journal of the Science of Food and Agriculture. 2019;99: 3879–3885. doi:10.1002/jsfa.9610

31. Corbett SW, Stern JS, Keesey RE. Energy expenditure in rats with diet-induced obesity. The American journal of clinical nutrition. 1986;44: 173–80. doi:10.1093/ajcn/44.2.173

32. Shen N, Yu X, Pan FY, Gao X, Xue B, Li CJ. An early response transcription factor, Egr-1, enhances insulin resistance in type 2 diabetes with chronic hyperinsulinism. Journal of Biological Chemistry. 2011;286: 14508–14515. doi:10.1074/jbc.M110.190165

33. Garnett KE, Chapman P, Chambers JA, Waddell ID, Boam DSW. Differential gene expression between Zucker Fatty rats and Zucker Diabetic Fatty rats: a potential role for the immediate- early gene Egr-1 in regulation of beta cell proliferation. Journal of molecular endocrinology. 2005;35: 13–25. doi:10.1677/jme.1.01792

34. Zhang J, Zhang Y, Sun T, Guo F, Huang S, Chandalia M, et al. Dietary obesity-induced Egr-1 in adipocytes facilitates energy storage via suppression of FOXC2. Scientific Reports. 2013;3: 1– 10. doi:10.1038/srep01476

35. Tiangang Li, Erika Owsley, Michelle Matozel, Peter Hsu, John Y.L. Chiang. Transgenic expression of CYP7A1 in the liver prevents high fat diet-induced obesity and insulin resistance in mice | The FASEB Journal. Pharmacology/Experimental Therapeutics. 2010 [cited 17 Jun 2020]. Available: https://www.fasebj.org/doi/abs/10.1096/fasebj.24.1_supplement.570.4

36. Gutiérrez-Juárez R, Pocai A, Mulas C, Ono H, Bhanot S, Monia BP, et al. Critical role of stearoyl-CoA desaturase - 1 (SCD1) in the onset of diet-induced hepatic insulin resistance. Journal of Clinical Investigation. 2006;116: 1686–1695. doi:10.1172/JCI26991

37. Mugabo Y, Zhao S, Seifried A, Gezzar S, Al-Mass A, Zhang D, et al. Identification of a mammalian glycerol-3-phosphate phosphatase: Role in metabolism and signaling in pancreatic β-cells and hepatocytes. Proceedings of the National Academy of Sciences of the United States of America. 2016;113: E430–E439. doi:10.1073/pnas.1514375113

119

38. Zhang M, Zhu X. miR-9-5p plays an important role in gestational diabetes mellitus (GDM) progression by targeting HK-2. Int J Clin Exp Med. 2018. Available: www.ijcem.com/

39. Jassal B, Matthews L, Viteri G, Gong C, Lorente P, Fabregat A, et al. The reactome pathway knowledgebase. Nucleic Acids Research. 2020;48. doi:10.1093/nar/gkz1031

40. Karst H, Steiniger J, Noack R, Steglich HD. Diet-induced thermogenesis in man: Thermic effects of single proteins, carbohydrates and fats depending on their energy amount. Annals of Nutrition and Metabolism. 1984;28: 245–252. doi:10.1159/000176811

41. Kvaløy K, Page CM, Holmen TL. Epigenome-wide methylation differences in a group of lean and obese women – A HUNT Study. Scientific Reports. 2018;8: 1–9. doi:10.1038/s41598-018- 34003-8

42. Chen J, He X, Huang J. Diet Effects in Gut Microbiome and Obesity. Journal of Food Science. 2014;79: R442-51. doi:10.1111/1750-3841.12397

43. Zhu C, Sawrey-Kubicek L, Bardagjy AS, Houts H, Tang X, Sacchi R, et al. Whole egg consumption increases plasma choline and betaine without affecting TMAO levels or gut microbiome in overweight postmenopausal women. Nutrition Research. 2020 [cited 26 Apr 2020]. doi:10.1016/j.nutres.2020.04.002

44. Bushnell B, Rood J, Singer E. BBMerge – Accurate paired shotgun read merging via overlap. PLoS ONE. 2017;12. doi:10.1371/journal.pone.0185056

45. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, et al. STAR: Ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29: 15–21. doi:10.1093/bioinformatics/bts635

46. Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2. Nature Methods. 2012;9: 357–359. doi:10.1038/nmeth.1923

47. Farrell D. smallrnaseq : short non coding RNA-seq analysis with Python. Bioarxiv. 2017; 110585. doi:10.1101/110585

48. Robinson MD, Oshlack A. A scaling normalization method for differential expression analysis of RNA-seq data. Genome Biology. 2010;11: R25. doi:10.1186/gb-2010-11-3-r25

49. Benjamini, Yoav ; Hochberg Y. Controlling the False Discovery Rate - a Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society Series B- Methodological 1995.pdf. Journal of the Royal Statistical Society Series B (Methological). 1995. doi:10.2307/2346101

50. Livak KJ, Schmittgen TD. Analysis of Relative Gene Expression Data Using Real-Time Quantitative PCR and the 2 C T Method. METHODS. 2001;25: 402–408. doi:10.1006/meth.2001.1262

120

Tables & Figures

Table 4-1 Differentially Expressed Genes

Ensembl_ID Genotype Tissue (ENSRNO) Symbol Gene Name L2FC P-value3

Adipose ZDF Downregulated G00000011039 Gch1 GTP cyclohydrolase 1 -2.71 2.1E-05

G00000040108 RGD1565355 similar to fatty acid translocase/CD36 -2.65 2.7E-07

G00000011024 Zdhhc20 zinc finger DHHC-type -2.49 1.2E-04 palmitoyltransferase 20

G00000006946 Arhgap9 Rho GTPase activating protein 9 -2.49 2.6E-06

G00000006715 Ccr1 C-C motif chemokine receptor 1 -2.39 7.4E-06

G00000032546 Dot1l DOT1 like histone lysine -2.12 1.9E-06 methyltransferase

G00000034230 Fcrl1 Fc receptor-like 1 -2.09 6.9E-05

G00000019283 P2ry2 purinergic receptor P2Y2 -2.07 1.2E-04

G00000022975 Nfam1 NFAT activating protein with ITAM -1.92 3.1E-04 motif 1

G00000013917 Igsf10 immunoglobulin superfamily, member -1.91 2.4E-05 10

G00000049115 Ccr5 C-C motif chemokine receptor 5 -1.89 4.3E-07

G00000015895 B4galt6 beta-1,4-galactosyltransferase 6 -1.84 5.1E-05

G00000020479 Pik3c2a phosphatidylinositol-4-phosphate 3- -1.83 2.2E-04 kinase, catalytic subunit type 2 alpha

G00000061379 C7 complement C7 -1.82 1.8E-04

G00000011927 Sdc3 syndecan 3 -1.82 2.3E-04

G00000026644 Glipr1 GLI pathogenesis-related 1 -1.77 6.4E-05

G00000011946 Ptn pleiotrophin -1.74 4.8E-05

G00000013922 Dok2 docking protein 2 -1.61 2.7E-04

G00000013526 Rassf4 Ras association domain family -1.60 3.1E-06 member 4

G00000001989 Alcam activated leukocyte cell adhesion -1.57 1.5E-05 molecule

G00000016643 Lpcat2 lysophosphatidylcholine -1.52 2.9E-04 acyltransferase 2

G00000003835 Slc43a2 solute carrier family 43 member 2 -1.52 1.6E-04

121

Table 4-1 – continued

G00000019077 Lipa lipase A, lysosomal acid type -1.51 3.8E-05

G00000009347 Arhgap25 Rho GTPase activating protein 25 -1.49 2.1E-04

G00000000257 Smpd3 sphingomyelin phosphodiesterase 3 -1.47 2.8E-04

G00000012616 Ppt1 palmitoyl-protein thioesterase 1 -1.41 2.7E-04

G00000009331 Hck HCK proto-oncogene, Src family -1.30 4.6E-05 tyrosine kinase

G00000010183 Gask1b golgi associated kinase 1B -1.26 2.7E-04

G00000017022 Cerk ceramide kinase -1.25 3.2E-04

G00000008465 Tmem176b transmembrane protein 176B -1.23 2.4E-04

G00000010208 Timp1 TIMP metallopeptidase inhibitor 1 -1.20 1.8E-04

ZDF Adipose Ensembl_ID Gene Symbol Gene Name L2FC P-value Upregulated (ENSRNO)

G00000015072 Ptgr1 prostaglandin reductase 1 1.15 2.2E-04

G00000010389 Ndrg2 NDRG family member 2 1.30 2.0E-04

G00000037446 Pxmp2 peroxisomal membrane protein 2 1.30 2.3E-04

G00000002896 Prdx6 peroxiredoxin 6 1.37 3.1E-04

G00000019328 Phgdh phosphoglycerate dehydrogenase 1.45 3.4E-04

G00000021316 Tmem98 transmembrane protein 98 1.48 2.1E-04

G00000046858 MGC109340 similar to Microsomal signal peptidase 1.52 9.7E-05 23 kDa subunit (SPase 22 kDa subunit) (SPC22/23)

G00000017012 Coq7 coenzyme Q7, hydroxylase 1.56 4.5E-05

G00000021524 Mrap melanocortin 2 receptor accessory 1.69 1.6E-04 protein

G00000017226 Slc2a4 solute carrier family 2 member 4 1.87 1.8E-04

G00000002579 Parm1 prostate androgen-regulated mucin protein1 1.89 7.0E-06

G00000001001 Retn resistin 1.95 3.0E-05

G00000008615 Mal2 mal, T-cell differentiation protein 2 2.09 1.6E-04

122

Table 4-1 – continued

G00000019412 Rhbg Rh family B glycoprotein 2.12 1.3E-05

G00000009715 Me1 malic enzyme 1 2.13 2.0E-06

G00000012404 Thrsp thyroid hormone responsive 2.43 2.4E-06

G00000019914 Tlcd3b TLC domain containing 3B 2.47 4.8E-10

G00000045636 Fasn fatty acid synthase 3.25 1.7E-11

G00000049911 LOC1025563 carbonyl reductase [NADPH] 1-like 4.33 6.0E-18 47

Lean Adipose Ensembl_ID Gene Symbol Gene Name L2FC P-value Downregulated (ENSRNO)

G00000016700 Tcf21 transcription factor 21 -2.82 3.2E-06

Lean Adipose Ensembl_ID Gene Symbol Gene Name L2FC P-value Upregulated (ENSRNO)

G00000013733 Ppp4r1 protein phosphatase 4, regulatory 2.68 2.4E-05 subunit 1

G00000009536 Pgp phosphoglycolate phosphatase 2.69 3.4E-05

G00000031789 Rangap1 RAN GTPase activating protein 1 2.97 3.9E-05

G00000005082 Irf6 interferon regulatory factor 6 3.25 1.3E-05

G00000031934 Enah ENAH, actin regulator 3.26 2.4E-05

G00000011296 Cenpn centromere protein N 3.34 2.8E-05

G00000056550 Epb41l4b erythrocyte membrane protein band 4.1 3.51 1.9E-05 like 4B

G00000021589 Nexmif neurite extension and migration factor 4.43 2.9E-05

G00000008713 Slc41a2 solute carrier family 41 member 2 4.51 4.8E-07

G00000033262 Reep6 receptor accessory protein 6 4.72 1.0E-05

G00000003098 Prom1 prominin 1 5.25 2.6E-06

G00000015403 Cd52 CD52 molecule 7.66 2.5E-07

123

Table 4-1 – continued

ZDF PFC Ensembl_ID Gene Symbol Gene Name L2FC P-value Downregulated (ENSRNO)

G00000033932 AY172581.22 AY172581.22-201 -5.21 1.1E-05 -201

G00000032112 AY172581.14 AY172581.14 -4.73 1.5E-05

ZDF PFC Ensembl_ID Gene Symbol Gene Name L2FC P-value Upregulated (ENSRNO)

None

Lean PFC Ensembl_ID Gene Symbol Gene Name L2FC P-value Downregulated (ENSRNO)

None

Lean PFC Ensembl_ID Gene Symbol Gene Name L2FC P-value Upregulated (ENSRNO)

None

ZDF Kidney Ensembl_ID Gene Symbol Gene Name L2FC P-value Downregulated (ENSRNO)

G00000020204 Srp19 signal recognition particle 19 -1.72 3.3E-05

G00000004794 Rtn1 reticulon 1 -2.01 5.1E-05

G00000055471 Ywhah tyrosine 3-monooxygenase/tryptophan -1.29 5.5E-05 5-monooxygenase activation protein, eta

G00000003357 Col3a1 collagen type III alpha 1 chain -1.46 6.4E-05

ZDF Kidney Ensembl_ID Gene Symbol Gene Name L2FC P-value Upregulated (ENSRNO)

G00000018237 Gstp1 glutathione S-transferase pi 1 2.04 5.2E-07

G00000018940 CNT1 solute carrier family 28 member 1 1.65 7.8E-05

Lean Kidney Ensembl_ID Gene Symbol Gene Name L2FC P-value Downregulated (ENSRNO)

G00000020151 Cdh1 cadherin 1 -1.08 1.4E-05

G00000013062 Cyp24a1 cytochrome P450, family 24, -1.30 4.3E-06 subfamily a, polypeptide 1

124

Table 4-1 – continued

G00000012956 Tgm2 transglutaminase 2 -1.34 3.5E-05

G00000004019 Phlda1 pleckstrin homology-like domain, -2.30 2.7E-08 family A, member 1

Lean Kidney Ensembl_ID Gene Symbol Gene Name L2FC P-value Upregulated (ENSRNO)

G00000029726 Gstm1 glutathione S-transferase mu 1 1.40 1.5E-05

G00000053811 Arg2 arginase 2 1.51 3.5E-05

G00000000576 Anapc16 anaphase promoting complex subunit 2.11 3.3E-05 16

ZDF Liver Ensembl_ID Gene Symbol Gene Name L2FC P-value Downregulated (ENSRNO)

G00000014320 Inhba inhibin subunit beta A -4.77 3.2E-12

G00000007923 Cgref1 cell growth regulator with EF hand -3.74 4.4E-09 domain 1

G00000004307 Tor3a torsin family 3 -3.55 6.1E-18

G00000034190 Ighm immunoglobulin heavy constant mu -3.38 1.4E-16

G00000003802 Pttg1 PTTG1 regulator of sister chromatid -3.35 1.2E-05 separation

G00000007060 Plin2 perilipin 2 -3.31 6.8E-28

G00000045636 Fasn fatty acid synthase -3.31 1.5E-10

G00000022256 Cxcl10 C-X-C motif chemokine ligand 10 -3.30 4.4E-04

G00000009019 Slc6a6 solute carrier family 6 member 6 -3.09 6.0E-11

G00000025691 Pla2g7 phospholipase A2 group VII -3.03 4.5E-06

G00000020035 Cyp17a1 cytochrome P450 -3.00 2.1E-09

G00000020480 Fads1 fatty acid desaturase 1 -2.93 1.5E-13

G00000000658 Acacb acetyl-CoA carboxylase beta -2.91 4.0E-18

G00000030154 Cyp4a2 cytochrome P450 -2.88 5.0E-04

125

Table 4-1 – continued

G00000021802 Isg15 ISG15 ubiquitin-like modifier -2.72 2.4E-03

G00000001052 Slc25a30 solute carrier family 25 -2.58 1.8E-09

G00000001963 Mx2 MX dynamin like GTPase 2 -2.56 2.1E-03

G00000040151 Sdr16c6 short chain dehydrogenase/reductase -2.55 6.3E-04 family 16C

G00000017914 Cavin3 caveolae associated protein 3 -2.51 1.4E-14

G00000006859 Insig1 insulin induced gene 1 -2.50 3.4E-13

G00000006204 Slc30a3 solute carrier family 30 member 3 -2.47 3.0E-07

G00000016353 Nim1k NIM1 serine/threonine protein kinase -2.40 3.6E-08

G00000016011 Plekhg1 pleckstrin homology and RhoGEF -2.40 7.8E-05 domain containing G1

G00000028137 Mki67 marker of proliferation Ki-67 -2.38 1.5E-04

G00000014476 Evl Enah/Vasp-like -2.37 3.4E-04

G00000008022 Apaf1 apoptotic peptidase activating factor 1 -2.36 7.1E-05

G00000053891 Phf11 PHD finger protein 11 -2.34 6.4E-08

G00000010819 Hspa4l heat shock protein family A (Hsp70) -2.32 6.9E-06 member 4 like

G00000021150 Plcb3 phospholipase C beta 3 -2.31 3.1E-05

G00000001414 Serpine1 serpin family E member 1 -2.27 1.2E-04

G00000016924 Acly ATP citrate lyase -2.25 5.5E-17

G00000045560 Gvin1 GTPase -2.25 2.1E-06

G00000020503 Cbln3 cerebellin 3 precursor -2.22 1.4E-06

G00000052444 Samd9 sterile alpha motif domain containing 9 -2.22 3.3E-04

G00000005209 Spred1 sprouty-related -2.21 1.7E-05

126

Table 4-1 – continued

G00000010888 Ankrd33b ankyrin repeat domain 33B -2.20 2.1E-06

G00000047218 Clic5 chloride intracellular channel 5 -2.20 1.7E-03

G00000009481 Ddhd1 DDHD domain containing 1 -2.19 2.0E-04

G00000022242 Cxcl9 C-X-C motif chemokine ligand 9 -2.16 1.3E-05

G00000008807 Rp1 RP1 -2.08 4.7E-05

G00000014426 Lox lysyl oxidase -2.07 1.9E-03

G00000015498 Il17rb interleukin 17 receptor B -2.07 2.3E-04

G00000051965 Smad4 SMAD family member 4 -2.07 8.7E-04

G00000017512 Aldh3b1 aldehyde dehydrogenase 3 family -2.05 1.0E-04

G00000057092 Slfn4 schlafen family member 4 -2.05 6.2E-06

G00000012685 Adck1 aarF domain containing kinase 1 -2.04 1.8E-03

G00000011268 Chd5 chromodomain helicase DNA binding -2.02 2.3E-03 protein 5

G00000032374 Paqr9 progestin and adipoQ receptor family -2.01 3.3E-14 member 9

G00000020272 Elapor1 endosome-lysosome associated -1.97 1.0E-04 apoptosis and autophagy regulator 1

G00000061118 LOC1025510 uncharacterized LOC102551095 -1.96 9.6E-05 95

G00000061527 Gck glucokinase -1.93 4.4E-07

G00000053460 Acot3 acyl-CoA thioesterase 3 -1.91 1.4E-04

G00000005043 Cpeb2 cytoplasmic polyadenylation element -1.91 2.0E-03 binding protein 2

G00000017332 Dapk2 death-associated protein kinase 2 -1.87 3.8E-04

G00000034013 Acaca acetyl-CoA carboxylase alpha -1.86 4.5E-05

127

Table 4-1 – continued

G00000017611 Tnp1 transition protein 1 -1.86 2.0E-03

G00000012603 Sestd1 SEC14 and spectrin domain containing -1.85 1.2E-03 1

G00000025558 Palm2 paralemmin 2 -1.84 5.7E-06

G00000018461 Pdgfrb platelet derived growth factor receptor -1.82 1.0E-03 beta

G00000016123 Rnf144b ring finger protein 144B -1.80 5.5E-17

G00000013111 Mettl3 methyltransferase-like 3 -1.78 6.7E-04

G00000045679 Apoa1 apolipoprotein A1 -1.78 1.1E-11

G00000001926 Cldn1 claudin 1 -1.78 1.8E-06

G00000005600 Nr4a2 nuclear receptor subfamily 4 -1.77 4.2E-04

G00000012148 Trio trio Rho guanine nucleotide exchange -1.76 7.0E-04 factor

G00000004626 Slc34a2 solute carrier family 34 member 2 -1.76 8.7E-05

G00000009360 Sh3bp1 SH3-domain binding protein 1 -1.74 2.2E-03

G00000010890 Bmp1 bone morphogenetic protein 1 -1.71 1.8E-07

G00000011820 Acp3 acid phosphatase 3 -1.69 1.1E-04

G00000007591 Slc45a3 solute carrier family 45 -1.68 8.9E-05

G00000006170 Bach2 BTB domain and CNC homolog 2 -1.68 1.2E-03

G00000028895 Rtp4 receptor (chemosensory) transporter -1.66 5.3E-05 protein 4

G00000002773 Rgs4 regulator of G-protein signaling 4 -1.65 5.0E-04

G00000007234 Cyp51 cytochrome P450 -1.64 1.2E-09

G00000020918 Ccnd1 cyclin D1 -1.64 7.0E-09

G00000028941 Zbed3 zinc finger -1.63 8.4E-06

128

Table 4-1 – continued

G00000012681 Lgals9 galectin 9 -1.63 2.8E-13

G00000001640 Tomm70 translocase of outer mitochondrial -1.63 2.6E-03 membrane 70

G00000009117 Otub2 OTU deubiquitinase -1.62 1.9E-04

G00000005726 Pclo piccolo (presynaptic cytomatrix -1.62 6.2E-04 protein)

G00000051171 G6pc glucose-6-phosphatase -1.61 1.6E-04

G00000016552 Hmgcs1 3-hydroxy-3-methylglutaryl-CoA -1.60 4.4E-15 synthase 1

G00000004577 Fez2 fasciculation and elongation protein -1.60 1.9E-04 zeta 2

G00000000547 Tspyl4 TSPY-like 4 -1.59 5.0E-04

G00000017120 Abhd2 abhydrolase domain containing 2 -1.59 1.9E-07

G00000015906 Tgif1 TGFB-induced factor homeobox 1 -1.56 1.1E-03

G00000008144 Irf1 interferon regulatory factor 1 -1.54 2.4E-06

G00000007319 Trib3 tribbles pseudokinase 3 -1.54 8.5E-06

G00000018467 Mitd1 microtubule interacting and trafficking -1.52 1.9E-03 domain containing 1

G00000023238 Dgkd diacylglycerol kinase -1.52 1.2E-05

G00000020776 Dhcr7 7-dehydrocholesterol reductase -1.51 1.1E-05

G00000033824 Gpd2 glycerol-3-phosphate dehydrogenase 2 -1.51 5.5E-04

G00000003442 Adora1 adenosine A1 receptor -1.50 2.3E-06

G00000025689 Abhd1 abhydrolase domain containing 1 -1.50 2.1E-07

G00000018198 Dapk1 death associated protein kinase 1 -1.48 5.3E-04

G00000029668 Wfdc21 WAP four-disulfide core domain 21 -1.45 1.6E-04

G00000006280 Pcsk9 proprotein convertase subtilisin/kexin -1.44 4.7E-07 type 9

129

Table 4-1 – continued

G00000008215 Trim47 tripartite motif-containing 47 -1.44 1.3E-04

G00000014013 Map4k4 mitogen-activated protein kinase -1.43 1.1E-05 kinase kinase kinase 4

G00000018627 Plekhb1 pleckstrin homology domain -1.43 8.9E-05 containing B1

G00000007713 Tmcc3 transmembrane and coiled-coil domain -1.43 2.6E-05 family 3

G00000000987 Ptcd1 pentatricopeptide repeat domain 1 -1.43 1.0E-03

G00000014702 Elovl2 ELOVL fatty acid elongase 2 -1.43 2.8E-11

G00000037595 Gpbp1l1 GC-rich promoter binding protein 1- -1.41 2.3E-04 like 1

G00000001205 Agpat3 1-acylglycerol-3-phosphate O- -1.41 3.8E-09 acyltransferase 3

G00000019776 Sh3gl3 SH3 domain containing GRB2 like 3 -1.40 4.1E-05

G00000012622 Mmp15 matrix metallopeptidase 15 -1.40 1.9E-03

G00000023856 Agxt alanine--glyoxylate and serine-- -1.39 7.5E-11 pyruvate aminotransferase

G00000042560 Bag4 BCL2-associated athanogene 4 -1.38 1.2E-03

G00000013841 Dcaf1 DDB1 and CUL4 associated factor 1 -1.38 2.1E-03

G00000019995 Dnajc18 DnaJ heat shock protein family -1.37 2.4E-04 (Hsp40) member C18

G00000019996 Slc16a1 solute carrier family 16 member 1 -1.36 1.6E-04

G00000023226 S100a10 S100 calcium binding protein A10 -1.35 2.2E-04

G00000006227 Ifih1 interferon induced with helicase C -1.35 7.5E-04 domain 1

G00000019005 Pde8a phosphodiesterase 8A -1.34 1.5E-03

G00000021259 Prnp prion protein -1.33 1.7E-03

G00000055909 Apoa4 apolipoprotein A4 -1.33 2.6E-10

130

Table 4-1 – continued

G00000008012 Abcb4 ATP binding cassette subfamily B -1.33 4.1E-10 member 4

G00000021405 Cyp2c7 cytochrome P450 -1.32 1.0E-08

G00000012181 Lpl lipoprotein lipase -1.31 5.3E-05

G00000056041 AABR07062570 -1.30 2.5E-03

G00000016815 Tmem135 transmembrane protein 135 -1.29 4.1E-05

G00000019422 Egr1 early growth response 1 -1.29 1.1E-05

G00000054077 Aabr07024870 -1.29 1.4E-03

G00000008194 Znfx1 zinc finger -1.28 5.5E-04

G00000009076 Ttpal alpha tocopherol transfer protein like -1.27 1.7E-03

G00000005825 Lyz2 lysozyme 2 -1.26 2.7E-05

G00000032293 Polg DNA polymerase gamma -1.26 2.9E-04

G00000008586 Aldh1l2 aldehyde dehydrogenase 1 family -1.26 3.5E-04

G00000010805 Fabp4 fatty acid binding protein 4 -1.25 2.5E-03

G00000016044 Mab21l3 mab-21 like 3 -1.25 1.0E-04

G00000000459 Psmb9 proteasome 20S subunit beta 9 -1.25 7.4E-04

G00000015124 Gpam glycerol-3-phosphate acyltransferase -1.25 1.4E-03

G00000020871 Ltbp4 latent transforming growth factor beta -1.22 2.1E-03 binding protein 4

G00000016516 Mbp myelin basic protein -1.22 1.2E-04

G00000007324 Plxna2 plexin A2 -1.22 2.4E-07

G00000001821 Adipoq adiponectin -1.21 2.7E-03

G00000020573 Efna1 ephrin A1 -1.19 1.7E-04

131

Table 4-1 – continued

G00000004606 Meis1 Meis homeobox 1 -1.19 8.3E-04

G00000001647 Ets2 ETS proto-oncogene 2 -1.17 9.1E-09

G00000059043 Itch itchy E3 ubiquitin protein ligase -1.17 1.4E-03

G00000006787 Dhcr24 24-dehydrocholesterol reductase -1.16 1.9E-12

G00000015121 N4bp1 Nedd4 binding protein 1 -1.15 3.0E-04

G00000042771 Apol3 apolipoprotein L -1.14 7.6E-08

G00000023664 Lepr leptin receptor -1.12 2.5E-03

G00000000451 RT1-Ba RT1 class II -1.12 1.1E-03

G00000012782 Cemip2 cell migration inducing hyaluronidase -1.11 1.0E-05 2

G00000014766 Galt galactose-1-phosphate -1.11 1.7E-05 uridylyltransferase

G00000014718 Acsl3 acyl-CoA synthetase long-chain family -1.11 1.4E-03 member 3

G00000017428 Map1b microtubule-associated protein 1B -1.10 1.7E-03

G00000018517 Trim21 tripartite motif-containing 21 -1.09 2.0E-03

G00000001426 Prkrip1 PRKR interacting protein 1 -1.09 2.2E-03

G00000028448 Elovl1 ELOVL fatty acid elongase 1 -1.08 2.2E-03

G00000005695 Mgp matrix Gla protein -1.07 1.2E-03

G00000017558 Tubb2a tubulin -1.07 7.2E-04

G00000012876 Slc6a13 solute carrier family 6 member 13 -1.07 7.1E-05

G00000018960 Syne1 spectrin repeat containing nuclear -1.07 7.3E-05 envelope protein 1

G00000017993 Abcb10 ATP binding cassette subfamily B -1.05 8.6E-05 member 10

G00000007545 Angptl4 angiopoietin-like 4 -1.04 4.1E-07

132

Table 4-1 – continued

G00000007990 Adipor2 adiponectin receptor 2 -1.04 1.1E-04

G00000020134 Upf1 UPF1 -1.03 5.1E-04

G00000027434 Fitm2 fat storage-inducing transmembrane -1.02 1.7E-05 protein 2

G00000048315 Eif2ak2 eukaryotic translation initiation factor -1.02 6.7E-05 2-alpha kinase 2

G00000005642 Frs2 fibroblast growth factor receptor -1.02 7.7E-04 substrate 2

G00000014604 Sigmar1 sigma non-opioid intracellular receptor -1.01 4.9E-09 1

G00000002175 Clock clock circadian regulator -1.01 2.4E-04

G00000042785 Sesn2 sestrin 2 -1.00 1.7E-05

G00000023463 Parp9 poly (ADP-ribose) polymerase family -0.99 3.5E-04

G00000043377 Fdps farnesyl diphosphate synthase -0.99 1.6E-03

G00000000593 Rev3l REV3 like -0.98 2.0E-03

G00000019283 P2ry2 purinergic receptor P2Y2 -0.98 1.0E-03

G00000024061 Rarb retinoic acid receptor -0.98 2.7E-03

G00000017220 Tcirg1 T-cell immune regulator 1 -0.98 3.3E-04

G00000021032 Sphk2 sphingosine kinase 2 -0.97 2.3E-05

G00000001585 Nrip1 nuclear receptor interacting protein 1 -0.97 1.1E-03

G00000003882 Cep350 centrosomal protein 350 -0.96 5.2E-04

G00000005292 Trip11 thyroid hormone receptor interactor 11 -0.96 2.4E-06

G00000046889 Dbi diazepam binding inhibitor -0.96 2.3E-05

G00000000664 Tpst2 tyrosylprotein sulfotransferase 2 -0.95 2.1E-03

G00000014900 Crem cAMP responsive element modulator -0.95 2.0E-03

133

Table 4-1 – continued

G00000024115 C6 complement C6 -0.93 2.9E-04

G00000030225 Clpx caseinolytic mitochondrial matrix -0.92 3.8E-05 peptidase chaperone subunit X

G00000038012 Commd6 COMM domain containing 6 -0.91 1.8E-04

G00000007302 Fbn1 fibrillin 1 -0.91 1.4E-03

G00000018420 Slc22a7 solute carrier family 22 member 7 -0.90 4.1E-04

G00000002635 Dexi Dexi homolog -0.89 1.6E-06

G00000007728 Gsdmd gasdermin D -0.89 2.3E-03

G00000026942 RGD1311595 similar to KIAA2026 protein -0.88 2.9E-05

G00000034066 Hspa8 heat shock protein family A (Hsp70) -0.87 3.0E-04 member 8

G00000019372 Pc pyruvate carboxylase -0.87 8.5E-06

G00000000177 Plpp2 phospholipid phosphatase 2 -0.87 9.6E-04

G00000056703 Atrx ATRX -0.86 2.0E-04

G00000016219 Vnn1 vanin 1 -0.86 1.5E-04

G00000014338 Slc25a25 solute carrier family 25 member 25 -0.86 6.6E-04

G00000013391 Sorbs2 sorbin and SH3 domain containing 2 -0.84 9.9E-06

G00000016692 Hsdl2 hydroxysteroid dehydrogenase like 2 -0.83 1.3E-04

G00000024145 Trim65 tripartite motif-containing 65 -0.83 5.9E-05

G00000010947 Mmp14 matrix metallopeptidase 14 -0.83 1.7E-03

G00000018584 Ptma prothymosin alpha -0.83 1.5E-05

G00000008274 Xpc XPC complex subunit -0.83 3.6E-04

G00000011261 Ttc14 tetratricopeptide repeat domain 14 -0.83 2.7E-03

134

Table 4-1 – continued

G00000047386 Smg1 SMG1 -0.82 2.7E-03

G00000007400 Srebf2 sterol regulatory element binding -0.82 4.0E-04 transcription factor 2

G00000028801 Gsap gamma-secretase activating protein -0.82 2.1E-03

G00000007700 Inhbc inhibin subunit beta C -0.81 5.3E-04

G00000013178 Cmip c-Maf-inducing protein -0.81 6.6E-04

G00000032394 Tymp thymidine phosphorylase -0.80 5.0E-04

G00000031709 Ppfibp1 PPFIA binding protein 1 -0.79 6.6E-04

G00000003020 Slc25a47 solute carrier family 25 -0.79 1.5E-03

G00000019450 Etf1 eukaryotic translation termination -0.78 1.6E-03 factor 1

G00000010497 RGD1305807 hypothetical LOC298077 -0.77 1.7E-05

G00000000184 Tmprss6 transmembrane serine protease 6 -0.75 3.7E-04

G00000004709 Foxn3 forkhead box N3 -0.73 2.2E-04

G00000007681 Brd3 bromodomain containing 3 -0.72 2.1E-03

G00000033593 Osbpl9 oxysterol binding protein-like 9 -0.72 7.9E-04

G00000002212 Hsd17b13 hydroxysteroid (17-beta) -0.70 5.2E-06 dehydrogenase 13

G00000053550 Itga1 integrin subunit alpha 1 -0.68 1.6E-03

G00000030700 COX3 cytochrome c oxidase subunit 3 -0.67 2.8E-04

G00000020425 Stim1 stromal interaction molecule 1 -0.66 1.0E-03

G00000057814 Nsdhl NAD(P) dependent steroid -0.66 2.0E-05 dehydrogenase-like

G00000056371 Pik3ca phosphatidylinositol-4 -0.66 1.5E-03

G00000016266 Mphosph10 M-phase phosphoprotein 10 -0.65 2.2E-03

135

Table 4-1 – continued

G00000015441 Il4r interleukin 4 receptor -0.65 1.8E-03

G00000009102 Fermt2 fermitin family member 2 -0.62 2.2E-03

G00000005015 Rabep1 rabaptin -0.62 1.8E-03

G00000020151 Cdh1 cadherin 1 -0.60 2.4E-03

G00000013135 Ptpn12 protein tyrosine phosphatase -0.58 2.7E-04

G00000057623 Copb1 COPI coat complex subunit beta 1 -0.53 4.7E-04

G00000011140 Prxl2a peroxiredoxin like 2A -0.51 2.0E-03

G00000018849 Tcerg1 transcription elongation regulator 1 -0.51 2.0E-03

G00000008305 Sc5d sterol-C5-desaturase -0.47 2.0E-03

G00000009263 Ifi27 interferon -0.47 2.2E-03

G00000004345 Daam1 dishevelled associated activator of -0.46 2.7E-03 morphogenesis 1

G00000016963 Trip12 thyroid hormone receptor interactor 12 -0.43 1.7E-03

ZDF Liver Ensembl_ID Gene Symbol Gene Name L2FC P-value Upregulated (ENSRNO)

G00000003119 Gc GC 0.45 1.1E-03

G00000000610 Cisd1 CDGSH iron sulfur domain 1 0.46 7.7E-04

G00000019629 Lamp1 lysosomal-associated membrane 0.50 8.8E-05 protein 1

G00000000701 Iscu iron-sulfur cluster assembly enzyme 0.51 4.0E-05

G00000037850 Mtarc2 mitochondrial amidoxime reducing 0.51 6.0E-04 component 2

G00000019048 Sod2 superoxide dismutase 2 0.54 2.8E-04

G00000007967 Sdhb succinate dehydrogenase complex iron 0.54 1.8E-03 sulfur subunit B

G00000013928 Dsp desmoplakin 0.54 1.5E-03

136

Table 4-1 – continued

G00000016794 Phyhd1 phytanoyl-CoA dioxygenase domain 0.55 2.0E-03 containing 1

G00000019626 Slc27a5 solute carrier family 27 member 5 0.55 8.8E-05

G00000028368 Etnk2 ethanolamine kinase 2 0.55 1.4E-03

G00000011535 Gcsh glycine cleavage system protein H 0.56 9.9E-04

G00000008921 Dynll2 dynein light chain LC8-type 2 0.56 1.5E-03

G00000030449 Gsta4 glutathione S-transferase alpha 4 0.56 1.1E-03

G00000018604 Tufm Tu translation elongation factor 0.59 2.2E-03

G00000017672 Akr1c14 aldo-keto reductase family 1 0.59 3.1E-04

G00000020994 Slc25a39 solute carrier family 25 0.59 7.3E-04

G00000047708 Gstz1 glutathione S-transferase zeta 1 0.59 1.1E-04

G00000013704 Cps1 carbamoyl-phosphate synthase 1 0.60 5.4E-04

G00000043404 Uroc1 urocanate hydratase 1 0.60 1.6E-05

G00000007395 Baat bile acid CoA:amino acid N- 0.60 5.3E-04 acyltransferase

G00000017577 Bphl biphenyl hydrolase like 0.60 6.8E-04

G00000007069 Adhfe1 alcohol dehydrogenase 0.62 4.9E-04

G00000023538 Aldh5a1 aldehyde dehydrogenase 5 family 0.62 4.9E-04

G00000006653 Slc38a4 solute carrier family 38 0.62 1.2E-04

G00000001333 Azgp1 alpha-2-glycoprotein 1 0.62 8.6E-06

G00000016339 Uox urate oxidase 0.63 2.8E-05

G00000061876 Tas1r2 taste 1 receptor member 2 0.63 2.4E-04

G00000006916 Sardh sarcosine dehydrogenase 0.63 8.6E-05

137

Table 4-1 – continued

G00000029549 Eci3 enoyl-Coenzyme A delta isomerase 3 0.63 8.9E-04

G00000048723 Pros1 protein S 0.64 4.9E-04

G00000009005 Slco2a1 solute carrier organic anion transporter 0.64 2.7E-05 family

G00000007839 Slc16a7 solute carrier family 16 member 7 0.64 8.2E-04

G00000010389 Ndrg2 NDRG family member 2 0.65 5.7E-04

G00000014165 Ssr1 signal sequence receptor subunit 1 0.65 1.0E-04

G00000029735 Pid1 phosphotyrosine interaction domain 0.65 1.9E-03 containing 1

G00000033466 Apon apolipoprotein N 0.65 1.1E-03

G00000000158 Cdo1 cysteine dioxygenase type 1 0.65 6.6E-06

G00000008364 Cat catalase 0.67 1.1E-03

G00000061883 Aqp9 aquaporin 9 0.68 1.3E-03

G00000021916 Slc16a12 solute carrier family 16 0.68 2.5E-03

G00000007743 Mgst1 microsomal glutathione S-transferase 1 0.68 1.8E-05

G00000003653 Fh fumarate hydratase 0.68 1.6E-03

G00000013223 Fah fumarylacetoacetate hydrolase 0.69 2.4E-04

G00000014700 Ttc36 tetratricopeptide repeat domain 36 0.69 8.4E-05

G00000030862 Atp6v1h ATPase H+ transporting V1 subunit H 0.69 4.9E-04

G00000030667 Ppm1b protein phosphatase 0.71 4.7E-06

G00000004139 Ndel1 nudE neurodevelopment protein 1-like 0.72 3.8E-05 1

G00000007927 Mettl7b methyltransferase like 7B 0.72 5.0E-05

G00000004147 Abca8a ATP-binding cassette 0.73 1.4E-03

138

Table 4-1 – continued

G00000029726 Gstm1 glutathione S-transferase mu 1 0.74 1.3E-03

G00000003370 Otc ornithine carbamoyltransferase 0.74 6.8E-06

G00000013039 Add1 adducin 1 0.74 4.2E-04

G00000014727 Fahd1 fumarylacetoacetate hydrolase domain 0.75 4.2E-04 containing 1

G00000059463 Slc39a1 solute carrier family 39 member 1 0.76 1.6E-03

G00000004302 Pah phenylalanine hydroxylase 0.76 3.4E-07

G00000029651 Rdh16 retinol dehydrogenase 16 0.76 8.2E-04

G00000028746 Gsto1 glutathione S-transferase omega 1 0.77 3.2E-04

G00000018426 NEWGENE_ apolipoprotein C1 0.77 1.3E-06 2134

G00000001053 Tmed2 transmembrane p24 trafficking protein 0.77 6.7E-04 2

G00000016173 Cyp1a2 cytochrome P450 0.77 6.7E-04

G00000004089 Enpp2 ectonucleotide 0.78 3.5E-04 pyrophosphatase/phosphodiesterase 2

G00000042274 Fbxo31 F-box protein 31 0.78 2.3E-03

G00000000186 Tst thiosulfate sulfurtransferase 0.78 8.6E-05

G00000048812 Gpx1 glutathione peroxidase 1 0.79 5.0E-04

G00000047986 Sult2a1 sulfotransferase family 2A member 1 0.79 2.5E-03

G00000006345 Sec61b SEC61 translocon subunit beta 0.79 6.2E-04

G00000009779 Krt8 keratin 8 0.79 2.2E-03

G00000006623 Cd302 CD302 molecule 0.80 1.5E-04

G00000005987 Suox sulfite oxidase 0.81 1.1E-03

139

Table 4-1 – continued

G00000061890 Ust5r integral membrane transport protein 0.81 2.3E-04 UST5r

G00000020879 Nags N-acetylglutamate synthase 0.81 3.3E-04

G00000008902 Pon1 paraoxonase 1 0.82 9.7E-07

G00000018904 Dtymk deoxythymidylate kinase 0.82 2.1E-03

G00000023116 Agmo alkylglycerol monooxygenase 0.82 4.0E-05

G00000047816 Ccs copper chaperone for superoxide 0.84 1.3E-04 dismutase

G00000012142 Glyat glycine-N-acyltransferase 0.84 5.6E-07

G00000021206 Plaat3 phospholipase A and acyltransferase 3 0.84 7.5E-04

G00000012962 Nudt16 nudix hydrolase 16 0.85 1.9E-04

G00000050315 Dcxr dicarbonyl and L-xylulose reductase 0.86 2.9E-06

G00000000024 Hebp1 heme binding protein 1 0.86 2.7E-04

G00000000386 Pbld1 phenazine biosynthesis-like protein 0.87 1.3E-05 domain containing 1

G00000007378 Acox2 acyl-CoA oxidase 2 0.87 7.0E-05

G00000003307 Gcdh glutaryl-CoA dehydrogenase 0.87 2.2E-08

G00000002205 Ociad1 OCIA domain containing 1 0.87 1.4E-03

G00000014645 Aldh7a1 aldehyde dehydrogenase 7 family 0.88 8.2E-08

G00000008638 Angptl3 angiopoietin-like 3 0.88 2.9E-09

G00000011351 Mat1a methionine adenosyltransferase 1A 0.89 3.6E-05

G00000009421 Ivd isovaleryl-CoA dehydrogenase 0.89 1.9E-09

G00000036894 Cisd3 CDGSH iron sulfur domain 3 0.89 4.0E-04

G00000014128 Ecsit ECSIT signaling integrator 0.90 1.6E-03

140

Table 4-1 – continued

G00000017619 Aldh1a1 aldehyde dehydrogenase 1 family 0.90 3.1E-05

G00000018662 Amacr alpha-methylacyl-CoA racemase 0.90 3.9E-07

G00000020000 Tmem219 transmembrane protein 219 0.90 5.2E-04

G00000001957 Sult1e1 sulfotransferase family 1E member 1 0.90 2.8E-06

G00000051860 Rnase4 ribonuclease A family member 4 0.91 1.3E-09

G00000014160 Tcp1 t-complex 1 0.91 2.2E-04

G00000048114 Echdc3 enoyl CoA hydratase domain 0.91 2.7E-07 containing 3

G00000003291 Creg1 cellular repressor of E1A-stimulated 0.92 1.3E-07 genes 1

G00000008837 Ass1 argininosuccinate synthase 1 0.92 7.7E-04

G00000018159 Anxa4 annexin A4 0.92 2.3E-04

G00000010993 Dpm1 dolichyl-phosphate 0.92 9.1E-04 mannosyltransferase subunit 1

G00000019982 Ethe1 ETHE1 0.92 2.4E-05

G00000023177 Esrp2 epithelial splicing regulatory protein 2 0.93 9.8E-07

G00000013409 Gclm glutamate cysteine ligase 0.93 3.0E-04

G00000001806 Fetub fetuin B 0.93 2.9E-04

G00000017291 Sord sorbitol dehydrogenase 0.94 7.2E-09

G00000053362 Gabarapl1 GABA type A receptor associated 0.94 1.4E-07 protein like 1

G00000021174 Macrod1 mono-ADP ribosylhydrolase 1 0.95 7.1E-05

G00000014268 Abca2 ATP binding cassette subfamily A 0.95 9.8E-04 member 2

G00000049771 Gstt1 glutathione S-transferase theta 1 0.96 8.4E-05

G00000011226 Timm8a1 translocase of inner mitochondrial 0.96 4.5E-06 membrane 8A1

141

Table 4-1 – continued

G00000005175 Sgpp1 sphingosine-1-phosphate phosphatase 0.97 2.0E-03 1

G00000049464 Cyp2c13 cytochrome P450 0.97 6.0E-10

G00000002210 Hsd17b11 hydroxysteroid (17-beta) 0.97 4.4E-10 dehydrogenase 11

G00000012786 Pgrmc1 progesterone receptor membrane 0.99 1.2E-07 component 1

G00000004327 Ddc dopa decarboxylase 0.99 4.8E-05

G00000046357 Adh5 alcohol dehydrogenase 5 (class III) 0.99 1.2E-11

G00000054049 Prelid2 PRELI domain containing 2 0.99 7.6E-04

G00000004442 Dglucy D-glutamate cyclase 0.99 1.6E-03

G00000014876 Lpin2 lipin 2 1.00 3.9E-04

G00000012911 Erlin1 ER lipid raft associated 1 1.00 6.8E-04

G00000055314 Msrb1 methionine sulfoxide reductase B1 1.00 1.1E-07

G00000006619 Dnajc9 DnaJ heat shock protein family 1.01 6.5E-04 (Hsp40) member C9

G00000018937 Gstm7 glutathione S-transferase 1.01 1.6E-04

G00000027016 Cobll1 cordon-bleu WH2 repeat protein-like 1 1.01 1.4E-04

G00000046007 Cldn3 claudin 3 1.02 2.8E-04

G00000033609 Irx1 iroquois homeobox 1 1.02 2.0E-03

G00000017777 Ahcy adenosylhomocysteinase 1.02 1.5E-05

G00000019180 Acsl4 acyl-CoA synthetase long-chain family 1.02 1.0E-08 member 4

G00000022932 Serhl2 serine hydrolase-like 2 1.03 1.5E-04

G00000016484 Gstk1 glutathione S-transferase kappa 1 1.03 1.5E-07

G00000003620 Fmo3 flavin containing dimethylaniline 1.04 1.7E-05 monoxygenase 3

142

Table 4-1 – continued

G00000032895 Cyp4f4 cytochrome P450 1.04 5.0E-08

G00000032737 F7 coagulation factor VII 1.05 2.1E-04

G00000023816 Aph1a aph-1 homolog A 1.05 1.6E-03

G00000015205 Cyb5a cytochrome b5 type A 1.06 9.6E-07

G00000008079 Ugp2 UDP-glucose pyrophosphorylase 2 1.06 4.1E-08

G00000011559 Cnn3 calponin 3 1.07 5.6E-05

G00000013484 Gsta1 glutathione S-transferase alpha-1 1.07 4.2E-10

G00000050595 Mup5 major urinary protein 5 1.07 1.5E-04

G00000026775 Pmpca peptidase 1.08 3.9E-04

G00000001338 Hpd 4-hydroxyphenylpyruvate dioxygenase 1.08 7.7E-06

G00000001618 Ripk4 receptor-interacting serine-threonine 1.09 2.3E-03 kinase 4

G00000000768 Ubd ubiquitin D 1.10 2.4E-05

G00000007508 Lrtm2 leucine-rich repeats and 1.10 1.8E-08 transmembrane domains 2

G00000031769 Chchd7 coiled-coil-helix-coiled-coil-helix 1.10 4.4E-04 domain containing 7

G00000013291 Cyp2c23 cytochrome P450 1.10 4.4E-07

G00000017188 Cyp27a1 cytochrome P450 1.11 1.7E-08

G00000025079 Fam126b family with sequence similarity 126 1.13 4.7E-04

G00000061215 Crym crystallin 1.14 2.1E-04

G00000017752 Mccc2 methylcrotonoyl-CoA carboxylase 2 1.16 1.8E-05

G00000016166 Pdlim1 PDZ and LIM domain 1 1.16 7.7E-07

G00000010079 Ca3 carbonic anhydrase 3 1.17 4.7E-10

143

Table 4-1 – continued

G00000013728 Polg2 DNA polymerase gamma 2 1.17 6.2E-04

G00000062298 Rpl13a ribosomal protein L13A 1.19 2.6E-03

G00000013751 Plpbp pyridoxal phosphate binding protein 1.19 2.3E-06

G00000001442 Por cytochrome p450 oxidoreductase 1.19 1.2E-09

G00000042253 Ecd ecdysoneless cell cycle regulator 1.20 6.0E-04

G00000020254 Per2 period circadian regulator 2 1.20 3.1E-04

G00000007949 Rgn regucalcin 1.21 8.8E-08

G00000003253 Qdpr quinoid dihydropteridine reductase 1.21 3.4E-09

G00000003515 Ephx1 epoxide hydrolase 1 1.22 7.9E-07

G00000011039 Gch1 GTP cyclohydrolase 1 1.23 2.8E-07

G00000038746 Bco2 beta-carotene oxygenase 2 1.24 6.3E-07

G00000005861 Hsd11b1 hydroxysteroid 11-beta dehydrogenase 1.24 4.3E-09 1

G00000000588 Slc16a10 solute carrier family 16 member 10 1.24 1.6E-05

G00000020202 Asrgl1 asparaginase and isoaspartyl peptidase 1.25 2.7E-03 1

G00000017826 Mtrr 5-methyltetrahydrofolate- 1.26 1.6E-03 homocysteine methyltransferase reductase

G00000033700 Bud23 BUD23 1.27 1.7E-03

G00000000281 Prodh1 proline dehydrogenase 1 1.28 2.1E-11

G00000042084 Acsm2 acyl-CoA synthetase medium-chain 1.30 3.4E-09 family member 2

G00000006972 Zfp189 zinc finger protein 189 1.30 1.5E-03

G00000027784 Tsku tsukushi 1.32 7.2E-04

144

Table 4-1 – continued

G00000012387 Glyatl2 glycine-N-acyltransferase-like 2 1.33 7.3E-07

G00000011714 Sat2 spermidine/spermine N1- 1.33 6.9E-05 acetyltransferase family member 2

G00000045799 Rup2 urinary protein 2 1.34 7.8E-04

G00000021924 Cyp2c22 cytochrome P450 1.35 9.3E-08

G00000018494 Ppp1r3c protein phosphatase 1 1.36 4.1E-11

G00000004693 Pbx1 PBX homeobox 1 1.36 1.4E-03

G00000001258 Snx8 sorting nexin 8 1.37 2.0E-04

G00000020698 Rnd2 Rho family GTPase 2 1.37 6.0E-05

G00000051227 AABR070484 AABR07048487.2 1.38 6.2E-04 87.2

G00000052810 Cyp2c11 cytochrome P450 1.39 2.0E-11

G00000012436 Adh6 alcohol dehydrogenase 6 (class V) 1.41 4.4E-15

G00000015936 Gng5 G protein subunit gamma 5 1.41 2.6E-03

G00000018413 Per3 period circadian regulator 3 1.42 1.6E-04

G00000016967 Hfe homeostatic iron regulator 1.42 2.9E-07

G00000001376 Mettl7a methyltransferase like 7A 1.43 3.1E-04

G00000056940 Cited2 Cbp/p300-interacting transactivator 1.44 1.5E-11

G00000015002 Abhd15 abhydrolase domain containing 15 1.44 1.5E-04

G00000032959 Adh7 alcohol dehydrogenase 7 (class IV) 1.45 7.6E-09

G00000050232 LOC680406 similar to Urinary protein 2 precursor 1.46 1.5E-09 (RUP-2)

G00000020700 Rnaseh2c ribonuclease H2 1.48 7.7E-04

G00000011635 Ces2e carboxylesterase 2E 1.49 2.9E-08

145

Table 4-1 – continued

G00000015354 Aox1 aldehyde oxidase 1 1.54 2.6E-12

G00000061450 Homer2 homer scaffold protein 2 1.54 2.5E-05

G00000009629 Car2 carbonic anhydrase 2 1.55 2.9E-05

G00000042111 Sult1c2a sulfotransferase family 1.55 2.3E-03

G00000057072 Slc12a3 solute carrier family 12 member 3 1.55 3.1E-04

G00000004009 Xpnpep2 X-prolyl aminopeptidase 2 1.57 1.1E-08

G00000013313 Nceh1 neutral cholesterol ester hydrolase 1 1.57 8.8E-07

G00000015438 LOC501233 LRRGT00080 1.58 1.2E-13

G00000015076 Cyp26b1 cytochrome P450 1.62 6.3E-04

G00000016456 Il33 interleukin 33 1.65 4.2E-18

G00000001766 Tfrc transferrin receptor 1.67 1.1E-04

G00000011718 C1rl complement C1r subcomponent like 1.68 1.2E-06

G00000013949 Idh2 isocitrate dehydrogenase (NADP(+)) 2 1.68 2.3E-16

G00000018740 Ugt1a6 UDP glucuronosyltransferase family 1 1.69 6.1E-14 member A6

G00000016807 Oat ornithine aminotransferase 1.72 1.2E-05

G00000025418 Armc9 armadillo repeat containing 9 1.74 4.5E-04

G00000023778 Gcnt2 glucosaminyl (N-acetyl) transferase 2 1.77 6.8E-05 (I blood group)

G00000056596 Alas1 5'-aminolevulinate synthase 1 1.80 2.3E-15

G00000046643 Cyp3a9 cytochrome P450 1.82 5.3E-04

G00000003260 Nr1i3 nuclear receptor subfamily 1 1.84 6.4E-05

G00000001158 Abcg1 ATP binding cassette subfamily G 1.86 3.0E-05 member 1

146

Table 4-1 – continued

G00000020250 Pcgf6 polycomb group ring finger 6 1.88 9.2E-04

G00000006420 Rbm38 RNA binding motif protein 38 1.89 2.0E-04

G00000012458 Cyp2e1 cytochrome P450 1.91 1.3E-19

G00000002258 Tmem150c transmembrane protein 150C 1.94 9.3E-05

G00000013982 Hsd17b2 hydroxysteroid (17-beta) 1.94 5.1E-04 dehydrogenase 2

G00000021027 Dbp D-box binding PAR bZIP transcription 1.94 3.2E-05 factor

G00000004437 Map2k6 mitogen-activated protein kinase 2.07 1.1E-08 kinase 6

G00000032246 Acsm3 acyl-CoA synthetase medium-chain 2.19 2.3E-16 family member 3

G00000014490 Bdh2 3-hydroxybutyrate dehydrogenase 2 2.21 2.5E-14

G00000036687 Alyref Aly/REF export factor 2.23 8.9E-04

G00000015519 Ces1d carboxylesterase 1D 2.24 6.9E-32

G00000009598 Ncaph2 non-SMC condensin II complex 2.41 3.8E-05

G00000043131 LOC1003600 urinary protein 1-like 2.43 4.9E-19 95

G00000034191 Fmo1 flavin containing dimethylaniline 2.46 2.8E-25 monoxygenase 1

G00000005985 Kcnma1 potassium calcium-activated channel 2.82 1.6E-04 subfamily M alpha 1

G00000011250 Inmt indolethylamine N-methyltransferase 2.90 3.9E-21

G00000058904 Tex13b testis expressed 13B 3.10 5.9E-15

G00000012772 Nqo1 NAD(P)H quinone dehydrogenase 1 3.11 1.5E-12

G00000001388 Sds serine dehydratase 3.29 1.7E-09

G00000056847 Gsta3 glutathione S-transferase alpha 3 3.40 3.1E-21

147

Table 4-1 – continued

G00000001242 Gstt3 glutathione S-transferase 3.66 1.2E-85

G00000051912 Acnat2 acyl-coenzyme A amino acid N- 3.74 8.9E-05 acyltransferase 2

G00000009488 Cyp7a1 cytochrome P450 family 7 subfamily 4.19 1.4E-18 A member 1

Lean Liver Ensembl_ID Gene Symbol Gene Name L2FC P-value Downregulated (ENSRNO)

G00000029668 Wfdc21 WAP four-disulfide core domain 21 -2.72 1.3E-04

G00000020480 Fads1 fatty acid desaturase 1 -2.53 4.0E-08

G00000006859 Insig1 insulin induced gene 1 -2.32 3.2E-05

G00000057557 Prlr prolactin receptor -2.27 2.3E-04

G00000055909 Apoa4 apolipoprotein A4 -1.93 2.3E-04

G00000030154 Cyp4a2 cytochrome P450 -1.75 2.2E-08

G00000019776 Sh3gl3 SH3 domain containing GRB2 like 3 -1.62 8.9E-05

G00000046889 Dbi diazepam binding inhibitor -1.61 1.1E-05

G00000014702 Elovl2 ELOVL fatty acid elongase 2 -1.60 1.8E-05

G00000032297 Msmo1 methylsterol monooxygenase 1 -1.56 3.6E-05

G00000007234 Cyp51 cytochrome P450 -1.56 4.9E-05

G00000020989 Tm7sf2 transmembrane 7 superfamily member -1.45 6.6E-05 2

Lean Liver Ensembl_ID Gene Symbol Gene Name L2FC P-value Upregulated (ENSRNO)

G00000001376 Mettl7a methyltransferase like 7A 1.16 1.7E-04

G00000048114 Echdc3 enoyl CoA hydratase domain 1.17 1.2E-04 containing 3

G00000023116 Agmo alkylglycerol monooxygenase 1.21 1.7E-04

148

Table 4-1 – continued

G00000002643 Ugdh UDP-glucose 6-dehydrogenase 1.29 7.6E-05

G00000015354 Aox1 aldehyde oxidase 1 1.33 1.2E-04

G00000004089 Enpp2 ectonucleotide 1.34 9.6E-05 pyrophosphatase/phosphodiesterase 2

G00000034191 Fmo1 flavin containing dimethylaniline 1.37 1.7E-04 monoxygenase 1

G00000013291 Cyp2c23 cytochrome P450 1.47 1.1E-04

G00000003809 Sat1 spermidine/spermine N1-acetyl 1.58 3.9E-05 transferase 1

G00000018740 Ugt1a6 UDP glucuronosyltransferase family 1 1.80 2.6E-05 member A6

G00000015519 Ces1d carboxylesterase 1D 1.94 5.6E-10

G00000033570 Arhgap8 Rho GTPase activating protein 8 2.03 1.5E-04

G00000051912 Acnat2 acyl-coenzyme A amino acid N- 2.05 1.1E-04 acyltransferase 2

G00000001158 Abcg1 ATP binding cassette subfamily G 2.21 1.1E-04 member 1

G00000001388 Sds serine dehydratase 2.29 8.0E-06

G00000047613 AABR070484 AABR07048463.1 2.34 1.4E-06 63.1

G00000013552 Scd stearoyl-CoA desaturase 2.36 4.8E-06

G00000001242 Gstt3 glutathione S-transferase 2.93 1.4E-09

G00000021924 Cyp2c22 cytochrome P450 3.22 5.2E-15

G00000009488 Cyp7a1 cytochrome P450 family 7 subfamily 3.36 1.5E-09 A member 1

149

Table 4-1 Differentially Expressed microRNAs

Genotype Tissue MicroRNA L2FC Non adjusted P-

p-value value

ZDF Adipose Downregulated rno-miR-221-3p -1.60 9.53E-05 0.007

ZDF PFC Upregulated rno-miR-29a-3p 0.59 0.0001 0.022

ZDF PFC Upregulated rno-miR-151-5p 0.89 0.0005 0.036

Lean Adipose Downregulated rno-miR-125a-5p -1.48 0.0022 0.069

Lean Adipose Downregulated rno-miR-125b-5p -1.78 0.0029 0.069

Lean Liver Upregulated rno-miR-9a-5p 1.89 9.08E-05 0.006

Lean Liver Upregulated rno-miR-181a-5p 1.10 0.0007 0.024

Lean Liver Upregulated rno-miR-10b-5p 1.37 0.0011 0.024

Lean Liver Downregulated rno-miR-192-5p -0.57 0.0013 0.024

150

Table 4-3 Correlational Mapping

Ensembl Rat Gene Non-adj Human Rat MicroRNA Tissue L2FC ID Gene Name Name p-value Symbol Symbol (RNOG)

miR- lean phosphoglycolate 125b-5p Pgp 2.69 0.00003 PGP 00000009536 Pgp adipose phosphatase (down regulated)

Cytochrome rno-miR- lean P450 Family 7 181a-5p CYP7A1 3.36 0.000000001 Cyp7a1 00000009488 Cyp7a1 liver Subfamily A (up regulated) Member 1

rno-miR- lean stearoyl-CoA 181a-5p Scd 2.35 0.0000048 Scd 00000013552 SCD liver desaturase (up regulated)

151

Table 4-4 Diet Formulations.

Ingredient (g/kg) CAS WE

Casein 200 -

Whole Egg2 - 435

Cornstarch 417 365

Corn Oil 183 -

Glucose Monohydrate 150 150

Mineral Mix 35 35

Vitamin Mix 10 10

Choline Bitartrate 2 2

L-methionine 3 3

Biotin (1%) - 0.4

Macronutrients (% total kcal)3

Protein 17 20

Carbohydrate 48 50 Fat 35 30

152

Table 4-5 mRNA Deseq2 Statistical Summary

153

Table 4-6 mRNA Raw Read Counts.

154

Table 4-7 MicroRNA Raw Read Counts.

155

Table 4-8 MicroRNA Raw Read Counts.

156

Table 4-9 KEGG/GO Analysis

157

Figure 4-1 Principal Component Analysis

158

Figure 4-2 - Volcano Plots.

159

Figure 4-3 Glutathione DEGs

160

Figure 4-4 qPCR Comparison with RNAseq Data

161

CHAPTER 5. IDENTIFICATION OF CONSERVED GENETIC FEATURES BETWEEN HUMANS AND DROSOPHILA IN AGING

Modified from a manuscript under review to Journal PLOS Computational Biology.

Authors & Affiliations

Joe Webb1, Andrew Bolstad2, Elizabeth McNeill1

1) Department of Food Science and Human Nutrition, Iowa State University, Ames, IA,

USA

2) Department of Electrical and Computer Engineering, Iowa State University, Ames, IA,

USA

Key Words: Aging, Differential Expression Analysis, RNA-seq, Prefrontal Cortex, XGBoost

Ethics approval and consent to participate

All studies contained within this manuscript received ethics approval from their respective IRB ethics committees prior to study initiation. More information regarding the individual study informed consent forms from all individuals participating is within

Supplemental Table 1.

Data and software availability

The data used in this study are publicly accessible through the National Institutes of

Health Sequence Read Archive (SRA). All Ascension numbers can be found in the supplementary data files, along with BioProject numbers in Table 1. The IPython notebook describing the workflow, python/R scripts and instructional files for this manuscript are available on GitHub at: https://github.com/joelwebb/AgingRNAseq.

162

Abstract

Despite the universal nature of aging, characterizing the molecular differences between healthy aging and neurological disease remains a challenging problem. Aging strongly affects the prefrontal cortex (PFC), which governs executive function, learning, and memory. Previous sequencing studies have demonstrated aging alters gene expression and methylation in the PFC, however the extent to which these changes are conserved across species is unknown. To examine conserved aging mechanisms, RNA Sequencing data from human PFC and fly heads were analyzed to determine if the transcriptomic signatures could predict biological ages within and across species or classify samples into young, middle-aged or old age groups. Interestingly, these analyses revealed that 42 conserved genes can accurately predict aging in Drosophila (R2=0.95) and in humans (R2=0.53). For classification analysis, the transcriptomic signatures were able to classify Drosophila with a mean accuracy of 99% and classify human samples with a mean accuracy of 88%. These data suggest that RNA seq data from the PFC or fly head modeled with

XGBoost can be used to predict chronological age, performing on par with previously published approaches using other types of sequencing data or different tissue types. Overall, these results demonstrate there are conserved aging-associated changes in the brain transcriptome that can be used to predict age and including data from model organisms may provide better insight into aging than studying a single species alone.

163

Introduction

Most basic research conducted on the molecular mechanisms of aging have examined single short-lived model organisms, such as mouse, and the resulting anti-aging therapeutics that have been developed have been primarily unsuccessful at protecting against the detrimental effects of aging [1]. Employing a comparative approach to study conserved aging phenotypes across multiple species may provide deeper insight into the molecular aging signatures than studying aging within a single species. Since aging is the strongest risk factor for developing neurodegenerative diseases, identification of conserved molecular signatures in the brain that can be used to predict lifespan would allow for development of novel therapeutic strategies to slow cognitive aging.

Age-related cognitive decline varies on an individual basis, and previous research has indicated that cognitive decline may be due to altered structure of the prefrontal cortex (PFC) [2] and altered white matter integrity [3]. The PFC is vulnerable to age-related morphology changes

[4], and the age-related alterations in the cortex occur across multiple species [5]. Cognitive function has been activity-regulated transcription [6], but the extent to which gene expression in the PFC is conserved across aging is unknown. A subset of age-related transcriptomic changes in the PFC are conserved between humans and mice [7] but identifying conserved age-related transcriptomic changes across other model organisms may uncover novel therapeutic pathways for preventing age-related cognitive decline or vulnerability to neurodegeneration.

Here, we combined previously published composed of publicly available RNA sequencing datasets across humans and Drosophila to identify conserved aging genes in the human PFC/fly head. The primary goal was to use machine learning to predict aging as a proxy for identifying novel aging associated genes. We focused on examining if any age-related transcriptomic changes are conserved across species and determine if the conserved genes can

164 effectively predict chronological age across and within species. Subsequently, we examined if these transcriptomic signatures could also accurately classify samples into broad age ranges of young, middle-aged, or old samples. Overall, this study sought to identify how age prediction based on RNA sequencing data can effectively model age, revealing underlying important aging genes that strongly predict chronological age.

Methods

Literature & Dataset Search

To identify available sequencing data, we searched for published data using the terms

“aging”, “RNA-seq”, “brain transcriptome”, “cortex”, and “heads” in articles on PubMed and in the NIH Sequence Read Archive (SRA) database. All eligible studies were published online before August 2019. The following inclusion criteria were used to select data for the analysis: 1) tissue including fly head or human prefrontal cortex; 2) age; 3) lack of disease and lack of treatment (except aging) or presence of control samples; and 4) Illumina format raw next- generation sequencing files. The following exclusion criteria were also used to select data for this study: 1) no controls or neurologically normal samples; 2) no age information or incubation temperature data for fly samples due to the differences in lifespan according to temperature; 3) presence of a treatment using injections, Traumatic Brain Injury or sham-surgery procedures; and 4) studies including only microarray data. When we encountered any overlapping data of the same study population included across multiple SRA submissions in more than one publication, only the most complete study was used for our analysis leaving only a single occurrence for each biological sample.

165

Data Acquisition

Publicly available data downloaded from the NIH SRA included in Bioprojects outlined in Table 1. Files were downloaded from the SRA and reads that were downloaded in SRA format were converted using fastq-dump from the NIH SRA-Toolkit [8]. Data were categorized by age for each species, representing young, middle-aged, and old as outlined in Supplemental Table 2.

Quality Control & Adapter Trimming

All reads were analyzed with Fastqc for quality control and reads with low quality scores

(average quality < 10) were discarded. Adapter sequences were trimmed using BBDUK [9].

Reads that matched known Truseq or Nextera adapters sequences were removed during trimming. Individual study manuscripts and supplemental data were examined to identify if reads were sequenced using a forward or reverse library preparation kit.

Alignment & Read Quantification

Reference fasta genome files and genome annotation gtf files were downloaded from the

Ensemble genome browser and Flybase.org browser. Human reads were aligned to the GRCh38 release 94 of the Ensemble Human Genome, and fly reads were aligned to the DMEL release 25

Flybase genome using the STAR v2.5.2 aligner [10]. Transcripts aligning to specific genes were counted using the STAR - - quantMode geneCounts function to map transcripts to each genome.

Files containing gene counts for all samples are included in as Supplemental Table 5 &

Supplemental Table 6.

166

Data Filtering & Quality Control

Bioinformatics processing and data wrangling were conducted in Python, while differential expression analysis was conducted using R. Following mapping genes were filtered out if they were not expressed in any samples or had fewer than 10 counts in for each species.

For comparison of orthologous genes, the DRSC Integrative Ortholog Prediction Tool was used to identify genes conserved across species. After removing non-orthologous genes, data filtered across all samples to remove genes with less than 10 rad counts.

KEGG/GO Analysis & Protein Interaction Maps

The Kyoto Encyclopedia of Genes and Genomes (KEGG) database was queried to identify the functional pathways corresponding to each gene in each species. Gene Ontology

(GO) Analysis was used to group genes according to cellular location, function and biological processes using the PANTHER (Protein ANalysis THrough Evolutionary Relationships)

Classification System Version 15 (http://www.pantherdb.org) for classifying genes into distinct protein pathways. Protein interaction maps were visualized with the STRING database

(https://string-db.org/). Maps represent physical protein interactions or indirect interactions and are based upon computational analysis across 24,584,628 proteins from 5,090 organisms.

Algorithm Selection

For algorithm selection, 13 models were selected for comparison using regression based on previously published algorithms for modelling sequencing data to predict age [14]. The top performing algorithm using default hyperparameters were then used for all subsequent analysis.

Hyperparameter optimization or adjustment was not performed on any algorithm during selection for regression analysis or classification.

167

Regression & Classification

To predict biological age, all data were split into a training set (75%) and a testing set

(25%) by randomly selecting samples using the train_test_split module in scikit-learn. For combined analyses across species, train-test splits equally stratified samples according to species and during individual species assessments, samples stratified according to age groups to prevent different age group class distributions from biasing the models. Age prediction models were fit using scikit-learn v0.20.3 [15] in Jupyter notebooks v.4.4.0 running Python 2.7.15rc1. Naive

Bayes,LDA, K-nearest neighbors, Logistic Regression, Linear regression, LASSO, Gradient

Boost, ADAboost, Huber Regression, ARDR Regression, Random Forest, and other statistical models were implemented using the default parameters in scikit-learn. Model performance was evaluated based on mean absolute error, R2, and median absolute error scores in sklearn. All analyses were conducted in batches of 1000 random sampling tests with replacement to estimate the mean for each model performance metric and a 95% confidence interval for R2. The mathematical modelling behind each classification and regression algorithm is beyond the scope of this publication, therefore for reads interested in more in-depth descriptions please reference

Hastie and colleague’s book, The Elements of Statistical Learning [16].

Results

Publicly Available Data Characteristics

The transcriptomic datasets composed of human, rat, and fly samples used in this study are described in Table 1. In total, 289 raw fastq files were downloaded from NCBI and processed to obtain between approximately 2 million and 40 million reads per sample. 86-94% of the

Drosophila reads, and 96% of the human reads were mapped to their respective reference sequences. In Drosophila, the reads mapped to approximately 17664 genes, and 58735 genes in

168 humans. Cross-species homologous analysis included a subset of data from each species, utilizing total 3347 orthologous genes where at least ¼ of the samples were not blank and were conserved across both species. Additional sequencing details about read length and other parameters can be found for each study used in this analysis under their respective bioproject descriptions by searching NCBI found in Table 1.

Algorithm Selection

First, we compared 13 distinct regression algorithms available in Python for predicting human chronological age as shown in Table 2. XGBoost outperformed all other regression algorithms with a meanR2 of 0.62 and a median absolute error of 8.28 years. Similarly, the standard gradient boosting regressor through SciKit-Learn also performed on par with the

XGBoost regression model with a mean R2 of 0.61 and a median absolute error of 8.71 years.

Previously it had been reported that an ensemble using LDA would perform the best on classifying human chronological age, where here we demonstrated that the LDA model showed a mean R2 of 0.13 and a median absolute error of 12.37 years. Therefore, the XGBoost regression algorithm was selected for downstream analysis.

Predicting Chronological Age

Figure 1 depicts how the XGBoost regression algorithm performs when applied to the human dataset and the fly dataset (trained on 75%, tested on 25%). For the human results, Figure

1 A shows a histogram of the R2 values across 1000 random sampling assessments, where XGB predicted chronological age on average with an R2 of 0.62, a median absolute error of 8.28 years. The regression plot contains a sample regression analysis with the average R2. The confusion matrix at the bottom of Figure 1 A depicts the average classification results from the

169 age group classification according to the age ranges in Supplemental Table 1. Figure 1 B depicts that in fly samples, the XGBoost regression algorithm outperformed all other age prediction models, with an average R2=0.89 and a median absolute error of 0.40 days. The confusion matrix at the bottom of Figure 1 A depicts the average classification results from the age group classification in fly samples according to the age ranges in Supplemental Table 1, which on average across 1000 trials was 99% accurate.

Subsequently we sought to examine which genomic features were the best predictors of chronological age in humans and Drosophila. Supplemental Table 3 shows the results for humans and Drosophila comparing different means of feature selection to predict age.

Surprisingly, looking at the genetic features that correlate with age, using the top 1000 genes to predict chronological age in humans and flies more accurately modeled aging. Figure 2 A shows how the top 1000 genes that correlate with human age, sorted by correlation coefficient, predict human age. Figure 2 B shows how the homologous genes in Drosophila for the top 1000 genes that correlate with human age, sorted by correlation coefficient, predict Drosophila age.

Interestingly, these human ages correlated genes more effectively predict Drosophila age than using all available genes as indicated in Supplemental Table 3.

Next, we sought to answer the question: “Do genes that strongly predict fly age also demonstrate predictive ability in human samples?” Figure 2 C shows how the top 1000 genes that correlate with Drosophila age, sorted by correlation coefficient, predict human age. Figure 2

D shows how the top 1000 genes that correlate with Drosophila age, sorted by correlation coefficient, predict Drosophila age. Interestingly, these homologous Drosophila age correlated genes more effectively predict human age than using all available homologous genes shared between flies and humans. Additional regression metrics are included in Supplemental Table 3.

170

After identifying that homologous genes shared predictive ability across species, we wanted to identify a smaller set of genes that could be validated in future animal studies to examine their role in aging. By cross comparing the top 1000 aging correlated features across each dataset, we identified an overlapping 41 shared conserved genes in both sets. In Figure 3, we demonstrate the utility of these conserved correlated features to predict human and

Drosophila age. Figure 3 A depicts the human results, where these 41 genes predicted chronological age with an average R2 =0.52 and mean absolute error of 9.62 years. Additionally, the age group classification using these top genes was 76% accurate across all age groups. Figure

3 B depicts the results for Drosophila, where XGBoost regression predicted chronological age with an average R2 =0.95 and mean absolute error of 0.49 days.

Age Group Classification

Next, we examined if using all available data from each species it was possible to classify samples into age bins as described in Supplemental Table 4. Human samples were grouped according to young (<30 years), middle-aged (30 – 59 years), or old (60 < years). We first assessed which age groups were the most effective using the XGBoost classifier and identified that these 30-year age bins were more accurate than other smaller or larger groups (Data not shown). Using all human genes, the XGBoost classifier was able to obtain an average accuracy of 73% over 1000 random sampling iterations. Using all available homologous genes between flies and humans, the average accuracy was 73%, while the top 1000 human age correlated genes achieved an average accuracy of 82%. Surprisingly, when only using only the 41 conserved aging correlated genes, an average accuracy of 76% was achieved. For Drosophila age categories were broken down into three age ranges: young (< 10 days), middle-aged (10 - 25 days), and old (> 25 days). Other feature selection methods are within Supplemental Table 4.

171

KEGG & Go Analyses

After identifying genes that were differentially regulated from aging, these genes were mapped to KEGG pathways shown in Supplemental Figure 2. KEGG analysis of the top 1000 human aging correlated genes included that the most represented protein class was metabolite interconversion enzymes (PC00262) with the highest category including oxidoreductase

(PC00176) enzymes, with over 100 genes in this category. The analysis of the top 1000 fly aging correlated genes corroborated this result, indicating that the most represented class of proteins was also metabolite interconversion enzymes (PC00262) with the highest category including oxidoreductase (PC00176) enzymes, The primary KEGG enrichment analyses of the top 41 conserved genes indicated that the primary protein pathway was the Gonadotropin-releasing hormone receptor pathway (P06664). Among the top 41 conserved aging genes, the highest represented class was metabolite interconversion enzymes (PC00262).

Discussion

In this study, we analyzed publicly available RNA sequencing of data from aging

Drosophila, and humans to identify aging associated genes by predicting chronological age. Our transcriptome profiling across species revealed 41 novel aging associated genes that strongly predict aging in Drosophila and humans. In addition, we demonstrated that the different features of gene expression from either the PFC in humans or Drosophila heads can be used to accurately predict chronological age using a XGBoost regressor comparable to other studies using transcriptomic data. Finally, we also demonstrated that these data can be used to accurately classify samples into age groups using the XGBoost Classifier.

First, we sought to identify an algorithm that would effectively model aging using human and Drosophila RNA sequencing data. Table 2 demonstrates that among human samples,

XGBoost outperformed other algorithms. Previously it has been reported that deep neural

172 networks [17] or ensemble algorithms using linear discriminant analysis (LDA) [14] may perform more accurately on classifying human transcriptomic data into distinct age groups. In the prefrontal cortex, it seems that LDA performs much worse than boosting algorithms, which may be due to the way that the LDA algorithm accounts for covariance structure to classify samples. Since the data used in this study display gender differences and heterogeneous tissue samples the large covariance might be leading the LDA algorithm to underperform in these data.

With a limited sample size of just over 100 human samples, deep learning approaches might not serve as an effective tool due to the overfitting on such a small training sample. If more data were available, this might serve as a more effective modelling approach.

Next, we compared common machine learning algorithms on predicting chronological age in humans and Drosophila using data from either the PFC or fly head. We demonstrated that the XGBoost regression algorithm outperformed other algorithms on the human and Drosophila datasets, achieving an average R2 of 0.62 in humans and R2 of 0.89 in Drosophila using all available genes. This model in humans provides slightly lower accuracy in predicting chronological age when compared to other tissue culture samples [14], in part due to the more heterogeneous cellular composition of the PFC. Our study utilized samples from PFC tissue containing numerous cell types (glia, neurons, astrocytes, oligodendrocytes) where using single tissue types such as fibroblasts may allow for less variation across samples. Other groups predicting chronological age using different data types such as methylation, blood metabolites or different statistical methods such as deep learning may provide more accurate age.

We demonstrated that using publicly available data, we can accurately estimate the age of

Drosophila samples across multiple studies using the XGBoost algorithm. Using transcriptomic data from fly heads, we achieved an R2 of 0.89, outperforming other methods of predicting

173

Drosophila age such as fluorometrically measuring pteridine 6-biopterin in the heads (R2 = 0.54)

[18]. Other groups focusing on using machine learning approaches from Drosophila transcriptomic data have utilized methods such as the PILGRIM algorithm [19] to identify novel aging genes but have not investigated if these genes provide predictive utility across species. Due to the nature of animal experiments collecting data, normally samples are all collected at a single specific age such as 10 days or 25 days of age. Therefore, these distinct groupings in Drosophila data are more like one another and do not provide continuous aging data unlike the human samples utilized in this study. Therefore, we additionally investigated how algorithms such as

Multinomial Logistic Regression perform, where these data were classified into distinct age groups 100% of the time. Since the focus of this paper was to model human aging to identify aging associated genes, we continued to utilize XGBoost for modelling fly data which outperformed Logistic Regression during smaller feature set comparisons.

To identify the genes that underlie the predictive capacity for predicting chronological age, we first ranked the genes according to their correlation with age as shown in Supplemental

Table 2. We noted in Supplemental Figure 1 that while examining the top 1000 human aging correlated features, the R2 and error metrics began to plateau after including 820 genes in the model, corresponding to correlation coefficient > 0.4. Therefore, we examined how these top

1000 human age correlated genes model chronological age. Surprisingly, including 1/16th the genes we were able to achieve an average R2 of 0.63 which was on par with including the entire set of human genes as shown in Supplemental Table 3. Similarly, in flies, the evaluation metrics began to plateau after the first 50 genes where the top 1000 resulted in an average R2 of 0.92, which was more accurate than including all the available data. Subsequently we examined if these genes could provide accurate age estimation across species, where we noted that the 257

174 homologs from the top 1000 human aging correlated genes predicted fly ages more accurately with an average R2 of 0.93. Using the 844homologs in humans from the top 1000 fly correlated genes, we were able to predict chronological age with an average R2 of 0.57. Finally, we compared these lists of predictive aging genes across species, identifying a gene set of 41 homologous genes in flies and humans. We turned our attention to examining how these genes model aging, noting that in humans they result in an average R2 of 0.52 and an average R2 of

0.95 in Drosophila.

When we examined the classes of the genes that highly predicted age in humans and

Drosophila, we identified that the most upregulated pathways were metabolite interconversion enzymes (PC00262) with the highest category including oxidoreductase (PC00176) enzymes. It is well established that alterations in compounds such as NAD+ that function in regulating oxidoreductase activity are altered with age [20], and that these alterations are also common during Alzheimer’s disease [21]. One gene that was implicated in this pathway was

ENSG00000119723 which encodes for the production Coenzyme Q10, where COQ10 supplementation is thought to combat increased oxidative stress during aging [22].

Recently Zullo et al. [23] showed that human longevity in humans is related to cortex transcriptomic signatures where genes underlying neural excitation and synaptic regulation are downregulated during aging. Downregulation of genes related to neural excitation and synaptic function. By comparing C. Elegans with humans, they showed that neural excitation increases as a function of age and inhibiting the excitation of neurons increases longevity. Even though the lifespan of humans compared to worms differs from 2 weeks to >80 years, additional evidence by Hamilton et al. [24] corroborates the similarities across worms to humans by showing that knocking out NADH dehydrogenase increased longevity, like McCarrol et al. [25].

175

Zahn et al. also compared data on aging transcriptomes showing that the genes underlying the electron transport chain pathway show decreased expression during aging across humans, mice, flies and worms [26]. Their study did not directly compare brain tissue samples across species, but instead looked for more universal aging signatures across different tissue types from previously published studies. Zhuang et al. compared aging between mice and humans across multiple tissues, demonstrating strong correlations in the transcriptomes of the hippocampus in humans and mice. Additionally, they uncovered that the gene GFAP, Glial fibrillary acidic protein, may be related to aging in both species.

Strengths & Limitations

The limitations of this study should be addressed. The current environment in machine learning faces issues with reproducibility [27], where studies conducted using the same random seeds across labs have been shown to produce widely different results where resulting data look like they are from completely different distributions [28]. To combat this issue, our models were run across 1000 random iterations where we reported averages and confidence intervals for each result. Additionally, genetic data is highly correlated in nature, which may explain why several classification models incorrectly predicted some subjects across several re-samplings, highlighting the utility of using newer machine learning models like XGBoost which are more resistant to the bias from multicollinearity. While multicollinearity may explain part of these results, it is unclear if subjects that were incorrectly classified across multiple random resampling’s displayed inherently different transcriptome expression profiles than other subjects at a similar chronological age, or if these errors were due to the inherent random nature of the model. Therefore, future studies should work toward determining why specific samples in an age category display different transcriptomic signatures, if any arise.

176

Another limitation of this study is that these may be biased toward males due to the nature of the current publicly available data, where nearly all samples used in this study were male. The large, unbalanced distribution of biological sexes may have skewed the data to overrepresent changes with aging specifically in males, since < 10% of the samples used in this analysis were female. This limitation could be addressed by future studies including more samples of both biological sexes and making all samples publicly available. Therefore, these results may be more applicable to conserved male aging than aging in both biological sexes.

Additionally, in this study the predictive ability of the human PFC to predict age did not surpass other previously published models using RNA sequencing techniques in sample types

[17]. Due to the heterogeneous cellular composition of the PFC and more so in the fly head, the larger variation between tissue samples could be due to different quantities of cellular differentiation in a tissue or different cell type composition compared against other studies using blood or cell culture techniques. Previously published ensemble models trained on cell culture samples have much lower variance between samples because they can account for inclusion of the same number of cells, same media conditions, the same carbon dioxide concentration, number of passages/differentiations, and cell type distributions. Even though these regression results may not outperform studies using more homogenous samples, the utility of studying heterogenous tissue samples may provide more useful insight than studying cell culture samples or blood samples by providing a more realistic picture of the transcriptomic landscape in a tissue.

Some of the key strengths of this study were that large sample sizes were achieved by combining

RNA seq samples across studies to improve the overall statistical power. Additionally, this also underscores the necessity for new statistical approaches to combine RNA sequencing data more effectively across multiple studies that were generated under different conditions, such as

177

Sequencing machines, RNA extraction protocols, library preparation kits, and sequencing depth.

Another strength of this study was the use of a comparative approach across multiple organisms to model aging, which may lend to novel insights over examining a single species alone.

Comparing our results with previously published sequencing studies that have examined aging in relation to transcriptomic or epigenetic biomarkers [17], we are able to achieve slightly lower R2 and MAE results on nearly 1/20th of the sample size while performing very accurately across species. Due to the way animal studies collect tissues at predefined age groups instead of heterogeneous ages across time, which may explain why a logistic regression classifier can very accurately predict fly and without performing accurately on human data. Since the human data used in this study was collected across a continuous age spectrum, it adds to the strength of our computational models. However, since we only focused on overlapping genes between rats, flies and humans, this study was not positioned to identify novel aging-related transcriptomic patterns within a single species. Additional comparative research studies should be conducted to identify more conserved pathways and mechanisms that may enable longer lifespans in multiple species.

Conclusion

In this study we discovered that when examining gene expression patterns between humans and Drosophila, 41 conserved genes were able to accurately predict aging across species. Several of these genes have been previously unassociated with aging, suggesting that studying aging through a comparative lens may shed light on previously unassociated pathways that are altered during aging. Our dataset combining publicly available transcriptomic data across multiple studies and two species, demonstrated successful prediction of chronological age using neurological transcriptomic signatures. By using machine learning to model aging, we identified novel similarities in aging signatures across humans and Drosophila emphasizing the necessity

178 for additional comparative aging research studies. Manipulating the expression of these conserved aging genes could potentially extend lifespan or enable development of novel anti- aging therapeutic compounds. Taken together, these data demonstrate that examining aging across multiple species may address the pitfalls of studying aging in a single model organism, uncovering new pathways that are important in cognitive health.

Author Contributions Study conception and design: EM, AB, and JW. Sample collection: JW. RNA extraction:

JW. Library construction and sequencing: JW. Analysis and interpretation of NGS data: JW.

Drafting manuscript: EM, AB, and JW. Critical revision: EM, AB, and JW. All authors read and approved the final manuscript.

Disclosures & Competing Interests

There are no competing interests to disclose.

Funding Information

This work is partially supported by the National Institutes of Health (NIH) grants and a

National Science Foundation GRFP. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

179

Acknowledgements

We would like to thank Dr. Peng Liu for her insightful comments during the planning stages of this project Kevin Cavallin & Dr. Mike Baker for his assistance sequencing the samples. Additionally, the authors would like to thank everyone who generated the data making this study possible.

References

1. Klimova B, Novotny M, Kuca K. Anti-Aging Drugs - Prospect of Longer Life? Curr Med Chem. 2018;25: 1946–1953.

2. Grady CL, McIntosh AR, Craik FIM. Task-related activity in prefrontal cortex and its relation to recognition memory performance in young and old adults. Neuropsychologia. 2005;43: 1466–1481.

3. O’Sullivan M, Jones DK, Summers PE, Morris RG, Williams SC, Markus HS. Evidence for cortical “disconnection” as a mechanism of age-related cognitive decline. Neurology. 2001;57: 632–638.

4. Morrison JH, Baxter MG. The ageing cortical synapse: hallmarks and implications for cognitive decline. Nat Rev Neurosci. 2012;13: 240–250.

5. Zhuang J, Zhang L, Dai S, Cui L, Guo C, Sloofman L, et al. Comparison of multi-tissue aging between human and mouse. Sci Rep. 2019;9: 6220.

6. West AE, Greenberg ME. Neuronal activity-regulated gene transcription in synapse development and cognitive function. Cold Spring Harb Perspect Biol. 2011;3. doi:10.1101/cshperspect.a005744

7. Loerch PM, Lu T, Dakin KA, Vann JM, Isaacs A, Geula C, et al. Evolution of the aging brain transcriptome and synaptic regulation. PLoS One. 2008;3: e3329.

8. Leinonen R, Sugawara H, Shumway M, International Nucleotide Sequence Database Collaboration. The sequence read archive. Nucleic Acids Res. 2011;39: D19–21.

9. Bushnell B, Rood J, Singer E. BBMerge - Accurate paired shotgun read merging via overlap. PLoS One. 2017;12: e0185056.

10. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29: 15–21.

180

11. Robinson MD, Oshlack A. A scaling normalization method for differential expression analysis of RNA-seq data. Genome Biol. 2010;11: R25.

12. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15: 550.

13. Benjamini Y, Hochberg Y, Benjamini Y, Hochberg Y, Benjamin Y. Controlling the false discovery rate - a practical and powerful approach to multiple testing. 1995 [cited 11 Mar 2020]. doi:10.2307/2346101

14. Fleischer JG, Schulte R, Tsai HH, Tyagi S, Ibarra A, Shokhirev MN, et al. Predicting age from the transcriptome of human dermal fibroblasts. Genome Biol. 2018;19: 221.

15. Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, et al. Scikit- learn: Machine Learning in Python. J Mach Learn Res. 2011;12: 2825–2830.

16. Hastie T, Tibshirani R, Friedman J. The Elements of Statistical Learning: Data Mining, Inference, and Prediction, Second Edition. Springer Science & Business Media; 2009.

17. Putin E, Mamoshina P, Aliper A, Korzinkin M, Moskalev A, Kolosov A, et al. Deep biomarkers of human aging: Application of deep neural networks to biomarker development. Aging . 2016;8: 1021–1033.

18. Robson SKA, Vickers M, Blows MW, Crozier RH. Age determination in individual wild- caught Drosophila serrata using pteridine concentration. J Exp Biol. 2006;209: 3155–3163.

19. Kacsoh BZ, Greene CS, Bosco G. Machine Learning Analysis Identifies Drosophila Grunge/Atrophin as an Important Learning and Memory Gene Required for Memory Retention and Social Learning. G3 . 2017;7: 3705–3718.

20. Johnson S, Imai S-I. NAD + biosynthesis, aging, and disease. F1000Res. 2018;7: 132.

21. Swerdlow RH. Brain aging, Alzheimer’s disease, and mitochondria. Biochim Biophys Acta. 2011;1812: 1630–1639.

22. Díaz-Casado ME, Quiles JL, Barriocanal-Casado E, González-García P, Battino M, López LC, et al. The Paradox of Coenzyme Q10 in Aging. Nutrients. 2019;11. doi:10.3390/nu11092221

23. Zullo JM, Drake D, Aron L, O’Hern P, Dhamne SC, Davidsohn N, et al. Regulation of lifespan by neural excitation and REST. Nature. 2019;574: 359–364.

24. Hamilton B, Dong Y, Shindo M, Liu W, Odell I, Ruvkun G, et al. A systematic RNAi screen for longevity genes in C. elegans. Genes Dev. 2005;19: 1544–1555.

25. McCarroll SA, Murphy CT, Zou S, Pletcher SD, Chin C-S, Jan YN, et al. Comparing genomic expression patterns across species identifies shared transcriptional profile in aging. Nat Genet. 2004;36: 197–204.

181

26. Zahn JM, Sonu R, Vogel H, Crane E, Mazan-Mamczarz K, Rabkin R, et al. Transcriptional profiling of aging in human muscle reveals a common aging signature. PLoS Genet. 2006;2: e115.

27. Islam R, Henderson P, Gomrokchi M, Precup D. Reproducibility of Benchmarked Deep Reinforcement Learning Tasks for Continuous Control. arXiv [cs.LG]. 2017. Available: http://arxiv.org/abs/1708.04133

28. Colas C, Sigaud O, Oudeyer P-Y. How Many Random Seeds? Statistical Power Analysis in Deep Reinforcement Learning Experiments. arXiv [cs.LG]. 2018. Available: http://arxiv.org/abs/1806.08295

182

Tables & Figures

Table 5-1: Study Characteristics.

Table 1 - Study Characteristics

Bioproject Species Sample Size Age Groups Tissue Type PRJEB7674 Human 14 young / middle-aged / old Prefrontal Cortex PRJNA322318 Human 18 young / middle-aged / old Prefrontal Cortex PRJNA394722 Human 20 young / middle Prefrontal Cortex PRJNA398545 Human 4 young Prefrontal Cortex PRJNA213747 Human 5 young Prefrontal Cortex PRJNA222268 Human 16 young / middle-aged / old Prefrontal Cortex PRJNA271929 Human 38 middle-aged / old Prefrontal Cortex PRJNA505319 Fly 15 young Head PRJNA388952 Fly 32 Young Head PRJNA270175 Fly 24 middle-aged/old Head PRJNA379297 Fly 6 middle-aged Head PRJNA432934 Fly 19 young / middle-aged Head PRJNA320747 Fly 12 old / young Head

183

Table 5-2 Regression Algorithm Performance Comparison.

Mean Median Mean Algorithm square absolute R2 (95% CI) R2 error error 34.6%- XGBoost 0.62 217.05 8.28 80.5% Gradient 36.71 - 0.61 218.12 8.72 Boosting 75.52 24.43 - Adaboost 0.58 235.78 9.27 78.04 Bagging 52.31 - 0.52 269.16 10.56 Regressor 52.32 Random 52.01 - 0.52 269.70 10.68 Forest 52.02 Extra Trees 47.84 - 0.48 290.43 10.75 Regressor 48.89 22.48 - KNN 0.38 341.94 14.10 60.16 Logistic 0.21 440.48 11.06 0.0 - 64.73 Regression 13.4 - LDA 0.13 481.82 12.37 21.84

Niave Bayes -0.08 605.21 15.15 -8.94 - 7.3

Linear -0.43 876.33 16.95 -12.74 - 2.4 Regression Huber -1.01 1115.84 18.40 -2.87 Regression ARDR -2.75 2084.02 16.16 -22.95 Regression

184

Table 5-3 Age Ranges for Age Group Classification in Humans and Drosophila.

Supplemental Table 1. Age Ranges for Group Classification

Young Middle Old Humans < 30 years 30 - 60 years > 60 years Drosophila < 10 days 10 - 25 days > 25 days

185

Table 5-4 Aging Correlated Genes.

Human Gene ID Aging Correlation Coef. Fly Gene ID Aging Correlation Coef. ENSG00000175287 0.648059301 FBgn0284244 0.708576291

ENSG00000167676 0.626566069 FBgn0033926 0.69414102 ENSG00000102878 0.609028595 FBgn0030040 0.69127608 ENSG00000132470 0.602381518 FBgn0039972 0.689038721 ENSG00000143772 0.598961673 FBgn0051469 0.686647611 ENSG00000162496 0.598228468 FBgn0029507 0.676604654 ENSG00000170439 0.596339161 FBgn0033574 0.668959599 ENSG00000214456 0.59463647 FBgn0037683 0.667441474 ENSG00000157343 0.59145007 FBgn0040099 0.666520746 ENSG00000184925 0.591284939 FBgn0005670 0.661148393 ENSG00000163702 0.58753572 FBgn0262717 0.660772239

186

Table 5-5 Regression Tables Predicting Chronological Age

Human Age Regression Results - XGB

Mean Median R2 Number of Algorithm Mean R2 square absolute (95% Genes error error CI) 31.7% - All human genes 15995 0.62 223.9 8.44 80.9% All Homologous 22.9% - 3347 0.54 259.99 9.60 Genes 77.8% Human Aging 32.0% - Correlated 1000 0.63 207.18 8.13 83.1% Genes Fly Aging 22.5% - Correlation 844 0.57 239.27 8.47 79.5% Genes Top 42 Homologous 23.0% - Aging 41 0.52 266.68 9.62 73.6% Correlated Genes

187

Table 5-5 Continued

Drosophila Age Regression Results - XGB

Mean Median Number R2 (95% Algorithm Mean R2 square absolute of Genes CI) error error 71.0% - All Fly genes 6789 0.89 29.23 0.40 99.0% All Homologous 66.6% - 3347 0.88 65.46 1.40 Genes 98.3% Human Aging 79.6% - Correlated 257 0.93 16.72 0.42 99.3% Genes Fly Aging 72.2% - Correlation 1000 0.92 20.17 0.55 99.3% Genes Top 41 Homologous 87.4% - Aging 41 0.95 13.11 0.49 99.1% Correlated Genes

Table 5-6 Classification Tables for Age Group Prediction

Human Age Classification Results - XGB

Average Average Average Average Number of Average Algorithm True True False False Genes Accuracy Positive Negative Positive Negative

All human genes 15995 0.79 23 46 6 12

All Homologous 3347 0.79 23 46 6 12 188 Genes

Human Aging 1000 0.82 24 47 5 11 Correlated Genes

Fly Aging 857 0.79 23 46 6 12 Correlation Genes

Top 41 Homologous 41 0.76 22 44 7 14 Aging Correlated Genes

Table 5-6 Continued

Drosophila Age Classification Results - XGB

Average Average Average Average Number Average Algorithm True True False False of Genes Accuracy Positive Negative Positive Negative

All Fly genes 6789 0.97 26 53 1 1

All Homologous 189 3347 0.94 25 51 2 3 Genes

Human Aging 257 0.98 27 53 0 1 Correlated Genes

Fly Aging Correlation 1000 0.98 26 53 1 1 Genes

Top 41 Homologous 41 0.99 27 54 0 0 Aging Correlated Genes

Table 5-7 Human Raw Read Counts.

190

191

Table 5-8 Drosophila Raw Read Counts.

192

Figure 5-1 Chronological Age Prediction using XGBoost across species.

193

Figure 5-2 Feature Selection using Aging Correlated Genes Across Species.

194

Figure 5-3 Homologous Aging Correlated Genes

195

Figure 5-4 : PANTHER Pathway Analysis.

196

CHAPTER 6. GENERAL CONCLUSIONS

Overall Summary & Conclusions

Chronic diseases including metabolic disorders like T2DM or aging-associated diseases are characterized by alterations in nutrient metabolism with concomitant changes in the transcriptome.[1] During healthy aging in rats, the transcriptome of the prefrontal cortex is thought to increase expression of immune responsive genes [2], while human aging data indicates the dorsolateral prefrontal cortex transcriptome downregulates genes encoding synaptic function [3]. Dietary patterns can play a central role in preventing or attenuating symptoms of chronic disease or, in contrast, directly lead to the development of chronic diseases[4]. Since whole eggs are a rich dietary source of protein, vitamins, and minerals, previous work has examined how egg consumption may play a key role in decreasing risk of chronic disease development [5]. Consuming whole eggs may improve bodyweight management by reducing weight gain [6], enhance micronutrient status, or modify microRNA expression underlying disease [7]. However, the literature surrounding consuming whole eggs remains controversial because some work has suggested that egg consumption can impair circulating glucose metabolism, or even increase the risk of developing cardiovascular disease in individuals with

T2DM[8]. During aging, consuming eggs may improve circulating vitamin D concentrations while also providing a readily available nutritious food at low cost [9]. The first two studies presented within this dissertation demonstrate that consuming diets rich in whole eggs may lead to improved transcriptomic profiles across the prefrontal cortex, kidney, liver, and adipose tissues in rats. The final study in this dissertation demonstrated that using transcriptome profiles from the prefrontal cortex can be used to model aging across multiple species using machine learning to identify novel aging associated genes.

197

Since dietary patterns affect chronic disease risk [4], the first study in this dissertation investigated whether consuming diets rich in whole egg would alter the transcriptomic profile of the prefrontal cortex. Using Sprague Dawley rats fed either AIN93G casein-based diets or whole egg-based diets that provided protein at 20% wt/wt for 2 weeks, it was identified that 53 genes were differentially expressed in the prefrontal cortex (PFC), along with 2 genes in the liver and

20 in the adipose tissue. Surprisingly, whole egg consumption for 2 weeks did not affect gene expression profiles in the kidney. The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses identified that the main pathways that were upregulated in the liver and brain corresponded to glutathione regulation, oxidative stress, and steroid hormone biosynthesis pathways. These results shed light on how identifying the exact nutrigenomic response to certain foods may provide insight into the precision nutrition/personalized medicine era, where food prescriptions with concurrent medicine dosages may be able to ameliorate chronic disease where each one, individually, may not.

Since chronic diseases alter the transcriptomic profiles across various tissues in rats [9] and humans [10], the second study in this dissertation sought to examine how whole egg consumption affects transcriptional profiles during T2DM. Zucker Diabetic Fatty (ZDF) rats and their lean controls were fed either AIN93G casein-based diets or whole egg-based diets that provided protein at 20% wt/wt for 8 weeks to determine if the same changes under normal conditions can be found during diabetes across the PFC, kidney, liver and adipose tissue. In the lean animals, similar gene expression profiles were discovered as the previous study, and surprisingly the most upregulated pathway from feeding whole egg to the diabetic rats was also glutathione regulation. ZDF rats develop low glutathione [11], where 8 weeks of consuming whole egg effectively reverses the reduction in glutathione transcriptional profiles in the liver.

198

Surprisingly, in the ZDF model, whole egg consumption did not affect the brain transcriptome, where a few changes were noted in the lean controls. These findings suggest that consuming whole eggs during T2DM may attenuate some of the oxidative stress that commonly occurs [12] by upregulating the glutathione pathways. Since whole eggs are rich in vitamins and nutrients such as choline and vitamin D, egg consumption may also improve vitamin D status during chronic disease. Future studies should consider examining the transcriptomic impact of each nutrient on these tissues further develop guidelines surrounding the nutrigenomic impact of dietary whole egg.

The last study in this dissertation aimed to identify healthy aging transcriptomic signatures in the prefrontal cortex across humans and Drosophila, where focusing on healthy aging through a comparative biology lens would improve the understanding of how aging impacts gene expression profiles. Results from this study showed that when performing cross- species analyses using aging correlated genes, 41 novel conserved genes were discovered to change as a function of age. Our data examining aging correlated genes within individual species corroborated previously published aging RNA sequencing studies where thousands of genes were differentially regulated as a function of age. Additionally, machine learning methods were successfully applied to the transcriptomic profiles to determine if these aging-associated genes predict biological age across species. These findings suggest that raw RNAseq data across 13 studies and 223 independent samples can be combined to improve the power of an analysis, as well as more effectively model aging than data from a single study. An interesting outcome of this study was that identifying genes that correlate with aging are more predictive of chronological age than genes identified by traditional methods such as differential expression analysis.

199

Strengths & Limitations

The strengths and limitations of the studies contained within this dissertation should be addressed. In the first study in this dissertation, we tested the novel idea of examining how whole egg consumption affects the transcriptomes of multiple tissues using Sprague Dawley rats. One of the key strengths of the second study was that we assessed multiple tissues, providing evidence that even consuming whole eggs for 2 weeks can alter the gene profiles of multiple tissues while accounting for food intake and body weight. One limitation of this study that should be addressed is the small sample size that was used to detect a treatment effect of whole egg on the kidney. Based on previous studies in our laboratory, we expected that if we had a sample size of 6 animals per group, this would yield 90% power, but we were unable to detect any changes in the kidney. However, our original power analysis did not account for the large interindividual variation so future studies employing this technique should increase the sample size to examine the effects of egg more closely on the kidney.

In the second study presented in this dissertation, we examined how T2DM may affect the way that whole egg consumption modifies transcriptomic profiles in the liver, kidney, prefrontal cortex, and adipose tissue. Surprisingly, we discovered that in the ZDF animals, whole egg consumption didn’t alter the expression of any genes in the prefrontal cortex while it altered genes in the kidney that we didn’t discover during the second study. One of the key strengths of this study was the use of the ZDF rats and their genetically similar lean controls. By utilizing these two breeds of rats together, we were able to tease apart the effects of whole egg consumption from the effects of diabetes on individual tissues. Since the ZDF rat is a leptin receptor deficient T2DM model, this does not lend wide applicability to all forms of diabetes

200 because leptin deficiency occurs in less than 5% of all cases of T2DM[13]. Thus, these results should be carefully examined before attempting to translate the results to humans with T2DM.

One main limitation across these two animal studies that should be discussed is the very large dose of whole egg that we utilized to identify these changes in gene expression and microRNA expression. The amount of whole egg used within these diets provided protein at 20% weight/weight, resulting in 43.5% of the total diet by weight which would translate roughly to

870 calories per day coming from eggs in a standard 2000 calorie diet or 12.5 eggs per day assuming 70 calories per large egg. Besides our preliminary data supporting the use of this large dose, using a lower dosage might have prevent identification of differences in the animals that were only consuming egg for 2 weeks. Ideally future studies from our laboratory will follow up these studies with a dose response study design to determine the lowest effective dose yielding increased expression in glutathione metabolism, as well as alterations in microRNA status.

On average, it has been reported that in America, males consume 5.9 eggs per week while females consume and 3.8 eggs per week on average[14], where a similar dose of egg in humans that was used in this study would translate to roughly 14 eggs per day. It has been recorded that among certain bodybuilding groups consume more than 50 eggs per day[15] while these studies demonstrated that consuming 14 eggs per day for as little as 2 weeks may protect against oxidative stress by modifying glutathione expression in multiple tissues. Since this level of egg consumption is roughly 2.5 - 4 times higher than the average egg consumption in the United

States, a combination approach of eating eggs with pharmaceutical drugs, may be a more feasible intervention. Nonetheless, our laboratory has previously reported that even smaller dosages at 7 eggs per day may be effective for weight management during T2DM[6].

201

In the final study, the novel approach of combining data from multiple publicly available

RNA sequencing studies to examine the impact of aging on the PFC was one key strength of the first study, effectively allowing for a large sample size that would be cost prohibitive for a single laboratory to sequence. More importantly, we demonstrated that the modelling techniques that successfully predicted biological age in multiple species was reproducible in our own fly samples. The limitations of this study should also be addressed. While there was a large sample size, there were still too few samples of both biological sexes due to the nature of the publicly available data. These results may be more applicable to males, while female samples composed fewer than 5% of our overall data. Additionally while one strength was analyzing the data through a comparative biology approach, the genes identified in this study may play a lesser role in regulating aging in each of these three species than other genes that change as a function of aging within a single species. Another major limitation of these results is that the analyses were limited to only using available data from samples with read counts greater than 1 for a given tissue. Since not all the genes within a species were included in this analysis, other genes that might be expressed to a lower extent within a tissue that contribute to aging phenotypes were unable to be assessed using the current available technologies.

202

Future Research

Future work examining egg consumption should aid in determining how whole egg consumption affects T2DM since the literature surrounding egg consumption has been controversial due to its contribution of dietary cholesterol content [16] and previously reported associations with cardiac disease [14]. In the last few months, a large study has suggested that in a large clinical trial across multiple continents eggs may not significantly contribute to these detrimental outcomes [17] suggesting that egg consumption might not confer the detrimental effects described when averaging across a large enough population. Outside of this controversy, many questions remain unanswered about egg consumption: 1) What mechanism underlies these changes in transcriptomic profiles across multiple tissues? 2) Are the microRNAs that were significantly altered from egg consumption purely endogenous or could they be contributed from the eggs? 3) Does the changing microRNA expression in these tissues during diabetes underlie the differences in weight gain? 4) What is the minimum effective dosage and duration of whole egg consumption required to achieve these benefits? Based on the current results presented in this dissertation, the mechanism by which whole egg consumption affects the transcriptome remains elusive. Future studies should consider employing a combinatorial approach whole egg with other functional pharmaceutical compounds that have been proven effective in attenuation of diabetic symptoms to optimize the potential benefits of egg consumption at a much lower dose.

As more studies emerge that focus on modelling aging, they should strive to use human subjects at the same chronological age to more successfully identify aging biomarkers that predict lifespan. Continued developments in this area would provide useful statistical tool for evaluating lifestyle changes or new therapeutics aimed at slowing the progression of aging. In

203 this final study, we created a computational approach to predicting age using RNA sequencing data from the prefrontal cortex across multiple species in order to identify aging-associated genes with strong predictive utility, which could allow future studies to estimate chronological age of humans or flies or identify how alterations in these genes affect lifespan. Outside of this work, several questions remain surrounding tools that successfully predict chronological age: 1) Are these same techniques applicable to RNA sequencing data from less invasive tissues and can similar levels of precision be achieved? 2) Would the addition of more data and more computationally intensive models such as deep neural networks provide better accuracy? 3) Are these results translatable to other species that defy traditional aging characteristics like naked mole rats? 4) Should a tool like this be created for use outside of the research setting, allowing consumers to “assess their chronological age” at home? Based on the results from the study in this dissertation, precisely estimating human chronological age remains a challenging problem and further developments will improve how age prediction can be used to identify aging associated changes in the transcriptome. Future studies examining this topic should work toward developing better statistical protocols for data normalization across different sequencing methods. These developments would afford new opportunities to combine data more easily across studies, allowing more reliable results and theoretically more affordable precision when predicting biological age.

204

References

1. Aramillo Irizar P, Schäuble S, Esser D, Groth M, Frahm C, Priebe S, et al. Transcriptomic alterations during ageing reflect the shift from cancer to degenerative diseases in the elderly. Nature Communications. 2018;9: 1–11. doi:10.1038/s41467-017-02395-2

2. Duggan MR, Joshi S, Tan YF, Slifker M, Ross EA, Wimmer M, et al. Transcriptomic changes in the prefrontal cortex of rats as a function of age and cognitive engagement. Neurobiology of Learning and Memory. 2019;163: 107035. doi:10.1016/j.nlm.2019.107035

3. de Jager PL, Ma Y, McCabe C, Xu J, Vardarajan BN, Felsky D, et al. Data descriptor: A multi- omic atlas of the human frontal cortex for aging and Alzheimer’s disease research. Scientific Data. 2018;5. doi:10.1038/sdata.2018.142

4. Bouchard-Mercier A, Paradis AM, Rudkowska I, Lemieux S, Couture P, Vohl MC. Associations between dietary patterns and gene expression profiles of healthy men and women: A cross-sectional study. Nutrition Journal. 2013;12: 24. doi:10.1186/1475-2891-12-24

5. Park SJ, Jung JH, Choi SW, Lee HJ. Association between egg consumption and metabolic disease. Korean Journal for Food Science of Animal Resources. Korean Society for Food Science of Animal Resources; 2018. pp. 209–223. doi:10.5851/kosfa.2018.38.2.209

6. Saande CJ, Webb JL, Curry PE, Rowling MJ, Schalinske KL. Dietary Whole Egg Reduces Body Weight Gain in a Dose-Dependent Manner in Zucker Diabetic Fatty Rats. The Journal of Nutrition. 2019;149: 1766–1775. doi:10.1093/jn/nxz143

7. Wade B, Cummins M, Keyburn A, Crowley TM. Isolation and detection of microRNA from the egg of chickens. BMC Research Notes. 2016;9. doi:10.1186/s13104-016-2084-5

8. Tran NL, Barraj LM, Heilman JM, Scrafford CG. Egg consumption and cardiovascular disease among diabetic individuals: A systematic review of the literature. Diabetes, Metabolic Syndrome and Obesity: Targets and Therapy. Dove Medical Press Ltd.; 2014. pp. 121–137. doi:10.2147/DMSO.S58668

9. Sárközy M, Szucs G, Fekete V, Pipicz M, Éder K, Gáspár R, et al. Transcriptomic alterations in the heart of non-obese type 2 diabetic Goto-Kakizaki rats. Cardiovascular Diabetology. 2016;15: 110. doi:10.1186/s12933-016-0424-3

10. Wu C, Chen X, Shu J, Lee CT. Whole-genome expression analyses of type 2 diabetes in human skin reveal altered immune function and burden of infection. Oncotarget. 2017;8: 34601–34609. doi:10.18632/oncotarget.16118

11. Raza H, John A, Howarth FC. Alterations in glutathione redox metabolism, oxidative stress, and mitochondrial function in the left ventricle of elderly zucker diabetic fatty rat heart. International Journal of Molecular Sciences. 2012;13: 16241–16254. doi:10.3390/ijms131216241

205

12. Asmat U, Abad K, Ismail K. Diabetes mellitus and oxidative stress—A concise review. Saudi Pharmaceutical Journal. Elsevier B.V.; 2016. pp. 547–553. doi:10.1016/j.jsps.2015.03.013

13. Wang B, Chandrasekera P, Pippin J. Leptin- and Leptin Receptor-Deficient Rodent Models: Relevance for Human Type 2 Diabetes. Current Diabetes Reviews. 2014;10: 131–145. doi:10.2174/1573399810666140508121012

14. Fuller NR, Sainsbury A, Caterson ID, Markovi TP. Egg consumption and human cardiometabolic health in people with and without diabetes. Nutrients. MDPI AG; 2015. pp. 7399– 7420. doi:10.3390/nu7095344

15. Chinese Bodybuilder Eats 70 Egg Whites Per Day Ahead Of First Show - Generation Iron Fitness & Bodybuilding Network. [cited 14 Mar 2020]. Available: https://generationiron.com/chinese-bodybuilder-eats-70-egg-whites-per-day-ahead-of-first-show/

16. Zhong VW, van Horn L, Cornelis MC, Wilkins JT, Ning H, Carnethon MR, et al. Associations of Dietary Cholesterol or Egg Consumption with Incident Cardiovascular Disease and Mortality. JAMA - Journal of the American Medical Association. 2019;321: 1081–1095. doi:10.1001/jama.2019.1572

17. Dehghan M, Mente A, Rangarajan S, Mohan V, Lear S, Swaminathan S, et al. Association of egg intake with blood lipids, cardiovascular disease, and mortality in 177,000 people in 50 countries. The American Journal of Clinical Nutrition. 2020 [cited 13 Feb 2020]. doi:10.1093/ajcn/nqz34

APPENDIX IACUC APPROVAL

206

0-1 - IACUC Approval Letter

This letter signifies the Institutional Animal Care and Use Committee’s approval of our research projects within this dissertation.