View of the Literature

Roles of Adipose Tissue-Derived Factors in Adipose Tissue Development

and Lipid Metabolism

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Jinsoo Ahn, M.S.

Graduate Program in Ohio State University Nutrition

The Ohio State University

2015

Dissertation Committee:

Kichoon Lee, Ph.D., Advisor

Earl H. Harrison, Ph.D.

Ramesh Selvaraj, Ph.D.

Ouliana Ziouzenkova, Ph.D.

Jinsoo Ahn

2015

Abstract

Obesity is a global trend and major risk factor for serious diseases including type 2 diabetes, heart disease, and hypertension. Obesity is characterized by excess fat accumulation, especially in the visceral area. The pathogenic effects related to common obesity are largely attributed to dysregulated secretion of adipokines followed by insulin resistance in peripheral tissues when adipose tissue mass is altered. White adipose tissue serves as a dynamic endocrine organ as well as a major energy reservoir for whole-body energy homeostasis. Adipokines influence various metabolic processes in the body including adipocyte differentiation; however, precise physiological roles of adipokines need to be further investigated. In addition, a large proportion of adipokines still needs to be identified.

Information from the gene expression omnibus (GEO) profile, a public repository for microarray data, combined with confirmatory studies on mRNA and protein expression were used to identify a novel adipose tissue-specific gene, chordin-like 1

(Chrdl1). Further analysis showed that Chrdl1 encodes a putative secreted protein which is a new adipokine. Chrdl1 expression increases during 3T3-L1 adipocyte development in vitro and mouse adipose tissue development in vivo. This pattern, combined with a dramatically increased lipid accumulation and adipocyte differentiation in Chrdl1- overexpressing 3T3-L1 cells suggests that Chrdl1 is a novel pro-adipogenic adipokine which plays stimulatory roles during adipocyte development.

Adipose tissue is also of great significance in the production of food animals.

Reduction of adipose tissue mass will lead to partitioning dietary nutrients towards muscle instead of storage as fat, which enhances feed efficiency and decreases production costs. The net decrease in adiposity is achieved by down-regulation of fat accumulation and up-regulation of fat breakdown or lipolysis. The regulation of lipolysis is likely to be applied to control fat content and increase meat production. The initial step of lipolysis is catalyzed by adipose triglyceride lipase (ATGL) which is the rate-limiting step in this process. Comparative gene identification-58 (CGI-58) and G0/G1 switch gene 2 (G0S2) are known to be an activator and an inhibitor of ATGL-mediated lipolysis, respectively.

In the current study, porcine G0S2 was abundantly expressed in adipose tissue and the liver, and its expression was spatially confined in fat cell fraction and temporally specified in late developmental stages, suggesting that G0S2 expression is related to adipocyte development (pro-adipogenic) as well as inhibition of ATGL (anti-lipolytic). In another study, the expression of bovine G0S2 and CGI-58 in muscle of high-marbled steers of Hanwoo, a Korean cattle, was investigated. The regulation of lipolysis in muscle has a potential to increase marbling (intramuscular fat) in beef carcasses. The expression of G0S2 and CGI-58 was significantly higher in longissimus dorsi muscle of high- marbled steers than in less-marbled bulls, suggesting that both genes are novel candidate biomarkers of marbling in beef cattle. In addition, G0S2 was highly expressed in intramuscular fat portion separated from the longissimus dorsi muscle, whereas the expresson of CGI-58 was significantly higher in the remaining portion consisting of the longissimus dorsi muscle without intramuscular fat. It suggests that G0S2 inhibits ATGL- mediated lipolysis in intramuscular fat while CGI-58 activates ATGL in lipids of

iii myocytes (intramyocellular triglycerides) which serves as an immediate source of energy during exercise.

In another recent study, RBP7 was identified as a novel adipose-specific gene in avian species through gene expression profiling after performing chicken microarray along with real-time PCR and Western blot analysis. Because the RBP7 promoter region was revealed to contain several potential binding sites for adipogenic transcription factors, transgenic quail containing a green fluorescent protein (GFP) gene under the control of the RBP7 promoter were generated by lentivirus-mediated gene transfer. GFP expression in transgenic quail was specific to adipose tissue and increased after adipocyte differentiation. These findings provide compelling evidence that the RBP7 promoter-gene construct can be used to overexpress target genes in adipose tissue.

In conclusion, this study has found a new visceral adipokine (Chrdl1), marbling biomarkers (G0S2 and CGI-58), and an adipose-specific gene (RBP7) and its promoter for overexpression of target genes in vivo. It has been shown that Chrdl1 increases lipid accumulation and adipocyte differentiation of 3T3L1 preadipocytes. Expressions of G0S2 and CGI-58, regulators of ATGL-mediated lipolysis, were significantly higher in longissimus dorsi muscle of high-marbled steers. Avian RBP7 gene showed an exclusive expression adipose tissue, and its promoter was used to generate transgenic quail expressing a target gene (GFP) in adipose tissue. Together, this study advances adipose tissue biology through identification of genes related to adipokine, marbling, and adipose-specific expression.

Acknowledgements

First, I would like to express my sincere gratitude to my advisor, Dr. Kichoon Lee, for offering me the opportunity to conduct this research. His continuous guidance and support helped me the entire time required to accomplish my doctoral research. I am grateful of sharing his experience, knowledge, and enthusiasm. Without his advice, this study would not have been successful.

I also would like to sincerely thank members of my dissertation committee, Dr.

Earl Harrison, Dr. Ramesh Selvaraj and Dr. Ouliana Ziouzenkova for their insightful comments and all the encouragements.

I would like to appreciate to Dr. Jeff Firkins, the director of my Interdisciplinary

Ph.D. Program in Nutrition, and Ms. Amanda Hargett, the program manager, for their instructions and advice from start to finish.

I would also thank all of previous and current members in the Lee Lab and my colleagues and fellow students: Ms. Yeunsu Suh, Dr. Sangsu Shin, Aishlin Lee, Xiang Li,

Dr. Yan Song, Shujin Yang, Dr. Young Min Choi, Jibin Zhang, Paula Chen, Elizabeth

Kim, Ju Yeon Park for their suggestions and kind assistance, and Dr. Henry Zerby, Dr.

Steven Moeller, Dr. Macdonald Wick, Dr. Joe Ottobre, Ann Ottobre, Ben Wenner, Mike

Cressman, Yang Cheng, and all other colleagues in the Dept. Animal Sciences for their support and collaborations.

Finally, I would like to give my gratefulness to my late father (Prof. Jong-Cheol

Ahn), mother (Seung-Hae Jung), and brother (Taesoo Ahn) for encouraging me to pursue this degree and supporting me throughout my life.

Vita

1999 ...... B.S. Biotechnology, Yonsei University,

Seoul, Korea

2011 ...... M.S. Foods and Nutrition, Ohio University

2011 to present ...... Graduate Research Associate,

Ohio State University Nutrition,

The Ohio State University

Publications

1. Ahn J, Oh SA, Suh Y, Moeller SL, Lee K. (2013) Porcine G0/G1 switch gene 2 (G0S2) expression is regulated during adipogenesis and short-term in-vivo nutritional interventions. Lipids, 48(3):209–18.

2. [Song Y, Ahn J], Suh Y, Davis ME, Lee K. (2013) Identification of novel tissue- specific genes by analysis of microarray databases: a human and mouse model. PLoS One, 8(5):e64483.

3. Ahn J, Li X, Choi YM, Shin S, Oh SA, Suh Y, Nguyen TH, Baik M, Hwang S, Lee K. (2014) Differential expressions of G0/G1 switch gene 2 and comparative gene identification 58 are associated with fat content in bovine muscle. Lipids, 49(1):1–14.

4. [Hassan A, Ahn J], Suh Y, Choi YM, Chen P, Lee K. (2014) Selenium promotes adipogenic determination and differentiation of chicken embryonic fibroblasts with regulation of genes in fatty acid uptake, triacylglycerol synthesis, and lipolysis. Journal of Nutritional Biochemistry, 25(8):858–67.

vii

5. Zhang J, Suh Y, Choi YM, Ahn J, Davis ME, Lee K. (2014) Differential expression of cyclin G2, cyclin dependent kinase inhibitor 2C and peripheral myelin protein 22 genes during adipogenesis. Animal, 8(5):800–9.

6. Choi YM, Suh Y, Ahn J, Lee K. (2014) Muscle hypertrophy in heavy weight Japanese quail line: Delayed muscle maturation and continued muscle growth with prolonged upregulation of myogenic regulatory factors. Poult Sci, 93(9):2271–7.

7. Ahn J, Shin S, Suh Y, Park JY, Hwang S, Lee K. (2015) Identification of the avian RBP7 gene as a new adipose-specific gene and RBP7 promoter-driven GFP expression in adipose tissue of transgenic quail. PLoS One, 10(4):e0124768.

Fields of study

Major Field: Ohio State University Nutrition

viii

Table of Contents

Abstract ...... ii

Acknowledgments...... v

Vita ...... vii

Table of Contents ...... ix

List of Tables ...... xiii

List of Figures ...... xiv

List of Abbreviations ...... xvi

Chapter 1: Introduction ...... 1

Chapter 2: Review of the literature ...... 4

2.1. Physiology and pathophysiology of adipose tissue ...... 4

2.2. Adipokines and endocrine functions of white adipose tissue ...... 7

2.3. Development of adipocytes ...... 9

2.3.1. Cell origin of adipocytes ...... 9

2.3.2. Commitment ...... 9

2.3.2.1. Mesenchymal stem cells (MSCs) ...... 10

2.3.2.2. Factors involved in the commitment step ...... 10

2.3.2.3. Bone Morphogenetic Protein (BMP) and commitment ...... 11

2.3.2.4. Wnt signaling and commitment ...... 12

2.3.2.5. Hedgehog signaling and commitment ...... 15

2.3.2.6. Zfp423 and commitment ...... 16

2.3.3. Terminal differentiation ...... 17

2.3.3.1. Preadipocyte cell line ...... 17

2.3.3.2. Growth arrest and induction of differentiation ...... 18

2.3.3.2. The roles of C/EBPβ and mitotic clonal expansion ...... 18

2.3.3.4. Early stage of transcriptional network ...... 20

2.3.3.5. Late stage of transcriptional network ...... 21

2.3.3.6. Bone Morphogenetic Protein (BMP) and the differentiation ...... 22

2.3.3.7. Wnt signaling and the differentiation ...... 23

2.4. The role of chordin-like 1 as a BMP regulator during adipogenesis ...... 24

2.4.1. Chordin ...... 24

2.4.2. Potential roles of Chordin-like 1 during adipogenesis ...... 24

2.5. Discovery of adipose-specific genes by cDNA microarray-based gene expression

Profiling……………………………………………………………………..…….26

Chapter 3: Expression of a novel adipose tissue-specific Chordin-like 1 gene and its influence on adipocyte development ...... 29

3.1. Abstract ...... 29

3.2. Introduction ...... 31

3.3. Materials and Methods ...... 32

3.4. Results ...... 37

3.5. Discussion ...... 41

3.6. Acknowledgements ...... 45

Chapter 4: Porcine G0/G1 Switch Gene 2 (G0S2) expression is regulated during adipogenesis and short-term in vivo nutritional interventions ...... 51

4.1. Abstract ...... 51

4.2. Introduction ...... 53

4.3. Materials and Methods ...... 55

4.4. Results ...... 60

4.5. Discussion ...... 65

4.6. Acknowledgements ...... 70

Chapter 5: Differential expressions of G0/G1 switch gene 2 and comparative gene identification-58 are Aassociated with fat content in bovine muscle……………………79

5.1. Abstract ...... 79

5.2. Introduction ...... 81

5.3. Materials and Methods ...... 82

5.4. Results ...... 88

5.5. Discussion ...... 92

5.6. Acknowledgements ...... 98

Chapter 6: Identification of the avian RBP7 gene as a new adipose-specific gene and

RBP7 promoter-driven GFP expression in adipose tissue of transgenic quail…………109

6.1. Abstract ...... 109

6.2. Introduction ...... 111

6.3. Materials and Methods ...... 112

6.4. Results ...... 120

6.5. Discussion ...... 123

6.6. Acknowledgements ...... 127

Chapter 7: Summary………….……………………………….…………..……………136

Chapter 8: Concluding remarks……………………………….…………..……………139

References ...... 141

xii

List of Tables

Table 6.1. Production of transgenic quail using lentivirus-mediated gene transfer into stage X quail embryos…………………………………………………………………135

xiii

List of Figures

Figure 2.1. Adipokines grouped according to their physiologic roles…..……….…….....9

Figure 2.2. BMP/smad signaling during the adipogenic commitment..………………....12

Figure 2.3. Wnt signaling during the adipogenic commitment and differentiation.....…..15

Figure 2.4. Hedgehog signaling during the adipogenic commitment..…………………..16

Figure 2.5. Zfp423-mediated up-regulation of adipogenesis…….…….…………….…..17

Figure 2.6. Adipocyte terminal differentiation and regulating factors…………………..22

Figure 2.7. Proposed inhibitory function of Chrdl1 against BMP signaling…….………26

Figure 3.1. Relative Chrdl1 mRNA expression in various tissues …...……..…………..46

Figure 3.2. Chrdl1 mRNA and protein expression ……………………………………...47

Figure 3.3. Developmental regulation of Chrdl1 expression ……………………………48

Figure 3.4. Increased adipocyte differentiation by Chrdl1 overexpression….…………..49

Figure 3.5. Alternative splicing of Chrdl1………………………………………………50

Figure 4.1. Comparison of G0S2 amino acid sequences………….……………………..71

Figure 4.2. Tissue distribution of porcine G0S2…………………………………..……..72

Figure 4.3. Porcine G0S2 gene expression in SV and fat cell fraction…...……………..73

Figure 4.4. Porcine G0S2 gene expression during adipocyte differentiation…….….…..74

Figure 4.5. Porcine G0S2 gene expression during adipose tissue development……..…..75

Figure 4.6. Porcine G0S2 gene expression during nutritional intervention.……………..76

Figure 4.7. Alignment of G0S2 nucleotide sequences.………………………..…………77

xiv

Figure 5.1.Alignment of G0S2 coding sequences………………....………………...…..99

Figure 5.2. Alignment of CGI-58 coding sequences ……………………..…………....100

Figure 5.3. Homology analysis of G0S2 amino acid sequences…………….…….……102

Figure 5.4. Homology analysis of CGI-58 amino acid sequences ....………………...... 103

Figure 5.5. Alternative splicing patterns of bovine CGI-58..……………..…………....105

Figure 5.6. Tissue distribution of bovine G0S2 and CGI-58 mRNA……….…….……106

Figure 5.7. The mRNA expressions of FABP4, G0S2 and CGI-58 in Hanwoo ….……107

Figure 6.1. RBP7 mRNA expression in various tissues…………………….…….……128

Figure 6.2. RBP7 amino acid sequences and protein expression ….……….…….……129

Figure 6.3. Analysis of RBP7 promoter ………………………..….……….…….….…130

Figure 6.4. GFP expression in various tissues of selected lines under UV light.....……131

Figure 6.5. GFP expression in SV and fat cell fractions and during primary adipocyte differentiation …………………………………………..………….……….……..……132

Figure 6.6. Alignment of RBP7 nucleotide sequences.………………….……………..133

Figure 6.7. GFP expression in various tissues under UV light.....…..…………………134

List of Abbreviations

ATGL…………………………………………..…...Adipose triglyceride lipase

BMP……………………………………….………Bone morphogenetic protein

C/EBP ……………………………………………..CCAAT/enhancer-binding protein

CGI-58 ...... ……Comparative gene identification-58

Chrdl1……………………………………………Chordin-like 1

FC ...... ……Fat cell

G0………………………………………………Generation 0

G1………………………………………………Generation 1

G2………………………………………………Generation 2

GEO………………………………………..Gene expression omnibus

G0S2 ...... G0/G1 Switch Gene 2

IMF ………………………………………..Intramuscular fat

IMTG…………………………………….Intramyocellular triglycerides

MSC………………………………….…Mesenchymal stem cell

PPAR ………………………………..Peroxisome proliferator-activated receptor gamma qPCR.………………………………….Quantitative real-time PCR

RBP7 ………………………………….Retinol .binding protein 7

TAG.…………………………………..Triacylglycerol

SV……………………………………….Stromal-vascular

xvi

Chapter 1: Introduction

The rapid increase in obesity over the past few decades has become a public health problem in most developed countries (Bleich et al., 2008). More than 1.4 billion adults worldwide were overweight, and at least 500 million of them were obese in 2008 (World

Health Organization, 2013). In the United States, nearly two-thirds of the adult population is overweight and approximately one-third is obese (Centers for Disease

Control and Prevention, 2013). It has been shown that obesity is highly associated with the incidence of a number of diseases, including type 2 diabetes, hypertension, dyslipidemia, and metabolic syndrome (Nguyen et al., 2008). Obesity is characterized by excess accumulation of adipose tissue, particularly in the visceral compartment, and dysregulated secretion of adipokines which leads to decreased insulin sensitivity in peripheral tissues, including skeletal muscle and the liver (Galic et al., 2010). Adipose tissue is a metabolically dynamic endocrine organ that plays a critical role in the maintenance of whole-body energy homeostasis. White adipose tissue, the predominant type of adipose tissue, was identified as a major site for secretion of bioactive peptides, collectively named adipokines, as well as storage of excess energy as triglycerides

(Kershaw & Flier, 2004).

Adipokines influence various metabolic processes, such as fatty acid metabolism, insulin sensitivity, adipocyte differentiation, angiogenesis, and inflammation (Dyck et al.,

2006). Among adipokines that have been identified, some of them, such as leptin, are

1 well-characterized, but their precise physiological roles need to be further investigated.

As noted previously, approximately 40% of expressed genes in adipose tissue that are currently unknown, and about 20% of these unknown genes may encode secretory proteins (Maeda et al., 1997; Kershaw & Flier, 2004). The identification of unknown genes that encode adipose-specific secretory proteins along with the continued characterization of known genes is likely to provide further insight into the endocrine function of adipose tissue. The pathogenic effects associated with obesity are possibly due to modifications in secretion of unknown adipokines when adipose tissue mass is altered.

In addition, adipose tissue is of great importance in the food animal industry, as decreasing adiposity is a main determinant for feed efficiency of partitioning nutrients from feed into food products, such as muscle. Considering feed as a major component of production costs, increasing feed efficiency by reducing adipose tissue mass has great economic value in the meat-producing enterprise. The net reduction of adipose tissue mass can be achieved by decreasing fat deposition and increasing the breakdown of fat

(lipolysis) in adipose tissue, which ultimately leads to repartitioning of energy from adipose tissue to other peripheral tissues and transferring nutrients from feed to muscle.

Genomic selection for improving feed efficiency is an important breeding objective

(Pryce et al., 2014). Beef cattle and pork production are significant components of agriculture and diets throughout the world. As global demand increases, in 2015, international beef and pork exports are forecasted to grow by nearly 2 percent to a record

9.9 million tons and 4 percent to 7.2 million tons, respectively (USDA, FAS., 2014). To meet the growing demand, beef and pork producers will need to increase meat production

2 by their animals. Regulation of lipolysis has recently gained attention for controlling fat content in domestic animal species. Lipolysis is composed of sequential catabolic enzymatic reactions. The first rate-limiting reaction is a hydrolysis of the first ester bond of stored triglycerides releasing fatty acids which is catalyzed by adipose triglyceride lipase (ATGL). The regulation of ATGL-mediated lipolysis through its activator, comparative gene identification-58 (CGI-58) (Lass et al., 2006), and its recently identified inhibitor, G0/G1 switch gene 2 (G0S2) (Yang et al., 2010b), has a great potential to efficiently regulate ATGL-mediated lipolysis.

The objectives of my current dissertation research were to identify a novel adipokine and to investigate its potential function in adipocyte differentiation and fat accumulation. This research provides candidate human obesity-linked genes, and it can also be applied to improve animal production. In my published research, expression of

CGI-58 and G0S2 genes of porcine and bovine species was analyzed which is linked to

ATGL-mediated lipolysis and the amount of fat content, including marbling. Lastly, overexpression of target genes in adipose tissue in vivo under the regulation of new promoter was investigated.

Chapter 2: Review of the literature

2.1. Physiology and pathophysiology of adipose tissue

Adipose tissue is divided into two functionally different types, which are brown and white adipose tissues in mammals. Brown adipose tissue is specialized for producing heat via non-shivering thermogenesis. In brief, energy from a proton gradient, which is derived from oxidation of fatty acids and glucose, is used for the generation of heat by uncoupling the oxidative process from ATP synthesis within the inner mitochondrial membrane through uncoupling protein 1 (UCP1; thermogenin) (Cannon and Nedergaard,

2004; Koppen and Kalkhoven, 2010). There is a high demand for oxygen supply due to abundance of mitochondria, and a large amount of capillaries in brown adipose tissue is responsible for the brown color (Koppen and Kalkhoven, 2010; Saely et al., 2012). Heat production in brown adipose tissue is essential for newborns and small mammals to survive cold stress because of their relatively large body surface compared to body volume, insufficient insulation by white adipose tissue, and inadequate muscle shivering

(Saely et al., 2012). The roles of some brown adipose tissues in adult humans around cervical and supraclavicular regions have yet to be determined, but they may protect vital organs by warming the blood supply (Gustafson, 2010; Saely et al., 2012). Brown adipose tissue is smaller than white adipose tissue, containing less triglyceride in multiple small droplets which mainly function as a fuel for thermogenesis (Lowell and Flier., 1997;

Trayhurn 2005).

White adipose tissue is a loose connective tissue, which is one of the most prevalent tissues in the human body that consists largely of lipid-laden adipocytes (Shen et al., 2003; Bjorndal et al., 2011). Besides adipocytes, adipose tissue contains adipocyte precursor cells (preadipocytes), fibroblasts, endothelial cells, vascular smooth muscle cells, blood cells, pericytes, mesenchymal stem cells, macrophages, and adipose tissue matrix (Schäffler et al., 2005). These various cell types, along with adipocytes which comprise about 25% of the total cell population in adipose tissue, give rise to complex roles of this tissue during metabolic responses. Studies are required to investigate how these individual cellular compartments interplay and function as a unit. White adipose tissue serves as a major long-term energy reservoir in mammals, storing excess energy as triglycerides in a large unilocular lipid droplet and releasing energy in the form of free fatty acids by hydrolyzing triglycerides in time of energy demand, which can be oxidized to generate ATP in mitochondria of other peripheral tissues (Koppen and Kalkhoven,

2010). Triglycerides are highly efficient for energy storage because these complex lipids are anhydrous and have a high energy density of 39 kJ g-1 (9 kcal g-1), in contrast with 17

KJ g-1 (4 kcal g-1) for proteins and carbohydrates (Trayhurn et al., 2005).

In order to maintain energy homeostasis, white adipose tissue dynamically responds to nutritional status within a short period of time, and it causes remodeling of the structure of adipose tissue. However, unlike healthy adipose tissue, pathological adipose tissue expansion in obese subjects can cause capillary rarefaction, limited angiogenesis, and subsequent hypoxia (Sun et al., 2011; Ouchi et al., 2011). In the presence of nutritional surplus, adipocytes are capable of expanding as much as 20-fold in diameter (hypertrophy), and new adipocytes are formed from preadipocytes present in

5 adipose tissue (Jo et al., 2009). White adipose tissue also provides thermal insulation and cushioning between organs (Ottaviani et al., 2011). In addition to the roles of white adipose tissue on the regulation of lipogenesis, lipolysis, insulation and cushioning, over the past two decades, white adipose tissue has been shown to be an active endocrine organ, and white adipocytes have been demonstrated to be secretory cells which release hormones or cytokines termed adipokines (Kershaw & Flier, 2004; Galic et al., 2010).

In the human, white adipose tissue can be divided into subcutaneous adipose tissue and visceral adipose tissue. Subcutaneous adipose tissue is found beneath the skin and is further divided into abdominal, femoral, and gluteal adipose tissues (Frühbeck,

2008). Visceral adipose tissue surrounds internal organs and is subdivided into intraperitoneal (e.g., omental and mesenteric white adipose tissues) and extraperitoneal

(e.g., perirenal) depots. Brown adipose tissue is found in supraclavicular and subscapular regions. In particular, omental and mesenteric visceral adipose tissues have been connected to the risk of developing obesity-related diseases (Bjørndal et al., 2011). In mice, subcutaneous adipose depots are composed of anterior and posterior subcutaneous

(e.g. inguinal white adipose tissues) depots. The visceral adipose depots, including mesenteric, retroperitoneal, epididymal (gonadal), and perirenal depots, are located in abdomen cavities and the thorax (Cinti, 2005). The pattern of adipose tissue distribution is affected by various factors, including gender, age, nutritional and hormonal status

(Kirkland et al., 2002; Shen et al., 2003). Visceral fat distribution in android obesity in males contributes to a greater risk for metabolic diseases than subcutaneous gluteofemoral fat distribution in gynoid obesity in females (Shi and Clegg, 2009). Strong association between individual adipose depots and pathological conditions implicates

6 specific functional differences between adipose depots. Although high-fat diet and hormonal changes in the menopausal transition contribute to accumulation of visceral fat in women (Yasmeen et al., 2013), the mechanisms underlying sex-dependent differences of fat distribution and visceral fat formation still remain unclear.

2.2. Adipokines and endocrine functions of white adipose tissue

Adipokines or adipocytokines are defined generally as bioactive peptides or cytokines secreted from adipose tissue that signal to peripheral organs and the immune system

(Matsuzawa et al., 1999; Baillargeon and Rose, 2006; Radin et al., 2009). Adipokines operate on target cells through the bloodstream in an endocrine pathway or act at the local level via autocrine and paracrine mechanisms (Baillargeon and Rose, 2006). A number of adipokines and their systemic or local physiological functions, including insulin sensitivity, inflammation, energy balance, adipocyte differentiation, and angiogenesis, have been reported (Blüher & Mantzoros, 2015).

The first described adipokine, adipsin, plays a role in immune response as a complement factor D, and its expression is impaired in obesity models (Cook et al., 1987;

Flier et al., 1987). Leptin has been discovered by Friedman‘s group and shown to be effective in weight loss in severe obesity with genetically inherited leptin deficiency; however, it showed little effect on common obesity and a mouse model of diet-induced obesity (Heymsfield et al., 1999; Dardeno et al., 2010). Impaired leptin signaling pathways that may be responsible for leptin resistance have not been fully understood.

Leptin is also involved in the regulation of appetite and food intake by stimulating the hypothalamus in an endocrine fashion (Kalra et al., 1999). Adiponectin encoded by the

AdipoQ gene is negatively correlated with obesity and display a variety of endocrine and autocrine/paracrine effects, including antidiabetic functions (Dadson et al., 2011).

Resistin is known to induce insulin resistance and is associated with inflammatory processes (Ouchi et al., 2011). Cytokines such as tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6) are involved in pro-inflammatory responses (Ouchi et al., 2011).

Vascular endothelial growth factor (VEGF) is responsible for angiogenesis and vascularization through an autocrine/paracrine manner (Lee et al., 2007). Recent proteomic profiling studies have characterized the human ‗adipokinome‘ from primary human adipocytes and reported 44 newly identified putative novel adipokines. Further functional studies showed that heme oxygenase-1 (HO-1) was significantly elevated in the plasma and visceral fat of obese patients, which may be a candidate adipokine for further research related to obesity and obesity-associated disorders (Lehr et al., 2012).

Although adipokines have great therapeutic potential for curing obesity and obesity-related diseases as the list of adipokines continues to grow, there is a need for further comprehensive research on their mechanism of reaction and clinical applications.

Hormones Immune system Anti-hyperglycemic Complement factors Leptin (e.g. adipsin) Adiponectin Vaspin Inflammation Visfatin Omentin Pro-inflammatory

cytokines Hyperglycemic (e.g. TNF-α, IL-6) Adipose tissue Resistin

Growth factors Enzymes VEGF Chemerin

Figure 2.1. Adipokines grouped according to their physiologic roles.

2.3. Development of adipocytes

2.3.1. Cell origin of adipocytes

Adipocytes are originated from pluripotent mesenchymal stem cells (MSCs), which reside primarily in the stroma of bone marrow and adipose tissue (Gronthos et al., 2001).

MSCs have the capacity to differentiate into several cell lineages including osteocytes, chondrocytes, adipocytes, and myocytes (Covas et al., 2008; Takada et al., 2009). The pluripotent progenitor cells become committed to preadipocytes, which are the precursors of adipocytes, when appropriately regulated and lose their capacity to develop into other lineages (commitment phase) (Gesta et al., 2007). The preadipocytes then differentiate into adipocytes under the influence of stimulatory signals following several stages: growth arrest, mitotic clonal expansion, and terminal differentiation.

2.3.2. Commitment

2.3.2.1. Mesenchymal stem cells (MSCs)

MSCs are derived from the mesoderm which begins to form when a layer of cells migrate between the endoderm and ectoderm during gastrulation. The mesodermal layer divides into the axial, intermediate, lateral plate, and paraxial mesoderm (Pourquie, 2001). Each of these regions is assumed to generate local MSCs and ultimately local adipose tissues.

Both white and brown adipocytes are derived from MSCs, but brown adipocytes are derived from Myf5 positive progenitor cells; whereas, white adipocytes are derived from

Myf5 negative progenitor cells (Koppen and Kalkhoven, 2010). Also, Myf5 positive progenitor cells give rise to myocytes. Brown adipocytes are therefore related to myocytes and muscle. Some thermogenic brown adipocytes, termed beige adipocytes, have been detected in white adipose tissue after specific conditions, such as chronic cold exposure (Koppen and Kalkhoven, 2010). Because of the short lifetime of MSCs in culture, the C3H10T1/2 line, which is established from C3H mouse embryos, has been commonly used as a pluripotent stem cell line. This cell line displays a fibroblast-like shape with long cytoplasmic tails in culture and is morphologically stable for more than

450 days in continuous culture (Reznikoff et al., 1973).

2.3.2.2. Factors involved in the commitment step

Commitment or determination is the process by which mesenchymal stem cell precursors of mesodermal origin are converted into preadipocytes, and therefore the adipocyte lineage. Little is known about the mechanisms controlling the commitment of MSCs into preadipocytes (Otto and Lane, 2005). Recently, several factors have been identified that

10 promote or inhibit the commitment of MSCs to the adipocyte lineage. These factors include bone morphogenetic protein (BMP) family members (BMP4 and BMP2), Wnt proteins, and Sonic hedgehog (Shh). Among them, BMP4, BMP2, and Wnts may activate the commitment, whereas Shh signaling has an inhibitory role on the commitment as described below (Tang et al., 2004; Huang et al., 2009; Bowers and Lane, 2008; Spinella-

Jaegle et al., 2001). Another study has proposed that up-regulation of PPARγ2 expression in preadipocytes by zing-finger protein 423 (Zfp423) may be a potential marker of commitment (Gupta et al., 2010).

2.3.2.3. Bone Morphogenetic Protein (BMP) and commitment

Bone morphogenetic proteins (BMPs) are a group of growth factors that belongs to the transforming growth factor β (TGFβ) superfamily. About 20 BMPs are characterized, and they were originally discovered as regulators of bone and cartilage formation, but their multi-functional properties have also been identified (Chen et al., 2004). BMPs signal through two types of serine/threonine kinase receptors: type I (BMPr1) and type II

(BMPr2) (Chen et al. 2004). In particular, it has been shown that the C3H10T1/2 cells express receptors BMPr1a and BMPr2, and binding of BMPs to the receptor complex

(i.e., BMPr1a:BMPr2) can induce the phosphorylation of downstream targets, smad1/5/8 and p38 MAP kinase. Then, smad1/5/8 creates a complex with smad4 and translocates to the nucleus where it regulates gene expression of a downstream target gene for lysyl oxidase (Lox) (Tang and Lane, 2012). Lysyl oxidase is an extracellular copper-containing amine oxidase that catalyzes cross-linking of collagen and elastin at lysine residues during extracellular matrix formation. Studies have demonstrated that BMP2 and BMP4

11 can induce the commitment of C3H10T1/2 cells to preadipocytes through induction of

Lox expression (Tang et al., 2004; Huang et al., 2009). However, the commitment was entirely inhibited by knockdown of Lox (Huang et al., 2009). This suggests that BMP- mediated Lox signaling plays a crucial role in the commitment of preadipocyte formation from pluripotent stem cells. However, whether Lox signaling is communicative or incompatible with the terminal differentiation process remains largely unknown. Future studies need to investigate if there are downstream mediators that may link BMP4/smad signaling to early adipogenic transcription factors (e.g. C/EBPβ) and late adipogenic transcription factors (e.g. C/EBPα and PPARγ) for terminal differentiation.

BMP2 BMP4

Extracellular BMPr1a:BMPr2

Cytosol Phosphorylation Smad1/5/8

Smad4 Translocation

Nucleus Smad4 transcription factor

? C/EBPs LOX PPARγ

Commitment Differentiation Mesenchymal Preadipocyte Stem Cell (MSC) Adipocyte

Figure 2.2. BMP/smad signaling during the adipogenic commitment. 12

2.3.2.4. Wnt signaling and commitment

Studies have shown that the role of Wnt signaling during the adipogenic commitment may be opposite from its role during differentiation, such that Wnts serve to increase the number of preadipocytes and promote commitment, but this signal may be repressed before new preadipocytes can undergo terminal differentiation into adipocytes, otherwise,

Wnts disrupt differentiation (Bowers and Lane, 2008). It has been shown that activation of both canonical Wnts (e.g. Wnt10b) and noncanonical Wnts (e.g. Wnt5a) inhibits terminal adipocyte differentiation through reducing the expression of C/EBPα and

PPARγ (Bowers and Lane, 2008; Song et al., 2014). Linkage between Wnt signaling and the commitment of MSCs has been suggested in microarray studies combined with RT-

PCR and Western blot analyses. Interestingly, proliferating preadipocytes (A33 cells) showed the following characteristics compared to their precursors (C3H10T1/2 stem cell): i) Dramatic up-regulation of activators of canonical Wnt signaling pathway (e.g. R- spondins-2 and -3), ii) accumulation of β-catenin, and iii) differential expression of

Lef/Tcf transcription factors (Bowers and Lane, 2008). This implicates that Wnt signaling is activated during the commitment of C3H10T1/2 stem cells to preadipocytes.

Wnt proteins are a family of secreted glycoproteins that regulate a variety of cellular responses such as cell fate determination, proliferation, and survival (Ross et al,

2000). Wnt signaling pathways are evolutionally conserved and are comprised of β- catenin-dependent canonical and β-catenin-independent noncanonical pathways (Komiya and Habas, 2008). In ―canonical‖ Wnt/β-catenin signaling, when Wnt ligands are absent,

β-catenin is recruited to a degradation complex, whereby glycogen synthase kinase-3β

(GSK-3β) phosphorylates β-catenin and primes it for ubiquitination and proteasomal

13 degradation (Takada et al., 2009; Tang and Lane 2012). Binding of canonical Wnt ligands (e.g. Wnt1, Wnt3a, Wnt5b, Wnt7a, and Wnt10b) to frizzled (FZD) receptors activates Wnt signaling, leading to dissociation of the degradation complex and it allows

β-catenin to accumulate and translocate into the nucleus, where it acts as a coactivator of the lymphoid-enhancing factor/T-cell factor (Lef/Tcf) transcription factors in order to activate Wnt target genes (Takada et al., 2009; Tang and Lane 2012).

β-catenin-independent ―noncanonical‖ pathways are mediated by several kinases, including mitogen-activated protein (MAP) kinase, protein kinase C (PKC), and calcium/calmodulin-dependent protein kinase II (CaMKII) (Takada et al., 2009). Binding of noncanonical Wnt ligands (e.g. Wnt4, Wnt5a, and Wnt11) to FZD promotes rapid cellular responses, such as changes in cell shape (Takada et al., 2009); however, the mechanism by which noncanonical Wnt singals are transmitted to the nucleus has not been fully identified.

In addition, it has been proposed that canonical Wnt signaling is present upstream of BMP4 expression, as C3H10T1/2 cells treated with Wnt3a showed up-regulation of

BMP4 expression (Jackson et al., 2005). However, it has not been demonstrated that the

BMP4/smad pathway is downstream of other Wnts. Future studies need to identify the relationship between Wnts and BMP4 signaling.

Canonical Noncanonical

Extracellular Wnt10b / Wnt3a Extracellular Wnt5a LRP Frizzled LRP Frizzled Dishevelled Dishevelled Cytosol Cytosol

β-catenin Ca++ PKC Translocation CaMKII

Nucleus β-catenin Nucleus + Lef/ Tcf transcription factor

BMP4 ? C/EBPα, PPARγ

Commitment Differentiation Mesenchymal Preadipocyte Stem Cell (MSC) Adipocyte

Figure 2.3. Wnt signaling during the adipogenic commitment and differentiation.

2.3.2.5. Hedgehog signaling and commitment

Hedgehog (Hh) signaling has an inhibitory effect on adipogenesis of C3H10T1/2 stem cells, as shown in studies using sonic hedgehog (Shh) that inhibited adipogenic differentiation of C3H10T1/2 cells in the presence and absence of BMP2, which has adipogenic potential (Zehentner et al., 2000; Spinella-Jaegle et al., 2001). These results suggest that the Shh protein may be antagonistic to pro-adipogenic BMP2 and block the commitment and/or differentiation of pluripotent C3H10T1/2 cells. However, the mechanisms linking Hh signaling and adipogenesis still remain unclear. 15

Extracellular Patched x Smoothened activated Cytosol

Ci/Gli intact Translocation

Nucleus Ci/Gli transcription factor

BMP2 Commitment Differentiation Mesenchymal Preadipocyte Stem Cell (MSC) Adipocyte

Figure 2.4. Hedgehog signaling during the adipogenic commitment.

2.3.2.6. Zfp423 and commitment

Searching for transcription factors specific to pro-adipogenic fibroblasts (Swiss 3T3 fibroblasts) has led to the identification of Zfp423 as a potential regulator of PPARγ2 during commitment (Gupta et al., 2010). This study showed that, by overexpressing

Zfp423, non-adipogenic NIH 3T3 fibroblasts can be transformed into adipogenic cells expressing an increased amount of PPARγ2, without a significant modification of upstream regulators of PPARγ2. It suggests that Zfp423 may be an alternative mediator of PPARγ2, which bypasses known pathways of transcription factors such as CREB,

KLF, GR, C/EBP β and -δ. It has been known that Zfp423 contains the smad-interacting domain and transcriptionally co-activates smad transcription factors through the 16

BMP/smad pathway (Gupta et al., 2010). However, future studies need to investigate the upstream regulatory pathways for the transcription factor, Zfp423, and it is unknown whether Zfp423 action on the transcription of PPARγ2 is direct or indirect.

? Extracellular

Cytosol

Nucleus Zfp423 transcription factor

Direct effect? PPARγ2

Commitment Differentiation Pro-adipogenic Preadipocyte fibroblasts Adipocyte

Figure 2.5. Zfp423-mediated up-regulation of adipogenesis.

2.3.3. Terminal differentiation

2.3.3.1. Preadipocyte cell line

Differentiation of preadipocytes into adipocytes has been extensively studied using established 3T3-L1 preadipocyte lines, which are derived from 17- to 19- day

17 disaggregated mouse embryos (Green and Kehinde, 1974). The clonal cell line contains a homogenous cellular population at the same differentiation stage, which probably provides equivalent responses to treatments. Electron micrographic studies have confirmed that ultrastructural features of differentiating 3T3-L1 cells are consistent with those of mature adipocytes in vivo (Novikoff et al., 1980).

2.3.3.2. Growth arrest and induction of differentiation

In order to induce differentiation, 3T3-L1 preadipocytes are grown to confluence at which they become growth-arrested in the G1 phase of the cell cycle by contact inhibition

(Waki et al., 2007). Standard differentiation inducing cocktail is composed of DMEM supplemented with high glucose, 10% fetal bovine serum, synthetic glucocorticoid dexamethasone (DEX), 3-isobutyl-1-methylxanthine (IBMX), and insulin (Waki et al.,

2007). Similar to β-adrenergic signaling, IBMX activates the cAMP signaling pathway and subsequently, phosphorylated protein kinase A (PKA) catalyzes phosphorylation and activation of cyclic AMP response element-binding protein (CREB) (Tang and Lane,

2012). CREB transcriptionally activates the expression of CCAAT/enhancer-binding protein beta (C/EBPβ) (Tang and Lane, 2012).

2.3.3.3. The roles of C/EBPβ and mitotic clonal expansion

The requirement of mitotic clonal expansion (MCE) has been controversial and some reports have suggested that 3T3-L1 preadipocytes can differentiate into adipocytes without MCE (Qiu et al., 2001). Recently, multiple recent studies have shown that MCE is a prerequisite for terminal differentiation of 3T3-L1 preadipocytes through blocking

18 both MCE and differentiation at the same time by using inhibitors of cell cycle or DNA synthesis (Yeh et al., 1995; Reichert & Eick, 1999; Tang et al., 2003a; Li et al., 2007b).

According to the latter view, approximately 16 hours after induction, 3T3-L1 preadipocytes reenter the cell cycle to convert from G1 to Synthesis (S) phase and undergo about two or three rounds of mitosis, termed mitotic clonal expansion (Tang et al., 2003a). Then, the cells exit the cell cycle and phenotypic changes occur.

Preadipocytes become spherical, accumulate lipid droplets filled with triglycerides, and display characteristics of adipocytes (Tang et al., 2003a).

The rapid up-regulation of C/EBPβ expression upon adipogenic induction plays a dual role: A prerequisite for MCE in the early stage as previously shown that mouse embryo fibroblasts (MEFs) from C/EBPβ-knockout mice do not undergo MCE and a transcriptional activator of C/EBPα and PPARγ in the late stage of differentiation (Tang et al., 2003b). Thus, targeting C/EBPβ has a substantial potential for regulation of adipogenesis. Although DNA replication during mitosis of MCE presumably increases the accessibility of the promoter region of C/EBPα and PPARγ (Cornelius et al., 1994), two master transcription factors for terminal adipocyte differentiation, the detailed molecular basis of the need for MCE before differentiation is still elusive.

Interestingly, human primary preadipocytes do not require MCE for adipogenesis

(Tomlinson et al., 2006), similar to C3H10T1/2 cells (Jefcoate et al., 2008). However, it is uncertain that this distinction is caused by a completed clonal expansion before sample collection or early expression of C/EBPα, an antimitotic factor, in human primary preadipocytes contributes to the removal of MCE (Tomlinson et al., 2006). Future studies need to verify the presence or absence of MCE in human preadipocytes, so that nutrient

19 regulators of MCE, such as curcumin, which showed an inhibitory role in MCE and adipocyte differentiation (Kim et al., 2011), can be tested in human preadipocytes.

2.3.3.4. Early stage of transcriptional network

The transcriptional cascade of adipocyte differentiation is regulated by an interplay of various transcription factors. The initial step of adipogenic transcriptional cascade is the induction of transcription factors C/EBPβ and -δ. The signaling pathway of cAMP/CREB is responsible for C/EBPβ expression, whereas the expression of C/EBPδ is mediated by glucocorticoid receptor (GR) signaling (Cao et al., 1991). When adipocyte differentiation is induced by treatment with adipogenic cocktails, the expression of C/EBPβ and -δ are rapidly increased and is followed by MCE, but the expression of C/EBPα and PPARγ is delayed until MCE is completed (Salma et al., 2006). Both the C/EBPα and PPARγ genes possess C/EBP regulatory elements at which C/EBPβ binds to activate transcription (Otto and Lane, 2005).

In addition to C/EBPβ and -δ, the early stage consists of other activators. It has been shown that IBMX not only activates cAMP/CREB signaling but also induces expression of KLF4, which in turn increases C/EBPβ expression (Birsoy et al., 2008).

Another zinc finger-containing transcription factor, Krox20, upregulates C/EBPβ expression (Chen et al., 2005) to a higher degree in combination with KLF4 (Birsoy et al.,

2008). In parallel with induction of C/EBPβ and -δ, STAT5A/B has been reported to be activated by factors in fetal bovine serum during the early phase of differentiation and regulate PPARγ expression (Nanbu-Wakao et al., 2002). It has also been shown that

STAT5A forms a complex with GR in the nucleus, and this association is highly regulated during differentiation (Floyd and Stephens, 2003; Baugh et al., 2007).

Interestingly, recent studies proposed that adipogenic signals converge to hotspots in enhancer regions during early adipogenesis to which multiple transcription factors transiently bind providing functional enhanceosomes, and these hotspots are inherited by late adipogenic transcription factors (Siersbaek et al., 2012). For example, STAT5A and

GR transiently co-occupy several chromatin regions during the early stages of adipocyte differentiation (Siersbaek et al., 2011). These cooperative bindings are likely to be important in programming of transcription during adipocyte differentiation. However, little is known about chromatin remodeling during adipocyte differentiation and the molecular mechanisms underlying the coordinate binding.

2.3.3.5. Late stage of transcriptional network

Most induced adipocyte genes contain C/EBP binding motif and PPAR regulatory element (PPRE) on their promoters through, which those genes are regulated by C/EBPα and PPARγ (Macdougald and Lane, 1995). Interestingly, at the PPARγ binding site,

PPRE and C/EBP binding motif shows co-localization, suggesting that this structural proximity on the chromatin plays a critical role in cross-talking and programming the late adipogenic differentiation (Siersbaek et al., 2012). The expression of C/EBPα is maintained by autoactivation and once both C/EBPα and PPARγ are expressed, they up- regulate each other to maintain expression levels (Otto and Lane, 2005). PPARγ forms heterodimers with the retinoid X receptor (RXR) before it binds to the response element on a target gene (Tang and Lane, 2012). It assumed that the regulation of differentiation

21 by C/EBPα takes place at an earlier stage compared to PPARγ, and PPARγ is the central regulator of adipogenesis (Otto and Lane, 2005).

Other transcriptional factors that regulate the late stages of adipocyte differentiation include SREBP-1c and KLF15. Attempts have been made to identify endogenous PPARγ ligands and SREBP-1c has been shown to cause cells to produce endogenous PPARγ ligands which are likely to be fatty acid derivatives (Kim et al.,

1998), and insulin promotes the expression levels of SREBP-1c (Yellaturu et al., 2009).

In addition, KLF15 has been shown to regulate PPARγ expression and induce adipocyte differentiation in 3T3-L1 cells (Mori et al., 2005).

Differentiation Preadipocyte Adipocyte Early stage Late stage

CREB IBMX, cAMP KLF4 C/EBPβ C/EBPα + Krox20

Dex (glucocorticoid) GR C/EBPδ PPARγ KLF15

FBS, growth factors STAT5A/B Ligand

Insulin SREBP-1c

Zfp423 PPARγ2

Figure 2.6. Adipocyte terminal differentiation and regulating factors.

2.3.3.6. Bone Morphogenetic Protein (BMP) and the differentiation

The role of BMPs on the differentiation phase is controversial. Studies have proposed that interaction with two distinct BMP receptors, BMPr1a and BMPr1b, may show different downstream pathways (Muruganandan et al., 2009). A signal from interaction with BMPr1a may converge at C/EBPα and PPARγ, and it regulates both adipocyte differentiation and osteoblastogenesis; however, unlike the transdifferentiation effect of

BMPr1a, BMPr1b signaling ends up with the activation of runt-related transcription factor 2 (RUNX2) and osterix (OSX), which are key determinants associated with osteoblastogenesis (Muruganandan et al., 2009). In contrast, as mentioned above, studies have suggested that BMP 2/4 signaling through the BMPr1a:BMPr2 complex leads to the up-regulation of expression of lysyl oxidase (Lox) (Huang et al., 2009). In addition, other reports have shown that lower concentrations of BMP2 led to differentiation of

C3H10T1/2 cells to adipocytes, while higher concentrations promoted differentiation of osteoblasts and chondrocytes (Wang et al., 1993; James 2013). Interestingly, recent studies have shown that overexpression of BMP3b, which regulates osteogenesis and the development of embryos, inhibited adipocyte differentiation, while its knockdown in

3T3-L1 cells resulted in enhanced adipogenesis (Hino et al., 2012). Future studies may need to investigate the downstream target of BMP3b and its association with transcription factors, PPARγ and C/EBPα.

2.3.3.7. Wnt signaling and the differentiation

Canonical Wnt/β-catenin signaling appears to inhibit adipocyte terminal differentiation by reducing the expression of C/EBPα and PPARγ (Takada et al., 2011). Also, it has been

23 reported that noncanonical ligand Wnt5a is a strong inducer of osteoblastogenesis and acts as a suppressor of PPARγ expression while decreasing the terminal differentiation

(Takada et al., 2011).

2.4. The role of chordin-like 1 as a BMP regulator during adipogenesis

2.4.1. Chordin

During the patterning of dorso-ventral axis in vertebrates and invertebrates, bone morphogenetic protein (BMP)/decapentaplegic (Dpp), a powerful ventralizing factor, signaling is conserved, and Chordin/Short gastrulation (Sog) proteins function as an antagonist of the signaling mechanism, in humans and fruit flies (Abreu et al., 2002). The inhibitory effect of chordin comes from conserved four cysteine-rich (CR) domains, and chordin directly binds to BMP through CR domains, thereby preventing BMP from activating BMP receptors (Abreu et al., 2002). The CR domains are typically 60-80 amino acids in length and are characterized by ten cysteines with a conserved spacing pattern. Between the first two cysteines, a glycine and a tryptophan residue are located.

Also, CXXCXC motif is in the middle, and CCXXC motif is in the C-terminus.

2.4.2. Potential roles of Chordin-like 1 during adipogenesis

Chordin-like 1 (Chrdl1) protein is a secreted protein that consists of three CR domains and, like chordin, it may inhibit BMP (Abreu et al., 2002). Observation of a strong dorsalizing pattern in chrdl1 mRNA microinjection studies using Xenopus embryos suggests that chrdl1 may block the effect of ventralizing BMP (Nakayama et al., 2001).

However, the binding affinity of chrdl1 compared to chordin needs to be determined,

24 considering the loss of one CR domain in chrdl1. It has been reported that the expression of chordin is restricted to the node and the notochord, but chrdl1 is expressed in various mesenchymes as well as the neural plate derivatives at a later stage of development

(Coffinier et al., 2001). This indicates that chrdl1 may act in not only axial patterning, but also in diverse processes possibly including adipogenesis.

Chrdl1 has previously been known as chordin-like, ventroptin, and neuralin-1

(Nakayama et al., 2001; Sakuta et al., 2001; Coffinier et al., 2001). Human chrdl1 cDNA encodes 458 amino acids (AA) containing a putative signal peptide. Multiple forms of human chrdl1 containing small deletions have been described. Mouse chrdl1 cDNA encodes a 447 AA (long-form) containing a signal peptide. The short-form of mouse chrdl1 protein with 333 AA in length is derived from alternative splicing and contains intact three CR domains.

Several previous studies have shown that bone morphogenetic protein 2 and 4

(BMP 2 and 4), which is involved in stem cell determination, induces commitment of

MSCs to preadipocytes (Huang et al., 2009). Chordin-like 1 (chrdl1) is known to be a secreted protein that binds to BMP2/4, antagonizing the function of BMP2/4 during skeletogenesis (Nakayama et al., 2001). However, it is completely unknown if chrdl1 plays a role in adipocyte development by suppressing the commitment step through inhibition of BMP2/4. In addition, it will be interesting to study another target mechanism of chrdl1 action which presumably antagonizes anti-adipogenic properties of

BMP3b while promoting adipogenic differentiation.

BMP2 BMP3b BMP4 Chrdl1 Extracellular ? ? BMPr1a:BMPr2

Cytosol Phosphorylation Smad1/5/8

Smad4 Translocation

Nucleus Smad4 transcription factor

? C/EBPs LOX PPARγ

Commitment Differentiation Mesenchymal Preadipocyte Stem Cell (MSC) Adipocyte

Figure 2.7. Proposed inhibitory function of Chrdl1 against BMP signaling.

2.5. Discovery of adipose-specific genes by cDNA microarray-based gene expression profiling

The cDNA microarray databases provide large-scale gene expression data based on mRNA abundance in various conditions including a variety of tissues, developmental processes, and physiological circumstances. Our previous studies using rat gene-filter microarray analysis have been successfully utilized to discover adipocyte-specific secretory factor (ADSF)/resistin by comparing expression profiles in a variety of tissues

(Kim et al., 2001). The adipose-specific gene expression pattern of ADSF/resistin was

26 confirmed by Northern blot analysis. The presence of N-terminal signal sequence and

Western blot analysis that detected the secreted protein of ADSF/resistin in the conditioned medium indicated that ADSF/resistin is a novel adipokine. Furthermore, transfection of ADSF/resistin into 3T3-L1 cells resulted in the inhibition of adipocyte differentiation, implying that ADSF/resistin functions as an inhibitor of adipogenesis.

Another previous microarray analysis in our lab that compared the gene expression profile in stromal-vascular (SV) cells, mainly containing preadipocytes with lipid-laden adipocytes, revealed that mouse interferon-stimulated gene 12b1 (ISG12b1) is a novel adipocyte-specific gene that is highly expressed in murine adipocytes and inhibits adipocyte differentiation via preventing mitochondrial biogenesis (Li et al., 2009). The results of real-time PCR showed that ISG12b1 mRNA expression was nearly 400-fold higher in fat cell fraction compared to SV cells. Moreover, adipose tissue-specific expression of ISG12b1 was determi]ned by real-time PCR and showed significantly greater expression of ISG12b1 in epididymal, inguinal, and brown adipose tissues (P <

0.001). Adenovirus-mediated ISG12b1 overexpression markedly inhibited adipocyte differentiation with few lipid droplets at post-induction day 10 as shown in Oil-Red O staining. It implicates that ISG12b1 is one of the anti-adipogenic factors. Recently, gene expression omnibus (GEO) profile database derived from DNA microarray have been used in our lab to identify novel tissue-specific genes across the human and mouse model

(Song et al., 2013). After processing GEO DataSet (GDS) from six tissues (kidney, liver, lung, heart, muscle, and adipose tissue) of the human (GDS3142) and mouse (GDS596), semi-quantitative PCR and real-time PCR confirmed three novel tissue-specific genes: heart-specific prune homolog 2 (PRUNE2), adipose tissue-specific activin A receptor,

27 type 1C (ACVR1C), and liver-specific amidohydrolase domain containing 1 (AMDHD1).

This approach combined with functional studies using existing data on GEO profiles derived from a variety of physiological conditions is a powerful platform to discover novel tissue-specific genes. In my dissertation research, GEO profile databases have been used to identify a novel adipose tissue-specific gene, and further studies have focused on its secretion and potential roles during adipocyte development, while providing a new adipokine that can be added to the current list of adipokines.

Chapter 3: Expression of a novel adipose tissue-specific Chordin-like 1 gene and its

influence on adipocyte development

Jinsoo Ahn1, 2, Yeunsu Suh1, and Kichoon Lee1, 2, *

1Department of Animal Sciences and 2 Interdisciplinary Ph.D. Program in Nutrition, The

Ohio State University, Columbus, OH, 43210, USA

3.1. Abstract

Chordin-like 1 (Chrdl1) is a bone morphogenetic protein (BMP) antagonist whose function has recently been described in optical development and neuronal differentiation.

In this study, through analyses of gene expression omnibus (GEO) profile followed by further confirmatory gene and protein expression studies, Chrdl1 was re-discovered as a novel candidate visceral fat-specific adipokine with therapeutic potentials for obesity treatment. During 3T3-L1 adipocyte development in vitro and postnatal mouse adipose tissue development in vivo, Chrdl1 gene and/or protein expression continued to increase.

This late-temporal expression pattern together with a drastic promotion of lipid accumulation and adipocyte differentiation in Chrdl1 overexpressing 3T3-L1 cells suggests that Chrdl1 plays stimulatory roles in adipocyte development as a potential endogenous pro-adipogenic factor. Through cysteine-rich domains, Chrdl1 binds to BMP, however, protein-protein interaction between Chrdl1 and BMP in adipocytes still remain

29 unclear. Thus, it will be important to identify binding partners of Chrdl1 during adipocyte developmental stages. The anti-angiogenic BMP-related function of Chrdl1 was demonstrated in human retinal pericytes, and clustered, regularly interspaced, palindromic repeats (CRISPR) system has been evaluated in various cells and organisms.

However, pro-angiogenic roles of Chrdl1 in adipocytes and Chrdl1 gene ablation in 3T3-

L1 cells using CRISPR system have not been studied. Future research on the function of

Chrdl1 during adipocyte angiogenesis and Chrdl1 gene ablation will help elucidate the roles of Chrdl1 during adipogenesis.

Keywords: chordin-like 1 (Chrdl1), bone morphogenetic protein (BMP), gene expression omnibus (GEO) profile, adipokine, obesity, lipid accumulation, adipocyte differentiation, pro-angiogenic, gene ablation

3.2. Introduction

Genetic factors involved in the development of obesity have been implicated in studies using identical twins who were raised separately, and recent studies have estimated that the heritability of common obesity is 40-70% (Maes et al., 1997; Loos, 2009; Speliotes et al., 2010). Discovery of novel genes that affect adipose tissue development in the general population will be critical for identifying mechanisms underlying the genetic predisposition to common obesity. Each human tissue, including adipose tissue, contains more than 25,000 genes, and most tissues exclusively express ~290 tissue-specific genes

(signature genes) on average (Yu et al., 2006); however, the number of genes that have been identified as being adipose tissue-specific is far below this average. For example, adiponectin has been widely studied after it was revealed as a secretory protein exclusively produced from adipose tissue (Hu et al., 1996). It has been shown that there is a strong negative correlation between plasma adiponectin concentrations in women and adipose tissue mass (Matsubara et al., 2002) and adiponectin improves insulin sensitivity in murine models of genetic and diet-induced obesity (Yamauchi et al., 2001). This suggests that up-regulation of gene expression of adipokines, such as adiponectin, and their secretion into circulation may be possible therapeutic targets for obesity. In this study, a potential adipokine Chordin-like 1 (Chrdl1) was investigated.

Chrdl1 protein, also named ventroptin and neuralin-1, is an antagonist of a secreted bone morphogenetic protein (BMP) (Nakayama et al., 2001; Sakuta et al., 2001;

Coffinier et al., 2001). It has recently been reported that loss-of-function mutations in

Chrdl1 have been described in X-linked megalocornea which is an inherited congenital disorder characterized by enlarged corneas and reduced central corneal thickness,

31 indicating that Chrdl1 plays an important role in anterior segment development (Webb et al., 2012). Another study has reported that overexpression of Chrdl1 promotes neuronal differentiation of rat spinal cord-derived neural stem cells by inhibiting BMP4 (Gao et al.,

2013). Given that there have been several reports about the stimulatory roles of BMPs

(BMP2 and 4) on the commitment of mesenchymal stem cells (mouse embryonic

C3H10T1/2 cells line) to the adipogenic lineage (Huang et al., 2009) and the inhibitory roles of BMP3b on adipocyte differentiation of 3T3-L1 preadipocytes (Hino et al., 2012), it will be interesting to investigate the function of Chrdl1 during adipocyte differentiation.

However, there have no reports about the function of Chrdl1 on adipogenesis.

In the current study, Chrdl1 was identified as a novel adipose-specific gene, which is highly expressed in epididymal visceral white adipose tissue of the mouse and has detectable secretion from this region, suggesting that Chrdl1 is a potential new adipokine produced from visceral fat. Further phenotypic analysis of Chrdl1 overexpression in 3T3-L1 has shown that Chrdl1 may have stimulatory effects on adipocyte differentiation. Taken together, Chrdl1 is a potential visceral adipokine that affects adipocyte differentiation.

3.3. Materials and Methods

Comparative analysis of microarray data

Gene Expression Omnibus (GEO) profiles for human and mouse mRNA transcripts were derived from GEO DataSets (GDSs) in the National Center for Biotechnology

Information (NCBI) archives: GDS596 and GDS3142 for 22,283 human gene expression profiles and 45,101 mouse gene expression profiles, respectively, in various tissues.

Those GEO profiles were sorted as previously described in our report to find adipose- specific genes (Song et al., 2013). Among them, a novel adipose-specific Chrdl1 gene was selected. Based on each GEO sample (GSM) in GEO profiles for Chrdl1 expressions in 14 representative tissues of two to four individual human and mice, relative Chrdl1 mRNA expressions in each tissue were displayed as shown in Fig 3.1.

Human and mouse mRNA and tissue samples

Human RNA and protein extracts of kidney, liver, lung, heart, muscle, and adipose tissue were purchased from Agilent Technologies (Santa Clara, CA, USA). For mouse RNA isolation, liver, brown adipose tissue (BAT), muscle, heart, lung, spleen, kidney, and white adipose tissue (WAT) were collected from 3-month-old FVB mice (n = 3-5) (Li et al., 2009). For obtaining mouse RNA extracts from stromal vascular (SV) and fat cell fractionation, inguinal (subcutaneous) white adipose tissue (WAT) was harvested from 1- month-old FVB mice (n = 3) (Li et al., 2009). For RNA isolation for in vivo developmental time-points of mice, 2-, 5-, 10-, 24-, and 49-d-old FVB mice (n = 3) were subjected to dissection for inguinal WAT.

Isolation of RNA, PCR, and quantitative real-time PCR

To isolate total RNA from tissues and SV and FC fractions, Trizol (Invitrogen Inc.,

Carlsbad, CA) was used according to the manufacturer‘s instruction. The isolated RNA was checked for quality using agarose gel electrophoresis. To measure the quantity of

RNA, a Nanodrop spectrophotometer (Thermo Scientific, Wilmington, DE) was used.

The RNA samples were stored at -80°C until use. Approximately 1 µg of RNA was

33 reverse-transcribed in a 20 µL total reaction to cDNA using Moloney murine leukemia virus (M-MLV) reverse transcriptase (Invitrogen). The thermal cycle of the reverse transcription was 65°C for 5 min, 37°C for 52 min, and 70°C for 15 min.

Exactly 1 µL of cDNA samples was used as templates for PCR in a 25 µL total reaction with Thermo Taq polymerase (New England Biolabs, Carlsbad, CA). The conditions for this reaction were 95°C for 1 min 30 s, 40 cycles of 94°C for 30 s, 57°C for 1 min, 72°C for 1 min, with an additional extension step at 68°C for 10 min. PCR products were separated by using 1% agarose gel electrophoresis. A forward primer for mouse Chrdl1 was 5‘-GTA GAA GAT GGA TGG CAT GAA ATA CAT C-3‘, a reverse primer for mouse Chrdl1 long-form was 5‘-CTG GCT GTT GTG TCT AAC AGT GGT-

3‘, and a reverse primer for the short-form was 5‘-TGG TTG AAT GAG ATT CAC ACA

TAG TC-3‘. A human Chrdl1 forward primer was 5‘-AGT GGA AAA TGG GAG GCA

TGA-3‘ and a reverse primer was 5‘-TGC TTC TCT GTT GGC AGG TTG-3‘. For real- time PCR (qPCR), 2 µL of cDNA was used as a template with the reaction system including AmpliTaq Gold polymerase (Applied Biosystems, Carlsbad, CA). The cycle parameters for qPCR were 95°C for 10 min, 40 cycles of 94°C for 15s, 57°C for 40s and

72°C for 30s, with an additional 82°C extension for 33s. The reference gene for normalization of gene expression levels was mouse cyclophilin (Cyc). The relative gene

-ΔΔ expression levels were calculated using 2 CT (CT: threshold cycle) method (Livak &

Schmittgen, 2001). Primer sequences for mouse Cyc, DLK1, FABP4, SCD1, and FAS are described in our previous report (Li et al., 2009). The primer set for Chrdl1 was as follows: a forward primer 5‘-GTA GAA GAT GGA TGG CAT GAA ATA CAT C-3‘ and a reverse primer 5‘ ACT CGG CTG CAA AGC ACA TT-3‘.

Construction of retroviral vectors and generation of stable cell line

The two forms of mouse Chrdl1 gene was cloned based on the sequences deposited in the

GenBank database: a long form (GenBank NM_001114385.1) and a short form

(GenBank NM_031258.3) with hemagglutinin (HA) tag in the C-terminal region. The two cloned isoforms with HA tags were subcloned into the retroviral vector pQCXIP

(Clontech Laboratories Inc., Mountain View, CA). The vector contains puromycin resistance (Purr). The retroviral packaging cell line Ecopack2TM-293 (Clontech

Laboratories Inc.) was transfected with the pQCXIP retroviral expression vector (empty vector control, vector with HA-tagged long-form, and vector with HA-tagged short-form) using Lipofectamine 2000 (Invitrogen). After 48 h of transfection, a viral soup

(conditioned media containing retroviruses) was collected, filtered, and snap-frozen at -

80° C.

The viral soup was diluted in a 1:1 ratio with a complete media containing

Dulbecco‘s modified Eagle‘s medium (DMEM, Gibco, Grand Island, NY) supplemented with 10 % fetal bovine serum (FBS, Gibco) and the mixture solution of penicillin and streptomycin (Pen Strep; Invitrogen). 3T3-L1 preadipocytes (American Type Culture

Collection, Manassas, VA, USA) were resuspended and cultured in the complete media at 37 °C in 5% CO2. On the day of infection, 3T3-L1 preadipocytes were infected with the diluted viral soup containing pQCXIP empty vector, the same vector with HA-tagged

Chrdl1 long-form, or HA-tagged short-form. Two days after infection the media was changed into the complete media with puromycin (2 µg/ ml) which were replenished every 2 days for stable cell selection within two weeks.

Cell culture and adipogenic differentiation

Stable 3T3-L1 cells were differentiated as previously described (Li et al., 2009). In detail, stable 3T3-L1 cells were cultured in the complete media and grown to confluence. After

2 d post-confluence (Day 0), differentiation of 3T3-L1 preadipocytes to adipocytes was induced by treating for 2 days with a standard differentiation media, which contains 1 µM dexamethasone, 62.5 µM isobutylmethylxanthine (IBMX), and 1 µg/ml insulin (Sigma-

Aldrich Co., St. Louis, MO, USA) additionally in the complete culture media. Two days after induction (Day 2), the differentiation media was changed to an insulin media, which is supplemented with 1 µg/ml insulin in the complete culture media. Two days later (Day

4), the insulin media was changed to the complete culture media for another 4 days and the media was changed every 2 days. Protein was isolated from the 3T3L1 cells at d 0, 4,

8, and 12 post-differentiation using ice-cold 1X lysis buffer (125 mM Tris-HCl pH 6.8,

0.5% SDS).

Western blot analysis

Western blot analysis was conducted as described previously in our report (Li et al.,

2012). Briefly, equal amounts of proteins were loaded onto 15% SDS-PAGE gels and separated proteins were wet-transferred to polyvinylidene difluoride (PVDF) membranes

(Bio-Rad, Hercules, CA). After blocking for 30 minutes in Tris-buffered saline with

Tween-20 (TBST) containing 4% nonfat dry milk, the membranes were incubated overnight at 4°C with primary antibodies: a rabbit polyclonal Chrdl1 antibody (1:1000;

Novus Biologicals, Littleton, Co, USA), a custom rabbit polyclonal Chrdl1 antibody raised against an immunogen including N-terminal amino acid residues 29–42 (1:3000;

AbClon, Seoul, Korea), mouse monoclonal HA tag antibody (1:5000; Cell signaling

Technology, Inc., Denvers, MA, USA), goat polyclonal FABP4 antibody (1:2000; R&D systems, Minneapolis, MN, USA), and mouse monoclonal α-tubulin antibody (1:1000;

Hybridoma Bank, Iowa City, IA, USA). The next day, after washing in TBST containing

4% nonfat dry milk, the membranes were incubated for 1-hour at room temperature in secondary antibodies: horseradish peroxidase-linked anti-rabbit IgG (1:5000; R&D systems), anti-mouse IgG (1:5000; Cell Signaling), and anti-goat IgG (1:5000; Santa

Cruz Biotechnology, Santa Cruz, CA, USA). The signals were detected with ECL plus reagents (GE Healthcare, Biosciences Pittsburgh, PA) and protein blots were visualized by exposing the membranes to X-ray films (GE Healthcare Biosciences).

Statistical Analysis

For comparison of gene expression in various tissues, one-way ANOVA followed by a

Fisher‘s protected least significant difference test was performed using SAS version 9.2

(SAS Institute Inc., Cary, NC). Comparison of other multiple means was conducted by one-way ANOVA followed by Tukey‘s post hoc test. Student‘s t-test was used to compare two means. The minimum level of significance was set as P< 0.05.

3.4. Results

Microarray Analyses show Chrdl1 is highly expressed in adipose tissue

Our comparative analysis of microarray data to identify genes highly expressed in adipose tissue showed that Chrdl1 expression is significantly high in adipose tissue (Fig.

3.1). Adipose tissue-specific expression of Chrdl1 gene was analyzed by statistical

37 analysis of microarray data derived from GDS596 for the human and GDS3142 for the mouse. The Chrdl1 gene ranked 81st in the human with 8.1-fold higher expression and

64th in the mouse with 9.1-fold higher expression in adipose tissue over average of all other tissues (P < 0.001). In addition, among genes identified to be highly expressed in adipose tissue, genes encoding secretory proteins were further selected and Chrdl1 was identified a gene that potentially produces a secretory protein.

Real-time PCR confirmed that Chrdl1 is a novel adipose-specific gene

The abundance of Chrdl1 was examined in various mouse tissues, and the results of real- time PCR and Western blot analysis showed that mRNA and protein expression of

Chrdl1 is significantly high in white adipose tissue (P < 0.001) (Fig. 3.2A). Chrdl1 mRNA expression in other tissues including liver, brown adipose tissue, muscle, heart, lung, spleen, and kidney was significantly low compared to the expression in white adipose tissue. Notably, Chrdl1 mRNA expression was significantly higher in epididymal white adipose tissue compared to inguinal white adipose tissue and brown adipose tissue

(P < 0.05) (Fig. 3.2C). Expression of mouse Ucp1, the brown fat-specific gene, was significantly higher in brown adipose tissue compared to epididymal and inguinal white adipose tissue (P < 0.05), indicating that the sample separation was successful (Fig. .2C).

In addition to the highest Chrdl1 mRNA expression in epididymal visceral adipose tissue, its secretion to media, after transfection with the pQCXIP retroviral expression vector containing HA-tagged long-form or HA-tagged short-form of Chrdl1 cDNA, has been detected by Western blot analysis using the HA tag antibodies (data not shown), suggesting that mouse Chrdl1 encodes a potential visceral fat-specific secretory protein.

Mouse Chrdl1 shows two alternatively spliced forms

RT-PCR shows that mouse Chrdl1 has two alternatively spliced forms (i.e. long-form and short-form), whereas isoforms of human Chrdl1 were not detected (Fig. 3.2B). Human

Chrdl1 nucleotide search in the NCBI webpage (http://www.ncbi.nlm.nih.gov/nuccore) showed that four transcript variants of human Chrdl1 have been deposited. Further

BLAST alignment (http://blast.ncbi.nlm.nih.gov/Blast.cgi) revealed that variant 2 displays a three nucleotide deletion and variant 3 shows a six nucleotide deletion in open reading frame (ORF) when compared to variant 1. These deletions do not cause a frame- shift, and they are too small to be detected by RT-PCR. Investigation of expressed sequence tag (EST) via BLAT search in the UCSC Genome Bioinformatics

(http://genome.ucsc.edu/cgi-bin/hgBlat?command=start) showed that variant 4 has deletion of exon 5 and 6 although the frequency of variant 4 was low among 110 EST sequences deposited in the BLAT (Figure 3.5). However, results from RT-PCR only demonstrated expression of the human Chrdl1 variant 1 (Fig. 3.2B). On the other hand, mouse Chrdl1 nucleotide search in the NCBI webpage showed that two transcript variants (long-form and short-form) have been reported, and this expected alternative splicing pattern was detected by RT-PCR (Fig. 3.2B). The short-form of mouse Chrdl1 has a shortened C-terminus, but alternative splicing does not affect cysteine-rich (CR) domains and three CR domains are intact in the short-form (Figure 3.5). Because CR1 and CR3 have shown to be the functional domains for BMP binding (Nakayama et al.,

2001), it suggests that mouse Chrdl1 protein may show a similar BMP binding activity as chordin protein.

Mouse Chrdl1 is highly expressed in fat cells and shows an increased expression pattern during development in vitro and in vivo

Fat cells and SV cells were fractionated from white adipose tissue, and fractionation was verified by high expression of a preadipocyte marker in SV cells (DLK1) and high expression of adipocyte markers in fat cells (FABP4, SCD1, and PPARγ) (Fig. 3.3A).

The expression level of Chrdl1 mRNA was nearly 3-fold higher in fat cells compared the level in SV cells (P < 0.05) (Fig. 3.3A). This indicates that Chrdl1 is highly expressed in the mature adipocytes rather than in preadipocytes.

To investigate the expression of Chrdl1 during adipose development in vitro, 3T3-

L1 cells were cultured and protein was harvested at days 0, 2, 4, 6, and 8. As FABP4, an adipocyte marker, shows an increased pattern of expression during adipocyte differentiation, Chrdl1 expression increased in later developmental stages (day 6 and 8)

(Fig. 3.3B). In addition, protein expression of Chrdl1 during 3T3-L1 cell differentiation was evaluated by Western blot analysis. Protein expression of mouse Chrdl1 short-form gradually increased, whereas the long-form was barely detectable (Fig. 3.3C).

Furthermore, in order to investigate the Chrdl1 expression during adipose development in vivo, adipose tissues at neonatal, weaning, and sexual maturation stages were collected from mice. We collected inguinal fat tissue from 2-, 5-, 10-, 24- and 49- day-old mice (n = 4 for each time point) and expressions of adipogenic marker genes were quantified by real time PCR. From days 2 to 49, the adipogenic markers, FAS and

SCD-1, increased and the preadipocyte marker, DLK1, decreased (Fig. 3.3D). The expression levels of Chrdl1 were not changed in adipose tissue of 2- to 10-day-old mice, but increased up to 2-fold at 24 and 49 days (Fig. 3.3D).

Retrovirus-mediated overexpression of Chrdl1 increases adipogenic differentiation of 3T3-L1 preadipocytes

To investigate the effect of Chrdl1 on adipogenic differentiation, 3T3-L1 stable cells infected with pQCXIP empty vector, the same vector with HA-tagged Chrdl1 long-form, and HA-tagged short-form were cultured and differentiated. At day 12 after differentiation, both cultures overexpressing Chrdl1 long- and short-forms displayed much more lipid accumulation in Oil Red O staining and significantly increased number of adipocytes in phase contrast microscopy (Fig. 3.4A). Interestingly, Chrdl1 long-form overexpressing cells showed more lipid accumulation than the short-form overexpressing cells, even though overexpression of Chrdl1 long-form is weaker than that of the short- form (Fig. 3.4B), suggesting that the long-form is more efficient in enhancing adipogenic differentiation. The adipogenic differentiation was further confirmed by FABP4 and

PPARγ protein expression at the differentiation time points (Day 8 and 12) with increased expressions compared to Day 0 and 4 (Fig. 3.4B). Both Chrdl1 long- and short- form overexpressing cells exhibited higher expressions of FABP4 and PPARγ protein than control cells with empty vector. Protein expression of Chrdl1 long- or short-form with HA tags was relatively consistent during the time points (Fig. 3.4B), possibly because of the effect of the strong CMV promoter in pQCXIP retroviral vector.

3.5. Discussion

Identification of candidate adipose-specific genes, which are mainly visceral fat-specific, provides potential targets for obesity treatment. The approach in this study that combines

GEO profiles based on microarray dataset with confirmatory analyses, including real-

41 time PCR and Western blot analysis, verified that Chrdl1 is a visceral white adipose- specific gene. Further sequence analysis for putative signal peptides and Western blot analysis for detecting secretion of Chrdl1 into cell media indicated that Chrdl1 is a novel adipokine that secretes protein. Taken together, this study provides a candidate visceral fat-specific adipokine that can be a potential regulator of lipid metabolism and adipose development.

Developmental regulation of expression of Chrdl1 was studied using 3T3-L1 cells as in vitro models and postnatal mouse tissues from day 2 through day 49 as in vivo models (Fig 3.3). The expression of Chrdl1 was inhibited at early stages but stimulated later during the development in both in vitro and in vivo models. This pattern was similar to pro-adipogenic genes including FABP4, SCD1, and FAS. Changes in other genes involved in lipid metabolism (e.g., glucose uptake and de novo lipogenesis) including glucose transporter 4 (GLUT4), acetyl-CoA carboxylase (ACC), and glycerol kinase and adipocyte differentiation will be important to investigate hypoglycemic and anti-diabectic potential of Chrdl1. Combined with results from overexpression of Chrdl1 in 3T3-L1 cells showing increased lipid accumulation and adipocyte differentiation (Fig 3.4), the late-temporal expression of Chrdl1 suggests that Chrdl1 may be a potential pro- adipogenic factor. In addition, significantly higher Chrdl1 expression was detected in the fat cell fraction containing mature adipocytes than in the stromal vascular fraction, which is mostly composed of preadipocytes (Fig 3.3). Analyses of protein-protein interaction or binding partners of Chrdl1 during the late stages of adipocyte differentiation will help clarify the roles of Chrdl1 in adipocyte development.

Chrdl1 has been known as one of the bone morphogenetic protein (BMP) inhibitors (Nakayama et al., 2001). BMP, a member of the transforming growth factor beta (TGF-β) superfamily, is known to play important roles in the development of various tissues, and the concentration of active BMP is tightly regulated by BMP antagonists, such as the chordin family (Rider & Mulloy, 2010). Interaction of the chordin family with BMPs is mediated through cysteine-rich (CR) domains, thereby preventing BMP from binding to its receptors (Abreu et al., 2002). Chrdl1 protein contains three CR domains and, among them, CR1 and CR3 are critical in binding to

BMP (Nakayama et al., 2001). Chrdl1 mutations located at highly conserved cysteine residues within CR1 and CR3 domains are predicted to abrogate interaction with BMPs.

According to recent studies, overexpression of BMP3b inhibited adipocyte differentiation of 3T3-L1 preadipocytes while its knockdown has led to increased adipocyte differentiation (Hino et al., 2012). Thus, it will be interesting to investigate whether

Chrdl1 interacts with BMP3b through CR domains to antagonize anti-adipogenic effects of BMP3b and promotes adipocyte differentiation.

Interestingly, studies have shown that hypoxia drives expression of Chrdl1 and vascular endothelial growth factor (VEGF) in human retinal pericytes, and Chrdl1 was secreted and bound to BMP4 (Kane et al., 2008). According to this study, Chrdl1 presumably plays a role in promoting angiogenesis in hypoxic conditions through up- regulation of VEGF and inhibition of anti-angiogenic BMP4 in human retinal pericytes.

In adipocytes, VEGF also promotes angiogenesis during adipocyte differentiation

(Hausman & Richardson, 2004). However, it is entirely unknown if Chrdl1 promotes angiogenesis during adipose tissue hypoxia in obesity through regulation of VEGF and

BMP4 similarly to human retinal pericytes. Future studies may need to investigate the function of Chrdl1 in adipocytes during hypoxia.

In this study, mouse Chrdl1 overexpression has led to increased adipogenesis. For the induction of differentiation, a reduced concentration of IBMX was used to show more drastic effects of Chrdl1 on the differentiation. Under this condition that is not permissive for full differentiation, lipid accumulation in control cells was minor, whereas, Chrdl1 overexpressing cells showed much higher lipid accumulation. To demonstrate an opposite effect to Chrdl1 overexpression, gene ablation through clustered, regularly interspaced, palindromic repeats (CRISPR) system can be considered. Previous technologies using zinc finger nucleases (ZFNs) and transcription activator-like effector nucleases (TALENs) had limitations such as high off-target frequency and demands for designing target-specific nucleases for protein-DNA interactions (Mandal et al., 2014).

The recently developed CRISPR system utilizes simple base-paring of a short guide RNA

(gRNA) to the target DNA followed by genome editing using CRISPR-associated protein

9 (Cas9). However, the CRISPR system has not been tested in 3T3-L1 cells, although this method worked efficiently in a variety of cell types and organisms (Sander & Joung,

2014). The CRISPR system is currently being evaluated in 3T3-L1 cells as a part of this current study.

In conclusion, Chrdl1 was discovered as a novel visceral fat-specific adipokine, and its late-temporal expression during adipocyte differentiation combined with a dramatic increase in lipid accumulation and adipocyte differentiation through overexpression of Chrdl1 suggests critical roles of Chrdl1 as a potential pro-adipogenic factor. Chrdl1 contains cysteine-rich domains that can mediate protein-protein

44 interactions with BMP, thereby antagonizing binding of BMP to its receptors, and it will be critical to identify binding partners of Chrdl1 during adipocyte developmental stages.

Future studies on the roles of Chrdl1 during angiogenesis in adipocytes and Chrdl1 gene ablation through CRISPR/Cas9 system will advance our understandings on the function of Chrdl1 during adipocyte development.

3.6. Acknowledgements

This project was supported by Agriculture and Food Research Initiative Competitive

Grant no. 2010-65206-20716 from the USDA National Institute of Food and Agriculture, and the National Research Foundation of Korea Agenda Program no. PJ009457.

Ovary Placenta Heart Testis Thymus

Pitauary gland *** Adipocyte Bone Marrow Adrenal gland Kidney Lung Brain Liver Muscle

0 1 2 3 4 5 6 7 8 9 Relative Human Chrdl1 Expression Value

Ovary Placenta Heart Testis Thymus

Pitauary gland *** Adipose tissue Bone Marrow Adrenal gland Kidney Lung Brain Liver Muscle

0 1 2 3 4 5 6 Relative Mouse Chrdl1 Expression Value

Figure 3.1. Relative Chrdl1 mRNA expression in various tissues according to GEO DataSets for the human (GDS596) and mouse (GDS3142)

A 4 D,*** *** 3 2

1 a Chrdl1 Mouse / Cyc 0

Chrdl1 long- form 1364 bp Chrdl1 short- form 1028 bp

Total Chrdl1

Mouse Cyc

Human Chrdl1

Human Cyc

1.2 1200 a b 1 1000

0.8 800 Chrdl1 / / Chrdl1Cyc - 0.6 600 b 0.4 b 400 0.2 Mouse Ucp1 Mouse Cyc / 200 Mouse T Mouse a a 0 0

Figure 3.2. A) Chrdl1 expression in various mouse tissues measured by real-time PCR and Western blot analysis with coomassie staining for a loading control. B) Mouse Chrdl1 long- and short-forms and Human Chrdl1 full-length detected by PCR. C) Mouse Chrdl1and Ucp1 expression in white (epididymal and inguinal) adipose tissue and brown adipose tissue.

A Chrdl1 / Cyc Dlk1 / Cyc Fabp4 / Cyc Scd1 / Cyc Pparg / Cyc 4 1.5 15 150 * 15 * * * 3 1 10 100 10 2 0.5 5 50 5 1 * 0 0 0 0 0 SV FC SV FC SV FC SV FC SV FC

B Chrdl1 / Cyc Fabp4 / Cyc C 5 2000 Day: 0 2 4 6 8 S L EV b d mChrdl1 L 4 b 1500 * 3 c mChrdl1 S 1000 2 a a a b 1 500 a a 0 0 0 2 4 6 8 0 2 4 6 8

D Chrdl1 / Cyc Dlk1 / Cyc Scd1 / Cyc Fas / Cyc 2.5 2000 1500 300 b b b 2 a b 1500 1000 200 1.5 a a 1000 1 a b 500 100 500 b 0.5 a a b b a a a a a a 0 0 0 0 2 5 10 24 49 2 5 10 24 49 2 5 10 24 49 2 5 10 24 49 Age (day) Age (day) Age (day) Age (day)

Figure 3.3. A) Mouse Chrdl1 mRNA expression in stromal vascular cells (SV) and fat cells (FC) B) Chrdl1 mRNA expression during 3T3-L1 adipocyte differentiation C) Chrdl1 protein expression during 3T3-L1 adipocyte differentiation. S: Short-form positive control, L: Long-form positive control, EV: Empty vector negative control, * : Non-specific band D) In vivo mouse Chrdl1 mRNA expression in during development. Dlk1 is a preadipocyte marker. Fabp4, Scd1, PPARγ, and FAS were used as adipocyte markers. Cyc was used as a loading control.

A EV L-Chrdl1 S-Chrdl1

ORO

Phase contrast

B L-Chrdl1-HA S-Chrdl1-HA

PPAR

FABP4

α-tubulin

Day 0 4 8 12 0 4 8 12 0 4 8 12 EV L S

Figure 3.4. A) Oil Red O staining and phase contrast microscopy of empty vector (EV) and Chrdl1 overexpressing 3T3-L1 cells B) Western blots of Chrdl1, adipogenic marker (FABP4 and PPARγ), and housekeeping protein (α-tubulin) during differentiation of EV and Chrdl1 overexpressing 3T3-L1adipocyte.

A ATG TAG

hChrdl1 Variant1 – 12 exons

ATG TAG

hChrdl1 Variant4 – Exon 5 & 6 deletion

B ATG TAG

mChrdl1 Long – 12 exons

ATG TGA

mChrdl1 Short – 9 exons

C TAG SP CR1 CR2 CR3

TGA SP CR1 CR2 CR3

Figure 3.5. Alternative splicing patterns of A) Human Chrdl1 and B) Mouse Chrdl1 according to the NCBI webpage (http://www.ncbi.nlm.nih.gov/nuccore) and UCSC Genome Bioinformatics (http://genome.ucsc.edu/cgi-bin/hgBlat?Command= start). C) Protein domains of mouse Chrdl1. SP: signal peptide, CR: cistein-rich. Arrow indicates a cleavage site.

Chapter 4: Porcine G0/G1 Switch Gene 2 (G0S2) expression is regulated during

adipogenesis and short-term in vivo nutritional interventions

Jinsoo Ahn1, 2, Shin-Ae Oh1, Yeunsu Suh1, Steven J. Moeller1 and Kichoon Lee1, 2, *

1Department of Animal Sciences and 2 Interdisciplinary Ph.D. Program in Nutrition, The

Ohio State University, Columbus, OH, 43210, USA

4.1. Abstract

Adipose triglyceride lipase (ATGL), catalyzing the initial step of hydrolysis of triacylglycerol (TAG) in adipocytes, has been known to be inhibited by G0/G1 switch gene 2 (G0S2). In this study, we report the porcine G0S2 cDNA and amino acid sequences as well as the expression level of porcine G0S2. The porcine G0S2 mRNA was abundantly expressed in adipose tissue and liver among various tissues. In adipose tissue, porcine G0S2 expression was 16-fold higher in the fat cell fraction than the stromal vascular fraction. The G0S2 level increased significantly during adipogenesis in vitro and in vivo. These data indicate that G0S2 expression is tightly associated with lipid accumulation and adipogenesis. Considering G0S2 as an inhibitor of cell proliferation, the relatively low levels of G0S2 in preadipocytes and adipose tissue of fetal and neonatal pigs compared to adipocytes and adult pigs may allow the fast cell proliferation rates.

Further studies showed that a short-term 24 h fast down-regulated G0S2 expression and

51 increased ATGL expression in adipose tissue; however, a long-term calorie restriction for

8 days had no influence on the level of G0S2 but increased ATGL expression. Therefore, porcine G0S2, which is both a negative regulator of ATGL-mediated lipolysis and cell proliferation in adipose tissue, can be down-regulated in vivo by a short-term 24 h fast followed by increased ATGL-mediated lipolysis.

Keywords: adipose triglyceride lipase (ATGL), lipolysis, triacylglycerol (TAG), G0/G1 switch gene 2 (G0S2), porcine, adipocyte differentiation, adipose tissue, fasting, calorie restriction

4.2. Introduction

Dietary fatty acids are mainly accumulated in lipid droplets in the cytoplasm of adipocytes in mammals as uncharged, neutral fat called triacylglycerol (TAG). TAG, a concentrated energy storage molecule, is hydrolyzed through lipolysis and mobilized into nonesterified fatty acids (NEFA) and glycerol which enter the circulation upon signals such as hormonal changes during fasting, an energy deficit after exercise, and intake of nutrients such as calcium (Duncan et al., 2007, Newsom et al., 2010). Lipolysis consists of sequential catabolic enzymatic reactions by three lipolytic enzymes known as adipose triglyceride lipase (ATGL), hormone-sensitive lipase (HSL), and monoacylglycerol

(MAG) lipase. ATGL hydrolyzes TAG 10-fold more than diacylglycerol (DAG). HSL is known to hydrolyze DAG more frequently than TAG (Zimmermannet al., 2004). HSL knock-out mice showed a normal body weight phenotype (Wang et al., 2001), which is probably because ATGL could hydrolyze TAG to reduce excess fat. HSL is mostly involved in the breakdown of DAG, followed by hydrolysis of monoacylglycerol (MAG) by MAG lipase resulting in the generation of free fatty acids (FFA). HSL-mediated lipolysis is activated through beta-adrenergic signaling and subsequent phosphorylation of both HSL and perilipin A, the surface protein of lipid droplets (Yang et al., 2010b).

Interruption of the process of lipolysis can lead to increased TAG content in lipid droplets.

Fat catabolism mediated by ATGL and its regulators is not fully understood. A recent in vitro study has shown that HeLa cells overexpressing murine ATGL had approximately 4-fold less TAG hydrolase activity in the presence of murine G0/G1 switch gene 2 (G0S2) (Yang et al., 2010b). The addition of extracts containing murine comparative gene identification-58 (CGI-58) into cell extracts of COS-7 cells expressing

53 murine ATGL, increased the hydrolase activity of ATGL approximately 20-fold (Lass et al., 2006). Another recent in vitro study demonstrated that murine G0S2 and CGI-58 can physically interact with a minimal active domain of murine ATGL up to Leu254 (patatin related region) which possesses TAG hydrolase activity (Cornaciu et al., 2011). These studies are based on in vitro TAG hydrolase activity assays and protein-protein interaction analyses. To examine the role of G0S2 in living organisms, our current study reports in vivo G0S2 expression in pigs under various developmental and nutritional conditions.

G0S2 was originally identified by its transient induction in lymphocytes during a lectin-induced re-entry into G1 phase from G0 phase favoring cell proliferation (Russell &

Forsdyke, 1991). A later study showed that murine G0S2 is predominantly expressed in adipose tissue and direct up-regulation of G0S2 by PPAR (peroxisome proliferator- activated receptor gamma) is associated with cell-cycle withdrawal in mouse 3T3L1 cells

(Zandbergen et al., 2005). Growth arrest during cell-cycle withdrawal leads to adipogenic differentiation of preadipocytes. More recently, it has been suggested that G0S2 inhibits hematopoietic stem cell (HSC) proliferation and promotes stem cell quiescence (G0 phase)

(Yamada et al., 2012). In addition, overexpression of G0S2 further increased quiescence in hematopoietic stem and progenitor cells (Yamada et al., 2012). Although it is generally accepted that growth arrest during cell-cycle withdrawal is required to initiate adipogenic differentiation of preadipocytes, the regulation of G0S2 expression in regard to cell cycle and differentiation has not been fully investigated in adipose tissue in vivo.

G0S2 is also involved in lipid metabolism as an inhibitor of ATGL-mediated lipolysis as described above. Our previous in vivo studies of the chicken and quail

54 revealed that while 24 h fasting increased ATGL expression in adipose tissue, G0S2 level in the tissue was decreased (Oh et al., 2011). As a result, less G0S2 expression could reduce the inhibitory effect of G0S2, increasing ATGL-mediated lipolysis during fasting.

A recent human study also showed that 72 h fasting significantly decreased G0S2 expression (Nielsen et al., 2011). However, to our knowledge, there have not been any in vivo studies regarding differences in the effects of short-term fasting and long-term calorie restriction on G0S2 expression.

In this study, porcine G0S2 expressions were examined in various pig tissues and under several conditions including in vitro adipogenic differentiation and in vivo adipose tissue development. Moreover, the effects of 24 h of short-term fasting and 8 days of long-term calorie restriction on porcine G0S2 levels were also investigated.

4.3. Materials and Methods

Experimental Animals

The Institutional Animal Care and Use Committee (IACUC) at the Ohio State University approved all animal care and use procedures. For cloning and sequencing of the G0S2 genes in adult Berkshire, Landrace, and Duroc pigs (120 days), adipose tissues were sampled and snap-frozen in dry ice before storing at −80˚C for total RNA isolation, reverse transcription polymerase chain reaction (RT-PCR), and gel extraction. To examine the tissue distribution of G0S2 gene expression, adipose tissue, heart, muscle, spleen, lung, liver, and kidney were collected from four adult Landrace pigs (120 days) and snap frozen in dry ice and kept at −80˚C for total RNA isolation and quantitative real-time PCR.

For the gene expression study during in vivo adipose development, adipose tissue samples from 105 days fetal, 6 days post-natal and 120 days adult Landrace pigs (three pigs per group) were collected, snap-frozen, and stored at −80˚C for total RNA isolation, reverse transcription (RT) and real-time PCR.

In order to obtain subcutaneous fat tissue samples from pigs after 24 h fasting as well as six hours after refeeding, biopsies were conducted at the OARDC Western

Agricultural Research Station, South Charleston, OH. For the adipose tissue biopsy, seven live Berskshire pigs (105 days old, approximately 100 kg) were assigned to each of three groups including the control, 24 h fasting, and 24 h fasting plus 6 h refeeding groups. Biopsy sites in the middle of the back of the pigs were washed with soap and warm water, scrubbed with Betadine solution for disinfection, and rinsed with 70% alcohol. Two milliliters of Lidocaine were applied for local anesthesia of the skin of the biopsy site. Each pig was biopsied once. The biopsy tissue samples were separated into skin, adipose tissue and muscle, and then immediately frozen and stored at -80°C for total

RNA isolation. Calorie restriction treatment was performed at Russell Agricultural

Research Center, USDA-ARS, Athens, GA as described in previous reports (Hart et al.,

2007; Deiuliis et al., 2008).

Cloning and sequencing of Berkshire, Landrace and Duroc pigs

Total RNA was isolated from adipose tissue of adult Berkshire, Duroc and Landrace pigs

(120 days) weighing between 80-100 kg using Trizol Reagent (Invitrogen, Carlsbad, CA,

USA) according to the manufacturer‘s instructions. Total RNA was checked by electrophoresis of 1 µl of RNA on 1.0% agarose gels stained with ethidium bromide.

56 cDNA was synthesized using 1 µg of total RNA, oligo dT and the Moloney murine leukemia virus reverse transcriptase (Invitrogen) in a 20 μl reaction volume. RT conditions for each cDNA synthesis were 65°C for 5 min, 37°C for 52 min, and 70°C for

15 min. The newly synthesized cDNAs were used as a template to amplify the entire coding sequence of porcine G0S2 gene by PCR using the forward (―F1‖ 5‘-

GCGGGCCTTCAGGTCA-3‘) and reverse (―R1‖ 5‘-CAGAAGCAAGTCCAAACGGA-

3‘) gene-specific primers which were designed based on the CX065684.1 EST sequence.

Conditions for the PCR were 2 min at 95°C, followed by 40 cycles of denaturation at

94°C for 30 sec, annealing at 55°C for 40 sec, and extension at 72°C for 1 min, and a final extension of 10 min at 72°C. AmpliTaq Gold polymerase (Applied Biosystems,

Foster City, CA) was used for the PCR reactions in a MJ Research PTC-200 thermal cycler (MJ Research Inc., South San Francisco, CA). The PCR products were separated by electrophoresis on a 1% agarose gel. Bands including the expected size (309 bp) of the entire coding sequence of porcine G0S2 (pG0S2) cDNA were excised, and the DNA in the gel was extracted using a Qiagen Gel Extraction Kit (Qiagen, Alameda City, CA).

The extracted DNA was ligated to the PCR 2.1 vector using the TOPO TA Cloning kit

(Invitrogen) according to the manufacturer‘s instructions. TOP 10 chemically competent cells (Invitrogen) were transformed with the PCR 2.1 plasmid including the pG0S2 insert.

Transformed cells were grown on a kanamycin agar plate containing X-gal. Isolated plasmids from positive colonies were sequenced by The Ohio State University

Sequencing Core Facility using an Applied Biosystems 3730 DNA analyzer (Foster City,

CA).

Primary Stromal - Vascular Cell Culture and Hormone Treatments

For the isolation of stromal-vascular (SV) cell fraction, which consists mostly of preadipocytes, subcutaneous adipose tissue (5 g) was collected from Landrace pigs (120 day old) slaughtered in the Department of Animal Sciences at The Ohio State University.

Excised adipose tissue was immediately transported to the laboratory in pre-warmed PBS.

Primary adipocytes were cultured using modifications of a previously described method

(Hausman & Poulos, 2004). In detail, razor blades were used to mince subcutaneous adipose tissues until fine pieces were obtained and then, the pieces were incubated with

3.2 mg/mL collagenase II (Sigma-Aldrich, St. Louis, MO) for 40 min to 1 hr in a shaking water bath (120 oscillations /min, 37°C) to separate individual cells. The digested material was passed through a 100µm pore size nylon mesh to eliminate large cell masses, and the floating adipocytes (fat cells, FC) were separated from the pellet of SV cells after the filtered cells were centrifuged for 10 min at 500 × g. Next, the SV cell pellet containing most of the preadipocytes was diluted in DMEM medium containing 10% fetal bovine serum (FBS) and the combination of the antibiotics penicillin and streptomycin (Invitrogen). The diluted cells were seeded on 12-well plates at a density of

3 × 106 cells/ml in the culture medium. Cells in the medium were maintained at 37°C in an incubator with 5% CO2 until confluence was reached. In order to induce adipocyte differentiation, cells were treated with dexamethasone (80 nM) for 2 days after confluence followed by addition of insulin (5 µg/ml), transferrin (5 µg/ml), isobutyl-1- methylxanthine (0.5 mM) and selenium (5 ng/ml) for the next 4 days. These cells were cultured in the differentiation media for 10 days and media was changed every 2 days.

Total RNA was extracted from the cultured cells at the induction of differentiation (Day 0)

58 and Day 3, 6, and 9 after differentiation for RT-PCR. To avoid RNA degradation, the

RNA samples were immediately stored at -80 °C.

Quantitative Real-Time PCR Measurement of Gene Expression

Quantitative real-time PCR (qPCR) was conducted to assess relative gene expression using AmpliTaq Gold polymerase (Applied Biosystems) and SYBR green I as a detection dye for a dissociation curve on the Stratagene MX3000P cycler (Agilent Technologies,

Inc.). To perform qPCR for porcine G0S2, forward (―F2‖ 5‘-AGAGCCCGGAGCCGAG-

3‘) and reverse (―R3‖ 5‘-CTGCACACGGTCTCCATCA-3‘) primers were designed.

Primers for qPCR for porcine delta-like 1 homolog (DLK1), PPAR , ATGL and cyclophilin (cyc) are described in our previous reports (Deiuliis et al., 2008; Li et al.,

2007a). Thermal profile was as follows: 95°C for 10 min, 40 cycles of 94°C for 30 s,

60°C for 60 s, and 82°C for 30 s. The relative gene expression level of target genes was calculated using the CT values for the internal control and target genes, as measured by the ABI software, and the method (Livak & Schmittgen, 2001). Cyc was used for normalization of mRNA level of each target gene.

Bioinformatics, Sequence Alignment, and Statistical Analysis

Sequence alignments for the G0S2 cDNAs and the deduced amino acid sequences were performed by using the ClustalX multiple alignment software (Thompson et al., 1994) in

BioEdit Sequence Alignment Editor (Version 7.0.9.0; http://www.mbio.ncsu.edu/BioEdit

/bioedit.html). Homology comparisons were performed using the Align Sequences

Nucleotide BLAST in the NCBI resources (http://www.ncbi.nlm.nih.gov/pubmed). 59

Statistical analyses were conducted by using statistical software programs as follows:

Comparison of two means was performed by a Student‘s t-test. One-way ANOVA built in SAS software program (SAS Institute, Inc.) was used to compare multiple means followed by Tukey‘s post hoc test. P-value was < 0.05 for the minimum level of significance.

4.4. Results

Nucleotide Sequence Comparison of G0S2 cDNAs

The sequencing analysis revealed that the entire coding sequence of G0S2 cDNA of the three pig breeds (Berkshire, Landrace, and Duroc) is 309 bp in length. There were no insertions or deletions, and G0S2 cDNA was highly conserved among the three pig breeds. In particular, the nucleotide sequences of Berkshire and Landrace show 100% homology, and both Berkshire and Landrace nucleotide sequences have 99% homology with Duroc nucleotide sequence. This difference in homology is due to two nucleotide differences at nucleotide position 265 and 266; nucleotide GT for Berkshire and Landrace, and AC for Duroc (Figure 4.7). The sequences for Berkshire, Landrace and Duroc were deposited in the GenBank database ( http://www.ncbi.nlm . n i h . go v / genbank) with accession numbers JQ_013998, JQ_013999, and JQ_014000, respectively.

The multiple nucleotide sequence comparison between species also showed that both Berkshire and Landrace G0S2 have 83% and 76% homology and Duroc G0S2 has

84% and 76% homology, to that of the human (GenBank NM_015714) and mouse

(GenBank NM_008059), respectively. Human and murine G0S2 cDNAs have 81%

60 homology. Therefore, G0S2 nucleotide sequences are highly conserved among those mammalian species.

Analysis of G0S2 Amino Acid Sequences and Protein Domains

Computer-generated putative G0S2 amino acid (AA) sequences were determined from the nucleotide sequences, and have open reading frames of 103 AA for Berkshire,

Landrace and Duroc swine (Figure 4.1). The AA sequences of Berkshire and Landrace are 100% homologous. Compared with Berkshire and Landrace, the AA sequence for

Duroc shows a 99% homology due to a single AA substitution of threonine (Thr) for valine (Val) in Berkshire and Landrace pigs at residue 89. Protein sequence comparison also showed that the AA sequences of both Berkshire and Landrace G0S2 have 66% and

61% similarity and Duroc G0S2 has 67% and 61% similarity, to those of the human

(GenBank NP_056529) and mouse (GenBank NP_032085), respectively. In addition, there is 78% homology between human and murine G0S2 proteins.

The putative hydrophobic domain of murine G0S2 protein is located in the central region (residues 27 to 42) that binds to the patatin domain in ATGL possessing TAG hydrolase activity (Wang et al., 2001). Human G0S2 has a similar hydrophobic domain between residue 27 and 42. The porcine G0S2 protein also possesses a hydrophobic domain between residue 27 and 42 and this domain is conserved when compared with the protein sequences of the human and mouse. However, within the hydrophobic domain, there is variation of six amino acids in the pig compared to the human and mouse (Leu at residue 28, Gly at residue 31, Leu at 32 residue, Phe at 35 residue, Leu at 36 residue and

Ala at 38 residue) as indicated in Figure 4.1. Most of those variations are substitutions

61 between amino acids with similar properties. Thus, G0S2 protein sequences are similar among those mammalian species and the hydrophobic domain is relatively conserved in the porcine breeds.

Gene Expression in Various Tissues

Tissue distributions of relative expression of G0S2 mRNA for the pig by using quantitative real-time PCR revealed that expression of porcine G0S2 was significantly high in adipose tissue and, to a lesser degree, in liver, compared to other tissues (Figure

4.2). G0S2 expressions in other tissues such as the heart, muscle, spleen, lung, and kidney were relatively low. A housekeeping gene (cyc) was used as a normalization control gene

(Figure 4.2).

Relative Gene Expressions in Fat cell (FC) and Stromal Vascular (SV) Fraction

To examine porcine G0S2 expression in adipose tissue, fat cell (FC) fraction and stromal vascular (SV) fraction of adipose tissue were separated by centrifugation. Unlike the floating FC fraction which contains adipocytes, the SV fraction can be obtained from pellets at the bottom of a test tube and contains mostly preadipocytes. First, cell fractionation was verified by using adipocyte and preadipocyte markers. DLK1, a preadipocyte marker, was expressed 15 times higher in SV fraction than in FC fraction (P

< 0.05), and ATGL, an adipocyte-specific gene was 11 times higher in FC fraction than in SV fraction (P < 0.01), indicating that fractionation was effective (Figure 4.3A and

3B). Next, G0S2 expression was approximately 16-fold greater in FC fraction compared

62 to SV fraction (P < 0.05) (Figure 4.3C), showing that pG0S2 is expressed higher in mature fat cells rather than preadipocytes.

Gene Expression Patterns during Adipogenic Differentiation and Adipose

Development

In order to determine if adipogenic differentiation of preadipocytes causes an increase in pG0S2 expression, porcine primary preadipocytes were cultured for differentiation. The level of DLK1 expression was significantly high at day 0 (the day for initiation of differentiation) (P < 0.05), but decreased from day 3 onwards (Figure 4.4A). PPAR was low at day 0 (the day for initiation of differentiation) and 3, but continued to increase at day 6 and 9 concomitantly with differentiation (Figure 4.4B). ATGL significantly increased at day 9 (P < 0.05) (Figure 4.4C). G0S2 expression was significantly higher at day 9 compared to day 0 and 3 (P < 0.05) (Figure 4.4D). In addition, during in vivo adipose development, DLK1 expression was significantly higher at the embryonic stage

(Embryonic day 105) (P < 0.05), but decreased from the neonatal stage (6 day) (Figure

4.5A). In contrast, PPAR , ATGL, and G0S2 levels increased significantly at adult stages

(120 days) (P < 0.05) as adipogenesis proceeded (Figure 4.5B, C, and D). Overall, pG0S2 expression increased as both in vitro adipocyte differentiation and in vivo adipose tissue development continued.

Effect of Nutritional Treatment on Gene Expression Levels

Nutritional regulation of G0S2 expression in adipose tissue was studied with a short-term fasting experiment consisting of a 24 h fasting group and a 6 h refeeding group, and a 63 long-term calorie restriction group with 8 days of food restriction. Fasting for 24 h increased the ATGL expression in adipose tissue by two-fold compared to control pigs with ad libitum feeding, and ATGL was brought down to the basal level six hours after refeeding (Figure 4.6A). Conversely, fasting decreased G0S2 expression level (P < 0.05), but the refeeding group tends to have an intermediate G0S2 level between that of the control and fasting group (Figure 4.6B). In addition, a control and a calorie restricted diet for 8 days prior to tissue collection were used for the long-term food restriction experiment as described in previous reports (Hart et al., 2007; Deiuliis et al., 2008). The control group was fed 3 kg/days on an ad libitum basis of a corn-soybean meal ration (14% crude protein) supplemented with vitamins and minerals, following the guidelines from

National Research Council (NRC, 1998). The calorie restricted group received 1kg/days of the same ration. One half of the rations were provided to each group twice daily at

0830 and 1630. The mean body weight (BW) of all twelve pigs (164 d of age) was 79 ± 4 kg initially. Six of these pigs were fed the control diet and the other six were fed the calorie restricted diet. By day 8, the control group gained 8.2 ± 0.6 kg of BW and 2 ± 1 mm of subcutaneous backfat (BF), but the calorie restricted group lost 2.5 ± 0.6 kg of

BW and 3 ± 0.6 mm, on average. Hourly blood samples on day 8 showed serum NEFA concentrations were significantly increased (P < 0.05) in the calorie restricted group at

0800, 0900, 1500, and 1600 samples. Both ATGL and DLK1 in the peri-renal fat depots were increased significantly after calorie restricted feeding (P < 0.01) (Figure 4.6C and

D). However, G0S2 expression in the peri-renal fat depots of pigs was not affected by the long-term calorie restriction (Figure 4.6E). Therefore, porcine G0S2 level was affected by short-term 24 h fasting, not by 8 day, long-term food restriction.

4.5. Discussion

According to our sequencing results, porcine, human, and murine G0S2 have the same length of nucleotide and AA sequences without any deletion or insertion. Moreover, porcine G0S2 nucleotide and AA sequences show high sequence homologies with human

G0S2 and, to a lesser degree, with murine G0S2. Especially, hydrophobic domains of porcine G0S2 show conserved AA sequences that are homologous to hydrophobic domains of the human, mouse and avian G0S2 (Oh et al., 2011). However, G0S2 does not have homologs in lower organisms such as Drosophila and Caenorhabditis elegans

(Zandbergen et al., 2005), although both having ATGL homologs, Brummer lipase

(Gronke et al., 2005) and ATGL-1 (Narbonne & Roy, 2009), respectively; suggesting

G0S2 in higher organisms may provide more complexity in regulation of ATGL mediated lipolysis.

Porcine G0S2 showed a comparatively conserved central hydrophobic domain in residues 27 to 42 with six amino acid substitutions compared to human and murine G0S2.

Among those substitutions, Val/Leu (at residue 28 and 32), Leu/Phe (at residue 35 and

36), or Val/Ala (at residue 38) substitutions are replacements between very hydrophobic amino acids, and Ser/Gly substitution at residue 31 is a switch between non- or less hydrophobic amino acids. Thus, it is plausible that the degree of hydrophobicity may be comparatively equal. Including the hydrophobic region, the first half of the porcine G0S2 shows a high level of similarity with human and murine G0S2, indicating that the first half may have important functional roles. In contrast, the second half has multiple sequence variations as mentioned above (Figure 4.1). A study showed that deletion in the area containing the hydrophobic region of murine G0S2 disrupts the ability of G0S2 to

65 interact with murine ATGL patatin-like domain, and conversely deletion of C-terminal region (residues 74-103) maintained the interaction with murine ATGL (Wang et al.,

2001). In addition, residues from 43 to 66 next to the hydrophobic domain are also highly conserved with the only a few exceptions. Residues 54 and 59 show only two variations in human and pigs, and residue 43, 54, and 64 displays three differences in mice and pigs, but the substitutions at residue 43 and 64 were replacements between hydrophobic amino acids; a substitution of methionine (Met) for valine (Val) and alanine (Ala) for Val. This suggests that the conserved residues from 43 to 66 may have an unknown functional role.

In addition, the area from residue 39 to 53 is the most conserved region; humans and pigs have an identical sequence, and only one variation is present between mice and pigs which occurs at residue 43 and was a replacement between hydrophobic amino acids as mentioned above. Thus, it would be interesting to investigate if residues from 27 to 66, including the previously identified hydrophobic domain (residue 27 - 42), may be functionally critical for the G0S2 activity.

Our sequence analysis also showed that there is a single nucleotide polymorphism

(SNP) in the porcine G0S2 nucleotide sequences at nucleotide 265 and 266; Duroc G0S2 nucleotides have AC, but Berkshire and Landrace G0S2 nucleotides have GT. The SNP results in changes in the G0S2 amino acid sequences at residue 89; Duroc has hydrophilic threonine (Thr), but Berkshire and Landrace have hydrophobic valine (Val).

Identification of functional DNA variants in pigs is important for marker-assisted breeding (Grisart et al., 2002); however, the SNP may not contribute to the alteration in the activity of G0S2, because the AA region from residue 75 to 93 including the corresponding residue 89 is highly variable compared to human and murine G0S2 (Figure

4.1), suggesting that this region may not be crucial for the function of G0S2. Also a previous study showed that the deletion of the C-terminal region of murine G0S2 from residue 74 to 103 did not affect G0S2 activity (Yang et al., 2010b). Therefore, it is not likely that the SNP affect G0S2 activity although future studies may need to confirm this.

Analysis of tissue distribution of porcine G0S2 mRNA by quantitative real-time

PCR showed that porcine G0S2 is highly expressed in adipose tissue and, to a lesser degree, in liver. Similar gene expression patterns are also observed in the mouse (Yang et al., 2010b) and the human G0S2 (GDS596 and GDS1096 at http://www.ncbi.nlm. nih.gov/geoprofiles). Recently, ATGL has been reported as a major hepatic lipase in mice

(Ong et al., 2011), thus G0S2 in mammalian liver may decrease ATGL-mediated lipolysis in the liver. Similarly, G0S2 in adipose tissue of mice decreases TAG hydrolase activity of ATGL in adipose tissue (Yang et al., 2010b). Abundant expression of G0S2 in the porcine adipose and liver suggests that G0S2 is mainly expressed in the tissues containing high amounts of TAG in order to regulate ATGL activity.

Adipose tissue fractionation showed that G0S2 is expressed at low levels in SV fraction compared to FC fraction. G0S2 expression was also low at day 0 and 3 during in vitro adipocyte differentiation and at embryo and neonatal stages during in vivo adipogenesis. G0S2 expressions were significantly high in the FC fraction (P < 0.05) and in later stages during in vitro adipocyte differentiation (Day 9) and in vivo adipose development (120 days adult) (Figure 4.3C, 4D and 5D). Considering the involvement of

G0S2 as a negative regulator in ATGL-mediated lipolysis, the differences in relative amounts of G0S2 expression could reflect the amount of TAG substrate in lipid droplets for ATGL in preadipocytes and adipocytes. Cells in both the SV fraction and early stages

67 during in vitro adipocyte differentiation (Day 0 and 3) and in vivo adipose development

(105 days fetal and 6 days post-natal) containing almost no or very small lipid droplets may not require high expression of ATGL and its regulators, G0S2 and CGI-58 (Li et al.,

2012), whereas, adipocytes filled with TAG inside of lipid droplets may require higher

ATGL and G0S2 expression.

Adipocyte precursor cells undergo preadipocyte proliferation and adipogenic differentiation, and only cells at a quiescent growth arrest state are capable of differentiating into adipocytes (Filipak et al., 1989; Gregoire, 2001). Given the inhibitory role of G0S2 on cell division and proliferation (Yamada et al., 2012), the low amount of

G0S2 in SV cells and early stages of in vitro and in vivo adipocyte differentiation may allow preadipocyte proliferation; whereas, adipocytes in the FC fraction and later stages during in vitro and in vivo adipocyte differentiation may need to have a high amount of

G0S2 in order to inhibit cell proliferation and maintain the quiescent stage that is required for initiating and/or maintaining adipogenic differentiation. Taken all together, these correlative G0S2 expressions with stages of adipocyte differentiation led us to propose the possibility that G0S2 could play dual roles in the mature adipocytes: the aforementioned maintenance of the quiescent stage, and negative regulation of ATGL- mediated lipolysis as an inhibitor of ATGL in mature adipocytes.

The G0S2 gene was originally identified as an up-regulated gene in response to

PPAR in vitro (Zandbergen et al., 2005), although it has not been clearly demonstrated whether PPAR directly binds to the promoter of the G0S2 to activate transcription of the

G0S2. The gradual increase in G0S2 expression during porcine adipogenesis in vitro and in vivo (Figure 4.4D and 5D) was correlated with increased PPAR expression (Figure 68

4.4B and 5B). This correlative expression pattern between the porcine PPAR and G0S2 gene further supports that the G0S2 gene is one of the downstream targets of PPAR

(Zandbergen et al., 2005). In addition, ATGL gene expression was also gradually up- regulated during both adipocyte differentiation and adipose tissue development in parallel with PPAR (Figure 4.4C and 5C). This is further supported by the finding that PPAR increased ATGL expression in vitro and in vivo (Kershaw et al., 2007; Liu et al., 2009).

These results indicate that both G0S2 and ATGL are regulated by PPAR . Without considering the activities of ATGL and G0S2 in response to hormonal and nutritional changes, induction of both genes by PPAR at the later stages of adipocyte differentiation could prepare the cells for regulation of lipolysis.

From a nutritional aspect, fasting (24 h) significantly increased ATGL expression level, while reducing the G0S2 level. Fasting has also shown the similar pattern in human and avian species, significantly enhancing ATGL protein level, but decreasing G0S2 expression (Oh et al., 2011; Nielsen et al., 2011). This may allow ATGL to hydrolyze

TAG, generating FFA as an energy source during the fasting period. These altered ATGL level returned to the basal level six hours after refeeding, although G0S2 level was not fully restored. Reasons for the incomplete recovery of G0S2 are not clear, but it is possible that other factors induced by fasting may regulate ATGL together with G0S2 in a short period of time so that full restoration of G0S2 may not be necessary during that time. In contrast, our previous report showed that quail G0S2 was completely restored 2 h after refeeding (Oh et al., 2011), suggesting that regulation of ATGL in quail may depend solely on G0S2. This difference may be explained by a higher metabolic rate of quail compared to pigs, which may enable G0S2 to be recovered promptly after fasting ends. 69

Additionally, long-term calorie restriction (8 days) increased the expression of ATGL.

This may allow breakdown of more fat in order to provide energy from stored peri-renal fat around the kidney. However, the expression level of G0S2 was not changed (Figure

4.6E). This suggests that G0S2 mRNA expression may not be regulated by long-term calorie restriction, whereas ATGL mRNA expression is controlled by long-term calorie restriction. Thus, short-term, 24 h fasting affects pG0S2 expression, but long-term food restriction does not influence the expression of pG0S2.

In conclusion, the sequencing analysis showed that the nucleotide and AA sequences of G0S2 of the mammalian species including the three breeds of pigs are conserved especially in the central hydrophobic domain. Our in vitro and in vivo developmental studies indicate that porcine G0S2 is adipose-specific and potentially related to the inhibition of preadipocyte proliferation promoting adipocyte differentiation.

In addition, the short-term 24 h fasting regulated G0S2, but G0S2 may not be controlled by long-term (2-week) calorie restriction. These findings give new insight for both the role of porcine G0S2 in adipose tissue during adipocyte differentiation and the regulations of G0S2 in vivo by nutritional treatments.

4.6. Acknowledgements

This work was supported by the National Research Foundation of Korea Grant funded by the Korean government (KRF-2009-220-F00006) and the Ohio Agricultural Research and Development Center Director‘s Associateship Program.

10 20 30 40 50 60 ....|....|....|....|....|....|....|....|....|....|....|....| Human 1 METVQELIPLAKEMMAQKRKGKMVKLYVLGSVLALFGVVLGLMETVCSPFTAARRLRDQE 60 Mouse 1 ..S...... PR..L...... V...... S...... 60 Berkshire 1 ...I...M.....L..H.PSR.L....L..GL..FL.A...... G....R. 60 Landrace 1 ...I...M.....L..H.PSR.L....L..GL..FL.A...... G....R. 60 Duroc 1 ...I...M.....L..H.PSR.L....L..GL..FL.A...... G....R. 60

70 80 90 100 ....|....|....|....|....|....|....|....|... Human 61 AAVAELQAALERQALQKQALQEKGKQQDTVLGGRALSNRQHAS 103 Mouse 61 ...V..RE.C.Q.S.H....LAG..A.E AT.CS....L..... 103 Berkshire 61 ...... R..R..K.PPEP.QL..D.RRQVGR.C....H.L... 103 Landrace 61 ...... R..R..K.PPEP.QL..D.RRQVGR.C....H.L... 103 Duroc 61 ...... R..R..K.PPEP.QL..D.RRQ .GR.C....H.L... 103

Figure 4.1. Comparison of deduced G0S2 amino acid sequences from mRNA of Human (GenBank assession NP_056529), Mouse (GenBank assession NP_032085) Berkshire (GenBank accession AFS_34519), Landrace (GenBank accession AFS_34520) and Duroc (GenBank accession AFS_34521) pigs. Residues are denoted as identical to the human sequence with periods. Similar residues are highlighted in grey, and different residues are emphasized in black. The conserved hydrophobic domain of G0S2 protein is located within the box positioning from residue 27 to 42.

35 a 30

15 pG0S2 / CYC pG0S2 / b 10

5 c c c c c

0 F H M Sp Lu Li K

Figure 4.2. Tissue-specific mRNA expression of G0S2 in the pig. Total RNAs were isolated from the fat, heart, muscle, spleen, lung, liver, and kidney. Porcine G0S2 mRNA expression was measured by quantitative real-time PCR (n = 4), using porcine cyc was as a normalization gene. Bars represent mean ± SEM. ANOVA followed by Tukey‘s post hoc test was performed to compare means and the three letters, a, b, and c indicate statistically significant differences at P < 0.05.

A B C 1.6 16 25

1.4 14 ** *

20 1.2 12

1 10 15 / CYC / 0.8 8

0.6 6 10

pG0S2 / CYC pG0S2 / pDLK1 / pDLK1 CYC 0.4 pATGL 4 5 0.2 * 2 0 0 0 1 2 SV FC SV FCFF SVSV FFFC

Figure 4.3. Relative gene expressions of DLK1, ATGL and G0S2 in the stromal vascular (SV) and fat cell (FC) fractions from pig adipose tissue. Porcine cyc gene expression was used to provide a ratio of gene expression. Each bar represents mean and SEM. Statistical significance from student's t-test is shown as: * P < 0.05; ** P < 0.01.

A B

1.6 25 a c 1.4

20 1.2 1

/ CYC / 15

0.8 γ

0.6 10 pDLK1 / pDLK1 CYC

0.4 pPPAR 5 a 0.2 b b b a 0 0 10 23 63 49 0 3 6 9 C D 10 b 3.5 9 b

8 7 2.5 ab

6 2 / CYC / 5 a a 4 1.5

3 1 pG0S2 / CYC / pG0S2 pATGL a 2 a 0.5 a 1 0 0 (Day) 0 3 6 9 (Day) 0 3 6 9 Primary adipocyte Primary adipocyte differentiation differentiation

Figure 4.4. Gene expressions of DLK1, PPAR , ATGL and G0S2 during in vitro pig primary adipocyte differentiation. Expression was reported as a ratio to porcine cyc gene expression. Total RNAs were isolated at induced differentiation (day 0) and after differentiation (day 3, 6, and 9). The bar represents mean SEM. ANOVA followed by Tukey‘s post hoc test was used to examine the differences among multiple means. The different letters (a-c) indicates differences at P < 0.05.

A B 1.4 a 16 b

1.2 14

12 1

10 / CYC /

0.8

γ 8 0.6 6

pDLK1 / pDLK1 CYC 0.4

pPPAR 4 a a 0.2 b b 2 0 0 Emb1 6d2 Adult3 Emb 6d Adult C D 180 b 20 b 160 18 16

140

14 120 12

/ CYC / 100 10 80 8

60 6 a

pG0S2 / CYC pG0S2 / pATGL 40 4 a 20 a a 2 0 0 Emb 6d Adult Emb1 6d2 Adult3 Adipose development Adipose development

Figure 4.5. DLK1, PPAR , ATGL and G0S2 expressions during in vivo adipose development. Porcine cyc gene expression was used as normalization controls. Total RNAs were extracted from 105-day old (embryo), 6-day old (neonatal) and 120-day old (adult) pig adipose tissues. Each bar shows mean SEM. ANOVA followed by Tukey‘s post hoc test was performed to compare means and the two letters, a and b indicate statistically significant differences at P < 0.05.

A B 2.5 b 2.5

a 2 2

a CYC 1.5 a 1.5 ab 1 1

b pATGL CYC / pATGL 0.5 pG0S2 / 0.5

0 0 1 2 3 1 2 3

C D E 14 2.5 2.5 12 ** ** 2

8 1.5 1.5

/ CYC /

6 1 1

pDLK1 CYC / pDLK1 pG0S2 / CYC pG0S2 / pATGL 0.5 0.5 2

0 0 0 Ad. Lib Restric Ad. Lib Restric Ad.Lib Restric

Figure 4.6. ATGL, G0S2, AND DLK1 gene expressions during short-term fasting and refeeding (A and B) and long-term food restriction (C, D, and E). A and B: Both ATGL and G0S2 expressions in pig subcutaneous fat after fasting for 24 hours and six hours after refeeding. ANOVA followed by Tukey‘s post hoc test was performed to compare three means, and the letters, a and b show statistically significance at P < 0.05. C, D, and E: DLK1, ATGL and G0S2 expressions in pig peri-renal fat. Control (3 kg/day of the corn-soybean meal ration) and calorie restricted diet (1 kg/day of the same ration) for 8 days were used as two treatments. Statistical significance from student's t-test is indicated as: ** P < 0.01. A ratio to porcine cyc gene expression was reported. Each bar represents mean SEM.

Figure 4.7. Multiple alignments of complete coding sequences of G0S2 from Human (GenBank accession NM_015714), Mouse (GenBank accession NM_008059) Berkshire (GenBank accession JQ_013998), Landrace (GenBank accession JQ_013999), and Duroc (GenBank accession JQ_014000). Identities to the human G0S2 nucleotide sequence are denoted by a period, and nucleotide differences among porcine species were indicated by an open box.

Figure 4.7. 10 20 30 40 50 60 ....|....|....|....|....|....|....|....|....|....|....|....| Human ATGGAAACGGTCCAGGAGCTGATCCCCCTGGCCAAGGAGATGATGGCCCAGAAGCGCAAG 60 Mouse ...... GT..G...... T...... G...... C.CGA 60 Berkshire ...... A...... G..T...... C...... C....C..GC 60 Landrace ...... A...... G..T...... C...... C....C..GC 60 Duroc ...... A...... G..T...... C...... C....C..GC 60

70 80 90 100 110 120 ....|....|....|....|....|....|....|....|....|....|....|....| Human GGGAAGATGGTGAAGCTGTACGTGCTGGGCAGCGTGCTGGCCCTCTTCGGCGTGGTGCTC 120 Mouse ...... C.A...... A...... T...... G...... T...... 120 Berkshire C.C...C...... C...... G..C...... T.TC...... C...... 120 Landrace C.C...C...... C...... G..C...... T.TC...... C...... 120 Duroc C.C...C...... C...... G..C...... T.TC...... C...... 120

130 140 150 160 170 180 ....|....|....|....|....|....|....|....|....|....|....|....| Human GGCCTGATGGAGACTGTGTGCAGCCCCTTCACGGCCGCCAGACGTCTGCGGGACCAGGAG 180 Mouse .....AG.T.....G...... T.....A...... C...... C.....A... 180 Berkshire ...... C...... G.C..C...... G.... 180 Landrace ...... C...... G.C..C...... G.... 180 Duroc ...... C...... G.C..C...... G.... 180

190 200 210 220 230 240 ....|....|....|....|....|....|....|....|....|....|....|....| Human GCAGCCGTGGCGGAGCTGCAGGCCGCCCTGGAGCGACAGGCTCTCCAGAAGCAAGCCCTG 240 Mouse ..T..A....T...... G..AA..GTGT..A.AG...T.C.....C.....G...... 240 Berkshire ..C...... C...... G...... G...... A....G.C..C.G...CG....A. 240 Landrace ..C...... C...... G...... G...... A....G.C..C.G...CG....A. 240 Duroc ..C...... C...... G...... G...... A....G.C..C.G...CG....A. 240

250 260 270 280 290 300 ....|....|....|....|....|....|....|....|....|....|....|....| Human CAGGAGAAAGGCAAGCAGCAGGACACGGTCCTCGGCGGCCGGGCCCTGTCCAACCGGCAG 300 Mouse .T..CAGG...... GCA.....GG..AC...GT..A...... CT...... 300 Berkshire .T...... G.A.....G..G.C.GGT..G..GG...T...... C...C...... T. 300 Landrace .T...... G.A.....G..G.C.GGT..G..GG...T...... C...C...... T. 300 Duroc .T...... G.A.....G..G.C.G....G..GG...T...... C...C...... T. 300

....|.... Human CACGCCTCC 309 Mouse ...... T 309 Berkshire ...... 309 Landrace ...... 309 Duroc ...... 309

Chapter 5: Differential expressions of G0/G1 switch gene 2 and comparative gene

identification-58 are Aassociated with fat content in bovine muscle

Jinsoo Ahn1, 2 , Xiang Li1, Young Min Choi1, Sangsu Shin1, Shin-Ae Oh1, Yeunsu Suh1,

Trang Hoa Nguyen3, Myunggi Baik4, Seongsoo Hwang5, and Kichoon Lee1, 2, *

1Department of Animal Sciences and 2 Interdisciplinary Ph.D. Program in Nutrition, The

Ohio State University, Columbus, OH, 43210, USA 3Department of Molecular

Biotechnology, Chonnam National University, Gwangju 500-757, Republic of Korea

4Department of Agricultural Biotechnology and Research Institute for Agriculture and

Life Sciences, Seoul National University, 1 Gwanak-ro, Gwanak-gu, Seoul 151-921,

Republic of Korea 5Animal Biotechnology Division, National Institute of Animal Science,

RDA, Seosuwon-ro 143-13, Suwon, Gyeonggi 441-706, Republic of Korea

5.1. Abstract

Regulation of lipolysis in muscle is a potential mechanism affecting marbling in beef carcasses and fat accumulation in muscles of humans, which is a known risk factor for type 2 diabetes. Adipose triglyceride lipase-mediated lipolysis is inhibited by G0/G1 switch gene 2 (G0S2) and co-activated by comparative gene identification-58 (CGI-58).

In this study, bovine G0S2 and CGI-58 were sequenced, and expressions of these genes were compared among various tissues and in muscles between bulls and steers with

79 different degrees of marbling. The protein coding sequences of bovine G0S2 and CGI-58 revealed breed-specific SNPs, causing two amino acid variations for each protein. Bovine

CGI-58 mRNA showed two isoforms from alternative splicing. The G0S2 gene was preferentially expressed in fat and, to a lesser degree, in the liver; whereas, CGI-58 was highly expressed in the muscle and fat (P < 0.05), suggesting their association with lipid metabolism in those tissues. The longissimus dorsi muscle (LM) of steers showed higher

FABP4, G0S2 and CGI-58 mRNA expression levels than the LM of bulls, implying the roles of those genes more in marbling of steers than in that of bulls. The G0S2 expression was markedly higher in the intramuscular fat (IMF) (P < 0.001); whereas, the CGI-58 expression was significantly higher in the pure muscle portion of the LM of steers (P <

0.01), suggesting that G0S2 and CGI-58 may regulate IMF and intramyocellular triglycerides (IMTG), respectively. Taken together, our data suggest that G0S2 and CGI-

58 are associated with fat content in bovine species.

Keywords: lipolysis, marbling, type 2 diabetes, bovine G0/G1 Switch Gene 2 (G0S2), bovine comparative gene identification-58 (CGI-58)

5.2. Introduction

Advancing the understanding of lipid metabolism in muscle is critical to the beef industry due to its implications in regulation of marbling and to people living with type 2 diabetes because of its association with insulin sensitivity. In order to further our knowledge, lipid droplet-associated proteins are of great interest and may reveal novel information about the basic mechanism of lipid metabolism. Quality grading on beef carcasses is primarily based on the degree of marbling (intramuscular fat; IMF) (Smith et al., 1984). Marbling scores have been shown to be associated with several factors, including nutritional management, age, breed, and genotype (Bennett & Williams, 1994; Wertz et al., 2002).

However, producing higher marbling deposition while maintaining a minimum level of undesirable subcutaneous fat and visceral fat has remained a challenge (Jeremiah, 1996;

Wood et al., 2008; Yang et al., 2010a). Therefore, identification of specific genes that regulate marbling is gaining attention (Lee et al., 2008). In type 2 diabetic patients, lipid accumulation in muscle impairs insulin action, as shown in studies demonstrating that intramyocellular triglycerides (IMTG) inversely correlate with insulin sensitivity (Krssak et al., 1999; Virkamäki et al., 2001). Taken together, lipid deposition in muscle has been shown to be associated with marbling in beef cattle as well as insulin resistance, leading to pathogenesis of type 2 diabetes in humans.

In addition to promoting triacylglycerol (TAG) synthesis, another way to enhance marbling is to reduce the loss of TAG in muscle tissue by inhibiting lipolysis.

Mobilization of deposited lipids is initiated by the action of lipolytic enzymes. Adipose triglyceride lipase (ATGL)-mediated lipolysis has been widely studied in several tissues, including adipose tissue and muscle, because of the role of ATGL as a rate-limiting

81 enzyme for hydrolyzing TAG (Zimmermann et al., 2004; Jenkins et al., 2004; Villena et al., 2004). ATGL has been shown not only to be elevated in adipose tissue of several species (Kershaw et al., 2006; Kim et al., 2006), but also to be expressed in non-adipose tissues, including bovine muscle (Deiuliis et al., 2010), connecting to the omnipresence of lipid droplets. ATGL-deficient mice showed significantly less TAG hydrolase activity and more TAG in skeletal muscle, suggesting a critical role of ATGL in lipolysis of skeletal muscle (Haemmerle et al., 2006).

As the role of ATGL on lipolysis emerges, regulators of ATGL have drawn particular interest. Among them, comparative gene identification-58/ α/β hydrolase domain-containing protein 5 (CGI-58/ABHD5) has been shown to stimulate TAG hydrolase activity of ATGL and, in particular, 4-fold increase of TAG hydrolase activity by CGI-58 in tissue lysates of murine skeletal muscle has been demonstrated (Lass et al.,

2006). To the contrary, the G0/G1 switch gene 2 (G0S2) is an inhibitor of ATGL- mediated lipolysis and is highly expressed in adipose tissue of the mouse, chicken, and pig (Yang et al., 2010b; Oh et al. 2011; Ahn et al., 2013). The objectives of this study are to investigate potential marbling associated-G0S2 and CGI-58 genes through analysis of sequence homology among several cattle breeds, reveal gene expression patterns, and examine an association between gene expressions of G0S2 and CGI-58 and marbling/IMF content in cattle.

5.3. Materials and Methods

Experimental Animals and Materials

All animal care and use procedures were approved by the Institutional Animal Care and

Use Committee (IACUC) at The Ohio State University. In order to clone and sequence bovine CGI-58 and G0S2, adipose tissue was collected from three adult male individuals

(450–600 kg, ~550 day old) of each cattle breed available at the time of this study, which includes Angus, Holstein, Jersey, and Hanwoo (3 cattle for each breed). For the investigation of mRNA expression in different bovine tissues, a set of tissues

(subcutaneous adipose tissue, heart, skeletal muscle, spleen, lung, liver, and kidney) was collected in a previous study using an Angus steer (~460 kg, ~550 day old) as a representative (Deiuliis et al., 2010). In detail, four Angus cattle were euthanized by captive blot stunning and exsanguination at the Ohio State University‘s Meat Science

Laboratory located in the Department of Animal Sciences, immediately followed by collection and snap-frozen in liquid N2 of the seven above tissues. Those tissues were stored at ‒80°C before RNA isolation. The expressions of FABP4, G0S2, and CGI-58 in the longissimus dorsi muscle (LM) of Hanwoo (Korean cattle) bulls and steers were evaluated as described below. Hanwoo steers were selected because of its high level of marbling. For Hanwoo experiment, tissues from a previous study were used (Bong et al.,

2012). Briefly, bulls were weaned at an average of 3 mo of age, and fed 30% concentrates/70% roughage. A portion of bulls was castrated at 6 mo of age. Cattle was fed with concentrate that consisted of 15% crude protein (CP)/71% total digestible nitrogen (TDN) from 6 to 14 mo of age, 13% CP/72% TDN‒‒ until 20 mo of age, and

11%/73% TDN after 21 mo of age. Roughage (mainly rice straw) was offered ad libitum.

The intact LM tissues at the 13th rib from a hot carcass were collected immediately after slaughter, frozen in liquid nitrogen, and stored at ‒80°C. We dissected both IMF and

83 pure muscle portion from the intact LM tissues in order to compare gene expression levels between the two tissues within steers. During IMF and muscle collection from the intact LM tissues, frozen tissues were placed on dry ice and covered with foil in order to maintain RNA integrity as previously described (Bong et al., 2012).

Isolation of Total RNA and the Preparation of cDNA

Trizol (Invitrogen Inc., Carlsbad, CA) was used to isolate total RNA in adipose tissue from the four breeds of cattle, as well as, from the bovine heart, muscle, spleen, lung, liver, and kidney tissues. The procedures were performed according to the instructions provided by the manufacturer. NanoDrop 100 (Thermo Scientific, Wilmington, DE) was used to assess the quantity of extracted RNA. The quality of the isolated RNA was evaluated by agarose gel electrophoresis. The total RNA samples were stored at ‒80°C.

Complementary DNA (cDNA) was obtained through reverse transcription (RT) using approximately 1µg of total RNA for each sample, according to the instructions of the manufacturer of Moloney murine leukemia virus (M-MLV) reverse transcriptase

(Invitrogen). The conditions of the RT reaction were 65°C for 5 min, 37°C for 52 min, and 79°C for 15 min.

Cloning and Sequencing of Bovine G0S2 and CGI-58 and Alternative Splicing

Patterns

The cDNA samples from three individuals from each of the four breeds of cattle were used as templates for cloning bovine G0S2 and CGI-58, respectively. Several sets of primers were designed according to the sequence information in The National Center for 84

Biotechnology Information (NCBI) database (NM_001192147.1 for G0S2 and

NM_001076063.2 for CGI-58/ABHD5). After preliminary trials, the set of primers

(G0S2 forward primer: 5‘-CAG CCC AGA TGG AGA CGG T-3‘; G0S2 reverse primer:

5‘-AGC AGC GAA CTC AGT CCC AAG T-3‘; CGI-58 forward primer: 5‘-GGC TAG

GAG GTA CTG TCT GAC AGT G-3‘; CGI-58 reverse primer: 5‘-GTT CTC GTG TCA

GAT TCC TTA CTC TG-3‘/ 5‘-GGC TTT CCC ACA TCA TCT GCA-3‘) was selected to amplify the entire sequence for bovine G0S2 and CGI-58. The primer set covers the coding sequences of G0S2 and CGI-58 from the first to the last exon and is capable of detecting any possible alternatively spliced variants.

Thermo Taq polymerase (New England Biolabs, Carlsbad, CA) was used during polymerase chain reaction (PCR) for bovine G0S2 and CGI-58 amplification. The conditions for this reaction were: 95°C for 1 min 30 s, 40 cycles of 94°C for 30 s, 57°C for 1 min, 72°C for 1 min, with an additional extension step at 68°C for 10 min.

Separation of PCR products was performed using 1% agarose gel electrophoresis. The major bands were excised for DNA extraction with Qiagen Gel Extraction Kit (Qiagen

Inc. Valencia, CA). The gel extraction products were incorporated into pCR 2.1 vectors using the TOPO TA Cloning Kit (Invitrogen), followed by transformation into TOP10 chemically competent cells (Invitrogen). LB kanamyacin/ ampicillin agar with X-gal for blue white selection of positive colonies was used to plate the transformation products.

The QIAprep Spin Miniprep Kit (Qiagen) was used to isolate the plasmid of the selected and cultured bacteria. Miniprep products were sent to The Ohio State University

Sequencing Core Facility to obtain the sequences of the targeted gene fragments, using an

Applied Biosystems 3730 DNA analyzer (Applied Biosystems, Carlsbad, CA).

The presence and patterns of alternative splicing of G0S2 and CGI-58 mRNAs were represented by using 1% agarose gel electrophoresis after amplification of those genes with the primer sets described above. The gel electrophoresis was only used to verify the existence of alternative splicing, but not quantification of the isoforms.

Quantitative Real-Time PCR

The relative mRNA expression levels of G0S2 and CGI-58 in various bovine tissues were assessed by quantitative real-time PCR (qPCR). For bovine G0S2, forward primer: 5‘-

GAA GCG AGA CCC CGG AGA-3‘ and reverse primer 5‘-TAG TTC AGT TCT AGA

AGT CGG TGG CTT-3‘ were used to detect the expression of the full-length G0S2 mRNA. For bovine CGI-58, a set of forward primer: 5‘-CAG TGA CGG AAT ACA

TCT ACC ACT G-3‘and reverse primer: GCC AAT TCG CTG GAG CAT-3‘ (partially spanning exon 5 and 6) was used to detect the full-length and both alternatively spliced variants of CGI-58 mRNA. Equal amounts of cDNA from each tissue sample were used as templates with the reaction system including AmpliTaq Gold polymerase (Applied

Biosystems, Foster City, CA), GeneAmp 10×PCR Buffer (containing 100-mM Tris-HCL, pH 8.3), and 500-mM KCL. SYBR green was used to detect the amplification of the products. Duplicate reactions were performed (25μL volume) on an ABI 7300 real-time

PCR instrument (Applied Biosystems). The conditions of the qPCR reaction were 95°C for 10 min, 40 cycles of 94°C for 15s, 57°C for 40 s, and 72°C for 30s, with an additional

82°C extension for 33s. A single major peak on the dissociation curves provided by the qPCR software indicated successful amplification. Bovine cyclophilin (CYC) was used as a reference gene for the normalization of the mRNA expression levels of G0S2 and CGI-

58 by virtue of a relatively constitutive expression of CYC among several tissues

(Deiuliis et al., 2010; Ghinis-Hozumi et al., 2011). The primer sequences used for CYC were 5'-GTG GTC ATC GGT CTC TTT GG-3' (forward primer) and 5'- CAC CGT AGA

TGC TCT TAC CTC-3' (reverse primer). For the tissue distribution, relative expression

-Δ of the target genes was calculated using the 2 CT (CT: threshold cycle) method (Livak &

Schmittgen, 2001).

To assess the expression of FABP4, G0S2, and CGI-58 in Hanwoo, intact LM tissue samples of 10 bulls and 10 steers were used from a previous study (Bong et al.,

2012). Moreover, the pure muscle and IMF portions from the intact LM of the steers were separated to evaluate gene expressions in both portions (Bong et al., 2012). qPCR was performed as described in a previous report using β-actin as a normalization control

(Jeong et al., 2013). The β-actin control gene was constitutively and uniformly expressed among all samples from intact LM of Hanwoo bulls and steers. The primer sequences for

FABP4 were 5'-GCT GCA CTT CTT TCT CAC CT-3' (forward primer) and 5'-TTC CTG

GTA GCA AAG CCC AC-3' (reverse primer). The primer sequences for G0S2 and CGI-

58 were the same as the primer set for mRNA expression levels of various bovine tissues as described above. The primer sequences for β-actin were 5'-AGC AAG CAG GAG

TAC GAT GAG T-3' (forward primer) and 5'-ATC CAA CCG ACT GCT GTC A-3'

(reverse primer). The level of gene expressions in Hanwoo was measured by relative

-ΔΔ quantification using the 2 CT method (Livak & Schmittgen, 2001). The efficiencies for each primer were 84.57% for G0S2, 87.75% for CGI-58, 83.48% for CYC, 88.61% for

FABP4, and 93.01% for β-actin, as assessed by a web-based qPCR efficiency calculator

(Thermo Scientific).

Bioinformatics and Statistical Analysis

For the analysis of gene and protein homologies of G0S2 and CGI-58 among species, the basic local alignment search tool (BLAST) from NCBI was used. Sequences for human and mouse G0S2 and CGI-58 were obtained from a NCBI resource (NM_015714.3 for human G0S2; NM_008059.3 for mouse G0S2; AF151816.1 for human CGI-58/ABDH5;

NM_026179.2 for mouse CGI-58/ABDH5). Sequencing information generated by the present study was used in the homology alignment to represent the bovine species.

ClustalX multiple alignment software (Thompson et al., 1994) with BioEdit Sequence

Alignment Editor (Version 7.0.9.0; http://www.mbio.ncsu.edu/bioedit/bioedit.html) was used for alignment and comparison of cDNA and AA sequences from human, mouse and bovine G0S2 and CGI-58. Translation of bovine G0S2 and CGI-58 cDNA sequences into

AA sequences was performed through the Open Reading Frame Finder (ORF finder)

(http://www.bioinformatics.org/sms2/orf_find.html).

For comparison of mRNA expression levels of bovine CGI-58 in different tissues,

Tukey‘s test was performed following One-way ANOVA using SAS version 9.1 (SAS

Inst. Inc., Cary, NC). Pvalues < 0.05 were considered significant. The general linear model (GLM) procedure of the SAS software (SAS institute, Cary, NC, USA) was used to compare the expression of FABP4, G0S2, and CGI-58 in Hanwoo.

5.4. Results cDNA and Amino Acid Sequences of Bovine G0S2 and CGI-58

Sequencing results revealed that full-length coding sequences of bovine G0S2 and CGI-

58 cDNAs are 294 base pairs (bp) with 2 exons and 1,047 bp with 7 exons, respectively.

The current sequencing results of Holstein and Jersey G0S2 cDNAs and Angus and

Jersey CGI-58 cDNA are 100% identical to the available NCBI information

(NM_001192147.1 for Hereford G0S2 and NM_001076063.2 for Hereford CGI-

58/ABHD5). However, there were minor differences between different breeds of cattle.

For G0S2, the 109th nucleotide is ―G‖ in Angus, Holstein and Jersey, while it is ―A‖ in

Hanwoo, and the 190th nucleotide is ―A‖ in Angus, whereas, it is ―G‖ in Holstein, Jersey, and Hanwoo (Fig. 5.1). Regarding CGI-58, the 167th nucleotide is ―G‖ in Angus and

Jersey, while it is ―A‖ in Holstein and Hanwoo, and the 359th nucleotide of Hanwoo is

―T‖, which is different from ―A‖ in Angus, Holstein and Jersey (Fig. 5.2). Taken together, both G0S2 and CGI-58 cDNAs are 99% homologous among the four breeds of Angus,

Holstein, Jersey and Hanwoo. Sequence variances were present among the breeds, but not within each breed. Furthermore, based on the homology comparison using the NCBI

Standard Nucleotide BLAST tool, bovine G0S2 (represented by Angus) shares an 86%,

79%, 77% and 73% homology with the pig (GenBank EB422575.1), human, chicken

(GenBank NM_001190924.3) and mouse, respectively. Bovine CGI-58 (represented by

Angus) shares a 93%, 92%, 87%, and 78% homology with pig (GenBank AY902463.1), human, mouse and chicken (GenBank HQ896422.1), respectively.

Bovine G0S2 and CGI-58 proteins consist of 98 and 348 amino acids, respectively, according to the translation from cDNA to amino acid sequence performed using an ORF finder (Fig. 5.3 and 5.4). Human and mouse G0S2 amino acid sequences are 103 amino acids in length (Fig. 5.3). Regarding CGI-58 amino acid sequences, the human and mouse have 349 and 351 amino acids, respectively (Fig. 5.4). The bovine

G0S2 proteins are five amino acids shorter than both human and mouse G0S2 proteins,

89 with deletions present in the C-terminal region from position 79 to 83 (Fig. 5.3). The bovine CGI-58 has a deletion at position 73 of mouse CGI-58 (or position 71 of human

CGI-58) (Fig. 5.4). The variations in nucleotides lead to amino acid differences among the four breeds of cattle. Regarding G0S2, glycine (Gly, G) at residue 37 in Angus,

Holstein, and Jersey is replaced by serine (Ser, S) in Hanwoo; and alanine (Ala, A) at residue 64 in Holstein, Jersey, and Hanwoo is changed to threonine (Thr, T) in Angus

(Fig. 5.3). For CGI-58, arginine (Arg, R) at residue 56 in Angus and Jersey is substituted by histidine (His, H) in Holstein and Hanwoo; and aspartic acid (Asp, D) at residue 120 in Angus, Holstein and Jersey is replaced by valine (Val, V) in Hanwoo (Fig. 5.4). Angus,

Holstein, Jersey and Hanwoo share a 98 – 100% and 99% homology regarding G0S2 and

CGI-58 amino acid sequences, respectively. Using Angus as the representative, bovine

G0S2 protein is 76%, 62%, 60%, and 49% homologous with the pig, human, mouse and chicken; and bovine CGI-58 protein is 95%, 94%, 93%, and 79% homologous with the human, pig, mouse, and chicken, respectively.

The Alternative Splicing of Bovine CGI-58 mRNA

PCR of bovine CGI-58 displayed three specific bands (full-length, ASI, and ASII) with sizes of 1110 bp, 654 bp, and 499 bp, respectively (Fig 5A). As the primers used in the cloning and amplification of bovine CGI-58 spanned the complete coding region of the gene, the two shorter transcripts detected by the agarose gel electrophoresis separation were likely to be the result of alternative splicing events within the mRNA editing of bovine CGI-58, even though the expression of both transcripts were less significant than that of the full-length. This speculation was confirmed by the sequencing result of the

90 two shorter fragments. Taken that the 1,110 bp of the full-length PCR product is covered by the primers, the observed 654 bp fragment (ASI) consists of an alternatively spliced variant with a deletion of 456 bp corresponding exon 2 and 3 (Fig. 5.5). Likewise, the

499 bp fragment (ASII) is the second alternatively spliced variant with a deletion of 611 bp covering 2nd, 3rd, and 4th exons (Fig. 5.5). It is predicted that the stop codon is located at 5th exon of ASII fragment, unlike the full-length and ASI with the stop codon at the 7th exon. On the other hand, bovine G0S2 did not display an alternative splicing pattern, as shown in the agarose gel electrophoresis, despite applying to PCR amplification using primers that cover both exon 1 and 2 (Data not shown).

Tissue Distribution of G0S2 and CGI-58 Expression

Tissue specific variations of G0S2 gene expression in Angus cattle were analyzed by qPCR. The levels of G0S2 expression were highest in subcutaneous adipose tissue and moderate in the liver, compared to other tissues (Fig. 5.6A). The expression of CGI-58 was also observed among different bovine tissues. In the bovine tissue distribution detected by qPCR, the mRNA expression of CGI-58 was remarkably higher in skeletal muscle, and showed moderated levels in subcutaneous adipose tissue compared to other tissues (Fig. 5.6B).

Expression of FABP4, G0S2 and CGI-58 in Muscle vs. IMF and Bulls vs. Steers

To measure the expression of G0S2 and CGI-58 in pure muscle and intramuscular fat

(IMF), pure muscle and IMF portions from the intact LM of steers of Hanwoo were separated. First, mRNA expression of FABP4 was measured as an adipose-specific gene

(Deiuliis et al., 2010). Expression of the FABP4 gene was 163-fold greater (P < 0.001) in the IMF compared to the muscle portion without IMF (Fig. 5.7A), indicating a successful separation of IMF and the pure muscle portion. In IMF, relative G0S2 expression was a

43-fold greater (P < 0.001) than the pure muscle portion of the intact LM (Fig. 5.7C).

Interestingly, muscle expressed approximately 3-fold greater (P < 0.01) amount of CGI-

58 expression than IMF (Fig. 5.7E).

Expressions of those genes were quantified in the intact LM of steers and bulls of

Hanwoo in order to evaluate the effect of more IMF in steers on the gene expression. The intact LM of steers displayed greater relative expressions of FABP4 (P < 0.01), G0S2 (P

< 0.05) and CGI-58 (P < 0.01) by 3-, 2-, and 1.7-folds, respectively, compared to those of bulls (Fig. 5.7B, D, and F).

5.5. Discussion

G0S2 and CGI-58 have been recently studied in humans, as well as in the domestic animal species, due to their important functions in the regulation of lipolysis. In adipose tissue, CGI-58 binds to the lipid droplet upon stimulation, associating with ATGL to perform the first step of breaking down stored TAG (Lass et al., 2006). These functions are supported by a series of features of the CGI-58 proteins. The main structure of CGI-

58/ABHD5 protein forms an ―α/β hydrolase-fold domain (ABHD)‖ with no intrinsic hydrolase activity, due to a replacement of nucleophilic serine (S) in the central consensus ―GXSXG‖ esterase/lipase motif to arginine (N) forming ―GXNXG‖ (Asn153 in humans and Asn155 in mice) (Schweiger et al., 2009). This change is also present in bovine CGI-58, as shown in Fig. 5.2 (Asn152). In addition, CGI-58 localizes to the lipid

92 droplet through its N-terminal extension (amino acids 1 – 30) containing a lipophilic tryptophan-rich domain (Gruber et al., 2010; Serr et al., 2011; Li et al., 2012). The tryptophan-rich domain is conserved in bovine CGI-58 (Fig. 5.4), implying a similar localization of bovine CGI-58.

A previous study showed that an alternatively spliced murine isoform with exon 2 and 3 deletion lost lipid droplet-binding or ATGL stimulation activity (Yang et al.,

2010c). In this study, two alternatively spliced variants of bovine CGI-58 were shown to exist with a relatively lower expression than that of the full-length. These variants did not possess the 2nd exon, which contains the tryptophan-rich lipid droplet binding domain, and might not have the ability to localize to the lipid droplet or stimulate ATGL during lipolysis, modifying the stimulatory action of bovine CGI-58 towards ATGL.

Furthermore, it has been suggested that a coenzyme A-dependent lysophosphatidic acid acyltransferase (LPAAT) function is linked to the ―HX4D‖ motif of CGI-58 at its C- terminal end (Yang et al., 2010c; Montero-Moran et al., 2010). These functional/structural domains are conserved among various species, such as the human, mouse, chicken, and pig (Li et al., 2012). The current study shows that alternatively spliced variant I (AS I) of bovine CGI-58 is 591 bp (196 AA) in length with the deletion of 2nd and 3rd exons. As the ―HX4D‖ motif for LPAAT activity of CGI-58 is contained in the 7th exon, LPAAT function of AS I is likely retained. However, AS II (436 bp in length) of bovine CGI-58 consists of only the 1st and 5 ‒ 7th exons, which leads to an

ORF shift causing a premature stop codon (TGA), resulting in a truncated 25 AA protein.

Consequently, LPAAT activity is not likely to be retained in this isoform. Further study of these alternatively spliced variants will be required to clarify these points.

G0S2 protein is directly bound to ATGL, inhibiting its hydrolase activity (Yang et al., 2010b). Comparative analysis of amino acid sequences among humans, mice, pigs, and avian species (Oh et al., 2011; Ahn et al., 2013) suggested that the conserved N- terminal half of G0S2 might possess critical functional roles on ATGL, compared with relatively variable C-terminal region. Bovine G0S2 also has a conserved N-terminal half and a deletion in variable C-terminal area, further confirming the important role of N- terminal region. This is supported by studies with deletion mutations in murine G0S2, showing that the hydrophobic region in the N-terminal half is a determinant for binding of G0S2 to ATGL (Yang et al., 2010b).

To relate expression of an adipogenic marker gene with different amounts of IMF deposition, LM tissue samples of Hanwoo bulls and steers were used. In our earlier study with these tissues, steer LM meats had a 3.7-fold higher IMF content compared to bulls

(11.0 ± 1.08 vs. 3.0 ± 0.28 for steers and bulls, respectively; P < 0.001) (Bong et al., 2012;

Jeong et al., 2013). In pigs and cattle, studies on linkage analysis suggested fatty acid binding protein 4 (FABP4) gene as a strong indicator for marbling (Estellé et al., 2006;

Wibowo et al., 2008; Lee et al., 2010). Our prior experiments revealed that FABP4 protein is dominantly expressed in adipose tissue among various tissues and organs in mice, pigs and cattle (Deiuliis et al., 2010; Shin et al., 2009), indicating FABP4 as an adipocyte and adipogenic marker. In this study, the pure muscle and IMF portions from the intact LM of the steers were separated to determine whether IMF deposition affects gene expression in the LM. Expression of the FABP4 gene was 163-fold greater (P <

0.001) in the IMF compared to the muscle portion without IMF (Fig. 7A), suggesting a positive correlation of FABP4 expression with the amounts of IMF and the degree of

94 marbling as well as an efficient separation of IMF from the pure muscle portion. In addition, greater expression of FABP4 gene in the LM of steers compared to bulls (Fig.

7B) further confirms a positive association of the FABP4 gene with marbling in cattle and pigs and fat deposition in poultry muscle (Wood et al., 2008; Jurie et al., 2007; Saez et al.,

2009; Li et al., 2008; Ye et al., 2010; Gardan et al., 2007; Zhao et al., 2009). Therefore, expression of the FABP4 gene may serve as a biomarker for the degree of marbling. In previous studies, we also found that mRNA levels of CCAAT/enhancer binding protein alpha (C-EBPα) and acetyl-CoA carboxylase (ACC) genes were higher in the IMF than in the pure muscle portion of intact LM tissues (Bong et al., 2012; Jeong et al., 2013). Our results demonstrate that adipogenic C-EBPα and lipogenic ACC mRNA levels are closely associated with marbling in cattle.

Recent studies with genetic alterations of the ATGL gene in the human and mouse showed an abnormal accumulation of fat in many tissues including adipose tissue, heart, liver, and muscle (Haemmerle et al., 2006; Fischer et al., 2007; Wu et al., 2011).

Increased IMF in ATGL knockout mice and humans with a perturbation of the ATGL gene suggest that regulation of ATGL-mediated lipolysis could be a potential application to increase marbling (Haemmerle et al., 2006; Nunes et al., 2012; Reilich et al., 2011).

However, despite the inhibitory and stimulatory roles of G0S2 and CGI-58 on ATGL, respectively, the roles of G0S2 and CGI-58 on regulation of fat accumulation in bovine muscle have not been fully understood.

The findings on higher expressions of G0S2 in bovine fat and liver (Fig. 5.6A) are in agreement with previous research on murine G0S2 expression, which showed that

G0S2 was highly expressed in adipose tissue and the liver (Yang et al., 2010b). Because

ATGL is the major rate-limiting lipase in adipose tissue (Zimmermann et al., 2004;

Villena et al., 2004) and hepatic ATGL is associated with hepatic TAG hydrolysis and decreased hepatic steatosis (Ong et al., 2011), a high G0S2 expression in adipose tissue and the liver may contribute to lipid metabolism through ATGL in those tissues. Since

IMF is an adipose tissue depot in muscle, G0S2 expression in IMF was further investigated. In this study, the amount of G0S2 expression was 43-fold greater in IMF compared to the pure muscle portion of the LM of the steers. Also, a significantly higher expression of G0S2 in the intact LM of steers than bulls could be associated with higher

IMF contents with a greater number of fat cells or more mature adipocytes in LM of the steers.

Interestingly, qPCR analysis revealed higher levels of CGI-58 expression in muscle than in other tissues (Fig. 5.6B). In addition, CGI-58 showed a significantly greater expression (P < 0.01) in the intact LM of the steers than the bulls (Fig. 5.7F), as well as, in the pure muscle portion of the intact LM of the steers (approximately 3-fold greater) than in IMF (Fig. 5.7E). These data suggest that the ATGL co-activator, CGI-58, may be involved in lipolysis more in myofiber than in IMF. This is supported by the findings from the studies with cultured human primary myotubes, that CGI-58 overexpression increased intracellular TAG hydrolysis; whereas, CGI-58 silencing reduced the hydrolysis, implying the role of CGI-58 on the regulation of intramyocellular triglycerides (IMTGs) in skeletal muscle (Badin et al., 2012).

Taken together, a predominant expression of G0S2 in adipose tissue may be linked to a significantly higher expression of G0S2 in IMF (adipose depot in muscle), and significantly higher expression of CGI-58 in muscle compared to other tissues may be

96 attributed to greater expression of CGI-58 in pure muscle containing IMTGs. However, greater expressions of both G0S2 and CGI-58 in the intact LM of the high-marbled steers than the bulls are paradoxical because CGI-58 co-activates lipid breakdown mediated by

ATGL instead of promoting fat accumulation, as in the case of G0S2, which inhibits

ATGL. This could be explained by regarding both G0S2 and CGI-58 not as causative factors, but as results of enhanced fat deposition in muscle. The increased expressions of

G0S2 and CGI-58 in the steer with more marbling could reflect the high amount of fat. It is consistent with previous studies that reported a positive correlation between an increased amount of TAG substrate and enhanced expressions of ATGL and its regulators

(G0S2 and CGI-58) during adipocyte differentiation and in fat cell fraction comparing to stromal vascular fraction (Ahn et al., 2013; Li et al., 2012). This idea is further supported by the fact that pigs, chicken, turkey, and quail with low levels of marbling display low expressions of G0S2 and CGI-58 in muscle (Oh et al., 2011; Ahn et al., 2013; Serr et al.,

2011; Li et al., 2012). Given the reflections of muscle fat content by amounts of G0S2 and CGI-58 expression, both genes could be novel biomarkers for marbling in cattle.

In conclusion, this study reported cDNA and amino acid sequences of G0S2 and

CGI-58 in four breeds of cattle (Angus, Holstein, Jersey, and Hanwoo) and determined bovine G0S2 and CGI-58 is highly homologous among each breed. Two unique alternatively spliced variants of bovine CGI-58 were identified. The mRNA expressions of both bovine G0S2 and CGI-58 were significantly higher in LM of the steers than in the bulls of Hanwoo, but G0S2 may have a significant role in the IMF portion, while CGI-58 may play a role in IMTGs of the muscle. The current research on single nucleotide polymorphisms (SNPs) in CGI-58 gives new insights in the regulators of muscle fat

97 accumulation, which can be potential factors that are linked to marbling. In support of this, loss-of-function mutations in the human CGI-58 gene is a causative factor of

Chanarin-Dorfman syndrome (neutral lipid storage disease), characterized by TAG deposition in multiple tissues including muscle (Lefèvre et al., 2001). More studies are needed to investigate the importance of G0S2 and CGI-58 genes in the bovine species.

Overall, this study provides new foundational information for future studies on bovine

G0S2 and CGI-58 and their regulation in the pathways of ATGL-mediated lipolysis.

5.6. Acknowledgements

This work was supported by the Ohio Agricultural Research and Development Center

Director‘s Associateship Program and two grants from the Next-Generation BioGreen 21

Program (No. PJ008191 and PJ009457), Rural Development Administration, Republic of

Korea.

10 20 30 40 50 60 ....|....|....|....|....|....|....|....|....|....|....|....| Human 1 ATGGAAACGGTCCAGGAGCTGATCCCCCTGGCCAAGGAGATGATGGCCCAGAAGCGCAAG 60 Mouse 1 ...... GT..G...... T...... G...... C.CGA 60 Angus 1 .....G...... T...... C...... G...... C..GC 60 Holstein 1 .....G...... T...... C...... G...... C..GC 60 Jersey 1 .....G...... T...... C...... G...... C..GC 60 Hanwoo 1 .....G...... T...... C...... G...... C..GC 60

70 80 90 100 110 120 ....|....|....|....|....|....|....|....|....|....|....|....| Human 61 GGGAAGATGGTGAAGCTGTACGTGCTGGGCAGCGTGCTGGCCCTCTTCGGCGTGGTGCTC 120 Mouse 61 ...... C.A...... A...... T...... G...... T...... 120 Angus 61 .C....C.....CGAA...... G...... G...... CC...... 120 Holstein 61 .C....C.....CGAA...... G...... G...... CC...... 120 Jersey 61 .C....C.....CGAA...... G...... G...... CC...... 120 Hanwoo 61 .C....C.....CGAA...... G...... G...... A...CC...... 120

130 140 150 160 170 180 ....|....|....|....|....|....|....|....|....|....|....|....| Human 121 GGCCTGATGGAGACTGTGTGCAGCCCCTTCACGGCCGCCAGACGTCTGCGGGACCAGGAG 180 Mouse 121 .....AG.T.....G...... T.....A...... C...... C.....A... 180 Angus 121 ...... G.C...... G...... G...... G.C..C.G...... G.GC... 180 Holstein 121 ...... G.C...... G...... G...... G.C..C.G...... G.GC... 180 Jersey 121 ...... G.C...... G...... G...... G.C..C.G...... G.GC... 180 Hanwoo 121 ...... G.C...... G...... G...... G.C..C.G...... G.GC... 180

190 200 210 220 230 240 ....|....|....|....|....|....|....|....|....|....|....|....| Human 181 GCAGCCGTGGCGGAGCTGCAGGCCGCCCTGGAGCGACAGGCTCTCCAGAAGCAAGCCCTG 240 Mouse 181 ..T..A....T...... G..AA..GTGT..A.AG...T.C.....C.....G...... 240 Angus 181 ..GA..C..A.C...... G...... G..G.------G..TCG....C. 225 Holstein 181 ..GA..C....C...... G...... G..G.------G..TCG....C. 225 Jersey 181 ..GA..C....C...... G...... G..G.------G..TCG....C. 225 Hanwoo 181 ..GA..C....C...... G...... G..G.------G..TCG....C. 225

250 260 270 280 290 300 ....|....|....|....|....|....|....|....|....|....|....|....| Human 241 CAGGAGAAAGGCAAGCAGCAGGACACGGTCCTCGGCGGCCGGGCCCTGTCCAACCGGCAG 300 Mouse 241 .T..CAGG...... GCA.....GG..AC...GT..A...... CT...... 300 Angus 226 .G....C...... C..T...GG.A....AG...T...... T. 285 Holstein 226 .G....C...... C..T...GG.A....AG...T...... T. 285 Jersey 226 .G....C...... C..T...GG.A....AG...T...... T. 285 Hanwoo 226 .G....C...... C..T...GG.A....AG...T...... T. 285

310 ....|....|.. Human 301 CACGCCTCCTAG 312 Mouse 301 ...... T..A 312 Angus 286 ...... 297 Holstein 286 ...... 297 Jersey 286 ...... 297 Hanwoo 286 ...... 297

Figure 5.1. Alignments of the entire coding sequence of G0S2 cDNA from Human (GenBank accession NM_015714.3), Mouse (GenBank accession NM_008059.3), Angus (GenBank accession JN_935375), Holstein (GenBank accession JN_935376), Jersey (GenBank accession JN_935377) and Hanwoo (GenBank accession JN_935378) cattle. Nucleotide sequences identical to the human G0S2 are indicated by dots. Nucleotide differences are highlighted with black. The dashes represent deletions of the nucleotide sequences. Two single nucleotide polymorphisms (SNPs) of Hanwoo and Angus at nucleotide 109 and 190 are indicated by filled arrows.

Figure 5.2. Multiple alignments of the entire coding sequence of CGI-58 cDNA from Human (GenBank accession AF_151816.1), Mouse (GenBank accession NM_026179.2), Angus (GenBank accession JX_446395), Holstein (GenBank accession JX_446396), Jersey (GenBank accession: KF_539976) and Hanwoo (GenBank accession JX_446397) cattle. Dots denote nucleotide residues identical to the human CGI-58. Differences in the sequence are highlighted with black and sequence deletions are represented by dashes. Filled arrows indicate single nucleotide polymorphisms (SNPs) of Angus and Hanwoo at nucleotide 173 and 368.

100

Figure 5.2.

10 20 30 40 50 60 70 80 90 100 110 120 130 140 ....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....| Human 1 ------ATGGCGGCGGAGGAGGAGGAGGTGGACTCTGCCGACACCGGAGAGAGGTCAGGATGGCTAACTGGTTGGCTCCCCACATGGTGCCCTACGTCTATATCACACCTTAAAGAAGCTGAAGAGAAGATGTTAAAATG 134 Mouse 1 ATGAAAGC.AT.....C...... G..A...G....T.GAG...... G..A..G.....T..T..C.....T..C..A....C...... A...... 140 Angus 1 ------...... A...... T.G....A....G..G...G.GA.C...... T...... A..A.CC...... A..T...... 134 Holstein 1 ------...... A...... T.G....A....G..G...G.GA.C...... T...... A..A.CC...... A..T...... 134 Jersey 1 ------...... A...... T.G....A....G..G...G.GA.C...... T...... A..A.CC...... A..T...... 134 Hanwoo 1 ------...... A...... T.G....A....G..G...G.GA.C...... T...... A..A.CC...... A..T...... 134

150 160 170 180 190 200 210 220 230 240 250 260 270 280 ....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....| Human 135 TGTGCCTTGCACATACAAAAAAGAACCTGTTCGTATATCTAATGGAAATAAAATATGGACACTGAAGTTCTCTCATAATATTTCAAATAAGACTCCACTTGTCCTTCTCCATGGTTTTGGAGGAGGTCTTGGGCTCTGGG 274 Mouse 141 ...C..C.....T.....G.....G.....G..C.....C...... C.G...... G....T...... C..C.....T.G...... G...... C..T...... A..T.... 280 Angus 135 ...C..C...... G...... C...... ---...... G...... C...... A...... 271 Holstein 135 ...C..C...... A...... G...... C...... ---...... G...... C...... A...... 271 Jersey 135 ...C..C...... G...... C...... ---...... G...... C...... A...... 271 Hanwoo 135 ...C..C...... A...... G...... C...... ---...... G...... C...... A...... 271

290 300 310 320 330 340 350 360 370 380 390 400 410 420 ....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....| Human 275 CACTGAATTTTGGAGATCTTTGCACCAACAGACCTGTCTATGCTTTTGACCTATTGGGTTTTGGACGAAGTAGTAGACCCAGGTTTGACAGTGATGCAGAAGAAGTGGAGAATCAGTTTGTGGAATCCATTGAAGAGTGG 414 Mouse 281 .C...... A...... AA.....G.T..G...... C...... C..C...A...... T...... G...... 420 Angus 272 ...... C...... T...... GC....C...... C...... A...... C...... 411 Holstein 272 ...... C...... T...... GC....C...... C...... A...... C...... 411 Jersey 272 ...... C...... T...... GC....C...... C...... A...... C...... 411 Hanwoo 272 ...... C...... T...... GC....C...... C...... A....T...... C...... 411

430 440 450 460 470 480 490 500 510 520 530 540 550 560 ....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....| Human 415 AGATGTGCCCTAGGATTGGACAAAATGATCTTGCTTGGGCACAACCTAGGTGGATTCTTGGCTGCTGCTTACTCGCTGAAGTACCCATCAAGGGTTAATCATCTCATTTTAGTGGAGCCTTGGGGTTTCCCTGAACGACC 554 Mouse 421 ...... CA.G...... A...... G..A..G...... C...... A...... G...C...... A.....A...... T.....G..... 560 Angus 412 C...... T...C...... A...... G...C...... G...... A...... C...... G...... TT...... 551 Holstein 412 C...... T...C...... A...... G...C...... G...... A...... C...... G...... TT...... 551 Jersey 412 C...... T...C...... A...... G...C...... G...... A...... C...... G...... TT...... 551 Hanwoo 412 C...... T...C...... A...... G...C...... G...... A...... C...... G...... TT...... 551

570 580 590 600 610 620 630 640 650 660 670 680 690 700 ....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....| Human 555 AGACCTTGCTGATCAAGACAGACCAATTCCAGTTTGGATCAGAGCCTTGGGAGCAGCATTGACTCCCTTTAACCCTTTAGCTGGCCTAAGGATTGCAGGACCCTTTGGTTTAAGTCTAGTGCAGCGTTTAAGGCCTGATT 694 Mouse 561 ...T...... G...... G...C.A..G...... C..G...... C...... T.....G...... G...... 700 Angus 552 ...... G...... G...... C...... G..A...... A...... 691 Holstein 552 ...... G...... G...... C...... G..A...... A...... 691 Jersey 552 ...... G...... G...... C...... G..A...... A...... 691 Hanwoo 552 ...... G...... G...... C...... G..A...... A...... 691

710 720 730 740 750 760 770 780 790 800 810 820 830 840 ....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....| Human 695 TCAAACGAAAGTATTCTTCAATGTTCGAAGACGATACTGTGACAGAATACATCTACCACTGTAATGTGCAGACTCCAAGTGGTGAGACAGCTTCAAAGAATATGACTATTCCTTATGGATGGGCAAAAAGGCCAATGCTC 834 Mouse 701 ....G..G.....C..C..T.....T.....T..C..G...... G...... A..A..C...... TC..A..C.....G...... G.....C...C...... T 840 Angus 692 ...... G...... T..C..T.....A.....G...... C...... T..G.TC..A..C...... C...... 831 Holstein 692 ...... G...... T..C..T.....A.....G...... C...... T..G.TC..A..C...... C...... 831 Jersey 692 ...... G...... T..C..T.....A.....G...... C...... T..G.TC..A..C...... C...... 831 Hanwoo 692 ...... G...... T..C..T.....A.....G...... C...... T..G.TC..A..C...... C...... 831

850 860 870 880 890 900 910 920 930 940 950 960 970 980 ....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|....| Human 835 CAGCGAATTGGTAAAATGCACCCTGACATTCCAGTTTCAGTGATCTTTGGCGCCCGATCCTGCATAGATGGCAATTCTGGCACCAGCATCCAGTCCTTACGACCACATTCATATGTGAAGACAATAGCTATTCTTGGGGC 974 Mouse 841 .....G..A...GGCT....T...... A...... C.....A...... AC.G.....GA.G..C..C...... T..C..C..C..... 980 Angus 832 ...... C...... C...... C...... G...... G...... AC...... G.....C...... G.....C.....G..A.. 971 Holstein 832 ...... C...... C...... C...... G...... G...... AC...... G.....C...... G.....C.....G..A.. 971 Jersey 832 ...... C...... C...... C...... G...... G...... AC...... G.....C...... G.....C.....G..A.. 971 Hanwoo 832 ...... C...... C...... C...... G...... G...... AC...... G.....C...... G.....C.....G..A.. 971

990 1000 1010 1020 1030 1040 1050 ....|....|....|....|....|....|....|....|....|....|....|....|....|....|....|. Human 975 AGGACATTATGTATATGCAGATCAACCAGAAGAATTCAACCAGAAAGTAAAGGAGATCTGCGACACTGTGGACTGA 1050 Mouse 981 G..G...... G...... G...... C...... C....A..A...... 1056 Angus 972 ...C...... G...... G..G.....C...... G...... T...... 1047 Holstein 972 ...C...... G...... G..G.....C...... G...... T...... 1047 Jersey 972 ...C...... G...... G..G.....C...... G...... T...... 1047 Hanwoo 972 ...C...... G...... G..G.....C...... G...... T...... 101 1047

10 20 30 40 50 60 ....|....|....|....|....|....|....|....|....|....|....|....| Human 1 METVQELIPLAKEMMAQKRKGKMVKLYVLGSVLALFGVVLGLMETVCSPFTAARRLRDQE 60 Mouse 1 ..S...... PR..L...... V...... S...... 60 Angus 1 ...... L....PSA.L.RM....G...... A...... A..G..A..G.R.ER. 60 Holstein 1 ...... L....PSA.L.RM....G...... A...... A..G..A..G.R.ER. 60 Jersey 1 ...... L....PSA.L.RM....G...... A...... A..G..A..G.R.ER. 60 Hanwoo 1 ...... L....PSA.L.RM....G.....SA...... A..G..A..G.R.ER. 60

70 80 90 100 ....|....|....|....|....|....|....|....|... Human 61 AAVAELQAALERQALQKQALQEKGKQQDTVLGGRALSNRQHAS 103 Mouse 61 ...V..RE.C.Q.S.H....LAG..A.EAT.CS....L..... 103 Angus 61 .TLT..R..RGES.PRE.-----..PLEA.Q.C...... L... 98 Holstein 61 .TL...R..RGES.PRE.-----..PLEA.Q.C...... L... 98 Jersey 61 .TL...R..RGES.PRE.-----..PLEA.Q.C...... L... 98 Hanwoo 61 .TL...R..RGES.PRE.-----..PLEA.Q.C...... L... 98

Figure 5.3. Homology analysis of the deduced G0S2 amino acid (AA) sequences from mRNA of Human (GenBank accession NP_056529), Mouse (GenBank accession NP_032085), Angus (GenBank accession AFS_34512), Holstein (GenBank accession AFS_34513), Jersey (GenBank accession AFS_34514) and Hanwoo (GenBank accession AFS_34515) cattle. Residues identical to the human sequence are denoted by dots. Different and similar residues are highlighted by black and grey, respectively. Deletions of sequences are denoted by a dash. The square box from residue 27 to 42 represents the conserved hydrophobic domain among the species. Two amino acid variations in Hanwoo and Angus are indicated by filled arrows on residues 37 and 64.

102

Figure 5.4. Comparison for homology of the deduced CGI-58 AA sequences from mRNA of Human (GenBank accession AF_151816.1), Mouse (GenBank accession NM_026179.2), Augus (GenBank accession AGM_61311.1), Holstein (GenBank accession AGM_61312.1), Jersey (Deduced from GenBank accession KF_539976) and Hanwoo (GenBank accession AGM_61313.1). Residues denoted by dots are identical to the human sequence. Different residues are emphasized with black, and similar residues are highlighted by grey. Dashes represent sequence deletions. The square box from residue 153 to 157 (GXNXG) represents a replacement of serine (S) of GXSXG esterase/lipase motif to arginine (N). The square box with dashed lines from residue 329 to 334 indicates conserved HXXXXD (HX4D) motif for lysophosphatidic acid acyltransferase (LPAAT) function. The filled arrows indicate amino acid variations among the cattle breeds. The open arrows show lipophilic tryptophan-rich region in the N-terminal end. The α/β hydrolase domain is underlined from the residue 50 to 350.

103

Figure 5.4.

10 20 30 40 50 60 ....|....|....|....|....|....|....|....|....|....|....|....| Human 1 --MAAEEEEVDSADTGERSGWLTGWLPTWCPTSISHLKEAEEKMLKCVPCTYKKEPVRIS 58 Mouse 1 MKAM.A...... A.GG...... T...... 60 Angus 1 --.....DG.N...AS...... T...... I...... 58 Holstein 1 --.....DG.N...AS...... T...... I...... H.. 58 Jersey 1 --.....DG.N...AS...... T...... I...... 58 Hanwoo 1 --.....DG.N...AS...... T...... I...... H.. 58

70 80 90 100 110 120 ....|....|....|....|....|....|....|....|....|....|....|....| Human 59 NGNKIWTLKFSHNISNKTPLVLLHGFGGGLGLWALNFGDLCTNRPVYAFDLLGFGRSSRP 118 Mouse 61 ...R....M...... S...... E..S.D...... 120 Angus 59 ...... L..-...... 117 Holstein 59 ...... L..-...... 117 Jersey 59 ...... L..-...... 117 Hanwoo 59 ...... L..-...... 117

130 140 150 GXNXG 160 170 180 ....|....|....|....|....|....|....|....|....|....|....|....| Human 119 RFDSDAEEVENQFVESIEEWRCALGLDKMILLGHNLGGFLAAAYSLKYPSRVNHLILVEP 178 Mouse 121 ...... R...... S...... 180 Angus 118 ...... G...... S...... 177 Holstein 118 ...... G...... S...... 177 Jersey 118 ...... G...... S...... 177 Hanwoo 118 ..V...... G...... S...... 177

190 200 210 220 230 240 ....|....|....|....|....|....|....|....|....|....|....|....| Human 179 WGFPERPDLADQDRPIPVWIRALGAALTPFNPLAGLRIAGPFGLSLVQRLRPDFKRKYSS 238 Mouse 181 ...... E...... 240 Angus 178 ...S...... E...... 237 Holstein 178 ...S...... E...... 237 Jersey 178 ...S...... E...... 237 Hanwoo 178 ...S...... E...... 237

250 260 270 280 290 300 ....|....|....|....|....|....|....|....|....|....|....|....| Human 239 MFEDDTVTEYIYHCNVQTPSGETASKNMTIPYGWAKRPMLQRIGKMHPDIPVSVIFGARS 298 Mouse 241 ...... F...... GL...... 300 Angus 238 ..D...... F...... 297 Holstein 238 ..D...... F...... 297 Jersey 238 ..D...... F...... 297 Hanwoo 238 ..D...... F...... 297

310 320 330HX 4D 340 350 ....|....|....|....|....|....|....|....|....|....|. Human 299 CIDGNSGTSIQSLRPHSYVKTIAILGAGHYVYADQPEEFNQKVKEICDTVD 349 Mouse 301 ...... K...... H... 351 Angus 298 .M..D...... Q...... D...... 348 Holstein 298 .M..D...... Q...... D...... 348 Jersey 298 .M..D...... Q...... D...... 348 Hanwoo 298 .M..D...... Q...... D...... 348

104

1000 bp CGI-58 850 bp * 650 bp AS I 500 bp AS II B ATG Stop AS I 1 2 3 4 5 6 7

ATG Stop AS II

1 2 3 4 5 6 7 Full length protein 1 4 5 6 7 AS I protein 1 5 AS II protein – Premature stop codon

Figure 5.5. Alternative splicing patterns of bovine CGI-58. A. The mRNA expressions of full-length CGI-58 and alternatively spliced variants (ASI and ASII) were detected by end-point PCR after reverse transcription. Based on sequencing, the * mark indicates nonspecific bands. B. Schema of alternative splicing patterns of ASI (long dashed line) and ASII (short dashed line). The 7 exons are illustrated by squares in between introns represented by the straight line. ASI protein is shorter than the full-length protein due to exon 2 and 3 deletion. ASII protein is shorter than both full-length and ASII proteins because of exon 2 – 4 deletion and ORF shift followed by occurrence of premature stop codon.

105

Figure 5.6. Tissue distribution of bovine G0S2 (a) and CGI-58 (b) mRNA. Total RNAs were isolated from adipose tissue (Ad), heart (H), muscle (M), spleen (Sp), lung (Lu), liver (Li), and kidney (K) of Angus cattle. The mRNA expression was measured by qPCR (n = 4). Bovine cyclophilin (CYC) was used as a housekeeping gene for normalization. Bars indicate means ± SEM. Statistically significant differences are marked by the three letters, a, b and c (P < 0.05). One-way ANOVA followed by Tukey‘s post hoc test was conducted to compare the means.

106

Figure 5.7. The mRNA expressions of FABP4, G0S2 and CGI-58 in the bulls and the steers of Hanwoo. A, C and E. FABP4, G0S2 and CGI-58 mRNA expressions in the pure muscle and intramuscular fat (IMF) portions separated from the intact longissimus dorsi muscle (LM) of the steers. B, D and F. FABP4, G0S2 and CGI-58 mRNA expressions in the LM muscle of the bulls and the steers. The mRNA expression levels were measured by qPCR (n = 10) and normalized to β-actin, a housekeeping gene. Means ± SEM is indicated by bars. Significant difference from the GLM procedure of SAS is shown as * P < 0.05, ** P < 0.01 and *** P < 0.001.

107

Figure 5.7.

108

Chapter 6: Identification of the avian RBP7 gene as a new adipose-specific gene and

RBP7 promoter-driven GFP expression in adipose tissue of transgenic quail

Jinsoo Ahn1, 2, Sangsu Shin1, 3, Yeunsu Suh1, Ju Yeon Park4, Seongsoo Hwang1, 5,

and Kichoon Lee1, 2, *

1Department of Animal Sciences and 2Interdisciplinary Ph.D. Program in Nutrition, The

Ohio State University, Columbus, OH, USA 3Life and Industry Convergence Research

Institute, Pusan National University, Miryang, Gyeongnam, Republic of Korea

4Department of Biological Sciences, The Ohio State University, Columbus, OH, USA

5Animal Biotechnology Division, National Institute of Animal Science, Suwon,

Gyeonggi, Republic of Korea

6.1. Abstract

The discovery of an increasing number of new adipose-specific genes has significantly contributed to our understanding of adipose tissue biology and the etiology of obesity and its related diseases. In the present study, comparison of gene expression profiles among various tissues was performed by analysis of chicken microarray data, leading to identification of RBP7 as a novel adipose-specific gene in chicken. Adipose-specific expression of RBP7 in the avian species was further confirmed at the protein and mRNA levels. Examination of the transcription factor binding sites within the chicken RBP7

109 promoter by Matinspector software revealed potential binding sites for adipogenic transcription factors. This led to the hypothesis that the RBP7 promoter can be utilized to overexpress a transgene in adipose tissue in order to further investigate the function of a transgene in adipose tissue. Several lines of transgenic quail containing a green fluorescent protein (GFP) gene under the control of the RBP7 promoter were generated using lentivirus-mediated gene transfer. The GFP expression in transgenic quail was specific to adipose tissue and increased after adipocyte differentiation. This expression pattern was consistent with endogenous RBP7 expression, suggesting the RBP7 promoter is sufficient to overexpress a gene of interest in adipose tissue at later developmental stages. These findings will lead to the establishment of a novel RBP7 promoter cassette which can be utilized for overexpressing genes of interest in adipose tissue in vivo to study the function of genes in adipose tissue development and lipid metabolism.

Keywords: promoters; retinol binding protein 7 (RBP7); green fluorescent protein (GFP); microarray; adipose-specific; adipose tissue development; transgenic quail

110

6.2. Introduction

Adipose tissue is a highly interactive organ that stores energy in the body to sustain energy balance. Identification of adipose-specific genes and use of their promoters that can drive expression of genes in adipose tissue of transgenic animal models are of particular interest in studying functions of genes in adipose growth and development in vivo. The results from the studies using transgenic animal models could provide in vivo evidence that ultimately leads to the potential application to ameliorate obesity and to improve production efficiency of food animals. However, it is difficult to adapt conventional promoters to modulate adipocyte gene expressions selectively in vivo for medical and agricultural purposes.

Retinol binding proteins (RBPs) are a family of transporters for fat-soluble retinol

(preformed vitamin A), which has diverse functions in vision, reproduction, and animal growth (Kanai et al., 1968). Among them, retinol binding protein (RBP7) is one of the cellular retinol binding proteins that play a role in cellular metabolism of retinol (Ross,

1993). In our microarray and real-time PCR data using chicken tissue samples, expression of the avian RBP7 gene was predominant in adipose tissue, which is a major reservoir for retinol. Further promoter analysis has shown that the RBP7 promoter contains regulatory elements for adipose-specific expression. Thus, the RBP7 promoter has been proposed as a novel adipose tissue-specific promoter. However, to our knowledge, there have been no reports on any transgenic livestock animal and poultry with adipose-specific expression of a transgene, beside our recent transgenic quail studies

(Shin et al., 2014).

111

In this study, RBP7 was identified as an adipose-specific gene in the avian species and transgenic quail expressing an enhanced green fluorescent protein (eGFP) gene in adipose tissue under the control of RBP7 promoter was generated in order to selectively target the transgene (eGFP) expression in adipose tissue. In addition to its adipose tissue specificity, the regulation of the transgene during the development of adipocytes was investigated. It has been suggested that the RBP7 promoter may stimulate GFP expression very strongly in the neck and abdominal adipose tissue of quail. Moreover,

GFP expression was significantly high in mature adipocytes, indicating a late-temporal control by the RBP7 promoter. The avian transgenesis used in this study to produce adipose-specific fluorescent transgenic quail can be utilized to drive spatiotemporally controlled overexpression of target transgenes to discover the function of unknown genes during adipose tissue development in vivo.

6.3. Materials and Methods

Animal use

All procedures for animal care and use were approved by The Ohio State University

Institutional Animal Care and Use Committee (IACUC). Quail, turkeys and chickens were maintained at the Ohio State Poultry House, The Ohio State University, Columbus,

OH. For tissue collection, birds were euthanized by CO2 inhalation.

Total RNA isolation and microarray analysis

Total RNA was isolated from chicken tissues using Trizol (Life Technologies Inc., Grand

Island, NY) according to the manufacturer‘s instructions. RNA quality and quantity were

112 assessed by gel electrophoresis and NanoDrop (NanoDrop Technologies, Wilmington,

DE), respectively. RNA samples for each tissue or organ, including fat, muscle, heart, lung, liver, kidney, brain, and spleen were pooled from three chickens and combined as one sample per tissue or organ to decrease individual variability. The pooled RNA samples were frozen and sent out to The Ohio State University Microarray Core Facility.

These RNA samples were reverse-transcribed into cDNA and labeled. The labeled cDNAs were hybridized to the Affymetrix GeneChip Chicken Genome array chip which covers 32,773 transcripts corresponding to 28,000 chicken genes. The raw data were gathered and converted by Dr. Lianbo Yu from the Center for Biostatistics at The Ohio

State University. Background correction and quantile normalization of the data were then conducted. The microarray results were returned as a relative gene expression per sample.

The microarray data saved in an Excel file were further processed for ranking genes that are abundantly expressed in adipose tissue by the procedures described in our previous report (Song et al., 2013).

Quantitative Real-Time PCR

To confirm adipose-specific expression of RBP7 that was most highly expressed in adipose tissue based on an analysis of our microarray data, relative mRNA expression levels of RBP7 in various chicken tissues (n=4) were assessed by quantitative real-time

PCR (qPCR). In addition, for the quantification of gene expression in fat cell fractions, subcutaneous adipose tissue was collected from chicken (n=4) and quail (n=3) and the fractionation of stromal-vascular cell (SV) and fat cell (FC) was performed as described previously (Ahn et al., 2013). Both various chicken tissues and SV and FC fractions were

113 gathered for RNA isolation using Trizol (Life Technologies Inc.). After conducting RNA isolation, RNA concentration was checked by nanodrop and 1 µg of total RNA for each sample was used for reverse transcription (RT) in 20 µL reaction following the instruction from the manufacturer of Moloney murine leukemia virus l(M-MLV) reverse transcriptase (Invitrogen, Carlsbad, CA). The conditions for RT were 65 °C for 5 min,

37 °C for 52 min, and 79 °C for 15 min. For qPCR, 2 µL of cDNA out of the 20 µL RT reaction were used as templates with the reaction system including AmpliTaq Gold polymerase (Applied Biosystems, Carlsbad, CA), GeneAmp 10×PCR Buffer (containing

100 mM Tris-HCl, pH 8.3), and 500-mM KCl. SYBR green was used to detect the amplification of products. Duplicate reactions were performed (25μL volume) on an ABI

7300 Real-Time PCR Instrument (Applied Biosystems). The condition of qPCR reaction was 95°C for 10 min, 40 cycles of 94°C for 15s, 57°C for 40s and 72°C for 30s, with an additional 82°C extension for 33s. A single major peak on the dissociation curves provided by the qPCR software indicated successful amplification. Chicken β-actin was used as a reference gene for the normalization of expression levels of RBP7 and other chicken genes. For the quantification of mRNA expression in quail, ribosomal protein

S13 (RPS13) was selected as a gene for normalization control, because of its stability described in previous reports (de Jonge et al., 2007; Serr et al., 2011). Relative expression

-ΔΔ of target genes was calculated using the 2 CT (CT: threshold cycle) method (Livak &

Schmittgen, 2001). The primer sequences for the chicken DLK1, β-actin and RPS13 were described in our previous reports (Lee et al., 2009; Yang et al., 2013). Other sets of primers used in this study are as follows: avian FABP4 forward primer 5‘- GGG CAC

CTG GAA GCT CCT T-3‘and reverse primer 5‘-TCT CAT CAA ACT CTT CAC CCA 114

GCT-3‘; avian SCD1 forward primer 5‘- GAA CAT CAA CCC ACG GGA GAA-3‘ and reverse primer 5‘-AGC CCC AGG AGG CAC ATG A-3‘; chicken PPARγ forward primer 5‘-ACA TAA AGT CCT TCC CGC TGA CC-3‘ and reverse primer 5‘-TCC AGT

GCG TTG AAC TTC ACA GC-3‘; avian RBP7 forward primer 5‘-CCC ACA GTC TAG

CAA TGC CTG T-3‘ and reverse primer 5‘-GTT TCC ATG TTG GAA CCA ATG CT-

3‘; eGFP forward primer 5‘-GCA TGG ACG AGC TGT ACA AGT AA-3‘ and reverse primer 5‘-CAT AAA GAG ACA GCA ACC AGG ATT-3‘.

Antibody production and Western blot analysis

An avian RBP7 polyclonal antibody was raised against an immunogen including amino acids 73–83 residues of chicken RBP7 protein in the rabbit by a custom antibody service

(AbClon, Seoul, Korea). Protein isolation and Western blot analysis were conducted following the procedures from our previous report (Li et al., 2012). In detail, after SDS-

PAGE and transfer to polyvinylidene fluoride (PVDF) membrane, the membrane was incubated overnight at 4°C with the RBP7 custom primary antibody at 1:3000 dilution in

1x Tris-buffered saline containing 0.05% tween-20 (TBST) with 4% non-fat dry milk.

After washing in 1x TBST, Western blots were incubated with Rabbit IgG HRP- conjugated secondary antibody (R&D systems Inc., Minneapolis, MN) for 1 h at room temperature. The membrane was washed with 1x TBST before the addition of Amersham

ECL plus Western Blotting Detection Reagents (GE Healthcare Biosciences, Pittsburgh,

PA), and the blots were then exposed to Hyperfilm (GE Healthcare Biosciences) to visualize target proteins.

115

Prediction of transcription factor binding sites

The DNA sequence of 5 kb upstream region from the start codon of the chicken RBP7

(cRBP7) gene was obtained from the chicken genome browser at the UCSC (University of California, Santa Cruz) Genome Bioinformatics site (http://genome.ucsc.edu). With the DNA sequence, the binding sites of adipose tissue-specific transcription factors were predicted by using the MatInspector program (Genomatix Software GmbH, Munich,

Germany). Among the 5 kb sequence, the 3 kb sequence from the start codon was used for a promoter as it had many binding sites of adipose tissue-specific transcription factors including PPARγ and C/EBPs.

Cloning of cRBP7 promoter followed by lentiviral vector and particle production

The 3 kb sequence in the 5‘ upstream region of cRBP7 gene was amplified by PCR with a primer set of cRBP7P3K-F1 with ClaI site (forward primer, 5‘-CGG TTA TCG ATG

CAA TCA AAA TGC CAC TGA A) and cRBP7P3K-R1 with PacI site (reverse primer,

5‘-CGG TTT TAA TTA ATG CTA GAC TGT GGG AAG GAG TTA-3‘), and cloned into pCR2.1-TOPO vector (Invitrogen). The pCR2.1 recombinant vector was then digested with two restriction enzymes, ClaI and PacI, producing the 3 kb promoter fragment. This 3 kb promoter replaced a RSV promoter of a previously constructed pLTReGW lentiviral vector containing eGFP (Shin et al., 2008) after removing the RSV promoter from the pLTReGW lentiviral vector with the same restriction enzymes, ClaI and PacI. The final vector designed to express eGFP gene specifically in adipose tissue being driven by the RBP7 promoter was named as pLT-RBP7p3k-eGFP. Lentiviral particles were produced by co-precipitation of calcium phosphate and pLT-RBP7p3k-

116 eGFP vector. In brief, on the day before transfection, 293 FT cells were plated on 100 mm culture dishes in the complete medium, which consisted of Dulbecco‘s Modified

Eagle Medium (DMEM; Life Technologies Inc.) supplemented with 10% fetal bovine serum (FBS; Life Technologies Inc.), 1% penicillin/streptomycin (pen/strep; Life

Technologies Inc.), 0.1 mM MEM non-essential amino acids (Life Technologies Inc.), and 1 mM MEM sodium pyruvate (Life Technologies Inc.). To prepare transfection solution, 9 µg of pLT-RBP7p3k-eGFP, 9 µg of ViraPower Packaging Mix (Life

Technologies Inc.), and 87 µl of 2M calcium solution (Clontech Laboratories Inc.,

Mountain View, CA) were added to a final volume of 700 µl of Sterile H2O (Clontech

Laboratories Inc.) and then 700 µl of 2 HEPES-Buffered Saline (HBS) (Clontech

Laboratories Inc.) were added dropwise while vortexing slowly. The transfection solution was then incubated at room temperature for 5 min and subsequently added dropwise to the complete medium. After 10 h of transfection, the medium was replenished with 5 ml of fresh complete medium. The supernatant was collected after 48 h and filtered through

0.22 µm pore size filters. The titer of lentiviral supernatants was measured by a standard

ELISA method using the Lenti-X p24 Rapid Titer Kit (Clontech Laboratories Inc.) after the non-concentrated viral supernatants were serially diluted, which resulted in 1‒10

IFU/ml of an end point titer (data not shown). The filtered supernatant was then pelleted by centrifugation at 25,000 rpm for 2 h with an ultracentrifuge (L7-65R,

Beckman Coulter, Fullerton, CA), resuspended in Opti-MEM as a 100 concentrated lentiviral particle soup, and stored as 40 µl aliquots at ‒80°C until use.

Lentivirus-mediated gene transfer into stage X embryos 117

Newly laid eggs from wild-type Japanese quail were collected and cleaned with 70% ethanol. Eggs were positioned with their sides facing upward for approximately 4 h at room temperature to move embryos to the side. A window of about 5 mm in diameter was then made on the side using fine tweezers and 2-3 µl of the 100 concentrated lentiviral particle soup was microinjected into the subgerminal cavity of stage X embryos with a fully formed area pellucida (Eyal-Giladi & Kochav, 1976). The window was then sealed twice with paraffin film, and the eggs were incubated with the pointy end down until hatching.

Hatching and mating after maturation

Of 184 stage X embryos microinjected with lentiviral particles containing pLT-

RBP7p3k-eGFP, 29 quail hatched (15.8%) after 17 days of incubation. Among them, 17 hatchings (9.2%) survived up to 7 weeks until they sexually matured. These mature quail were mated with counter-sex wild-type quail, except for two cases in which two generation 0 (G0) quail were mated to each other, to construct 15 G0 founder lines. As a result, 30 transgenic generation 1 (G1) quail were produced from seven G0 founder lines.

The success rate of transgenesis was between 5.1 % and 14.3% for the seven lines (Table

1).

Detection of the transgene in quail by PCR

To confirm the integration of the transgene in G1 and generation 2 (G2) quail, genomic

DNA was extracted from the blood of hatched quail and subjected to genotyping PCR.

Two primer sets of forward primers (5‘- AAT CGC AAA ACC AGC AAG AAA -3‘ and

118

5‘- TGA CTG TAA CTC CTT CCC ACA GTC -3‘) and reverse primers (5‘- TTC AGT

GGC ATT TTG ATT GCA -3‘and 5‘- TGA ACT TGT GGC CGT TTA CGT -3‘) for

~300 bp fragments were used to detect the transgene construct by PCR. The PCR was performed in a volume of 25 µl, containing 200 ng of genomic DNA, 10 mM dNTP, 2.5

µl of 10 reaction buffer, 0.4 µM of forward and reverse primers, and 0.4 U of Taq DNA polymerase by using a MJ Research PTC-200 thermal cycler (MJ Research Inc., South

San Francisco, CA). Cycle parameters for this PCR consisted of 2 min at 95°C, followed by 35 cycles of denaturation at 94°C for 30 sec, annealing at 55°C for 40 sec, extension at

72°C for 1 min, and a final extension of 10 min at 72°C. For negative and positive controls, genomic DNA extracted from wild-type quail and 4 pg of pLT-RBP7p3k-eGFP plasmid DNA were used as templates, respectively.

Analysis of GFP expression

A total of seven 6-wk-old transgenic G2 quail and two wild-type quail were euthanized by CO2 inhalation, and their organs were collected including neck fat, abdominal fat, muscle, heart, lung, liver, kidney, spleen, brain, intestine, abdominal skin, and wing skin.

For organ imaging of eGFP, its expression in transgenic quail was examined using a long-wave UV lamp (Blak-Ray B 100 AP/R, UVP, Upland CA, USA, radiation range

315−400 nm, peak at 365 nm). For cell culture studies, after primary stromal-vascular cells were cultured and differentiated as described in our previous report (Oh et al., 2011), adipocytes and undifferentiated preadipocytes were visualized under a fluorescence microscope at post-differentiation day 3 and analyzed using AxioVision software (Carl

Zeiss, Göttingen, Germany).

119

Statistical analysis

Statistical analyses were conducted by using statistical software programs as follows:

Comparison of two means was performed by Student‘s t-test. One-way ANOVA built in

SAS 9.4 software program (SAS Institute Inc.) was used to compare multiple means followed by a Tukey‘s post hoc test. P-value was < 0.05 for the minimum level of significance.

6.4. Results

RBP7 is a novel adipose-specific gene discovered in comparative microarray analysis

To identify novel adipose-specific genes, an analysis of chicken microarray was performed comparing expression levels of genes among various tissues. RBP7 was identified as the first-ranked gene among 28,000 chicken genes (data not shown), by showing 238-fold higher expression in chicken adipose tissue compared to an average value in other chicken tissues (Fig. 6.1A). This result was confirmed by qPCR, which showed 120-fold and 150-fold higher RBP7 expression in chicken subcutaneous and abdominal adipose tissue compared to an average level in other chicken tissues (Fig.

6.1B). Western blot analysis further revealed an exclusive expression pattern of the chicken RBP7 (cRBP7) 15.57-kDa protein in neck (or subcutaneous) and abdominal adipose tissue, which is highly consistent in other avian species, quail and turkey (Fig.

6.2B). Predicted amino acid sequence from quail RBP7 nucleotide sequence deposited to

GenBank by this study (GenBank accession number KP026122) exhibited significant similarity to chicken and turkey sequences (Figure 6.6), with only two amino acid

120 differences in both cases, and an epitope sequence detected by our custom RBP7 antibody is conserved among avian species (Fig. 6.2A). According to our comparative study of three avian species, RBP7 was discovered as a novel adipose-specific gene whose mRNA and protein expression is specific to adipose tissue.

A promoter region of RBP7 gene contains sequence elements for adipose-specific expression

In light of specificity of RBP7 for adipose tissue, its promoter region was analyzed to find cis-acting elements that could be involved in adipose development. As predicted by the MatInspector program, the 3kb upstream promoter region from the start codon of the cRBP7 gene was shown to have many cis-acting sequence elements that contain binding sites for PPAR and C/EBP which are major transcription factors responsible for adipose-specific expression of multiple genes during adipose development (Fig. 6.3A). It suggests that the endogenous cRBP7 promoter not only contains regulatory elements for the adipose-specific expression of cRBP7 gene, but also can be used to express target transgenes such as GFP gene in an adipose-specific manner.

Adipose-specific expression of GFP under the control of RBP7 promoter in transgenic quail lines

To prove adipose-specific regulation of RBP7 promoter in vivo, transgenic quail containing cRBP7 promoter–driven GFP transgene were generated. In particular, seven

G1 transgenic lines were obtained as shown in genotyping PCR using primers designed to detect two transgene junctions, RRE-cRBP7 promoter and eGFP-WPRE portions of

121 the lentiviral vector construct (Fig. 6.3B and C). Among the seven transgenic lines, offspring in four lines (G2 transgenic quail) were born normally in a Mendelian ratio and showed strong GFP expression in neck and abdominal adipose tissue under the UV light with relatively even intensity across the different white adipose depots (Fig. 6.4A and B), indicating adipose-specific expression of GFP under the control of RBP7 promoter.

Expression of both endogenous RBP7 gene and GFP transgene increases during development of adipocytes in vivo and in vitro

To analyze RBP7 expression during avian adipogenesis in vivo, mRNA expression of chicken RBP7 in stromal-vascular (SV) cells containing mostly preadipocytes and fat cells was quantified by qPCR. Subcutaneous adipose tissue of the chicken was collected and fractionated into SV cells and fat cells. Successful fractionation was confirmed by examining marker gene expressions. As shown in Figure 6.4A, DLK1, a preadipocyte marker, was expressed 8 times higher in SV cells than in fat cells, and FABP4 and

PPARγ, adipocyte markers were 10 and 3 times higher in fat cells than in SV cells, respectively, indicating effective fractionation. Chicken RBP7 mRNA expression was approximately 17-fold greater in fat cells compared to SV cells (Fig. 6.5A), showing that cRBP7 is expressed higher in mature adipocytes rather than preadipocytes.

To compare expression patterns between the endogenous RBP7 gene and the

RBP7 promoter driven-GFP transgene during in vivo adipose development of transgenic quail, subcutaneous adipose tissue of two G2 transgenic quail lines were fractionated into

SV cells and fat cells (Fig. 6.5B). Adipocyte markers, FABP4 and SCD1 were highly expressed in fat cells compared to SV cells, indicating successful fractionation. RBP7

122 mRNA expression was significantly high in fat cells. Similarly, the target transgene (GFP gene) was also significantly highly expressed in fat cells.

In order to investigate the temporal expression of RBP7 and GFP during the development of adipocytes from transgenic quail in vitro, primary preadipocytes from the two selected transgenic quail lines were cultured and induced to be differentiated to adipocytes. The primary cell culture has shown that GFP expression is much more prominent in adipocytes than in neighboring preadipocytes (Fig. 6.5C) indicating an adipocyte-specific GFP expression.

6.5. Discussion

In general, most major tissues preferentially express approximately 100-500 tissue- specific genes (Yu et al., 2006). Until recently, however, only a few genes have been identified as being adipose tissue-specific. Therefore, the discovery of new adipose- specific genes is essential to understand and advance adipose tissue biology. In the present study, we analyzed a chicken DNA microarray dataset followed by further confirmatory studies as described in our previously established method (Song et al., 2013) to identify adipose-specific genes. This has led us to identify a novel adipose-specific

RBP7 gene and to hypothesize that the RBP7 promoter can be utilized to overexpress a transgene in adipose tissue for future investigation of the function of a transgene in adipose tissue development and lipid metabolism.

Retinol binding protein 7 (RBP7) was identified in this study as a novel adipose- specific gene in the avian species and it is a member of the intracellular lipid-binding protein family. The RBP7 gene has also been named cellular retinol binding protein IV

123

(CRBP IV) in the human and cellular retinol-binding protein, type III (CRBP-III) in the mouse (Vogel et al., 2001; Folli et al., 2002). Because dietary retinol (ROH) is a hydrophobic compound, its esterification to retinyl esters (RE) or combination with water-soluble retinol binding protein (RBP) is required for storage in the aqueous cytosol

(Noy, 2000). It has been shown that RBP1 plays a role in maintaining retinol storage in the liver and RBP2 is involved in retinoid metabolism in the intestine (Vogel et al., 2001).

In addition, murine RBP7 has shown to be associated with retinol uptake and metabolism

(Noy, 2000; Piantedosi et al., 2005) and 15-20 % of total retinoid derivatives including retinols are stored in adipose tissue (Villarroya et al., 2004). Based on these studies, it was assumed that the level of avian RBP7 may be high in adipose tissue where retinols are present with great abundance. Our results regarding adipose-specific expression of

RBP7 support this assumption and RBP7 is a potential adipose-specific retinol-binding protein in the avian species. Furthermore, based on studies showing that retinoids are stored in adipocytes rather than in stromal-vascular fraction (Tsutsumi et al., 1992), higher expression of avian RBP7 in adipocytes than in stromal-vascular cells (Fig. 6.5A) suggests the role of RBP7 on retinol metabolism in mature adipocytes rather than preadipocytes.

Considering adipose-specific expression of avian RBP7, it was hypothesized that the chicken RBP7 promoter may drive the expression of a transgene in an adipose- specific manner. Transgene expression targeting adipose tissue in transgenic mouse models have been achieved using the adipose protein 2 (aP2/FABP4) promoter (Ross et al., 1990; Graves et al., 1992; Lee et al., 2003; Halberg et al., 2009; Lee et al., 2013).

However, it has been shown that the activity of the aP2 promoter can be found in other

124 cell types, such as macrophages and cardiac myocytes, or other tissues including the heart, lung, and muscle (Lee et al., 2013). In addition, an uneven distribution of expression of target genes across different fat pads has been reported (Halberg et al., 2009). Therefore, there has been an increasing demand for discovering new robust adipose-specific promoters. There have been attempts to generate an alternative adipose-specific promoter cassette such as a 4.9-kb adiponectin promoter cassette, using mouse models (Wang et al.,

2010). However, adiponectin gene expression was not as significant as RBP7 in chicken adipose tissue in our microarray analysis (data not shown), leading to an exclusion of adiponectin promoter cassette for the generation of transgenic avian species. In the current study, the exclusive expression of avian RBP7 protein in adipose tissue along with the presence of several binding sites for adipose-specific transcription factors,

PPAR and C/EBP in the promoter region of RBP7 suggests the RBP7 promoter will be a reliable promoter for driving overexpression of a transgene specifically in adipose tissue.

Quail has been developed as a model system for conducting genetic modifications due to its high egg-laying capacity, a short incubation period, and an earlier sexual maturation (Shin et al., 2014; Poynter et al., 2009). Particularly, Japanese quail (Coturnix c. japonica) have been used to study the function of target genes in adipose tissue in light of the fact that de novo lipogenesis occurring primarily in the liver and secondarily in adipose tissue is conserved in both humans and avian species; whereas, it mainly takes place in adipose tissue in many other animals (Shin et al., 2014; Serr et al., 2009). Also, there is no obvious difference in fatty acid synthesis and breakdown between humans and avian species (Serr et al., 2011; Oh et al., 2011; Bergen & Mersmann, 2005). Therefore, it

125 is useful to generate a transgenic quail model to identify adipose-specific promoters that increase gene expression in adipose tissue and to study the development of adipose tissue in developing quail. In addition, lentiviral vectors have been used for gene therapy and adopted for the generation of transgenic quail, because of their efficient infection mechanisms without further manipulations (Shin et al., 2008; Sato & Lansford, 2013).

Obtaining advantages of lentivirus, transgenic quail could be generated to improve the efficiency of germline integration of GFP reporter gene under the regulation of RBP7 promoter. The expression of GFP was relatively even across different fat pads. Much weaker or no GFP images generated from tissues of some lines may be due to undesirable chromosome integration or an insertion of truncated transgene constructs into a chromosome (Figure 6.7). In addition, during the development of adipose tissue, GFP expression was significantly high in mature adipocytes suggesting the regulatory role of

RBP7 promoter in mature adipocytes.

In summary, the expression of GFP transgene under the regulation of RBP7 promoter follows the adipose-specific tissue distribution pattern of the expression of endogenous RBP7 gene as well as its predominant expression pattern in adipocytes over preadipocytes within adipose tissue. These results suggest that the RBP7 promoter present in the transgene construct used for the generation of transgenic quail contains sufficient information to ensure developmental stage- and tissue-specific activity of the

RBP7 promoter. Fundamental understandings gained from these findings will establish the construct of RBP7 promoter and a transgene as a new tool to overexpress transgenes specifically in adipose tissue in vivo, targeting adipocytes at later stages of development.

It will lead to elucidating the function of a transgene in fat accretion in adipose tissue,

126 thereby providing potential target genes for developing anti-obesity drugs and strategies and for selecting superior lines of livestock with less fat accretion and improved carcass quality.

6.6. Acknowledgment

This project was supported by Agriculture and Food Research Initiative Competitive

Grant no. 2010-65206-20716 from the USDA National Institute of Food and Agriculture, the National Research Foundation of Korea Agenda Program no. PJ009457, and SEEDS:

The Ohio Agricultural Research and Development Center (OARDC) Research

Enhancement Competitive Grants Program no. 2014-090 at The Ohio State University.

127

A 300

250 200 150 RBP7 expression 100

50 Relative Relative 0 1 2 3 4 5 6 7 8

180 *** 160 ***

140

120 actin

- 100 β 80 60

RBP7/ 40 20 0 1 2 3 4 5 6 7 8

Figure 6.1. A) Relative RBP7 mRNA expression values from chicken cDNA microarray (n=3, pooled from three chickens for each tissue). B) Measurement of RBP7 mRNA expression in chicken tissues by qPCR (n=4) which is normalized to β-actin as a housekeeping gene. S-Fat: subcutaneous fat, A-Fat: abdominal fat, T-Mus: thigh muscle, P-Mus: pectoralis major muscle.

128

10 20 30 40 50 60 A ....|....|....|....|....|....|....|....|....|....|....|....| Chicken 1 MPVDFSGTWNLVSNDNFEGYMTALGIDFATRKIAKMLKPQKVIKQDGDSFSIHTTSTFRD 60 Turkey 1 ...... 60 Quail 1 ...... 60 Human 1 ..A.L....T.L.S...... L...... L...... E.N....T...N.SL.N 60 Mouse 1 ..A.L...... L.S...... L...... L...... E.N....T.Q.C.SL.N 60

70 80 90 100 110 120 ....|....|....|....|....|....|....|....|....|....|....|....| Chicken 61 YMLQFKIGEEFEEDNKGLDNRKCKSLVTWDNDKLICIQAGEKKNRGWTHWLEGDDLHLEL 120 Turkey 61 ...... D.....V...... 120 Quail 61 ...... V...... 120 Human 61 .FVK..V....D...R...... I....R.T...K...... I...K....M 120 Mouse 61 .LVK..V...... T.....E....T.V.R...... S..I...Q....M 120

130 ....|....|.... Chicken 121 RCENQVCKQVFKKA 134 Turkey 121 .....I...... R. 134 Quail 121 ...... R. 134 Human 121 F..G.....T.QR. 134 Mouse 121 F..G.....T.QR. 134

B Chicken 15 kDa RBP7

50 kDa α-tubulin SF AF TM PM H Lu Li Ki

Quail 15 kDa RBP7

50 kDa α-tubulin

SF AF M H Lu Li Ki Sp Br Int

Turkey 15 kDa RBP7

50 kDa α-tubulin SF AF TM PM H Lu Li Ki

Figure 6.2. A) Comparison of RBP7 amino acid sequences of chicken (GenBank accession number XP_417606.4), turkey (XP_003212313.1), quail (corresponding to KP026122), human (NP443192.1) and mouse (NP071303.1). An epitope sequence detected by a custom RBP7 antibody is underlined from residue 73 to 83. B) Western blot analysis of RBP7 protein expression in chicken, quail and turkey tissues, respectively. α- tubulin was used as a loading control. SF: subcutaneous fat, AF: abdominal fat, TM: thigh muscle, PM: pectoralis major muscle, H: heart, Lu: lung, Li: liver, Ki: kidney, Sp: spleen, Br: brain, Int: intestine.

129

A RBP7 promoter (3kb) RBP7 gene ATG 5’ Exon1 Exon2 Exon3 Exon4 3’ C/EBP PPAR :RXR PPAR

B ClaI PacI cPPT

LTR RRE cRBP7p3k (~3kb) eGFP WPRE LTR

347 bp 328 bp C K1 K2 K4 K5 K6 K7 K8 WT +

400 bp 300 bp 328 bp

400 bp 300 bp 347 bp

Figure 6.3. A) RBP7 promoter with regulatory elements interacting with major transcription factors in adipose tissue. B) Schematic representation of the pLT-RBP7p3k- eGFP vector. Restriction sites, ClaI and PacI are depicted. LTR, long terminal repeat; , packaging signal; RRE, rev-response element; cPPT, central polypurine tract of HIV-1; eGFP, enhanced green fluorescent protein; WPRE, woodchuck hepatitis virus posttranscriptional regulatory element. The arrows represent the location of forward and reverse primers used for genotyping PCR. C) Genotyping PCR of transgenic and wild- type quail using the two sets of primers depicted in the figure of B). WT, wild-type quail; +: pLT-RBP7p3k-eGFP plasmid DNA (positive control).

130

Figure 6.4. A) GFP expression in selected quail lines under normal and UV light. 1. Neck fat; 2. Abdominal fat; 3. Muscle; 4. Heart; 5. Lung; 6. Liver; 7. Kidney; 8. Spleen; 9. Brain; 10. Intestine; 11. Abdominal skin; and 12. Wing skin. B) Comparison of GFP expression in neck fat and abdominal fat tissues.

131

RBP7 DLK1 FABP4 PPARγ A 25 10 14 3.5 ** ** ** ** 12 3

actin 20 8 -

β 10 2.5 15 6 8 2 10 4 6 1.5 4 1 5 2 2 0.5 expression/ mRNA 0 0 0 0 SV FC SV FC SV FC SV FC eGFP SCD1 FABP4 RBP7 8 50 120 70

B * * * 60 40 ** 100 6 50 80 30 40 4 60 20 30 40 2 20 10 20 10 expression/RPS13 mRNA 0 0 0 0 SV FC SV FC SV FC SV FC eGFP SCD1 FABP4 RBP7 30 100 40 25

*** * * * 25 80 20 30 20 60 15 15 20 40 10 10 10 5 20 5 expression/RPS13 mRNA 0 0 0 0 SV FC SV FC SV FC SV FC

C K5 K7

Phase contrast

GFP

Figure 6.5. A) Quantification of RBP7, DLK1, FABP4, and PPARγ mRNA expression in stromal- vascular (SV) and fat cell (FC) fractions from chicken adipose tissue by qPCR (n=4). β-actin expression was used as a normalization control. B) qPCR analysis of SV and FC fractions from adipose tissue of selected transgenic G2 quail for mRNA expression of GFP, SCD1, FABP4, and RBP7 (n=3). Ribosomal protein S13 (RPS13) was selected as a reference gene. C) Phase contrast and GFP images of primary preadipocytes and adipocytes from K5 and K7 transgenic G2 quail.

132

10 20 30 40 50 60 ....|....|....|....|....|....|....|....|....|....|....|....| Chicken 1 ATGCCTGTGGATTTCAGTGGAACCTGGAACCTTGTCAGTAATGACAACTTCGAAGGCTAC 60 Turkey 1 ...... T...... 60 Quail 1 ...... T...... 60 Human 1 .....C.CC..CC....C..T..T....C...GC....C.GC...... G...... 60 Mouse 1 .....A.CA..CC....C..T...... T...C....C.GC...... G...... 60

70 80 90 100 110 120 ....|....|....|....|....|....|....|....|....|....|....|....| Chicken 61 ATGACAGCTTTAGGTATTGACTTTGCAACACGCAAGATAGCAAAAATGCTGAAGCCTCAG 120 Turkey 61 ...... A...... 120 Quail 61 ...... T...... 120 Human 61 ...CTG..CC...... C..T..T..A.....C..GT...... A... 120 Mouse 61 ...CTG..CC.G...... T..T.....T..C..GT...... A... 120

130 140 150 160 170 180 ....|....|....|....|....|....|....|....|....|....|....|....| Chicken 121 AAAGTGATTAAACAGGATGGAGATTCATTTTCCATCCATACCACAAGCACATTCAGAGAC 180 Turkey 121 ..G.....C...... T...... G...... 180 Quail 121 ...... C...... T.....G...... G...... 180 Human 121 ...... G.G...A....G.....T...A...... C..G.AC....GCC.A..GA.. 180 Mouse 121 ...... G.G..AA....G..C..C..CA...... G..GTGC....GCC....GA.. 180

190 200 210 220 230 240 ....|....|....|....|....|....|....|....|....|....|....|....| Chicken 181 TATATGCTCCAATTTAAAATTGGTGAAGAGTTTGAAGAAGACAACAAAGGCCTGGATAAC 240 Turkey 181 ..C...... A...... T...... T...... 240 Quail 181 ..C...... C...... T...... 240 Human 181 ..CT.TG.GA...... G....A.....A.....T.....T....G...... C... 240 Mouse 181 ..CC.TG.AA....C...G....A...... G..G..T...... 240

250 260 270 280 290 300 ....|....|....|....|....|....|....|....|....|....|....|....| Chicken 241 AGAAAATGCAAGAGCCTGGTTACCTGGGACAATGACAAACTTATCTGCATCCAGGCTGGG 300 Turkey 241 ...... TG...... TG...... 300 Quail 241 .....G...... TG...... 300 Human 241 ...... TT...... T...... GG..C.C...T...... AAG..A 300 Mouse 241 ...... C...... G...... C.CT...G.A...AGA..A 300

310 320 330 340 350 360 ....|....|....|....|....|....|....|....|....|....|....|....| Chicken 301 GAGAAGAAGAACAGGGGCTGGACTCACTGGCTCGAAGGAGATGACCTCCATTTGGAGCTT 360 Turkey 301 ...... 360 Quail 301 ...... T...... C...... 360 Human 301 ..A...... A...... C..T...A...... CA.A.....CC....AA.G 360 Mouse 301 ...... A.....A...... GC...... A...... G..CC.G.....CC....AA.G 360

370 380 390 400 ....|....|....|....|....|....|....|....|....| Chicken 361 CGTTGTGAGAATCAAGTATGTAAGCAAGTATTCAAGAAAGCTTAA 405 Turkey 361 ...... A...... G...... 405 Quail 361 ...... C...... G.....G. 405 Human 361 TTC.....AGG...... G..C..A..GAC....C...G...C.G. 405 Mouse 361 TTC..C..AGGC..G..G..C...... ACC...C...G...C.G. 405

Figure 6.6. Alignments of RBP7 nucleotide sequences of chicken (GenBank accession number XM_417606.4), turkey (XM_003212265.1), quail (KP026122), human (NM052960) and mouse (NM022020.2).

133

Figure 6.7. A) Selected quail lines without GFP expression. 1. Neck fat; 2. Abdominal fat; 3. Muscle; 4. Heart; 5. Lung; 6. Liver; 7. Kidney; 8. Spleen; 9. Brain; 10. Intestine; 11. Abdominal skin; and 12. Wing skin. B) Measurement of GFP expression in neck fat and abdominal fat tissues.

134

G1 G0 founder quail lines G2 (Offspring of G1, K) Male Female (J) Eggs Hatched Transgenic Transgenic quail ID G0-8836 WT J2 167 14 2 (14.3%) K1,K13 G0-9201 WT J4 100 41 5 (12.2%) K3,K8,K12,K29,K30 WT G0-9203 J7 132 78 8 (10.3%) K2,K9,K15,K16,K18,K19,K25,K26 WT G0-8897 J10 88 59 3 (5.1%) K6,K24,K27 WT G0-9204 J11 74 53 5 (9.4%) K7,K11,K17,K20,K28 WT G0-9202 J12 92 42 4 (9.5%) K4,K10,K14,K22 G0-9198 G0-9205 J14 83 42 3 (7.1%) K5,K21,K23

Table 6.1. Production of transgenic quail using lentivirus-mediated gene transfer into stage X quail embryos

135

Chapter 7: Summary

In this study, analyses of microarray database in Gene Expression Omnibus (GEO) led to identification of Chrd1l as a new adipokine that is highly expressed in visceral white adipose tissue. It was shown that Chrdl1 expression increases during the differentiation of

3T3-L1 preadipocytes and the development of mouse adipose tissue. Overexpression of

Chrdl1 in 3T3-L1 cells resulted in a dramatic promotion of lipid accumulation and adipocyte differentiation. Taken together, Chrdl1 is a novel visceral adipokine and increases adipocyte differentiation and lipid accumulation.

In addition, genes related to ATGL-mediated lipolysis have been investigated.

Porcine G0S2, an ATGL inhibitor, was sequenced, and its hydrophobic domain for binding to ATGL was found to be conserved across species. Expression of porcine G0S2 was high in the fat cell fraction of adipose tissue and significantly increased during adipogenesis in vitro and in vivo, suggesting the role of G0S2 during adipocyte development. A short-term 24 h fasting down-regulated G0S2 expression and increased

ATGL expression in adipose tissue of Berkshire pigs. Thus, this study showed that a short-term fasting could be one of the ways to maximize ATGL-mediated lipolysis.

In bovine species, G0S2 and CGI-58, an ATGL co-activator, were sequenced, and their hydrophobic domains were found to be conserved, suggesting their interaction with

ATGL is maintained in cattle. Importantly, SNPs were found in functional domains of

G0S2 and CGI-58 in high-marbled cattle, Hanwoo. In relation to marbling, activities of

136 these SNPs need to be further investigated. The longissimus dorsi muscle (LM) of

Hanwoo steers showed significantly higher mRNA expression of G0S2 and CGI-58 than that of Hanwoo bulls. Given that Hanwoo steers show a higher degree of marbling than

Hanwoo bulls, G0S2 and CGI-58 are both potential biomarkers for marbling.

Furthermore, G0S2 expression was markedly higher in the intramuscular fat (IMF) of

LM of Hanwoo steers; whereas, CGI-58 expression was significantly higher in the pure muscle portion mRNA where intramyocellular triglycerides (IMTG) are present, suggesting adipose depot-specific regulation of ATGL-mediated lipolysis by G0S2 and

CGI-58.

Lastly, microarray and Western blot analyses showed that, in avian species, RBP7 mRNA and protein expressions were predominant in adipose tissue. Its promoter region was further investigated, and many binding sites for transcription factors related to adipose development were found in the region. Using this promoter region, transgenic quail expressing a target gene (GFP) under the regulation of RBP7 promoter were generated by lentivirus-mediated gene transfer. The GFP expression in transgenic quail was specific to adipose tissue and increased during adipocyte differentiation, suggesting that the RBP7 promoter is sufficient to overexpress a target gene in adipose tissue at later developmental stages. Thus, the RBP7 promoter with a target gene can be used to find the in vivo function of a target gene in adipose tissue.

In brief, Chrdl1 is a new adipokine that may affect visceral fat accumulation, porcine G0S2 binds to and inhibit ATGL but is down-regulated during short-term fasting, bovine G0S2 and CGI-58 are marbling biomarkers that show depot-specific expression

137 patterns, and RBP7 is an avian adipose-specific gene, and its promoter can be used to overexpress target genes in adipose tissue of quail to investigate the function of genes.

138

Chapter 8: Concluding remarks

Obesity is the world-wide trend that has reached epidemic proportions. Because obesity is a main risk factor of many serious diseases and current strategies to overcome this global health problem are unsuccessful, identification of novel critical factors associated with fat accumulation gains enormous attention. In particular, excess fat accretion around the visceral fat compartment has been linked to the risk of developing obesity-related diseases. Obesity is also characterized by uncontrollable secretion of adipose-derived bioactive peptides, termed adipokines, thereby developing insulin resistance in neighboring tissues, such as skeletal muscle. Besides the role of adipose tissue in storing excessive energy in the form of triglycerides, its critical function as an endocrine organ is attributed to the presence and interplay of significant adipokines for regulating various physiological processes, including insulin sensitivity and adipocyte differentiation. There have been many attempts to establish new adipokines as a therapeutic target; however, well-known adipokines, such as leptin showed limited applications to congenital obesity.

Therefore, research on novel adipokines that can be utilized to treat common obesity is important. In my current study, Chrdl1 has been discovered as a new potential adipokine that is specific to visceral adipose tissue and promotes adipocyte differentiation. Future research on Chrdl1 will need to identify the mechanism of stimulatory Chrdl1 action during adipogenesis and its application for obesity research. In relation to the inhibitory roles of G0S2 on ATGL-mediated lipolysis, this study on porcine G0S2 provides new

139 insights on the regulation of lipolysis upon developmental and nutritional interventions.

Muscle meat production has great economic value around the world. Reducing fat accumulation will lead to effective conversion of feed into muscle and decrease production costs. Considering that G0S2 and CGI-58 may play a role in intramuscular fat

(IMF) and intramyocellular lipids (IMCL) in bovine muscle, respectively, the current research on SNPs in G0S2 and CGI-58 can be applied towards selective breeding for marbling. In another study, RBP7 has been identified as being adipose-specific whose promoter region can be utilized to overexpress a gene of interest specifically in adipose tissue. The RBP7 promoter can be applied to screen candidate genes related to fat accretion that leads to decreased feed efficiency in animals as well as obesity in humans.

In the future, our efforts on delineating precise functions of these complex arrays of genes are warranted.

140

References

Abreu JG, Coffinier C, Larrain J, Oelgeschlager M, De Robertis EM (2002) Chordin-like CR domains and the regulation of evolutionarily conserved extracellular signaling systems. Gene. 287:39-47.

Ahn J, Oh SA, Suh Y, Moeller SJ, Lee K (2013) Porcine G0/G1 switch gene 2 (G0S2) expression is regulated during adipogenesis and short-term in-vivo nutritional interventions. Lipids 48:209–18.

Badin PM, Loubière C, Coonen M, Louche K, Tavernier G, Bourlier V, et al. (2012) Regulation of skeletal muscle lipolysis and oxidative metabolism by the co-lipase CGI-58. J Lipid Res 53:839–48.

Baillargeon J, Rose DP (2006) Obesity, adipokines, and prostate cancer. Int J Oncol. 28: 737-45.

Baugh JE Jr, Floyd ZE, Stephens JM (2007) The modulation of STAT5A/GR complexes during fat cell differentiation and in mature adipocytes. Obesity. 15(3):583-90.

Bennett GL, Williams CB (1994) Implications of genetic changes in body composition on beef production systems. J Anim Sci 72:2756‒63.

Bergen WG, Mersmann HJ (2005) Comparative aspects of lipid metabolism: impact on contemporary research and use of animal models. J Nutr. 135: 2499-502.

Birsoy K, Chen Z, Friedman J (2008) Transcriptional regulation of adipogenesis by KLF4. Cell Metab. 7(4):339-47.

Bjørndal B, Burri L, Staalesen V, Skorve J, Berge RK (2011) Different Adipose Depots: Their Role in the Development of Metabolic Syndrome and Mitochondrial Response to Hypolipidemic Agents. J Obes. doi:10.1155/2011/490650.

Bleich S, Cutler D, Murray C, Adams A (2008) Why is the developed world obese? Annu Rev Public Health. 29:273-95.

Bong JJ, Jeong JY, Rajasekar P, Cho YM, Kwon EG, Kim HC, et al. (2012) Differential expression of genes associated with lipid metabolism in Longissimus dorsi of Korean bulls and steers. Meat Sci 91:284–93.

141

Bowers RR, Lane MD (2008) Wnt signaling and adipocyte lineage commitment. Cell Cycle. 7(9):1191-6.

Blüher M, Mantzoros CS (2015) From leptin to other adipokines in health and disease: facts and expectations at the beginning of the 21st century. Metabolism. 64(1):131-45.

Cannon B, Nedergaard J (2004) Brown Adipose Tissue: Function and Physiological Significance. Physiol Rev. 84: 277-359.

Cao Z, Umek RM, McKnight SL (1991) Regulated expression of three C/EBP isoforms during adipose conversion of 3T3-L1 cells. Genes Dev. 5(9):1538-52.

Centers for Disease Control and Prevention (2013) Obesity and overweight. FastStats. [updated Nov 2013]. Available from: http://www.cdc.gov/nchs/fastats/overwt.htm.

Chen D, Zhao M, Mundy GR (2004) Bone morphogenetic proteins. Growth Factors. 22(4):233-41.

Chen Z, Torrens JI, Anand A, Spiegelman BM, Friedman JM (2005) Krox20 stimulates adipogenesis via C/EBPbeta-dependent and -independent mechanisms. Cell Metab. 1(2):93-106.

Coffinier C, Tran U, Larrain J, De Robertis EM (2001) Neuralin-1 is a novel Chordin- related molecule expressed in the mouse neural plate. Mechanisms of Development 100: 119-22.

Cook KS, Min HY, Johnson D, Chaplinsky RJ, Flier JS, Hunt CR, et al. (1987) Adipsin: a circulating serine protease homolog secreted by adipose tissue and sciatic nerve. Science. 237(4813):402-5.

Cornaciu I, Boeszoermenyi A, Lindermuth H, Nagy HM, Cerk IK, Ebner C, et al. (2011) The minimal domain of adipose triglyceride lipase (ATGL) ranges until leucine 254 and can be activated and inhibited by CGI-58 and G0S2, respectively. PLos ONE 6(10):1-11

Cornelius P, MacDougald OA, Lane MD (1994) Regulation of adipocyte development. Annu Rev Nutr. 14:99-129.

Covas DT, Panepucci RA, Fontes AM, Silva WA, Orellana MD, et al. (2008) Multipotent mesenchymal stromal cells obtained from diverse human tissues share functional properties and gene-expression profile with CD146+ perivascular cells and fibroblasts. Exp. Hematol. 36:642–54.

Dadson K, Liu Y, Sweeney G (2011) Adiponectin action: a combination of endocrine and autocrine/paracrine effects. Frontiers in Endocrinology. 2(62) doi: 10.3389/fendo.2011. 00062 .

142

Dardeno TA, Chou SH, Moon HS, Chamberland JP, Fiorenza CG, Mantzoros CS (2010) Leptin in human physiology and therapeutics. Front Neuroendocrinol. 31(3):377-93. de Jonge HJ, Fehrmann RS, de Bont ES, Hofstra RM, Gerbens F, Kamps WA, et al. (2007) Evidence based selection of housekeeping genes. PLoS One 2: e898.

Deiuliis J, Shin J, Bae D, Azain MJ, Barb R, Lee K (2008) Developmental, hormonal, and nutritional regulation of porcine adipose triglyceride lipase (ATGL). Lipids 43:215- 225

Deiuliis J, Shin J, Murphy E, Kronberg SL, Eastridge ML, Suh Y, Yoon JT, Lee K (2010) Bovine adipose triglyceride lipase is not altered and adipocyte fatty acid-binding protein is increased by dietary flaxseed. Lipids 4:963–73

Duncan RE, Ahmadian M, Jaworski K, Sarkadi-Nagy E, Sul HS (2007) Regulation of lipolysis in adipocytes. Annu. Rev. Nutr 27:79-101

Dyck DJ, Heigenhauser GJ, Bruce CR (2006) The role of adipokines as regulators of skeletal muscle fatty acid metabolism and insulin sensitivity. Acta Physiol (Oxf). 86(1):5- 16.

Estellé J, Pérez-Enciso M, Mercadé A, Varona L, Alves E, Sánchez A, et al. (2006) Characterization of the porcine FABP5 gene and its association with the FAT1 QTL in an Iberian by Landrace cross. Anim Genet 37:589–91.

Eyal-Giladi H, Kochav S (1976) From cleavage to primitive streak formation: a complementary normal table and a new look at the first stages of the development of the chick. I. General morphology. Dev Biol 49: 321-37.

Filipak M, Estervig DN, Tzen C-Y, Minoo P, Hoerl BJ, Maercklein PB, et al. (1989) Integrated control of proliferation and differentiation of mesenchymal stem cells. Environmental Health Perspectives 80:117-25.

Fischer J, Lefèvre C, Morava E, Mussini JM, Laforêt P, Negre-Salvayre A, et al. (2007) The gene encoding adipose triglyceride lipase (PNPLA2) is mutated in neutral lipid storage disease with myopathy. Nat Genet 39:28–30.

Flier JS, Cook KS, Usher P, Spiegelman BM (1987) Severely impaired adipsin expression in genetic and acquired obesity. Science. 237(4813):405-8.

Floyd ZE, Stephens JM (2003) STAT5A promotes adipogenesis in nonprecursor cells and associates with the glucocorticoid receptor during adipocyte differentiation. Diabetes. 52(2):308-14.

143

Folli C, Calderone V, Ramazzina I, Zanotti G, Berni R (2002) Ligand binding and structural analysis of a human putative cellular retinol-binding protein. J Biol Chem. 277: 41970-7.

Frühbeck G (2008) Overview of adipose tissue and its role in obesity and metabolic disorders. Methods Mol Biol. 456:1-22.

Galic S, Oakhill JS, Steinberg GR (2010) Adipose tissue as an endocrine organ. Mol Cell Endocrinol. 316(2):129-39.

Gao WL, Zhang SQ, Zhang H, Wan B, Yin ZS (2013) Chordin-like protein 1 promotes neuronal differentiation by inhibiting bone morphogenetic protein-4 in neural stem cells. Mol Med Rep. 7(4):1143-8.

Gardan D, Louveau I, Gondret F (2007) Adipocyte- and heart-type fatty acid binding proteins are both expressed in subcutaneous and intramuscular porcine (Sus scrofa) adipocytes. Comp Biochem Physiol B Biochem Mol Biol 148:14–9.

Gesta S, Tseng Y-H, Kahn CR (2007) Developmental Origin of Fat: Tracking Obesity to Its Source. Cell. 131(2):242-56.

Ghinis-Hozumi Y, González-Gallardo A, González-Dávalos L, Antaramian A, Villarroya F, Shimada A, et al. (2011) Bovine sirtuins: initial characterization and expression of sirtuins 1 and 3 in liver, muscle, and adipose tissue. J Anim Sci 89:2529–36.

Graves RA, Tontonoz P, Platt KA, Ross SR, Spiegelman BM (1992) Identification of a fat cell enhancer: analysis of requirements for adipose tissue-specific gene expression. J Cell Biochem 49: 219-24.

Green H, Kehinde O (1974) Sublines of mouse 3T3 cells that accumulate lipid. Cell. 1(3):113-6.

Gregoire FM (2001) Adipocyte differentiation: From fibroblast to endocrine cells. Experimental Biology and Medicine 226:997-1002.

Grisart B, Coppieters W, Farnir F, Karim L, Ford C, Berzi P, et al. (2002) Positional candidate cloning of a QTL in dairy cattle: Identification of a missense mutation in the bovine DGAT1 gene with major effect on milk yield and composition. Genome Res 12:222-31.

Gronke S, Mildner A, Fellert S, Tennagels N, Petry S, Muller G, et al. (2005) Brummer lipase is an evolutionary conserved fat storage regulator in Drosophila. Cell Metab 1(5):323-30.

144

Gronthos S, Franklin DM, Leddy HA, Robey PG, Storms RW, Gimble JM (2001) Surface Protein Characterization of Human Adipose Tissue-Derived Stromal Cells. J Cell Physiol. 189:54-63.

Gruber A, Cornaciu I, Lass A, Schweiger M, Poeschl M, Eder C, et al. (2010) The N- terminal region of comparative gene identification-58 (CGI-58) is important for lipid droplet binding and activation of adipose triglyceride lipase. J Biol Chem 285:12289–98.

Gustafson B (2010) Adipose Tissue, Inflammation and Atherosclerosis, J of Atherosclerosis and Thrombosis. 17:1-10.

Gupta RK, Arany Z, Seale P, Mepani RJ, Ye L, Conroe HM, et al. (2010) Transcriptional control of preadipocyte determination by Zfp423. Nature. 25:464(7288) 619-23.

Haemmerle G, Lass A, Zimmermann R, Gorkiewicz G, Meyer C, Rozman J, et al. (2006) Defective lipolysis and altered energy metabolism in mice lacking adipose triglyceride lipase. Science 312:734-7.

Halberg N, Khan T, Trujillo ME, Wernstedt-Asterholm I, Attie AD, Sherwani S, et al. (2009) Hypoxia-inducible factor 1alpha induces fibrosis and insulin resistance in white adipose tissue. Mol Cell Biol. 29: 4467-83

Hart HA, Azain MJ, Hausman GJ, Reeves DE, Barb CR (2007) Failure of short-term feed restriction to affect luteinizing hormone and leptin secretion or subcutaneous adipose tissue expression of leptin in the prepuberal gilt. Can J Anim Sci 87:191-7.

Hausman GJ, Poulos S (2004) Recruitment and differentiation of intramuscular preadipocytes in stromal-vascular cell cultures derived from neonatal pig semitendinosus muscles. J Anim Sci 82:429-37.

Hausman GJ, Richardson RL (2004) Adipose tissue angiogenesis. J Anim Sci. 82(3):925- 34.

Heymsfield SB, Greenberg AS, Fujioka K, Dixon RM, Kushner R, Hunt T, Lubina JA, Patane J, Self B, Hunt P, McCamish M (1999) Recombinant leptin for weight loss in obese and lean adults: a randomized, controlled, dose-escalation trial. JAMA. 282(16):1568-75.

Hino J, Miyazawa T, Miyazato M, Kangawa K (2012) Bone morphogenetic protein-3b (BMP-3b) is expressed in adipocytes and inhibits adipogenesis as a unique complex. Int J Obes (Lond). 36(5):725-34.

Hu E, Liang P, Spiegelman BM (1996) AdipoQ is a novel adipose-specific gene dysregulated in obesity. J Biol Chem. 271(18):10697-703.

145

Huang H, Song T-J, Li X, Hu L, He Q, Liu M, et al. (2009) BMP signaling pathway is required for commitment of C3H10T1/2 pluripotent stem cells to the adipocyte lineage Proc Natl Acad Sci 106(31): 12670-5.

Jackson A, Vayssière B, Garcia T, Newell W, Baron R, Roman-Roman S, et al. (2005) Gene array analysis of Wnt-regulated genes in C3H10T1/2 cells. Bone. 36(4):585-98.

James AW (2013) Review of Signaling Pathways Governing MSC Osteogenic and Adipogenic Differentiation. Scientifica (Cairo). 2013:684736.

Jefcoate CR, Wang S, Liu X (2008) Methods that resolve different contributions of clonal expansion toadipogenesis in 3T3-L1 and C3H10T1/2 cells. Methods Mol Biol. 456:173- 93.

Jenkins CM, Mancuso DJ, Yan W, Sims HF, Gibson B, Gross RW (2004) Identification, cloning, expression, and purification of three novel human calcium-independent phospholipase A2 family members possessing triacylglycerol lipase and acylglycerol transacylase activities. J Biol Chem 279:48968–75.

Jeong JY, Kim JS, Nguyen TH, Lee H-J, Baik M (2013) Wnt/beta-catenin signaling and adipogenic genes are associated with intramuscular fat content in the Longissimus dorsi muscle of Korean cattle. Anim Genet 44: 627-35.

Jeremiah LE (1996) The influence of subcutaneous fat thickness and marbling on beef. Palatability and consumer acceptability. Food Res Int 29:513‒20.

Jo J, Gavrilova O, Pack S, Jou W, Mullen S, Sumner AE, et al. (2009) Hypertrophy and/or hyperplasia: Dynamics of adipose tissue growth. PLos Computational Biology. 5(3):1-11.

Jurie C, Cassar-Malek I, Bonnet M, Leroux C, Bauchart D, Boulesteix P, et al. (2007) Adipocyte fatty acid-binding protein and mitochondrial enzyme activities in muscles as relevant indicators of marbling in cattle. J Anim Sci 85:2660–9.

Kalra SP, Dube MG, Pu S, Xu B, Horvath TL, Kalra PS (1999) Interacting appetite- regulating pathways in the hypothalamic regulation of body weight. Endocr Rev. 20(1):68-100.

Kanai M, Raz A, Goodman DWS (1968) Retinol-binding protein: the transport protein for vitamin A in human plasma. J Clin Invest.; 47: 2025-44.

Kane R, Godson C, O'Brien C (2008) Chordin-like 1, a bone morphogenetic protein-4 antagonist, is upregulated by hypoxia in human retinal pericytes and plays a role in regulating angiogenesis. Mol Vis. 14:1138-48.

146

Kershaw EE, Flier JS (2004) Adipose tissue as an endocrine organ. J Clin Endocrinol Metab. 89(6):2548-56.

Kershaw EE, Hamm JK, Verhagen LA, Peroni O, Katic M, Flier JS (2006) Adipose triglyceride lipase: function, regulation by insulin, and comparison with adiponutrin. Diabetes 55:148–57

Kershaw EE, Schupp M, Guan H-P, Gardner NP, Lazar MA, Flier JS (2007) PPAR regulates adipose triglyceride lipase in adipocytes in vitro and in vivo. Am J Physiol Endocrinol Metab 293: E1736-45.

Kim CY, Le TT, Chen C, Cheng JX, Kim KH (2011) Curcumin inhibits adipocyte differentiation throughmodulation of mitotic clonal expansion. J Nutr Biochem. 22(10):910-20.

Kim JY, Tillison K, Lee JH, Rearick DA, Smas CM (2006) The adipose tissue triglyceride lipase ATGL/PNPLA2 is downregulated by insulin and TNF-alpha in 3T3- L1 adipocytes and is a target for transactivation by PPAR gamma. Am J Physiol Endocrinol Metab 291:E115–27.

Kim KH, Lee K, Moon YS, Sul HS (2001) A cysteine-rich adipose tissue-specific secretory factor inhibits adipocyte differentiation. J Biol Chem. 276(14):11252-6.

Kim JB, Wright HM, Wright M, Spiegelman BM. (1998) ADD1/SREBP1 activates PPARgamma through the production of endogenous ligand. Proc Natl Acad Sci U S A. 95(8):4333-7.

Kirkland JL, Tchkonia T, Pirtskhalava T, Han J, Karagiannides I (2002) Adipogenesis and aging: does aging make fat go MAD? Exp Gerontol. 37(6):757-67.

Komiya Y, Habas R (2008) Wnt signal transduction pathways. Organogenesis 4(2):68-75.

Koppen A, Kalkhoven E (2010) Brown vs white adipocytes: The PPARc coregulator story. FEBS Letters. 584: 3250–3259.

Krssak M, Falk Petersen K, Dresner A, DiPietro L, Vogel SM, Rothman DL, et al. (1999) Intramyocellular lipid concentrations are correlated with insulin sensitivity in humans: a 1H NMR spectroscopy study. Diabetologia 42:113‒6.

Lass A, Zimmermann R, Haemmerle G, Riederer M, Schoiswohl G, Schweiger M, et al. (2006) Adipose triglyceride lipase-mediated lipolysis of cellular fat stores is activated by CGI-58 and defective in Chanarin-Dorfman Syndrome. Cell Metab. 3(5):309-19.

147

Lee K, Shin J, Latshaw JD, Suh Y, Serr J. (2009) Cloning of adipose triglyceride lipase complementary deoxyribonucleic acid in poultry and expression of adipose triglyceride lipase during development of adipose in chickens. Poult Sci 88: 620-30.

Lee K, Villena JA, Moon YS, Kim KH, Lee S, Kang C, et al. (2003) Inhibition of adipogenesis and development of glucose intolerance by soluble Pref-1. J Clin Invest. 111: 453-61.

Lee KY, Russell SJ, Ussar S, Boucher J, Vernochet C, Mori MA, et al. (2013) Lessons on conditional gene targeting in mouse adipose tissue. Diabetes. 62: 864-74.

Lee S, Chen TT, Barber CL, Jordan MC, Murdock J, Desai S, et al. (2007) Autocrine VEGF Signaling Is Required for Vascular Homeostasis. Cell. 130(4):691-703.

Lee SH, Cho YM, Lee SH, Kim BS, Kim NK, Choy YH, et al. (2008) Identification of marbling-related candidate genes in M. longissimus dorsi of high- and low marbled Hanwoo (Korean Native Cattle) steers. BMB Reports 41:846‒51.

Lee SH, van der Werf JH, Lee SH, Park EW, Oh SJ, Gibson JP, et al. (2010) Genetic polymorphisms of the bovine fatty acid binding protein 4 gene are significantly associated with marbling and carcass weight in Hanwoo (Korean Cattle). Anim Genet 41: 442–4.

Lefèvre C, Jobard F, Caux F, Bouadjar B, Karaduman A, Heilig R, et al. (2001) Mutations in CGI-58, the gene encoding a new protein of the esterase/lipase/thioesterase subfamily, in Chanarin-Dorfman syndrome. Am J Hum Genet 69:1002–12.

Li B, Shin J, Lee K (2009) Interferon-stimulated gene ISG12b1 inhibits adipogenic differentiation and mitochondrial biogenesis in 3T3-L1 cells. Endocrinology. 150(3): 1217-24.

Li B, Zerby HN, Lee K (2007a) Heart fatty acid binding protein is up-regulated during porcine adipocyte development. J Anim Sci 85:1651-9.

Li WJ, Li HB, Chen JL, Zhao GP, Zheng MQ, Wen J (2008) Gene expression of heart- and adipocyte-fatty acid-binding protein and correlation with intramuscular fat in Chinese chickens. Anim Biotechnol 19:189–93.

Li X, Kim JW, Grønborg M, Urlaub H, Lane MD, Tang QQ (2007b) Role of cdk2 in the sequential phosphorylation/activation of C/EBPbeta during adipocyte differentiation. Proc Natl Acad Sci U S A. 104(28):11597-602.

Li X, Suh Y, Kim E, Moeller SJ, Lee K (2012) Alternative splicing and developmental and hormonal regulation of porcine comparative gene identification-58 (CGI-58) mRNA. J Anim Sci. 90(12):4346-54.

148

Liu L-F, Purushotham A, Wendel AA, Koba K, DeIuiis J, Lee K, et al. (2009) Regulation of adipose triglyceride lipase by rosiglitazone. Diabetes, Obesity and Metabolism 11:131- 142.

Livak KJ, Schmittgen TD (2001) Analysis of relative gene expression data using real- -ΔΔ quantitative PCR and the 2 CT. Methods. 25: 402-408.

Lowell BB, Flier JS (1997) Brown adipose tissue, β3-Adrenergic receptors, and obesity. Annu. Rev. Med. 48:307-16.

Maeda K, Okubo K, Shimomura I, Mizuno K, Matsuzawa Y, Matsubara K (1997) Analysis of an expression profile of genes in the human adipose tissue. Gene. 190(2): 227-35.

Mandal PK, Ferreira LM, Collins R, Meissner TB, Boutwell CL, Friesen M, et al. (2014) Efficient ablation of genes in human hematopoietic stem and effector cells using CRISPR/Cas9. Cell Stem Cell. 15(5):643-52.

Matsubara M, Maruoka S, Katayose S (2002) Inverse relationship between plasma adiponectin and leptin concentrations in normal-weight and obese women. Eur J Endocrinol. 147(2):173-80.

Matsuzawa Y, Funahashi T, Nakamura T (1999) Molecular Mechanism of Metabolic Syndrome X: Contribution of Adipocytokines·Adipocyte-derived Bioactive Substances. Annu. N.Y. Acad. Sci. 892:146–54.

Montero-Moran G, Caviglia JM, McMahon D, Rothenberg A, Subramanian V, Xu Z, et al. (2010) CGI-58/ABHD5 is a coenzyme A-dependent lysophosphatidic acid acyltransferase. J Lipid Res 51:709–19.

Mori T, Sakaue H, Iguchi H, Gomi H, Okada Y, Takashima Y, Nakamura K, Nakamura T, Yamauchi T, Kubota N, Kadowaki T, Matsuki Y, Ogawa W, Hiramatsu R, Kasuga M (2005) Role of Krüppel-like factor 15 (KLF15) in transcriptional regulation of adipogenesis. J Biol Chem. 280(13):12867-75.

Muruganandan S, Roman AA, Sinal CJ (2009) Adipocyte differentiation of bone marrow-derived mesenchymal stem cells: cross talk with the osteoblastogenic program. Cell Mol Life Sci. 66(2):236-53.

Nakayama N, Han C-Y, Scully S, Nishinakamura R, He C, Zeni L, et al. (2001) A novel chordin-like protein inhibitor for bone morphogenetic proteins expressed preferentially in mesenchymal cell lineages. Developmental Biology. 232:372-87.

Nanbu-Wakao R, Morikawa Y, Matsumura I, Masuho Y, Muramatsu MA, Senba E, Wakao H (2002) Stimulation of 3T3- L1 adipogenesis by signal transducer and activator

149 of transcription 5. Mol Endocrinol. 16(7):1565-76.

Narbonne P, Roy R (2009) Caenorhabditis elegans dauers need LKB1/AMPK to ration lipid reserves and ensure long-term survival. Nature 457:210-4.

Newsom SA, Schenk S, Thomas KM, Harber MP, Knuth ND, Goldenberg N, et al. (2010) Energy deficit after exercise augments lipid mobilization but does not contribute to the exercise-induced increase in insulin sensitivity. J. Appl. Physiol 108:554-60

Nguyen NT, Magno CP, Lane KT, Hinojosa MW, Lane JS (2008) Association of hypertension, diabetes, dyslipidemia, and metabolic syndrome with obesity: findings from the National Health and Nutrition Examination Survey, 1999 to 2004. J Am Coll Surg. 207(6):928-34.

Nielsen TS, Vendelbo MH, Jessen N, Pedersen SB, Jorgensen JO, Lund S, et al. (2011) Fasting, but not exercise, increases adipose triglyceride lipase (ATGL) protein and reduces G(0)/G(1) switch gene 2 (G0S2) protein and mRNA content in human adipose tissue. J Clin Endocrinol Metab 96(8):E1293-7.

Noy N (2000) Retinoid-binding proteins: mediators of retinoid action. Biochem J 348: 481-95.

Nunes PM, van de Weijer T, Veltien A, Arnts H, Hesselink MK, Glatz JF, et al. (2012) Increased intramyocellular lipids but unaltered in vivo mitochondrial oxidative phosphorylation in skeletal muscle of adipose triglyceride lipase-deficient mice. Am J Physiol Endocrinol Metab 303:E71–81.

Oh SA, Suh Y, Pang MG, Lee K (2011) Cloning of avian G(0)/G(1) switch gene 2 genes and developmental and nutritional regulation of G(0)/G(1) switch gene 2 in chicken adipose tissue. J Anim Sci 89:367-75.

Ong KT, Mashek MT, Bu SY, Greenberg AS, Mashek DG (2011) Adipose triglyceride lipase is a major hepatic lipase that regulates triacylglycerol turnover and fatty acid signaling and partitioning. Hepatology 53(1):116-26.

Ottaviani E, Malagoli D, Franceschi C (2011) The evolution of the adipose tissue: A neglected enigma. General and Comparative Endocrinology 174:1-4.

Otto TC, Lane MD (2005) Adipose Development: From Stem Cell to Adipocyte. Crit Rev in Biochem and Mol Biol. 40:229-42.

Ouchi N, Parker JL, Lugus JJ, Walsh K (2011) Adipokines in inflammation and metabolic disease Nat Rev Immunol. 11(2):85-97.

150

Piantedosi R, Ghyselinck N, Blaner WS, Vogel S (2005) Cellular retinol-binding protein type III is needed for retinoid incorporation into milk. J Biol Chem. 280: 24286-92.

Pourquie O (2001) Vertebrate Somatogenesis. Annu. Rev. Cell Dev. Biol. 17:311-50.

Poynter G, Huss D, Lansford R (2009) Japanese quail: an efficient animal model for the production of transgenic avians. Cold Spring Harb Protoc. 2009(1):pdb.emo112.

Pryce JE, Wales WJ, de Haas Y, Veerkamp RF, Hayes BJ (2014) Genomic selection for feed efficiency in dairy cattle. Animal. 8(1):1-10.

Qiu Z, Wei Y, Chen N, Jiang M, Wu J, Liao K (2001) DNA synthesis and mitotic clonal expansion is not a required step for 3T3-L1 preadipocyte differentiation into adipocytes. J Biol Chem. 276(15):11988-95.

Radin MJ, Sharkey LC, Holycross BJ (2009) Adipokines: a review of biological and analytical principles and an update in dogs, cats, and horses. Vet Clin Pathol. 38(2):136- 56.

Reichert M, Eick D (1999) Analysis of cell cycle arrest in adipocyte differentiation. Oncogene. 18(2):459-66.

Reilich P, Horvath R, Krause S, Schramm N, Turnbull DM, Trenell M, et al. (2011) The phenotypic spectrum of neutral lipid storage myopathy due to mutations in the PNPLA2 gene. J Neurol 258:1987–97.

Reznikoff CA, Brankow DW, Heidelberger C (1973) Establishment and Characterization of a Cloned Line of C3H Mouse Embryo Cells Sensitive to Post- confluence Inhibition of Division. Cancer Res. 33:3231-8.

Russell L, Forsdyke DR (1991) A human putative lymphocyte G0/G1 switch gene containing a CpG-rich island encodes a small basic protein with the potential to be phosphorylated. DNA and Cell Biol 10(8):581-91

Rider CC, Mulloy B (2010) Bone morphogenetic protein and growth differentiation factor cytokine families and their protein antagonists. Biochem J. 429(1):1-12.

Ross AC. (1993) Cellular metabolism and activation of retinoids: roles of cellular retinoid binding proteins. FASEB J. 7: 317-27.

Ross SE, Hemati N, Longo KA, Bennett CN, Lucas PC, Erickson RL, et al. (2000) Inhibition of Adipogenesis by Wnt Signaling. Science. 11;289(5481):950-3.

151

Ross SR, Graves RA, Greenstein A, Platt KA, Shyu HL, Mellovitz B, et al. (1990) A fat- specific enhancer is the primary determinant of gene expression for adipocyte P2 in vivo. Proc Natl Acad Sci U S A. 87: 9590-4.

Saely CH, Geiger K, Drexel H (2012) Brown versus White Adipose Tissue: A Mini- Review. Gerontology. 58:15-23.

Saez G, Davail S, Gente`s G, Hocquette JF, Jourdan T, Degrace P, et al. (2009) Gene expression and protein content in relation to intramuscular fat content in Muscovy and Peking ducks. Poult Sci 88:2382–91.

Sakuta H, Suzuki R, Takahashi H, Kato A, Shintani T, Iemura Si, et al. (2001) Ventroptin: a BMP-4 antagonist expressed in a double-gradient pattern in the retina. Science. 293(5527):111-5.

Salma N, Xiao H, Imbalzano AN (2006) Temporal recruitment of CCAAT/enhancer- binding proteins to early and late adipogenic promoters in vivo. J Mol Endocrinol. 36(1):139-51.

Sander JD, Joung JK (2014) CRISPR-Cas systems for editing, regulating and targeting genomes. Nat Biotechnol. 32(4):347-55.

Sato Y, Lansford R (2013) Transgenesis and imaging in birds, and available transgenic reporter lines. Dev Growth Differ. 55: 406-21.

Schäffler A, Schölmerich J, Büchler C (2005) Mechanisms of Disease: adipocytokines and visceral adipose tissue—emerging role in intestinal and mesenteric diseases. Nat Clin Pract Gastroenterol Hepatol. 2(2): 103-11.

Schweiger M, Lass A, Zimmermann R, Eichmann TO, Zechner R (2009) Neutral lipid storage disease: genetic disorders caused by mutations in adipose triglyceride lipase/PNPLA2 or CGI-58/ABHD5. Am J Physiol Endocrinol Metab 297:E289–96.

Serr J, Suh Y, Lee K (2009) Regulation of adipose triglyceride lipase by fasting and refeeding in avian species. Poult Sci 88: 2585-91.

Serr J, Suh Y, Lee K (2011) Cloning of comparative gene identification-58 gene in avian species and investigation of its developmental and nutritional regulation in chicken adipose tissue. J Anim Sci 89:3490–500.

Shen W, Wang Z, Punyanita M, Lei J, Sinav A, Kral JG, et al. (2003) Adipose Tissue Quantification by Imaging Methods: A Proposed Classification. Obes Res. 11(1): 5-16.

152

Shin J, Li B, Davis ME, Suh Y, Lee K (2009) Comparative analysis of fatty acid-binding protein 4 promoters: conservation of peroxisome proliferator-activated receptor binding sites. J Anim Sci 87:3923–34.

Shin S, Choi YM, Han JY, Lee K (2014) Inhibition of lipolysis in the novel transgenic quail model overexpressing G0/G1 switch gene 2 in the adipose tissue during feed restriction.PLoS One. 9: e100905. Shin SS, Kim TM, Kim SY, Kim TW, Seo HW, Lee SK, et al. (2008) Generation of transgenic quail through germ cell-mediated germline transmission. FASEB J. 22: 2435- 44.

Siersbæk R, Nielsen R, John S, Sung MH, Baek S, Loft A, Hager GL, Mandrup S (2011) Extensive chromatin remodelling and establishment of transcription factor 'hotspots' during early adipogenesis. EMBO J. 30(8):1459-72.

Siersbæk R, Nielsen R, Mandrup S (2012) Transcriptional networks and chromatin remodeling controlling adipogenesis. Trends Endocrinol Metab. 23(2):56-64.

Smith GC, Carpenter ZL, Cross HL, Murphey CE, Abraham HC, Savell JW, et al. (1984) Relationship of USDA marbling groups to palatability of cooked beef. J Food Qual 7: 289-308

Song K, Wang S, Mani M, Mani A (2014) Wnt signaling, de novo lipogenesis, adipogenesis and ectopic fat. Oncotarget. 5(22):11000-3.

Song Y, Ahn J, Suh Y, Davis ME, Lee K (2013) Identification of novel tissue-specific genes by analysis ofmicroarray databases: a human and mouse model. PLoS One. 8(5):e64483.

Spinella-Jaegle S, Rawadi G, Kawai S, Gallea S, Faucheu C, Mollat P, et al. (2001). Sonic hedgehog increases the commitment of pluripotent mesenchymal cells into the osteoblastic lineage and abolishes adipocytic differentiation. J. Cell Sci. 114:2085-94.

Sun K, Kusminski CM, Scherer PE (2011). Adipose tissue remodeling and obesity. J Clin Invest. 121(6):2094-101.

Takada I, Kouzmenko P, Kato S (2009) Wnt and PPARγ signaling in osteoblastogenesis and adipogenesis. Nat Rev Rheumatol. 5:442-7.

Tang QQ, Otto TC, Lane MD (2003a) Mitotic clonal expansion: a synchronous process required for adipogenesis. Proc Natl Acad Sci U S A. 100(1):44-9.

Tang QQ, Otto TC, Lane MD (2003b) CCAAT enhancer-binding protein beta is required for mitotic clonal expansion during adipogenesis. Proc Natl Acad Sci U S A. 100(3):850- 5.

153

Tang QQ, Otto TC, Lane MD (2004) Commitment of C3H10T1/2 pluripotent stem cells to the adipocyte lineage. Proc Natl Acad Sci 101:9607-11.

Tang QQ, Lane MD (2012) Adipogenesis: From Stem Cell to Adipocyte. Annu. Rev. Biochem. 81:715-36.

Thompson JD, Higgins DG, Gibson TJ (1994) CLUSTAL W: Improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position- specific gap penalties and weight matrix choice. Nucleic Acids Res 22:4673-80.

Tomlinson JJ, Boudreau A, Wu D, Atlas E, Haché RJ (2006) Modulation of early human preadipocyte differentiation by glucocorticoids. Endocrinology. 147(11):5284-93.

Trayhurn P (2005) Endocrine and signalling role of adipose tissue: new perspectives on fat. Acta Physiol Scand. 184: 285–293.

Tsutsumi C, Okuno M, Tannous L, Piantedosi R, Allan M, Goodman DS, et al. (1992) Retinoids and retinoid-binding protein expression in rat adipocytes. J Biol Chem. 267: 1805-10.

USDA, FAS. 2014. Livestock and Poultry: World Markets and Trade (http://www.fas. usda.gov/data/livestock-and-poultry-world-markets-and-trade, 17 October 2014). World Production, Markets, and Trade Reports.

Villarroya F, Iglesias R, Giralt M. (2004) Retinoids and retinoid receptors in the control of energy balance: novel pharmacological strategies in obesity and diabetes. Curr Med Chem. 11: 795-805.

Villena JA, Roy S, Sarkadi-Nagy E, Kim KH, Sul HS (2004) Desnutrin, an adipocyte gene encoding a novel patatin domain-containing protein, is induced by fasting and glucocorticoids: ectopic expression of desnutrin increases triglyceride hydrolysis. J Biol Chem 279:47066–75.

Virkamäki A, Korsheninnikova E, Seppälä-Lindroos A, Vehkavaara S, Goto T, Halavaara J, et al. (2001) Intramyocellular lipid is associated with resistance to in vivo insulin actions on glucose uptake, antilipolysis, and early insulin signaling pathways in human skeletal muscle. Diabetes 50:2337‒43.

Vogel S, Mendelsohn CL, Mertz JR, Piantedosi R, Waldburger C, Gottesman ME, et al. (2001) Characterization of a new member of the fatty acid-binding protein family that binds all-trans-retinol. J Biol Chem 276: 1353-60.

Waki H, Park KW, Mitro N, Pei L, Damoiseaux R, Wilpitz DC, et al. (2007) Small Molecule Harmine Is an Antidiabetic Cell-Type-Speciﬁc Regulator of PPARγ Expression. Cell Metab. 5(5):357-70.

154

Wang EA, Israel DI, Kelly S, Luxenberg DP (1993) Bone morphogenetic protein-2 causes commitment and differentiation in C3H10T1/2 and 3T3 cells. Growth Factors 9: 57 -71.

Wang SP, Laurin N, Himms-Hagen J, Rudnicki MA, Levy E, Robert MF, et al. (2001) The adipose tissue phenotype of hormone-sensitive lipase deficiency in mice. Obes. Res 9:119-28.

Wang ZV, Deng Y, Wang QA, Sun K, Scherer PE. (2010) Identification and characterization of a promoter cassette conferring adipocyte-specific gene expression. Endocrinology. 151: 2933-9.

Webb TR, Matarin M, Gardner JC, Kelberman D, Hassan H, Ang W, et al. (2012) X- linked megalocornea caused by mutations in CHRDL1 identifies an essential role for ventroptin in anterior segment development. Am J Hum Genet. 90(2):247-59.

Wertz AE, Berger LL, Walker PM, Faulkner DB, McKeith FK, Rodriguez-Zas SL (2002) Early-weaning and postweaning nutritional management affect feedlot performance, carcass merit, and the relationship of 12th-rib fat, marbling score, and feed efficiency among Angus and Wagyu heifers. J Anim Sci 80:28‒37.

Wibowo TA, Gaskins CT, Newberry RC, Thorgaard GH, Michal JJ, Jiang Z (2008) Genome assembly anchored QTL map of bovine chromosome 14. Int J Bio Sci 4:406–14.

Wood JD, Enser M, Fisher AV, Nute GR, Sheard PR, Richardson RI, et al. (2008) Fat deposition, fatty acid composition and meat quality: A review. Meat Sci 78:343–58.

World Health Organization (2013) Obesity and overweight. Fact Sheet N°311. [updated Mar 2013]. Available from: http://www.who.int/mediacentre/factsheets/fs311/en/.

Wu JW, Wang SP, Alvarez F, Casavant S, Gauthier N, Abed L, et al. (2011) Deficiency of liver adipose triglyceride lipase in mice causes progressive hepatic steatosis. Hepatology 54:122–32.

Yamada T, Park CS, Burns A, Nakada D, Lacorazza HD (2012) The cytosolic protein G0S2 maintains quiescence in hematopoietic stem cells. PLoS ONE 7(5):e38280.

Yamauchi T, Kamon J, Waki H, Terauchi Y, Kubota N, Hara K, et al. (2001) The fat- derived hormone adiponectin reverses insulin resistance associated with both lipoatrophy and obesity. Nat Med. 7(8):941-6.

Yang KX, Ma JW, Guo YM, Guo TF, Zhao YG, Ding NS, et al. (2010a) Correlations between fat depot traits and fatty acid composition in abdominal subcutaneous adipose tissue and longissimus muscle: Results from a White Duroc × Erhualian intercross F2 population. J Anim Sci 88:3538–45.

155

Yang S, Suh Y, Choi YM, Shin S, Han JY, Lee K. (2013) Loss of fat with increased adipose triglyceride lipase-mediated lipolysis in adipose tissue during laying stages in quail. Lipids. 48: 13-21.

Yang X, Lu X, Lombès M, Rha GB, Chi YI, Guerin TM, et al. (2010b) The G(0)/G(1) switch gene 2 regulates adipose lipolysis through association with adipose triglyceride lipase. Cell Metab. 11(3):194-205.

Yang X, Lu X, Liu J (2010c) Identification of a novel splicing isoform of murine CGI-58. FEBS Lett 584:903–10.

Yasmeen R, Reichert B, Deiuliis J, Yang F, Lynch A, Meyers J, et al. (2013) Autocrine function of aldehyde dehydrogenase 1 as a determinant of diet- and sex-specific differences in visceral adiposity. Diabetes. 62(1):124-36.

Ye MH, Chen JL, Zhao GP, Zheng MQ, Wen J (2010) Associations of A-FABP and H- FABP markers with the content of intramuscular fat in Beijing-You chicken. Anim Biotechnol 21:14–24.

Yeh WC, Bierer BE, McKnight SL (1995) Rapamycin inhibits clonal expansion and adipogenic differentiation of 3T3-L1 cells. Proc Natl Acad Sci U S A. 92(24):11086-90.

Yellaturu CR, Deng X, Park EA, Raghow R, Elam MB (2009) Insulin enhances the biogenesis of nuclear sterol regulatory element-binding protein (SREBP)-1c by posttranscriptional down-regulation of Insig-2A and its dissociation from SREBP cleavage-activating protein (SCAP).SREBP-1c complex. J Biol Chem. 284(46):31726-34.

Yu X, Lin J, Zack DJ, Qian J (2006) Computational analysis of tissue-specific combinatorial gene regulation: predicting interaction between transcription factors in human tissues. Nucleic Acids Res. 34(17):4925-36.

Zhao SM, Ren LJ, Chen L, Zhang X, Cheng ML, Li WZ, et al. (2009) Differential expression of lipid metabolism related genes in porcine muscle tissue leading to different intramuscular fat deposition. Lipids 44:1029–37.

Zandbergen F, Mandard S, Escher P, Tan NS, Patsouris D, Jatkoe T, et al. (2005) The G0/G2 switch gene 2 is a novel PPAR target gene. Biochem. J. 392:313-24.

Zehentner BK, Leser U, Burtscher H (2000) BMP-2 and sonic hedgehog have contrary effects on adipocyte-like differentiation of C3H10T1/2 cells. DNA Cell Biol. 19(5):275- 81.

Zimmermann R, Strauss JG, Haemmerle G, Schoiswohl G, Birner-Gruenberger R, Riederer M, et al. (2004) Fat mobilization in adipose tissue is promoted by adipose triglyceride lipase. Science 306:1383-6.

156