ROLE OF THE DROPLET PROTEOME IN THE PATHOGENESIS OF INSULIN RESISTANCE IN THE LIVER

A thesis submitted in fulfilment of requirements for the degree

Doctor of Philosophy

from

Faculty of Medicine and Health The University of Sydney

by

Shilpa Nagarajan

2019

i

Statement of Originality

‘This is to certify that to the best of my knowledge, the content of this thesis is my own work. This thesis has not been submitted for any degree or other purposes. All the assistance received in preparing in this thesis and sources have been acknowledged.’

Shilpa Nagarajan

ii Acknowledgements

The truth about doing a PhD is that the carefully curated experiments and analysis you see in this thesis are only the tip of the iceberg. In reality, there are many more moments excluded from this thesis – such as all the failed experiments that tested my resilience, the joy from obtaining promising results after months of optimization, and the discussions that reminded me that I really don’t know what I don’t know. Between the ups and the downs, I owe it all to the people that were right there with me throughout this journey.

To my supervisor, Andrew. I don’t think I could have picked a better mentor to jumpstart my career. Your commitment to research integrity and quality and your drive for growth and balance are unmatched. Regardless of how much you had on your plate, you always followed through – whether it be my constant questions, anxiety about my data or advice on the realities of being a scientist. It was, and always will be, a true honor learning from you.

To my every-changing lab family, thank you for keeping me sane and reminding me that I am more than my results. To the OG lipid lab members Seher and Tim, thank you for taking me under your wing and making Sydney feel a little more like home. To Bianca and Atqiya, thank you for you being some of the most kind, amazing, selfless and competent workhorses the lab has ever seen. To Toni, Andi, Nikki and Meghna, thank you for always making me laugh and sticking through those long hours with me. I hope we always remember that it’s always just a quick squirt squirt in the Lipid Lab, am I right Andrew?

To my parents, Appa and Amma. I am so grateful for the unconditional support, patience and trust you had in me while I pursued this new endeavor abroad. To Appa, thank you for your unwavering curiosity in my work and my field. Despite the technical jargon, you have always understood the meaning behind the science. Amma, thank you for being the emotional stability I needed to get through the hardest parts of my degree. Thank you especially for praying that my experiments would work that one long weekend in June 2018, because they did, and everything was better after that. You both have worked so hard to provide this opportunity for me and I hope I have made you proud. This doctorate is as much yours as it is mine (maybe not AS much).

And finally, to my absolute rock, Nikhil. To all the times you comforted me when nothing was working, listened to all of my crazy jargon when I got excited about theories, for all the phone calls you made to keep me company when I walked home late from the lab – I thank you. You believed in me when I didn’t and that means more than words.

iii

A person’s willpower is like water. Nothing can stop it. It will pass through mountains, and pour down waterfalls, but it will always end up where it needs to be

iv Table of Contents

STATEMENT OF ORIGINALITY ...... II ACKNOWLEDGEMENTS ...... III AUTHORSHIP ATTRIBUTION STATEMENT (CHAPTER 2)...... IX AUTHORSHIP ATTRIBUTION STATEMENT (CHAPTER 4)...... X LIST OF ABBREVIATIONS ...... XI PUBLICATIONS AND PRESENTATIONS THAT HAVE ARISEN AS A DIRECT RESULT OF THIS THESIS ...... XV PUBLICATIONS THAT HAVE ARISEN IN CONCURRENCE WITH THIS THESIS...... XVII ABSTRACT ...... XVIII LIST OF FIGURES ...... XX LIST OF TABLES ...... XXI INTRODUCTION ...... 1

1.1. INTRODUCTION ...... 2 1.2. NON-ALCOHOLIC FATTY LIVER ...... 3 General Characteristics of NAFL(D) ...... 3 1.2.2 Prevalence of NAFL(D) ...... 3 1.2.3 Pathophysiology of NAFL(D) ...... 4 1.2.3.1 Mechanisms of TAG Accumulation ...... 5 1.2.3.1.1 Hepatic Fatty Acid Uptake ...... 5 1.2.3.1.2 Intracellular Fatty Acid Activation and Transport ...... 6 1.2.3.1.3 Hepatic de novo Lipogenesis ...... 7 1.2.3.1.4 Lipid Esterification into ...... 8 1.2.3.2 Mechanisms of TAG Disposal ...... 10 1.2.3.2.1 Secretion as VLDL ...... 10 1.2.3.2.2 Intrahepatic Lipolysis ...... 11 1.2.3.2.3 Hepatic b-oxidation ...... 11 1.2.3.3 Co-morbidities ...... 12 1.2.3.3.1 Obesity ...... 12 1.2.3.3.2 Cardiovascular Disease ...... 13 1.2.3.3.3 Insulin Resistance...... 14 1.2.3.3.4 Complexities to Uncouple NAFLD, IR and Obesity ...... 15 1.3. LINKING WITH IR SIGNALLING ...... 16 1.3.1 Hepatic Glucose Output ...... 17 1.3.2. Ubiquitin-Proteasomal Degradation ...... 18 1.3.3. Bioactive Lipid Signalling ...... 20 1.3.3.1. Diacylglycerols ...... 20 1.3.3.2. Ceramides ...... 21 1.3.4. Leaky Lipid Droplets...... 21

1.4. LIPID DROPLETS ...... 22

v 1.4.1. General Properties ...... 23 1.4.2. The Lipid Droplet Life Cycle ...... 24 1.4.2.1. Biogenesis and Growth ...... 24 1.4.2.2. Lipid Droplets as Dynamic Interactors ...... 26 1.4.2.2.1. Contact Sites and Transport of Lipid Droplets ...... 27 1.4.2.2.2. Transfer with Endoplasmic Reticulum ...... 28 1.4.2.2.3. Mitochondrial and Peroxisomal Lipid Droplet Catabolism ...... 28 1.4.2.2.4. Lipid Exchange between Lipid Droplets ...... 29 1.4.3. Lipid Droplet Proteome in Metabolic Disease ...... 29 1.4.3.1. Lipids Droplet Proteome and Non-Alcoholic Fatty Liver...... 30 1.4.3.2. Lipid Droplet Proteome and Insulin Resistance ...... 31 1.4.3.2.1. Regulation of LD by Insulin...... 31

1.5. SUMMARY AND THESIS AIMS...... 33

CHAPTER 2 ...... 35 CHAPTER 3 ...... 48

3.1. INTRODUCTION ...... 49

3.2 METHODS...... 51 3.2.1 Isolation of lipid droplets from rat livers ...... 52 3.2.1.1 Homogenization ...... 52 3.2.1.2 Sucrose Gradient Fractionation ...... 52 3.2.1.3 Sample Delipidation ...... 52 3.2.2 Proteomics ...... 52 3.2.2.1 Protein Preparation ...... 52 3.2.2.2 Protein Digestion ...... 53 3.2.2.3 Cartridge Desalting ...... 53 3.2.2.4 Tandem Mass Tag Labelling ...... 53 3.2.2.5 Stage-Tip Desalting ...... 54 3.2.2.6 Phosphopeptide Enrichment...... 54 3.2.2.7 Liquid Chromatography-Tandem Mass Spectrometry ...... 55 3.2.2.7.1 Total Proteome...... 55 3.2.2.7.2 Total Phosphoproteome ...... 55 3.2.2.8 Protein Identification and Quantification ...... 55 3.2.2.8.1 Total Proteome...... 55 3.2.2.8.2 Total Phosphoproteome ...... 56 3.2.2.9 Bioinformatic Analysis ...... 56 3.2.2.9.1 Normalisation ...... 56 3.2.2.9.2 Determination of Novel Rattus norvegicus Phosphosites ...... 56 3.2.2.9.3 Enriched Pathway Analysis ...... 56 3.2.2.9.4 Scenario Modelling ...... 57 3.2.3 Western Blotting ...... 57 3.2.3.1 Protein Quantification ...... 57 3.2.3.2 SDS-Polyacrylamide Gel Electrophoresis ...... 58

vi 3.2.3.3 Protein Transfer to a PVDF Membrane ...... 58 3.2.3.4 Antibody Probing ...... 59 3.2.3.5 Imaging with Bio-Rad ChemiDoc MP...... 60 3.2.3.6 Stripping and Re-Probing ...... 60 3.3 RESULTS ...... 61 3.3.1 Hepatic Lipid Droplet Morphology and Purity ...... 61 3.3.2 Intra-Group Variability and Reproducibility in Proteome and Phosphoproteome ...... 61 3.3.3 Quantitative Analysis of Proteome and Phosphoproteome ...... 64 3.3.4 Differential Expression Analysis of LD-Associated Proteome ...... 66 3.3.5 Differential Expression Analysis of LD-Associated Phosphoproteome ...... 67 3.3.6 Modelling Insulin Resistance via Binary Differential Expression ...... 70 3.3.7 Modelling Insulin Resistance via Biologically Relevant Scenarios ...... 73 3.3.7.1 Identification of “Insulin Resistant” Proteins ...... 73 3.3.7.2 Identification of “Insulin Resistant” Phosphosites ...... 84 3.4 DISCUSSION ...... 93 3.4.1 Proteomic Characterization of LDs ...... 94 3.4.1.1 Influence of HFD on LD Proteome ...... 94 3.4.1.2 Influence of Insulin on LD Proteome ...... 95 3.4.2 Phosphoproteomic Characterization of LDs ...... 96 3.4.3 Phospho-Proteomics and HFD-Induced Insulin Resistance...... 97

CHAPTER 4 ...... 100 CHAPTER 5 ...... 138

5.1 INTRODUCTION ...... 139

5.2 METHODS...... 141 5.2.1 Culture ...... 141 5.2.1.1 Cell Maintenance ...... 141 5.2.1.2 Treatment of Cells ...... 141 5.2.1.2.1 Insulin Treatment ...... 141 5.2.1.2.2 FA Treatment ...... 141 5.2.1.2.3 Peroxisome Proliferator-Activated Receptor (PPAR) a Agonist Treatment ...... 141 5.2.1.3 siRNA-Mediated UBXD8 Knockdown ...... 142 5.2.2 Western Blotting ...... 142 5.2.2.1 Protein Extraction and Quantification ...... 142 5.2.2.2 Antibody Probing ...... 143 5.2.3 Radiolabel Metabolic Analysis ...... 144 5.2.3.1 Extracellular Fatty Acid Metabolism ...... 144 5.2.3.1.1 FA Oxidation ...... 144 5.2.3.2 Protein Quantification ...... 145 5.2.4 Lipid Extraction ...... 145 5.2.5 Immunofluorescence Microscopy ...... 145 5.2.6 Subcellular Fractionation ...... 146

vii 5.2.6.1 Treatment and Homogenization of Hepatocytes ...... 146 5.2.6.2 Sucrose Gradient Fractionation ...... 147 5.2.6.3 Immunoblotting Fractions ...... 147 5.2.7 Mass Spectrometry ...... 148 5.2.7.1 Sample Preparation ...... 148 5.2.7.2 Stage-Tipping ...... 148 5.2.7.3 LC-MS/MS Acquisition ...... 148 5.2.7.4 MaxQuant Analysis ...... 149 5.2.8 Statistical Analysis ...... 149

5.3 RESULTS ...... 150 5.3.1 UBXD8 is Abundant in Mouse Liver ...... 150 5.3.2 Upregulation and Translocation of UBXD8 Upon Insulin Stimulation ...... 151 5.3.3 Upregulation and Translocation of UBXD8 Upon Lipid Treatment ...... 152 5.3.4 UBXD8 Knockdown Attenuates ATGL Protein Expression ...... 153 5.3.5 UBXD8 Knockdown Promotes TAG Accumulation in Hepatocytes ...... 154 5.3.6 UBXD8 is Associated with G0S2 Protein Expression ...... 155 5.3.7 UBXD8 Regulates Mitochondrial Fatty Acid Oxidation ...... 156 5.3.8 UBXD8 KD Downregulates Oxidation and Upregulates FA Synthesis Pathways ...... 159 5.3.9 UBXD8 Co-Localizes with Ester Pools Upon Lipid Treatment ...... 164 5.3.10 UBXD8 is Not Related to Expression of Key Cholesterol-Related Proteins ...... 165

5.4 DISCUSSION ...... 167 5.4.1 UBD8 and TAG Homeostasis ...... 167 5.4.2 UBXD8 and Oxidation ...... 169

CHAPTER 6 ...... 173

6.1 OVERVIEW ...... 174 6.1.1 Cooperative Regulation of Insulin Signalling: A PTM Story...... 174 6.1.2 Coordinated Interplay of the ‘Lipolysome’: An Interactor Story ...... 180 6.1.3 Modelling Metabolic Disease: Influence on Proteome ...... 182 6.1.4 Conclusion ...... 184

REFERENCES ...... 186

viii Authorship Attribution Statement (Chapter 2)

Chapter 2 of this thesis is published as Nagarajan S.R.*, Brandon A.E., McKenna J.A., Shtein, H.C., Nguyen T.Q., Suryana E., Poronnik P., Cooney G.J., Saunders D.N., Hoy A.J. Insulin and diet-induced changes in the ubiquitin-modified proteome of rat liver. PLoS ONE, 2016.

Contribution to manuscript Authors Designed the study Hoy A.J., Saunders D.N. Extracted the data Nagarajan S.R., Brandon A.E., McKenna J.A., Shtein H.C., Nguyen T.Q., Suryana E., Pronnik P. Analysed the data Nagarajan S.R., Hoy A.J., McKenna J.A., Saunders D.N., Cooney G.J. Wrote and edited the manuscript Nagarajan S.R., Brandon A.E., McKenna J.A., Saunders D.N., Cooney G.J., Hoy A.J.

As supervisor for the candidature upon which this thesis is based, I can confirm that the authorship statement above is correct.

______Dr. Andrew James Hoy Date

ix Authorship Attribution Statement (Chapter 4)

Chapter 4 of this thesis is published as Nagarajan S.R.*, Paul-Heng M., Krycer J.R., Fazakerley D.J., Sharland A.F., Hoy A.J. Lipid and glucose metabolism in hepatocyte cell lines and primary mouse hepatocytes: A comprehensive resource for in vitro studies of hepatic metabolism. Am J Physiol Endocrinol Metab, 2019.

Contribution to manuscript Authors Designed the study Nagarajan S.R., Hoy A.J. Isolated primary hepatocytes Paul M., Sharland A.F. Extracted the data Nagarajan S.R. Analysed the data Nagarajan S.R., Hoy A.J. Wrote and edited the manuscript Nagarajan S.R., Hoy A.J., Krycer J.R., Fazakerley D.J., Sharland A.F.

As supervisor for the candidature upon which this thesis is based, I can confirm that the authorship statement above is correct.

______Dr. Andrew James Hoy Date

x List of Abbreviations

ACAT1 acetyl-CoA acetyltransferase, mitochondrial ACBP acyl carrier binding proteins ACC acetyl-CoA carboxylases ACSL acyl-CoA synthetases AGPAT 1-acylglycerol-3-phosphate O-acyltransferase AMPK AMP-activated protein kinase ASO antisense oligonucleotide ATGL adipose triglyceride BCA Bicinchoninic Assay BMI body mass index BSA bovine serum albumin CD36 cluster of differentiation 36 CERS ceramide synthase CGI58 comparative identification-58 CIDE cell death-inducing DFF45-like effector CPT1 carnitine palmitoyltransferase-1 CVD cardiovascular diseases DAG diacylglycerols DCD dermcidin DE differentially expressed DE differentially expressed DGAT diacylglycerol acyltransferases DHA docosahexaenoic acids DMEM-F12 Dulbecco’s modified Eagle’s medium and Ham’s F-12 DNL de novo lipogenesis DUB deubiquitinating E1 ubiquitin-activating E2 ubiquitin-conjugating enzyme E3 ubiquitin ligase eLD expanding LDs EPA eicosapentaenoic acids ERP72 endoplasmic reticulum

xi ETC electron transport chain FABP fatty acid binding proteins FA-CoAs fatty acyl-coenzyme A FAF2 fas-asociated factor 2 FA fatty acids FASN fatty acid synthase FATP5 fatty acid transport protein FAT fatty acid translocases FFA free fatty acids FIT2 fat-storage-inducing transmembrane 2 FoxO1 forkhead box protein 1 Fsp27 fat-specific protein 27 G0S2 G0/G1 switch gene 2 G3P glycerol-3-phosphate G-3-P glycerol-3-phosphate GLUT glucose transporter GPAT glycerol-3-phosphate acyltransferase GSK3B glycogen synthase kinase 3 beta GSTa5 glutathione S-transferase alpha 5 HCC hepatocellular carcinoma HDL high-density HFSD high-fat, high-sucrose diet HLM hypotonic lysis medium HMGCR 3-hydroxy-3-methyglutaryl-CoA reductase HSL hormone-sensitive lipase IDL intermediate-density iLD initial LDs IR insulin signalling JNK Jun NH2-termnal kinase KEGG Kyoto encyclopaedia of and genomes LC-MS/MS liquid chromatography-tandem mass spectrometry LDL low-density LDs lipid droplets

xii LIRKO liver-specific insulin receptor KO LPA lysophosphatidic acid LPLs MAGL monoacylglyceride lipase MAG monoacylgerols MAPK mitogen-activated protein kinase MGAT1 monoacylglycerol-O-acyltransferase 1 MTP microsomal triglyceride transfer protein NAFLD non-alcoholic fatty liver disease Na-K ATPase plasma membrane NASH non-alcoholic steatohepatitis NEFA non-esterified fatty acids P/S penicillin and streptomycin PA phosphatidic acid PBS phosphate-buffered saline PC2 highest variance PCA perchloric acid PCA principal components analysis PDHK1 pyruvate dehydrogenase kinase isozyme 1 PEPCK phosphoenolpyruvate carboxykinase PKA protein kinase A PKC protein kinase C PLIN perilipin PMH primary mouse hepatocytes PPAR peroxisome proliferator-activated receptor PPARs peroxisome proliferator-activated receptors PSM peptide spectral match PTM post-translational modification PUFA polyunsaturated FAs PVDF polyvinylidene fluoride RIPA radioimmunoprecipitation assay RUV-2 remove unwanted variation S100A10 S100 calcium-binding protein A10

xiii SCD stearoyl-CoA desaturases SDB-RPS styrenedivinylbenzene-reverse phase sulfonate SDS sodium dodecyl sulphate SOCS suppressors of cytokine signalling TAG triglyceride TBS tris buffered saline TCA tricarboxylic acid cycle TEAB triethylammonium bicarbonate TFA trifluoroacetic acid UBXD8 UBX domain containing 8 UCH ubiquitin C-terminal UPS ubiquitin proteasome system USP ubiquitin-specific proteases VLDL very-low density

xiv Publications and presentations that have arisen as a direct result of this thesis

Nagarajan S.R.*, Paul M., Krycer J.R., Fazakerley D.J., Sharland A.F., Hoy A.J. Lipid and glucose metabolism in hepatocyte cell lines and primary mouse hepatocytes: A comprehensive resource for in vitro studies of hepatic metabolism. Am J Physiol Endocrinol Metab, 2019.

Nagarajan S.R.*, Brandon A.E., McKenna J.A., Shtein, H.C., Nguyen T.Q., Suryana E., Poronnik P., Cooney G.J., Saunders D.N., Hoy A.J. Insulin and diet-induced changes in the ubiquitin-modified proteome of rat liver. PLoS ONE, 2016.

Nagarajan S.R.*, Su Z., Brandon A.E., Shtein H., Nguyen T., Chaudhuri R., Parker B.J., Suryana E., James D.E., Cooney G.J., Fazakerley D.J., Hoy A.J. Lipid droplet proteome reveals role of key proteins in the pathogenesis of hepatic insulin resistance. Presented at: Charles Perkins Centre Biology Domain Seminar Series; 2018 November 15; Sydney, NSW, Australia.

Nagarajan S.R.*, Su Z., Brandon A.E., Shtein H., Nguyen T., Chaudhuri R., Parker B.J., Suryana E., James D.E., Cooney G.J., Fazakerley D.J., Hoy A.J. Lipid droplet proteome reveals role of key proteins in the pathogenesis of hepatic insulin resistance. Presented at: Keystone Symposia, Organ Crosstalk in Obesity and NAFLD; 2018 January 21-25; Keystone, CO, United States of America.

Nagarajan S.R.*, Su Z., Brandon A.E., Shtein H., Nguyen T., Chaudhuri R., Parker B.J., Suryana E., James D.E., Cooney G.J., Fazakerley D.J., Hoy A.J. Lipid droplet proteome reveals role of key proteins in the pathogenesis of hepatic insulin resistance. Presented at: Charles Perkins Centre Early and Mid-Career Researchers Annual Symposium; 2017 October 5; Sydney, NSW, Australia.

Nagarajan S.R.*, Su Z., Brandon A.E., Shtein H., Nguyen T., Chaudhuri R., Parker B.J., Suryana E., James D.E., Cooney G.J., Fazakerley D.J., Hoy A.J. Lipid droplet proteome reveals role of key proteins in the pathogenesis of hepatic insulin resistance. Presented at: Australian Diabetes Society, Annual Scientific Meeting; 2016 August 24-26; Gold Coast, QLD, Australia.

Nagarajan S.R.*, Su Z., Brandon A.E., Shtein H., Nguyen T., Chaudhuri R., Parker B.J., Suryana E., James D.E., Cooney G.J., Fazakerley D.J., Hoy A.J. Lipid droplet proteome reveals

xv role of key proteins in the pathogenesis of hepatic insulin resistance. Presented at: The Federation of American Societies for Experimental Biology, Lipid Droplets: Dynamic Organelles in Metabolism and Beyond; 2016 July 24-29; Snowmass, CO, United States of America.

Nagarajan S.R.*, Brandon A.E., McKenna J.A., Shtein, H.C., Nguyen T.Q., Suryana E., Poronnik P., Cooney G.J., Saunders D.N., Hoy A.J. Protein ubiquitination in rat liver is insulin sensitive and is altered in insulin resistant high-fat fed rats. Presented at: Chris O’Brien Life House Research Symposium; 2015 September 22; Sydney, NSW, Australia.

xvi Publications that have arisen in concurrence with this thesis

Nassar Z.D., Nagarajan S.R., Hoy A.J., Butler L.M., HSP90 inhibition activates fatty acid oxidative metabolism in prostate cancer cells. (Under preparation).

Balaban S., Nassar Z., Zhang A., Hosseini-Beheshti E., Centenera M., Schreuder M., Lin H., Aishah A., Varney B., Liu-Fu F., Lee L., Nagarajan S.R., Shearer R., Hardie R., Raftopulos N., Kakani M., Saunders D., Holst J., Horvath L., Hoy A.J. Extracellular fatty acids are a major contributor to lipid synthesis in prostate cancer. Mol. Cancer. Res. 2018.

Joshi R., Faruqui N., Nagarajan S.R., Rampatige R., Martiniuk A., Gouda H. Reporting of ethics in peer-reviewed verbal autopsy studies: a systematic review. Int. J. Epidemiol. 2018.

Cairns R., Alvarez-Guaita A., Martinez-Saludes I., Wason S.J., Hanh J., Nagarajan S.R., Hosseini-Beheshti E., Monastyrskaya K., Hoy A.J., Buechler C., Enrich C., Rentero C., Grewal T. Role of hepatic Annexin A6 in fatty acid-induced lipid droplet formation. Exp. Cell. Res, 2017.

xvii Abstract

Non-alcoholic fatty liver disease is defined by the accumulation of intrahepatic triglycerides stored in metabolically active lipid droplets (LDs). Hundreds to thousands of proteins associate on the surface of LDs to regulate the storage and utilisation of the neutral lipids stored inside. Further, proteins from non-canonical processes such as oxidative stress and proteasomal degradation also reside on LDs and contribute to the diverse and dynamic functions of these cytosolic organelles. Hepatocytes are important regulators of glucose, lipid and cholesterol flux and thus accumulation of protein-rich LDs in the liver can disrupt molecular signalling and metabolism that is often linked with whole-body insulin resistance. Resident LD-associated proteins that participate in lipolysis have been of particular interest in elucidating the pathogenesis of insulin resistance in fatty liver disease; however, single target gain- and loss- of function studies of these regulators have provided largely contradictory results. Additionally, there is emerging evidence for the diverse regulation of these proteins, such as translocation, post-translational activation and molecular interactions, that govern their function at the LD more than absolute expression. High throughput proteomic and phosphoproteomic research have revealed large changes in the composition and post- translational modifications of LD proteins following a high-fat diet challenge, most of which have not been pathologically described. Moreover, little attention has been paid to the influence of normal and dysregulated insulin action on the proteome, phosphoproteome and ubiquitome within fatty liver.

Therefore, the overall aim of this project was to quantify high-fat diet and insulin-sensitive changes in the abundance, ubiquitination and of cellular and LD-associated proteins and to identify novel regulation of candidates in the pathogenesis of non-alcoholic fatty liver disease and insulin resistance.

Using a quantitative proteomics approach, this work provided comprehensive profiles on proteins and post-translational modifications of whole livers and LDs from chow or fat-fed rats with or without acute insulin stimulation. The study is the first to show that the ubiquitome is sensitive to insulin and diet and is differentially regulated in steatotic and insulin resistant livers. Further, LD-associated proteins are responsive to insulin signalling and demonstrate dramatic changes in abundance and phosphorylation with hepatic insulin resistance. In particular, phosphorylation and/or ubiquitination of several known lipolytic regulators, such as

xviii ATGL and Plin2 differed more than their total protein expression levels at the LD, suggesting previously unknown multi-layered regulation in hepatocytes. Finally, characterisation of the LD-associated protein, UBXD8, revealed a novel role in selectively promoting PPARa- mediated FA oxidation by regulating various components of the hepatic lipolysome.

Collectively, the results described in this thesis demonstrates that post-translational modifications, subcellular localization and protein-protein interactions are underestimated regulators of key lipolytic processes in the liver. Understanding their dysregulation is important in elucidating the pathogenesis of insulin resistance within fatty liver disease.

xix List of Figures

Figure 1.1 Pathogenesis of NAFLD...... 2 Figure 1.2 Summary of lipid metabolism in the liver...... 5 Figure 1.3 Triglyceride Synthesis Pathway ...... 9 Figure 1.4 Relationships between IR and Hepatic TAG Accumulation ...... 16 Figure 1.5 Ubiquitin-proteasome system...... 19 Figure 1.6 Lipid droplet biogenesis, growth and expansion...... 26 Figure 1.7 LD contact sites with organelles...... 27 Figure 3.1 Animal workflow for LD proteomic and phosphoproteomic enrichment ...... 51 Figure 3.2 Purification of hepatic lipid droplets...... 61 Figure 3.3 Quality control of proteome and phosphoproteome ...... 63 Figure 3.4 Total proteome, phosphoproteome and associated enriched pathway analysis. .... 65 Figure 3.5 Enriched pathway analysis of differentially expressed proteins...... 67 Figure 3.6 Enriched pathway analysis of differentially expressed phosphosites...... 69 Figure 3.7 Volcano plots of proteome and phosphoproteome...... 72 Figure 3.8 Modelling of insulin-resistance in hepatic lipid droplet proteome...... 75 Figure 3.9 Modelling of insulin resistance in hepatic lipid droplet phosphoproteome ...... 85 Figure 5.1 Tissue distribution of UBXD8 protein expression...... 150 Figure 5.2 Quantification and Visualization of UBXD8 in Response to Insulin...... 151 Figure 5.3 Quantification of UBXD8 Protein Expression In Response to Oleate ...... 152 Figure 5.4 ATGL Protein Expression with UBXD8 KD ...... 153 Figure 5.5 TAG Content in UBXD8 Knocked Down Hepatocytes ...... 154 Figure 5.6 Lipolytic and Lipogenesis Protein Expression ...... 156 Figure 5.7 Role of UBXD8 in Fatty Acid Oxidation ...... 158 Figure 5.8 Differentially Expressed Proteins Associated with UBXD8 KD ...... 160 Figure 5.9 UBXD8 Intracellular Movement Across Cholesterol Pools ...... 165 Figure 5.10 Cholesterol-Related Protein Expression with UBXD8 KD ...... 166 Figure 5.11 Molecular Interactors of UBXD8 and Associated Lipid Phenotype ...... 169 Figure 5.12 Model of UBXD8 in FA Oxidation ...... 171 Figure 6.1 Proposed Model of Ubiquitin Dysregulation of PFKL ...... 176 Figure 6.2 Proposed Model of PTM Dysregulation of Plin2 ...... 179

xx List of Tables Table 3.1. Filters used to identify proteins and phosphosites in scenarios 1-4...... 57 Table 3.2. Primary antibodies used in this chapter...... 59 Table 3.3 Secondary antibodies used in this chapter...... 59 Table 3.4 List of hepatic lipid-droplet associated proteins in scenario 1 ...... 76 Table 3.5. List of hepatic lipid-droplet associated proteins in scenario 2 ...... 77 Table 3.6. List of hepatic lipid-droplet associated proteins in scenario 3 ...... 79 Table 3.7. List of hepatic lipid-droplet associated proteins in scenario 4 ...... 82 Table 3.8. List of hepatic lipid-droplet associated phosphosites in scenario 1 ...... 86 Table 3.9. List of hepatic lipid-droplet associated phosphosites in scenario 2 ...... 87 Table 3.10. List of hepatic lipid-droplet associated phosphosites in scenario 3 ...... 90 Table 3.11. List of hepatic lipid-droplet associated phosphosites in scenario 4 ...... 92 Table 3.12. Lipolytic regulators identified as ‘insulin resistant’...... 92 Table 5.1 Primary antibodies used in this chapter...... 143 Table 5.2 Secondary antibodies used in this chapter...... 144 Table 5.3. List of downregulated proteins in AML12 hepatocytes with UBXD8 KD ...... 161 Table 5.4. List of upregulated proteins in AML12 hepatocytes with UBXD8 KD ...... 163

xxi

Introduction

1

1.1. Introduction Non-alcoholic fatty liver disease (NAFLD) is a metabolic condition largely fuelled by the alarming epidemic of obesity. Although the prevalence of NAFLD has not been well established, systematic reviews predict that about 20-30% of individuals in Western countries have NAFLD, with an increasing number of affected of patients observed in Eastern countries (Younossi et al. 2016). Historically, the condition is referred to as a spectrum that includes hepatic steatosis, steatohepatitis (NASH), fibrosis, cirrhosis, hepatocellular carcinoma (HCC) and liver failure (Fig 1.1); however, clinical studies have shown that simple steatosis practically never develops into NASH (Bedogni et al. 2014). Therefore, ‘NAFL(D)’ is used in this thesis as a broader term for fatty liver, as suggested by the American Gastroenterological Association (Chalasani et al. 2012) and is defined as intrahepatic triglyceride content greater than 5% (Kleiner et al. 2005). The accumulation of lipids in insulin-sensitive organs, such as the liver, is often associated with perturbed lipid and glucose homeostasis, which is characteristic of insulin resistance and is almost universally presented in patients with NAFL (Marchesini et al. 1999). Over 76% of type 2 diabetics, defined as long-term insulin resistance, have been reported to have NAFL(D), whereas diabetes was identified in over 23% of NAFL(D) cases, indicating a strong link between the two conditions. A necessary feature of their coexistence is an imbalance of uptake, synthesis, oxidation and export of fatty acids.

Figure 1.1 Pathogenesis of NAFLD. The majority of NAFL(D) patients develop steatosis, or TAG accumulation in the liver, without progressing further. A small subset may develop non-alcoholic steatohepatitis (NASH), which may develop into cirrhosis and hepatocellular carcinoma. Figure adapted from (Cohen et al. 2011).

In the liver, esterified lipids are primarily stored in dynamic organelles called lipid droplets (LDs). The LD is composed of a neutral lipid core surrounded by a monolayer of phospholipids and associated proteins involved in energy homeostasis, particularly lipid flux, protein trafficking and signal transduction (Vanni 2017). It is likely that hormonal and transcriptional

2 regulation of proteins in close proximity to stored accessible lipids in LDs underlie the metabolic pathology that leads to NAFL(D), and therefore dysfunctions in this organelle make it a promising direction for research. This Introductory Chapter presents an overview of the pathogenesis of NAFL and discusses available theories linking excess triglycerides with impaired insulin action, followed by an in depth analysis of LD physiology and its role in health and disease.

1.2. Non-Alcoholic Fatty Liver

General Characteristics of NAFL(D) A liver is classified as steatotic when there is more than 5% of triglycerides accumulated within hepatocytes in the absence of any other secondary causes such as medications or alcohol (Bedogni et al. 2014). While liver biopsies are the gold-standard for the diagnosis of NAFL, liver ultrasonography is often employed as a less invasive method that can be used in the general population outside of specialist liver centres. When ultrasonography is used, the minimum detection for accurate diagnosis starts from an intrahepatic triglyceride content of 10% (Adams et al. 2005, Machado and Cortez-Pinto 2013). The majority of fatty liver patients present with macrovesicular steatosis, defined as lipid accumulation in few medium to large lipid droplets that displace nuclei to the periphery, as opposed to numerous smaller droplets (Tandra et al. 2011). Treatment options for fatty liver include weight loss, exercise, bariatric surgery and pharmacological therapy aimed at treating comorbidities of NAFL, such as insulin resistance and dyslipidaemia (Tolman and Dalpiaz 2007). Patients with fatty liver have increased mortality due to cardiovascular disease despite relatively passive histological progression (Chalasani et al. 2012). For the <4% of individuals with fatty liver that do progress to cirrhotic and fibrotic stages of liver injury, liver-related mortality is more common (Boutari et al. 2018, Mahady and Adams 2018). Due to the variability in stage classification, assessment and treatment, prognosis is unclear. There is currently no known or approved cure for NAFL or NASH-related outcomes.

1.2.2 Prevalence of NAFL(D) NAFL(D) is the most common cause of chronic liver disease in the world (Marchesini et al. 2008), with over 66 million individuals affected with NAFL in the United States alone (Estes et al. 2018). The global prevalence of NAFL(D) is estimated to be 24%, similar to the prevalence in North America, with higher estimates in the Middle East (32%), South America

3 (30%) and Asia (27%), and lower estimates in Europe (24%) and Africa (14%) (Younossi et al. 2016). In Australia, there are a limited number of epidemiological studies on NAFL(D) in the general population and even fewer with accurate tools. However, a recent meta-analysis estimated a prevalence of 25% in Caucasians and over 40% in ethnic groups (Mahady and Adams 2018). Hispanics are especially susceptible to increased hepatic fat content, which is evidenced by a high frequency of the PNPLA3-148M allele that is strongly associated with NAFL(D) (Romeo et al. 2008). Further, the prevalence of NAFL(D) steadily increases with age, with 22% in individuals in their 30’s and up to 34% in individuals in their 70’s (Younossi et al. 2018), similar to the pattern of age-related prevalence for obesity (Ng et al. 2014). Additionally, men with NAFL(D) tend to be in their youth or middle-age, whereas women display higher prevalence in the development of fatty liver post-menopause, likely when the protective effects of oestrogen decrease (Fan et al. 2005, Eguchi et al. 2012, Hamaguchi et al. 2012, Ballestri et al. 2017, Summart et al. 2017). It is important to note that studies that measure liver enzyme levels to diagnose NAFL markedly underestimate its prevalence compared to tools such as sonography or biopsy and it is likely that prevalence in all categories are, in reality, significantly higher than reported (Khodadoostan et al. 2016).

1.2.3 Pathophysiology of NAFL(D) NAFL is primarily characterized by excess accumulation of triglycerides within the liver. Although considerable quantities of fatty acids are handled by the liver, less than 5% are stored in the form of triglycerides under normal physiological conditions (Dowman et al. 2010). Free fatty acids, which are esterified with glycerol to form triglycerides, can arise from dietary sources, adipose tissue lipolysis and via de novo lipogenesis using substrates such as pyruvate, acetate and amino acids (Ipsen et al. 2018). In contrast, hepatic fatty acid levels are balanced by beta oxidation, fatty acid re-esterification for storage into lipid droplets or exported with cholesterol into circulation as very-low density lipoproteins (Fig 1.2). Therefore, an accumulation of triglycerides can be attributed to an increase in delivery and synthesis or a decrease in oxidation and export.

4 1.2.3.1 Mechanisms of TAG Accumulation 1.2.3.1.1 Hepatic Fatty Acid Uptake Over 90-95% of lipids derived from dietary sources are composed of triglycerides, while the remaining 5%-10% consists of phospholipids, free fatty acids, sterols and fat soluble vitamins (Iqbal and Hussain 2009). In the small intestine, these meal-derived lipids are broken down

Figure 1.2 Summary of lipid metabolism in the liver. Accumulation of TAG in the liver can occur from either an increased influx or decreased efflux of FFA. Lipids derived from adipose tissue lipolysis and dietary sources can be taken up by the liver to be converted into TAG for storage in LDs. TAG can then be repackaged and secreted as VLDL or oxidized in mitochondria to generate energy or ketone bodies. Figure adapted from (Cohen et al. 2011). into fatty acids and monoglycerides for absorption into enterocytes where they are re- synthesized as TAG and secreted into circulation in small lipoprotein transport particles called (Rui 2014). -TAGs are then subjected to hydrolysis by lipoprotein lipases (LPLs) at priority tissues such as adipose and muscle to release non-esterified FAs (NEFAs) for storage and oxidation, respectively (Lambert and Parks 2012). Remaining TAG in chylomicron remnants, as well as released FAs that were not taken up by adipose or muscle tissues, are then transported to hepatocytes for LPL-mediated lipolysis and fatty acid uptake. Interestingly, morbidly obese patients with hepatic steatosis have increased protein levels and activity of hepatic LPL before bariatric surgery compared to control patients (Pardina et al. 2009). Similarly, liver-specific LPL overexpression increased intrahepatic TAG content and

5 liver-specific insulin resistance due to increased delivery of FAs (Kim et al. 2001), demonstrating the importance of LPLs in regulating hepatic lipid homeostasis.

During fasting, the major contributor of NEFAs to the liver are those released from intracellular lipolysis of TAG pools in adipose tissue (Frayn 2002). In the postprandial state, hormonal regulation by insulin inhibits adipose-derived FA contribution to hepatocytes in healthy individuals (Chakrabarti et al. 2013). Stable isotope labelling of NEFAs in NAFL(D) patients prior to liver biopsy revealed over 50% of hepatic TAG arose from systemic adipose-derived NEFAs (Donnelly et al. 2005). Although there was no control group in the study, there is strong evidence from other studies demonstrating the lack of insulin suppression of adipose tissue lipolysis in patients with hepatic steatosis and related insulin resistance (Thorne et al. 2010, Schweiger et al. 2017).

FAs primarily enter hepatocytes via CD36/fatty acid translocases (FATs), fatty acid transport protein 2 (FATP2), FATP4 and FATP5. FATP5 is exclusively expressed in the liver and loss of function in mice decreased hepatic long-chain FA uptake and lower intracellular TAG and FFA levels (Doege et al. 2006). Further, liver specific deletion of CD36 protected mice fed a 60% high-fat diet from developing hepatic steatosis (Wilson et al. 2016). Likewise in , the histological grade of steatosis in 227 NAFLD patients correlated with progressive increases in soluble CD36 levels in serum (Garcia-Monzon et al. 2014), demonstrating a positive relationship between fatty acid uptake and hepatic steatosis.

1.2.3.1.2 Intracellular Fatty Acid Activation and Transport Once taken up, FAs must be activated by long chain acyl-CoA synthetases (ACSLs) in order to form acyl-CoAs and enter almost any downstream metabolic pathway, including TAG synthesis and beta oxidation. Coenzyme A, or CoA, is a highly conserved polar molecule with an adenosine 3’,5’-disphosphate moiety that increases the affinity of CoA to enzymes, thereby acting as a necessary substrate for numerous FA enzymatic reactions, including elongases, desaturases, dehydrogenases, , and many more (Davaapil et al. 2014). It has been hypothesized that different ACSL isoforms target FAs of specific chain-length groups and channel newly formed FA-CoAs into distinct tissue-specific downstream pathways. For instance, ACSL 1 and 5 are highly expressed in the liver and target 12 to 20 carbon FAs to channel into synthesis of phospholipids and cholesterol, or incorporation of exogenous FAs

6 into TAG, respectively (Mashek et al. 2007). Furthermore, ACSL isoforms also exhibit organelle specificity, whereby ACSL1 is detected in the mitochondria, ACSL3 and 4 have been identified in lipase-activated LDs, and ACSL5 and 6 are enriched in cholesterol-rich regions of the plasma membrane (Soupene and Kuypers 2008). While these enzymes are important for the downstream synthesis of complex lipids from fatty acids (Bu and Mashek 2010), their role in hepatic steatosis and metabolic disease remains largely unsettled (Ellis et al. 2010, Yan et al. 2015).

To increase the solubility of hydrophobic molecules such as long-chain FA and FA-CoAs in aqueous environments, fatty acid binding proteins (FABPs) and acyl carrier binding proteins (ACBPs) facilitate their transport in the cytosol (Wang et al. 2015). Although they do not have any catalytic activity themselves, the importance of binding proteins in FA utilisation is reflected in the impairment of FA uptake and oxidation pathways in liver-specific FABP null mice (Newberry et al. 2003), consistent with decreased FABP protein levels found in steatotic rodent models (Guzman et al. 2013). This is likely due to the role that FABPs play in FA- mediated ligand activation of peroxisome proliferator-activated receptor alpha (PPARa) transcription factors and their downstream effects on lipid metabolism (Tan et al. 2002). Additionally, FABPs and ACBPs both promote the increase in intracellular LCFA-CoA pool size by binding to the end-products of ACSLs and preventing its feedback inhibition (Huang et al. 2005). This collectively functions as a cytoprotectant from lipotoxic free FAs (Gajda et al. 2013).

1.2.3.1.3 Hepatic de novo Lipogenesis The most striking result of Donnelly’s stable isotope study in NAFL(D) patients (2005) was the increase in intrahepatic triglyceride content that originated from de novo lipogenesis (DNL), with almost 30% from DNL in individuals with NAFL(D) compared to less than 5% in healthy individuals (Parks 2002). Lipogenesis is primarily driven by dietary carbohydrates, whereby the main glycolytic product, pyruvate, is broken down in the mitochondria by pyruvate dehydrogenase complex to provide acetyl-CoA as the carbon source for DNL, thereby linking glycolysis to lipogenesis (Rui 2014). Although there are other lipogenic tissues, such as adipose and mammary glands, the major site for DNL in response to acute or prolonged carbohydrate feeding in humans is the liver (Diraison et al. 2003).

7 The rate-limiting step of DNL is the conversion of acetyl-CoA to malonyl-CoA, which is catalysed by acetyl-CoA carboxylases (ACC). ACC1 is found in tissues where synthesis is important, such as adipose and mammary tissue, whereas ACC2 is enriched in oxidative tissues such as skeletal muscle and , while the liver is highly enriched for both of these isoforms. Despite distinct enrichment, ACC1 and 2 are largely compensatory and thus evidence of their role in NAFL pathogenesis has been controversial. For example, antisense oligonucleotide (ASO)-mediated double KD of ACC1 and 2 in mice led to decreased hepatic fat and increased insulin sensitivity as expected (Savage et al. 2006). However, liver-specific ACC double KO in mice in another study unexpectedly revealed increased hepatic fat accumulation, decreased fatty acid oxidation and increased protein acetylation, likely due to activation of compensatory fat storage pathways (Chow et al. 2014).

Following carboxylation, malonyl-CoA is converted to palmitate, a 16-carbon saturated fatty acid, by fatty acid synthase (FAS). Palmitate can be elongated by acyl-CoA elongase (Elovl) to form long-chain FAs or desaturated by stearoyl-CoA desaturases (SCD) to form mono- and polyunsaturated FAs (Drag et al. 2017). Interestingly, several studies have demonstrated a decrease in n-3 and n-6 polyunsaturated FAs (PUFAs) in NAFL patients (Arendt et al. 2015), and that supplementation of long PUFAs derived from essential FAs can reduce hepatic steatosis after 12 months (Capanni et al. 2006). Likewise, ELOVL5 deletion contributed to the pathogenesis of NAFL in mice (Moon et al. 2009), while hepatic SCD1 activity in humans showed the inverse effect (Kotronen et al. 2009). Collectively, these studies suggest that more FA desaturation may be required when not provided by dietary sources.

1.2.3.1.4 Lipid Esterification into Triglycerides When FAs are not immediately required for cellular processes, they are esterified to glycerol and stored as TAG in LDs. In hepatocytes, glycerol is derived as glycerol-3-phosphate (G3P) from three main processes: 1) glycolysis, forming G3P from a critical branchpoint of glucose breakdown by glycerophosphate dehydrogenase; 2) partial reverse glycolysis or glyceroneogenesis, from the conversion of lactate and pyruvate and; 3) uptake of circulating glycerol from other tissues and conversion to G3P by glycerokinase (Festuccia et al. 2003). FA-CoAs are added to G3P sequentially (Fig 1.3), each step catalysed by a specific enzyme. G3P acyltransferases (GPATs) and acylglycerol-3-phosphate acyltransferases (AGPATs) reside primarily in the ER and mitochondria but can also be found on newly synthesized LDs

8 (Wilfling et al. 2013, Wilfling et al. 2014). They generate lysophosphatidic acid (LPA) and phosphatidic acid (PA), respectively, by acylation of G3P (Takeuchi and Reue 2009). Although these enzymes are important for TAG synthesis, their role in the pathogenesis of hepatic steatosis is unclear. When investigating TAG synthesis in hepatocytes isolated from GPAT1, 2, 3 or 4 KO mice, GPAT1 was the only isoform that diverge de novo synthesized FAs away from oxidation and incorporated into TAG in the liver (Wendel et al. 2013). Consequently, the role of GPAT1 in aggravating NAFL has since been suggested, however ob/ob mice lacking GPAT1 were not protected from steatosis when fed a high-fat diet long-term (Yazdi et al. 2008), indicating a different culprit in the pathway.

Diacylglycerols (DAGs) act as an important branchpoint of glycerophospholipid formation and precursor for TAG synthesis, and are one of the most studied lipid intermediates due to their involvement in various metabolic reactions as signalling molecules. Diacylglycerols (DAGs) can then be formed via the dephosphorylation of PA by lipin, otherwise known as PA -1, or via the esterification of monoacylgerols (MAGs) by MAG O-acyltransferase 1 (MGAT1). Although DAGs are thought to play a pivotal role in lipotoxicity and IR (discussed in detail in 1.3.3.1), their position as intermediates in lipid synthesis and breakdown pathways makes it difficult to distinctly associate their pool size with impaired lipid metabolism.

Figure 1.3 Triglyceride Synthesis Pathway Triglyceride (TAG) synthesis encompasses numerous enzymes in a multi-step pathway. Glycerol-3-phosphate (G-3-P) acyltransferase (GPAT) adds a FA-CoA to G-3-P to form lysophosphatidic acid (LPA), followed by phosphatidic acid (PA) by 1-acylglycerol-3-phosphate O-acyltransferase (AGPAT), and diacylglycerol (DAG) by lipin. DAG can also be made from the addition of an FA-CoA to monoacylglycerol (MAG). DAG is then converted to TAG by DAG acyltransferases (DGAT).

The final pivotal step of TAG synthesis is catalysed by DAG acyltransferases (DGATs) that add the final FA-CoA to generate TAG for safe storage in lipid droplets or for very-low density lipoproteins (VLDL) synthesis (Yen et al. 2008). Although the two isoforms, DGAT 1 and 2, are generally compensatory, DGAT2 is more potent and is associated with more TAG accumulation when overexpressed than DGAT1 (Zammit 2013). Furthermore, DGAT2 is

9 proposed to be primarily involved in glycerolipid synthesis whereas DGAT1 may play a more important role in re-esterification of hydrolysed TAG, suggesting that DGAT2 is a critical regulator of DNL rates in the development of hepatic steatosis (Monetti et al. 2007, Yen et al. 2008). The source-dependent use of FAs by the two DGAT isoforms can be linked to their differential proximity to key proteins., Specifically, DGAT2 has been found to interact with enzymes that catalyse the de novo synthesis or transport of FAs and MAGs, such as SCD1, MGAT2 and FATP1 (Wurie et al. 2012, Xu et al. 2012, Jin et al. 2014). Interestingly, high fat- fed mice with decreased DGAT2 expression, and not DGAT1, was associated with reduced hepatic TAG levels (Choi et al. 2007), similar to studies done in humans with two or more copies of a minor DGAT2 mutation (Kantartzis et al. 2009).

1.2.3.2 Mechanisms of TAG Disposal 1.2.3.2.1 Triglyceride Secretion as VLDL Triglycerides in the liver can be delivered to adipose or muscle tissue by secretion as very low- density lipoproteins (VLDL), defined as triglyceride-rich, cholesterol-poor transport particles similar in composition to lipid droplets (Mahley et al. 1984). Lipoproteins are characterized by the attachment of core structural (Apo) that are distinct to each lipoprotein category, with ApoB-100 identifying hepatic VLDL particles (Feingold and Grunfeld 2000). The formation of nascent VLDL begins with the translocation of ApoB-100 on ribosomes at the ER and subsequent lipidation of the protein by microsomal triglyceride transfer protein (MTP) (Choi and Ginsberg 2011). VLDLs undergo further maturation in the Golgi apparatus by receiving a bulk of lipids from cytosolic lipid droplets before eventually moving to the plasma membrane for secretion into the . Once in circulation, VLDLs become progressively more dense as triglycerides get released as FAs into tissues. As the proportion of lipoprotein cholesterol content increases, particles are gradually defined as low-density (LDL), intermediate-density (IDL) and high-density (HDL) lipoproteins. Individuals with familial hypobetalipoproteinaemia, characterized by a mutation in the ApoB gene, have decreased VLDL-TAG secretion and increased intrahepatic TAG levels (Schonfeld et al. 2003). In contrast, NAFL patients actually report a heightened rate of VLDL-TAG secretion, despite increased intracellular TAG stores (Adiels et al. 2006). Interestingly, individuals with normal intrahepatic TAG levels exhibit progressive increases in VLDL-TAG secretion with greater FA delivery to the liver (Fabbrini et al. 2008). In contrast, NAFL patients are able to upregulate VLDL-TAG synthesis and secretion only up to a certain ceiling, thereafter capping the export

10 of lipids and resulting in increased intracellular storage and extracellular export of TAG (Fabbrini et al. 2008). Therefore, TAG disposal is an important mechanism that the liver utilizes in order to offset increased FA availability in hepatic steatosis.

1.2.3.2.2 Intrahepatic Lipolysis During times of energy deprivation, TAG stores can be hydrolysed to release FAs for signalling, stimulate b oxidation, and ketone body generation (Hodson and Frayn 2011). Neutral lipases on the surface of lipid droplets work in a stepwise manner to catalyse the degradation of TAG. The successive hydrolysis of TAG to DAG and monoacylglycerols (MAG) is catalysed by adipose triglyceride lipase (ATGL), followed by hormone-sensitive lipase (HSL) and finally monoacylglyceride lipase (MAGL), respectively. Overexpression of ATGL and HSL in ob/ob mice ameliorates hepatic steatosis by increasing TAG breakdown leading to the upregulation of b oxidation (Reid et al. 2008). In the liver, the rate-limiting lipolytic enzyme ATGL is regulated by two proteins: the ATGL coactivator, comparative gene identification-58 (CGI-58) and the inhibitor G0/G1 switch gene 2 (G0S2). Humans with CGI- 58 autosomal recessive disease and ASO-mediated knockdown of CGI-58 in mice decreased lipolysis and increased hepatic steatosis (Yamaguchi and Osumi 2009, Cantley et al. 2013). Similarly, liver overexpression of G0S2 in mice resulted in excess TAG in the liver, suggesting that ATGL-mediated availability of FFAs, regulated by activators and inhibitors, is important for in the pathogenesis of hepatic steatosis (Wang et al. 2013, Zhang et al. 2014). Finally, after FAs are released from hydrolysis by cytoplasmic lipases, LDs localize and fuse with mitochondria to enable efficient oxidative metabolism and avoid lipotoxicity (Jagerstrom et al. 2009, Rambold et al. 2015), as discussed in section 1.2.3.2.3.

1.2.3.2.3 Hepatic b-oxidation The liver has one of the highest metabolic rates of any organ in the body, nearly 20 times more than the resting rate of skeletal muscle and 50 times greater than that of adipose tissue (Klein and Jeejeebhoy 2002). Long-chain FAs from adipose tissue lipolysis or degradation of LD TAG stores in the liver are converted to acyl-CoAs before being sequentially broken down in the mitochondria (Fabbrini et al. 2010). Carnitine palmitoyltransferase-1 (CPT1) and 2 (CPT2) are present on the outer and inner mitochondrial membranes, respectively, and facilitate the transport of FA-CoAs into the mitochondria by converting them into acylcarnitines (Nassir and Ibdah 2014). Through a series of dehydrogenation, hydration

11 and cleavage reactions, two-carbon units of acetyl-CoA are progressively removed from FA-

CoA each cycle to generate the coenzymes NADH and FADH2. The acetyl-CoA can then be subsequently used to generate ketone bodies or be completely oxidized by entering the tricarboxylic acid cycle (TCA) (Hodson and Frayn 2011). The TCA cycle generates additional

NADH and FADH2 that are substrates for the electron transport chain to generate usable chemical energy in the form of ATP. The electron transport chain contains five complexes that play critical roles in oxygen consumption and energy generation. A current therapeutic target for the treatment of NAFL is the stimulation of peroxisome proliferator-activated receptors (PPARs), which are ligand-activated transcriptions factors, via agonists (Seo et al. 2008). Specifically, PPARa and g modulate oxidative genes such as CPT1 and 2 to enhance fatty acid oxidation and improve steatosis. Furthermore, overexpression of the rate-limiting oxidative gene CPT1a in mice significantly attenuated hepatic steatosis by 35% and was associated with a 70% increase in beta-oxidation and 60% decrease in TAG secretion (Stefanovic-Racic et al. 2008). Interestingly, 2H and 13C NMR analysis in African Americans and Hispanics with NAFLD showed higher rates of lipolysis associated with a 2-fold increase in mitochondrial oxidative metabolism and production of reactive oxygen species, especially in the progression of NASH (Sunny et al. 2011). It is possible that FA oxidation represents two sides of a coin in distinguishing simple steatosis from patients that progress to NASH in the pathogenesis of NAFLD.

1.2.3.3 Co-morbidities 1.2.3.3.1 Obesity Obesity is a multifaceted and generally preventable metabolic disease, affecting over 650 million individuals and almost 25% of the worldwide population in 2016; nearly triple the prevalence since 1975 (World Health Organization 2018). Although not the most precise measurement of body fat proportion, obesity is commonly defined as having a body mass index (BMI) greater than or equal to 30 kg/m2 height (Ofei 2005). Individuals with obesity have been long associated with a disproportionately higher risk of developing comorbid conditions, including fatty liver disease. In a meta-analysis of 21 cohort studies with over 380,000 participants, obese individuals had a 3.5-fold greater risk of developing NAFL compared to control patients and demonstrated a clear dose-dependent relationship between NAFL risk and BMI (Li et al. 2016). In morbidly obese Australian patients, the prevalence of NAFL increases to a staggering 83-96% of the cohort when diagnosed by liver biopsy (Dixon et al. 2001, Ooi

12 et al. 2018). This not only demonstrates a strong association between obesity and NAFL, but also suggests the influence of diagnostic tools in determining the true prevalence of NAFL. However, it is not uncommon for lean individuals to develop NAFL, with roughly 7-19% of lean individuals diagnosed with the disease worldwide, compared to roughly 29-76% in obese individuals (Kumar and Mohan 2017). While there are a few distinct characteristics of lean NAFL individuals compared to their obese counterparts, such as a preponderance to the female sex, younger age and lower risk for NAFLD progression, they still demonstrate higher BMI, blood pressure and prevalence of metabolic syndrome compared to healthy individuals (Kumar and Mohan 2017). Research in Asian populations with lean and obese NAFL patients indicates a unifying feature between the two groups linked to localisation of adiposity (Mohan et al. 2009, Kumar and Mohan 2017). In particular, centralised or abdominal obesity from fat accumulation in the visceral compartment, commonly measured by waist circumference, is a better predictor of NAFL and IR than generalized adiposity or obesity, thereby providing a likely distinction between “metabolically obese” versus “physically obese” cases (Chitturi et al. 2011, Mirza 2011, Ha et al. 2015). Although race and genetics likely play a role in determining variability in fat distribution, it is clear that there is a strong relationship between obesity, specifically central obesity, and NAFL development.

1.2.3.3.2 Cardiovascular Disease There is an increased prevalence for the development of cardiovascular diseases (CVD), including and coronary heart disease, in patients with NAFL. Hepatic steatosis has been diagnosed in various cardiovascular conditions, including 20% of cerebrovascular disease (Targher et al. 2007), 43% of ischaemic stroke patients (Moshayedi et al. 2014), 51% of mild coronary stenosis patients and 100% of patients with three affected coronary arteries (Choi et al. 2013). Additionally, a longitudinal study on roughly 1000 Finnish participants found that the presence of severe hepatic steatosis was a predictor for the occurrence of a cardiovascular event within a 19-year follow-up period, whereby 29% of NAFL patients experienced coronary or a stroke compared to 14% of non-NAFL patients (Pisto et al. 2014). In particular, individuals with NAFLD are reported to have increased carotid intima-media thickness, coronary artery calcification, arterial stiffness and impaired flow-mediated vasodilation (Oni et al. 2013), all of which are reported to be predictive of future CVD and stroke events (Darabian et al. 2013). The association of CVD and hepatic lipid accumulation can be explained by reports of hypertriglyceridemia, increased VLDL secretion and localized

13 lipolysis of lipoproteins in conjunction with heightened release of thrombotic factors such as fibrinogen and inflammatory factors such as IL-6 and TNF-alpha (Linton et al. 2000). The inflammatory response is thought to occur due to a build-up of apoB-containing lipoproteins in the walls of arteries, promoting atherogenic and atherothrombotic consequences. Not only has epidemiological research shown that NAFL can precede and exacerbate CVD events, but longitudinal cohort studies have proven CVD to be the most common cause of death among NAFL patients, especially those within the 45-54 age range (Adams et al. 2005, Dunn et al. 2008). Early patient education on the importance of recognizing symptoms of co-pathologies of NAFL can improve cardiovascular mortality and patient outcomes.

1.2.3.3.3 Insulin Resistance Lean and obese patients with NAFLD almost universally exhibit hepatic IR, which substantially increases the risk of developing subsequent type 2 diabetes mellitus by 2 to 5-fold (Armstrong et al. 2014). Conversely, the estimated prevalence of NAFLD in diabetic patients is reportedly between 60-87% depending on the country, compared to 20-30% in the general population (Targher et al. 2007, Prashanth et al. 2009, Chalasani et al. 2012, Doycheva et al. 2013). IR is defined as a whole organism or target organ’s attenuated response to insulin, resulting in increased insulin production and secretion by the pancreas in order to maintain blood glucose levels in the healthy range (Shanik et al. 2008). When beta cells can no longer produce enough insulin to overcome the demand, blood glucose levels remain elevated and T2DM ensues.

Insulin has multiple direct and indirect effects on the gluconeogenic and lipogenic pathways of the liver that contribute to insulin resistance (Brown and Goldstein 2008). In the postprandial state, insulin directly acts on forkhead box protein 1 (FoxO1) to inhibit the transcription of gluconeogenic genes such as PEPCK and G6Pase to reduce hepatic glucose output. Insulin also simultaneously stimulates glycogen synthase to promote glycogen synthesis (Matsumoto et al. 2006). When hepatic glycogen stores reach maximum capacity, insulin acts on SREBP-1c to enhance transcription of lipogenic genes such as ACC and FAS for conversion of glucose into FAs and TAG via DNL (Foretz et al. 1999). Further, hepatic gluconeogenesis is indirectly inhibited by suppressing adipose tissue lipolysis and the subsequent release of NEFAs for hepatic uptake (Bergman 2000). In ob/ob mice which are insulin resistant, the FoxO1 pathway regulating gluconeogenesis is impaired, while the SREBP-1c pathway in charge of lipid

14 synthesis remains insulin sensitive, thereby causing high levels of hepatic gluconeogenesis and lipogenesis (Shimomura et al. 2000). Excess DNL-derived TAG is secreted in VLDL to other tissues, such as adipose and muscle, further exacerbating systemic whole-body insulin resistance. A time-course comparison of muscle, adipose and liver insulin resistance supported the notion that lipid accumulation initiates insulin resistance at the liver after just 1 week of high fat-feeding, followed by peripheral tissues after three weeks (Turner et al. 2013). Although selective hepatic insulin resistance can lead to elevated TAG levels in the liver and increase risk for NAFL, it is likely that central obesity and TAG accumulation in the liver initiates and provokes hepatic insulin resistance, which will be discussed in greater detail in the following section.

1.2.3.3.4 Complexities to Uncouple NAFLD, IR and Obesity The coexistence of elevated lipid levels, basal or fasting insulin levels and IR make it particularly difficult to determine unidirectional causality. Although the concept of obesity- induced IR causing hyperinsulinemia is well accepted, there is also evidence supporting the causal role of hyperinsulinemia in the pathogenesis of IR and obesity (Fig 1.4) (Shanik et al. 2008). Increased pancreatic secretion of insulin causes desensitization of the target cell to insulin (Gavin et al. 1974). Mice transfected with extra copies of the insulin gene displayed two to four times higher plasma insulin levels compared to controls in the basal state, which were associated with reduced binding of the insulin receptor, increased postprandial glucose, and elevated TAG levels (Marbani and Roth 1996). Desensitization of the insulin receptor to insulin, due to complete inactivation of Foxa2 (Wolfrum et al. 2004), is therefore thought to cause IR, which has been directly associated with hepatic lipid accumulation and NAFL. Indeed, hyperinsulinemia in obese Zucker rat hepatocytes increased CD36 mRNA, translocation of CD36 protein to the plasma membrane and subsequent hepatic FA uptake, all of which preceded obesity and NAFL (Buque et al. 2012, Steneberg et al. 2015).

Admittedly, the relationship between IR and NAFL is cyclical due to the nature of their pathogenesis and is therefore difficult to determine causality. While IR can be induced by elevated visceral FA levels, it can also provoke or exacerbate lipid accumulation, thereby acting as a cause and consequence simultaneously. Furthermore, hyperinsulinemia acts as both a compensatory mechanism of impaired insulin signalling (IR), as well as a driver of insulin desensitization and lipid synthesis. In reality, the direction of causality is not as important as

15 the factors that affect the known features of metabolic disease, particularly hormonal regulation of energy metabolism. The effect of insulin on LD homeostasis, the availability of lipid intermediates such as DAGs and ceramides, and the regulation of ubiquitination in metabolism are all key pieces of a larger and more complex puzzle.

Figure 1.4 Relationships between IR and Hepatic TAG Accumulation Obesity is associated with increased energy input that is primarily stored as lipids in adipose tissue. When adipose reaches its capacity, lipid accumulation occurs in ectopic tissues such as liver and muscle in the form of TAG. Fatty liver is intimately linked with insulin resistance in the liver and whole body insulin resistance. Insulin resistance is associated with a lack of suppression of adipose tissue lipolysis and hepatic glucose production while selectively activating DNL, thereby exacerbating lipid accumulation and blood glucose levels. Impaired insulin action and elevated blood glucose levels signal for more secretion of insulin from beta cells, which causes further desensitization of the body to insulin.

1.3. Linking Lipids with IR Signalling Obesity has long been correlated with insulin resistance. More than 50 years ago, Randle and colleagues (Randle et al. 1963) postulated that lipids and carbohydrate metabolic pathways were related in the form of a cycle and that this relationship played a role in impairing insulin sensitivity in muscle and adipose tissues. However, studies have shown that steatotic liver can also become insulin resistant in the form of dysregulated hepatic glucose production. There are four distinct, yet interrelated, hypotheses for the mechanisms that contribute to its pathogenesis. Elevated lipid levels in peripheral tissues can lead to increased FFA flux into the liver and affect gluconeogenic and lipogenic pathways. Moreover, increased degradation of key metabolic proteins due to obesity may impair intracellular signalling. Additionally, active

16 lipid accumulation in the form of DAGs and ceramides can act as signals to trigger novel kinases and subsequent perturbations. Finally, faulty LD dynamics may lead to increased hepatic FA availability and impede cellular homeostasis. The following sections describe these hypotheses in detail. 1.3.1 Hepatic Glucose Output The liver is one of the three main insulin-sensitive tissues in the body and is integral in facilitating the insulin-mediated suppression of gluconeogenesis, a process that is impaired in IR. Elevated postprandial levels of insulin bind to insulin receptors in hepatocytes to activate Akt and induce the downstream phosphorylation of FOXO1 (Matsumoto et al. 2007). FOXO1 is a that promotes the transcription of key gluconeogenic genes such as PEPCK and G6Pase (Wang et al. 2014). Phosphorylated FOXO1 is then translocated from the nucleus to the cytosol, causing subsequent suppression on gluconeogenesis. Meanwhile, PEPCK is the rate-limiting enzyme that converts oxaloacetate to phosphoenolpyruvate, while G6Pase catalyses the last step in the generation of glucose from glucose-6-phosphate.

Although insulin-mediated suppression of gluconeogenesis is well accepted, the mechanism controlling hepatic glucose output remains to be defined. It was previously believed that insulin had a linear and direct effect on gluconeogenesis and output from the liver; however, insulin secreted from pancreatic beta cells can circulate either directly or indirectly in relation to the liver (Bradley et al. 1993, Lewis et al. 1996). When insulin is secreted into the portal vein, it reaches the liver immediately due to the lack of endothelial barrier in hepatic capillaries and acts on suppressing gluconeogenesis (Cherrington 1999). In contrast, insulin that is targeted for peripheral tissues such as adipose or muscle work on decreasing gluconeogenic precursors available to the liver, thereby acting indirectly (Sindelar et al. 1997). Interestingly, the rate of suppression of glucose output in the liver and the rate of increased glucose disposal in peripheral tissues is similar, suggesting an indirect extrahepatic effect. This explanation, named the Single Gateway Hypothesis, postulates that insulin may travel to adipose tissue and inhibit lipolysis, thereby decreasing plasma FFA levels and subsequently reducing hepatic gluconeogenesis (Bergman et al. 1993, Rebrin et al. 1995). Thus, the single gateway for insulin in this context is the endothelium of extrahepatic tissues. However, the validity of this hypothesis has been questioned several times with the characterisation of liver-specific insulin receptor KO (LIRKO) mice, whereby loss of the insulin receptor in the liver causes hyperinsulinemia and severe IR (Michael et al. 2000). Elevated plasma insulin levels in LIRKO

17 mice provides further evidence for the direct binding of insulin to its hepatic receptor and the subsequent receptor-mediated degradation of insulin at the liver, responsible for almost 75- 80% of total insulin clearance (Duckworth et al. 1998). It is likely that both direct and indirect effects are critical for normal insulin action and downstream metabolic homeostasis in the liver.

1.3.2. Ubiquitin-Proteasomal Degradation The ubiquitin-proteasome system is a tightly coordinated process required to preserve cellular homeostasis, and its dysregulation can lead to inappropriate degradation of specific proteins implicated in the pathogenesis of IR. Ubiquitination involves three steps regulated by distinct classes of enzymes to attach a ubiquitin molecule to the target protein for degradation (Fig 1.5) (Hershko and Ciechanover 1998). Ubiquitin is first activated by ubiquitin-activating enzyme (E1), then transferred to the cysteine site of ubiquitin-conjugating enzyme (E2), which catalyses the transfer of ubiquitin to ubiquitin ligase (E3), through which it is finally linked to residues of the protein target (Stewart et al. 2016). Additional ubiquitin molecules are added to one of the seven lysine residues on the primary ubiquitin, eventually forming a chain of numerous ubiquitin chains to signal the proteasome for degradation which can occur in a variety of ways (Schrader et al. 2009). One mechanism of signalling degradation occurs via the recognition of the polyubiquitinated substrate by 19S regulatory proteins and transfer to 20S proteasomal regulatory particles for cleavage into short peptides (Finley 2009). Alternatively, ubiquitinated substrates can be transported to the proteasome by adaptor proteins that can bind to both ubiquitin chains and the proteasome (Elsasser et al. 2004). Although most proteins require at least monoubiquitylation, there are substrates that have an intrinsic high affinity for the proteasome and do not require ubiquitination for proteasomal degradation. Ubiquitinated substrates can be deubiquitinated by ubiquitin C-terminal hydrolases (UCH) and ubiquitin- specific proteases (USP) and are involved in disassembling polyubiquitin chains and regulating signal transduction of a variety of pathways (Wilkinson 2000).

18

Figure 1.5 Ubiquitin-proteasome system. Proteins are degraded by the 26S proteasome via recognition of a multi-ubiquitin chain attached to substrate proteins. Ubiquitin molecules are added to substrates by the successive action of E1-3 enzymes. Ubiquitin molecules can be removed from substrate proteins by deubiquitinating enzymes (DUBs).

Expression levels of several proteins involved in the ubiquitin-proteasome system are altered in patients with obesity, NAFL and IR. In particular, overall total intracellular protein proteolysis is increased in metabolic disease (Balasubramanyam et al. 2005, Ishii et al. 2015). The downregulation of deubiquitinating enzymes USP 4, 10 and 18 protein levels in hepatic steatosis and IR suggest inappropriate degradation of key proteins that regulate lipid accumulation and insulin action (An et al. 2017, Luo et al. 2018, Zhao et al. 2018). Furthermore, E3 ubiquitin ligases such as MG53, Nedd4 and Trim32 specifically target IRS1/2, reduce its phosphorylation and downstream AKT/FOXO1 signalling (Yang et al. 2016). Increased proteasomal activity is likely associated with obesity and subsequent chronic low-grade inflammation (Wunderlich et al. 2013, Pereira and Alvarez-Leite 2014). Suppressors of cytokine signalling (SOCS) 1 and 3 proteins are E3 ubiquitin ligases that upregulate with severe lipid accumulation and, like other E3 enzymes, affect downstream insulin signalling (Rui et al. 2002). Interestingly, insulin action is associated with inhibition of degradation, and this is impaired in IR, thereby exacerbating abnormal proteasomal activity on other protein targets (Russell-Jones and Umpleby 1996). These targets include key lipogenic regulators such

19 as PPARa and SREBP and likely contribute to the pathogenesis of selective IR in the liver and further steatosis (Hirano et al. 2001, Blanquart et al. 2002).

1.3.3. Bioactive Lipid Signalling The unifying pathological feature of both lean and obese NAFL is the increase in FFA flux into the liver. FAs can be converted to an array of distinct lipid species that not only serve as building blocks for cell membranes or sources of energy, but act as pathophysiological mediators of intracellular processes. Bioactive lipids influence cell signalling functions and can be divided into four main families depending on their biochemical structure and function: classical eicosanoids such as prostaglandins, specialized pro-resolving mediators such as docosahexaenoic acids (DHA) and eicosapentaenoic acids (EPA), glycerolipids or sphingolipids such as ceramides and DAGs, and cannabinoid-receptor specific eicosanoids called endocannabinoids (Chiurchiu et al. 2018). Currently, ceramides and DAGs are considered two putative regulators of lipid-induced hepatic IR.

1.3.3.1. Diacylglycerols Intrahepatic DAG content is a strong predictor of whole-body and liver-specific IR in obese and NAFLD patients, even more so than BMI, ER stress, cytokine concentrations and long- chain FA-CoA content (Kumashiro et al. 2011). Indeed, pharmacological treatment to reduce total intrahepatic DAG levels in rodent models significantly attenuated IR, regardless of distinct species pools (Perry et al. 2013, Tao et al. 2014, Perry et al. 2015). DAGs are versatile lipids that can be generated from a myriad of metabolic reactions, each resulting in distinct isomers of DAG depending on the enzymes and precursors involved. For instance, TAGs have three possible sites for lipase action which can result in three different isomers of DAG: sn- 1,2, sn-2,3 or the less common rac-1,3, which has unknown stereochemistry (Eichmann and Lass 2015). In contrast, precursors such as phospholipids and PAs which utilize /transferases or , respectively, can only result in sn-1,2 DAG isoforms. Although DAGs can exist in three different stereo/regioisoforms and originate from FAs of a variety of lengths and saturation states, research indicates that the majority of DAG species correlate positively with hepatic IR (Wang et al. 2014, Luukkonen et al. 2016). When the cytosolic content of over 13 distinct DAG species were measured in insulin sensitive and insulin resistant livers of obesity-induced NAFL patients, all species except C20:4-C20:5 and C18:0 were increased in the insulin resistant samples (Ter Horst et al. 2017).

20

DAGs are proposed to impair insulin signalling through the activation of novel or atypical protein kinase C (PKC) isoforms. DAG-mediated activation of PKCs can directly phosphorylate residues and IRS-1 and the insulin receptor and consequently inhibit downstream PI3-kinase/Akt pathway (Schmitz-Peiffer and Biden 2008). Furthermore, PKCs act upstream by activating other kinases such as Jun NH2-termnal kinase (JNK), p42/44 mitogen-activated protein kinase (MAPK), and inhibitor of nuclear factor kB kinase (IKK)-b to specifically modulate insulin signalling (Khoshnan et al. 2000, Aguirre et al. 2002). Although PKCs as therapeutic targets seem promising, KO and KD studies show contrasting results. Inhibition of certain PKC isoforms have improved insulin sensitivity in mostly short- term HFD studies (Samuel et al. 2004, Samuel et al. 2007, Schmitz-Peiffer et al. 2007) but not long-term models, likely due to alternative compensatory isoforms (Gao et al. 2007, Schmitz- Peiffer and Biden 2008).

1.3.3.2. Ceramides Apart from DAGs, liver ceramide levels have also been reported to be elevated in NAFLD patients (Yetukuri et al. 2007) and promote IR through a similar mechanism to DAGs. Ceramides can be produced from the re-acylation of sphingosine or from the condensation of palmitate and serine, both of which use ceramide synthase (CERS) to catalyse the final step (Levy and Futerman 2010). The generation of ceramides of different chain lengths is facilitated by 6 distinct CERS isoforms, some more associated with the development of obesity and IR than others (Kanety et al. 1996, Chavez et al. 2003, Stratford et al. 2004). Inhibition of its endogenous synthesis in obese rodents reverses IR (Holland et al. 2007) via restored Akt phosphorylation (Holland and Summers 2008). In particular, the C14-C16 fatty acyl chains generated by CERS6 specifically has been shown to block PKC-mediated Akt translocation for downstream insulin signalling (Reali et al. 2017), although it is unknown whether its inhibition rescues insulin action. More research into liver-specific isoforms of CERS and IR in rodents and humans need to done to understand its role in NAFL.

1.3.4. Leaky Lipid Droplets From a reductionist point of view, steatosis occurs from an imbalance in synthesis and breakdown of TAG; however, the difficulty in elucidating the aetiology of NAFL(D) is that a single universal pathway responsible for its pathogenesis is highly unlikely. The majority of

21 research on intracellular hepatic FA trafficking in NAFL and IR have focused on FFA uptake and storage, but some evidence suggests that compensatory FA disposal may also cause harm (Mashek 2013). The inappropriate release of FAs from LDs via lipolysis can thus be referred to as the ‘leaky LD hypothesis’ for the pathogenesis of IR in NAFL. Indeed, sequestration of TAG in LDs highly likely protects against lipotoxicity and thus, the attempt by hepatocytes to alleviate steatosis by releasing FAs into the cytosol may have deleterious consequences (Listenberger et al. 2003). This is especially true for long-chain saturated FAs such as stearate and palmitate, which are known to cause lipotoxic stress, proinflammatory signalling and mitochondrial dysregulation in hepatocytes (Listenberger et al. 2003, Ogawa et al. 2018). Although the direct relationship between dysregulated LD lipolysis and pathogenesis of disease in non-adipocytes has not been extensively studied, intrahepatic lipolysis has been shown to be 50% greater in individuals with NAFLD in one tracer study (Sunny et al. 2011). However, a look at the downstream effects of lipolytic pathways, namely VLDL-TAG secretion and oxidation can allude to leaky LDs. For example, although it is well known that VLDL secretion is heightened in NAFLD participants (Fabbrini et al. 2008), 70% of the FAs that contribute to the VLDL-TAG are derived from the re-esterification of FAs released from hepatic TAG lipolysis (Gibbons and Wiggins 1995). Further, systemic sources of FAs such as those released from adipose tissue lipolysis made up 43% of that incorporated into VLDL-TAG in NAFLD patients, and is therefore less important in hepatic FA utilization than hepatic DNL or TAG- derived FAs (57%) (Fabbrini et al. 2008). Similarly, intrahepatic TAG lipolysis is also a source of FAs for mitochondria oxidation. While many studies have reported impaired hepatic FA oxidation in NAFLD (Hashimoto et al. 2000, Ibdah et al. 2005, Zhang et al. 2007, Stefanovic- Racic et al. 2008, Rector et al. 2010), others have described excessive lipolysis-induced TCA cycling in hepatic steatotic and IR liver (Sunny et al. 2011, Satapati et al. 2012) and greater 13 incorporation of U- C-labelled FAs into CO2 in patients’ breaths (Hodson et al. 2010), suggesting heightened mitochondrial oxidation of hepatic-derived FAs in patients with NAFL. Therefore, it is likely that impaired control of FA release from LDs (i.e. leaky LDs) in the liver is an underestimated perturbation in the pathogenesis of NAFL and IR, alongside other pathways known to be impaired including FA uptake and oxidation.

1.4. Lipid Droplets Intracellular storage of neutral lipids in LDs is a process that is evolutionarily conserved in almost all organisms, including prokaryotes and eukaryotes (Thiam and Beller 2017). These

22 complex organelles vary in size, spatial organization, number and composition across species and cell types. In mammalian adipocytes for example, LDs are generally unilocular and large, whereas hepatocytes have multiple smaller LDs (Natarajan et al. 2017). Since the discovery of proteins on LDs in the 1990’s, studies have demonstrated their importance in managing FA utilization, ER and oxidative stress, protein turnover and even in modulating immune responses towards pathogens (Khatchadourian and Maysinger 2009, Hapala et al. 2011, Welte 2015, Bersuker and Olzmann 2017, den Brok et al. 2018). Extensive research has been specifically conducted on the role of LDs in the pathogenesis of metabolic diseases, such as obesity, NAFLD, and type 2 diabetes due to the central role that lipids and proteins in LDs play in TAG esterification, lipolysis, oxidation and secretion. This section will summarize what is currently known about LD structure, formation and interactions and elaborate on the relationship between LDs and disease.

1.4.1. General Properties LDs are composed of a phospholipid monolayer coated with an array of proteins that surround a hydrophobic core of neutral lipids. Although the lipid composition may differ between species and cell types, the majority of LDs contain TAG, cholesterol esters (CE) and retinyl esters (Khor et al. 2013). Recently, Senkal and colleagues (Senkal et al. 2017) reported the sequestration of acylceramides in hepatic LDs for the first time in mice, indicating a potentially novel mechanism by which LDs exacerbate IR in NAFL. Hepatocytes generally have multiple smaller LDs approximately 10 µm in diameter, but can reach up to 100 µm in steatotic livers (Kochan et al. 2015). The lipid monolayer consists of free cholesterol and a variety of phospholipids that are thought to help with maintaining LD stability and facilitate interactions with other organelles (Bartz et al. 2007). Over 50 different species of phosphatidylcholine make up 50-60% of the phospholipid content of the LD membrane, followed by roughly 45 types of phosphatidylethanolamine, and up to nine species of phosphatidylinositol (Krahmer et al. 2011). Interestingly, LD surfaces have less cholesterol and more phospholipids than other membranes, However, during periods of LD growth or shrinkage, membranes can accumulate more lipid intermediates such as DAG, FAs and sterols to act as fluidifiers (Thiam et al. 2013).

A large number of proteins associated with LD surfaces have been identified across species and cell types in the last two decades. There are thought to be 50 core proteins that are found in the majority of mammalian LDs and can act as distinguishing markers of this organelle. The

23 perilipin (PLIN) family are integral LD proteins that upregulate with increased LD growth and abundance (Itabe et al. 2017). Out of the five isoforms, PLIN2, 3 and 5 are expressed in the liver and have a variety of functions related to LD generation and stabilization. Overexpression of PLIN2 in liver promoted TAG storage in LDs and protected against autophagy in lipid-rich environments (Tsai et al. 2017). Similarly, PLIN5 facilitates lipid storage by disrupting the interaction of ATGL with its coactivator, CGI-58 in order to prevent lipolysis (Wang et al. 2015). This protective affect against lipotoxicity exacerbates hepatic steatosis without affecting insulin resistance (Trevino et al. 2015). There has also been a large number of additional proteins that have been identified on LDs, ranging from 8-200 proteins in and yeast to generally over 1000 in humans and rodents (Jolivet et al. 2004, Wu et al. 2009, Xie et al. 2010, Ding et al. 2012, Horn et al. 2013, Krahmer et al. 2013, Currie et al. 2014, Khan et al. 2015). The protein composition of LDs largely depends on the environment and life cycle stage they are in. For instance, LDs can be considered nascent or mature, localize in the cytosol or the nucleus, be small or steatotic in size, and interact with mitochondria for FA oxidation or ER for lipid synthesis; all of which can influence LD proteome.

1.4.2. The Lipid Droplet Life Cycle 1.4.2.1. Biogenesis and Growth As described in 1.2.3.1.3, TAGs can be synthesized via the DNL pathway or through re- esterification of MAGs and DAGs, both of which are catalysed by DGAT enzymes in the final step (Yen et al. 2008). DGAT1 is exclusively localized at the ER and primarily synthesises TAG from the re-esterification of recycled DAG, whereas DGAT2 is found in both the ER and LDs and is mainly involved in DNL-mediated TAG generation from the esterification of newly-formed DAG (Yen et al. 2008). TAG is formed at the ER wherever these two enzymes are found in excess and they are deposited into the ER bilayer until the membrane forms a bulge of lipids called an ‘oil lens’ (Hamilton and Small 1981, Wilfling et al. 2014) (Fig 1.6). Lenses are formed to reduce the costs of entropy when disrupting the ER bilayer during TAG formation. The ER is composed of different domains that vary in their curvature, where sheets are considered flat with no curvature and tubules have curvature in one or both directions of the leaflet (Nixon-Abell et al. 2016). Lens and subsequent LD formation occurs in tubules and therefore proteins that are involved in maintaining ER tubule structure can affect LD biogenesis (Jacquier et al. 2011). For example, atlastin is an ER protein that is critical for the fusion of ER network tubules and is required for LD expansion from the ER (Klemm et al. 2013).

24 Knockdown of atlastin is associated with less dense ER networks, reduced TAG content, and formation of smaller LDs with diameter less than half of controls. Likewise, LD budding and expansion from the ER are affected by lipid composition and ER proteins, although the exact mechanisms remain unclear. Fat-storage-inducing transmembrane 2 (FIT2) is an ER protein that colocalizes with newly emerging LDs and was recently reported to have lipid phosphate phosphatase activity associated with modified LD phospholipid composition (Becuwe et al. 2018). Membrane phospholipids are key components in adjusting LD curvature for emergence from the ER. Further, FIT2 regulates DAG levels in LDs as an additional point of control in LD expansion , whereby a reduction in DAG is required for budding to occur (Choudhary et al. 2018). As more TAG is deposited in between the bilayer, the contact angle because the newly forming LD and the ER increases until eventually fission occurs (Thiam and Foret 2016). It is hypothesized that seipins are required to form ER membranes with extreme curvature, such as the angle required for budding of newly formed LDs (Han et al. 2015). Although the full extent of seipin’s function remains unclear, they were recently reported to collect and transfer phosphatidylamines to phosphatases to reduce surface tension (Ben M'barek et al. 2017). Elevated levels of phosphatidylamines in membranes of emerging LDs increases ER surface tension, thereby blocking LD growth (Yan et al. 2018). SThe importance of seipins in maintain appropriate phospholipid composition for LD biogenesis is supported by the presence of aberrant LDs in seipin-deficient HeLa cells (Yan et al. 2018).

Proteins can associate with LDs by either migrating to LDs via ER-LD membrane bridges, defined as class I proteins, or by targeting to the LD surface from the cytosol, termed class II proteins (Kory et al. 2016). Membrane bridges can be stabilized by Arf and COPI/ protein machinery, allowing for the relocation of TAG synthesis enzymes such as DGAT2 and GPAT4 from the ER to the LD surface (Wilfling et al. 2013, Wilfling et al. 2014). Nascent LDs with localized enzymes for lipid synthesis are called expanding LDs (eLDs), which are also initial LDs (iLDs) that are now specialized . TAG synthesis enzymes on eLDs facilitate the expansion and growth of LDs that have separated from the ER. When lipolysis and subsequent LD shrinkage is required, class I proteins generally remain bound due to their higher affinity compared to class II proteins (Rowe et al. 2016). In contrast, class II proteins can associate to LDs via hydrophobic domains such as the amphipathic helices in PLIN proteins that act as LD-binding motifs (Li et al. 2017). Other unclassified proteins may also bind to LDs using lipid anchors, such as the palmitoylation of ELMOD2 (Suzuki et al. 2015)

25 or via protein-protein interactions, such as the Jabba-mediated LD association of histones in Drosophila (Li et al. 2012).

Figure 1.6 Lipid droplet biogenesis, growth and expansion. TAG is synthesized and accumulated within the bilayer membrane of the ER, causing an oil lens to bulge. The lens expands until an LD forms and buds off the membrane as nascent LDs. Figure taken from (Walther et al. 2017).

1.4.2.2. Lipid Droplets as Dynamic Interactors Traditionally considered as mere depots for fat storage, LDs are now known as dynamic organelles with a multitude of functions, some of which entail association with other key players (Fig 1.7). Being a major metabolic hubs of the cell, LDs require tight communication with almost every organelle in the cell to maintain energy homeostasis. For example, contact between LDs and ER are primarily concerned with lipid synthesis and protein storage, whereas interactions with mitochondria and peroxisomes are necessary for FA catabolism and ATP generation. Furthermore, LDs interact with each other to share neutral lipid pools, mediate pressure differences or alter their lipase interface. Being one of the most active and social organelles, transport and trafficking of LDs are a key method of communication during nutrient load, cellular stress and much more.

26

Figure 1.7 LD contact sites with organelles. Molecular tethers are utilized by LDs to link with organelles, particularly mitochondria, peroxisome, ER and other LDs, to modulate lipid metabolism. Figure taken from (Schuldiner and Bohnert 2017).

1.4.2.2.1. Contact Sites and Transport of Lipid Droplets The majority of organelles, such as the ER, mitochondria and nucleus, consist of an aqueous lumen with at least one amphipathic bilayer membrane to separate the core from the cytosol. In contrast, LDs have a concentrated hydrophobic core with a single phospholipid monolayer to separate the lipids from the outer hydrophilic environment. Unsurprisingly, the distinct anatomy of LDs from other organelles necessitates alternative mechanisms for LD organellar contact instead of regular fission and fusion. Contact sites are positions between two organelles that mediate the transfer of lipids, ions or proteins without compromising the structural integrity of either organelle (Schuldiner and Bohnert 2017). Regardless of the organelle, almost all contact sites enlist molecular tethers to increase stability and stay in close proximity to each other, usually between 10-70 nm. Additionally, proteins may also reside at contact sites to facilitate a specific type of interorganellar transfer (called effectors) or to adapt features of contact site such as size and abundance (called regulators) (Eisenberg-Bord et al. 2016). Interactions with membrane-bound organelles are possible via bidirectional LD movement on microtubules (Welte et al. 2005). Indeed, LDs engage with the cytoskeleton and even house motor proteins on its surface to aid in moving LD cargo to and from organelles. Although LD interactions and transport are fairly similar, there are distinctions that can be made between the various organelles.

27 1.4.2.2.2. Protein Transfer with Endoplasmic Reticulum While biogenesis of LDs occurs at the ER, recent findings suggest that it is possible to form lipidic bridges between the LD and ER, even after droplets have reached full maturation (Schuldiner and Bohnert 2017). Lipidic bridges are defined as hemifusion intermediates that are continuous with the LD monolayer and the outer leaflet of the ER membrane bilayer. It is hypothesized that the majority of LD-associated proteins are ER residents and protein targeting between the two organelles requires contact via these bridges (Wilfling et al. 2013, Wilfling et al. 2014). Arf1/COP1 activity can result in the formation of nano-LDs via the removal of phospholipids from the LD surface (Wilfling et al. 2014). Subsequent increase in LD surface tension allows the formation of membrane bridges between the LD and ER when in close proximity. Proteins targeted for LD localization can then move from ER to LDs by crossing the bridge. Similar studies using chemicals to raise surface tension in cells without Arf/COP1 are consistent with this model (Wilfling et al. 2014). Contact between the ER and LD are crucial in supplying an LD proteome that is sufficient to meet energy demands. Failure to appropriately curate and provide key metabolic proteins for lipid homeostasis can cause serious defects (Grippa et al. 2015).

1.4.2.2.3. Mitochondrial and Peroxisomal Lipid Droplet Catabolism TAG stored in LDs can be hydrolysed by lipases to release energy-rich FFAs to the mitochondria for ATP generation via b-oxidation. Release of FAs into the cytosol at high concentrations can be cytotoxic to the cell, therefore it is beneficial for LDs to maintain close contact with mitochondria during times of energy deprivation (Klecker et al. 2017).

Studies on the exact mechanism of LD-mitochondria contact are relatively sparse; however, there are several proteins reported to play a role in their interaction. The bona fide LD protein PLIN5 is expressed in oxidative tissues such as heart, skeletal muscles, adipocytes and the liver and has been reported to localise to the mitochondria when stimulated by the PKA activator forskolin (Wang et al. 2011). Indeed, PLIN5 overexpression induced more mitochondrial recruitment to LDs in cardiac myocytes, supposedly via the 20 amino acids on the C-terminal of the protein (Wang et al. 2011). Similarly in adipose tissue, PLIN1 knockdown rather than PLIN5 was associated with a reduction in LD-mitochondria interaction as evidenced by decreased colocalization of PLIN1 with the outer mitochondrial membrane proteins (Boutant

28 et al. 2017). Although other PLIN proteins are not reported to affect LD contact with mitochondria, it is possible that the PLIN-specific mechanism may be dependent on the tissue.

Peroxisomes also play a critical role in FA utilization, including complete b-oxidation, TCA and ETC in yeast and plants, and partial oxidation of very long chain FAs in eukaryotes (Poirier et al. 2006). Interestingly, peroxisomes initiate contact with LDs in a unique manner using protrusions called pexopodia (Binns et al. 2006). These protrusions are enriched for oxidative proteins and can intrude into the LD core to allow for intimate involvement of the two organelles in lipid breakdown.

1.4.2.2.4. Lipid Exchange between Lipid Droplets Interaction between multiple LDs, called a homotypic contact, is not uncommon, despite the lack of information on the protein machinery involved in facilitating their association (Murphy et al. 2009). One of the only players known are a family of cell death-inducing DFF45-like effector (CIDE) proteins which appear to be localized at contact sites between LDs (Gong et al. 2011). It is hypothesized that CIDE-A proteins tether LDs via an amphipathic a-helix that is present on two domains; however, it is uncertain whether CIDE-B, the liver isoform, possesses the same features (Gong et al. 2009, Barneda et al. 2015). It is important to note that CIDE-mediated LD-LD interactions always result in directional fusion of the droplets, whereby the neutral lipids from the smaller droplet are transferred to the larger droplet (Gong et al. 2011). Exchange of lipid content is likely due to higher internal pressure in smaller LDs compared to larger ones. While CIDE proteins enable focal contact between LDs, fat-specific protein 27 (Fsp27) has been shown to increase the number of LDs greater than 5 µm in diameter by almost 8-fold in adipocytes (Gong et al. 2011). Together, CIDE and Fsp27 mediate LD growth via lipid transfer of interacting LDs with size disparity. More research is required to ascertain whether LDs can interact with each other without fusion.

1.4.3. Lipid Droplet Proteome in Metabolic Disease Given the importance of LD-associated proteins in lipid homeostasis, it is of great significance to understand how these proteins are altered in NAFL and hepatic insulin resistance. Although there are numerous proteomic studies of the LD, only a handful have been investigated in models of liver disease. The following section outlines the advancements of this niche field and identifies the major holes that are left to investigate.

29

1.4.3.1. Lipids Droplet Proteome and Non-Alcoholic Fatty Liver Recent proteomic investigations of LDs have been revealing the intricacies and dynamics of hepatic LDs in a variety of environments, especially in periods of excess lipid supply. Although lipolytic regulators are the most prominent LD proteins examined in the pathogenesis of NAFL (discussed in detail in section 1.2.3), several proteomic studies have reported that the majority of LD proteins that alter in abundance or localization with HFD-induced hepatic steatosis are not primarily involved in lipid metabolism. In patients with simple steatosis, pathways such as retinol, xenobiotic, drug and linoleic acid metabolism were increased compared to patients without fatty liver (Su et al. 2014). Similarly in high fat-fed mice, the LD proteome revealed upregulation of ribosomal, ER and xenobiotic metabolism while glucose metabolism and liver X receptor signalling were downregulated (Khan et al. 2015). Whole organellar protein differences allow researchers to identify novel and unexpected candidates that could cause more disruption in lipid metabolism than obvious targets. For example, S100 calcium-binding protein A10 (S100A10) was found to be one of 101 proteins that were enriched in hepatic LDs by 10-fold with high-fat feeding (Liu et al. 2017). Although historically associated with the endo- and exocytosis of neurotransmitters (Milosevic et al. 2017), knockdown of S100A10 led to increased hepatic steatosis by inhibiting lipid transport and trafficking (Liu et al. 2017). In another proteomic study, the oxidoreductase 17b-hydroxysteroid dehydrogenase 13 (17b-HSD 13) was identified on LDs in human steatotic liver and reported to induce lipogenesis and promote steatosis in vitro (Horiguchi et al. 2008, Su et al. 2014). Despite the value of high throughput research in understanding the role novel LD proteins in NAFL, the vast majority of studies take a more targeted and selective approach using known regulators.

It is important to note that lipid stimulation intrinsically associated with large changes in protein localization, activation and association, even in non-diseased states (Krahmer et al. 2018). In fact, with gradual diet-induced hepatic lipid accumulation across 12 weeks, over 20% of proteins are re-localized, with 6% of all proteins moving towards LDs. It is hypothesized that expanding LDs have larger surface areas and therefore proteins may move to LDs as a way to regulate protein depletion, acting as a ‘protein sink’. In steatotic mice, LDs were found to have increased abundance of contact site proteins, potentially facilitating the transfer of proteins across organelles to LDs. If this were true, LDs would consequently be at the forefront of regulating a vast number of pathways that can be indirectly linked to symptoms of metabolic

30 syndrome. Thus, more research into the role of LDs in mediating proteomic shifts in metabolic disease is required to discover truly novel players that connect these pathways.

1.4.3.2. Lipid Droplet Proteome and Insulin Resistance The intracellular accumulation of TAG in hepatocytes is strongly associated with the development of IR (see section 1.2.3.3.3 for more detail). However, work over the past several decades has shown that perturbed pathways in NAFL and IR are inextricably linked, making it hard to identify specific players involved in impaired insulin signalling at the LD level without being confounded by the effect of lipids. Since insulin resistance is fundamentally defined by impaired insulin action, investigating how the LD proteome is altered with stimulation by this hormone is a logical step forward in this field. Strikingly, there are no known high throughput studies that have explored the relationship between insulin stimulation and proteomic activation, translocation or interaction at the whole LD level, and how they change in the insulin resistant setting. Instead, much of the literature has focused on targeted investigation into the insulin-stimulated regulation of major lipolytic regulators.

1.4.3.2.1. Regulation of LD Proteins by Insulin Several proteins critical for lipolysis are phosphorylated by protein kinase A (PKA), a downstream regulator of the cyclic AMP second messenger system (Duncan et al. 2007) and inhibited by AMP-activated protein kinase (AMPK). In particular, HSL is phosphorylated at sites Ser563, Ser659 and Ser660 by PKA as well as the synthetic lipolytic activator isoproterenol and the catecholamines glucagon and adrenaline (Anthonsen et al. 1998, Roepstorff et al. 2004, Duncan et al. 2007). In contrast, Ser 565 is phosphorylated to inhibit HSL lipolytic activity, demonstrating the strict regulation of phosphorylation events. Additionally, PLIN1 is phosphorylated by PKA to facilitate the translocation of HSL to the LD surface (Shen et al. 2009). Similarly, PKA phosphorylation of Ser406 on ATGL and Ser239 on CGI58 further promote TAG hydrolysis in adipose tissue (Pagnon et al. 2012, Sahu-Osen et al. 2015). Insulin inhibits these pathways by Akt-mediated phosphorylation of 3B, an inhibitor of cAMP levels and PKA activity (DiPilato et al. 2015). It is important to note that the majority of these phosphorylation events are reported in adipocytes and have not been elucidated in hepatocytes, most likely due to the relatively greater role that adipocyte lipolysis plays in obesity-induced IR (see section 1.3.1). Although the role that insulin plays in regulating lipid and glucose metabolism is largely known, the exact pathways

31 and mechanistic events involved are largely unknown for the vast number of proteins, including key metabolic regulators. Understanding insulin and diet effects both independently and as a whole on LD proteins can provide great insight into their dysregulation in insulin resistance and NAFL, and eventually help identify potential drug targets.

32 1.5. Summary and Thesis Aims NAFL(D) is defined by the accumulation of TAG in the liver and is often associated with the development of IR. NAFL(D) or obesity-induced IR is primarily characterized by sustained hepatic glucose production, DNL and adipose-specific TAG lipolysis with insulin stimulation. The dysregulation of other pathways that affect lipid homeostasis, such as FA uptake, b- oxidation and VLDL-TAG secretion have also been implicated in the pathogenesis of NAFL(D) and IR. The central hub where the majority of these processes function is at cytosolic LDs which are dynamic organelles that store neutral lipids such as CE, TAG and possibly acylceramides. LDs host 100-1000s of proteins on its surface that not only facilitate FA utilization but also mediate an array of other pathways that have no straightforward connection to classical LD function. Many LD-associated proteins have been previously known to be residents of other organelles, further emphasizing the complexity of LD behaviour interorganellar interactions. In selected proteins, insulin and diet have been shown to alter proteomic translocation, activation and interaction at LDs but no known studies have investigated these changes at a whole LD proteome level.

The thesis contains four research projects that are intended to be read independently but weave together to provide new knowledge on the role of proteomic, phosphoproteomic and ubiquitomic changes in HFD-induced IR in the liver. The first research project (Chapter 2) focuses on insulin and diet effects of ubiquitin-proteasomal protein degradation in the liver to investigate changes in protein abundance of key metabolic regulators in NAFL or IR. The second research project (Chapter 3) delves a little deeper to understand insulin and diet effects on protein translocation and phosphorylation/activation at the LD level, and how this alters with hepatic IR. To assist with the validation of high-throughput data, the metabolic physiology of a number of hepatocyte cell lines were characterized to identify the most suitable cell line to use for validation, which the results are described in Chapter 4. The chosen cell line was then used to characterize the function of a novel protein candidate from Chapter 3 in lipid and glucose metabolic pathways to elucidate its role in the pathogenesis of metabolic disease. Together, this thesis acts as the first step-wise analysis into and insulin-induced changes in whole and LD protein regulation and how it contributes to NAFL and IR.

33

34

Chapter 2

Role of Ubiquitination in Obesity-Induced Insulin Resistance in the Liver

35 RESEARCH ARTICLE Insulin and diet-induced changes in the ubiquitin-modified proteome of rat liver

Shilpa R. Nagarajan1☯, Amanda E. Brandon2,3☯¤, Jessie A. McKenna4,5☯, Harrison C. Shtein1, Thinh Q. Nguyen1, Eurwin Suryana2, Philip Poronnik1, Gregory J. Cooney2,3¤, Darren N. Saunders4,5*, Andrew J. Hoy1*

1 Discipline of Physiology, School of Medical Sciences & Bosch Institute, Charles Perkins Centre, University of Sydney, Sydney, NSW, Australia, 2 Diabetes and Metabolism Division, Garvan Institute of Medical Research, Darlinghurst, NSW, Australia, 3 St Vincent’s Clinical School, Faculty of Medicine, University of New South Wales, Sydney, NSW, Australia, 4 Kinghorn Cancer Centre, Garvan Institute of Medical a1111111111 Research, Darlinghurst, NSW, Australia, 5 School of Medical Sciences, Faculty of Medicine, University of a1111111111 New South Wales, Sydney, NSW, Australia a1111111111 a1111111111 ☯ These authors contributed equally to this work. a1111111111 ¤ Current address: Discipline of Physiology, School of Medical Sciences & Bosch Institute, Charles Perkins Centre, University of Sydney, Sydney, NSW, Australia. * [email protected] (AJH); [email protected] (DNS)

OPEN ACCESS Abstract

Citation: Nagarajan SR, Brandon AE, McKenna JA, Ubiquitin is a crucial post-translational modification regulating numerous cellular processes, Shtein HC, Nguyen TQ, Suryana E, et al. (2017) but its role in metabolic disease is not well characterized. In this study, we identified the in Insulin and diet-induced changes in the ubiquitin- modified proteome of rat liver. PLoS ONE 12(3): vivo ubiquitin-modified proteome in rat liver and determined changes in this ubiquitome e0174431. https://doi.org/10.1371/journal. under acute insulin stimulation and high-fat and sucrose diet-induced insulin resistance. We pone.0174431 identified 1267 ubiquitinated proteins in rat liver across diet and insulin-stimulated condi- Editor: Manlio Vinciguerra, University College tions, with 882 proteins common to all conditions. KEGG pathway analysis of these proteins London, UNITED KINGDOM identified enrichment of metabolic pathways, TCA cycle, glycolysis/gluconeogenesis, fatty Received: December 19, 2016 acid metabolism, and carbon metabolism, with similar pathways altered by diet and insulin Accepted: March 8, 2017 resistance. Thus, the rat liver ubiquitome is sensitive to diet and insulin stimulation and this is perturbed in insulin resistance. Published: March 22, 2017

Copyright: © 2017 Nagarajan et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which Introduction permits unrestricted use, distribution, and reproduction in any medium, provided the original The liver is exquisitely insulin-sensitive, and plays a critical role in glucose and lipid home- author and source are credited. ostasis as well as detoxification. Dysregulation of glucose and lipid metabolism in liver is a Data Availability Statement: All relevant data are major factor in the pathogenesis of metabolic diseases including type 2 diabetes and non-alco- within the paper and its Supporting Information holic fatty liver disease (NAFLD). A primary characteristic of these metabolic diseases is the files. accumulation of excess lipid in the liver, known as fatty liver or hepatic steatosis [1]. This sce- Funding: AJH is supported by a Helen and Robert nario is linked with impaired whole-body and hepatic insulin-stimulated glucose metabolism Ellis Postdoctoral Research Fellowship from the [2]. High-fat feeding studies with rodents demonstrates that impairment of insulin action Sydney Medical School Foundation and funding in the liver precedes the development of insulin resistance in other glucoregulatory tissues, from the University of Sydney. DNS is supported including skeletal muscle and adipose tissue [3, 4]. Fatty liver reduces tolerance to ischemic by the National Health and Medical Research Council (project grant GNT1052963), a Neale injury, impaired regeneration capacity [5] and increases the risk of developing hepatocellular Daniher MND Research Grant from the Motor carcinoma [6]. As such, fatty liver is an initiating factor in the development of a range of Neurone Disease Research Institute of Australia, pathologies.

PLOS ONE | https://doi.org/10.1371/journal.pone.0174431 March 22, 2017 1 / 14

36 Rat liver ubiquitome and insulin resistance

Guest Family Fellowship, and Mostyn Family Transcriptomic [7], lipidomic [3], and proteomic [8] studies have provided significant Foundation. insight into the mechanisms that link fatty liver and insulin resistance. Ubiquitin is a crucial Competing interests: The authors have declared post-translational modification that regulates cellular signaling in numerous processes such as that no competing interests exist. metabolism, transcription, translation, vesicle transport and apoptosis [9]. Dysfunction of the ubiquitin proteasome system (UPS) is observed in multiple diseases including cancer [10] but its role in metabolic disease is relatively poorly defined. Therefore, we aimed to identify the in vivo ubiquitin-modified proteome (ubiquitome) in rat liver and determine changes in this ubi- quitome under acute insulin stimulation and in high-fat diet-induced insulin resistance.

Materials and methods Animals All surgical and experimental procedures performed were approved by the Animal Ethics Committee (Ethics # 14/07, Garvan Institute/St. Vincent’s Hospital) and were in accordance with the National Health and Medical Research Council of Australia’s guidelines on animal experimentation. Sixteen adult male Wistar rats (Animal Resources Centre, Perth, Australia) were commu- nally housed at 22 ± 0.5˚C with a controlled 12:12h light-dark cycle. Half were fed ad libitum a standard rodent diet (Rat Maintenance Diet; Gordons Specialty Feeds, Sydney, Australia) con- taining (10% fat, 69% carbohydrate, and 21% (w/w) protein) while the other half was fed a high- fat, high-sucrose diet (HFSD; 45% energy from fat, 30% energy from sucrose) made in-house [11]. After 3 weeks of feeding, indwelling catheters were implanted as described elsewhere [12].

Insulin infusion All animals were fasted for 5 h and randomly assigned for acute insulin stimulation or tissue collected in the basal state, eight rats in each group. Insulin (Actrapid, Novo Nordisk, Copen- hagen, Denmark) was infused at a rate of 0.5 U/kg/h, and a 30% (wt/vol) variable glucose infusion was started at 4 min to maintain euglycemia, based upon [13]. In these rats, plasma samples were taken at 0, 1, 3, 5, 7.5 and 10 min for the measurement of insulin and glucose concentrations only. At 10 min, rats were euthanized with an overdose of sodium pentobarbi- tone (Troy Laboratories, Australia), liver rapidly dissected, freeze-clamped with aluminum tongs precooled in liquid nitrogen, and stored at −80˚C for subsequent analysis. Epidydimal fat pass mass was used as an index of adiposity.

Ubiquitome analysis Approximately 3–4 grams of liver was lysed in a 1:4 ratio (w:vol) of tissue and lysis buffer (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 0.5% NP-40, 1 mM DTT, 1X EDTA-free protease inhibi- tor cocktail, 10 mM N-ethylmaleimide, 1mM sodium orthovanadate) using a POLYTRON hand homogenizer. The protein concentration was determined using a BCA assay and 500 μg of total protein was used per sample to immunopurify mono- and poly-ubiquitinated proteins using specialized ubiquitin affinity matrix (VIVAbind Ubiquitin Kit, VIVA Bioscience). After substan- tial washing to remove residual detergent, beads were digested for 30 min at 27˚C, then reduced with 1mM DTT and left to digest overnight at room temperature with sequencing-grade trypsin (5 μg/mL, Promega), as described previously [14]. Samples were alkylated with 5mg/mL iodoa- cetamide and protease digestion terminated with trifluoroacetic acid. Trypsinized eluents were collected after brief centrifugation then purified and desalted using self-packed tips with 6 layers of C18 Empore disks (Pacific Laboratory Products), then dried in a SpeedVac. Samples were then resuspended in 12 μL 5% formic acid, 2% acetonitrile and stored at -80˚C.

PLOS ONE | https://doi.org/10.1371/journal.pone.0174431 March 22, 2017 2 / 14

37

Rat liver ubiquitome and insulin resistance

For MS analysis, 5 μL of each of the peptide samples were loaded and separated along a C18 column (400 mm, 75 μm ID, 3 μm silica beads) and introduced by nanoelectrospray into an LTQ Orbitrap Velos Pro coupled to an Easy-nLC HPLC (Thermo Fisher). Tandem mass spec- trometry data was collected for the top 10 most abundant ions per scan over a 140-minute time gradient. The order of data collection was randomized to interchange between biological conditions with BSA run between each sample to minimize temporal bias. MS/MS raw files were analyzed using MaxQuant (v1.2.7.4) [15] against the Uniprot Rat database using the Andromeda search engine integrated into MaxQuant [16]. A false discovery rate of 1% was tolerated for protein, peptide, and sites, and one missed cleavage was allowed. MaxQuant output data were filtered to remove contaminants, reverse hits, proteins only iden- tified by site and proteins with < 2 unique peptides, and all further analysis was performed using filtered data in our Pegasus statistical workflow [17] (S1 Fig). An individual protein was defined as present under a particular condition if it was detected in a minimum of two repli- cates. Analysis of individual replicates (S2 and S3 Figs) showed a good degree of reproducibil- ity across individual replicates within treatment groups. Protein-protein interactions and KEGG pathway analysis among the resulting protein list was analyzed using STRING (v10) [18] (with a confidence score of 0.700) and Cytoscape (v3.1.1) [19].

Analytical methods Blood and plasma glucose levels were determined by an immobilized glucose oxidase method (YSI 2300; Yellow Springs Instruments, Yellow Springs, OH, USA). Plasma insulin was mea- sured by ELISA (Ultra-Sensitive Mouse ELISA, Crystal Chem, USA). Liver triacylglycerols

(TAGs) were extracted using the method of Folch [20], lipids were dried under N2 gas, resus- pended in absolute ethanol and quantified using an enzymatic colorimetric method (GPO-- PAP reagent; Roche Diagnostics).

Immunoblot analysis Protein extraction and immunoblots from liver homogenates were performed as previously described [21]. Antibodies used were anti-phospho-Ser473 Akt and Atk (Cell Signaling Tech- nology, Danvers, MA) and anti-Ubiquitin (Abcam, UK). Densitometry was performed using IPLab Gel software (Signal Analytics Corporation, Vienna, VA, USA).

Statistical analysis Differences among relevant groups were assessed using unpaired Student’s t-test or Two-Way ANOVA using Tukey-Kramer post hoc tests as appropriate noted in figure legends. P < 0.05 was considered significant. Data are reported as mean ± seM.

Results and discussion HFSD leads to increased adiposity, increased liver triacylglycerol content and impaired insulin-stimulated Akt phosphorylation in liver The HFSD fed rat is a well-established model of obesity and insulin resistance [22, 23]. In this study, 4 weeks of high-fat feeding increased body mass (Fig 1A), absolute epidydimal fat pad mass (Fig 1B) and percent epidydimal fat pad mass (Chow: 3.2 ± 0.2%; HFD: 6.0 ± 0.3%; P<0.00001), and liver TAG content (Fig 1C) compared to chow-fed controls. Importantly, there were no differences in these outputs within diet groups for those animals that did or did not receive insulin stimulation (Fig 1A–1C).

PLOS ONE | https://doi.org/10.1371/journal.pone.0174431 March 22, 2017 3 / 14

38 Rat liver ubiquitome and insulin resistance

Fig 1. High-fat, high-sucrose feeding results in obesity and insulin resistance in male rats. (A) Body mass, (B) epidydimal fat pad mass, and (C) liver triacylglycerol content of rats fed a high fat diet for 4 weeks or chow control. * P < 0.05 vs Chow for each condition by Two-Way ANOVA followed by Tukey’s Multiple Comparisons test. (D) Plasma insulin and (E) blood glucose levels in rats infused with insulin for 10 mins at 0.5 U/kg/h. (F) Liver Akt phosphorylation. * P < 0.05 vs Basal for each condition, # P < 0.05 vs Chow Insulin by Two-Way ANOVA followed by Tukey’s Multiple Comparisons test. Data are mean ± SEM, n = 4. https://doi.org/10.1371/journal.pone.0174431.g001

Acute insulin infusion increased plasma insulin levels and at 10 mins post-infusion there was no difference between Chow and HFSD groups (Fig 1D). Further, there was no difference in blood glucose levels between groups at the end of the insulin infusion (Fig 1E). Insulin infu- sion increased liver Akt phosphorylation in both the chow and HFSD-fed groups but this response was blunted in the HFSD-diet group (P = 0.04; Fig 1F), indicating hepatic insulin resistance. There was no difference in total Akt levels between groups (Chow: 1.24 ± 0.23 a.u.; HFSD: 1.24 ± 0.09 a.u; P = 0.99). Calculating the ratio of phosphorylated Akt to total Akt pro- tein levels resulted in similar findings, albeit with slightly less statistical significance (P = 0.055) as a consequence of combining the error of both measurements in a small number of animals (data not shown). Collectively, these data demonstrate the effectiveness of the HFSD diet to increase adiposity leading to a decreased response in liver for the same rise in circulat- ing insulin.

Global ubiquitome of rat liver Efforts to understand regulation of key metabolic functions of the liver to date have focused mostly on post-translational modifications such as phosphorylation [24–26] and acetylation

PLOS ONE | https://doi.org/10.1371/journal.pone.0174431 March 22, 2017 4 / 14

39 Rat liver ubiquitome and insulin resistance

[27, 28]. However, recent advances in affinity purification and proteomics techniques have facilitated systematic assessment of the ubiquitome in vivo. To more closely examine changes in the ubiquitin-modified liver proteome in rats following acute insulin stimulation and/or 4 weeks HFSD feeding, we applied AP-MS/MS using ubiquitin-affinity matrix to enrich for ubi- quitinated proteins and ubiquitin binding proteins (Fig 2A). Immunoblot analysis of protein ubiquitination in representative samples of rat liver lysates showed a characteristic broad smear representing a complex mixture of ubiquitinated proteins in all conditions (Fig 2B). Label-free quantitative analysis of total peptide intensities showed no significant difference in the overall abundance of proteins isolated by ubiquitin affinity purification in each condition (Fig 2B). We identified a total of 1279 proteins following Ub-affinity purification from liver lysates across all four experimental conditions (Chow ± insulin, HFSD ± insulin). Of these, 831 proteins have been previously identified in the rat ubiquitome (total 7024; [29]). Hence, we have identified 448 new ubiquitinated proteins in rat liver (Fig 2C). Detailed lists are pro- vided in S1 Table and a protein-protein interaction map based upon STRING analysis is pro- vided in S2 Fig. Fig 2D shows the top 30 biological pathways represented by these proteins using ontology analysis based upon KEGG pathways. Many represent known pathways central to liver biology such as glucose, and fatty acid metabolic pathways, steroid hor- mone biosynthesis and drug metabolism (Fig 2D). Strikingly, we observe ubiquitination of all but 1 enzyme (glucose-6-phosphatase) involved in glycolysis/gluconeogenesis in rat liver (Fig 2E). This widespread post-translational modification suggests that ubiquitination is a key regu- lator of gluconeogenesis/glycolysis, acting in concert with the established transcriptional regu- lation of key genes (including glucose-6-phosphatate and PEPCK) [30]. Crosstalk between acetylation and ubiquitination is known to regulate the key gluconeogenesis enzyme PEPCK [31]. As our proteomics approach cannot discriminate between different ubiquitin chain topologies, validation studies beyond the scope of this work will be necessary to determine downstream functional consequences of ubiquitination (i.e. degradation via K48 chains) on regulation of signaling and metabolic pathways.

The liver ubiquitome of rats is sensitive to insulin stimulation The liver is a key insulin-sensitive tissue that contributes to the regulation of whole-body glu- cose homeostasis and is highly susceptible to the deleterious influences of high-fat diet [3, 4]. Of the 1279 proteins identified in this study, 788 ubiquitinated proteins were detected in chow fed animals, with 868 proteins ubiquitinated following acute insulin stimulation (Fig 3A). In HFSD rats, we identified 864 ubiquitinated proteins and 817 proteins following insulin stimu- lation in this group (S1 Table). Finally, we identified 686 ubiquitinated proteins in common across the 4 conditions. To assess the potential role of the UPS in insulin action in the liver, we evaluated the specific influences of insulin on the liver ubiquitome in the chow fed rat. 745 proteins were detected in common between chow and chow + insulin conditions (Fig 3B). Of the 868 Ub-associated pro- teins identified in liver following i.v. insulin stimulation, 123 (14%) were unique to this condi- tion (S2 Table), compared to the 43 (5% of 788 total) Ub-associated proteins unique to the basal state. STRING analysis of protein-protein interaction (PPI) networks within the set of differentially ubiquitinated proteins (n = 166) following insulin stimulation identified enrich- ment for components of OxPhos, migration signaling, focal adhesion, UPS and Ras Signaling as putative targets of ubiquitination following insulin stimulation (Fig 3C). Further, KEGG pathway analysis of these proteins identified enrichment of components of NAFLD, endo- cytosis, PI3K-Akt signaling and fatty acid metabolism (Fig 3D). Ubiquitination of proteins involved in fatty acid degradation and metabolism following insulin stimulation suggests a

PLOS ONE | https://doi.org/10.1371/journal.pone.0174431 March 22, 2017 5 / 14

40 Rat liver ubiquitome and insulin resistance

Fig 2. Rat liver ubiquitome. (A) Experimental design of the ubiquitomic analysis rat liver. (B) Immunoblot analysis of protein ubiquitination in representative samples of rat liver lysates. (C) Venn diagram of previously identified protein in this study and that in the known rat ubiquitome. (D) Ontology analysis of identified ubiquitinylated proteins. (E) Gluconeogenesis/Glycolysis biochemical pathway with ubiquitinylated enzymes marked in red. https://doi.org/10.1371/journal.pone.0174431.g002

PLOS ONE | https://doi.org/10.1371/journal.pone.0174431 March 22, 2017 6 / 14

41 Rat liver ubiquitome and insulin resistance

Fig 3. Insulin stimulation alters the ubiquitome of the rat liver. (A) Venn diagram of identified proteins in each group. (B) Venn diagram of proteins ubiquitinated in Chow compared to Chow + Insulin. (C) STRING analysis of differentially ubiquitinated proteins. (D) Ontology analysis of identified ubiquitinylated proteins. https://doi.org/10.1371/journal.pone.0174431.g003

PLOS ONE | https://doi.org/10.1371/journal.pone.0174431 March 22, 2017 7 / 14

42 Rat liver ubiquitome and insulin resistance

Fig 4. High-fat, high-sucrose diet alters the ubiquitome of the rat liver. (A) Venn diagram of identified proteins ubiquitinated in HFSD compared to HFSD + Insulin. (B) Ontology analysis of identified ubiquitinylated

PLOS ONE | https://doi.org/10.1371/journal.pone.0174431 March 22, 2017 8 / 14

43 Rat liver ubiquitome and insulin resistance

proteins in HFSD compared to HFSD + Insulin. (C) STRING analysis of differentially ubiquitinated proteins. (D). Ontology analysis of identified ubiquitinated proteins responsive to insulin in Chow but not in HFSD. https://doi.org/10.1371/journal.pone.0174431.g004

novel mechanism contributing to insulin inhibition of fatty acid catabolism (i.e. beta oxida- tion) and stimulation of fatty acid synthesis [32]. Targeted analysis of protein ubiquitination has demonstrated that insulin stimulation leads to ubiquitination of a range of key insulin signaling intermediates which acts as a negative feedback mechanism. For example, acute insulin stimulation of FAO hepatocyte cells in- creased IRS-2 ubiquitination and lowered IRS-2 protein levels [33]. Conversely, insulin stimu- lation of primary rat hepatocytes reportedly reduced protein ubiquitination [34]. Insulin stimulation of mouse hepatocytes expressing HA-tagged ubiquitination increased APPL1 ubi- quitination in a TRAF6-mediated mechanism [35]. Our systematic approach to identify the insulin-sensitive rat liver ubiquitome did not detect IRS-1, IRS-2 or APPL1, possibly reflecting sensitivity limitations or differences between in vitro and in vivo models. Collectively, we have identified a broad range of liver proteins with altered ubiquitination in response to acute in vivo insulin stimulation. These will provide an important basis for ongoing targeted studies to better understand regulatory mechanisms underpinning insulin action in the liver.

The rat liver ubiquitome is sensitive to high-fat diet The liver is highly susceptible to the deleterious influences of high-fat diet, including dysregu- lated glucose and lipid homeostasis [3, 4]. Altered signal transduction, including via differen- tial protein phosphorylation, are proposed to underpin these systemic phenotypes [36]. However, other post-translational modifications such as ubiquitin are also likely to be impor- tant regulators of signaling and metabolic pathways as a consequence of HFSD feeding. We observed 744 ubiquitinated proteins in common between chow and HFSD animals (Fig 4A). Of the 864 ubiquitinated proteins identified in HFSD, 120 (14%) proteins were unique to the HFSD group. This compared to 44 (6%) of the 788 ubiquitinated proteins unique to chow-fed animals (Fig 4A). Analysis of KEGG pathways and PPI networks within this set of differentially ubiquitinated proteins identified enrichment of proteins involved in RNA metabolism, actin cytoskeleton, and amino acid and fatty acid metabolism (Fig 4B and 4C). These diet-induced changes in the liver ubiquitome differ from observations made follow- ing exposure of HepG2 hepatocellular carcinoma cells to single fatty acids in vitro. For exam- ple, ubiquitin levels in HepG2 cells (detected by western blot) were not altered by exposure to the polyunsaturated fatty acid linoleate with or without the addition of exogenous FABP4 [37], which is increased in the plasma of obese patients [38]. Conversely, exposing HepG2 cells to the saturated fatty acid palmitate did lead to ubiquitination of IR, IRS-1 and Akt [39]. Palmi- tate does have significant effects on cell biology, but interestingly many of these effects are lost with the addition of other unsaturated fatty acids such as linoleate or oleate [40]. Further, the use of single fatty acids to increase extracellular lipid levels does not model the environment of HFSD [41]. We have performed similar analyses of the binary comparisons between HFSD vs HFSD + Insulin (S5 Fig), and Chow + Insulin vs HFSD + Insulin (S6 Fig). Briefly, 56 proteins were uniquely ubiquitinated in the HFSD + Insulin group compared to the HFSD group, whereas 103 were identified exclusively in the HFSD group compared to HFSD + Insulin (S5A Fig). Additionally, only 46 proteins were uniquely identified in the HFSD + Insulin group com- pared to the Chow + Insulin groups, whilst 97 proteins were only ubiquitinated in the Chow + Insulin group (S6A Fig). These proteins are involved in a diverse range of metabolic path- ways (S5B and S6B Fig).

PLOS ONE | https://doi.org/10.1371/journal.pone.0174431 March 22, 2017 9 / 14

44 Rat liver ubiquitome and insulin resistance

Insulin resistance is defined as a blunted response by cells to insulin stimulation. E3 ubiqui- tin ligases have been implicated in the pathogenesis of insulin resistance by modulating insulin signaling (see review [42]). Much of the data underlying this model comes from targeted anal- ysis of ubiquitination of individual insulin signaling intermediates [43, 44], rather than the sys- tematic analysis of global changes in ubiquitome in pre-clinical models of insulin resistance presented here. Our dataset provided a unique opportunity to characterize changes in the liver ubiquitome associated with insulin resistance. We identified 310 ubiquitin modified proteins that were responsive to insulin stimulation in chow or HFSD animals (S3 Table). 99 (40%) of these 310 proteins responded to insulin stimulation in the chow group but failed to respond in the HFSD group. Notably, these proteins are involved in oxidative phosphorylation and fatty acid metabolism (Fig 4D), highlighting these pathways as key targets of ubiquitin in insulin resistance. For example, one of these fatty acid metabolism proteins was ACSL4 which is increased in patients with NAFLD compared to controls [45]. Further, we identified altered ubiquitination of subunits of complex 2 and 5 of the electron transport chain, which is consis- tent with a role for altered oxidative phosphorylation in the development of liver insulin resis- tance [46]. Conversely, 89 proteins (36%) responded to insulin stimulation in the HFSD group but not in the chow group, with 61 proteins (24%) responding to insulin in both groups. Together with our identification of ubiquitination of all but one of the enzymes in glycolytic/ gluconeogenic pathways, these novel data suggest that ubiquitination is a key regulator of the pathogenesis of liver insulin resistance and may play a role in fatty liver associated diseases.

Conclusion Herein, we have described the systematic characterization of the rat liver ubiquitome and report the effects of acute in vivo insulin stimulation and high-fat, high-sucrose diet. Specifi- cally, we observed ubiquitination of proteins involved in key metabolic pathways, in particular gluconeogenesis/glycolysis, oxidative phosphorylation and fatty acid metabolism. Hence, ubi- quitination is likely a novel mechanism acting at multiple levels to regulate whole-body eugly- cemia and lipidemia. Further, widespread changes in the ubiquitin modified proteome may mediate the pathogenesis of fatty acid-associated diseases.

Supporting information S1 Fig. Statistical workflow. Flow diagram showing data analysis method from mass spec- trometry raw data. (TIF) S2 Fig. Proteomics data variability analysis. (A) Histograms showing peptide abundance dis- tributions in individual replicates for each condition. (B) Multiple regression analysis of pep- tide abundance in individual replicates. (TIF) S3 Fig. Venn analysis of proteins identified in each replicate. (TIF) S4 Fig. Protein-protein interactome of the total liver ubiquitome. (TIF) S5 Fig. Insulin stimulation alters the hfsd ubiquitome of the rat liver. (A) Venn diagram of proteins ubiquitinated in High-Fat Sucrose Diet (HFSD) compared to HFSD + Insulin. (B) Ontology analysis of identified ubiquitinylated proteins. (C) STRING analysis of differentially

PLOS ONE | https://doi.org/10.1371/journal.pone.0174431 March 22, 2017 10 / 14

45 Rat liver ubiquitome and insulin resistance

ubiquitinated proteins. (TIF) S6 Fig. Effects of diet on insulin stimulated ubiquitome of the rat liver. (A) Venn diagram of proteins ubiquitinated in Chow + Insulin compared to High-Fat Sucrose Diet (HFSD) + Insulin. (B) Ontology analysis of identified ubiquitinylated proteins. (C) STRING analysis of differentially ubiquitinated proteins. (TIF) S1 Table. Proteins identified in the rat liver ubiquitome for each condition. (XLSX) S2 Table. Ubiquitinated proteins unique to each condition. (XLSX) S3 Table. Proteins ubiquitinated in response to insulin. (XLSX)

Acknowledgments The authors wish to thank staff of the Mass Spectrometry Core Facility at the University of Sydney for their expertise and advice.

Author Contributions Conceptualization: DNS AJH. Data curation: SRN AEB JAM DNS AJH. Formal analysis: SRN AEB JAM HCS TQN ES DNS AJH. Funding acquisition: PP DNS AJH. Investigation: SRN AEB JAM HCS TQN ES. Methodology: SRN AEB JAM PP DNS AJH. Project administration: DNS AJH. Resources: PP GJC DNS AJH. Software: JAM DNS. Supervision: GJC DNS AJH. Validation: DNS AJH. Visualization: DNS AJH. Writing – original draft: DNS AJH. Writing – review & editing: SRN AEB JAM GJC DNS AJH.

References 1. Lonardo A, Ballestri S, Marchesini G, Angulo P, Loria P. Nonalcoholic fatty liver disease: a precursor of the metabolic syndrome. Dig Liver Dis. 2015; 47(3):181–90. https://doi.org/10.1016/j.dld.2014.09.020 PMID: 25739820 2. Byrne CD, Targher G. NAFLD: a multisystem disease. J Hepatol. 2015; 62(1 Suppl):S47–64.

PLOS ONE | https://doi.org/10.1371/journal.pone.0174431 March 22, 2017 11 / 14

46 Rat liver ubiquitome and insulin resistance

3. Turner N, Kowalski GM, Leslie SJ, Risis S, Yang C, Lee-Young RS, et al. Distinct patterns of tissue-spe- cific lipid accumulation during the induction of insulin resistance in mice by high-fat feeding. Diabetolo- gia. 2013; 56(7):1638–48. https://doi.org/10.1007/s00125-013-2913-1 PMID: 23620060 4. Kraegen EW, Clark PW, Jenkins AB, Daley EA, Chisholm DJ, Storlien LH. Development of muscle insu- lin resistance after liver insulin resistance in high-fat-fed rats. Diabetes. 1991; 40(11):1397–403. Epub 1991/11/01. PMID: 1936601 5. Tashiro H, Kuroda S, Mikuriya Y, Ohdan H. Ischemia-reperfusion injury in patients with fatty liver and the clinical impact of steatotic liver on hepatic surgery. Surgery today. 2014; 44(9):1611–25. https://doi. org/10.1007/s00595-013-0736-9 PMID: 24078000 6. Charrez B, Qiao L, Hebbard L. Hepatocellular carcinoma and non-alcoholic steatohepatitis: The state of play. World journal of gastroenterology WJG. 2016; 22(8):2494–502. PubMed Central PMCID: PMCPMC4768195. https://doi.org/10.3748/wjg.v22.i8.2494 PMID: 26937137 7. Toye AA, Dumas ME, Blancher C, Rothwell AR, Fearnside JF, Wilder SP, et al. Subtle metabolic and liver gene transcriptional changes underlie diet-induced fatty liver susceptibility in insulin-resistant mice. Diabetologia. 2007; 50(9):1867–79. https://doi.org/10.1007/s00125-007-0738-5 PMID: 17618414 8. Schmid GM, Converset V, Walter N, Sennitt MV, Leung KY, Byers H, et al. Effect of high-fat diet on the expression of proteins in muscle, adipose tissues, and liver of C57BL/6 mice. Proteomics. 2004; 4 (8):2270–82. https://doi.org/10.1002/pmic.200300810 PMID: 15274121 9. Iconomou M, Saunders DN. Systematic approaches to identify E3 ligase substrates. Biochem J. 2016; 473(22):4083–101. https://doi.org/10.1042/BCJ20160719 PMID: 27834739 10. Popovic D, Vucic D, Dikic I. Ubiquitination in disease pathogenesis and treatment. Nat Med. 2014; 20 (11):1242–53. https://doi.org/10.1038/nm.3739 PMID: 25375928 11. Wright LE, Brandon AE, Hoy AJ, Forsberg GB, Lelliott CJ, Reznick J, et al. Amelioration of lipid-induced insulin resistance in rat skeletal muscle by overexpression of Pgc-1beta involves reductions in long- chain acyl-CoA levels and oxidative stress. Diabetologia. 2011; 54(6):1417–26. https://doi.org/10.1007/ s00125-011-2068-x PMID: 21331471 12. Bruce CR, Hoy AJ, Turner N, Watt MJ, Allen TL, Carpenter K, et al. Overexpression of carnitine palmi- toyltransferase-1 in skeletal muscle is sufficient to enhance fatty acid oxidation and improve high-fat diet-induced insulin resistance. Diabetes. 2009; 58(3):550–8. Epub 2008/12/17. PubMed Central PMCID: PMCPMC2646053. https://doi.org/10.2337/db08-1078 PMID: 19073774 13. Frangioudakis G, Ye JM, Cooney GJ. Both saturated and n-6 polyunsaturated fat diets reduce phos- phorylation of insulin receptor substrate-1 and protein kinase B in muscle during the initial stages of in vivo insulin stimulation. Endocrinology. 2005; 146(12):5596–603. https://doi.org/10.1210/en.2005-0481 PMID: 16150913 14. Turriziani B, Garcia-Munoz A, Pilkington R, Raso C, Kolch W, von Kriegsheim A. On-beads digestion in conjunction with data-dependent mass spectrometry: a shortcut to quantitative and dynamic interaction proteomics. Biology (Basel). 2014; 3(2):320–32. PubMed Central PMCID: PMCPMC4085610. 15. Cox J, Mann M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat Biotechnol. 2008; 26(12):1367–72. https:// doi.org/10.1038/nbt.1511 PMID: 19029910 16. Cox J, Neuhauser N, Michalski A, Scheltema RA, Olsen JV, Mann M. Andromeda: a peptide search engine integrated into the MaxQuant environment. J Proteome Res. 2011; 10(4):1794–805. https://doi. org/10.1021/pr101065j PMID: 21254760 17. Croucher DR, Iconomou M, Hastings JF, Kennedy SP, Han JZ, Shearer RF, et al. Bimolecular comple- mentation affinity purification (BiCAP) reveals dimer-specific protein interactions for ERBB2 dimers. Sci Signal. 2016; 9(436):ra69. https://doi.org/10.1126/scisignal.aaf0793 PMID: 27405979 18. Szklarczyk D, Franceschini A, Wyder S, Forslund K, Heller D, Huerta-Cepas J, et al. STRING v10: pro- tein-protein interaction networks, integrated over the tree of life. Nucleic Acids Res. 2015; 43(Database issue):D447–52. PubMed Central PMCID: PMCPMC4383874. https://doi.org/10.1093/nar/gku1003 PMID: 25352553 19. Smoot ME, Ono K, Ruscheinski J, Wang PL, Ideker T. Cytoscape 2.8: new features for data integration and network visualization. Bioinformatics. 2011; 27(3):431–2. PubMed Central PMCID: PMCPMC3031041. https://doi.org/10.1093/bioinformatics/btq675 PMID: 21149340 20. Folch J, Lees M, Sloane Stanley GH. A simple method for the isolation and purification of total lipides from animal tissues. J Biol Chem. 1957; 226(1):497–509. PMID: 13428781 21. Hoy AJ, Bruce CR, Cederberg A, Turner N, James DE, Cooney GJ, et al. Glucose infusion causes insu- lin resistance in skeletal muscle of rats without changes in Akt and AS160 phosphorylation. Am J Phy- siol Endocrinol Metab. 2007; 293(5):E1358–64. https://doi.org/10.1152/ajpendo.00133.2007 PMID: 17785505

PLOS ONE | https://doi.org/10.1371/journal.pone.0174431 March 22, 2017 12 / 14

47

Chapter 3

Role of Lipid Droplet Proteome and Phosphoproteome in Obesity-Induced Insulin Resistance in the Liver

48 3.1. Introduction

Obesity is characterized by chronic excess intracellular accumulation of lipids that is often associated with a higher risk of metabolic disease, including cardiovascular disease and IR. Subcutaneous adipose tissue has an abundant, yet limited, capacity to store exogenous lipids, causing the redistribution of FAs to tissues when primary lipid stores are capped (Engin 2017). Excessive deposition of TAG in insulin-sensitive organs such as skeletal muscle and liver is associated with impaired insulin signalling and often leads to IR and type 2 diabetes (Boren et al. 2013). This relationship is particularly strong in the liver, where temporal differences in tissue-specific IR point towards the liver as the site of early onset (Turner et al. 2013). This association is further supported by the high prevalence of type 2 diabetes in patients with NAFL(D) and NASH, with 23% and 44% worldwide, respectively (Younossi et al. 2016).

While it is evident that increased lipid availability attenuates the metabolic competence of the liver, studies looking into the role of cytoplasmic lipid droplets (LDs) in the pathogenesis of hepatic IR have yielded relatively unclear results. These highly coordinated organelles store the majority of cellular neutral lipids and host a myriad of lipolytic regulators such as adipose triglyceride lipase (ATGL), comparative gene identification-58 (CGI58), G0/G1 switch 2 (G0S2), hormone sensitive lipase (HSL), and perilipins (PLIN) (Krahmer et al. 2013). Genetic or chemical inhibition of ATGL, the primary protein involved in TAG hydrolysis, increased steatosis despite improving glucose clearance in some mice studies (Hoy et al. 2011, Schweiger et al. 2017), while decreasing hepatic insulin signalling in genetic models of ATGL deficiency (Kienesberger et al. 2009). Moreover, increased expression of the established ATGL inhibitor, G0S2, exacerbates diet-induced IR in some studies (Sugaya and Satoh 2017), yet also restored insulin sensitivity in others (Wang et al. 2013, Zhang et al. 2014). In clinical settings however, obese and NAFLD patients with IR have reduced mRNA and protein levels of ATGL, its co- activator CGI58, and the TAG/DAG/cholesteryl ester lipase, HSL (Jocken et al. 2007, Kato et al. 2008). The controversy in the field largely represents the complexity of LD dynamics and neutral lipid homeostasis, such as protein-protein interactions, protein translocation and hormonal stimulation.

Mass spectrometry-based proteomics has developed into a premier tool for investigating pathophysiology at a systems biology level, allowing for the simultaneous measurement of subcellular localization and post-translational modifications of proteins (Cox and Mann 2011).

49 Several studies have assessed the hepatic LD proteome of patients with NAFL(D) (Su et al. 2014, Khan et al. 2015, Liu et al. 2017); however, changes in the movement and phosphorylation of the LD proteome in response to insulin in normal and insulin-resistant states have not been explored thus far. Defining the insulin-stimulated proteome and phosphoproteome of rat liver LDs and how it is perturbed in IR animals could assist in the delineation of LD dynamics in a stepwise manner.

The aim of the experiments and analysis described in this chapter were to determine 1) insulin and diet-induced changes in the LD proteome and phosphoproteome and 2) compare them to an insulin resistant model using various biologically-relevant analytical measures.

50 3.2 Methods Samples for experiments described in this chapter were derived from animal models described in detail in Chapter 2. Methodology pertaining to mass spectrometry was performed by collaborators in the David James Metabolic Cybernetics Lab at the Charles Perkins Centre, The University of Sydney. Specifically, Dr. Zhiduan Su and Dr. Benjamin Parker performed the mass spectrometry (section 4.2.2.3-4.2.2.9) and Dr. Rima Chaudhuri performed statistical normalisation of the proteomic and phosphoproteomic datasets (section 4.2.2.10.1).

X 8 X 8

Chow-fed HFD-fed Basal Insulin infusion Basal Insulin infusion

X 4 X 4 X 4 X 4

Lipid Lipid droplets droplets

Protein digestion and TMT10-labelling

TiO2 enriched Proteome phosphoproteome

128N/C y t

i 126 129N/C

s 131

n m/z 130N/C e

t 127N/C n y I t i s n e t n I

m/z MS analysis Figure 3.1 Animal workflow for LD proteomic and phosphoproteomic enrichment Sixteen male adult Wistar rats were fed either a standard rodent chow or HF diet for 3 weeks. Following a 5 hour fast, half of the animals from each group were randomly assigned for acute insulin stimulation for 10 minutes. Once euthanized, liver tissue was extracted and LDs were isolated using a sucrose density gradient. LDs were then delipidated and prepared for LC—

MS/MS analysis using TMT10 labelling and TiO2 to enrich the phosphoproteome. Analysis of raw data was performed using MaxQuant.

51

3.2.1 Isolation of lipid droplets from rat livers 3.2.1.1 Homogenization Fresh livers were placed in specimen jars on ice filled with cold hypotonic lysis medium (HLM; 20 mM Tris-Cl pH 7.4, 1 mM EDTA, 10 mM NaF supplemented with protease and phosphatase inhibitors (Astral Scientific)) at a ratio of 4 ml medium/ gram of tissue. Livers were finely cut with scissors before being transferred to a Potter-Elvehjem tissue homogenizer for hand-driven homogenization by eight gentle strokes of an ice-cold pestle. Homogenates were then transferred to a 50 ml tube and centrifuged for 10 mins at 1000 × g at 4°C (Beckman Coulter). The resulting supernatant was then transferred to microfuge tubes for storage at -80°C as whole liver lysates or to fresh ultracentrifuge tubes for further processing.

3.2.1.2 Sucrose Gradient Fractionation A sucrose gradient was prepared by thoroughly mixing 3 ml of liver homogenate with 1 ml of HLM containing 60% sucrose in a 13.2 ml ultracentrifuge tube (Beckman Coulter). Another 5ml and 4 ml of HLM containing 5% and 0% sucrose, respectively, were layered on top. Tubes were carefully weighed and matched within 0.02 g of each other for balancing. Samples were centrifuged for 30 mins at 28000 × g at 4°C (P55ST2-636 spinout rotor, Hitachi). The rotor was allowed to coast to a stop with a deceleration of “F” (free). Once at rest, a pipet with a wide opening was used to carefully transfer 1.5 ml of the floating fat layer, containing lipid droplets, into a 15 ml tube (Corning).

3.2.1.3 Sample Delipidation Tubes containing isolated lipid droplet fractions were filled with 10 volumes of ice-cold acetone and mixed by inversion several times to remove excess lipids from samples. Samples were incubated overnight at -20°C and centrifuged the next day for 1 hour at 4300 × g at 4°C to pellet the LD-associated proteins. A nitrogen sample concentrator was used to evaporate the residual acetone and dehydrate the pellet. Samples were then stored at -80°C until use.

3.2.2 Proteomics 3.2.2.1 Protein Preparation Delipidated protein samples were resuspended in 400 µl of urea buffer (6 M urea, 2 M thiourea, 25mM triethylammonium bicarbonate (TEAB), and phosphatase inhibitors) and vortexed thoroughly. Samples were then sonicated in an ultrasonic water bath (Sonorex Digitec,

52 Bandelin) set at room temperature for 5 mins and then sonicated further using an ultrasonic homogenizer with the following parameters: 55 second working time, 10 second pulse with 5 second breaks. After spinning proteins at 15,000 x g for 20 mins at room temperature, samples were transferred into a new QubitÔ assay tube (Thermo Fisher Scientific) and quantified using the Qubit 30 Fluorometer (Life Technologies).

3.2.2.2 Protein Digestion For digestion, 800 µg of protein was added to 10 mM dithiothreitol (DTT) and incubated at 37°C for 1 hour. After 1 hour, 25 mM indole acetic acid was added to the samples and incubated at room temperature for 40 mins in the dark and stopped with 10 mM DTT. Finally, LysineC was added to the samples at a LysineC:protein ratio of 1:50 and incubated at 30°C for 2 hours. Samples were then diluted five times in TEAB and trypsinized at 37°C overnight. The next day, 1% formic acid was added to the digested samples to decrease the pH and quench the digestion.

3.2.2.3 Cartridge Desalting To separate salts from proteins, samples were desalted using the Sep-Pak Vac 1cc tC18 cartridge kit (Waters). To activate C18, 1 ml of methanol was added to each column and washed once with 1 ml of elution buffer containing 80% acetonitrile (ACN) and 0.1% trifluoroacetic acid (TFA). The columns were then washed three times with 1 ml of 0.1% TFA loading buffer. Peptides were resuspended in 1 ml of the loading buffer and loaded onto the column twice. After washing the columns three times with 1 ml loading buffer, peptides were eluted using 500 µl of elution buffer and collected. Samples were then dried using a vacuum concentrator (Thermo Fisher Scientific). Peptides were resuspended in 30 µl of 100 mM TEAB and quantified using Qubit to ensure post-digestion efficiency before labelling.

3.2.2.4 Tandem Mass Tag Labelling Tandem Mass Tag (TMT) -10plex mass tags (Thermo Fisher Scientific) were used for the quantification and identification of proteins for mass spectrometry. TMT label reagent was added to 200 µg of peptides from each sample and incubated at room temperature. After 1 hour, 8 µl of 5% hydroxylamine was added to the samples and incubated for 15 mins to stop the reaction. Samples were combined in a microfuge tube containing a total of 2 mg peptides and stored at -80°C. Two assays were performed to accommodate for all conditions.

53

3.2.2.5 Stage-Tip Desalting Peptide samples were further cleaned by stage-tip desalting prior to LC-MS/MS. Methods are essentially the same as those in section 4.2.2.3; however, volumes used were 10 times less.

3.2.2.6 Phosphopeptide Enrichment Phosphopeptide enrichment consisted of titanium dioxide followed by sequential elution from immobilized metal ion affinity chromatography and fractionation by hydrophilic interaction liquid chromatography (Engholm-Keller et al. 2012). Peptides were resuspended in 1 ml of titanium dioxide loading buffer (1 M glycolic acid, 80% acetonitrile, 5% TFA), and a 20 µg aliquot was saved for total proteomic analysis. Titanium dioxide beads (15 mg in 150 µl acetonitrile; GL Science, Japan) were added to the peptide mixture and rotated at room temperature for 20 min. The mixture was centrifuged at 10000 x g for 1 min and the resulting supernatant was applied to a second aliquot of titanium dioxide (7.5 mg in 75 µl acetonitrile) and rotated at room temperature for 20 min. The mixture was centrifuged at 10000 x g for 1 min and the supernatant was applied to a third aliquot of titanium dioxide (4 mg in 40 µl acetonitrile) and rotated at room temperature for 20 min. The beads were washed with 100 µL titanium dioxide loading buffer followed by 80% acetonitrile, 2% TFA and finally 16% acetonitrile, 0.4% TFA. The beads were dried briefly by vacuum centrifugation and eluted with 50 µL of 1% ammonium hydroxide by shaking at room temperature for 15 min. The titanium dioxide elution slurry was loaded onto a C8-plugged micro-column and eluted with gentle air pressure to trap beads. The beads were eluted with an additional 50 µL 1% ammonium hydroxide and the elution was pooled. Enriched phosphopeptides were acidified to a final concentration of 10% formic acid and dried by vacuum centrifugation. The enriched phosphopeptides were resuspended in 500 µl of 50% acetonitrile, 0.2% TFA and rotated with 50 µl of Fe(III)-IMAC beads (Sigma-Aldrich) at room temperature for 45 min. The IMAC slurry was loaded onto a crushed GeLoader microcolumn and eluted with gentle air pressure to trap beads. Mono-phosphorylated peptides were eluted with 50% acetonitrile, 0.1% TFA followed by 20% acetonitrile, 1% TFA; pooled with the flow-through and dried by vacuum centrifugation. Multi-phosphorylated peptides were eluted with 1% ammonium hydroxide, acidified to a final concentration of 10% formic acid, 0.1% TFA and desalted with C18 microcolumns. Enriched mono-phosphorylated peptides were subjected to another round of titanium dioxide as described above. The enriched mono-phosphorylated peptides and non-

54 phosphorylated peptides were fractionated on in-house packed TSK-amide HILIC column as described previously (Palmisano et al. 2010).

3.2.2.7 Liquid Chromatography-Tandem Mass Spectrometry 3.2.2.7.1 Total Proteome Mass Spectrometry (MS)-based total proteomic analysis was performed on an Easy nLC-1000 UHPLC coupled to a Q-Exactive mass spectrometer in positive polarity mode. Peptides were separated using an in-house packed 75 µm x 40 cm column (1.9 µm particle size, C18AQ) with a gradient of 5-30% ACN containing 0.1% FA over 90 min at 250 nL/min at 55ºC. The MS1 scan was acquired from 300-1750 m/z (70000 resolution, 3e6 AGC, 100 ms injection time) followed by MS/MS data-dependent acquisition of the top 20 ions with HCD (35,000 resolution, 1e5 AGC, 120 ms injection time, 25 NCE, 2.0 m/z isolation width).

3.2.2.7.2 Total Phosphoproteome MS-based phosphoproteomic analysis was performed on a Dionex chromatography coupled to a Q-Exactive plus mass spectrometer in positive polarity mode. Peptides were separated using an in-house packed 75 µm x 40 cm column (1.9 µm particle size, C18AQ) with a gradient of 10-40% ACN containing 0.1% FA over 120 min at 300 nL/min at 55 ºC. The MS1 scan was acquired from 300-1750 m/z (70000 resolution, 3e6 AGC, 100 ms injection time) followed by MS/MS data-dependent acquisition of the top 20 ions with HCD (17500 resolution, 5e5 AGC, 120 ms injection time, 32 NCE, 1.2 m/z isolation width).

3.2.2.8 Protein Identification and Quantification 3.2.2.8.1 Total Proteome Raw data was processed using MaxQuant (v1.5.1.3) against a Uniprot rat database (12/2014, 35,579 entries) with default settings and minor changes: Oxidation of methionine and acetylation of protein N-terminus were set as variable modifications and carbamidomethylation of cysteine was set as a fixed modification. TMT10-plex of isobaric labels was enabled and reporter mass tolerance was set to 0.001 Da. The “second peptides searching” was enabled. The peptide match tolerance was set to 20 ppm and 4.5 pm for first and main searches respectively and MS/MS match tolerance set to 20 ppm. Both peptide spectral match (PSM) and protein FDR were set to 1%.

55 3.2.2.8.2 Total Phosphoproteome Raw data was processed using MaxQuant (v1.5.1.3) (Cox and Mann 2008) against a Uniprot rat database (12/2014, 35,579 entries) with default settings and minor changes: Oxidation of methionine, acetylation of protein N-terminus and phosphorylation of serine, and tyrosine were set as variable modifications and carbamidomethylation of cysteine was set as a fixed modification. TMT10-plex of isobaric labels was enabled and reporter mass tolerance was set to 0.001 Da. The “second peptides searching” was enabled. The peptide match tolerance was set to 20 ppm and 4.5 pm for first and main searches respectively and MS/MS match tolerance set to 20 ppm. Both peptide spectral match (PSM) and protein FDR were set to 1%.

3.2.2.9 Bioinformatic Analysis 3.2.2.9.1 Normalisation Data was normalised using the two-step Remove Unwanted Variation (RUV-2) method, as previously described (Gagnon-Bartsch and Speed 2012). Briefly, RUV utilizes the curation of positive and negative control proteins based on hepatic LD association to adjust for systematic errors and normalise data.

3.2.2.9.2 Determination of Novel Rattus norvegicus Phosphosites The proportion of novel phosphosites found in Rattus norvegicus in this study was determined by overlaying the 2286 phosphosite dataset with the known Rattus norvegicus phosphosite database (562 entries for rat, within a database of 46,248 total entries across all species) curated by PhosphositePlusÒ (PhosphoSite, Cell Signalling Technology). A Venn diagram was created using the Core Graphic Module on GeneVenn (http://genevenn.sourceforge.net/) to determine the rat phosphosite lists for known, overlapping and novel phosphosites.

3.2.2.9.3 Enriched Pathway Analysis For total proteomic and phosphoproteomic pathway analysis, Uniprot IDs of all quantified proteins were submitted to Search Tool for the Retrieval of Interacting Genes/Proteins ((STRING) version 10.5) for Kyoto Encyclopaedia of Genes and Genomes (KEGG) pathway enrichment analysis. For diet and insulin-specific pathway analysis, significant differentially expressed (DE) proteins and phosphoproteins were submitted for Core Canonical Pathway Analysis using Ingenuity Pathway Analysis (Qiagen).

56

3.2.2.9.4 Scenario Modelling Differentially expressed protein and phosphosite data was used to systematically identify “insulin-resistant” proteins and sites through biologically relevant scenario filters. For example, to determine proteins that normally moved to the lipid droplet with insulin- stimulation but did not move to the lipid droplet due to perturbed insulin signaling in a high- fat diet-induced insulin resistant state (Scenario 1), the following filters were used: 1) log2(Fold Change) greater than 0 and p value less than 0.05 in the chow-fed insulin stimulated condition compared to chow-fed basal condition and 2) p value greater than 0.05 in high-fat diet-fed insulin stimulated condition compared to high-fat diet-fed basal condition. Similar filters were used to identify proteins and phosphosites that fit scenarios 2-4 (Table 4.1).

Table 3.1. Filters used to identify proteins and phosphosites in scenarios 1-4. CI vs CB HI vs HB

log2Fold change > 0 Scenario 1 p > 0.05 p < 0.05

log2Fold change > 0 Scenario 2 p > 0.05 p < 0.05

log2Fold change < 0 Scenario 3 p < 0.05 p > 0.05

log2Fold change < 0 Scenario 4 p > 0.05 p < 0.05

3.2.3 Western Blotting 3.2.3.1 Protein Quantification Whole liver and LD isolated lysates were transferred to a clear 96-well plate (Greiner) and diluted with water to a factor of 25 for protein quantification using the Bicinchoninic Assay (BCA) Protein Kit (PierceÔ, Thermo Fisher Scientific). A standard curve ranging from 0 to 1 mg/ml protein was prepared to interpolate the absorbance of the samples. After the reagent was added, the plate was incubated at 37°C for 30 mins and absorbance measured at 560 nm using Magellan software (TECAN Infinite M1000 PRO, Tecan Trading, AG, Switzerland). After

57 protein determination, samples of equal concentration were prepared using 6x Laemmli sample buffer (Cold Spring Harbor) and heated at 50°C for 10 mins or 75°C for 7 mins with gentle agitation (Thermomixer 5436, Eppendorf) to denature soluble proteins and allow antibody access to the epitope.

3.2.3.2 SDS-Polyacrylamide Gel Electrophoresis Proteins were separated based upon their size by using 10-12% hand-casted resolving gels with a 4% stacking layer made of the following stock ingredients: 40% acrylamide (Thermo Fisher), 1.5 M Tris pH 8.8 for resolving or 0.5 M Tris pH 6.8 for stacking, 10% sodium dodecyl sulphate (SDS), 10% ammonium persulphate and TEMED (Bio-Rad Laboratories). Gels were poured into 1.5 mm spaced casting plates and set with 10 or 15-well 1.5 mm combs. After polymerization, combs were removed, and wells were filled with 1x Tris--SDS running buffer (Astral Scientific). Equivalent amount of protein (30-40 µg) were loaded onto the gels alongside molecular weight markers ranging from 10 to 250 kD (Precision Plus ProteinÔ KaleidoscopeÔ Prestained Standards, Bio-Rad Laboratories) and run at 90V until the samples stacked, after which it was increased to 160V until samples were appropriately separated.

3.2.3.3 Protein Transfer to a PVDF Membrane Polyvinylidene fluoride (PVDF) membranes with 0.45 µm pores (ImmobilonÒ-P, Thermo Fisher Scientific) were activated in methanol for 60 seconds and then submerged in tris-glycine transfer buffer (Astral Scientific) for 5 mins along with two extra thick filter papers (Thermo Fisher Scientific). For semi-dry transfers, a tight sandwich was created in the following order: filter paper > gel > membrane > filter paper; and transferred using the Trans-Blot Turbo Transfer System (Bio-Rad Laboratories) at 25V for 30-60 mins. For wet transfers, a tight sandwich was created using mini gel holder cassettes in the following order: red side of cassette > foam pad > filter paper > gel > membrane > filter paper > foam pad > clear side of cassette. For both methods, bubbles were carefully removed using a roller. Cassettes were lodged into mini trans-blot cores alongside a bio-ice cooling unit (Bio-Rad Laboratories) and ran at 90V for 90 mins surrounded in ice. Wet transfers were primarily used for proteins larger than 100 kD.

58 3.2.3.4 Antibody Probing Membranes were blocked in 3% BSA diluted in 1 x Tris buffered saline (TBS; Astral Scientific) containing 0.1% Tween-20 (Sigma) for 60 mins at room temperature on a rocker. For primary antibody incubation, membranes were submerged in an antibody solution made of 3% BSA diluted in TBST containing primary antibody and kept overnight at 4°C with gentle agitation (Table 3.2).

Table 3.2. Primary antibodies used in this chapter. A list of proteins of interest that were visualized for abundance using western blotting. Antibody Catalogue # Dilution Species Company anti-ATGL 2138S 1:1000 Rabbit CST anti-CytoC 11940S 1:1000 Rabbit CST anti-ERP72 5033S 1:1000 Rabbit CST anti-G0S2 SC-133423 1:500 Rabbit SCBT anti-Na/K-ATPase 3010S 1:1000 Rabbit CST anti-Plin1 Ab3526 1:1000 Rabbit Abcam anti-Rab11 3539 1:1000 Rabbit CST CST – Cell Signaling Technologies (CST); SCBT – Santa Cruz Biotechnology.

After overnight primary antibody incubation, membranes were kept at room temperature for 30 mins to allow for further binding before washing with TBST every 15 mins for 1 hour on a shaker. Membranes were then incubated in solution made of 3% BSA in TBST containing a selected secondary antibody for 1 hour on a rocker (Table 3.3). The following table is a list of all the secondary antibodies used in this chapter. Finally, membranes were washed again every 15 mins for at least 1 hour on a shaker.

Table 3.3 Secondary antibodies used in this chapter. A list of secondary antibodies that were used to visualize abundance of proteins of interest using western blotting. Antibody Catalogue # Dilution Species Company Rabbit 7074P2 1:5000 Goat CST CST – Cell Signaling Technologies (CST).

59 3.2.3.5 Imaging with Bio-Rad ChemiDoc MP Proteins were visualized after incubation with primary and secondary antibodies by placing membranes onto a transparent sheet onto the imaging platform and carefully pressed using a roller to remove bubbles and excess TBST. Membranes were evenly coated with Luminata Crescendo Western HRP (Merck) for at least 60 seconds before imaging using the Bio-Rad ChemiDoc MP Imaging System (Bio-Rad Laboratories). Quantification of protein abundance was performed using densitometry analysis using Image Lab software 4.1 (Bio-Rad Laboratories).

3.2.3.6 Stripping and Re-Probing Restore PLUS Western Blot Stripping Buffer (Thermo Fisher Scientific) was used to remove primary and secondary antibodies. Membranes were incubated for 10 mins with gentle shaking, followed by 10 mins of PBS, and then 10 mins of TBST. Primary antibody removal was confirmed by incubating membranes in ECL then imaged using the ChemiDoc MP Imaging System for at least 5 mins. Membranes were then blocked with 3% BSA in TBST for 1 hour on the rocker before re-probing with a new primary antibody overnight.

60 3.3 Results

3.3.1 Hepatic Lipid Droplet Morphology and Purity LDs were isolated from extracted livers of basal or acutely insulin-stimulated male Wistar rats that had been fed either a chow diet or HFD for 4 weeks to determine the hepatic LD-associated proteome and phosphoproteome (Fig 3.1). Rats fed a high-fat diet had greater accumulation of LDs in the liver compared to rats fed a chow diet (Fig 3.2A). This supports the other biochemical measures presented in Chapter 2 to demonstrate that rats fed a high-fat diet had fatty livers.

Rat liver LD fractions did not contain adipocyte-derived LDs (Plin1) or widely-used markers of other organelles, including endoplasmic reticulum (ERP72), mitochondrial (Cytochrome C), plasma membrane (Na-K ATPase) and recycling endosomal (Rab11; Fig 3.2B). Importantly, the LD fractions were enriched for the lipid droplet protein ATGL and its co-regulator G0S2 (Fig 3.2B). Overall, these data demonstrate that the floating fractions obtained following exposure to a discontinuous sucrose-gradient contain LDs.

Figure 3.2 Purification of hepatic lipid droplets. (A) Haematoxylin and eosin stain of lipid droplets in chow (top) and high-fat fed (bottom) rats (B) Immunoblot analysis of lipid droplet purity.

3.3.2 Intra-Group Variability and Reproducibility in Proteome and Phosphoproteome Proteins from isolated LDs were identified using mass spectrometry. Principal components analysis (PCA) of LC-MS/MS spectra from all replicates across 4 conditions was performed to visualize variance within the datasets. The PCA score plot for LD-associated proteins

61 showed that replicates within each condition associated together to form clusters, indicating greater intergroup rather than intragroup variability (Fig 3.3A). Furthermore, the PC coordinate system reduces dimensionality to indicate the direction of the highest variance (PC1), shown on the x axis, followed by the second highest variance (PC2) shown on the y-axis. Therefore, the majority of the variance between samples in Fig 3.3A were attributed to diet, as illustrated by the separation of chow-fed and HFD-fed replicates along the PC1 axis. Interestingly, the PCA score plot for LD-associated phosphosites showed an even greater separation between chow and HFD clusters, while basal and insulin-stimulated chow clusters overlapped slightly (Fig 3.3B). In general, replicates within a condition clustered together and the greater part of the variance were attributed to treatment conditions.

Normalisation of data was performed to further reduce intragroup variance between biological replicates. Proteomic and phosphoproteomic box plots of log2 transformed quantile normalized peptide intensities are shown in Fig 3.3C and D respectively. After normalisation, distribution of transformed protein intensity had similar shapes (box) and extents (whiskers). Further, normalisation requires equivalent datasets across conditions and thus, the horizontal width of each box plot, which is proportional to the number of proteins, is identical between all conditions. Reproducibility of the proteomic and phosphoproteomic datasets were also quantified using standardisation correlation coefficients from duplicates (Fig 3.3E and F respectively). A correlation coefficient, r, close to 1.0 indicates high reproducibility, as seen with the proteome (r = 0.998) and phosphoproteome (r = 0.980).

A multiscatter analysis was also performed to assess the reproducibility of the protein and phosphosite quantification between biological replicates within each condition, where each dot represents one protein or phosphosite. A median Pearson correlation coefficient of 0.901 was found for the LD-associated proteome (Fig 3.3G) and 0.822 for the LD-associated phosphoproteome (Fig 3.3H), indicating relatively higher biological variability particularly for the phosphoproteome, likely due to the low replicate number. It is important to note that while median values are less likely to skew results caused by outliers, they are also less likely to be representative of the results when there are only four replicates per conditions.

62

Figure 3.3 Quality control of proteome and phosphoproteome (A) Principal component analysis of proteome and (B) phosphoproteome. (C) Post-normalization box plots of proteome and (D) phosphoproteome. (E) Standardisation correlation of proteome and (F) phosphoproteome. (G) Multiscatter plots of proteome and (H) phosphoproteome.

63 3.3.3 Quantitative Analysis of Proteome and Phosphoproteome LC-MS/MS analysis identified 4831 LD-associated proteins based on fragmentation spectra of peptides obtained by the mass spectrometer (Fig 3.4A). Of those identified, 4819 proteins were quantified using signal intensities in order to infer relative protein abundance profiles over multiple conditions. Proteins that were not found in all 4 replicates and across all 4 conditions were then removed to allow for more comprehensive comparisons, which resulted in a final 4115 LD-associated proteins identified in all 16 rats. Of the proteins removed from analysis, no condition-specific clustering of proteins was present (i.e. identified in all 4 replicates of 1 condition). Pathway analysis on the proteome using STRING revealed enrichment of expected LD processes such as insulin signalling pathway, peroxisome proliferator-activated receptor (PPAR) signalling pathway and fatty acid metabolism (Fig 3.4B).

Alongside proteomic analysis, 8616 LD protein-associated phosphosites were identified after phosphopeptide enrichment to determine normal and perturbed, diet and insulin-stimulated post-translational modifications (PTMs; Fig 3.4C). Of those identified, 6875 phosphosites were quantified, with 2286 sites found across all 16 rats. Similar to the proteome, no condition- specific clustering of phosphosites was present. This list was narrowed down further to reveal 2141 phosphorylation sites on 1362 proteins that were defined as having the highest confidence in localisation (Class I). Class I sites were obtained by filtering for phosphosites that had at least a 75% probability for the phospho-group (localization probability > 0.75) and that the difference in score between the first and second possible localizations was at least 5 or higher (probability localization score difference > 5) (Olsen et al. 2006). It should be noted that post- translational modification (PTM)-level data will always have higher variance than data at the protein-level, and this is part of the limitations of having a relatively lower number of replicates. This can be seen in the 67% reduction in quantified phosphosites when filtering for those found in all 16 rats (Fig 3.4C). The list of class I phosphosites were then compared to those found in the published online database of known phosphorylation sites hosted by PhosphoSitePlus to determine the number of novel phosphosites identified in this study. The comparison revealed 53 phosphosites that have previously been reported in other studies, and 2087 newly identified phosphosites in Rattus norvegicus (Fig 3.4D). It should be noted that the majority of these “novel” phosphosites could have already been reported as orthologous sites in the more commonly studied species, Mus musculus, and would therefore not be entirely undiscovered. Phosphoamino acid analysis was also performed to determine if the distribution of phosphorylated serine, threonine and tyrosine was comparable to previously published data.

64 In accordance with classic studies by Hunter and Sefton (1980) and Olsen et al. (2006), phosphorylated serine sites were the most abundant with 90.47% of Class I phosphosites, followed by threonine (8.92%) and tyrosine (0.5%; Fig 3.4E). Finally, pathway analysis on the Class I phosphoproteome using STRING revealed enrichment of commonly phosphorylated pathways, such as insulin signalling pathway and AMPK signalling pathway (Fig 3.4F).

Figure 3.4 Total proteome, phosphoproteome and associated enriched pathway analysis. (A) Identification and quantification of proteome (B) Enrichment pathway analysis of total proteome found across all 16 rats (C) Identification and quantification of phosphorylome (D) Venn diagram of previously identified phosphosites in Rattus norvegicus and phosphosites identified in this study (E) Proportion of phosphoamino acid in Class I phosphorylome (F) Enrichment pathway analysis of total phosphoproteome found across all 16 rats.

65 3.3.4 Differential Expression Analysis of LD-Associated Proteome Differential expression analysis was performed to determine how the relative abundance of proteins changed across the 4 conditions. Differentially expressed (DE) proteins were defined as proteins whose abundance at the LD significantly changed between two conditions (p < 0.05). The change in protein abundance at the LD can be attributed to the synthesis of new proteins, the translocation of proteins to or from the LD or a combination of both. The abundance of 63 proteins at the LD increased and 177 proteins decreased in response to insulin stimulation (Fig 3.5A). Furthermore, 105 proteins increased and 203 proteins decreased at the LD in response to high-fat feeding. Strikingly, both treatments resulted in roughly double the number of proteins that decreased in abundance at the LD compared to those that increased in abundance. In contrast, insulin stimulation in the HFD-induced insulin resistant state was associated with a similar number of proteins with increased and decreased abundance at the LD (132 increased, 129 decreased).

Canonical pathway analysis was performed to determine the functional enrichment of these DE proteins (Fig 3.5B). The top significantly increased pathways in response to insulin were oxidative phosphorylation, mitochondrial dysfunction and sirtuin signalling, all of which encompass the same group of mitochondrial proteins of the electron transport chain (ETC). While the majority of the LD fraction is free from impurity (Fig 3.2B), the crude LD isolation method is likely contain some proteins from other organelles, consistent with previously reported LD proteomic studies (Zhang et al. 2011, Crunk et al. 2013, Khan et al. 2015). It is also important to note that the frequent crosstalk between LDs and mitochondria is not only to direct fatty acids for breakdown in times of starvation, but also to move fatty acids back to LDs for re-acylation to prevent lipotoxic mitochondrial dysfunction (Schrauwen et al. 2010, Bosma et al. 2012, Gao and Goodman 2015). This is further supported by the upregulation of the same pathways with HFD (Fig 3.5C). Strikingly, 3 of the top 10 pathways that decreased with insulin (Fig 3.5B), and 5 of the top 10 pathways that decreased with HFD (Fig 3.5C) are related to cholesterol biosynthesis, which is consistent with downregulation of de novo cholesterol synthesis when hepatic cholesterol needs have been capped.

66

Figure 3.5 Enriched pathway analysis of differentially expressed proteins. (A) Number of differentially expressed proteins represented as up- or downregulated between any two conditions (B) Enriched pathway analysis of proteins that were significantly up- or downregulated with high-fat diet in basal conditions (C) Enriched pathway analysis of proteins that were significantly up- or downregulated with insulin stimulation in chow-fed mice.

3.3.5 Differential Expression Analysis of LD-Associated Phosphoproteome Alongside proteomic analysis, identification of DE phosphorylated sites revealed 55 phosphosites that increased and 86 phosphosites that decreased in abundance at the LD with insulin stimulation (Fig 3.6A). Similarly, 41 phosphosites increased and 86 phosphosites decreased at the LD with high fat feeding. Therefore, insulin stimulation and HFD was associated with a net decrease in phosphorylation at the lipid droplet which is similar to DE analysis at the proteomic level (Fig 3.5A). Interestingly, there were five times as many detectable phosphosites that increased at the LD than decreased, despite a lack of change in protein abundance in insulin stimulated HF-fed rats. Phosphosites that significantly changed in response to insulin showed an enrichment for participating in mTOR signalling, AMPK

67 signalling and TAG synthesis pathways (Fig 3.6B). Although this upregulation is consistent with insulin signalling, it is unclear how these pathways interact at the LD. These same pathways as well as insulin receptor signalling pathway were also enriched in the LD phosphoproteome of liver LDs from animals fed a HFD (Fig 3.6C). Furthermore, phosphorylation of TAG degradation enzymes HSL and PNPLA3 were increased with HFD. It is important to note that many regulated proteins, such as HSL, have multiple antagonistic phosphorylation sites, and levels of phosphorylation may be dependent on protein levels. Consequently, pathway analysis that is largely dependent on protein IDs may not be the most informative method of investigation. In order to tackle one of these complexities, the phosphoproteome, containing 1304 unique phosphorylated proteins, was overlapped with the proteome, containing 4068 proteins, to reveal 784 proteins found in both datasets (Fig 3.6D). Of those 784 proteins identified, DE phosphosites that had no significant changes in abundance at the proteomic level were mapped, with 36 sites that were phosphorylated and 56 sites that were dephosphorylated upon insulin stimulation (Fig 3.6E). Moreover, 24 sites were phosphorylated and 56 sites were dephosphorylated with high-fat diet feeding. Taken together, these results suggest a role for PTM modifications in the regulation of LD dynamics, regardless of changes at the proteome-level.

68

Figure 3.6 Enriched pathway analysis of differentially expressed phosphosites. (A) Number of differentially expressed phosphosites represented as up- or downregulation between any two conditions (B) Enriched pathway analysis of proteins that were significantly up- or downregulated with high-fat diet (C) Enriched pathway analysis of proteins that were significantly up- or downregulated with insulin stimulation (D) Venn diagram of proteins found in both the proteomic and phosphoproteomic datasets (E) Number of differentially expressed phosphosites with insignificant protein fold change.

69 3.3.6 Modelling Insulin Resistance via Binary Differential Expression The data was visualised using volcano plots to further define the changes to the LD proteome and phosphoproteome in insulin stimulated and insulin resistant rat liver. Volcano plots assist in visualizing large datasets by highlighting proteins with high fold change and statistical power. A notable responder to insulin stimulation in chow fed animals (i.e. insulin-stimulated chow-fed vs chow basal) was glutathione S-transferase alpha 5 (GSTa5; Fig 3.7A), which has been reported to be upregulated with insulin stimulation in rat hepatocytes (Kim et al. 2003), and preferentially bound to TAG-enriched LDs vs CE-enriched LDs (Khor et al. 2014). However, GST1a5 was not detected in the phosphoproteome. In contrast, two phosphosites of AKT1s1 (otherwise known as PRAS40) were in the top 5 phosphosites with greatest statistical power (Fig 3.7B), which is consistent with previously reported studies on its phospho- regulation by insulin and localisation at the LD (Hao et al. 2014). These results provide insight into the insulin-stimulated processes involved on LDs in healthy rats.

Perturbed insulin signalling can be viewed two ways in this study using binary comparisons: 1) insulin signalling in normal (Chow-Insulin condition) versus insulin resistant (HFD-Insulin condition) rats, and 2) insulin resistant rats with acute insulin stimulation (HFD-Insulin condition) compared to basal (HFD-Basal condition). The protein levels of A4 (ApoA4) on LDs was different between the insulin stimulated, chow-fed animals and the insulin stimulated, HFD-fed animals (Fig 3.7C). ApoA4 is a chylomicron marker that is increased in response to insulin stimulation and meal feeding and inhibits hepatic gluconeogenesis (Li et al. 2014). Here, ApoA4 increased in abundance at the LD in the ‘perturbed insulin signalling’ state compared to normal, suggesting an expansion of the role of the LD in potentially regulating glucose metabolism.

The phosphorylation of Ser185 on the serine/threonine protein kinase WNK1 had the greatest statistical power and second highest fold-change between insulin stimulated, chow-fed animals and the insulin stimulated, HFD-fed animals (Fig 3.7D). WNK1 is directly phosphorylated by PKB/Akt in response to insulin (Jiang et al. 2005) and its phosphorylation (Ser2032) has been identified as a marker of simple steatosis (i.e. compared to normal and NASH samples) (Wattacheril et al. 2017). This is consistent with its increased abundance at the LD in our model of insulin resistant steatotic liver (Fig 3.7D).

70 Analysis of the insulin-responsive proteomes of HFD-fed rats, which are characterised by perturbed insulin signalling, identified dermcidin (DCD) as a striking outlier, with the highest significance and greatest decrease in fold change (Fig 3.7E). Although DCD is commonly known as a antimicrobial peptide secreted from sweat glands, adenovirus-mediated knockdown of DCD in mouse adipose tissue resulted in increased HSL and decreased Plin1 mRNA levels (Kim et al. 2008). Interestingly, this study also observed translocation of DCD and HSL away from LDs in response to insulin stimulation in HFD-fed mice.

On the other hand, the analysis of the phosphoproteome revealed a family of spectrin phosphosites (spectrin alpha sites Ser1039 and Ser1031, and spectrin beta site Ser2234) that were increased in response to insulin in HFD-fed rats (Fig 3.7F). Spectrin is a scaffolding protein that localises to LDs in rat liver (Turro et al. 2006), interacts with various phospholipids (Maksymiw et al. 1987, Janmey and Lindberg 2004, Grzybek et al. 2006), and functions as a liaison between lipid uptake and LD growth in Drosophila melanogaster (Diaconeasa et al. 2013). Conceivably, the increased phosphorylation of spectrin proteins in response to insulin stimulation in the liver of HFD-fed rats suggests greater lipid uptake, potentially from the lack of suppression of adipose lipolysis that is a characteristic of obesity-induced insulin resistance (Kim et al. 2017).

71

Figure 3.7 Volcano plots of proteome and phosphoproteome.

The volcano plots show protein or phosphosite abundance at the lipid droplet based on -log10(p-value) and fold change. Proteins or sites that were insignificant (p > 0.05), significant (0.01 > p < 0.05) or very significant (p < 0.01) were indicated as black-filled circles, dark grey-filled circles, and red-filled circles, respectively. Key lipolytic regulators (PLIN2, PLIN3, PLIN4, PLIN5, CGI-58, G0S2, PNPLA3, ATGL and HSL) were distinguished according to the legend. The plots include binary comparisons of (A) the proteome and (B) the phosphoproteome of chow-fed acutely insulin stimulated rats compared to basal, (C) the proteome and (D) the phosphoproteome of HFD-fed acutely insulin-stimulated rats compared to chow-fed rats, and (E) the proteome and (F) the phosphoproteome of HFD-fed acutely insulin-stimulated rats compared to basal.

72 Volcano plots are a simple and effective way to discover changes in large data-sets. The changes, or lack thereof, of known key regulators of lipolysis were mapped on the volcano plots presented in Fig 3.7. The majority of proteins (Plin2, Plin5, G0S2, Pnpla3, ATGL and HSL) were found in both the proteomic and phosphoproteomic datasets; however, Plin3, Plin4 and CGI-58 were not detected in the enriched phosphoproteome (Fig 3.7). In chow fed rats, Plin2 protein levels remained unchanged in response to insulin stimulation (Fig 3.7A) but phosphorylation of Ser254, a novel Plin2 phosphosite, was decreased (Fig 3.7B). Similarly, Plin2 protein levels did not change (Fig 3.7E) but phosphorylation of Ser253 and Ser417 was significantly decreased (Fig 3.7F) with insulin stimulation in HFD-fed rats. Interestingly, Plin5 protein levels remained constant in all conditions (Fig 3.7A, C, E) but phosphorylation of Ser163 was increased in insulin stimulated, HFD-fed rats compared to chow animals (Fig 3.7D) and decreased phosphorylation in high fat-fed insulin stimulated rats compared to basal (Fig 3.7F). The discrepancy suggests a diet vs insulin effect on Plin5 phosphorylation, specifically upregulation with HFD, and downregulation with insulin in HFD-induced insulin resistance. HSL contains many well characterised phosphosites including Ser565, Ser565, Ser659 or Ser660 (Hornbeck 2015), which were all detected in this study. Somewhat remarkably, the phosphorylation of these sites did not differ in any of the comparisons (Fig 3.7), but protein abundance was decreased with insulin stimulation in HFD-fed rats compared to insulin- stimulated, chow-fed animals (Fig 3.7C). Taken together, acute insulin stimulation and/or HFD feeding had little effect on the protein levels of key LD-associated regulators of lipolysis, whereas there were greater changes in the phosphorylation patterns.

3.3.7 Modelling Insulin Resistance via Biologically Relevant Scenarios 3.3.7.1 Identification of “Insulin Resistant” Proteins The two-dimensional aspect of differential expression analysis is limited in its ability to clearly illustrate the behaviour of proteins in multiple conditions simultaneously. This is especially important when investigating insulin resistance, which is defined as a condition where cells fail to respond normally to the insulin. To identify proteins that fit this definition, four potential scenarios were created on the basis of protein abundance or translocation at the LD (Fig 3.8). Each of the scenarios represent a distinct difference in insulin-stimulated protein behaviour at the LD in healthy animals vs HFD-induced insulin resistant animals. For example, scenario 1 captures proteins that normally increase in abundance at the LD with insulin stimulation, but fail to in insulin resistant liver. Similarly, in scenario 2 proteins normally do not increase in

73 abundance at LDs with insulin stimulation, but they do in the insulin resistant setting. Protein lists were generated by using DE fold change and p-values for two specific comparisons 1) chow-insulin vs chow-basal and 2) HFD-insulin vs HFD-basal. There were 42 proteins that fit into scenario 1 (Table 4.2), 87 proteins into scenario 2 (Table 4.3), 130 proteins in scenario 3 (Table 4.4), and 79 proteins in scenario 4 (Table 4.5).

Interestingly, all members of the electron transport chain (ETC) were identified as ‘insulin resistant’ despite residing in the mitochondrial. These results are likely indicative of changes in interorganellar contact between LD and mitochondria in HFD-induced IR compared to control. Although numerous studies have described the relationship between dysregulated mitochondrial oxidation and insulin sensitivity (Goodpaster 2013, Montgomery and Turner 2015, Fazakerley et al. 2018), the majority have not explored this correlation in hepatocytes. Further, most of the ETC proteins appear in Scenario 1, whereas insulin stimulation has been shown to reduce FA oxidation in the liver (see Chapter 4), potentially due to altered LD- mitochondria contact. Alternatively, many proteasomal proteins were also identified as “insulin resistant”, and their translocation to the LD during impaired signalling may be indicative of increased ubiquitination of proteins at the LD. This is consistent with published reports on the role of LDs as storage depots for degradation (Cermelli et al. 2006, Bersuker and Olzmann 2017). Markedly, no lipases or known lipolytic regulators, such as ATGL, HSL, CGI58 or G0S2, were identified in any of the proteomic scenarios, consistent with results presented in Fig 3.7.

74

Figure 3.8 Modelling of insulin-resistance in hepatic lipid droplet proteome. A list of 338 LD-associated insulin resistant proteins were identified through systematic filtering. An increase or decrease in abundance at the LD is defined as p < 0.05 with fold change greater than or less than 1. No change in abundance was defined as p > 0.05. Comparisons were made between 1) chow-insulin versus chow-basal and 2) HFD-insulin vs HFD-basal.

75 Table 3.4 List of hepatic lipid-droplet associated proteins in scenario 1

Name Symbol Uniprot ID Apoptosis inducing factor mitochondria associated 1 Aifm1 Q924M6 Aly/REF export factor Alyref D3ZXH7 ATP synthase subunit beta Atp5b G3V6D3 ATP synthase subunit epsilon, mito Atp5e P29418 ATP synthase F(0) complex subunit B1, mito Atp5f1 P19511 ATP synthase subunit e, mito Atp5i P29419 ATPase, H trnasporting, lysosomal V1 subunit G1 Atp6v1g1 B2GUV5 ATPase inhibitor, mito Atpif1 Q03344 BRICK1, SCAR/WAVE actin nucleating complex subunit Brk1 D3ZUP5 Chaperonin containing Tcp1, subunit 6B (zeta 2) Cct6b Q6AYJ7 Isoform 2 of cyclin-dependent kinase 12 Cdk12 Q3MJK5-2 Cytochrome c oxidase subunit 4 isoform 1, mito Cox4i1 P10888 Cytochrome c oxidase subunit 6C-2 Cox6c2 P11951 Density regulated re-initiation and release factor Denr B0BNB2 Enoyl-CoA hydratase domain-containing protein 3, mito Echdc3 Q3MIE0 Eukaryotic translation initiation factor 3 Eif3d Q6AYK8 Eukaryotic translation initiation factor 4E family member 2 Eif4e2 A0A096MK14 Fumarylacetoacetase Fah P25093 Family wih sequence similarity 134 member A Fam134a Q3MHU5 Haloacid dehalogenase-like domain-containing protein 2 Hdhd2 Q6AYR6 Heterogeneous nuclear ribonucleoprotein A0 Hnrnpa0 F1M3H8 IQ motif-containing GTPase activating protein 1 Iqgap1 G3V7Q7 Importin subunit alpha Kpna3 Q56R18 LOC100362640 LOC100362640 X1WI37 LOC100910934 LOC100910934 Q4V797 Macrophage migration inhibitory factor Mif P30904 Meiosis-specific nuclear structural protein 1 Mns1 Q6AXQ8 3-mercaptopyruvate sulfurtransferase Mpst P97532 NADH dehydrogenase (ubiquinone) 1 alpha subcomplex subunit 2 Ndufa2 D3ZS58 NADH dehydrogenase (ubiquinone) 1 alpha Ndufa4 B2RZD6 Poly(A) binding protein, nuclear 1, isoform CRA_a Pabpn1 G3V7Z8 Proteosome subunit alpha 7 Psma7 F1LSQ6 Proteosome subunit beta type-7 Psmb7 Q9JHW0 Prostaglandin G/H synthase 1 Ptgs1 Q66HK3 Glycogen phosphorylase, liver form Pygl P09811 RGD1564804 RGD1564804 D4A7G9 60S ribosomal protein L34 Rpl34 B2RZD4 Ribosomal protein S29 Rps29 P62275 Succinate dehydrogenase (ubiquinone) flavoprotein subunit, mito Sdha Q920L2 TatD DNase domain-containing 3 Tatdn3 M0R5U3 Transketolase Tkt P50137 Tetratricopeptide repeat domain 9C Ttc9c Q6P5P3 In scenario 1, proteins are listed as insulin-resistant if their protein abundance at the LD significantly increases with insulin stimulation in the chow-fed group but does not change in abundance with insulin stimulation in the HFD-fed group.

76

Table 3.5. List of hepatic lipid-droplet associated proteins in scenario 2

Name Symbol Uniprot ID ATP binding cassette subfamily A member 6 Abca6 M0R890 Ankyrin repeat and SOCS box containing 13 Asb13 D3ZEJ6 ATP synthase subunit d, mito Atp5h P31399 ATP synthase-coupling factor 6, mito Atp5j P21571 BCL2 interacting protein 2 Bnip2 D3ZGE0 Cingulin Cgn D4A4X4 Charged multivesicular body protein 5 Chmp5 Q4QQV8 Cytokine induced apoptosis inhibitor 1 Ciapin1 Q5XID1 Calponin-3 Cnn3 P37397 Collagen alpha-2(I) chain Col1a2 P02466 Cytochrome c oxidase subunit 5B, mito Cox5b P12075 Carboxypeptidase B2 Cpb2 Q9EQV9 Cytoplasmic polyadenylation element binding protein 3 Cpeb3 D3Z9M2 CWF19-like 1, cell cycle control (S. pombe) (Predicted) Cwf19l1 D3Z863 Cytochrome P450 2C11 Cyp2c11 P08683 DEAH (Asp-Glu-Ala-His) box polypeptide 38 (Predicted), isoform CRA_a Dhx38 D4A321 Dual specificity phosphatase 19 (Predicted), isoform CRA_b Dusp19 D4A8F3 Eukaryotic translation initiation factor 2B subunit epsilon Eif2b5 Q64350 Family with sequence similarity 98 member C Fam98c F1LQ27 Dimethylaniline monooxygenase [N-oxide-forming] 3 Fmo3 Q9EQ76 Isoform 2 of Fibronectin Fn1 P04937-2 Gamma-glutamylaminecyclotransferase Ggact Q4KM86 Vitamin K-dependent gamma-carboxylase Ggcx F7FCW3 synthetase Glul P09606 LOC688393 protein Gpn1 B1WBZ7 General transcription factor IIIC, polypeptide 5 Gtf3c5 A1L1K6 Hydroxyacid oxidase 2 Hao2 Q07523 Hemoglobin alpha, adult chain 2 Hba1 B1H216 HD domain containing 2 Hddc2 D3ZKT8 Histone H2B Hist1h2bl M0R4L7 Insulin-like growth factor-binding protein 4 Igfbp4 P21744 IlvB acetolactate synthase like Ilvbl D4ACG2 Type I inositol 3,4-bisphosphate 4-phosphatase Inpp4a D3ZAN1 Inositol polyphosphate-5-phosphatase K Inpp5k Q5XIU8 Lactamase, beta (Predicted) Lactb D3ZFJ6 Ragulator complex protein LAM Lamtor3 Q5U204 Legumain Lgmn Q5PPG2 LOC100364175 LOC100364175 Q5FVP9 Peptidylprolyl cis/trans isomerase, NIMA-interacting 1 LOC364561 B0BNL2 LOC366772 LOC366772 Q5VLR6

77 LOC501233 LOC501233 F1LQJ4 LSM2 homolog, U6 small nuclear RNA associated Lsm2 Q6MG66 Putative MAGE-like protein Maged2 Q9EPI7 Mast cell protease 1 Mcpt1 P09650 Mast cell protease 2 Mcpt2 P00770 Magnesium dependent phosphatase 1 Mdp1 D4A4U3 Multifunctional protein ADE2 Paics P51583 PDGFA associated protein 1 Pdap1 Q62785 Polymerase (RNA) II (DNA directed) polypeptide J (Predicted) Polr2j D3ZQI0 Polymerase I and transcript release factor Ptrf G3V8L9 Sodium/nucleoside cotransporter 2 Q91WW8 Q91WW8 RNA binding motif protein 26 Rbm26 D3ZRC3 RGD1309730 RGD1309730 B2RYT8 Ribosomal protein S6 kinase Rps6ka3 D3Z8E0 Ras-related 2 Rras2 Q5BJU0 Stromal cell derived factor 2 Sdf2 D4A4H5 Succinate dehydrogenase [ubiquinone] iron-sulfur subunit, mito Sdhb P21913 SEC63 homolog, protein translocation regulator Sec63 D4A2Z6 Selenoprotein SEPT.15 Q923V8 SET nuclear proto-oncogene Set B0BMV1 Splicing factor 3a subunit 3 Sf3a3 Q4KLI7 Solute carrier family 43 member 3 Slc43a3 D4A832 Solute carrier organic anion transporter family member 1A1 Slco1a1 P46720 Beta-synuclein Sncb Q63754 Small nuclear ribonucleoprotein F Snrpf D4AAT4 Sorting necin family member 21 Snx21 F1M3C9 Sorting necin family member 30 Snx30 D4A060 SPRY domain containing 7 Spryd7 Q5M7T2 Stomatin Stom Q5XI04 modifying factor 2 Sumf2 Q5BK78 Transgelin Tagln P31232 Tubulin folding cofactor E like Tbcel Q5PQJ7 Serotransferrin Tf P12346 Transforming growth factor beta-1-induced transcript 1 protein Tgfb1i1 Q99PD6 Mitochondrial import inner membrane translocase subunit Timm8a Q9WVA1 Tight junction protein 3 Tjp3 D3Z8G7 Transmembrane protein 106B Tmem106b Q6AYA5 Tropomyosin 1 Tpm1 P04692-5 TNFRSF1A associated via death domain Tradd D3ZZC5 Cdc42-interacting protein 4 Trip10 M0R3P9 Transthyretin Ttr P02767 Taxilin gamma Txlng Q4FZU5 Upstream binding transcription factor Ubtf P25977-2 UBX domain-containing protein 1 Ubxn1 Q499N6

78 Urinary protein 2 UP2 P81828 Cytochrome b-c1 complex subunit 7 Uqcrb B2RYS2 Cytochrome b-c1 complex subunit Rieske, mito Uqcrfs1 P20788 In scenario 2, proteins are listed as insulin-resistant if their protein abundance does not change at the LD with insulin stimulation in the chow-fed group, but increases in abundance at the LD with insulin stimulation in the HFD-fed group.

Table 3.6. List of hepatic lipid-droplet associated proteins in scenario 3

Name Symbol Uniprot ID Acyl-CoA-binding domain-containing protein 6 Acbd6 Q5RJK8 Long-chain-fatty-acid--CoA ligase 3 Acsl3 Q63151 ARP10 actin-related protein 10 homolog (S. cerevisiae) Actr10 Q5M9F7 Alcohol dehydrogenase 1 Adh1 F1LSR9 Afamin Afm G3V9R9 Alpha-2-HS-glycoprotein Ahsg P24090 Apoptosis-inducing factor, mitochondria-associated 2 Aifm2 D4AA14 Annexin A5 Anxa5 P14668 AP-2 complex subunit mu Ap2m1 P84092 Apolipoprotein A-IV Apoa4 P02651 Isoaspartyl peptidase/L-asparaginase Asrgl1 Q8VI04 Autophagy-related 2A Atg2a D3ZT64 Lys-63-specific deubiquitinase BRCC36 Brcc3 B2RYM5 Bromodomain-containing 4 Brd4 D3ZGX8 Basic leucine zipper and W2 domain-containing protein 2 Bzw2 Q9WTT7 Complement C1r subcomponent C1r D4A1T6 Complement C3 C3 M0RBJ7 C4b-binding protein beta chain C4bpb A0A5C5 Complement component C6 C6 Q811M5 Oxidation resistance protein 1 (Fragment) C7 F1M7F8 Complement component 8, alpha polypeptide (Predicted) C8a D3ZWD6 Carbonic anhydrase 2 Ca2 P27139 Carbonic anhydrase 3 Ca3 P14141 Coiled-coil domain-containing 9 Ccdc9 M0RA86 CD2 antigen (Cytoplasmic tail) binding protein 2 (Predicted), isoform CRA Cd2bp2 B4F786 Protein-glutamate O-methyltransferase CF211 Q6AYT5 B-factor, properdin Cfb Q6MG74 Complement inhibitory factor H Cfh Q91YB6 Complement factor I Cfi Q9WUW3 Clusterin Clu Q6P7S6 COMM domain containing 9 Commd9 Q3MIE7 Crystallin, zeta (Quinone reductase)-like 1 Cryzl1 Q5XI39 RNA polymerase II subunit A C-terminal domain phosphatase Ctdp1 D3ZMS2 Isoform 3 of NADH-cytochrome b5 reductase 3 Cyb5r3 P20070-3 N(G),N(G)-dimethylarginine dimethylaminohydrolase 2 Ddah2 Q6MG60

79 DEAD (Asp-Glu-Ala-Asp) box polypeptide 58 (Predicted) Ddx58 D4ADT5 Type I iodothyronine deiodinase Dio1 P24389 Protein kintoun Dnaaf2 F1LMB5 Eukaryotic translation initiation factor 6 Eif6 Q3KRD8 ELMO domain-containing protein 2 Elmod2 D3ZNV6 Erlin-2 Erlin2 B5DEH2 ERO1-like protein alpha Ero1l Q8R4A1 Exosome complex component CSL4 Exosc1 M0R5B6 Coagulation factor X F10 Q63207 Prothrombin F2 G3V843 FAS-associated factor 1 Faf1 Q924K2 FAS-associated factor 2 Faf2 Q5BK32 Fam45a protein (Fragment) Fam45a B2RZA3 Family with sequence similarity 96, member A Fam96a Q5RJS3 F-box and WD-40 domain protein 11 (Predicted) Fbxw11 D3ZXC6 Fibrinogen beta chain Fgb P14480 Fggy carbohydrate kinase domain-containing protein Fggy F1LSP9 Fibronigen-like protein Fgl1 Q8K583 Flotillin-1 Flot1 Q9Z1E1 Glucose-6-phosphatase G6pc A0A096MK74 Group specific component Gc Q68FY4 Glyoxalase domain-containing protein 4 Glod4 Q5I0D1 Guanine nucleotide binding protein beta 2 (Fragment) Gnb2 Q45QL6 Glucosamine 6-phosphate N-acetyltransferase Gnpnat1 B1H249 Golgi reassembly stacking protein 2 Gorasp2 Q68G33 Gelsolin 1 Gsn Q68FP1 Beta-globin chain Hbe2 Q63067 Dynein, axonemal, assembly factor 5 Heatr2 G3V943 Histidine-rich glycoprotein Hrg Q99PS8 Estradiol 17-beta-dehydrogenase 11 Hsd17b11 Q6AYS8 Heat shock 70 kDa protein 1A/1B Hspa1a Q07439 Hsp70-binding protein 1 Hspbp1 Q6IMX7 Ifi47 protein Ifi47 Q6NYB8 Insulin-like growth factor-binding protein complex acid labile subunit (Fragment) Igfals F1LRE2 Il1rap protein Il1rap Q4V8K9 Importin 9 (Predicted) Ipo9 D4A857 Immunity-related GTPase family M protein Irgm Q6AYC2 Integrin subunit alpha 2b Itga2b D3ZAC0 Integrin alpha-4 Itga4 D3ZMQ3 Inter-alpha trypsin inhibitor heavy chain H2 Itih2 D3ZFH5 Isoform 2 of Kinesin-like protein KIF1B Kif1b O88658-2 Kallikrein B, plasma 1 Klkb1 Q5FVS2 Kininogen-2 Kng2 Q5M8A0

80 Lecithin cholesterol acyltransferase Lcat O35849 Lipopolysaccharide responsive and beige-like anchor protein Lrba F1LZG6 Lymphocyte specific 1, isoform CRA Lsp1 Q4QQV6 O-acetyl-ADP-ribose deacetylase MACROD1 Macrod1 G3V8V6 Mitotic spindle assemble checkpoint protein MAD1 Mad1l1 D3ZIV3 tRNA (guanine-N(7)-)-methyltransferase Mettl1 M0R637 Murinoglobulin-1 Mug1 Q03626 Myosin light chain 1/3, skeletal muscle isoform Myl1 P02600 Nicotinate phosphoribosyltransferase Naprt1 G3V709 Nitric oxide synthase adaptor protein c NOS1AP D5LG85 Neurogranin Nrgn Q04940 Nuclear pore complex protein Nup214 Nup214 D4ACK1 NUS1 dehydrodolichyl diphosphate synthase Nus1 D3ZFM4 Poly (A) polymerase alpha (Predicted), isoform CRA Papola D3ZK96 Programmed cell death 6 (Predicted), isoform CRA Pdcd6 G3V7W1 3-phosphoinositide-dependent protein kinase 1 Pdpk1 O55173 Phosphoacetylglucosamine mutase Pgm3 B2RYN0 Phosphatidylinositol 3-kinase regulatory subunit beta Pik3r2 Q5FVS6 PITH domain containing 1 Pithd1 D4ABS5 Peptidyl-prolyl cis-trans isomerase Ppil1 Q4KLI4 E3 ubiquitin-protein ligase PPP1R11 Ppp1r11 Q6MFY6 Vitamin K-dependent protein Z Proz D3ZKM4 Proteasome subunit beta type-2 Psmb2 P40307 Ras-related protein Rab-18 Rab18 Q5EB77 Ras-related protein Rab-32 Rab32 D4AEG2 Ras-related protein Rab-7L1 Rab7l1 Q66HQ8 Rab-like protein 6 Rabl6 D3ZKQ4 Retinoblastoma-binding protei 5 Rbbp5 D3ZC01 CapZ-ineracting protein Rcsd1 F1M4V3 Retinol dehydrogenase 10 Rdh10 Q80ZF7 RGD1306556 RGD1306556 F1M8B8 E3 ubiquitin-protein ligase RNF123 Rnf123 F1LMI7 Serpin family A member 4 Serpina4 Q5M8C3 Serpin family F member 2 Serpinf2 Q68FT8 SH2-B PH domain containing signaling mediator 1, isoform CRA Sh2b1 M0R617 Carbohydrate kinase-like, isoform CRA Shpk Q3MID4 Spartin Spg20 E9PT90 Signal transducing adaptor family member 2 Stap2 Q5BK28 Metalloreductase STEAP4 Steap4 Q4V8K1 Transcription elongation factor B polypeptide 2 Tceb2 P62870 Transmembrane protein 263 Tmem263 B2RZ08 Transmembrane protein 33 Tmem33 Q9Z142 Torsin-1A Tor1a Q68G38

81 tRNA (guanine (26)-N(2))-dimethyltransferase Trmt1 Q5U4E4 Ubiquitin/ISG15-conjugating enzyme E2 L6 Ube2l6 Q4V8J2 Ubiquitin carboxyl-terminal hydrolase 33 Usp33 Q569A1 Vacuolar protein-sorting associated protein 13C Vps13c D4A4K4 Vacuolar protein-sorting associated protein 37B Vps37b D3ZU64 WW domain binding protein 2, isoform CRA Wbp2 G3V721 Protein yuppee-like 5 Ypel5 D4A4Q3 Terminal uridylultransferase 4 Zcchc11 F1M466 Zinc finger protein 330 Zfp330 D3ZSN4 In scenario 3, proteins are listed as insulin-resistant if their protein abundance decreases at the LD with insulin stimulation in the chow-fed group, but does not change in protein abundance at the LD with insulin stimulation in the HFD-fed group.

Table 3.7. List of hepatic lipid-droplet associated proteins in scenario 4

Name Symbol Uniprot ID Steroid 5 alpha-reductase (Fragment) A1E0X5 A1E0X5 Abhydrolase domain containing 2 (Predicted) Abhd2 D4A7W1 Annexin 7 Anxa7 Q8VIN2 Armc8 protein Armc8 B4F7A2 Ataxin-10 Atxn10 Q9ER24 Uncharacterized protein B1H247 B1H247 BET1 homolog Bet1 Q62896 Complement component C9 C9 Q62930 CDK2-associated and cullin domain-containing protein 1 Cacul1 Q5XI53 Coiled-coil domain-containing protein 47 Ccdc47 Q5U2X6 Cyclin-dependent kinase 4 (Fragment) Cdk4 Q71VC8 Cytidine/uridine monophosphate kinase 2 Cmpk2 D3ZC63 Coatomer subunit alpha Copa G3V6T1 Casein kinase II subunit alpha Csnk2a1 P19139 DDHD domain containing 1 Ddhd1 Q3ZAU5 Neutrophil antibiotic peptide NP-3B DEF3B Q9Z1F1 Protein Dr1 Dr1 Q5XI68 Down syndrome critical region gene 3 (Predicted), isoform CRA Dscr3 E9PU42 Eukaryotic translation initiation factor 3 subunit M Eif3m D3ZAZ0 Echinoderm microtubule associated protein like 4 (Predicted), isoform CRA Eml4 F1LT71 Isoform 2 of Endoplasmic reticulum aminopeptidase 1 Erap1 Q9JJ22-2 SEC3-like 1 Exoc1 Q4V8H2 Exosome component 3 Exosc3 D3ZT51 Family with sequence similarity 115, member C Fam115c M0RBL7 Family with sequence similarity 160, member B1 Fam160b1 D4A3I5 F-box only protein 4 (Predicted) Fbxo4 D3ZIH4 Fermitin family member 3 Fermt3 B2GVB9 FIP1-like 1 Fip1l1 B6RIU0 Isoform 2 of Fragile X mental retardation protein 1 homolog Fmr1 Q80WE1-2

82 GIMAP9 Gimap9 Q3ZAV4 GNAS complex Gnas B0BMW4 Golgin subfamily A member 2 Golga2 F1LSS0 Golgi phosphoprotein 3 Golph3 Q569C9 Histone deacetylase 1 Hdac1 Q4QQW4 Heme oxygenase 1 Hmox1 P06762 Importin 7 (Predicted), isoform CRA Ipo7 D4AE96 Interleukin-1 receptor-associated kinase 4 Irak4 D4A7K4 Interferon, alpha-inducible protein (Clone IFI-15K) (Predicted) Isg15 D4A3X3 T-kininogen 2 KNT2 P08932 Leucine carboxyl methyltransferase 1 Lcmt1 G3V7V9 Hormone-sensitive lipase Lipe P15304-2 LOC100912386 LOC100912386 M0RAK8 M0R9D9 M0R9D9 M0R9D9 Dual specificity mitogen-activated protein kinase kinase 2 Map2k2 P36506 Metallo-beta-lactamase domain-containing 2 Mblac2 D4A249 Histidine protein methyltransferase 1 homolog Mettl18 Q4KM84 Metallothionein 1 (Fragments) Mt1f Q7M082 Metallothionein Mt1m D3ZHV3 Myotubularin related protein 2 (Predicted), isoform CRA Mtmr2 D3ZA31 Myosin V1 Myo6 D4A5I9 Nuclear cap-binding protein subunit 2 Ncbp2 B1WC40 Isoform Cytoplasmic of Cysteine desulfurase, mitochondrial Nfs1 Q99P39-2 Aminopeptidase-like 1 (Predicted) Npepl1 D4A3E2 Aminopeptidase Npepps F1M9V7 Nuclear pore complex protein Nup54 Nup54 P70582 Alpha-1-acid glycoprotein Orm1 P02764 Peflin Pef1 Q641Z8 Peroxisome assembly factor 2 Pex6 P54777 Parvalbumin alpha Pvalb P02625 Pyridine nucleotide-disulfide oxidoreductase domain-containing protein 2 Pyroxd2 Q68FT3 RGD1309995 RGD1309995 D4A7I6 RGD1310429 RGD1310429 D3ZEJ9 N-terminal kinase-like protein Scyl1 Q5M9F8 Sodium-coupled neutral amino acid transporter 4 Slc38a4 Q9EQ25 Smu-1 suppressor of mec-8 and unc-52 homolog (C. elegans) Smu1 G3V702 Signal recognition particle 54 kDa protein Srp54 Q6AYB5 Non-specific serine/threonine protein kinase Stk38 G3V9Q4 Sulfotransferase Sult2a1 M3ZCQ0 Similar to CG9339-PA (Predicted) Tbc1d24 D4A3Z3 Tia1 cytotoxic granule-associated RNA-binding protein-like 1 Tial1 Q5BJN3 Mitochondrial import inner membrane translocase subunit Tim8 B Timm8b P62078

83 FLN29 gene product, isoform CRA Trafd1 G3V959 Similar to tripartite motif protein 47 Trim47 D3ZA22 Ubiquitin-activating enzyme E1-like (Predicted) Uba7 D3ZEU5 Ube2g2 protein Ube2g2 B2RYD0 Voltage-dependent anion-selective channel protein 3 Vdac3 Q9R1Z0 Similar to vacuolar protein sorting 26 homolog, isoform CRA_a Vps26b B1WBS4 WD repeat domain phosphoinositide-interacting protein 4 Wdr45 Q5U2Y0 Similar to zinc finger FYVE domain containing 19 Zfyve19 Q499R6 In scenario 4, proteins are listed as insulin-resistant if their protein abundance does not change at the LD with insulin stimulation in the chow-fed group, but decreases in abundance at the LD with insulin stimulation in the HFD-fed group.

3.3.7.2 Identification of “Insulin Resistant” Phosphosites Protein phosphorylation is most meaningful when the protein levels are known, thus phosphosites were overlaid with the proteomic data. It should be noted that the illustration in Fig 3.9 only represents one of multiple interpretations of the phosphorylation data. An increase in a particular phosphosite at the LD could be due to increased phosphorylation, as shown in Fig 3.9, or due to translocation of previously phosphorylated protein to the LD (not shown). A list of 252 LD-associated ‘insulin resistant’ phosphosites were identified following data filtering through the four scenarios (Fig 3.9). In general, there were 37 phosphosites on 36 different proteins that fitted scenario 1, 26 of which were found in the proteomic dataset as well (Table 4.8). Similarly, 118 phosphosites on 96 phosphoproteins were identified in scenario 2, 64 of which were in the proteomic dataset (Table 4.9). In scenario 3, 77 phosphosites on 70 phosphoproteins with 52 of them in the proteomics dataset were identified (Table 4.10). Finally, 20 phosphosites on 19 phosphoproteins fit scenario 4, 14 of which were quantified in the proteomic dataset (Table 4.11).

In contrast to the proteome, phosphosites identified as having abnormal phosphorylation with acute insulin treatment in the HFD-induced IR group generally belonged to proteins involved in diverse and non-canonical LD processes, such as translation factors and scaffolding proteins. The exception to this observation was the identification of phosphosites on several lipolytic regulators listed in scenario 3 and 4 (Table 4.12). Despite a lack of change in the proteome, phosphorylation of Ser422 on PNPLA2 (otherwise known as ATGL) was found to decrease with insulin stimulation at the LD, but remained there in the HFD-fed animals. Although Ser422 has previously been identified as a phosphosite on ATGL in human, mouse and rat studies (Lundby et al. 2012, Wilson-Grady et al. 2013, Bian et al. 2014), its role in modulating ATGL activity has not been characterised. However, published reports in non-hepatocytes on

84 the effect of alternative phosphosites on ATGL indicate insulin responsiveness and regulation of LD localization (Chakrabarti et al. 2013, Xie et al. 2014). Additionally, phosphorylation of Ser410 on PNPLA3 and Ser253 and S417 on Plin2 were found to not respond to insulin in chow-fed animals, but decreased in those fed a HFD (Table 4.12). The phosphorylation of Ser410 on PNPLA3 and Ser417 on Plin2 are both novel points of regulation that are unidentified in any species, while Ser253 on Plin2 has been identified in a proteogenomic study using human breast cancer (Mertins et al. 2016). These results suggest that a key regulator of lipolysis, ATGL, is phospho-regulated in response to acute insulin stimulation in hepatocytes, and this cannot be accounted for by changes in protein abundance at the LD. Further, phosphorylation of ATGL, PNPLA3 and Plin2 in response to insulin is perturbed in obese and insulin-resistant rats, suggesting an underestimated role for altered PTMs of proteins involved in lipolysis in the pathogenesis of IR.

Figure 3.9 Modelling of insulin resistance in hepatic lipid droplet phosphoproteome LD-associated insulin resistant phosphosites were identified through systematic filtering. An increase or decrease in phosphosites at the LD is defined as p < 0.05 with fold change greater than or less than 1. No change was defined as p > 0.05. Comparisons were made between 1) chow-insulin versus chow-basal and 2) HFD-insulin vs HFD-basal.

85 Table 3.8. List of hepatic lipid-droplet associated phosphosites in scenario 1

Name Symbol Uniprot ID AA Site 40S ribosomal protein S6 (Fragment) Rps6 M0RD75 S 233 Afadin Afdn O35889 S 1282 Alanine and glycine-rich protein-like LOC103694328 D3ZBC4 S 274 Betaine--homocysteine S-methyltransferase 1 Bhmt O09171 S 330 Cytoplasmic polyadenylation element binding protein 4 (Predicted) Cpeb4 D3ZKL3 S 97 Enhancer of mRNA-decapping protein 4 Edc4 Q3ZAV8 S 924 Enthoprotin Clint1 Q6DGF2 T 308 EPS8-like 2 (Predicted), isoform CRA Eps8l2 F7DLY1 S 528 Erythrocyte membrane protein band 4.1 Epb41 D3ZKF7 S 535 Eukaryotic translation initiation factor 4 gamma, 1 Eif4g1 D4AD15 S 1140 Golgi brefeldin A-resistant guanine nucleotide exchange factor 1 Gbf1 F1M8X9 S 1786 Isoform 4 of Tuberin Tsc2 P49816-4 S 939 KH domain-containing, RNA-binding, signal transduction-associated protein 1 Khdrbs1 Q91V33 S 38 KN motif and ankyrin repeat domains 1 Kank1 D4AE58 S 318 La ribonucleoprotein domain family, member 1 Larp1 F1M062 S 627 Lipin 1 Lpin1 Q5XIM8 S 328 Membrane-associated guanylate kinase, WW and PDZ domain-containing protein 3 Magi3 F1M7S0 S 702 Mitochondrial antiviral-signaling protein Mavs Q66HG9 S 152 MLLT1, super elongation complex subunit Mllt1 M0R462 S 307 Neuroprotective protein 8 Zwint B3SVE5 S 206 Phosphoinositide kinase, FYVE-type zinc finger containing Pikfyve D3ZYT8 S 1753 Pumilio 2 Pum2 D3ZQL8 S 181 RCG40648, isoform CRA_b Sec23ip G3V8Q8 S 749 Rho guanine nucleotide exchange factor 5 Arhgef5 E9PT59 S 699 RNA-binding protein 3 Ranbp3 M0R5Q3 S 58 Scaffold attachment factor B1 Safb M0RBF0 S 367 Scaffold attachment factor B1 Safb M0RBF0 S 466 Serine/ repetitive matrix protein 1 Srrm1 B2RYB3 S 657 Serine/arginine repetitive matrix protein 2 Srrm2 F1LRJ2 S 231 Serine/threonine-protein kinase PRP4 homolog Prpf4b Q5RKH1 S 387 Serine/threonine- 1 regulatory subunit 10 Ppp1r10 F1LQE9 S 313 Splicing factor, arginine/serine-rich 19 Scaf1 Q63624 S 495 Symplekin Sympk Q561R4 S 293 Tetratricopeptide repeat domain 4 Ttc4 Q5XI93 S 51 Thyroid hormone receptor-associated protein 3 Thrap3 Q5M7V8 S 405 Ubiquitin-conjugating enzyme E20 Ube2o F1M403 S 892 Zinc finger CCCH domain-containing protein 18 Zc3h18 Q6TQE1 S 887 In scenario 1, phosphosites are listed as insulin-resistant if their intensity at the LD increases with insulin stimulation in the chow-fed group, but does not change at the LD with insulin stimulation in the HFD-fed group. AA and site refer to the amino acid residue and position that is phosphorylated on that protein.

86

Table 3.9. List of hepatic lipid-droplet associated phosphosites in scenario 2

Name Symbol Uniprot ID AA Site A kinase anchor protein 13 Akap13 F1M3G7 T 932 A-kinase anchor protein 2 Akap2 F1LPQ9 S 789 Actin-binding LIM protein 1 Ablim1 Q3KR72 S 304 Adducin 3 (Gamma), isoform CRA Add3 D3ZCH7 S 648 Afadin Mllt4 O35889 S 1090 Aldehyde oxidase 3 Aox3 Q5QE80 S 196 Alpha-parvin Parva G3V818 S 19 AP2-associated protein kinase 1 Aak1 P0C1X8 S 679 DNA-(apurinic or apyrimidinic site) lyase Apex1 Q99PF3 S 13 Arf-GAP with coiled-coil, ANK repeat and PH domain-containing protein 2 Acap2 Q5FVC7 S 384 ATPase WRNIP1 Wrnip1 Q8CG07 S 153 B-cell CLL/lymphoma 6 Bcl6 B2GUV8 S 334 BUD13 homolog Bud13 G3V8F3 S 135 Calcium-transporting ATPase Atp2b1 D3ZAX3 S 1142 CD2AP/CMS Cd2ap Q80ZE7 S 233 Desmuslin Synm Q810D0 S 1042 Epithelial protein lost in neoplasm Lima1 F1LR10 S 488 Epithelial splicing regulatory protein 2 Esrp2 B2RYJ8 S 562 Etoposide-induced protein 2.4 homolog Ei24 Q4KM77 S 47 Eukaryotic translation initiation factor 4E binding protein 2 Eif4ebp2 Q497A9 S 65 FERM, RhoGEF and pleckstrin domain- containing protein 1 Farp1 F1LYQ8 T 24 Gap junction beta-1 protein Gjb1 P08033 S 266 General transcription factor IIF subunit 2 Gtf2f2 Q63489 S 248 Glycogen phosphorylase, brain form (Fragment) Pygb P53534 S 15 Glycogen phosphorylase, liver form Pygl P09811 S 15 GTP-binding protein 1 Gtpbp1 D2XV59 S 24 H2-K region expressed gene 4, rat orthologue Slc39a7 Q6MGB4 S 276 Hepatocyte nuclear factor 1-alpha Hnf1a P15257 S 247 High mobility group nucleosomal binding domain 3, isoform CRA Hmgn3 M0R5I3 S 6 Interferon regulatory factor 2-binding protein-like Irf2bpl M0R567 S 320 Isoform 2 of E3 ubiquitin-protein ligase TRIP12 Trip12 F1LP64-2 S 1069 Isoform 2 of ERBB receptor feedback inhibitor 1 Errfi1 P05432-2 S 256 Kinesin 13B Kif13b Q70AM4 S 1718 Kinesin 13B Kif13b Q70AM4 S 1719 Lamin A, isoform CRA Lmna G3V8L3 S 390 Lamin A, isoform CRA Lmna G3V8L3 S 22 Lamin A, isoform CRA Lmna G3V8L3 S 392 Lamina-associated polypeptide 2, isoform beta Tmpo Q62733 S 155

87 Leukocyte common antigen Ptprc Q6LDZ3 S 633 Leukocyte common antigen Ptprc Q6LDZ3 S 898 Map3k3 protein Map3k3 B5DF98 S 250 Methyl-CpG-binding protein 2 (Fragment) Mecp2 F1LWH6 S 359 Monocarboxylate transporter 1 Slc16a1 P53987 S 213 Nuclear factor 1 Nfic O70188 S 333 Nuclear factor 1 Nfib O70187 S 265 Nuclear ubiquitous casein and cyclin-dependent kinase substrate 1 Nucks1 Q9EPJ0 S 39 Nuclear ubiquitous casein and cyclin-dependent kinase substrate 1 Nucks1 Q9EPJ0 S 229 Palmdelphin Palmd Q4KM62 S 371 Palmdelphin Palmd Q4KM62 S 376 Paralemmin-1 Palm Q920Q0 T 145 Plectin 4 N/A Q6S3A1 S 4153 Polyhomeotic like 3 (Drosophila) (Predicted) Phc3 D3ZS50 S 109 Polymerase I and transcript release factor Ptrf G3V8L9 S 42 Adipogenesis-associated, Mth938 domain- containing Aamdc D3ZZQ4 S 48 Apoptotic chromatin condensation inducer 1 Acin1 E9PST5 S 1004 Apoptotic chromatin condensation inducer 1 Acin1 E9PST5 S 216 Neuroblast differentiation-associated protein Ahnak M0R9D5 S 136 Neuroblast differentiation-associated protein Ahnak M0R9D5 S 5211 Neuroblast differentiation-associated protein Ahnak M0R9D5 S 94 Neuroblast differentiation-associated protein Ahnak M0R9D5 S 217 Cingulin Cgn D4A4X4 S 91 CCR4-NOT transcription complex, subunit 3 Cnot3 D3ZUV9 S 299 Cordon-bleu WH2 repeat protein-like 1 Cobll1 F1M124 S 567 Eukaryotic translation initiation factor 4 gamma, 1 Eif4g1 D4AD15 S 1588 Erythrocyte membrane protein band 4.1-like 2 Epb41l2 D3ZM69 S 58 Erbin Erbb2ip M0R5K2 S 571 Family with sequence similarity 114, member A1 Fam114a1 D4A777 S 119 WASH complex subunit 2C Washc2c F1LPG9 S 387 Ubiquitin carboxyl-terminal hydrolase MINDY-1 Fam63a Q5BJQ2 S 16 H1 histone family, member X H1fx D3ZIX4 S 134 Family with sequence similarity 234, member A Fam234a Q5M7W6 S 21 LOC100911677 LOC100911677 M0R7L9 S 590 LOC100911677 LOC100911677 M0R7L9 S 594 LOC680377 Anp32a M0R9D0 T 13 Leucine-rich repeat-continaing 47 Lrrc47 F1LT49 S 429 MLLT1, super elongation complex subunit Mllt1 M0R462 T 306 Par-3 family cell polarity regulator beta Pard3b F1LW22 S 338 Pinin, desmosome-associated protein Pnn D3ZAY8 S 346 Serine/threonine-protein phosphatase 2A 56 kDa regulatory subunit Ppp2r5e D3ZHI9 S 32

88 Uncharacterized protein N/A F1LT30 S 149 RNA-binding protein 27 Rbm27 F1LWJ2 S 100 Ras-related GTP-binding C Rragc Q0D2L6 S 381 Protein transport protein sec16 Sec16a D3ZN76 S 1365 SWI/SNF-related, matrix-associated, actin- dependent regulator of chromatin, subfamily c, member 2 Smarcc2 D4A510 S 302 Spectrin beta chain Sptbn1 G3V6S0 S 2334 Spectrin beta chain Sptbn1 G3V6S0 S 2096 Spectrin beta chain Sptbn1 G3V6S0 S 2158 SFRS protein kinase 2 Srpk2 B1WBT4 S 10 Serine/arginine repetitive matrix protein 2 Srrm2 F1LRJ2 T 965 Serine/arginine repetitive matrix protein 2 Srrm2 F1LRJ2 S 1383 Serine/arginine repetitive matrix protein 2 Srrm2 F1LRJ2 S 1481 Serine/arginine repetitive matrix protein 2 Srrm2 F1LRJ2 S 772 Serine/arginine repetitive matrix protein 2 Srrm2 F1LRJ2 S 1723 RCG61762, isoform CRA_d Srsf7 D4A720 S 183 START domain containing 10, isoform CRA_b Stard10 Q5BJN1 S 283 TBC1 domain family member 22B Tbc1d22b F1LNA5 S 40 Tight junction protein 1 Tjp1 F1M4A0 S 915 Tight junction protein 1 Tjp1 F1M4A0 S 166 Tensin 1 Tns1 F1LN42 S 1365 Trafficking protein particle complex subunit 12 Trappc12 M0R780 S 232 USP6 N-terminal like (Predicted), isoform CRA_a Usp6nl D3ZL57 S 701 USP6 N-terminal like (Predicted), isoform CRA_a Usp6nl D3ZL57 S 392 Zinc finger CCCH type-containing 13 Zc3h13 E9PSN4 S 1271 Rat glutathione S-transferase Gstm1 Q6LDP3 S 199 Regulator of microtubule dynamics protein 2 Rmdn2 Q498D5 S 121 RNA polymerase-associated protein LE Leo1 Q641X2 S 642 Serine/threonine-protein kinase WNK1 Wnk1 Q9JIH7 S 1776 Sharpin Sharpin Q9EQL9 S 307 Sister chromatid cohesion protein PDS5 homolog B Pds5b D3ZU56 S 1187 Sorbin and SH3 domain-containing protein 1 (Fragment) Sorbs1 F1M866 S 329 Sorbin and SH3 domain-containing protein 1 (Fragment) Sorbs1 F1M866 S 330 Spectrin alpha chain, non-erythrocytic 1 Sptan1 Q6IRK8 S 1029 Spectrin alpha chain, non-erythrocytic 1 Sptan1 Q6IRK8 S 1031 Starch-binding domain-containing protein 1 Stbd1 Q5FVN1 S 192 UBX domain-containing protein 1 Ubxn1 Q499N6 S 188 Vimentin Vim P31000 S 72 Zinc finger CCCH domain-containing protein 18 Zc3h18 Q6TQE1 S 531 Zinc finger CCCH domain-containing protein 18 Zc3h18 Q6TQE1 S 529

89 In scenario 2, phosphosites are listed as insulin-resistant if their intensity at the LD does not change with insulin stimulation in the chow-fed group, but increases at the LD with insulin stimulation in the HFD-fed group. AA and site refer to the amino acid residue and position that is phosphorylated on that protein.

Table 3.10. List of hepatic lipid-droplet associated phosphosites in scenario 3

Name Symbol Uniprot ID AA Site 1,4-alpha-glucan-branching enzyme Gbe1 A0A096MJY6 S 133 3 beta-hydroxysteroid dehydrogenase type 5 Hsd3b5 P27364 S 372 Ac1873 Fga Q7TQ70 S 46 Adaptor-related protein complex AP-4, beta 1 Ap4b1 D4AD35 S 593 Alpha-2-HS-glycoprotein Ahsg P24090 S 309 Alpha-2-HS-glycoprotein Ahsg P24090 S 313 Ankyrin repeat and SAM domain containing 1 Anks1a D4AC12 S 640 Ankyrin repeat and SAM domain containing 1 Anks1a D4AC12 S 638 Antisecretory factor Psmd4 O88321 T 267 Arf-GAP domain / FG repeat-containing protein 1 Agfg1 F1M9N7 T 179 Arf-GAP domain / FG repeat-containing protein 1 Agfg1 F1M9N7 T 177 Arsenite methyltransferase As3mt Q8VHT6 S 46 Autophagy related 2B Atg2b F1MAF8 S 379 Autophagy related 2B Atg2b F1MAF8 S 496 Calcium-regulated heat stable protein 1 Carhsp1 Q9WU49 S 32 Calmodulin Calm1 P62161 S 102 CAP-Gly domain-containing linker protein 2 Clip2 O55156 S 353 cGMP-inhibited 3',5'-cyclic phosphodiesterase A Pde3a Q62865 S 310 Cysteine-rich protein 2 Crip2 P36201 S 104 Cytochrome P450 2A1 Cyp2a1 P11711 S 130 Dihydropyrimidinase-related protein 2 Dpysl2 P47942 S 522 Dihydropyrimidine dehydrogenase [NADP(+)] Dpyd O89000 S 905 Diphosphoinositol pentakisphosphate kinase 2 Ppip5k2 A0A096MK18 S 593 DNA damange-inducible 1 homolog 2 Ddi2 A0A096MJP9 S 85 DNA fragmentation factor subunit beta Cad D4A8A0 S 1406 Dopey family member 1 Dopey1 D3ZY24 S 1261 Elongation factor for RNA polymerase II Ell D4A753 S 310 Epidermal growth factor receptor pathway substrate 15-like 1 Eps15l1 D3ZJR1 S 255 Epithelial protein lost in neoplasm Lima1 F1LR10 S 225 Eukaryotic translation initiation factor 4B Eif4b Q5RKG9 S 459 F-box only protein 30 Fbxo30 Q5XI67 T 385 FAM122A Fam122a Q6AYT4 S 188 FAS-associated factor 1 Faf1 F1LSQ0 S 22 FCH domain only protein 2 Fcho2 D3ZYR1 S 403 Fructose-1,6-bisphosphatase 1 Fbp1 P19112 S 353 Glucocorticoid receptor Gr A8IRI3 S 154 High density lipoprotein binding protein (Vigilin) Hdlbp Q3KRF2 S 645

90 Isoform 2 of Adenylate kinase 2, mitochondrial Ak2 P29410-2 T 30 Isoform 2 of Alpha-endosulfine Ensa P60841-2 S 109 Isoform 2 of Programmed cell death protein 4 Pdcd4 Q9JID1-2 S 313 Kinesin family member 16B Kif16b D3ZFN9 S 662 LOC100361087 LOC100361087 D4A2R8 S 69 LOC100912115 LOC100912115 D3ZVL1 S 606 Methyltransferase like 7A Mettl7a Q3KRE2 S 218 Nuclear receptor binding SET domain protein 1 (Predicted), isoform CRA Nsd1 D4AA06 S 2161 Nuclear ubiquitous casein and cyclin-dependent kinase substrate 1 Nucks1 Q9EPJ0 S 181 O-phosposeryl-tRNA(Sec) selenium transferase Sepsecs B2GV97 S 469 Optineurin Optn Q8R5M4 S 346 Optineurin Optn Q8R5M4 S 217 Oxidation resistance protein 1 Oxr1 Q4V8B0 S 83 Patatin-like phospholipase domain-containing protein 2 Pnpla2 P0C548 S 422 Patched domain containing 1, isoform CRA_b Ptchd1 D3ZX73 S 672 Phosphoinositide Plcg1 G3V845 S 1248 Protein kinase C eta type (Fragment) Prkch R9PXS6 S 196 regulatory subunit 12 Ppp1r12a D4ACS0 S 526 Protein phosphatase 1 regulatory subunit 12 Ppp1r12a D4ACS0 S 527 Pseudouridylate synthase 7-like Pus7l D4ABN2 S 92 RCSD domain containing 1, isoform CRA_b Rcsd1 F1M4V3 S 53 RGD1305117 RGD1305117 A0A059NZV6 S 173 Rho GTPase-activating protein 25 Arhgap25 D3ZGL1 S 363 Rhophillin, Rho GTPase binding protein 2 Rhpn2 D4A8N7 T 655 Rhophillin, Rho GTPase binding protein 2 Rhpn2 D4A8N7 S 654 Ribosome-binding protein 1 Rrbp1 F1M5X1 S 135 Serine/threonine-protein kinase 24 (Fragment) Stk24 H9KVF3 T 329 SIK family kinase 3 Sik3 M0RD40 S 671 Spermatogenesis-defective protein 39 homolog Vipas39 Q5PQN6 S 119 Sterile alpha and TIR motif containing 1 (Predicted) Sarm1 D3ZUM2 S 572 Sterol-4-alpha-carboxylate 3-dehydrogenase, decarboxylating Nsdhl Q5PPL3 T 17 T-complex protein 1 subunit epsilon Cct5 Q68FQ0 S 485 Trio and actin binding 1a (Fragment) Triobp A2TIS7 S 170 Ubiquitin carboxyl-terminal hydrolase Usp5 D3ZVQ0 S 785 Ubiquitin carboxyl-terminal hydrolase 10 Usp10 Q3KR59 S 361 Unconventional myosin-IXb Myo9b Q63358 S 1649 Vacuolar protein sorting 13 homolog D Vps13d D3ZKC6 S 2455 Zero beta-globin (Fragment) N/A Q63011 S 89 Zinc finger protein 148 Znf148 Q62806 S 784 Zinc finger Rab-binding domain-containing 2 Zranb2 O35986 S 290

91 In scenario 3, phosphosites are listed as insulin-resistant if their intensity at the LD decreases in intensity with insulin stimulation in the chow-fed group, but does not change at the LD with insulin stimulation in the HFD-fed group. AA and site refer to the amino acid residue and position that is phosphorylated on that protein.

Table 3.11. List of hepatic lipid-droplet associated phosphosites in scenario 4

Name Symbol Uniprot ID AA Site 60S ribosomal protein L13 Rpl13 P41123 S 140 Acyl-CoA synthetase long-chain family member 5 Acsl5 Q6IN15 S 408 Arginine/serine-rich protein 1 Rsrp1 Q5U2S0 S 17 C3G protein Rapgef1 Q9QYV3 S 203 Cullin 4A Cul4a B2RYJ3 S 10 Deltex E3 ubiquitin ligase 3L Dtx3l D3Z8X6 S 9 Interferon-inducible double-stranded RNA- dependent protein kinase activator A Prkra G3V7J2 T 20 Lupus La protein homolog Ssb Q66HM7 S 92 Patatin-like phospholipase domain-containing 3 Pnpla3 D3Z9J9 S 410 Perilipin 2 Plin2 Q5U2U5 S 253 Perilipin 2 Plin2 Q5U2U5 S 417 PPFIA-binding protein 2 Ppfibp2 G3V9H6 T 517 Protein transport protein sec16 Sec16a D3ZN76 S 1311 Pseudopodium-enriched atypical kinase 1 Peak1 D4A563 S 281 RalBP1-associated Eps domain-containing protein 2 Reps2 F1M0T5 S 172 RGD1564606 (Fragment) RGD1564606 F1LT35 S 43 RGD1565685 RGD1565685 B1WBZ6 T 315 RNA polymerase II-associated protein 1 Rpap1 Q3T1I9 S 201 Serine/threonine-protein kinase D1 Prkd1 Q9WTQ1 S 403 Superoxide dismutase [Cu-Zn] Sod1 Q6LDS4 S 104 In scenario 1, phosphosites are listed as insulin-resistant if their intensity at the LD does not change with insulin stimulation in the chow-fed group, but decreases at the LD with insulin stimulation in the HFD-fed group. AA and site refer to the amino acid residue and position that is phosphorylated on that protein.

Table 3.12. Lipolytic regulators identified as ‘insulin resistant’

Name Symbol Phosphosite Scenario CI vs CB HI vs HB Patatin-like phospholipase PNPLA2 S422 3 ¯ n.s domain-containing protein 2 Patatin-like phospholipase PNPLA3 S410 4 n.s ¯ domain-containing protein 3

Perilipin 2 Plin2 S253 4 n.s ¯

Perilipin 2 Plin2 S417 4 n.s ¯

92 3.4 Discussion

LDs are dynamic and specialized organelles that govern the storage and turnover of neutral lipids, namely triglycerides and sterol esters, and can include retinol esters and acylceramides in certain cell types (Blaner et al. 2009, Senkal et al. 2017). Surrounding this lipid core is a monolayer of phospholipids that are decorated by a catalogue of proteins involved in traditional pathways such as energy metabolism (Greenberg et al. 2011, Walther and Farese 2012), and non-canonical processes such as viral, bacterial and parasitic propagation (Jackson et al. 2004, Samsa et al. 2009, Daniel et al. 2011). These non-canonical examples of LD-associated processes collectively highlight the diverse functions that occur at the LD. These ubiquitous lipid depots are conserved across nearly all organisms, including bacteria and plants, and are found in the majority of cell types, particularly adipose and liver (Yang et al. 2012). Research into the biology of LDs in a broad range of cell types has dramatically increased in the past decade due to its demonstrated role in the pathogenesis of obesity, NAFL, IR and associated cardiovascular complications (Xu et al. 2018). Proteomic analysis of LDs have identified hundreds to thousands of proteins localized at their surface, including enzymes involved in lipid metabolism, structural proteins for LD biogenesis and inter-organelle contact, and signalling proteins implicated in a number of diseases (Ducharme and Bickel 2008, Arrese et al. 2014, Lin et al. 2017, Schuldiner and Bohnert 2017). Moreover, studies have shown clear differences in the proteome of LDs in models of hepatic steatosis and IR compared to control (van Loon and Goodpaster 2006, Bostrom et al. 2007, Su et al. 2014, Khan et al. 2015). The aim of the experiments described in this chapter was to systematically elucidate the influence of acute insulin stimulation and diet on the LD proteome and phosphoproteome in order to unravel the actions of known candidates while also identifying novel targets in the pathogenesis of NAFL and IR. The key results obtained from this descriptive analysis revealed that 1) the abundance of neutral lipases at the LD did not differ between healthy and HFD-induced insulin resistant animals, but 2) the levels of the co-regulators of ATGL activity were altered in a manner that suggests dysregulation and increased ATGL activity in IR, and 3) did demonstrate a significant impairment in phosphorylation between the conditions, suggesting the importance of post-translational modifications in metabolic diseases.

93 3.4.1 Proteomic Characterization of LDs The analysis presented in the results of this chapter identified 4116 proteins associated with hepatic LDs across all 16 rats, many of which have been previously reported in other LD proteomic studies. Although it is not uncommon to identify over a thousand proteins on mammalian liver or even insect LDs (Khan et al. 2015, Rosch et al. 2016, Kramer et al. 2018), this number is generally larger than that reported by others (Zhang et al. 2011, Khor et al. 2014, Konige et al. 2014, Saka et al. 2015, Wang et al. 2015). This discrepancy may be due to the higher sensitivity of the mass spectrometer used for this study, or because of the variability in LD isolation protocols in the field. In particular, LDs were isolated from whole liver tissue that encompasses a heterogenous cell population of 80% hepatocytes and 20% vitamin A-storing stellate cells, macrophagic Kupffer cells and sinusoidal endothelial cells. (Damania et al. 2014). The proteome of LDs differ between cell types in order to cater to distinct functions of that cell, such as the retinol hydrolases found in LDs of hepatic stellate cells to aid in retinol metabolism (Molenaar et al. 2017). Further, LDs are highly gregarious organelles that form strong interactions with other organelles and this can contribute to contaminants that are not bona fide LD proteins, despite the post-isolation assessment of the purity of enriched LD fractions by western blot. Additionally, it is impossible to imagine that all 4115 proteins are found on every LD at any given time but rather populate in distinct LD populations that are diverse and multifunctional even within a single cell. For instance, nucleating LDs are rich in proteins involved in TAG synthesis such as DGAT and budding/scission such as Arf1/COP1 (Wilfling et al. 2014). In contrast, expanding LDs contain proteins involved in fusion and pore formation such as Rab8 and CIDE proteins (Hashemi and Goodman 2015, Xu et al. 2016). Additionally, catabolic LDs associated with lysosomes or mitochondria are high in Atg7, ATP synthase subunits and COX proteins (Hashemi and Goodman 2015, Benador et al. 2018). As such, all of these proteins from distinct subpopulations are found in the LD proteome of this study. Thus, the dynamic nature of LDs allow them to host a large variety of proteins to facilitate diverse functions.

3.4.1.1 Influence of HFD on LD Proteome High fat feeding for 4 weeks led to changes in the abundance of a number of LD-associated proteins, particularly the prominent LD marker in hepatocytes, PLIN2 (Itabe et al. 2017). Increase in the availability of lipids increased the number and size of LDs in the liver and was therefore associated with elevated expression of LD-associated PLIN2. This is consistent with

94 proteomic studies in hepatocytes, myoblasts and enterocytes that administered a chronic fat challenge rather than short-term lipid supply, as the latter is generally associated with an increase in PLIN3 rather than PLIN2 (Crunk et al. 2013, Beilstein et al. 2016, Chen et al. 2016). Additionally, several pathways that were upregulated with HFD feeding participate in TAG catabolism, which is consistent with previous studies reporting increased fat oxidation when fed a HFD (Schrauwen et al. 2000, Cole et al. 2011, Sikder et al. 2018). Amongst these upregulated oxidative or lipophagic proteins is PLIN5, a member of the perilipin family known to promote interactions between LDs and mitochondria (Bosma et al. 2012). PLIN5 is correlated with higher oxidation rates fueled by increased lipolysis (Bosma et al. 2012, Kimmel and Sztalryd 2014), consistent with studies that have demonstrated increased interorganellar contact related to oxidation in HFD-fed rats (Krahmer et al. 2018). While elevated availability of FFAs are known to promote increased fatty acid oxidation rates and increase the expression of proteins that facilitate oxidative phosphorylation (Schrauwen et al. 2000), it is unlikely that plasma FFAs are driving this change. Indeed, the rate of FA oxidation is governed by glycogen stores rather than lipids in hepatocytes and muscles (Randle et al. 1963) and thus a diet lacking in carbohydrates may boost FA oxidation and elevate oxidative phosphorylation. Therefore, at rest, the FAs that enter oxidation when on a HFD are likely derived from TAG stores rather than plasma (Schrauwen et al. 2000). The regulators that facilitate the release of TAG-derived FAs, such as ATGL and HSL, did not change in abundance at the LD; however, the LD protein CGI58 was significantly upregulated with a HFD. As described in section 1.2.3.2.2, CGI58 binds to ATGL to promote its activation and stimulate TAG hydrolytic activity. Therefore, molecular interactions rather than total expression levels of lipases are regulatory drivers of FA release, which likely go on to act as substrates for oxidation.

3.4.1.2 Influence of Insulin on LD Proteome Insulin primarily promotes TAG synthesis in hepatocytes while suppressing TAG lipolysis in adipocytes. Despite this fundamental regulatory pathway, the effect of insulin signaling on the LD proteome has not been explored in any other cell type. In the current study, lipolytic regulators such as ATGL, HSL, CGI58 and G0S2 did not change in abundance at the LD of livers in response to insulin stimulation in the chow-fed group. While acute stimulation is unlikely to be not long enough to see changes in total protein expression, it is likely that primary regulation of lipolytic activity is via protein-protein interactions at the LD such as that between ATGL and G0S2 in response to insulin (Wang et al. 2013, Zhang et al. 2016). Unexpectedly in this study, half the pathways enriched with insulin treatment were related to the

95 downregulation of cholesterol synthesis and lipoprotein metabolism. In particular, lanosterol synthase and cholestenol delta-isomerase, enzymes involved in the last step of the mevalonate pathway and the second last step of the Kandutsch-Russell pathway in cholesterol synthesis respectively, were both decreased with insulin stimulation (Sharpe and Brown 2013, Cerqueira et al. 2016). Similarly, proteins responsible for generating cholesteryl ester-rich lipoproteins from intestinal absorption of dietary substrates such as apolipoprotein A-4 and lecithin- cholesterol acyltransferase were also downregulated with insulin (Wang et al. 2015, Zannis et al. 2017). Although insulin is known to decrease levels of cAMP, which in turn promotes the activation of cholesterol synthesis (Bathaie et al. 2017), it appears that 10 minutes was insufficient to see these expected physiological responses. Further, cholesterol synthesis traditionally occurs at the ER and thus the extent of these effects at the LD, which has a large pool of cholesteryl esters, is unknown. It is known that insulin’s stimulatory effect in rat hepatocytes takes approximately 6 hours to measure changes in the expression of the rate- limiting enzyme, HMGCR, and roughly 24 hours to observe differences in incorporation of acetate into cholesterol (Wiss and Wiss 1977). It is likely that the downregulation of cholesterol synthesis enzymes was not reflective of the post-insulin treatment. Although not all proteins require such long effect times, these results present an interesting challenge in analyzing proteomic data in response to acute hormonal treatment.

3.4.2 Phosphoproteomic Characterization of LDs The major mechanism that insulin utilizes to control the activity of downstream proteins is reversible phosphorylation. Briefly, insulin binds to its receptor on the cell surface and results in receptor autophosphorylation of tyrosine residues, thereby transducing the phosphorylation of substrates via their tyrosine kinase activity (Kruger et al. 2008). These substrates then activate PI3K, which initiates PKB/Akt signalling, leading to downstream phospho-stimulation or inhibition of metabolic pathways, such as activation of glycogen synthesis and attenuation of lipolysis and gluconeogenesis (Bevan 2001). In eukaryotic cells, phosphorylation most often occurs on the serine residue (84%), followed by threonine (15%) and tyrosine (<1%) (Humphrey et al. 2015), relatively similar to the proportions found in this study. Although protein phosphorylation is not exclusive to the insulin signalling pathway, insulin-sensitive phosphorylation of proteins at the LD has been largely unexplored. However, the majority of phosphoproteomic studies have focused on upstream effectors of insulin signalling in the whole cell (Kruger et al. 2008, Parker et al. 2015, Zhang et al. 2017) rather than downstream targets which associate with the LD. In the current study, 2141 class I phosphosites on 1362 LD-

96 associated proteins were quantified from the isolated livers of all 16 rats. Over 97% of these phosphosites were considered novel sites in rattus norvegicus which are less characterised than mus musculus when it comes to proteomic and phosphoproteomic research. It is likely that the majority of these novel phosphosites are orthologs of phosphosites identified in mus musculus. The number of identified phosphosites was far greater than that reported in a previous phosphoproteomic study performed on LDs in HeLa cells, whereby 8 sites on 7 proteins were quantified after overnight oleate stimulation (Bartz et al. 2007). Of these 7 proteins, only 4 were identified in the current study - ACC1, PLIN2, ATGL and PLA2G4A. In particular, both studies pinpointed the phosphorylation of ATGL on Ser422 in rat liver, otherwise known as Ser428 in human HeLa cells (Bartz et al. 2007), and has been identified in over 45 mass spectrometry studies in human, rat and mouse total lysates (Hornbeck 2015) and validated in mouse adipocytes (Lundby et al. 2012). In contrast, all other phosphosites were not described in this study, potentially due to differing phosphorylation events in response to oleate versus insulin treatment.

3.4.3 Phospho-Proteomics and HFD-Induced Insulin Resistance The pathogenesis of NAFL and IR is associated with dysregulation of numerous pathways in the metabolism of lipids and glucose, causing major disorder in protein activity, distribution and intermolecular interactions. Almost 20% of proteins in hepatocytes completely shift their subcellular localization as a result of hepatic steatosis, which is in contrast to their behaviour in healthy organs (Krahmer et al. 2018). Further, 6% of proteins move away from important processes in other organelles and are drawn toward LDs in steatosis (Krahmer et al. 2018), demonstrating the importance of not only total protein abundance but protein localization as well. A high proportion of proteins residing on LDs were also shown to be phosphorylated or have uncharacterised phosphorylation sites compared to organelles such as peroxisome, Golgi apparatus, endosomes, lysosomes and mitochondria, despite enrichment of fewer absolute phosphosites (Krahmer et al. 2018). Thus, the movement and abundance of phosphorylated proteins suggests the importance of post-translational modifications in organelles such as LDs, and that steatosis could be linked to aberrant phosphorylation events. Indeed, analysis of 54 kinase substrates in insulin-mediated pathways of adipose tissue revealed differentiated protein phosphorylation in patients with NAFLD, particularly those sites phosphorylated by PKCd, AKT and SHC (Calvert et al. 2007). In the liver, 12 phosphorylation sites were identified to distinguish NASH patients from those with simple steatosis (Wattacheril et al. 2017). Since the

97 hallmark feature of NAFL is the accumulation of LDs in hepatocytes, it is conceivable that many of these aberrant phosphorylation events occur at the LD.

In the current study, many of the regulators of lipolysis that have been suggested to participate in the pathogenesis of NAFL, namely ATGL, HSL, CGI58 and G0S2, were not changed in abundance at the LD between chow and HFD-fed rats. These results are consistent with subcellular proteomic analysis performed by Krahmer et al. (2018), whereby canonical LD markers were localized to LDs independent of diet in hepatocytes. Interestingly, these lipolytic regulators had the greatest fold change in phosphosite intensity at the LD and revealed disparities between normal and ‘insulin-resistant’ states that could not be attributed to changes at the proteomic level. In particular, ATGL was phosphorylated on Ser422, a site which has not been extensively characterized despite being a known phosphosite that has been identified in various rat and mouse tissues (Lundby et al. 2012, Humphrey et al. 2013, Wilson-Grady et al. 2013). Data presented in this chapter suggest that Ser422 is an insulin-sensitive phosphosite that is affected by impaired insulin action in HFD-induced IR. In contrast, Ser410 on PNPLA3 and Ser253 and Ser417 on PLIN2 are not sensitive to insulin stimulation but unexpectedly increase in intensity at the LD when stimulated with insulin on a HFD. Ser253 on PLIN2 has been previously reported (Mertins et al. 2016) but its function remains to be determined, while the other two phosphosites on PNPLA3 and PLIN2 are novel. PNPLA3 is one of the strongest candidate biomarkers for NAFL but its underlying mechanism still remains unclear (Romeo et al. 2008, Graff et al. 2013, Salameh et al. 2016). In the liver, PNPLA3 is a resident LD protein with predicted TAG hydrolase capacities. However, after generating a PNPLA3 mutant via substitution of methionine for isoleucine at residue 148 (I148M), the hydrolase activity was lost and resulted in hepatic fat accumulation (He et al. 2010). However, both gain and loss-of- function studies revealed no change in intrahepatic TAG levels, lending to the hypothesis that the PNPLA3 mutant is a neomorph with new functions (Chen et al. 2010, Basantani et al. 2011). The novel phosphosite identified on PNPLA3 in the current study may present a new perspective on its regulation in hepatocytes; specifically, the phosphorylation that occurs in HFD-induced IR. Finally, a number of lipolytic regulators increased in phosphorylation with HFD-feeding only and were not insulin-sensitive in either diet group. One of these regulators was G0S2, which was identified as a phosphoprotein for the first time in this study. The phosphoregulation of G0S2 is conceivable considering that ATGL, HSL and CGI58 are all known phosphoproteins.

98 Collectively, these results point to novel phosphosites and new perspectives on the importance of post-translational modifications in regulating LD proteins. Specifically, studies into the total expression levels of canonical LD proteins have presented contradictory results and this suggests the importance of alternative reversible regulation such as phosphorylation. Consistent with previous findings, abundance levels of key lipid regulators does not appear to be a key regulatory mechanism when compared to “activation” by phosphorylation, or translocation and protein-protein interactions that may arise from being phosphorylated. This study provides the first step in establishing new regulatory mechanisms for lipid metabolism at the LD and its role in the progression of NAFL and IR.

99

Chapter 4

Comprehensive Characterisation of Hepatocytes for Functional In vitro Validation

100 Lipid and Glucose Metabolism in Hepatocyte Cell Lines and Primary Mouse Hepatocytes: A comprehensive resource for in vitro studies of hepatic metabolism

Shilpa R. Nagarajan 1, Moumita Paul-Heng 2, James R. Krycer 3, Daniel J. Fazakerley 3, Alexandra F. Sharland 2, Andrew J. Hoy 1 #

1 Discipline of Physiology, School of Medical Sciences & Bosch Institute, Charles Perkins Centre, Faculty of Medicine and Health, The University of Sydney, NSW 2006, Australia. 2 Discipline of Surgery, Central Clinical School & Bosch Institute, Charles Perkins Centre, Faculty of Medicine and Health, University of Sydney, NSW 2006, Australia. 3 School of Life and Environmental Sciences, Charles Perkins Centre, Faculty of Science, The University of Sydney, NSW 2006, Australia.

Running Head: Basal and Insulin Stimulated Metabolism in Hepatocytes

# Address for Correspondence: Andrew J. Hoy, Discipline of Physiology, The Hub (D17), Charles Perkins Centre, University of Sydney, NSW, Australia 2006. Phone: +61 2 9351 2514, Email: [email protected]

101 Abstract

The liver is a critical tissue for maintaining glucose, fatty acid and cholesterol homeostasis. Primary hepatocytes represent the gold standard for studying the mechanisms controlling hepatic glucose, lipid and cholesterol metabolism in vitro. However, access to primary hepatocytes can be limiting and therefore other immortalized hepatocyte models are commonly used. Here we describe substrate metabolism of cultured AML12, IHH and PH5CH8 cells, hepatocellular carcinoma-derived HepG2s and primary mouse hepatocytes (PMHs) to identify which of these cell lines most accurately phenocopy PMH basal and insulin-stimulated metabolism. Insulin-stimulated glucose metabolism in PH5CH8 cells, and to a lesser extent AML12 cells, responded most similarly to PMH. Notably, glucose incorporation in HepG2 cells were 14-fold greater than PMH. The differences in glucose metabolic activity were not explained by differential protein expression of key regulators of these pathways, for example glycogen synthase and glycogen content. In contrast, fatty acid metabolism in IHH cells was the closest to PMHs, yet insulin-responsive fatty acid metabolism in AML12 and HepG2 cells was most similar to PMH. Finally, incorporation of acetate into intracellular free cholesterol was comparable for all cells to PMH; however, insulin-stimulated glucose conversion into lipids and the incorporation of acetate into intracellular cholesterol esters were strikingly different between PMHs and all tested cell lines. In general, AML12 cells most closely phenocopied PMH in vitro energy metabolism. However, the cell line most representative of PMHs differed depending on the mode of metabolism being investigated, and so careful consideration is needed in model selection.

102 Keywords: Fatty acid metabolism; glucose metabolism; cell lines; protein expression; insulin action

Abbreviations:

ACAT1, Acetyl-CoA acetyltransferase, mitochondrial; ACC1, Acetyl-CoA carboxylase, ATGL, Adipose ; CD36, Cluster of differentiation 36; CGI58, Comparative gene identification; CPT1a, Carnitine palmitoyltransferase I; DGAT, Diacylglycerol acyltransferase; DNL, De novo lipogenesis; FASN, Fatty acid synthase; FATP5, Fatty acid transport protein; G0S2, G0/G1 switch gene 2; GLUT, Glucose transporter; GSK3B, Glycogen synthase kinase 3 beta; HCC, Hepatocellular carcinoma; HMGCR, 3- hydroxy-3-methyglutaryl-CoA reductase; HSL, Hormone sensitive lipase; LDL, Low-density lipoproteins; NAFLD, Non-alcoholic fatty liver disease; NEFA, Non-esterified fatty acids; PDHK1, Pyruvate dehydrogenase kinase isozyme 1; PLIN, Perilipin; PMH, Primary mouse hepatocytes; TAG, Triacylglycerol; TCA, Tricarboxylic acid; VLDL, Very low-density lipoproteins

103 INTRODUCTION

The liver plays an essential role in substrate homeostasis and detoxification. In fact, the liver ensures that roughly 70 functions within 12 prime metabolic domains, including carbohydrate, lipid and lipoprotein metabolism, progress continuously, in periodic cycles, or upon necessity (2008). For example, 33% of post-prandial glucose is cleared by the liver and can be broken down to pyruvate via glycolysis, oxidized through the TCA cycle to produce ATP, stored as glycogen, or converted into fatty acids or amino acids through de novo synthetic pathways that use glycolytic products (Moore et al. 2012, Sharabi et al. 2015). Conversely, whole-body euglycemia is maintained in the fasted state by the production of glucose in the liver via glycogen breakdown or the synthesis of new glucose using lactate, pyruvate or other substrates such as amino acids from extrahepatic tissues. The liver also plays a critical role in lipid homeostasis. This includes the incorporation of meal-derived triacylglycerols (TAG) and circulating non-esterified fatty acids into intracellular TAG pools for storage in lipid droplets or secreted as very low-density lipoproteins (VLDL). Moreover, the liver performs peroxisomal and mitochondrial oxidation of fatty acids to produce ATP and ketone bodies which can be transported to extrahepatic tissues for energy when carbohydrates are scarce (Mannaerts et al. 1979). Additionally, de novo synthesized cholesterol in the liver can be esterified for storage in lipid droplets, packaged and secreted as LDLs, or serve as a source for a diverse array of cellular functions including the synthesis of bile acids, steroid hormones and vitamins (see review (Dietschy et al. 1993)). These pathways are regulated by a range of hormones, in particular insulin, which stimulates glycolysis and glycogenesis, de novo fatty acid and lipid synthesis and suppresses gluconeogenesis, glycogenolysis and lipolysis (Rui 2014).

Studying the mechanisms controlling these crucial metabolic functions of the liver, not only in healthy physiology but also in perturbed conditions such as obesity, insulin resistance, non- alcoholic fatty liver disease (NAFLD) or other metabolism-linked pathologies, requires appropriate and robust models that closely resemble normal physiology. The gold standard model for studying hepatocyte biology in vitro is isolated primary human or rodent hepatocytes, which require access to fresh human or animal liver tissue. Despite the optimization of cell culture conditions, primary hepatocytes present the added challenge of having limited ex vivo proliferative capacities and eventually lose their hepatic phenotype (Ramboer et al. 2015). Consequently, many studies have used immortalized hepatocyte cell lines such as hepatocellular carcinoma HepG2 and Huh7 cells. While these cell lines present

104 with liver-like functions including albumin synthesis (Hosoya et al. 2014), and VLDL and triglyceride secretion (Clayton et al. 2005), these neoplasm-derived cells exhibit a phenotype closer to liver cancer than healthy primary tissue. In fact, one study reported that out of 25 hepatocellular carcinoma cell lines, the profile of HepG2 and Huh7 cells are amongst the most highly correlative to primary hepatocellular carcinoma tumors (Chen et al. 2015), re-enforcing the notion that these are ideal cells to study hepatocellular carcinoma biology and not normal hepatocyte biology. At present, non-cancerous hepatocyte cell lines for the study of hepatic metabolism are poorly characterized and are often overlooked as an alternative to primary cells for in vitro hepatocyte research.

The aim of this study was to characterize glucose and lipid metabolism in a range of immortalized hepatocyte cell-lines to identify non-cancerous hepatocyte cells that reflect (i.e. phenocopy) the metabolism of primary mouse hepatocytes. This comprehensive comparison of in vitro hepatocyte models provides an important resource for the field of hepatocyte biology and pathophysiology.

105 METHODS

Cell Culture AML12 and IHH cells were kindly provided by Prof Rob Parton (Institute for Molecular Bioscience, University of Queensland) and A/Prof Carsten Schmitz-Peiffer (Garvan Institute of Medical Research) respectively and cultured in Dulbecco’s Modified Eagle Medium supplemented with F12 (DMEM-F12; Life Tech), containing 11 mM glucose. PH5CH8 and HepG2 cells were kindly provided by A/Prof Susan McLennan (University of Sydney) and were cultured in DMEM (Life Tech), containing 5 mM glucose or 25 mM glucose respectively. All cell culture media were supplemented with 10% fetal calf serum (Hyclone) and 1% penicillin/streptomycin (Sigma-Aldrich). All cell lines were seeded at > 98% viability.

Animals and primary hepatocyte isolation Primary hepatocytes were isolated from 8-12 week old male C57BL/6 mice which were normal (untransduced) controls for in vivo studies of transgene expression following administration of AAV vectors (approved by the University of Sydney Animal Ethics Committee under protocols 2013/5946 and 2017/1253). Hepatocytes which were surplus to need for the gene expression studies were made available immediately following isolation. Mice were not fasted prior to hepatocyte isolations which were performed between 9 and 11 am. Groups of 5 mice were maintained on a standard 12 hour light-dark cycle in individually-ventilated cages (Tecniplast) with autoclaved 1/8-inch corn-cob bedding (Andersons Lab Bedding, Maumee, OH), crinkled paper nesting material and red perspex domes, cardboard tubes, and aspen wood blocks (Able Scientific, Perth, Australia) as enrichment. They had access to standard rodent chow (irradiated rat and mouse diet, Specialty Feeds, Glen Forrest, Australia), and water ad libitum.

Midline laparotomy was performed under anesthesia and the IVC identified. A 22G cannula was introduced distal to the renal bifurcation and secured using a 5-0 silk tie. Retrograde perfusion of the liver was achieved via the IVC cannula. The hepatic portal vein was transected to allow outflow of perfusate. The liver was sequentially perfused with the following solutions at a flow rate of 5 ml/min (administered using a Masterflex L/S 7528-30, Thermo Fisher Scientific, USA); firstly with 25 ml of Hanks’ Balanced Salt Solution (HBSS, Lonza, USA), then with 25 ml of HBSS with 0.5 mM EDTA (Sigma-Aldrich, USA), followed by 25 ml of

HBSS and finally with 25 ml of HBSS plus 5mM CaCl2 (Sigma-Aldrich, USA) and 0.05% of collagenase Type IV (Gibco, USA). All solutions were warmed to 37°C. The gall bladder was removed and discarded. Lobes of the liver were collected taking care to avoid damage and

106 transferred into a Petri dish containing RPMI 1640 medium (Lonza) with 2% FCS (RPMI/FCS2) at 4°C and gently agitated to disperse the hepatocytes. The hepatocyte slurry was transferred to a 50 ml conical tube and washed with RPMI/FCS2 by centrifugation at 50 x g for 3 minutes. The hepatocyte pellet was gently resuspended in 15 ml PBS at room temperature, then 9 ml of isotonic Percoll PLUS (GE Healthcare Life Sciences, Australia) added and mixed. The mixture was centrifuged at 500 x g for 15 minutes at room temperature with no brake. Debris and excess solution were aspirated, and the hepatocyte pellet washed twice as above. Viable hepatocytes were resuspended in RPMI containing 10% FCS and counted on a hemocytometer using Trypan Blue exclusion. Cells were then seeded with DMEM plating media supplemented with F12 and containing 5 mM glucose, 10% fetal calf serum, 1% penicillin/streptomycin (P/S), 400 nM dexamethasone, 100 nM insulin and 1.5 nM epidermal growth factor (Turpin et al. 2011).

Lipid Loading Cells were incubated in DMEM containing 5 mM of glucose supplemented with 1% P/S and 10% FBS with the addition of 0, 150, 300 or 450 µM of oleate or a lipid mixture of palmitate:oleate:linoleate (1:2:1 v/v/v) for 24 h.

Insulin Stimulation Immunoblots: Cells were serum-starved for 2 h in DMEM containing 5 mM glucose supplemented with 1% P/S and 0.2% BSA then replaced with fresh serum-free medium containing 0 and 100 nM of insulin and incubated for 20 min. Glucose production: Cells were serum-starved for 2 h in DMEM containing 5 mM glucose supplemented with 1% P/S and 0.2% BSA then replaced with Krebs buffer (Fazakerley et al.

2015) supplemented with 1 mM CaCl2, 20 mM sodium lactate, 2 mM sodium pyruvate and increasing doses of insulin (0, 0.1, 1 nM) incubated for 2 h.

Fatty Acid and Glucose Metabolism Cells were incubated in low glucose (5.5 mM) DMEM medium consisting of 0.5 mM oleate, 1 mM carnitine, 2% BSA, 0.5 µCi/ml [1-14C]-oleate or 0.2 µCi/ml [U-14C]-glucose (Perkin Elmer) with or without 100 nM insulin and incubated for 4 h. Fatty acid and glucose oxidation

14 was determined by measuring CO2 in half of the culture media by the addition of an equal 14 volume of 1 M perchloric acid and liberated CO2 trapped in 1 N sodium hydroxide.

107 Cells were harvested in PBS to determine [1-14C]-oleate or [U-14C]-glucose incorporation into complex lipids using a Folch extraction of cellular lipids (Folch et al. 1957). 14C activity in the upper aqueous phase was defined as acid soluble metabolites. Secreted TAG into the media was determined using a Folch extraction of the remaining half of the media. Intracellular and extracellular (media) lipids were resuspended in chloroform:methanol (2:1), spiked with glycerol tripalmitin (Sigma) and separated by TLC using hexane:isopropyl ether:acetic acid (60:40:3 v/v/v) as the developing solvent. 14C activity in the TAG bands determined by liquid scintillation counting. Fatty acid uptake was defined as the sum of total lipid synthesis (14C activity in the organic phase following Folch extraction), oxidation, acid soluble metabolites, and extracellular TAG synthesis. Total 14C-glucose incorporation was defined as the sum of 14C activity in aqueous and organic phases following a Folch extraction of cell lysates and oxidation.

Glycogen Synthesis and Quantification Cells were scraped and lysed in 1M KOH and heated to 65°C for 30 min. Lysates were then diluted in saturated Na2SO4 and absolute ethanol followed by centrifugation at 15,000 g for 15 minutes at 4°C. The supernatant was discarded and the pellet was resuspended in water followed by 95% ethanol. Supernatant was discarded once more and the glycogen pellet digested with amyloglucosidase (Sigma A1602) at 37°C in 1 ml of 0.25 M acetate buffer. After overnight incubation, samples were vortexed and a glucose assay performed or the [U-14C]- glucose incorporated into glycogen determined.

Glycogenolysis Cells were seeded in high glucose (25 mM) DMEM medium to promote maximum glycogen storage. After 24 hours, cells were incubated in 0.5 ml of phosphate buffered saline (Astral Scientific) without or with 10 µg/ml glucagon (Sigma Aldrich) for 1 hour and harvested in 300 µl of KOH (Farghali et al. 2008). Glycogenolysis was determined by extracting glycogen from lysates and performing a glucose assay on extracted samples.

Biochemical Measures Cell triacylglycerols and cholesterol were extracted using a Folch extraction (Folch et al. 1957) and quantified using an enzymatic colorimetric method for TAG (GPO-PAP reagent, Roche Diagnostics) or fluorometric method for cholesterol (Amplex Red Cholesterol, Thermo Fisher). Cell protein content was determined using Pierce Micro BCA protein assay (Thermo

108 Fisher Scientific). Media glucose levels and cell glucosyl units from glycogen quantified using the HK glucose assay (Sigma GAHK20).

Cholesterol Synthesis Cells were incubated in medium containing 0.5 mM oleate, 1 mM carnitine, 0.2% BSA, 0.2 µCi/ml [1-14C]-acetate with or without 100 nM insulin and incubated for 2 h. Cell lysates and media samples were exposed to chloroform: methanol (2:1) as described above to extract lipids and resuspended in chloroform: methanol (2:1) spiked with cholesterol (Sigma) and cholesteryl oleate (Sigma). Fractions were separated by TLC in a solvent system of chloroform: methanol: water (60:30:5, v/v/v) to 30% of the plate, followed by diethyl ether: benzene: ethanol: acetic acid (40:50:2:0.2, v/v/v/v) to 60% of the plate, and hexane: diethyl ether: acetic acid (80:20:1.5, v/v/v) to 90% of the plate. 14C activity in the cholesterol and cholesterol ester bands determined by liquid scintillation counting.

Western Blot Analysis Cells were scraped, homogenized in RIPA lysis buffer (65 mM Tris, 150 mM NaCl, 5 mM EDTA, 1% NP-40, 0.5% Na-Deoxycholate, 0.1% SDS, 10% glycerol, 1M DTT, protease and phosphatase inhibitors) lysed using a 29G syringe and centrifuged at 15,000 g for 15 minutes. Equal amounts of solubilized proteins were loaded on 10% SDS-PAGE gels and transferred onto polyvinylidene fluoride membranes (Merck) Membranes were incubated at room temperature for 1 h with blocking buffer (TBS, PHM 4.5, with 0.1% Tween 20 (TBST) and 3% nonfat milk or BSA). Membranes were incubated overnight at 4°C with a specific primary antibody in TBST and 3% BSA. Antibodies used were as follows: pY1162/3-insulin receptor (1:1000, 44804G, Invitrogen), insulin receptor β (1:1000, 3025S, CST), pS473-AKT (1:1000, 9271, CST), AKT (1:1000, 4685S, CST), pS256-FOXO1 (1:1000, 9461, CST), FOXO1 (1:1000, 2880, CST), pS9-GSK3β (1:1000, 9323S, CST), GSK3β (1:1000, 9315S, CST), GS (1:1000, 3893S, CST), GLUT1 (1:1000, kindly gifted by Prof. David James, University of Sydney), GLUT2 (1:1000, 400061, Merck), Mitomix (1:500, ab110413, Mitosciences), PDHK1 (1:1000, 3820S, CST), ATGL (1:1000, 2138S, CST), HSL (1:1000, 4107S, CST), CD36 (1:1000, 14347, CST), CPT1α (1:1000, ab128568, Abcam), DGAT1 (1:1000, ab54037, Abcam), DGAT2 (1:1000, ab59493, Abcam), SLC27A5 (1:1000, HPA007292, Sigma), FASN (1:1000, 3180, CST), ACC1 (1:1000, 3676, CST), HMGCR (1:1000, PA5-37367, Thermo Fisher Scientific), ACAT1 (1:1000, SC-69836, Santa Cruz Biotechnology) GAPDH (1:1000, 2118S, CST). Subsequently, membranes were incubated with secondary HRP-coupled

109 antibodies (CST). Membranes were washed again for 1 h and incubated with enhanced chemiluminescence reagent (Merck) for 1 min prior to visualization using the ChemiDoc System. Data was analyzed using the ImageLab 5.2 version software (Bio-Rad Laboratories, Hercules, USA).

Statistical Analysis All statistical analyses were performed using GraphPad Prism 6.0 for Mac or 7.02 for PC (GraphPad Software Inc., San Diego, CA). Differences among groups were assessed with appropriate statistical tests noted in figure legends. P < 0.05 was considered significant. Data are presented as mean + SEM.

110 RESULTS

Hepatocyte Glucose Storage

The liver is a critical regulator of glucose homeostasis. In particular, liver glycogen is a major contributor to the maintenance of euglycemia during fasting and exercise, and depletion of this pool is associated with the onset of exercise-induced fatigue (Baldwin et al. 1975). Firstly, we measured basal glycogen levels in primary mouse hepatocytes (PMH) and compared these levels to non-carcinoma hepatocyte cells (AML12, IHH, PH5CH8) and the widely used HepG2 cells (Table 1). PH5CH8 and HepG2 had similar levels of glycogen compared to PMH (Figure 1A). Insulin stimulation for 4 h increased glycogen content in all hepatocytes except IHH (P=0.1) and HepG2 cells (Figure 1A). The differences in basal glycogen content did not correlate with glycogen synthase protein levels (Figure 1B; R2=0.009, P=0.5) or protein levels of insulin receptor β (Figure 1C). However, insulin-stimulated increases in glycogen content correlated with the change in Akt phosphorylation (i.e. no change in HepG2) in response to insulin, although this was not the case for phosphorylation of the insulin receptor β, and Akt substrates FOXO1 or GSK3β (Figure 1C).

Hepatocyte Glucose Utilization

Next, we assessed the cellular fate of extracellular 14C-glucose to determine whether differences in glycogen content were explained by variances in intracellular handling of extracellular glucose in the basal and insulin stimulated states. Total 14C-glucose incorporation was defined as the sum of 14C activity in aqueous and organic phases following a Folch 14 extraction of cell lysates and CO2 generated, although this did not account for labeled glucose secreted back into the media. IHH and PH5CH8 cells had the same basal rates of glucose incorporation compared to PMH, whilst AML12 cells had very low basal glucose incorporation rate (P=0.99 by two-way ANOVA followed by Tukey’s Multiple Comparisons test; P<0.0001 by unpaired Student’s t-test vs PH; Figure 1D). Most strikingly and as expected, glucose incorporation in the hepatocellular carcinoma HepG2 cells was 14-fold greater compared to PMH. HepG2 cells were also the only cells to increase glucose incorporation in response to insulin stimulation (Figure 1D). GLUT1 and GLUT2 protein levels did not correlate with glucose incorporation (Figure 1E).

We then examined whether the discrepancies between glucose incorporation and glycogen content was associated with glucose oxidation rates. AML12 and HepG2 cells had lower basal glucose oxidation rates, whilst IHH had higher rates, compared to PMH (Figure 1F). Only

111 PMH responded to insulin stimulation to increase glucose oxidation although AML12 cells showed a strong trend for an increase (P=0.08 by two-way ANOVA followed by Tukey’s Multiple Comparisons test, P<0.0001 Student’s t-test; Figure 1F). The differences in glucose oxidation were inversely associated with pyruvate dehydrogenase kinase 1 (PDHK1) protein levels (Figure 1G; R2=0.42, P<0.0001), which phosphorylates pyruvate dehydrogenase to inhibit its activity and thereby reduce glucose oxidation rates. However, glucose oxidation rates did not associate with protein levels of subunits of the mitochondrial electron transport complexes (Figure 1H).

Hepatocyte Glucose Output

Hepatocytes are the primary cell population that generates glucose from a range of gluconeogenic substrates including pyruvate and lactate. PMH had greater glucose production in basal conditions compared to the other cell lines (Figure 2A) and were the only cell type to show dose-dependent decreases in absolute glucose production with increasing concentrations of insulin (Figure 2A). However, when normalized to basal, all cells exhibited a ~20-30% decrease in glucose production with 1 nM of insulin (Figure 2B), including AML12 cells (P=0.44 by two-way ANOVA followed by Tukey’s Multiple Comparisons test, P<0.001 by one-way ANOVA followed by Tukey’s Multiple Comparisons test for individual cells; data not shown).

During times of fasting, hepatocytes respond to glucagon stimulation to mobilize glucose from glycogen stores. PMH, AML12 and PH5CH8 cells had lower glycogen levels following glucagon stimulation whilst IHH and HepG2 cells did not (Figure 2C). Collectively, these measures of glucose metabolism highlight striking differences between the range of hepatocyte cells assessed, with no single immortalized cell line resembling PMH in basal conditions. However, PH5CH8 cells, followed by AML12 cells, most closely phenocopied PMH in terms of basal and insulin-stimulated changes in glycogen content, glucose incorporation , glucose oxidation and glycogenolysis. In particular, HepG2 cells exhibited classical Warburg metabolism with increased glucose incorporation and no compensatory increase in oxidation and have basal and insulin stimulated glucose metabolism that is different to PMH.

Hepatocyte Fatty Acid Uptake & Storage

The liver is a critical regulator of whole-body lipid homeostasis, in particular fatty acid-based complex lipids (i.e. TAG). Basal stores of intracellular TAGs in IHH and HepG2 cells were

112 similar to PMH, whilst PH5CH8 and AML12 cells TAG stores were 4-fold and 9-fold greater than PMH, respectively (Figure 3A).

Dietary fatty acids contained in TAG and adipose tissue-derived non-esterified fatty acids are the main sources for TAG pools in the liver. In fact, ~60% of hepatic TAG is derived via fatty acid uptake in humans with metabolic disorders such as NAFLD (Donnelly et al. 2005). In this study, we defined fatty acid uptake as the sum of 14C activity in total stored lipids, extracellular

TAG, acid soluble metabolites and CO2. AML12 and HepG2 cells had a similar rate of fatty acid uptake compared to PMH, whilst IHH and PH5CH8 cells had significantly lower rates (Figure 3B). The rate of uptake did not correlate with the expression of key fatty acid transport protein FATP5 (R2=0.02, P=0.3) but were somewhat explained by CD36 levels in all cells except IHH (Figure 3C). Only PMH and HepG2 cells were able to increase fatty acid uptake with acute insulin stimulation (Figure 3B).

Hepatocyte Fatty Acid Utilization

In the liver, fatty acids are esterified and stored as TAG in lipid droplets. PMH had the highest rate of fatty acid esterification as complex lipids compared to the other cells, with IHH cells having relatively very low rates (Figure 3D). The vast majority of extracellular oleate was stored as TAG, specifically ~70% in PMH and ~45% on average for the other cells (Figure 3D). Insulin stimulation increased the rate of extracellular oleate stored as complex lipids in PMH, AML12 and HepG2 cells (Figure 3D). Newly synthesized TAG can also be secreted as very-low density lipoproteins for distribution to extrahepatic tissues. The basal incorporation of extracellular oleate into secreted TAG was much greater in AML12 and HepG2 compared to PMH (Figure 3E). As expected, 4 h insulin stimulation suppressed TAG secretion in AML12 and HepG2 cells, as well as PMH (P>0.99 by two-way ANOVA followed by Tukey’s Multiple Comparisons test, P=0.01 Student’s t-test). However, this effect was not observed in IHH and PH5CH8 cells (Figure 3E). TAG synthesis somewhat correlated with DGAT1 but not DGAT2 protein levels (Figure 3F; DGAT1: R2=0.19, P=0.004). Specifically, high TAG secretion in AML12 cells relative to PMH corresponded with increased expression of DGAT2 and decreased expression of DGAT1 (Figure 3F). While DGAT1 and DGAT2 are primarily responsible for TAG anabolism, the lipolytic enzymes ATGL and HSL are crucial regulators of TAG catabolism (Ong et al. 2011). AML12 cells had greater protein levels of ATGL compared to PMH, whilst all cells had lower levels of HSL protein compared to PMH (Figure

113 3G). These patterns did not correlate with basal TAG levels (Figure 3A) or with the rate of oleate incorporation into intracellular TAG (Figure 3D).

PMH had the greatest rate of extracellular-derived fatty acid oxidation compared to the other cells (Figure 3H). HepG2 cells had the lowest basal fatty acid oxidation rate, which was ~2% of basal fatty acid oxidation rate of PMH, potentially due to the higher rates of glucose incorporation (Figure 1F). These rates did not align with the expression of CPT1a protein levels (Figure 3I). As expected, insulin suppressed fatty acid oxidation in PMH; however, this effect was not observed in the other cell lines (Figure 3H).

We further examined fatty acid metabolism by fatty acid handling of cells by determining the relative distribution of extracellular-derived fatty acids into oxidation, intracellular esterification into total lipids and extracellular TAG pools. Despite having the lowest absolute rates of fatty acid uptake and metabolism (Figure 3B, D, E, H), IHH cells were most similar to PMH in terms of fatty acid handling (Figure 3J). In contrast, AML12 and HepG2 cells predominantly partitioned fatty acids into TAG that was secreted into the media, whereas PH5CH8 cells partitioned fatty acids into intracellular storage (Figure 3J).

We next assessed the response of hepatocytes to varying concentrations of oleate and a physiologically relevant FA mixture. All immortalized cell lines accumulated TAG in a dose- dependent manner significantly more than PMH, with PH5CH8 and AML12 cells increasing TAG levels the greatest (Figure 4A). When the media was supplemented with the fatty acid mixture of 1:2:1 palmitate:oleate:linoleate, representing the predominant fatty acid species in human plasma (Bondia-Pons et al. 2007, Quehenberger et al. 2010), AML12 cells stored almost 4 times as much TAG than other hepatocytes, while TAG levels in IHH cells dropped (Figure 4B) due to the toxicity of the FA mixture (data not shown, visual inspection by microscopy). Interestingly, IHH and HepG2 cells had the greatest fold-change in TAG content with increasing concentrations of oleate after accounting for differences in the basal levels (Figure 4C), whereas PH5CH8 and HepG2 cells showed the greatest fold-change when exposed to FA mixture (Figure 4D). Only PMH had a similar increase in TAG levels when cultured in media supplemented with either oleate or the fatty acid mixture (Figure 4E-I). The differences in the ability of cells to accumulate extracellular fatty acids into intracellular TAG was correlated with the protein levels of DGAT1 and DGAT2 (Figure 3F), but not ATGL and HSL (Figure 3G).

114

Hepatocyte de novo Lipid Synthesis

The liver is a major lipogenic tissue converting excess glucose carbons (and other non-lipid nutrients) into lipids for long-term storage. The basal rate of glucose incorporation into intracellular lipids was lower in AML12, PH5CH8 and HepG2 cells but higher in IHH cells compared to PMH (Figure 5A). Only PMH increased the rate of glucose conversion into lipids in response to insulin stimulation (Figure 5A). Similar patterns were observed when assessing the incorporation of glucose into intracellular TAG (Figure 5A). We also observed cell-specific differences in the rate of the incorporation of glucose into TAG secreted into the media. Specifically, IHH and PH5CH8 cells had greater rates, whilst AML12 and HepG2 cells were not different from PMH (Figure 5B). Interestingly, insulin stimulation did not alter this pattern (Figure 5B), although PH5CH8 cells showed a strong trend for an increase in glucose into secreted TAG (P=0.058). These measures of de novo lipid synthesis did not correlate with the protein levels of the key enzymes, FASN and ACC1 (Figure 5C).

Hepatocyte Cholesterol Metabolism

Cholesterol homeostasis is tightly controlled by the liver both in terms of both synthesis and excretion as bile. AML12, IHH and HepG2 cells had similar basal levels of intracellular cholesterol as PMH, while PH5CH8 had higher levels (Figure 6A). All hepatocytes except for IHH cells had similar levels of cholesterol ester (Figure 6A). Basal rates of cholesterol synthesis from acetate that was incorporated into membranes as free cholesterol was similar for AML12, IHH and PH5CH8 cells (Figure 6B), whilst the rate of newly synthesized cholesterol that was exported into the media was greater in IHH, PH5CH8 and HepG2 cells compared to PMH (Figure 6C). The rate of cholesterol synthesis did not correlate with HMG- CoA reductase protein levels (Figure 6D). PMH incorporated newly synthesized cholesterol into intracellular cholesterol esters at greater rates than the 4 immortalized cell lines (Figure 6B) but the rate of incorporation into cholesterol esters that were exported into the media was similar between PMH and AML12, IHH and HepG2 cells (Figure 6C). Cholesterol ester synthesis is catalyzed by ACAT1 and PMHs had the highest protein levels compared to the cell lines (Figure 6D).

115 DISCUSSION

The liver plays a vital role in nutrient homeostasis and dysregulation of hepatic metabolic pathways are central to many pathologies including NAFLD, HCC and type 2 diabetes. The aim of this study was to identify non-malignant hepatocyte cell lines that most closely recapitulates the metabolic activities of PMH. Here, we present data on the metabolism of glucose, fatty acids and cholesterol, and protein levels of key regulators of these pathways in normal hepatocyte cell lines and compared them to PMH and the commonly used hepatocellular carcinoma HepG2 cells. Overall, no one cell line resembled the glucose, fatty acid and cholesterol metabolic pathways of PMH. However, in general, the normal hepatocyte cell lines more closely phenocopy PMH than the widely used HepG2 cells, specifically, the basal and insulin-stimulated glucose storage, uptake oxidation and production of PH5CH8 and AML12 cells (Table 1 and 2). Insulin-stimulated oleate uptake and esterification into intracellular TAG and suppression of lipid secretion in AML12 and HepG2 cells were very similar to PMH (Table 2). Collectively, these observations highlight striking heterogeneity in the nutrient metabolism in a range of hepatocyte cells that was not accounted for by differences in the protein expression of key regulators of these diverse pathways and that choice of cell line is vital if using these cell lines as a surrogate for PMHs.

HepG2 cells are one of the most prevalent cell lines used to investigate hepatocyte metabolism and broader aspects of hepatocyte biology. Our data suggest that, in most instances, HepG2 cells do not represent a good model for PMH metabolism. Here, we observed striking differences in glucose incorporation rates in HepG2 cells that were 14-fold greater compared to PMH and the other cell lines (Figure 1D), which correlated with GLUT2 protein levels. This is in line with the use of fluorodeoxyglucose in positron emission tomography for the diagnosis of HCC (Haug 2017). HepG2 cells also had the lowest rate of fatty acid oxidation (Figure 2H) and somewhat surprisingly reduced de novo lipogenesis as measured by the conversion of glucose into lipids (Figure 5A), despite having the highest protein levels of ACC1 (Figure 5C), which has been previously reported (Nelson et al. 2017). These differences are likely driven by the unsurprising close resemblance of HepG2 cells to primary hepatocellular carcinoma (HCC) samples (Chen et al. 2015). Specifically, the authors compared over 1000 cancer cell lines, including 25 HCC cell lines, to 200 HCC tumor samples, and HepG2 cells were one of the top HCC cell lines where gene expression highly correlated with that in primary tumors (Chen et al. 2015). More recently, correlations between HepG2, Hep3B and Huh7 cell lines

116 and human liver; hierarchical cluster analysis and principle component analysis all identified dramatic differences in the microsomal proteome (12). Collectively, these results demonstrate that HepG2 cell metabolism is strikingly different to that of non-malignant hepatocyte lines and PMH, and therefore caution is required in the extrapolation of metabolic activity observed in HepG2 cells to normal hepatocyte metabolism; however, they remain an obvious, and appropriate, choice for analyzing HCC biology.

The tight regulation of glucose levels by the liver is crucial in preventing the pathogenic mechanisms underlying chronic diseases such as type 2 diabetes and NAFLD. While multiple glucose-increasing hormones exist, insulin is the only known hormone to lower blood glucose levels and activate glucose-utilizing pathways. The liver is regulated by insulin in a number of cell autonomous actions, such as the direct insulin signaling required for lipogenesis, as well as non-cell autonomous actions, such as the indirect suppression of glucose production by reducing adipose-derived NEFA (Titchenell et al. 2016). These in vitro models are invaluable in teasing apart these indirect and direct effects that are usually interwoven in whole animal studies. In this study, PMH and the 3 normal hepatocyte cells exhibited insulin-independent glucose incorporation as expected. The protein levels of the glucose transporter GLUT2, which is not insulin-sensitive (Fukumoto et al. 1988, Gould et al. 1991), were detectable in all 3 normal hepatic cell lines but PMH had very low levels of GLUT2 protein expression, likely due to downregulation of GLUT2 mRNA which occurs during the first 24 hours of primary hepatocyte culturing (Kim and Ahn 1998). Further, insulin increased glycogen levels in PMH and the 3 normal hepatocyte cell lines, which was associated with increased phosphorylation of GSK3β (Figure 1C). The phosphorylation of GSK3β reduces its kinase activity, leading to increased glycogen synthase activity (Peak et al. 1998).

Insulin plays a critical role in the liver by suppressing gluconeogenesis; an important characteristic when defining insulin resistant hepatic models (Mortimore 1963, Edgerton et al. 2017). In this study, PMH had a ~30% decrease in glucose production with 1 nM insulin treatment similar to that of IHH, PH5CH8 and HepG2 cells. Interestingly, human and mouse hepatocarcinoma cells are reported to have a loss of gluconeogenic capabilities due to decreased PEPCK and G6Pase enzymes as a consequence of abnormal regulation of 11β- HSD1/11β-HSD2 (Ma et al. 2013). It is possible that this discrepancy is due to differences between glucocorticoid- versus insulin-regulated gluconeogenesis, rather than glucose production capacities as a whole (Ma et al. 2013). Overall, PH5CH8 cells, followed by AML12

117 cells, present themselves as the best model to use to model hepatic basal and insulin-stimulated glucose metabolism in the non-disease state. While elevated blood glucose levels are the primary sign of insulin resistance and type 2 diabetes, increased levels of triglycerides have become a well-established hallmark amongst these patients as well, demonstrating the link between glucose and lipid metabolism (Nikkila 1974, Greenfield et al. 1982, Dunn 1990). When glucose is abundant, insulin promotes the conversion of glycolytic substrates into fatty acids in a process known as de novo lipogenesis (DNL) (De Freitas and Depocas 1965). While none of the cell lines exhibited an insulin-stimulated increase in DNL as consistently as PMH, PH5CH8 cells showed a strong trend for increased incorporation of radiolabeled glucose into all lipids, but not TAG specifically, after acute insulin treatment. The availability of acetyl- CoA as a lipogenic precursor appears to explain the differences in DNL rates better than the availability of catalytic DNL enzymes FASN and ACC1 in these cells, likely due to the central role that PDHK1 plays in metabolic flexibility between oxidation and biosynthesis (Strumilo 2005, Zhang et al. 2014). Apart from changes in DNL, uptake of fatty acids in excess of metabolic requirements is a common precursor to obesity, NAFLD and type 2 diabetes. Fatty acid uptake from either dietary fat or adipose tissue lipolysis occurs mainly through CD36 or FATP5 proteins at the cell membrane (Febbraio et al. 1999, Doege et al. 2006). All cell lines had relatively low rates of oleate uptake and only HepG2 cells managed to exhibit insulin- stimulated increases. Variance in intracellular esterification between cell lines was largely reflected by the differences in oleate uptake, and therefore AML12 and HepG2 cells exhibited the closest phenotype to primary cells. Although compensatory roles for triglyceride synthesis by DGAT1 and DGAT2 have previously been shown, their products have different preferential pathways. DGAT1 is primarily involved in the re-esterification of lipolysis-derived DAG to ultimately support VLDL maturation, whereas triglycerides generated by DGAT2 are preferentially used for fatty acid oxidation (Li et al. 2015). Similarly in this study, lower levels of DGAT2 in PMH occur alongside increased fatty acid oxidation, whereas the lower levels of DGAT1 in AML12 cells correlated with greater TAG storage and secretion than other cell lines (Li et al. 2015). Collectively, AML12 cells and HepG2 cells, exemplify normal fatty acid metabolism in hepatocytes and closely resemble that of PMH.

Finally, the liver is the central organ involved in maintaining cholesterol homeostasis. In this study, all hepatocytes had similar intracellular free cholesterol content and synthesis rates from acetate, most likely attributed to the comparable protein expression of the rate-limiting cholesterol synthesis enzyme, HMGCR. Interestingly, the ratio between HMGCR and ACAT1

118 appears to explain the variability in the proportion and synthesis of cholesterol ester within the total cholesterol pool better than the expression of either protein alone, or by HSL protein expression. For example, PMHs had greater ACAT1 to HMGCR protein expression and this is reflected in the ~20-fold increase in the rate of conversion of acetate into cholesterol ester, compared to free cholesterol. Conversely, free cholesterol in IHH cells was ~97% of the total cholesterol pool compared to 70% in PMHs. This is supported by the increased HMGCR to ACAT1 protein expression in these cells. Strikingly, PMHs preferentially secreted cholesterol ester and TAG rather than storing them, which none of the immortalized cell lines phenocopied. A general limitation of correlating metabolic flux with protein expression data is the disregard of the complexity of enzyme regulation in metabolic studies. Although the central dogma of biology is to link protein abundance with enzyme activity and function, there are numerous processes beyond absolute protein concentration that contribute to establishing the activity level of a protein (Liu et al. 2016). These include 1) allosteric modulation, such as the careful balance between activator CGI-58 and inhibitor G0S2 in regulating ATGL lipase activity (Lu et al. 2010), 2) modulation of a protein’s half-life, like the rapid ER-associated proteasomal degradation of HMGCR (Jo and Debose-Boyd 2010), 3) protein translocation, such as the PLIN-mediated HSL translocation from the cytosol to lipid droplets for maximal lipolytic activity (Sztalryd et al. 2003), 4) post-translational modifications, like the 6 different types of modifications affecting CD36 stability and intracellular trafficking (Lauzier et al. 2011, Luiken et al. 2016), and 5) compensatory mechanisms, such as the numerous other flux-controlling points beyond HMGCR in the cholesterol synthesis pathway (Tsai et al. 2012, Sharpe and Brown 2013). Therefore, caution is required when interpreting basal levels of protein abundance of key regulators of metabolic pathways. A potential technical limitation of our protein expression analysis is difference in antibody avidity or affinity. To the best of our knowledge, the antibodies we selected were raised against sequences that are conserved in both human and mouse.

In conclusion, we reveal that no single immortalized hepatocyte cell line broadly resembled isolated PMH in terms of glucose, fatty acid and cholesterol metabolism in the basal or insulin- stimulated states (Table 2 and 3). Our results indicate that PH5CH8 or AML12 cells most closely exemplify basal and insulin-stimulated glucose metabolism of PMH, whereas AML12 cells most closely approximates basal and insulin-stimulated lipid metabolism in PMHs (Table 4). As such, our study can assist in making an appropriate choice of in vitro models for distinct hepatocyte substrate metabolism studies. It remains to be determined how these particular cell

119 lines correlate with other aspects of hepatocyte biology and what are the factors, beyond protein levels of rate limiting enzymes, that underpin the metabolic pleomorphism exhibited by these hepatocytes. Overall, this study has improved our understanding of the similarities and differences between the substrate metabolism of PMH, the commonly used hepatoma cell line HepG2 and non-malignant hepatocyte cell lines, of importance when developing physiologically relevant in vitro models of liver pathologies such as NAFLD. This study comprehensively examined key metabolic pathways of glucose, fatty acid and cholesterol metabolism, providing a valuable resource for selecting hepatic cell lines that are most relevant to mimic a specific metabolic phenotype in liver models.

120 AUTHOR CONTRIBUTIONS

SN, JRK, DJF, AJH conceived and designed the experiments; SN, MP performed the experiments; SN, JRK, DJF, AJH analyzed the data; SN, AJH wrote the manuscript; MP, JRK, DJF, AFS edited the manuscript.

ACKNOWLEDGEMENTS We thank Bianca Varney, Mariam F. Adebayo, Meghna S. Kakani and Nikki L. Raftopulos for technical support.

GRANTS This work was supported by the following funding – AJH is supported by a Robinson Fellowship from the University of Sydney and was supported by a Helen and Robert Ellis Postdoctoral Research Fellowship from the Sydney Medical School Foundation, and funding from the Sydney Medical School and University of Sydney. MP-H holds an Earl Owen Fellowship through the Sydney Medical School Foundation, and also receives support from the Myee Codrington Medical Research Foundation. JRK was supported by an NHMRC Early Career Fellowship (APP1072440). The contents of the published material are solely the responsibility of the authors and do not reflect the views of the NHMRC.

DISCOLUSURES The authors declare there is no conflict of interest in relation to this work.

121 TABLE 1. HEPATOCYTE CHARACTERISTICS

Primary AML12 IHH PH5CH8 HepG2 Hepatocytes (Wu et al. (Schippers et (Noguchi (Knowles and (C57BL/6) 1994) al. 1997) and Aden 1983) Hirohashi 1996) Organism Mus musculus Mus Homo sapiens Homo Homo sapiens musculus sapiens Tissue Liver Liver Liver Liver Liver Disease: None None None None Hepatocellular carcinoma Age: 8-12 weeks 3 months 59 years 58 years 15 years Gender: Male Male Male Male Male

122 TABLE 2. DENSITOMETRIC QUANTITATION OF PROTEIN LEVELS IN HEPATOCYTES

AML12 IHH PH5CH8 HEPG2 Glucose Metabolism GLUT1 0.41 (0.09) # 0.49 (0.17) # 2.44 (0.62) * 0.42 (0.23) # GLUT2 2.08 (0.20) * 1.97 (0.26) # 2.07 (0.37) # 1.79 (0.15) # GS 1.11 (0.38) 3.47 (1.08) 4.10 (1.36) 3.58 (1.74) PDHK1 2.17 (0.23) # 0.72 (0.19) 0.75 (0.56) 2.41 (0.52) * Electron Transport Chain Complex I 0.65 (0.23) 0.48 (0.27) 1.18 (0.35) 0.97 (0.47) Complex II 0.71 (0.10) 0.42 (0.18) 0.57 (0.28) 0.68 (0.34) Complex III 0.72 (0.13) 0.33 (0.09) * 0.33 (0.09) * 0.46 (0.13) * Complex IV 0.49 (0.19) * 0.47 (0.06) * 0.56 (0.08) # 0.69 (0.12) # Complex V 0.57 (0.11) * 0.48 (0.03) * 0.59 (0.05) * 0.53 (0.04) * Lipid Metabolism ACC1 0.72 (0.11) 0.30 (0.07) # 0.36 (0.03) # 1.43 (0.74) ATGL 2.39 (0.31) * 1.02 (0.25) 0.72 (0.06) 0.38 (0.04) # CD36 1.37 (0.35) 2.40 (0.70) 0.83 (0.46) 1.86 (0.51) CPT1a 0.95 (0.12) 0.60 (0.04) # 1.92 (0.25) * 1.20 (0.33) DGAT1 0.29 (0.04) * 0.72 (0.15) 0.75 (0.09) 0.38 (0.01) * DGAT2 9.32 (2.30) # 12.07 (4.22) * 6.45 (1.87) # 0.39 (0.19) FASN 1.07 (0.17) 0.92 (0.39) 0.63 (0.14) 1.17 (0.27) FATP5 0.72 (0.28) 0.83 (0.34) 1.54 (1.05) 3.59 (0.96) * HSL 0.95 (0.29) 0.15 (0.04) * 0.21 (0.16) * 0.19 (0.07) * Cholesterol Metabolism ACAT1 0.51 (0.16) * 0.06 (0.01) * 0.04 (0.03) * 0.13 (0.10) * HMGCR 1.35 (0.21) 1.06 (0.20) 1.00 (0.13) 0.63 (0.22) Data is mean (SEM) fold change from PMH. * P ≤ 0.05 compared to PMH by one-way ANOVA followed by Tukey’s Multiple Comparisons test. # P ≤ 0.05 compared to PMH by unpaired Student’s t-test.

123 TABLE 3. BASAL REPONSES OF HEPATOCYTES

AML12 IHH PH5CH8 HEPG2 Glucose Metabolism Glycogen Content - - ++ ++ Glucose Incorporation - ++ ++ - Glucose Oxidation - - + - Gluconeogenesis - - - - Fatty Acid Metabolism TAG Content - + - ++ Dose TAG (Oleate) - - - - Dose TAG (Mixture) - - - - Oleate Uptake ++ - - + Oleate Oxidation - - - - Oleate Intra- - - - - Esterification Oleate Extra- - - - - Esterification De novo Lipogenesis Glucose into Intra-Lipids - - - - Glucose into Extra- - - - - Lipids Cholesterol Metabolism Cholesterol Content + + - + Cholesterol Ester + - + + Content Acetate into Intra-CE - - - - Acetate into Extra-CE - - - ++ Acetate into Intra-FC ++ ++ + - Acetate into Extra-FC - - - -

The ranking of basal metabolic responses of hepatocytes in comparison to primary hepatocyte. ++ indicates almost exact levels to PMH, + indicates similar levels to PMH, - indicates dissimilar levels compared to PMH. Abbreviations: TAG, triacylglycerol, Intra, intracellular, Extra, extracellular, CE, cholesterol ester, FC, free cholesterol.

124 TABLE 4. HORMONE-STIMULATED REPONSES OF HEPATOCYTES

PMH AML12 IHH PH5CH8 HEPG2 Glucose Metabolism Glycogen Content + ++ - ++ - Glucose Incorporation ++ ++ ++ ++ - Glucose Oxidation ++ - - - - Gluoneogenesis + - + + + Glycogenolysis a ++ ++ - ++ - Fatty Acid Metabolism Oleate Uptake ++ - - - + Oleate Oxidation + - - - - Oleate Intra- ++ ++ - - ++ Esterification Oleate Extra- - ++ - - ++ Esterification De novo Lipogenesis Glucose into Intra- ++ - - - - Lipids Glucose into Extra------Lipids

The ranking of insulin or aglucagon-stimulated metabolic responses of hepatocytes. ++ indicates expected significant response with great fold change, + indicates expected significant response with mild fold change, - indicates no change. Abbreviations: Intra, intracellular, Extra, extracellular

125 REFERENCES: 1. Biochemistry and Functions of the Liver. In: Hepatology Textbook and AtlasSpringer, Berlin, Heidelberg, 2008. 2. Baldwin KM, Fitts RH, Booth FW, Winder WW, and Holloszy JO. Depletion of muscle and liver glycogen during exercise. Protective effect of training. Pflugers Arch 354: 203-212, 1975. 3. Bondia-Pons I, Molto-Puigmarti C, Castellote AI, and Lopez-Sabater MC. Determination of conjugated linoleic acid in human plasma by fast gas chromatography. J Chromatogr A 1157: 422-429, 2007. 4. Chen B, Sirota M, Fan-Minogue H, Hadley D, and Butte AJ. Relating hepatocellular carcinoma tumor samples and cell lines using gene expression data in translational research. BMC Med Genomics 8 Suppl 2: S5, 2015. 5. Clayton RF, Rinaldi A, Kandyba EE, Edward M, Willberg C, Klenerman P, and Patel AH. Liver cell lines for the study of hepatocyte functions and immunological response. Liver Int 25: 389-402, 2005. 6. De Freitas AS, and Depocas F. Fatty Acid and Glyceride Glycerol Synthesis from Glucose during High Rates of Glucose Uptake in the Intact Rat. Can J Biochem 43: 437-450, 1965. 7. Dietschy JM, Turley SD, and Spady DK. Role of liver in the maintenance of cholesterol and low density lipoprotein homeostasis in different animal species, including humans. J Lipid Res 34: 1637-1659, 1993. 8. Doege H, Baillie RA, Ortegon AM, Tsang B, Wu Q, Punreddy S, Hirsch D, Watson N, Gimeno RE, and Stahl A. Targeted deletion of FATP5 reveals multiple functions in liver metabolism: alterations in hepatic lipid homeostasis. Gastroenterology 130: 1245-1258, 2006. 9. Donnelly KL, Smith CI, Schwarzenberg SJ, Jessurun J, Boldt MD, and Parks EJ. Sources of fatty acids stored in liver and secreted via lipoproteins in patients with nonalcoholic fatty liver disease. J Clin Invest 115: 1343-1351, 2005. 10. Dunn FL. Hyperlipidemia in diabetes mellitus. Diabetes Metab Rev 6: 47-61, 1990. 11. Edgerton DS, Kraft G, Smith M, Farmer B, Williams PE, Coate KC, Printz RL, O'Brien RM, and Cherrington AD. Insulin's direct hepatic effect explains the inhibition of glucose production caused by insulin secretion. JCI Insight 2: e91863, 2017.

126 12. Farghali H, Hodis J, Kutinova-Canova N, Potmesil P, Kmonickova E, and Zidek Z. Glucose release as a response to glucagon in rat hepatocyte culture: involvement of NO signaling. Physiol Res 57: 569-575, 2008. 13. Fazakerley DJ, Naghiloo S, Chaudhuri R, Koumanov F, Burchfield JG, Thomas KC, Krycer JR, Prior MJ, Parker BL, Murrow BA, Stockli J, Meoli CC, Holman GD, and James DE. Proteomic Analysis of GLUT4 Storage Vesicles Reveals Tumor Suppressor Candidate 5 (TUSC5) as a Novel Regulator of Insulin Action in Adipocytes. Journal of Biological Chemistry 290: 23528-23542, 2015. 14. Febbraio M, Abumrad NA, Hajjar DP, Sharma K, Cheng W, Pearce SF, and Silverstein RL. A null mutation in murine CD36 reveals an important role in fatty acid and lipoprotein metabolism. J Biol Chem 274: 19055-19062, 1999. 15. Folch J, Lees M, and Sloane Stanley GH. A simple method for the isolation and purification of total lipides from animal tissues. J Biol Chem 226: 497-509, 1957. 16. Fukumoto H, Seino S, Imura H, Seino Y, Eddy RL, Fukushima Y, Byers MG, Shows TB, and Bell GI. Sequence, tissue distribution, and chromosomal localization of mRNA encoding a human glucose transporter-like protein. Proc Natl Acad Sci U S A 85: 5434-5438, 1988. 17. Gould GW, Thomas HM, Jess TJ, and Bell GI. Expression of human glucose transporters in Xenopus oocytes: kinetic characterization and substrate specificities of the erythrocyte, liver, and brain isoforms. Biochemistry 30: 5139-5145, 1991. 18. Greenfield MS, Doberne L, Rosenthal M, Vreman HJ, and Reaven GM. Lipid metabolism in non-insulin-dependent diabetes mellitus: effect of glipizide therapy. Arch Intern Med 142: 1498-1500, 1982. 19. Haug AR. Imaging of primary liver tumors with positron-emission tomography. Q J Nucl Med Mol Imaging 61: 292-300, 2017. 20. Hosoya S, Watanabe M, Ueno M, and Adachi E. Hepatoblastoma-derived cells (HepG2) used to restore polarized hepatocyte morphology by culturing them in 3D collagen gels. The Kitasato Medical Journal 44: 139-147, 2014. 21. Jo Y, and Debose-Boyd RA. Control of cholesterol synthesis through regulated ER- associated degradation of HMG CoA reductase. Crit Rev Biochem Mol Biol 45: 185- 198, 2010. 22. Kim JW, and Ahn YH. CCAAT/enhancer binding protein regulates the promoter activity of the rat GLUT2 glucose transporter gene in liver cells. Biochem J 336 ( Pt 1): 83-90, 1998.

127 23. Knowles BB, and Aden DP. Human hepatoma derived cell line, process for preparation thereof, and uses therefor edited by Office USPaTThe Wistar Institute of Anatomy and Biology, Philadelphia, Pa. , 1983. 24. Lauzier B, Merlen C, Vaillant F, McDuff J, Bouchard B, Beguin PC, Dolinsky VW, Foisy S, Villeneuve LR, Labarthe F, Dyck JR, Allen BG, Charron G, and Des Rosiers C. Post-translational modifications, a key process in CD36 function: lessons from the spontaneously hypertensive rat heart. J Mol Cell Cardiol 51: 99-108, 2011. 25. Li C, Li L, Lian J, Watts R, Nelson R, Goodwin B, and Lehner R. Roles of Acyl- CoA:Diacylglycerol Acyltransferases 1 and 2 in Triacylglycerol Synthesis and Secretion in Primary Hepatocytes. Arterioscler Thromb Vasc Biol 35: 1080-1091, 2015. 26. Liu Y, Beyer A, and Aebersold R. On the Dependency of Cellular Protein Levels on mRNA Abundance. Cell 165: 535-550, 2016. 27. Lu X, Yang X, and Liu J. Differential control of ATGL-mediated lipid droplet degradation by CGI-58 and G0S2. Cell Cycle 9: 2719-2725, 2010. 28. Luiken JJ, Chanda D, Nabben M, Neumann D, and Glatz JF. Post-translational modifications of CD36 (SR-B2): Implications for regulation of myocellular fatty acid uptake. Biochim Biophys Acta 1862: 2253-2258, 2016. 29. Ma R, Zhang W, Tang K, Zhang H, Zhang Y, Li D, Li Y, Xu P, Luo S, Cai W, Ji T, Katirai F, Ye D, and Huang B. Switch of glycolysis to gluconeogenesis by dexamethasone for treatment of hepatocarcinoma. Nat Commun 4: 2508, 2013. 30. Mannaerts GP, Debeer LJ, Thomas J, and De Schepper PJ. Mitochondrial and peroxisomal fatty acid oxidation in liver homogenates and isolated hepatocytes from control and clofibrate-treated rats. J Biol Chem 254: 4585-4595, 1979. 31. Moore MC, Coate KC, Winnick JJ, An Z, and Cherrington AD. Regulation of hepatic glucose uptake and storage in vivo. Adv Nutr 3: 286-294, 2012. 32. Mortimore GE. Effect of insulin on release of glucose and urea by isolated rat liver. Am J Physiol 204: 699-704, 1963. 33. Nelson ME, Lahiri S, Chow JD, Byrne FL, Hargett SR, Breen DS, Olzomer EM, Wu LE, Cooney GJ, Turner N, James DE, Slack-Davis JK, Lackner C, Caldwell SH, and Hoehn KL. Inhibition of hepatic lipogenesis enhances liver tumorigenesis by increasing antioxidant defence and promoting cell survival. Nat Commun 8: 14689, 2017. 34. Nikkila EA. Plasma triglycerides in human diabetes. Proc R Soc Med 67: 662-665, 1974.

128 35. Noguchi M, and Hirohashi S. Cell lines from non-neoplastic liver and hepatocellular carcinoma tissue from a single patient. In Vitro Cell Dev Biol Anim 32: 135-137, 1996. 36. Ong KT, Mashek MT, Bu SY, Greenberg AS, and Mashek DG. Adipose triglyceride lipase is a major hepatic lipase that regulates triacylglycerol turnover and fatty acid signaling and partitioning. Hepatology 53: 116-126, 2011. 37. Peak M, Rochford JJ, Borthwick AC, Yeaman SJ, and Agius L. Signalling pathways involved in the stimulation of glycogen synthesis by insulin in rat hepatocytes. Diabetologia 41: 16-25, 1998. 38. Quehenberger O, Armando AM, Brown AH, Milne SB, Myers DS, Merrill AH, Bandyopadhyay S, Jones KN, Kelly S, Shaner RL, Sullards CM, Wang E, Murphy RC, Barkley RM, Leiker TJ, Raetz CR, Guan Z, Laird GM, Six DA, Russell DW, McDonald JG, Subramaniam S, Fahy E, and Dennis EA. Lipidomics reveals a remarkable diversity of lipids in human plasma. J Lipid Res 51: 3299-3305, 2010. 39. Ramboer E, Vanhaecke T, Rogiers V, and Vinken M. Immortalized Human Hepatic Cell Lines for In Vitro Testing and Research Purposes. Methods Mol Biol 1250: 53-76, 2015. 40. Rui L. Energy metabolism in the liver. Compr Physiol 4: 177-197, 2014. 41. Schippers IJ, Moshage H, Roelofsen H, Muller M, Heymans HS, Ruiters M, and Kuipers F. Immortalized human hepatocytes as a tool for the study of hepatocytic (de- )differentiation. Cell Biol Toxicol 13: 375-386, 1997. 42. Sharabi K, Tavares CD, Rines AK, and Puigserver P. Molecular pathophysiology of hepatic glucose production. Mol Aspects Med 46: 21-33, 2015. 43. Sharpe LJ, and Brown AJ. Controlling cholesterol synthesis beyond 3-hydroxy-3- methylglutaryl-CoA reductase (HMGCR). J Biol Chem 288: 18707-18715, 2013. 44. Strumilo S. Short-term regulation of the mammalian pyruvate dehydrogenase complex. Acta Biochim Pol 52: 759-764, 2005. 45. Sztalryd C, Xu G, Dorward H, Tansey JT, Contreras JA, Kimmel AR, and Londos C. Perilipin A is essential for the translocation of hormone-sensitive lipase during lipolytic activation. J Cell Biol 161: 1093-1103, 2003. 46. Titchenell PM, Quinn WJ, Lu M, Chu Q, Lu W, Li C, Chen H, Monks BR, Chen J, Rabinowitz JD, and Birnbaum MJ. Direct Hepatocyte Insulin Signaling Is Required for Lipogenesis but Is Dispensable for the Suppression of Glucose Production. Cell Metab 23: 1154-1166, 2016.

129 47. Tsai YC, Leichner GS, Pearce MM, Wilson GL, Wojcikiewicz RJ, Roitelman J, and Weissman AM. Differential regulation of HMG-CoA reductase and Insig-1 by enzymes of the ubiquitin-proteasome system. Mol Biol Cell 23: 4484-4494, 2012. 48. Turpin SM, Hoy AJ, Brown RD, Rudaz CG, Honeyman J, Matzaris M, and Watt MJ. Adipose triacylglycerol lipase is a major regulator of hepatic lipid metabolism but not insulin sensitivity in mice. Diabetologia 54: 146-156, 2011. 49. Wu JC, Merlino G, and Fausto N. Establishment and characterization of differentiated, nontransformed hepatocyte cell lines derived from mice transgenic for transforming growth factor alpha. Proc Natl Acad Sci U S A 91: 674-678, 1994. 50. Zhang S, Hulver MW, McMillan RP, Cline MA, and Gilbert ER. The pivotal role of pyruvate dehydrogenase kinases in metabolic flexibility. Nutr Metab (Lond) 11: 10, 2014.

130 FIGURE LEGENDS

Figure 1: Regulation of insulin-stimulated glucose metabolism in hepatocytes. (A) Basal and 100 nM insulin stimulated glycogen content (three independent experiments performed in triplicate except for primary hepatocytes (two independent experiments)) (B) Representative immunoblots of glycogen synthase in hepatocytes cultured in basal media (two independent experiments performed in duplicate). (C) Representative immunoblots of phospho-IR, IR, phospho-Akt, Akt, phospho-FOXO1, FOXO1, phospho-GSK3β and GSK3β in hepatocytes cultured basal media or 100 nM insulin for 30 mins (three independent experiments performed

131 in triplicate). (D) Basal and 100 nM insulin stimulated glucose incorporation into hepatocytes (three independent experiments performed in triplicate except for primary hepatocytes (two independent experiments)). Glucose incorporation was defined as the sum of 14C activity in aqueous and organic phases following a Folch extraction of cell lysates and oxidation (E) Representative immunoblots of GLUT1 and GLUT2 in hepatocytes cultured in basal media (two independent experiments performed in duplicate). (F) Basal and 100 nM insulin stimulated glucose oxidation (three independent experiments performed in triplicate except for primary hepatocytes (two independent experiments)) (G) Representative immunoblots of PDHK1 in hepatocytes cultured in basal media (two independent experiments performed in duplicate). (H) Representative immunoblots of protein subunits in the mitochondrial complexes (complex II-30 kDa, complex III-Core protein 2, complex IV-subunit 1, and complex V-alpha subunit) in hepatocytes cultured in basal media (two independent experiments performed in duplicate). Data are presented as mean ± SEM. A, D, F *P ≤ 0.05 compared to primary hepatocytes; #P ≤ 0.05 compared to basal for each cell type by two-way ANOVA followed by Tukey’s Multiple Comparisons test. PMH: Primary mouse hepatocytes

Figure 2: Hormone-stimulated glucose production in hepatocytes. (A) Absolute basal and insulin stimulated (2 h) glucose production in hepatocytes and (B) expressed as percent from basal. (C) Basal and 10 µg/ml glucagon (1 h) stimulated glycogen content (three independent experiments performed in triplicate). Data are presented as mean ± SEM. A-B *P ≤ 0.05 compared to primary hepatocytes; $P ≤ 0.05 main effect for Insulin by two-way ANOVA followed by Tukey’s Multiple Comparisons test. C #P ≤ 0.05 compared to basal for each cell type by two-way ANOVA followed by Tukey’s Multiple Comparisons test. . PMH: Primary mouse hepatocytes; Gluc: Glucagon

132

133 Figure 3: Fatty acid partitioning in hepatocytes. (A) TAG content of hepatocytes cultured in basal media (three independent experiments performed with six technical replicates each except for primary hepatocytes (two independent experiments)). (B) Basal and 100 nM insulin stimulated oleate uptake into hepatocytes (three independent experiments performed in triplicate except for primary hepatocytes (two independent experiments)). Fatty acid uptake was defined as the sum of total lipid synthesis (14C activity in the organic phase following Folch extraction), oxidation, acid soluble metabolites, and extracellular TAG synthesis. (C) Representative immunoblot of CD36 and FATP5 in hepatocytes cultured in basal media (two independent experiments performed in duplicate). (D) Basal and 100 nM insulin stimulated oleate incorporation into intracellular and (E) secreted lipids (three independent experiments performed in triplicate except for primary hepatocytes (two independent experiments)). (F) Representative immunoblots DGAT1, DGAT2, and (G) ATGL and HSL in hepatocytes cultured in basal media (two independent experiments performed in duplicate). (H) Basal and 100 nM insulin stimulated oleate oxidation (three independent experiments performed in triplicate except for primary hepatocytes (two independent experiments)). (I) Representative immunoblots CPT1α in hepatocytes cultured in basal media (two independent experiments performed in duplicate). (J) Relative basal fatty acid partitioning in hepatocytes where oleate uptake was expressed as 100% and the relative contribution of intracellular other lipids (total lipid synthesis – TAG synthesis) and TAG synthesis, oxidation, and extracellular TAG synthesis presented. Data are presented as mean ± SEM. A-B, D-E, H *P ≤ 0.05 compared to primary hepatocytes; #P ≤ 0.05 compared to basal for each cell type by two-way ANOVA followed by Tukey’s Multiple Comparisons test. PMH: Primary mouse hepatocytes

134

Figure 4: Effect of extracellular oleate and FA mixture on TAG content in hepatocytes. Cells were treated with a concentration range (0-450 µM) of (A) oleate alone or (B) a FA mix (1:2:1 palmitate:oleate:linoleate) for 24 hours then assayed for TAG content expressed as absolute values or (C-D) as fold change from basal (three independent experiments performed in triplicate except for primary hepatocytes (two independent experiments)). (E) primary mouse hepatocytes, (F) AML12, (G) IHH, (H) PH5CH8, and (I) HepG2 cells were treated with a concentration range (0-450 µM) of oleate alone or a FA mix (1:2:1 palmitate:oleate:linoleate) for 24 hours then assayed for TAG content (three independent experiments performed in triplicate except for primary hepatocytes (two independent experiments)). Data are presented as mean ± SEM. A-D *P ≤ 0.05 compared to primary hepatocytes; $P ≤ 0.05 for main effect of FA dose by two-way ANOVA followed by Dunnett's Multiple Comparison test. E-I $P ≤ 0.05 for main effect of FA dose; #P ≤ 0.05 compared to Oleate by two-way ANOVA followed by Dunnett's Multiple Comparison test. PMH: Primary mouse hepatocytes

135

Figure 5: Insulin-stimulated de novo lipogenesis in hepatocytes. (A) Basal and 100 nM insulin stimulated glucose incorporation into intracellular and (B) secreted lipids. (C) Representative immunoblots for FASN and ACC in hepatocytes cultured in basal media (two independent experiments performed in duplicate). Data are presented as mean ± SEM. A-B *P ≤ 0.05 compared to primary hepatocytes; #P ≤ 0.05 compared to basal for each cell type by two-way ANOVA followed by Tukey’s Multiple Comparisons test. PMH: Primary mouse hepatocytes

136

Figure 6: Cholesterol and cholesterol ester synthesis and content in hepatocytes. (A) Basal total cholesterol and cholesterol ester content (three independent experiments performed in triplicate except for primary hepatocytes (two independent experiments)) (B) Absolute rates of acetate incorporation into intracellular and (C) secreted free cholesterol and cholesterol ester (three independent experiment performed in triplicate except for primary hepatocytes (two independent experiments)). (D) Representative immunoblot of HMG-CoA reductase (HMGCR) in hepatocytes cultured in basal media (two independent experiments performed in duplicate). Data are presented as mean ± SEM. *P ≤ 0.05 compared to primary hepatocytes by one-way ANOVA followed by Dunnett’s Multiple Comparisons test. PMH: Primary mouse hepatocytes

137

Chapter 5

Identification and Novel Functional Characterisation of LD-Associated Candidate in the Pathogenesis of NAFL and IR

138 5.1 Introduction

Metabolic diseases rooted in TAG overabundance, such as obesity, metabolic syndrome and NAFL(D), are aptly coined ‘diseases of excessive LDs’. These diseases are often associated with insulin resistance, type 2 diabetes and cardiovascular complications. Although cells such as hepatocytes, adipocytes and myocytes likely use LDs to protect themselves from lipotoxicity, intracellular lipid accumulation can exceed the storage capacity of these cell populations and lead to dysfunction. This delicate balance of protective storage versus mobilisation for energy generation is likely mediated by the hundreds to thousands of proteins bound to the surface of LDs, alongside substrate availability. LD-associated proteins are involved in a large variety of functions, ranging from lipases participating in lipid catabolism and synthesis to proteins involved in DNA damage prevention under stressful situations (Zhang et al. 2017). Therefore, identifying and characterizing novel proteins that localise to LDs and how they respond to diet and hormone stimulation is paramount to ultimately understanding their role in the pathogenesis of metabolic disease and potential for therapeutic targeting.

The preeminent challenge of cell biology is the ability to unveil complex regulation of proteins under distinct and physiologically relevant conditions on a large scale. High throughput mass spectrometry-based proteomics has been useful for the quantitative analysis of protein abundance, protein-protein interactions and post-translational modifications. As discussed in Chapter 3, mass spectrometry was utilized to investigate the influence of acute in vivo insulin stimulation on protein phosphorylation and abundance on LDs in the livers of HFD-induced insulin resistant rats. Apart from HSL, all known participants of lipolysis did not significantly change in abundance at the LD between healthy and insulin-resistant animals. This observation suggests the need to broaden the understanding of key players in LD metabolism and identify new candidates that link LD biology to the pathogenesis of NAFL and IR. One candidate that was identified from curation of the hepatic LD proteome was the ubiquitin binding protein named UBX domain containing 8 (UBXD8), also known as FAF2.

Through systematic filtering, UBXD8 was identified as an ‘insulin resistant’ LD protein that is expressed in rat liver. Though a pool of UBXD8 is ubiquitously found on LDs, it is primarily an ER-associated protein that is known to migrate to LDs upon stimulation with unsaturated FAs (Kim et al. 2013). UBXD8 has three main domains including the UAS domain for FA

139 binding, the UBX domain required for interaction with the segregase p97, and the UBA domain to form polyubiquitin chains (Buchberger 2002). The UBA domain allows UBXD8 to trigger rapid proteasomal degradation of key metabolic proteins such as INSIG-1, which is known to regulate cholesterol concentrations, the major VLDL constituent ApoB-100 and the rate- limiting enzyme in sterol biosynthesis, HMGCR (Lee et al. 2008, Suzuki et al. 2012, Loregger et al. 2017). UBXD8 also inhibits LD turnover by separating ATGL from its coactivator CGI58 and inhibiting its activity in HEK cells (Olzmann et al. 2013). Collectively, these observations suggest that UBXD8 is an important regulator of lipid and sterol metabolism and may be dysregulated in LD-related diseases. The following chapter describes the changes in lipid phenotype and molecular interactions of UBXD8 under distinct hormonal, transcriptional and dietary regulation. Collectively, these analyses attempt to further outline its potential as a key regulator of lipid metabolism and as a novel candidate in the pathogenesis of NAFL(D) and IR.

140 5.2 Methods

5.2.1 Cell Culture 5.2.1.1 Cell Maintenance AML12 mouse hepatocytes kindly provided by Professor Rob Parton (Institute for Molecular Bioscience, University of Sydney) were cultured in a 1:1 mixture of Dulbecco’s modified Eagle’s medium and Ham’s F-12 medium (DMEM-F12) supplemented with GlutaMAXÔ, sodium pyruvate and 2.4 g/L sodium bicarbonate (GibcoÒ, Life Technologies), 10% fetal bovine serum (FBS; In Vitro Technologies) and 1% of penicillin and streptomycin (P/S; GibcoÒ, Life Technologies).

5.2.1.2 Treatment of Cells 5.2.1.2.1 Insulin Treatment AML12 hepatocytes were serum-starved for 2 hours in low glucose DMEM (Gibco, Life Technologies) supplemented with 0.2% bovine serum albumin (BSA; Bovogen) and 1% P/S followed by fresh serum-free media containing 0, 10 or 100 nM of insulin (Sigma) incubated for 30 minutes, as indicated.

5.2.1.2.2 FA Treatment Oleate and linoleate were diluted in absolute ethanol to make 100 mM stock concentration and stored at -30°C, whereas palmitate was resuspended in absolute ethanol to make 100 mM stock concentration and made fresh on the day. Before treatment, FAs are added to serum-containing DMEM for 30 minutes to allow FAs to solubilize in media. AML12 hepatocytes were treated with 300 µM of FA(s) supplementation in low glucose DMEM supplemented with 10% FBS and 1% P/S for 3, 6, 16 or 24 hours, as indicated.

5.2.1.2.3 Peroxisome Proliferator-Activated Receptor (PPAR) a Agonist Treatment The PPARa agonist, WY-14643 (Sigma), was resuspended in DMSO to make a 10 mM stock concentration. AML12 hepatocytes were treated with 20 µM WY-14643 in low glucose DMEM, supplemented with 2% BSA for 6 hours if measuring protein, or 4 hours if measuring any other parameters.

141

5.2.1.3 siRNA-Mediated UBXD8 Knockdown AML12 hepatocytes were seeded at 10-20% confluence in 12-well plates (Corning) at least 24 hours before being transiently transfected with 20 nM ON-TARGETplus Mouse FAF2 siRNA (Dharmacon) using RNase-free water (Dharmacon), OptiMEM I reduced serum medium supplemented with GlutaMAXÔ, HEPES and 2.4 g/L sodium bicarbonate (GibcoÒ, Life Technologies) and LipofectamineÔ RNAiMAX Reagent (Invitrogen) according to manufacturer’s instructions. A stable ON-TARGETplus Non-Targeting Pool siRNA was used as a negative control (Dharmacon). Cells were gently washed twice with warm phosphate- buffered saline (PBS; Astral Scientific) 4 hours after transfection and replaced with low glucose DMEM supplemented with GlutaMAXÔ, sodium pyruvate (GibcoÒ, Life Technologies) and FBS for 48 hours.

5.2.2 Western Blotting 5.2.2.1 Protein Extraction and Quantification Cells were washed twice with ice-cold PBS then lysed with 80 µl of radioimmunoprecipitation assay (RIPA) buffer (65 mM Tris, 150 mM NaCl, 5 mM EDTA 1% (v/v) Nonidet P40, 0.5% (w/v) sodium deoxycholate, 0.1% (w/v) SDS, 10% glycerol, pH 7.4) supplemented with 1 x EDTA-free turbocharged protease and phosphatase inhibitor cocktail (Astral Scientific). Adherent cells were scraped using a cell scraper (Corning) for 20 seconds in all directions and lysed 8 times with a 0.3 ml 29G insulin syringe (BD) on ice. Cell lysates were transferred to 1.5 ml microfuge tubes and spun in a Heraeus Fresco 21 Centrifuge (Thermo Fisher Scientific) at 15 000 × g for 15 minutes at 4°C. The resulting supernatant was kept on ice and transferred to a clear 96-well plate (Greiner) and diluted with water to a factor of 25 for protein quantification. Bicinchoninic Assay (BCA) Protein Kit (PierceÔ, Thermo Fisher Scientific) was used to prepare a standard curve ranging from 0 to 1 mg/ml protein to interpolate the absorbance of the samples. After the reagent was added, the plate was incubated at 37°C for 30 minutes and the absorbance at 560 nm was determined using Magellan software (TECAN Infinite M1000 PRO, Tecan Trading, AG, Switzerland). After protein determination, samples of equal concentration were prepared using 6x Laemmli sample buffer and heated at 50°C for 10 minutes or 75°C for 7 minutes with gentle agitation (Thermomixer 5436, Eppendorf) to denature soluble proteins and allow antibody access to the epitope. Between 30-40 µg of protein was loaded onto a polyacrylamide gel for best resolution.

142 Refer to Section 4.2 for detailed descriptions on SDS-polyacrylamide gel electrophoresis (4.2.3.2), protein transfer to a PVDF membrane (4.2.3.3), imaging with Bio-Rad ChemiDoc MP (4.2.3.5) and stripping and re-probing of membranes (4.2.3.6). For information on antibodies used in this chapter, refer below.

5.2.2.2 Antibody Probing Membranes were blocked in 3% BSA diluted in 1 x Tris buffered saline (TBS; Astral Scientific) containing 0.1% Tween-20 (TBST; Sigma) for 60 minutes at room temperature on a rocker. For primary antibody incubation, membranes were submerged in an antibody solution made of 3% BSA diluted in TBST containing primary antibody and kept overnight at 4°C with gentle agitation (Table 5.1).

Table 5.1 Primary antibodies used in this chapter. Antibody Catalogue # Dilution Species Company anti-ABHD5 ab73551 1:1000 Rabbit Abcam anti-ACAT2 13294S 1:1000 Rabbit CST anti-ACC1 3676S 1:1000 Rabbit CST anti-ATG7 8558T 1:1000 Rabbit CST anti-ATGL 2138S 1:1000 Rabbit CST anti-CPT1a ab128568 1:1000 Mouse Abcam anti-CYC 11940S 1:1000 Rabbit CST Anti-DGAT2 ab59493 1:1000 Goat Abcam anti-ERP72 5033S 1:1000 Rabbit CST anti-ETEA 34945S 1:1000 Rabbit CST anti-FASN 3180S 1:1000 Rabbit CST anti-GAPDH 2118S 1:1000 Rabbit CST anti-G0S2 SC-133423 1:500 Rabbit SCBT anti-HMGCR PA-37367 1:1000 Rabbit Thermo Fisher Scientific anti-LC3A/B 12741T 1:1000 Rabbit CST anti-LDLR ab52818 1:1000 Rabbit Abcam anti-pAKT(Ser473) 4058S 1:1000 Rabbit CST anti-OXPAT PA1-46215 1:1000 Rabbit Thermo Fisher Scientific anti-PPARa C26H12 1:1000 Rabbit CST A list of proteins of interest that were visualized for abundance using western blotting. CST – Cell Signaling Technologies (CST); SCBT – Santa Cruz Biotechnology.

143

After overnight primary antibody incubation, membranes were kept at room temperature for 30 minutes to allow for further binding before washing with fresh TBST for 15 minutes for a total of 1 hour on a shaker. Membranes were then incubated in solution made of 3% BSA in TBST containing a selected secondary antibody for 1 hour on a rocker (Table 5.2). The following table is a list of all the secondary antibodies used in experiments described in this chapter. Finally, membranes were washed again every 15 minutes for at least 1 hour on a shaker.

Table 5.2 Secondary antibodies used in this chapter. Antibody Catalogue # Dilution Species Company Rabbit 7074P2 1:5000 Goat CST Mouse 7076 1:5000 Horse CST Goat SC-2020 1:5000 Donkey SCBT

A list of secondary antibodies that were used to visualize abundance of proteins of interest using western blotting. CST – Cell Signaling Technologies (CST); SCBT – Santa Cruz Biotechnology.

5.2.3 Radiolabel Metabolic Analysis 5.2.3.1 Extracellular Fatty Acid Metabolism AML12 hepatocytes were incubated in 0.5 µCi/ml 1-[14C]-oleate (Perkin Elmer), 0.5 mM cold oleate, 1 mM carnitine and 2% BSA in low glucose DMEM (Gibco) for 4 hours to measure short-term changes in FA metabolism following drug or hormone stimulation. Endpoints 14 measured were 1-[ C]-oleate incorporation into CO2.

5.2.3.1.1 FA Oxidation After 4 hours, media was transferred into 25 ml borosilicate glass liquid scintillation counting vials (Perkin Elmer) containing 1:1 1 M perchloric acid (PCA): media and a 1.5 ml microfuge tube filled with 500 µl of 1 N sodium hydroxide (NaOH). Vials were capped immediately in

14 order to capture [ C]-CO2 after being liberated from acidified media by PCA. Capped vials were incubated on a shaker at room temperature for 2 hours before transferring the NaOH into scintillation vials containing 3 ml of scintillant for measurement of 14C activity by Tri-Carb 2810TR Liquid Scintillation Analyzer (Perkin Elmer).

144 5.2.3.2 Protein Quantification Cells were washed twice with cold PBS and replaced with 300 µl PBS. Adherent cells were scraped using a cell scraper (Corning) for 20 seconds in all directions and lysed 8 times with a 0.3 ml 29G insulin syringe (BD). Cell lysates were transferred to 2 ml microfuge tubes and 25 µl was added to a clear 96-well plate (Greiner) for protein quantification. Bicinchoninic Assay (BCA) Protein Kit (PierceÔ, Thermo Fisher Scientific) was used to prepare a standard curve ranging from 0 to 1 mg/ml protein to interpolate the absorbance of the samples. After the reagent was added, the plate was incubated at 37°C and read at 560 nm after 30 minutes using Magellan software (TECAN Infinite M1000 PRO, Tecan Trading, AG, Switzerland).

5.2.4 Lipid Extraction Adherent AML12 hepatocytes were washed twice with ice-cold PBS and scraped for 20 seconds in 300 – 600 µl of PBS (Astral Scientific) using cell scrapers (Corning). Cells were then lysed with 8 strokes of a 0.3 ml 29G insulin syringe (BD) and transferred to 12 x 75 mm glass tubes (Sigma). Samples were vortexed and protein concentration was measured using the protocol mentioned in section 5.2.3.3. Then, 3 volumes of chloroform: methanol solution in the ratio 2:1 was added to each glass tube and subsequently vortexed every 15 minutes for 1 hour under the fume hood to separate the sample into organic and inorganic phases. After 1 hour, samples were spun at 1000 x g for 10 minutes (Eppendorf). Lipids in the lower phase were then extracted into a new glass tube and concentrated in a Dri-Block sample concentrator (Techne) under a stream of nitrogen gas at 40°C until the chloroform: methanol is evaporated. Lipids in the remaining powdered pellet were resuspended in 10 µl of absolute ethanol and vortexed for 20 seconds. For TAG, concentration was quantified using GPO-PAP reagent (Roche Diagnostics) and absorbance was measured at 492 nm on a Tecan Infinite M1000 Pro. To quantify cholesterol concentrations, the Amplex Red cholesterol kit (Sigma) was used as per manufacturer’s instructions.

5.2.5 Immunofluorescence Microscopy AML12 hepatocytes were seeded at 5% confluence on circular 12 mm No. 1.5 coverslips (Thermo Fisher Scientific) and treated with 300 µM oleate supplemented with 10% FCS and 1% P/S for 24 hours to induce LD formation. Cells were washed with PBS twice, fixed with 4% paraformaldehyde for 20 minutes at room temperature and then washed twice again.

145 If visualizing UBXD8 and LDs, cells were blocked in an antibody diluent mixture containing 0.5% BSA and 0.5% saponin for 1 hour at room temperature, followed by overnight incubation with 1:1000 UBXD8 primary antibody (see Table 5.1) at 4°C. The next day, cells were rinsed three times with PBS and incubated in 5 ug/ml fluorochrome-conjugated secondary Alexa Fluor Far Red 647 antibody (A3273, goat anti-rabbit, Invitrogen) and 1:1000 BODIPY 493/503 (Life Technologies) in PBS for 1 hour in the dark at room temperature. If visualizing LDs only, an antibody diluent mixture containing 0.5% BSA, 0.5% saponin and 1:1000 BODIPY 493/503 in PBS was directly added to the cells for 1 hour at room temperature for LD staining.

The plate was kept in a hand-made humidified chamber using wet paper towels and protected from light. After 1 hour, cells were washed twice with PBS and mounted onto 1.0-1.22 mm glass slides (Livingstone) using mounting media with DAPI (Duolink In Situ Mounting Media with DAPI, Sigma) or without (ProlongÒ Diamond Antifade Mountant). Clear nail polish (Sally Hansen) was added to the edges to seal the coverslips and let dry for 15 minutes. Stained cells were observed under a Zeiss Axio Scope upright microscope and obtained using DAPI, FITC and Cy5 filters under 40X objective and 100X objective with Type-F immersion oil (Olympus). Images were optimized and captured using Zen 2012 (Blue Edition) software (Zeiss) and color-coded using ImageJ.

5.2.6 Subcellular Fractionation 5.2.6.1 Treatment and Homogenization of Hepatocytes AML12 hepatocytes were seeded at 70% confluence into 5 x 100-mm dishes per condition early in the morning and treated with 300 µM oleate for 16 hours later that afternoon, or for 30 minutes the next morning. Cells were washed twice with PBS, scraped in 3 ml PBS for 20 seconds in all directions using cell scrapers (Corning) and combined to make 15 ml per condition. Cells were centrifuged in Eppendorf 5810 (Eppendorf) for 10 mins at 500 × g at 4°C and the resultant pellet was resuspended in five times the cell pellet volume in ice-cold hypotonic lysis medium (HLM) (20 mM Tris-Cl pH 7.4, 1 mM EDTA, 10 mM NaF, supplemented with protease and phosphatase inhibitors (Astral Scientific) using a pipet with a wide opening. Cells were homogenized using eight strokes of a hand-driven Potter-Elvehjem tissue homogenizer on ice and centrifuged again for 10 mins at 1000 × g at 4°C. The supernatant and thick fat layer were collected into a separate tube and protein quantification was performed using the BCA Protein Kit (PierceÔ, Thermo Fisher Scientific).

146

5.2.6.2 Sucrose Gradient Fractionation A density gradient was prepared in 5.1-ml ultracentrifuge tube (Hitachi) by adding 2 ml of cell lysate with approximately 4 mg of protein and mixing it with 1 ml of ice-cold HLM buffer containing 80% sucrose using a pipet tip with a wide opening. Once the density-adjusted cell lysate was thoroughly mixed, 1.5 ml of ice-cold HLM containing 20% sucrose was gently layered over the sample, followed by 1 ml of HLM buffer containing 10% sucrose over the previous layer. Finally, 500 ul of HLM containing 0% sucrose was added to fill the tube to the brim. Tubes were balanced within 0.02g of each other and centrifuged for 2.5 hours at 82,700 × g at 4C in a P55ST2-636 spinout rotor (Hitachi). The rotor was allowed to coast to a stop with a deceleration of “F” (free). Once spun, a pipet with a wide opening was used to carefully transfer 420 ul of the most superficial layer at a time to get 12 individual fractions.

5.2.6.3 Immunoblotting Fractions

Equal volumes were taken from each fraction and prepared for western blot. Samples were denatured and reduced with 6X sample buffer and boiled on a heating block with gentle agitation for 5 mins at 100°C. Fractionated samples were loaded onto a 15-well 12% gel and run and semi-dry transferred according to section 4.2.3.3. Membranes were stained with 0.1% w/v Ponceau S (Sigma) and 5% w/v acetic acid in ddH2O for 5 minutes and de-stained with ddH2O for 15-30 minutes to confirm satisfactory protein abundance at each fraction before incubation with primary antibody. Cytochrome C, ERP72 and ATGL were used as organellar markers to track UBXD8 movement with different treatments.

147 5.2.7 Mass Spectrometry Methodology pertaining to mass spectrometry (5.2.7.2-4) was performed by Dr. Mark Larance and Dr. Dylan Hearney from the Charles Perkins Centre at The University of Sydney.

5.2.7.1 Sample Preparation AML12 cells were seeded at 5% confluence in 6-well plates before transfection with siRNA for 48 hours to knock down UBXD8 (see section 5.2.1.3). Once cells reached 80% confluence, plates were washed with PBS twice at room temperature and replaced with 1 ml of denaturing lysis buffer (4% SDS, 20 mM sodium phosphate, 100 mM NaCl, 10 mM NaF, protease inhibitor cocktail (Astral Scientific)) Extracts were then transferred to Eppendorf tubes and heated at 65°C for 10 mins and sonicated using the QSonica in the following order: 30 seconds ON, 20 seconds OFF for 10 mins at 80% amplitude at room temperature. Lysates were centrifuged at 18,000 x g for 10 mins at > 18°C. Protein quantification was then performed on the supernatant, as described in section 5.2.3.3.

5.2.7.2 Stage-Tipping To separate salts from proteins, samples were desalted using styrenedivinylbenzene-reverse phase sulfonate (SDB-RPS) stage tipping. To prepare disks for solid phase extraction, 200 ul tips were mounted into a 3D-printed holder over a 96-well deep well plate. The following solutions were added, centrifuged at 1000 x g for 3 mins and eluted before drying peptides in a PCR plate for 1 hour: 100% acetonitrile, 30% methanol with 1% TFA, 0.2% TFA, 1% TFA, 0.2% TFA, 99% isopropanol with 1% TFA and 5% ammonium hydroxide with 80% acetonitrile. After drying, peptides were resuspended in 50 mM triethylammonium bicarbonate in water. Disulfide bonds were reduced using 5 mM DTT and incubated on a thermomixer at 1000 RPM for 10 min at 95°C. Samples were then alkylated with chloroacetamide and incubated on a thermomixer in the same manner. Trypsin was added in a ratio of 1:20 to digest samples and samples were incubated for 16 hours at 37°C at 500 RPM. Digestion was halted the next day using a final concentration of 1% TFA.

5.2.7.3 LC-MS/MS Acquisition The Dionex RSLCnano uHPLC (Thermo Fisher Scientific) was used to inject peptides directly onto a 45 cm x 75 µm C18 analytic column with a 10 µm pulled tip. Peptides were resolved over a gradient ranging from 5% to 40% acetonitrile for 15 to 140 mins with a flow rate of 300

148 nL per minute. Electrospray ionization at 2.3 kV was employed to ionize peptides and tandem mass spectrometry analysis was performed used a Q-Exactive HF or HFX mass spectrometer (Thermo Fisher Scientific) using HCD fragmentation.

5.2.7.4 MaxQuant Analysis MaxQuant (version 1.5.7.10) (Tyanova et al. 2016), with the integrated search engine Andromeda (Cox et al. 2011), was used to analyze raw data from the mass spectrometer. A false discovery rate of 1% was employed using a target-decoy based strategy and used a database downloaded on April 18, 2018 containing 42, 325 protein sequence entries. The following parameters were set: mass tolerance 4.5 ppm for precursor ions and 20 ppm for MS/MS mass tolerance, semi-specific N-ragged trypsin for enzyme specificity, variable modifications for deamidation, oxidation and acetylation and fixed modification for carbamidomethyl. Statistical tests were performed using R software (version 3.4.3).

5.2.8 Statistical Analysis All data were statistically analyzed using Graphpad Prism version 7 (GraphPad Software). Unpaired Mann-Whitney t-tests and one- or two-way ANOVA tests were conducted appropriately and described in detail under each figure legend. Post-hoc analyses were performed using Tukey’s and Sidak’s multiple comparisons tests, whereby p < 0.05 was considered significant.

149 5.3 Results

5.3.1 UBXD8 is Abundant in Mouse Liver The protein expression levels of UBXD8 across organs in mouse tissue from 15-week old male and female mice kindly donated by Professor Greg Cooney were determined using immunoblotting including brain, heart, lungs, diaphragm, liver, kidney, quadriceps femoris muscle, gastrocnemius muscle, plasma, subcutaneous adipose, retroperitoneal adipose and epidydimal tissue. UBXD8 protein levels in the liver was the highest among all tissues assessed, followed by kidney and the brain (Fig 5.1). This is consistent with the literature, where the majority of studies on UBXD8 are performed in hepatocyte and renal cell lines such as Huh7, HepG2 and HEK293 cells (Suzuki et al. 2012, Olzmann et al. 2013, Loregger et al. 2017). Strikingly, UBXD8 was largely undetected in other oxidative tissues such as heart and skeletal muscle and was not detected in plasma, indicating that it is not a circulating protein. There were no sex differences in UBXD8 protein levels in tissues, aside from adipose and diaphragm tissue. The tissue distribution of UBXD8 in murine samples are consistent with data obtained in the Human Protein Atlas (Uhlen et al. 2015), suggesting evolutionary conservation of tissue expression profiles.

Figure 5.1 Tissue distribution of UBXD8 protein expression. Raw and quantified immunoblots of UBXD8 protein expression levels across 12 tissues in 15-week old male and female mice, specifically brain, heart, lungs, diaphragm, liver, kidney, quadriceps femoris muscle, gastrocnemius muscle, plasma subcutaneous adipose, retroperitoneal adipose and epidydimal tissue.

150 5.3.2 Upregulation and Translocation of UBXD8 Upon Insulin Stimulation In the results described in Chapter 3, UBXD8 was identified as an insulin-sensitive LD- associated protein. To confirm whether insulin treatment modifies UBXD8 protein expression in hepatocytes, AML12 cells were stimulated with insulin and lysates were immunoblotted. Phosphorylation of AktSer473 increased (p < 0.01) in AML12 hepatocytes treated with 10 nM and 100 nM insulin after 30 minutes (Fig 5.2A), and abundance of UBXD8 total protein levels significantly increased (p = 0.04) with 100 nM of insulin after 6 hours of stimulation (Fig 5.2B), with a main effect of insulin dose (p < 0.01). To determine if protein abundance corresponds with translocation of proteins to LDs, AML12 hepatocytes were cultured for 24 hours in 300 µM of oleate to induce LD formation followed by 30 minutes of stimulation with 100 nM insulin. Insulin stimulation resulted in translocation of UBXD8 to LDs (Fig 5.2C).

Figure 5.2 Quantification and Visualization of UBXD8 in Response to Insulin. (A) Representative blots of pAKT, UBXD8 and GAPDH with 0, 10 and 100 nM of insulin stimulation. UBXD8 and pAKT protein expression was normalized to GAPDH. Data are presented as mean + SEM; *p < 0.05 compared to 0 nM treatment by two-way ANOVA followed by Tukey’s Multiple Comparisons test (3 independent experiments performed in duplicate) (B) Representative blots of UBXD8 and GAPDH protein expression with 0, 10 and 100 nM of insulin stimulation over 30 minutes and 6 hours (3 independent experiments) (C) Merged images representing the translocation of UBXD8 to the lipid droplet after 30 mins of stimulation with 100 nM insulin (red = UBXD8 stained with Alexa Fluor FAR RED 647, green = LDs stained with BODIPY493/503). The white line in the lower right hand image is 100 µm relative to the view.

151 5.3.3 Upregulation and Translocation of UBXD8 Upon Lipid Treatment

Figure 5.3 Quantification of UBXD8 Protein Expression In Response to Oleate (A) Representative immunoblots of UBXD8 and GAPDH protein expression with 3- and 16-hour treatment of 300 µM oleate, linoleate or a mixture of palmitate: oleate: linoleate (P:O:L) in a 1:2:1 ratio (three independent experiments). Quantification of (B) Cytochrome C (mitochondria), ERp72 (ER) and PLIN2 (lipid droplets) protein expression (C) total triglyceride (TAG) levels and (D) total UBXD8 protein levels across 12-fractions obtained by fractionation of sucrose density gradient.

Treatment with FAs is an alternative extracellular stimuli that stimulates TAG synthesis in hepatocytes, as previously described in Chapter 4. In HEK293 cells, UBXD8 was reported to sense and polymerize in response to elevated levels of unsaturated FAs (Kim et al. 2013). To corroborate these results in hepatocytes, AML12 cells were incubated in 300 µM oleate, linoleate or a 1:2:1 mixture of palmitate, oleate and linoleate for 3 hours or overnight followed by protein level determination. All three overnight fatty acid treatments increased total UBXD8 protein levels with overnight but not 3 hour of fatty acid treatment (Fig 5.3A).

In HEK293 cells, treatment with unsaturated FAs leads to the translocation of UBXD8 from the ER to LDs to inhibit ATGL-mediated lipolysis (Kim et al. 2013) . To determine whether this is also evident in hepatocytes, organelles were separated into 12 distinct fractions using sucrose density-gradient ultracentrifugation and identified by standard organellar markers cytochrome C for mitochondria, ERp72 for ER and PLIN2 for lipid droplets. A greater

152 percentage of total cellular PLIN2 protein expression was enriched in the top 3 fractions and was noticeably separated from fractions enriched with ERp72 (fractions 8-9) and cytochrome C (fractions 9-12; Fig 5.3B). TAG levels were then measured to confirm the colocalization of PLIN2 to lipid droplets with varying lipid incubation times. TAG levels were not different between 0 and 30 mins of oleate stimulation, likely due to the changes being below the sensitivity of the biochemical assay, but did increase 2-fold in the top 3 fractions with overnight incubation, consistent with PLIN2 localization (Fig 5.3C). In the control and acute oleate treatment conditions, UBXD8 was primarily enriched in fractions 8-9 along with the ER marker ERp72 (Fig 5.3D). In contrast, overnight treatment with oleate was associated with a larger distribution of UBXD8 protein levels across fractions 1-8 (Fig 5.3D), consistent with intracellular movement away from the ER and towards LDs (Olzmann et al. 2013).

5.3.4 UBXD8 Knockdown Attenuates ATGL Protein Expression When associated with LDs, UBXD8 segregates ATGL from its activator CGI58 and directly binds and inhibits ATGL at the LD (Olzmann et al. 2013); however, loss-of-function of UBXD8 studies assessing ATGL biology have not been reported. Unexpectedly, ATGL protein levels were decreased by almost 50% with UBXD8 KD in AML12 hepatocytes (p = 0.01; Fig 5.4). This suggests that UBXD8 may be involved in stabilizing ATGL total protein levels.

Figure 5.4 ATGL Protein Expression with UBXD8 KD AML12 hepatocytes were seeded and treated with UBXD8 siRNA for 48 hours before protein visualization. Data are representative immunoblots and quantification for UBXD8, ATGL and GAPDH with scrambled and UBXD8 siRNA (eight independent experiments in duplicates) and are presented as mean + SEM; *p < 0.05 compared to scrambled by unpaired student’s t-test.

153 5.3.5 UBXD8 Knockdown Promotes TAG Accumulation in Hepatocytes

Figure 5.5 TAG Content in UBXD8 Knocked Down Hepatocytes AML12 hepatocytes were seeded and transfected with UBXD8 siRNA for 48 hours before incubation with 300 µM of oleate for 24 hours. (A) Lipids from hepatocytes were extracted and TAG was quantified by spectrophotometry. Data are representative of 3 independent experiments done in triplicate and presented as mean + SEM; *p < 0.05 compared to scrambled by unpaired student’s t-test. (B) UBXD8-KD AML12 hepatocytes treated with oleate as indicated were fixed and stained with BODIPY 493/503 (neutral lipids) and subjected to immunofluorescent microscopy analysis.

To confirm whether the reduction in ATGL protein expression affected TAG lipolysis, TAG content was measured in UBXD8 KD AML12 hepatocytes. After 24 hours of treatment with 300 µM of oleate, TAG content was greater by 28% in UBXD8 KD cells compared to scrambled control (p < 0.01; Fig 5.5A), consistent with reduced ATGL protein levels. To confirm these results, lipid-loaded AML12 cells treated with scrambled or UBXD8 siRNA were stained with BODIPY 493/503 to visualize LDs. Consistent with TAG levels (Fig 5.5A), UBXD8 KD resulted in accumulation of LDs compared to control cells (Fig 5.5B). Together, these results indicate that a reduction in UBXD8 protein expression in AML12 hepatocytes is associated with decreased ATGL-mediated TAG lipolysis.

154 5.3.6 UBXD8 is Associated with G0S2 Protein Expression Lipase activity of ATGL is influenced by its interaction with CGI58, which acts as co-activator, and G0S2, which acts as a co-suppressor (Lu et al. 2010) (Fig 5.6A) and so the protein levels of these key regulators of ATGL activity were measured. UBXD8 KD decreased G0S2 protein levels by 75% (p = 0.002) but did not significantly alter abundance of CGI58 or HSL (Fig 5.6B). The concurrent lowering of the two proteins known to inhibit ATGL activity suggests a novel and intimate relationship between UBXD8 and G0S2. In HeLa cells, G0S2 was reportedly bound to ATGL regardless of activity or coactivator binding, and the loss of ATGL led to its degradation (Lu et al. 2010), consistent with results obtained in this study. Additionally, PLIN5 is an LD-associated protein that promotes TAG storage by binding to CGI58 and inhibiting lipolysis. Interestingly, UBXD8-KD increased PLIN5 protein expression (p = 0.03) despite the reduction in other lipolytic inhibitors. PLIN5 sequestration of CGI58 is reported to increase its stability at LDs, and thus the elevated levels of CGI58 following the loss of ATGL may contribute to increased PLIN5 protein expression(Granneman et al. 2011). To determine the role of UBXD8 in lipid turnover beyond lipolysis, the levels of lipogenic proteins were also measured. FASN protein abundance was modestly decreased with UBXD8 KD (p = 0.06; Fig 5.6C), likely associated with the role of UBXD8 in modulating SREBP transcription of specific lipogenic target genes, such as FASN (Lee et al. 2010). Meanwhile, UBXD8 KD did not affect ACC1 or DGAT2 protein expression (Fig 5.6C). These results suggest that UBXD8 regulates TAG turnover exclusively via the lipolytic pathway and that

UBXD8 may utilize G0S2 to facilitate its inhibition of ATGL.

155

Figure 5.6 Lipolytic and Lipogenesis Protein Expression (A) A schematic diagram of the players involved in lipid droplet turnover. Lipolytic regulators are illustrated in green while lipogenic enzymes are illustrated in red. The abundance of proteins related to (B) lipolysis and (C) lipogenesis were quantified. Data is presented as mean + SEM; *p < 0.05 compared to scrambled control by unpaired student’s t-test from eight independent experiments performed in duplicates.

5.3.7 UBXD8 Regulates Mitochondrial Fatty Acid Oxidation FAs released by ATGL-mediated LD degradation can be oxidized by mitochondria to generate useable energy. CPT1a is the central regulator of oxidation as it facilitates the transport FAs across the mitochondrial membranes and into the β-oxidation pathway. In AML12 hepatocytes

156 cultured in basal media, CPT1a protein levels were decreased in UBXD8 KD cells (p < 0.01; Fig 5.7A), consistent with increased TAG content and decreased ATGL and G0S2 protein expression observed in these cells (Fig 5.4-6). Additionally, PPARa is a transcription factor that can be activated by FA ligands and regulates the transcription of genes involved in mitochondrial FA oxidation (McMullen et al. 2014). Likewise, UBXD8 KD caused a decrease in PPARa protein expression (p < 0.01; Fig 5.7A). These results suggest that UBXD8 not only regulates lipolytic proteins, but also plays a role in mediating the expression of oxidative proteins. To determine whether the reduction in CPT1a and PPARa affects FA oxidation, incorporation of radiolabelled oleate into CO2 was measured. As expected, the rate of oxidation was markedly decreased with UBXD8 KD (p < 0.01; Fig 5.7B). Further, treatment with the PPARa agonist WY1463 partially rescued the rate of FA oxidation (Fig 5.7B). These results were consistent with the recovery of downstream targets of oxidation, such as CPT1a, but not upstream regulators such as ATGL (Fig 5.7C). Collectively, these results confirms that UBXD8 inhibits oxidation upstream of CPT1a and is mediated by altered ATGL-PPARa signalling.

157

Figure 5.7 Role of UBXD8 in Fatty Acid Oxidation (A) Representative and quantification of immunoblots CPT1a and PPARa in scrambled or UBXD8 KD AML12 hepatocytes. Data are presented as mean + SEM; p < 0.05 compared to scrambled by unpaired student’s t-test from eight independent experiments performed in duplicate. (B) Basal and WY14643 stimulated oleate oxidation (three independent experiments performed in triplicate) in AML12 hepatocytes treated with scrambled or UBXD8 siRNA. Data are presented as mean + SEM; p < 0.05 compared to scrambled by two-way ANOVA followed by Tukey’s Multiple Comparisons test. (C) Representative immunoblots UBXD8, ATGL and CPT1a in scrambled or UBXD8 KD AML12 hepatocytes treated with the PPARa agonist WY14643 as indicated (two independent experiments performed in duplicate). (D) Quantification of immunoblots ATG7 and LC3 in scrambled or UBXD8 KD AML12 hepatocytes. Data are presented as mean + SEM; p < 0.05 compared to scrambled by unpaired student’s t-test from eight independent experiments performed in duplicate.

158 In mammalian liver, there are two major pathways for LD degradation controlled by PPARa: mitochondrial and peroxisomal (micro) catabolism or lysosomal (macro) lipid autophagy (lipophagy) (Schulze et al. 2017). Key lipophagy-related proteins were quantified in UBXD8 KD cells to determine if UBXD8 influences a broad array of PPARa regulated pathways. Interestingly, both ATG7 and LC3 protein levels were not altered by UBXD8 loss-of-function (Fig 5.7D), despite being PPARa target genes. These results demonstrate that UBXD8’s effects to regulate ATGL proteins levels and TAG lipolysis selectively PPARa-CPT1a signalling and fatty acid oxidation but not PPARa-autophagy pathways.

5.3.8 UBXD8 KD Downregulates Oxidation and Upregulates FA Synthesis Pathways The loss of function of UBXD8 resulted in changes in key proteins involved in lipolysis and FA oxidation (Fig 5.6-7). High throughput mass spectrometry was used to elucidate the broader patterns in the proteome of UBXD8 KD cells. Hepatocytes were treated with siRNA for 48 hours before performing mass spectrometry to identify groups of proteins and general pathways that were affected by UBXD8 KD. In AML12 cells, 133 proteins were differentially expressed with UBXD8 siRNA treatment (Fig 5.8); 67 of which were downregulated (Table 5.3) and 66 of which were upregulated (Table 5.4) compared to control. In particular, UBXD8 KD was associated with a reduction in mitochondrial and peroxisomal oxidation proteins, such as CPT2, ATP synthase, and the peroxisomal targeting receptor, PEX5. However, proteins such as ATGL, PPARa, CPT1a and G0S2 were not enriched sufficiently to be detected in all samples by the mass spectrometer. Further, UBXD8 KD enhanced the expression of key proteins involved in FA synthesis and LD generation. The protein most significantly upregulated was atlastin, an ER-embedded GTPase essential for ensuring structural integrity of ER tubules required for LD budding and expansion (Klemm et al. 2013). Similarly, proteins involved in generating very long unsaturated FAs such as FA desaturase and long-chain FA elongase were also increased with UBXD8 KD. Interestingly, two enzymes in the cholesterol biosynthesis pathway, squalene monooxygenase and sterol reductase, were significantly upregulated with UBXD8 KD, whereas sterol carrier protein, a cholesterol transport protein localized in peroxisomes and mitochondria, was downregulated. Collectively, these results suggest an overarching role for UBXD8 in lipid and cholesterol metabolism and provide further evidence that UBXD8 expression is associated with lipid catabolism rather than anabolism.

159

Figure 5.8 Differentially Expressed Proteins Associated with UBXD8 KD AML12 hepatocytes were treated with scrambled or UBXD8 siRNA for 48 hours and loaded onto a mass spectrometer for identification and quantification of differentially expressed proteins. Data is presented as log fold change of UBXD8 KD proteins compared to scrambled control from 5 independent experiments.

160 Table 5.3. List of downregulated proteins in AML12 hepatocytes with UBXD8 KD

Name Symbol Uniprot ID LogP Log2FC FAS-associated factor 2 Faf2 Q3TDN2 2.50 -2.37 Lysosomal PPT2 Ppt2 E9Q0T0 1.64 -0.78 dCTP pyrophosphatase 1 Dctpp1 Q9QY93 1.67 -0.76 Son of sevenless homolog 1 Sos1 Q62245 1.36 -0.58 Large subunit GTPase 1 homolog Lsg1 Q3UM18 1.40 -0.41 Enhancer of rudimentary homolog Erh P84089 1.31 -0.38 Interferon-induced, double-stranded RNA-activated protein kinase Eif2ak2 Q03963 2.12 -0.38 Zinc finger protein 706 Znf706 Q9D115 1.67 -0.36 Vacuolar protein sorting-associated protein 28 homolog Vps28 Q9D1C8 2.35 -0.31 COX assembly mitochondrial protein 2 homolog Cmc2 Q8K199 1.68 -0.31 Protein S100-A11 S100a11 P50543 1.35 -0.30 Aftiphilin Aftph Q80WT5-2 1.77 -0.30 Bromodomain-containing protein 7 Brd7 O88665 1.33 -0.28 Non-specific serine/threonine protein kinase;Serine/threonine- protein kinase PAK 1 Pak1 G5E884 1.35 -0.27 PHD finger protein 3 Phf3 B2RQG2 2.04 -0.26 ADP-ribose pyrophosphatase, mitochondrial Nudt9 Q8BVU5 1.73 -0.24 Adipocyte plasma membrane-associated protein Apmap Q9D7N9 1.38 -0.24 Non-specific lipid-transfer protein Scp2 P32020 1.33 -0.24 Regulation of nuclear pre-mRNA domain-containing protein 1B Rprd1b A0A0R4J195 1.44 -0.23 Tropomyosin 1, alpha, isoform CRA_k Tpm1 G5E8R2 1.59 -0.22 RalBP1-associated Eps domain-containing protein 1 Reps1 D3Z2E3 1.40 -0.22 Heterogeneous nuclear ribonucleoprotein U-like protein 2 Hnrnpul2 Q00PI9 1.94 -0.20 Protein FAM195B Fam195b Q3UGS4 1.74 -0.20 Prenylcysteine oxidase-like Pcyox1l Q8C7K6 1.50 -0.19 LIM domain-containing protein 1 Limd1 Q9QXD8 1.34 -0.19 Negative elongation factor A Nelfa Q8BG30 1.33 -0.19 LIM domain and actin-binding protein 1 Lima1 Q9ERG0 1.39 -0.19 SWI/SNF-related matrix-associated actin-dependent regulator of chromatin subfamily E member 1 Smarce1 O54941 1.45 -0.18 Thioredoxin domain-containing protein 12 Txndc12 Q9CQU0 1.31 -0.18 Carnitine O-palmitoyltransferase 2, mitochondrial Cpt2 P52825 1.40 -0.18 RNA polymerase-associated protein LEO1 Leo1 Q5XJE5 2.11 -0.18 EF-hand domain-containing protein D2 Efhd2 Q8C845 1.52 -0.17 Trinucleotide repeat-containing gene 6B protein Tnrc6b Q8BKI2 1.46 -0.17 Cadherin-6 Cdh6 P97326 1.34 -0.17 Glutathione synthetase Gss P51855 1.30 -0.16 Peroxisomal targeting signal 1 receptor Pex5 O09012-2 1.33 -0.16 Oxidation resistance protein 1 Oxr1 E9Q0A7 1.93 -0.16 Bromodomain-containing protein 1 Brd1 E9Q412 1.51 -0.15 Leucine-rich repeat protein SHOC-2 Shoc2 O88520 1.50 -0.15 Tyrosine-protein phosphatase non-receptor type 11 Ptpn11 P35235 1.68 -0.15 Lamina-associated polypeptide 2, isoforms alpha/zeta Tmpo Q61033 2.85 -0.15 Interferon regulatory factor 2-binding protein 2 Irf2bp2 E9Q1P8 1.46 -0.14 Leucine-rich repeat-containing protein 47 Lrrc47 E9PV22 2.49 -0.14

161 Glutathione S-transferase P 1 Gstp1 P19157 1.91 -0.14 Diphosphoinositol polyphosphate phosphohydrolase 2 Nudt4 Q8R2U6 1.34 -0.14 28S ribosomal protein S11, mitochondrial Mrps11 Q3U8Y1 1.42 -0.14 Membrane-associated progesterone receptor component 1 Pgrmc1 O55022 1.41 -0.13 Probable RNA polymerase II nuclear localization protein Slc7a6os Q7TPE5 1.31 -0.12 PDZ domain-containing protein 8 Pdzd8 B9EJ80 1.31 -0.12 cAMP-dependent protein kinase type I-alpha regulatory subunit Prkar1a Q9DBC7 1.37 -0.12 55 kDa erythrocyte membrane protein Mpp1 P70290 1.63 -0.12 Heterogeneous nuclear ribonucleoprotein H2 Hnrnph2 P70333 1.54 -0.11 Cleavage and polyadenylation specificity factor subunit 6 Cpsf6 H3BJ30 1.55 -0.11 X-ray repair cross-complementing protein 5 Xrcc5 P27641 1.34 -0.11 S-phase kinase-associated protein 1 Skp1 Q9WTX5 1.99 -0.10 tRNA-splicing ligase RtcB homolog Rtcb Q99LF4 1.68 -0.10 Transmembrane 9 superfamily member 3 Tm9sf3 Q9ET30 1.45 -0.10 Protein disulfide-isomerase P4hb P09103 1.85 -0.09 Protein ERGIC-53 Lman1 Q9D0F3 1.34 -0.09 Nucleosome assembly protein 1-like 1 Nap1l1 Q3TF41 1.43 -0.09 Plexin-B2 Plxnb2 B2RXS4 1.40 -0.08 Serpin B6 Serpinb6 Q60854 1.53 -0.08 Keratin, type I cytoskeletal 19 Krt19 CON__P19001 1.31 -0.08 Ubiquitin carboxyl-terminal hydrolase 47 Usp47 A0A1L1SV73 1.61 -0.08 Peptidyl-prolyl cis-trans isomerase NIMA-interacting 1 Pin1 Q9QUR7 1.41 -0.07 Microtubule-associated protein RP/EB family member 1 Mapre1 Q61166 1.41 -0.06 ATP synthase subunit beta, mitochondrial Atp5b P56480 1.99 -0.05

162 Table 5.4. List of upregulated proteins in AML12 hepatocytes with UBXD8 KD

Name Symbol Uniprot ID LogP Log2FC Vesicle transport protein SEC20 Bnip1 Q6QD59 1.31 0.57 Fatty acid desaturase 3 Fads3 Q9JJE7 1.60 0.46 Squalene monooxygenase Sqle P52019 1.96 0.46 Growth hormone-inducible transmembrane protein Ghitm Q91VC9 2.13 0.45 SAGA-associated factor 29 homolog Ccdc101 Q9DA08 2.35 0.44 Delta(24)-sterol reductase Dhcr24 Q8VCH6 1.91 0.44 Atlastin-2 Atl2 Q6PA06 2.69 0.37 Ubiquitin-like protein 3 Ubl3 Q9Z2M6 2.17 0.35 Elongation of very long chain fatty acids protein 1 Elovl1 Q9JLJ5 1.30 0.34 Prostaglandin G/H synthase 2 Ptgs2 Q05769 1.41 0.34 Protein transport protein Sec61 subunit alpha isoform 1 Sec61a1 P61620 1.98 0.31 Septin-10 Sep10 A0A0R4J233 1.36 0.31 CDGSH iron-sulfur domain-containing protein 2 Cisd2 D3Z3X4 1.45 0.30 Midasin Mdn1 A2ANY6 1.47 0.30 DnaJ homolog subfamily B member 12 Dnajb12 Q8C4C9 2.28 0.29 Renin receptor Atp6ap2 Q9CYN9 1.77 0.28 Integrator complex subunit 4 Ints4 Q8CIM8 1.89 0.27 40S ribosomal protein S17 Rps17 P63276 1.71 0.26 U3 small nucleolar RNA-associated protein 18 homolog Utp18 Q5SSI6 1.77 0.24 U3 small nucleolar ribonucleoprotein protein IMP4 Imp4 Q8VHZ7 1.88 0.22 Cell division cycle protein 23 homolog Cdc23 A0A0R4J1W7 1.48 0.22 Cyclin-dependent kinase 2 Cdk2 P97377-2 1.89 0.22 HEAT repeat-containing 1 Heatr1 G3X9B1 1.44 0.21 Unconventional prefoldin RPB5 interactor Uri1 A0A0U1RPG9 1.32 0.21 H2 subunit A Rnaseh2a Q9CWY8 1.76 0.21 Calcium-binding protein 39 Cab39 Q06138 1.42 0.21 Apoptosis regulator BAX Bax A0A1B0GT81 1.35 0.21 Surfeit locus protein 4 Surf4 Q64310 2.06 0.21 Protein MAK16 homolog Mak16 Q8BGS0-2 1.34 0.20 Exportin-1 Xpo1 Q6P5F9 2.26 0.20 mRNA turnover protein 4 homolog Mrto4 Q9D0I8 1.94 0.20 Ancient ubiquitous protein 1 Aup1 Q3U3K9 1.50 0.20 40S ribosomal protein S3 Rps3 P62908 1.61 0.20 AP-1 complex subunit gamma-1 Ap1g1 Q8CBB7 1.47 0.20 Vacuolar protein sorting-associated protein 33B Vps33b P59016 1.32 0.20 Keratin, type II cytoskeletal 7 Krt7 CON__Q9DCV7 1.45 0.19 Serine/threonine-protein phosphatase 2A 65 kDa regulatory subunit A beta isoform Ppp2r1b Q3TTF6 1.55 0.19 Hsp70-binding protein 1 Hspbp1 Q99P31 1.84 0.19 40S ribosomal protein S28 Rps28 G3UYV7 1.55 0.18 THO complex subunit 2 Thoc2 B1AZI6 1.73 0.18 60S ribosome subunit biogenesis protein NIP7 homolog Nip7 Q9CXK8 1.76 0.17 40S ribosomal protein S18 Gm10260 F6YVP7 1.53 0.17 Melanoma-associated antigen G1 Ndnl2 Q9CPR8 1.43 0.16

163 Alpha-parvin Parva Q3UF75 1.95 0.16 26S proteasome non-ATPase regulatory subunit 6 Psmd6 Q99JI4 1.97 0.15 40S ribosomal protein S4, X isoform Rps4x P62702 2.52 0.15 40S ribosomal protein S2 Rps2 P25444 1.44 0.15 Exportin-2 Cse1l Q9ERK4 1.57 0.15 60S ribosomal protein L12 Rpl12 P35979 2.12 0.14 40S ribosomal protein S5 Rps5 Q91V55 1.59 0.13 Importin subunit beta-1 Kpnb1 P70168 1.70 0.13 mRNA cap guanine-N7 methyltransferase Rnmt Q9D0L8 1.67 0.13 T-complex protein 1 subunit eta Cct7 P80313 1.85 0.12 Retinoblastoma-binding protein 5 Rbbp5 A0A0R4J2B6 1.46 0.12 Myb-binding protein 1A Mybbp1a Q7TPV4 1.42 0.12 Phenylalanine--tRNA ligase alpha subunit Farsa Q8C0C7 1.87 0.12 Nuclear protein localization protein 4 homolog Nploc4 P60670 1.74 0.12 Serine hydroxymethyltransferase Shmt2 Q3TFD0 1.33 0.12 Ribosome biogenesis regulatory protein homolog Rrs1 Q9CYH6 1.50 0.11 Polypeptide N-acetylgalactosaminyltransferase 3 Galnt3 P70419 1.46 0.11 MCG21756, isoform CRA_b Nup205 B9EJ54 1.43 0.10 Prohibitin-2 Phb2 O35129 1.42 0.09 Coatomer subunit beta Copb1 Q9JIF7 1.84 0.09 T-complex protein 1 subunit delta Cct4 P80315 1.45 0.08 SUMO-activating enzyme subunit 1 Sae1 Q9R1T2 1.43 0.06 Dolichyl-diphosphooligosaccharide--protein glycosyltransferase subunit 1 Rpn1 Q91YQ5 1.34 0.04

5.3.9 UBXD8 Co-Localizes with Cholesterol Ester Pools Upon Lipid Treatment To corroborate the mass spectrometry data on the role of UBXD8 in sterol metabolism, cholesterol and cholesteryl ester levels were quantified in subcellular fractions of AML12 hepatocytes. In these cells, the majority of the total cholesterol pool is composed of CE, regardless of lipid availability (Fig 5.9). CE accumulated in two pools located in fractions 5 (~4-6) and 8 (~7-10) in cells cultured in basal media; the latter pool likely representing sterol biogenesis in the ER (Fig 5.9A; organelle markers presented in Fig 5.3B). With overnight oleate stimulation, CE levels increased in fractions 1-6, including the formation of two equal peaks in fractions 5 (~3-6) and 8 (~6-9) (Fig 5.9B). CE levels in Fractions 1-3 is representative of increased cholesterol storage in LDs (Fig 5.3B-C) while fractions 6-9 is enriched for the ER. It is unknown what organelle(s) are enriched in fractions 3-6; however, it is hypothesized to be a Golgi or Golgi-related fraction due to its relative density compared to LDs (more dense) and ER/mitochondria (less dense) (Krahmer et al. 2018). Interestingly, UBXD8 protein expression peaked precisely where CE levels peaked in both basal and lipid-treated cells, suggesting a close relationship between UBXD8 and cholesterol formation, storage or transport.

164

Figure 5.9 UBXD8 Intracellular Movement Across Cholesterol Pools Quantification of (A) total UBXD8 protein expression and cholesterol levels across 12-fractions obtained by fractionation of sucrose density gradient in AML12 hepatocytes in basal media and (B) after overnight incubation with 300 µM of oleate.

5.3.10 UBXD8 is Not Related to Expression of Key Cholesterol-Related Proteins Cholesterol is synthesized de novo by a series of reactions in the mevalonate pathway which is primarily regulated by the ER-bound rate-limiting enzyme 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR) (Ness 2015). Free cholesterol is stored in plasma/organelle membranes or esterified to form CE which stored in LDs. Cholesterol is converted to CE by members of the ACAT family consisting of ACAT1 and ACAT2. While ACAT1 is expressed ubiquitously, ACAT2 is largely expressed in the liver and intestine and aids in lipoprotein secretion (Lee et al. 2000). UBXD8 KD did not alter the protein levels of HMGCR (p = 0.06) or ACAT2 (p = 0.16) in AML12 hepatocytes (Fig 5.10), despite demonstrating a strong downward trend. These results indicate an insignificant role for UBXD8 in regulating cholesterol and CE synthesis. UBXD8 KO studies in mice have shown decreased lipoprotein transport (Imai et al. 2015), therefore the levels of the lipoprotein receptor LDLR were measured. In AML12 hepatocytes, LDLR protein expression was unchanged in the absence of UBXD8. Collectively, these results indicate that UBXD8 associates with pools of CE away from the ER but is not involved in lipoprotein endocytosis. Further, it does not strongly correlate with HMGCR and ACAT2 protein expression, despite mass spectrometry data pointing towards a role in cholesterol metabolism. These results are inconsistent with published reports on UBXD8’s role in HMGCR degradation in Drosophila melanogaster and blunted VLDL secretion in UBXD8 KO mice synthesis (Suzuki et al. 2012, Imai et al. 2015).

165

Figure 5.10 Cholesterol-Related Protein Expression with UBXD8 KD AML12 hepatocytes were seeded and treated with UBXD8 siRNA for 48 hours before protein visualization. Data are representative immunoblots and quantification for HMGCR, ACAT2 and LDLR with scrambled and UBXD8 siRNA (eight independent experiments in duplicates) and are presented as mean + SEM; *p < 0.05 compared to scrambled by unpaired student’s t-test.

166 5.4 Discussion Metabolic disorders such as obesity, NAFL and IR are widespread in Western countries and are becoming increasingly prevalent in developing nations. An excessive amount of intracellular lipid, and subsequent LD biogenesis, is a hallmark feature of these disorders (Khan et al. 2015, Engin 2017). Thus, understanding the precise players involved in LD formation and utilization will provide important insights into the pathophysiology of LD dysregulation commonly observed in these populations. Although impairment of several neutral lipolytic LD proteins, such as ATGL, HSL, CGI58, G0S2 and PLIN proteins, are reported to cause or alleviate NAFL, they are often explored in isolation rather than as interacting and compensatory partners (Brown et al. 2010, Ong et al. 2011, McManaman et al. 2013, Sugaya and Satoh 2017). In the results described in this chapter, UBXD8, a protein known to inhibit ATGL activity at the LD, was further characterized under the lens of lipid storage, oxidation and molecular regulation to position it in the grand scheme of LD physiology. In particular, knockdown of UBXD8 in AML12 hepatocytes revealed an unexpected decrease in lipolysis, causing subsequent TAG accumulation and reduction in FA oxidation. These affects were associated with major changes in the protein expression of LD-associated metabolic players and regulators of mitochondrial fatty acid oxidation.

5.4.1 UBD8 and TAG Homeostasis UBXD8 performs a significant role in controlling intracellular FA availability. In the absence of unsaturated FAs, UBXD8 promotes the proteolytic activation of SREBP-1 which in turn allows SREBP-1 to enter the nucleus and activate the transcription of FA synthesis genes (Lee et al. 2008). Consequently, ATGL remains bound to its lipolytic coactivator CGI58 at the LD and continues to hydrolyse TAG into DAGs and FAs (Olzmann et al. 2013). In the presence of FAs, UBXD8 is reported to polymerize and migrate to LDs to segregate ATGL from its coactivator and subsequently inhibit its activity in HEK293 and SV589 cells (Lee et al. 2010, Kim et al. 2013, Olzmann et al. 2013). It was therefore hypothesized that knockdown of UBXD8 will remove the anti-lipolytic signal and thereby increase ATGL activity. However, results described in this chapter demonstrate that KD of UBXD8 in AML12 hepatocytes significantly lowers ATGL protein expression, resulting in a 2-fold increase in TAG content and associated accumulation of cytosolic LDs. Interestingly, these observations were consistent with studies reporting decreased acetate incorporation into FAs in fibroblasts in the absence of UBXD8 (Lee et al. 2010) and macrovesicular steatosis in mice with liver-specific

167 UBXD8-KO (Imai et al. 2015). However, these reported observations were attributed to impaired TAG synthesis and secretion, rather than a reduced TAG lipolysis. In contrast, UBXD8 KD in AML12 cells was not associated with changes in protein expression of lipid synthesis enzymes, such as DGAT2 and ACC1, but may influence the transcription of the SREBP-1 target gene, FASN. Strikingly, overexpression of S-peptide tagged UBXD8 generated the same lipid phenotype in HEK cells (Olzmann et al. 2013) as UBXD8 KD in AML12 cells. Specifically, UBXD8 overexpression led to a reduction in lipolysis that was strongly linked to decreased association of CGI58 with ATGL and greater LD and TAG levels. These results were amplified in the presence of oleate, whereby total protein levels of UBXD8 were increased, likely due to polymerization, leading to migration to LDs to control ATGL- mediated lipolysis. Despite the inhibition of ATGL by UBXD8, protein expression of ATGL was elevated in the presence of oleate (Lee et al. 2010). It has been demonstrated that oleate stimulation prevents proteasomal degradation of ATGL, regardless of its activity (Olzmann et al. 2013), presumably as a marker for LD generation or a secondary mechanism to limit uncontrolled lipid synthesis when required. An interesting development to this story was the novel finding that UBXD8 protein expression was regulated by insulin stimulation and that acute signalling caused its migration to LDs in hepatocytes, similar to that of oleate treatment. In previous studies, FOXO1 has been shown to directly stimulate ATGL transcription and promote TAG catabolism in adipose tissue, while insulin inhibits this pathway (Zhang et al. 2016). It is therefore conceivable that insulin may stimulate the migration of an endogenous inhibitor of ATGL to further suppress its activity in the presence of lipids. The similarities and differences in lipid phenotypes arising from KD and overexpression experiments across various treatment conditions is likely explained by nuances in the protein expression of molecular interactors of ATGL.

Emerging evidence indicates that different cell types have distinct molecular regulators of lipid metabolism, particularly around ATGL-mediated lipolysis. In adipocytes, the interaction between ATGL and its coactivator CGI58 is inhibited by Plin1 sequestration and stabilization of CGI58 in the fed state (Subramanian et al. 2004). In HEK cells, UBXD8 segregates ATGL from CGI58 to inhibit lipolytic activity. Although plausible, previous studies investigating UBXD8 have largely ignored the role that G0S2 plays in hepatic modulation of TAG catabolism, and thus its association with UBXD8 is completely unknown. The results described in this chapter demonstrated a concomitant decrease in G0S2 protein expression with UBXD8 KD, despite the increase in TAG levels. This may be explained by the ability of ATGL to

168 stabilize and retard the proteasomal degradation of G0S2 post-transcriptionally, regardless of the presence of CGI58 or FA accumulation (Heckmann et al. 2016) (Fig 5.11). In other words, ATGL decreases the ubiquitination of G0S2 protein and thus the reduction in ATGL expression following UBXD8 KD consequently allows for the degradation of G0S2 protein, but not mRNA. Although it is apparent that G0S2 is still a binding partner and regulator of ATGL in hepatocytes, it is unclear how it functions alongside UBXD8. When looking through the lens of insulin signalling in hepatocytes, studies show that insulin inhibits FOXO1-mediated promotion of ATGL activity and suppression of G0S2 activity, thereby attenuating TAG lipolysis (Zhang et al. 2016). Similarly, UBXD8 protein levels were increased and UBXD8 translocated to LDs upon insulin stimulation, presumably to inhibit ATGL activity. With two potent inhibitors of ATGL activity regulated by similar hormonal stimulation, it is likely that UBXD8 and G0S2 associate to regulate lipolysis via ATGL.

Figure 5.11 Molecular Interactors of UBXD8 and Associated Lipid Phenotype ATGL is coactivated by CGI58 to promote lipolysis and inactivated by G0S2 to promote TAG storage. UBXD8 is known to dissociate ATGL from CGI58 to prevent lipolysis, whereas the role of G0S2 is unknown. It is proposed that UBXD8 promotes inhibition of lipolysis through the formation of a G0S2 binding complex with ATGL.

5.4.2 UBXD8 and Oxidation In addition to its role as a TAG hydrolase, ATGL was reported to selectively partition FA products into b-oxidation pathways. In the liver, gain and loss of function studies on ATGL modulate production of both CO2 and acid soluble oxidative intermediates but do not alter the rate of VLDL synthesis or secretion (Reid et al. 2008, Ong et al. 2011, Wu et al. 2011). Although mechanisms that discern FA transport to mitochondria over lipoproteins are currently unknown, newly hydrolysed FAs can act as ligands for transcriptional regulation of oxidative

169 genes via PPARa activation, further promoting energy production (Ong et al. 2014). PPARa expression is elevated in tissues with high oxidative capacities, such as muscle and liver and its targets genes are involved in mitochondrial and peroxisomal oxidation, FA transport, and FA utilization for gluconeogenic pathways (Burri et al. 2010). In particular, expression levels of PPARa and downstream target genes such as CPT1a are attenuated in the absence of ATGL (Ong et al. 2011). This was consistent in AML12 hepatocytes in the absence of UBXD8, whereby a reduction in ATGL protein expression led to decreased PPARa and CPT1a protein levels. These results were associated with reduced rates of b-oxidation in UBXD8 KD cells and could only be partially rescued with PPARa agonism. Thus, UBXD8 expression modulates FA availability by regulating ATGL activity and these FAs control b-oxidation by acting as 1) ligands for PPARa activation and transcription of key oxidative genes such as CPT1a and as 2) direct substrates for the citric acid cycle in b-oxidation (Chakravarthy et al. 2009).

Apart from ATGL, little is known about the independent role that other lipolytic regulators play in modulating mitochondrial oxidation in hepatocytes. Due to their relationship with ATGL, expression levels of both regulators have been associated with changes in oxidation (Liu et al. 2009, Brown et al. 2010, Zhang et al. 2014). For example, ASO-mediated CGI58 KD in mice lowered β-hydroxybutyrate levels, indicating less hepatic oxidation (Brown et al. 2010). Interestingly, CGI58 plays an additional role in the selective autophagic degradation of mitochondria, otherwise known as mitophagy (Zhang et al. 2014). Specifically, CGI58 overexpression promoted mitochondrial fission and translocation of key proteins involved in mitophagy to the mitochondria, suggesting an independent function in bioenergetics beyond ATGL regulation (Zhang et al. 2014). While few studies have reported mitophagy in models of hepatic TAG accumulation (Liu et al. 2009), the expression levels of CGI58 as well as other autophagy markers such as ATG7 and LC3 were not altered with UBXD8 KD. In contrast, G0S2 expression markedly decreased by 75% in UBXD8 KD cells in comparison to the almost 50% reduction in ATGL levels, suggesting additional functions beyond ATGL inhibition. While overexpression of G0S2 caused decreased oxidation rates in primary mouse hepatocytes (Kim et al. 2013, Zhang et al. 2014), studies have reported utilization of both PPARa- dependent and independent pathways (Zhang et al. 2014, El-Assaad et al. 2015). In particular, ablation of G0S2 in primary mouse hepatocytes led to increased expression of some PPARa target genes such as CPT1a, but caused no alterations in the expression of PPARa or other PPARa-sensitive genes like ACOT1 (Zhang et al. 2014, El-Assaad et al. 2015). These results

170 suggest an uncoupling of G0S2 from PPARa-regulated FA oxidation. Interestingly, G0S2 has been identified as a mitochondrial regulator of hypoxia, whereby it can directly bind to mitochondrial proteins such as BCL2 and F0/F1 ATPase and promote oxidative phosphorylation and ATP production (Welch et al. 2009, Kioka et al. 2014). Through the dual and antagonistic role of G0S2 in mediating oxidation, it is conceivable that G0S2 can regulate oxidation independently, and that the influence of UBXD8 on G0S2 protein expression acts as an additional regulator of oxidation (Fig 5.12). This model provides a third alternative to the role that UBXD8 plays in modulating FA oxidation in hepatocytes.

Figure 5.12 Model of UBXD8 in FA Oxidation UBXD8 binds to ATGL and inhibits its ability to hydrolyse TAG. Released FAs can act as ligands for PPARa activation in the nucleus to promote the transcription of oxidative genes or as substrates for oxidation in the mitochondria. Binding of UBXD8 to ATGL may have an additional role of retarding its proteasomal degradation in hepatocytes. UBXD8 is proposed to bind to G0S2 to regulate inhibition of ATGL and directly modulate FA oxidation under particular conditions.

In the fed state, insulin action may also add a secondary layer of regulation to FA oxidation in the liver. In AML12 hepatocytes, insulin stimulation elevates UBXD8 protein expression. Although its role in insulin-suppression of FA oxidation was not assessed, it is plausible that UBXD8 aids in the regulation of this pathway. For example, expression of UBXD8 and G0S2 and hepatic stimulation with insulin have all been independently associated with the inhibition of lipolysis (Yang et al. 2010, Olzmann et al. 2013). Further, G0S2 mRNA was increased in a time-dependent manner when stimulated with insulin in hepatocytes (Zhang et al. 2016), which

171 is similar to UBXD8 patterns described in this chapter. Strikingly, overexpression of G0S2 at LDs decreased FA oxidation (Wang et al. 2013), but promotes oxidative phosphorylation in hypoxic environments, suggesting G0S2’s opposing roles in modulating oxidation (Welch et al. 2009, Kioka et al. 2014). Taken together, these results strongly point towards a working relationship between UBXD8 and G0S2 in basal, lipid and insulin-stimulated conditions. Collectively, insulin action on lipid metabolism may employ UBXD8 to regulate FA oxidation in hepatocytes using key molecular partners. Future experiments should focus on identifying the mechanism by which UBXD8 affects FA oxidation. Measuring oxidation rates of cells with UBXD8 KD in the presence and absence of FAs can help determine if UBXD8 directly inhibits FA oxidation at the mitochondria or indirectly via ATGL-mediated lipolysis. Validation of this proposed model needs to be confirmed in order to pinpoint the interactors that are key to UBXD8’s role in LD physiology.

172

Chapter 6

General Discussion

173 6.1 Overview The increasing prevalence of NAFL(D) is an emerging health concern due to its strong association with metabolic diseases such as obesity, type 2 diabetes, metabolic syndrome and cardiovascular disease, all of which are clinical and economic burdens to health care systems (Younossi et al. 2016). NAFL(D) pathophysiology is localized intracellularly, where hepatic lipid deposition is characterised by cytosolic LD accumulation (Green and Hodson 2014). The molecular networks involved in regulating intracellular metabolism are connected to essentially every other process, such as cell cycle, detoxification, and redox balance, and as a result, perturbations in metabolic signalling have widespread consequences. One of the most pervasive impairments of NAFL(D) is the dysregulation of insulin action, which has profound effects on whole-body energy storage, distribution and utilisation (Rui 2014). Through its activation and secretion from the pancreas, insulin regulates protein abundance, activity, translocation and interactions at target tissues using post-translation modifications (PTMs) (Virkamaki et al. 1999, Luiken et al. 2002, Mounier and Posner 2006). In NAFL(D), diminished insulin sensitivity is known to be associated with an accumulation of intracellular lipids stored in LDs and hence exploring the relationship between the two can be elucidated using proteomic and PTM analysis (Zhang et al. 2013). While high-throughput studies investigating hepatic steatosis exist, few have explored this relationship at the LD level. In reality, LDs are metabolic hubs that connect cellular signalling with energy substrates via the dynamic network of PTM-sensitive proteins associated with the LD (Arrese et al. 2014, Krahmer et al. 2018). Explaining how the proteomic circuitry goes awry in close proximity to large pools of TAG and neutral lipid stores is important to extend the understanding of the pathogenesis of NAFL and IR.

6.1.1 Cooperative Regulation of Insulin Signalling: A PTM Story There are currently more than 450 different types of PTMs that can occur in complex organisms, with ubiquitination and phosphorylation being two of the most common (UniProt Consortium 2017). PTMs are advantageous in initiating critical processes such as hormone signalling because they provide a relatively rapid and reversible manner of regulating pathways (seconds) over other types of molecular controls such as gene expression and protein translation (minutes to hours) (Venne et al. 2014). To date, protein phosphorylation is the most studied PTM in NAFL and IR due to the pivotal role that Akt/PKB phosphorylation plays in insulin signalling and energy metabolism. Once insulin binds to its receptor at the cell surface

174 and triggers a signalling cascade, Akt is phosphorylated at Thr308 by PDK1 and Ser473 by mTORC-2 for maximal activation (Risso et al. 2015). Accordingly, this stimulates the phosphorylation of intermediary signalling proteins such as FOXO1 and GSK3b as well as other downstream targets (Dong et al. 2008). However, in reality, there is considerable crosstalk between different types of PTMs in the regulation of metabolic flux. In fact, 37% of all modified proteins in any given cell are regulated by at least two different PTMs and are highly conserved across mouse and human models (Minguez et al. 2012). The complex interplay between phosphorylation and ubiquitination in insulin signalling is an added cooperative layer of regulation, and can occur at multiple levels (Venne et al. 2014). For example, phosphorylated Akt is tagged for degradation following Lys48 ubiquitination, which is a known signal for proteasomal targeting of proteins (Suizu et al. 2009, Bae et al. 2012). Another example is Lys63-linked ubiquitination of proteins. In particular, ubiquitination of Lys63 on Akt by the E3 ligase TRAF6, stimulates Akt phosphorylation and membrane recruitment leading to activation of downstream insulin signalling (Yang et al. 2009, Cheng et al. 2013). Further, a lack of Akt phosphorylation at Ser473 and Thr308 promotes its degradation by the ubiquitin-proteasome system (Wei et al. 2018). Thus, ubiquitination of Akt can regulate, and be regulated, by the phosphorylation of Akt, demonstrating a clear example of PTM coupling in insulin signalling.

The complex interplay between ubiquitination and phosphorylation is, however, not exclusive to Akt. One of the main findings presented in Chapter 2 was that almost all enzymes responsible for glycolysis/gluconeogenesis and related pathways are ubiquitinated, thereby indicating the importance of ubiquitin in regulating this fundamental metabolic process. Strikingly, all 15 of these enzymes have been reported to be phosphorylated and even acetylated based on proteomic and in vitro validation studies (Lundby et al. 2012, Lundby et al. 2012, Wagner et al. 2012). For example, phosphofructokinase (PFKL), the rate-limiting bifunctional enzyme of glycolysis/gluconeogenesis, is phospho-activated in response to insulin or adrenergic stimulation and phosphorylates fructose-6-phosphate to generate fructose-1,6-bisphosphate (Ausina et al. 2018). Similar to Akt, unphosphorylated PFKL is sensed by an unknown E3 ligase and is subsequently degraded by the ubiquitin proteasomal pathway (Swaney et al. 2013). Interestingly, data presented in Chapter 2 indicate that PFKL is inappropriately ubiquitinated in response to insulin in the HFD-fed rats but not the chow-fed rats (Fig 6.1). It is possible that diminished insulin signalling in the obesity-induced IR animals cannot

175 stimulate phosphorylation of PFKL and is therefore degraded, consistent with recent data in cardiomyocytes obtained from HFD-fed animals (Bockus et al. 2017).

Figure 6.1 Proposed Model of Ubiquitin Dysregulation of PFKL A conceptual model based on ubiquitomic data obtained from basal and insulin-stimulated chow and HF-fed rats. In chow-fed rats, insulin facilitates the phosphorylation of PFKL in order to stimulate glycolysis. Ubiquitin targets unphosphorylated PFKL for proteasomal degradation. In HF-fed animals, ubiquitin targets phosphorylated PFKL for unknown reasons and prevents propose glucose utilisation.

Many proteins involved in insulin-stimulated lipid metabolism were both ubiquitinated at the whole liver level (see Chapter 2) and phosphorylated at the LD (see Chapter 3) in response to acute insulin treatment or diet in the liver, such as Plin2, ACBD5, ACSL4 and Decr1. Due to its more defined role in regulating LD biology, this section will focus on Plin2. Plin2 is a constitutive LD protein thought to promote LD expansion by stimulating lipid and sterol synthesis proteins such as MGAT, DGAT and ACAT1 while decreasing the expression of lipases such as HSL and ATGL at the LD in muscle and fibroblasts (McIntosh et al. 2012, MacPherson et al. 2013). In the absence of Plin2, a 60% reduction in TAG content was observed in MEF cells, whereas TAG, CE and PC levels increased in Plin2-overexpressing cells, consistent with LD composition (McIntosh et al. 2012, Tsai et al. 2017). Plin2’s role in promoting LD expansion is likely regulated by multiple independent yet related mechanisms. In particular, two of the three PLIN2 phosphosites detected in Chapter 3 were identified as ‘insulin-resistant’ after systematic filtering. While PLIN2 was dephosphorylated on Ser254 with insulin stimulation in both chow and HF-fed rats, Ser253 and Ser417 were dephosphorylated with insulin in the HFD fed group only. The functional consequence of Plin2

176 phosphorylation, or any other PTM, has largely been unexplored, despite the identification of numerous PTM sites (Hornbeck 2015). Emerging evidence indicates that pan-phosphorylation of Plin2 reduces its own expression and stimulates ATGL-mediated lipolysis in NIH3T3 cells (Kaushik and Cuervo 2016), suggesting a positive relationship between Plin2 phosphorylation and TAG hydrolysis. Therefore, it is conceivable that Ser254 phosphorylation of Plin2 promotes lipolysis in the absence of insulin, when energy is poor. Although no specific phosphosites have been characterised yet, AMP-activated protein kinase (AMPK) is a known upstream kinase of Plin2 and is primarily activated by a nutrient deficit status, thereby promoting the release of FAs from TAG stores and shunting them towards mitochondrial oxidation for energy production (Long and Zierath 2006). Therefore, the dephosphorylation of Ser254 observed with insulin stimulation may act as a potential insulin-responsive regulatory mechanism to inhibit lipolysis in energy-rich conditions.

The tight regulation of Plin2 phosphorylation is facilitated by chaperone-mediated autophagy, whereby Plin2 is transported by chaperone complexes and engulfed by lysosomes in AMPK- active conditions (Nedelsky et al. 2008, Kaushik and Cuervo 2016). Interestingly, in results presented in Chapter 2 of this thesis, Plin2 was ubiquitinated in chow-fed rats under basal conditions (i.e. not ubiquitinated following insulin stimulation) which is identical to the Ser254 phosphorylation pattern of Plin2 (Fig 6.2). Although chaperone-mediated autophagy and the ubiquitin-proteasome system are two uniquely discrete processes, they possess strikingly similar signalling cascades that often have parallel activation (Nedelsky et al. 2008). For example, the deletion of key autophagy genes such as Atg5 and 7 in mice lead to accumulation of ubiquitinated proteins despite normal proteasome function, suggesting that the lysosome feeds on non-ubiquitinated and ubiquitinated proteins (Komatsu et al. 2006). In other words, ubiquitination of Plin2 may occur prior to chaperone-mediated autophagy. This makes sense because the ubiquitin-proteasome system is traditionally thought to be the primary route for degradation of proteins with short half-lives under acute chronic starvation, such as Plin2 (Ciechanover et al. 2000, Vabulas and Hartl 2005). In light of this, it is conceivable that it is ubiquitin that directly targets phosphorylated Plin2 for degradation in a similar manner to ubiquitin-targeting of phosphorylated Akt, as described above.

Although Ser254 phosphorylation of Plin2 remains insulin-responsive with HFD-feeding, a number of other dysregulated PTMs on Plin2 occur in this HFD-induced IR condition. Namely, HFD-feeding increases the phosphorylation of Ser253 and Ser417 and this is associated with

177 increased protein abundance at the LD. If phosphorylation of Ser254 is hypothesized to signal its degradation, it is possible that phosphorylation of Ser253 and Ser417 may prevent its degradation, despite the likely continued ubiquitination of phosphorylated Ser254. Consequently, increased association of Plin2 to LDs prevents lipase-mediated TAG breakdown and thus promotes the development of steatotic liver in a HF environment. Inversely, Plin2-/- mice fed a western diet had a marked reduction in TAG and cholesterol levels in the liver and prevented the development of obesity and IR (Libby et al. 2016). Further, its deletion was associated with suppression of DNL and cholesterol synthesis gene and protein expression. Therefore, elevated Plin2 expression with HFD-feeding is consistent with the steatotic pathology observed. Furthermore, Plin2 remained ubiquitinated in the HF-fed insulin- stimulated group, despite a complete lack of phosphorylation. This could present as an alternative mode of regulating Plin2 biology. Alternatively, ubiquitin may also be able to target Plin2 without phosphorylation, as occurs with the majority of proteins, and the conditions in which this is required could be dysregulated with impaired insulin action (Nguyen et al. 2013). As such, inappropriate degradation of Plin2 following insulin stimulation in a HF environment (i.e. IR) would be predicted to allow lipases access to their substrates leading to increased release FFAs from large TAG stores. Thus, it is conceivable that lipotoxic FFAs may overload the mitochondrial oxidation system, thereby causing oxidative stress and the production of reactive oxygen species that are commonly associated with NAFL pathology (Cichoz-Lach and Michalak 2014).

178

Figure 6.2 Proposed Model of PTM Dysregulation of Plin2 A conceptual model based on ubiquitomic, proteomic and phosphoproteomic data obtained from basal and insulin-stimulated chow and HF-fed rats. In the basal energy-poor state, Plin2 is phosphorylated and ubiquitinated for degradation, thereby allowing lipase sequestration on LDs for FA release. In the fed state, insulin stimulation inhibits phosphorylation of Plin2. With HF-feeding, phosphorylation on alternative sites prevents its degradation and promotes hepatic steatosis. The potential role of persistent ubiquitination of Plin2 with insulin stimulation under HFD remains unclear.

It is important to note that these results are derived from mass spectrometry data mining and thus targeted in vitro measures are required to substantiate these hypotheses. Instead, high- throughput datasets such as those presented in this thesis act as launching pads in modelling data in novel and creative ways with a biologically sound foundation. Combining proteomic data with multiple PTM datasets, in combination with biochemical changes associated with lipid homeostasis, allows us to postulate complex mechanisms at a larger scale, and thus presents as a valuable resource in understanding diseases of impaired signalling. Future studies should focus on the validation of novel and known phosphosites identified in the LD, particularly those sites on regulators of lipolysis such as G0S2, Plin2, Plin5 and ATGL.

179 6.1.2 Coordinated Interplay of the ‘Lipolysome’: An Interactor Story Apart from PTMs, differential regulation of protein activity can occur through a multitude of mechanisms, such as subcellular translocation and protein-protein interactions, each bringing a new layer of complexity. It is conceivable that proteins that have multiple types of regulation are critical nodes for controlling essential processes. This is evident by the relatively high conservation (>60% across >80 eukaryotes) of multiple types of PTMs found specifically on metabolic enzymes such as ATGL, suggesting the importance of efficiently utilising energy sources (Venne et al. 2014). A fitting example is the lipase primarily responsible for the first step of lipolysis, ATGL, and its long list of modifications and binding partners that modulate its lipolytic activity and thereby govern the storage of energy-dense substrates in hepatocytes until required (Yang et al. 2010, Granneman et al. 2011, Wang et al. 2011, MacPherson et al. 2013, Olzmann et al. 2013, Grahn et al. 2014, Kim et al. 2016). Although ATGL-mediated hydrolysis of TAG occurs in nearly all tissues, the large network of regulators that control its activity, sometimes referred to as a ‘lipolysome’, is quite variable between cell types and is largely cell autonomous (Lass et al. 2011). Indeed, the majority of proteins that are considered lipolytic regulators have been characterized in adipocytes and myocytes, with limited knowledge existing in hepatocytes. For instance, Plin1 is a known PKA-activated transporter of HSL to LDs and facilitates TAG hydrolysis in adipocytes (Miyoshi et al. 2006). However, Plin1 is not expressed in hepatocytes and it is uncertain whether other perilipins, such as Plin2 or Plin5 compensate for this, despite having vastly different associations and functions (Kimmel and Sztalryd 2014, Tsai et al. 2017, Wei et al. 2018). Therefore, identification of the hepatic lipolysome and characterisation of its regulation are crucial in elucidating lipolytic dysregulation in NAFL.

Two of the most commonly discussed regulators of ATGL activity are its activator, CGI58 and inhibitor, G0S2. The two proteins bind to different domains on ATGL and are therefore not competitive, despite acting in opposition to each other (Lu et al. 2010). While CGI58 binds to the hydrophobic C-terminal domain used to target ATGL to LDs, G0S2 binds to its patatin- like domain and blocks substrate accessibility, irrespective of CGI58 binding. (Lu et al. 2010). Additionally, PLIN5 decreases ATGL activity by competitively sequestering CGI58, thereby inhibiting ATGL-mediated lipolysis and protecting mitochondria from lipotoxic overload (Wang et al. 2015). More recently, UBXD8 was identified as another binding partner of ATGL, where it separates ATGL and CGI58 via its p97 segregase domain to inhibit lipolysis (Olzmann et al. 2013); In results described in Chapter 5, UBXD8 was identified as a novel regulator of

180 FA oxidation, alongside its role in regulating ATGL activity. Interestingly, UBXD8 knockdown in AML12 hepatocytes caused a marked reduction in G0S2, PPARa and CPT1a protein expression and an increase in Plin5 expression, thereby linking ATGL-mediated FA release with transcriptional regulation of FA oxidation. Together, these data illustrate a multi- protein complex consisting of at least five independent LD-associated proteins that govern ATGL activity in hepatocytes.

Traditionally, unrestrained lipolysis is associated with hepatic steatosis and IR solely from the standpoint of adipose tissue, where leaky LDs in adipocytes supply excess substrates to the liver for TAG synthesis (Schweiger et al. 2017). However, studies have reported the influence of insulin on ATGL mRNA or lipolysis in hepatocytes, suggesting insulin-stimulated regulation of FA release in the liver (Beynen et al. 1981, Zhang et al. 2016). The mechanism that ATGL uses in the tight regulation of protein-protein interactions lie in PTMs such as phosphorylation and ubiquitination. In Chapter 3, mass spectrometry data revealed very little change in ATGL protein abundance at LDs across all conditions, but a significant decrease in Ser422 phosphorylation with insulin stimulation. Further, the insulin-stimulated phosphorylation of ATGL was absent in HFD-fed animals, likely affecting lipolytic activity. It is conceivable that phosphorylation of ATGL at this phosphosite in response to insulin promotes the association of ATGL with G0S2, and diminished insulin signalling in IR is insufficient to appropriately suppress lipolysis. Interestingly, ATGL was not identified in the rat liver ubiquitome, as described in Chapter 3 of this thesis. In other cell types, ATGL has been reported to have a short half-life and associates with proteins that have the ability to signal and degrade ATGL via the ubiquitin-proteasome system (Olzmann et al. 2013), however this was not observed in hepatocytes in this study. The coactivation of CGI58 and ATGL is unlikely to depend on phospho- or ubiquitin-regulation of CGI58 in hepatocytes, despite being a validated target of PKA phosphorylation in Plin1-expressing adipocytes (Sahu-Osen et al. 2015). It is possible that the phospho-activated separation of CGI58 from Plin1 is not necessary in cells that lack Plin1 such as hepatocytes, especially because phosphorylation is not known to affect ATGL-mediated LD turnover. The role of PTMs in the association of UBXD8 with ATGL is a little more controversial, whereby mass spectrometry data identified decreased abundance of UBXD8 at the LD in healthy insulin stimulated animals, whereas immunofluorescence data suggested LD targeting with insulin stimulation. Considering the established role of UBXD8 in inhibiting ATGL-mediated lipolysis, it is likely that UBXD8

181 translocates to LDs with insulin stimulation to inhibit this process in the energy-rich fed state that insulin treatment mimics. Thus, while ATGL itself is mediated by a variety of regulators, each of those regulators is also independently regulated by other factors, thereby demonstrating the complexity of LD dynamics.

Collectively, ATGL protein activity can be regulated by protein abundance, protein-protein interactions and PTMs in response to hormone stimulation and altered diet composition. In this thesis, several layers of regulation were measured through thoughtful modelling of this complexity. Furthermore, the ATGL binding protein UBXD8 was further characterized to shed light on the importance of ATGL translocation, degradation and interaction in facilitating lipolysis in hepatocytes.

6.1.3 Modelling Metabolic Disease: Influence on Proteome The complex network of control measures for intracellular protein regulation requires considerable sensitivity to their microenvironment to distinguish subtle changes. Thus, representative models of metabolism and disease that generate these microenvironments are essential for obtaining accurate data. In the experiments described in this thesis, the HFD-fed rat was the primary model of fatty liver and IR. This is a well described model that exhibits classical aspects of short-term high-fat and high-sucrose feeding such as hepatic steatosis and IR but not oxidative stress or intralobular inflammation (Nakamura and Terauchi 2013). The liver ubiquitome and liver LD proteome and phosphoproteome were defined in this model as well as their response to acute insulin stimulation. Although rodents are often preferred for diet studies of metabolism, obesity and NAFLD, selection of the most appropriate model often involves trading one advantage in favour for another. For instance, NAFLD is a spectrum disease encompassing NAFL, NASH, fibrosis, cirrhosis and HCC; however, C57BL/6 mice and Wistar rats fed a diet composed of 40%-70% fat for less than a year do not progress from NAFL to NASH (Ragab et al. 2015, Lau et al. 2017). Rather, a methionine and choline deficient diet causes oxidative stress and impairs lipid metabolism leading to NAFL/NASH, but these diets do not induce obesity or IR, which are often associated with NAFL(D) (Xu et al. 2010, Rinella et al. 2016). However, studies have shown that longer HFD-feeding (> 1 year) in C57BL/6 mice and even less time (> 6 months) in B6/129 mice can lead to an increased degree of livery injury, albeit not comparable to methionine and choline deficient diets (Asgharpour

182 et al. 2016). Thus, variability can result from rodent species, proportion and type of fat in the diet and duration of treatment.

The specific constrains on steatotic pathology in the liver can have a significant impact on proteomic, phosphoproteomic and ubiquitomic data obtained. The HFD-fed rat is one of many models of fatty liver. Alcoholic fatty liver disease (AFLD) is a related liver pathology to NAFL with similar pathological spectra and associated comorbidities. AFLD has greater oxidative stress due to ethanol intake, greater inflammatory infiltration and thus greater progression to steatohepatitis compared to NAFL (Toshikuni et al. 2014). To date, the LD proteome of AFLD has not been reported. The whole liver proteome of a mouse model of AFLD was described by Carr et al. (2014), which identified Plin2 as the sole member of the ‘lipolysome’ to have altered protein abundance in AFLD liver. Similarly, the LD phosphoproteome of the AFLD has yet to be described; however, it is predicted that other PTMs linked to oxidative stress, specifically carbonylation, nitrosylation and ubiquitination leading to decreased anti-oxidant functions, are likely to be most abundant and biologically important in this setting (Newton et al. 2009). Another pathology that is characterised by hepatic lipid accumulation and IR is hepatitis C. The LD proteome of hepatitis C-infected Huh7 hepatocytes was enriched with proteins involved in RNA binding, cytoplasmic stress granule and vesicle pathways (Rosch et al. 2016). Strikingly, members of the ‘lipolysome’ including ATGL, CGI58 and PNPLA3 were depleted in infected cells, suggesting alternative mechanisms in the development of hepatic steatosis in the presence of hepatitis C compared to HFD-fed rodents. Collectively, variable models of fatty liver all accumulate LDs in hepatocytes regardless of general adiposity and that these LDs show changes in protein expression of lipolytic regulators as a pathogenic feature of the disease. Additionally, LD proteins are highly modified by PTMs beyond phosphorylation and ubiquitination in numerous models of fatty liver, suggesting their importance in dysregulated lipid flux. Further, AFLD and hepatitis C exhibit hepatic IR; however, the common features of the LD proteome and PTMs in response to acute insulin stimulation across these varied models of fatty liver remains to be determined. Moreover, no studies have assessed the proteomic and phosphoproteomic differences in the liver between IR and insulin sensitive NAFL(D) models. The pathophysiology of lipid accumulation and IR are often so intimately linked that results from such a study would help delineate the LD proteome in steatosis from IR.

The validation of mass spectrometry data using in vitro systems also requires careful consideration of the metabolic sensitivity of specific cell lines. In liver research, the majority

183 of in vitro studies use HepG2 and Huh7 cells, two hepatic carcinoma cell lines, for non- cancerous work (Hewitt et al. 2007, Reid et al. 2008, Kuo et al. 2012, Jin et al. 2018). Although these cell lines possess various features that can be useful, such as pronounced glucose uptake, they exhibit many markers of tumour metabolism that are likely to be predominant in the phenotype. To tackle this challenge, the metabolic phenotype of three normal human and mouse hepatic immortalised cell lines were measured against the gold-standard model of primary mouse hepatocytes and the commonly used model, HepG2 cells. Results described in Chapter 4 demonstrated remarkable baseline differences in FA, glucose and cholesterol intracellular utilization and variable sensitivity to hormonal or lipid treatment. AML12 cells, a mouse hepatocyte cell line, were chosen as the model to characterise UBXD8 function in Chapter 5 due to their consistent sensitivity to insulin treatment, and their ability to store and oxidize lipid and glucose appropriately (glucose data not shown in thesis). Although AML12 cells were the best immortalised cell line to use for UBXD8 validation, it is important to note that there are a number of individualised factors that can influence or even improve experimental readouts for each cell line. This includes but is not limited to the confluence of cells when exposed to treatment, culture volume and media composition, incubation time and treatment toxicity, all of which can affect cell lines differently (Dong et al. 2008, Wright Muelas et al. 2018). For example, AML12, IHH and HepG2 cells were able to handle higher concentrations of unsaturated FAs compared to PH5CH8, which died if treated above a certain concentration. However, this was remedied when PH5CH8 cells were seeded at a higher confluence in order to 1) account for their lower growth rate and 2) reduce the ratio of FA availability per individual cell, thereby reducing the toxicity of the treatment to these cells. These nuances in physiological environments during treatment can feedback into the way that proteins of interest respond under particular conditions. Thus, thorough characterisation or investigation of available in vivo and in vitro models for a specific set of experiments is required before choosing one out of convenience or familiarity.

6.1.4 Conclusion In conclusion, results presented in this thesis demonstrated for the first time that the ubiquitome and LD proteome and phosphoproteome are sensitive to insulin and diet and are perturbed in HFD-induced IR. In particular, multiple levels of molecular and PTM regulation work together to tightly modulate expression and activity of key lipases at the LD. Further, these regulatory processes were, in general, more dysregulated than total protein expression itself in the disease

184 state. Finally, UBXD8 has been characterized as a new member of the ‘lipolysome’ and an integral regulator of lipid metabolism by controlling lipase-mediated FA oxidation in hepatocytes. Collectively, this thesis describes the complexity of protein dysregulation in NAFL and IR and presents novel perspectives in the investigation of its pathogenesis.

185 References

Adams, L. A., J. F. Lymp, J. St Sauver, S. O. Sanderson, K. D. Lindor, A. Feldstein and P. Angulo (2005). "The natural history of nonalcoholic fatty liver disease: a population- based cohort study." Gastroenterology 129(1): 113-121.

Adiels, M., M. R. Taskinen, C. Packard, M. J. Caslake, A. Soro-Paavonen, J. Westerbacka, S. Vehkavaara, A. Hakkinen, S. O. Olofsson, H. Yki-Jarvinen and J. Boren (2006). "Overproduction of large VLDL particles is driven by increased liver fat content in man." Diabetologia 49(4): 755-765.

Aguirre, V., E. D. Werner, J. Giraud, Y. H. Lee, S. E. Shoelson and M. F. White (2002). "Phosphorylation of Ser307 in insulin receptor substrate-1 blocks interactions with the insulin receptor and inhibits insulin action." J Biol Chem 277(2): 1531-1537.

An, S., L. P. Zhao, L. J. Shen, S. Wang, K. Zhang, Y. Qi, J. Zheng, X. J. Zhang, X. Y. Zhu, R. Bao, L. Yang, Y. X. Lu, Z. G. She and Y. D. Tang (2017). "USP18 protects against hepatic steatosis and insulin resistance through its deubiquitinating activity." Hepatology 66(6): 1866-1884.

Anthonsen, M. W., L. Ronnstrand, C. Wernstedt, E. Degerman and C. Holm (1998). "Identification of novel phosphorylation sites in hormone-sensitive lipase that are phosphorylated in response to isoproterenol and govern activation properties in vitro." J Biol Chem 273(1): 215-221.

Arendt, B. M., E. M. Comelli, D. W. Ma, W. Lou, A. Teterina, T. Kim, S. K. Fung, D. K. Wong, I. McGilvray, S. E. Fischer and J. P. Allard (2015). "Altered hepatic gene expression in nonalcoholic fatty liver disease is associated with lower hepatic n-3 and n-6 polyunsaturated fatty acids." Hepatology 61(5): 1565-1578.

Armstrong, M. J., L. A. Adams, A. Canbay and W. K. Syn (2014). "Extrahepatic complications of nonalcoholic fatty liver disease." Hepatology 59(3): 1174-1197.

Arrese, E. L., F. Z. Saudale and J. L. Soulages (2014). "Lipid Droplets as Signaling Platforms Linking Metabolic and Cellular Functions." Lipid Insights 7: 7-16.

Asgharpour, A., S. C. Cazanave, T. Pacana, M. Seneshaw, R. Vincent, B. A. Banini, D. P. Kumar, K. Daita, H. K. Min, F. Mirshahi, P. Bedossa, X. Sun, Y. Hoshida, S. V. Koduru, D. Contaifer, Jr., U. O. Warncke, D. S. Wijesinghe and A. J. Sanyal (2016). "A diet-induced animal model of non-alcoholic fatty liver disease and hepatocellular cancer." J Hepatol 65(3): 579-588.

Ausina, P., D. Da Silva, D. Majerowicz, P. Zancan and M. Sola-Penna (2018). "Insulin specifically regulates expression of liver and muscle phosphofructokinase isoforms." Biomed Pharmacother 103: 228-233.

Bae, S., S. Y. Kim, J. H. Jung, Y. Yoon, H. J. Cha, H. Lee, K. Kim, J. Kim, I. S. An, J. Kim, H. D. Um, I. C. Park, S. J. Lee, S. Y. Nam, Y. W. Jin, J. H. Lee and S. An (2012). "Akt is negatively regulated by the MULAN E3 ligase." Cell Res 22(5): 873-885.

186

Balasubramanyam, M., R. Sampathkumar and V. Mohan (2005). "Is insulin signaling molecules misguided in diabetes for ubiquitin-proteasome mediated degradation?" Mol Cell Biochem 275(1-2): 117-125.

Baldwin, K. M., R. H. Fitts, F. W. Booth, W. W. Winder and J. O. Holloszy (1975). "Depletion of muscle and liver glycogen during exercise. Protective effect of training." Pflugers Arch 354(3): 203-212.

Ballestri, S., F. Nascimbeni, E. Baldelli, A. Marrazzo, D. Romagnoli and A. Lonardo (2017). "NAFLD as a Sexual Dimorphic Disease: Role of Gender and Reproductive Status in the Development and Progression of Nonalcoholic Fatty Liver Disease and Inherent Cardiovascular Risk." Adv Ther 34(6): 1291-1326.

Barneda, D., J. Planas-Iglesias, M. L. Gaspar, D. Mohammadyani, S. Prasannan, D. Dormann, G. S. Han, S. A. Jesch, G. M. Carman, V. Kagan, M. G. Parker, N. T. Ktistakis, J. Klein-Seetharaman, A. M. Dixon, S. A. Henry and M. Christian (2015). "The brown adipocyte protein CIDEA promotes lipid droplet fusion via a phosphatidic acid- binding amphipathic helix." Elife 4: e07485.

Bartz, R., W. H. Li, B. Venables, J. K. Zehmer, M. R. Roth, R. Welti, R. G. Anderson, P. Liu and K. D. Chapman (2007). "Lipidomics reveals that adiposomes store ether lipids and mediate phospholipid traffic." J Lipid Res 48(4): 837-847.

Bartz, R., J. K. Zehmer, M. Zhu, Y. Chen, G. Serrero, Y. Zhao and P. Liu (2007). "Dynamic activity of lipid droplets: protein phosphorylation and GTP-mediated protein translocation." J Proteome Res 6(8): 3256-3265.

Basantani, M. K., M. T. Sitnick, L. Cai, D. S. Brenner, N. P. Gardner, J. Z. Li, G. Schoiswohl, K. Yang, M. Kumari, R. W. Gross, R. Zechner and E. E. Kershaw (2011). "Pnpla3/Adiponutrin deficiency in mice does not contribute to fatty liver disease or metabolic syndrome." J Lipid Res 52(2): 318-329.

Bathaie, S. Z., M. Ashrafi, M. Azizian and F. Tamanoi (2017). "Mevalonate Pathway and Human Cancers." Curr Mol Pharmacol 10(2): 77-85.

Becuwe, M., L. M. Bond, N. Mejhert, S. Boland, S. D. Elliott, M. Cicconet, X. N. Liu, M. M. Graham, T. C. Walther and R. V. Farese (2018). "FIT2 is a lipid phosphate phosphatase crucial for endoplasmic reticulum homeostasis." bioRxiv.

Bedogni, G., V. Nobili and C. Tiribelli (2014). "Epidemiology of fatty liver: an update." World J Gastroenterol 20(27): 9050-9054.

Beilstein, F., V. Carriere, A. Leturque and S. Demignot (2016). "Characteristics and functions of lipid droplets and associated proteins in enterocytes." Exp Cell Res 340(2): 172- 179.

Ben M'barek, K., D. Ajjaji, A. Chorlay, S. Vanni, L. Foret and A. R. Thiam (2017). "ER Membrane Phospholipids and Surface Tension Control Cellular Lipid Droplet Formation." Dev Cell 41(6): 591-604.e597.

Benador, I. Y., M. Veliova, K. Mahdaviani, A. Petcherski, J. D. Wikstrom, E. A. Assali, R. Acin-Perez, M. Shum, M. F. Oliveira, S. Cinti, C. Sztalryd, W. D. Barshop, J. A.

187 Wohlschlegel, B. E. Corkey, M. Liesa and O. S. Shirihai (2018). "Mitochondria Bound to Lipid Droplets Have Unique Bioenergetics, Composition, and Dynamics that Support Lipid Droplet Expansion." Cell Metab 27(4): 869-885.e866.

Bergman, R. N. (2000). "Non-esterified fatty acids and the liver: why is insulin secreted into the portal vein?" Diabetologia 43(7): 946-952.

Bergman, R. N., D. C. Bradley and M. Ader (1993). "On insulin action in vivo: the single gateway hypothesis." Adv Exp Med Biol 334: 181-198.

Bersuker, K. and J. A. Olzmann (2017). "Establishing the lipid droplet proteome: Mechanisms of lipid droplet protein targeting and degradation." Biochim Biophys Acta Mol Cell Biol Lipids 1862(10 Pt B): 1166-1177.

Bevan, P. (2001). "Insulin signalling." J Cell Sci 114(Pt 8): 1429-1430.

Beynen, A. C., A. J. Van der Molen and M. J. Geelen (1981). "Control of lipolysis in isolated hepatocytes by insulin and glucagon." Horm Metab Res 13(3): 183-184.

Bian, Y., C. Song, K. Cheng, M. Dong, F. Wang, J. Huang, D. Sun, L. Wang, M. Ye and H. Zou (2014). "An enzyme assisted RP-RPLC approach for in-depth analysis of human liver phosphoproteome." J Proteomics 96: 253-262.

Binns, D., T. Januszewski, Y. Chen, J. Hill, V. S. Markin, Y. Zhao, C. Gilpin, K. D. Chapman, R. G. Anderson and J. M. Goodman (2006). "An intimate collaboration between peroxisomes and lipid bodies." J Cell Biol 173(5): 719-731.

Blaner, W. S., S. M. O'Byrne, N. Wongsiriroj, J. Kluwe, D. M. D'Ambrosio, H. Jiang, R. F. Schwabe, E. M. Hillman, R. Piantedosi and J. Libien (2009). "Hepatic stellate cell lipid droplets: a specialized lipid droplet for retinoid storage." Biochim Biophys Acta 1791(6): 467-473.

Blanquart, C., O. Barbier, J. C. Fruchart, B. Staels and C. Glineur (2002). "Peroxisome proliferator-activated receptor alpha (PPARalpha ) turnover by the ubiquitin- proteasome system controls the ligand-induced expression level of its target genes." J Biol Chem 277(40): 37254-37259.

Bockus, L. B., S. Matsuzaki, S. S. Vadvalkar, Z. T. Young, J. R. Giorgione, M. F. Newhardt, M. Kinter and K. M. Humphries (2017). "Cardiac Insulin Signaling Regulates Glycolysis Through Content and Activity." J Am Heart Assoc 6(12).

Bondia-Pons, I., C. Molto-Puigmarti, A. I. Castellote and M. C. Lopez-Sabater (2007). "Determination of conjugated linoleic acid in human plasma by fast gas chromatography." J Chromatogr A 1157(1-2): 422-429.

Boren, J., M. R. Taskinen, S. O. Olofsson and M. Levin (2013). "Ectopic lipid storage and insulin resistance: a harmful relationship." J Intern Med 274(1): 25-40.

Bosma, M., R. Minnaard, L. M. Sparks, G. Schaart, M. Losen, M. H. de Baets, H. Duimel, S. Kersten, P. E. Bickel, P. Schrauwen and M. K. Hesselink (2012). "The lipid droplet

188 coat protein perilipin 5 also localizes to muscle mitochondria." Histochem Cell Biol 137(2): 205-216.

Bostrom, P., L. Andersson, M. Rutberg, J. Perman, U. Lidberg, B. R. Johansson, J. Fernandez- Rodriguez, J. Ericson, T. Nilsson, J. Boren and S. O. Olofsson (2007). "SNARE proteins mediate fusion between cytosolic lipid droplets and are implicated in insulin sensitivity." Nat Cell Biol 9(11): 1286-1293.

Boutant, M., S. S. Kulkarni, M. Joffraud, J. Ratajczak, M. Valera-Alberni, R. Combe, A. Zorzano and C. Canto (2017). "Mfn2 is critical for brown adipose tissue thermogenic function." Embo j 36(11): 1543-1558.

Boutari, C., P. Lefkos, V. G. Athyros, A. Karagiannis and K. Tziomalos (2018). "Nonalcoholic Fatty Liver Disease vs. Nonalcoholic Steatohepatitis: Pathological and Clinical Implications." Curr Vasc Pharmacol 16(3): 214-218.

Bradley, D. C., R. A. Poulin and R. N. Bergman (1993). "Dynamics of hepatic and peripheral insulin effects suggest common rate-limiting step in vivo." Diabetes 42(2): 296-306.

Brown, J. M., J. L. Betters, C. Lord, Y. Ma, X. Han, K. Yang, H. M. Alger, J. Melchior, J. Sawyer, R. Shah, M. D. Wilson, X. Liu, M. J. Graham, R. Lee, R. Crooke, G. I. Shulman, B. Xue, H. Shi and L. Yu (2010). "CGI-58 knockdown in mice causes hepatic steatosis but prevents diet-induced obesity and glucose intolerance." J Lipid Res 51(11): 3306-3315.

Brown, M. S. and J. L. Goldstein (2008). "Selective versus total insulin resistance: a pathogenic paradox." Cell Metab 7(2): 95-96.

Bu, S. Y. and D. G. Mashek (2010). "Hepatic long-chain acyl-CoA synthetase 5 mediates fatty acid channeling between anabolic and catabolic pathways." J Lipid Res 51(11): 3270- 3280.

Buchberger, A. (2002). "From UBA to UBX: new words in the ubiquitin vocabulary." Trends Cell Biol 12(5): 216-221.

Buque, X., A. Cano, M. E. Miquilena-Colina, C. Garcia-Monzon, B. Ochoa and P. Aspichueta (2012). "High insulin levels are required for FAT/CD36 plasma membrane translocation and enhanced fatty acid uptake in obese Zucker rat hepatocytes." Am J Physiol Endocrinol Metab 303(4): E504-514.

Burri, L., G. H. Thoresen and R. K. Berge (2010). "The Role of PPARalpha Activation in Liver and Muscle." PPAR Res 2010.

Calvert, V. S., R. Collantes, H. Elariny, A. Afendy, A. Baranova, M. Mendoza, Z. Goodman, L. A. Liotta, E. F. Petricoin and Z. M. Younossi (2007). "A systems biology approach to the pathogenesis of obesity-related nonalcoholic fatty liver disease using reverse phase protein microarrays for multiplexed cell signaling analysis." Hepatology 46(1): 166-172.

Cantley, J. L., T. Yoshimura, J. P. Camporez, D. Zhang, F. R. Jornayvaz, N. Kumashiro, F. Guebre-Egziabher, M. J. Jurczak, M. Kahn, B. A. Guigni, J. Serr, J. Hankin, R. C. Murphy, G. W. Cline, S. Bhanot, V. P. Manchem, J. M. Brown, V. T. Samuel and G.

189 I. Shulman (2013). "CGI-58 knockdown sequesters diacylglycerols in lipid droplets/ER-preventing diacylglycerol-mediated hepatic insulin resistance." Proc Natl Acad Sci U S A 110(5): 1869-1874.

Capanni, M., F. Calella, M. R. Biagini, S. Genise, L. Raimondi, G. Bedogni, G. Svegliati- Baroni, F. Sofi, S. Milani, R. Abbate, C. Surrenti and A. Casini (2006). "Prolonged n- 3 polyunsaturated fatty acid supplementation ameliorates hepatic steatosis in patients with non-alcoholic fatty liver disease: a pilot study." Aliment Pharmacol Ther 23(8): 1143-1151.

Carr, R. M., G. Peralta, X. Yin and R. S. Ahima (2014). "Absence of perilipin 2 prevents hepatic steatosis, glucose intolerance and ceramide accumulation in alcohol-fed mice." PLoS One 9(5): e97118.

Cermelli, S., Y. Guo, S. P. Gross and M. A. Welte (2006). "The lipid-droplet proteome reveals that droplets are a protein-storage depot." Curr Biol 16(18): 1783-1795.

Cerqueira, N. M., E. F. Oliveira, D. S. Gesto, D. Santos-Martins, C. Moreira, H. N. Moorthy, M. J. Ramos and P. A. Fernandes (2016). "Cholesterol Biosynthesis: A Mechanistic Overview." Biochemistry 55(39): 5483-5506.

Chakrabarti, P., J. Y. Kim, M. Singh, Y. K. Shin, J. Kim, J. Kumbrink, Y. Wu, M. J. Lee, K. H. Kirsch, S. K. Fried and K. V. Kandror (2013). "Insulin inhibits lipolysis in adipocytes via the evolutionarily conserved mTORC1-Egr1-ATGL-mediated pathway." Mol Cell Biol 33(18): 3659-3666.

Chakravarthy, M. V., I. J. Lodhi, L. Yin, R. R. Malapaka, H. E. Xu, J. Turk and C. F. Semenkovich (2009). "Identification of a physiologically relevant endogenous ligand for PPARalpha in liver." Cell 138(3): 476-488.

Chalasani, N., Z. Younossi, J. E. Lavine, A. M. Diehl, E. M. Brunt, K. Cusi, M. Charlton and A. J. Sanyal (2012). "The diagnosis and management of non-alcoholic fatty liver disease: practice Guideline by the American Association for the Study of Liver Diseases, American College of Gastroenterology, and the American Gastroenterological Association." Hepatology 55(6): 2005-2023.

Chavez, J. A., T. A. Knotts, L. P. Wang, G. Li, R. T. Dobrowsky, G. L. Florant and S. A. Summers (2003). "A role for ceramide, but not diacylglycerol, in the antagonism of insulin signal transduction by saturated fatty acids." J Biol Chem 278(12): 10297- 10303.

Chen, B., M. Sirota, H. Fan-Minogue, D. Hadley and A. J. Butte (2015). "Relating hepatocellular carcinoma tumor samples and cell lines using gene expression data in translational research." BMC Med Genomics 8 Suppl 2: S5.

Chen, W., B. Chang, L. Li and L. Chan (2010). "Patatin-like phospholipase domain-containing 3/adiponutrin deficiency in mice is not associated with fatty liver disease." Hepatology 52(3): 1134-1142.

Chen, X., S. Xu, S. Wei, Y. Deng, Y. Li, F. Yang and P. Liu (2016). "Comparative Proteomic Study of Fatty Acid-treated Myoblasts Reveals Role of Cox-2 in Palmitate-induced Insulin Resistance." Sci Rep 6: 21454.

190 Cheng, K. K., K. S. Lam, Y. Wang, D. Wu, M. Zhang, B. Wang, X. Li, R. L. Hoo, Z. Huang, G. Sweeney and A. Xu (2013). "TRAF6-mediated ubiquitination of APPL1 enhances hepatic actions of insulin by promoting the membrane translocation of Akt." Biochem J 455(2): 207-216.

Cherrington, A. D. (1999). "Banting Lecture 1997. Control of glucose uptake and release by the liver in vivo." Diabetes 48(5): 1198-1214.

Chitturi, S., V. W. Wong and G. Farrell (2011). "Nonalcoholic fatty liver in Asia: Firmly entrenched and rapidly gaining ground." J Gastroenterol Hepatol 26 Suppl 1: 163- 172.

Chiurchiu, V., A. Leuti and M. Maccarrone (2018). "Bioactive Lipids and Chronic Inflammation: Managing the Fire Within." Front Immunol 9: 38.

Choi, C. S., D. B. Savage, A. Kulkarni, X. X. Yu, Z. X. Liu, K. Morino, S. Kim, A. Distefano, V. T. Samuel, S. Neschen, D. Zhang, A. Wang, X. M. Zhang, M. Kahn, G. W. Cline, S. K. Pandey, J. G. Geisler, S. Bhanot, B. P. Monia and G. I. Shulman (2007). "Suppression of diacylglycerol acyltransferase-2 (DGAT2), but not DGAT1, with antisense oligonucleotides reverses diet-induced hepatic steatosis and insulin resistance." J Biol Chem 282(31): 22678-22688.

Choi, D. H., S. J. Lee, C. D. Kang, M. O. Park, D. W. Choi, T. S. Kim, W. Lee, B. R. Cho, Y. H. Kim, B. K. Lee, D. R. Ryu and J. W. Lee (2013). "Nonalcoholic fatty liver disease is associated with coronary artery disease in Koreans." World J Gastroenterol 19(38): 6453-6457.

Choi, S. H. and H. N. Ginsberg (2011). "Increased very low density lipoprotein (VLDL) secretion, hepatic steatosis, and insulin resistance." Trends Endocrinol Metab 22(9): 353-363.

Choudhary, V., G. Golani, A. S. Joshi, S. Cottier, R. Schneiter, W. A. Prinz and M. M. Kozlov (2018). "Architecture of Lipid Droplets in Endoplasmic Reticulum Is Determined by Phospholipid Intrinsic Curvature." Curr Biol 28(6): 915-926.e919.

Chow, J. D., R. T. Lawrence, M. E. Healy, J. E. Dominy, J. A. Liao, D. S. Breen, F. L. Byrne, B. M. Kenwood, C. Lackner, S. Okutsu, V. R. Mas, S. H. Caldwell, J. L. Tomsig, G. J. Cooney, P. B. Puigserver, N. Turner, D. E. James, J. Villen and K. L. Hoehn (2014). "Genetic inhibition of hepatic acetyl-CoA carboxylase activity increases liver fat and alters global protein acetylation." Mol Metab 3(4): 419-431.

Cichoz-Lach, H. and A. Michalak (2014). "Oxidative stress as a crucial factor in liver diseases." World J Gastroenterol 20(25): 8082-8091.

Ciechanover, A., A. Orian and A. L. Schwartz (2000). "Ubiquitin-mediated proteolysis: biological regulation via destruction." Bioessays 22(5): 442-451.

Clayton, R. F., A. Rinaldi, E. E. Kandyba, M. Edward, C. Willberg, P. Klenerman and A. H. Patel (2005). "Liver cell lines for the study of hepatocyte functions and immunological response." Liver Int 25(2): 389-402.

191 Cohen, J. C., J. D. Horton and H. H. Hobbs (2011). "Human fatty liver disease: old questions and new insights." Science 332(6037): 1519-1523.

Cole, M. A., A. J. Murray, L. E. Cochlin, L. C. Heather, S. McAleese, N. S. Knight, E. Sutton, A. A. Jamil, N. Parassol and K. Clarke (2011). "A high fat diet increases mitochondrial fatty acid oxidation and uncoupling to decrease efficiency in rat heart." Basic Res Cardiol 106(3): 447-457.

Consortium, T. U. (2017). "UniProt: the universal protein knowledgebase." Nucleic Acids Research 45(D1).

Cox, J. and M. Mann (2008). "MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification." Nat Biotechnol 26(12): 1367-1372.

Cox, J. and M. Mann (2011). "Quantitative, high-resolution proteomics for data-driven systems biology." Annu Rev Biochem 80: 273-299.

Cox, J., N. Neuhauser, A. Michalski, R. A. Scheltema, J. V. Olsen and M. Mann (2011). "Andromeda: a peptide search engine integrated into the MaxQuant environment." J Proteome Res 10(4): 1794-1805.

Crunk, A. E., J. Monks, A. Murakami, M. Jackman, P. S. Maclean, M. Ladinsky, E. S. Bales, S. Cain, D. J. Orlicky and J. L. McManaman (2013). "Dynamic regulation of hepatic lipid droplet properties by diet." PLoS One 8(7): e67631.

Currie, E., X. Guo, R. Christiano, C. Chitraju, N. Kory, K. Harrison, J. Haas, T. C. Walther and R. V. Farese, Jr. (2014). "High confidence proteomic analysis of yeast LDs identifies additional droplet proteins and reveals connections to dolichol synthesis and sterol acetylation." J Lipid Res 55(7): 1465-1477.

Damania, A., E. Jain and A. Kumar (2014). "Advancements in in vitro hepatic models: application for drug screening and therapeutics." Hepatol Int 8(1): 23-38.

Daniel, J., H. Maamar, C. Deb, T. D. Sirakova and P. E. Kolattukudy (2011). "Mycobacterium tuberculosis uses host triacylglycerol to accumulate lipid droplets and acquires a dormancy-like phenotype in lipid-loaded macrophages." PLoS Pathog 7(6): e1002093.

Darabian, S., M. Hormuz, M. A. Latif, S. Pahlevan and M. J. Budoff (2013). "The role of carotid intimal thickness testing and risk prediction in the development of coronary atherosclerosis." Curr Atheroscler Rep 15(3): 306.

Davaapil, H., Y. Tsuchiya and I. Gout (2014). "Signalling functions of coenzyme A and its derivatives in mammalian cells." Biochem Soc Trans 42(4): 1056-1062.

De Freitas, A. S. and F. Depocas (1965). "Fatty Acid and Glyceride Glycerol Synthesis from Glucose during High Rates of Glucose Uptake in the Intact Rat." Can J Biochem 43: 437-450. den Brok, M. H., T. K. Raaijmakers, E. Collado-Camps and G. J. Adema (2018). "Lipid Droplets as Immune Modulators in Myeloid Cells." Trends Immunol 39(5): 380-392.

192 Diaconeasa, B., G. H. Mazock, A. P. Mahowald and R. R. Dubreuil (2013). "Genetic studies of spectrin in the larval fat body of Drosophila melanogaster: evidence for a novel lipid uptake apparatus." Genetics 195(3): 871-881.

Dietschy, J. M., S. D. Turley and D. K. Spady (1993). "Role of liver in the maintenance of cholesterol and low density lipoprotein homeostasis in different animal species, including humans." J Lipid Res 34(10): 1637-1659.

Ding, Y., L. Yang, S. Zhang, Y. Wang, Y. Du, J. Pu, G. Peng, Y. Chen, H. Zhang, J. Yu, H. Hang, P. Wu, F. Yang, H. Yang, A. Steinbuchel and P. Liu (2012). "Identification of the major functional proteins of prokaryotic lipid droplets." J Lipid Res 53(3): 399- 411.

DiPilato, L. M., F. Ahmad, M. Harms, P. Seale, V. Manganiello and M. J. Birnbaum (2015). "The Role of PDE3B Phosphorylation in the Inhibition of Lipolysis by Insulin." Mol Cell Biol 35(16): 2752-2760.

Diraison, F., V. Yankah, D. Letexier, E. Dusserre, P. Jones and M. Beylot (2003). "Differences in the regulation of adipose tissue and liver lipogenesis by carbohydrates in humans." J Lipid Res 44(4): 846-853.

Dixon, J. B., P. S. Bhathal and P. E. O'Brien (2001). "Nonalcoholic fatty liver disease: predictors of nonalcoholic steatohepatitis and liver fibrosis in the severely obese." Gastroenterology 121(1): 91-100.

Doege, H., R. A. Baillie, A. M. Ortegon, B. Tsang, Q. Wu, S. Punreddy, D. Hirsch, N. Watson, R. E. Gimeno and A. Stahl (2006). "Targeted deletion of FATP5 reveals multiple functions in liver metabolism: alterations in hepatic lipid homeostasis." Gastroenterology 130(4): 1245-1258.

Dong, J., C. F. Mandenius, M. Lubberstedt, T. Urbaniak, A. K. Nussler, D. Knobeloch, J. C. Gerlach and K. Zeilinger (2008). "Evaluation and optimization of hepatocyte culture media factors by design of experiments (DoE) methodology." Cytotechnology 57(3): 251-261.

Dong, X. C., K. D. Copps, S. Guo, Y. Li, R. Kollipara, R. A. DePinho and M. F. White (2008). "Inactivation of hepatic Foxo1 by insulin signaling is required for adaptive nutrient homeostasis and endocrine growth regulation." Cell Metab 8(1): 65-76.

Donnelly, K. L., C. I. Smith, S. J. Schwarzenberg, J. Jessurun, M. D. Boldt and E. J. Parks (2005). "Sources of fatty acids stored in liver and secreted via lipoproteins in patients with nonalcoholic fatty liver disease." J Clin Invest 115(5): 1343-1351.

Dowman, J. K., J. W. Tomlinson and P. N. Newsome (2010). "Pathogenesis of non-alcoholic fatty liver disease." Qjm 103(2): 71-83.

Doycheva, I., N. Patel, M. Peterson and R. Loomba (2013). "Prognostic implication of liver histology in patients with nonalcoholic fatty liver disease in diabetes." J Diabetes Complications 27(3): 293-300.

193 Drag, J., A. Gozdzialska, M. Knapik-Czajka, A. Gawedzka, K. Gawlik and J. Jaskiewicz (2017). "Effect of high carbohydrate diet on elongase and desaturase activity and accompanying gene expression in rat's liver." Genes Nutr 12: 2.

Ducharme, N. A. and P. E. Bickel (2008). "Lipid droplets in lipogenesis and lipolysis." Endocrinology 149(3): 942-949.

Duckworth, W. C., R. G. Bennett and F. G. Hamel (1998). "Insulin degradation: progress and potential." Endocr Rev 19(5): 608-624.

Duncan, R. E., M. Ahmadian, K. Jaworski, E. Sarkadi-Nagy and H. S. Sul (2007). "Regulation of lipolysis in adipocytes." Annu Rev Nutr 27: 79-101.

Dunn, F. L. (1990). "Hyperlipidemia in diabetes mellitus." Diabetes Metab Rev 6(1): 47-61.

Dunn, W., R. Xu, D. L. Wingard, C. Rogers, P. Angulo, Z. M. Younossi and J. B. Schwimmer (2008). "Suspected nonalcoholic fatty liver disease and mortality risk in a population- based cohort study." Am J Gastroenterol 103(9): 2263-2271.

Edgerton, D. S., G. Kraft, M. Smith, B. Farmer, P. E. Williams, K. C. Coate, R. L. Printz, R. M. O'Brien and A. D. Cherrington (2017). "Insulin's direct hepatic effect explains the inhibition of glucose production caused by insulin secretion." JCI Insight 2(6): e91863.

Eguchi, Y., H. Hyogo, M. Ono, T. Mizuta, N. Ono, K. Fujimoto, K. Chayama and T. Saibara (2012). "Prevalence and associated metabolic factors of nonalcoholic fatty liver disease in the general population from 2009 to 2010 in Japan: a multicenter large retrospective study." J Gastroenterol 47(5): 586-595.

Eichmann, T. O. and A. Lass (2015). "DAG tales: the multiple faces of diacylglycerol-- stereochemistry, metabolism, and signaling." Cell Mol Life Sci 72(20): 3931-3952.

Eisenberg-Bord, M., N. Shai, M. Schuldiner and M. Bohnert (2016). "A Tether Is a Tether Is a Tether: Tethering at Membrane Contact Sites." Dev Cell 39(4): 395-409.

El-Assaad, W., K. El-Kouhen, A. H. Mohammad, J. Yang, M. Morita, I. Gamache, O. Mamer, D. Avizonis, N. Hermance, S. Kersten, M. L. Tremblay, M. A. Kelliher and J. G. Teodoro (2015). "Deletion of the gene encoding G0/G 1 switch protein 2 (G0s2) alleviates high-fat-diet-induced weight gain and insulin resistance, and promotes browning of white adipose tissue in mice." Diabetologia 58(1): 149-157.

Ellis, J. M., J. L. Frahm, L. O. Li and R. A. Coleman (2010). "Acyl-coenzyme A synthetases in metabolic control." Curr Opin Lipidol 21(3): 212-217.

Elsasser, S., D. Chandler-Militello, B. Muller, J. Hanna and D. Finley (2004). "Rad23 and Rpn10 serve as alternative ubiquitin receptors for the proteasome." J Biol Chem 279(26): 26817-26822.

Engholm-Keller, K., P. Birck, J. Storling, F. Pociot, T. Mandrup-Poulsen and M. R. Larsen (2012). "TiSH--a robust and sensitive global phosphoproteomics strategy employing a combination of TiO2, SIMAC, and HILIC." J Proteomics 75(18): 5749-5761.

194 Engin, A. (2017). "Fat Cell and Fatty Acid Turnover in Obesity." Adv Exp Med Biol 960: 135- 160.

Engin, A. B. (2017). "What Is Lipotoxicity?" Adv Exp Med Biol 960: 197-220.

Estes, C., H. Razavi, R. Loomba, Z. Younossi and A. J. Sanyal (2018). "Modeling the epidemic of nonalcoholic fatty liver disease demonstrates an exponential increase in burden of disease." Hepatology 67(1): 123-133.

Fabbrini, E., B. S. Mohammed, F. Magkos, K. M. Korenblat, B. W. Patterson and S. Klein (2008). "Alterations in adipose tissue and hepatic lipid kinetics in obese men and women with nonalcoholic fatty liver disease." Gastroenterology 134(2): 424-431.

Fabbrini, E., S. Sullivan and S. Klein (2010). "Obesity and nonalcoholic fatty liver disease: biochemical, metabolic, and clinical implications." Hepatology 51(2): 679-689.

Fan, J. G., J. Zhu, X. J. Li, L. Chen, L. Li, F. Dai, F. Li and S. Y. Chen (2005). "Prevalence of and risk factors for fatty liver in a general population of Shanghai, China." J Hepatol 43(3): 508-514.

Farghali, H., J. Hodis, N. Kutinova-Canova, P. Potmesil, E. Kmonickova and Z. Zidek (2008). "Glucose release as a response to glucagon in rat hepatocyte culture: involvement of NO signaling." Physiol Res 57(4): 569-575.

Fazakerley, D. J., R. Chaudhuri, P. Yang, G. J. Maghzal, K. C. Thomas, J. R. Krycer, S. J. Humphrey, B. L. Parker, K. H. Fisher-Wellman, C. C. Meoli, N. J. Hoffman, C. Diskin, J. G. Burchfield, M. J. Cowley, W. Kaplan, Z. Modrusan, G. Kolumam, J. Y. Yang, D. L. Chen, D. Samocha-Bonet, J. R. Greenfield, K. L. Hoehn, R. Stocker and D. E. James (2018). "Mitochondrial CoQ deficiency is a common driver of mitochondrial oxidants and insulin resistance." Elife 7.

Fazakerley, D. J., S. Naghiloo, R. Chaudhuri, F. Koumanov, J. G. Burchfield, K. C. Thomas, J. R. Krycer, M. J. Prior, B. L. Parker, B. A. Murrow, J. Stockli, C. C. Meoli, G. D. Holman and D. E. James (2015). "Proteomic Analysis of GLUT4 Storage Vesicles Reveals Tumor Suppressor Candidate 5 (TUSC5) as a Novel Regulator of Insulin Action in Adipocytes." Journal of Biological Chemistry 290(39): 23528-23542.

Febbraio, M., N. A. Abumrad, D. P. Hajjar, K. Sharma, W. Cheng, S. F. Pearce and R. L. Silverstein (1999). "A null mutation in murine CD36 reveals an important role in fatty acid and lipoprotein metabolism." J Biol Chem 274(27): 19055-19062.

Feingold, K. R. and C. Grunfeld (2000). Introduction to Lipids and Lipoproteins. Endotext. L. J. De Groot, G. Chrousos, K. Dungan et al. South Dartmouth (MA), MDText.com, Inc.

Festuccia, W. T., N. H. Kawashita, M. A. Garofalo, M. A. Moura, S. R. Brito, I. C. Kettelhut and R. H. Migliorini (2003). "Control of glyceroneogenic activity in rat brown adipose tissue." Am J Physiol Regul Integr Comp Physiol 285(1): R177-182.

Finley, D. (2009). "Recognition and processing of ubiquitin-protein conjugates by the proteasome." Annu Rev Biochem 78: 477-513.

195 Folch, J., M. Lees and G. H. Sloane Stanley (1957). "A simple method for the isolation and purification of total lipides from animal tissues." J Biol Chem 226(1): 497-509.

Foretz, M., C. Guichard, P. Ferre and F. Foufelle (1999). "Sterol regulatory element binding protein-1c is a major mediator of insulin action on the hepatic expression of glucokinase and lipogenesis-related genes." Proc Natl Acad Sci U S A 96(22): 12737- 12742.

Frayn, K. N. (2002). "Adipose tissue as a buffer for daily lipid flux." Diabetologia 45(9): 1201- 1210.

Fukumoto, H., S. Seino, H. Imura, Y. Seino, R. L. Eddy, Y. Fukushima, M. G. Byers, T. B. Shows and G. I. Bell (1988). "Sequence, tissue distribution, and chromosomal localization of mRNA encoding a human glucose transporter-like protein." Proc Natl Acad Sci U S A 85(15): 5434-5438.

Gagnon-Bartsch, J. A. and T. P. Speed (2012). "Using control genes to correct for unwanted variation in microarray data." Biostatistics 13(3): 539-552.

Gajda, A. M., Y. X. Zhou, L. B. Agellon, S. K. Fried, S. Kodukula, W. Fortson, K. Patel and J. Storch (2013). "Direct comparison of mice null for liver or intestinal fatty acid- binding proteins reveals highly divergent phenotypic responses to high fat feeding." J Biol Chem 288(42): 30330-30344.

Gao, Q. and J. M. Goodman (2015). "The lipid droplet-a well-connected organelle." Front Cell Dev Biol 3: 49.

Gao, Z., Z. Wang, X. Zhang, A. A. Butler, A. Zuberi, B. Gawronska-Kozak, M. Lefevre, D. York, E. Ravussin, H. R. Berthoud, O. McGuinness, W. T. Cefalu and J. Ye (2007). "Inactivation of PKCtheta leads to increased susceptibility to obesity and dietary insulin resistance in mice." Am J Physiol Endocrinol Metab 292(1): E84-91.

Garcia-Monzon, C., O. Lo Iacono, J. Crespo, M. Romero-Gomez, J. Garcia-Samaniego, M. Fernandez-Bermejo, A. Dominguez-Diez, J. Rodriguez de Cia, A. Saez, J. L. Porrero, J. Vargas-Castrillon, E. Chavez-Jimenez, S. Soto-Fernandez, A. Diaz, R. Gallego- Duran, A. Madejon and M. E. Miquilena-Colina (2014). "Increased soluble CD36 is linked to advanced steatosis in nonalcoholic fatty liver disease." Eur J Clin Invest 44(1): 65-73.

Gavin, J. R., 3rd, J. Roth, D. M. Neville, Jr., P. de Meyts and D. N. Buell (1974). "Insulin- dependent regulation of insulin receptor concentrations: a direct demonstration in cell culture." Proc Natl Acad Sci U S A 71(1): 84-88.

Gibbons, G. F. and D. Wiggins (1995). "Intracellular triacylglycerol lipase: its role in the assembly of hepatic very-low-density lipoprotein (VLDL)." Adv Enzyme Regul 35: 179-198.

Gong, J., Z. Sun and P. Li (2009). "CIDE proteins and metabolic disorders." Curr Opin Lipidol 20(2): 121-126.

196 Gong, J., Z. Sun, L. Wu, W. Xu, N. Schieber, D. Xu, G. Shui, H. Yang, R. G. Parton and P. Li (2011). "Fsp27 promotes lipid droplet growth by lipid exchange and transfer at lipid droplet contact sites." J Cell Biol 195(6): 953-963.

Goodpaster, B. H. (2013). "Mitochondrial deficiency is associated with insulin resistance." Diabetes 62(4): 1032-1035.

Gould, G. W., H. M. Thomas, T. J. Jess and G. I. Bell (1991). "Expression of human glucose transporters in Xenopus oocytes: kinetic characterization and substrate specificities of the erythrocyte, liver, and brain isoforms." Biochemistry 30(21): 5139-5145.

Graff, M., K. E. North, N. Franceschini, A. P. Reiner, M. Feitosa, J. J. Carr, P. Gordon-Larsen, M. K. Wojczynski and I. B. Borecki (2013). "PNPLA3 gene-by-visceral adipose tissue volume interaction and the pathogenesis of fatty liver disease: the NHLBI family heart study." Int J Obes (Lond) 37(3): 432-438.

Grahn, T. H., R. Kaur, J. Yin, M. Schweiger, V. M. Sharma, M. J. Lee, Y. Ido, C. M. Smas, R. Zechner, A. Lass and V. Puri (2014). "Fat-specific protein 27 (FSP27) interacts with adipose triglyceride lipase (ATGL) to regulate lipolysis and insulin sensitivity in human adipocytes." J Biol Chem 289(17): 12029-12039.

Granneman, J. G., H. P. Moore, E. P. Mottillo, Z. Zhu and L. Zhou (2011). "Interactions of perilipin-5 (Plin5) with adipose triglyceride lipase." J Biol Chem 286(7): 5126-5135.

Green, C. J. and L. Hodson (2014). "The influence of dietary fat on liver fat accumulation." Nutrients 6(11): 5018-5033.

Greenberg, A. S., R. A. Coleman, F. B. Kraemer, J. L. McManaman, M. S. Obin, V. Puri, Q. W. Yan, H. Miyoshi and D. G. Mashek (2011). "The role of lipid droplets in metabolic disease in rodents and humans." J Clin Invest 121(6): 2102-2110.

Greenfield, M. S., L. Doberne, M. Rosenthal, H. J. Vreman and G. M. Reaven (1982). "Lipid metabolism in non-insulin-dependent diabetes mellitus: effect of glipizide therapy." Arch Intern Med 142(8): 1498-1500.

Grippa, A., L. Buxo, G. Mora, C. Funaya, F. Z. Idrissi, F. Mancuso, R. Gomez, J. Muntanya, E. Sabido and P. Carvalho (2015). "The seipin complex Fld1/Ldb16 stabilizes ER- lipid droplet contact sites." J Cell Biol 211(4): 829-844.

Grzybek, M., A. Chorzalska, E. Bok, A. Hryniewicz-Jankowska, A. Czogalla, W. Diakowski and A. F. Sikorski (2006). "Spectrin-phospholipid interactions. Existence of multiple kinds of binding sites?" Chem Phys Lipids 141(1-2): 133-141.

Guzman, C., M. Benet, S. Pisonero-Vaquero, M. Moya, M. V. Garcia-Mediavilla, M. L. Martinez-Chantar, J. Gonzalez-Gallego, J. V. Castell, S. Sanchez-Campos and R. Jover (2013). "The human liver fatty acid binding protein (FABP1) gene is activated by FOXA1 and PPARalpha; and repressed by C/EBPalpha: Implications in FABP1 down-regulation in nonalcoholic fatty liver disease." Biochim Biophys Acta 1831(4): 803-818.

Ha, Y., N. Seo, J. H. Shim, S. Y. Kim, J. A. Park, S. Han, K. W. Kim, E. Yu, K. M. Kim, Y. S. Lim, H. C. Lee, Y. H. Chung and Y. S. Lee (2015). "Intimate association of visceral

197 obesity with non-alcoholic fatty liver disease in healthy Asians: A case-control study." J Gastroenterol Hepatol 30(11): 1666-1672.

Hamaguchi, M., T. Kojima, A. Ohbora, N. Takeda, M. Fukui and T. Kato (2012). "Aging is a risk factor of nonalcoholic fatty liver disease in premenopausal women." World J Gastroenterol 18(3): 237-243.

Hamilton, J. A. and D. M. Small (1981). "Solubilization and localization of triolein in phosphatidylcholine bilayers: a 13C NMR study." Proc Natl Acad Sci U S A 78(11): 6878-6882.

Han, S., D. D. Binns, Y. F. Chang and J. M. Goodman (2015). "Dissecting seipin function: the localized accumulation of phosphatidic acid at ER/LD junctions in the absence of seipin is suppressed by Sei1p(DeltaNterm) only in combination with Ldb16p." BMC Cell Biol 16: 29.

Hao, J., F. Li, W. Liu, Q. Liu, S. Liu, H. Li and H. Duan (2014). "Phosphorylation of PRAS40- Thr246 involved in renal lipid accumulation of diabetes." J Cell Physiol 229(8): 1069- 1077.

Hapala, I., E. Marza and T. Ferreira (2011). "Is fat so bad? Modulation of endoplasmic reticulum stress by lipid droplet formation." Biol Cell 103(6): 271-285.

Harbor, C. S. (2015). "Laemmli Sample Buffer (4x)." Cold Spring Harbor Protocols, from http://cshprotocols.cshlp.org/content/2015/8/pdb.rec084533.full?text_only=true.

Hashemi, H. F. and J. M. Goodman (2015). "The life cycle of lipid droplets." Curr Opin Cell Biol 33: 119-124.

Hashimoto, T., W. S. Cook, C. Qi, A. V. Yeldandi, J. K. Reddy and M. S. Rao (2000). "Defect in peroxisome proliferator-activated receptor alpha-inducible fatty acid oxidation determines the severity of hepatic steatosis in response to fasting." J Biol Chem 275(37): 28918-28928.

Haug, A. R. (2017). "Imaging of primary liver tumors with positron-emission tomography." Q J Nucl Med Mol Imaging 61(3): 292-300.

He, S., C. McPhaul, J. Z. Li, R. Garuti, L. Kinch, N. V. Grishin, J. C. Cohen and H. H. Hobbs (2010). "A sequence variation (I148M) in PNPLA3 associated with nonalcoholic fatty liver disease disrupts triglyceride hydrolysis." J Biol Chem 285(9): 6706-6715.

Heckmann, B. L., X. Zhang, A. M. Saarinen and J. Liu (2016). "Regulation of G0/G1 Switch Gene 2 (G0S2) Protein Ubiquitination and Stability by Triglyceride Accumulation and ATGL Interaction." PLoS One 11(6): e0156742.

Hershko, A. and A. Ciechanover (1998). "The ubiquitin system." Annu Rev Biochem 67: 425- 479.

Hewitt, N. J., M. J. Lechon, J. B. Houston, D. Hallifax, H. S. Brown, P. Maurel, J. G. Kenna, L. Gustavsson, C. Lohmann, C. Skonberg, A. Guillouzo, G. Tuschl, A. P. Li, E. LeCluyse, G. M. Groothuis and J. G. Hengstler (2007). "Primary hepatocytes: current understanding of the regulation of metabolic enzymes and transporter proteins, and

198 pharmaceutical practice for the use of hepatocytes in metabolism, enzyme induction, transporter, clearance, and hepatotoxicity studies." Drug Metab Rev 39(1): 159-234.

Hirano, Y., M. Yoshida, M. Shimizu and R. Sato (2001). "Direct demonstration of rapid degradation of nuclear sterol regulatory element-binding proteins by the ubiquitin- proteasome pathway." J Biol Chem 276(39): 36431-36437.

Hodson, L. and K. N. Frayn (2011). "Hepatic fatty acid partitioning." Curr Opin Lipidol 22(3): 216-224.

Hodson, L., S. E. McQuaid, S. M. Humphreys, R. Milne, B. A. Fielding, K. N. Frayn and F. Karpe (2010). "Greater dietary fat oxidation in obese compared with lean men: an adaptive mechanism to prevent liver fat accumulation?" Am J Physiol Endocrinol Metab 299(4): E584-592.

Holland, W. L., J. T. Brozinick, L. P. Wang, E. D. Hawkins, K. M. Sargent, Y. Liu, K. Narra, K. L. Hoehn, T. A. Knotts, A. Siesky, D. H. Nelson, S. K. Karathanasis, G. K. Fontenot, M. J. Birnbaum and S. A. Summers (2007). "Inhibition of ceramide synthesis ameliorates glucocorticoid-, saturated-fat-, and obesity-induced insulin resistance." Cell Metab 5(3): 167-179.

Holland, W. L. and S. A. Summers (2008). "Sphingolipids, insulin resistance, and metabolic disease: new insights from in vivo manipulation of sphingolipid metabolism." Endocr Rev 29(4): 381-402.

Horiguchi, Y., M. Araki and K. Motojima (2008). "17beta-Hydroxysteroid dehydrogenase type 13 is a liver-specific lipid droplet-associated protein." Biochem Biophys Res Commun 370(2): 235-238.

Horn, P. J., C. N. James, S. K. Gidda, A. Kilaru, J. M. Dyer, R. T. Mullen, J. B. Ohlrogge and K. D. Chapman (2013). "Identification of a new class of lipid droplet-associated proteins in plants." Physiol 162(4): 1926-1936.

Hornbeck, P. V. Z., B., Murray B., Kornhauser, J.M., Latham, V., Skrzypek, E. (2015). "PhosphoSitePlus, 2014: mutations, PTMs and recalibrations." Nucleic Acids Research 43: D512-520.

Hosoya, S., M. Watanabe, M. Ueno and E. Adachi (2014). "Hepatoblastoma-derived cells (HepG2) used to restore polarized hepatocyte morphology by culturing them in 3D collagen gels." The Kitasato Medical Journal 44(2): 139-147.

Hoy, A. J., C. R. Bruce, S. M. Turpin, A. J. Morris, M. A. Febbraio and M. J. Watt (2011). "Adipose triglyceride lipase-null mice are resistant to high-fat diet-induced insulin resistance despite reduced energy expenditure and ectopic lipid accumulation." Endocrinology 152(1): 48-58.

Huang, H., B. P. Atshaves, A. Frolov, A. B. Kier and F. Schroeder (2005). "Acyl-coenzyme A binding protein expression alters liver fatty acyl-coenzyme A metabolism." Biochemistry 44(30): 10282-10297.

Humphrey, S. J., D. E. James and M. Mann (2015). "Protein Phosphorylation: A Major Switch Mechanism for Metabolic Regulation." Trends Endocrinol Metab 26(12): 676-687.

199 Humphrey, S. J., G. Yang, P. Yang, D. J. Fazakerley, J. Stockli, J. Y. Yang and D. E. James (2013). "Dynamic adipocyte phosphoproteome reveals that Akt directly regulates mTORC2." Cell Metab 17(6): 1009-1020.

Hunter, T. and B. M. Sefton (1980). "Transforming gene product of Rous sarcoma virus phosphorylates tyrosine." Proc Natl Acad Sci U S A 77(3): 1311-1315.

Ibdah, J. A., P. Perlegas, Y. Zhao, J. Angdisen, H. Borgerink, M. K. Shadoan, J. D. Wagner, D. Matern, P. Rinaldo and J. M. Cline (2005). "Mice heterozygous for a defect in mitochondrial trifunctional protein develop hepatic steatosis and insulin resistance." Gastroenterology 128(5): 1381-1390.

Imai, N., M. Suzuki, K. Hayashi, M. Ishigami, Y. Hirooka, T. Abe, G. Shioi, H. Goto and T. Fujimoto (2015). "Hepatocyte-Specific Depletion of UBXD8 Induces Periportal Steatosis in Mice Fed a High-Fat Diet." PLoS One 10(5): e0127114.

Ipsen, D. H., J. Lykkesfeldt and P. Tveden-Nyborg (2018). "Molecular mechanisms of hepatic lipid accumulation in non-alcoholic fatty liver disease." Cell Mol Life Sci 75(18): 3313-3327.

Iqbal, J. and M. M. Hussain (2009). "Intestinal lipid absorption." Am J Physiol Endocrinol Metab 296(6): E1183-1194.

Ishii, M., A. Maeda, S. Tani and M. Akagawa (2015). "Palmitate induces insulin resistance in human HepG2 hepatocytes by enhancing ubiquitination and proteasomal degradation of key insulin signaling molecules." Arch Biochem Biophys 566: 26-35.

Itabe, H., T. Yamaguchi, S. Nimura and N. Sasabe (2017). "Perilipins: a diversity of intracellular lipid droplet proteins." Lipids Health Dis 16(1): 83.

Jackson, K. E., N. Klonis, D. J. Ferguson, A. Adisa, C. Dogovski and L. Tilley (2004). "Food vacuole-associated lipid bodies and heterogeneous lipid environments in the malaria parasite, Plasmodium falciparum." Mol Microbiol 54(1): 109-122.

Jacquier, N., V. Choudhary, M. Mari, A. Toulmay, F. Reggiori and R. Schneiter (2011). "Lipid droplets are functionally connected to the endoplasmic reticulum in Saccharomyces cerevisiae." J Cell Sci 124(Pt 14): 2424-2437.

Jagerstrom, S., S. Polesie, Y. Wickstrom, B. R. Johansson, H. D. Schroder, K. Hojlund and P. Bostrom (2009). "Lipid droplets interact with mitochondria using SNAP23." Cell Biol Int 33(9): 934-940.

Janmey, P. A. and U. Lindberg (2004). "Cytoskeletal regulation: rich in lipids." Nat Rev Mol Cell Biol 5(8): 658-666.

Jiang, Z. Y., Q. L. Zhou, J. Holik, S. Patel, J. Leszyk, K. Coleman, M. Chouinard and M. P. Czech (2005). "Identification of WNK1 as a substrate of Akt/protein kinase B and a negative regulator of insulin-stimulated mitogenesis in 3T3-L1 cells." J Biol Chem 280(22): 21622-21628.

200 Jin, Y., P. J. McFie, S. L. Banman, C. Brandt and S. J. Stone (2014). "Diacylglycerol acyltransferase-2 (DGAT2) and monoacylglycerol acyltransferase-2 (MGAT2) interact to promote triacylglycerol synthesis." J Biol Chem 289(41): 28237-28248.

Jin, Y., Y. Tan, L. Chen, Y. Liu and Z. Ren (2018). "Reactive Oxygen Species Induces Lipid Droplet Accumulation in HepG2 Cells by Increasing Perilipin 2 Expression." Int J Mol Sci 19(11).

Jo, Y. and R. A. Debose-Boyd (2010). "Control of cholesterol synthesis through regulated ER- associated degradation of HMG CoA reductase." Crit Rev Biochem Mol Biol 45(3): 185-198.

Jocken, J. W., D. Langin, E. Smit, W. H. Saris, C. Valle, G. B. Hul, C. Holm, P. Arner and E. E. Blaak (2007). "Adipose triglyceride lipase and hormone-sensitive lipase protein expression is decreased in the obese insulin-resistant state." J Clin Endocrinol Metab 92(6): 2292-2299.

Jolivet, P., E. Roux, S. D'Andrea, M. Davanture, L. Negroni, M. Zivy and T. Chardot (2004). "Protein composition of oil bodies in Arabidopsis thaliana ecotype WS." Plant Physiol Biochem 42(6): 501-509.

Kanety, H., R. Hemi, M. Z. Papa and A. Karasik (1996). "Sphingomyelinase and ceramide suppress insulin-induced tyrosine phosphorylation of the insulin receptor substrate- 1." J Biol Chem 271(17): 9895-9897.

Kantartzis, K., F. Machicao, J. Machann, F. Schick, A. Fritsche, H. U. Haring and N. Stefan (2009). "The DGAT2 gene is a candidate for the dissociation between fatty liver and insulin resistance in humans." Clin Sci (Lond) 116(6): 531-537.

Kato, M., N. Higuchi and M. Enjoji (2008). "Reduced hepatic expression of adipose tissue triglyceride lipase and CGI-58 may contribute to the development of non-alcoholic fatty liver disease in patients with insulin resistance." Scand J Gastroenterol 43(8): 1018-1019.

Kaushik, S. and A. M. Cuervo (2016). "AMPK-dependent phosphorylation of lipid droplet protein PLIN2 triggers its degradation by CMA." Autophagy 12(2): 432-438.

Khan, S. A., E. E. Wollaston-Hayden, T. W. Markowski, L. Higgins and D. G. Mashek (2015). "Quantitative analysis of the murine lipid droplet-associated proteome during diet- induced hepatic steatosis." J Lipid Res 56(12): 2260-2272.

Khatchadourian, A. and D. Maysinger (2009). "Lipid droplets: their role in nanoparticle- induced oxidative stress." Mol Pharm 6(4): 1125-1137.

Khodadoostan, M., B. Shariatifar, N. Motamedi and H. Abdolahi (2016). "Comparison of liver enzymes level and sonographic findings value with liver biopsy findings in nonalcoholic fatty liver disease patients." Adv Biomed Res 5: 40.

Khor, V. K., R. Ahrends, Y. Lin, W. J. Shen, C. M. Adams, A. N. Roseman, Y. Cortez, M. N. Teruel, S. Azhar and F. B. Kraemer (2014). "The proteome of cholesteryl-ester- enriched versus triacylglycerol-enriched lipid droplets." PLoS One 9(8): e105047.

201 Khor, V. K., W. J. Shen and F. B. Kraemer (2013). "Lipid droplet metabolism." Curr Opin Clin Nutr Metab Care 16(6): 632-637.

Khoshnan, A., D. Bae, C. A. Tindell and A. E. Nel (2000). "The physical association of protein kinase C theta with a lipid raft-associated inhibitor of kappa B factor kinase (IKK) complex plays a role in the activation of the NF-kappa B cascade by TCR and CD28." J Immunol 165(12): 6933-6940.

Kienesberger, P. C., D. Lee, T. Pulinilkunnil, D. S. Brenner, L. Cai, C. Magnes, H. C. Koefeler, I. E. Streith, G. N. Rechberger, G. Haemmerle, J. S. Flier, R. Zechner, Y. B. Kim and E. E. Kershaw (2009). "Adipose triglyceride lipase deficiency causes tissue-specific changes in insulin signaling." J Biol Chem 284(44): 30218-30229.

Kim, H., H. Zhang, D. Meng, G. Russell, J. N. Lee and J. Ye (2013). "UAS domain of Ubxd8 and FAF1 polymerizes upon interaction with long-chain unsaturated fatty acids." J Lipid Res 54(8): 2144-2152.

Kim, J. K., J. J. Fillmore, Y. Chen, C. Yu, I. K. Moore, M. Pypaert, E. P. Lutz, Y. Kako, W. Velez-Carrasco, I. J. Goldberg, J. L. Breslow and G. I. Shulman (2001). "Tissue- specific overexpression of causes tissue-specific insulin resistance." Proc Natl Acad Sci U S A 98(13): 7522-7527.

Kim, J. W. and Y. H. Ahn (1998). "CCAAT/enhancer binding protein regulates the promoter activity of the rat GLUT2 glucose transporter gene in liver cells." Biochem J 336 ( Pt 1): 83-90.

Kim, J. Y., A. Nasr, H. Tfayli, F. Bacha, S. F. Michaliszyn and S. Arslanian (2017). "Increased Lipolysis, Diminished Adipose Tissue Insulin Sensitivity, and Impaired beta-Cell Function Relative to Adipose Tissue Insulin Sensitivity in Obese Youth With Impaired Glucose Tolerance." Diabetes 66(12): 3085-3090.

Kim, K. A., S. O. Ka, W. S. Moon, H. K. Yi, Y. H. Lee, K. B. Kwon, J. W. Park and B. H. Park (2008). "Effect of dermcidin, an antimicrobial peptide, on body fat mobilization in normal mice." J Endocrinol 198(1): 111-118.

Kim, K. H., Y. T. Jeong, H. Oh, S. H. Kim, J. M. Cho, Y. N. Kim, S. S. Kim, D. H. Kim, K. Y. Hur, H. K. Kim, T. Ko, J. Han, H. L. Kim, J. Kim, S. H. Back, M. Komatsu, H. Chen, D. C. Chan, M. Konishi, N. Itoh, C. S. Choi and M. S. Lee (2013). "Autophagy deficiency leads to protection from obesity and insulin resistance by inducing Fgf21 as a mitokine." Nat Med 19(1): 83-92.

Kim, S. J., T. Tang, M. Abbott, J. A. Viscarra, Y. Wang and H. S. Sul (2016). "AMPK Phosphorylates Desnutrin/ATGL and Hormone-Sensitive Lipase To Regulate Lipolysis and Fatty Acid Oxidation within Adipose Tissue." Mol Cell Biol 36(14): 1961-1976.

Kim, S. K., K. J. Woodcroft and R. F. Novak (2003). "Insulin and glucagon regulation of glutathione S-transferase expression in primary cultured rat hepatocytes." J Pharmacol Exp Ther 305(1): 353-361.

Kimmel, A. R. and C. Sztalryd (2014). "Perilipin 5, a lipid droplet protein adapted to mitochondrial energy utilization." Curr Opin Lipidol 25(2): 110-117.

202 Kioka, H., H. Kato, M. Fujikawa, O. Tsukamoto, T. Suzuki, H. Imamura, A. Nakano, S. Higo, S. Yamazaki, T. Matsuzaki, K. Takafuji, H. Asanuma, M. Asakura, T. Minamino, Y. Shintani, M. Yoshida, H. Noji, M. Kitakaze, I. Komuro, Y. Asano and S. Takashima (2014). "Evaluation of intramitochondrial ATP levels identifies G0/G1 switch gene 2 as a positive regulator of oxidative phosphorylation." Proc Natl Acad Sci U S A 111(1): 273-278.

Klecker, T., R. J. Braun and B. Westermann (2017). "Lipid Droplets Guard Mitochondria during Autophagy." Dev Cell 42(1): 1-2.

Klein, S. and K. N. Jeejeebhoy (2002). "The malnourished patient: Nutritional assessment and management." Sleisenger & Fordtran’s Gastrointestinal and Liver Disease: 265-285.

Kleiner, D. E., E. M. Brunt, M. Van Natta, C. Behling, M. J. Contos, O. W. Cummings, L. D. Ferrell, Y. C. Liu, M. S. Torbenson, A. Unalp-Arida, M. Yeh, A. J. McCullough and A. J. Sanyal (2005). "Design and validation of a histological scoring system for nonalcoholic fatty liver disease." Hepatology 41(6): 1313-1321.

Klemm, R. W., J. P. Norton, R. A. Cole, C. S. Li, S. H. Park, M. M. Crane, L. Li, D. Jin, A. Boye-Doe, T. Y. Liu, Y. Shibata, H. Lu, T. A. Rapoport, R. V. Farese, Jr., C. Blackstone, Y. Guo and H. Y. Mak (2013). "A conserved role for atlastin GTPases in regulating lipid droplet size." Cell Rep 3(5): 1465-1475.

Knowles, B. B. and D. P. Aden (1983). Human hepatoma derived cell line, process for preparation thereof, and uses therefor U. S. P. a. T. Office, The Wistar Institute of Anatomy and Biology, Philadelphia, Pa. . 4,393,133.

Kochan, K., E. Maslak, C. Krafft, R. Kostogrys, S. Chlopicki and M. Baranska (2015). "Raman spectroscopy analysis of lipid droplets content, distribution and saturation level in Non-Alcoholic Fatty Liver Disease in mice." J Biophotonics 8(7): 597-609.

Komatsu, M., S. Waguri, T. Chiba, S. Murata, J. Iwata, I. Tanida, T. Ueno, M. Koike, Y. Uchiyama, E. Kominami and K. Tanaka (2006). "Loss of autophagy in the central nervous system causes neurodegeneration in mice." Nature 441(7095): 880-884.

Konige, M., H. Wang and C. Sztalryd (2014). "Role of adipose specific lipid droplet proteins in maintaining whole body energy homeostasis." Biochim Biophys Acta 1842(3): 393-401.

Kory, N., R. V. Farese, Jr. and T. C. Walther (2016). "Targeting Fat: Mechanisms of Protein Localization to Lipid Droplets." Trends Cell Biol 26(7): 535-546.

Kotronen, A., T. Seppanen-Laakso, J. Westerbacka, T. Kiviluoto, J. Arola, A. L. Ruskeepaa, M. Oresic and H. Yki-Jarvinen (2009). "Hepatic stearoyl-CoA desaturase (SCD)-1 activity and diacylglycerol but not ceramide concentrations are increased in the nonalcoholic human fatty liver." Diabetes 58(1): 203-208.

Krahmer, N., R. V. Farese, Jr. and T. C. Walther (2013). "Balancing the fat: lipid droplets and human disease." EMBO Mol Med 5(7): 973-983.

Krahmer, N., Y. Guo, F. Wilfling, M. Hilger, S. Lingrell, K. Heger, H. W. Newman, M. Schmidt-Supprian, D. E. Vance, M. Mann, R. V. Farese, Jr. and T. C. Walther (2011).

203 "Phosphatidylcholine synthesis for lipid droplet expansion is mediated by localized activation of CTP:phosphocholine cytidylyltransferase." Cell Metab 14(4): 504-515.

Krahmer, N., M. Hilger, N. Kory, F. Wilfling, G. Stoehr, M. Mann, R. V. Farese, Jr. and T. C. Walther (2013). "Protein correlation profiles identify lipid droplet proteins with high confidence." Mol Cell Proteomics 12(5): 1115-1126.

Krahmer, N., B. Najafi, F. Schueder, F. Quagliarini, M. Steger, S. Seitz, R. Kasper, F. Salinas, J. Cox, N. H. Uhlenhaut, T. C. Walther, R. Jungmann, A. Zeigerer, G. H. H. Borner and M. Mann (2018). "Organellar Proteomics and Phospho-Proteomics Reveal Subcellular Reorganization in Diet-Induced Hepatic Steatosis." Dev Cell 47(2): 205- 221.e207.

Kramer, D. A., A. D. Quiroga, J. Lian, R. P. Fahlman and R. Lehner (2018). "Fasting and refeeding induces changes in the mouse hepatic lipid droplet proteome." J Proteomics 181: 213-224.

Kruger, M., I. Kratchmarova, B. Blagoev, Y. H. Tseng, C. R. Kahn and M. Mann (2008). "Dissection of the insulin signaling pathway via quantitative phosphoproteomics." Proc Natl Acad Sci U S A 105(7): 2451-2456.

Kumar, R. and S. Mohan (2017). "Non-alcoholic Fatty Liver Disease in Lean Subjects: Characteristics and Implications." J Clin Transl Hepatol 5(3): 216-223.

Kumashiro, N., D. M. Erion, D. Zhang, M. Kahn, S. A. Beddow, X. Chu, C. D. Still, G. S. Gerhard, X. Han, J. Dziura, K. F. Petersen, V. T. Samuel and G. I. Shulman (2011). "Cellular mechanism of insulin resistance in nonalcoholic fatty liver disease." Proc Natl Acad Sci U S A 108(39): 16381-16385.

Kuo, Y. T., T. H. Lin, W. L. Chen and H. M. Lee (2012). "Alpha-lipoic acid induces adipose triglyceride lipase expression and decreases intracellular lipid accumulation in HepG2 cells." Eur J Pharmacol 692(1-3): 10-18.

Lambert, J. E. and E. J. Parks (2012). "Postprandial metabolism of meal triglyceride in humans." Biochim Biophys Acta 1821(5): 721-726.

Lass, A., R. Zimmermann, M. Oberer and R. Zechner (2011). "Lipolysis - a highly regulated multi-enzyme complex mediates the catabolism of cellular fat stores." Prog Lipid Res 50(1): 14-27.

Lau, J. K., X. Zhang and J. Yu (2017). "Animal models of non-alcoholic fatty liver disease: current perspectives and recent advances." J Pathol 241(1): 36-44.

Lauzier, B., C. Merlen, F. Vaillant, J. McDuff, B. Bouchard, P. C. Beguin, V. W. Dolinsky, S. Foisy, L. R. Villeneuve, F. Labarthe, J. R. Dyck, B. G. Allen, G. Charron and C. Des Rosiers (2011). "Post-translational modifications, a key process in CD36 function: lessons from the spontaneously hypertensive rat heart." J Mol Cell Cardiol 51(1): 99- 108.

Lee, J. N., H. Kim, H. Yao, Y. Chen, K. Weng and J. Ye (2010). "Identification of Ubxd8 protein as a sensor for unsaturated fatty acids and regulator of triglyceride synthesis." Proc Natl Acad Sci U S A 107(50): 21424-21429.

204 Lee, J. N., X. Zhang, J. D. Feramisco, Y. Gong and J. Ye (2008). "Unsaturated fatty acids inhibit proteasomal degradation of Insig-1 at a postubiquitination step." J Biol Chem 283(48): 33772-33783.

Lee, R. G., M. C. Willingham, M. A. Davis, K. A. Skinner and L. L. Rudel (2000). "Differential expression of ACAT1 and ACAT2 among cells within liver, intestine, kidney, and adrenal of nonhuman ." J Lipid Res 41(12): 1991-2001.

Levy, M. and A. H. Futerman (2010). "Mammalian ceramide synthases." IUBMB Life 62(5): 347-356.

Lewis, G. F., B. Zinman, Y. Groenewoud, M. Vranic and A. Giacca (1996). "Hepatic glucose production is regulated both by direct hepatic and extrahepatic effects of insulin in humans." Diabetes 45(4): 454-462.

Li, C., L. Li, J. Lian, R. Watts, R. Nelson, B. Goodwin and R. Lehner (2015). "Roles of Acyl- CoA:Diacylglycerol Acyltransferases 1 and 2 in Triacylglycerol Synthesis and Secretion in Primary Hepatocytes." Arterioscler Thromb Vasc Biol 35(5): 1080-1091.

Li, C., X. Luo, S. Zhao, G. K. Siu, Y. Liang, H. C. Chan, A. Satoh and S. S. Yu (2017). "COPI- TRAPPII activates Rab18 and regulates its lipid droplet association." Embo j 36(4): 441-457.

Li, L., D. W. Liu, H. Y. Yan, Z. Y. Wang, S. H. Zhao and B. Wang (2016). "Obesity is an independent risk factor for non-alcoholic fatty liver disease: evidence from a meta- analysis of 21 cohort studies." Obes Rev 17(6): 510-519.

Li, X., M. Xu, F. Wang, A. B. Kohan, M. K. Haas, Q. Yang, D. Lou, S. Obici, W. S. Davidson and P. Tso (2014). "Apolipoprotein A-IV reduces hepatic gluconeogenesis through nuclear receptor NR1D1." J Biol Chem 289(4): 2396-2404.

Li, Z., K. Thiel, P. J. Thul, M. Beller, R. P. Kuhnlein and M. A. Welte (2012). "Lipid droplets control the maternal histone supply of Drosophila embryos." Curr Biol 22(22): 2104- 2113.

Libby, A. E., E. Bales, D. J. Orlicky and J. L. McManaman (2016). "Perilipin-2 Deletion Impairs Hepatic Lipid Accumulation by Interfering with Sterol Regulatory Element- binding Protein (SREBP) Activation and Altering the Hepatic Lipidome." J Biol Chem 291(46): 24231-24246.

Lin, J. L., J. Zhu and I. Wheeldon (2017). "Synthetic Protein Scaffolds for Biosynthetic Pathway Colocalization on Lipid Droplet Membranes." ACS Synth Biol 6(8): 1534- 1544.

Linton, M. F., P. G. Yancey, S. S. Davies, W. G. J. Jerome, E. F. Linton and K. C. Vickers (2000). The Role of Lipids and Lipoproteins in Atherosclerosis. Endotext. L. J. De Groot, G. Chrousos, K. Dungan et al. South Dartmouth (MA), MDText.com, Inc.

Listenberger, L. L., X. Han, S. E. Lewis, S. Cases, R. V. Farese, Jr., D. S. Ory and J. E. Schaffer (2003). "Triglyceride accumulation protects against fatty acid-induced lipotoxicity." Proc Natl Acad Sci U S A 100(6): 3077-3082.

205 Liu, H. Y., J. Han, S. Y. Cao, T. Hong, D. Zhuo, J. Shi, Z. Liu and W. Cao (2009). "Hepatic autophagy is suppressed in the presence of insulin resistance and hyperinsulinemia: inhibition of FoxO1-dependent expression of key autophagy genes by insulin." J Biol Chem 284(45): 31484-31492.

Liu, M., R. Ge, W. Liu, Q. Liu, X. Xia, M. Lai, L. Liang, C. Li, L. Song, B. Zhen, J. Qin and C. Ding (2017). "Differential proteomics profiling identifies LDPs and biological functions in high-fat diet-induced fatty livers." J Lipid Res 58(4): 681-694.

Liu, M., R. Ge, W. Liu, Q. Liu, X. Xia, M. Lai, L. Liang, C. Li, L. Song, B. Zhen, J. Qin and C. Ding (2017). "Differential proteomics profiling identifies lipid droplet proteins and biological functions in high fat diet induced fatty livers." Journal of Lipid Research.

Liu, Y., A. Beyer and R. Aebersold (2016). "On the Dependency of Cellular Protein Levels on mRNA Abundance." Cell 165(3): 535-550.

Long, Y. C. and J. R. Zierath (2006). "AMP-activated protein kinase signaling in metabolic regulation." J Clin Invest 116(7): 1776-1783.

Loregger, A., M. Raaben, J. Tan, S. Scheij, M. Moeton, M. van den Berg, H. Gelberg-Etel, E. Stickel, J. Roitelman, T. Brummelkamp and N. Zelcer (2017). "Haploid Mammalian Genetic Screen Identifies UBXD8 as a Key Determinant of HMGCR Degradation and Cholesterol Biosynthesis." Arterioscler Thromb Vasc Biol 37(11): 2064-2074.

Lu, X., X. Yang and J. Liu (2010). "Differential control of ATGL-mediated lipid droplet degradation by CGI-58 and G0S2." Cell Cycle 9(14): 2719-2725.

Luiken, J. J., D. Chanda, M. Nabben, D. Neumann and J. F. Glatz (2016). "Post-translational modifications of CD36 (SR-B2): Implications for regulation of myocellular fatty acid uptake." Biochim Biophys Acta 1862(12): 2253-2258.

Luiken, J. J., D. J. Dyck, X. X. Han, N. N. Tandon, Y. Arumugam, J. F. Glatz and A. Bonen (2002). "Insulin induces the translocation of the fatty acid transporter FAT/CD36 to the plasma membrane." Am J Physiol Endocrinol Metab 282(2): E491-495.

Lundby, A., K. Lage, B. T. Weinert, D. B. Bekker-Jensen, A. Secher, T. Skovgaard, C. D. Kelstrup, A. Dmytriyev, C. Choudhary, C. Lundby and J. V. Olsen (2012). "Proteomic analysis of lysine acetylation sites in rat tissues reveals organ specificity and subcellular patterns." Cell Rep 2(2): 419-431.

Lundby, A., A. Secher, K. Lage, N. B. Nordsborg, A. Dmytriyev, C. Lundby and J. V. Olsen (2012). "Quantitative maps of protein phosphorylation sites across 14 different rat organs and tissues." Nat Commun 3: 876.

Luo, P., C. Qin, L. Zhu, C. Fang, Y. Zhang, H. Zhang, F. Pei, S. Tian, X. Y. Zhu, J. Gong, Q. Mao, C. Xiao, Y. Su, H. Zheng, T. Xu, J. Lu and J. Zhang (2018). "Ubiquitin-Specific Peptidase 10 (USP10) Inhibits Hepatic Steatosis, Insulin Resistance, and Inflammation Through Sirt6." Hepatology.

Luukkonen, P. K., Y. Zhou, S. Sadevirta, M. Leivonen, J. Arola, M. Oresic, T. Hyotylainen and H. Yki-Jarvinen (2016). "Hepatic ceramides dissociate steatosis and insulin

206 resistance in patients with non-alcoholic fatty liver disease." J Hepatol 64(5): 1167- 1175.

Ma, R., W. Zhang, K. Tang, H. Zhang, Y. Zhang, D. Li, Y. Li, P. Xu, S. Luo, W. Cai, T. Ji, F. Katirai, D. Ye and B. Huang (2013). "Switch of glycolysis to gluconeogenesis by dexamethasone for treatment of hepatocarcinoma." Nat Commun 4: 2508.

Machado, M. V. and H. Cortez-Pinto (2013). "Non-invasive diagnosis of non-alcoholic fatty liver disease. A critical appraisal." J Hepatol 58(5): 1007-1019.

MacPherson, R. E., S. V. Ramos, R. Vandenboom, B. D. Roy and S. J. Peters (2013). "Skeletal muscle PLIN proteins, ATGL and CGI-58, interactions at rest and following stimulated contraction." Am J Physiol Regul Integr Comp Physiol 304(8): R644-650.

Mahady, S. E. and L. A. Adams (2018). "Burden of non-alcoholic fatty liver disease in Australia." J Gastroenterol Hepatol 33 Suppl 1: 1-11.

Mahley, R. W., T. L. Innerarity, S. C. Rall, Jr. and K. H. Weisgraber (1984). "Plasma lipoproteins: apolipoprotein structure and function." J Lipid Res 25(12): 1277-1294.

Maksymiw, R., S. F. Sui, H. Gaub and E. Sackmann (1987). "Electrostatic coupling of spectrin dimers to phosphatidylserine containing lipid lamellae." Biochemistry 26(11): 2983- 2990.

Mannaerts, G. P., L. J. Debeer, J. Thomas and P. J. De Schepper (1979). "Mitochondrial and peroxisomal fatty acid oxidation in liver homogenates and isolated hepatocytes from control and clofibrate-treated rats." J Biol Chem 254(11): 4585-4595.

Marbani, S. L. and J. Roth (1996). "Transgenic hyperinsulinemia: A mouse model of insulin resistance and glucose intolerance without obesity." Lessons from Animal Diabetes VI. Ed. E. Shafrir 6.

Marchesini, G., M. Brizi, A. M. Morselli-Labate, G. Bianchi, E. Bugianesi, A. J. McCullough, G. Forlani and N. Melchionda (1999). "Association of nonalcoholic fatty liver disease with insulin resistance." Am J Med 107(5): 450-455.

Marchesini, G., S. Moscatiello, S. Di Domizio and G. Forlani (2008). "Obesity-associated liver disease." J Clin Endocrinol Metab 93(11 Suppl 1): S74-80.

Mashek, D. G. (2013). "Hepatic fatty acid trafficking: multiple forks in the road." Adv Nutr 4(6): 697-710.

Mashek, D. G., L. O. Li and R. A. Coleman (2007). "Long-chain acyl-CoA synthetases and fatty acid channeling." Future Lipidol 2(4): 465-476.

Matsumoto, M., S. Han, T. Kitamura and D. Accili (2006). "Dual role of transcription factor FoxO1 in controlling hepatic insulin sensitivity and lipid metabolism." J Clin Invest 116(9): 2464-2472.

Matsumoto, M., A. Pocai, L. Rossetti, R. A. Depinho and D. Accili (2007). "Impaired regulation of hepatic glucose production in mice lacking the forkhead transcription factor Foxo1 in liver." Cell Metab 6(3): 208-216.

207 McIntosh, A. L., S. Senthivinayagam, K. C. Moon, S. Gupta, J. S. Lwande, C. C. Murphy, S. M. Storey and B. P. Atshaves (2012). "Direct interaction of Plin2 with lipids on the surface of lipid droplets: a live cell FRET analysis." Am J Physiol Cell Physiol 303(7): C728-742.

McManaman, J. L., E. S. Bales, D. J. Orlicky, M. Jackman, P. S. MacLean, S. Cain, A. E. Crunk, A. Mansur, C. E. Graham, T. A. Bowman and A. S. Greenberg (2013). "Perilipin-2-null mice are protected against diet-induced obesity, adipose inflammation, and fatty liver disease." J Lipid Res 54(5): 1346-1359.

McMullen, P. D., S. Bhattacharya, C. G. Woods, B. Sun, K. Yarborough, S. M. Ross, M. E. Miller, M. T. McBride, E. L. LeCluyse, R. A. Clewell and M. E. Andersen (2014). "A map of the PPARalpha transcription regulatory network for primary human hepatocytes." Chem Biol Interact 209: 14-24.

Mertins, P., D. R. Mani, K. V. Ruggles, M. A. Gillette, K. R. Clauser, P. Wang, X. Wang, J. W. Qiao, S. Cao, F. Petralia, E. Kawaler, F. Mundt, K. Krug, Z. Tu, J. T. Lei, M. L. Gatza, M. Wilkerson, C. M. Perou, V. Yellapantula, K. L. Huang, C. Lin, M. D. McLellan, P. Yan, S. R. Davies, R. R. Townsend, S. J. Skates, J. Wang, B. Zhang, C. R. Kinsinger, M. Mesri, H. Rodriguez, L. Ding, A. G. Paulovich, D. Fenyo, M. J. Ellis and S. A. Carr (2016). "Proteogenomics connects somatic mutations to signalling in breast cancer." Nature 534(7605): 55-62.

Michael, M. D., R. N. Kulkarni, C. Postic, S. F. Previs, G. I. Shulman, M. A. Magnuson and C. R. Kahn (2000). "Loss of insulin signaling in hepatocytes leads to severe insulin resistance and progressive hepatic dysfunction." Mol Cell 6(1): 87-97.

Milosevic, A., T. Liebmann, M. Knudsen, N. Schintu, P. Svenningsson and P. Greengard (2017). "Cell- and region-specific expression of depression-related protein p11 (S100a10) in the brain." J Comp Neurol 525(4): 955-975.

Minguez, P., L. Parca, F. Diella, D. R. Mende, R. Kumar, M. Helmer-Citterich, A. C. Gavin, V. van Noort and P. Bork (2012). "Deciphering a global network of functionally associated post-translational modifications." Mol Syst Biol 8: 599.

Mirza, M. S. (2011). "Obesity, Visceral Fat, and NAFLD: Querying the Role of Adipokines in the Progression of Nonalcoholic Fatty Liver Disease." ISRN Gastroenterol 2011: 592404.

Miyoshi, H., S. C. Souza, H. H. Zhang, K. J. Strissel, M. A. Christoffolete, J. Kovsan, A. Rudich, F. B. Kraemer, A. C. Bianco, M. S. Obin and A. S. Greenberg (2006). "Perilipin promotes hormone-sensitive lipase-mediated adipocyte lipolysis via phosphorylation-dependent and -independent mechanisms." J Biol Chem 281(23): 15837-15844.

Mohan, V., S. Farooq, M. Deepa, R. Ravikumar and C. S. Pitchumoni (2009). "Prevalence of non-alcoholic fatty liver disease in urban south Indians in relation to different grades of glucose intolerance and metabolic syndrome." Diabetes Res Clin Pract 84(1): 84- 91.

208 Molenaar, M. R., A. B. Vaandrager and J. B. Helms (2017). "Some Lipid Droplets Are More Equal Than Others: Different Metabolic Lipid Droplet Pools in Hepatic Stellate Cells." Lipid Insights 10: 1178635317747281.

Monetti, M., M. C. Levin, M. J. Watt, M. P. Sajan, S. Marmor, B. K. Hubbard, R. D. Stevens, J. R. Bain, C. B. Newgard, R. V. Farese, Sr., A. L. Hevener and R. V. Farese, Jr. (2007). "Dissociation of hepatic steatosis and insulin resistance in mice overexpressing DGAT in the liver." Cell Metab 6(1): 69-78.

Montgomery, M. K. and N. Turner (2015). "Mitochondrial dysfunction and insulin resistance: an update." Endocr Connect 4(1): R1-r15.

Moon, Y. A., R. E. Hammer and J. D. Horton (2009). "Deletion of ELOVL5 leads to fatty liver through activation of SREBP-1c in mice." J Lipid Res 50(3): 412-423.

Moore, M. C., K. C. Coate, J. J. Winnick, Z. An and A. D. Cherrington (2012). "Regulation of hepatic glucose uptake and storage in vivo." Adv Nutr 3(3): 286-294.

Mortimore, G. E. (1963). "Effect of insulin on release of glucose and urea by isolated rat liver." Am J Physiol 204(4): 699-704.

Moshayedi, H., R. Ahrabi, A. Mardani, S. Sadigetegad and M. Farhudi (2014). "Association between non-alcoholic fatty liver disease and ischemic stroke." Iran J Neurol 13(3): 144-148.

Mounier, C. and B. I. Posner (2006). "Transcriptional regulation by insulin: from the receptor to the gene." Can J Physiol Pharmacol 84(7): 713-724.

Murphy, S., S. Martin and R. G. Parton (2009). "Lipid droplet-organelle interactions; sharing the fats." Biochim Biophys Acta 1791(6): 441-447.

Nakamura, A. and Y. Terauchi (2013). "Lessons from mouse models of high-fat diet-induced NAFLD." Int J Mol Sci 14(11): 21240-21257.

Nassir, F. and J. A. Ibdah (2014). "Role of mitochondria in nonalcoholic fatty liver disease." Int J Mol Sci 15(5): 8713-8742.

Natarajan, S. K., K. Rasineni, M. Ganesan, D. Feng, B. L. McVicker, M. A. McNiven, N. A. Osna, J. L. Mott, C. A. Casey and K. K. Kharbanda (2017). "Structure, Function and Metabolism of Hepatic and Adipose Tissue Lipid Droplets: Implications in Alcoholic Liver Disease." Curr Mol Pharmacol 10(3): 237-248.

Nedelsky, N. B., P. K. Todd and J. P. Taylor (2008). "Autophagy and the ubiquitin-proteasome system: collaborators in neuroprotection." Biochim Biophys Acta 1782(12): 691-699.

Nelson, M. E., S. Lahiri, J. D. Chow, F. L. Byrne, S. R. Hargett, D. S. Breen, E. M. Olzomer, L. E. Wu, G. J. Cooney, N. Turner, D. E. James, J. K. Slack-Davis, C. Lackner, S. H. Caldwell and K. L. Hoehn (2017). "Inhibition of hepatic lipogenesis enhances liver tumorigenesis by increasing antioxidant defence and promoting cell survival." Nat Commun 8: 14689.

209 Ness, G. C. (2015). "Physiological feedback regulation of cholesterol biosynthesis: Role of translational control of hepatic HMG-CoA reductase and possible involvement of oxylanosterols." Biochim Biophys Acta 1851(5): 667-673.

Newberry, E. P., Y. Xie, S. Kennedy, X. Han, K. K. Buhman, J. Luo, R. W. Gross and N. O. Davidson (2003). "Decreased hepatic triglyceride accumulation and altered fatty acid uptake in mice with deletion of the liver fatty acid-binding protein gene." J Biol Chem 278(51): 51664-51672.

Newton, B. W., W. K. Russell, D. H. Russell, S. K. Ramaiah and A. Jayaraman (2009). "Liver proteome analysis in a rodent model of alcoholic steatosis." J Proteome Res 8(4): 1663-1671.

Ng, M., T. Fleming, M. Robinson, B. Thomson, N. Graetz, C. Margono, E. C. Mullany, S. Biryukov, C. Abbafati, S. F. Abera, J. P. Abraham, N. M. Abu-Rmeileh, T. Achoki, F. S. AlBuhairan, Z. A. Alemu, R. Alfonso, M. K. Ali, R. Ali, N. A. Guzman, W. Ammar, P. Anwari, A. Banerjee, S. Barquera, S. Basu, D. A. Bennett, Z. Bhutta, J. Blore, N. Cabral, I. C. Nonato, J. C. Chang, R. Chowdhury, K. J. Courville, M. H. Criqui, D. K. Cundiff, K. C. Dabhadkar, L. Dandona, A. Davis, A. Dayama, S. D. Dharmaratne, E. L. Ding, A. M. Durrani, A. Esteghamati, F. Farzadfar, D. F. Fay, V. L. Feigin, A. Flaxman, M. H. Forouzanfar, A. Goto, M. A. Green, R. Gupta, N. Hafezi-Nejad, G. J. Hankey, H. C. Harewood, R. Havmoeller, S. Hay, L. Hernandez, A. Husseini, B. T. Idrisov, N. Ikeda, F. Islami, E. Jahangir, S. K. Jassal, S. H. Jee, M. Jeffreys, J. B. Jonas, E. K. Kabagambe, S. E. Khalifa, A. P. Kengne, Y. S. Khader, Y. H. Khang, D. Kim, R. W. Kimokoti, J. M. Kinge, Y. Kokubo, S. Kosen, G. Kwan, T. Lai, M. Leinsalu, Y. Li, X. Liang, S. Liu, G. Logroscino, P. A. Lotufo, Y. Lu, J. Ma, N. K. Mainoo, G. A. Mensah, T. R. Merriman, A. H. Mokdad, J. Moschandreas, M. Naghavi, A. Naheed, D. Nand, K. M. Narayan, E. L. Nelson, M. L. Neuhouser, M. I. Nisar, T. Ohkubo, S. O. Oti, A. Pedroza, D. Prabhakaran, N. Roy, U. Sampson, H. Seo, S. G. Sepanlou, K. Shibuya, R. Shiri, I. Shiue, G. M. Singh, J. A. Singh, V. Skirbekk, N. J. Stapelberg, L. Sturua, B. L. Sykes, M. Tobias, B. X. Tran, L. Trasande, H. Toyoshima, S. van de Vijver, T. J. Vasankari, J. L. Veerman, G. Velasquez- Melendez, V. V. Vlassov, S. E. Vollset, T. Vos, C. Wang, X. Wang, E. Weiderpass, A. Werdecker, J. L. Wright, Y. C. Yang, H. Yatsuya, J. Yoon, S. J. Yoon, Y. Zhao, M. Zhou, S. Zhu, A. D. Lopez, C. J. Murray and E. Gakidou (2014). "Global, regional, and national prevalence of overweight and obesity in children and adults during 1980- 2013: a systematic analysis for the Global Burden of Disease Study 2013." Lancet 384(9945): 766-781.

Nguyen, L. K., W. Kolch and B. N. Kholodenko (2013). "When ubiquitination meets phosphorylation: a systems biology perspective of EGFR/MAPK signalling." Cell Commun Signal 11: 52.

Nikkila, E. A. (1974). "Plasma triglycerides in human diabetes." Proc R Soc Med 67(7): 662- 665.

Nixon-Abell, J., C. J. Obara, A. V. Weigel, D. Li, W. R. Legant, C. S. Xu, H. A. Pasolli, K. Harvey, H. F. Hess, E. Betzig, C. Blackstone and J. Lippincott-Schwartz (2016). "Increased spatiotemporal resolution reveals highly dynamic dense tubular matrices in the peripheral ER." Science 354(6311).

210 Noguchi, M. and S. Hirohashi (1996). "Cell lines from non-neoplastic liver and hepatocellular carcinoma tissue from a single patient." In Vitro Cell Dev Biol Anim 32(3): 135-137.

Ofei, F. (2005). "Obesity - a preventable disease." Ghana Med J 39(3): 98-101.

Ogawa, Y., K. Imajo, Y. Honda, T. Kessoku, W. Tomeno, S. Kato, K. Fujita, M. Yoneda, S. Saito, Y. Saigusa, H. Hyogo, Y. Sumida, Y. Itoh, K. Eguchi, T. Yamanaka, K. Wada and A. Nakajima (2018). "Palmitate-induced lipotoxicity is crucial for the pathogenesis of nonalcoholic fatty liver disease in cooperation with gut-derived endotoxin." Sci Rep 8(1): 11365.

Olsen, J. V., B. Blagoev, F. Gnad, B. Macek, C. Kumar, P. Mortensen and M. Mann (2006). "Global, in vivo, and site-specific phosphorylation dynamics in signaling networks." Cell 127(3): 635-648.

Olzmann, J. A., C. M. Richter and R. R. Kopito (2013). "Spatial regulation of UBXD8 and p97/VCP controls ATGL-mediated lipid droplet turnover." Proc Natl Acad Sci U S A 110(4): 1345-1350.

Ong, K. T., M. T. Mashek, S. Y. Bu, A. S. Greenberg and D. G. Mashek (2011). "Adipose triglyceride lipase is a major hepatic lipase that regulates triacylglycerol turnover and fatty acid signaling and partitioning." Hepatology 53(1): 116-126.

Ong, K. T., M. T. Mashek, N. O. Davidson and D. G. Mashek (2014). "Hepatic ATGL mediates PPAR-alpha signaling and fatty acid channeling through an L-FABP independent mechanism." J Lipid Res 55(5): 808-815.

Oni, E. T., A. S. Agatston, M. J. Blaha, J. Fialkow, R. Cury, A. Sposito, R. Erbel, R. Blankstein, T. Feldman, M. H. Al-Mallah, R. D. Santos, M. J. Budoff and K. Nasir (2013). "A systematic review: burden and severity of subclinical cardiovascular disease among those with nonalcoholic fatty liver; should we care?" Atherosclerosis 230(2): 258-267.

Ooi, G. J., A. Earnest, W. W. Kemp, P. R. Burton, C. Laurie, A. Majeed, N. Johnson, C. McLean, S. K. Roberts and W. A. Brown (2018). "Evaluating feasibility and accuracy of non-invasive tests for nonalcoholic fatty liver disease in severe and morbid obesity." Int J Obes (Lond).

Organization, W. H. (2018). "Obesity and Overweight." from http://www.who.int/news- room/fact-sheets/detail/obesity-and-overweight.

Pagnon, J., M. Matzaris, R. Stark, R. C. Meex, S. L. Macaulay, W. Brown, P. E. O'Brien, T. Tiganis and M. J. Watt (2012). "Identification and functional characterization of protein kinase A phosphorylation sites in the major lipolytic protein, adipose triglyceride lipase." Endocrinology 153(9): 4278-4289.

Palmisano, G., S. E. Lendal, K. Engholm-Keller, R. Leth-Larsen, B. L. Parker and M. R. Larsen (2010). "Selective enrichment of sialic acid-containing glycopeptides using titanium dioxide chromatography with analysis by HILIC and mass spectrometry." Nat Protoc 5(12): 1974-1982.

Pardina, E., J. A. Baena-Fustegueras, R. Llamas, R. Catalan, R. Galard, A. Lecube, J. M. Fort, M. Llobera, H. Allende, V. Vargas and J. Peinado-Onsurbe (2009). "Lipoprotein

211 lipase expression in livers of morbidly obese patients could be responsible for liver steatosis." Obes Surg 19(5): 608-616.

Parker, B. L., G. Yang, S. J. Humphrey, R. Chaudhuri, X. Ma, S. Peterman and D. E. James (2015). "Targeted phosphoproteomics of insulin signaling using data-independent acquisition mass spectrometry." Sci Signal 8(380): rs6.

Parks, E. J. (2002). "Dietary carbohydrate's effects on lipogenesis and the relationship of lipogenesis to blood insulin and glucose concentrations." Br J Nutr 87 Suppl 2: S247- 253.

Peak, M., J. J. Rochford, A. C. Borthwick, S. J. Yeaman and L. Agius (1998). "Signalling pathways involved in the stimulation of glycogen synthesis by insulin in rat hepatocytes." Diabetologia 41(1): 16-25.

Pereira, S. S. and J. I. Alvarez-Leite (2014). "Low-Grade Inflammation, Obesity, and Diabetes." Curr Obes Rep 3(4): 422-431.

Perry, R. J., T. Kim, X. M. Zhang, H. Y. Lee, D. Pesta, V. B. Popov, D. Zhang, Y. Rahimi, M. J. Jurczak, G. W. Cline, D. A. Spiegel and G. I. Shulman (2013). "Reversal of hypertriglyceridemia, fatty liver disease, and insulin resistance by a liver-targeted mitochondrial uncoupler." Cell Metab 18(5): 740-748.

Perry, R. J., D. Zhang, X. M. Zhang, J. L. Boyer and G. I. Shulman (2015). "Controlled-release mitochondrial protonophore reverses diabetes and steatohepatitis in rats." Science 347(6227): 1253-1256.

Pisto, P., M. Santaniemi, R. Bloigu, O. Ukkola and Y. A. Kesaniemi (2014). "Fatty liver predicts the risk for cardiovascular events in middle-aged population: a population- based cohort study." BMJ Open 4(3): e004973.

Poirier, Y., V. D. Antonenkov, T. Glumoff and J. K. Hiltunen (2006). "Peroxisomal beta- oxidation--a metabolic pathway with multiple functions." Biochim Biophys Acta 1763(12): 1413-1426.

Prashanth, M., H. K. Ganesh, M. V. Vima, M. John, T. Bandgar, S. R. Joshi, S. R. Shah, P. M. Rathi, A. S. Joshi, H. Thakkar, P. S. Menon and N. S. Shah (2009). "Prevalence of nonalcoholic fatty liver disease in patients with type 2 diabetes mellitus." J Assoc Physicians India 57: 205-210.

Quehenberger, O., A. M. Armando, A. H. Brown, S. B. Milne, D. S. Myers, A. H. Merrill, S. Bandyopadhyay, K. N. Jones, S. Kelly, R. L. Shaner, C. M. Sullards, E. Wang, R. C. Murphy, R. M. Barkley, T. J. Leiker, C. R. Raetz, Z. Guan, G. M. Laird, D. A. Six, D. W. Russell, J. G. McDonald, S. Subramaniam, E. Fahy and E. A. Dennis (2010). "Lipidomics reveals a remarkable diversity of lipids in human plasma." J Lipid Res 51(11): 3299-3305.

Ragab, S. M., S. Abd Elghaffar, T. H. El-Metwally, G. Badr, M. H. Mahmoud and H. M. Omar (2015). "Effect of a high fat, high sucrose diet on the promotion of non-alcoholic fatty liver disease in male rats: the ameliorative role of three natural compounds." Lipids Health Dis 14: 83.

212 Ramboer, E., T. Vanhaecke, V. Rogiers and M. Vinken (2015). "Immortalized Human Hepatic Cell Lines for In Vitro Testing and Research Purposes." Methods Mol Biol 1250: 53- 76.

Rambold, A. S., S. Cohen and J. Lippincott-Schwartz (2015). "Fatty acid trafficking in starved cells: regulation by lipid droplet lipolysis, autophagy, and mitochondrial fusion dynamics." Dev Cell 32(6): 678-692.

Randle, P. J., P. B. Garland, C. N. Hales and E. A. Newsholme (1963). "The glucose fatty-acid cycle. Its role in insulin sensitivity and the metabolic disturbances of diabetes mellitus." Lancet 1(7285): 785-789.

Reali, F., M. J. Morine, O. Kahramanogullari, S. Raichur, H. C. Schneider, D. Crowther and C. Priami (2017). "Mechanistic interplay between ceramide and insulin resistance." Sci Rep 7: 41231.

Rebrin, K., G. M. Steil, L. Getty and R. N. Bergman (1995). "Free fatty acid as a link in the regulation of hepatic glucose output by peripheral insulin." Diabetes 44(9): 1038- 1045.

Rector, R. S., J. P. Thyfault, G. M. Uptergrove, E. M. Morris, S. P. Naples, S. J. Borengasser, C. R. Mikus, M. J. Laye, M. H. Laughlin, F. W. Booth and J. A. Ibdah (2010). "Mitochondrial dysfunction precedes insulin resistance and hepatic steatosis and contributes to the natural history of non-alcoholic fatty liver disease in an obese rodent model." J Hepatol 52(5): 727-736.

Reid, B. N., G. P. Ables, O. A. Otlivanchik, G. Schoiswohl, R. Zechner, W. S. Blaner, I. J. Goldberg, R. F. Schwabe, S. C. Chua, Jr. and L. S. Huang (2008). "Hepatic overexpression of hormone-sensitive lipase and adipose triglyceride lipase promotes fatty acid oxidation, stimulates direct release of free fatty acids, and ameliorates steatosis." J Biol Chem 283(19): 13087-13099.

Rinella, M. E., Z. Lominadze, R. Loomba, M. Charlton, B. A. Neuschwander-Tetri, S. H. Caldwell, K. Kowdley and S. A. Harrison (2016). "Practice patterns in NAFLD and NASH: real life differs from published guidelines." Therap Adv Gastroenterol 9(1): 4-12.

Risso, G., M. Blaustein, B. Pozzi, P. Mammi and A. Srebrow (2015). "Akt/PKB: one kinase, many modifications." Biochem J 468(2): 203-214.

Roepstorff, C., B. Vistisen, M. Donsmark, J. N. Nielsen, H. Galbo, K. A. Green, D. G. Hardie, J. F. Wojtaszewski, E. A. Richter and B. Kiens (2004). "Regulation of hormone- sensitive lipase activity and Ser563 and Ser565 phosphorylation in human skeletal muscle during exercise." J Physiol 560(Pt 2): 551-562.

Romeo, S., J. Kozlitina, C. Xing, A. Pertsemlidis, D. Cox, L. A. Pennacchio, E. Boerwinkle, J. C. Cohen and H. H. Hobbs (2008). "Genetic variation in PNPLA3 confers susceptibility to nonalcoholic fatty liver disease." Nat Genet 40(12): 1461-1465.

Rosch, K., M. Kwiatkowski, S. Hofmann, A. Schobel, C. Gruttner, M. Wurlitzer, H. Schluter and E. Herker (2016). "Quantitative Lipid Droplet Proteome Analysis Identifies Annexin A3 as a Cofactor for HCV Particle Production." Cell Rep 16(12): 3219-3231.

213 Rowe, E. R., M. L. Mimmack, A. D. Barbosa, A. Haider, I. Isaac, M. M. Ouberai, A. R. Thiam, S. Patel, V. Saudek, S. Siniossoglou and D. B. Savage (2016). "Conserved Amphipathic Helices Mediate Lipid Droplet Targeting of Perilipins 1-3." J Biol Chem 291(13): 6664-6678.

Rui, L. (2014). "Energy metabolism in the liver." Compr Physiol 4(1): 177-197.

Rui, L., M. Yuan, D. Frantz, S. Shoelson and M. F. White (2002). "SOCS-1 and SOCS-3 block insulin signaling by ubiquitin-mediated degradation of IRS1 and IRS2." J Biol Chem 277(44): 42394-42398.

Russell-Jones, D. L. and M. Umpleby (1996). "Protein anabolic action of insulin, growth hormone and insulin-like growth factor I." Eur J Endocrinol 135(6): 631-642.

Sahu-Osen, A., G. Montero-Moran, M. Schittmayer, K. Fritz, A. Dinh, Y. F. Chang, D. McMahon, A. Boeszoermenyi, I. Cornaciu, D. Russell, M. Oberer, G. M. Carman, R. Birner-Gruenberger and D. L. Brasaemle (2015). "CGI-58/ABHD5 is phosphorylated on Ser239 by protein kinase A: control of subcellular localization." J Lipid Res 56(1): 109-121.

Saka, H. A., J. W. Thompson, Y. S. Chen, L. G. Dubois, J. T. Haas, A. Moseley and R. H. Valdivia (2015). "Chlamydia trachomatis Infection Leads to Defined Alterations to the Lipid Droplet Proteome in Epithelial Cells." PLoS One 10(4): e0124630.

Salameh, H., M. A. Hanayneh, M. Masadeh, M. Naseemuddin, T. Matin, A. Erwin and A. K. Singal (2016). "PNPLA3 as a Genetic Determinant of Risk for and Severity of Non- alcoholic Fatty Liver Disease Spectrum." J Clin Transl Hepatol 4(3): 175-191.

Samsa, M. M., J. A. Mondotte, N. G. Iglesias, I. Assuncao-Miranda, G. Barbosa-Lima, A. T. Da Poian, P. T. Bozza and A. V. Gamarnik (2009). "Dengue virus capsid protein usurps lipid droplets for viral particle formation." PLoS Pathog 5(10): e1000632.

Samuel, V. T., Z. X. Liu, X. Qu, B. D. Elder, S. Bilz, D. Befroy, A. J. Romanelli and G. I. Shulman (2004). "Mechanism of hepatic insulin resistance in non-alcoholic fatty liver disease." J Biol Chem 279(31): 32345-32353.

Samuel, V. T., Z. X. Liu, A. Wang, S. A. Beddow, J. G. Geisler, M. Kahn, X. M. Zhang, B. P. Monia, S. Bhanot and G. I. Shulman (2007). "Inhibition of protein kinase Cepsilon prevents hepatic insulin resistance in nonalcoholic fatty liver disease." J Clin Invest 117(3): 739-745.

Satapati, S., N. E. Sunny, B. Kucejova, X. Fu, T. T. He, A. Mendez-Lucas, J. M. Shelton, J. C. Perales, J. D. Browning and S. C. Burgess (2012). "Elevated TCA cycle function in the pathology of diet-induced hepatic insulin resistance and fatty liver." J Lipid Res 53(6): 1080-1092.

Savage, D. B., C. S. Choi, V. T. Samuel, Z. X. Liu, D. Zhang, A. Wang, X. M. Zhang, G. W. Cline, X. X. Yu, J. G. Geisler, S. Bhanot, B. P. Monia and G. I. Shulman (2006). "Reversal of diet-induced hepatic steatosis and hepatic insulin resistance by antisense oligonucleotide inhibitors of acetyl-CoA carboxylases 1 and 2." J Clin Invest 116(3): 817-824.

214 Schippers, I. J., H. Moshage, H. Roelofsen, M. Muller, H. S. Heymans, M. Ruiters and F. Kuipers (1997). "Immortalized human hepatocytes as a tool for the study of hepatocytic (de-)differentiation." Cell Biol Toxicol 13(4-5): 375-386.

Schmitz-Peiffer, C. and T. J. Biden (2008). "Protein kinase C function in muscle, liver, and beta-cells and its therapeutic implications for type 2 diabetes." Diabetes 57(7): 1774- 1783.

Schmitz-Peiffer, C., D. R. Laybutt, J. G. Burchfield, E. Gurisik, S. Narasimhan, C. J. Mitchell, D. J. Pedersen, U. Braun, G. J. Cooney, M. Leitges and T. J. Biden (2007). "Inhibition of PKCepsilon improves glucose-stimulated insulin secretion and reduces insulin clearance." Cell Metab 6(4): 320-328.

Schonfeld, G., B. W. Patterson, D. A. Yablonskiy, T. S. Tanoli, M. Averna, N. Elias, P. Yue and J. Ackerman (2003). "Fatty liver in familial hypobetalipoproteinemia: triglyceride assembly into VLDL particles is affected by the extent of hepatic steatosis." J Lipid Res 44(3): 470-478.

Schrader, E. K., K. G. Harstad and A. Matouschek (2009). "Targeting proteins for degradation." Nat Chem Biol 5(11): 815-822.

Schrauwen, P., V. Schrauwen-Hinderling, J. Hoeks and M. K. Hesselink (2010). "Mitochondrial dysfunction and lipotoxicity." Biochim Biophys Acta 1801(3): 266- 271.

Schrauwen, P., A. J. Wagenmakers, W. D. van Marken Lichtenbelt, W. H. Saris and K. R. Westerterp (2000). "Increase in fat oxidation on a high-fat diet is accompanied by an increase in triglyceride-derived fatty acid oxidation." Diabetes 49(4): 640-646.

Schuldiner, M. and M. Bohnert (2017). "A different kind of love - lipid droplet contact sites." Biochim Biophys Acta Mol Cell Biol Lipids 1862(10 Pt B): 1188-1196.

Schulze, R. J., A. Sathyanarayan and D. G. Mashek (2017). "Breaking fat: The regulation and mechanisms of lipophagy." Biochim Biophys Acta Mol Cell Biol Lipids 1862(10 Pt B): 1178-1187.

Schweiger, M., M. Romauch, R. Schreiber, G. F. Grabner, S. Hutter, P. Kotzbeck, P. Benedikt, T. O. Eichmann, S. Yamada, O. Knittelfelder, C. Diwoky, C. Doler, N. Mayer, W. De Cecco, R. Breinbauer, R. Zimmermann and R. Zechner (2017). "Pharmacological inhibition of adipose triglyceride lipase corrects high-fat diet-induced insulin resistance and hepatosteatosis in mice." Nat Commun 8: 14859.

Senkal, C. E., M. F. Salama, A. J. Snider, J. J. Allopenna, N. A. Rana, A. Koller, Y. A. Hannun and L. M. Obeid (2017). "Ceramide Is Metabolized to Acylceramide and Stored in Lipid Droplets." Cell Metab 25(3): 686-697.

Seo, Y. S., J. H. Kim, N. Y. Jo, K. M. Choi, S. H. Baik, J. J. Park, J. S. Kim, K. S. Byun, Y. T. Bak, C. H. Lee, A. Kim and J. E. Yeon (2008). "PPAR agonists treatment is effective in a nonalcoholic fatty liver disease animal model by modulating fatty-acid metabolic enzymes." J Gastroenterol Hepatol 23(1): 102-109.

215 Shanik, M. H., Y. Xu, J. Skrha, R. Dankner, Y. Zick and J. Roth (2008). "Insulin resistance and hyperinsulinemia: is hyperinsulinemia the cart or the horse?" Diabetes Care 31 Suppl 2: S262-268.

Sharabi, K., C. D. Tavares, A. K. Rines and P. Puigserver (2015). "Molecular pathophysiology of hepatic glucose production." Mol Aspects Med 46: 21-33.

Sharpe, L. J. and A. J. Brown (2013). "Controlling cholesterol synthesis beyond 3-hydroxy-3- methylglutaryl-CoA reductase (HMGCR)." J Biol Chem 288(26): 18707-18715.

Shen, W. J., S. Patel, H. Miyoshi, A. S. Greenberg and F. B. Kraemer (2009). "Functional interaction of hormone-sensitive lipase and perilipin in lipolysis." J Lipid Res 50(11): 2306-2313.

Shimomura, I., M. Matsuda, R. E. Hammer, Y. Bashmakov, M. S. Brown and J. L. Goldstein (2000). "Decreased IRS-2 and increased SREBP-1c lead to mixed insulin resistance and sensitivity in livers of lipodystrophic and ob/ob mice." Mol Cell 6(1): 77-86.

Sikder, K., S. K. Shukla, N. Patel, H. Singh and K. Rafiq (2018). "High Fat Diet Upregulates Fatty Acid Oxidation and Ketogenesis via Intervention of PPAR-gamma." Cell Physiol Biochem 48(3): 1317-1331.

Sindelar, D. K., C. A. Chu, M. Rohlie, D. W. Neal, L. L. Swift and A. D. Cherrington (1997). "The role of fatty acids in mediating the effects of peripheral insulin on hepatic glucose production in the conscious dog." Diabetes 46(2): 187-196.

Soupene, E. and F. A. Kuypers (2008). "Mammalian long-chain acyl-CoA synthetases." Exp Biol Med (Maywood) 233(5): 507-521.

Stefanovic-Racic, M., G. Perdomo, B. S. Mantell, I. J. Sipula, N. F. Brown and R. M. O'Doherty (2008). "A moderate increase in carnitine palmitoyltransferase 1a activity is sufficient to substantially reduce hepatic triglyceride levels." Am J Physiol Endocrinol Metab 294(5): E969-977.

Steneberg, P., A. G. Sykaras, F. Backlund, J. Straseviciene, I. Soderstrom and H. Edlund (2015). "Hyperinsulinemia Enhances Hepatic Expression of the Fatty Acid Transporter Cd36 and Provokes Hepatosteatosis and Hepatic Insulin Resistance." J Biol Chem 290(31): 19034-19043.

Stewart, M. D., T. Ritterhoff, R. E. Klevit and P. S. Brzovic (2016). "E2 enzymes: more than just middle men." Cell Res 26(4): 423-440.

Stratford, S., K. L. Hoehn, F. Liu and S. A. Summers (2004). "Regulation of insulin action by ceramide: dual mechanisms linking ceramide accumulation to the inhibition of Akt/protein kinase B." J Biol Chem 279(35): 36608-36615.

Strumilo, S. (2005). "Short-term regulation of the mammalian pyruvate dehydrogenase complex." Acta Biochim Pol 52(4): 759-764.

Su, W., Y. Wang, X. Jia, W. Wu, L. Li, X. Tian, S. Li, C. Wang, H. Xu, J. Cao, Q. Han, S. Xu, Y. Chen, Y. Zhong, X. Zhang, P. Liu, J. A. Gustafsson and Y. Guan (2014).

216 "Comparative proteomic study reveals 17beta-HSD13 as a pathogenic protein in nonalcoholic fatty liver disease." Proc Natl Acad Sci U S A 111(31): 11437-11442.

Subramanian, V., A. Rothenberg, C. Gomez, A. W. Cohen, A. Garcia, S. Bhattacharyya, L. Shapiro, G. Dolios, R. Wang, M. P. Lisanti and D. L. Brasaemle (2004). "Perilipin A mediates the reversible binding of CGI-58 to lipid droplets in 3T3-L1 adipocytes." J Biol Chem 279(40): 42062-42071.

Sugaya, Y. and H. Satoh (2017). "Liver-specific G0 /G1 switch gene 2 (G0s2) expression promotes hepatic insulin resistance by exacerbating hepatic steatosis in male Wistar rats." J Diabetes 9(8): 754-763.

Suizu, F., Y. Hiramuki, F. Okumura, M. Matsuda, A. J. Okumura, N. Hirata, M. Narita, T. Kohno, J. Yokota, M. Bohgaki, C. Obuse, S. Hatakeyama, T. Obata and M. Noguchi (2009). "The E3 ligase TTC3 facilitates ubiquitination and degradation of phosphorylated Akt." Dev Cell 17(6): 800-810.

Summart, U., B. Thinkhamrop, N. Chamadol, N. Khuntikeo, M. Songthamwat and C. S. Kim (2017). "Gender differences in the prevalence of nonalcoholic fatty liver disease in the Northeast of Thailand: A population-based cross-sectional study." F1000Res 6: 1630.

Sunny, N. E., E. J. Parks, J. D. Browning and S. C. Burgess (2011). "Excessive hepatic mitochondrial TCA cycle and gluconeogenesis in humans with nonalcoholic fatty liver disease." Cell Metab 14(6): 804-810.

Suzuki, M., T. Murakami, J. Cheng, H. Kano, M. Fukata and T. Fujimoto (2015). "ELMOD2 is anchored to lipid droplets by palmitoylation and regulates adipocyte triglyceride lipase recruitment." Mol Biol Cell 26(12): 2333-2342.

Suzuki, M., T. Otsuka, Y. Ohsaki, J. Cheng, T. Taniguchi, H. Hashimoto, H. Taniguchi and T. Fujimoto (2012). "Derlin-1 and UBXD8 are engaged in dislocation and degradation of lipidated ApoB-100 at lipid droplets." Mol Biol Cell 23(5): 800-810.

Swaney, D. L., P. Beltrao, L. Starita, A. Guo, J. Rush, S. Fields, N. J. Krogan and J. Villen (2013). "Global analysis of phosphorylation and ubiquitylation cross-talk in protein degradation." Nat Methods 10(7): 676-682.

Sztalryd, C., G. Xu, H. Dorward, J. T. Tansey, J. A. Contreras, A. R. Kimmel and C. Londos (2003). "Perilipin A is essential for the translocation of hormone-sensitive lipase during lipolytic activation." J Cell Biol 161(6): 1093-1103.

Takeuchi, K. and K. Reue (2009). "Biochemistry, physiology, and genetics of GPAT, AGPAT, and lipin enzymes in triglyceride synthesis." Am J Physiol Endocrinol Metab 296(6): E1195-1209.

Tan, N. S., N. S. Shaw, N. Vinckenbosch, P. Liu, R. Yasmin, B. Desvergne, W. Wahli and N. Noy (2002). "Selective cooperation between fatty acid binding proteins and peroxisome proliferator-activated receptors in regulating transcription." Mol Cell Biol 22(14): 5114-5127.

217 Tandra, S., M. M. Yeh, E. M. Brunt, R. Vuppalanchi, O. W. Cummings, A. Unalp-Arida, L. A. Wilson and N. Chalasani (2011). "Presence and significance of microvesicular steatosis in nonalcoholic fatty liver disease." J Hepatol 55(3): 654-659.

Tao, H., Y. Zhang, X. Zeng, G. I. Shulman and S. Jin (2014). "Niclosamide ethanolamine- induced mild mitochondrial uncoupling improves diabetic symptoms in mice." Nat Med 20(11): 1263-1269.

Targher, G., L. Bertolini, R. Padovani, S. Rodella, R. Tessari, L. Zenari, C. Day and G. Arcaro (2007). "Prevalence of nonalcoholic fatty liver disease and its association with cardiovascular disease among type 2 diabetic patients." Diabetes Care 30(5): 1212- 1218.

Ter Horst, K. W., P. W. Gilijamse, R. I. Versteeg, M. T. Ackermans, A. J. Nederveen, S. E. la Fleur, J. A. Romijn, M. Nieuwdorp, D. Zhang, V. T. Samuel, D. F. Vatner, K. F. Petersen, G. I. Shulman and M. J. Serlie (2017). "Hepatic Diacylglycerol-Associated Protein Kinase Cepsilon Translocation Links Hepatic Steatosis to Hepatic Insulin Resistance in Humans." Cell Rep 19(10): 1997-2004.

Thiam, A. R. and M. Beller (2017). "The why, when and how of lipid droplet diversity." J Cell Sci 130(2): 315-324.

Thiam, A. R., R. V. Farese, Jr. and T. C. Walther (2013). "The biophysics and cell biology of lipid droplets." Nat Rev Mol Cell Biol 14(12): 775-786.

Thiam, A. R. and L. Foret (2016). "The physics of lipid droplet nucleation, growth and budding." Biochim Biophys Acta 1861(8 Pt A): 715-722.

Thorne, A., P. Lofgren and J. Hoffstedt (2010). "Increased visceral adipocyte lipolysis--a pathogenic role in nonalcoholic fatty liver disease?" J Clin Endocrinol Metab 95(10): E209-213.

Titchenell, P. M., W. J. Quinn, M. Lu, Q. Chu, W. Lu, C. Li, H. Chen, B. R. Monks, J. Chen, J. D. Rabinowitz and M. J. Birnbaum (2016). "Direct Hepatocyte Insulin Signaling Is Required for Lipogenesis but Is Dispensable for the Suppression of Glucose Production." Cell Metab 23(6): 1154-1166.

Tolman, K. G. and A. S. Dalpiaz (2007). "Treatment of non-alcoholic fatty liver disease." Ther Clin Risk Manag 3(6): 1153-1163.

Toshikuni, N., M. Tsutsumi and T. Arisawa (2014). "Clinical differences between alcoholic liver disease and nonalcoholic fatty liver disease." World J Gastroenterol 20(26): 8393-8406.

Trevino, M. B., D. Mazur-Hart, Y. Machida, T. King, J. Nadler, E. V. Galkina, A. Poddar, S. Dutta and Y. Imai (2015). "Liver Perilipin 5 Expression Worsens Hepatosteatosis But Not Insulin Resistance in High Fat-Fed Mice." Mol Endocrinol 29(10): 1414-1425.

Tsai, T. H., E. Chen, L. Li, P. Saha, H. J. Lee, L. S. Huang, G. S. Shelness, L. Chan and B. H. Chang (2017). "The constitutive lipid droplet protein PLIN2 regulates autophagy in liver." Autophagy 13(7): 1130-1144.

218 Tsai, Y. C., G. S. Leichner, M. M. Pearce, G. L. Wilson, R. J. Wojcikiewicz, J. Roitelman and A. M. Weissman (2012). "Differential regulation of HMG-CoA reductase and Insig- 1 by enzymes of the ubiquitin-proteasome system." Mol Biol Cell 23(23): 4484-4494.

Turner, N., G. M. Kowalski, S. J. Leslie, S. Risis, C. Yang, R. S. Lee-Young, J. R. Babb, P. J. Meikle, G. I. Lancaster, D. C. Henstridge, P. J. White, E. W. Kraegen, A. Marette, G. J. Cooney, M. A. Febbraio and C. R. Bruce (2013). "Distinct patterns of tissue-specific lipid accumulation during the induction of insulin resistance in mice by high-fat feeding." Diabetologia 56(7): 1638-1648.

Turpin, S. M., A. J. Hoy, R. D. Brown, C. G. Rudaz, J. Honeyman, M. Matzaris and M. J. Watt (2011). "Adipose triacylglycerol lipase is a major regulator of hepatic lipid metabolism but not insulin sensitivity in mice." Diabetologia 54(1): 146-156.

Turro, S., M. Ingelmo-Torres, J. M. Estanyol, F. Tebar, M. A. Fernandez, C. V. Albor, K. Gaus, T. Grewal, C. Enrich and A. Pol (2006). "Identification and characterization of associated with lipid droplet protein 1: A novel membrane-associated protein that resides on hepatic lipid droplets." Traffic 7(9): 1254-1269.

Tyanova, S., T. Temu and J. Cox (2016). "The MaxQuant computational platform for mass spectrometry-based shotgun proteomics." Nat Protoc 11(12): 2301-2319.

Uhlen, M., L. Fagerberg, B. M. Hallstrom, C. Lindskog, P. Oksvold, A. Mardinoglu, A. Sivertsson, C. Kampf, E. Sjostedt, A. Asplund, I. Olsson, K. Edlund, E. Lundberg, S. Navani, C. A. Szigyarto, J. Odeberg, D. Djureinovic, J. O. Takanen, S. Hober, T. Alm, P. H. Edqvist, H. Berling, H. Tegel, J. Mulder, J. Rockberg, P. Nilsson, J. M. Schwenk, M. Hamsten, K. von Feilitzen, M. Forsberg, L. Persson, F. Johansson, M. Zwahlen, G. von Heijne, J. Nielsen and F. Ponten (2015). "Proteomics. Tissue-based map of the human proteome." Science 347(6220): 1260419.

Vabulas, R. M. and F. U. Hartl (2005). "Protein synthesis upon acute nutrient restriction relies on proteasome function." Science 310(5756): 1960-1963. van Loon, L. J. and B. H. Goodpaster (2006). "Increased intramuscular lipid storage in the insulin-resistant and endurance-trained state." Pflugers Arch 451(5): 606-616.

Vanni, S. (2017). "Intracellular Lipid Droplets: From Structure to Function." Lipid Insights 10: 1178635317745518.

Venne, A. S., L. Kollipara and R. P. Zahedi (2014). "The next level of complexity: crosstalk of posttranslational modifications." Proteomics 14(4-5): 513-524.

Virkamaki, A., K. Ueki and C. R. Kahn (1999). "Protein-protein interaction in insulin signaling and the molecular mechanisms of insulin resistance." J Clin Invest 103(7): 931-943.

Wagner, S. A., P. Beli, B. T. Weinert, C. Scholz, C. D. Kelstrup, C. Young, M. L. Nielsen, J. V. Olsen, C. Brakebusch and C. Choudhary (2012). "Proteomic analyses reveal divergent ubiquitylation site patterns in murine tissues." Mol Cell Proteomics 11(12): 1578-1585.

Walther, T. C., J. Chung and R. V. Farese, Jr. (2017). "Lipid Droplet Biogenesis." Annu Rev Cell Dev Biol 33: 491-510.

219 Walther, T. C. and R. V. Farese, Jr. (2012). "Lipid droplets and cellular lipid metabolism." Annu Rev Biochem 81: 687-714.

Wang, C., Y. Zhao, X. Gao, L. Li, Y. Yuan, F. Liu, L. Zhang, J. Wu, P. Hu, X. Zhang, Y. Gu, Y. Xu, Z. Wang, Z. Li, H. Zhang and J. Ye (2015). "Perilipin 5 improves hepatic lipotoxicity by inhibiting lipolysis." Hepatology 61(3): 870-882.

Wang, F., A. B. Kohan, C. M. Lo, M. Liu, P. Howles and P. Tso (2015). "Apolipoprotein A- IV: a protein intimately involved in metabolism." J Lipid Res 56(8): 1403-1418.

Wang, G., H. L. Bonkovsky, A. de Lemos and F. J. Burczynski (2015). "Recent insights into the biological functions of liver fatty acid binding protein 1." J Lipid Res 56(12): 2238-2247.

Wang, H., M. Bell, U. Sreenivasan, H. Hu, J. Liu, K. Dalen, C. Londos, T. Yamaguchi, M. A. Rizzo, R. Coleman, D. Gong, D. Brasaemle and C. Sztalryd (2011). "Unique regulation of adipose triglyceride lipase (ATGL) by perilipin 5, a lipid droplet- associated protein." J Biol Chem 286(18): 15707-15715.

Wang, H., U. Sreenivasan, H. Hu, A. Saladino, B. M. Polster, L. M. Lund, D. W. Gong, W. C. Stanley and C. Sztalryd (2011). "Perilipin 5, a lipid droplet-associated protein, provides physical and metabolic linkage to mitochondria." J Lipid Res 52(12): 2159- 2168.

Wang, M., J. Hayakawa, K. Yang and X. Han (2014). "Characterization and quantification of diacylglycerol species in biological extracts after one-step derivatization: a shotgun lipidomics approach." Anal Chem 86(4): 2146-2155.

Wang, W., S. Wei, L. Li, X. Su, C. Du, F. Li, B. Geng, P. Liu and G. Xu (2015). "Proteomic analysis of murine testes lipid droplets." Sci Rep 5: 12070.

Wang, Y., Y. Zhang, H. Qian, J. Lu, Z. Zhang, X. Min, M. Lang, H. Yang, N. Wang and P. Zhang (2013). "The g0/g1 switch gene 2 is an important regulator of hepatic triglyceride metabolism." PLoS One 8(8): e72315.

Wang, Y., Y. Zhou and D. T. Graves (2014). "FOXO transcription factors: their clinical significance and regulation." Biomed Res Int 2014: 925350.

Wattacheril, J., K. L. Rose, S. Hill, C. Lanciault, C. R. Murray, K. Washington, B. Williams, W. English, M. Spann, R. Clements, N. Abumrad and C. R. Flynn (2017). "Non- alcoholic fatty liver disease phosphoproteomics: A functional piece of the precision puzzle." Hepatol Res 47(13): 1469-1483.

Wei, S., S. Liu, X. Su, W. Wang, F. Li, J. Deng, Y. Lyu, B. Geng and G. Xu (2018). "Spontaneous development of hepatosteatosis in perilipin-1 null mice with adipose tissue dysfunction." Biochim Biophys Acta Mol Cell Biol Lipids 1863(2): 212-218.

Wei, Y., J. Zhou, H. Yu and X. Jin (2018). "AKT phosphorylation sites of Ser473 and Thr308 regulate AKT degradation." Biosci Biotechnol Biochem: 1-7.

220 Welch, C., M. K. Santra, W. El-Assaad, X. Zhu, W. E. Huber, R. A. Keys, J. G. Teodoro and M. R. Green (2009). "Identification of a protein, G0S2, that lacks Bcl-2 domains and interacts with and antagonizes Bcl-2." Cancer Res 69(17): 6782-6789.

Welte, M. A. (2015). "Expanding roles for lipid droplets." Curr Biol 25(11): R470-481.

Welte, M. A., S. Cermelli, J. Griner, A. Viera, Y. Guo, D. H. Kim, J. G. Gindhart and S. P. Gross (2005). "Regulation of lipid-droplet transport by the perilipin homolog LSD2." Curr Biol 15(14): 1266-1275.

Wendel, A. A., D. E. Cooper, O. R. Ilkayeva, D. M. Muoio and R. A. Coleman (2013). "Glycerol-3-phosphate acyltransferase (GPAT)-1, but not GPAT4, incorporates newly synthesized fatty acids into triacylglycerol and diminishes fatty acid oxidation." J Biol Chem 288(38): 27299-27306.

Wilfling, F., J. T. Haas, T. C. Walther and R. V. Farese, Jr. (2014). "Lipid droplet biogenesis." Curr Opin Cell Biol 29: 39-45.

Wilfling, F., A. R. Thiam, M. J. Olarte, J. Wang, R. Beck, T. J. Gould, E. S. Allgeyer, F. Pincet, J. Bewersdorf, R. V. Farese, Jr. and T. C. Walther (2014). "Arf1/COPI machinery acts directly on lipid droplets and enables their connection to the ER for protein targeting." Elife 3: e01607.

Wilfling, F., H. Wang, J. T. Haas, N. Krahmer, T. J. Gould, A. Uchida, J. X. Cheng, M. Graham, R. Christiano, F. Frohlich, X. Liu, K. K. Buhman, R. A. Coleman, J. Bewersdorf, R. V. Farese, Jr. and T. C. Walther (2013). "Triacylglycerol synthesis enzymes mediate lipid droplet growth by relocalizing from the ER to lipid droplets." Dev Cell 24(4): 384-399.

Wilkinson, K. D. (2000). "Ubiquitination and deubiquitination: targeting of proteins for degradation by the proteasome." Semin Cell Dev Biol 11(3): 141-148.

Wilson, C. G., J. L. Tran, D. M. Erion, N. B. Vera, M. Febbraio and E. J. Weiss (2016). "Hepatocyte-Specific Disruption of CD36 Attenuates Fatty Liver and Improves Insulin Sensitivity in HFD-Fed Mice." Endocrinology 157(2): 570-585.

Wilson-Grady, J. T., W. Haas and S. P. Gygi (2013). "Quantitative comparison of the fasted and re-fed mouse liver phosphoproteomes using lower pH reductive dimethylation." Methods 61(3): 277-286.

Wiss, O. and V. Wiss (1977). "Influence of insulin and glucagon on the cholesterol synthesis in rat liver in vivo. Evidence against a rate limiting function of the HMG-coenzyme A reductase." Helv Chim Acta 60(6): 1967-1968.

Wolfrum, C., E. Asilmaz, E. Luca, J. M. Friedman and M. Stoffel (2004). "Foxa2 regulates lipid metabolism and ketogenesis in the liver during fasting and in diabetes." Nature 432(7020): 1027-1032.

Wright Muelas, M., F. Ortega, R. Breitling, C. Bendtsen and H. V. Westerhoff (2018). "Rational cell culture optimization enhances experimental reproducibility in cancer cells." Sci Rep 8(1): 3029.

221 Wu, J. C., G. Merlino and N. Fausto (1994). "Establishment and characterization of differentiated, nontransformed hepatocyte cell lines derived from mice transgenic for transforming growth factor alpha." Proc Natl Acad Sci U S A 91(2): 674-678.

Wu, J. W., S. P. Wang, F. Alvarez, S. Casavant, N. Gauthier, L. Abed, K. G. Soni, G. Yang and G. A. Mitchell (2011). "Deficiency of liver adipose triglyceride lipase in mice causes progressive hepatic steatosis." Hepatology 54(1): 122-132.

Wu, Y. B., J. Dai, X. L. Yang, S. J. Li, S. L. Zhao, Q. H. Sheng, J. S. Tang, G. Y. Zheng, Y. X. Li, J. R. Wu and R. Zeng (2009). "Concurrent quantification of proteome and phosphoproteome to reveal system-wide association of protein phosphorylation and gene expression." Mol Cell Proteomics 8(12): 2809-2826.

Wunderlich, C. M., N. Hovelmeyer and F. T. Wunderlich (2013). "Mechanisms of chronic JAK-STAT3-SOCS3 signaling in obesity." Jakstat 2(2): e23878.

Wurie, H. R., L. Buckett and V. A. Zammit (2012). "Diacylglycerol acyltransferase 2 acts upstream of diacylglycerol acyltransferase 1 and utilizes nascent and de novo synthesized fatty acids in HepG2 cells." Febs j 279(17): 3033-3047.

Xie, X., P. Langlais, X. Zhang, B. L. Heckmann, A. M. Saarinen, L. J. Mandarino and J. Liu (2014). "Identification of a novel phosphorylation site in adipose triglyceride lipase as a regulator of lipid droplet localization." Am J Physiol Endocrinol Metab 306(12): E1449-1459.

Xie, X., Z. Yi, B. Bowen, C. Wolf, C. R. Flynn, S. Sinha, L. J. Mandarino and C. Meyer (2010). "Characterization of the Human Adipocyte Proteome and Reproducibility of Protein Abundance by One-Dimensional Gel Electrophoresis and HPLC-ESI-MS/MS." J Proteome Res 9(9): 4521-4534.

Xu, N., S. O. Zhang, R. A. Cole, S. A. McKinney, F. Guo, J. T. Haas, S. Bobba, R. V. Farese, Jr. and H. Y. Mak (2012). "The FATP1-DGAT2 complex facilitates lipid droplet expansion at the ER-lipid droplet interface." J Cell Biol 198(5): 895-911.

Xu, S., X. Zhang and P. Liu (2018). "Lipid droplet proteins and metabolic diseases." Biochim Biophys Acta Mol Basis Dis 1864(5 Pt B): 1968-1983.

Xu, W., L. Wu, M. Yu, F. J. Chen, M. Arshad, X. Xia, H. Ren, J. Yu, L. Xu, D. Xu, J. Z. Li, P. Li and L. Zhou (2016). "Differential Roles of Cell Death-inducing DNA Fragmentation Factor-alpha-like Effector (CIDE) Proteins in Promoting Lipid Droplet Fusion and Growth in Subpopulations of Hepatocytes." J Biol Chem 291(9): 4282- 4293.

Xu, Z. J., J. G. Fan, X. D. Ding, L. Qiao and G. L. Wang (2010). "Characterization of high-fat, diet-induced, non-alcoholic steatohepatitis with fibrosis in rats." Dig Dis Sci 55(4): 931-940.

Yamaguchi, T. and T. Osumi (2009). "Chanarin-Dorfman syndrome: deficiency in CGI-58, a lipid droplet-bound coactivator of lipase." Biochim Biophys Acta 1791(6): 519-523.

Yan, R., H. Qian, I. Lukmantara, M. Gao, X. Du, N. Yan and H. Yang (2018). "Human SEIPIN Binds Anionic Phospholipids." Dev Cell 47(2): 248-256.e244.

222 Yan, S., X. F. Yang, H. L. Liu, N. Fu, Y. Ouyang and K. Qing (2015). "Long-chain acyl-CoA synthetase in fatty acid metabolism involved in liver and other diseases: an update." World J Gastroenterol 21(12): 3492-3498.

Yang, L., Y. Ding, Y. Chen, S. Zhang, C. Huo, Y. Wang, J. Yu, P. Zhang, H. Na, H. Zhang, Y. Ma and P. Liu (2012). "The proteomics of lipid droplets: structure, dynamics, and functions of the organelle conserved from bacteria to humans." J Lipid Res 53(7): 1245-1253.

Yang, W. L., J. Wang, C. H. Chan, S. W. Lee, A. D. Campos, B. Lamothe, L. Hur, B. C. Grabiner, X. Lin, B. G. Darnay and H. K. Lin (2009). "The E3 ligase TRAF6 regulates Akt ubiquitination and activation." Science 325(5944): 1134-1138.

Yang, X., X. Lu, M. Lombes, G. B. Rha, Y. I. Chi, T. M. Guerin, E. J. Smart and J. Liu (2010). "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. D., D. X. Xiang and Y. Y. Yang (2016). "Role of E3 ubiquitin ligases in insulin resistance." Diabetes Obes Metab 18(8): 747-754.

Yazdi, M., A. Ahnmark, L. William-Olsson, M. Snaith, N. Turner, F. Osla, M. Wedin, A. K. Asztely, A. Elmgren, Y. M. Bohlooly, S. Schreyer and D. Linden (2008). "The role of mitochondrial glycerol-3-phosphate acyltransferase-1 in regulating lipid and glucose homeostasis in high-fat diet fed mice." Biochem Biophys Res Commun 369(4): 1065-1070.

Yen, C. L., S. J. Stone, S. Koliwad, C. Harris and R. V. Farese, Jr. (2008). "Thematic review series: glycerolipids. DGAT enzymes and triacylglycerol biosynthesis." J Lipid Res 49(11): 2283-2301.

Yetukuri, L., M. Katajamaa, G. Medina-Gomez, T. Seppanen-Laakso, A. Vidal-Puig and M. Oresic (2007). "Bioinformatics strategies for lipidomics analysis: characterization of obesity related hepatic steatosis." BMC Syst Biol 1: 12.

Younossi, Z., Q. M. Anstee, M. Marietti, T. Hardy, L. Henry, M. Eslam, J. George and E. Bugianesi (2018). "Global burden of NAFLD and NASH: trends, predictions, risk factors and prevention." Nat Rev Gastroenterol Hepatol 15(1): 11-20.

Younossi, Z. M., D. Blissett, R. Blissett, L. Henry, M. Stepanova, Y. Younossi, A. Racila, S. Hunt and R. Beckerman (2016). "The economic and clinical burden of nonalcoholic fatty liver disease in the United States and Europe." Hepatology 64(5): 1577-1586.

Younossi, Z. M., A. B. Koenig, D. Abdelatif, Y. Fazel, L. Henry and M. Wymer (2016). "Global epidemiology of nonalcoholic fatty liver disease-Meta-analytic assessment of prevalence, incidence, and outcomes." Hepatology 64(1): 73-84.

Zammit, V. A. (2013). "Hepatic triacylglycerol synthesis and secretion: DGAT2 as the link between glycaemia and triglyceridaemia." Biochem J 451(1): 1-12.

Zannis, V. I., S. Su and P. Fotakis (2017). "Role of apolipoproteins, ABCA1 and LCAT in the biogenesis of normal and aberrant high density lipoproteins." J Biomed Res.

223 Zhang, C., E. L. Klett and R. A. Coleman (2013). "Lipid signals and insulin resistance." Clin Lipidol 8(6): 659-667.

Zhang, C., L. Yang, Y. Ding, Y. Wang, L. Lan, Q. Ma, X. Chi, P. Wei, Y. Zhao, A. Steinbuchel, H. Zhang and P. Liu (2017). "Bacterial lipid droplets bind to DNA via an intermediary protein that enhances survival under stress." Nat Commun 8: 15979.

Zhang, D., Z. X. Liu, C. S. Choi, L. Tian, R. Kibbey, J. Dong, G. W. Cline, P. A. Wood and G. I. Shulman (2007). "Mitochondrial dysfunction due to long-chain Acyl-CoA dehydrogenase deficiency causes hepatic steatosis and hepatic insulin resistance." Proc Natl Acad Sci U S A 104(43): 17075-17080.

Zhang, H., Y. Wang, J. Li, J. Yu, J. Pu, L. Li, H. Zhang, S. Zhang, G. Peng, F. Yang and P. Liu (2011). "Proteome of skeletal muscle lipid droplet reveals association with mitochondria and apolipoprotein a-I." J Proteome Res 10(10): 4757-4768.

Zhang, J., D. Xu, J. Nie, R. Han, Y. Zhai and Y. Shi (2014). "Comparative gene identification- 58 (CGI-58) promotes autophagy as a putative lysophosphatidylglycerol acyltransferase." J Biol Chem 289(47): 33044-33053.

Zhang, S., M. W. Hulver, R. P. McMillan, M. A. Cline and E. R. Gilbert (2014). "The pivotal role of pyruvate dehydrogenase kinases in metabolic flexibility." Nutr Metab (Lond) 11(1): 10.

Zhang, W., S. Y. Bu, M. T. Mashek, O. S. I, Z. Sibai, S. A. Khan, O. Ilkayeva, C. B. Newgard, D. G. Mashek and T. G. Unterman (2016). "Integrated Regulation of Hepatic Lipid and Glucose Metabolism by Adipose Triacylglycerol Lipase and FoxO Proteins." Cell Rep 15(2): 349-359.

Zhang, X., X. Xie, B. L. Heckmann, A. M. Saarinen, T. A. Czyzyk and J. Liu (2014). "Targeted disruption of G0/G1 switch gene 2 enhances adipose lipolysis, alters hepatic energy balance, and alleviates high-fat diet-induced liver steatosis." Diabetes 63(3): 934-946.

Zhang, Y., Y. Zhang and Y. Yu (2017). "Global Phosphoproteomic Analysis of Insulin/Akt/mTORC1/S6K Signaling in Rat Hepatocytes." J Proteome Res 16(8): 2825-2835.

Zhao, Y., L. Gao, L. Xu, R. Tong, N. Lin, Y. Su, Y. Yan, Y. Gao, J. He, L. Kong, A. Yuan, Y. Zhuge, F. Wang and J. Pu (2018). "Ubiquitin-specific protease 4 is an endogenous negative regulator of metabolic dysfunctions in nonalcoholic fatty liver disease." Hepatology.

224