Characterisation of the first specific inhibitor of synthase 1

Xin Ying Lim

A thesis submitted for the fulfilment of the requirements for the degree of Doctor of Philosophy (PhD)

Prince of Wales Clinical School Faculty of Medicine University of New South Wales

May 2018

THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet

Surname or Family name: Lim

First name: Xin Ying

Abbreviation for degree as given in the University calendar: PhD

School: Prince of Wales Clinical School Faculty: Faculty of Medicine

Title: Characterisation of the first specific inhibitor of

Abstract (348 words): The signalling lipid ceramide has been implicated in the pathogenesis of obesity-related insulin resistance. Ceramide is synthesized by a family of six ceramide synthases in mammals (CerS1-6), each of which preferentially utilises fatty acids of particular acyl chain length (C14-C28). In (SkM), C18 ceramide, which is synthesized almost exclusively by ceramide synthase 1 (CerS1), has been suggested to mediate systemic insulin resistance in obese humans and rodent models. Therefore, targeted reduction of SkM C18 ceramide may improve detrimental metabolic outcomes induced by a high fat diet (HFD). This thesis describes the characterisation of a new, selective inhibitor of CerS1, P053. P053 inhibits CerS1 with an IC50 of 0.5 micromolar and selectively reduced C18 ceramide in cultured cells. Lipidomic profiling revealed that P053 administration specifically reduced C18 ceramide levels in SkM of mice fed with standard chow or HFD for 4-6 weeks. The reduction of SkM C18 ceramide was coupled to a compensatory increase in C24 , which are synthesized by CerS2, and loss of triacylglycerols. No other complex lipid species were affected by P053 treatment. At the physiological level, P053 impeded fat deposition induced by a HFD. Despite its effect on adiposity, P053 treatment had no effect on glucose tolerance and insulin sensitivity, assessed by glucose tolerance test and hyperinsulinemic-euglycemic clamps. Unexpectedly, prolonged CerS1 inhibition with P053 improved mitochondrial oxidative capacity in SkM. This effect was associated with significant increases in mitochondrial respiratory chain complex levels, increased activity of TCA cycle and β-oxidation , and increased mitochondrial respiratory capacity in SkM. Enhanced mitochondrial metabolism may therefore underlie the reduced lipid accretion in mice treated with P053. A significant positive correlation between SkM C18 ceramide and body adiposity was identified. In contrast, SkM C24 ceramides were inversely correlated with adiposity. Thus, P053 may reduce adiposity both directly by reducing SkM C18 ceramide and indirectly through the associated increase in C24 ceramides. In conclusion, we have generated the first selective CerS1 inhibitor. Results from this thesis revealed a potential new role for CerS1 as an endogenous inhibitor of mitochondrial oxidative function in SkM and regulator of whole-body adiposity.

Declaration relating to disposition of project thesis/dissertation

I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral theses only).

……………………………………………… ……………………………………..……… ……….……………………...… …………… ……… ….… Signature Witness Date

The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research.

FOR OFFICE USE ONLY Date of completion of requirements for Award:

I

ORIGINALITY STATEMENT

‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’

Signed ……………………………………………......

Date ……………………………………………......

II

Abstract

The signalling lipid ceramide has been implicated in the pathogenesis of obesity-related insulin resistance. Ceramide is synthesized by a family of six ceramide synthases in mammals (CerS1-6), each of which preferentially utilises fatty acids of particular acyl chain length (C14-C28). In skeletal muscle (SkM), C18 ceramide, which is synthesized almost exclusively by ceramide synthase 1 (CerS1), has been suggested to mediate systemic insulin resistance in obese humans and rodent models. Therefore, targeted reduction of SkM C18 ceramide may improve detrimental metabolic outcomes induced by a high fat diet (HFD).

This thesis describes the characterisation of a new, selective inhibitor of CerS1, P053. P053 inhibits CerS1 with an IC50 of 0.5 micromolar and selectively reduced C18 ceramide in cultured cells. Lipidomic profiling revealed that P053 administration specifically reduced C18 ceramide levels in SkM of mice fed with standard chow or HFD for 4-6 weeks. The reduction of SkM C18 ceramide was coupled to a compensatory increase in C24 ceramides, which are synthesized by CerS2, and loss of triacylglycerols. No other complex lipid species were affected by P053 treatment.

At the physiological level, P053 impeded fat deposition induced by a HFD. Despite its effect on adiposity, P053 treatment had no effect on glucose tolerance and insulin sensitivity, assessed by glucose tolerance test and hyperinsulinemic-euglycemic clamps. Unexpectedly, prolonged CerS1 inhibition with P053 improved mitochondrial oxidative capacity in SkM. This effect was associated with significant increases in mitochondrial respiratory chain complex levels, increased activity of TCA cycle and β-oxidation enzymes, and increased mitochondrial respiratory capacity in SkM. Enhanced mitochondrial metabolism may therefore underlie the reduced lipid accretion in mice treated with P053. A significant positive correlation between SkM C18 ceramide and body adiposity was identified. In contrast, SkM C24 ceramides were inversely correlated with adiposity. Thus, P053 may reduce adiposity both directly by reducing SkM C18 ceramide and indirectly through the associated increase in C24 ceramides.

In conclusion, we have generated the first selective CerS1 inhibitor. Results from this thesis revealed a potential new role for CerS1 as an endogenous inhibitor of mitochondrial oxidative function in SkM and regulator of whole-body adiposity.

III

Publications and presentations arising from this thesis

Publications: • Lim, X. Y., Pickford, R., and Don, A. S. (2016) Assaying Ceramide Synthase Activity In Vitro and in Living Cells Using Liquid Chromatography-Mass Spectrometry. Methods Mol Biol 1376, 11-22 • *Turner, N., *Lim, X. Y., Toop, H. D., Brandon, A., Osborne, B., Taylor, E. N., Fiveash, C. E., Teo, J., McEwen, H. P., Govindaraju, H., Couttas, T. A., Das, A., Kowalski, G. M., Bruce, C. R., Fath, T., Schmitz-Peiffer, C., Cooney, G., Montgomery, M. K., Morris, J. C., Don, A. S. (2018) A selective inhibitor of ceramide synthase 1 reveals a novel role in fat metabolism. Nature Communications (Accepted for publication) *Co-first authors

Presentations: • Lim, X. Y., Toop, H., Montgomery, M. K., Turner, N., Morris, J., Don, A. S. (2016) Title: Lipidomic characterization of the first specific inhibitor of Ceramide Synthase 1: Implications for Insulin Resistance. The Australian and New Zealand Metabolomics Conference, 30th March- 1st April, Melbourne, Australia. (oral presentation) • Lim, X. Y., Toop, H., Montgomery, M. K., Turner, N., Morris, J., Don, A. S. (2016) Title: Lipidomic characterization of the first specific inhibitor of Ceramide Synthase 1: Implications for Metabolic diseases. ComBio 2016, 3rd- 7th October, Queensland, Australia. (oral presentation) • Lim, X. Y., Toop, H., Montgomery, M. K., Fiveash, C., Osborne, B., Fath, T., Morris, J., Turner, N., Don, A. S. (2016) Title: Lipidomic characterization of the first specific inhibitor of Ceramide Synthase 1: Implications for Metabolic diseases. 3rd Australian Lipid Meeting, 21st -22nd November, Melbourne, Australia. (poster presentation; best poster prize) • Lim, X. Y., Toop, H., Montgomery, M. K., Morris, J., Turner, N., Don, A. S. (2017) Title: A potent and specific Ceramide Synthase 1 inhibitor: Implications for Metabolic Diseases. Keystone Symposia: Lipidomics and Bioactive Lipids in Metabolism and Disease, 26th February- 2nd March, California, USA. (poster presentation)

IV

Publications arising in conjunction with this thesis

• Couttas, T. A., Lim, X. Y., and Don, A. S. (2015) A three-step assay for ceramide synthase activity using a fluorescent substrate and HPLC. Lipids 50, 101-109

• Montgomery, M. K., Brown, S. H., Lim, X. Y., Fiveash, C. E., Osborne, B., Bentley, N. L., Braude, J. P., Mitchell, T. W., Coster, A. C., Don, A. S., Cooney, G. J., Schmitz-Peiffer, C., and Turner, N. (2016) Regulation of glucose homeostasis and insulin action by ceramide acyl-chain length: A beneficial role for very long- chain species. Biochim Biophys Acta 1861, 1828-1839

V

Acknowledgements

I dedicate this dissertation to my parents. Without their sacrifices I would have never make it this far. I am grateful for all their love and support. Mom and Dad, both of you have been extremely understanding and supportive. All these years, both of you never doubted me and are always concerned with my wellbeing. Mom, you have taught me perseverance and how to push myself. Dad, you have never asked for anything but for me to do my best and be happy. To Mom and Dad, thank you for making me the strong person I am today. I love you both very much!

To my best friend and partner: Luka. I am glad I have you throughout all these years. Despite the ups and downs, I know you will support me no matter what. I can’t thank you enough for all the things you do that make our lives easier. For that, I want to say thanks and I love you.

To my supervisor, Anthony Don, thank you all your support and guidance. I have learned a great deal from you. I really appreciate all the opportunities and independence you gave me. It has been a great experience to be on your team. To my co-supervisor, Nigel Turner, thank you for being supportive and kind throughout my PhD. I am grateful for all the help you have offered me. I have definitely learnt a lot from both of you.

I have met many amazing people during my PhD years. All of you have contributed to the journey over the last few years. A special thanks to Phil, John, Timmy, Swapna, Jai, Diego, Joyce, Weini, Regina, Collin, Hazem, Ashwin, Nupur, Julie, Andrea, Magda, Amanda, Brenna, Corrine, Hemna and all friends and colleagues in ACP Lowy.

Lastly, I want to thank Prince of Wales Clinical School, UNSW, and my supervisors for the scholarships and funding support over the years.

VI

List of Figures

Figure 1. 1. Overview of triacylglycerol synthesis pathways...... 7

Figure 1. 2. Overview of lipogenesis in hepatocytes and adipocytes...... 8

Figure 1. 3. Overview of lipolysis and FA metabolism...... 10

Figure 1. 4. β-oxidation pathway for the breakdown of FA and the TCA cycle...... 11

Figure 1. 5. The electron transport chain...... 12

Figure 1. 6. The de novo synthesis pathway of ...... 14

Figure 1. 7. Basic sphingolipid structures...... 16

Figure 1. 8. Dendrogram of the known mouse and human LASS family members and yeast Lag1p/Lac1p...... 18

Figure 1. 9. Schematic illustration of ceramide action in metabolic tissues...... 36

Figure 2. 1. Chromatograms for AAL(S) and P053...... 49

Figure 2. 2. LC-MS Chromatograms of lipids from mouse quadriceps muscles...... 52

Figure 3. 1. Chromatogram of a defined mixture of ceramides...... 72

Figure 3. 2. Standard curves used for quantification of ceramide, HexCer and SM...... 76

Figure 3. 3. Chromatograms showing peaks for D7-labelled C24:1 dihydroceramide products...... 77

Figure 3. 4. FB1 treatment inhibits D7-ceramides formation...... 78

Figure 4. 1. Compound structures and selectivity against CerS1...... 82

Figure 4. 2. Activity of overexpressed CerS isoforms...... 84

Figure 4. 3. Activity of CerS isoforms as a function of P053 concentration...... 85

Figure 4. 4. Kinetic analysis of CerS1 inhibition by P053...... 86

Figure 4. 5. P053 selectively inhibits de novo synthesis of deuterated C18:0 ceramide.87

VII

Figure 4. 6. RNA expression for CerS1 in a panel of cell lines...... 88

Figure 4. 7. P053 selectively reduced endogenous C18:0 ceramides in HEK293 cells. 89

Figure 4. 8. P053 selectively reduced endogenous C18:0 HexCer in HEK293 cells. ... 90

Figure 4. 9. P053 selectively reduced endogenous C18:0 SM in HEK293 cells...... 91

Figure 4. 10. Effect of long term P053 treatment on endogenous ceramide and SM levels...... 92

Figure 4. 11. Effect of P053 on HEK293 cell viability...... 93

Figure 4. 12. Ceramide composition of primary mouse myotubes and myoblasts...... 94

Figure 4. 13. Ceramide composition of C2C12 cells...... 94

Figure 5. 1. Mouse tissues CerS1 mRNA expression...... 102

Figure 5. 2. P053 bioavailability...... 102

Figure 5. 3. C18:0 ceramide levels in quadriceps SkM and ...... 103

Figure 5. 4. Tissue ceramide profiles in P053 pilot study...... 104

Figure 5. 5. Lipidome of quadriceps SkM...... 106

Figure 5. 6. Sphingolipids in quadriceps SkM...... 108

Figure 5. 7. Ceramide profiles in quadriceps SkM...... 108

Figure 5. 8. profiles in quadriceps SkM...... 109

Figure 5. 9. Ceramides and sphingomyelin profiles in gastrocnemius SkM...... 110

Figure 5. 10. Dihydrosphingosine, sphingosine and S1P levels in quadriceps SkM. ...111

Figure 5. 11. Phospholipids in quadriceps SkM...... 112

Figure 5. 12. Diacylglycerols and triacylglycerols in quadriceps SkM...... 113

Figure 5. 13. Liver ceramide and sphingomyelin levels...... 114

Figure 5. 14. Epididymal adipose tissue ceramide and sphingomyelin levels...... 115

VIII

Figure 5. 15. Ceramide levels as a function of total TG levels in quadriceps SkM...... 117

Figure 5. 16. C24 ceramide synthase activity in quadriceps SkM of P053 treated mice...... 119

Figure 5. 17. P053 does not affect brain ceramide and sphingomyelin levels...... 120

Figure 5. 18. P053 uptake in tissues and plasma...... 121

Figure 6. 1. Body composition and adiposity of animals...... 129

Figure 6. 2. Food intake is not affected by P053 treatment...... 130

Figure 6. 3. Effect of HFD and P053 treatment on glucose metabolism...... 131

Figure 6. 4. P053 did not prevent system insulin resistance in HFD-fed mice...... 133

Figure 6. 5. P053 did not prevent the reduction in glucose uptake into muscle and adipose tissues caused by a HFD...... 134

Figure 6. 6. Effect of P053 on SkM palmitate oxidative capacity...... 136

Figure 6. 7. P053 increases respiratory complex subunit levels in SkM but not liver. .138

Figure 6. 8. P053 increases mRNA levels for respiratory complex subunits in SkM. ..139

Figure 6. 9. P053 on oxidative activities in SkM and liver of mice...... 140

Figure 6. 10. SkM ceramide levels as a function of body fat in mice...... 142

IX

List of Tables

Table 1. 1. mRNA expression levels for CerS1-6 in mouse tissues and their substrate specificity towards fatty acyl-CoA with various chain lengths...... 17

Table 1. 2. Characteristics of CerS null mouse models...... 20

Table 2. 1. Parameter settings for P053 detection using LC-MS/MS...... 49

Table 2. 2. HPLC gradient for untargeted lipidomic profiling...... 50

Table 2. 3. Source running parameters for QExactive Plus mass spectrometer...... 51

Table 2. 4. LipidSearch lipid identification and quantitation parameters...... 53

Table 2. 5. LipidSearch software alignment parameters...... 54

Table 2. 6. Primer sequences used in qPCR amplification reactions...... 56

Table 2. 7. List of antibodies used for western blotting...... 66

Table 3. 1. Exact mass, column elution times and precursor and product ion masses [M+H] for dihydroceramide and ceramide species...... 71

Table 3. 2. List of sphingolipid metabolites analysed with LC-MS/MS...... 74

Table 4. 1. Calculated IC50 values for FB1, FTY720, AAL(S), G024, and P053 on multiple CerS isoforms...... 85

Table 5. 1. Correlations between quadriceps SkM ceramides and TG (4-week cohort)...... 116

Table 5. 2. Correlations between quadriceps SkM ceramides and TG (6-week cohort)...... 118

Table 6. 1. P053 on blood cell parameters...... 130

Table 6. 2. Characteristics of mice undergoing hyperinsulinemic-euglycemic clamp. 132

Table 6. 3. Correlations between SkM ceramides and body adiposity...... 141

X

Table A1. Statistically significant discoveries from multiple comparison analysis of 4- week quadriceps SkM lipidomic data...... 155

Table A2. Statistically significant discoveries from multiple comparison analysis of 6- week quadriceps SkM lipidomic data...... 158

Table A3. Statistically significant discoveries from multiple comparison analysis of 4- week liver lipidomic data...... 161

XI

Table of Contents

Thesis/Dissertation Sheet ...... I

ORIGINALITY STATEMENT ...... II

Abstract ...... III

Publications and presentations arising from this thesis ...... IV

Publications arising in conjunction with this thesis ...... V

Acknowledgements ...... VI

List of Figures ...... VII

List of Tables ...... X

Table of Contents ...... XII

Abbreviations ...... XVII

Chapter 1: Literature Review ...... 1

1.1. Obesity and Metabolic Syndrome ...... 1 1.1.1. Global Health Burden...... 1 1.1.2. Pathophysiology ...... 2 1.1.3. Treatment ...... 3

1.2. Lipotoxicity and Obesity ...... 3 1.2.1. Lipid mediators of lipotoxicity ...... 4

1.3. Lipid metabolism, a brief overview ...... 6 1.3.1. Anabolic metabolism ...... 6 1.3.2. Catabolic metabolism ...... 9

1.4. Introduction to sphingolipids ...... 12 1.4.1. De novo synthesis pathway ...... 12 1.4.2. Sphingolipid structure and nomenclature ...... 15

1.5. Ceramide synthases ...... 16 1.5.1. CerS-deficient mouse models ...... 19 1.5.2. CerS1 ...... 20 1.5.3. CerS2 ...... 22

XII

1.5.4. CerS3 ...... 22 1.5.5. CerS4 ...... 23 1.5.6. CerS5 ...... 23 1.5.7. CerS6 ...... 24

1.6. Ceramide metabolism ...... 24 1.6.1. Ceramide and ...... 24 1.6.2. Ceramide, insulin signalling and glucose metabolism ...... 26 1.6.3. Ceramide and mitochondrial functions ...... 27

1.7. Ceramide and Obesity complications ...... 28 1.7.1. Ceramide and obesity-induced inflammation ...... 28 1.7.2. Ceramide, adiponectin and FGF21 ...... 29 1.7.2. Ceramide and vascular functions ...... 30 1.7.3. Ceramide and brain lipotoxicity ...... 30 1.7.4. Ceramide and liver health ...... 32 1.7.5. Ceramide and pancreatic β-cell function ...... 33 1.7.6. Ceramide and insulin resistance ...... 33

1.8. Pharmacological inhibition of ceramide synthesis ...... 37 1.8.1. SPT inhibitor: Myriocin ...... 37 1.8.2. DEGS1 inhibitor: Fenretinide ...... 37 1.8.3. CerS inhibitor: Fumonisin B1 ...... 38 1.8.4. Fingolimod, FTY720...... 38 1.8.5. CerS isoform-specific inhibitor ...... 39

1.9. Aims of thesis ...... 40

Chapter 2: Methods ...... 41

2.1. Compounds ...... 41

2.2. Cell Culture Methods...... 41 2.2.1. Cell Culture conditions...... 41 2.2.2. Flow cytometry ...... 43 2.2.3. Transfection ...... 43

2.3. Ceramide synthase activity assays...... 44

2.4. Lipid extraction ...... 45

XIII

2.4.1. Tissue samples ...... 45 2.4.2. Cultured cells ...... 46 2.4.3. Plasma samples ...... 47

2.5. Liquid Chromatography Mass Spectrometry ...... 48 2.5.1. Targeted LC-MS/MS ...... 48 2.5.2. Untargeted lipidomic profiling ...... 50 2.5.3. Data processing and calculations ...... 52

2.6. Animal and Diet composition ...... 54 2.6.1. Assessment of body weight and body composition ...... 55 2.6.2. Tissue measurements ...... 55 2.6.3. Gene expression analysis using qPCR ...... 55 2.6.4. Assessment of metabolic parameters ...... 57

2.7. Analysis of insulin sensitivity by hyperinsulinemic-euglycemic clamp ...... 58 2.7.1. Dual cannulation surgery ...... 58 2.7.2. Hyperinsulinemic-euglycemic Clamp ...... 58 2.7.3. Hyperinsulinemic-euglycemic Clamp analytical methods ...... 59

2.8. Substrate utilization assays...... 61 2.8.1. Ex vivo skeletal muscle palmitate oxidation ...... 61 2.8.2. Fresh muscle homogenate palmitate oxidation...... 62 2.8.3. Palmitate oxidation in primary hepatocytes ...... 62

2.9. Enzyme activity assays ...... 63 2.9.1. (CS)...... 63 2.9.2. β-hydroxyacyl-CoA-dehydrogenase (β-HAD) ...... 63

2.10. Analysis of mitochondrial function in situ in permeabilized muscle fibres 64

2.11. Western Blotting ...... 65

2.12. Statistics...... 67

Chapter 3: Assaying Ceramide Synthase Activity in vitro and in Living Cells using Liquid Chromatography-Tandem Mass Spectrometry ...... 68

3.1. Introduction ...... 68

3.2. Aims ...... 69

XIV

3.3. Methods ...... 69 3.3.1. Assaying CerS activity in vitro using LC-MS/MS ...... 69 3.3.2. Assaying CerS activity in living cells ...... 73 3.3.2. Extension of LC-MS/MS method ...... 74

3.4. Results ...... 77 3.4.1. Validation of in vitro CerS activity assay ...... 77 3.4.2. Validation of cell-based CerS assay...... 78

3.5. Discussion ...... 78

Chapter 4: In vitro characterization of CerS1 inhibition by P053 ...... 81

4.1. Introduction ...... 81

4.2. Aims ...... 83

4.3. Results ...... 84 4.3.1. Validation of CerS overexpression ...... 84

4.3.2. IC50 of P053 ...... 84 4.3.3. Mode of inhibition of CerS1 by P053 ...... 86 4.3.4. P053 inhibits C18 ceramide synthase activity in cultured ...... 87 4.3.5. Effect of P053 treatment on endogenous ceramide levels ...... 88 4.3.6. Effect of P053 treatment on endogenous Hexosyl-ceramide levels ...... 90 4.3.7. Effect of P053 treatment on endogenous sphingomyelin levels ...... 91 4.3.8. Effect of long-term P053 treatment on sphingolipid levels ...... 92 4.3.9. P053 toxicity ...... 93 4.3.10. Ceramide profiling in cultured skeletal muscle cells ...... 94

4.4. Discussion ...... 95

Chapter 5: Lipidomic characterization of P053 in vivo ...... 100

5.1. Introduction ...... 100

5.2. Aims ...... 101

5.3. Results ...... 101 5.3.1. CerS1 expression in various tissues ...... 101 5.3.2. P053 bioavailability and pharmacokinetics ...... 102 5.3.3. Pilot in vivo study ...... 103

XV

5.3.4. Effects of P053 on the lipidome of skeletal muscle ...... 105 5.3.5. Effects of P053 on levels of sphingolipids in skeletal muscle ...... 107 5.3.6. Effects of P053 on levels of phospholipids in skeletal muscle ...... 112 5.3.7. Effects of P053 on levels of skeletal muscle diacylglycerol and triacylglycerol levels ...... 113 5.3.8. Effects of P053 on lipidomic profiles in other organs ...... 114 5.3.9. Correlation between levels of ceramides and TG in skeletal muscle ...... 116 5.3.10. Effects of P053 on CerS2 activity ...... 119 5.3.11. P053 does not affect brain ceramide and SM profiles ...... 120 5.3.12. Uptake of P053 ...... 121

5.4. Discussion ...... 122

Chapter 6: Physiological effects of CerS1 inhibition by P053 on glucose and fat metabolism in vivo ...... 126

6.1. Introduction ...... 126

6.2. Aims ...... 127

6.3. Results ...... 128 6.3.1. Physiological effects of HFD and P053 treatment in vivo ...... 128 6.3.2. Effects of HFD and P053 treatment on glucose tolerance in vivo ...... 130 6.3.3. Effects of HFD and P053 on whole body insulin action in vivo ...... 131 6.3.4. Effects of HFD and P053 on muscle insulin resistance ...... 133 6.3.5. Effects of P053 on muscle lipid oxidative capacity ...... 135 6.3.6. Effect of P053 on mitochondrial markers ...... 137 6.3.7. Correlations between muscle ceramides and body adiposity ...... 141

6.4. Discussion ...... 143

Chapter 7: Summary and Future Directions ...... 148

7.1. Potential limitations and future directions ...... 149

7.2. Overall therapeutic potential of targeting ceramide metabolism ...... 152

Appendix ...... 155

References:...... 163

XVI

Abbreviations ACC Acetyl-CoA carboxylase Acetyl-CoA Acetyl-coenzyme A AdipoR Adiponectin receptor ADP Adenosine diphosphate AMPK AMP-activated protein kinase ATP Adenosine triphosphate BAT Brown adipose tissue BMI Body Mass Index BSA Bovine serum albumin C1P Ceramide 1-phosphate CerS Ceramide synthase CNS Central nervous system CPT Carnitine palmitoyl- 1 CS Citrate synthase CVD Cardiovascular disease DEGS Dihydroceramide desaturase DG Diacylglycerol DhCer Dihydroceramide DMSO Dimethyl sulfoxide Drp-1 Dynamin related protein-1 EDL Extensor digitorum longus EpiWAT Epididymal white adipose tissue ER ERK Mitogen activated protein kinase/extracellular regulated kinase FA Fatty acids FATP Fatty acid transport protein FB1 Fumonisin B1 FGF21 Fibroblast growth factor 21 FTY720 Fingolimod G3P Glycerol-3-phosphate GalCer Galactosyl-ceramide GCS Glucosylceramide synthase

XVII

GluCer Glucosyl-ceramide GLUT Glucose transporter GRP78 Chaperone glucose-regulated protein 78 GTT Glucose tolerance test HDL Low high-density lipoprotein HexCer Hexosyl-ceramide HFD High fat diet HNSCC Head and neck squamous cell carcinoma HOX -like HPLC High-performance liquid chromatography IngWAT Inguinal white adipose tissue IRS Insulin receptor substrate JNK c-Jun N-terminal kinase LacCer Lactosylceramide LAG1p Longevity assurance gene 1 LASS Longevity Assurance LC-MS/MS Liquid Chromatography tandem mass spectrometry MG Monoacylglycerol MS Mass spectrometry MTBE Methyl-tert-butyl ether NAFLD Non-alcoholic fatty liver disease NEFA Non-esterified fatty acid NF-κB/IKK Nuclear factor kappa-light-chain-enhancer of activated B cells/ IκB kinase NLRP3 Nucleotide-binding domain, leucine-rich-containing family, pyrin domain containing 3 PI3K Phosphatidylinositide-3-kinase PKC Protein kinase C PP Protein phosphatase PPARγ Peroxisome proliferator-activated receptor gamma qPCR Real-time PCR RT Retention time S1P Sphingosine 1-phosphate SkM Skeletal muscle

XVIII

SM Sphingomyelin SMase Sphingomyelinases SPHK Sphingosine kinase SPT Serine palmitoyltransferase T2D Type 2 Diabetes TCA Tricarboxylic acid cycle TG Triacylglycerol TLC Thin layer chromatography TLR Toll-like receptor TNF-α Tumour necrosis factor-α VLDL Very low density lipoprotein WAT White adipose tissue β-HAD β-hydroxyacyl-CoA-dehydrogenase

XIX Chapter 1

Chapter 1: Literature Review

1.1. Obesity and Metabolic Syndrome

The world is experiencing an epidemic of obesity and metabolic syndrome, associated with an increased risk of all-cause and cardiovascular mortality. Despite the identification of a few associated with obesity and obesity-induced disorders1-4, the prevalence of obesity is believed to be mainly driven by environmental effects such as diet and exercise. Intake of high caloric and fatty foods combined with a sedentary lifestyle contribute to the development of obesity and related diseases. Obese individuals are at a high risk of developing various chronic diseases such as Type 2 diabetes (T2D), insulin resistance, cardiovascular diseases (CVD) and dyslipidaemia. These comorbidities are collectively referred to as the metabolic syndrome. Metabolic syndrome is diagnosed when there is a presence of three or more of these risk factors5.

1.1.1. Global Health Burden

Across the globe, the prevalence of obesity has more than doubled between 1980 and 2014. The World Health Organisation reported that more than 1.9 billion adults, 18 years and older, were overweight in 2014 and over 600 million of them were obese. Being overweight and obese is associated with a higher mortality rate than being underweight. Body mass index (BMI) is a simple measure of weight-for-height ratio and is generally used to classify overweight and obesity in adults. An adult with a BMI greater than or equal to 25 is considered overweight. A person with BMI of 30 or more is considered obese. Although BMI is a regular measurement for obesity among both sexes and all ages of adults, it does not indicate the level of adiposity6. In accordance with the obesity trends in the United States, it has been projected that approximately 30- 40% of adults have metabolic syndrome by age 65. The high prevalence of metabolic syndrome is associated with an increased risk of developing CVD and T2D7. Diabetes is a chronic disease defined by the inability of the pancreas to produce enough insulin or an inability of the body to effectively utilise the insulin that the pancreas produces. The International Diabetes Federation reported an estimated 415 million people worldwide living with diabetes, which is around 8.8% of all adults aged between 20 and 79. If this

1 Chapter 1

trend continues, by 2040, 642 million people, or one in ten adults, will have diabetes. Increased obesity is highly associated with the rising rates of diabetes worldwide8. Together, they place a huge burden on the global economy and health care expenditure.

1.1.2. Pathophysiology

An important component of metabolic syndrome and T2D associated with obesity is insulin resistance. Insulin resistance is defined by the diminished ability of insulin sensitive tissues to increase glucose uptake in response to insulin9. The pancreatic β- cells secrete more insulin to compensate for the lack of response, resulting in hyperinsulinemia. Over time, hyperglycemia will occur as the body’s insulin supply fails to keep up with demand. Eventually, the symptoms worsen and T2D develops10. One of the major contributors to the development of insulin resistance is an over- abundance of circulating free fatty acids (FA), released from expanded intra-abdominal fat mass, as a result of obesity11. Free FA reduce insulin sensitivity in muscle by reducing insulin-mediated glucose uptake12. In the liver, free FA also increase the production of glucose and triglycerides (also known as triacylglycerol, TG), and the secretion of cholesterol associated with very low density lipoproteins (VLDL)13-15. High blood pressure, or hypertension, is another pathophysiological feature of metabolic syndrome. Many obese individuals with insulin resistance also suffer from hypertension16. Weight reduction has been shown to improve both conditions17. A third core feature of metabolic syndrome in obese individuals is dyslipidaemia, characterized by high TG and low high-density lipoprotein (HDL) cholesterol levels. It is strongly associated with raised atherogenic VLDL cholesterol in the plasma18. This results in increased risk of CVD19. Intra-abdominal fat accumulation and large waist circumference also play a direct intermediary role in metabolic syndrome progression. Free FA released from expanded fat mass have been shown to interfere with hepatic insulin clearance13,14. The transformation of metabolically active intra-abdominal adipose tissues into a fat store affects the homeostasis of adipocytokine secretion, which affects energy homeostasis. Besides the factors mentioned above, there are many other conditions that provide clues to the pathophysiology of metabolic syndrome such as genetic variation, prolonged exposure to glucocorticoids and chronic stress16.

2 Chapter 1

1.1.3. Treatment

Treatment for obesity and metabolic syndrome varies greatly between individuals. Common effective treatment options include lifestyle changes such as weight reduction, diet improvement and increased physical activity. Surgical and pharmacological approaches can be considered for individuals with more severe symptoms where the preventative measures will not have adequate impact on the manifestation of the disorder. There is not one recognized method for tackling metabolic syndrome, instead the disorder is treated according to the individual features, primarily insulin resistance. A modest weight reduction of only 5-10% has been shown to reduce the risk of developing diabetes20. Exercise coupled with diet changes also produce positive effects such as stabilizing blood glucose levels, blood pressure, blood lipids and reducing overall adiposity. As for morbidly obese individuals, bariatric surgery has been demonstrated to be an effective treatment. Pharmacological approaches are preferred when treating individual risk factors. The most common being Metformin, for treating T2D by suppressing liver glucose production, angiotensin receptor blockers for hypertension, statins for lowering LDL cholesterol levels and orlistat, an inhibitor of gastrointestinal lipase, for weight management5.

1.2. Lipotoxicity and Obesity

Obesity is a type of metabolic stress that is typically associated with lipid accumulation in adipose tissue. However, prolonged nutrient excess through intake of high caloric and high fat diet (HFD) can lead to lipid influx that exceeds the adipose tissue storage capacity. Excess lipid influx can result in ectopic deposition of harmful lipids. The organs that are typically affected by lipotoxicity include the heart, liver, kidneys and skeletal muscle (SkM). The toxic effects of lipid accumulation in ectopic tissues are known as lipotoxicity, a term coined by Roger Unger in the early 1990s. It was hypothesized that over-supply of lipid is the early pathogenesis for metabolic disorders and diabetes21,22. Lipotoxicity has been further associated with the development of metabolic disorders such as non-alcoholic fatty liver disease (NAFLD), insulin resistance and cardiomyopathy23. There are two possible means whereby ectopic accumulation of lipid may occur. Firstly, enlarged adipocytes in obese adipose tissue have diminished ability to store fat in the form of TG, resulting in an increase in

3 Chapter 1

circulating free FA. However, this theory is controversial as in rodent models, insulin resistance in the liver has been shown to be established within one week of high fat feeding, while the expansion of adipose tissue continues several weeks or months24. The second possible mechanism is that the oversupply of FA obtained from diet leads to synthesis of lipotoxic lipid species such as diacylglycerol (DG) and ceramide25-28, which impair cellular metabolic functions. The following section reviews a few of the well- studied lipid mediators in lipotoxicity.

1.2.1. Lipid mediators of lipotoxicity

Elevated plasma free FA and TG levels are usually present in obesity29, probably resulting from free FA spillover from enlarged adipose tissues, and possibly reduced clearance of free FA from the plasma. Furthermore, once circulating free FA levels are elevated, they can exhibit inhibitory effects on insulin’s anti-lipolytic action on metabolic organs such as the liver and SkM30,31. Free FA have been shown to disrupt insulin-stimulated glucose uptake due to defects in insulin signalling. Besides that, elevated lipid influx can have an impact on cellular organelle structures. For instance, the endoplasmic reticulum (ER), a major centre for lipid synthesis and modification, is an important organelle that is sensitive to energy-related stresses. Chronic nutrient excess can cause dysregulated lipid synthesis in the ER, resulting in alteration of ER membrane phospholipid composition. These changes ultimately lead to prolonged ER stress, disrupted calcium signalling and decreased translation of ER-associated proteins32,33. Prolonged ER stress is highly associated with chronic inflammation via activation of the unfolded protein response, which intersects with multiple inflammatory pathways34. One of these pathways involves activation of pro- inflammatory NF- κB/IKK (nuclear factor kappa-light-chain-enhancer of activated B cells/ IκB kinase)35, which has been implicated as a contributing factor in insulin resistance. Other markers of ER stress include activation of inflammatory kinases such as protein kinase R and c- Jun N-terminal kinase (JNK), and activation of NLRP3 (nucleotide-binding domain, leucine-rich-containing family, pyrin domain containing 3) inflammasome34,35.

Chronic low-grade inflammation is one of the hallmarks of obesity-associated diseases and takes place in several metabolic tissues such the SkM, adipose tissue and liver. For instance, uncontrolled release of free FA can lead to inflammation of the adipose tissue, characterized by activation of toll-like receptors (TLR), increased release of cytokines, 4 Chapter 1

such as monocyte chemoattractant protein-1 and tumour necrosis factor-α (TNF-α), recruitment of immune effectors including macrophages and T cells, and decreased release of the anti-inflammatory adipokine adiponectin36-40. Without intervention, chronic lipotoxicity continues to drive inflammatory responses in adipose and other tissues.

Besides FA, other bioactive lipid intermediates are also associated with lipotoxicity. One of them is DG, which is heavily implicated as a mediator of insulin resistance. DG is generated at multiple subcellular sites such as the ER, plasma membrane and lipid droplets41. Because of the excess influx of FA, the ability to convert them to TG for storage is overwhelmed and this leads to the generation and accumulation of DG, an intermediate in the TG synthesis pathway25. HFD rich in oleate or linoleate unsaturated FA has also been shown to cause increased DG content in SkM and insulin resistance31. Intracellular accumulation of DG is thought to activate novel protein kinase C (PKC) isoforms, such as PKC-θ and PKC-ε42. PKC family members play vital roles in many signalling pathways, which control essential cellular functions, such as cell growth43, but may inhibit insulin receptor signalling, as described below.

Insulin signalling is initiated by the binding of insulin to its receptor at the plasma membrane. Insulin receptor is a membrane-spanning glycoprotein receptor with two extracellular α subunits and two intracellular β subunits. Binding of insulin to the α subunits results in a conformational change, which activates tyrosine kinase domains on the β subunits and subsequent auto-phosphorylation of tyrosine kinase residues in the β subunits. Tyrosine phosphorylation and activation of the docking proteins, insulin receptor substrate (IRS) then recruit and interact with phosphatidylinositide-3-kinase (PI3K) to induce downstream effects, which eventually lead to translocation of glucose transporter 4 (GLUT4) to the plasma membrane, facilitating glucose uptake44. The activation of PKCs by DG impairs the cascade of insulin signalling as PKCs, mainly PKC-θ and PKC-ε, phosphorylate IRS on a different site blocking IRS phosphorylation by insulin receptor45,46. Consequently, translocation of GLUT4 does not occur and glucose uptake is prevented.

Another heavily studied lipotoxic mediator is ceramide, which is the focus of this dissertation. Accumulation of ceramide has been reported in tissues of obese individuals47-49. Oversupply of substrates for ceramide synthesis, such as dietary 5 Chapter 1

palmitate, is thought to be the major source of ceramide accumulation in obesity. Ceramide belongs to the sphingolipid family and is a bioactive lipid with multiple signalling roles including activation of inflammatory pathways and inhibition of insulin actions. Ceramide can inhibit insulin-mediated stimulation of Akt/protein kinase B, an important serine/threonine kinase in anabolic signalling50-52. Furthermore, accumulation of ceramide in pancreatic β-cells is likely to contribute to apoptosis, suppression of insulin gene transcription and ER stress53-56. Ceramide has also been implicated in mitochondrial dysfunction and obesity-induced chronic inflammation57-61. The mechanisms of ceramide in metabolic diseases are discussed in detail later.

1.3. Lipid metabolism, a brief overview

A brief overview of lipid metabolism is required to lay the foundations for subsequent sections of this literature review, as well as the result and discussion chapters. Energy reserves are stored in adipose tissue in the form of TG. The free FA generated from the breakdown of TG can be used as substrates for lipid and membrane synthesis or as mediators for cell signalling. The complete oxidation of free FA in the mitochondria can be used to generate energy in the form of ATP. It involves the tricarboxylic acid cycle (TCA) and the electron transport chain, also known as the oxidative phosphorylation pathway.

1.3.1. Anabolic metabolism

FA are one of the major classes of molecules obtained through diet, along with protein and carbohydrates. As described above, FA themselves are important cellular signalling molecules. Once ingested, dietary fat usually in the form of TG is digested by pancreatic lipase into free FA and glycerol. Short-chain FA can be absorbed directly into the bloodstream but most FA are re-esterified into TG for transport in the form of lipoproteins, called chylomicrons. At the endothelial surface of capillaries, TG in chylomicrons are converted by lipoprotein lipase back into free FA and glycerol, which are then taken up by the adipose tissue for lipogenesis. Lipogenesis is the esterification of glycerol with free FA to form TG through the monoacylglycerol (MG) or glycerol phosphate pathways (Figure 1.1). TG are stored in the form of lipid droplets. Lipogenesis also occurs in the liver where FA are synthesized from acetyl-coenzyme A (acetyl-CoA) derived from glucose metabolism through a cascade of reactions (Figure

6 Chapter 1

1.2). FA synthesized in the liver can subsequently be esterified with glycerol to form TG, that are packaged in VLDLs and secreted back into the circulation. TG can then be released from VLDLs in the circulation through lipoprotein lipase activity in the capillary beds of various tissues. TG released can be stored or used for energy production62. Similar processes with slight modifications occur in other metabolic tissues such as SkM as well. An overview of lipogenesis in hepatocytes and adipocytes is shown in Figure 1.2.

Figure 1. 1. Overview of triacylglycerol synthesis pathways. The two metabolic pathways involved in the synthesis of TG. The monoacylglycerol (MG) pathway begins with the acylation of MG with fatty acyl-CoA catalysed by MG (MGAT). This pathway is involved in the dietary fat absorption in the small intestine. The glycerol phosphate pathway is a de novo TG synthesis pathway, which happens in most tissues. This pathway begins with the acylation of glycerol-3- phosphate (G3P) with fatty acyl-CoA by G3P acyltransferase (GPAT), producing lysophosphatidic acid (LPA). This is followed sequentially by further acylation by LPA acyltransferase (LPAAT) and dephosphorylation by phosphatidic acid (PA) phosphorylase (PP) to yield diacylglycerol (DG). The two pathways share the final step in converting DG into TG by DG acyltransferase (DGAT).

7 Chapter 1

Figure 1. 2. Overview of lipogenesis in hepatocytes and adipocytes.

(A) Glucose is transported into hepatocytes through glucose transporter 2 (GLUT2) for the generation of pyruvate through glycolysis. Pyruvate is then oxidized by pyruvate dehydrogenase in the mitochondria into acetyl-CoA. The acetyl-CoA enters the TCA cycle generating citrate, which exits into the cytosol where ATP citrate (ACL) convert it into acetyl-CoA again. In the cytosol, acetyl-CoA is carboxylated into malonyl-CoA by acetyl-CoA carboxylase (ACC) and used to generate acyl-CoA through FA synthase (FAS). Glucose can also enter the pentose phosphate pathway generating NADPH to be used in acyl-CoA production. The acyl-CoA produced through the above means and the intermediates of the glycolytic pathway, glycerol-3- phosphate (G3P) are used to synthesize triacylglycerol (TG) and VLDL. TG are then transported to the adipocytes for storage in the form of VLDL and chylomicrons (CM). Lipoprotein lipases (LPL) hydrolyse TG in VLDL and CM into fatty acids (FA) to be transported into the cell through various FA transport proteins. In the cytosol of adipocytes, acyl-CoA synthases utilize FA to generate acyl-CoA. Glucose transporter 4 (GLUT4) is responsible for the uptake of glucose into the cytosol of adipocytes. Similar to the hepatocytes, glucose is used to generate G3P and FA through glycolysis and de novo lipogenesis respectively. TG generated using acyl-CoA and G3P can then be stored in lipid droplets. Image adapted from63.

8 Chapter 1

1.3.2. Catabolic metabolism

Lipolysis is the process where TG are hydrolytically cleaved to generate free FA and glycerol. Lipolysis is well characterized in the adipose tissues and is mainly driven by three enzymes, adipose TG lipase, hormone-sensitive lipase and MG lipase (Figure 1.3 A). The glycerol generated from lipolysis can be converted into glyceraldehyde 3- phosphate, then enter the glycolytic pathway for energy production. An overview of FA metabolism is shown in Figure 1.3 B. Free FA from lipolysis either diffuse or are transported across the plasma membrane64. FA transporters have been implicated in metabolic diseases as they play an important role in facilitating and regulating FA uptake into cells. A few of the transporters that have been identified are CD36, a family of FA transport proteins (FATP 1-6), and plasma membrane-associated FA-binding proteins65. FA oxidation (β-oxidation) first involves the addition of CoA by acyl-CoA synthetases in the cytosol. The resultant fatty acyl-CoA can then be transferred to the mitochondria via carnitine palmitoyl-transferase (CPT) where β-oxidation takes place (Figure 1.4 A). The end products of β-oxidation are acetyl-CoA, a shorter fatty acyl-

CoA, FADH2 and NADH. Acetyl-CoA is then used as a fuel source for the TCA cycle

(Figure 1.4 B) while FADH2 and NADH can be used for ATP generation via the electron transport chain (Figure 1.5)62.

9 Chapter 1

Figure 1. 3. Overview of lipolysis and FA metabolism.

(A) The breakdown of TG begins with the hydrolysis of TG by adipose triglyceride lipase (ATGL) or hormone-sensitive lipase (HSL) to generate DG and non-esterified free FA (NEFA). HSL is then responsible for the hydrolysis of DG generating NEFA and MG. Finally, MG lipase (MGL) cleaves MG into glycerol and NEFA. (B) FA primarily enter a cell via various FA transporter proteins. Once in the cytosol, FA are used to generate acyl-CoA, catalysed by fatty acyl-CoA synthases (FACS). Carnitine palmitoyl-transferase 1 (CPT1) then convert the acyl- CoA into acyl-carnitine, which can then be transported by carnitine (CAT) across the inner mitochondrial membrane by exchanging acyl-carnitine for carnitine. CPT2 then converts acyl-carnitine back into acyl-CoA before entering the FA β-oxidation pathway, resulting in the generation of acetyl-CoA. The acetyl-CoA then enters the TCA cycle. NADH and FADH2 are generated by both β-oxidation and the TCA cycle, which are then used by the electron transport chain to produce energy in the form of ATP. Image adapted from63.

10 Chapter 1

(A)

Figure 1. 4. β-oxidation pathway for the breakdown of FA and the TCA cycle.

(A) β-oxidation involves a repeated cascade of four reactions whereby fatty acyl-CoA is broken down to generate acetyl-CoA, NADH and FADH2 that are used to generate energy via the TCA cycle and oxidation phosphorylation pathway. The β-oxidation cycle continues until all the carbons of the original FA have been used to generate acetyl-CoA. (B) Acetyl-CoA generated from nutrient oxidation enters the TCA cycle, which consists of a series of reactions that 62 generate NADH and FADH2. Images adapted from

11 Chapter 1

Figure 1. 5. The electron transport chain.

Complex I to V of the electron transport chain are located on the inner mitochondrial membrane. Each complex is a large multi-subunit assembly of proteins. They catalyse the oxidation of NADH and FADH2, which drives electron transport through the respiratory complexes producing a proton gradient across the inner mitochondrial membrane. ATP is synthesized when protons flow back into mitochondrial matrix through complex V. Image adapted from 62.

1.4. Introduction to sphingolipids

Sphingolipids are a diverse class of lipids found in all eukaryotes. First discovered in brain extracts in 1884 by Thudichum, sphingolipids are named after the mythological Sphinx due to their enigmatic nature66. In mammals, the sphingolipid backbone generally consists of eighteen carbon amino-alcohols termed sphingoid bases (Figure 1.6), which are synthesized in the ER. Variation of the acyl chain length, the degree of saturation of FA chains, and glycosylation of the hydroxyl headgroup give rise to the vast diversity of sphingolipids (Figure 1.6), each contributing to various cellular functions including cell membrane structure, cell-cell recognition, modulation of receptor signalling, and transcription regulation67,68.

1.4.1. De novo synthesis pathway

In mammalian cells, the sphingolipid metabolic pathway begins with de novo biosynthesis of ceramide in the ER. The first step in the synthesis pathway is the condensation of serine and palmitoyl-CoA by the enzyme serine palmitoyltransferase (SPT), producing 3-keto dihydrosphingosine, which is then reduced to dihydrosphingosine (Figure 1.6). Ceramide synthases (CerS1-6) then catalyse the N- 12 Chapter 1

acylation of dihydrosphingosine, producing dihydroceramide (dhCer). DhCer is rapidly desaturated by dhCer desaturase (DEGS1/2), forming ceramide. Ceramide is the lipid precursor for subsequent synthesis of complex sphingolipids. Ceramide is not a single entity but is instead a class of molecules with varying length and degree of unsaturation in the variable N-acyl chain (Figure 1.7). Ceramide transfer protein or cytoplasmic vesicles shuttle ceramides formed in the ER to the , where ceramide can be (i) phosphorylated by ceramide kinase, forming ceramide 1-phosphate (C1P); (ii) galactosylated, forming galactosyl-ceramide (GalCer), which can subsequently become sulfatide with the addition of sulfate groups; (iii) converted into sphingomyelin (SM) by enzymatic transfer of a phosphocholine head group by SM synthases; or (iv) glucosylated by glucosylceramide synthase (GCS), yielding glucosyl-ceramide (GluCer). GluCer can further be modified with sequential addition of different monosaccharide units to form complex glycosphingolipids within the globoside and ganglioside families. Ceramide can also be catabolised by ceramidases, producing sphingosine. Sphingosine kinases (SPHK1 or 2) subsequently phosphorylate sphingosine, forming sphingosine 1-phosphate (S1P). At this point, S1P can either be secreted, inducing various cell signalling responses, or degraded by S1P lyase. Degradation of S1P by S1P lyase is the only known “exit” of carbon from the sphingolipid pathway, yielding ethanolamine phosphate and 2-hexadecanal, which can be recycled into palmitoyl-CoA and feed back into the de novo synthesis pathway again.

13 Chapter 1

Figure 1. 6. The de novo synthesis pathway of sphingolipids.

Abbreviations: ASA: arylsulfatase A; CDase: ceramidase; aSMase: acid sphingomyelinase; Cer: ceramide; C1P: ceramide 1-phosphate; CERK: ceramide kinase; CerS: ceramide synthase; CERT: ceramide transfer protein; CGT: ceramide galactosyltransferase; CST: cerebroside sulfotransferase; DEGS1/2: dihydroceramide desaturase 1 or 2; DhCer: dihydroceramide; DhSph: dihydrosphingosine; GalCer: galactosylceramide; GCS: glucosylceramide synthase; GluCer: glucosylceramide; GSL: glycosphingolipid; HD: 2-hexadecanal; KDS: 3- ketosphinganine reductase; LacCer: lactosylceramide; LacCer-S: lactosylceramide synthase; nSMase: Neutral sphingomyelinase; P-CoA: palmitoyl-CoA; PE: ethanolamine phosphate; S1P: sphingosine 1-phosphate; S1PR: sphingosine 1-phosphate receptor; SGPL: S1P lyase; SGPP1/2: sphingosine 1-phosphate phosphatase 1 or 2; SM: sphingomyelin; SPHK1/2 : sphingosine kinases 1 or 2; Sph: sphingosine; SPT: serine palmitoyltransferase. Image adapted from 69.

Ceramide is the central metabolite of the sphingolipid family. Ceramide may be formed through a number of other pathways in addition to the de novo synthesis pathway. The first is the SM hydrolytic pathway (Figure 1.6). This pathway involves the hydrolysis of the phosphocholine head group of SM by a family of sphingomyelinases (SMase) either at the plasma membrane or in the lysosome, forming ceramide. There are five known SMases, termed acid, neutral or alkaline SMase, which are distinguished by their pH

14 Chapter 1

optima for enzymatic activity70. Neutral SMases71 usually reside at the plasma membrane, whereas the acid SMases72 are localised in organelles such as lysosomes or at the extracellular surface of plasma membranes. Dietary SM can be hydrolysed by alkaline SMase in the digestive tract and liver73. SM is also the most abundant sphingolipid species, therefore its capacity as a source of the total ceramide pool is significant.

The second pathway is the salvage pathway (Figure 1.6). Ceramide can be produced from the catabolism of other sphingolipids species through a series of events. Higher- order sphingolipids such as SM and glycosphingolipids can be degraded into ceramides within the lysosomes by SMases and glycosidases. The ceramides generated are degraded into FA and sphingosine before entering the cytosol. Sphingosines are then converted back into ceramide by CerSs. The cycle continues as the ceramides generated are used for production of other sphingolipid species as needed.

1.4.2. Sphingolipid structure and nomenclature

Sphingolipid structure is defined by the sphingoid base backbone. The variable N-acyl chain added to the sphingoid base by CerS can vary in headgroups, alkyl chain length and the number and position of double bonds and hydroxyl groups (Figure 1.7)74. The simple nomenclature for the sphingoid base is to provide the number of hydroxyl groups followed by the chain length and number of double bonds. For complex sphingolipids, information about the fatty acyl chain can be given in the second half of the abbreviation. For example, a C16:0 ceramide is given an abbreviation of Cer(d18:1/16:0). The first half, d18:1 gives information about the sphingoid base; d represents two (di) hydroxyl groups; a chain length of 18 carbon atoms and one double bond. The second half, 16:0 gives information about the amide-linked fatty acid chain, consisting of sixteen carbon atoms with no double bonds. A C16:0 dihydro-ceramide (C16 dhCer) is abbreviated Cer(d18:0/16:0), as it lacks a double bond in the sphingoid base. The amide linked fatty acyl chain can vary further in number of carbon atoms, number of double bonds and presence of hydroxyl groups75. The most common sphingoid base for sphingolipids is d18:1. To simplify, sphingolipid species will be abbreviated Ca:b from herein, where a represents the number of carbons in the fatty acyl chain and b gives the number of double bonds, e.g. C18:1 Cer. Dihydro sphingolipid species will be given an abbreviation “dh”, e.g. C18:1 dhCer76. 15 Chapter 1

Figure 1. 7. Basic sphingolipid structures.

(A) The sphingoid base, sphingosine is linked by an amide bond to a fatty acid tail of variable length (14- 28 carbon). If the “-X” is a hydrogen (H), then the resulting SL is ceramide. If the “- X” is a sugar, then the resulting sphingolipid is a glycosphingolipid. (B) C18:0 Cer. (C) Addition of a galactose to “-X” yields C18:0 GalCer. Different headgroups added to “-X” yield the different sphingolipid classes.

1.5. Ceramide synthases

The dominant feature distinguishing different forms of ceramide is variation in the length of their fatty acyl chain. In mammalian cells, the 18-carbon sphingoid base is the most prevalent, but 16- and 20-carbon sphingoid bases do exist. There are six known mammalian CerSs, CerS1-6, which catalyse the N-acylation of dihydrosphingosine or sphingosine. Each CerS isoforms preferentially transfers different length FAs, producing ceramide species with different fatty acyl chain length (Table 1.1). For example, CerS1 preferentially transfers 18-carbon stearic acid, forming C18:0 ceramide. CerS2 is selective for very long chain FA, ranging from 20-26 carbons in length, whilst CerS5 and CerS6 are selective for the 16-carbon FA, palmitate. Therefore, the ceramide pool may vary from organ to organ and cell to cell, in proportion to the relative expression of the different CerS isoforms77-80. Hydroxylation of the N-acyl FA chain is typically found in the kidneys and brain81. This further adds to the complexity of the ceramide family.

16 Chapter 1

Table 1. 1. mRNA expression levels for CerS1-6 in mouse tissues and their substrate specificity towards fatty acyl-CoA with various chain lengths. Adapted from82. Enzyme Acyl-chain-length Tissue mRNA expression profile Name specificity CerS1 C18:0 Brain, skeletal muscle, testis CerS2 C20:0 - 26:0 , liver CerS3 C22:0 - 26:0 Skin, testis CerS4 C18:0 - 20:0 Low expression in all tissues, more in skin, heart and liver CerS5 C16:0 Low expression in all tissues CerS6 C14:0 - 16:0 Low expression in all tissues

The CerSs, originally known as Longevity Assurance (LASS) proteins, share homology with the yeast protein, longevity assurance gene 1 (LAG1p)83-85. It was named LAG1 because its deletion was found to prolong the life cycle of the yeast S. cerevisiae86. Another close homolog of LAG1 was identified, named LAC187. In 2002, the first mammalian CerS, UOG-1 was discovered84. It was demonstrated that over-expression of UOG-1 increases ceramide synthesis, more specifically C18:0 ceramide synthesis in human embryonic kidney (HEK-293) cells. It was later discovered through homology analysis that there are other mammalian LAG homologs87,88. There are now six known mammalian CerS isoforms. The similarity and distinction of the LASS/CerS proteins can be viewed in the form of a dendrogram demonstrating how CerS1/Lass1 is the most distinct isoform in terms of its primary amino acid sequence79 (Figure1.8).

17 Chapter 1

Figure 1. 8. Dendrogram of the known mouse and human LASS family members and yeast Lag1p/Lac1p. Sequences from the LASS family members of mouse, human and yeast were aligned using ClustalW. The GenBank accession numbers for human LASS family members are M62302 (hLASS1), BC010032 (hLASS2), BC010032 (hLASS3), AK022151(hLASS4), BC03 2565 (hLASS5) and BC030800 (hLASS6), and those for yeast Lag1p/Lac1p members are NP_011860 (sLag1p) and NP_012917(sLac1p). The scale bar indicates the genetic distance for 0.1 amino acid substitution per site. Image from79.

There has been little work done on the structure of CerS. In fact, there are no full structural models for any of the CerS. However, it is known that the LAG1p motif conserved in all CerS isoforms contains two histidine residues that are critical for catalysis or substrate binding89,90. Mutation of the two histidine residues resulted in the loss of catalytic activity of mammalian CerS190 and yeast Lag191. The for dihydrosphingosine shows similarity across all CerS irrespective of the fatty acyl CoA substrate80. Futerman et. al. had shed light on possible modes of substrate binding using a sphingosine analogue, FTY720. It is suggested that there may be two dihydrosphingosine binding sites, which act allosterically with respect to one another. It may also be possible that CerS themselves form dimers92. Mammalian CerS can have 5- 8 transmembrane domains79,91. CerS2-6 all have a homeobox-like (HOX) domain, whilst CerS1 does not. The HOX domain is highly conserved in transcription factors associated with developmental regulation93. Most CerSs do not require the HOX-like domain for catalytic activity, except for CerS5-6, which share a high conserved 12 amino acid region at the C-terminal end of the HOX-like domain94. More recently, Voelzmann et al. have reported that the nuclear Drosophila CerS, named Schlank, can

18 Chapter 1

regulate ceramide synthesis and fat metabolism through its HOX-like domain independent of the common catalytic LAG1p motif95.

1.5.1. CerS-deficient mouse models

A panel of CerS deficient mouse models has been established in the past decade. Each of the mouse models lacking specific CerS isoforms displays overlapping phenotypes with certain human disease pathologies. Hence, studies using these mouse models will provide better understanding of the human diseases in which CerS and ceramide are involved (Table 1.2).

19 Chapter 1

Table 1. 2. Characteristics of CerS null mouse models.

Mouse Tissue Phenotypes References • death in • Lipofuscin accumulation CerS1 • Neuronal apoptosis and shrinkage of cerebellum Brain 96,97 null and forebrain • Cerebellar ataxia • Reduced -associated glycoprotein • Myelin sheath defects Brain 71,98 • Vacuolization and gliosis • Increased airflow resistance and lung volume Lung • Foamy macrophage infiltration 99 CerS2 • Inflammation null • Increased membrane fluidity • Hepatoadenoma Liver • Hypoglycaemia and insulin resistance 71,100-105 • Mitochondrial dysfunction • Increased hepatocytes turnover • Lack of continuous extracellular lipid lamellae • Hyperkeratosis, deficient cornification CerS3 Skin • Increased susceptibility to candida albicans 106 null infections • Death after birth from trans-epidermal water loss CerS4 • Altered lipid composition of sebum Skin 107 null • Progressive hair loss and alopecia • Prevented diet-induced obesity Whole • Protected from obesity-induced glucose tolerance CerS5 Body and insulin resistance 108 null Adipose • Improved adipose tissue health after high fat tissue feeding • Behavioural abnormalities in clasping of hind Brain CerS6 limbs and habituation deficit 58,109 null Whole • Protected from high-fat-diet-induced obesity and body glucose intolerance

1.5.2. CerS1

CerS1, as mentioned above is functionally and structurally distinct from other CerS. It is the most closely related to yeast Lag1, which functions without a HOX-like domain94. CerS1 has a relatively short half-life. CerS1 turnover happens rapidly through

20 Chapter 1

ubiquitination and proteasomal degradation under stresses including ultra-violet radiation and drug treatment110. Proteolytic cleavage and degradation of the N-terminus of CerS1 leaves a 17kDa fragment that subsequently translocates from the ER to the Golgi111.

According to the mRNA expression profiles of CerS in various human and mouse tissues (Table 1.1), CerS1 is highly expressed in the brain and SkM112. The high expression of CerS1 in the brain is consistent with the profiles of neuronal sphingolipid species, consisting mainly of C18:0 and C22:0-26:0 lipid species113. CerS1 is essential for cerebellar development and regulation of the associated behaviour97. Ablation of neuronal CerS1 in mice resulted in neuronal apoptosis, foliation defects and progressive shrinkage of the cerebellum, with a 60% decrease in gangliosides levels in the forebrain and cerebellum97. The decrease in C18:0 chain sphingolipids was coupled with functional and behavioural deficits in motor coordination, locomotion, and exploration of novel objects. Another study with two CerS1 knockout mouse models also demonstrated similar degeneration of cerebellar Purkinje neurons and build-up of lipofuscin96.

In addition to brain function, CerS1 has been implicated in head and neck squamous cell carcinoma (HNSCC) in the literature. Decreased C18:0 ceramide level was reported in the majority of HNSCC tumour tissues114,115. Increased generation of C18:0 ceramide by mouse LAG1 has been shown to inhibit the growth of UM-SCC-22A squamous cell carcinoma cell line via induction of apoptotic cell death, which is mediated through mitochondrial dysfunction and modulation of telomerase activity114. Interestingly, C16:0 ceramide produced by CerS6 promotes survival in HNSCC114,115. CerS1 has also been found to sensitize various cell types to such as , , , gemcitabine, and imatinib through generation of C18:0 ceramide and caspase activation116,117.

C18:0 ceramide produced by CerS1 has also been implicated in SkM insulin resistance. Accumulation of C18:0 ceramide is prevalent in SkM of insulin-resistant obese individuals and diet-induced obese rodent models24,48.

21 Chapter 1

1.5.3. CerS2

CerS2 synthesizes very long chain ceramide species. CerS2 mRNA is expressed in most tissues with highest expression in the kidneys and liver82. In mouse brain, CerS2 is predominantly expressed in the Schwann cells and oligodendrocytes, especially along the tracts113,118. CerS2 is found to be upregulated during the first 3 weeks of mouse development when there is active myelination113. This implies a plausible role for very long chain ceramide produced by CerS2 in myelin production. CerS2 has also been implicated in autophagy. Knock-down of CerS2 can lead to induction of autophagy119. Furthermore, elevated ceramide levels have been implicated in both acute and chronic lung injury models. In this regard, very long chain ceramide has been shown to be vital for lung health and CerS2 dysfunction can lead to pulmonary diseases and airway inflammation99. Another important feature of very long chain ceramide and CerS2 is their impact on the biophysical properties of membranes. Genetic ablation of CerS2 can lead to changes in membrane curvature, which in turn lead to morphology modifications promoting membrane fusion, tubule formation and vesicle adhesion102. This demonstrates the importance of very long chain ceramide and CerS2 functions on cellular processes dependent on membrane structure102.

CerS2 haploinsufficiency leads to reduced FA oxidation and increased susceptibility to liver steatosis and insulin resistance59,103. Very long chain ceramide has been shown to be insulin-sensitizing59. This insulin-sensitizing function of CerS2 contrasts with CerS6, which promotes insulin resistance (discussed below).

1.5.4. CerS3

CerS3 synthesizes the longest forms of ceramide and is found mainly in the testes and skin120,121. Its high expression in increases further during terminal differentiation122. CerS3 synthesizes hydroxyl-ceramides that are important in structuring and maintaining the water permeability barrier function of the skin123-125. In the testes, CerS3 was suggested to be involved in androgen production and sperm formation120. CerS3 mRNA expression is up-regulated more than 700-fold in the germ cells during juvenile testicular maturation. The increased expression of CerS3 mRNA corresponds with the increase of long chain glycosphingolipids that are essential during spermatogenesis121.

22 Chapter 1

1.5.5. CerS4

CerS4 synthesizes ceramide with 18:0-22:0 fatty acyl tails and is expressed highly in leukocytes, skin, liver and the heart112. The absolute mRNA expression of CerS4 is the lowest among all other CerS82. CerS4 has been reported to regulate the epidermal ceramide pool that is essential to hair follicle stem cell homeostasis and cycling, mediated through Wnt and bone morphogenetic protein signalling126. Lack of CerS4 expression in CerS4-/- mice models led to a decrease in C20:0 1,2-alkane diols and wax esters in the sebum and resulted in alopecia107. The CerS4 deficient mice also presented dilated and obstructed pilary canals, resulting in epidermal tissue destruction. This points to another CerS isoform being involved in skin homeostasis, other than CerS3. CerS4 has also been implicated in the pathobiology of cancer. In liver cancer, CerS4 was suggested to regulate cell proliferation and tumour growth in vitro and in vivo. Knockdown of CerS4 reduced the size and volume of tumours and affected the NF-κB signalling pathway127.

1.5.6. CerS5

CerS5 was the first CerS to be purified, in 2005128. It is studied comprehensively due to its ability to synthesize pro-apoptotic C16:0 ceramide. Kitatani and colleagues reported that the accumulation of C16:0 ceramide formed from the PKC-dependent sphingolipid salvage pathway was at least in part derived from CerS5 activity129. The accumulated ceramide could modulate p38-MAPK signalling via protein phosphatase (PP) 1A activation, resulting in an inhibitory effect on p38130. CerS5 also plays a role in cellular stress responses131. CerS5 mRNA expression was increased along with acid SMase activation upon hypoxic injury of neuronal precursor cells, which eventually ended with apoptosis. This response can be partially rescued by the knockdown of CerS5 and acid SMase, which hinder the increase in ceramide131. These data support the idea that CerS5 and acid SMase work together in the salvage pathway to generate C16:0 ceramide and induce cell death. In the context of metabolic diseases, CerS5 has been found to mediate lipid-induced autophagy and hypertrophy in cardiomyctes132. C16:0 ceramide produced by CerS5 was also implicated as a culprit in diet-induced disruption of glucose homeostasis in white adipose tissue (WAT)108. Deletion of CerS5 improved glucose tolerance.

23 Chapter 1

1.5.7. CerS6

CerS6 catalyses the addition of C14:0 and C16:0 FA to dihydrosphingosine. CerS6 shares high homology with CerS5. A microarray study has implied that CerS6 is also involved in early embryonic development133. Some studies have suggested the involvement of CerS6 in cancer aetiology. One study suggested that CerS6 is regulated by estrogen receptor signalling, and that its mRNA expression was raised in breast cancer tissues134-136. Overexpression of CerS6 can have a tumour cell killing effect through CD95 activation whereas its knockdown repressed CD95 activation in pancreatic, ovarian and hepatic cancer cells137. Similar to CerS5, CerS6 has also been implicated in obesity-induced complications. Recently, a report indicated that C16:0 ceramide production by CerS6 promotes weight gain and glucose intolerance, and its deficiency improves the condition by increasing lipid utilization in brown adipose tissue (BAT) and liver58,59.

1.6. Ceramide metabolism

Ceramide is a bioactive lipid that has been implicated in mediating multiple cellular processes, including apoptosis, arrest, stress responses, inflammatory signalling and glucose metabolism. As described above, ceramide can be generated through the de novo synthesis pathway, the SM hydrolytic pathway and the salvage pathway. Ceramide mechanisms of action are described below with respect to the key cellular processes.

1.6.1. Ceramide and apoptosis

One of the earliest reports on ceramide action was on the regulation of cell survival, where addition of C2:0 ceramide analogues induced marked DNA fragmentation in U- 937 human cells138. Ceramide can directly bind and activate the protein phosphatases, PP1A and PP2A in vitro139. Activation of PP2A by ceramide can promote apoptosis through the inactivation of Akt, a central anti-apoptotic signalling node. This promotes the activation of pro-apoptotic Bax and Bad131,140-142. Furthermore, ceramide-induced activation of PP1A promotes apoptosis via alternative splicing of Caspase-9 and Bcl-x143. Another pro-apoptotic target of ceramide is PKCξ. The direct binding of ceramide to PKCξ can lead to formation of a

24 Chapter 1

pro-apoptotic complex with prostate apoptosis response-4144 and/or coordinate gene expression through the activation of NF-κB145.

Aside from activation of apoptotic mediators, ceramide may also mediate apoptosis processes through modification of membrane structure. Ceramide in the plasma membrane segregates into ceramide-enriched domains, forming sizeable lipid rafts146- 148. Different cell receptors such as CD40 can oligomerize in these ceramide raft domains149. It has been reported that upon activation of the TNF superfamily receptors such as CD95 or CD40 on the surface of T and B cells, there was rapid ceramide formation in the span of just one minute150-152. Ceramide-enriched lipid rafts help facilitate the clustering of these receptors, resulting in the recruitment of downstream signalling proteins: TNF receptor-associated factor following activation of CD40, and Fas-associated death domain and caspase 8 following activation of CD95153,154. Ceramide can also shape membrane curvature and stability, forming non-lamellar structures. Studies have suggested that ceramide induces membrane fusion, vesicle budding and efflux through these non-lamellar structures155,156. Vectorial membrane budding initiated by the production of ceramide by SMase in the non-lamellar structures is physiologically relevant157. It is a phagocytic process dependent on the activity of cell surface acid SMase158. In general, this type of ceramide formation facilitates changes in the membrane physical state, inducing membrane fusion or fission. In the effector phase of apoptosis, neutral SMase hydrolyses large amounts of SM into ceramide. The loss of SM and cholesterol in the membrane leads to membrane blebbing and vesicle shedding159. Note that ceramide can also diffuse through lipid bilayers in giant unilamellar vesicles160. Interestingly, the diffusion rate differs between different ceramide species, as determined by the length of the fatty acyl tails.

Ceramide has also been implicated in mitophagy. In excess, mitophagy is pro-apoptotic. C18:0 ceramide has been shown to target autophagolysosomes to mitochondria, resulting in lethal mitophagy161. This was shown with both an exogenous C18:0 ceramide analogue, and endogenous C18:0 ceramide synthesized by CerS1. The process implicated is regulated by LC3B-containing autophagolysosomes interacting with ceramide on mitochondrial membranes. This interaction was regulated based on a dynamin related protein 1 (Drp-1) dependent mitochondrial fission. These data suggest

25 Chapter 1

that ceramide can serve as a receptor anchor for LC3B autophagolysosomes on the mitochondrial membranes.

Contrary to the well accepted pro-apoptotic role for ceramide, its recruitment of kinase suppressor of Ras to the plasma membrane is necessary to stimulate the classical mitogen activated protein kinase/extracellular regulated kinase (ERK1/2) pathway in response to Epidermal Growth Factor162. Treatment with ceramide analogues increased kinase suppressor of Ras autophosphorylation and its ability to transactivate Raf-1 and subsequent ERK1/2 activation163. Although ceramide is well known for its pro-apoptotic effects, the premise that all ceramides produce a pro- apoptotic response is debatable. For instance, C16:0 ceramide synthesized by CerS6 promotes survival in HNSCC164 while C18:0 ceramide synthesized by CerS1 can inhibit tumour cell growth114,117.

1.6.2. Ceramide, insulin signalling and glucose metabolism

Studies have suggested that ceramides may negatively regulate insulin signalling and glucose metabolism. Initial studies were done in 3T3-L1 adipocytes and C2C12 SkM myotubes165,166. Exogenous addition of short-chain ceramide was shown to reduce mRNA expression of GLUT4 and inhibit glucose uptake. Further studies demonstrated that elevated cellular ceramide levels affect glucose metabolism mainly through Akt inhibition50,167. This was not achieved through the inhibition of PI3K, upstream of Akt pathway166,168. Inhibition of Akt by ceramide is mainly driven by two distinct mechanisms. Firstly, ceramide promotes Akt dephosphorylation via PP2A activation. As stated above, ceramide can directly bind and activate PP2A. PP2A dephosphorylates Akt at Ser473 and Thr308, the two important phosphorylation sites for its full activation51,52. Secondly, ceramide can prevent Akt translocation, through the activation of PKCζ52,169. The variations of the two mechanisms are dependent on cell type and also the amount of caveolae in the cell, particularly because the PKCζ-mediated pathway demands functional caveolae, as PKCζ directly interacts with caveolin169. It has been demonstrated that inhibition of Akt by ceramide is diminished in adipocytes lacking caveolin-1169. Akt activation is crucial for GLUT4 translocation to the plasma membrane170, its inhibition can impact insulin-stimulated glucose metabolism, which may ultimately lead to hyperglycemia, a characteristic of insulin resistance. Therefore, ceramide is seen as a lipotoxic mediator impacting insulin signaling. 26 Chapter 1

Although it has been demonstrated that ceramide does not impact PI3K. A recent study reported that prolonged accumulation of ceramide in C2C12 and human myotubes can negatively impact PI3K activity171. Impaired PI3K activity was a result of protein kinase

R activation in a c-Jun NH2-terminal kinase-dependent manner. Protein kinase R activation leads to serine phosphorylation of IRS-1 and inhibition, which in turn prevents the recruitment and activation of PI3K. This mechanism is an important aspect as significant impaired IRS-1-associated PI3K activity was found in SkM of insulin- resistant humans172. Furthermore, it has been shown that short-term treatment (2 h) with C2:0 ceramide increased inhibitory phosphorylation of IRS-1 at Ser636/639 as a result of upstream activation of ribosomal S6 kinase173. This chain of reactions presumably leads to PI3K inhibition.

Ceramide may also impact glucose metabolism through pyruvate dehydrogenase, the rate-limiting enzyme of glucose oxidation174. It was reported that C6:0 ceramide treatment increased mRNA expression of pyruvate dehydrogenase kinase 4 in AC16 human cardiac myocytes175. Pyruvate dehydrogenase kinase 4 phosphorylates and inhibits pyruvate dehydrogenase. Furthermore, studies have shown that treatment of isolated mouse hearts with myriocin, a SPT inhibitor, reduced ceramide content and led to increased rates of glucose oxidation28,175.

1.6.3. Ceramide and mitochondrial functions

Ceramide synthesis can occur in the mitochondria and its accumulation in this organelle may impact mitochondrial functions. Studies to date suggest that elevated intracellular ceramide levels negatively impact mitochondrial function by interfering with mitochondrial respiration processes. C2:0 or C16:0 ceramide treatment on isolated rat heart mitochondria leads to cytochrome c release, reduced respiratory complex I activity and reduced pyruvate-stimulated respiration176,177. Another study reported that the inhibition of respiration in isolated rat heart mitochondria following ceramide treatment was a result of inhibition of respiratory complex III activity176,177. In addition, in CerS2 null mouse liver, C16:0 ceramide was found to inhibit respiratory complex IV activity and induced reactive oxygen species formation, resulting in chronic oxidative stress101. The same effect was not seen with C24:0 ceramide. In addition, treatment of H9c2 myoblasts with carboxymethyl lysine, an advanced glycation end product, increased intracellular ceramide levels and led to inhibition of respiratory complex II activity and 27 Chapter 1

impaired mitochondrial respiration. The negative effects were prevented via myriocin pre-treatment178.

Recent studies have also suggested that JNK induces expression of ceramide biosynthesis enzymes in the mitochondria179. Inhibition of JNK leads to reduced ceramide levels and improved mitochondrial oxidative phosphorylation function180. Mitochondrial fission is suggested to be another mechanism of ceramide-mediated mitochondrial dysfunction as ceramide treatment of C2C12 myotubes led to increased mitochondrial fission, which was associated with elevated expression of Drp-1181. Treatment with Drp-1inhibitor, Mdivi-1 on permeabilized mouse red gastrocnemius muscle prevented inhibition of mitochondrial respiration mediated by ceramide accumulation181. Of interest, lactosylceramide (LacCer), a glycosphingolipid derivative of ceramide has also been implicated as the culprit for mitochondrial defects in diabetic hearts182. Increased levels of LacCer were accompanied by reduced calcium retention capacity and respiration in the mitochondria of diabetic heart tissues. Furthermore, exogenous LacCer added to baseline mitochondria was able to reproduce the diabetic phenotype. The authors suggested that LacCer may be one of the early causative factors of mitochondrial dysfunction in Type 1 diabetes.

1.7. Ceramide and Obesity complications

1.7.1. Ceramide and obesity-induced inflammation

Chronic low-grade inflammation is a hallmark of obesity. Excess saturated FA from the diet upregulate cytokine synthesis and secretion through receptor signaling involved in innate immunity183,184. The inflamed environment, in concert with excess lipid influx, is likely to drive sphingolipid production. Lipidomic screening has revealed that inflammatory mediators such as TLR4 and TNF-α increased sphingolipid levels without affecting neutral glycerolipid levels, including the lipotoxic DG60. Correlational studies indicate close relationships between plasma ceramides and the severity of insulin resistance in unison with circulating cytokines, such as interleukin-6. Interestingly, administration of TNF-α into C57BL/6J mice upregulates ceramide synthesis in adipose tissue through SM hydrolysis and de novo synthesis pathways185. In relation to this finding, Holland et al. reported a correspondence between ceramide synthesis and inflammatory status converging on the TLR4 pathway without involvement of TNF-α

28 Chapter 1

signaling27. Another study demonstrated that increased ceramide content by lipopolysaccharides or FA overload was prevented in SkM and liver in genetic mice lacking functional TLR4, suggesting ceramide production in inflammatory responses is TLR4 dependent27.

Macrophage infiltration in adipose tissue is another phenomenon of obesity-induced inflammation39. A recent report has demonstrated that VLDL receptor is elevated in macrophages in obese adipose tissue, which promotes inflammation by increased C16:0 ceramide production and facilitation of M1-like macrophage polarization186. Consequently, increased ceramide stimulates the MAPK and NF-κB pathways promoting pro-inflammatory responses. Another mechanism of ceramide-mediated inflammation in diabetes and insulin resistance is through activation of the NLRP3 inflammasome187. It was demonstrated that NLRP3 mediates caspase 1 cleavage upon ceramide stimulation in adipose tissues and macrophages, and thus partly gives rise to insulin resistance. Of interest, C1P was recently associated with inflammatory signalling of immune cells. One study has shown that ceramide kinase deficiency in mice prevented diet-induced weight gain and associated glucose intolerance188. Ceramide kinase deficiency also decreases monocyte chemoattractant protein-1 signalling in adipose tissue macrophages, reducing adipocyte inflammation188.

1.7.2. Ceramide, adiponectin and FGF21

Adiponectin is an insulin-sensitizing and cardioprotective adipokine189. Transgenic overexpression of adiponectin in obese mice produces a remarkable metabolic phenotype whereby animals accumulate fat in adipose tissues but are protected from metabolic disorders including insulin resistance190. It was suggested that the protection from metabolic disorders is attributed to adiponectin’s ability to activate AMP-activated protein kinase (AMPK)191. Adiponectin was also found to reduce ceramide content and associated apoptotic responses by increasing ceramidase activity. Inhibition of ceramidase activity negated adiponectin-mediated anti-apoptotic properties189. In these reports, AMPK inhibition has no impact on the biological actions of increased adiponectin-induced ceramidase activity, suggesting that the two pathways act independently.

29 Chapter 1

Similar to adiponectin, fibroblast growth factor 21 (FGF21) induces beneficial metabolic responses ameliorating hyperglycemia192. FGF21 treatment was found to reduce hepatic ceramide content, increase glucose infusion rates, and reduce hepatic glucose output during hyperinsulinemic-euglycemic clamp studies in obese mice193. These effects of FGF21 were abolished in obese adiponectin-deficient mice. FGF21 was also found to induce expression of adiponectin while simultaneously reducing ceramide levels. These data suggest that FGF21-mediated ceramide reduction is induced by adiponectin expression.

1.7.2. Ceramide and vascular functions

CVD is one of the leading causes of death in insulin-resistant or diabetic people194. SM, GluCer and ceramide have all been implicated in atherogenic processes. Pharmacological manipulations of sphingolipid levels have been shown to improve atherosclerosis195. The exact sphingolipids involved in plaque formation is unclear but it is likely ceramide and other sphingolipids all contribute through different actions. Ceramide certainly contributes to dyslipidaemia196, retention of oxidised lipids197 and monocyte adhesion to the vascular walls198. Hypertension is often present in diet- induced ceramide accumulation and subsequent vascular wall remodelling. Ceramide can also inhibit phosphorylation of nitric oxide synthase and Akt through PP2A activation and reduce subsequent vasodilatory outcomes199,200. Another underlying mechanism could be ceramide-mediated reduction in membrane fluidity, a trait of hypertension in rats201.

Circulating plasma ceramide has been associated with ischemic heart disease202, atherosclerotic plaque formation203, stroke, myocardial infarction204 and hypertension. Plasma or serum ceramide concentrations have been used as a prediction tool for major cardiovascular events205,206. MayoClinic has also recently launched a blood test for assessing CVD risk using plasma ceramide. The test offered, CERAM measures plasma ceramide including the highly relevant C18:0, C16:0 and C24:1 ceramides.

1.7.3. Ceramide and brain lipotoxicity

Most obesity-related pathologies are diet induced. Our central nervous system (CNS) oversees regulation of appetites, food intake and energy metabolism. As described

30 Chapter 1

previously, lipotoxicity occurs through ER stress and inflammation. The same can occur in the CNS. It has been demonstrated that ER stress in the hypothalamus also induces insulin resistance and resistance, which leads to weight gain207,208. In particular, ceramide accumulation has been reported to play a role in the impaired regulation of energy balance in the CNS. One study has demonstrated that hypothalamic lipotoxicity, characterized by ER stress and reduced sympathetic tone to the BAT can be induced by exogenous ceramides209. The hypothalamic lipotoxicity pathologies eventually led to reduced thermogenesis and weight gain independent of feeding. Moreover, overexpression of the chaperone glucose-regulated protein 78 (GRP78) in the hypothalamic ventromedial nuclei abolished ceramide action on ER stress and improved overall BAT thermogenesis and glucose homeostasis210. Collectively, these studies reveal a signalling linkage between ceramides, hypothalamic lipotoxicity and BAT thermogenesis in the context of obesity. Of note, overexpression of GRP78 in obese Zucker rats, which ameliorated ER stress, did not impact ceramide content suggesting that ER stress is a downstream effect of ceramide accumulation210.

Recent discoveries indicate that malonyl-CoA can regulate hypothalamic control in energy balance211-213. The adipokine, leptin was shown to regulate feeding and increase malonyl-CoA concentrations in the hypothalamic arcuate nucleus through inhibition of AMPK and subsequent activation of acetyl-CoA carboxylase (ACC)214. AMPK is an energy sensing protein that controls ACC activity and FAS expression. The elevation in malonyl-CoA leads to an anorexigenic effect, as malonyl-CoA inhibits CPT-1c activity215,216. Related to this, a hypothalamic specific CPT-1c knockout mouse model exhibited reduced food intake and weight gain216. Interestingly, CPT-1c knockout mice had less ceramides in their arcuate nuclei of the hypothalamus and the opposite was observed when CPT-1c was overexpressed217. Additionally, infusion of C6:0 ceramides to the arcuate nuclei led to inhibition of leptin-induced anorexigenic effects. Treatment with myriocin to block de novo ceramide synthesis also resulted in reduced food intake and body weight in mice217. Collectively, these data demonstrated that ceramide in the arcuate nucleus may have a novel role in the hypothalamic control of feeding and energy homeostasis.

31 Chapter 1

1.7.4. Ceramide and liver health

In addition to insulin resistance, ceramide has been implicated in NAFLD and hepatic steatosis. NAFLD is characterized by TG accumulation in the liver218 and progresses into steatosis, characterised by liver inflammation and fibrosis. NAFLD is a major health problem as it predisposes individuals to CVD, cirrhosis and liver cancer219. These conditions are associated with impaired hepatic lipid oxidation218. Studies have shown that hepatic ceramides positively correlate with hepatic steatosis and/or insulin resistance in mice and humans49,220. Myriocin and the DEGS1 inhibitor fenretinide have been shown to alleviate hepatic steatosis in mouse models221-224. The mechanism behind this is unlikely the rescue of insulin-mediated Akt signalling via reduced ceramide content. One possible mechanism involved could be related to the inhibition of mitochondrial respiratory activity by ceramide. Mice treated with myriocin presented increased mitochondrial activity and enhanced oxygen consumption rates28,59. In particular, C16:0 ceramide synthesized by CerS6 was identified as the culprit for steatohepatitis. This was supported by the work of Turpin et al., who showed that depletion of CERS6 from the liver protected obese mice from insulin resistance and steatohepatitis58. Interestingly, CERS2 depletion resulted in increased CerS6 and C16:0 ceramide levels, reduced β-oxidation, and predisposed mice to HFD-induced steatohepatitis59. A subsequent study further supports the role of C16:0 ceramide in liver steatosis108. Mice lacking CERS5 have lessened C16:0 ceramide levels in the liver. These mice present with improved glucose metabolism, insulin sensitivity, and are protected from hepatic steatosis. These studies suggest the major steatosis effect appeared to result from impairment of lipid oxidation through the inhibition of mitochondrial respiratory complexes by C16:0 ceramide. A study on the effect of ceramide on heart mitochondrial functions supports this supposition176.

Based on the aforementioned system, a study further investigated the role of ceramides in the liver by inducing expression of acid ceramidase in transgenic C57BL/6J mice, which resulted in the same physiological outcome225. The protection from hepatic steatosis appeared to be a result of reduced hepatic lipid uptake. It was also demonstrated that ceramide can induce translocation of CD36 to the plasma membrane through activation of PKCζ225. Thus, reduced ceramide content results in reduced translocation of CD36 for FA uptake into hepatocytes. Likewise, inducible expression

32 Chapter 1

of adiponectin receptor (AdipoR) leads to reduced hepatic ceramide content and protection against hepatic steatosis and insulin resistance via upregulation of AdipoR- mediated ceramidase activation226.

Recently, the poorly studied metabolite acylceramide was reported to be stored in lipid droplets in steatotic liver of mice fed with a HFD rich in oleate227. The generation of acylceramide involves the formation of a complex consists of CerS, fatty acyl-CoA synthase 5 (ACSL5), and DGAT2 at the ER membrane and on lipid droplets. The inhibition of acylceramide formation from ceramide resulted in ceramide accumulation and ceramide-mediated apoptosis. This study proposed a new molecular pathway with biological significance in steatotic liver apoptosis by controlling ceramide content via acylceramide generation227.

1.7.5. Ceramide and pancreatic β-cell function

Ceramide has been implicated in pancreatic β-cell damage in T2D. Shimabukuro et al. demonstrated that Zucker diabetic fatty rats have increased SPT expression and activity in their pancreatic islets, associated with significantly higher ceramide content compared to the lean non-diabetic controls53. Treatment with L-cycloserine, a SPT inhibitor, reduced ceramide content and decreased apoptosis in the islets of these rats. Furthermore, in cultured human islet MIN6 cells, C14:0, C16:0 and C24:0 ceramides have been reported as the critical ceramide species that affect islet cell viability and insulin secretion under the influence of palmitate loading54. In another study, treatment of INS-1 pancreatic β-cells with adiponectin improved cell viability through ceramidase activity, which lowers ceramide levels189. However, another study also reported that SPT and DEGS1 inhibition in MIN6 cells could not protect against a palmitate-induced deficiency in glucose-stimulated insulin secretion228. In spite of all these studies, there is no direct evidence that localized manipulation of ceramide content in β-cells can protect against T2D independent of metabolic influences or effects in other locales.

1.7.6. Ceramide and insulin resistance

Accumulation of ceramide can induce insulin resistance in SkM, liver, and WAT229. The role of ceramide in insulin resistance first emerged when it was shown to inhibit insulin-stimulated glucose transport. As described previously in section 1.6.2., ceramide

33 Chapter 1

can inhibit insulin signaling through activation of PP2A and subsequent dephosphorylation of Akt, or through activation of PKCζ, which prevents translocation of Akt50-52,169. These impact GLUT4 translocation and insulin-stimulated glucose uptake. Pharmacological inhibition or genetic knockdown of SPT, CerS and DEGS1, in cultured cells treated with excess palmitate, was able to rescue insulin-mediated Akt signalling in the presence of ceramide221,230. Up-regulation of ceramidase in liver of transgenic C57BL/6J mice could also reduce hepatic ceramide content and achieve the same results225.

Based on these observations, multiple groups subsequently demonstrated that insulin resistance caused by leptin deficiency, lard infusion, HFD and/or dexamethasone can be prevented through pharmacological inhibition of SPT, DEGS1 and CerS in rats and mice28,199,222,231-233. Insulin resistance in mouse models can also be reduced via genetic ablation of SPT subunit 2, DESG1, or CerS6, accompanied by beneficial metabolic features such as reduced hepatic ceramide content, protection from hepatic steatosis and increased energy expenditure58,231,234.

Recently, studies have reported that insulin resistance caused by high fat feeding can be improved by decreasing C16:0 ceramide synthesis through ablation of CerS5 and CerS6 genes in the liver and adipose tissue58,59,108. In cultured hepatocytes, C16:0 ceramide can impede insulin signalling and reduce β-oxidation of FA59. CerS6-dependent C16:0 ceramide accumulation also promotes weight gain58. Interestingly, C24:0 ceramide synthesized by CerS2 is protective against insulin resistance59,103,235. Following the discovery that C16:0 ceramide, synthesized by CerS5/6, is the culprit for inducing insulin resistance in the liver or adipose tissue, we now have more insights into the physiological significance of individual ceramide species in different organs. In regard to SkM insulin resistance, C18:0 ceramide has been shown to associate with inhibition of insulin signalling in studies involving obese insulin resistant and insulin sensitive human subjects, insulin resistant mouse models, and exercise interventions in T2D subjects24,48,236. In SkM, CerS1 is the main CerS isoform and synthesizes C18:0 ceramide. Increased human SkM C18:0 ceramide has been reported in insulin-resistant postmenopausal women with obesity237.

Earlier studies have indicated that circulating ceramide may play a role in insulin resistance as well. Ceramide in circulating LDL particles can induce insulin resistance 34 Chapter 1

in vitro and in vivo238. This is an interesting observation as it indicates the possibility that ceramide packaged into LDLs either in the adipose tissue or liver could account for insulin resistance. Similarly, circulating C18:0 ceramide, although low in abundance, was positively correlated with visceral fat and inversely with metabolic flexibility48. In the same study, SkM C18:0 ceramide was positively correlated with visceral fat, systolic and diastolic blood pressure. Both the circulating and SkM C18:0 ceramides are inversely correlated with adiponectin48. Furthermore, in this study, genes encoding enzymes of the sphingolipid pathway were up-regulated in SkM of both obese and non- obese insulin resistant subjects. This suggests that de novo synthesis of ceramide may be associated with insulin resistance independent of obesity. As discussed earlier, lowering the ceramide concentration may have insulin-sensitizing effects. In this regard, exercise and caloric restriction have been shown to be effective in lowering C18:0 ceramide in SkM, resulting in improved insulin sensitivity239. However, caloric restriction alone was not adequate to reduce SkM C18:0 ceramide. In another study, patients that went through bariatric surgery and exercise intervention displayed reduced SkM C18:0 ceramide with enhanced insulin sensitivity240. Likewise, bariatric surgery alone could not reduce SkM C18:0 ceramide. Interestingly, acute exercise can also prevent FA-induced insulin resistance in non-obese women, associated with a significant reduction in SkM ceramide content241. Collectively, these studies highlight the importance of exercise intervention as a treatment in insulin resistance regardless of obesity.

The glycosphingolipid, GM3 was also shown to inhibit insulin activation of IRS-1 in adipocytes, when added exogenously 242. In agreement with this, GM3 synthase knockout mouse models were protected from HFD-induced insulin resistance243. Another glycosphingolipid, GluCer, is considered a lipotoxic agent and has been implicated in insulin resistance as well. Pharmacological inhibition of GCS could improve glucose tolerance and insulin sensitivity in liver and muscle of ob/ob mice, Zucker Diabetic Fatty rats and diet-induced obese mice244-247. Patients with Gaucher disease are characterized by accumulation of GluCer, and are predisposed to develop insulin resistance248. An inhibitor was developed for treating Gaucher disease targeting GCS244. This inhibitor, used in rodent models, has shown beneficial effects such as improved insulin sensitivity, resolution of hepatic steatosis and reduced inflammation in

35 Chapter 1

adipose tissues244-246,249. Both ceramide and GluCer are established independent antagonists of insulin signalling in vitro249.

Collectively, all the above mentioned in vitro and in vivo studies have demonstrated the crucial role of ceramide in insulin resistance and how individual ceramide species differ in function in different tissues. A schematic illustration of ceramide action in metabolic tissues is shown in Figure 1.9.

Figure 1. 9. Schematic illustration of ceramide action in metabolic tissues.

De novo synthesis of ceramide is driven by intake of free FA either diffusing through the plasma membrane or transported by CD36 or other FA transporters (not shown in diagram)64,65. Ceramide synthesis is also driven by TLR4-mediated upregulation of SPT expression and/or TNF-α-induced hydrolysis of SM by SMases. The ceramide produced mediates ER stress, inhibits lipid oxidation at the mitochondria, and inhibit insulin-stimulated Akt activation. FGF21 regulates ceramide levels via increased adiponectin expression. Adiponectin reduces ceramide content by inducing ceramidase activity to degrade ceramide into sphingosine. Abbreviation: AdipoQ: adiponectin precursor gene; AdipoR, adiponectin receptor; INSR: insulin receptor; Sptlc: SPT genes. Image adapted from229.

36 Chapter 1

1.8. Pharmacological inhibition of ceramide synthesis

The conclusion from all the metabolic consequences of ceramide accumulation discussed above is that ceramide is now appreciated as an important modulator of cellular metabolism. Manipulation of ceramide metabolism and synthesis presented an overall potent effect on improving multiple metabolic complications in both animal models and human subjects. For the purpose of this project, the focus will be on pharmacological inhibition of ceramide synthesis.

1.8.1. SPT inhibitor: Myriocin

Myriocin, also known as thermozymocidin or ISP-1, was first discovered from the thermophilic fungi, Mycelia sterilia and Myriococcum albomyces250,251. Myriocin was found to be a potent antifungal agent. In the 1990s, myriocin was re-isolated and found to be an immunosuppressant252-254. The molecular target was identified as SPT, the rate limiting first step in de novo sphingolipid synthesis. SPT inhibition with myriocin depletes cells of sphingolipids including ceramide and protect against insulin resistance, dyslipidaemia and weight gain both in vitro and in vivo28,222,231. However, due to the immunosuppressive properties of myriocin252, it might not be suitable for all subjects with obesity-related diseases. Further to this, not all ceramide species should be considered the same. Recent studies have suggested that very long chain ceramides can be insulin sensitizing, hence beneficial for affected subjects59,103,235.

1.8.2. DEGS1 inhibitor: Fenretinide

Fenretinide (N-(4-hydroxyphenyl) retinamide) is an orally-available synthetic phenylretinamide analogue of retinol (vitamin A). It has been suggested as a potential anti-cancer drug. When used as a cancer treatment, fenretinide can slow tumour growth and promote cell death by triggering stress responses and accumulation of ceramides 255,256. Interestingly, fenretinide usage in rodent models and humans can have insulin- sensitizing effects257-259. The precise mechanism of fenretinide was not known until recent indication of structural similarity between fenretinide and dhCer260,261. This structural similarity allows fenretinide to bind to DEGS1 and exerts competitive irreversible inhibition. In vitro, fenretinide can inhibit ceramide formation and protect against lipid-induced insulin signalling impairment in isolated SkM strips and cultured

37 Chapter 1

myotubes. In vivo, fenretinide treatment can improve glucose homeostasis in diet- induced obese mice221. Interestingly, fenretinide treatment also significantly reduced TG in liver, indicating a possible reversal of hepatic steatosis. In conclusion, modulation of ceramide synthesis by DEGS1 inhibition with fenretinide can be explored as a therapeutic means for associated metabolic diseases.

1.8.3. CerS inhibitor: Fumonisin B1

Fumonisin B1 (FB1) is a mycotoxin and a member of a family of toxins, fumonisins. It is produced by the fungus Fusarium verticillioides, which commonly occurs in maize and other agricultural products262. FB1 is structurally similar to ceramide and is characterized as a CerS inhibitor in a competitive manner towards both dihydrosphingosine and fatty acyl CoA at CerS active sites263,264. FB1 is a potent inhibitor in vitro inhibiting all CerS isoforms. The potential clinical use of FB1 in metabolic disorders is limited as it is neurotoxic, hepatotoxic and nephrotoxic in animals265-267. It has also been identified as a potential carcinogen. Although FB1 can be a potent CerS inhibitor and can modulate ceramide synthesis for the purpose of improving glucose homeostasis54, for the above toxicity reasons, FB1 is not a suitable treatment option for all subjects. Just like myriocin, inhibiting all ceramide synthesis might not be ideal since very long chain ceramides have beneficial effects for liver insulin signalling.

1.8.4. Fingolimod, FTY720

Fingolimod, also known as FTY720 is the first oral drug approved for the treatment of multiple sclerosis268. FTY720 is a structural analogue of sphingosine. The FTY720 pro- drug is phosphorylated by SPHKs, forming FTY720-phosphate, which is an agonist at S1P receptors269-271. FTY720 is derived from myriocin and retains its immunosuppressive properties272. The primary mechanism of action of FTY720 as an immunosuppressant is to sequester lymphocytes in the secondary lymphoid organs (lymph nodes), thereby inducing lymphopenia. This is mediated through the action of FTY720-phosphate as a super-agonist of S1P1 receptors on lymphocytes273,274. Since then, FTY720 has been found to have other molecular targets and functions. FTY720 has been reported to inhibit CerS activity in an uncompetitive manner towards dihydrosphingosine and non-competitive manner towards fatty acyl CoAs92,275. Most

38 Chapter 1

recently, FTY720 treatment has been shown to reduce SkM ceramide content and improve glucose tolerance in HFD fed mice276. FTY720 treatment also prevented TG accumulation in SkM induced by HFD in mice276. In vitro, FTY720 increases TG lipolysis277. Despite all that, FTY720 is immunosuppressive and should be used with caution. Its inhibition of CerS isoforms is not selective, inhibiting multiple isoforms at the same time, including the beneficial CerS2 isoform.

1.8.5. CerS isoform-specific inhibitor

Recent findings established that C16:0 ceramide, synthesized by CerS6 is responsible for promoting insulin resistance in the liver58,59. On the contrary, CerS2 synthesizes very long chain ceramides that are protective against insulin resistance59,235. Therefore, the development of specific inhibitors of individual CerS isoforms is of research and clinical importance. One possible approach is to modify existing analogues that have CerS inhibitory properties such as FB1 or FTY720. In this regard, Schiffmann et al. have attempted to create inhibitors for specific CerS isoforms by modifying the amine group of FTY720 and its ether analogue (O-FTY720)278. There is still a need for a highly specific CerS inhibitor with nanomolar potency. Recently, our group established that AAL(S), a non-phosphorylatable analogue of FTY720, and its benzyl tail derivative G024, exhibited limited inhibitory specificity for CerS1279. However, the selectivity for CerS1 is weak and the inhibitors failed to reduce C18:0 ceramide content in cultured cells. Further modifications to the benzyl tail of G024 were made, giving rise to the potent and selective inhibitor P053, which is the subject of this thesis.

39 Chapter 1

1.9. Aims of thesis

In summary, the literature presented has identified the global health burden of metabolic syndrome, particularly obesity-related diseases. In this regard, ceramide accumulation certainly plays an important role in the various pathologies described. Recent discoveries illustrate the importance of identifying and characterising the role of individual CerS isoforms and ceramide species in metabolic disease. Following previous studies on the inhibitory properties of AAL(S) and G024 towards CerS1, our group made modifications to the benzyl tail of G024 and created a potent and selective CerS1 inhibitor, P053. With the advancement of techniques relating to mass spectrometry (MS), I aimed to characterize this inhibitor in vitro and in vivo, and its effects on metabolic physiology. The specific aims of this thesis were: a) To develop MS methods for assaying CerS activity in vitro and in cultured cells. b) To characterize the inhibitory properties of P053 on CerS isoforms, particularly CerS1, in vitro and in cell culture. c) To characterize the efficacy of P053 at lowering ceramide content in vivo, and its effect on the lipidome of SkM, liver, brain, and adipose tissue. d) To determine whether P053 restores muscle insulin sensitivity in vivo in mice fed a HFD. e) To investigate the molecular basis behind the effect of C18:0 ceramide reduction on insulin sensitivity and overall lipid levels in the body.

40 Chapter 2

Chapter 2: Methods

This section will outline general methods applied throughout the dissertation.

2.1. Compounds

FTY720 was purchased from Cayman Chemical Company, USA. AAL(S) and G024 were synthesized as described previously279. Compound P053 was synthesized by Dr. Hamish Toop and Elysha Taylor from Prof. Jonathan Morris’s lab. P053 is one of the non-phosphorylatable derivative of FTY720. For biochemistry and cell treatments, the compounds were prepared as stock solutions at 10 mM in dimethyl sulfoxide (DMSO), then diluted in water or cell culture medium.

2.2. Cell Culture Methods

2.2.1. Cell Culture conditions

Cell culture reagents were purchased from Life Technologies, Australia. Tissue culture flasks and plates were purchased from Nunc, Thermo Fisher Scientific, Australia. All cell culture work was performed aseptically in a Class II Biosafety Cabinet. Routine mycoplasma testing was performed. All testing came back negative in our cultures.

The human embryonic kidney cell line, HEK293, and human glioblastoma cell line, U- 251 (both from American Type Culture Collection), were cultured in Dulbecco’s Modified Eagle Medium (DMEM), supplemented with 10 % fetal calf serum and 2 mM

L-glutamine, in a humidified atmosphere with 5 % CO2 at 37ºC. Cells were passaged as they near 80 % confluency. Cells were washed with PBS, and then detached with 1x trypsin-EDTA. Culture medium was used to neutralise the trypsin-EDTA. Dissociated cells were collected into conical falcon tubes and centrifuged at 1,200 r.p.m. for 3 min and resuspended to desired dilution or cell density in fresh culture medium for plating or subculturing. Cell counting was performed to determine seeding density or total cell numbers for MS data normalization. Viable cell counts were obtained using trypan blue stain (Life Technologies, Australia) and counting using a hemocytometer.

41 Chapter 2

Mouse myoblast cell line, C2C12 was cultured in DMEM: Ham’s F12 (1:1 v/v), supplemented with 10 % horse serum and 2 mM L-glutamine. The cultures were kept at low density and not allowed to reach confluence. Medium was changed every two days. To induce differentiation, cells were seeded at 90-100 % confluency and culture medium was changed with differentiation medium (DMEM: F12, 1:1 (v/v)) containing 2 % horse serum. The cells differentiated rapidly, forming contractile myotubes over the span of 3-7 days. During this time, cells were replenished with fresh differentiation medium every 24 h. Myotube formation was monitored microscopically daily to determine when the cells would be ready for treatment and processing. Primary muscle satellite cells or myoblasts were cultured on matrigel coated plates in DMEM: F12 medium (1:1 v/v) supplemented with 20 % horse serum, 1 % Penicillin/Streptomycin. The cultures were maintained daily with addition of 5 ng/mL of mouse growth factor, bFGF and media changed every other day. Satellite cells were subcultured at 50-60 % confluency and re-plated at 5000 cells/ cm2. To induce differentiation into myotubes, culture media was changed with differentiation medium (DMEM: F12, 1:1 (v/v)) containing 2 % horse serum and 1 % Penicillin/Streptomycin. The cells became fully differentiated into myotubes after 48-72 h.

Mouse cortical neurons were prepared from E16 pups280 and cultured in Neurobasal medium supplemented with 2 % (v/v) B27 supplement and 0.25 % (v/v) GlutaMax. The neurons were seeded at 5×105 cells/well in poly-D-lysine coated 6-well plates. After two weeks, the cells were ready for treatment and subsequent processing. The neurons were kindly prepared and provided by members of A/Prof Thomas Fath’s lab at UNSW.

Primary hepatocytes were prepared from livers of C57BL/6J strain mice (8-14 weeks old) and kindly provided by A/Prof Nigel Turner’s lab at UNSW. Primary hepatocytes were cultured on plates coated with 0.5 µg/cm2 rat tail Collagen Type 1 (BD Biosciences) in M199 adherence media with 5.5 mM glucose containing the following supplements: 100 U/mL penicillin/streptomycin, 10% BSA, Ultroser G (Pall Corp, USA), 100 nM dexamethasone (Sigma Aldrich, Australia), 100 nM insulin (Actrapid, Novo-Nordisk, Denmark). After 4 h, media was changed with fresh basal culture media (M199 media supplemented with 100 U/mL Penicillin/ Streptomycin, and 100 nM dexamethasone). For FA treatments, basal culture media was supplemented with 2 % BSA, 300 µM palmitic acid, 300 µM oleic acid and 150 µM linoleic acid (total 750 µM

42 Chapter 2

combined FA). This is to simulate the effects of high physiological concentrations of long-chain FA as previously recommended281. Vehicle control media was prepared by adding ethanol to basal culture media with 2 % BSA. Both FA and vehicle control media were allowed to conjugate for 2 h at 55 °C. Media was filtered, sterilised and allowed to cool to 37 °C before use. The cells were incubated with FA or vehicle control media for 16 hours before further analysis.

2.2.2. Flow cytometry

Flow cytometry was used to assess toxicity of compounds treatment in HEK293 cells. HEK293 cells were seeded onto 6-well plates at 2 ×105 cells per well and treated with P053 for 72 h. At the end of treatment, cells were harvested as described previously with trypsin-EDTA and washed with PBS. Cells were then resuspended in 300 µL of 2 µg/ml propidium iodide (Invitrogen, Australia) in PBS for 10 min on ice. A positive control (cells treated with vehicle control) and unstained control were also used. Cell counting and propidium iodide stained cells were determined on a BD FACSCanto II flow cytometer (BD Biosciences, USA). A total of 10,000 cells were counted over a 30- 50 s collection period, and at a constant medium to high flow rate. Propidium iodide staining analysis was performed on the FlowJo software (TreeStar Inc., USA). Propidium iodide positive represented dead cells and negative represented live cells.

2.2.3. Transfection

HEK293 or U-251 were cultured in 75 cm2 flasks until 60 % confluency. Prior to plasmid transfection, the culture medium was changed to 12 mL of Opti-MEM and combined with 500 µL of Opti-MEM that had been pre-incubated with 20 µL of Fugene 6 transfection reagent (Promega, Australia) and 10 µg of plasmid, according to the manufacturer’s instructions. Plasmids used in this thesis include human CerS1 and CerS6 Open Reading Frames (ORFs) with N-terminal Flag tags in pCMVexSVneo vectors; human CerS2 and CerS5 ORFs in pCMV6-XL6 vectors; and human CerS4 ORF in pCMV6-XL5 vector. CerS2, CerS5 and CerS4 plasmids were purchased from Origene, USA. CerS1 and CerS6 plasmids were kindly given by Prof. Besim Ogretmen from Medical University of South Carolina. After 16 h of incubation in Opti-MEM transfection mixture, the medium was changed to normal culture medium for a further 24 h incubation. The transfected cells were harvested with trysin-EDTA, neutralised, 43 Chapter 2

and centrifuged at 1,200 r.p.m. for 3 min to collect the cell pellet. The cell pellet was washed with PBS and centrifuged again. HEK293 cells were transfected to overexpress CerS1, CerS2, CerS5 and CerS6 while U-251 cells were transfected to overexpress CerS4. For each transfection, there was a relevant flask of cells that went through the same procedure with only the absence of plasmid to act as a control.

Murine CerS1, 2, and 5 were overexpressed in monkey kidney fibroblast, COS-7 cells prepared by A/Prof Carsten Schmitz-Peiffer. The CerS and control (LacZ) adenoviruses were manufactured and verified as described previously282. Briefly, the concentrated viral stocks for either control or CerSs were prepared in HEK293 cells and purified with caesium chloride density gradient centrifugation. A 150 cm2 flask of COS-7 cells was transfected with either murine CerS1, 2, and 5 adenoviruses or LacZ control adenovirus. A day after transfection, the cells were washed extensively with PBS for 15 times to rid of active free viruses. The transfected cells were then harvested as described earlier.

Transfection efficiency was verified via CerS activity assay of protein lysates prepared from the transfected cells, compared to the control cells.

2.3. Ceramide synthase activity assays

CerS activity assays were carried out using protein extracts of HEK293 cells exogenously expressing human CerS1, 2, and 6; U251 cells expressing human CerS4; COS-7 cells expressing murine CerS1, 2, and 5; or protein extracts prepared from mouse tissue. Briefly, cells were washed once with PBS and scraped directly into lysis buffer. Alternatively, cells can be detached with trypsin/EDTA solution, washed with PBS, and then pelleted by centrifugation at 200 × g for 5 min prior to lysis. The cells were lysed via sonication in ice water bath using a Bioruptor (Diagenode, Belgium), set to High intensity, with a 30 s on/ 30 s off cycle. The homogenate is centrifuged for 10 min at 800 × g to pellet debris. The supernatant is transferred to new tube and the protein concentration is measured using a BCA assay kit (Pierce, Thermo Scientific, USA). The lysates are stored in aliquots at -80 °C until required.

CerS activity was assessed using a high-performance liquid chromatography (HPLC) based assay with a fluorescent NBD-dihydrosphingone substrate as described283. Briefly, the reaction volume of each sample was 25 µL comprising 20 mM HEPES, 25

44 Chapter 2

mM KCl, 2 mM MgCl2, 0.02 % BSA, 0.5 mM DTT, 10 µM NBD-dihydrosphingosine (Avanti Polar Lipids, USA), and 25 µM of fatty acyl CoA substrate. C18:0-CoA (Sigma Aldrich) was used as the fatty acyl substrate for CerS1 and CerS4; C24:1-CoA (Avanti Lipids) for CerS2; and C16:0-CoA (Sigma Aldrich) for CerS5 and CerS6. Reactions were started with the addition of 2.5-10 µg protein lysate and run at 37 ºC for 30 min, then stopped with the addition of four volumes methanol. Reactions were stored at - 20°C for at least 2 h then centrifuged at 14,000 r.p.m. for 10 min at 4 °C to remove any precipitate formed. Supernatant (~100 µL) was transferred to glass HPLC vials.

Reaction product peaks were quantified using a Finnigan Surveyor HPLC system (Thermo Scientific, USA) connected to a Shimadzu RF-10AXL fluorescence detector (Shimadzu, Japan). Each sample (20 µL) was injected and resolved through a 3´150mm C8-XDB HPLC column (Agilent Technologies, USA). Data was acquired over a 12- min chromatography run using a two-solvent system at a flow rate of 0.5 mL/min: solvent A, 0.2 % formic acid, 2 mM ammonium formate in water; solvent B, 0.2 % formic acid, 1 mM ammonium formate in methanol. The 12-min gradient starts at 20:80 A/B, increasing to 5.95 A/B over 2 min, then to 100 % B from 2 to 8 min. The gradient is then held at 100 % B for 2 min, before re-equilibration to 20:80 A/B for another 2 min. Peak areas were integrated using Xcalibur software and converted to pmoles of product using a standard curve ranging from 0.2 to 20 pmoles.

2.4. Lipid extraction

2.4.1. Tissue samples

All tissue samples used for lipid extraction were prepared in the same manner. At the conclusion of animal studies, mice were euthanized and the tissues were rapidly dissected and freeze-clamped prior to storage in cryovials in the -80ºC freezer. Prior to lipid extraction, tissues were powderized using a Cell Crusher device (Cellcrusher, Cork, Ireland). To prevent thawing of the samples, the Cell Crusher device was maintained at below freezing temperatures using liquid nitrogen. The Cell Crusher was decontaminated thoroughly between samples. It is essential to powderize the tissues prior to processing to ensure each sample used can represent the organ as a whole. For lipid extraction, 20-30 mg of powderized tissues were weighed into pre-labelled 2-mL

45 Chapter 2

tough tubes containing CK14 (1.4-mm ceramic) beads. The exact tissue weight was recorded for the purpose of data normalization later on. The tissues were then homogenized in 0.2 mL of methanol containing 0.01% butylated hydroxytoluene (BHT), using a Precellys 24 homogenizer and Cryolys cooling unit (Bertin Technologies) at 6,500 r.p.m. for two 20 s cycles at 4ºC. The homogenates were transferred to 15 mL glass tubes and spiked with a methanol mixture containing the following internal standards: 5 nmole 19:0/19:0 phosphatidylcholine and 2 nmole each of 17:0/17:0 phosphatidylethanolamine, 17:0/17:0 phosphatidylserine, 17:0/17:0 d5- DG, 17:0/17:0 phosphatidylglycerol, 14:0/14:0/14:0/14:0 cardiolipin, 18:1/12:0 SM, 18:1/17:0 ceramide, 18:1/12:0 GluCer, 18:1/12:0 LacCer, 18:1/12:0 sulfatide (all from Avanti Polar Lipids, USA), and 17:0/17:0/17:0 TG (Cayman Chemical, USA). To ensure there was no tissue debris left in the 2-mL tough tubes, 0.3 mL 0.01 % BHT- methanol was used to wash the beads and tubes interior. The methanol was pooled together with the homogenate in the 15-mL glass tubes (Corning, Australia) to a final volume of 500 µL. Methyl-tert-butyl ether (MTBE; 1.7 mL) was added to the glass tubes and the samples were vortexed and sonicated in a sonicating ice water bath (Thermoline Scientific, Australia) for 30 min. Phase separation was induced by addition of 417 µL of MS-grade water followed by vortexing and centrifugation at 1,000 g for 10 min (Megafuge 1.0R, Heraeus, Thermo Scientific, USA) . The upper organic phase was collected into 5-mL glass tubes (Thermo Scientific, Australia). The lower phase was re- extracted as mentioned above with solvent mixture, whose composition was equivalent to MTBE/methanol/water (10:3:2.5, v/v/v). The upper phase was pooled together in the 5-mL glass tube and dried under vacuum in a Savant SC210 SpeedVac (Thermo Scientific, Australia). The dried extract was reconstituted in 500 µL methanol, or 400 µL 80 % methanol/ 0.2 % formic acid if measuring P053 uptake. After vortexing well and centrifuging at 4,000 r.p.m. at 4 ºC for 5 min, 300 µL supernatant was transferred into a glass HPLC vial (Thermo Scientific, Australia) and stored at -20 °C until analysis.

2.4.2. Cultured cells

Cells were cultured as described in section 2.2. Extra wells were seeded for counting the total number of cells in the well for data normalisation later. At the end of treatments, culture plates were placed on ice and each well was gently washed with ice cold PBS

46 Chapter 2

twice. PBS was aspirated completely and 300 µL of methanol was added to each well for cell scraping. An additional 300 µL of methanol was used to wash the wells and pooled together into 15-mL glass tubes to a final volume of 600 µL. The samples were then spiked with 50 µL of internal standard mixture: 250 pmoles of C17:0 ceramide, C12:0 SM and C12:0 GluCer in methanol. MTBE (2.17 mL) was added to each sample followed by thorough vortexing and sonication in sonicating ice water bath for 30 min. Phase separation was induced via addition of 542 µL of MS-grade water and centrifugation for 10 min at 1,000 g. The upper phase was transferred to 5-mL glass tubes. The lower phase was re-extracted with solvent mixture to a final composition of MTBE/methanol/water (10:3:2.5, v/v/v). Upper phase was pooled together and dried under vacuum in a Savant SC210 SpeedVac. The dried extract was reconstituted in 400 µL of methanol or mobile phase solvent dependent on subsequent MS analysis method. After vortexing well and centrifuging at 4,000 r.p.m. at 4 ºC for 5 min, 300 µL supernatant was transferred into a glass HPLC vial and stored at -20 °C until analysis.

2.4.3. Plasma samples

For measuring P053 uptake, plasma samples (20 µL) stored at -80 ºC were thawed on ice. To each sample in 1.5 mL Eppendorf tubes, 200 µL of methanol, containing 20 pmoles AAL(S) as the internal standard, was added. The mixtures were vortexed, then sonicated in ice water using a Bioruptor (Diagenode, Belgium) for 2 min at 30 s intervals. Extracts were centrifuged at 1,400 r.p.m. for 20 min at 4 ºC to clear debris. Supernatants were transferred to 5 mL glass tubes. The insoluble material was re- extracted with 1 mL 80 % methanol as above, and the supernatant combined with the first extract. The extracts were dried under vacuum in a Savant SC210 SpeedVac (Thermo Scientific, USA). The dried extract was reconstituted in 200 µL of 80 % methanol/ 0.2 % formic acid and stored at -20 °C until analysis. On the day of analysis, the extracts were vortexed thoroughly and centrifuged at 300 g for 20 min before transferring 100 µL of the supernatant to glass HPLC vials.

47 Chapter 2

2.5. Liquid Chromatography Mass Spectrometry

2.5.1. Targeted LC-MS/MS

Liquid Chromatography tandem MS (LC-MS/MS) was performed in the Bioanalytical Mass Spectrometry Facility (BMSF) at UNSW. The LC-MS/MS method developed for lipid profiling is discussed in detail in chapter 3.

Briefly, targeted quantification of Cer, HexCer and SM lipid species in cultured cells or tissue lipid extracts was performed by LC-MS/MS on a TSQ Quantum Access triple quadrupole mass spectrometer (ThermoFisher Scientific)284 with an Accela UPLC system. The TSQ Quantum Access was coupled to an electrospray ionisation source and run in positive ion mode. Each extract (20 µL) was injected and separated on a Zorbax XDB-C8 HPLC column (5 µm pore size, 150 mm x 3.0 mm, Agilent Technologies, USA) using a 10-min isocratic elution at a flow rate of 0.5 mL/min. Mobile phase consisted of 1 mM ammonium formate in methanol with 0.2 % formic acid.

For targeted analysis of P053, extracts (20 µL) were separated using the same C8 column using a 6.5 min binary HPLC gradient: 0 min, 80 % B (20 % A); 2 min, 90 % B; 5 min, 90 % B; 6.5 min 80 % B. Mobile phase A: 2 mM ammonium formate/ 0.2 % formic acid in water; mobile phase B: 1 mM ammonium formate/ 0.2 % formic acid in methanol. The flow rate was 500 µL/min and column oven temperature 30 °C. An external standard curve was constructed using P053 at 0.1, 1, 100 and 1000 nM, with peak areas normalised to the AAL(S) internal standard for quantification of P053 in plasma samples for pharmacokinetic analysis or peak areas only for uptake of P053 into various tissue and plasma samples from mice treated with P053 for 2 weeks. Standard curves were run at the beginning and end of the sequence run. The extraction efficiency for P053 using the above protocol was verified using control plasma extracts (20 µL each), spiked with or without 5 pmoles of P053 compound and 20 pmoles of AAL(S) as internal standard and subjected to the same extraction procedure. The positive and negative control extracts were run using the above LC-MS/MS method under the parameters stated in Table 2.1. Examples of the chromatogram for P053 and AAL(S) can be seen on Figure 2.1. Analysis for P053-phosphate, C18:0 S1P, C18:0 sphingosine and C18:0 dihydrosphingosine were performed on TSQ Altis Triple Quadrupole mass spectrometer (ThermoFisher Scientific) by adapting the same chromatography. The

48 Chapter 2

parameters for P053-phosphate, C18:0 S1P, C18:0 sphingosine and C18:0 dihydrosphingosine are stated in Table 2.1.

Table 2. 1. Parameter settings for P053 detection using LC-MS/MS. Internal standards used are highlighted in red. Abbreviations: Sph, sphingosine; dhSph, dihydrosphingosine; S1P, sphingosine 1-phosphate.

Metabolite Precursor m/z Product m/z Collision Energy (eV) AAL(S) 294.0 161.1 16 P053 354.0 159.0 31 P053-P 434.3 159.0 31 C17:0 Sph 286.1 268.0 17 C18:0 Sph 300.2 264.1 20 C18:0 dhSph 302.5 284.1 18 C17:0 S1P 366.1 250.1 23 C18:0 S1P 380.1 264.1 23

Figure 2. 1. Chromatograms for AAL(S) and P053.

M/z 294.0 (AAL(S) internal standard) and 354.0 (P053) events from the same 100 nM external standard at the beginning (A) and end (B) of MS sequence run performed on TSQ Access. Abbreviations: RT, retention time; AA, peak area detected using Thermo Xcalibur software. 49 Chapter 2

2.5.2. Untargeted lipidomic profiling

Untargeted lipidomic profiling was performed on a QExactive Plus mass spectrometer with heated electrospray ionization probe and a Dionex UltiMate 2000 LC pump (ThermoFisher Scientific). Lipid extracts from tissues were diluted 1:10 with methanol. Diluted extracts (10 µL) were then resolved on a 2.1×100 mm Waters Acquity C18 UPLC column (1.7 µm pore size) with a VanGuard pre-column (Waters corporation, USA) using a 30-min gradient in which mobile phase A consisted of acetonitrile: water (60: 40), 10 mM ammonium formate, 0.1 % formic acid; and mobile phase B was isopropanol:acetonitrile (90: 10), 10 mM ammonium formate, 0.1 % formic acid285 (Table 2.2). The flow rate was 0.26 mL/min and column oven was set at 55 ºC. The temperature of the autosampler tray was set to 10 ºC. Data was acquired in data dependent acquisition (DDA) mode according to the manufacturer’s software. The source tuning parameters for both positive and negative ion mode are listed in Table 2.3. For quality control, a standard mix was prepared and included in each run to be sure of the proper chromatographic elution time. Additional lipids (18:0/18:0 PE, 16:0/18:1 PC, 16:0/16:0 DG, 16:0/16:0/18:0 TG; all from Avanti Polar Lipids and Cayman Chemical, USA) were added to the usual sphingolipid external and internal standard mixture (Table 3.2). Adequate blanks were run between samples and standards to ensure no contamination or carry over of samples during sequence run.

Table 2. 2. HPLC gradient for untargeted lipidomic profiling.

Retention (min) %B 0.0 32.0 1.5 32.0 4.0 45.0 5.0 52.0 8.0 58.0 11.0 66.0 14.0 70.0

18.0 75.0 21.0 97.0 25.0 97.0 30.0 32.0

50 Chapter 2

Table 2. 3. Source running parameters for QExactive Plus mass spectrometer.

Spray voltage (+/-) 4000 V Capillary temp. (+/-) 250 Sheath gas (+/-) 20 Aux gas (+/-) 5 S-Lens RF level 60 DDA acquisition parameters MS1 (+/-) 30 min acquisition time 1 mscan, AGC target value 3e6 Max ion injection time 50 ms Resolution 70,000 Scan range (Profile) 400 – 1200 m/z (+) 500 - 1650 m/z (-) MS2 (+/-) 30 min acquisition time 1 mscan, AGC target value 2e5 (+) 1e5 (-) Max ion injection time 110 ms (+) 45 ms (-) Resolution 17,500 Spectrum data type Profile Isolation width 1.4 m/z Collision energy 30 (+) 15, 27 (-) DDA Loop count Top 10 ions from MS1 scan

An exclusion list of background ions was used based on a solvent blank for each run. The method developed was adapted from 286 with the help from Dr. Russ Pickford, senior staff from the BMSF at UNSW. Examples of the chromatograms for a quadriceps lipid extract from a control C57BL/6J mouse can be seen in Figure 2.2. All MS data were acquired and visualised using Xcalibur software.

51 Chapter 2

Figure 2. 2. LC-MS Chromatograms of lipids from mouse quadriceps muscles. Figure shows chromatograms of the same C57BL/6J mouse quadriceps lipid extracts in positive mode (A) and negative mode (B). Chromatograms are visualised in Xcalibur software.

2.5.3. Data processing and calculations

For data collected from the TSQ Quantum Access, visualisation and quantification of peaks was performed using Thermo Fisher’s XCalibur software v2.2 SP1.

For data collected from the QExactive Plus, LipidSearch v4.1.30 software (ThermoFisher Scientific) was used for chromatogram alignment, peak identification and integration according to the available settings within the software. Firstly, LipidSearch software identifies lipids based on its own internal library of masses and fragment ions covering 18 classes of lipids and 66 subclasses, which can be selected by the user. The software then aligns the chromatogram or lipid peaks of samples to accurately quantify MS1 peak profiles. Table 2.4 and Table 2.5 show the parameters used for lipid identification, quantification and alignment in this thesis via LipidSearch. After alignment, each lipid peak and its identification were manually verified in the LipidSearch software interface, using a diagnostic product ion and any additional ions that confer acyl chain composition. Rejected IDs are marked and can be excluded in final exported Excel data spreadsheet. Although LipidSearch software offers a

52 Chapter 2

normalisation tool, the data generated in this thesis were exported to an Excel spreadsheet for normalisation of each lipid to its class-specific internal standard. Lipid concentrations were calculated relative to the amount of internal standard (nmoles or pmoles) and normalised to tissue weight (mg).

Table 2. 4. LipidSearch lipid identification and quantitation parameters. Abbreviation: RT, retention time; ppm, parts per million; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PG, phosphatidylglycerol; PI, phosphatidylinositol; PA, phosphatidic acid; SM, sphingomyelin; DG, diacylglycerol; TG, triacylglycerol; CL, cardiolipin; Cer, ceramide; CerG, glycosphingolipid; Che, cholesterol ester.

Database General and Q Exactive Search type Product Experiment type LC-MS Precursor tolerance 5.0 ppm Product tolerance 8.0 ppm Intensity threshold 1.0% Execute Quantitation On M/z tolerance (quantitation) -/+ 5.0 ppm RT range (min) (quantitation) -/+ 0.5 Top rank filter On Main node filter Main isomer peak m-Score threshold 5.0 c-Score threshold 2.0 FA priority On ID Quality filter Check A, B, C and D Target class PC, PE, PI, PS, PG, PA, SM, Cer, CerG, TG, DG, D5DG, CL, Che

Ion adducts (+/-) +H, + NH4, and +2H (+) -H, -2H, +HCOO (-)

53 Chapter 2

Table 2. 5. LipidSearch software alignment parameters. Abbreviation: RT, retention time.

Search type Product Experiment type LC-MS Alignment method Max (or Mean) RT tolerance 0.15 min Calculate unassigned peak area On Filter type New filter Toprank filter On Main node filter Main isomer peak

m-Score threshold 5.0

ID quality filter A, B, and C only

2.6. Animal and Diet composition

Mice used in this thesis were obtained from the Animal Resource Centre in Perth (WA, Australia). All animals were housed at 22 ± 1ºC with a controlled 12:12 h light-dark cycle. The mice had ad libitum access to water and either chow (8 % calories from fat, 21 % calories from protein, 71 % calories from carbohydrate; Gordon’s Specialty Stock Feeds, Yanderra, NSW, Australia) or HFD made in house287 (45 % calories from fat (lard), 20 % calories from protein, 35 % calories from carbohydrates, 4.7 kcal/g; based on Rodent Diet no. D12451 Research Diets Inc., New Brunswick, NJ, USA). For the cohorts in this thesis, male C57BL/6J mice at 10-12 weeks old were used and randomly allocated to remain on the low-fat chow diet or to receive a HFD. The cohorts lasted from 4-6 weeks. The experiments were approved by the UNSW animal care and ethics committee (ACEC 15/48B), and followed guidelines issued by the National Health and Medical Research Council of Australia.

Animal cohorts consisted of 40-48 male C57BL/6J mice randomly allocated into groups to receive either chow or HFD over a course of 4-6 weeks. Within each diet group, mice received either P053 or vehicle control administration. At the same time the mice were given HFD, P053 (5 mg/kg in drinking water) or vehicle control (2 % DMSO in

54 Chapter 2

drinking water) administration commenced. The method of administration was oral gavage.

2.6.1. Assessment of body weight and body composition

Body weight measurements were taken weekly or twice-weekly and used for calculation of drug dose for oral gavage. Fat and lean body mass were measured in mice using an EchoMRI-900 Body Composition Analyser (EchoMRI Corporation Pte Ltd, Singapore) in accordance to the manufacturer’s instructions and normalized to body weight.

2.6.2. Tissue measurements

Organ tissue crude weights were measured at cull. All tissues were then freeze-clamped and stored in cryovials at -80ºC until further analysis.

2.6.3. Gene expression analysis using qPCR

Total RNA was purified from mouse tissues with TRIzol reagent (Sigma-Aldrich), followed by DNase treatment (Promega), and reverse transcribed using random hexamer primers with FirstStrand cDNA synthesis kit (Roche), according to the manufacturer’s instructions. PCR (40 cycles) was performed with an MX3000P thermocycler (Agilent) and PowerUp SYBR Green Master Mix (Applied Biosystems).

55 Chapter 2

Table 2. 6. Primer sequences used in qPCR amplification reactions.

Gene Forward primer (5’ to 3’) Reverse primer (5’ to 3’)

18s ribosomal RNA (18s GTAACCCGTTGAACCCC CCATCCAATCGGTAGTA rRNA) ATT GCG Ribosomal protein L13 AGGAGGCGAAACAAGTC GGAGACTGGCAAAAGCC (RPL13) CAC TTAAAG Ribosomal protein, large, P0 GGCTCCAAGCAGATGCA CCTGATAGCCTTGCGCAT (36B4) GCAG CATGG Ceramide Synthase 1 CCACCACACACATCTTTC GGAGCAGGTAAGCGCAG (CerS1) GG TAG ATGCTCCAGACCTTGTAT CTGAGGCTTTGGCATAG (CerS2) GACT ACAC CATTTATTATCGCGGCCC TGTTGGGTTGTTTGATCC Cytochrome B (CytB) TA TG NADH dehydrogenase 1b TTTTCTCACGCGGAGCTT ATAAAGAAGGCTTGACG subcomplex 5 (NDUFB5) TC ACAAACA Cytochrome C, somatic GCAAGCATAAGCCTGGA TTGTTGGCATCTGTGTAA (CYCS) CCAAA GAGAATC Cytochrome C oxidase CTAGCCGCAGGCATTACT TGCCCAAAGAATCAGAA subunit I (CoxI) AT CAG Cytochrome C oxidase GCCGACTAAATCAAGCA CAATGGGCATAAAGCTA subunit II (CoxII) ACA TGG Cytochrome C oxidase GCTGCATCTGTGAAGAG CAGCTTGTAATGGGTTCC subunit 5B (Cox5b) GACAAC ACAGT ATP synthase subunit O TTCTCCTTAGATGCAGCA AGGCCCTTTGCCAAGCTT (ATP5o) GAGTACA Mitochondrial-encoded GACGAACATGAACCCTA TACGGCTCCAGCTCATAG ATP synthase 6 (ATP6) AT T

For CerS1 and CerS2, the relative amount of each gene product was determined using a standard curve of Ct against input cDNA. CerS1 or CerS2 gene expression was normalised to 18s RNA. For analysis of respiratory complex subunits, relative gene expression was calculated according to delta-delta CT method288 and normalised to the average of the housekeeping genes RPL13 and 36B4. 56 Chapter 2

2.6.4. Assessment of metabolic parameters

2.6.4.1. GTT

Glucose tolerance tests (GTT) were carried out following a 6 h fast. The mice received a dose of glucose (2 g/kg) by oral gavage (oGTT). Blood glucose levels were monitored with a hand held glucometer (Accu-Check, Roche, Australia) from the tail-tip for 90 min following glucose administration. Blood insulin levels prior to and during GTT were measured using an Ultra-Sensitive Insulin ELISA kit (Crystal Chem, Downers Grove, IL, USA) according to the manufacturer’s protocol.

2.6.4.3. Blood analysis

Blood collection was done either from tail-tip for sampling on live mice, or by cardiac puncture at the time of sacrifice. EDTA or heparin coated collection tubes were used. For plasma collection, the blood samples were centrifuged at 1,000 g for 10 min at 4 ºC. The top layer of plasma was carefully transferred into a separate tube and stored at -80 ºC until further analysis.

2.6.4.4. NEFA assay

NEFA content was assessed by an enzymatic colorimetric technique (NEFA-C, Wako Pure Chemical Industries, Japan). The NEFA assay was performed immediately upon the first thaw cycle of the plasma samples. In the presence of added CoA synthase enzyme, CoA is acylated by FA in the sample. The acyl-CoA formed is then oxidized by acyl-CoA oxidase, which produces hydrogen peroxide in the presence of peroxidase, resulting in the condensation of 3-methy-N-ethyl-N (β-hydroxyethyl)-aniline with 4- aminoantipyrine. The resultant pigment can then be measured at 550 nm on a microplate reader (SpectraMax, Molecular Devices, USA) and calculated against a free FA standard curve (0-10 mM).

57 Chapter 2

2.6.4.5. Blood cell count

Blood samples were taken at endpoint and analysed using an ActDiff hemato analyser- 424 (Beckman Coulter) using equine settings. Red blood cell, platelet, haemoglobin, haematocrit and white blood cell counts were recorded.

2.7. Analysis of insulin sensitivity by hyperinsulinemic-euglycemic clamp

All hyperinsulinemic-euglycemic clamp procedures were conducted with the help of Dr. Amanda Brandon according to published methods289,290.

2.7.1. Dual cannulation surgery

All surgical procedures were conducted on heating pads using sterile technique. Dual cannulation was performed on mice under general anaesthesia (4% isoflurane for induction; 1-2% during procedure) and under aseptic conditions289,290. Once anaesthetised, the surgical site was shaved and a small incision made with a scalpel. Catheters were inserted into the left carotid artery and right jugular vein. Free catheter ends were tunnelled subcutaneously, externalised at the neck and sealed. Mice received the analgesic/ anti-inflammatory ketoprofen (5 mg/kg) at the end of surgery. Mice were then singly housed and monitored daily. Catheters were flushed every 1-2 days with heparinised saline to maintain patency.

2.7.2. Hyperinsulinemic-euglycemic Clamp

Hyperinsulinemic-euglycemic clamps were performed 6-8 days after dual cannulation surgery289,290, and at 6 weeks after commencing a HFD or chow control diet, with or without P053 (5 mg/kg/day).On the day of experimentation, conscious mice were fasted for approximately 5 h, and fitted with extension catheters to enable infusion and sampling. Mice received a [3-3H]-glucose (0.05 µCi/min; Perkin Elmer, Australia) infusion for a period of 90 min. This allowed the calculation of whole body glucose uptake. After this basal period, the hyperinsulinemic clamp was initiated with a primed- continuous infusion of insulin (Actrapid, Novo Nordisk, Denmark) at 24 mU/kg bolus followed by 6 mU/kg/min. All animals were given the same dose of insulin based on 27 g mouse body weight and the rate of [3-3H]-glucose infusion was increased to 0.1 µCi/min. Euglycemia (~8 mM) was maintained during the clamp procedure by 58 Chapter 2

measuring blood glucose levels every 10 min. When necessary, 25 % glucose was infused to maintain euglycemia. Once euglycaemia was established, a bolus of 10 µCi of 2[14C]-deoxyglucose (Perkin Elmer, Australia) was administered for determination of tissue specific glucose uptake. Blood samples were taken at 2, 5, 10, 15, 20 and 30 min after the tracer bolus entered the bloodstream. At the conclusion of the tracer period, mice were euthanised and organs removed, snap frozen in liquid nitrogen and stored at - 80 °C for further analysis.

During the clamp procedure, blood and plasma glucose levels were determined using a hand held glucometer (Accu-Check, Roche, Australia). Plasma samples collected were used to determine insulin levels using an Ultra Sensitive Insulin ELISA kit (Crystal Chem, Downers Grove, IL, USA) and for measuring NEFAs as described in section 2.6.4.4.

2.7.3. Hyperinsulinemic-euglycemic Clamp analytical methods

2.7.3.1. Tracer dose

The dose of tracer injected into each animal was accurately determined from the before and after weight of the syringe. Total tracer activity was determined by counting 100 µL of a 1:1000 dilution of tracer in scintillation fluid (Ultima Gold XR, Packard Biosciences, The Netherlands) using a liquid scintillation counter (Tri-Carb liquid scintillation counter, Perkin Elmer, MA, USA).

2.7.3.2. Rate of tracer disappearance from plasma (Rd)

During the tracer period, at each time points, 10 µL plasma was taken and deproteinised by mixing with 20 µL 5 % ZnSO4 and 20 µL saturated Ba(OH)2. To determine the Rd the deproteinised sample was centrifuged (8000 g, 6 min, at room temperature) and the amount of 3H and 14C activity present in 15 µL of the supernatant was determined by addition of scintillation fluid and counted in a liquid scintillation counter (Tri-Carb liquid scintillation counter, Perkin Elmer, MA, USA). The rate of basal and clamp glucose disappearance (Rd) was determined using steady-state equations.

59 Chapter 2

3 14 To calculate the Rd (in mg/kg/min) disappearance curves for plasma H and C, the measured reduction in plasma radioactivity was fitted to a double exponential equation, which was integrated to determine area under the curve to the final time point (30 min) and to estimate the area to infinity: Cp. dose R = ( ) ∫ Cp ∗ t dt

In this equation, Cp is the plasma glucose concentration (mM), dose is the disintegration ( ) per minute (dpm) of tracer administered, ∫ Cp ∗ t dt is the area under the tracer disappearance curve to infinity.

2.7.3.3. Rate of glucose uptake (Rg’) in muscle tissues

Glucose uptake into tissues was calculated by measuring the relative levels of 2-deoxy- D-glucose (2DG)-6-phosphate in tissues. To determine glucose uptake, approximately 30 mg of powdered tissue was homogenised in 1 mL of distilled water and centrifuged at 13,000 r.p.m for 10 min at 4 °C. An aliquot of 400 µL of the supernatant was counted directly in scintillation fluid using the scintillation counter to determine 2[14C]DG and 2[14C]DG-6-phosphate (2[14C]DGP) radioactivity. To determine the non- phosphorylated glucose, another 400 µL of supernatant was treated with 200 µL of 0.3

N ZnSO4 and 200 µL of 0.3 N Ba(OH)2 (Sigma Aldrich), vortexed and then centrifuged at 13,000 r.p.m for 10 min at 4 °C. This removes 2[14C]DGP and any tracer incorporated into glycogen prior to scintillation counting to determine 2[14C]DG 14 radioactivity. The difference between the two aliquots is the 2[ C]DGPtissue. Glucose uptake into tissues was determined via the following equation: 2[ C]DGP R = × [arterial glucose] AUC2[ C]DG

14 14 In this equation, 2[ C]DGPtissue is the 2[ C]DGP radioactivity in the tissue (in dpm/g), 14 14 AUC 2[ C]DGplasma is the area under the plasma 2[ C]DG disappearance curve (in dpm/min/mL), and [arterial glucose] is the average blood glucose (in mmol/L)291.

60 Chapter 2

2.8. Substrate utilization assays

2.8.1. Ex vivo skeletal muscle palmitate oxidation

To examine whole muscle palmitate oxidation, soleus (SOL) and extensor digitorum longus (EDL) muscles were dissected quickly and carefully tendon to tendon. Extra precautions were taken to preserve the tendon attachments at the distal ends of the muscle strips to ensure that muscle fibres were intact. All procedures following dissection were carried out in pre-warmed 20 mL scintillation vials in a shaking water bath at 30 ºC. All buffers and atmosphere within all vials were pre-gassed and maintained at 20 % CO2/ 5 % O2. As soon as the muscle was dissected, it was placed in a petri dish of modified Krebs-Henseleit buffer (118 mM NaCl, 4.7 mM KCl, 2.5 mM

CaCl2, 1.2 mM KH2PO4, 1.2 mM MgSO4.7H2O, 15 mM NaHCO3) containing 4 % FA- free BSA, 5 mM glucose, and 0.5 mM palmitate, giving a palmitate-to-BSA molar ratio of 1:1. The muscle strips were secured on plastic support board using 29 G needles (one at each tendon) and then transferred to vials containing modified Krebs buffer. After a 30-min incubation, the pinned muscle strips were transferred to vials containing 1.5 mL modified Krebs buffer + 0.5 µCi/ mL 14C-palmitate for another 60 min at 30 ºC in a shaking water bath. An open Eppendorf tube containing 100 µL of 1 M KOH was included in the vial to capture CO2. At the end of the incubation period, muscle strips were removed, dabbed dry to remove excess medium, and snap-frozen to be stored in the -80 ºC freezer for further processing. Immediately after removal of muscle strips, 400 µL of 1 M perchloric acid was added to the incubation medium and the vial was 14 capped tightly. CO2 released was captured in the Eppendorf containing 1 M KOH over two hours and measured later by adding 100 µL of the KOH solution into scintillation fluid and counted using a liquid scintillation counter (Tri-Carb liquid scintillation counter, Perkin Elmer, MA, USA).

To determine 14C-radiolabeled palmitate incorporation into intramuscular TG, muscles strips were weighed and then homogenized using a Polytron (Kinematica, Littau- Lucerne, Switzerland) in 1.5 mL 2:1 chloroform-methanol (v/v). The homogenates were then sonicated in ice water for 15 min, after which 0.3 mL of distilled water was added, samples were vortexed, and centrifuged at 350 g for 10 min to induce phase separation. The upper aqueous phase was quantified by liquid scintillation counting to determine the amount of 14C-labeled oxidative intermediates. This represents a two-fold correction

61 Chapter 2

factor for exogenous 14C palmitate oxidation292. The lower organic phase was collected and dried under stream of nitrogen gas, then reconstituted in hexane and spotted on a Silica Gel 60 F254 thin layer chromatography (TLC) plate. The TLC plate was developed in hexane/diethyl ether/glacial acetic acid (85:15:1), then air dried. 14C TG bands were visualised using iodine vapour staining and determined according to the external TG standard band. The individual TG bands were marked on the plate with a scalpel and scraped into vials for liquid scintillation counting (Tri-Carb liquid scintillation counter, Perkin Elmer, MA, USA).

2.8.2. Fresh muscle homogenate palmitate oxidation

Tibialis muscles were dissected quickly, weighed and placed in vials containing ice- cold homogenisation buffer (250 mM Sucrose, 10 mM Tris-HCl, 1 mM EDTA, pH 7.4). A 5 % homogenate (1:19, w/v) was prepared using a Polytron device (Kinematica AG, Switzerland). A volume of 50 µL of the homogenate was added to a 7-mL scintillation vial containing an Eppendorf tube with 100 µL 1 M NAOH, followed by addition of 450 µL of pre-warmed oxidation medium (111 mM Sucrose, 11.1 mM Tris-HCl, 5.56 mM KH2PO4, 1.11 mM MgCl2, 88.9 mM KCl, 0.22 mM EDTA, 1.11 mM DTT, 2.22 mM ATP, 0.33 % FA-free BSA, 2.22 mM L-carnitine, 0.056 mM CoA, 0.11 mM malate, 0.2 mM palmitate, 0.5 µCi/mL [14C]-palmitate, pH 7.4) to initiate the reaction. The vials were sealed rapidly and incubated in a shaking water bath for 90 min at 30 ºC. Reactions were stopped by adding 100 µL of 1 M perchloric acid into the vials as fast as possible to prevent loss of CO2 released. The vials were left for 2 h to allow trapping of 14 14 14 CO2 in the NaOH. C-labelled CO2, as well as C-labelled acid soluble metabolites present in the acidified supernatant, were quantified by scintillation counting (Tri-Carb liquid scintillation counter, Perkin Elmer, MA, USA).

2.8.3. Palmitate oxidation in primary hepatocytes

To assess palmitate oxidation in hepatocytes cultures, cells were seeded in 6-well culture dishes at 4×105 cells/well and cultured overnight with or without treatment with FA or drug. After 16 h, spent media was removed and cells were washed twice with PBS and incubated for 1 h in 1 mL of palmitate oxidation media (Basal M199 media with 2 % BSA, conjugated to 200 µM palmitate and supplemented with 0.5 µCi 14C- palmitate) with or without drug. After 1 h incubation, media was removed and placed in

62 Chapter 2

a scint vial with 400 µL 1 M perchloric acid and an Eppendorf tube containing 100 µL

1M NaOH for CO2 trapping at room temperature for 2 h. The NaOH was then removed from the tube and used for liquid scintillation counting. Acidified media was centrifuged at 13,000 r.p.m to remove debris, and supernatant containing acid soluble metabolites were collected and used for liquid scintillation counting as well. Cells were washed with PBS twice and collected in 300 µL 1 M KOH using a cell scraper. The cell lysates were assessed for protein concentration using the BCA assay kit (Pierce, Thermo Scientific, USA).

2.9. Enzyme activity assays

For enzyme activity assays, liver and tibialis muscle were homogenized 1:19 (w/v) in 50 mM Tris-HCl, 1 mM EDTA, 0.1 % Triton X-100, pH 7.4. using a Polytron device. The homogenates were subjected to three freeze-thaw cycles and centrifuged for 10 min at 7,000 g at 4 ºC. Protein concentration was determined using the BCA assay kit (Pierce, Thermo Scientific, USA) and the homogenates were used for further assays or stored in a -80 ºC freezer.

2.9.1. Citrate synthase (CS)

An aliquot of 10 µL of a 1:6 dilution of the protein homogenate prepared as described above was pipetted into a 96-well microplate, with 240 µL of citrate synthase reaction buffer (100 mM Tris-HCl, 1 mM MgCl2, 1 mM EDTA, 0.1 mM 5,5’-dithio-bis (2- nitrobenzoic acid) (DTNB), 300 µM acetyl-CoA, pH 8.2) added to each well. The reaction was initiated with 500 µM oxaloacetate (prepared in reaction buffer), and followed at 412 nm (SpectraMax, Molecular Devices, USA) for 2- 3 min to measure the rate of production of 2-nitro-5-thiobenzoate from DTNB, ε = 13.6 mm−1cm−1.

2.9.2. β-hydroxyacyl-CoA-dehydrogenase (β-HAD)

An aliquot of 10 µL of protein homogenate prepared was pipetted into a 96-well microplate, with 240 µL of β-HAD reaction buffer (50 mM imidazole, 1.2 mM EDTA, 0.18 mM NADH, pH 7.4) and pre-heated to 30 ºC. The reaction was initiated with 50 µL of 0.6 mM acetoacetyl-CoA (prepared in 50 mM imidazole). The rate of NADH

63 Chapter 2

oxidation, ε = 6.22 mm−1cm−1 was measured by following the reaction at 340 nm (SpectraMax, Molecular Devices, USA) for 2 – 3 min.

2.10. Analysis of mitochondrial function in situ in permeabilized muscle fibres

Muscle fibres were prepared using EDL muscle as described previously293. Briefly, EDL muscle strips were removed from mice quickly and placed on a petri dish containing ice-cold isolation medium (10 mM Ca-EGTA buffer, 0.1 µM free concentration of calcium, 20 mM imidazole, 20 mM taurine, 49 mM K-MES, 3 mM

K2HPO4, 9.5 mM MgCl2, 5.7 mM ATP, 15 mM phosphocreatine, 1 µM leupeptin, pH 7.1). Under a microscope, extra-sharp forceps were used to dissect the muscle strips along fibres to form muscle fibre bundles, taking extra care not to mechanically damage the fibres. The fibre bundles were transferred to a vial containing 2 mL of isolation medium and 50 µg/mL saponin and mixed gently at 4 °C for 20 min. Permeabilized muscle fibres were then transferred into a tube containing the mitochondrial medium

(0.5 mM EGTA, 3 mM MgCl2.6H2O, 20 mM taurine, 10 mM KH2PO4, 20 mM HEPES, 1 mg/mL FA-free BSA, 60 mM potassium-lactobionate, 0.3 mM DTT, 110 mM mannitol, pH 7.1), mixed gently at 4 °C for 5 min to wash out saponin and ATP. This washing step was repeated 2-3 times to completely remove saponin and other metabolites. The wet weights of fibre samples were measured prior to respirometric measurements. Oxygen consumption was measured polarographically using a Clark- type electrode (Rank Brothers, Cambridge, UK) at 37 °C and data recorded using a PowerLab (ADinstruments, NZ). A substrate/inhibitor titration was performed for analysis of respiratory complexes I, II and IV as follows: (1) Measurement of complex I respiration using complex I substrates, 10 mM glutamate and 5 mM malate, activated by 2 mM ADP; (2) Inhibition of complex I with 0.5 mM rotenone, a specific inhibitor of complex I. This was followed by measurement of complex II respiration by addition of complex II substrate, 10 mM succinate; (3) Inhibition of complex III with 5 mM antimycin A, a specific inhibitor of complex III. This was followed by measurement of complex IV respiration by addition of artificial substrate, 0.5 mM TMPD (N,N,N¢,N¢- tetramethyl-p-phenylenediamine dihydrochloride) and 2 mM ascorbate. All measurements were calculated in nmol oxygen/ min/ mg muscle fibres.

64 Chapter 2

2.11. Western Blotting

Tissue lysates for western blotting were prepared by homogenizing tissues in RIPA buffer (100 mM NaCl, 10 mM Tris, pH 7.4, 1 % Triton X-100, 0.5 % sodium deoxycholate, 0.1 % SDS, 1 mM EDTA, 10 % Glycerol) containing complete protease inhibitor cocktail (Roche) and phosphatase inhibitors (3 mM b-glycerophosphate, 1 mM sodium orthovanadate, 5 mM sodium fluoride) by bead-beating in a Precellys 24 at 4 °C as described in detailed previously. Extracts were centrifuged at 6500 r.p.m. for 10 min at 4 °C to clear insoluble debris. Supernatant was collected and protein concentrations determined with the BCA assay kit.

For western blotting, protein lysates prepared were diluted in water, 4 × LDS sample buffer and 10 × reducing agent (Life Technologies, USA) to a final concentration of 1-2 mg/mL protein, 1x sample buffer and 1× reducing agent. All samples were heated at 70 ºC for 10 min except when blotting for mitochondrial complexes, for which samples were heated to only 37 ºC for 5-10 min. Samples (15 µg) were resolved on Bolt 4-12 % Bis-Tris Plus polyacrylamide gels (Life Technologies, USA). Novex Sharp or SeeBlue Plus2 pre-stained protein standards (Life Technologies, USA) were used as molecular weight markers. A common loading control was included on every gel to account for variation in relative band intensities between different blots. Proteins on the gel were then transferred onto a PVDF membrane (Merck Millipore, USA) in a Bolt transfer module at 20 V for 1 h. The membrane was blocked with 5 % skim milk solution for 1 h at room temperature, washed 3 times at 5 min intervals with Tris-buffered saline solution containing 0.05 % Tween 20 (TBST), probed with primary antibody overnight at 4 ºC with rocking (Table 2.7), and washed 3 times with TBST to remove excess primary antibody before incubation with the appropriate secondary antibody (1:5000, prepared in 5 % skim milk solution) for 1 h at room temperature. Antibody-antigen binding was detected using ECL reagent (GE Healthcare, Australia) and imaged on a Fujifilm Las-4000 CCD camera. Bands were quantified by densitometry with Fuji ImageQuantTL software and normalized to housekeeping protein bands and loading control on each gel. Only when necessary, membranes were stripped with mild stripping buffer solution composed of 1.5 % w/v glycine, 0.1 % w/v SDS and 1 % v/v Tween 20. Membranes were incubated with secondary antibody and re-imaged to check for

65 Chapter 2

stripping efficiency, then blocked and probed with a new antibody. Primary antibodies with very strong binding affinity were used last.

Table 2. 7. List of antibodies used for western blotting. Primary Catalogue Antibody Source Secondary antibody antibody number dilution Anti-mouse IgG, HRP-linked Complex I: subunit antibody, #7076S; 1:1000 ab110242 NADH Abcam, UK Cell Signaling dehydrogenase Technology, USA Anti-mouse IgG, Complex II: HRP-linked succinate Abcam, UK antibody, #7076S; 1:1000 ab14714 dehydrogenase Cell Signaling subunit B Technology, USA Anti-mouse IgG, Complex III: subunit HRP-linked Core 2 Ubiquinol- Abcam, UK antibody, #7076S; 1:1000 ab14745 cytochrome c Cell Signaling reductase Technology, USA Anti-mouse IgG, Complex IV: Cell Signaling HRP-linked Cytochrome c Technology, antibody, #7076S; 1:1000 11967S oxidase USA Cell Signaling Technology, USA Anti-mouse IgG, HRP-linked Complex V: ATP Abcam, UK antibody, #7076S; 1:1000 ab14748 synthase α subunit Cell Signaling Technology, USA Anti-rabbit IgG, Santa Cruz HRP-linked pan 14-3-3 (K19) Biotechnology, antibody, #7074S; 1:1000 sc-629 USA Cell Signaling Technology, USA

66 Chapter 2

2.12. Statistics

In general, all data are either reported as means ± SEM or SD. Statistical significance was accepted at P < 0.05. All statistics were analysed using GraphPad Prism software (Prism 7).

In chapter 4, statistical significance was measured by one-way ANOVA followed by a Dunnett’s post hoc test or by multiple t-tests with Holme-Sidak post-test. comparing each treatment with vehicle control.

For lipidomic data presented in Chapter 5, three comparisons were made for each lipid measured: chow control vs chow + P053, chow control vs HFD control, and HFD control vs HFD + P053. Data were log transformed, then subjected to two-tailed t-tests adjusted for multiple comparisons using GraphPad PRISM (Benjamini, Krieger and Yekutieli correction, Q = 1%).

For individual measures in chapter 5 and 6, two-way ANOVA was performed with diet as one variable and drug treatment as the other. P values for the main effect of diet (i.e. chow vs HFD) and drug (i.e. vehicle vs P053) are reported. Fisher’s Least Significant Difference test was used where post-tests were applied. Data was also analysed with unpaired t-tests where appropriate. Spearman analysis was used to assess correlation between ceramide species and TG levels in SkM; SkM ceramides and adipose tissue mass or % body fat.

67 Chapter 3

Chapter 3: Assaying Ceramide Synthase Activity in vitro and in Living Cells using Liquid Chromatography-Tandem Mass Spectrometry

3.1. Introduction

Over recent decades, there have been major technological advances in MS. MS is a popular technique for proteomic studies including protein identification, structure determination and quantification. Similarly, MS can be used to quantify small-molecule metabolites such as lipids in biological samples such as body fluids, tissues and cell cultures. One of the most widely used techniques is electrospray ionization LC-MS/MS. Traditionally, lipids can be separated using TLC. Although it offers a rapid screening tool, modern LC is a better way of separating lipids than TLC and when coupled with MS, the sensitivity in detecting and quantifying lipid species is improved. In an LC- MS/MS system, eluants from the HPLC are fed to an electrospray in which a high voltage is applied to create an aerosol to produce ions. The ions produced will pass through into the tandem mass spectrometer and undergo separation according to their m/z in the first stage (MS1). Selected precursor ions with specific m/z are then fragmented to produce unique product ions that can be detected in the second stage (MS2). Therefore, LC-MS/MS is a sensitive and accurate technique for the quantification of lipid species.

For the purpose of this thesis, a simple LC-MS/MS method has been optimized for assaying CerS activity in cells or tissue lysates, or in cultured cells (in situ). Compared to traditional radioactivity- or fluorescence-based assays assessed by TLC, LC-MS/MS method allows accurate detection and quantification of ceramide and dihydroceramide products. Similar LC-MS/MS-based CerS assays have been previously described using C17:0 sphingoid base dihydrosphingosine as substrates294,295. The alternative method

68 Chapter 3

described herein utilizes deuterated dihydrosphingosine as substrates. The LC-MS/MS method has also been published as a book chapter296.

3.2. Aims

The overall aim of the work presented in this chapter was to create and optimize a sensitive working method for measuring CerS activity using LC-MS/MS.

3.3. Methods

3.3.1. Assaying CerS activity in vitro using LC-MS/MS

To assay CerS activity in vitro, crude extracts containing cell membranes were firstly prepared from tissues or cultured cells. The preparation method of protein extracts was as described in section 2.3. Protein extracts served as the enzyme source for CerS reactions containing deuterated substrate (D7-dihydrosphingosine or sphingosine) and fatty acyl-CoA substrates. The product formed is dependent on the fatty acyl-CoA substrates used. The method for CerS assay is as described in section 2.3 with the only differences being the usage of deuterated dihydrosphingosine instead of NBD fluorescent dihydrosphingosine and addition of 50 pmoles C17:0 ceramide internal standard into the four volumes of methanol used to stop the reaction.

To quantify ceramide products formed from CerS assay, selected reaction monitoring mode is employed on the mass spectrometer after positive mode electrospray ionization. The settings described in this chapter are for a Thermo Fisher Scientific Quantum Access triple quadrupole MS coupled to a Thermo Fisher Scientific Accela UPLC. Two columns (Agilent 3 ´ 150mm XDB-C8 column, 5 µM pore size and Agilent Poroshell 120, 2.1 ´ 150mm SB-C18 column, 2.7 µm pore size) are tested with this MS method. Precursor and product ion m/z are indicated in Table 3.1 for both columns. To measure D7-labelled ceramide products, m/z for both precursor and product ions are increased by 7 mass units compared to naturally occurring ceramides. The column oven is kept at 30 ºC and the flow rate is 0.5 mL/ min using the HPLC mobile phase (methanol containing 0.2 % formic acid (v/v) and 1 mM ammonium formate). Each sample (20 µL) is resolved on the column for a total of 9 min for the C8 column and 15 min for the C18 column. Ceramide products elute at different times from the HPLC column (Table 3.1).

69 Chapter 3

The same injection volume (20 µL) was used for external standards and methanol blanks. Adequate blanks were run between samples to ensure no contamination or carry over of samples during sequence run. For each MS analysis sequence run, samples were analysed in a random order. External standards were run at the beginning and end of the sequence to ensure satisfactory performance of the instruments. The external standard curve consisted of the following concentrations in a serial dilution: 8000 nM, 2000 nM, 500 nM, 125 nM, 32 nM, 8 nM, and 2 nM. Each of the external standard dilution contains the same amount of internal standards as the samples.

70 Chapter 3

Table 3. 1. Exact mass, column elution times and precursor and product ion masses [M+H] for dihydroceramide and ceramide species. Ceramidea Molecular Precursor Product Collision C8 C18 Weight m/z m/z Energy column column [M+H]c [M+H]c (eV) elution elution time time (min) (min) Dihydroceramides d18:0/16:0 539.53 540.5 522.5 17 3.1 3.3 (547.5) (529.5) d18:0/18:0 567.55 568.6 550.6 17 3.6 4.3 (575.6) (557.6) d18:0/24:0 651.65 652.7 634.7 17 5.8 10.9 (659.7) (641.7) d18:0/24:1 649.64 650.6 632.6 17 5.0 8.0 (657.6) (639.6) Ceramides d18:1/17:0b 551.53 552.5 264.1 30 3.2 3.4 d18:1/16:0 537.51 538.5 264.1 30 3.0 3.0 (545.5) (271.1) d18:1/18:0 565.54 566.5 264.1 30 3.4 3.9 (573.5) (271.1) d18:1/20:0 593.57 594.6 264.1 30 3.9 5.3 (601.6) (271.1) d18:1/22:0 621.61 622.6 264.1 30 4.6 7.2 (629.6) (271.1) d18:1/24:0 649.64 650.6 264.1 30 5.5 9.9 (657.6) (271.1) d18:1/24:1 647.62 648.6 264.1 30 4.7 7.3 (655.6) (271.1)

Elution times are given for a 3 ´ 150 mm Agilent XDB-C8 column (5 µm pore size) and an Agilent Poroshell 120, 2.1 ´ 150 mm SB-C18 column (2.7 µm pore size). Mass/charge (m/z) values in parentheses are for D7-labeled compounds formed in the reactions. Note that triple quadrupole mass spectrometers are generally only accurate to unit mass (whole number). a d18:0 forms are commonly referred to as dihydroceramides; d18:1 forms are referred to as ceramides b d18:1/17:0 ceramide is the C17:0 ceramide internal standard. c Reactions or cell incubations with D7-dihydrosphingosine result in formation of D7-labeled forms of dihydroceramide and ceramide. The m/z values for these are given in parentheses.

71 Chapter 3

An example chromatogram of metabolites detected in a defined mixture of ceramide standards (each ceramide at 500 nM) is shown in Figure 3.2.

Figure 3. 1. Chromatogram of a defined mixture of ceramides. Peaks generated from individual SRM scans for C17:0, C16:0, C18:0, C20:0, C22:0 and C24:0 ceramides detected in 500nM external standard. Chromatograms were generated using Xcalibur software. Abbreviation: Cer, ceramide; RT, retention time; AA, peak area.

For quantifying metabolites analysed with LC-MS/MS, the vendor software (Xcalibur for Thermo MS systems) was used. Peak areas were expressed as a ratio relative to the C17:0 ceramide or other relevant internal standards for each sample. Similarly, the external standard peak areas are expressed as ratio to the same amount of internal standards. The resulting standard curves were used to quantify each metabolite measured in pmoles.

72 Chapter 3

3.3.2. Assaying CerS activity in living cells

To assay CerS activity in living cells, cells are seeded on the preceding day at the desired density as described above (section 2.2). The cells are pre-treated with the desired drugs or genetic manipulations before incubation for 1 h with 500 nM D7- dihydrosphingosine. After 1 h incubation, cells are washed once with 1´ PBS to remove serum proteins and culture medium. The cells are held on ice, and 0.4 mL 75 % isopropanol containing 50 pmoles C17:0 ceramide internal standard is added to each well. A cell scraper is used to collect the cells into the solvent mixture, then the cell extracts are transferred to a 15-mL glass tube. The culture wells are then washed with another 0.4 mL 75 % isopropanol, and this extract is combined with the first. To the collected mixture, 1.2 mL ethyl acetate is added. The mixture is vortexed thoroughly, then sonicated for 2 h in a sonicating water bath (Thermoline Scientific, Australia) with mixing every 30 min. Cell extracts are centrifuged at 3,700 g for 10 min to pellet insoluble debris. The supernatant is transferred to a 5-mL glass tube. A second extraction is performed on the insoluble debris with another 2 mL ethyl acetate- isopropanol-water (6:3:1, v/v/v) solvent for another 1 h in the sonicating water bath. Extract is centrifuged and supernatant is combined with the first. The combined supernatant is dried down under vacuum in a SpeedVac evaporator as described in section 2.3.1.2. Dried extracts are reconstituted with vigorous vortexing in 0.25 mL HPLC mobile phase. The 5-mL tubes were sealed and centrifuged at 3,700 g for 10 min to pellet any insoluble material, and 0.2 mL of the supernatants are transferred to glass HPLC vials with 300 µL inserts. Sample vials are stored at 4°C until MS analysis. Quantification of D7-dihydroceramides and D7-ceramides is as described above. In living cells, D7-dihydroceramides formed from loaded D7-dihydrosphingosine are rapidly converted to D7-ceramides by DEGS. Therefore, it is important to quantify D7- ceramides as well as D7-dihydroceramides.

73 Chapter 3

3.3.2. Extension of LC-MS/MS method

For the cell-based assay, the LC-MS/MS method developed above was extended to include profiling of hexosyl-ceramide (HexCer) and SM, downstream metabolites of ceramide (Table 3.2). A list of precursor and product ions added to the method can be seen on Table 3.2. To quantify other lipid species, relevant internal standards and external standards have been added to the mixture, as shown in Table 3.2.

Table 3. 2. List of sphingolipid metabolites analysed with LC-MS/MS. Components of the internal standards (red) and external standards (blue) are highlighted in the list. Abbreviations: Cer, ceramides; dh-, dihydro-.

Metabolites Precursor ion Product ion Collision (m/z) (m/z) energy (eV) Ceramides (Cer) C17:0 Cer 552.7 264.1 30 C14:0 Cer 510.6 264.1 30 C14:0 dhCer 512.6 494.6 17 C16:0 Cer 538.6 264.1 30 C16:0 dhCer 540.4 522.2 17 C18:1 Cer 564.7 264.1 30 C18:0 Cer / C18:1 dhCer 566.7 264.1 30 C18:0 dhCer 568.7 550.7 15 C20:1 Cer 592.7 264.1 30 C20:0 Cer 594.7 264.1 30 C22:1 Cer 620.7 264.1 30 C22:0 Cer 622.7 264.1 30 C23:1 Cer 634.7 264.1 30 C23:0 Cer 636.7 264.1 30 C24:1 Cer 648.8 264.1 30 C24:0 Cer 650.8 264.1 30 C24:1 dhCer 650.9 632.9 17 C24:0 dhCer 652.9 634.9 23 C25:1 Cer 662.8 264.1 30 C25:0 Cer 664.8 264.1 30 C26:1 Cer 676.8 264.1 30 C26:0 Cer 678.8 264.1 30 Hexocylceramides (HexCer)

74 Chapter 3

C12:0 HexCer 644.6 264.1 35 C14:0 HexCer 672.7 264.1 35 C16:0 HexCer 700.7 264.1 35 C18:1 HexCer 726.7 264.1 35 C18:0 HexCer 728.7 264.1 35 C20:1 HexCer 754.7 264.1 35 C20:0 HexCer 756.7 264.1 35 C22:1 HexCer 782.7 264.1 35 C22:0 HexCer 784.7 264.1 35 C24:1 HexCer 810.8 264.1 35 C24:0 HexCer 812.7 264.1 35 Sphingomyelin (SM) C12:0 SM 647.5 184.0 35 C14:0 SM 675.5 184.0 35 C14:0 dhSM 677.5 184.0 35 C16:1 SM 701.6 184.0 35 C16:0 SM / C16:1 dhSM 703.6 184.0 35 C16:0 dhSM 705.6 184.0 35 C18:1 SM 729.6 184.0 35 C18:0 SM / C18:1 dhSM 731.6 184.0 35 C18:0 dhSM 733.6 184.0 35 C20:1 SM 757.6 184.0 35 C20:0 SM / C20:1 dhSM 759.6 184.0 35 C20:0 dhSM 761.6 184.0 35 C22:1 SM 785.6 184.0 35 C22:0 SM / C22:1 dhSM 787.6 184.0 35 C22:0 dhSM 789.6 184.0 35 C24:1 SM 813.7 184.0 35 C24:0 SM / C24:1 dhSM 815.7 184.0 35 C24:0 dhSM 817.7 184.0 35

The data is processed as described previously with Xcalibur software. An example of standards curves used to quantify each metabolite is shown in Figure 3.2. For the lipid species with no components in the external standard mixture, the nearest chain length external standard present was used instead (Table 3.2).

75 Chapter 3

Figure 3. 2. Standard curves used for quantification of ceramide, HexCer and SM. Standard curves showing the mean peak area ratio (external standard (ESTD): internal standard (ISTD)) of C18:0 ceramide, C18:0 HexCer and C18:0 SM in the external standard solutions ranging from 0.039- 160.0 pmoles on column (20 µL injection volume). At least N = 2 for each concentration from each MS run. (A-C) and (D-F) represent standard curves generated from two separate MS runs on the Thermo Fisher Scientific Quantum Access mass spectrometer. Error bars represent standard deviation (SD). Data was fitted to a log-log line using GraphPad Prism software. Error bars are not plotted where the error bars would be shorter than the height of symbols.

76 Chapter 3

3.4. Results

3.4.1. Validation of in vitro CerS activity assay

An example reaction assaying C24:1 ceramide synthase activity in mouse liver homogenates is showed in Figure 3.3. As expected, product formation is inhibited with the well characterised ceramide synthase inhibitor FB1297.

Figure 3. 3. Chromatograms showing peaks for D7-labelled C24:1 dihydroceramide products. Reactions were set up with 25 µg mouse liver homogenate, and D7-dihydrosphingosine (10 µM) and C24:1-CoA (50 µM) as substrates. Chromatograms shown in (B) are for a reaction that included the CerS inhibitor FB1 (10 µM), illustrating the inhibition of product (D7-C24:1 dihydroceramide) formation relative to the vehicle control reaction (A). Chromatograms for the C17:0 ceramide internal standard are also shown for reference.

77 Chapter 3

3.4.2. Validation of cell-based CerS assay

To validate the method developed for assaying CerS activity in living cells, U-251 glioblastoma cells were pre-treated with either vehicle control or 5 µM FB1 for 1 h, prior to a 1 h incubation with 500 nM D7-dihydrosphingosine substrate. As expected, LC-MS/MS analysis on cell extracts showed that D7-dihydrosphingosine is converted to D7-dihydroceramides and D7-ceramides in U-251 cells. This conversion is inhibited with 5 µM FB1 treatment (Figure 3.4).

150 Vehicle Fumonisin B1

cells 100 5

50 pmoles/ 10

0

C16:0C18:0 C24:0 C24:1 C16:0C18:0 C20:0 C22:0 C22:1C24:0 C24:1

D7-dihydroceramides D7-ceramides Figure 3. 4. FB1 treatment inhibits D7-ceramides formation. Formation of D7- dihydroceramides and D7-ceramides is inhibited with 5 µM FB1. Column graphs show mean ± SD (error bars) for D7-dihydroceramides and D7-ceramides levels, expressed as pmoles/ 105 cells, N = 4 per group. Unpaired t-tests were performed for vehicle versus FB1 treated groups (P < 0.05). FB1 significantly reduced all D7-dihydroceramides and ceramides except for D7-C18:0 and C20:0 ceramides.

3.5. Discussion

The current study presents an analytical method that is efficient for assaying CerS activity by targeted quantification of ceramide species. Both the in vitro and cell-based CerS assay were also validated using FB1, a known CerS inhibitor. Samples treated with FB1 had significantly less ceramide products formed (Figure 3.3-3.4). Assaying CerS activity in living cells with deuterated dihydrosphingosine coupled with LC-

78 Chapter 3

MS/MS has several key advantages when compared to traditional radioactive assays or fluorescence-based assays. Traditionally, radioactive assays are used to assess CerS activity. Tritiated dihydrosphingosine or sphingosine is used as the substrate and TLC is used to resolve tritiated ceramide products79,80. Although tritiated substrates are safe when used properly, they can be dangerous if handled inappropriately. Another alternative radioactive CerS assay has also been described using 14C-labeled FA substrate298.These radioactive assays have to be conducted with extra care and the use and storage of radioactive chemicals must be monitored closely as well. Radioactive work also involves additional paperwork and regulations. Tritiated or 14C-labeled ceramides on the TLC plate can be quantified by scraping the bands into vials and quantified for their radioactivity by scintillation counting. Alternatively, the TLC plate can be exposed to a phosphorimager screen and imaged for densitometry. CerS activity can also be assayed using fluorescent assays. The principle of a fluorescent assay is the same and the fluorescent ceramide product formed can be resolved and visualized on the TLC plate or detected through a fluorescent detector coupled to an HPLC system299. A major disadvantage of the fluorescent approach is that the fluorescent signal can fade if samples are not protected from laboratory lights. Hence, extra precautions need to be taken. Plus, long term storage of the samples for second analysis or recounts may be a problem. Using fluorescent substrates for assaying CerS activity in living cells may not be ideal as the fluorescent signal may not be preserved by the time the cells are treated and extracted for downstream analysis. In this regard, the utilization of deuterated labelled substrates in current method allows monitoring of the deuterated products formed within a given amount of time or under various treatments. Therefore, assays with deuterated substrates such as D7-dihydrosphingosine assessed by LC-MS/MS are a better option as there is no risk of losing signals and the samples can be stored for a longer time if needed.

LC-MS/MS is a better approach for quantifying ceramide formed from CerS assays. Although TLC-based approach can distinguish very long chain ceramide from long chain ceramide and dihydroceramide, it cannot undoubtedly distinguish a C24:0 from C24:1 or C16:0 from C18:0 dihydroceramide species. LC-MS/MS approach has a clear advantage in this matter as it can easily distinguish ceramide species of different chain length based on their masses and retention time. The lipid identification can be

79 Chapter 3

confirmed by running specific external standards. However, the LC-MS/MS approach might not be accessible for every lab as it requires access to highly specialized equipment and expertise. The HPLC run time for each sample can take up to 15 min per sample. Hence, the machine time required for a typical experiment is usually a full working day and possibly an overnight booking in order to run sample replicates, external standards and regular blank washes. This approach is impractical especially when MS equipment is centralized and shared between multiple users. Access and machine time can also be costly, as is commonly the case with specialized core equipment. However, if budget permits and with the right expertise, LC-MS/MS is a highly accurate and sensitive approach for quantifying ceramide formation. Although the method developed in this chapter is specific for Thermo Fisher TSQ Quantum Access, the HPLC conditions can be carried over onto different instrumentations and with the right optimization, the method can be tailored according to the needs of the experiment.

For the purpose of quantifying other sphingolipids, the method has been extended to include SM and HexCer species (section 3.3.2). The extension of the LC-MS/MS method is used for targeted quantification of ceramide, SM and HexCer in cultured cells or tissues extracts in subsequent chapters.

80 Chapter 4

Chapter 4: In vitro characterization of CerS1 inhibition by P053

4.1. Introduction

Ceramide is the central metabolite of the sphingolipid family. Ceramide is structurally comprised of a sphingosine or dihydrosphingosine backbone with a variable fatty acyl side-chain (Figure 1.7)77,78. Ceramide itself serves as a backbone for synthesis of other sphingolipids such as SM, GluCer, GalCer and other complex glycosphingolipids69,77. They are potent cellular and physiological signalling molecules regulating ER stress, apoptosis and insulin signalling78,103,138,164. As a bioactive lipid, ceramide can directly activate protein kinases and phosphatases mediating cellular signalling processes139,300- 302 and possesses biophysical properties regulating membrane fluidity and permeabilization102,303,304. Accumulation of lipotoxic ceramide has been implicated in metabolic distress and insulin resistance58,59,108,236. Researchers have been manipulating ceramide levels through genetic or pharmacological approaches to improve disease outcome in mouse models221,225,230,231. However, there are no specific inhibitors of individual CerS isoforms.

The biological properties of ceramide are heavily influenced by variation in the fatty acyl-chain of ceramide. In the context of insulin resistance, C16:0 ceramide has been implicated as the major culprit in liver and adipose tissue insulin resistance58,59,108. C16:0 ceramide antagonizes insulin receptor signalling and has a negative impact on energy expenditure via b-oxidation. Interestingly, very long chain ceramides, particularly C24:0 ceramides synthesized by CerS2 are protective against insulin resistance. Another ceramide species implicated in insulin resistance is C18:0, synthesized by CerS1. CerS1 is highly expressed in the SkM. Muscle C18:0 ceramide has been correlated with impaired insulin signalling, visceral fat and blood pressure48.

Due to the significant negative impact of long chain ceramides, selective inhibition of CerS isoforms, particularly CerS5, CerS6 and/or CerS1 would be valuable for improving metabolic health. Currently, there is a lack of potent and selective isoform-

81 Chapter 4

specific CerS inhibitors278. Interestingly, FTY720, an FDA approved drug for the treatment of multiple sclerosis has been reported to inhibit CerSs. FTY720 itself is an analogue of the endogenous lipid sphingosine305. FTY720 can be phosphorylated in vivo and acts as a potent agonist of S1P receptors. The non-phosphorylated form of FTY720 demonstrates non-selective inhibition of CerSs, as an off-target effect92,275. Due to its immunosuppressive properties, FTY720 is not an ideal drug for inhibiting CerSs in the context of insulin resistance studies. Following this finding, our lab had previously synthesized a series of non-phosphorylatable, chiral FTY720 analogues (Figure 4.1).

Figure 4. 1. Compound structures and selectivity against CerS1. (A) Structures of FTY720, AAL(S), G024 and P053. (B) CerS1 activity, expressed as % of control, in the presence of 1 µM AAL(S), G024 or P053. Graphs show mean ± SD (N = 3 for each concentration). The data were obtained by A/Prof Anthony Don and Dr. Hamish Toop.

Among the analogues tested, AAL(S) was previously shown to be reasonably potent in inhibiting C18:0 ceramide synthase activity of CerS1-expressing cell extracts, achieving

82 Chapter 4

50% inhibition at 1 µM. However, AAL(S) showed limited selectivity for CerS1 over other CerS isoforms279. AAL(S) is also highly cytotoxic and therefore not a suitable molecule for studying CerS functions. Further modifications were made to the alkyl tail of AAL(S), resulting in the benzyl tail analogue G024. While we were able to eliminate the cytotoxic properties of AAL(S) with G024, there were still limited inhibitory properties and specificity for CerS1. Neither AAL(S) nor G024 were found to reduce C18:0 ceramide levels in living cells. Using the Topliss tree as a guide306,307, subsequent modifications to the benzyl tail of G024 were made, resulting in the identification of P053 [(S)-2-amino-4-(4-(3,4-dichlorobenzyloxy)phenyl)-2-methylbutan-1-ol].

4.2. Aims

The overall aim of the work presented in this chapter was to characterize the inhibitory properties of P053 on CerS isoforms, particularly CerS1, in vitro and in cell culture.

The specific aims of this chapter: 1. To examine the potency and mode of inhibition of P053. 2. To identify a suitable cell culture model for assaying effects of P053 in vitro. 3. To investigate the selectivity of P053 for inhibition of CerS1 and its effects on cellular sphingolipid levels.

83 Chapter 4

4.3. Results

4.3.1. Validation of CerS overexpression

Overexpression of mouse and human CerS isoforms in cultured cells was examined using our fluorescent CerS activity assay283. C18:0-CoA was used as the fatty acyl substrate for CerS1 and CerS4; C24:1-CoA for CerS2; and C16:0-CoA for CerS5 and CerS6. Peak areas of NBD-C16:0, C18:0 and C24:1 dihydroceramides from reactions using CerS-expressing lysates are plotted relative to that of respective untransfected- control cell lysates (Figure 4.2). Overexpression of CerS isoforms greatly increased CerS activity compared to control.

3000

2000

1000 Activity (% of control) 0

hCerS1mCerS1hCerS2mCerS2hCerS4mCerS5hCerS6 CerS isoforms Figure 4. 2. Activity of overexpressed CerS isoforms.

Graphs show mean ± SEM (N = 3 for each concentration) of CerS activity, expressed as fold- change relative to the respective untransfected control. All overexpressed CerS isoforms showed increased activity. Abbreviation: mCerS, murine CerS; hCerS, human CerS.

4.3.2. IC50 of P053

The IC50 for inhibition of all CerS isoforms with P053, AAL(S), G024, FTY720 and FB1 compounds was determined using the CerS activity assay. G024 is the benzyl-tail derivative prior to P053 and was included for comparison. The IC50 for inhibition of CerS1 with P053 was 0.54 µM, an order of magnitude lower than that of G024. The

IC50 for inhibition of CerS2, CerS4, CerS5 and CerS6 by P053 were all ³ 7 µM, demonstrating a strong selectivity for CerS1 over other CerS isoforms (Figure 4.3 and

84 Chapter 4

Table 4.1). The IC50 for P053 on CerS1 was comparable to that of the potent but non- selective CerS inhibitor Fumonisin B1 (FB1)297, which inhibited all CerS isoforms with sub-micromolar IC50 (Table 4.1).

Figure 4. 3. Activity of CerS isoforms as a function of P053 concentration.

Graphs show mean ± SEM (N = 3 for each concentration). Abbreviation: mCerS, murine CerS; hCerS, human CerS.

Table 4. 1. Calculated IC50 values for FB1, FTY720, AAL(S), G024, and P053 on multiple CerS isoforms. Values are mean ± SEM (N = 3 for each concentration). Abbreviation: mCerS, murine CerS; hCerS, human CerS.

IC50 (µM) Fatty Enzyme acyl-CoA FB1 FTY720 AAL(S) G024 P053 0.22 ± 9.94 ± 1.89 ± 6.04 ± 0.54 ± hCerS1 C18:0 0.053 0.069 0.050 0.095 0.056 5.61 ± 1.62 ± 0.46 ± mCerS1 C18:0 - - 0.046 0.049 0.080 0.15 ± 19.11 ± 28.13 ± 14.45 ± 28.55 ± hCerS2 C24:1 0.076 0.102 0.170 0.087 0.152 10.61 ± 15.56 ± 18.51 ± mCerS2 C24:1 - - 0.130 0.121 0.120 0.92 ± 26.04 ± 7.81 ± 17.87 ± 17.16 ± hCerS4 C18:0 0.058 0.175 0.192 0.119 0.088 0.50 ± 15.01 ± 12.38 ± 24.29 ± 7.20 ± mCerS5 C16:0 0.054 0.069 0.064 0.216 0.097 0.29 ± 5.81 ± 7.52 ± 8.53 ± 11.39 ± hCerS6 C16:0 0.052 0.197 0.096 0.125 0.165

85 Chapter 4

4.3.3. Mode of inhibition of CerS1 by P053

To study the mode of CerS1 inhibition by P053, CerS1 activity was measured using varying concentrations of C18:0-CoA or dihydrosphingosine substrates, and in the presence or absence of P053. Data for CerS1 activity as a function of dihydrosphingosine or C18:0-CoA substrates concentration was plotted using the allosteric sigmoidal model available in GraphPad Prism. As shown previously by Lahiri et al.92 for CerS2, CerS1 activity as a function of substrate concentration is sigmoidal, particularly for the C18:0-CoA substrate, indicative of a cooperative binding model.

P053 reduces maximal reaction rate (Vmax) without notably affecting substrate affinity,

Khalf (Figure 4.4), indicating that it is a non-competitive inhibitor.

Figure 4. 4. Kinetic analysis of CerS1 inhibition by P053.

CerS1 activity as a function of (A) dihydrosphingosine or (B) C18:0-CoA concentration. For (A) C18:0-CoA concentration was held at 25 µM, whilst in (B) dihydrosphingosine concentration was held at 10 µM. Calculated Vmax and Khalf values are below the graphs. Graphs show mean ± SEM (N = 3 for each concentration), and r2 of all fits are ≥ 0.95. Non-competitive inhibition was confirmed in three independent experiments for each substrate.

86 Chapter 4

4.3.4. P053 inhibits C18 ceramide synthase activity in cultured neurons

To investigate whether P053 can inhibit C18:0 ceramide synthesis in live cells, primary mouse cortical neurons were treated with varying concentrations of P053 for 2 h followed by incubation with 500 nM D7-dihydrosphingosine for 1 h. LC-MS/MS analysis showed that P053 treatment reduced de novo synthesis of D7-C18:0 ceramide from D7-dihydrosphingosine in a dose-dependent manner. At 1 µM, P053 reduced mean D7-C18:0 levels by 66.7 % compared to vehicle control (p < 0.0001, Dunnett’s post hoc test, Figure 4.5). This reduction was more efficient than FTY720 treatment at the same dosage with only 29.1 % reduction compared to vehicle control. This result confirmed that P053 is cell permeable and can inhibit CerS1 activity in living cells.

Figure 4. 5. P053 selectively inhibits de novo synthesis of deuterated C18:0 ceramide.

Graph shows the mean ± SEM for (A) D7-C18:0 ceramide levels and (B) D7-C16:0 ceramide levels, expressed as percentage of vehicle control, using combined data of two independent cell treatments and experiments (N = 3 per treatment per experiment). Statistical significance was measured by one-way ANOVA followed by a Dunnett’s post hoc test, comparing each treatment with vehicle control (0 µM). **** p-value < 0.0001.

87 Chapter 4

4.3.5. Effect of P053 treatment on endogenous ceramide levels

After verifying that P053 is cell permeable and can inhibit CerS1 activity in a live cell assay using cortical neurons, a suitable cell line was chosen for investigating the potency and specificity of P053 in reducing endogenous ceramides. According to publicly available data on RNA expression profiles of CerS1 in various cell lines (Cell Atlas), most immortalised cell lines have very low CerS1 expression (Figure 4.6). HEK293 and an endometrium adenocarcinoma cell line (AN3-CA) have the highest CerS1 RNA expression profiles. HEK293 cells also produce more C18:0 ceramides than other immortalised cell lines that our lab previously tested. Therefore, HEK293 cells were used in subsequent cell-based experiments. HEK293 cells were treated for 24 h with vehicle control or P053 (10 nM, 30 nM, 100 nM and 300 nM). P053 significantly reduced C18:0 ceramide at just 30 nM and achieved 53 % reduction at 100 nM (Figure 4.7). No other forms of ceramide were reduced by P053 treatment, but a minor increase in C24:0 ceramide was observed.

Figure 4. 6. RNA gene expression for CerS1 in a panel of cell lines.

Publicly available RNA-seq data on 56 cell lines was used to estimate the transcript abundance of CerS1. The abundance of gene expression is reported in transcripts per million (TPM). For each cell line, the average TPM value for replicate samples were used as the abundance score. Image credit: Human Protein Atlas, available from https://www.proteinatlas.org/ENSG00000223802-CERS1/cell - rna.

88 Chapter 4

200 10nM

160 ++ 30nM 100nM 120 # * * 300nM 80

40 Ceramides (% of control) 0 16:0 18:0 20:0 22:0 24:0 24:1 ceramide species

Figure 4. 7. P053 selectively reduced endogenous C18:0 ceramides in HEK293 cells. Graph shows the mean ± SEM for endogenous ceramide levels, expressed as percentage of vehicle control, using combined data of two independent cell treatments and experiments (N = 4 per treatment per experiment). Statistical significance was measured by one-way ANOVA followed by a Dunnett’s post hoc test, comparing each treatment with vehicle control; +, P < 0.05; #, P < 0.01; *, P < 0.001.

89 Chapter 4

4.3.6. Effect of P053 treatment on endogenous hexosyl-ceramide levels

In addition to reducing endogenous C18:0 ceramides in HEK293 cells, P053 selectively reduced C18:0 HexCer, derived from C18:0 ceramide (Figure 4.8). At only 30 nM, P053 reduced C18:0 HexCer by 38.7 %, reaching 62.1 % reduction at 100 nM.

200 10nM 160 30nM # 100nM 120 ** * 300nM 80

40 HexCer (% of control) 0 16:0 18:0 20:0 22:0 24:0 24:1 HexCer species

Figure 4. 8. P053 selectively reduced endogenous C18:0 HexCer in HEK293 cells. Graph shows the mean ± SEM for endogenous HexCer levels, expressed as percentage of vehicle control from combined data of two independent cell treatments and experiments (N = 4 per treatment per experiment). Statistical significance was measured by one-way ANOVA followed by a Dunnett’s post hoc test, comparing each treatment with vehicle control; #, P < 0.01; *, P < 0.001.

90 Chapter 4

4.3.7. Effect of P053 treatment on endogenous sphingomyelin levels

SM is another downstream product derived from ceramide. Similar to HexCer, P053 treatment significantly reduced 18 carbon chain SM species. Both C18:0 and C18:1 SM were reduced with P053 treatment in a dose-dependent manner (Figure 4.9). At 100 nM, P053 reduced C18:0 SM by 36.3 %.

200 10nM 30nM 160 100nM 300nM 120 ** ** 80

SM (% of control) 40

0 14:0 16:0 16:1 18:0 18:1 20:0 22:0 22:1 24:0 24:1 SM species Figure 4. 9. P053 selectively reduced endogenous C18:0 SM in HEK293 cells.

Graph shows the mean ± SEM for endogenous SM levels, expressed as percentage of vehicle control. Combined data of two independent cell treatments and experiments (N = 4 per treatment per experiment) shown. Statistical significance was measured by one-way ANOVA followed by a Dunnett’s post hoc test, comparing each treatment with vehicle control; *, P < 0.001.

91 Chapter 4

4.3.8. Effect of long-term P053 treatment on sphingolipid levels

The long-term effect of P053 treatment on endogenous ceramide and SM levels was examined. HEK293 cells and cortical mouse neurons were treated with vehicle control or 0.5 µM P053 for 24, 48 and 72 h. In HEK293 cells, 0.5 µM achieved a 65.8 % reduction in endogenous C18:0 ceramide levels after 24 h, reaching 77 % at 72 h. C18:0 SM was also reduced, reflecting the lack of C18:0 ceramide for SM synthesis (Figure 4.10). In neurons, there was only a 17.4 % reduction in C18:0 ceramide after 24 h of P053 treatment. Interestingly, C18:0 ceramide was not further reduced at 48 and 72 hours. Similarly, P053 treatment did not significantly affect C18:0 SM levels (Figure 4.10).

Figure 4. 10. Effect of long term P053 treatment on endogenous ceramide and SM levels. Graphs show the mean ± SEM for endogenous ceramide and SM levels, expressed as percentage of vehicle control (N = 4 per treatment per experiment) in (A-B) HEK293 cells and (C-D) cortical mouse neurons. Statistical significance was calculated by multiple t-tests with Holme-Sidak post-test, comparing each treatment to a matched vehicle control; +, P < 0.05; #, P < 0.01; *, P < 0.001.

92 Chapter 4

4.3.9. P053 toxicity

To determine whether P053 is cytotoxic, HEK293 cell viability upon P053 treatment was examined. HEK293 cells were treated with 0, 0.1, 0.3, 1 and 3 µM P053 for 72 h, followed by flow cytometry to measure the percentage of dead cells by propidium iodide staining. The result indicated that P053 had no effect on HEK293 cell viability at or below 1 µM (Figure 4.11).

Figure 4. 11. Effect of P053 on HEK293 cell viability.

Graph shows the mean ± SEM for the percentage of non-viable HEK293 cells, as determined by propidium iodide (PI) staining following 72 h treatment with P053. Combined data from two independent experiments, each with 4 separate cell treatments are shown. No statistical significance was identified by one-way ANOVA followed by a Dunnett’s post hoc test.

93 Chapter 4

4.3.10. Ceramide profiling in cultured skeletal muscle cells

Given the high expression of CerS1 in SkM82, I next sought to identify a suitable muscle cell culture model for studying P053. In primary mouse myoblasts and myotubes, C16:0 ceramide is the dominant long chain ceramide species (Figure 4.12). This contradicts muscle tissue ceramide composition, where C18:0 ceramide is the predominant species. In immortalized C2C12 cell line, there was very little C18:0 ceramide relative to C16:0 and very long chain ceramides (Figure 4.13). Therefore, neither primary myotubes nor C2C12 cells are suitable models for investigating effects of P053 in vitro.

Figure 4. 12. Ceramide composition of primary mouse myotubes and myoblasts.

Levels of major ceramide species of (A) differentiated primary mouse myotubes and (B) primary myoblasts, expressed as pmoles per 105 cells. Graph shows the mean ± SEM, N = 3.

Figure 4. 13. Ceramide composition of C2C12 cells. Levels of major ceramide species of C2C12 cells, expressed as pmoles per 105 cells. Graph shows mean ± SEM, N = 4.

94 Chapter 4

4.4. Discussion

Recently, studies have shown that different forms of ceramide, synthesized by different CerS isoforms, can regulate distinct physiological processes78,308. However, these studies are often hindered by the lack of isoform-selective CerS inhibitors. Commonly, genetic approaches such as CerS isoform knock out or knock down, or pharmacological approaches using myriocin, are used in in vivo studies58,59,105,108,196,224,309. Although potent in blocking biosynthesis of ceramides, myriocin inhibits SPT, the rate limiting first step of the de novo synthesis pathway, affecting all forms of ceramide and other sphingolipids. Results from this chapter describe the first potent and selective inhibitor of any CerS isoform, showing that P053 selectively inhibits CerS1 both in vitro and in situ.

P053 demonstrates ≥ 15-fold selectivity for CerS1 over other CerS isoforms in in vitro

CerS assays with an IC50 in the nanomolar range (Table 4.1). P053 (IC50 = 0.54 µM ±

0.056) is also ≥ 18-fold more potent than FTY720 (IC50 = 9.94 µM ± 0.069) and ≥ 11- fold more potent than G024 (IC50 = 6.04 µM ± 0.095) on CerS1. When investigating the mode of inhibition, the adequacy of the fit of the enzyme kinetic data using the Michaelis-Menten or allosteric-sigmoidal models was assessed by performing the replicates test in Prism. Allosteric-sigmoidal model was the preferred model in the replicates test. The results of a comparison of fits by performing extra sum of squares F tests supported the allosteric sigmoidal model as well (p < 0.0001). Non-regression and allosteric sigmoidal fitting in Prism generates the following analysis:

Vmax, Khalf and Hill coefficients. Khalf is the concentration of substrate that produces a half-maximal enzyme velocity, while Hill coefficients give an indication of the presence of in substrate binding. CerS isoforms are bi-substrate enzymes. They require the binding of both dihydrosphingosine and fatty acyl-CoA for generation of ceramides. Therefore, the co-operative binding model, which is particularly notable when varying the concentration of the fatty acyl-CoA substrate, may indicate that this substrate aids binding of the dihydrosphingosine substrate, perhaps through a conformational change.

Results from this chapter indicate that P053 is a non-competitive inhibitor with respect to either the dihydrosphingosine or C18:0 fatty acyl-CoA substrate. It is possible that 95 Chapter 4

P053 binds to an allosteric lipid-binding site on CerS1. Of note, both CerS1 and CerS4 activities were assayed using C18:0 fatty acyl substrate, indicating that inhibition by P053 is highly specific to the CerS1 isoform and not the specific fatty acyl substrate used to measure enzyme activity. The Hill coefficient generated for CerS1 activity with varying C18:0-CoA concentrations without the P053 inhibitor was > 1.0. This indicates positive cooperativity where binding of C18:0-CoA may facilitate binding of subsequent ligands at other sites. However, the hill coefficient for CerS1 activity with varying dihydrosphingosine concentrations and in the absence of P053 was 1.0, suggesting independent binding of ligand. This may be because there is plenty of C18:0-CoA, so the dihydrosphingosine substrate is able to bind more readily. Hence, the working model here might indicate that the binding of fatty acyl-CoA substrate is required prior to the binding of dihydrosphingosine substrates. The mode of inhibition may be more complex when considering the possibility of CerS1 forming dimers or multimers that may interact allosterically. This theory is supported by findings from Laviad et al.310 where CerS homodimers were reported under cross-linking or non- denaturing conditions. Their data suggested that CerS2 can form a heterodimer with CerS5, and the activity of CerS2 depends on CerS5 catalytic activity. Although similar findings have not been reported for CerS1, there is a possibility that CerS1 may form heterodimers as well. Currently, there is no structural data or suitable homology model for the CerS family. Hence, it is difficult to confidently conclude which substrate binds first to form an enzyme-substrate complex and how CerS1 may interact with dimers or multimers. Future investigation such as cross-linking of either substrate to CerS1 will help to clarify the inhibitory mechanism of P053. Regardless, P053 appears to inhibit CerS1 activity in a non-competitive manner.

To characterize P053 in cultured cells, mouse cortical neurons were tested first, as CerS1 is highly expressed in neuronal cultures113. As expected, in cultured cortical neurons, a 2 h pre-treatment with P053 significantly reduces the ability of CerS1 to synthesize deuterated C18:0 ceramide (Figure 4.6). However, due to lack of convenience and availability of primary neurons, a transformed cell line with adequate CerS1 expression was used for further experimentation. According to the Human Protein Atlas311, a Swedish-based program aiming to map all human proteins in cells, tissues and organs, the HEK293 cell line has the highest RNA expression of CerS1

96 Chapter 4

across a diverse panel of 56 human cell lines (Figure 4.7). From past experience of our lab, HEK293 cells produce more C18:0 ceramide than other cell lines available. In HEK293 cells, incubation with P053 for 24 h successfully inhibited endogenous C18:0 ceramide production without notably affecting levels of other ceramide species and their derivatives, HexCer and SM (Figure 4.8- 4.10). The reduction in C18:0 HexCer and C18:0 SM was not surprising as depletion of their precursor, C18:0 ceramide would have a direct impact on their abundance. The extent of reduction in C18:0 SM was only 36.3% at 100 nM of P053 while there was more than 50% reduction in C18:0 ceramide and C18:0 HexCer. This may suggest a slower turnover rate for SM. CerS1 inhibition with P053 was also well dissected from non-specific pro-apoptotic properties of sphingoid base analogues279,312,313, as the compound inhibited C18:0 ceramide synthesis at a concentration at least one order of magnitude lower than those required to promote apoptosis in HEK293 cells (Figure 4.12).

Results from the long-term treatment of HEK293 cells with P053 (24, 48 and 72 h) demonstrated that CerS1 inhibition did not eliminate endogenous 18-carbon chain ceramide and SM despite the efficacy of the 0.5 µM dosage and longer incubation time (Figure 4.11). The levels of C18:0 ceramide and derivatives were not significantly lower at 48- and 72-hour time points. It is therefore possible that P053 can be metabolised within the cells. Further experiments are needed to verify whether P053 is continuing to inhibit CerS1 activity at 48- and 72-hour time points. Another possibility is that C18:0 ceramides were generated by CerS4. CerS1 inhibition via P053 might lead to complementary increased activity of CerS4, which also synthesizes C18:0 ceramide. Interestingly, long term P053 treatment did not reduce C18:0 ceramide levels in cortical neurons (< 20 % reduction) to the same extent as in HEK293 cells (> 60 % reduction). This suggests that neurons may require C18:0 ceramide for survival, hence can possibly override CerS1 inhibition via other means. Another more likely explanation is that the rate of synthesis of new lipids is relatively slow in mature cultured neurons, which are post-mitotic cells. The precise role of C18:0 ceramide in neuronal survival is unknown but it has been shown that C18:0 ceramide positively influences glutamate exocytosis in the PC12 cell line derived from a pheochromocytoma of the rat adrenal medulla314. The secreted neurotransmitter glutamate can exert its effects by binding to or activating various cell surface receptors and facilitate ion exchange between neuronal synapses315.

97 Chapter 4

Hence, changes in ceramide levels may influence neurotransmission. C18:0 ceramide is also required for the synthesis of gangliosides, which are essential for central nervous system functions. Ablation of neuronal CerS1 in genetic mouse models can deplete gangliosides and expression of myelin-associated glycoprotein in oligodendrocytes, ultimately leading to behavioural functional deficits in mice96,97. Interestingly, ectopic expression of CerS2 in neurons of CerS1-deficient mice was shown to suppress behavioural deficits brought on by CerS1 deficiency316. Although CerS2 expression could not replenish C18:0 ceramide, it led to the reduction of long-chain sphingoid base substrates. It was therefore hypothesized that accumulation of these substrates may be the underlying cause for neuronal cell death in CerS1-deficient mice as shown in cultured primary neurons where treatment with dihydrosphingosine substrates can induce neurite fragmentation316. Collectively, these findings showed the importance of C18:0 ceramide for neuronal health. Therefore, it is important to consider the possible undesirable effects of CerS1 inhibition by P053 when planning future in vivo studies.

CerS1 is highly expressed in SkM78,82. C18:0 ceramide generated from CerS1 in the muscle is associated with inhibition of insulin signalling in insulin-resistant mouse models and obese insulin-resistant human subjects24,48,236,317. A reduction of C18:0 ceramide via CerS1 inhibition may bring beneficial effects on the metabolic functions of SkM. With the intent to study the effect CerS1 inhibition via P053 on SkM metabolic functions, a SkM cell culture was needed. C2C12 is an immortalized mouse myoblast cell line commonly used for studying metabolic functions of SkM. It was developed for in vitro studies and can be differentiated into multinucleated myotubes under low serum conditions318. Accumulation of ceramide has previously been reported in C2C12 cells, which accounts for inhibition of the insulin-stimulated Akt/PKB signalling pathway and reduced glucose uptake and anabolic metabolism319-321. Unfortunately, ceramide profiling data generated in the current study suggests that C2C12 is not a suitable cell culture model as C16:0 ceramide is the dominant long chain ceramide species with very little C18:0 ceramide. It is the same case for primary mouse myoblasts and differentiated myotubes. Given the extremely low level of C18:0 ceramide in these cells, it would be futile to use them for studying CerS1 inhibition by P053. One of the possible reasons for the lack of C18:0 ceramide in these cultured muscle cells may be due to the formulation of the culture medium. Although the culture medium used and

98 Chapter 4

culturing protocol has been established, there may be certain ingredients in the medium such as horse serum, which inhibit or down-regulate CerS1 activities. Future studies will investigate how different culture medium and supplementation can affect CerS1 activities in cultured muscle cells.

Although a suitable SkM culture model could not be identified, data generated from experiments using HEK293 cells and neurons provides pharmacological evidence that endogenous CerS1 is indeed highly selective for C18 fatty acyl substrates, as treatment with P053 resulted in selective reduction of C18:0 ceramide. In summary, P053 is the first potent, isoform-selective CerS inhibitor, specifically targeting CerS1. In the following chapter, the potency, selectivity, and bioavailability of P053 for in vivo use will be investigated.

99 Chapter 5

Chapter 5: Lipidomic characterization of P053 in vivo

5.1. Introduction

Chapter 4.0 investigated the in vitro effects of P053 on CerS activities and on the sphingolipid composition in cultured cells. Although a suitable SkM cell model was not established, results from the previous chapter concluded that P053 is an effective and specific CerS1 inhibitor in vitro. In this chapter, the in vivo effects of P053 on organ lipidomes are investigated and discussed.

As discussed in the literature review, ceramide has been described as a putative lipotoxic lipid species in the context of obesity-related pathologies. Accumulation of ceramide has been reported in various organs and serum of insulin-resistant human subjects and animal models24,47,48,60,232,239,317,322-324. The role of ceramide in inhibition of insulin action has been well described in cell culture models indicating that ceramide acts through PP2A and PKC, which subsequently hinder insulin-mediated Akt signalling50-52,169. Pharmacological inhibition of ceramide synthesis via SPT inhibitor, myriocin, or DEGS1 inhibitor, ferentinide, was able to rescue insulin-stimulated Akt signalling in cultured cells28,222,231,257-259. Accumulation of ceramide in vivo has also been reported in various animal models of diseases such as hypertension, NAFLD, diabetes and insulin resistance53,58,201,218. All these animal models were induced by high fat feeding in either mice or rats.

Recently, studies have implicated the importance of individual ceramide species in the context of metabolic disorders, with C16:0 ceramide being the culprit in liver and adipose tissue under HFD fed conditions58,59,236. Another ceramide species implicated in insulin resistance is C18:0 ceramide in SkM78,82. In SkM, CerS1 is the main CerS isoform and synthesizes C18:0 ceramide, a major contributor to the organ’s ceramide composition. It is now known that high fat feeding can lead to increased CerS1 expression and accumulation of C18:0 ceramide in SkM of mice and is linked to reduced glucose tolerance and impaired insulin action48,236,317. In humans, SkM C18:0

100 Chapter 5

ceramide is positively correlated with visceral fat and blood pressure but negatively correlated with adiponectin, an insulin-sensitizing adipokine48. Therefore, one would think that selective reduction of C18:0 ceramide in SkM may be a possible intervention for treatment of insulin resistance.

5.2. Aims

Since P053 has shown striking potency in the inhibition of CerS1 and reduction of C18:0 ceramide in vitro, I sought to investigate the potency of P053 in vivo, and how the lipidome of SkM and other relevant organ tissues was affected by high fat feeding and P053 administration. Overall, the aim of this chapter was to characterize the lipidome under P053 treatment in vivo.

The specific aims for this chapter are: 1. To determine the bioavailability of P053 in vivo 2. To characterize the efficacy of P053 at reducing C18:0 ceramide levels in vivo 3. To investigate the global changes of the lipidome under P053 treatment

5.3. Results

5.3.1. CerS1 expression in various tissues

To confirm the high expression of CerS1 in brain and SkM as previously reported82, a panel of mouse tissues were screened for CerS1 mRNA levels normalized to 18S rRNA. The expression profile confirmed previous literature and showed that CerS1 is highly expressed in brain and SkM, with minimal or no expression in kidney, liver, heart, and adipose tissues (Figure 5.1). CerS2 expression was higher in tissues lacking CerS1 expression when compared to SkM and brain. This confirms that these tissue samples did contain mRNA, and that CerS1 expression in these tissues was beyond the limit of detection.

101 Chapter 5

Figure 5. 1. Mouse tissues CerS1 mRNA expression.

(A) CerS1 mRNA expression was determined in a panel of mouse tissues by qPCR (N = 3 mice; mean ± SEM). Levels are normalised to 18S rRNA. (B) Results for CerS2 are also shown, to demonstrate the presence of RNA in samples from tissues lacking CerS1 expression. Abbreviation: EpiWAT: epididymal white adipose tissue; BAT: brown adipose tissue.

5.3.2. P053 bioavailability and pharmacokinetics

To investigate the in vivo efficacy of P053, bioavailability was tested by administering a single 5 mg/kg dose by oral gavage to male C57BL/6J mice. The dosage was chosen based on the reported dosage for FTY720 in previous literature276. A maximal plasma concentration of 20 nM was achieved 4 h after administration, with a plasma half-life of 28 h (Figure 5.2).

25

20

15

10

5 Concentration (nM) 0 0 12 24 36 48 60 72 time (h) Figure 5. 2. P053 bioavailability.

Plasma samples were collected at 0, 1, 4, 8, 24, 48 and 72 h time points following a single 5 mg/kg oral dose of P053 (N = 5 mice; mean ± SD). P053 concentration was determined by LC- MS/MS.

102 Chapter 5

5.3.3. Pilot in vivo study

A pilot study was conducted using male C57BL/6J mice following 7 days administration of vehicle control or P053 at 1 mg/kg or 5 mg/kg by oral gavage and fed with chow diet. Tissues were collected at end point for ceramide profiling. C18:0 ceramide levels were examined in muscle and brain. P053 administration at 1 mg/kg had no effect on C18:0 ceramide levels (Figure 5.3). At 5 mg/kg, there was a 31% reduction in C18:0 ceramide in quadriceps SkM. No effect was observed in brain.

600 Vehicle 1mg/kg P053 5mg/kg P053 400

200 pmoles/mg tissue 0 Muscle Brain

Tissue type Figure 5. 3. C18:0 ceramide levels in quadriceps SkM and brain.

Graph shows mean ± SEM of C18:0 ceramide levels, expressed as pmoles per mg tissue in quadriceps SkM and brain, following 7 day daily adminstration of P053. N = 5 for each group.

The ceramide profiles for liver, kidney and heart were also examined (Figure 5.4). The 1 mg/kg group was excluded from the graphs due to absence of any effect compared to vehicle control. 7 days of P053 administration at 5 mg/kg significantly reduced C18:0 ceramide in the quadriceps muscle by 31% (adjusted P = 0.0019) (Figure 5.4A). There were no other significant changes for other ceramide species in SkM and in other tissues (Figure 5.4 B-E).

103 Chapter 5

Figure 5. 4. Tissue ceramide profiles in P053 pilot study.

Graphs show mean ± SEM of ceramide levels, expressed as pmoles per mg tissue in (A) quadriceps SkM, (B) brain, (C) Kidney, (D) Liver and (E) Heart, N = 5 per group. Statistical significance was calculated by multiple t-tests with Holme-Sidak post-test, comparing each treatment group to respective vehicle control; **, P < 0.001.

104 Chapter 5

5.3.4. Effects of P053 on the lipidome of skeletal muscle

After verifying that P053 is bioavailable through oral gavage and is active in vivo, inhibiting C18:0 ceramide synthesis in quadriceps, another two in vivo cohorts were conducted to investigate how P053 administration affects the lipidome of mice when fed with HFD. The physiological glucose regulation studies from the animal cohorts are discussed in the next chapter. Results in this chapter will focus on the lipidomic characterization of P053 in vivo. Male C57BL/6J mice were administered 5 mg/kg of P053 or vehicle daily for 4 or 6 weeks. Lipid extracts were prepared from quadriceps SkM collected at the end point of 4- or 6-weeks and subjected to comprehensive untargeted lipidomic profiling. Using LipidSearch software, 312 and 353 individual species were identified from MS data of 4-week and 6-week SkM samples, respectively.

In the 4-week cohort, of 312 individual lipid species manually verified following lipidomic profiling, only 18:1/18:0 and 18:2/18:0 ceramides were significantly reduced in both chow and HFD groups (Figure 5.5 A-B), at P < 0.01, after adjusting for multiple comparisons (Table A1). Interestingly, 18:1/24:1 ceramide was significantly increased by P053 treatment in the HFD group (Figure 5.5 B). In the 6-week cohort, of 353 species, 18:1/18:0, 18:2/18:0 and 18:1/19:0 ceramides were significantly reduced in both chow and HFD groups (Figure 5.5 C-D, Table A2). Similar to the 4-week cohort, very long chain ceramides (18:1/24:0, 18:0/24:1 and 18:1/24:1) were significantly increased by P053 treatment in the HFD group (Figure 5.5 D).

105 Chapter 5

Figure 5. 5. Lipidome of quadriceps SkM.

Volcano plots showing P value plotted against fold change for each of 312 and 353 lipids in quadriceps SkM of P053-treated relative to vehicle-treated mice. (A-B) are lipidomic data of SkM from 4-week cohort and (C-D) are from 6-week cohort. (A, C) contained data of chow diet groups and (B, D) of HFD groups. N = 8-10 per group. Lipids significantly altered in P053 vs vehicle are shown in red and labelled by name. Three comparisons were made for each lipid measured: chow control vs chow + P053, chow control vs HFD control, and HFD control vs HFD + P053. Data were log transformed, then subjected to two-tailed t-tests adjusted for multiple comparisons using GraphPad Prism (Benjamini, Krieger and Yekutieli correction, Q = 1%). Abbreviation: Cer, ceramide; HFD, high fat diet; SM, sphingomyelin; PG, phosphatidylglycerol.

106 Chapter 5

5.3.5. Effects of P053 on levels of sphingolipids in skeletal muscle

P053 treatment significantly reduced total ceramide content in quadriceps tissues in both cohorts (P < 0.001 by two-way ANOVA; Figure 5.6 A-B). This was entirely attributed to decreased C18 ceramide (Figure 5.7). There was no significant diet or P053 treatment effect on the total GluCer and SM levels in SkM (Figure 5.6). Only 6, not 4, weeks of HFD feeding caused an accumulation of C18:0 ceramide in quadriceps (Figure 5.7 B). High fat feeding also led to significant reduction of very long chain ceramides, which was rescued by P053 treatment in the HFD group (Figure 5.7 A-B).

The ten most abundant SM species were graphed in Figure 5.8. SM species are presented as sum composition, given as an ID by LipidSearch software when only the lipid class, the sum of carbon atoms and total number of double bonds can be identified with high confidence (based on the accurate mass and characteristic fragment ions). None of the SM species was significantly affected by P053 treatment in either chow or HFD groups. Only 38:1 and 40:1 SM were significantly affected by high fat feeding (P < 0.001). Although not significant, a minor decrease in 36:1 SM can be observed in P053-treated groups. Based on the sum composition and abundance pattern, 36:1 SM is likely 18:1/18:0 SM. The very long chain SM species are likely identified as 42:0, 42:1, 42:2 and 42:3 assuming the backbone is 18-carbon in length. To investigate whether the effect of P053 on C18:0 ceramide is also present in SkM other than quadriceps, ceramide and SM profiles of gastrocnemius muscle were examined with targeted MS analysis. As predicted, C18 ceramide species were significantly reduced by P053 treatment in gastrocnemius muscle of mice fed with either chow and HFD for 6 weeks (Figure 5.9). The only SM species significantly affected in gastrocnemius muscle was C18:1 SM, significantly reduced in chow + P053 group. Similar to quadriceps, 6 weeks of high fat feeding significantly increased C18:0 ceramide levels.

Ceramides are formed de novo from the condensation of dihydrosphingosine with a variable length fatty acid. The inhibition of CerS1 did not significantly affect dihydrosphingosine levels in SkM (Figure 5.10 A). However, there was a significant increase in S1P levels in the HFD + P053 group (Figure 5.10 C). Although not significant, there was an increase in sphingosine levels in P053 treated groups especially the HFD group (Figure 5.10 B). Increased sphingosine levels can lead to higher S1P

107 Chapter 5

production due to the availability of substrates for sphingosine kinases, which phosphorylate sphingosine to produce S1P.

Figure 5. 6. Sphingolipids in quadriceps SkM. Total ceramide, GluCer and SM in quadriceps SkM from (A) 4-week cohort and (B) 6-week cohort. N = 10 for each group except Chow + P053, where N = 8 and HFD + P053, where N = 9 of 4-week cohort (A). Graphs show mean ± SEM, total levels of lipids expressed as pmoles/ mg tissue. Results were analysed by two-way ANOVA, with Fisher’s exact post-test to compare individual groups (+, P < 0.01; #, P < 0.001; *, P < 0.0001). Abbreviation: Cer, ceramide; GluCer, glucosyl-ceramide; SM, sphingomyelin.

Figure 5. 7. Ceramide profiles in quadriceps SkM. Levels of major ceramide species in quadriceps SkM of mice fed chow or HFD with vehicle or 5 mg/kg P053 for 4 weeks (A) and 6 weeks (B), expressed as pmoles/ mg tissue. Graphs show the mean ± SEM, N = 8-10 per group. Statistical significance (#, P < 0.01; *, P < 0.001) was determined by two-tailed t-tests adjusted for multiple comparisons (i.e. for all 312 or 353 verified lipid species).

108 Chapter 5

Figure 5. 8. Sphingomyelin profiles in quadriceps SkM.

Levels of major SM species in quadriceps SkM of mice fed chow or HFD, with vehicle or 5 mg/kg P053 for 4 weeks (A) and 6 weeks (B), expressed as pmoles/mg tissue. Graphs show the mean ± SEM, N = 8-10 per group. Statistical significance (*, P < 0.001) was determined by two-tailed t-tests adjusted for multiple comparisons using GraphPad Prism (Benjamini, Krieger and Yekutieli correction, Q = 1%).

109 Chapter 5

Figure 5. 9. Ceramides and sphingomyelin profiles in gastrocnemius SkM.

Levels of major ceramide and SM species in gastrocnemius SkM of mice fed chow or HFD, with vehicle or 5 mg/kg P053 for 6 weeks, expressed as pmoles/mg tissue. Graphs show the mean ± SEM, N = 10 per group. Statistical significance (#, P < 0.01; *, P < 0.001) was determined by two-tailed t-tests adjusted for multiple comparisons using GraphPad Prism (Benjamini, Krieger and Yekutieli correction, Q = 1%).

110 Chapter 5

Figure 5. 10. Dihydrosphingosine, sphingosine and S1P levels in quadriceps SkM.

Levels of dihydrosphingosine (A), sphingosine (B) and S1P (C) in SkM of mice fed chow or HFD with vehicle or 5mg/kg P053 for 4 weeks, expressed as pmoles/mg tissue. Graphs show the mean ± SEM, N = 8-10 per group. Statistical significance (*, P < 0.001) was determined by two-tailed t-tests adjusted for multiple comparisons using GraphPad Prism (Benjamini, Krieger and Yekutieli correction, Q = 1%).

111 Chapter 5

5.3.6. Effects of P053 on levels of phospholipids in skeletal muscle

P053 treatment did not affect levels of common phospholipids from the phosphatidylcholine, phosphatidylethanolamine, phosphatidylserine, and phosphatidylinositol families (Figure 5.11). Only total phosphatidylserine was affected by high fat feeding in the 6-week cohort. Phospholipid species affected by diet are listed in Table A1 and A2.

Figure 5. 11. Phospholipids in quadriceps SkM.

Levels of phospholipid classes in quadriceps SkM (A) 4-week cohort and (B) 6-week cohort. N = 10 for each group except Chow + P053, where N = 8 and HFD + P053, where N = 9, of 4- week cohort (A). Graphs show mean ± SEM, total levels of lipids expressed as nmoles/ mg tissue. Results were analysed by two-way ANOVA, with Fisher’s post-test to compare individual groups (+, P < 0.05; #, P < 0.001). Abbreviation: PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PG, phosphatidylglycerol; PI, phosphatidylinositol.

112 Chapter 5

5.3.7. Effects of P053 on levels of skeletal muscle diacylglycerol and triacylglycerol levels

The total levels of most TG species were reduced by more than 50 % in quadriceps SkM of the HFD + P053 compared to HFD + vehicle group, as seen in volcano plots for both 4- and 6-week cohorts (Figure 5.12 B). High fat feeding for 4 and 6 weeks significantly increased total TG levels in SkM by 65% and 87% respectively when compared to respective chow control group (P < 0.05 by two-way ANOVA). P053 treatment did not affect SkM TG levels in mice fed with normal chow (Figure 5.12 B). Interestingly, DG levels were not significantly affected by diet and P053 treatment in both cohorts (Figure 5.12 A). TGs are synthesized directly from DG. Lipidomic results therefore suggest that P053 does not directly interfere with TG synthesis from DG.

Figure 5. 12. Diacylglycerols and triacylglycerols in quadriceps SkM.

Total DG and TG in SkM of mice fed chow or HFD with vehicle or 5mg/kg P053 for 4 weeks (left panel) and 6 weeks (right panel), N = 8-10 per group. Graphs show mean ± SEM of total (A) DG or (B) TG, expressed as nmoles/ mg tissue. Results were analysed by two-way ANOVA, with Fisher’s post-test to compare individual groups (+, P < 0.05). Abbreviation: DG, diacylglycerol; TG, triacylglycerol.

113 Chapter 5

5.3.8. Effects of P053 on lipidomic profiles in other organs

Of 265 lipids measured in liver of mice treated with P053 fed with chow or HFD for 4 weeks, none was significantly affected by P053 treatment (Table A3). Forty lipid species were significantly affected by diet (Q < 0.01 after adjusting for multiple comparisons; Table A3), including C22:0 ceramide, C24:1 ceramide and 38:1 SM (Figure 5.13). Similar to the liver, P053 did not significantly impact on levels of C18:0 ceramide, nor other ceramide and SM species in epididymal adipose tissues (Figure 5.14). The lack of an effect of P053 on C18:0 ceramide levels in liver and adipose tissue reflects the lack of CerS1 expression in these tissues (Figure 5.1).

Figure 5. 13. Liver ceramide and sphingomyelin levels.

Levels of (A) ceramide and (B) sphingomyelin species in liver of mice fed chow or HFD with vehicle or 5 mg/kg P053 for 4 weeks, expressed as pmoles/ mg tissue. Graphs show the mean ± SEM, N = 10 per group. Statistical significance (#, P < 0.01; *, P < 0.001) was determined by two-tailed t-tests adjusted for multiple comparisons. Abbreviation: SM, sphingomyelin.

114 Chapter 5

Figure 5. 14. Epididymal adipose tissue ceramide and sphingomyelin levels.

Levels of (A) ceramide and (B) SM species in epididymal adipose tissue of mice fed chow or HFD with vehicle or 5 mg/kg P053 for 4 weeks, expressed as pmoles/ mg tissue. Graphs show the mean ± SEM, N = 10 per group. Statistical significance (#, P < 0.01) was determined by two-tailed t-tests adjusted for multiple comparisons.

115 Chapter 5

5.3.9. Correlation between levels of ceramides and TG in skeletal muscle

Intrigued by the reduced TG levels in quadriceps of P053-treated groups, correlations between individual ceramides and TG levels was examined, using Spearman correlation analysis. Separate correlation analysis was carried out for SkM ceramide profiles and total TG from 4-week (Table 5.1) and 6-week (Table 5.2) cohorts. This analysis was performed both including and excluding the mice treated with P053, due to the confounding effect of P053 on muscle ceramides. In both cohorts, SkM very long chain ceramides (C24:0 + C24:1) were inversely correlated with SkM TG, with Spearman coefficients (r) > -0.5 and P < 0.05. Although not statistically significant by two-tailed tests, C18:0 ceramide showed positive correlation (r > 0.3) with TG levels when P053- treated mice were excluded. Figure 5.15 shows representative graphs of ceramide levels expressed as a function of total TG levels in SkM of mice from the 4-week cohort.

Table 5. 1. Correlations between quadriceps SkM ceramides and TG (4-week cohort). Spearman correlation co-efficient (r) and P values are shown. Significant associations are in blue font. Mice from 4-week cohort were used for this analysis (N =38 mice, both chow and HFD group). Both data including and excluding P053-treated mice are shown. P values were derived from two-tailed tests. Including P053-treated mice Excluding P053-treated mice Spearman Spearman Ceramide P value P value coefficient, r coefficient, r Total -0.0174 0.9211 -0.1703 0.4993 16:0 -0.2123 0.2208 0.0623 0.8040 18:0 0.1798 0.3013 0.3354 0.1736 20:0 -0.0702 0.6887 -0.1042 0.6806 22:0 -0.2401 0.1648 -0.1311 0.6042 24:0 -0.4284 0.0103 -0.4530 0.0590 24:1 -0.5464 0.0007 -0.5748 0.0126 24:0 + 24:1 -0.5073 0.0019 -0.5232 0.0259

116 Chapter 5

Figure 5. 15. Ceramide levels as a function of total TG levels in quadriceps SkM.

C18:0, C24:0 and C24:1 ceramide levels in SkM of mice from 4-week cohort were plotted against total TG level in SkM. Data in (A-C) includes P053-treated mice while (D-F) exclude P053-treated mice (A-C, N = 38 mice; D-F, N =18 mice). Spearman analysis was used to assess correlations between ceramide and TG. P values were derived from two-tailed tests. Abbreviation: r, spearman’s coefficient.

117 Chapter 5

Table 5. 2. Correlations between quadriceps SkM ceramides and TG (6-week cohort). Spearman correlation co-efficient (r) and P values are shown. Significant associations are in blue font. Mice from 6-week cohort were used for this analysis (N =40 mice, both chow and HFD group). Both data including and excluding P053-treated mice are shown. P values were derived from two-tailed tests.

Including P053-treated mice Excluding P053-treated mice Ceramide Spearman P value Spearman P value coefficient, r coefficient, r Total -0.3321 0.8388 0.1519 0.5227 16:0 0.0885 0.5871 -0.0994 0.6767 18:0 0.0169 0.9176 0.3865 0.0923 20:0 -0.1452 0.3713 -0.1171 0.6229 22:0 0.0192 0.9062 -0.286 0.2215 24:0 -0.2518 0.1170 -0.4902 0.0282 24:1 -0.0694 0.6703 -0.3504 0.1299 24:0 + 24:1 -0.2325 0.1488 -0.5346 0.0152

118 Chapter 5

5.3.10. Effects of P053 on CerS2 activity

Following the discovery of negative correlations between very long chain ceramides and TG levels, the next question is whether there is a change in CerS2 activity with P053 treatment that led to increased very long chain ceramides. C24 CerS activity of SkM from mice treated for 4 weeks with P053 or vehicle control was not significantly affected by P053 treatment (Figure 5.16). Therefore, the elevated levels of very long chain ceramides were not directly associated with C24 CerS activity. The very long chain ceramides upregulation may be a consequence of increased sphingoid base substrate availability due to the reduction in CerS1 activity.

Figure 5. 16. C24 ceramide synthase activity in quadriceps SkM of P053 treated mice.

(A) C24 CerS activity of SkM from mice treated for 4 weeks with P053 or vehicle control. N = 10 for all groups except chow + P053 group, where n = 8. Two-way ANOVA revealed no significant effect of diet (P = 0.1132) or P053 treatment (P = 0.4931) on C24 CerS activity.

119 Chapter 5

5.3.11. P053 does not affect brain ceramide and SM profiles

Data so far suggests that P053 is effective in tissues where CerS1 is highly expressed. Although CerS1 is highly expressed in the brain (Figure 5.1), P053 did not affect C18:0 ceramide and sphingomyelin levels in cerebellum tissues (Figure 5.17). This may be due to the lack of efficiency of P053 uptake into the brain.

Figure 5. 17. P053 does not affect brain ceramide and sphingomyelin levels.

(A) Ceramide and (B) SM levels in the cerebellum of vehicle- or P053-treated mice on a chow or HFD for 4 weeks. N = 10 mice per group, except for chow vehicle group (N = 9). Mean ± SEM shown. No statistical significance was determined by multiple t-tests using GraphPad Prism (Benjamini, Krieger and Yekutieli correction, Q = 1%) comparing each treatment group to respective vehicle control. Abbreviation: SM, sphingomyelin.

120 Chapter 5

5.3.12. Uptake of P053

LC-MS/MS analysis revealed that P053 uptake into the brain was 4-fold and 6-fold lower than the quadriceps and gastrocnemius muscles respectively, of mice administered 5 mg/kg P053 for two weeks (Figure 5.18 A). Whether this is due to low uptake of P053 across the blood brain barrier is unclear, but it indicates that treatment with P053 is unlikely to cause neurological issues previously reported in CerS1 knockout mice. Uptake of P053 was also considerably lower in both epididymal and inguinal adipose tissues. P053 concentration was the highest in liver and kidney (4-5 fold higher than SkM), the two main organs for drug clearance. P053 appeared to largely remain in circulation, with an average of 289 pmoles/mL in the plasma (Figure 5.18 B).

Figure 5. 18. P053 uptake in tissues and plasma.

P053 was quantified, using LC-MS/MS, in various tissues (A) and plasma (B) of mice administered 5 mg/kg P053 daily for 2 weeks (N = 5 mice). No P053 was detected in mice administered vehicle only. Abbreviations: EpiWAT, epididymal white adipose tissue; IngWAT, inguinal white adipose tissue; Quads, quadriceps muscles; Gastroc, gastrocnemius muscles.

121 Chapter 5

5.4. Discussion

The previous chapter explored the potency and efficacy of P053 in vitro. Results suggest that P053 is a selective and potent inhibitor of CerS1 activity. This chapter investigated P053 in in vivo mouse models with implications for insulin resistance. C18:0 ceramide, as the most abundant form of ceramide in SkM, has been implicated in SkM insulin resistance78,82. Elevated C18:0 ceramide has been correlated to impairments in insulin action in studies comparing obese insulin resistant and insulin sensitive subjects, exercise interventions in T2D and induction of insulin resistance in mice48,236,317.

With the intent to study C18:0 ceramide and CerS1 in SkM insulin resistance, this study involved mice fed a HFD that is common in the field. The HFD mouse studies commenced after verifying that P053 is bioavailable in vivo (Figure 5.2), reaching a maximal plasma concentration of 20 nM 4 h after administration of a single 5 mg/kg P053 dose via oral gavage. A pilot study involving 15 mice administered vehicle, 1 or 5 mg/kg P053 for one week indicated that at 1 mg/kg P053 was ineffective (Figure 5.3). At 5 mg/kg, P053 reduced C18:0 ceramide content in quadriceps SkM (Figure 5.4) but not in brain, where CerS1 is also highly expressed (Figure 5.1). Given that CerS1 inhibition was not apparent at 1 mg/kg of P053, the mice in subsequent studies were dosed with 5 mg/kg of P053.

CerS1 is highly expressed in the SkM, comparable to expression levels in the brain. Other organs such as liver and adipose tissue, relevant for metabolic studies, have little CerS1 expression (Figure 5.1). Since P053 is highly potent and selective for CerS1, the possibility of it targeting SkM is high, provided that the brain will not be affected. CerS1 global knockout mice develop cerebellar degeneration and impairment in neurodevelopmentally regulated behaviour96,97. Previous studies with FTY720 treatment at 5 mg/kg were able to reduce SkM ceramide content and improve glucose tolerance in HFD fed mice276. However, FTY720 was not selective, reducing levels of several ceramide species at the same time. This study set out to investigate whether selective inhibition of CerS1 in SkM via P053 treatment has a beneficial effect on SkM insulin resistance. This chapter focuses on the lipidome of SkM and other organ tissues. The physiological data are discussed in Chapter 6. 122 Chapter 5

Untargeted lipidomic profiling was performed on SkM tissues collected at the end of 4- week and 6-week animal studies. Lipidomic results indicate that P053 is highly selective in reducing C18:0 ceramide in quadriceps regardless of diet (Figure 5.7). P053 treatment did not affect levels of other sphingolipids, including those with C18:0 fatty acyl chains, nor common phospholipids from the phosphatidylcholine, phosphatidylethanolamine, phosphatidylserine, and phosphatidylinositol families. Interestingly, 4 weeks of HFD feeding did not induce an accumulation of C18:0 ceramide when compared to chow vehicle control group but 6 weeks of HFD feeding did (P < 0.001). The duration of high fat feeding may be important in causing accumulation of ceramide. To promote accumulation of ceramide comparable to that of obese insulin-resistant human subjects, a longer duration of high fat feeding may be needed. Furthermore, this finding matches other reports in which C18:0 ceramide accumulation is not always present with high fat feeding in both animal models and humans25,325,326. There was also a lack of lipotoxic DG accumulation in our data (Figure 5.12A) although HFD feeding clearly increased TG content in SkM (Figure 5.12B). One possible explanation for the lack of ceramide or DG accumulation compared to other literature may be the choice of HFD. In particular, it has been reported that ceramide accumulation in SkM is dependent on the composition of HFD. For instance, a HFD rich in oleate or linoleate unsaturated FA lead to increased DG content in SkM but not ceramide31. Diets rich in palmitate on the other hand lead to accumulation of ceramide in SkM28,231. In addition, some of the studies in the field use a HFD with 60% calories coming from fat while the HFD in our studies was modest with only 45% calories coming from fat108,224.

With a prominent effect on SkM C18:0 ceramide levels, one would expect the sphingolipid species derived from C18:0 ceramide to be affected by P053 treatment. Surprisingly, P053 treatment had no significant effect on major GluCer and SM species in SkM (Table A1 and A2). Despite the lack of effect on GluCer and SM species, P053 treatment appeared to increase S1P levels in SkM of mice treated with P053 in both diet groups (Figure 5.10). One likely explanation is that by inhibiting ceramide synthesis with P053, there was a modest increase in sphingosine or dihydrosphingosine substrates. This contributes to a much more pronounced increase in S1P levels as the sphingosine is rapidly phosphorylated. S1P is around 10-fold lower in abundance than

123 Chapter 5

sphingosine. Hence, a small increase in sphingosine is sufficient to noticeably increase S1P levels. S1P is a well-known bioactive lipid implicated in biological processes such as proliferation, differentiation and survival69. In SkM, S1P has been implicated as a mediator of SkM regeneration acting through distinct S1P receptors and is regulated by multiple growth factors and cytokines327,328. S1P has also been implicated in SkM insulin resistance329. However, its role in SkM insulin signalling seems to be divergent. Several studies indicate that increased activity of SPHK1 and the formation of S1P contributes to the development of insulin resistance while SkM cell culture studies have demonstrated that S1P can increase glucose uptake through trans-activation of insulin receptor328,330,331. Increased S1P levels in SkM in our current study could be mechanistically important and should be investigated in future studies.

Lipidomic profiling data from this chapter showed that a HFD can have different effects on the ceramide profiles of SkM, liver and adipose tissues. The differences are likely due to the expression pattern of CerS isoforms. P053 treatment did not reduce C18:0 ceramide levels in liver and WAT despite the uptake of P053 into these tissues. This reflects the lack of CerS1 expression in the liver and WAT. It is possible that CerS4 is the major contributor to the C18 ceramide pool in these tissues as CerS4 can synthesize C18:0 ceramide as well80. Similar to the pilot study, P053 treatment did not affect C18:0 ceramide levels in cerebellum of mice administered the drug for 4 weeks (Figure 5.17). This is likely a consequence of 4-fold lower levels of the compound in brain tissue compared to quadriceps (Figure 5.18A). P053 uptake was the highest in kidney, followed by liver, gastrocnemius and quadriceps. The high concentration of P053 in liver and kidney may be related to hepatic and renal clearance of the drug. LC-MS/MS analysis showed that most administered P053 was retained in plasma. Uptake of P053 was very low into the WAT, which reflects the lack of effect on their ceramide profiles.

With inhibition of CerS1 and reduction of SkM C18:0 ceramide, there was an increase in very long chain ceramides, particularly C24:0 and C24:1. The reduction of very long chain ceramides in HFD groups was consistent in SkM, liver and WAT in this study. The reduction of C24:1 ceramide is in line with previous findings24. Very long chain ceramides are synthesized by CerS2. However, the increased very long chain ceramides in SkM tissues in this study were not associated with increased C24 ceramide synthase

124 Chapter 5

activity (Figure 5.16). This is most likely due to CerS1 inhibition freeing up sphingoid base substrates for other CerSs, such as CerS2, resulting in increased very long chain ceramides. This is in agreement with the lack of a notable increase in dihydrosphingosine levels.

Recently, the C18:0 ceramide content of SkM was positively correlated with visceral fat mass in humans48. Also, the nuclear Drosophila CerS named Schlank has been reported to regulate fat metabolism through ceramide synthesis95. Interestingly, FTY720 treatment has been reported to prevent TG accumulation in SkM of mice fed a HFD276. These studies together suggest that ceramide may play a role in fat metabolism apart from insulin resistance pathology. Data from the current study suggests similar findings. P053 treatment significantly reduced total TG content in HFD fed mice compared to the vehicle control group. In our study, C24 ceramides were inversely correlated with TG content in the SkM (Table 5.1 and Table 5.2). DG levels were unaffected by P053 treatment, suggesting that P053 does not directly inhibit TG synthesis from DG. These correlations suggest a potential relationship between ceramide and TG content and warrant further studies. The relationship between ceramide content and fat metabolism is explored in the next chapter.

Overall, results from this chapter demonstrate that P053 is effective in reducing C18:0 ceramides in SkM. C18:0 ceramide levels were significantly reduced with P053 treatment in quadriceps SkM tissues. Future investigations are warranted to explore how P053 is internalized and taken up into SkM and other organs, metabolized or excreted, and whether the animals will develop tolerance towards the drug over time. In addition, further studies should be carried out to investigate whether P053 may affect RNA and protein expression of CerS isoforms in various organs and how this may relate to the lipidome.

125 Chapter 6

Chapter 6: Physiological effects of CerS1 inhibition by P053 on glucose and fat metabolism in vivo

6.1. Introduction

The previous chapter investigated the potency of P053 in vivo and its effect on the lipidome of SkM. P053 is bioavailable and successfully inhibited C18:0 ceramide production in SkM of mice fed with either chow or HFD. The effect is only apparent in SkM but not in liver and WAT. Reduction of C18:0 ceramide levels was accompanied by complementary increased in very long chain ceramide species and loss of SkM TG content.

Ceramide, a lipotoxic agent is now recognized as one of the significant mediators in obesity-induced metabolic pathologies such as whole-body and tissue-specific insulin resistance. As discussed in Chapter 1, C16:0 ceramide has been identified as the ceramide species that mediates the pathophysiology of insulin resistance in liver and adipose tissues58,59,101,108. Part of the lipotoxic effects of C16:0 ceramide was due to inhibition of FA b-oxidation in liver and adipose tissue58,59. As for SkM insulin resistance, the important ceramide species at play is C18:0 ceramide. SkM is the major organ for energy homeostasis as it is heavily involved in insulin-stimulated glucose uptake, usage, disposal and storage. Therefore, SkM is of particular interest for studies in insulin resistance. High fat feeding has been shown to result in increased CerS1 expression and accumulation of C18:0 ceramide in SkM of mice and is linked to reduced glucose tolerance and impaired insulin action48,236,317. These observations in animal models appear to be clinically relevant to humans. Various human studies involving obese insulin-resistant and insulin-sensitive human subjects, and exercise interventions in T2D subjects also demonstrated that C18:0 ceramide was associated with inhibition of insulin signalling48,236. SkM C18:0 ceramide is also positively correlated with visceral fat, systolic and diastolic blood pressure but inversely correlated

126 Chapter 6

with insulin-sensitizing adiponectin48. Therefore, lowering C18:0 ceramide in SkM may protect against fatty diet-induced metabolic dysfunction.

Currently, there is a lack of isoform-selective CerS inhibitor. Therefore, the most commonly used pharmacological means for reducing ceramide content is myriocin, which inhibits SPT at the first step of the sphingolipid synthesis pathway. Administration of myriocin to Zucker fatty rats, ob/ob mice and HFD-induced obese mice successfully reduced SkM ceramide content and produced significant improvements in glucose and insulin tolerance28,222,231. However, due to the immunosuppressive properties of myriocin252, its potential as an intervention for treating humans is questionable. Administration of FTY720 has also been shown to prevent ceramide accumulation in SkM and improve homeostasis demonstrated by an overall improvement in whole-body glucose tolerance, increased insulin-stimulated glucose uptake in muscle and a reduction in plasma insulin276. Similar to myriocin, FTY720 is immunosuppressive and can induce lymphopenia272, thus, it may not be an ideal treatment for lowering ceramide content in human. In this regard, P053 has shown potency in selective inhibition of CerS1 and lowering C18:0 ceramide levels in SkM in mice, as presented in Chapter 5. Work in this chapter seek to investigate whether CerS1 inhibition in SkM by P053 can provide protection against fat-induced metabolic dysfunction.

6.2. Aims

Previously shown by Turner et al.24, 4-6 weeks of high fat feeding is sufficient to induce whole-body and muscle insulin resistance. Work in this chapter sought to investigate the effect P053 treatment on the metabolic physiology of mice fed a HFD for 4-6 weeks.

The specific aims for this chapter are: 1. To determine if P053 treatment has a beneficial effect on glucose tolerance and insulin action in mice fed a HFD. 2. To determine if P053 prevents weight gain in mice fed a HFD. 3. To determine if P053 treatment has a beneficial effect on fatty acid metabolism and mitochondrial function.

127 Chapter 6

6.3. Results

6.3.1. Physiological effects of HFD and P053 treatment in vivo

Four weeks of high-fat feeding resulted in a significant increase in body mass (Figure 6.1A). High-fat feeding also increased the weight of epididymal, inguinal and brown adipose tissues compared to animals fed a chow diet (Figure 6.1 B-D) but had no effect on quadriceps, gastrocnemius and liver mass (Figure 6.1 E-G). EchoMRI measurements of body composition revealed a significant diet effect on fat mass and lean mass as a percentage of body mass (Figure 6.1 H-I). Interestingly, P053 treatment significantly reduced whole body fat mass and increased lean mass of animals fed a HFD. P053 treatment also significantly reduced the mass of individual adipose deposits in mice fed a HFD. This was not due to any effect of P053 on food intake (Figure 6.2). P053 treatment for two weeks also had no significant effect on red and white blood cell numbers, blood platelets, haemoglobin and haematocrit (Table 6.1), indicating that P053 does not cause lymphopenia like its precursor, FTY720.

128 Chapter 6

Two-way ANOVA statistical significance Body Fat Lean EpiWAT IngWAT BAT Gastroc Quad Liver mass mass mass (g) (g) (g) (g) (g) (g) (g) (%) (%) Diet **** *** **** ** ns ns ns **** *** Treat ns ** * ns ns ns ns * * ment

Figure 6. 1. Body composition and adiposity of animals. (A) Total body mass, (B) epididymal WAT (EpiWAT) mass, (C) inguinal WAT (IngWAT) mass, (D) brown adipose tissue (BAT) mass, (E) gastrocnemius (gastroc) mass, (F) quadriceps (quad) mass, (G) liver mass, (H) fat mass as a % of body mass, (I) lean mass as a % of body mass, in mice fed chow or HFD with vehicle or 5 mg/kg P053 for 4 weeks (N = 10 per group, mean ± SEM). ANOVA results are shown in the results table. Statistical significance was assessed by two-way ANOVA with diet as one variable and treatment (P053 or vehicle) as the other. P values for the main effect of diet (i.e. chow vs HFD) and treatment (i.e. vehicle vs P053) for each measure are reported in the table (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant).

129 Chapter 6

150 Vehicle P053 100

50 /cage/ week Food intake (g) 0 Chow HFD

Figure 6. 2. Food intake is not affected by P053 treatment.

Weekly food intake per cage of 4 mice, monitored on 5-6 independent occasions over the course of the P053 treatment period. Two-way ANOVA revealed a significant effect of diet (P < 0.0001) but not P053 treatment (P = 0.1483) on food intake.

Table 6. 1. P053 on blood cell parameters. Effect of P053 treatment on white blood cells, red blood cells, blood haemoglobin, haematocrit, and platelets of mice treated with either vehicle control or 5mg/kg P053 for two weeks. Values are mean ± SD of N = 5 animals per group. No statistical significance was determined by unpaired t-tests.

Chow Vehicle Chow P053 9 White blood cell (x10 /L) 3.620 ± 1.204 3.990 ± 1.517 12 Red blood cell (x10 /L) 8.709 ± 0.551 8.855 ± 0.209 Haemoglobin (g/L) 132.8 ± 7.193 134.6 ± 1.350 Haematocrit (L/L) 0.427 ± 0.023 0.432 ± 0.008 9 Platelets (x10 /L) 865.6 ± 109.0 808.2 ± 228.141

6.3.2. Effects of HFD and P053 treatment on glucose tolerance in vivo

There is a close association between excess adiposity and insulin resistance332. As expected, HFD fed animals displayed elevated fasting insulin (Figure 6.3A) and became glucose intolerant due to defects in glucose disposal (Figure 6.3B-C). Previous studies have shown that inhibition of ceramide synthesis can have beneficial effects on glucose tolerance276. However, the reduction of SkM C18:0 ceramide and overall adiposity through P053 treatment had no impact on glucose tolerance.

130 Chapter 6

Figure 6. 3. Effect of HFD and P053 treatment on glucose metabolism.

(A) Fasting blood insulin and (B) blood glucose levels following an oral glucose load in mice fed chow or HFD with vehicle or 5 mg/kg P053 for 6 weeks (N = 10 per group, mean ± SEM). (C) Incremental area under the curve (iAUC) from the oGTT shown in (B). P-values derived from two-way ANOVA for each measure are shown in the table (**, P < 0.01; ****, P < 0.0001; ns, not significant).

6.3.3. Effects of HFD and P053 on whole body insulin action in vivo

To further assess the effect of P053 on whole body insulin action, mice fed a normal chow or HFD, and treated with vehicle control or P053 were subjected to hyperinsulinemic-euglycemic clamps.

The effects of HFD and P053 treatment on the whole body-parameters of the mice that underwent hyperinsulinemic-euglycemic clamps can be seen in Table 6.2. Due to poor recoveries from dual cannulation surgeries, several HFD fed animals were either euthanized or did not survive clamping procedures. Hence, only 12 HFD-fed animals (6 per treatment group) were successfully clamped while there were 17 chow-fed animals successfully clamped from this cohort. Similar to the previous cohort, high fat feeding

131 Chapter 6

for 6 weeks significantly increased the body mass and fasting insulin levels of mice. During the clamp procedure, insulin levels were significantly elevated in all groups compared to their respective basal insulin levels. Plasma glucose was clamped within 7.5- 8.5 mM (Table 6.2). While there was suppression of plasma NEFAs under clamp conditions, there was no significant diet or P053 treatment difference.

During the hyperinsulinemic-euglycemic clamp, HFD-fed mice had a reduced glucose infusion rate (GIR), indicative of impairment in whole body insulin action (Figure 6.4

A). The rate of whole body glucose disappearance, Rd was also compromised in HFD- fed mice compared to chow-fed mice (Figure 6.4 B). Collectively, these results clearly demonstrated that whole body system insulin resistance occurred in response to 6 weeks of high fat feeding. However, P053 treatment showed no effect in improving whole body insulin resistance.

Table 6. 2. Characteristics of mice undergoing hyperinsulinemic-euglycemic clamp. Characteristics of mice undergoing hyperinsulinemic-euglycemic clamp. Mice were fed chow or HFD with vehicle or 5 mg/kg P053 for 6 weeks. Data are mean ± SEM. A significant effect of diet was determined by two-way ANOVA. P values are indicated as *, P < 0.05 and **, P < 0.01 in the table. Diet Chow HFD Treatment Vehicle P053 Vehicle P053 N 9 8 6 6 25.94 ± 26.58 ± 29.42 ± 28.18 ± Body mass 0.532 0.428 1.295** 1.108 Fasting plasma 13.88 ± 16.62 ± 22.31 ± 18.79 ± insulin (mU/L) 1.577 2.075 3.706 * 2.620 Plasma insulin 93.62 ± 90.00 ± 78.552 ± 93.07 ± during clamp 12.950 9.465 8.117 10.219 (mU/L) Basal plasma 8.64 ± 7.36 ± 8.57 ± 9.73 ± glucose (mM) 0.435 0.420 0.353 0.336 Plasma glucose 8.41 ± 7.59 ± 7.64 ± 8.24 ± during clamp (mM) 0.203 0.181 0.329 0.121 1.18 ± 1.47 ± 1.18 ± 1.17 ± Basal NEFA (mM) 0.069 0.101 0.142 0.235 NEFA during 0.50 ± 0.48 ± 0.57 ± 0.58 ± clamp (mM) 0.054 0.092 0.076 0.115

132 Chapter 6

Figure 6. 4. P053 did not prevent systemic insulin resistance in HFD-fed mice.

(A) Glucose infusion rate (GIR) and (B) whole-body glucose disappearance (Rd) were significantly lowered in mice fed a HFD, treated with either vehicle control or P053 for 6 weeks (N= 6-9 per group). Significant diet effects assessed by two-way ANOVA are indicated under the figures (*, P < 0.05; ***, P < 0.001).

6.3.4. Effects of HFD and P053 on muscle insulin resistance

To determine if P053 treatment altered muscle insulin action, glucose uptake into both quadriceps and soleus muscles was examined. Under clamp conditions, glucose uptake into quadriceps and soleus muscles of HFD-fed mice was reduced by 35 % and 39 %, respectively, compared to chow-fed mice (Figure 6.5 A-B). No significant effect of P053 on glucose uptake into SkM was observed. Next, we examined in vivo glucose uptake into epididymal and inguinal WAT, where the loss of their gross weights was observed previously (Figure 6.1). The uptake of glucose into epididymal and inguinal WAT of HFD fed mice was reduced by 68 % and 63 %, respectively, compared to chow-fed mice (Figure 6.5 C-D). Similar to the muscles, P053 treatment did not have an effect on the uptake of glucose into adipose tissues.

133 Chapter 6

Quad Soleus EpiWAT IngWAT Diet * ** *** *** Treatment ns ns ns ns Figure 6. 5. P053 did not prevent the reduction in glucose uptake into muscle and adipose tissues caused by a HFD. Glucose uptake into (A) quadriceps muscle, (B) soleus muscle, (C) epididymal white adipose, and (D) inguinal white adipose tissues of mice fed a chow or HFD with vehicle control or P053 treatment for 6 weeks (N= 6-9 per group). Statistical significance was determined by two-way ANOVA, with results shown in tables shown at the bottom of the figure: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant.

134 Chapter 6

6.3.5. Effects of P053 on muscle lipid oxidative capacity

Although P053 had no notable effect on whole-body and SkM insulin resistance, the effects of P053 on SkM TG and body fat mass were significant. Previous studies have reported that the lipotoxic effects of C16:0 ceramides are partly due to inhibition of FA β-oxidation in the liver and adipose tissues59,101. In this regard, the SkM FA oxidative capacity was investigated. Muscle homogenates from chow and HFD-fed mice treated with P053 displayed increased oxidation of 14C-palmitate (Figure 6.6A), suggesting that enhanced lipid oxidation may underlie the reduced lipid accretion in response to P053.

To differentiate between inhibition of TG synthesis and enhanced lipid oxidation, the 14 TG synthesis from C-palmitate was quantified, as well as production of CO2 and acid soluble metabolites (i.e. β-oxidation intermediates), in soleus muscle strips from mice treated for 2 weeks with 5 mg/kg P053. There was no change in the incorporation of 14C-palmitate into 14C-TG during the 1 h incubation (Figure 6.6B). However, the oxidation of 14C-palmitate was significantly increased in soleus muscles from P053- treated mice (Figure 6.6C, P = 0.02).

To further investigate whether the above results were due to an acute effect of P053, isolated soleus muscle strips were treated in vitro for 1.5 h with 1 µM P053 prior to incubation with 14C-palmitate substrate. Acute treatment with P053 did not affect palmitate oxidation in soleus muscle (Figure 6.6D), suggesting that prolonged CerS1 inhibition is likely priming SkM to oxidise FA. To assess the acute effect of the compound in non-CerS1 dominant tissue type, primary mouse hepatocytes were isolated, cultured and treated with 0.5 µM P053 for 16 h, with or without FA treatment, prior to palmitate oxidation assay. P053 treatment did not affect palmitate oxidation in primary hepatocytes (Figure 6.6E).

135 Chapter 6

Figure 6. 6. Effect of P053 on SkM palmitate oxidative capacity. (A) 14C-palmitate oxidation in SkM homogenates from mice treated for 6 weeks with P053 or vehicle (N = 10 mice), with significant P053 treatment effect only (P = 0.01, measured by two- way ANOVA). (B) 14C-palmitate incorporation into TG, and (C) 14C-palmitate oxidation in isolated soleus muscle strips taken from mice treated for two weeks with 5 mg/kg P053 or vehicle control (N = 5 mice). (D) 14C-palmitate oxidation in soleus muscle strips pre-treated for 1.5 h in vitro with 1 µM P053 or vehicle control (N = 5 mice). Statistical significance in B-D was determined by two-tailed t-test (results where P < 0.05 are indicated on the graph). (E) 14C- palmitate oxidation in primary hepatocytes treated with 0.5 µM P053 or vehicle control, with or without FA treatment for 16 hours (N = 3, treatments were prepared in duplicate wells and taken as average). No statistical significance was determined by two-way ANOVA for measures in (E).

136 Chapter 6

6.3.6. Effect of P053 on mitochondrial markers

β-oxidation occurs in the mitochondria. It has been reported that ceramides can directly influence mitochondrial function, as well as mitochondrial morphology and turnover59,161,181. In this regard, the current study demonstrated significant increases in a number of mitochondrial markers, including elevated protein levels of respiratory chain complex proteins in SkM of P053-treated mice (Figure 6.7A). All respiratory chain complex protein levels were significantly increased except for Complex V. Protein levels for complex I, III and V subunits were significantly reduced with a HFD. The effects of P053 on mitochondrial complex markers was specific to SkM, as there was no increase in any of the complexes in the liver (Figure 6.7B). This result reflected the results from previous chapter in which P053 has no effect on lipid levels in the liver, where CerS1 expression and C18:0 ceramides are low. In addition, mRNA expression of respiratory complex subunits in SkM of P053 treatment mice was assessed. The subunits of complex assessed included NADH dehydrogenase 1b subcomplex 5 (NDUFB5; subunit of complex I), cytochrome B (CytB; subunit of complex III), somatic cytochrome c (CYCS; electron carrier between complex III and IV), cytochrome c oxidase subunit I and II (CoxI and CoxII; subunits of complex IV), cytochrome c oxidase subunit 5b (Cox5b; subunit of complex IV), ATP synthase subunit O (ATP5o; subunit of complex V), and mitochondrial-encoded ATP synthase 6 (ATP6; subunit of complex V). All of the above subunits were significantly increased with P053 treatment except CoxI (Figure 6.8).

In addition to increased markers of mitochondrial complexes, there was an increased activity of citrate synthase (CS) and β-hydroxyacyl coenzyme A dehydrogenase (βHAD) (P = 0.06) was observed in SkM homogenates but not in liver homogenates of P053-treated mice (Figure 6.9A-D). The activity of CS is a common biomarker for mitochondrial content and muscle oxidative capacity. In agreement with mitochondrial markers measured in the current study, the maximal activity of respiratory complexes I and IV was significantly increased in permeabilized EDL muscle fibres of mice treated with P053 for 3 weeks (Figure 6.9 E).

137 Chapter 6

Two-way ANOVA statistical significance, SkM Complex I Complex II Complex III Complex IV Complex V Diet *** ns *** ns ** Treatment ** *** * * ns

Two-way ANOVA statistical significance, Liver Complex I Complex II Complex III Complex IV Complex V Diet ns * ns ns ns Treatment ns ns ns ns *

Figure 6. 7. P053 increases respiratory complex subunit levels in SkM but not liver.

(A) Western blots of respiratory complex subunits in SkM and (B) liver homogenates of vehicle- or P053-treated mice on chow or HFD diet. Quantified densitometry is shown beside the representative blots (N = 10 mice for all groups except the chow + P053 group in SkM only, for which N = 8). Statistical significance was determined by two-way ANOVA, with results shown in tables shown at the bottom of the figure: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant.

138 Chapter 6

Two-way ANOVA statistical significance

NDUFB5 CytB CYCS CoxI CoxII Cox5b ATP5o ATP6

Diet ns ns ns ns ns ns ns ns Treatment *** * * ns * * * **

Figure 6. 8. P053 increases mRNA levels for respiratory complex subunits in SkM.

Relative mRNA expression of respiratory complex subunits was determined by qPCR, in SkM of mice fed chow or HFD with vehicle or 5 mg/kg P053 for 6 weeks (N = 8 for chow groups and N= 10 for HFD groups; mean ± SEM). Levels are normalised to 36B4 and Rpl13 housekeeping genes. Statistical significance was determined by two-way ANOVA, with results shown in tables shown at the bottom of the figure: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. Abbreviation: NDUFB5, NADH dehydrogenase 1b subcomplex 5; CytB, cytochrome B; CYCS, somatic cytochrome c; CoxI, cytochrome c oxidase subunit I; CoxII, cytochrome c oxidase subunit II; Cox5b, cytochrome c oxidase subunit 5b; ATP5o, ATP synthase subunit O; ATP6, mitochondrial-encoded ATP synthase 6.

139 Chapter 6

Two-way ANOVA statistical significance CS βHAD Tissue SkM Liver SkM Liver Diet * * * * Treatment ** ns ns ns

Figure 6. 9. P053 on oxidative enzyme activities in SkM and liver of mice.

Citrate synthase (CS) and β-hydroxyacyl coenzyme A dehydrogenase (βHAD) activity in SkM (A and B) and liver (C and D) homogenates (N = 9 - 10 mice per group). Statistical significance was determined by two-way ANOVA, with results shown in tables shown at the bottom of the figure: *, P < 0.05; **, P < 0.01; ns, not significant. (E) Respiratory activities of mitochondrial complexes I (CI), II (CII) and IV (CIV) measured in isolated permeabilized EDL muscle fibres of mice treated with vehicle or 5 mg/kg P053 for 3 weeks. The respiration for each complex

(nmol O2/min/mg tissue) is normalized to the respective vehicle control sample for each substrate. Statistical significance in E was determined by two-tailed t-test (results where P <

0.05 are indicated on the graph as * for CI and CIV).

140 Chapter 6

6.3.7. Correlations between muscle ceramides and body adiposity

Intrigued by the overall reduced body adiposity with P053 treatment in HFD fed mice, correlations between SkM ceramide content and body adiposity were examined, using Spearman correlation analysis. Only the mice that had not been treated with P053 were used for this analysis due to the pronounced effect of P053 on SkM ceramide content. Correlation analysis indicated that C18:0 ceramide levels in the SkM were positively correlated with body fat percentage (%) and WAT mass (Table 6.3 and Figure 6.10). Interestingly, C24:0 and C24:1 ceramides were inversely correlated with body fat. These correlations suggest a potential relationship between ceramide content of SkM and whole-body adiposity.

Table 6. 3. Correlations between SkM ceramides and body adiposity. Spearman correlation co-efficients (r) and P values are shown. Significant associations are in blue font. Body fat % as determined by EchoMRI, whilst WAT weight is the combined weight of inguinal and epididymal adipose pads. Mice from both chow and HFD groups were used for this analysis (N = 38 mice). Mice treated with P053 were not included in this analysis, due to the confounding effect of P053 on muscle ceramides.

Body fat (%) WAT weight Ceramide r P r P Total 0.010 0.54 0.138 0.40 18:1/16:0 -0.156 0.34 -0.164 0.31 18:1/18:0 0.367 0.020 0.393 0.012 18:1/20:0 -0.197 0.22 -0.030 0.85 18:1/22:0 0.209 0.196 0.267 0.096 18:1/24:0 -0.524 0.0005 -0.464 0.0025 18:1/24:1 -0.5353 0.0004 -0.428 0.0058

141 Chapter 6

Figure 6. 10. SkM ceramide levels as a function of body fat in mice.

C18:0, C24:0 and C24:1 ceramide levels in SkM of mice were plotted against body fat percentage (%) as determined by EchoMRI (A-C) and the combined weight of inguinal and epididymal adipose pads (WAT weight; D-F). Mice from both chow and HFD groups were used for this analysis (N = 38 mice). Mice treated with P053 were not included in this analysis, due to the confounding effect of P053 on muscle ceramides. P values were derived from two-tailed tests. Abbreviation: r, spearman’s coefficient.

142 Chapter 6

6.4. Discussion

Previous studies have implicated C18:0 ceramide in muscle as a driver of muscle insulin resistance and poor glucose disposal from blood. Work from this chapter indicated that selective inhibition of CerS1 and reduction of C18:0 ceramide levels in SkM did not prevent HFD-induced glucose intolerance and insulin resistance but significantly reduced fat deposition in mice fed a HFD. Results from this chapter also highlight a new role for CerS1 in the regulation of mitochondrial oxidative capacity. Chronic CerS1 inhibition with P053 treatment primed SkM to metabolise palmitate.

In the current study, high fat feeding was sufficient to induce a two-fold increase in percentage body fat in mice. In rodent models, glucose intolerance and insulin resistance pathologies usually develop within 3-4 weeks of high fat feeding24. In this regard, mice fed a HFD in the current study were glucose intolerant and had elevated fasting blood insulin levels. Hyperinsulinemic-euglycemic clamp results showed that whole body insulin resistance was evident in the high fat fed animals. This condition was not affected by P053 treatment. HFD also significantly affected glucose uptake into muscle and adipose tissues. Although P053 treatment appeared to slightly improve glucose uptake into the muscle, this effect was not statistically significant. As described in the literature, blocking total sphingolipid synthesis with myriocin protects against insulin resistance and hepatic steatosis in rodents fed a HFD196,224,309, as does genetic ablation of Cers5 and CerS6 in liver and adipose tissue58,108. Previous literatures have also reported improved muscle glucose uptake in mice fed a HFD treated with FTY720 or overexpressing SPHK1276,333. In these reports, the total muscle ceramide levels were reduced including C18:0 ceramide and other long chain to very long chain species. However, CerS1 inhibition with P053, which was shown in the previous chapter to cause a large reduction in C18 ceramide levels in SkM, did not prevent glucose intolerance, whole-body insulin resistance and muscle glucose uptake. Thus, it seems that other ceramide species synthesized by CerS2, CerS5 and CerS6 may play a much more significant role than C18:0 ceramide synthesized by CerS1 in the regulation of whole-body glucose homeostasis and insulin action. Therefore, the associations between SkM C18:0 ceramide and insulin resistance in humans may be correlative48,317,334, but not causal. However, it is possible that SkM C18:0 ceramide in humans is more physiologically significant in its contribution to insulin action and glucose homeostasis,

143 Chapter 6

as animal studies are not always directly translatable to humans. One reason may be that humans are exposed to a mixed diet composed of fats, carbohydrates and proteins. Although the rodent HFD diet is composed of mostly fat, the nutrition values are more well defined, designed to provide all the necessary nutrients to the rodents living in controlled housing. Humans are also much more variable in the degree of physical activity and genetic variance than the mice used for these studies. Another important consideration in the current study is that HFD feeding only induced a modest increase in C18:0 ceramide in the SkM. A longer period of high fat feeding may be needed to induce a more significant impact on glucose metabolism before investigating the possible effect of CerS1 inhibition by P053 on glucose homeostasis. On the other hand, whole body glucose intolerance and muscle insulin resistance were clearly apparent in the mice fed a HFD in this study, which indicates that accumulation of C18 ceramide in muscle is not necessary for these physiological responses to a HFD.

Despite the lack of effect on glucose homeostasis, CerS1 inhibition with P053 significantly impeded fat deposition in mice fed a HFD when compared to the control group. Interestingly, the reduced percentage body fat and WAT weights correlate positively with SkM C18:0 ceramide levels and inversely with SkM C24:0 ceramide levels. Homogenates prepared from SkM of mice treated with P053 for 6 weeks demonstrated increased palmitate oxidation when compared to the vehicle control group. Increased palmitate oxidation was also observed in isolated muscle strips of mice treated with P053 for two weeks. This appeared to require a prolonged period of CerS1 inhibition, as acute treatment with P053 did not increase palmitate oxidation. The effect of CerS1 inhibition on FA oxidation is reflected in significantly increased protein levels of mitochondrial electron transport chain complexes I, II, III, and IV in the SkM. The effect was not evident in the liver, where CerS1 expression is markedly lower. Western blot results also demonstrated a significant diet effect for complexes I, III and V in the SkM of mice. High fat feeding reduced the protein levels of these complexes. In agreement with increased protein levels of mitochondrial complexes, P053 treatment increased the activity of respiratory complexes I, II, and IV in permeabilized muscle fibres. Additionally, P053 treatment enhanced the activities of β-HAD and CS. β-HAD catalyses the oxidation of straight-chain 3-hydroxyacyl-CoAs as part of the β oxidation pathway while CS catalyses the condensation reaction of the two-carbon acetate residue

144 Chapter 6

from acetyl-CoA and four-carbon oxaloacetate to form citrate in the first step of the TCA cycle. Increased activities of CS and β-HAD suggest increased β-oxidation of FA and increased flow of acetyl-CoA into the TCA cycle. Overall, these data suggest increased FA metabolism in the SkM of P053-treated mice.

Increased FA metabolism in SkM offers a reasonable explanation for the reduced fat deposition in P053-treated mice on a HFD. Accelerated fat metabolism in SkM may slow down fat deposition induced by HFD as the FA are being used as substrates for FA oxidation, thus reducing adiposity. However, to the best of my knowledge, there are no other studies demonstrating that increased fat metabolism specifically in SkM can lead to reduced adiposity in mice fed with HFD. In fact, prior studies have shown that enhancing FA oxidation in SkM through genetic deletion of ACC2 or SkM-specific expression of CPT-1 did not reduce adiposity335-337. The difference of the current study with the above reports is that the increase in FA oxidation was coupled with higher protein levels of respiratory complex subunits and enzyme activities. Our results provide evidence indicating that C18:0 ceramide, synthesized by CerS1 suppresses respiratory complex activity in SkM. The specific mechanism through which prolonged CerS1 inhibition influences mitochondrial function still requires further investigation. C18:0 ceramide may mediate its influence on mitochondrial function or content through mitophagy. C18:0 ceramide has been shown to induce mitophagy by targeting autophagolysosomes to mitochondria161. This was shown with either exogenous C18:0 ceramide analogues targeting the mitochondria or overexpression of CerS1 increasing de novo synthesis of C18:0 ceramides161. In addition, C16 ceramide has been shown to induce mitochondrial fission in SkM cells, thus reducing their oxidative capacity181. Therefore, reduction of C18:0 ceramide levels through CerS1 inhibition by P053 may aid mitochondrial functions and oxidative capacity by reducing mitophagy and blocking mitochondrial fission. Future work should investigate whether C18:0 ceramide and CerS1 can influence the regulation of mitochondrial biogenesis and whether P053 may reduce mitophagy.

The reduction of C18:0 ceramide with P053 treatment was accompanied by a modest elevation of C24:0 and C24:1 ceramides, which are synthesized by CerS2. The elevation of very long chain ceramides was not associated with increased CerS2 activity

145 Chapter 6

and is most likely due to CerS1 inhibition freeing up sphingoid base substrates for CerS2 as discussed in chapter 5. C24:0 and C24:1 ceramides may contribute just as much to the regulation of FA metabolism as long chain (C16:0 and C18:0) ceramides. Transcripts of genes involved in the regulation of FA metabolism are down-regulated in the liver of CerS2 null mice104. Downregulated expression of these genes is associated with reduced β-oxidation of FA59. The absence of CerS2 was also shown to reduce respiratory enzyme activities and increase mitochondrial oxidative stress101. These effects were reported to result from elevation of C16:0 ceramide in the absence of very long chain ceramide synthesis59,101. Data from the current study indicates that C18:0 ceramide in SkM, similar to C16:0 ceramide in liver, exerts negative effects on respiratory enzyme levels and activities and FA oxidation.

The role of ceramide in regulating fat metabolism and storage has been demonstrated in several prior studies. Genetic ablation of the Drosophila CerS homologue Schlank reduced TG storage in fat body338. It was shown that Schlank positively regulates FA synthesis by promoting the expression of the sterol-regulatory element-binding protein and its downstream target genes. Furthermore, Schlank prevents lipolysis by downregulating TG lipase expression338. Other studies have demonstrated that inhibition of de novo ceramide synthesis with the SPT inhibitor myriocin reduces adiposity in rodents fed a HFD196,224,309. In humans, a positive correlation between SkM C18:0 ceramide and visceral fat mass, liver fat, and blood pressure has been established317,339. Similarly, the current study indicated that SkM C18:0 ceramide levels are positively correlated with whole body adiposity in mice. Enhanced fat metabolism following CerS1 inhibition may be related to increased C24:0 and C24:1 ceramides instead, in accordance with the very strong correlations between very long chain ceramides in SkM and whole body fat mass. Although increased fat metabolism in SkM offers a reasonable explanation for the loss of fat mass in this study, the possibility that P053 has an off-target effect on adipose tissues or liver causing reduced adiposity cannot be ruled out at this stage. The expression of CerS1 is markedly lower in liver and adipose tissue. Thus, loss of adiposity is highly unlikely to result from CerS1 inhibition in these tissues. Further investigation is warranted to decipher/determine the cellular mechanism through which CerS1 inhibition enhances fat metabolism.

146 Chapter 6

In summary, work from this chapter indicated a potential role of CerS1 and C18:0 ceramide in fat storage by acting as a brake on mitochondrial FA oxidation. The current results also demonstrated that C18:0 ceramide and CerS1 activity are not requisite for the development of whole-body insulin resistance and glucose intolerance in mice following high-fat feeding.

147 Chapter 7

Chapter 7: Summary and Future Directions

Metabolic diseases in association with obesity are becoming alarmingly common throughout the world. Such an increase in the prevalence of obesity and metabolic complications presents a challenge for the human and economic resources of health service organizations and governments. One of the key defects of obesity is insulin resistance, which has been suggested as the underlying cause of metabolic syndrome5. In this regard, ceramide has been identified as a lipotoxic mediator of insulin resistance. Recent discoveries emphasize that different forms of ceramide, synthesized by different CerS isoforms, regulate distinct physiological processes in insulin resistance. These studies illustrate the importance of identifying and characterising the physiological role of individual CerS isoforms and ceramide species. Research on the role of specific CerS isoforms is limited to genetic manipulations, as there is currently a lack of isoform- selective CerS inhibitors. Researchers seeking to pharmacologically inhibit ceramide synthesis in vivo have generally used myriocin, which inhibits SPT, the rate limiting step in the de novo synthesis of all sphingolipids196,224,309.

Studies in this thesis describe the discovery and characterisation of the first potent, isoform-selective CerS inhibitor, P053, which specifically targets CerS1. Results from this thesis have demonstrated that P053 is highly selective for CerS1 over other CerS isoforms. P053 potently and very specifically reduced C18:0 ceramide levels in cultured cells and mouse SkM. The selective inhibition of CerS1 by P053 was also dissected from non-specific pro-apoptotic properties of sphingoid base analogues, as the compound inhibited C18:0 ceramide synthesis at concentrations at least one order of magnitude lower than those required to promote apoptosis in HEK293 cells. The mode of inhibition of P053 appeared to be non-competitive with respect to either sphingosine or C18:0 fatty acyl-CoA. In vivo, CerS1 inhibition with P053 in SkM significantly impeded body fat deposition in mice fed a HFD but did not prevent glucose intolerance or improve insulin sensitivity as previously demonstrated. Studies from this thesis also uncovered a new role for CerS1 and C18:0 ceramide as a regulator of mitochondrial function. Chronic CerS1 inhibition was shown to prime SkM to metabolise FA. The

148 Chapter 7

increase in oxidative capacity was accompanied by elevated levels of mitochondrial respiratory complex subunits, markers of the TCA cycle, and respiratory complex activity. Finally, this thesis revealed a significant positive correlation between SkM C18:0 ceramide and body adiposity. A very strong inverse correlation between SkM very long chain ceramide and body adiposity was also observed.

7.1. Potential limitations and future directions

Results from this thesis demonstrated that CerS1 inhibition by P053 had no effect on glucose uptake and insulin sensitivity in vivo. However, the molecular mechanism of glucose uptake or insulin signaling was not investigated in detail. Instead of using animal models, cell culture models could be a more cost-effective and rational approach. Unfortunately, a suitable muscle cell culture model was not established for the work of this thesis. C16:0 ceramide is the dominant ceramide species in both C2C12 and primary mouse myotubes. Previous literature regarding the role of ceramide on insulin signalling mechanisms was reported in cultured myotubes166,319,320,340. When individual ceramide species were reported, C16:0 ceramide remained the dominant species in these cultures including L6, a rat myoblast cell line319,320,341. Therefore, none of the above cell culture models are suitable for studying a CerS1 specific inhibitor. An important aspect to consider is that these findings about ceramide on SkM insulin signalling using cell culture models may not be as physiologically relevant as C18:0 ceramide, synthesized by CerS1, is the dominant species in both human and mouse SkM tissue48,319. Therefore, in vivo models are the preferred approach for studying SkM insulin resistance. The alternative may be genetic manipulation or altering the cell culture conditions for cultured SkM cells to restore physiological levels of C18:0 ceramide relative to other forms of ceramide.

An interesting observation from the current study was the complementary increase in C24:0 and C24:1 ceramides in both HEK293 cells and SkM of mice following inhibition of CerS1. This is likely a consequence of increased availability of sphingoid base substrates for CerS2-catalysed C24 ceramide synthesis in the absence of CerS1 activity, consistent with two distinct genetic mouse models of CerS1 deficiency97,316. There is also the possibility of a competing relationship between CerS1 and CerS2. A previous study reported that human CerS2 can form heterodimers with CerS5/6 and that

149 Chapter 7

its activity is dependent on CerS5310. There are currently no studies indicating similar findings for CerS1 but it is possible CerS1 can interact with other CerS isoforms and influence their catalytic activities. Future study can explore this possibility and whether this relationship is tissue-specific.

P053 has proven to be bioavailable and a potent CerS1 inhibitor. Upon gavage, P053 was distributed to various major tissues but remained at high levels in the circulation. The current study has not explored the route of P053 excretion, although there are indications that P053 may be cleared through the liver and kidneys as its concentration was the highest in these two tissues. The majority of drugs are eliminated by pathways that involve the kidneys or the liver. Future work should investigate the excretion of P053, including whether it is excreted in its original chemical state or undergoes chemical modification to biologically active or inactive metabolites. Future work should also include lipidomic profiling of blood plasma. Plasma ceramides have been correlated with the severity of insulin resistance, hypertension and CVD202-204. Plasma or serum ceramide concentrations have been used as a prediction tool for CVD risk and major cardiovascular events205,206. It would be interesting to see if P053 has any effect on plasma ceramides. In particular, circulating C18:0 ceramide has been correlated with increased visceral fat and reduced metabolic flexibility48. P053 may exert metabolic benefits by lowering circulating C18:0 ceramide levels. Another possibility to explore is whether treatment with P053 has any influence over the packaging of ceramide in LDL particles secreted from the liver and adipose tissue into the circulation. Ceramides in LDL have been associated with insulin resistance238.

Although the current profiling method is able to identify hundreds of lipid species and is reproducible, it is stringent and potentially biased. The LC method and sensitivity of the MS could be tuned to facilitate better identification of other lipid classes, including lysophospholipids, gangliosides, plasmologens, cholesterol esters and free FA. The current or newly generated data can be further assessed using computational and informatics approaches to study the lipidome in the context of known metabolic pathways and established pathophysiological responses. Systems biology approaches to study metabolic diseases are becoming increasingly popular with the advances in computational and mathematical modelling of complex biological systems. In this regard, proteomic analysis or RNA sequencing on samples from the current studies 150 Chapter 7

could provide a broader picture on the effect of CerS1 inhibition on metabolic pathways. The combination of proteomic, transcriptomic and lipidomic data would allow the study of the interactions between ceramide synthesis and metabolic pathways.

Due to the exciting consequences of CerS1 inhibition on fat deposition in this thesis, future research should also explore the relationship between C18:0 ceramide and adipose tissue functions, particularly on lipogenesis, lipolysis and adipogenesis. Adipose tissue expansion induced by high-fat feeding can occur through enlargement of adipocytes or an increase in adipocyte numbers by means of adipogenesis342. Adipogenesis is the development of fully differentiated mature adipocytes from their precursor cells. Adipose tissue not only plays an important role in fat storage and energy homeostasis, it is an endocrine organ that can secrete adipokines and numerous pro- and anti-inflammatory cytokines. Earlier literature has indicated an inverse relationship between adipocyte differentiation and ceramide concentrations in 3T3-L1, a pre- adipocyte cell line343. However, it is unknown in this study which particular ceramide species are involved. Ceramide 1-phosphate has also been associated with adipogenesis. Ceramide kinase is upregulated during differentiation of 3T3-L1 cells344. Noteworthy, knockdown of ceramide kinase resulted in a reduction of TG in adipocytes, decreased lipid droplet formation, blockade of leptin secretion and reduced expression of peroxisome proliferator-activated receptor gamma (PPARγ), a known master regulator of adipogenesis344. In vivo, ceramide kinase deficiency has been shown to prevent diet- induced weight gain, glucose intolerance in mice and adipocyte inflammation188. Therefore, the reduction of ceramides may directly lower levels of C1P and its effects on adipose tissue. Furthermore, it was demonstrated that ceramide may play a role in adipocyte browning. Ectopic ceramides were shown to inhibit the browning of beige adipocytes345. Further studies demonstrated that whole body or adipose tissue-specific inhibition of SPT in mice significantly reduced adipose sphingolipids, including ceramide, and resulted in altered adipose morphology and metabolism309. The effects included increased brown and beige adipocytes numbers, improved insulin sensitivity and mitochondrial activity309. The exact ceramide and sphingolipid species responsible were not fully dissected, but evidence suggests that C16:0 ceramide is the culprit for adipose tissue metabolic dysfunction58,59,108. Although C16:0 ceramide has been identified as the culprit in adipose tissue metabolic dysfunction, a HFD-induced increase in C16:0 ceramide58,346 was not always apparent309. This discrepancy still 151 Chapter 7

requires further investigation. Regardless, the effect of CerS1 inhibition in reducing fat is evident in this thesis and there is a possibility that P053 acts directly on adipose tissues. Thus, future research on the role of CerS1 inhibition on adipose tissue morphology and metabolism is warranted.

HFD feeding for 4-6 weeks in this thesis was sufficient to significantly increase the body fat composition of mice. However, the increase in 18:0 ceramide in SkM of mice was modest with a 6-week high fat feeding. Although 4-6 weeks of HFD feeding was adequate to induce whole-body insulin resistance in this study, high fat feeding is often maintained for 12-20 weeks to yield a more profound insulin resistance and glucose intolerance in the literature58,309,346. This thesis also utilized the more common experimental prevention paradigm where P053 was given concurrently with the HFD. Future research could employ a longer HFD feeding period and a reversal paradigm, in which P053 is administered after the animals display weight gain and metabolic dysfunction. C57BL/6 mice, used in this thesis, are considered an obesity prone strain, and are commonly used in obesity research. However, mouse strains can differ in their inherent propensities to develop metabolic diseases as shown previously347. The degree of SkM C18:0 ceramide content accumulation after high fat feeding also varies between mouse strains348. Thus, the variation between mouse strains should be taken into consideration when studying the role of ceramides in obesity in the future. Besides diet- induced obese rodent models, future testing of the role of CerS1 and other CerS isoforms should employ ob/ob mice or the fatty Zucker rat, which have a marked obesity phenotype.

7.2. Overall therapeutic potential of targeting ceramide metabolism

Based on the available evidence on the influence of ceramide on metabolic diseases, strategies aimed at reducing tissue ceramide levels may help obese or diabetic patients achieve improvements in glycemic control and insulin sensitivity. Moreover, targeting ceramide synthesis may help protect against CVD in humans202-206.

Although SkM C18:0 ceramide has been associated with various adverse metabolic phenotypes in humans, the current study presented no benefits on glycemic control and insulin sensitivity from CerS1 inhibition using P053 in mice. Also, CerS1 inhibition

152 Chapter 7

could have severe neurological consequences, previously reported in CerS1 knockout mice. These findings question the therapeutic potential of targeting CerS1 alone. As discussed previously, C16:0 ceramide appears to be a more damaging ceramide species in the context of metabolic diseases59,78,140,181. Thus, a CerS5/CerS6 inhibitor may prove more effective. Alternatively, an inhibitor targeting C16:0 and C18:0 ceramide synthesis without sacrificing the very long chain ceramides, which are insulin sensitizing, may prove the best approach59,103,235. In this regard, modification of current analogues with CerS inhibitory properties may generate other CerS isoform-specific inhibitors. Reducing ceramide content appears to be a promising approach that is translatable to humans as numerous studies have indicated accumulation of ceramides in tissue of obese and/or T2D humans47,237,322,349. However, it is important to note that some studies report a discrepancy in the correlation between SkM ceramide content and overall insulin sensitivity239,326. The reasons behind this discrepancy are unknown but highlight that there are multiple mediators of insulin resistance pathologies. Nonetheless, pharmacologically targeting CerSs and ceramide synthesis may still be a clinically significant approach.

Apart from targeting CerSs, inhibition of ceramide synthesis through other pharmacological or genetic means may have translational potentials as well. The majority of previous studies target SPT via myriocin, thereby inhibiting de novo ceramide biosynthesis28,222,231. However, relatively understudied in comparison is targeting of other enzymes involved in the synthesis and degradation of ceramide, which may represent potential therapeutic targets for reversal or prevention of obesity- induced metabolic diseases. For instance, fenretinide, a DEGS1 inhibitor used in obese mice and premenopausal women at high risk for breast cancer, can have insulin- sensitizing effects257-259. These effects may be mediated through the inhibition of ceramide formation from dihydroceramide, therefore protecting against ceramide- mediated insulin signalling impairments. In addition, reduction of ceramide levels can be achieved through enhanced flux into downstream generation of S1P. In this regard, transgenic mice overexpressing SPHK1 fed a HFD for 6 weeks demonstrated a striking improvement in SkM and whole-body insulin sensitivity compared with their wild-type littermates333. The effects were likely attributed to an attenuated intramuscular ceramide accumulation, including C18:0 ceramide. Despite the beneficial metabolic outcomes of targeting DEGS1 and SPHK1, the loss of ceramide content was not specific to the 153 Chapter 7

harmful long chain species. Therefore, targeting specific CerS isoforms would be advantageous with minimized risk of losing insulin-sensitizing very long chain ceramides. Other targets to consider include acid ceramidase225 and FGF21193. Improvement in glycemia was previously shown in transgenic mice overexpressing acid ceramidase, which promotes degradation of hepatic ceramides225. Treatment with FGF21 also improved glycemia via increase in adiponectin secretion and subsequent reduction of hepatic ceramide193. Thus, targeting ceramide metabolism does appear to be a promising approach to pursue further for the potential treatment of obesity-related metabolic disorders in humans.

In conclusion, work from this thesis has described and verified the first isoform- selective CerS inhibitor, specifically targeting CerS1 with nanomolar potency. P053 is potent and remained selective in vitro and in vivo, specifically targeting SkM ceramide content. Lipidomic profiling uncovers a potential complementary relationship between C18:0 ceramide, synthesized by CerS1, and very long chain ceramides, synthesized by CerS2. The balance of these ceramide species appeared to correlate with SkM TG content and whole body adiposity. This thesis also supports a model in which CerS1 activity and C18:0 ceramide in SkM are not requisite for the development of glucose intolerance and insulin resistance in mice fed a HFD, regardless of the positive effect on adiposity. Studies from this thesis also indicated that the effect of CerS1 inhibition on adiposity was likely due to enhanced mitochondrial function including increased FA oxidation and respiration. The marked effect on tissue-specific and whole-body adiposity by P053 not only warrants further investigation but further strengthens the notion that targeting ceramide synthesis may be a viable pharmacological option for treating obesity. P053 will provide a valuable resource for researchers seeking to investigate the functions of CerS1 in physiology and pathology.

154 Appendix

Appendix

Table A1. Statistically significant discoveries from multiple comparison analysis of 4-week quadriceps SkM lipidomic data. Three main comparisons were made for each lipid measured in quadriceps SkM of vehicle- or P053-treated mice on chow or HFD diet for 4 weeks (N = 10 mice for all groups except the chow + P053 group in SkM only, for which N = 8) using untargeted lipid profiling approach. Discoveries from multiple comparison analysis (Benjamini, Krieger and Yekutieli correction, Q = 1%) of 312 lipid species with statistical significance of P < 0.01 are shown. Abbreviation: Diff, difference; Cer, ceramide; CerG1, glucosyl- ceramide; SM, sphingomyelin; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PG, phosphatidylglycerol; PI, phosphatidylinositol; DG, diacylglycerol; TG, triacylglycerol.

(G1) Chow Vehicle vs (G2) Chow P053 Mean Mean Lipid ID P value Diff. SE of Diff. t ratio q value G1 G2 Cer(d18:1/18:0)+H <0.00001 3.457 2.892 0.565 0.067 8.380 0.000090 Cer(d18:2/18:0)+H <0.00001 0.961 0.281 0.680 0.104 6.525 0.001100 (G1) HFD Vehicle vs (G2) HFD P053 Cer(d18:1/18:0)+H <0.00001 3.479 2.826 0.653 0.088 7.394 0.000330 Cer(d18:2/18:0)+H <0.00001 1.022 0.192 0.829 0.133 6.224 0.001450 Cer(d18:1/24:1)+H 0.00009 0.968 1.390 -0.422 0.083 5.092 0.009420 (G1) Chow Vehicle vs (G2) HFD Vehicle PC(40:9)+H <0.00001 2.383 0.342 2.041 0.115 17.760 <0.00001 PS(19:0/22:6)-H <0.00001 1.604 0.338 1.266 0.079 16.000 <0.00001 Cer(d18:1/19:0)+H <0.00001 0.006 -0.923 0.929 0.066 14.110 <0.00001 PS(18:0/22:4)-H <0.00001 1.827 2.864 -1.037 0.074 13.920 <0.00001 PI(17:0/22:6)-H <0.00001 -0.022 -1.341 1.319 0.101 13.010 <0.00001 PI(18:1/22:6)-H <0.00001 2.178 0.980 1.198 0.099 12.140 <0.00001 PE(18:0p/22:4)+H <0.00001 1.519 2.739 -1.220 0.109 11.220 <0.00001 PI(19:0/20:4)-H <0.00001 1.066 -0.177 1.243 0.112 11.120 <0.00001 PC(39:7)+H <0.00001 3.149 2.012 1.137 0.106 10.680 <0.00001 PS(17:0/22:6)-H <0.00001 1.357 0.512 0.845 0.088 9.556 <0.00001 PI(17:0/20:4)-H <0.00001 0.999 0.133 0.867 0.095 9.112 <0.00001 PS(18:0/22:5)-H <0.00001 3.223 3.810 -0.587 0.069 8.477 <0.00001

155 Appendix

PI(16:0/22:6)-H <0.00001 2.442 1.775 0.667 0.080 8.315 <0.00001 PI(18:0/22:4)-H <0.00001 0.868 1.693 -0.824 0.105 7.832 <0.00001 PC(38:3)+H <0.00001 2.988 4.181 -1.193 0.164 7.286 0.000010 CL(74:8)-H <0.00001 2.146 2.742 -0.597 0.082 7.250 0.000010 PE(18:0p/20:4)+H <0.00001 3.129 4.090 -0.961 0.133 7.207 0.000010 PE(22:5/22:6)+H <0.00001 3.773 2.644 1.129 0.161 7.001 0.000020 PE(18:0/22:5)+H <0.00001 4.71 5.544 -0.834 0.119 6.995 0.000020 PI(18:2/20:4)-H <0.00001 0.479 -0.256 0.736 0.106 6.970 0.000020 PC(40:5)+H <0.00001 2.853 4.031 -1.177 0.173 6.801 0.000020 PI(18:1/22:5)-H <0.00001 0.878 0.210 0.669 0.099 6.734 0.000030 PC(38:2)+H <0.00001 2.027 2.550 -0.523 0.080 6.518 0.000040 CL(80:15)-H <0.00001 1.013 0.117 0.896 0.139 6.446 0.000040 PE(18:1/24:7)+H <0.00001 2.599 3.337 -0.738 0.115 6.432 0.000040 PC(38:4)+H <0.00001 5.903 6.847 -0.945 0.147 6.425 0.000040 PS(22:6/22:6)-H <0.00001 0.829 -0.160 0.989 0.154 6.424 0.000040 PC(44:12)+H <0.00001 1.143 -0.133 1.276 0.205 6.230 0.000060 PS(16:0/20:4)-H <0.00001 -0.005 0.543 -0.549 0.089 6.197 0.000060 PS(22:5/22:6)-H <0.00001 2.24 1.366 0.874 0.143 6.124 0.000070 CL(80:16)-H <0.00001 1.67 0.804 0.866 0.143 6.069 0.000070 PE(22:6/22:6)+H 0.00001 2.392 1.042 1.350 0.223 6.046 0.000070 PE(18:0p/22:5)+H 0.00001 3.229 4.045 -0.816 0.136 6.017 0.000070 PC(34:4)+H 0.00001 2.776 3.468 -0.692 0.115 6.014 0.000070 PI(18:1/20:4)-H 0.00001 3.475 2.827 0.649 0.109 5.967 0.000080 CL(72:8)-H 0.00001 3.412 4.166 -0.753 0.128 5.890 0.000090 CL(68:4)-H 0.00002 1.261 0.789 0.472 0.081 5.834 0.000100 PC(38:8)+H 0.00002 0.823 0.111 0.712 0.122 5.817 0.000100 PI(16:0/18:2)-H 0.00002 2.182 1.501 0.680 0.118 5.745 0.000110 PC(40:7)+H 0.00002 6.162 5.174 0.988 0.172 5.734 0.000110 TG(18:0/16:0/18:1)+NH4 0.00002 6.313 7.613 -1.300 0.229 5.683 0.000120 CL(83:8)-H 0.00002 1.077 1.698 -0.621 0.110 5.666 0.000120 TG(18:0/18:0/18:1)+NH4 0.00002 3.545 4.720 -1.175 0.208 5.649 0.000120 PS(16:0/22:5)-H 0.00003 1.391 1.721 -0.331 0.060 5.531 0.000150 PC(39:6)+H 0.00003 3.769 2.425 1.344 0.243 5.529 0.000150 PI(18:1/18:1)-H 0.00003 0.311 -0.417 0.728 0.133 5.474 0.000160 SM(d37:1)+H 0.00003 1.086 0.265 0.821 0.150 5.467 0.000160 PS(18:0/20:4)-H 0.00005 2.415 2.970 -0.555 0.105 5.300 0.000230 PS(18:1/22:6)-H 0.00005 2.356 2.009 0.348 0.066 5.250 0.000250 PC(40:7p)+H 0.00006 2.536 1.682 0.854 0.163 5.224 0.000260 PI(18:0/18:1)-H 0.00007 1.582 0.988 0.594 0.115 5.148 0.000300

156 Appendix

CL(76:13)-H 0.00008 0.224 -0.868 1.091 0.215 5.076 0.000340 PE(18:0/22:4)+H 0.00009 2.187 3.266 -1.079 0.214 5.037 0.000370 TG(18:0/18:1/18:1)+NH4 0.0001 6.964 8.216 -1.251 0.251 4.987 0.000400 Cer(d18:1/23:0)+H 0.0001 0.044 -0.447 0.491 0.099 4.965 0.000410 PI(18:1/20:3)-H 0.00014 1.235 0.701 0.534 0.111 4.811 0.000570 PC(42:10)+H 0.00015 3.229 2.326 0.903 0.189 4.789 0.000580 PE(18:1p/22:6)+H 0.00017 5.768 5.070 0.698 0.148 4.718 0.000670 PC(44:11)+H 0.00018 0.977 -0.088 1.065 0.226 4.706 0.000670 CL(80:10)-H 0.00018 -0.786 0.107 -0.893 0.190 4.697 0.000670 CL(72:9)-H 0.00018 1.609 1.223 0.387 0.082 4.693 0.000670 PC(40:10)+H 0.00022 2.169 1.187 0.981 0.213 4.605 0.000800 PE(19:0/22:6)+H 0.00022 4.525 3.689 0.837 0.182 4.595 0.000810 CL(74:7)-H 0.00024 1.681 2.111 -0.431 0.094 4.556 0.000870 PC(40:4)+H 0.00027 1.193 2.600 -1.407 0.312 4.508 0.000950 Cer(d18:1/24:1)+H 0.00029 1.363 0.968 0.395 0.088 4.472 0.001010 PI(18:1/18:2)-H 0.00033 1.261 0.638 0.622 0.141 4.422 0.001110 PC(40:6e)+H 0.00039 0.764 1.446 -0.682 0.157 4.347 0.001290 PE(16:1/22:6)+H 0.00043 3.683 3.114 0.569 0.132 4.305 0.001400 PC(42:6)+H 0.00045 0.615 1.322 -0.707 0.165 4.278 0.001460 PC(36:1)+H 0.00048 4.575 4.925 -0.349 0.082 4.249 0.001540 CL(68:5)-H 0.00054 1.495 1.175 0.320 0.076 4.197 0.001690 PC(40:8p)+H 0.00055 -0.530 -1.286 0.756 0.180 4.193 0.001690 PC(31:0)+H 0.00062 3.001 1.600 1.401 0.339 4.133 0.001910 DG(18:1/22:6)+NH4 0.00065 4.406 3.405 1.002 0.243 4.118 0.001950 PE(18:0/20:4)+H 0.00071 5.917 6.580 -0.663 0.163 4.077 0.002110 PE(17:0/22:6)+H 0.00072 3.97 3.061 0.910 0.224 4.068 0.002120 PC(38:6)+H 0.00089 8.715 8.108 0.607 0.153 3.975 0.002570 PE(16:0p/16:0)+H 0.00101 1.055 1.414 -0.359 0.092 3.917 0.002900 PC(33:2)+H 0.00103 3.647 4.203 -0.557 0.142 3.909 0.002910 CL(70:5)-H 0.00113 2.45 2.116 0.334 0.086 3.866 0.003160 CL(76:10)-H 0.00138 1.994 1.629 0.364 0.097 3.776 0.003810 PI(20:3/20:4)-H 0.00196 -0.330 -0.886 0.556 0.154 3.620 0.005330 TG(18:0/16:0/16:0)+NH4 0.0021 4.446 5.291 -0.845 0.236 3.588 0.005670 PI(16:0/16:0)-H 0.00227 0.590 0.007 0.583 0.164 3.554 0.006020 TG(20:0/18:1/18:1)+NH4 0.00229 3.806 4.591 -0.785 0.221 3.550 0.006020 CL(78:14)-H 0.00234 1.522 1.167 0.355 0.100 3.540 0.006080 Cer(d18:1/24:0)+H 0.00266 0.432 0.121 0.311 0.089 3.483 0.006830 PI(16:0/20:3)-H 0.00351 0.459 0.061 0.398 0.119 3.357 0.008890 PC(35:3)+H 0.00353 2.072 0.897 1.175 0.351 3.354 0.008890

157 Appendix

Table A2. Statistically significant discoveries from multiple comparison analysis of 6-week quadriceps SkM lipidomic data. Three main comparisons were made for each lipid measured in quadriceps SkM of vehicle- or P053-treated mice on chow or HFD diet for 6 weeks (N = 10 mice for all groups) using untargeted lipid profiling approach. Discoveries from multiple comparison analysis (Benjamini, Krieger and Yekutieli correction, Q = 1%) of 353 lipid species with statistical significance of P < 0.01 are shown. Abbreviation: Cer, ceramide; CerG1, glucosyl-ceramide; SM, sphingomyelin; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PG, phosphatidylglycerol; PI, phosphatidylinositol; DG, diacylglycerol; TG, triacylglycerol.

(G1) Chow Vehicle vs (G2) Chow P053 Mean Mean SE of Lipid ID P value Diff. t ratio q value G1 G2 Diff. Cer(d18:1/19:0)+H <0.00001 -0.946 -1.468 0.522 0.073 7.148 0.000320 Cer(d18:2/18:0)+H <0.00001 1.084 0.528 0.557 0.080 6.924 0.000320 Cer(d18:1/18:0)+H 0.00002 3.305 2.900 0.406 0.070 5.812 0.001950 PG(16:0/16:0)+NH4 0.00009 1.805 1.401 0.405 0.081 5.023 0.007790 (G1) HFD Vehicle vs (G2) HFD P053 Cer(d18:1/19:0)+H <0.00001 -2.066 -2.874 0.808 0.066 12.330 <0.00001 Cer(d18:2/18:0)+H <0.00001 0.967 0.038 0.929 0.089 10.480 <0.00001 Cer(d18:1/18:0)+H <0.00001 3.584 2.798 0.785 0.087 9.062 <0.00001 Cer(d18:0/24:1)+H <0.00001 -3.118 -2.494 -0.624 0.101 6.183 0.000680 Cer(d18:1/24:0)+H 0.00001 0.026 0.382 -0.356 0.061 5.875 0.001020 SM(d35:1)+H 0.00002 2.306 1.889 0.418 0.072 5.773 0.001050 Cer(d18:1/24:1)+H 0.00015 0.662 0.975 -0.313 0.065 4.787 0.007370 (G1) Chow Vehicle vs (G2) HFD Vehicle PE(18:0p/22:4)+H <0.00001 1.017 2.844 -1.827 0.072 25.440 <0.00001 PE(18:3/22:6)+H <0.00001 2.691 0.893 1.798 0.083 21.710 <0.00001 PS(18:0/22:4)-H <0.00001 2.347 3.717 -1.371 0.077 17.900 <0.00001 PE(16:0p/22:4)+H <0.00001 2.232 3.602 -1.370 0.079 17.390 <0.00001 PE(18:0/22:4)+H <0.00001 1.997 3.760 -1.763 0.102 17.360 <0.00001 Cer(d18:1/19:0)+H <0.00001 -0.946 -2.066 1.120 0.078 14.330 <0.00001 PE(20:5/22:6)+H <0.00001 0.726 -1.011 1.737 0.125 13.940 <0.00001 SM(d37:1)+H <0.00001 1.650 0.367 1.282 0.095 13.430 <0.00001 PE(18:0/22:5)+H <0.00001 4.780 5.860 -1.081 0.088 12.300 <0.00001 PS(19:0/22:6)-H <0.00001 2.618 1.210 1.408 0.117 12.040 <0.00001 PE(18:0p/20:4)+H <0.00001 3.229 4.133 -0.904 0.077 11.740 <0.00001

158 Appendix

PE(22:6/22:6)+H <0.00001 2.494 1.396 1.098 0.102 10.810 <0.00001 PE(18:0p/20:3)+H <0.00001 0.076 0.955 -0.880 0.084 10.430 <0.00001 PE(15:0/22:6)+H <0.00001 1.603 0.590 1.014 0.098 10.370 <0.00001 PG(22:6/22:6)+NH4 <0.00001 0.119 -1.287 1.407 0.136 10.350 <0.00001 CL(72:8)-H <0.00001 4.566 5.547 -0.981 0.095 10.340 <0.00001 PE(18:0/20:4)+H <0.00001 5.641 6.311 -0.670 0.066 10.090 <0.00001 PS(18:0/22:5)-H <0.00001 4.084 4.983 -0.899 0.090 10.030 <0.00001 PE(22:5/22:6)+H <0.00001 3.899 2.765 1.134 0.115 9.872 <0.00001 PE(16:1/22:6)+H <0.00001 3.942 3.095 0.847 0.087 9.709 <0.00001 PE(18:1/22:6)+H <0.00001 6.175 5.272 0.903 0.096 9.374 <0.00001 PE(16:0/22:5)+H <0.00001 3.973 4.830 -0.857 0.098 8.748 <0.00001 PE(20:0p/20:4)+H <0.00001 0.479 1.283 -0.804 0.094 8.583 <0.00001 PE(16:2/22:6)+H <0.00001 -0.529 -1.237 0.708 0.083 8.529 <0.00001 CL(74:8)-H <0.00001 3.149 3.908 -0.759 0.091 8.310 <0.00001 PE(20:1/22:6)+H <0.00001 2.852 1.988 0.864 0.106 8.195 <0.00001 PI(18:1/22:6)-H <0.00001 4.113 2.982 1.130 0.139 8.162 <0.00001 PS(18:0/20:4)-H <0.00001 3.017 3.724 -0.707 0.087 8.115 <0.00001 PE(18:0p/22:5)+H <0.00001 3.316 4.106 -0.790 0.098 8.050 <0.00001 PI(18:0/22:4)-H <0.00001 2.515 3.775 -1.260 0.159 7.919 <0.00001 PE(18:0p/16:0)+H <0.00001 0.818 1.420 -0.603 0.077 7.867 <0.00001 PS(17:0/22:6)-H <0.00001 2.183 1.413 0.771 0.102 7.546 <0.00001 PI(19:0/20:4)-H <0.00001 2.994 1.867 1.128 0.166 6.808 0.000020 PE(18:1p/22:6)+H <0.00001 5.769 5.223 0.546 0.081 6.732 0.000020 PI(17:0/22:6)-H <0.00001 1.752 0.645 1.107 0.167 6.612 0.000030 PE(17:0/22:6)+H <0.00001 2.727 1.509 1.219 0.185 6.593 0.000030 PE(18:2/20:4)+H <0.00001 2.303 2.857 -0.555 0.088 6.298 0.000050 CL(83:8)-H 0.00001 1.800 2.901 -1.102 0.183 6.007 0.000080 CL(74:7)-H 0.00001 2.523 3.104 -0.582 0.098 5.909 0.000100 PS(16:0/20:4)-H 0.00002 0.660 1.344 -0.683 0.117 5.821 0.000110 PE(22:1/20:4)+H 0.00002 -0.512 -0.089 -0.423 0.073 5.784 0.000120 CL(80:10)-H 0.00002 -0.508 1.183 -1.691 0.296 5.707 0.000130 PE(16:0/20:4)+H 0.00002 4.428 4.894 -0.466 0.083 5.636 0.000150 PG(16:0/14:0)+NH4 0.00003 0.236 1.103 -0.867 0.155 5.602 0.000160 PE(20:0/20:4)+H 0.00003 -0.504 0.458 -0.962 0.173 5.552 0.000170 PE(20:1/18:1)+H 0.00003 0.824 1.263 -0.439 0.080 5.465 0.000200 PE(16:0p/14:0)+H 0.00004 -1.148 -0.689 -0.459 0.086 5.338 0.000260 CL(80:16)-H 0.00006 3.078 2.338 0.740 0.141 5.234 0.000310 PE(18:0/20:3)+H 0.00006 2.643 3.132 -0.489 0.094 5.227 0.000310 PE(16:0/22:6)+H 0.00006 6.427 5.705 0.722 0.138 5.218 0.000310

159 Appendix

CL(68:4)-H 0.00006 2.263 1.804 0.459 0.088 5.215 0.000310 CL(76:13)-H 0.00006 1.154 0.299 0.856 0.164 5.214 0.000310 PG(18:1/18:2)+NH4 0.00007 0.287 1.143 -0.856 0.168 5.109 0.000380 CL(74:11)-H 0.00008 3.724 4.110 -0.386 0.076 5.067 0.000390 PI(18:1/22:5)-H 0.00008 2.521 1.731 0.790 0.156 5.064 0.000390 SM(d38:1)+H 0.00008 0.532 -0.024 0.556 0.110 5.063 0.000390 PG(16:0/18:2)+NH4 0.00008 2.978 3.643 -0.665 0.131 5.062 0.000390 PE(24:4/22:6)+H 0.00010 0.020 0.530 -0.509 0.102 4.988 0.000450 PE(14:0/20:4)+H 0.00010 -0.653 -0.009 -0.644 0.130 4.951 0.000480 PG(18:2/18:2)+NH4 0.00011 -1.235 -0.424 -0.811 0.165 4.902 0.000520 Cer(d18:1/18:0)+H 0.00014 3.305 3.584 -0.278 0.058 4.797 0.000650 CL(72:7)-H 0.00017 4.591 4.938 -0.347 0.073 4.734 0.000730 PE(16:0/16:0)+H 0.00021 1.407 1.010 0.397 0.086 4.616 0.000930 PS(16:0/22:5)-H 0.00024 2.333 2.837 -0.503 0.110 4.561 0.001030 PE(16:1/16:1)+H 0.00031 0.083 -0.327 0.410 0.092 4.444 0.001310 PS(22:6/22:6)-H 0.00033 1.815 0.824 0.991 0.225 4.415 0.001380 TG(18:0/16:0/18:0)+N 0.00034 2.762 3.953 -1.191 0.271 4.401 0.001400 H4 SM(d40:1)+H 0.00036 2.796 3.157 -0.361 0.082 4.386 0.001430 SM(d18:1/23:3)+H 0.00039 -2.993 -2.656 -0.337 0.078 4.346 0.001540 Cer(d18:1/25:0)+H 0.00054 -2.400 -2.808 0.408 0.097 4.197 0.002110 PE(20:0/22:6)+H 0.00072 1.552 1.154 0.398 0.098 4.069 0.002770 PI(18:2/20:4)-H 0.00075 2.524 1.899 0.625 0.154 4.052 0.002830 PE(18:2/22:6)+H 0.00083 5.439 5.161 0.278 0.070 4.005 0.003100 CerG1(d18:1/16:0)+H 0.00103 -1.129 -1.852 0.723 0.185 3.908 0.003780 PI(18:1/18:1)-H 0.00104 1.953 1.272 0.681 0.174 3.905 0.003780 PE(16:0p/16:0)+H 0.00119 1.177 1.436 -0.259 0.067 3.844 0.004270 Cer(d18:0/24:0)+H 0.00124 -2.108 -2.496 0.388 0.101 3.826 0.004390 Cer(d18:1/24:1)+H 0.00140 0.947 0.662 0.285 0.076 3.772 0.004880 PE(16:0/16:1)+H 0.00145 2.171 1.842 0.330 0.088 3.756 0.004990 PS(22:5/22:6)-H 0.00152 3.020 2.215 0.805 0.216 3.733 0.005190 Cer(d18:0/24:1)+H 0.00184 -2.759 -3.118 0.358 0.098 3.648 0.006200 TG(18:0/16:0/16:0)+N 0.00200 3.435 4.492 -1.057 0.293 3.609 0.006670 H4 CL(70:5)-H 0.00208 3.429 3.119 0.310 0.086 3.594 0.006820 PI(17:0/20:4)-H 0.00242 2.780 2.270 0.509 0.145 3.524 0.007820 CL(74:9)-H 0.00244 3.730 4.029 -0.298 0.085 3.522 0.007820 PS(18:1/22:6)-H 0.00265 3.131 2.747 0.385 0.110 3.485 0.008390 PE(18:1/18:1)+H 0.00280 3.241 4.354 -1.113 0.322 3.459 0.008770 PE(18:2p/22:6)+H 0.00302 2.156 1.639 0.518 0.151 3.424 0.009370

160 Appendix

Table A3. Statistically significant discoveries from multiple comparison analysis of 4-week liver lipidomic data. Three main comparisons were made for each lipid measured in liver of vehicle- or P053-treated mice on chow or HFD diet for 4 weeks (N = 10 mice for all groups) using untargeted lipid profiling approach. Discoveries from multiple comparison analysis (Benjamini, Krieger and Yekutieli correction, Q = 1%) of 265 lipid species with statistical significance of P < 0.01 are shown. For this data set, there were only discoveries for chow vehicle vs HFD vehicle comparison. Abbreviation: Diff, difference; Cer, ceramide; CerG1, glucosyl-ceramide; SM, sphingomyelin; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PG, phosphatidylglycerol; PI, phosphatidylinositol; DG, diacylglycerol; TG, triacylglycerol.

(G1) Chow Vehicle vs (G2) HFD Vehicle Mean Mean SE of Lipid ID P value G1 G2 Diff. Diff. t ratio q value PE(16:1/18:2)+H <0.00001 4.137 3.139 0.998 0.101 9.924 <0.00001 PC(34:0e)+H <0.00001 0.356 1.072 -0.716 0.089 8.024 0.00003 SM(d36:0)+H <0.00001 0.663 1.876 -1.213 0.170 7.133 0.00006 PE(16:0/20:5)+H <0.00001 5.292 3.116 2.176 0.307 7.096 0.00006 PC(35:2)+H <0.00001 4.322 3.242 1.080 0.152 7.088 0.00006 PC(37:2)+H <0.00001 3.096 1.950 1.146 0.170 6.752 0.00010 CerG1(d18:1/22:0)+H <0.00001 3.097 3.918 -0.821 0.124 6.601 0.00010 PC(36:5)+H <0.00001 6.140 5.320 0.820 0.124 6.590 0.00010 PE(18:0p/20:4)+H <0.00001 4.124 5.072 -0.948 0.151 6.294 0.00016 CerG1(d38:1)+H 0.00003 1.254 2.097 -0.843 0.152 5.542 0.00068 PI(18:0/18:2)+NH4 0.00003 3.981 2.427 1.553 0.284 5.473 0.00071 PE(18:0/22:5)+H 0.00005 3.258 4.410 -1.152 0.216 5.336 0.00087 SM(d38:1)+H 0.00005 3.198 4.298 -1.100 0.208 5.292 0.00089 PC(39:7)+H 0.00008 2.717 1.894 0.823 0.162 5.098 0.00125 PE(17:0/18:1)+H 0.00008 2.494 1.735 0.759 0.150 5.049 0.00129 PI(18:0/20:2)+NH4 0.00009 2.999 1.197 1.803 0.360 5.006 0.00133 PC(40:4)+H 0.00010 2.487 3.176 -0.689 0.138 4.979 0.00133 PC(44:12)+H 0.00010 1.399 0.566 0.832 0.168 4.943 0.00133 PC(33:0)+H 0.00011 2.481 1.624 0.857 0.174 4.925 0.00133 PE(18:0/22:4)+H 0.00011 3.031 3.771 -0.740 0.151 4.902 0.00133 SM(d43:1)+H 0.00018 1.797 0.874 0.923 0.197 4.693 0.00200 PC(30:1)+H 0.00026 0.805 -0.065 0.870 0.193 4.521 0.00279

161 Appendix

PE(18:1/18:2)+H 0.00029 6.769 6.130 0.640 0.143 4.475 0.00296 PC(35:1)+H 0.00035 3.792 3.291 0.501 0.114 4.396 0.00337 PC(34:0)+H 0.00039 3.704 4.192 -0.488 0.112 4.350 0.00359 Cer(d18:1/24:1)+H 0.00044 2.153 1.540 0.613 0.143 4.292 0.00392 PE(18:0/18:2)+H 0.00053 5.482 4.658 0.825 0.196 4.209 0.00454 PS(18:0/22:4)+H 0.00067 2.319 4.011 -1.692 0.413 4.099 0.00559 SM(d40:0)+H 0.00080 0.971 1.687 -0.716 0.178 4.024 0.00638 PE(16:0/20:1)+H 0.00086 4.653 4.213 0.440 0.110 3.987 0.00648 PC(37:3)+H 0.00086 2.829 1.765 1.064 0.267 3.987 0.00648 SM(d42:4)+H 0.00091 3.492 4.153 -0.661 0.167 3.964 0.00649 SM(d43:2)+H 0.00095 0.947 0.151 0.796 0.202 3.946 0.00649 PI(18:0/20:3)+NH4 0.00095 5.807 3.897 1.910 0.484 3.944 0.00649 PE(16:0/16:1)+H 0.00106 3.958 2.772 1.187 0.305 3.894 0.00706 PC(39:6)+H 0.00113 3.768 2.798 0.970 0.251 3.865 0.00732 PI(18:0/18:1)+NH4 0.00121 3.159 2.153 1.006 0.262 3.837 0.00757 PC(40:2)+H 0.00125 0.962 0.479 0.483 0.126 3.823 0.00762 PC(36:4)+H 0.00146 7.782 8.642 -0.860 0.229 3.752 0.00869 Cer(d18:1/22:0)+H 0.00170 1.501 2.001 -0.499 0.136 3.683 0.00989

162

References:

1 Barroso, I., Luan, J., Sandhu, M. S., Franks, P. W., Crowley, V., Schafer, A. J., O'Rahilly, S. & Wareham, N. J. Meta-analysis of the Gly482Ser variant in PPARGC1A in type 2 diabetes and related phenotypes. Diabetologia 49, 501- 505, doi:10.1007/s00125-005-0130-2 (2006). 2 Lindroos, M. M., Majamaa, K., Tura, A., Mari, A., Kalliokoski, K. K., Taittonen, M. T., Iozzo, P. & Nuutila, P. m.3243A>G mutation in mitochondrial DNA leads to decreased insulin sensitivity in skeletal muscle and to progressive beta-cell dysfunction. Diabetes 58, 543-549, doi:10.2337/db08-0981 (2009). 3 Ling, C., Poulsen, P., Simonsson, S., Ronn, T., Holmkvist, J., Almgren, P., Hagert, P., Nilsson, E., Mabey, A. G., Nilsson, P., Vaag, A. & Groop, L. Genetic and epigenetic factors are associated with expression of respiratory chain component NDUFB6 in human skeletal muscle. The Journal of clinical investigation 117, 3427-3435, doi:10.1172/JCI30938 (2007). 4 Ronn, T., Poulsen, P., Tuomi, T., Isomaa, B., Groop, L., Vaag, A. & Ling, C. Genetic variation in ATP5O is associated with skeletal muscle ATP50 mRNA expression and glucose uptake in young twins. PloS one 4, e4793, doi:10.1371/journal.pone.0004793 (2009). 5 Han, T. S. & Lean, M. E. A clinical perspective of obesity, metabolic syndrome and cardiovascular disease. JRSM Cardiovasc Dis 5, 2048004016633371, doi:10.1177/2048004016633371 (2016). 6 WHO. Obesity and overweight, (2016). 7 Ford, E. S. Risks for all-cause mortality, cardiovascular disease, and diabetes associated with the metabolic syndrome: a summary of the evidence. Diabetes Care 28, 1769-1778 (2005). 8 Ogurtsova, K., da Rocha Fernandes, J. D., Huang, Y., Linnenkamp, U., Guariguata, L., Cho, N. H., Cavan, D., Shaw, J. E. & Makaroff, L. E. IDF Diabetes Atlas: Global estimates for the prevalence of diabetes for 2015 and 2040. Diabetes Res Clin Pract 128, 40-50, doi:10.1016/j.diabres.2017.03.024 (2017).

163

9 Chisholm, D. J., Campbell, L. V. & Kraegen, E. W. Pathogenesis of the insulin resistance syndrome (syndrome X). Clin Exp Pharmacol Physiol 24, 782-784 (1997). 10 Saltiel, A. R. New perspectives into the molecular pathogenesis and treatment of type 2 diabetes. Cell 104, 517-529 (2001). 11 Marchesini, G., Moscatiello, S., Di Domizio, S. & Forlani, G. Obesity- associated liver disease. The Journal of clinical endocrinology and metabolism 93, S74-80, doi:10.1210/jc.2008-1399 (2008). 12 Kim, J. K., Wi, J. K. & Youn, J. H. Plasma free fatty acids decrease insulin- stimulated skeletal muscle glucose uptake by suppressing glycolysis in conscious rats. Diabetes 45, 446-453 (1996). 13 Bugianesi, E., McCullough, A. J. & Marchesini, G. Insulin resistance: a metabolic pathway to chronic liver disease. Hepatology 42, 987-1000, doi:10.1002/hep.20920 (2005). 14 Cano, A., Ciaffoni, F., Safwat, G. M., Aspichueta, P., Ochoa, B., Bravo, E. & Botham, K. M. Hepatic VLDL assembly is disturbed in a rat model of nonalcoholic fatty liver disease: is there a role for dietary coenzyme Q? J Appl Physiol (1985) 107, 707-717, doi:10.1152/japplphysiol.00297.2009 (2009). 15 Lam, T. K., van de Werve, G. & Giacca, A. Free fatty acids increase basal hepatic glucose production and induce hepatic insulin resistance at different sites. American journal of physiology. Endocrinology and metabolism 284, E281-290, doi:10.1152/ajpendo.00332.2002 (2003). 16 Kaur, J. A comprehensive review on metabolic syndrome. Cardiol Res Pract 2014, 943162, doi:10.1155/2014/943162 (2014). 17 Sjostrom, C. D., Lissner, L., Wedel, H. & Sjostrom, L. Reduction in incidence of diabetes, hypertension and lipid disturbances after intentional weight loss induced by bariatric surgery: the SOS Intervention Study. Obes Res 7, 477-484 (1999). 18 Gallagher, E. J., Leroith, D. & Karnieli, E. The metabolic syndrome--from insulin resistance to obesity and diabetes. Med Clin North Am 95, 855-873, doi:10.1016/j.mcna.2011.06.001 (2011). 19 Miller, M., Stone, N. J., Ballantyne, C., Bittner, V., Criqui, M. H., Ginsberg, H. N., Goldberg, A. C., Howard, W. J., Jacobson, M. S., Kris-Etherton, P. M., Lennie, T. A., Levi, M., Mazzone, T., Pennathur, S., American Heart 164

Association Clinical Lipidology, T., Prevention Committee of the Council on Nutrition, P. A., Metabolism, Council on Arteriosclerosis, T., Vascular, B., Council on Cardiovascular, N. & Council on the Kidney in Cardiovascular, D. Triglycerides and cardiovascular disease: a scientific statement from the American Heart Association. Circulation 123, 2292-2333, doi:10.1161/CIR.0b013e3182160726 (2011). 20 Knowler, W. C., Barrett-Connor, E., Fowler, S. E., Hamman, R. F., Lachin, J. M., Walker, E. A., Nathan, D. M. & Diabetes Prevention Program Research, G. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. The New England journal of medicine 346, 393-403, doi:10.1056/NEJMoa012512 (2002). 21 Unger, R. H. Lipotoxicity in the pathogenesis of obesity-dependent NIDDM. Genetic and clinical implications. Diabetes 44, 863-870 (1995). 22 Lee, Y., Hirose, H., Ohneda, M., Johnson, J. H., McGarry, J. D. & Unger, R. H. Beta-cell lipotoxicity in the pathogenesis of non-insulin-dependent diabetes mellitus of obese rats: impairment in adipocyte-beta-cell relationships. Proceedings of the National Academy of Sciences of the United States of America 91, 10878-10882 (1994). 23 Unger, R. H., Clark, G. O., Scherer, P. E. & Orci, L. Lipid homeostasis, lipotoxicity and the metabolic syndrome. Biochim Biophys Acta 1801, 209-214, doi:10.1016/j.bbalip.2009.10.006 (2010). 24 Turner, N., Kowalski, G. M., Leslie, S. J., Risis, S., Yang, C., Lee-Young, R. S., Babb, J. R., Meikle, P. J., Lancaster, G. I., Henstridge, D. C., White, P. J., Kraegen, E. W., Marette, A., Cooney, G. J., Febbraio, M. A. & Bruce, C. R. Distinct patterns of tissue-specific lipid accumulation during the induction of insulin resistance in mice by high-fat feeding. Diabetologia 56, 1638-1648, doi:10.1007/s00125-013-2913-1 (2013). 25 Itani, S. I., Ruderman, N. B., Schmieder, F. & Boden, G. Lipid-induced insulin resistance in human muscle is associated with changes in diacylglycerol, protein kinase C, and IkappaB-alpha. Diabetes 51, 2005-2011 (2002). 26 Kumashiro, N., Erion, D. M., Zhang, D., Kahn, M., Beddow, S. A., Chu, X., Still, C. D., Gerhard, G. S., Han, X., Dziura, J., Petersen, K. F., Samuel, V. T. & Shulman, G. I. Cellular mechanism of insulin resistance in nonalcoholic fatty

165

liver disease. Proceedings of the National Academy of Sciences of the United States of America 108, 16381-16385, doi:10.1073/pnas.1113359108 (2011). 27 Holland, W. L., Bikman, B. T., Wang, L. P., Yuguang, G., Sargent, K. M., Bulchand, S., Knotts, T. A., Shui, G., Clegg, D. J., Wenk, M. R., Pagliassotti, M. J., Scherer, P. E. & Summers, S. A. Lipid-induced insulin resistance mediated by the proinflammatory receptor TLR4 requires saturated fatty acid-induced ceramide biosynthesis in mice. The Journal of clinical investigation 121, 1858- 1870, doi:10.1172/JCI43378 (2011). 28 Ussher, J. R., Koves, T. R., Cadete, V. J., Zhang, L., Jaswal, J. S., Swyrd, S. J., Lopaschuk, D. G., Proctor, S. D., Keung, W., Muoio, D. M. & Lopaschuk, G. D. Inhibition of de novo ceramide synthesis reverses diet-induced insulin resistance and enhances whole-body oxygen consumption. Diabetes 59, 2453-2464, doi:10.2337/db09-1293 (2010). 29 Reaven, G. M., Hollenbeck, C., Jeng, C. Y., Wu, M. S. & Chen, Y. D. Measurement of plasma glucose, free fatty acid, lactate, and insulin for 24 h in patients with NIDDM. Diabetes 37, 1020-1024 (1988). 30 Boden, G., Cheung, P., Stein, T. P., Kresge, K. & Mozzoli, M. FFA cause hepatic insulin resistance by inhibiting insulin suppression of glycogenolysis. American journal of physiology. Endocrinology and metabolism 283, E12-19, doi:10.1152/ajpendo.00429.2001 (2002). 31 Yu, C., Chen, Y., Cline, G. W., Zhang, D., Zong, H., Wang, Y., Bergeron, R., Kim, J. K., Cushman, S. W., Cooney, G. J., Atcheson, B., White, M. F., Kraegen, E. W. & Shulman, G. I. Mechanism by which fatty acids inhibit insulin activation of insulin receptor substrate-1 (IRS-1)-associated phosphatidylinositol 3-kinase activity in muscle. The Journal of biological chemistry 277, 50230-50236, doi:10.1074/jbc.M200958200 (2002). 32 Fu, S., Yang, L., Li, P., Hofmann, O., Dicker, L., Hide, W., Lin, X., Watkins, S. M., Ivanov, A. R. & Hotamisligil, G. S. Aberrant lipid metabolism disrupts calcium homeostasis causing liver endoplasmic reticulum stress in obesity. Nature 473, 528-531, doi:10.1038/nature09968 (2011). 33 Fu, S., Fan, J., Blanco, J., Gimenez-Cassina, A., Danial, N. N., Watkins, S. M. & Hotamisligil, G. S. Polysome profiling in liver identifies dynamic regulation of endoplasmic reticulum translatome by obesity and fasting. PLoS Genet 8, e1002902, doi:10.1371/journal.pgen.1002902 (2012). 166

34 Nakamura, T., Furuhashi, M., Li, P., Cao, H., Tuncman, G., Sonenberg, N., Gorgun, C. Z. & Hotamisligil, G. S. Double-stranded RNA-dependent protein kinase links pathogen sensing with stress and metabolic homeostasis. Cell 140, 338-348, doi:10.1016/j.cell.2010.01.001 (2010). 35 Hu, P., Han, Z., Couvillon, A. D., Kaufman, R. J. & Exton, J. H. Autocrine tumor necrosis factor alpha links endoplasmic reticulum stress to the membrane death receptor pathway through IRE1alpha-mediated NF-kappaB activation and down-regulation of TRAF2 expression. Molecular and cellular biology 26, 3071-3084, doi:10.1128/MCB.26.8.3071-3084.2006 (2006). 36 Huh, J. Y., Kim, J. I., Park, Y. J., Hwang, I. J., Lee, Y. S., Sohn, J. H., Lee, S. K., Alfadda, A. A., Kim, S. S., Choi, S. H., Lee, D. S., Park, S. H., Seong, R. H., Choi, C. S. & Kim, J. B. A novel function of adipocytes in lipid antigen presentation to iNKT cells. Molecular and cellular biology 33, 328-339, doi:10.1128/MCB.00552-12 (2013). 37 Shi, H., Kokoeva, M. V., Inouye, K., Tzameli, I., Yin, H. & Flier, J. S. TLR4 links innate immunity and fatty acid-induced insulin resistance. The Journal of clinical investigation 116, 3015-3025, doi:10.1172/JCI28898 (2006). 38 Kanda, H., Tateya, S., Tamori, Y., Kotani, K., Hiasa, K., Kitazawa, R., Kitazawa, S., Miyachi, H., Maeda, S., Egashira, K. & Kasuga, M. MCP-1 contributes to macrophage infiltration into adipose tissue, insulin resistance, and hepatic steatosis in obesity. The Journal of clinical investigation 116, 1494- 1505, doi:10.1172/JCI26498 (2006). 39 Hotamisligil, G. S., Shargill, N. S. & Spiegelman, B. M. Adipose expression of tumor necrosis factor-alpha: direct role in obesity-linked insulin resistance. Science 259, 87-91 (1993). 40 Oh, D. Y., Morinaga, H., Talukdar, S., Bae, E. J. & Olefsky, J. M. Increased macrophage migration into adipose tissue in obese mice. Diabetes 61, 346-354, doi:10.2337/db11-0860 (2012). 41 Eichmann, T. O. & Lass, A. DAG tales: the multiple faces of diacylglycerol-- stereochemistry, metabolism, and signaling. Cellular and molecular life sciences : CMLS 72, 3931-3952, doi:10.1007/s00018-015-1982-3 (2015). 42 Schmitz-Peiffer, C., Browne, C. L., Oakes, N. D., Watkinson, A., Chisholm, D. J., Kraegen, E. W. & Biden, T. J. Alterations in the expression and cellular localization of protein kinase C isozymes epsilon and theta are associated with 167

insulin resistance in skeletal muscle of the high-fat-fed rat. Diabetes 46, 169-178 (1997). 43 Steinberg, S. F. Structural basis of protein kinase C isoform function. Physiol Rev 88, 1341-1378, doi:10.1152/physrev.00034.2007 (2008). 44 Sano, H., Kane, S., Sano, E., Miinea, C. P., Asara, J. M., Lane, W. S., Garner, C. W. & Lienhard, G. E. Insulin-stimulated phosphorylation of a Rab GTPase- activating protein regulates GLUT4 translocation. The Journal of biological chemistry 278, 14599-14602, doi:10.1074/jbc.C300063200 (2003). 45 Li, Y., Soos, T. J., Li, X., Wu, J., Degennaro, M., Sun, X., Littman, D. R., Birnbaum, M. J. & Polakiewicz, R. D. Protein kinase C Theta inhibits insulin signaling by phosphorylating IRS1 at Ser(1101). The Journal of biological chemistry 279, 45304-45307, doi:10.1074/jbc.C400186200 (2004). 46 Samuel, V. T., Liu, Z. X., Wang, A., Beddow, S. A., Geisler, J. G., Kahn, M., Zhang, X. M., Monia, B. P., Bhanot, S. & Shulman, G. I. Inhibition of protein kinase Cepsilon prevents hepatic insulin resistance in nonalcoholic fatty liver disease. The Journal of clinical investigation 117, 739-745, doi:10.1172/JCI30400 (2007). 47 Adams, J. M., 2nd, Pratipanawatr, T., Berria, R., Wang, E., DeFronzo, R. A., Sullards, M. C. & Mandarino, L. J. Ceramide content is increased in skeletal muscle from obese insulin-resistant humans. Diabetes 53, 25-31 (2004). 48 Tonks, K. T., Coster, A. C., Christopher, M. J., Chaudhuri, R., Xu, A., Gagnon- Bartsch, J., Chisholm, D. J., James, D. E., Meikle, P. J., Greenfield, J. R. & Samocha-Bonet, D. Skeletal muscle and plasma lipidomic signatures of insulin resistance and overweight/obesity in humans. Obesity (Silver Spring) 24, 908- 916, doi:10.1002/oby.21448 (2016). 49 Luukkonen, P. K., Zhou, Y., Sadevirta, S., Leivonen, M., Arola, J., Oresic, M., Hyotylainen, T. & Yki-Jarvinen, H. Hepatic ceramides dissociate steatosis and insulin resistance in patients with non-alcoholic fatty liver disease. J Hepatol 64, 1167-1175, doi:10.1016/j.jhep.2016.01.002 (2016). 50 Summers, S. A., Garza, L. A., Zhou, H. & Birnbaum, M. J. Regulation of insulin-stimulated glucose transporter GLUT4 translocation and Akt kinase activity by ceramide. Molecular and cellular biology 18, 5457-5464 (1998). 51 Stratford, S., Hoehn, K. L., Liu, F. & Summers, S. A. Regulation of insulin action by ceramide: dual mechanisms linking ceramide accumulation to the 168

inhibition of Akt/protein kinase B. The Journal of biological chemistry 279, 36608-36615, doi:10.1074/jbc.M406499200 (2004). 52 Powell, D. J., Hajduch, E., Kular, G. & Hundal, H. S. Ceramide disables 3- phosphoinositide binding to the pleckstrin homology domain of protein kinase B (PKB)/Akt by a PKCzeta-dependent mechanism. Molecular and cellular biology 23, 7794-7808 (2003). 53 Shimabukuro, M., Higa, M., Zhou, Y. T., Wang, M. Y., Newgard, C. B. & Unger, R. H. Lipoapoptosis in beta-cells of obese prediabetic fa/fa rats. Role of serine palmitoyltransferase overexpression. The Journal of biological chemistry 273, 32487-32490 (1998). 54 Manukyan, L., Ubhayasekera, S. J., Bergquist, J., Sargsyan, E. & Bergsten, P. Palmitate-induced impairments of beta-cell function are linked with generation of specific ceramide species via acylation of sphingosine. Endocrinology 156, 802-812, doi:10.1210/en.2014-1467 (2015). 55 Boslem, E., MacIntosh, G., Preston, A. M., Bartley, C., Busch, A. K., Fuller, M., Laybutt, D. R., Meikle, P. J. & Biden, T. J. A lipidomic screen of palmitate- treated MIN6 beta-cells links sphingolipid metabolites with endoplasmic reticulum (ER) stress and impaired protein trafficking. The Biochemical journal 435, 267-276, doi:10.1042/BJ20101867 (2011). 56 Lei, X., Zhang, S., Emani, B., Barbour, S. E. & Ramanadham, S. A link between endoplasmic reticulum stress-induced β-cell apoptosis and the group VIA Ca2+- independent phospholipase A2 (iPLA2β). Diabetes, Obesity and Metabolism 12, 93-98, doi:10.1111/j.1463-1326.2010.01270.x (2010). 57 Bruce, C. R., Hoy, A. J., Turner, N., Watt, M. J., Allen, T. L., Carpenter, K., Cooney, G. J., Febbraio, M. A. & Kraegen, E. W. Overexpression of Carnitine Palmitoyltransferase-1 in Skeletal Muscle Is Sufficient to Enhance Fatty Acid Oxidation and Improve High-Fat Diet–Induced Insulin Resistance. Diabetes 58, 550-558, doi:10.2337/db08-1078 (2009). 58 Turpin, S. M., Nicholls, H. T., Willmes, D. M., Mourier, A., Brodesser, S., Wunderlich, C. M., Mauer, J., Xu, E., Hammerschmidt, P., Bronneke, H. S., Trifunovic, A., LoSasso, G., Wunderlich, F. T., Kornfeld, J. W., Bluher, M., Kronke, M. & Bruning, J. C. Obesity-induced CerS6-dependent C16:0 ceramide production promotes weight gain and glucose intolerance. Cell metabolism 20, 678-686, doi:10.1016/j.cmet.2014.08.002 (2014). 169

59 Raichur, S., Wang, S. T., Chan, P. W., Li, Y., Ching, J., Chaurasia, B., Dogra, S., Ohman, M. K., Takeda, K., Sugii, S., Pewzner-Jung, Y., Futerman, A. H. & Summers, S. A. CerS2 haploinsufficiency inhibits beta-oxidation and confers susceptibility to diet-induced steatohepatitis and insulin resistance. Cell metabolism 20, 687-695, doi:10.1016/j.cmet.2014.09.015 (2014). 60 de Mello, V. D., Lankinen, M., Schwab, U., Kolehmainen, M., Lehto, S., Seppanen-Laakso, T., Oresic, M., Pulkkinen, L., Uusitupa, M. & Erkkila, A. T. Link between plasma ceramides, inflammation and insulin resistance: association with serum IL-6 concentration in patients with coronary heart disease. Diabetologia 52, 2612-2615, doi:10.1007/s00125-009-1482-9 (2009). 61 Youm, Y. H., Adijiang, A., Vandanmagsar, B., Burk, D., Ravussin, A. & Dixit, V. D. Elimination of the NLRP3-ASC inflammasome protects against chronic obesity-induced pancreatic damage. Endocrinology 152, 4039-4045, doi:10.1210/en.2011-1326 (2011). 62 Berg, J. M., Tymoczko, J. L. & Stryer, L. Biochemistry. 7th edn, (W.H. Freeman, 2012). 63 Saponaro, C., Gaggini, M., Carli, F. & Gastaldelli, A. The Subtle Balance between Lipolysis and Lipogenesis: A Critical Point in Metabolic Homeostasis. Nutrients 7, 9453-9474, doi:10.3390/nu7115475 (2015). 64 Kamp, F. & Hamilton, J. A. How fatty acids of different chain length enter and leave cells by free diffusion. Prostaglandins, leukotrienes, and essential fatty acids 75, 149-159, doi:10.1016/j.plefa.2006.05.003 (2006). 65 Schwenk, R. W., Holloway, G. P., Luiken, J. J., Bonen, A. & Glatz, J. F. Fatty acid transport across the cell membrane: regulation by fatty acid transporters. Prostaglandins, leukotrienes, and essential fatty acids 82, 149-154, doi:10.1016/j.plefa.2010.02.029 (2010). 66 Thudichum, J. L. W. A Treatise on the Chemical Constitution of Brain. (Bailiere, Tindall and Cox, 1884). 67 Hannun, Y. A. & Obeid, L. M. Principles of bioactive lipid signalling: lessons from sphingolipids. Nat Rev Mol Cell Biol 9, 139-150, doi:10.1038/nrm2329 (2008). 68 Sabbadini, R. A. Targeting sphingosine-1-phosphate for cancer therapy. Br J Cancer 95, 1131-1135, doi:10.1038/sj.bjc.6603400 (2006).

170

69 Don, A. S., Lim, X. Y. & Couttas, T. A. Re-configuration of sphingolipid metabolism by oncogenic transformation. Biomolecules 4, 315-353, doi:10.3390/biom4010315 (2014). 70 Hannun, Y. A., Luberto, C. & Argraves, K. M. Enzymes of sphingolipid metabolism: from modular to integrative signaling. Biochemistry 40, 4893-4903 (2001). 71 Imgrund, S., Hartmann, D., Farwanah, H., Eckhardt, M., Sandhoff, R., Degen, J., Gieselmann, V., Sandhoff, K. & Willecke, K. Adult ceramide synthase 2 (CERS2)-deficient mice exhibit myelin sheath defects, cerebellar degeneration, and hepatocarcinomas. J Biol Chem 284, 33549-33560, doi:10.1074/jbc.M109.031971 (2009). 72 Santana, P., Pena, L. A., Haimovitz-Friedman, A., Martin, S., Green, D., McLoughlin, M., Cordon-Cardo, C., Schuchman, E. H., Fuks, Z. & Kolesnick, R. Acid sphingomyelinase-deficient human lymphoblasts and mice are defective in radiation-induced apoptosis. Cell 86, 189-199 (1996). 73 Duan, R. D. Alkaline sphingomyelinase: an old enzyme with novel implications. Biochim Biophys Acta 1761, 281-291, doi:10.1016/j.bbalip.2006.03.007 (2006). 74 Pruett, S. T., Bushnev, A., Hagedorn, K., Adiga, M., Haynes, C. A., Sullards, M. C., Liotta, D. C. & Merrill, A. H., Jr. Biodiversity of sphingoid bases ("sphingosines") and related amino alcohols. Journal of lipid research 49, 1621- 1639, doi:10.1194/jlr.R800012-JLR200 (2008). 75 Sandhoff, R., Geyer, R., Jennemann, R., Paret, C., Kiss, E., Yamashita, T., Gorgas, K., Sijmonsma, T. P., Iwamori, M., Finaz, C., Proia, R. L., Wiegandt, H. & Grone, H. J. Novel class of glycosphingolipids involved in male fertility. The Journal of biological chemistry 280, 27310-27318, doi:10.1074/jbc.M502775200 (2005). 76 Schnaar, R. L., Suzuki, A. & Stanley, P. in Essentials of Glycobiology (eds A. Varki et al.) (2009). 77 Merrill, A. H., Jr. Sphingolipid and glycosphingolipid metabolic pathways in the era of sphingolipidomics. Chem Rev 111, 6387-6422, doi:10.1021/cr2002917 (2011). 78 Park, J. W., Park, W. J. & Futerman, A. H. Ceramide synthases as potential targets for therapeutic intervention in human diseases. Biochim Biophys Acta 1841, 671-681, doi:10.1016/j.bbalip.2013.08.019 (2014). 171

79 Mizutani, Y., Kihara, A. & Igarashi, Y. Mammalian Lass6 and its related family members regulate synthesis of specific ceramides. The Biochemical journal 390, 263-271, doi:10.1042/BJ20050291 (2005). 80 Lahiri, S., Lee, H., Mesicek, J., Fuks, Z., Haimovitz-Friedman, A., Kolesnick, R. N. & Futerman, A. H. Kinetic characterization of mammalian ceramide synthases: determination of K(m) values towards sphinganine. FEBS Lett 581, 5289-5294, doi:10.1016/j.febslet.2007.10.018 (2007). 81 Hama, H. Fatty acid 2-Hydroxylation in mammalian sphingolipid biology. Biochim Biophys Acta 1801, 405-414, doi:10.1016/j.bbalip.2009.12.004 (2010). 82 Laviad, E. L., Albee, L., Pankova-Kholmyansky, I., Epstein, S., Park, H., Merrill, A. H., Jr. & Futerman, A. H. Characterization of ceramide synthase 2: tissue distribution, substrate specificity, and inhibition by sphingosine 1- phosphate. The Journal of biological chemistry 283, 5677-5684, doi:10.1074/jbc.M707386200 (2008). 83 Riebeling, C., Allegood, J. C., Wang, E., Merrill, A. H., Jr. & Futerman, A. H. Two mammalian longevity assurance gene (LAG1) family members, trh1 and trh4, regulate dihydroceramide synthesis using different fatty acyl-CoA donors. The Journal of biological chemistry 278, 43452-43459, doi:10.1074/jbc.M307104200 (2003). 84 Venkataraman, K., Riebeling, C., Bodennec, J., Riezman, H., Allegood, J. C., Sullards, M. C., Merrill, A. H., Jr. & Futerman, A. H. Upstream of growth and differentiation factor 1 (uog1), a mammalian homolog of the yeast longevity assurance gene 1 (LAG1), regulates N-stearoyl-sphinganine (C18- (dihydro)ceramide) synthesis in a fumonisin B1-independent manner in mammalian cells. The Journal of biological chemistry 277, 35642-35649, doi:10.1074/jbc.M205211200 (2002). 85 Guillas, I., Jiang, J. C., Vionnet, C., Roubaty, C., Uldry, D., Chuard, R., Wang, J., Jazwinski, S. M. & Conzelmann, A. Human homologues of LAG1 reconstitute Acyl-CoA-dependent ceramide synthesis in yeast. The Journal of biological chemistry 278, 37083-37091, doi:10.1074/jbc.M307554200 (2003). 86 D'Mello N, P., Childress, A. M., Franklin, D. S., Kale, S. P., Pinswasdi, C. & Jazwinski, S. M. Cloning and characterization of LAG1, a longevity-assurance gene in yeast. The Journal of biological chemistry 269, 15451-15459 (1994).

172

87 Jiang, J. C., Kirchman, P. A., Zagulski, M., Hunt, J. & Jazwinski, S. M. Homologs of the yeast longevity gene LAG1 in Caenorhabditis elegans and human. Genome Res 8, 1259-1272 (1998). 88 Venkataraman, K. & Futerman, A. H. Do longevity assurance genes containing Hox domains regulate cell development via ceramide synthesis? FEBS letters 528, 3-4 (2002). 89 Winter, E. & Ponting, C. P. TRAM, LAG1 and CLN8: members of a novel family of lipid-sensing domains? Trends Biochem Sci 27, 381-383 (2002). 90 Spassieva, S., Seo, J. G., Jiang, J. C., Bielawski, J., Alvarez-Vasquez, F., Jazwinski, S. M., Hannun, Y. A. & Obeid, L. M. Necessary role for the Lag1p motif in (dihydro)ceramide synthase activity. The Journal of biological chemistry 281, 33931-33938, doi:10.1074/jbc.M608092200 (2006). 91 Kageyama-Yahara, N. & Riezman, H. Transmembrane topology of ceramide synthase in yeast. Biochem J 398, 585-593, doi:10.1042/BJ20060697 (2006). 92 Lahiri, S., Park, H., Laviad, E. L., Lu, X., Bittman, R. & Futerman, A. H. Ceramide synthesis is modulated by the sphingosine analog FTY720 via a mixture of uncompetitive and noncompetitive inhibition in an Acyl-CoA chain length-dependent manner. The Journal of biological chemistry 284, 16090- 16098, doi:10.1074/jbc.M807438200 (2009). 93 Gehring, W. J., Affolter, M. & Burglin, T. Homeodomain proteins. Annual review of biochemistry 63, 487-526, doi:10.1146/annurev.bi.63.070194.002415 (1994). 94 Mesika, A., Ben-Dor, S., Laviad, E. L. & Futerman, A. H. A new functional motif in Hox domain-containing ceramide synthases: identification of a novel region flanking the Hox and TLC domains essential for activity. The Journal of biological chemistry 282, 27366-27373, doi:10.1074/jbc.M703487200 (2007). 95 Voelzmann, A., Wulf, A. L., Eckardt, F., Thielisch, M., Brondolin, M., Pesch, Y. Y., Sociale, M., Bauer, R. & Hoch, M. Nuclear Drosophila CerS Schlank regulates lipid homeostasis via the homeodomain, independent of the lag1p motif. FEBS letters 590, 971-981, doi:10.1002/1873-3468.12125 (2016). 96 Zhao, L., Spassieva, S. D., Jucius, T. J., Shultz, L. D., Shick, H. E., Macklin, W. B., Hannun, Y. A., Obeid, L. M. & Ackerman, S. L. A deficiency of ceramide biosynthesis causes cerebellar purkinje cell and lipofuscin

173

accumulation. PLoS Genet 7, e1002063, doi:10.1371/journal.pgen.1002063 (2011). 97 Ginkel, C., Hartmann, D., vom Dorp, K., Zlomuzica, A., Farwanah, H., Eckhardt, M., Sandhoff, R., Degen, J., Rabionet, M., Dere, E., Dormann, P., Sandhoff, K. & Willecke, K. Ablation of neuronal ceramide synthase 1 in mice decreases ganglioside levels and expression of myelin-associated glycoprotein in oligodendrocytes. The Journal of biological chemistry 287, 41888-41902, doi:10.1074/jbc.M112.413500 (2012). 98 Ben-David, O., Pewzner-Jung, Y., Brenner, O., Laviad, E. L., Kogot-Levin, A., Weissberg, I., Biton, I. E., Pienik, R., Wang, E., Kelly, S., Alroy, J., Raas- Rothschild, A., Friedman, A., Brugger, B., Merrill, A. H., Jr. & Futerman, A. H. Encephalopathy caused by ablation of very long acyl chain ceramide synthesis may be largely due to reduced galactosylceramide levels. The Journal of biological chemistry 286, 30022-30033, doi:10.1074/jbc.M111.261206 (2011). 99 Petrache, I., Kamocki, K., Poirier, C., Pewzner-Jung, Y., Laviad, E. L., Schweitzer, K. S., Van Demark, M., Justice, M. J., Hubbard, W. C. & Futerman, A. H. Ceramide synthases expression and role of ceramide synthase-2 in the lung: insight from human lung cells and mouse models. PloS one 8, e62968, doi:10.1371/journal.pone.0062968 (2013). 100 Pewzner-Jung, Y., Park, H., Laviad, E. L., Silva, L. C., Lahiri, S., Stiban, J., Erez-Roman, R., Brugger, B., Sachsenheimer, T., Wieland, F., Prieto, M., Merrill, A. H., Jr. & Futerman, A. H. A critical role for ceramide synthase 2 in liver homeostasis: I. alterations in lipid metabolic pathways. The Journal of biological chemistry 285, 10902-10910, doi:10.1074/jbc.M109.077594 (2010). 101 Zigdon, H., Kogot-Levin, A., Park, J. W., Goldschmidt, R., Kelly, S., Merrill, A. H., Jr., Scherz, A., Pewzner-Jung, Y., Saada, A. & Futerman, A. H. Ablation of ceramide synthase 2 causes chronic oxidative stress due to disruption of the mitochondrial respiratory chain. The Journal of biological chemistry 288, 4947- 4956, doi:10.1074/jbc.M112.402719 (2013). 102 Silva, L. C., Ben David, O., Pewzner-Jung, Y., Laviad, E. L., Stiban, J., Bandyopadhyay, S., Merrill, A. H., Jr., Prieto, M. & Futerman, A. H. Ablation of ceramide synthase 2 strongly affects biophysical properties of membranes. Journal of lipid research 53, 430-436, doi:10.1194/jlr.M022715 (2012).

174

103 Park, J. W., Park, W. J., Kuperman, Y., Boura-Halfon, S., Pewzner-Jung, Y. & Futerman, A. H. Ablation of very long acyl chain sphingolipids causes hepatic insulin resistance in mice due to altered detergent-resistant membranes. Hepatology 57, 525-532, doi:10.1002/hep.26015 (2013). 104 Pewzner-Jung, Y., Brenner, O., Braun, S., Laviad, E. L., Ben-Dor, S., Feldmesser, E., Horn-Saban, S., Amann-Zalcenstein, D., Raanan, C., Berkutzki, T., Erez-Roman, R., Ben-David, O., Levy, M., Holzman, D., Park, H., Nyska, A., Merrill, A. H., Jr. & Futerman, A. H. A critical role for ceramide synthase 2 in liver homeostasis: II. insights into molecular changes leading to hepatopathy. The Journal of biological chemistry 285, 10911-10923, doi:10.1074/jbc.M109.077610 (2010). 105 Park, W. J., Park, J. W., Erez-Roman, R., Kogot-Levin, A., Bame, J. R., Tirosh, B., Saada, A., Merrill, A. H., Jr., Pewzner-Jung, Y. & Futerman, A. H. Protection of a ceramide synthase 2 null mouse from drug-induced liver injury: role of gap junction dysfunction and connexin 32 mislocalization. The Journal of biological chemistry 288, 30904-30916, doi:10.1074/jbc.M112.448852 (2013). 106 Jennemann, R., Rabionet, M., Gorgas, K., Epstein, S., Dalpke, A., Rothermel, U., Bayerle, A., van der Hoeven, F., Imgrund, S., Kirsch, J., Nickel, W., Willecke, K., Riezman, H., Grone, H. J. & Sandhoff, R. Loss of ceramide synthase 3 causes lethal skin barrier disruption. Hum Mol Genet 21, 586-608, doi:10.1093/hmg/ddr494 (2012). 107 Ebel, P., Imgrund, S., Vom Dorp, K., Hofmann, K., Maier, H., Drake, H., Degen, J., Dormann, P., Eckhardt, M., Franz, T. & Willecke, K. deficiency in mice causes lipid alterations in sebum and results in alopecia. The Biochemical journal 461, 147-158, doi:10.1042/BJ20131242 (2014). 108 Gosejacob, D., Jager, P. S., Vom Dorp, K., Frejno, M., Carstensen, A. C., Kohnke, M., Degen, J., Dormann, P. & Hoch, M. Is Essential to Maintain C16:0-Ceramide Pools and Contributes to the Development of Diet-induced Obesity. The Journal of biological chemistry 291, 6989-7003, doi:10.1074/jbc.M115.691212 (2016). 109 Ebel, P., Vom Dorp, K., Petrasch-Parwez, E., Zlomuzica, A., Kinugawa, K., Mariani, J., Minich, D., Ginkel, C., Welcker, J., Degen, J., Eckhardt, M., Dere, 175

E., Dormann, P. & Willecke, K. Inactivation of ceramide synthase 6 in mice results in an altered sphingolipid metabolism and behavioral abnormalities. The Journal of biological chemistry 288, 21433-21447, doi:10.1074/jbc.M113.479907 (2013). 110 Sridevi, P., Alexander, H., Laviad, E. L., Pewzner-Jung, Y., Hannink, M., Futerman, A. H. & Alexander, S. Ceramide synthase 1 is regulated by proteasomal mediated turnover. Biochim Biophys Acta 1793, 1218-1227, doi:10.1016/j.bbamcr.2009.04.006 (2009). 111 Sridevi, P., Alexander, H., Laviad, E. L., Min, J., Mesika, A., Hannink, M., Futerman, A. H. & Alexander, S. Stress-induced ER to Golgi translocation of ceramide synthase 1 is dependent on proteasomal processing. Experimental cell research 316, 78-91, doi:10.1016/j.yexcr.2009.09.027 (2010). 112 Levy, M. & Futerman, A. H. Mammalian ceramide synthases. IUBMB Life 62, 347-356, doi:10.1002/iub.319 (2010). 113 Becker, I., Wang-Eckhardt, L., Yaghootfam, A., Gieselmann, V. & Eckhardt, M. Differential expression of (dihydro)ceramide synthases in mouse brain: oligodendrocyte-specific expression of CerS2/Lass2. Histochem Cell Biol 129, 233-241, doi:10.1007/s00418-007-0344-0 (2008). 114 Koybasi, S., Senkal, C. E., Sundararaj, K., Spassieva, S., Bielawski, J., Osta, W., Day, T. A., Jiang, J. C., Jazwinski, S. M., Hannun, Y. A., Obeid, L. M. & Ogretmen, B. Defects in cell growth regulation by C18:0-ceramide and longevity assurance gene 1 in human head and neck squamous cell carcinomas. The Journal of biological chemistry 279, 44311-44319, doi:10.1074/jbc.M406920200 (2004). 115 Karahatay, S., Thomas, K., Koybasi, S., Senkal, C. E., Elojeimy, S., Liu, X., Bielawski, J., Day, T. A., Gillespie, M. B., Sinha, D., Norris, J. S., Hannun, Y. A. & Ogretmen, B. Clinical relevance of ceramide metabolism in the pathogenesis of human head and neck squamous cell carcinoma (HNSCC): attenuation of C(18)-ceramide in HNSCC tumors correlates with lymphovascular invasion and nodal metastasis. Cancer letters 256, 101-111, doi:10.1016/j.canlet.2007.06.003 (2007). 116 Min, J., Mesika, A., Sivaguru, M., Van Veldhoven, P. P., Alexander, H., Futerman, A. H. & Alexander, S. (Dihydro)ceramide synthase 1 regulated sensitivity to cisplatin is associated with the activation of p38 mitogen-activated 176

protein kinase and is abrogated by sphingosine kinase 1. Molecular cancer research : MCR 5, 801-812, doi:10.1158/1541-7786.MCR-07-0100 (2007). 117 Senkal, C. E., Ponnusamy, S., Rossi, M. J., Bialewski, J., Sinha, D., Jiang, J. C., Jazwinski, S. M., Hannun, Y. A. & Ogretmen, B. Role of human longevity assurance gene 1 and C18-ceramide in chemotherapy-induced cell death in human head and neck squamous cell carcinomas. Molecular cancer therapeutics 6, 712-722, doi:10.1158/1535-7163.MCT-06-0558 (2007). 118 Couttas, T. A., Kain, N., Daniels, B., Lim, X. Y., Shepherd, C., Kril, J., Pickford, R., Li, H., Garner, B. & Don, A. S. Loss of the neuroprotective factor Sphingosine 1-phosphate early in Alzheimer's disease pathogenesis. Acta Neuropathol Commun 2, 9, doi:10.1186/2051-5960-2-9 (2014). 119 Spassieva, S. D., Mullen, T. D., Townsend, D. M. & Obeid, L. M. Disruption of ceramide synthesis by CerS2 down-regulation leads to autophagy and the unfolded protein response. The Biochemical journal 424, 273-283, doi:10.1042/BJ20090699 (2009). 120 Mizutani, Y., Kihara, A. & Igarashi, Y. LASS3 (longevity assurance homologue 3) is a mainly testis-specific (dihydro)ceramide synthase with relatively broad substrate specificity. The Biochemical journal 398, 531-538, doi:10.1042/BJ20060379 (2006). 121 Rabionet, M., van der Spoel, A. C., Chuang, C. C., von Tumpling-Radosta, B., Litjens, M., Bouwmeester, D., Hellbusch, C. C., Korner, C., Wiegandt, H., Gorgas, K., Platt, F. M., Grone, H. J. & Sandhoff, R. Male germ cells require polyenoic sphingolipids with complex glycosylation for completion of meiosis: a link to ceramide synthase-3. The Journal of biological chemistry 283, 13357- 13369, doi:10.1074/jbc.M800870200 (2008). 122 Mizutani, Y., Kihara, A., Chiba, H., Tojo, H. & Igarashi, Y. 2-Hydroxy- ceramide synthesis by ceramide synthase family: enzymatic basis for the preference of FA chain length. Journal of lipid research 49, 2356-2364, doi:10.1194/jlr.M800158-JLR200 (2008). 123 Coderch, L., Lopez, O., de la Maza, A. & Parra, J. L. Ceramides and skin function. Am J Clin Dermatol 4, 107-129 (2003). 124 Han, X. & Cheng, H. Characterization and direct quantitation of cerebroside molecular species from lipid extracts by shotgun lipidomics. Journal of lipid research 46, 163-175, doi:10.1194/jlr.D400022-JLR200 (2005). 177

125 Ponec, M., Weerheim, A., Kempenaar, J., Mulder, A., Gooris, G. S., Bouwstra, J. & Mommaas, A. M. The formation of competent barrier lipids in reconstructed human epidermis requires the presence of vitamin C. The Journal of investigative dermatology 109, 348-355 (1997). 126 Peters, F., Vorhagen, S., Brodesser, S., Jakobshagen, K., Bruning, J. C., Niessen, C. M. & Kronke, M. Ceramide synthase 4 regulates stem cell homeostasis and hair follicle cycling. The Journal of investigative dermatology 135, 1501-1509, doi:10.1038/jid.2015.60 (2015). 127 Chen, J., Li, X., Ma, D., Liu, T., Tian, P. & Wu, C. Ceramide synthase-4 orchestrates the cell proliferation and tumor growth of liver cancer in vitro and in vivo through the nuclear factor-kappaB signaling pathway. Oncol Lett 14, 1477-1483, doi:10.3892/ol.2017.6365 (2017). 128 Lahiri, S. & Futerman, A. H. LASS5 is a bona fide dihydroceramide synthase that selectively utilizes palmitoyl-CoA as acyl donor. The Journal of biological chemistry 280, 33735-33738, doi:10.1074/jbc.M506485200 (2005). 129 Becker, K. P., Kitatani, K., Idkowiak-Baldys, J., Bielawski, J. & Hannun, Y. A. Selective inhibition of juxtanuclear translocation of protein kinase C betaII by a negative feedback mechanism involving ceramide formed from the salvage pathway. The Journal of biological chemistry 280, 2606-2612, doi:10.1074/jbc.M409066200 (2005). 130 Kitatani, K., Idkowiak-Baldys, J., Bielawski, J., Taha, T. A., Jenkins, R. W., Senkal, C. E., Ogretmen, B., Obeid, L. M. & Hannun, Y. A. Protein kinase C- induced activation of a ceramide/protein phosphatase 1 pathway leading to dephosphorylation of p38 MAPK. The Journal of biological chemistry 281, 36793-36802, doi:10.1074/jbc.M608137200 (2006). 131 Jin, J., Hou, Q., Mullen, T. D., Zeidan, Y. H., Bielawski, J., Kraveka, J. M., Bielawska, A., Obeid, L. M., Hannun, Y. A. & Hsu, Y. T. Ceramide generated by sphingomyelin hydrolysis and the salvage pathway is involved in hypoxia/reoxygenation-induced Bax redistribution to mitochondria in NT-2 cells. The Journal of biological chemistry 283, 26509-26517, doi:10.1074/jbc.M801597200 (2008). 132 Russo, S. B., Baicu, C. F., Van Laer, A., Geng, T., Kasiganesan, H., Zile, M. R. & Cowart, L. A. Ceramide synthase 5 mediates lipid-induced autophagy and

178

hypertrophy in cardiomyocytes. The Journal of clinical investigation 122, 3919- 3930, doi:10.1172/JCI63888 (2012). 133 Weinmann, A., Galle, P. R. & Teufel, A. LASS6, an additional member of the longevity assurance gene family. Int J Mol Med 16, 905-910 (2005). 134 Ruckhaberle, E., Rody, A., Engels, K., Gaetje, R., von Minckwitz, G., Schiffmann, S., Grosch, S., Geisslinger, G., Holtrich, U., Karn, T. & Kaufmann, M. Microarray analysis of altered sphingolipid metabolism reveals prognostic significance of sphingosine kinase 1 in breast cancer. Breast cancer research and treatment 112, 41-52, doi:10.1007/s10549-007-9836-9 (2008). 135 Schiffmann, S., Sandner, J., Birod, K., Wobst, I., Angioni, C., Ruckhaberle, E., Kaufmann, M., Ackermann, H., Lotsch, J., Schmidt, H., Geisslinger, G. & Grosch, S. Ceramide synthases and ceramide levels are increased in breast cancer tissue. Carcinogenesis 30, 745-752, doi:10.1093/carcin/bgp061 (2009). 136 Erez-Roman, R., Pienik, R. & Futerman, A. H. Increased ceramide synthase 2 and 6 mRNA levels in breast cancer tissues and correlation with sphingosine kinase expression. Biochemical and biophysical research communications 391, 219-223, doi:10.1016/j.bbrc.2009.11.035 (2010). 137 Walker, T., Mitchell, C., Park, M. A., Yacoub, A., Graf, M., Rahmani, M., Houghton, P. J., Voelkel-Johnson, C., Grant, S. & Dent, P. Sorafenib and vorinostat kill colon cancer cells by CD95-dependent and -independent mechanisms. Molecular pharmacology 76, 342-355, doi:10.1124/mol.109.056523 (2009). 138 Obeid, L. M., Linardic, C. M., Karolak, L. A. & Hannun, Y. A. Programmed cell death induced by ceramide. Science 259, 1769-1771 (1993). 139 Chalfant, C. E., Szulc, Z., Roddy, P., Bielawska, A. & Hannun, Y. A. The structural requirements for ceramide activation of serine-threonine protein phosphatases. J Lipid Res 45, 496-506, doi:10.1194/jlr.M300347-JLR200 (2004). 140 Gao, X., Lowry, P. R., Zhou, X., Depry, C., Wei, Z., Wong, G. W. & Zhang, J. PI3K/Akt signaling requires spatial compartmentalization in plasma membrane microdomains. Proceedings of the National Academy of Sciences of the United States of America 108, 14509-14514, doi:10.1073/pnas.1019386108 (2011). 141 Colombini, M. Membrane channels formed by ceramide. Handb Exp Pharmacol, 109-126, doi:10.1007/978-3-7091-1368-4_6 (2013). 179

142 Xin, M. & Deng, X. Protein phosphatase 2A enhances the proapoptotic function of Bax through dephosphorylation. The Journal of biological chemistry 281, 18859-18867, doi:10.1074/jbc.M512543200 (2006). 143 Chalfant, C. E., Rathman, K., Pinkerman, R. L., Wood, R. E., Obeid, L. M., Ogretmen, B. & Hannun, Y. A. De novo ceramide regulates the alternative splicing of caspase 9 and Bcl-x in A549 lung adenocarcinoma cells. Dependence on protein phosphatase-1. The Journal of biological chemistry 277, 12587- 12595, doi:10.1074/jbc.M112010200 (2002). 144 Wang, G., Silva, J., Krishnamurthy, K., Tran, E., Condie, B. G. & Bieberich, E. Direct binding to ceramide activates protein kinase Czeta before the formation of a pro-apoptotic complex with PAR-4 in differentiating stem cells. The Journal of biological chemistry 280, 26415-26424, doi:10.1074/jbc.M501492200 (2005). 145 Lozano, J., Berra, E., Municio, M. M., Diaz-Meco, M. T., Dominguez, I., Sanz, L. & Moscat, J. Protein kinase C zeta isoform is critical for kappa B-dependent promoter activation by sphingomyelinase. The Journal of biological chemistry 269, 19200-19202 (1994). 146 Venkataraman, K. & Futerman, A. H. Ceramide as a second messenger: sticky solutions to sticky problems. Trends in cell biology 10, 408-412 (2000). 147 Kolesnick, R. N., Goni, F. M. & Alonso, A. Compartmentalization of ceramide signaling: physical foundations and biological effects. J Cell Physiol 184, 285- 300, doi:10.1002/1097-4652(200009)184:3<285::AID-JCP2>3.0.CO;2-3 (2000). 148 Holopainen, J. M., Subramanian, M. & Kinnunen, P. K. Sphingomyelinase induces lipid microdomain formation in a fluid phosphatidylcholine/sphingomyelin membrane. Biochemistry 37, 17562-17570, doi:10.1021/bi980915e (1998). 149 Segui, B., Andrieu-Abadie, N., Adam-Klages, S., Meilhac, O., Kreder, D., Garcia, V., Bruno, A. P., Jaffrezou, J. P., Salvayre, R., Kronke, M. & Levade, T. CD40 signals apoptosis through FAN-regulated activation of the sphingomyelin- ceramide pathway. The Journal of biological chemistry 274, 37251-37258 (1999). 150 Grassme, H., Jekle, A., Riehle, A., Schwarz, H., Berger, J., Sandhoff, K., Kolesnick, R. & Gulbins, E. CD95 signaling via ceramide-rich membrane rafts. 180

The Journal of biological chemistry 276, 20589-20596, doi:10.1074/jbc.M101207200 (2001). 151 Paris, F., Grassme, H., Cremesti, A., Zager, J., Fong, Y., Haimovitz-Friedman, A., Fuks, Z., Gulbins, E. & Kolesnick, R. Natural ceramide reverses Fas resistance of acid sphingomyelinase(-/-) hepatocytes. J Biol Chem 276, 8297- 8305, doi:10.1074/jbc.M008732200 (2001). 152 Grassme, H., Jendrossek, V., Bock, J., Riehle, A. & Gulbins, E. Ceramide-rich membrane rafts mediate CD40 clustering. Journal of immunology 168, 298-307 (2002). 153 Hueber, A. O., Bernard, A. M., Herincs, Z., Couzinet, A. & He, H. T. An essential role for membrane rafts in the initiation of Fas/CD95-triggered cell death in mouse thymocytes. EMBO Rep 3, 190-196, doi:10.1093/embo- reports/kvf022 (2002). 154 Vidalain, P. O., Azocar, O., Servet-Delprat, C., Rabourdin-Combe, C., Gerlier, D. & Manie, S. CD40 signaling in human dendritic cells is initiated within membrane rafts. The EMBO journal 19, 3304-3313, doi:10.1093/emboj/19.13.3304 (2000). 155 Goni, F. M., Sot, J. & Alonso, A. Biophysical properties of sphingosine, ceramides and other simple sphingolipids. Biochemical Society transactions 42, 1401-1408, doi:10.1042/BST20140159 (2014). 156 Castro, B. M., Prieto, M. & Silva, L. C. Ceramide: a simple sphingolipid with unique biophysical properties. Prog Lipid Res 54, 53-67, doi:10.1016/j.plipres.2014.01.004 (2014). 157 Holopainen, J. M., Angelova, M. I. & Kinnunen, P. K. Vectorial budding of vesicles by asymmetrical enzymatic formation of ceramide in giant liposomes. Biophysical journal 78, 830-838, doi:10.1016/S0006-3495(00)76640-9 (2000). 158 Grassme, H., Gulbins, E., Brenner, B., Ferlinz, K., Sandhoff, K., Harzer, K., Lang, F. & Meyer, T. F. Acidic sphingomyelinase mediates entry of N. gonorrhoeae into nonphagocytic cells. Cell 91, 605-615 (1997). 159 Tepper, A. D., Ruurs, P., Wiedmer, T., Sims, P. J., Borst, J. & van Blitterswijk, W. J. Sphingomyelin hydrolysis to ceramide during the execution phase of apoptosis results from phospholipid scrambling and alters cell-surface morphology. The Journal of cell biology 150, 155-164 (2000).

181

160 Lopez-Montero, I., Rodriguez, N., Cribier, S., Pohl, A., Velez, M. & Devaux, P. F. Rapid transbilayer movement of ceramides in phospholipid vesicles and in human erythrocytes. J Biol Chem 280, 25811-25819, doi:10.1074/jbc.M412052200 (2005). 161 Sentelle, R. D., Senkal, C. E., Jiang, W., Ponnusamy, S., Gencer, S., Selvam, S. P., Ramshesh, V. K., Peterson, Y. K., Lemasters, J. J., Szulc, Z. M., Bielawski, J. & Ogretmen, B. Ceramide targets autophagosomes to mitochondria and induces lethal mitophagy. Nature chemical biology 8, 831-838, doi:10.1038/nchembio.1059 (2012). 162 Yin, X., Zafrullah, M., Lee, H., Haimovitz-Friedman, A., Fuks, Z. & Kolesnick, R. A ceramide-binding C1 domain mediates kinase suppressor of ras membrane translocation. Cell Physiol Biochem 24, 219-230, doi:10.1159/000233248 (2009). 163 Zhang, Y., Yao, B., Delikat, S., Bayoumy, S., Lin, X. H., Basu, S., McGinley, M., Chan-Hui, P. Y., Lichenstein, H. & Kolesnick, R. Kinase suppressor of Ras is ceramide-activated protein kinase. Cell 89, 63-72 (1997). 164 Senkal, C. E., Ponnusamy, S., Bielawski, J., Hannun, Y. A. & Ogretmen, B. Antiapoptotic roles of ceramide-synthase-6-generated C16-ceramide via selective regulation of the ATF6/CHOP arm of ER-stress-response pathways. FASEB journal : official publication of the Federation of American Societies for Experimental Biology 24, 296-308, doi:10.1096/fj.09-135087 (2010). 165 Long, S. D. & Pekala, P. H. Lipid mediators of insulin resistance: ceramide signalling down-regulates GLUT4 gene transcription in 3T3-L1 adipocytes. The Biochemical journal 319 ( Pt 1), 179-184 (1996). 166 Schmitz-Peiffer, C., Craig, D. L. & Biden, T. J. Ceramide generation is sufficient to account for the inhibition of the insulin-stimulated PKB pathway in C2C12 skeletal muscle cells pretreated with palmitate. The Journal of biological chemistry 274, 24202-24210 (1999). 167 Hajduch, E., Balendran, A., Batty, I. H., Litherland, G. J., Blair, A. S., Downes, C. P. & Hundal, H. S. Ceramide impairs the insulin-dependent membrane recruitment of protein kinase B leading to a loss in downstream signalling in L6 skeletal muscle cells. Diabetologia 44, 173-183, doi:10.1007/s001250051596 (2001).

182

168 Chavez, J. A., Knotts, T. A., Wang, L. P., Li, G., Dobrowsky, R. T., Florant, G. L. & Summers, S. A. A role for ceramide, but not diacylglycerol, in the antagonism of insulin signal transduction by saturated fatty acids. The Journal of biological chemistry 278, 10297-10303, doi:10.1074/jbc.M212307200 (2003). 169 Hajduch, E., Turban, S., Le Liepvre, X., Le Lay, S., Lipina, C., Dimopoulos, N., Dugail, I. & Hundal, H. S. Targeting of PKCzeta and PKB to caveolin-enriched microdomains represents a crucial step underpinning the disruption in PKB- directed signalling by ceramide. The Biochemical journal 410, 369-379, doi:10.1042/BJ20070936 (2008). 170 Sussman, M. A., Volkers, M., Fischer, K., Bailey, B., Cottage, C. T., Din, S., Gude, N., Avitabile, D., Alvarez, R., Sundararaman, B., Quijada, P., Mason, M., Konstandin, M. H., Malhowski, A., Cheng, Z., Khan, M. & McGregor, M. Myocardial AKT: the omnipresent nexus. Physiol Rev 91, 1023-1070, doi:10.1152/physrev.00024.2010 (2011). 171 Hage Hassan, R., Pacheco de Sousa, A. C., Mahfouz, R., Hainault, I., Blachnio- Zabielska, A., Bourron, O., Koskas, F., Gorski, J., Ferre, P., Foufelle, F. & Hajduch, E. Sustained Action of Ceramide on the Insulin Signaling Pathway in Muscle Cells: IMPLICATION OF THE DOUBLE-STRANDED RNA- ACTIVATED PROTEIN KINASE. The Journal of biological chemistry 291, 3019-3029, doi:10.1074/jbc.M115.686949 (2016). 172 Samuel, V. T. & Shulman, G. I. Mechanisms for insulin resistance: common threads and missing links. Cell 148, 852-871, doi:10.1016/j.cell.2012.02.017 (2012). 173 Hsieh, C. T., Chuang, J. H., Yang, W. C., Yin, Y. & Lin, Y. Ceramide inhibits insulin-stimulated Akt phosphorylation through activation of Rheb/mTORC1/S6K signaling in skeletal muscle. Cellular signalling 26, 1400- 1408, doi:10.1016/j.cellsig.2014.03.004 (2014). 174 Jaswal, J. S., Keung, W., Wang, W., Ussher, J. R. & Lopaschuk, G. D. Targeting fatty acid and carbohydrate oxidation--a novel therapeutic intervention in the ischemic and failing heart. Biochim Biophys Acta 1813, 1333-1350, doi:10.1016/j.bbamcr.2011.01.015 (2011). 175 Park, T. S., Hu, Y., Noh, H. L., Drosatos, K., Okajima, K., Buchanan, J., Tuinei, J., Homma, S., Jiang, X. C., Abel, E. D. & Goldberg, I. J. Ceramide is a

183

cardiotoxin in lipotoxic cardiomyopathy. Journal of lipid research 49, 2101- 2112, doi:10.1194/jlr.M800147-JLR200 (2008). 176 Di Paola, M., Cocco, T. & Lorusso, M. Ceramide interaction with the respiratory chain of heart mitochondria. Biochemistry 39, 6660-6668 (2000). 177 Gudz, T. I., Tserng, K. Y. & Hoppel, C. L. Direct inhibition of mitochondrial respiratory chain complex III by cell-permeable ceramide. The Journal of biological chemistry 272, 24154-24158 (1997). 178 Nelson, M. B., Swensen, A. C., Winden, D. R., Bodine, J. S., Bikman, B. T. & Reynolds, P. R. Cardiomyocyte mitochondrial respiration is reduced by receptor for advanced glycation end-product signaling in a ceramide-dependent manner. Am J Physiol Heart Circ Physiol 309, H63-69, doi:10.1152/ajpheart.00043.2015 (2015). 179 Masharani, U. B., Maddux, B. A., Li, X., Sakkas, G. K., Mulligan, K., Schambelan, M., Goldfine, I. D. & Youngren, J. F. Insulin resistance in non- obese subjects is associated with activation of the JNK pathway and impaired insulin signaling in skeletal muscle. PloS one 6, e19878, doi:10.1371/journal.pone.0019878 (2011). 180 Yu, J., Novgorodov, S. A., Chudakova, D., Zhu, H., Bielawska, A., Bielawski, J., Obeid, L. M., Kindy, M. S. & Gudz, T. I. JNK3 signaling pathway activates ceramide synthase leading to mitochondrial dysfunction. The Journal of biological chemistry 282, 25940-25949, doi:10.1074/jbc.M701812200 (2007). 181 Smith, M. E., Tippetts, T. S., Brassfield, E. S., Tucker, B. J., Ockey, A., Swensen, A. C., Anthonymuthu, T. S., Washburn, T. D., Kane, D. A., Prince, J. T. & Bikman, B. T. Mitochondrial fission mediates ceramide-induced metabolic disruption in skeletal muscle. The Biochemical journal 456, 427-439, doi:10.1042/BJ20130807 (2013). 182 Novgorodov, S. A., Riley, C. L., Yu, J., Keffler, J. A., Clarke, C. J., Van Laer, A. O., Baicu, C. F., Zile, M. R. & Gudz, T. I. Lactosylceramide contributes to mitochondrial dysfunction in diabetes. Journal of lipid research 57, 546-562, doi:10.1194/jlr.M060061 (2016). 183 Hotamisligil, G. S. Inflammation and metabolic disorders. Nature 444, 860-867, doi:10.1038/nature05485 (2006). 184 Hotamisligil, G. S., Murray, D. L., Choy, L. N. & Spiegelman, B. M. Tumor necrosis factor alpha inhibits signaling from the insulin receptor. Proceedings of 184

the National Academy of Sciences of the United States of America 91, 4854- 4858 (1994). 185 Samad, F., Hester, K. D., Yang, G., Hannun, Y. A. & Bielawski, J. Altered adipose and plasma sphingolipid metabolism in obesity: a potential mechanism for cardiovascular and metabolic risk. Diabetes 55, 2579-2587, doi:10.2337/db06-0330 (2006). 186 Shin, K. C., Hwang, I., Choe, S. S., Park, J., Ji, Y., Kim, J. I., Lee, G. Y., Choi, S. H., Ching, J., Kovalik, J. P. & Kim, J. B. Macrophage VLDLR mediates obesity-induced insulin resistance with adipose tissue inflammation. Nat Commun 8, 1087, doi:10.1038/s41467-017-01232-w (2017). 187 Vandanmagsar, B., Youm, Y. H., Ravussin, A., Galgani, J. E., Stadler, K., Mynatt, R. L., Ravussin, E., Stephens, J. M. & Dixit, V. D. The NLRP3 inflammasome instigates obesity-induced inflammation and insulin resistance. Nat Med 17, 179-188, doi:10.1038/nm.2279 (2011). 188 Mitsutake, S., Date, T., Yokota, H., Sugiura, M., Kohama, T. & Igarashi, Y. Ceramide kinase deficiency improves diet-induced obesity and insulin resistance. FEBS letters 586, 1300-1305, doi:10.1016/j.febslet.2012.03.032 (2012). 189 Holland, W. L., Miller, R. A., Wang, Z. V., Sun, K., Barth, B. M., Bui, H. H., Davis, K. E., Bikman, B. T., Halberg, N., Rutkowski, J. M., Wade, M. R., Tenorio, V. M., Kuo, M. S., Brozinick, J. T., Zhang, B. B., Birnbaum, M. J., Summers, S. A. & Scherer, P. E. Receptor-mediated activation of ceramidase activity initiates the pleiotropic actions of adiponectin. Nat Med 17, 55-63, doi:10.1038/nm.2277 (2011). 190 Kim, J. Y., van de Wall, E., Laplante, M., Azzara, A., Trujillo, M. E., Hofmann, S. M., Schraw, T., Durand, J. L., Li, H., Li, G., Jelicks, L. A., Mehler, M. F., Hui, D. Y., Deshaies, Y., Shulman, G. I., Schwartz, G. J. & Scherer, P. E. Obesity-associated improvements in metabolic profile through expansion of adipose tissue. The Journal of clinical investigation 117, 2621-2637, doi:10.1172/JCI31021 (2007). 191 Yamauchi, T., Kamon, J., Minokoshi, Y., Ito, Y., Waki, H., Uchida, S., Yamashita, S., Noda, M., Kita, S., Ueki, K., Eto, K., Akanuma, Y., Froguel, P., Foufelle, F., Ferre, P., Carling, D., Kimura, S., Nagai, R., Kahn, B. B. & Kadowaki, T. Adiponectin stimulates glucose utilization and fatty-acid oxidation 185

by activating AMP-activated protein kinase. Nat Med 8, 1288-1295, doi:10.1038/nm788 (2002). 192 Kharitonenkov, A. & Larsen, P. FGF21 reloaded: challenges of a rapidly growing field. Trends Endocrinol Metab 22, 81-86, doi:10.1016/j.tem.2010.11.003 (2011). 193 Holland, W. L., Adams, A. C., Brozinick, J. T., Bui, H. H., Miyauchi, Y., Kusminski, C. M., Bauer, S. M., Wade, M., Singhal, E., Cheng, C. C., Volk, K., Kuo, M. S., Gordillo, R., Kharitonenkov, A. & Scherer, P. E. An FGF21- adiponectin-ceramide axis controls energy expenditure and insulin action in mice. Cell metabolism 17, 790-797, doi:10.1016/j.cmet.2013.03.019 (2013). 194 Summers, S. A. & Nelson, D. H. A role for sphingolipids in producing the common features of type 2 diabetes, metabolic syndrome X, and Cushing's syndrome. Diabetes 54, 591-602 (2005). 195 Hojjati, M. R., Li, Z., Zhou, H., Tang, S., Huan, C., Ooi, E., Lu, S. & Jiang, X. C. Effect of myriocin on plasma sphingolipid metabolism and atherosclerosis in apoE-deficient mice. The Journal of biological chemistry 280, 10284-10289, doi:10.1074/jbc.M412348200 (2005). 196 Kasumov, T., Li, L., Li, M., Gulshan, K., Kirwan, J. P., Liu, X., Previs, S., Willard, B., Smith, J. D. & McCullough, A. Ceramide as a mediator of non- alcoholic Fatty liver disease and associated atherosclerosis. PloS one 10, e0126910, doi:10.1371/journal.pone.0126910 (2015). 197 Li, W., Yang, X., Xing, S., Bian, F., Yao, W., Bai, X., Zheng, T., Wu, G. & Jin, S. Endogenous ceramide contributes to the transcytosis of oxLDL across endothelial cells and promotes its subendothelial retention in vascular wall. Oxid Med Cell Longev 2014, 823071, doi:10.1155/2014/823071 (2014). 198 Gao, D., Pararasa, C., Dunston, C. R., Bailey, C. J. & Griffiths, H. R. Palmitate promotes monocyte atherogenicity via de novo ceramide synthesis. Free Radic Biol Med 53, 796-806, doi:10.1016/j.freeradbiomed.2012.05.026 (2012). 199 Zhang, Q. J., Holland, W. L., Wilson, L., Tanner, J. M., Kearns, D., Cahoon, J. M., Pettey, D., Losee, J., Duncan, B., Gale, D., Kowalski, C. A., Deeter, N., Nichols, A., Deesing, M., Arrant, C., Ruan, T., Boehme, C., McCamey, D. R., Rou, J., Ambal, K., Narra, K. K., Summers, S. A., Abel, E. D. & Symons, J. D. Ceramide mediates vascular dysfunction in diet-induced obesity by PP2A-

186

mediated dephosphorylation of the eNOS-Akt complex. Diabetes 61, 1848- 1859, doi:10.2337/db11-1399 (2012). 200 Wu, Y., Song, P., Xu, J., Zhang, M. & Zou, M. H. Activation of protein phosphatase 2A by palmitate inhibits AMP-activated protein kinase. The Journal of biological chemistry 282, 9777-9788, doi:10.1074/jbc.M608310200 (2007). 201 Kronke, M. Biophysics of ceramide signaling: interaction with proteins and phase transition of membranes. Chem Phys Lipids 101, 109-121 (1999). 202 Pan, W., Yu, J., Shi, R., Yan, L., Yang, T., Li, Y., Zhang, Z., Yu, G., Bai, Y., Schuchman, E. H., He, X. & Zhang, G. Elevation of ceramide and activation of secretory acid sphingomyelinase in patients with acute coronary syndromes. Coron Artery Dis 25, 230-235, doi:10.1097/MCA.0000000000000079 (2014). 203 Cheng, J. M., Suoniemi, M., Kardys, I., Vihervaara, T., de Boer, S. P., Akkerhuis, K. M., Sysi-Aho, M., Ekroos, K., Garcia-Garcia, H. M., Oemrawsingh, R. M., Regar, E., Koenig, W., Serruys, P. W., van Geuns, R. J., Boersma, E. & Laaksonen, R. Plasma concentrations of molecular lipid species in relation to coronary plaque characteristics and cardiovascular outcome: Results of the ATHEROREMO-IVUS study. Atherosclerosis 243, 560-566, doi:10.1016/j.atherosclerosis.2015.10.022 (2015). 204 Tarasov, K., Ekroos, K., Suoniemi, M., Kauhanen, D., Sylvanne, T., Hurme, R., Gouni-Berthold, I., Berthold, H. K., Kleber, M. E., Laaksonen, R. & Marz, W. Molecular lipids identify cardiovascular risk and are efficiently lowered by simvastatin and PCSK9 deficiency. The Journal of clinical endocrinology and metabolism 99, E45-52, doi:10.1210/jc.2013-2559 (2014). 205 Havulinna, A. S., Sysi-Aho, M., Hilvo, M., Kauhanen, D., Hurme, R., Ekroos, K., Salomaa, V. & Laaksonen, R. Circulating Ceramides Predict Cardiovascular Outcomes in the Population-Based FINRISK 2002 Cohort. Arterioscler Thromb Vasc Biol 36, 2424-2430, doi:10.1161/ATVBAHA.116.307497 (2016). 206 Laaksonen, R., Ekroos, K., Sysi-Aho, M., Hilvo, M., Vihervaara, T., Kauhanen, D., Suoniemi, M., Hurme, R., Marz, W., Scharnagl, H., Stojakovic, T., Vlachopoulou, E., Lokki, M. L., Nieminen, M. S., Klingenberg, R., Matter, C. M., Hornemann, T., Juni, P., Rodondi, N., Raber, L., Windecker, S., Gencer, B., Pedersen, E. R., Tell, G. S., Nygard, O., Mach, F., Sinisalo, J. & Luscher, T. F. Plasma ceramides predict cardiovascular death in patients with stable coronary 187

artery disease and acute coronary syndromes beyond LDL-cholesterol. Eur Heart J 37, 1967-1976, doi:10.1093/eurheartj/ehw148 (2016). 207 Zhang, X., Zhang, G., Zhang, H., Karin, M., Bai, H. & Cai, D. Hypothalamic IKKbeta/NF-kappaB and ER stress link overnutrition to energy imbalance and obesity. Cell 135, 61-73, doi:10.1016/j.cell.2008.07.043 (2008). 208 Ozcan, L., Ergin, A. S., Lu, A., Chung, J., Sarkar, S., Nie, D., Myers, M. G., Jr. & Ozcan, U. Endoplasmic reticulum stress plays a central role in development of leptin resistance. Cell metabolism 9, 35-51, doi:10.1016/j.cmet.2008.12.004 (2009). 209 Contreras, C., Gonzalez-Garcia, I., Martinez-Sanchez, N., Seoane-Collazo, P., Jacas, J., Morgan, D. A., Serra, D., Gallego, R., Gonzalez, F., Casals, N., Nogueiras, R., Rahmouni, K., Dieguez, C. & Lopez, M. Central ceramide- induced hypothalamic lipotoxicity and ER stress regulate energy balance. Cell reports 9, 366-377, doi:10.1016/j.celrep.2014.08.057 (2014). 210 Contreras, C., Gonzalez-Garcia, I., Seoane-Collazo, P., Martinez-Sanchez, N., Linares-Pose, L., Rial-Pensado, E., Ferno, J., Tena-Sempere, M., Casals, N., Dieguez, C., Nogueiras, R. & Lopez, M. Reduction of Hypothalamic Endoplasmic Reticulum Stress Activates Browning of White Fat and Ameliorates Obesity. Diabetes 66, 87-99, doi:10.2337/db15-1547 (2017). 211 Obici, S., Feng, Z., Arduini, A., Conti, R. & Rossetti, L. Inhibition of hypothalamic carnitine palmitoyltransferase-1 decreases food intake and glucose production. Nat Med 9, 756-761, doi:10.1038/nm873 (2003). 212 He, W., Lam, T. K., Obici, S. & Rossetti, L. Molecular disruption of hypothalamic nutrient sensing induces obesity. Nat Neurosci 9, 227-233, doi:10.1038/nn1626 (2006). 213 Lopez, M., Lelliott, C. J., Tovar, S., Kimber, W., Gallego, R., Virtue, S., Blount, M., Vazquez, M. J., Finer, N., Powles, T. J., O'Rahilly, S., Saha, A. K., Dieguez, C. & Vidal-Puig, A. J. Tamoxifen-induced anorexia is associated with fatty acid synthase inhibition in the ventromedial nucleus of the hypothalamus and accumulation of malonyl-CoA. Diabetes 55, 1327-1336 (2006). 214 Gao, S., Kinzig, K. P., Aja, S., Scott, K. A., Keung, W., Kelly, S., Strynadka, K., Chohnan, S., Smith, W. W., Tamashiro, K. L., Ladenheim, E. E., Ronnett, G. V., Tu, Y., Birnbaum, M. J., Lopaschuk, G. D. & Moran, T. H. Leptin activates hypothalamic acetyl-CoA carboxylase to inhibit food intake. Proceedings of the 188

National Academy of Sciences of the United States of America 104, 17358- 17363, doi:10.1073/pnas.0708385104 (2007). 215 Wolfgang, M. J., Cha, S. H., Sidhaye, A., Chohnan, S., Cline, G., Shulman, G. I. & Lane, M. D. Regulation of hypothalamic malonyl-CoA by central glucose and leptin. Proceedings of the National Academy of Sciences of the United States of America 104, 19285-19290, doi:10.1073/pnas.0709778104 (2007). 216 Wolfgang, M. J., Kurama, T., Dai, Y., Suwa, A., Asaumi, M., Matsumoto, S., Cha, S. H., Shimokawa, T. & Lane, M. D. The brain-specific carnitine palmitoyltransferase-1c regulates energy homeostasis. Proceedings of the National Academy of Sciences of the United States of America 103, 7282-7287, doi:10.1073/pnas.0602205103 (2006). 217 Gao, S., Zhu, G., Gao, X., Wu, D., Carrasco, P., Casals, N., Hegardt, F. G., Moran, T. H. & Lopaschuk, G. D. Important roles of brain-specific carnitine palmitoyltransferase and ceramide metabolism in leptin hypothalamic control of feeding. Proceedings of the National Academy of Sciences of the United States of America 108, 9691-9696, doi:10.1073/pnas.1103267108 (2011). 218 Kim, C. H. & Younossi, Z. M. Nonalcoholic fatty liver disease: a manifestation of the metabolic syndrome. Cleve Clin J Med 75, 721-728 (2008). 219 Neuschwander-Tetri, B. A. & Caldwell, S. H. Nonalcoholic steatohepatitis: summary of an AASLD Single Topic Conference. Hepatology 37, 1202-1219, doi:10.1053/jhep.2003.50193 (2003). 220 Yetukuri, L., Katajamaa, M., Medina-Gomez, G., Seppanen-Laakso, T., Vidal- Puig, A. & Oresic, M. Bioinformatics strategies for lipidomics analysis: characterization of obesity related hepatic steatosis. BMC Syst Biol 1, 12, doi:10.1186/1752-0509-1-12 (2007). 221 Bikman, B. T., Guan, Y., Shui, G., Siddique, M. M., Holland, W. L., Kim, J. Y., Fabrias, G., Wenk, M. R. & Summers, S. A. Fenretinide prevents lipid-induced insulin resistance by blocking ceramide biosynthesis. The Journal of biological chemistry 287, 17426-17437, doi:10.1074/jbc.M112.359950 (2012). 222 Yang, G., Badeanlou, L., Bielawski, J., Roberts, A. J., Hannun, Y. A. & Samad, F. Central role of ceramide biosynthesis in body weight regulation, energy metabolism, and the metabolic syndrome. American journal of physiology. Endocrinology and metabolism 297, E211-224, doi:10.1152/ajpendo.91014.2008 (2009). 189

223 Correnti, J. M., Juskeviciute, E., Swarup, A. & Hoek, J. B. Pharmacological ceramide reduction alleviates alcohol-induced steatosis and hepatomegaly in adiponectin knockout mice. Am J Physiol Gastrointest Liver Physiol 306, G959- 973, doi:10.1152/ajpgi.00395.2013 (2014). 224 Kurek, K., Piotrowska, D. M., Wiesiolek-Kurek, P., Lukaszuk, B., Chabowski, A., Gorski, J. & Zendzian-Piotrowska, M. Inhibition of ceramide de novo synthesis reduces liver lipid accumulation in rats with nonalcoholic fatty liver disease. Liver Int 34, 1074-1083, doi:10.1111/liv.12331 (2014). 225 Xia, J. Y., Holland, W. L., Kusminski, C. M., Sun, K., Sharma, A. X., Pearson, M. J., Sifuentes, A. J., McDonald, J. G., Gordillo, R. & Scherer, P. E. Targeted Induction of Ceramide Degradation Leads to Improved Systemic Metabolism and Reduced Hepatic Steatosis. Cell metabolism 22, 266-278, doi:10.1016/j.cmet.2015.06.007 (2015). 226 Holland, W. L., Xia, J. Y., Johnson, J. A., Sun, K., Pearson, M. J., Sharma, A. X., Quittner-Strom, E., Tippetts, T. S., Gordillo, R. & Scherer, P. E. Inducible overexpression of adiponectin receptors highlight the roles of adiponectin- induced ceramidase signaling in lipid and glucose homeostasis. Mol Metab 6, 267-275, doi:10.1016/j.molmet.2017.01.002 (2017). 227 Senkal, C. E., Salama, M. F., Snider, A. J., Allopenna, J. J., Rana, N. A., Koller, A., Hannun, Y. A. & Obeid, L. M. Ceramide Is Metabolized to Acylceramide and Stored in Lipid Droplets. Cell metabolism 25, 686-697, doi:10.1016/j.cmet.2017.02.010 (2017). 228 Watson, M. L., Macrae, K., Marley, A. E. & Hundal, H. S. Chronic effects of palmitate overload on nutrient-induced insulin secretion and autocrine signalling in pancreatic MIN6 beta cells. PloS one 6, e25975, doi:10.1371/journal.pone.0025975 (2011). 229 Chaurasia, B. & Summers, S. A. Ceramides - Lipotoxic Inducers of Metabolic Disorders. Trends Endocrinol Metab 26, 538-550, doi:10.1016/j.tem.2015.07.006 (2015). 230 Watson, M. L., Coghlan, M. & Hundal, H. S. Modulating serine palmitoyl transferase (SPT) expression and activity unveils a crucial role in lipid-induced insulin resistance in rat skeletal muscle cells. The Biochemical journal 417, 791- 801, doi:10.1042/BJ20081149 (2009).

190

231 Holland, W. L., Brozinick, J. T., Wang, L. P., Hawkins, E. D., Sargent, K. M., Liu, Y., Narra, K., Hoehn, K. L., Knotts, T. A., Siesky, A., Nelson, D. H., Karathanasis, S. K., Fontenot, G. K., Birnbaum, M. J. & Summers, S. A. Inhibition of ceramide synthesis ameliorates glucocorticoid-, saturated-fat-, and obesity-induced insulin resistance. Cell metabolism 5, 167-179, doi:10.1016/j.cmet.2007.01.002 (2007). 232 Fillmore, N., Keung, W., Kelly, S. E., Proctor, S. D., Lopaschuk, G. D. & Ussher, J. R. Accumulation of ceramide in slow-twitch muscle contributes to the development of insulin resistance in the obese JCR:LA-cp rat. Exp Physiol 100, 730-741, doi:10.1113/EP085052 (2015). 233 Dekker, M. J., Baker, C., Naples, M., Samsoondar, J., Zhang, R., Qiu, W., Sacco, J. & Adeli, K. Inhibition of sphingolipid synthesis improves dyslipidemia in the diet-induced hamster model of insulin resistance: evidence for the role of sphingosine and sphinganine in hepatic VLDL-apoB100 overproduction. Atherosclerosis 228, 98-109, doi:10.1016/j.atherosclerosis.2013.01.041 (2013). 234 Li, Z., Zhang, H., Liu, J., Liang, C. P., Li, Y., Li, Y., Teitelman, G., Beyer, T., Bui, H. H., Peake, D. A., Zhang, Y., Sanders, P. E., Kuo, M. S., Park, T. S., Cao, G. & Jiang, X. C. Reducing plasma membrane sphingomyelin increases insulin sensitivity. Molecular and cellular biology 31, 4205-4218, doi:10.1128/MCB.05893-11 (2011). 235 Montgomery, M. K., Brown, S. H., Lim, X. Y., Fiveash, C. E., Osborne, B., Bentley, N. L., Braude, J. P., Mitchell, T. W., Coster, A. C., Don, A. S., Cooney, G. J., Schmitz-Peiffer, C. & Turner, N. Regulation of glucose homeostasis and insulin action by ceramide acyl-chain length: A beneficial role for very long- chain sphingolipid species. Biochim Biophys Acta 1861, 1828-1839, doi:10.1016/j.bbalip.2016.08.016 (2016). 236 Bergman, B. C., Brozinick, J. T., Strauss, A., Bacon, S., Kerege, A., Bui, H. H., Sanders, P., Siddall, P., Wei, T., Thomas, M. K., Kuo, M. S. & Perreault, L. Muscle sphingolipids during rest and exercise: a C18:0 signature for insulin resistance in humans. Diabetologia, 1-14, doi:10.1007/s00125-015-3850-y (2016). 237 Coen, P. M., Dube, J. J., Amati, F., Stefanovic-Racic, M., Ferrell, R. E., Toledo, F. G. & Goodpaster, B. H. Insulin resistance is associated with higher intramyocellular triglycerides in type I but not type II myocytes concomitant 191

with higher ceramide content. Diabetes 59, 80-88, doi:10.2337/db09-0988 (2010). 238 Boon, J., Hoy, A. J., Stark, R., Brown, R. D., Meex, R. C., Henstridge, D. C., Schenk, S., Meikle, P. J., Horowitz, J. F., Kingwell, B. A., Bruce, C. R. & Watt, M. J. Ceramides contained in LDL are elevated in type 2 diabetes and promote inflammation and skeletal muscle insulin resistance. Diabetes 62, 401-410, doi:10.2337/db12-0686 (2013). 239 Dube, J. J., Amati, F., Toledo, F. G., Stefanovic-Racic, M., Rossi, A., Coen, P. & Goodpaster, B. H. Effects of weight loss and exercise on insulin resistance, and intramyocellular triacylglycerol, diacylglycerol and ceramide. Diabetologia 54, 1147-1156, doi:10.1007/s00125-011-2065-0 (2011). 240 Coen, P. M., Menshikova, E. V., Distefano, G., Zheng, D., Tanner, C. J., Standley, R. A., Helbling, N. L., Dubis, G. S., Ritov, V. B., Xie, H., Desimone, M. E., Smith, S. R., Stefanovic-Racic, M., Toledo, F. G., Houmard, J. A. & Goodpaster, B. H. Exercise and Weight Loss Improve Muscle Mitochondrial Respiration, Lipid Partitioning, and Insulin Sensitivity After Gastric Bypass Surgery. Diabetes 64, 3737-3750, doi:10.2337/db15-0809 (2015). 241 Schenk, S. & Horowitz, J. F. Acute exercise increases triglyceride synthesis in skeletal muscle and prevents fatty acid-induced insulin resistance. The Journal of clinical investigation 117, 1690-1698, doi:10.1172/JCI30566 (2007). 242 Tagami, S., Inokuchi Ji, J., Kabayama, K., Yoshimura, H., Kitamura, F., Uemura, S., Ogawa, C., Ishii, A., Saito, M., Ohtsuka, Y., Sakaue, S. & Igarashi, Y. Ganglioside GM3 participates in the pathological conditions of insulin resistance. The Journal of biological chemistry 277, 3085-3092, doi:10.1074/jbc.M103705200 (2002). 243 Yamashita, T., Hashiramoto, A., Haluzik, M., Mizukami, H., Beck, S., Norton, A., Kono, M., Tsuji, S., Daniotti, J. L., Werth, N., Sandhoff, R., Sandhoff, K. & Proia, R. L. Enhanced insulin sensitivity in mice lacking ganglioside GM3. Proceedings of the National Academy of Sciences of the United States of America 100, 3445-3449, doi:10.1073/pnas.0635898100 (2003). 244 Aerts, J. M., Ottenhoff, R., Powlson, A. S., Grefhorst, A., van Eijk, M., Dubbelhuis, P. F., Aten, J., Kuipers, F., Serlie, M. J., Wennekes, T., Sethi, J. K., O'Rahilly, S. & Overkleeft, H. S. Pharmacological inhibition of

192

glucosylceramide synthase enhances insulin sensitivity. Diabetes 56, 1341-1349, doi:10.2337/db06-1619 (2007). 245 Bijl, N., Sokolovic, M., Vrins, C., Langeveld, M., Moerland, P. D., Ottenhoff, R., van Roomen, C. P., Claessen, N., Boot, R. G., Aten, J., Groen, A. K., Aerts, J. M. & van Eijk, M. Modulation of glycosphingolipid metabolism significantly improves hepatic insulin sensitivity and reverses hepatic steatosis in mice. Hepatology 50, 1431-1441, doi:10.1002/hep.23175 (2009). 246 van Eijk, M., Aten, J., Bijl, N., Ottenhoff, R., van Roomen, C. P., Dubbelhuis, P. F., Seeman, I., Ghauharali-van der Vlugt, K., Overkleeft, H. S., Arbeeny, C., Groen, A. K. & Aerts, J. M. Reducing glycosphingolipid content in adipose tissue of obese mice restores insulin sensitivity, adipogenesis and reduces inflammation. PloS one 4, e4723, doi:10.1371/journal.pone.0004723 (2009). 247 Zhao, H., Przybylska, M., Wu, I. H., Zhang, J., Siegel, C., Komarnitsky, S., Yew, N. S. & Cheng, S. H. Inhibiting glycosphingolipid synthesis improves glycemic control and insulin sensitivity in animal models of type 2 diabetes. Diabetes 56, 1210-1218, doi:10.2337/db06-0719 (2007). 248 Fuller, M. Sphingolipids: the nexus between Gaucher disease and insulin resistance. Lipids Health Dis 9, 113, doi:10.1186/1476-511X-9-113 (2010). 249 Chavez, J. A., Siddique, M. M., Wang, S. T., Ching, J., Shayman, J. A. & Summers, S. A. Ceramides and glucosylceramides are independent antagonists of insulin signaling. The Journal of biological chemistry 289, 723-734, doi:10.1074/jbc.M113.522847 (2014). 250 Kluepfel, D., Bagli, J., Baker, H., Charest, M. P. & Kudelski, A. Myriocin, a new antifungal antibiotic from Myriococcum albomyces. J Antibiot (Tokyo) 25, 109-115 (1972). 251 Aragozzini, F., Manachini, P. L., Craveri, R., Rindone, B. & Scolastico, C. Structure of thermozymocidin. Experientia 28, 881-882 (1972). 252 Fujita, T., Inoue, K., Yamamoto, S., Ikumoto, T., Sasaki, S., Toyama, R., Yoneta, M., Chiba, K., Hoshino, Y. & Okumoto, T. Fungal metabolites. Part 12. Potent immunosuppressant, 14-deoxomyriocin, (2S,3R,4R)-(E)-2-amino-3,4- dihydroxy-2-hydroxymethyleicos-6-enoic acid and structure-activity relationships of myriocin derivatives. J Antibiot (Tokyo) 47, 216-224 (1994). 253 Miyake, Y., Kozutsumi, Y., Nakamura, S., Fujita, T. & Kawasaki, T. Serine palmitoyltransferase is the primary target of a sphingosine-like 193

immunosuppressant, ISP-1/myriocin. Biochemical and biophysical research communications 211, 396-403, doi:10.1006/bbrc.1995.1827 (1995). 254 Chen, J. K., Lane, W. S. & Schreiber, S. L. The identification of myriocin- binding proteins. Chem Biol 6, 221-235, doi:10.1016/S1074-5521(99)80038-6 (1999). 255 Wu, J. M., DiPietrantonio, A. M. & Hsieh, T. C. Mechanism of fenretinide (4- HPR)-induced cell death. Apoptosis : an international journal on programmed cell death 6, 377-388 (2001). 256 Veronesi, U., Mariani, L., Decensi, A., Formelli, F., Camerini, T., Miceli, R., Di Mauro, M. G., Costa, A., Marubini, E., Sporn, M. B. & De Palo, G. Fifteen-year results of a randomized phase III trial of fenretinide to prevent second breast cancer. Ann Oncol 17, 1065-1071, doi:10.1093/annonc/mdl047 (2006). 257 Preitner, F., Mody, N., Graham, T. E., Peroni, O. D. & Kahn, B. B. Long-term Fenretinide treatment prevents high-fat diet-induced obesity, insulin resistance, and hepatic steatosis. American journal of physiology. Endocrinology and metabolism 297, E1420-1429, doi:10.1152/ajpendo.00362.2009 (2009). 258 Yang, Q., Graham, T. E., Mody, N., Preitner, F., Peroni, O. D., Zabolotny, J. M., Kotani, K., Quadro, L. & Kahn, B. B. Serum retinol binding protein 4 contributes to insulin resistance in obesity and type 2 diabetes. Nature 436, 356- 362, doi:10.1038/nature03711 (2005). 259 Johansson, H., Gandini, S., Guerrieri-Gonzaga, A., Iodice, S., Ruscica, M., Bonanni, B., Gulisano, M., Magni, P., Formelli, F. & Decensi, A. Effect of fenretinide and low-dose tamoxifen on insulin sensitivity in premenopausal women at high risk for breast cancer. Cancer research 68, 9512-9518, doi:10.1158/0008-5472.CAN-08-0553 (2008). 260 Wang, H., Maurer, B. J., Liu, Y. Y., Wang, E., Allegood, J. C., Kelly, S., Symolon, H., Liu, Y., Merrill, A. H., Jr., Gouaze-Andersson, V., Yu, J. Y., Giuliano, A. E. & Cabot, M. C. N-(4-Hydroxyphenyl)retinamide increases dihydroceramide and synergizes with dimethylsphingosine to enhance cancer cell killing. Molecular cancer therapeutics 7, 2967-2976, doi:10.1158/1535- 7163.MCT-08-0549 (2008). 261 Rahmaniyan, M., Curley, R. W., Jr., Obeid, L. M., Hannun, Y. A. & Kraveka, J. M. Identification of dihydroceramide desaturase as a direct in vitro target for

194

fenretinide. The Journal of biological chemistry 286, 24754-24764, doi:10.1074/jbc.M111.250779 (2011). 262 Marasas, W. F. O., Nelson, P. E. & Toussoun, T. A. Toxigenic Fusarium species, identity and mycotoxicology. 263 Wang, E., Norred, W. P., Bacon, C. W., Riley, R. T. & Merrill, A. H., Jr. Inhibition of sphingolipid biosynthesis by fumonisins. Implications for diseases associated with Fusarium moniliforme. The Journal of biological chemistry 266, 14486-14490 (1991). 264 Delgado, A., Casas, J., Llebaria, A., Abad, J. L. & Fabrias, G. Inhibitors of sphingolipid metabolism enzymes. Biochim Biophys Acta 1758, 1957-1977, doi:10.1016/j.bbamem.2006.08.017 (2006). 265 Gelineau-van Waes, J., Rainey, M. A., Maddox, J. R., Voss, K. A., Sachs, A. J., Gardner, N. M., Wilberding, J. D. & Riley, R. T. Increased sphingoid base-1- phosphates and failure of neural tube closure after exposure to fumonisin or FTY720. Birth Defects Res A Clin Mol Teratol 94, 790-803, doi:10.1002/bdra.23074 (2012). 266 Gelderblom, W. C., Abel, S., Smuts, C. M., Marnewick, J., Marasas, W. F., Lemmer, E. R. & Ramljak, D. Fumonisin-induced hepatocarcinogenesis: mechanisms related to cancer initiation and promotion. Environmental health perspectives 109 Suppl 2, 291-300 (2001). 267 Bondy, G., Suzuki, C., Barker, M., Armstrong, C., Fernie, S., Hierlihy, L., Rowsell, P. & Mueller, R. Toxicity of fumonisin B1 administered intraperitoneally to male Sprague-Dawley rats. Food Chem Toxicol 33, 653-665 (1995). 268 Cohen, J. A., Barkhof, F., Comi, G., Hartung, H. P., Khatri, B. O., Montalban, X., Pelletier, J., Capra, R., Gallo, P., Izquierdo, G., Tiel-Wilck, K., de Vera, A., Jin, J., Stites, T., Wu, S., Aradhye, S., Kappos, L. & Group, T. S. Oral fingolimod or intramuscular interferon for relapsing multiple sclerosis. The New England journal of medicine 362, 402-415, doi:10.1056/NEJMoa0907839 (2010). 269 Paugh, S. W., Payne, S. G., Barbour, S. E., Milstien, S. & Spiegel, S. The immunosuppressant FTY720 is phosphorylated by sphingosine kinase type 2. FEBS letters 554, 189-193, doi:10.1016/S0014-5793(03)01168-2 (2003).

195

270 Billich, A., Bornancin, F., Dévay, P., Mechtcheriakova, D., Urtz, N. & Baumruker, T. Phosphorylation of the Immunomodulatory Drug FTY720 by Sphingosine Kinases. Journal of Biological Chemistry 278, 47408-47415, doi:10.1074/jbc.M307687200 (2003). 271 Sanchez, T., Estrada-Hernandez, T., Paik, J.-H., Wu, M.-T., Venkataraman, K., Brinkmann, V., Claffey, K. & Hla, T. Phosphorylation and Action of the Immunomodulator FTY720 Inhibits Vascular Endothelial Cell Growth Factor- induced Vascular Permeability. Journal of Biological Chemistry 278, 47281- 47290, doi:10.1074/jbc.M306896200 (2003). 272 Adachi, K., Kohara, T., Nakao, N., Arita, M., Chiba, K., Mishina, T., Sasaki, S. & Fujita, T. Design, synthesis, and structure-activity relationships of 2- substituted-2-amino-1,3-propanediols: Discovery of a novel immunosuppressant, FTY720. Bioorganic & Medicinal Chemistry Letters 5, 853-856, doi:http://dx.doi.org/10.1016/0960-894X(95)00127-F (1995). 273 Cinamon, G., Matloubian, M., Lesneski, M. J., Xu, Y., Low, C., Lu, T., Proia, R. L. & Cyster, J. G. Sphingosine 1-phosphate receptor 1 promotes B cell localization in the splenic marginal zone. Nature immunology 5, 713-720, doi:10.1038/ni1083 (2004). 274 Mandala, S., Hajdu, R., Bergstrom, J., Quackenbush, E., Xie, J., Milligan, J., Thornton, R., Shei, G. J., Card, D., Keohane, C., Rosenbach, M., Hale, J., Lynch, C. L., Rupprecht, K., Parsons, W. & Rosen, H. Alteration of lymphocyte trafficking by sphingosine-1-phosphate receptor agonists. Science 296, 346-349, doi:10.1126/science.1070238 (2002). 275 Berdyshev, E. V., Gorshkova, I., Skobeleva, A., Bittman, R., Lu, X., Dudek, S. M., Mirzapoiazova, T., Garcia, J. G. & Natarajan, V. FTY720 inhibits ceramide synthases and up-regulates dihydrosphingosine 1-phosphate formation in human lung endothelial cells. The Journal of biological chemistry 284, 5467-5477, doi:10.1074/jbc.M805186200 (2009). 276 Bruce, C. R., Risis, S., Babb, J. R., Yang, C., Lee-Young, R. S., Henstridge, D. C. & Febbraio, M. A. The sphingosine-1-phosphate analog FTY720 reduces muscle ceramide content and improves glucose tolerance in high fat-fed male mice. Endocrinology 154, 65-76, doi:10.1210/en.2012-1847 (2013). 277 Moon, M. H., Jeong, J. K., Lee, J. H., Park, Y. G., Lee, Y. J., Seol, J. W. & Park, S. Y. Antiobesity activity of a sphingosine 1-phosphate analogue FTY720 196

observed in adipocytes and obese mouse model. Experimental & molecular medicine 44, 603-614, doi:10.3858/emm.2012.44.10.069 (2012). 278 Schiffmann, S., Hartmann, D., Fuchs, S., Birod, K., Ferreiros, N., Schreiber, Y., Zivkovic, A., Geisslinger, G., Grosch, S. & Stark, H. Inhibitors of specific ceramide synthases. Biochimie 94, 558-565, doi:10.1016/j.biochi.2011.09.007 (2012). 279 Toop, H. D., Don, A. S. & Morris, J. C. Synthesis and biological evaluation of analogs of AAL(S) for use as ceramide synthase 1 inhibitors. Org Biomol Chem 13, 11593-11596, doi:10.1039/c5ob01931a (2015). 280 Fath, T., Ke, Y. D., Gunning, P., Gotz, J. & Ittner, L. M. Primary support cultures of hippocampal and substantia nigra neurons. Nature protocols 4, 78- 85, doi:10.1038/nprot.2008.199 (2009). 281 Watt, M. J., Hoy, A. J., Muoio, D. M. & Coleman, R. A. Distinct roles of specific fatty acids in cellular processes: implications for interpreting and reporting experiments. American journal of physiology. Endocrinology and metabolism 302, E1-3, doi:10.1152/ajpendo.00418.2011 (2012). 282 Frangioudakis, G., Diakanastasis, B., Liao, B. Q., Saville, J. T., Hoffman, N. J., Mitchell, T. W. & Schmitz-Peiffer, C. Ceramide accumulation in L6 skeletal muscle cells due to increased activity of ceramide synthase isoforms has opposing effects on insulin action to those caused by palmitate treatment. Diabetologia 56, 2697-2701, doi:10.1007/s00125-013-3035-5 (2013). 283 Couttas, T., Lim, X. & Don, A. A Three-Step Assay for Ceramide Synthase Activity Using a Fluorescent Substrate and HPLC. Lipids 50, 101-109, doi:10.1007/s11745-014-3969-5 (2015). 284 Wong, J. W., Abuhusain, H. J., McDonald, K. L. & Don, A. S. MMSAT: automated quantification of metabolites in selected reaction monitoring experiments. Analytical chemistry 84, 470-474, doi:10.1021/ac2026578 (2012). 285 Hu, C., van Dommelen, J., van der Heijden, R., Spijksma, G., Reijmers, T. H., Wang, M., Slee, E., Lu, X., Xu, G., van der Greef, J. & Hankemeier, T. RPLC- ion-trap-FTMS method for lipid profiling of plasma: method validation and application to mutant mouse model. Journal of proteome research 7, 4982- 4991, doi:10.1021/pr800373m (2008). 286 Bird, S. S., Marur, V. R., Sniatynski, M. J., Greenberg, H. K. & Kristal, B. S. Serum lipidomics profiling using LC-MS and high-energy collisional 197

dissociation fragmentation: focus on triglyceride detection and characterization. Analytical chemistry 83, 6648-6657, doi:10.1021/ac201195d (2011). 287 Turner, N., Hariharan, K., TidAng, J., Frangioudakis, G., Beale, S. M., Wright, L. E., Zeng, X. Y., Leslie, S. J., Li, J. Y., Kraegen, E. W., Cooney, G. J. & Ye, J. M. Enhancement of muscle mitochondrial oxidative capacity and alterations in insulin action are lipid species dependent: potent tissue-specific effects of medium-chain fatty acids. Diabetes 58, 2547-2554, doi:10.2337/db09-0784 (2009). 288 Kim, W. S., Hsiao, J. H., Bhatia, S., Glaros, E. N., Don, A. S., Tsuruoka, S., Shannon Weickert, C. & Halliday, G. M. ABCA8 stimulates sphingomyelin production in oligodendrocytes. The Biochemical journal 452, 401-410, doi:10.1042/BJ20121764 (2013). 289 Ayala, J. E., Bracy, D. P., McGuinness, O. P. & Wasserman, D. H. Considerations in the design of hyperinsulinemic-euglycemic clamps in the conscious mouse. Diabetes 55, 390-397 (2006). 290 Brandon, A. E., Stuart, E., Leslie, S. J., Hoehn, K. L., James, D. E., Kraegen, E. W., Turner, N. & Cooney, G. J. Minimal impact of age and housing temperature on the metabolic phenotype of Acc2-/- mice. J Endocrinol 228, 127-134, doi:10.1530/JOE-15-0444 (2016). 291 Ayala, J. E., Bracy, D. P., Julien, B. M., Rottman, J. N., Fueger, P. T. & Wasserman, D. H. Chronic treatment with sildenafil improves energy balance and insulin action in high fat-fed conscious mice. Diabetes 56, 1025-1033, doi:10.2337/db06-0883 (2007). 292 Dyck, D. J., Peters, S. J., Glatz, J., Gorski, J., Keizer, H., Kiens, B., Liu, S., Richter, E. A., Spriet, L. L., van der Vusse, G. J. & Bonen, A. Functional differences in lipid metabolism in resting skeletal muscle of various fiber types. Am J Physiol 272, E340-351, doi:10.1152/ajpendo.1997.272.3.E340 (1997). 293 Kuznetsov, A. V., Veksler, V., Gellerich, F. N., Saks, V., Margreiter, R. & Kunz, W. S. Analysis of mitochondrial function in situ in permeabilized muscle fibers, tissues and cells. Nat Protoc 3, 965-976, doi:10.1038/nprot.2008.61 (2008). 294 Spassieva, S., Bielawski, J., Anelli, V. & Obeid, L. M. Combination of C(17) sphingoid base homologues and mass spectrometry analysis as a new approach

198

to study sphingolipid metabolism. Methods in enzymology 434, 233-241, doi:10.1016/S0076-6879(07)34012-3 (2007). 295 Novgorodov, S. A., Riley, C. L., Keffler, J. A., Yu, J., Kindy, M. S., Macklin, W. B., Lombard, D. B. & Gudz, T. I. SIRT3 Deacetylates Ceramide Synthases: IMPLICATIONS FOR MITOCHONDRIAL DYSFUNCTION AND BRAIN INJURY. The Journal of biological chemistry 291, 1957-1973, doi:10.1074/jbc.M115.668228 (2016). 296 Lim, X. Y., Pickford, R. & Don, A. S. Assaying Ceramide Synthase Activity In Vitro and in Living Cells Using Liquid Chromatography-Mass Spectrometry. Methods Mol Biol 1376, 11-22, doi:10.1007/978-1-4939-3170-5_2 (2016). 297 Merrill, A. H., Jr., van Echten, G., Wang, E. & Sandhoff, K. Fumonisin B1 inhibits sphingosine (sphinganine) N-acyltransferase and de novo sphingolipid biosynthesis in cultured neurons in situ. The Journal of biological chemistry 268, 27299-27306 (1993). 298 Bose, R. & Kolesnick, R. Measurement of ceramide synthase activity. Methods in enzymology 322, 378-382 (2000). 299 Couttas, T. A., Lim, X. Y. & Don, A. S. A three-step assay for ceramide synthase activity using a fluorescent substrate and HPLC. Lipids 50, 101-109, doi:10.1007/s11745-014-3969-5 (2015). 300 Kajimoto, T., Shirai, Y., Sakai, N., Yamamoto, T., Matsuzaki, H., Kikkawa, U. & Saito, N. Ceramide-induced apoptosis by translocation, phosphorylation, and activation of protein kinase Cdelta in the Golgi complex. The Journal of biological chemistry 279, 12668-12676, doi:10.1074/jbc.M312350200 (2004). 301 Aschrafi, A., Franzen, R., Shabahang, S., Fabbro, D., Pfeilschifter, J. & Huwiler, A. Ceramide induces translocation of protein kinase C-alpha to the Golgi compartment of human embryonic kidney cells by interacting with the C2 domain. Biochim Biophys Acta 1634, 30-39 (2003). 302 Kashiwagi, K., Shirai, Y., Kuriyama, M., Sakai, N. & Saito, N. Importance of C1B domain for lipid messenger-induced targeting of protein kinase C. The Journal of biological chemistry 277, 18037-18045, doi:10.1074/jbc.M111761200 (2002). 303 Lee, H., Rotolo, J. A., Mesicek, J., Penate-Medina, T., Rimner, A., Liao, W. C., Yin, X., Ragupathi, G., Ehleiter, D., Gulbins, E., Zhai, D., Reed, J. C., Haimovitz-Friedman, A., Fuks, Z. & Kolesnick, R. Mitochondrial ceramide-rich 199

macrodomains functionalize Bax upon irradiation. PloS one 6, e19783, doi:10.1371/journal.pone.0019783 (2011). 304 Stiban, J., Caputo, L. & Colombini, M. Ceramide synthesis in the endoplasmic reticulum can permeabilize mitochondria to proapoptotic proteins. Journal of lipid research 49, 625-634, doi:10.1194/jlr.M700480-JLR200 (2008). 305 Bigaud, M., Guerini, D., Billich, A., Bassilana, F. & Brinkmann, V. Second generation S1P pathway modulators: research strategies and clinical developments. Biochim Biophys Acta 1841, 745-758, doi:10.1016/j.bbalip.2013.11.001 (2014). 306 Topliss, J. G. Utilization of operational schemes for analog synthesis in drug design. J Med Chem 15, 1006-1011 (1972). 307 Topliss, J. G. A manual method for applying the Hansch approach to drug design. J Med Chem 20, 463-469 (1977). 308 Hannun, Y. A. & Obeid, L. M. Many ceramides. The Journal of biological chemistry 286, 27855-27862, doi:10.1074/jbc.R111.254359 (2011). 309 Chaurasia, B., Kaddai, V. A., Lancaster, G. I., Henstridge, D. C., Sriram, S., Galam, D. L., Gopalan, V., Prakash, K. N., Velan, S. S., Bulchand, S., Tsong, T. J., Wang, M., Siddique, M. M., Yuguang, G., Sigmundsson, K., Mellet, N. A., Weir, J. M., Meikle, P. J., Bin, M. Y. M. S., Shabbir, A., Shayman, J. A., Hirabayashi, Y., Shiow, S. T., Sugii, S. & Summers, S. A. Adipocyte Ceramides Regulate Subcutaneous Adipose Browning, Inflammation, and Metabolism. Cell metabolism 24, 820-834, doi:10.1016/j.cmet.2016.10.002 (2016). 310 Laviad, E. L., Kelly, S., Merrill, A. H., Jr. & Futerman, A. H. Modulation of ceramide synthase activity via dimerization. The Journal of biological chemistry 287, 21025-21033, doi:10.1074/jbc.M112.363580 (2012). 311 Thul, P. J., Akesson, L., Wiking, M., Mahdessian, D., Geladaki, A., Ait Blal, H., Alm, T., Asplund, A., Bjork, L., Breckels, L. M., Backstrom, A., Danielsson, F., Fagerberg, L., Fall, J., Gatto, L., Gnann, C., Hober, S., Hjelmare, M., Johansson, F., Lee, S., Lindskog, C., Mulder, J., Mulvey, C. M., Nilsson, P., Oksvold, P., Rockberg, J., Schutten, R., Schwenk, J. M., Sivertsson, A., Sjostedt, E., Skogs, M., Stadler, C., Sullivan, D. P., Tegel, H., Winsnes, C., Zhang, C., Zwahlen, M., Mardinoglu, A., Ponten, F., von Feilitzen, K., Lilley, K. S., Uhlen, M. & Lundberg, E. A subcellular map of the human proteome. Science 356, doi:10.1126/science.aal3321 (2017). 200

312 Kagedal, K., Zhao, M., Svensson, I. & Brunk, U. T. Sphingosine-induced apoptosis is dependent on lysosomal proteases. The Biochemical journal 359, 335-343 (2001). 313 Hannun, Y. A. & Bell, R. M. Lysosphingolipids inhibit protein kinase C: implications for the sphingolipidoses. Science 235, 670-674 (1987). 314 Tang, N., Ong, W. Y., Zhang, E. M., Chen, P. & Yeo, J. F. Differential effects of ceramide species on exocytosis in rat PC12 cells. Exp Brain Res 183, 241- 247, doi:10.1007/s00221-007-1036-7 (2007). 315 Meldrum, B. S. Glutamate as a neurotransmitter in the brain: review of physiology and pathology. J Nutr 130, 1007S-1015S (2000). 316 Spassieva, S. D., Ji, X., Liu, Y., Gable, K., Bielawski, J., Dunn, T. M., Bieberich, E. & Zhao, L. Ectopic expression of ceramide synthase 2 in neurons suppresses neurodegeneration induced by ceramide synthase 1 deficiency. Proceedings of the National Academy of Sciences of the United States of America 113, 5928-5933, doi:10.1073/pnas.1522071113 (2016). 317 Haus, J. M., Kashyap, S. R., Kasumov, T., Zhang, R., Kelly, K. R., Defronzo, R. A. & Kirwan, J. P. Plasma ceramides are elevated in obese subjects with type 2 diabetes and correlate with the severity of insulin resistance. Diabetes 58, 337- 343, doi:10.2337/db08-1228 (2009). 318 Yaffe, D. & Saxel, O. Serial passaging and differentiation of myogenic cells isolated from dystrophic mouse muscle. Nature 270, 725-727 (1977). 319 Mahfouz, R., Khoury, R., Blachnio-Zabielska, A., Turban, S., Loiseau, N., Lipina, C., Stretton, C., Bourron, O., Ferre, P., Foufelle, F., Hundal, H. S. & Hajduch, E. Characterising the inhibitory actions of ceramide upon insulin signaling in different skeletal muscle cell models: a mechanistic insight. PloS one 9, e101865, doi:10.1371/journal.pone.0101865 (2014). 320 Park, M., Kaddai, V., Ching, J., Fridianto, K. T., Sieli, R. J., Sugii, S. & Summers, S. A. A Role for Ceramides, but Not , as Antagonists of Insulin Signaling and Mitochondrial Metabolism in C2C12 Myotubes. The Journal of biological chemistry 291, 23978-23988, doi:10.1074/jbc.M116.737684 (2016). 321 Chavez, J. A. & Summers, S. A. Characterizing the effects of saturated fatty acids on insulin signaling and ceramide and diacylglycerol accumulation in 3T3-

201

L1 adipocytes and C2C12 myotubes. Arch Biochem Biophys 419, 101-109 (2003). 322 Amati, F., Dube, J. J., Alvarez-Carnero, E., Edreira, M. M., Chomentowski, P., Coen, P. M., Switzer, G. E., Bickel, P. E., Stefanovic-Racic, M., Toledo, F. G. & Goodpaster, B. H. Skeletal muscle triglycerides, diacylglycerols, and ceramides in insulin resistance: another paradox in endurance-trained athletes? Diabetes 60, 2588-2597, doi:10.2337/db10-1221 (2011). 323 Dube, J. J., Amati, F., Stefanovic-Racic, M., Toledo, F. G., Sauers, S. E. & Goodpaster, B. H. Exercise-induced alterations in intramyocellular lipids and insulin resistance: the athlete's paradox revisited. American journal of physiology. Endocrinology and metabolism 294, E882-888, doi:10.1152/ajpendo.00769.2007 (2008). 324 Brozinick, J. T., Hawkins, E., Hoang Bui, H., Kuo, M. S., Tan, B., Kievit, P. & Grove, K. Plasma sphingolipids are biomarkers of metabolic syndrome in non- human primates maintained on a Western-style diet. Int J Obes (Lond) 37, 1064- 1070, doi:10.1038/ijo.2012.191 (2013). 325 Serlie, M. J., Meijer, A. J., Groener, J. E., Duran, M., Endert, E., Fliers, E., Aerts, J. M. & Sauerwein, H. P. Short-term manipulation of plasma free fatty acids does not change skeletal muscle concentrations of ceramide and glucosylceramide in lean and overweight subjects. The Journal of clinical endocrinology and metabolism 92, 1524-1529, doi:10.1210/jc.2006-2347 (2007). 326 Skovbro, M., Baranowski, M., Skov-Jensen, C., Flint, A., Dela, F., Gorski, J. & Helge, J. W. Human skeletal muscle ceramide content is not a major factor in muscle insulin sensitivity. Diabetologia 51, 1253-1260, doi:10.1007/s00125- 008-1014-z (2008). 327 Ieronimakis, N., Pantoja, M., Hays, A. L., Dosey, T. L., Qi, J., Fischer, K. A., Hoofnagle, A. N., Sadilek, M., Chamberlain, J. S., Ruohola-Baker, H. & Reyes, M. Increased sphingosine-1-phosphate improves muscle regeneration in acutely injured mdx mice. Skelet Muscle 3, 20, doi:10.1186/2044-5040-3-20 (2013). 328 Ross, J. S., Hu, W., Rosen, B., Snider, A. J., Obeid, L. M. & Cowart, L. A. Sphingosine kinase 1 is regulated by peroxisome proliferator-activated receptor alpha in response to free fatty acids and is essential for skeletal muscle

202

interleukin-6 production and signaling in diet-induced obesity. The Journal of biological chemistry 288, 22193-22206, doi:10.1074/jbc.M113.477786 (2013). 329 Fayyaz, S., Japtok, L. & Kleuser, B. Divergent role of sphingosine 1-phosphate on insulin resistance. Cell Physiol Biochem 34, 134-147, doi:10.1159/000362990 (2014). 330 Wang, J., Badeanlou, L., Bielawski, J., Ciaraldi, T. P. & Samad, F. Sphingosine kinase 1 regulates adipose proinflammatory responses and insulin resistance. American journal of physiology. Endocrinology and metabolism 306, E756-768, doi:10.1152/ajpendo.00549.2013 (2014). 331 Rapizzi, E., Taddei, M. L., Fiaschi, T., Donati, C., Bruni, P. & Chiarugi, P. Sphingosine 1-phosphate increases glucose uptake through trans-activation of insulin receptor. Cellular and molecular life sciences : CMLS 66, 3207-3218, doi:10.1007/s00018-009-0106-3 (2009). 332 Hocking, S., Samocha-Bonet, D., Milner, K. L., Greenfield, J. R. & Chisholm, D. J. Adiposity and insulin resistance in humans: the role of the different tissue and cellular lipid depots. Endocrine reviews 34, 463-500, doi:10.1210/er.2012- 1041 (2013). 333 Bruce, C. R., Risis, S., Babb, J. R., Yang, C., Kowalski, G. M., Selathurai, A., Lee-Young, R. S., Weir, J. M., Yoshioka, K., Takuwa, Y., Meikle, P. J., Pitson, S. M. & Febbraio, M. A. Overexpression of sphingosine kinase 1 prevents ceramide accumulation and ameliorates muscle insulin resistance in high-fat diet-fed mice. Diabetes 61, 3148-3155, doi:10.2337/db12-0029 (2012). 334 Bergman, B. C., Brozinick, J. T., Strauss, A., Bacon, S., Kerege, A., Bui, H. H., Sanders, P., Siddall, P., Wei, T., Thomas, M. K., Kuo, M. S. & Perreault, L. Muscle sphingolipids during rest and exercise: a C18:0 signature for insulin resistance in humans. Diabetologia 59, 785-798, doi:10.1007/s00125-015-3850- y (2016). 335 Hoehn, K. L., Turner, N., Swarbrick, M. M., Wilks, D., Preston, E., Phua, Y., Joshi, H., Furler, S. M., Larance, M., Hegarty, B. D., Leslie, S. J., Pickford, R., Hoy, A. J., Kraegen, E. W., James, D. E. & Cooney, G. J. Acute or chronic upregulation of mitochondrial fatty acid oxidation has no net effect on whole- body energy expenditure or adiposity. Cell metabolism 11, 70-76, doi:10.1016/j.cmet.2009.11.008 (2010).

203

336 Olson, D. P., Pulinilkunnil, T., Cline, G. W., Shulman, G. I. & Lowell, B. B. Gene knockout of Acc2 has little effect on body weight, fat mass, or food intake. Proc Natl Acad Sci U S A 107, 7598-7603, doi:10.1073/pnas.0913492107 (2010). 337 Vavrova, E., Lenoir, V., Alves-Guerra, M. C., Denis, R. G., Castel, J., Esnous, C., Dyck, J. R., Luquet, S., Metzger, D., Bouillaud, F. & Prip-Buus, C. Muscle expression of a malonyl-CoA-insensitive carnitine palmitoyltransferase-1 protects mice against high-fat/high-sucrose diet-induced insulin resistance. American journal of physiology. Endocrinology and metabolism 311, E649-660, doi:10.1152/ajpendo.00020.2016 (2016). 338 Bauer, R., Voelzmann, A., Breiden, B., Schepers, U., Farwanah, H., Hahn, I., Eckardt, F., Sandhoff, K. & Hoch, M. Schlank, a member of the ceramide synthase family controls growth and body fat in Drosophila. The EMBO journal 28, 3706-3716, doi:10.1038/emboj.2009.305 (2009). 339 Weir, J. M., Wong, G., Barlow, C. K., Greeve, M. A., Kowalczyk, A., Almasy, L., Comuzzie, A. G., Mahaney, M. C., Jowett, J. B., Shaw, J., Curran, J. E., Blangero, J. & Meikle, P. J. Plasma lipid profiling in a large population-based cohort. Journal of lipid research 54, 2898-2908, doi:10.1194/jlr.P035808 (2013). 340 Chavez, J. A., Holland, W. L., Bar, J., Sandhoff, K. & Summers, S. A. Acid ceramidase overexpression prevents the inhibitory effects of saturated fatty acids on insulin signaling. The Journal of biological chemistry 280, 20148-20153, doi:10.1074/jbc.M412769200 (2005). 341 Obanda, D. N., Hernandez, A., Ribnicky, D., Yu, Y., Zhang, X. H., Wang, Z. Q. & Cefalu, W. T. Bioactives of Artemisia dracunculus L. mitigate the role of ceramides in attenuating insulin signaling in rat skeletal muscle cells. Diabetes 61, 597-605, doi:10.2337/db11-0396 (2012). 342 Rutkowski, J. M., Stern, J. H. & Scherer, P. E. The cell biology of fat expansion. The Journal of cell biology 208, 501-512, doi:10.1083/jcb.201409063 (2015). 343 Choi, K. M., Lee, Y. S., Choi, M. H., Sin, D. M., Lee, S., Ji, S. Y., Lee, M. K., Lee, Y. M., Yun, Y. P., Hong, J. T. & Yoo, H. S. Inverse relationship between adipocyte differentiation and ceramide level in 3T3-L1 cells. Biol Pharm Bull 34, 912-916 (2011).

204

344 Ordonez, M., Presa, N., Trueba, M. & Gomez-Munoz, A. Implication of Ceramide Kinase in Adipogenesis. Mediators Inflamm 2017, 9374563, doi:10.1155/2017/9374563 (2017). 345 Jiang, C., Xie, C., Lv, Y., Li, J., Krausz, K. W., Shi, J., Brocker, C. N., Desai, D., Amin, S. G., Bisson, W. H., Liu, Y., Gavrilova, O., Patterson, A. D. & Gonzalez, F. J. Intestine-selective farnesoid X receptor inhibition improves obesity-related metabolic dysfunction. Nat Commun 6, 10166, doi:10.1038/ncomms10166 (2015). 346 Shah, C., Yang, G., Lee, I., Bielawski, J., Hannun, Y. A. & Samad, F. Protection from high fat diet-induced increase in ceramide in mice lacking plasminogen activator inhibitor 1. The Journal of biological chemistry 283, 13538-13548, doi:10.1074/jbc.M709950200 (2008). 347 Montgomery, M. K., Hallahan, N. L., Brown, S. H., Liu, M., Mitchell, T. W., Cooney, G. J. & Turner, N. Mouse strain-dependent variation in obesity and glucose homeostasis in response to high-fat feeding. Diabetologia 56, 1129- 1139, doi:10.1007/s00125-013-2846-8 (2013). 348 Montgomery, M. K., Brown, S. H. J., Mitchell, T. W., Coster, A. C. F., Cooney, G. J. & Turner, N. Association of muscle lipidomic profile with high-fat diet- induced insulin resistance across five mouse strains. Sci Rep 7, 13914, doi:10.1038/s41598-017-14214-1 (2017). 349 Straczkowski, M., Kowalska, I., Baranowski, M., Nikolajuk, A., Otziomek, E., Zabielski, P., Adamska, A., Blachnio, A., Gorski, J. & Gorska, M. Increased skeletal muscle ceramide level in men at risk of developing type 2 diabetes. Diabetologia 50, 2366-2373, doi:10.1007/s00125-007-0781-2 (2007).

205