PCT1 REGULATES PHOSPHATIDYLCHOLINE SYNTHESIS IN RESPONSE TO CHANGES IN SURFACE CURVATURE ELASTIC STRESS SENSED ON THE INNER NUCLEAR MEMBRANE     YU-CHEN WEI QUEENS’ COLLEGE UNIVERSITY OF CAMBRIDGE THIS DISSERTATION IS SUBMITTED FOR THE DEGREE OF DOCTOR OF PHILOSOPHY JULY 2018                                               i   Abstract Pct1 regulates phosphatidylcholine synthesis in response to changes in surface curvature elastic stress sensed on the inner nuclear membrane. Yu-Chen Wei Cell and organelle membranes consist of a complex mixture of phospholipids that determine their size, shape, and function. Among the distinct types of phospholipids found in membranes of living organisms, phosphatidylcholine (PC) is the most abundant. The rate- limiting step of the predominant pathway for PC synthesis in eukaryotic cells is catalysed by the enzyme, CTP: phosphocholine cytidylyltransferase α (CCTα or PCYT1A). CCTα has a critical role in lipid metabolism and also has direct clinical relevance as mutations in CCTα result in an interesting spectrum of human diseases, such as lipodystrophy with fatty liver, growth plate dysplasia and cone-rod related dystrophy. Numerous biochemical and structural studies on purified CCTα have revealed its membrane-bound activation and suggested that it acts as a lipid compositional sensor, yet the in vivo mechanism of how CCTα senses and regulates PC levels in membranes remains unclear. Here I show that in budding yeast Saccharomyces cerevisiae, Pct1, the yeast homolog of CCT, is intranuclear and translocates to the nuclear membrane in response to changes in membrane properties and the need for membrane PC synthesis. By aligning imaging with lipidomic analysis and data-driven modelling, Pct1 membrane association is demonstrated to correlate with membrane stored curvature elastic stress estimates. Furthermore, this process occurs inside the nucleus, although nuclear localization signal mutants can compensate for the loss of endogenous Pct1. These data suggest an ancient mechanism by which CCT senses lipid packing defects and regulates phospholipid homeostasis from the nucleus. Additionally, I identified the importance of mammalian CCT in early adipogenesis and investigated the enzymatic function of PCYT1A mutants in fibroblasts from lipodystrophic patients. The allele Val142Met is evaluated to be the main cause of loss-of-function in the compound heterozygous mutations by using yeast survival assay. These results collectively provide preliminary evidence for the pathogenicity of PCYT1A mutations in adipose tissue. From yeast to humans, this study uncovers the critical role of Pct1/CCT in maintaining the internal membrane environment. ii    Declaration I hereby declare that this dissertation: Is the result of my own work and includes nothing which is the outcome of work done in collaboration except as declared in the Preface and specified in the text; Is not substantially the same as any that I have submitted, or, is being concurrently submitted for a degree or diploma or other qualification at the University of Cambridge or any other University or similar institution except as declared in the Preface and specified in the text. I further state that no substantial part of my dissertation has already been submitted, or, is being concurrently submitted for any such degree, diploma or other qualification at the University of Cambridge or any other University or similar institution; Does not exceed the word limit of 60,000 (excluding figures, tables, appendices and bibliography) prescribed by the Degree Committee of the Faculty of Clinical Medicine. Yu-Chen Wei July 2018             iii   Acknowledgements I would like to express my great appreciation to Professor David Savage, my supervisor, for his patient guidance, enthusiastic encouragement and constructive recommendations during the planning and development of this research work. This thesis would not have been possible without his willingness to give his time and support so generously. I am grateful to Dr Symeon Siniossoglou for his professional advice that helps me overcome numerous obstacles through the research. I am also indebted to Dr Antonio Daniel Barbosa for his substantial critiques on this project. My sincere gratitude are also extended to Dr Albert Koulman for his help in doing the mass spectrometry data analysis, and to Dr Marcus Dymond, who helped calculate the stored curvature elastic stress. In addition, I wish to thank my laboratory mates who have been the source of both friendship and cooperation: Dr Amandine Girousse, whom I have learnt most of the basic but essential laboratory techniques from; Dr Afreen Haider, who was always glad to provide invaluable insights and review for this programme; Dr Satish Patel, with whom I have discussed on this project, and constantly being inspired; Dr KoiNi Lim, for her assistance in handling technical problems. Nothing is more important than a friendly and fun working environment created by Dr Anna Alvarez Guaita, Ms Liang Dong and Mr Michael Mimmack. I would also like to extend my thanks to the staff of Institute of Metabolic Science for their help in keeping everything running smoothly, and also to Cambridge where the greatest joy is found in the pursuit of knowledge. Finally, I wish to thank my family, my parents and sister for their unwavering love and belief in me throughout my PhD study. A deep gratitude to my parents, thank you for supporting me along my PhD journey.           iv    A Brief Note on Terminology CTP: phosphocholine cytidylyltransferase is often abbreviated to CCT. Saccharomyces cerevisiae express one CCT enzyme, called Pct1. Higher eukaryotes express two CCT enzymes: CCTα and CCTβ. CCTα is also known as PCYT1A in mammals; CCT1 in Drosophila. CCTβ is also known as PCYT1B in mammals; CCT2 in Drosophila.       v   Table of Contents Abstract ...................................................................................................................................... i  Declaration ................................................................................................................................ ii  Acknowledgements .................................................................................................................. iii  A Brief Note on Terminology ................................................................................................. iv  Abbreviations ........................................................................................................................... xi  Amino Acids ........................................................................................................................... xvi  Chapter 1  General Introduction .......................................................................................... 1  1.1  Membrane composition and function .......................................................................... 1  1.1.1  Common lipids found in membranes ................................................................... 1  1.1.1.1  Phospholipids .................................................................................................... 4  1.1.1.2  Sphingolipids: ceramide and sphingoglycolipids ........................................... 11  1.1.1.3  Sterols ............................................................................................................. 11  1.1.2  Cell membrane proteins ...................................................................................... 12  1.2  Phospholipids in membranes in eukaryotic cells ....................................................... 14  1.2.1  The diversity of membranes ............................................................................... 14  1.2.2  The asymmetry of the lipid bilayer .................................................................... 16  1.3  The importance of phosphatidylcholine in membranes ............................................. 17  1.4  General phospholipid biosynthesis pathway .............................................................. 18  1.4.1  Diacylglycerol and CDP-diacylglycerol ............................................................. 18  1.4.2  Biosynthesis of phospholipids from diacylglycerol ........................................... 20  1.4.3  Biosynthesis of phospholipids from CDP-diacylglycerol .................................. 22  1.5  CTP: phosphocholine cytidylyltransferase ................................................................ 24  1.5.1  Domain structure, localisation and expression ................................................... 24  1.5.1.1  The localisation and tissue-specific expression .............................................. 25  1.5.1.2  Characterisation of domains ........................................................................... 26  1.5.2  Regulation of CCT ............................................................................................. 30  1.5.2.1  Transcriptional regulation ............................................................................... 30  1.5.2.2  Post-translational regulation ........................................................................... 30  1.5.3  Membrane-binding amphipathic helices ............................................................ 39  1.5.3.1  Classical amphipathic helices and ALPS motifs ............................................ 39  vi    1.5.3.2  CCT has ALPS like motifs involved in membrane remodelling .................... 40  1.6  Evidence for the physiological importance of PCYT1A ........................................... 44  1.6.1  Congenital lipodystrophy and fatty liver disease ............................................... 44  1.6.2  Spondylometaphyseal dysplasia with cone-rod dystrophy ................................ 45  1.6.3  Leber congenital amaurosis ............................................................................... 45  1.7  Overview: experimental aims .................................................................................... 46  Chapter 2  Materials and Methods .................................................................................... 47  2.1  Saccharomyces cerevisiae cell culture ...................................................................... 47  2.1.1  Saccharomyces cerevisiae strains ...................................................................... 47  2.1.2  Saccharomyces cerevisiae growth ..................................................................... 48  2.1.3  Lithium acetate transformation of yeast ............................................................. 49  2.2  Cloning strategies ...................................................................................................... 50  2.2.1  Extraction of yeast genomic DNA ..................................................................... 50  2.2.2  Yeast DNA plasmids .......................................................................................... 50  2.2.3  PCR-based PCT1 gene modification ................................................................. 54  2.2.4  GeneArt cloning ................................................................................................. 55  2.2.5  Site-directed mutagenesis .................................................................................. 58  2.2.6  In-Fusion cloning ............................................................................................... 59  2.2.7  Amplification and expression of the PCYT1A gene from cDNA constructs ..... 60  2.3  Cloning techniques overview .................................................................................... 63  2.3.1  Polymerase chain reaction ................................................................................. 63  2.3.2  Agarose gel electrophoresis ............................................................................... 63  2.3.3  Agarose gel extraction ....................................................................................... 63  2.3.4  Restriction enzyme DNA digestion ................................................................... 64  2.3.5  Alkaline phosphatase treatment ......................................................................... 64  2.3.6  DNA Ligation .................................................................................................... 64  2.3.7  Bacterial transformation ..................................................................................... 64  2.3.8  Colony PCR ....................................................................................................... 65  2.3.9  Plasmid DNA purification ................................................................................. 65  2.3.10  Plasmid DNA sequencing .................................................................................. 66  2.4  Fluorescence microscopy .......................................................................................... 66  2.5  Confocal microscopy ................................................................................................. 66  2.5.1  Confocal microscopy of Saccharomyces cerevisiae .......................................... 66  vii   2.5.2  Quantification of lipid droplet area .................................................................... 67  2.5.3  Photobleaching assay .......................................................................................... 67  2.5.4  Quantification of GFP tagged protein localisation ............................................. 67  2.6  Immunoblotting of yeast protein ................................................................................ 68  2.6.1  Preparation of yeast lysate for western blotting ................................................. 68  2.6.2  SDS-PAGE electrophoresis ................................................................................ 68  2.6.3  Western blotting ................................................................................................. 68  2.7  Saccharomyces cerevisiae screening assay ............................................................... 69  2.7.1  5-Fluoroorotic acid assay ................................................................................... 69  2.7.2  Phleomycin assay ............................................................................................... 70  2.8  Saccharomyces cerevisiae lipidomic profiling analysis ............................................ 70  2.8.1  Lipid profiling analysis ....................................................................................... 70  2.8.2  Estimates of the lipid contribution to membrane stored curvature elastic (SCE) stress of cells ..................................................................................................................... 71  2.8.3  Estimate of the total membrane stored curvature elastic (SCE) stress of cells (referred to herein as PSCE) ................................................................................................ 73  2.8.4  Brief summary of data driven model .................................................................. 74  2.9  Isolation of Saccharomyces cerevisiae lipid droplets ................................................ 75  2.10  Mammalian cell culture ............................................................................................. 76  2.11  siRNA transfection and knockdown of target genes.................................................. 77  2.12  Lipid (Oil Red O) staining and quantification ........................................................... 77  2.13  Immunoblotting from cell extracts............................................................................. 78  2.14  PEMT activity measurement...................................................................................... 79  2.15  Lipid extraction from fibroblast cells ........................................................................ 79  2.16  Thin-layer chromatography ....................................................................................... 80  2.17  RNA Isolation, reverse transcription and RT-PCR.................................................... 80  2.17.1  RNA isolation and purification .......................................................................... 80  2.17.2  Reverse transcription polymerisation chain reaction .......................................... 80  2.17.3  Quantitative-PCR ............................................................................................... 81  2.18  Statistical analysis ...................................................................................................... 81  Chapter 3  The Yeast PCYT1A Homolog, Pct1, Modulating Phosphatidylcholine Homeostasis by a Membrane Dependent on/off Switch Mechanism ................................. 82  3.1  Abstract ...................................................................................................................... 82  viii    3.2  Introduction ............................................................................................................... 82  3.2.1  Evolutionary conservation of lipid metabolism pathways ................................. 82  3.2.2  Membrane curvature in cell biology: integration of molecular mechanisms ..... 84  3.3  Aims .......................................................................................................................... 86  3.4  Results ....................................................................................................................... 87  3.4.1  Saccharomyces cerevisiae as a suitable model organism for studying the on/off switch mechanism of PCYT1A ........................................................................................ 87  3.4.2  Pct1 localisation to the nucleus and mobility within the nucleoplasm .............. 89  3.4.3  Pct1 localisation on the inner nuclear membrane in yeast Saccharomyces cerevisiae .......................................................................................................................... 91  3.4.4  Glucose depletion inducing dissociation of Pct1 from the nuclear membrane independently of the glucose signalling machinery.......................................................... 93  3.4.5  Manipulation of the localisation of Pct1 by supplementation of choline ........ 110  3.4.6  Implementation of PE methylation pathway- and Kennedy pathway- specific knockout yeast strains in the study of Pct1 function ...................................................... 113  3.4.7  Pct1 nuclear membrane localisation regulated by substrate availability ......... 117  3.4.8  Phospholipid composition affecting Pct1 membrane affinity .......................... 120  3.4.9  Membrane elastic stress determing Pct1 membrane localisation ..................... 129  3.4.10  Lack of dependence of Pct1 membrane localisation on the source of PC synthesis .......................................................................................................................... 130  3.4.11  Overall PC composition is unaffected by the source of PC synthesis ............. 134  3.4.12  Manipulating the catalytic domain of Pct1 has no effect on membrane association of Pct1 .......................................................................................................... 138  3.4.13  Mutations disrupting the amphipathic helix of Pct1 interferes with the ability to synthesise PC .................................................................................................................. 144  3.4.14  DNA damage protection function of Pct1 in the nucleus ................................ 150  3.5  Discussion ............................................................................................................... 152  3.5.1  Proposed membrane targeting mechanism of Pct1 .......................................... 152  3.5.2  Conical phospholipids involved in PC regulation............................................ 152  3.5.3  Mammalian PCYT1A also associated with the nuclear envelope ................... 153  3.5.4  Influence of glucose availability on Pct1 membrane-binding ......................... 156  3.5.5  Scientific rationale for the intranuclear localisation of Pct1/PCYT1A ............ 157  3.5.6  Conclusions ...................................................................................................... 158  Chapter 4  The Importance of Nuclear Compartmentalisation of Pct1....................... 159  ix   4.1  Abstract .................................................................................................................... 159  4.2  Introduction .............................................................................................................. 159  4.2.1  Classical nuclear localisation signals ............................................................... 159  4.2.2  Overview of nuclear import pathways ............................................................. 160  4.2.3  Interaction of importin- with the NLS ........................................................... 161  4.2.4  Evolutionary analysis of the PCYT1A and PCYT1B genes .............................. 162  4.2.5  Prediction and alignment of classical nuclear localisation signals in human PCYT1A and yeast Pct1 proteins .................................................................................... 163  4.3  Aims ......................................................................................................................... 165  4.4  Results ...................................................................................................................... 165  4.4.1  The N-terminal basic stretch 60PRKRRRL66 is required for Pct1 nuclear localisation ...................................................................................................................... 165  4.4.2  Mutation in the nuclear localisation sequence of Pct1 resulting in plasma membrane localisation .................................................................................................... 172  4.4.3  The Pct1 NLS mutant is mobile at the plasma membrane but does not shuttle between the plasma membrane and the nucleus ............................................................. 174  4.4.4  NLS mutant compensatation for the loss of the nuclear Pct1 in vivo ............... 175  4.4.5  Disruption of Pct1 nuclear localisation does not prevent phospholipid composition changes or membrane stored curvature stress (SCE) maintenance. ........... 180  4.4.6  Ability of NLS mutant of Pct1 to rescue DNA damage sensitivity ................. 186  4.5  Discussion ................................................................................................................ 187  4.5.1  Ability of CCT NLS mutant to rescue cells with wild-type CCT deficiency .. 187  4.5.2  Cytoplasmic Pct1 can replace the nuclear Pct1 but PCYT1B cannot replace the function of PCYT1A ....................................................................................................... 190  4.5.3  Conclusions ...................................................................................................... 191  Chapter 5  Initial Characterisation of Disease-Related Human PCYT1A Mutations in Cultured Cells ....................................................................................................................... 192  5.1  Abstract .................................................................................................................... 192  5.2  Introduction .............................................................................................................. 192  5.2.1  Animal models of PCYT1A deficiency ............................................................. 192  5.2.2  Human PCYT1A mutations and the related clinical phenotype ........................ 193  5.2.3  Pathways for PC biosynthesis in humans ......................................................... 194  5.3  Aims ......................................................................................................................... 197  x    5.4  Results ..................................................................................................................... 197  5.4.1  The role of PCYT1A in mammalian cells ....................................................... 197  5.4.1.1  Effect of Pcyt1a knockdown on adipocyte differentiation ........................... 197  5.4.1.2  PCYT1A mutations impairing the Kennedy pathway ................................... 201  5.4.1.3  PEMT does not compensate for deficits in the Kennedy pathway ............... 206  5.4.2  Characterisation of PCYT1A in yeast Saccharomyces cerevisiae ................... 207  5.4.2.1  A yeast expression system for human mutant PCYT1A alleles .................... 207  5.4.2.2  PCYT1A mutant alleles reveal different phenotypes in the cho2∆opi3∆pct1∆ yeast strain ................................................................................................................... 210  5.5  Discussion ............................................................................................................... 213  5.5.1  The role of PCYT1A in adipogenesis .............................................................. 213  5.5.2  The impact of PCYT1A mutations on PC and related metabolites in primary patient cells ..................................................................................................................... 214  5.5.3  Implications of the mutations on PCYT1A: expression and function .............. 215  5.5.4  Conclusions ...................................................................................................... 217  Chapter 6  General Discussion ......................................................................................... 218  6.1  Pct1 is intranuclear and relocates to the nuclear membrane in response to the need for membrane PC synthesis ................................................................................................ 218  6.2  Compensation by nuclear localisation signal (NLS) mutants for the loss of endogenous Pct1 ................................................................................................................. 220  6.3  Disease-related human PCYT1A mutations impair PC synthesis in primary patient cells and adequate PC synthesis is essential for adipogenesis ........................................... 221  6.4  Future directions ...................................................................................................... 223  6.4.1  Characterisation of the nuclear membrane: PL composition and stored curvature elastic stress .................................................................................................................... 223  6.4.2  Validation of the DNA damage protection function of yeast Pct1/human PCYT1A ......................................................................................................................... 224  6.4.3  Investigation of Cpt1 localisation .................................................................... 226  6.4.4  Characterisation of the phenotype of cells affected by disease-associated PCYT1A mutations ......................................................................................................... 227  6.5  Concluding remarks ................................................................................................ 228  References ............................................................................................................................. 229  Appendix: Publication ......................................................................................................... 256 xi   Abbreviations AdoHCys (SAH) S-adenosyl homocysteine AdoMet (SAM) S-adenosyl methionine ADR 1-acyldihydroxyacetone phosphate reductase AGPAT 1-acylglycerol-3-pohsphate-acyltransferase AH Amphipathic helices AI Auto-inhibitory segment ALPS Amphipathic lipid-packing sensor aP2 Adipocyte protein 2 BHMT Betaine-homocysteine methyltransferase cAMP Cyclic adenosine monophosphate CCT CTP: phosphate cytidylyltransferase CCT CTP: phosphate cytidylyltransferase A, encoded by PCYT1A CCT CTP: phosphate cytidylyltransferase B, encoded by PCYT1B cDNA Complementary DNA C-domain Catalytic domain CDP Cytidine diphosphate CDP-Cho Cytidinediphosphate choline CDP-DAG Cytidinediphosphate-diacylglycerol CDP-Etn Cytidinediphosphate ethanolamine CDS Cytidinediphosphate-diacylglycerol synthase cER Cortical endoplasmic reticulum CGL Congenital generalized lipodystophy Cho Choline Cho1 Phosphatidylserine synthase Cho2 Phosphatidylethanolamine methyltransferase CK Choline kinase CL Cardiolipin (Diphosphadiylglycerol) CLD Congenital lipodystrophy CLD-FL Congenital lipodystrophy and severe fatty liver disease CLS Cardiolipin synthase CMP Cytidine monophosphate CPT CDP-choline:1,2-diacylglycerol cholinephosphotransferase xii    CRD Cone-rod dystrophy CTP Cytidine triphosphate DAG Diacylglycerol DGAT Diacylglycerol acyltransferase DHAPAT  Dihydroxyacetone phosphate acyltransferase DIC Differential interference contrast DMEM Dulbecco's modified eagle medium DNA Deoxyribonucleic acid dNTP Deoxynucleotide triphosphates DSB Double strand breaks DTT Dithiothreitol ECL Enhanced chemiluminescence ECT Ethanolamine-phosphate cytidylyltransferase (CTP: phosphoethanolamine cytidylyltransferase) EDTA Ethylenediaminetetraacetic acid EK Ethanolamine kinase EMSC Ear mesenchymal stem cells EPT (CEPT) CDP-ethanolamine:1,2-diacylglycerol ethanolaminephosphotransferase (choline/ethanolamine phosphotransferase) ER Endoplasmic reticulum Etn Ethanolamine Exp Exponential phase FA Fatty acid FBS Fetal bovine serum FLIP Fluorescence loss in photobleaching FOA 5-Fluoroorotic acid FPL Familial partial lipodystrophy FRAP Fluorescence recovery after photobleaching G-3-P Glycerol-3-phospate GAPDH Glyceraldehyde 3-phosphate dehydrogenase GFP Green fluorescent protein GPAT Glycerol-3-phosphate acyltransferase GTP Guanosine-5'-triphosphate xiii   HDL High-density lipoproteins INM Inner nuclear membrane Ist2 Increased sodium tolerance protein LCA Leber congenital amaurosis LDs Lipid droplets LPCAT Lysophosphatidylcholine acyltransferase LysoPC Lysophosphatidylcholine mAH Membrane-inducible amphipathic helix MAT Methionine adenosyltransferase M-domain Membrane-binding domain MMS Methyl methanesulfonate MS Methionine synthase MSP Major sperm protein NADPH Nicotinamide adenine dinucleotide phosphate NAFLD Non-alcoholic fatty liver disease NE Nuclear envelope NLS Nuclear localisation signals NM Nuclear membrane NPCs Nuclear pore complexes N-region N-terminal region Nups Nucleoporins OA Oleic acid ONM Outer nuclear membrane OPI3 Phosphatidyl-N-methylethanolamine methyltransferase ORO Oil Red O PA Phophatidic acid PAGE Polyacrylamide gel electrophoresis PAP Phosphatidic acid phosphatase PBS Phosphate buffered saline PC Phosphatidylcholine P-Cho Phosphocholine PCP Phosphatidylcholine: ceramide cholinephosphotransferase PCR Polymerase chain reaction Pct1 CTP: phosphate cytidylyltransferase xiv    (Cholinephosphate cytidylyltransferase) PDME (DMPE) Phosphatidyl-dimethylethanolamine  (Dimethylphosphatidylethanolamine) PDS Post-diauxic shift phase PE Phosphatidylethanolamine PEMT Phosphatidylethanolamine N-methyltransferase P-Etn Phosphoethanolamine PG Phosphatidylglycerol PGP Phosphatidylglycerol phosphate PGP-Pase Phosphatidylglycerol phosphate phosphatase PGPS Phosphatidylglycerol phosphate synthase PI Phosphatidylinositol PIPs Phosphorylated phosphatidylinositol derivatives PIS Phosphatidylinositol synthase PL Phospholipid PLA Phospholipase A PLB Phospholipase B PLC Phospholipase C PLD Phospholipase D PM Plasma membrane PMME (MMPE) Phosphatidyl-monomethylethanolamine  (Monomethylphosphatidylethanolamine) P-region C-terminal phosphorylated domain PS Phosphatidylserine PSD Phosphatidylserine decarboxylase PSS1 Phosphatidylserine synthase-1 PSS2 Phosphatidylserine synthase-2 RanGAP Ran GTPase-activating protein RanGEF Ran guanine nucleotide exchange factor RNA Ribonucleic acid RT Room temperature SCE stress Stored curvature elastic stress Scs2 Suppressor of choline sensitivity 2 proteins SD Standard deviation xv   SDS Sodium dodecyl sulfate siRNA Small interfering RNA SM Sphingomyelin SMD Spondylometaphyseal dysplasias SMD-CRD Spondylometaphyseal dysplasia with cone-rod dystrophy SMS-1 Sphingomyelin synthase-1 TAG Triacylglycerol TAGL Triacylglycerol lipase Tcb TriCalBins, Three calcium and lipid binding proteins TLC Thin-layer chromatography VAP Vesicle-associated membrane protein-associated protein VLDL Very low-density lipoproteins WT Wild-type YPD Yeast extract peptone dextrose xvi    Amino Acids Alanine Ala A Arginine Arg R Asparagine Asn N Aspartic acid Asp D Cysteine Cys C Glutamic acid Glu E Glutamine Gln Q Glycine Gly G Histidine His H Isoleucine Ile I Leucine Leu L Lysine Lys K Methionine Met M Phenylalanine Phe F Proline Pro P Serine Ser S Threonine Thr T Tryptophan Trp W Tyrosine Tyr Y Valine Val V 1    Chapter 1 General Introduction 1.1 Membrane composition and function The cell membrane is surrounded by a cell wall in bacteria, fungi, algae and plant cells. The major components of the cell wall are polysaccharides, which form a strong fabric by linking to one another. Although the cell wall provides strength and rigidity for the cell, it is absent in animal cells and protozoa. The latter are surrounded by biological membranes composed of lipids and proteins that together form hydrophobic barriers that segregate the cellular and sub-cellular compartments (or organelles) of living organisms from the environment. These well-defined compartments limit the distribution of aqueous macromolecules and metabolites and, as a result, facilitate distinct chemical processes by concentrating specific chemicals. The semi-permeable membranes are stable and also flexible due to fundamentally being made up of a phospholipid bilayer, which allows proteins to move freely across the surface. Accumulations of a particular lipid composition in the membrane can change curvature or attract proteins with specific targeting domains, e.g. caveolae. This section summarises the molecular components from which the cellular membranes are constructed. 1.1.1 Common lipids found in membranes The lipids that constitute the cell membrane can be divided into four classes: phospholipids, sphingolipids, glycolipids and sterols (Figure 1.1A), which amount to approximately 50% of the mass of most animal cell membranes (the rest is protein). All of the lipid molecules in cell membranes are amphipathic (Figure 1.1B); one end (the ‘polar head group’) is hydrophilic; the remainder is a hydrophobic tail (the ‘nonpolar fatty acid chain’ or ‘hydrocarbon chain’). Being amphipathic, lipid molecules spontaneously form micelles or bilayers in aqueous environments, that is with the hydrophilic heads towards the surrounding aqueous medium and the hydrophobic tails tending to be buried away from the water to form a non-polar environment. In the lipid bilayer (Figure 1.2), the hydrophilic heads interact favourably with solvents at each surface of the bilayer and the hydrophobic tails aggregate together in the interior. This conformational assembly is a result of the thermodynamic drive for the system to minimise contact between the nonpolar portions of the lipids and polar water. 2    Membrane lipids Phospholipids Glycerophospholipids Sphingophospholipids Sphingolipids Ceramide Glycolipids Sphingoglycolipids Sterols Typical lipids Schematic diagram Glycerophospholipids See also figure 1.3 and table 1.1. Sphingophospholipids  Sphingomyelin (shown) A B Head-group Phosphate Glycerol Fa tty ac id Fa tty ac id Fat ty aci d Sphingosine Head-group Phosphate 3    Figure 1.1 Schematic diagrams of the major classes of membrane lipids present in eukaryotic cells. Figure 1.2 A simplified representation of a lipid bilayer in an aqueous environment. Ceramide Sphingoglycolipids  Cerebrosides (Ceraimide momosaccharide)  Globosides (di- ,tri- or tetrasaccharide)  Gangliosides (complex oligosaccharide) Sterols  Cholesterol  Ergosterol (fungi) Hydrophilic head Hydrophobic tails (fatty acids) Amphipathic lipids H2O H2O H2O H2O H2O H2O H Saccharide Fa tty ac id Fa tty ac id Sphingosine Sphingosine HO Polycyclic aromatic hydrocarbons Hydrocarbon chain 4    1.1.1.1 Phospholipids The most common membrane lipids, phospholipids, include glycerophospholipids and sphingophopholipids (Figure 1.1). Glycerophospholipids are composed of a head group, a phosphate group, glycerol and two fatty acid chains. Glycerol is a three-carbon molecule that functions as the scaffold of these membrane lipids. Two fatty acyl chains are attached through ester bonds at the sn-1 and sn-2 positions, and a phosphate group is attached at the sn-3 position of the glycerol core (Figure 1.3). Glycerophospholipids can have many different combinations of fatty acids of varying lengths and saturation attached at the sn-1 and sn-2 positions. Various head-group substitutes are linked via the phosphate residue. The head-group moieties (Fahy et al., 2011) can be a base (choline, ethanolamine), a polyol (inositol, glycerol, phosphatidylglycerol), an amino acid (serine), or hydrogen (Table 1.1). Figure 1.3 A general structure of glycerophospholipid. sn-3 sn-2 sn-1 Head-group substituents Phosphate Glycerol Fatty acyl chains 5    Table 1.1 Typical glycerophospholipids (related to Figure 1.3). Name of head-group substituents Formula of head-group substituents Phospholipid name (Abbreviate) Net charge (at pH 7) Choline Phosphatidylcholine (PC) 0 Ethanolamine Phosphatidylethanol- amine (PE) 0 Inositol Phosphatidylinositol (PI) -4 Glycerol Phosphatidylglycerol (PG) -1 Phosphatidylglycerol Diphosphatidylglycerol (Cardiolipin; CL) 0 Serine Phosphatidylserine (PS) -1 Hydrogen Phosphatidic acid (PA) -1 The predominant phosphatidylcholine (PC) and phosphatidylethanolamine (PE) are zwitterions with no net charge at physiological pH because the polar head-group introduces a positively- charged amine group, whilst the phosphate group retains a negative charge. The phosphatidylinositol (PI) molecule, with a net charge of -4, contains a negatively-charged phosphate group and a unique head-group inositol in which the 3’, 4’, and 5’ carbons can be phosphorylated by PI-kinases. PI and the other negatively-charged phosphatidylglyerol (PG), phosphatidylserine (PS) and phosphatidic acid (PA) typically comprise 10–20 mol % of membrane lipids. These anionic lipids can be exploited by cells for a variety of purposes. For example, charged PI plays a role in response to extracellular signals and helps to target small 6    GTPases with polybasic regions to the cytosolic face of the plasma membrane (Heo et al., 2006). If the negatively-charged PS is exposed on the outer leaflet of the plasma membrane, macrophages will recognise this as a cellular apoptosis signal and engulf the cell (Dickey and Faller, 2008). In glycerophospholipids, the fatty acyl chains are varied in terms of length and degree of saturation. Some fatty acyl chains are unbranched and fully saturated (contains no double bonds); others contain one or more double bonds. The most commonly occurring fatty acids have unbranched chains of 16 or 18 carbons (Table 1.2.1). A simplified but unambiguous nomenclature for these compounds specifies the chain length and number of double bonds, separated by a colon; for instance, 16:0 is for the 16-carbon saturated palmitic acid, and 18:1 is for the 18-carbon oleic acid, with one double bond. In this notation, PC (34:1) implies a PC molecule having carboxylate substituents at sn-1 and sn-2 positions on glycerol with a total of 34 carbons and a single site of unsaturation. The possible acyl combinations of PC (34:1) could be i) one chain of palmitic acid (16:0) and one chain of oleic acid (18:1), or ii) one chain of stearic acid (18:0) and one chain of palmitoleic acid (16:1). Another example is that the fatty acyl chains of PC (34:2) could consist of i) one chain of palmitoleic acid (16:1) and one chain of oleic acid (18:1), or ii) one chain of palmitic acid (16:0) and one chain of linoleic acid (18:2). Note that (16:2), (14:2) or (20:2) are unusual fatty acids. Table 1.2.2 contains the reported fatty acid composition of several yeast strains. The most common fatty acids include (16:0), (18:0), (18:1), (16:1) and (18:2). The composition of fatty acids varies across yeast species. Oleic acid (18:1) and palmitoleic acid (16:1) are the most abundant fatty acids in Saccharomyces cerevisiae. The chain length of fatty acids is synthesised by successive additions of a two-carbon unit- acetyl CoA; thus, the fatty acyl chain predominantly comprises an even number of carbon atoms. Each chain is generally between 12 and 20 carbons in length. A shorter chain length reduces the tendency of the hydrocarbon tails to attract one another so it increases membrane fluidity at lower temperature (the transition of the membrane from a liquid state to a rigid crystalline state is more difficult). Fatty acyl chains can contain up to six double bonds. The presence of double bonds in the fatty acid creates kinks (Small, 1984) that reduce tight packing. Hence, membranes rich in unsaturated fatty acids tend to be more fluid. The double bond geometry is usually cis in nature, which means that the carbon moieties lie on the same side of the double bond. When there is more than one double bond, each cis-configuration is separated by a single methylene group. 7    Table 1.2.1 Common fatty acids. FA Common name Systemic name Structurea Saturated fatty acids 16:0 Palmitic acid Hexadecanoic acid CH3(CH2)14COOH 18:0 Stearic acid Octadecanoic acid CH3(CH2)16COOH 14:0 Myristic acid Tetradecanoic acid CH3(CH2)12COOH Unsaturated fatty acids 18:1 Oleic acid 9-Octadecanoic acid CH3(CH2)7CH=CH(CH2)7COOH 16:1 Palmitoleic acid 9-Hexadecanoic acid CH3(CH2)5CH=CH(CH2)7COOH 18:2 Linoleic acid 9,12-Octadecanoic acid CH3(CH2)4(CH=CHCH2)2(CH2)6COOH a Source: The PubChem Project. https://pubchem.ncbi.nlm.nih.gov/ 8    Table 1.2.2 Fatty acid profiles of several yeast strains, mean and (range)a Species 16:0 18:0 18:1 16:1 18:2 othersb Saccharomyces cerevisiae 11.07 (8.86–14.16) 1.58 (0.95–2.23) 24.92 (21.16–28.68) 55.31 (50.46–62.01) ND c 6.48 Candida albicans 15.37 (11.26-19.48) 3.31 (2.45–4.17) 34.74 (29.95–39.53) 13.43 (11.20–15.66) 26.58 (23.13–30.03) 3.01 Candida famata 13.52 (11.10–17.22) 2.60 (2.00–3.90) 49.31 (44.21–54.11) 2.87 (1.94–4.72) 27.89 (24.15–29.95) 2.40 Candida glabrata 8.99 (7.41–10.57) 4.69 (3.74–5.98) 30.01 (27.05–32.97) 50.45 (45.58–55.33) ND c 2.58 Candida guilliermondii 14.82 (11.38–16.97) 2.65 (1.94–3.87) 43.09 (40.94–45.89) 8.90 (6.52–11.24) 28.77 (24.10–33.67) 2.83 Candida pseudotropicalis 17.49 (14.36–20.62) 1.94 (1.60–2.28) 34.67 (30.69–38.65) 12.77 (10.60–14.94) 28.83 (24.70–32.96) 8.29 Candida krusei 15.52 (13.40–17.67) 1.26 (0.88–1.64) 54.46 (51.34–57.58) 4.29 (3.14–5.55) 18.39 (15.40–21.38) 3.97 Candida lambica 12.85 (9.95–15.95) 0.95 (0.65–1.40) 37.15 (33.94–40.45) 18.43 (15.18–22.63) 26.08 (23.20–29.72) 4.66 Candida lusitaniae 16.31 (15.58–17.61) 1.87 (1.24–2.35) 40.45 (38.35–42.41) 6.87 (4.88–7.73) 32.95 (30.41–35.91) 1.92 Candida parapsilosis 14.78 (12.03–17.53) 3.52 (2.15–4.89) 51.84 (47.81–55.87) 3.06 (2.84–4.28) 22.45 (19.34–25.56) 2.63 Candida tropicalis 16.95 (14.45–19.39) 4.86 (3.74–5.98) 44.22 (38.59–49.85) 9.80 (7.67–11.93) 22.32 (18.95–25.69) 2.92 Candida sphaerica 13.75 (10.95–16.90) 1.45 (1.10–1.90) 29.95 (27.06–33.07) 18.50 (14.55–23.62) 31.53 (27.40–34.83) 3.80 Candida utilis 16.40 (13.73–19.36) 1.46 (0.90–1.80) 47.85 (43.83–51.16) 2.65 (2.15–4.07) 29.08 (25.10–33.70) 1.88 Candida valida 12.70 (10.35–16.64) 1.29 (0.75–1.90) 47.70 (42.78–52.23) 13.01 (10.76–16.79) 21.18 (18.82–25.05) 3.30 Cryptococcus neoformans 15.27 (11.35–19.02) 4.50 (2.55–6.28) 37.50 (34.35–40.03) ND c 41.80 (36.60–47.90) 1.90 Trichosporon spp. 15.42 (12.07–18.91) 2.10 (1.56–3.00) 34.37 (31.17–38.57) 5.18 (3.49–6.98) 36.75 (31.85–41.73) 3.41 a Data are adapted from (El Menyawi et al., 2000). b Most analyses indicate presence of (10:0), (12:0), (14:0), (14:1), (17:0), (17:1) fatty acids. c ND: Not detected. 9    Sphingophospholipids (Figure 1.1B) are a class of lipids containing a backbone of sphingoid base; for example, the most common sphingomyelin consists of sphingosine, a fatty acid tail and either a phosphocholine or phosphoethanolamine group (Fahy et al., 2011). Sphingomyelins (SM) have physical properties similar to those of glycerophospholipids. However, the fatty acyl chain of sphingomyelin is usually more saturated than that of other phospholipids.  Sphingomyelin is predominantly found in the plasma membrane, and is particularly enriched in the myelin sheath of the nervous system. Sphingomyelin is not present in yeast (Sawai et al., 2000), though the sphingolipid metabolic pathways and corresponding enzymes are conserved across fungi, plant and animal kingdoms. The phospholipid shape, frequently, used to describe the volume occupied by phospholipids, depends on the relative volumes of their polar head groups and fatty acyl chains. PC, PI, PG and SM have a nearly cylindrical molecular geometry and can spontaneously self-organise as a planar bilayer. Conversely, PE, PS, PA, cardiolipin (CL) and diacylglycerol (DAG) assume a conical shape because their polar head group has a smaller cross-section than the tail (Figure 1.4 A and B). The inclusion of conically shaped lipids in cylindrical lipid bilayers imposes a curvature stress onto the membrane, which is characterised by negative spontaneous curvature (phospholipids bend to make the hydrophilic head-group come closer together), and provides membranes with potential for tubulation, budding, fission and fusion (van Meer et al., 2008). These processes are essential for key physiological functions of living cells, including cell division, intracellular membrane trafficking and biological reproduction. An additional consequence of low PC in lipid membranes is the frequency of lipid packing defects (Figure 1.4 C). These ‘packing defects’ can be recognised by hydrophobic surfaces of protein helices, which enable specific proteins to associate with the lipid membrane (Thiam et al., 2013).   10    Figure 1.4 Cross-sectional shapes of phospholipids and lipid bilayer organisation. (A) Schematic representation of the molecular shapes of PC and PE. PC with a bulky polar head group has a molecular shape similar to a cylinder. PE with a smaller polar head group has a conical shape. (B) Phospholipids can be classed as cylinders (e.g., PC, PI and PG) and cones (e.g., PE, PS, PA and DAG), depending on the relative volumes of their polar head groups and fatty acyl chains. (C) (i) A monolayer with lipids with smaller polar head groups has high interfacial curvature (negative curvature). The scheme also shows such molecules might generate high hydrophobicity exposure and high surface packing defects in bilayers. On the contrary, (ii) lipids with bulky polar head groups have a lower propensity for interfacial curvature and hence generate flattened bilayers. Adapted from (Bigay and Antonny, 2012; Jouhet, 2013). 11    1.1.1.2 Sphingolipids: ceramide and sphingoglycolipids Sphingosine with a fatty acyl chain is categorised as a ceramide (Figure 1.1), which is an important component of the stratum corneum intercellular lipid lamellae. The stratum corneum is the outermost epidermal layer and is a compact layer of dead skin cells and lipids. It is the major barrier to protect the living cells beneath it from the external environment. Ceramide in the stratum corneum is required for the creation of the surface water barrier (Coderch et al., 2003). Ceramide can also participate in several cellular processes such as cell proliferation, angiogenesis, apoptosis, inflammation, and others (Hannun and Obeid, 2008). Sphingoglycolipids consist of a saccharide residue in glycosidic linkage to the C-1 hydroxyl group of the ceramide moiety (Figure 1.1B). They are a subclass of glycolipids found in the cell membranes of animals. In microbes and plants, the major subclass of glycolipids is glycoglycerolipids, which have a different structure with saccharides linked to the C-3 hydroxyl of DAG. Sphingoglycolipids can be divided into three categories based on the carbohydrate moieties: 1) cerebroside is characterised by the linking of ceramide to a monosaccharide unit such as glucose, galactose or sulfatides; 2) globoside is characterised by the linking of ceramide to di-, tri- or tetrasaccharides; and 3) ganglioside is characterised by the inclusion of one or more sialic acids linked to complex oligosaccharides. The importance of sphingoglycolipids lies primarily in cell recognition and transmembrane signalling because of their glycan markers (Hakomori, 1993; Hakomori and Igarashi, 1995). Sphingoglycolipids are usually expressed in the outer leaflet of the plasma membrane with their glycans facing the external milieu. The sphingoglycolipids mediate cell–cell interactions via binding to counterpart sphingoglycolipid molecules on opposing plasma membranes or through recognition by sphingoglycolipid-binding proteins. In addition, some sphingoglycolipids directly modulate the activities of receptor tyrosine kinase families and protein kinase C, leading to cis regulation in the same membrane (D'Angelo et al., 2013). 1.1.1.3 Sterols Sterols are another lipid component of animal cell membranes. One important sterol is cholesterol, which is composed of a polar hydroxyl group, a hydrophobic ring structure region with four hydrocarbon rings and a hydrophobic hydrocarbon tail (Figure 1.1B). Due to its chemical composition, its small hydroxyl groups are in close proximity and attract the polar head 12    groups of neighbouring lipids, whilst its rigid hydrophobic chains interact with the fatty acyl chains of the phospholipids. The increased density of the hydrophobic section reduces the motion of the neighbouring acyl chains; thus, cholesterol molecules spanning the cell membrane reduce membrane fluidity. At the same time, this helps to keep cell membranes from becoming stiff/freezing by preventing phospholipids from being too closely packed together. Generally, cholesterols decrease fluidity at high temperatures and increase fluidity at low temperatures (de Meyer and Smit, 2009; Monteiro et al., 2014). Cholesterol molecules, although less abundant than glycerophospholipids, account for about 10–20% of the lipids in mammalian cell membranes (Table 1.3), especially the cell plasma membrane, in which they are found in higher concentrations. Ergosterol is equivalent to mammalian cholesterol and is the main sterol found in fungal cells (Klug and Daum, 2014). Table 1.3 Lipid composition of a common nucleated mammalian cell Percentage of total lipidsa Phosphatidylcholine 45–55 Phosphatidylethanolamine 15–25 Phosphatidylinositol 10–15 Phophatidylglyerol <1 Cardiolipin 2–5 Phosphatidylserine 5–10 Phosphatidic acid 1–2 Sphingomyelin 5–10 Sphingoglycolipids 2–5 Cholesterol 10–20 a Data are adapted from (Vance, 2015) claimed as averaged from several sources. 1.1.2 Cell membrane proteins The first complete model of membrane structure was proposed by Jonathan Singer and Garth Nicolson (Singer and Nicolson, 1972). Since then scientists have viewed membranes as a fluid mosaic in which proteins are embedded into a lipid bilayer. The lipid bilayer is a dynamic two- dimensional fluid and recent research has added much detail to this model. In addition to individual lipid molecules that can rotate about their long axis, their hydrocarbon chains are 13    flexible, and both lipid molecules and proteins can diffuse laterally within each monolayer. The phospholipid molecules are able to migrate from one leaflet to the other (known as ‘flip-flop’ transition) by the aid of transmembrane protein flippases, floppases and scramblases. Membrane proteins account for roughly half the mass of most cellular membranes and they have a number of specific functions. These proteins are divided into two general classes based on their relationship with the membrane. Integral membrane proteins are directly embedded within the lipid bilayer and most pass through the membrane. Peripheral membrane proteins are not inserted into the lipid bilayer but are indirectly associated with the membrane, usually by interplay with integral membrane proteins. Many of the integral proteins found in membranes expose largely non-polar surfaces to the hydrophobic core of the membrane. Membrane-bound proteins can often be identified from their amino acid composition: the membrane-spanning portions are usually α-helical regions of around 20 nonpolar residues (De Marothy and Elofsson, 2015). The membrane-spanning regions are often repeated; in other words, some transmembrane proteins stick out of and back into the membrane multiple times. The regions that stick out on both sides of the membrane are typically hydrophilic. Transmembrane proteins are highly functionalised. For example, cell membrane receptor proteins help cells communicate with the outside environment through interaction with a molecule (ligand) and then trigger a conformational change in the protein that transmits a signal to intracellular second messenger molecules. The ligands might be hormones, neurotransmitters or other signalling molecules.  Glycoproteins (proteins with carbohydrates attached) help in molecular recognition processes and cell–cell interactions. Transport proteins are selective for the molecules they move. Some of these channels transport molecules across cell membranes by concentration or electrochemical gradients. Others transport molecules against these gradients; usually this is coupled to ATP hydrolysis as a source of energy. Interestingly, a transmembrane protein is able to exploit anionic lipids around it to attract positively-charged ions near the protein opening and hence facilitate transport of the ions (Lee, 2004). Some membrane proteins are structural proteins, tethered to cytoskeletal elements, which function to support the cell shape. Cytoskeletal elements or extracellular ligands can inversely trigger the clustering of regional membrane components; these are called lipid rafts. Lipid rafts are small but specialised areas in membranes where particular lipids (primarily sphingolipids and cholesterol) and certain proteins 14    accumulate.  These specialised membrane microdomains are in highly ordered array, but do not hinder their lateral movement in the membrane bilayer (Simons and Ehehalt, 2002). Raft clustering is thought to serve as an organising centre for special functions, such as the assembly of more stable structures, and the regulation of neurotransmission and receptor trafficking (Korade and Kenworthy, 2008; Pike, 2009). Although they are more common in the plasma membrane, the existence of raft-like domains in intracellular membranes has not been entirely excluded. 1.2 Phospholipids in membranes in eukaryotic cells 1.2.1 The diversity of membranes In contrast to prokaryotes, eukaryotic cells possess not only an outer plasma membrane (PM), but also membrane-bound organelles, such as the nucleus, mitochondria, endoplasmic reticulum (ER), vacuoles, lysosomes and Golgi apparatus, and in plants chloroplasts. Most of these subcellular structures are enclosed by a single lipid bilayer like the PM, whilst the nucleus, mitochondria and chloroplast are surrounded by double lipid bilayers. For the nucleus, the inner and outer nuclear membranes are continuous structures, and the outer nuclear membrane is also continuous with the ER. The double membrane enclosures of mitochondria and chloroplasts are discontinuous but play an essential role in oxidative phosphorylation (synthesis of ATP), which is similar to certain bacterial systems. It is thought to be related to the evolutionary origins of these organelles. The membranes of the different organelles vary in composition and function. The versatility of biological membranes is dependent on the types of lipids and proteins that compose the membranes. Their different biophysical properties relate in part to their distinct phospholipid composition (see Table 1.4; the values are approximate since each isolated membrane is probably contaminated by other organelle membranes). PC constitutes the predominant phospholipid (PL) proportion in all these subcellular membrane systems. SM and cholesterol/ergosterol are more abundant in the PM compared with other organelles. The molar ratio of cholesterol/phospholipid in the PM is about ten times greater than in the ER and mitochondria. Also, in yeast, the difference is five-fold. This means that PMs are packed at a higher density and can thus protect cells against external mechanical stress. CL is located almost exclusively in the inner mitochondrial membrane and is involved in generating ATP through the electron transfer process. The enrichment of CL in this specific membrane may occur because this is the location in which 15    CL biosynthesis occurs. Furthermore, CL as a non-bilayer phospholipid used to accommodate electron transport complexes and optimise their activities (Paradies et al., 2014). The non-bilayer phospholipids are cone‐shaped molecules, such as CL and PE, which introduce curvature stress in the bilayer membrane and have been shown to impact the assembly and activity of mitochondrial respiratory chain complexes (Basu Ball et al., 2018). The content of PE is higher in the inner and the outer membranes of mitochondria compared with that of other organelles. Since PE and CL are the major phospholipids of the bacterial membrane (Sohlenkamp and Geiger, 2016), it also reinforces the hypothesis that mitochondria are descendants of bacteria that established a symbiotic relationship with early eukaryotic cells. Membrane/phospholipid biogenesis is needed to accommodate cell growth and replication, as well as other specific cellular events. A common example of membrane biogenesis takes place in the nuclear envelope which elongates with chromosome segregation during mitosis. For another instance, the ER and Golgi apparatus expands during the maturation of naïve B-lymphocytes into mature plasma cells (Fagone et al., 2007). Table 1.4 Phospholipid composition of organelles (% total phospholipids)a Mitochondria Lipid ER Inner Outer Lysosomes Nuclei Golgi PM PC 57 41 49 42 52 45 43 PE 21 38 34 21 25 17 21 PI 9 2 9 6 4 9 7 CL 0 16 5 0 0 0 0 PS 4 1 1 1 6 4 4 SM 4 2 2 16 6 12 23 Others 5 <1 <1 14 7 13 Chol/PLb 0.07 0.06 0.49 - 0.15 0.76 Erg/PLc 0.1 0.1 - - - 0.5 a Approximate phospholipid compositions of several subcellular organelles isolated from rat liver. Data are adapted from (Vance, 2015). b Approximate Chol (cholesterol) to phospholipid (PL) molar ratios. Data are derived from a. c Approximate Erg (ergosterol) to phoshpolipid molar ratios in yeast. Data are adapted from (van Meer et al., 2008). 16    The endoplasmic reticulum is the major site where PC, PE, PI, PS and PA  are generated (van Meer et al., 2008). These phospholipids do not simply remain in the ER membrane; instead, they are spread and exchanged throughout the endomembrane system in an organised fashion. This also applies to the Golgi biosynthesised SM (van Meer et al., 2008). The distinct phospholipid composition of organelle membranes is maintained through several mechanisms, some of which are discussed below. 1.2.2 The asymmetry of the lipid bilayer Apart from the unbalanced distribution of phospholipids between organelle membranes, phospholipids are also asymmetrically distributed across the two halves of the membrane bilayer (Fadeel and Xue, 2009; van Meer, 2011; van Meer et al., 2008). PE, PI and PS are more commonly found on the inner leaflet of the Golgi, endosomal and plasma membranes, whereas PC and SM are mainly present on the external leaflet (Devaux and Morris, 2004; Fadeel and Xue, 2009). PC and SM together constitute 87% of the external leaflet and 23% of the inner leaflet in these organelle membranes (Cooley et al., 2013; Virtanen et al., 1998). Since PS and PI are negatively-charged phospholipids, their predominance creates a net negative charge in the inner leaflet. Moreover, the asymmetric distribution of PS in the plasma membrane serves important physiological functions, for example, in distinguishing between living and dead cells for macrophage recognition, as well as in initiating the blood clotting process: in the plasma membrane of the platelet, the presence of PS in the outer leaflet enhances thrombin formation (Bevers et al., 1982). The translocation of PS, in this case, is performed by a ‘floppase’ that transfers phospholipids from the cytosolic to the exoplasmic face, or a ‘scramblase’ that transfers phospholipids non-specifically in both directions between the two monolayers. In other situations, phospholipids are translocated by the following mechanisms (van Meer et al., 2008): 1) A ‘flippase’ transports phospholipid from the exoplasmic to the cytosolic face; 2) ATP-dependent aminophospholipid transporters selectively transport PS and PE from the non-cytosolic leaflet to the cytosolic leaflet in Golgi and plasma membrane; 3) Through transfer proteins at the contact sites between the two cytosolic surfaces of organelles; and (4) By vesicle-mediated transport systems which move phospholipids between cellular organelles. The asymmetric distribution of various lipids between the two bilayer leaflets is one of the mechanisms which generate spontaneous curvature (curvature stress) in biomembranes. As 17    previously mentioned in section 1.1.1.1, phospholipids have intrinsic three-dimensional shapes resulting from the relative size of their head-groups, the length of their acyl chains and the degree of saturation of their acyl chains. Accumulation of phospholipids with similar shapes in one leaflet will generate a curvature change as the lipids will aggregate and bend to minimise the lipid-water interface (surface tension). Due to the two monolayers coupling together, one monolayer bending affects the other monolayer. Therefore, the overall spontaneous curvature will reflect the net effect of these forces. More specifically, the spontaneous curvature of a lipid bilayer is equal to the difference between the spontaneous curvatures of its inner and outer leaflets (McMahon and Boucrot, 2015). By changing the composition of artificial liposomes, for example, increasing the unsaturation of acyl chains, the liposome enlarges lipid packing defects and simultaneously generates spontaneous curvature (Bigay and Antonny, 2012). Changes in the lipid composition between bilayer membranes by lipid flippases induce local membrane curvature, which is supported by the evidence of vesicular budding in the endolysosomal pathway (Ruaud et al., 2009) and in spermatogenesis (Xu et al., 2009). The displacement of aminophospholipids, such as PE and PS, from the exoplasmic to the cytoplasmic leaflet of a biological membrane increases the spontaneous curvature of the bilayer, thus helping deformation of the membrane during vesicle budding and spermatogenesis.   1.3 The importance of phosphatidylcholine in membranes Phosphatidylcholine (PC) naturally assembles into stable bilayers in an aqueous environment due to its cylindrical shape and is ideally suited to its role as the building block (‘brick’) of membranes, hence its predominance in almost all eukaryotic membranes. PC is the most abundant membrane lipid accounting for approximately 40–50% of total PL present in all organelles of mammalian cells (van Meer et al., 2008). PC constitutes the essential phospholipid component in cell membranes including the PM and other bilayered membranes surrounding organelles, such as the ER and mitochondria. It is also the major PL in the surface monolayer of lipid droplets, lipoproteins as well as in bile and pulmonary surfactant (Alberts B, 2002; Jacobs et al., 2008). The importance of PC was demonstrated using genetically modified mice possessing a deletion of exon 5–6 in the Pcyt1a gene, which disrupts the expression of the critical enzyme CCT involved in the CDP-choline pathway required for PC biosynthesis, and notably a dysfunction in this enzyme resulted in early embryonic lethality (Wang et al., 2005). This 18    requirement for PC is believed to be associated with cell growth (Guo et al., 2008; Niebergall and Vance, 2012) and membrane biogenesis (Esko et al., 1982). The requirement for PC in the maintenance of membrane integrity is balanced by the need for other phospholipids with a range of head groups and varied acyl chain composition, thus subcellular organelles can retain their ability to change shape, undergo fission/fusion events, bud off or accept vesicles, and to retain unique surface-identifying features. PC is also important for signal transduction since its catabolic derivatives diacylglycerol, phosphatidic acid and lysophosphatidylcholine (LysoPC, a PC partially hydrolyzed by phospholipaseA which removes one fatty acyl chain) are lipid second messengers, which mediate multiple cellular responses, such as activating protein kinase C (Sando and Chertihin, 1996; Shirai and Saito, 2002) or LysoPC inducing phagocyte recruitment during apoptosis (Lauber et al., 2003). 1.4 General phospholipid biosynthesis pathway 1.4.1 Diacylglycerol and CDP-diacylglycerol Both diacylglycerol (DAG) and CDP-diacylglyerol (CDP-DAG) are important phospholipid precursors synthesised from phosphatidic acid (PA) in most eukaryotes (Foster, 2013; Vance, 2015). PA is produced either from glycerol-3-phosphate or from dihydroxyacetone phosphate (Figure 1.5). Glycerol-3-phosphate acyltransferase (GPAT) catalyses the first acylation of glycerol-3-phosphate and converts it into 1-acylglycerol-3-phosphate (also called lyso- phosphatidic acid). 1-acylglyerol-3-phosphate is subsequently further acylated to phosphatidic acid by 1-acylglycerol-3-phosphate-acyltransferase (AGPAT). Finally, phosphatidic acid phosphatase (PAP) dephosphorylates PA and generates diacylglycerol on the ER membranes. Diacylglycerol can be converted into PC, PE and triacylglycerol (TAG). In the alternative pathway for PA synthesis, dihydroxyacetone phosphate is acylated with acyl-CoA in the presence of dihydroxyacetone phosphate acyltransferase  (DHAPAT) to produce 1-acyl dihydroxyacetone phosphate. Then, 1-acyldihydroxyacetone phosphate reductase (ADR) promotes a reduction of 1- acyl dihydroxyacetone phosphate and results in 1-acylglycerol-3-phosphate and then PA. Biosynthesis of CDP-diacylglycerol is catalysed by an enzyme CDP-diacylglycerol synthase (CDS), which involves condensation of phosphatidic acid and cytidine triphosphate (CTP) in the 19    ER or mitochondria membranes. The phospholipids PI, PG and CL are directly generated from CDP-diacylglycerol. Figure 1.5 Biosynthesis of diacylglycerol and CDP-diacylglycerol. Diacylglycerol and CDP-diacylglycerol are respectively formed from phosphatidic acid. GPAT, glycerol 3-phosphate acyltransferase; AGPAT, 1-acylglycerol-3-phosphate-acyltransferase; PAP, phosphatidic acid phosphatase; CDS, CDP-diacylglycerol synthase; DHAPAT, dihydroxyacetone phosphate acyltransferase; ADR, 1-acyldihydroxyacetone phosphate reductase; PA, phosphatidic acid; DAG, diacylglycerol; PC, phosphatidylcholine; PE, phosphatidylethanolamine; TAG, triacylglycerol; PI, phosphatidylinositol; PG, phosphatidylglycerol; CL; cardiolipin; NADPH, nicotinamide adenine dinucleotide phosphate. 20    1.4.2 Biosynthesis of phospholipids from diacylglycerol Diacylglycerol is involved in the CDP-choline pathway and the CDP-ethanolamine pathway (Gibellini and Smith, 2010). These pathways are widely known as the Kennedy pathway. Eugene Kennedy first elucidated these pathways in the 1950’s (Kennedy and Weiss, 1956). In the CDP- choline pathway (Figure 1.6), CDP-choline is esterified with diacylglycerol by the CDP-choline: 1,2-diacylglycerol cholinephosphotransferase (CPT) to produce PC. The rate-limiting enzyme of this pathway is CTP: phosphocholine cytidylyltransferase (CCT), which converts phosphocholine to CDP-choline. PC can alternatively be synthesised from PE via the enzyme PE methyltransferase (PEMT). SM is synthesised by the transfer of phosphocholine moiety of PC to the acceptor ceramide (Tafesse et al., 2006). This step is mainly catalysed by the Golgi enzyme SM synthase-1 (SMS1). PE is made from the CDP-ethanolamine pathway (Figure 1.6). In parallel with CPT function, CDP-ethanolamine is esterified with diacylglycerol by CDP-ethanolamine: 1,2-diacylglycerol ethanolaminephosphotransferase (EPT) to generate PE. PS (Pereira et al., 2012) is synthesised by a head-group exchange reaction from either PE or PC via different PS synthases, PSS2 and PSS1 respectively. PS is synthesised in a specialised domain of the ER, known as mitochondria- associated membranes (MAM). The enzyme PS decarboxylase (PSD) in the inner mitochondrial membrane converts the PS to PE. As illustrated in figure 1.6, diacylglycerol can be converted into TAG by diacylglycerol acyltransferase (DGAT) (Ahmadian et al., 2007). Conversely, TAG is hydrolysed to diacylglycerol by triacylglycerol lipase (TAGL). TAG is not incorporated into membranes but is deposited in lipid droplets as an energy reserve. It is insoluble in water. 21    Figure 1.6 Biogenesis of phospholipids from diacylglycerol. PC, PE, PS and SM are biosynthesised directly or indirectly from dicaylglycerol. PSS1. PS synthase-1; PSS2, PS synthase-2; PSD, PS decarboxylase; EK, ethanolamine kinase; ECT, CTP: phosphoethanolamine cytidylyltransferase; EPT, CDP-ethanolamine:1,2-diacylglycerol ethanolaminephosphotransferase; CK, choline kinase; CCT, CTP: phosphocholine cytidylyltransferase; CPT, CDP-choline:1,2-diacylglycerol cholinephosphotransferase; PEMT, PE methyltransferase; SMS1, SM synthase-1; DGAT, diacylglycerol acyltransferase; TAGL, triacylglycerol lipase. 22    1.4.3 Biosynthesis of phospholipids from CDP-diacylglycerol In most eukaryotic cells, PI is made from CDP-diacylglycerol and inositol via PI synthase (PIS) in the ER membranes (Figure 1.7). Multiple phosphorylated PI derivatives (PIPs) are derived from the PI precursor. Various PIPs are synthesised by the phosphatidylinositol phosphate kinase family (PIPK, which generates PI-4,5-P2; PI3K, which generates PI-3-P, PI-3,4-P2 and PI-3,4,5- P3) and/or some phosphatases. PIPs have seven phosphorylated forms and play key roles in cell signalling events (Falkenburger et al., 2010). PG is made from CDP-diacylglycerol by two steps (Li et al., 2016; Vance, 2015). Firstly, phosphatidylglycerol phosphate (PGP) is generated from CDP-diacylglycerol and glycerol-3- phosphate (G-3-P) by PGP synthase (PGPS). Secondly, PGP phosphatase (PGP-Pase) dephosphorylates PGP to produce PG (Figure 1.7). Both PGP and PGP-Pase are mainly active in the ER and to a lesser extent in the mitochondria. For the synthesis of CL (Figure 1.7), PG couples with a second molecule of CDP-diacylglycerol in a reaction catalysed by cardiolipin synthase (CLS), an enzyme restricted to mitochondrial inner membranes. In eukaryotes, ‘nascent’ CL is remodelled to become ‘mature’ CL by phospholipases and transacylases. The ‘mature’ CL is enriched in polyunsaturated fatty acids (Tian et al., 2012). Unlike eukaryotes, the prokaryotic CL biosynthesis reaction is a reversible transesterification via the condensation of two molecules PG and the synthesised CL is not required for further remodelling. The phospholipid CL has an unusual dimeric structure with two-phosphate head- group and four acyl chains (Table 1.1). 23    Figure 1.7 Biogenesis of phospholipids from CDP-diacylglycerol. PI, PG and CL are made from CDP-diacylglycerol. PIS, PI synthase; PIPs, phosphorylated PI derivatives; PGP, phosphatidylglycerol phosphate; PGPS, PGP synthase; PGP-Pase, PGP phosphatase; CLS, cardiolipin synthase; G-3-P, glycerol-3-phospate; CDP-DAG, CDP- diacylglycerol; CDP, cytidine diphosphate; CMP, cytidine monophosphate. 24    1.5 CTP: phosphocholine cytidylyltransferase 1.5.1 Domain structure, localisation and expression CTP: phosphocholine cytidylyltransferase (CCT) has four functional domains, a N-terminal region, a catalytic (C) domain, a membrane-binding (M) domain and a phosphorylated (P)- region (Figure 1.8). CCT is encoded by two genes, PCYT1A and PCYT1B, and there are four major mammalian isoforms, CCT, CCT1, CCT2 and CCT3.    In humans, PCYT1A gives rise to CCTα while PCYT1B encodes for CCTβ1 and CCTβ2. In mice, Pcyt1a encodes CCT while Pcyt1b encodes CCTβ2 and CCTβ3. The CCTα and CCTβ proteins are essentially identical in the catalytic and membrane-binding domains but differ at the N- and C-termini (Cornell and Northwood, 2000; Karim et al., 2003; Lykidis et al., 1999). The predicted CCT1 lacks the C- terminal tail. CCTβ3 is 28 residues shorter at the N-terminus than CCTα. CCTβ2 and CCTβ3 initiate transcription with different first exons (Karim et al., 2003). The α isoform, unlike the β, bears an N-terminal nuclear localisation signal (NLS) in the N-region. Evidence indicates that the NLS is found as a segment of amino acid residues 8 to 28 within CCT (Chen and Mallampalli, 2009; Wang et al., 1995). Deletion of the NLS increases the proportion of CCT in the cytoplasm, but has no effect on the activity of PC biosynthesis in CCT-deficient CHO cells (Wang et al., 1995).   Figure 1.8 Structural domains of mammalian CCT isoforms. There are four CCT isoforms, CCT, CCT1, CCT2 and CCT3. The ubiquitously expressed CCT has been described as having four domains. An N-terminal region housing its nuclear localization signal (NLS) sequence is followed by a ∼150-residue catalytic domain (C-domain), a 25    ∼60-residue membrane-binding domain (M-domain), and a ~50-residue phosphorylated tail (P- region). The NLS sequence is absent in all of the CCTisoforms. Adapted from (Cornell and Northwood, 2000; Karim et al., 2003; Lykidis et al., 1999). 1.5.1.1 The localisation and tissue-specific expression Both CCT and CCT are amphitropic, that is, the enzymes are reversibly capable of taking an inactive soluble form and an active membrane-bound form. Localisation studies have shown that CCTα appears to be a nuclear protein and translocates to the nuclear envelope when PC levels decrease (Aitchison et al., 2015; Cornell and Northwood, 2000; Cornell and Ridgway, 2015); CCT is cytosolic and targets to ER membrane (Lykidis et al., 1999) probably when activated.  CCTα is expressed at high levels in all murine tissues and is most predominant in the liver, kidney and heart (Karim et al., 2003). The expression level of the CCTβ isoforms is much lower than CCT though CCT is also expressed in most tissues, especially with relatively high levels in brain and reproductive tissues (Jackowski et al., 2004; Karim et al., 2003). CCTβ1 is expressed at a higher level than CCTβ2 in the liver and only CCTβ1 is expressed in the placenta. In contrast, CCTβ2 is expressed at a higher level than CCTβ1 in the brain. Neither the CCTβ1 nor the CCTβ2 variant is expressed in adult lung (Lykidis et al., 1999). CCTβ3 is expressed in the same tissues as CCTβ2 but CCTβ3 is predominant in the testes (Jackowski et al., 2004). CCTα is important for survival and total deficiency of Pcyt1a in mice is lethal (Wang et al., 2005). The targeted deletion of hepatic Pcyt1a in mice reduced hepatic CCT activity by 85%,  resulting in a reduction in the production of plasma PC, cholesterol, TAG,  very low-density lipoprotein (VLDL) and high-density lipoprotein (HDL) (Jacobs et al., 2008). The re-introduction of hepatic CCTthrough adenoviral delivery restored hepatic PC concentration as well as plasma VLDL and HDL levels (Jacobs et al., 2008). This indicates that CCTα also plays a crucial role in hepatic lipid metabolism and lipoprotein production. In contrast to Pcyt1a knockouts, mice with global deletion on Pcyt1b survived (Jackowski et al., 2004). A mild neurological phenotype was reported in these knockout mice, although there was no apparent abnormality in brain development. An impairment in neurite branching was observed when neurons from the knockout mice were cultured ex vivo (Carter et al., 2008; Jackowski et al., 2004). Further analysis of these knockout mice showed defective ovarian follicle development in females and impaired testicular spermatogenesis in males (Jackowski et al., 2004).    26    1.5.1.2 Characterisation of domains The N-terminal region is non-conserved and disordered. Within the N-region, the NLS helps to translocate the enzyme to the nucleus. The N-region is more accessible to protease digestion in the membrane-bound form compared with the soluble form, suggesting that the N-region may interact with the membrane-binding domain at least in its soluble form (Bogan et al., 2005). The N-region is involved in the interface of the two CCT monomers. Among the seven cysteines in rat CCT, only the substitution of Cys37 to Ser led to a monomeric CCT, suggesting that Cys37 in the N-region is the predominant residue in forming the dimer interface (Xie et al., 2004). A yeast two-hybrid analyses showed an interaction between the catalytic domain of CCT when expressed alone, and the interaction was strengthened when CCT expressed the N-region as well as the catalytic domain (Xie et al., 2004). The catalytic domain is highly conserved in CCT among species; additionally, this domain shares conserved features across the cytidylyltransferase superfamily, which includes CCT, ethanolamine phosphate cytidylyltransferase (ECT) and glycerol-3-phosphate cytidylyltransferase (GCT) (Kalmar et al., 1994; Park et al., 1997). There are three motifs found conserved within the catalytic domains of the cytidylyltransferase superfamily: HXGH, RTXGISTT and RYVDEVI (Weber et al., 1999).  The crystal structure and mutagenesis studies indicate that both the HXGH and RTXGISTT motifs are involved in substrate/CTP-binding, and the RYVDEVI motif is at the homodimer interface (Bork et al., 1995; Dunne et al., 1996; Helmink et al., 2003; Veitch and Cornell, 1996). The crystal structure demonstrated mammalian CCT as a homodimer, in which the key contact interface is the conserved 140RYVD143 motif located in the catalytic domain (Lee et al., 2009). Within the dimer interface, the terminal nitrogen atoms of the Arg140 amino acid residue in one chain directly interact with the carbonyl oxygen at Val142 in the other chain. The contact point between two CCT units is expected to be essential for stabilising the protein structure (Payne et al., 2014). The catalytic activity of cytidylyltransferases was reduced when mutations occurred within the catalytic conserved motifs, for example, mutagenesis of the glycine and histidines in the HXGH motif of CCT led to a reduction in Vmax and an increase in the Km for CTP (Park et al., 1997; Veitch et al., 1998; Weber et al., 1999). 27    The catalytic domain is followed by the membrane-binding domain (the M-domain). The M- domain contains three parts: a positively-charged N-terminus, a non-conserved middle segment and a conserved amphipathic C-terminal segment of ~22 residues (auto-inhibitory motif, AI motif) (Cornell and Taneva, 2006) (Figure 1.9A). The unstructured region has positively-charged amino acids (Figure 1.9B), which attract anionic phospholipids on the lipid bilayer, and promote surface localisation of CCT. The auto-inhibitory region can shield the catalytic residues and silence CCT (Cornell, 2016). The M-domain is mostly random-coil in the CCT inactive state (Figure 1.9B), and the unstructured region fold into a long amphipathic -helix structure (Figure 1.9C) in the presence of certain lipids, such as cone-shaped or anionic lipids, and become active (Cornell, 2016; Johnson and Cornell, 1994; Taneva et al., 2003). A series of truncation mutants and proteolysis studies have revealed that the putative amphipathic -helix segment within this domain is important for membrane intercalation (Craig et al., 1994; Wang and Kent, 1995b; Yang et al., 1995). Subsequent studies photolabelling CCT with a hydrophobic probe (presented in artificial lipid vesicles) indicated that CCT can be inserted into the hydrophobic core of the membrane (Johnson et al., 1997), suggesting that CCT lies flat with the membrane-induced - helices in the plane of the phospholipid interface. A CCT truncation mutant lacking the putative amphipathic helices was shown to be soluble and constitutively active even in the absence of lipids while wild-type CCT required lipids for activation  (Friesen et al., 1999; Huang et al., 2013; Wang and Kent, 1995b). Furthermore, the rate of PC synthesis of the CCT truncation mutant in CCT-deficient CHO cells is much higher than in the case of the wild-type enzyme.  In other words, CCT is activated if the inhibitory constraint is removed either by truncation or by membrane- binding.   28    Figure 1.9 Membrane-binding domain of CCT. (A) Sequences of the M-domain from R. norvegicus CCT (P19836), D. melanogaster CCT2 (Q9W0D9), C. elegans CCT1 (P49583) and A. thaliana CCT1 (Q9ZV56). The colour code is as follows: apolar, black; polar and neutral, green; basic, blue; acidic, salmon; Gly and Pro, orange. Arrows above the sequence indicate the Ser260, Ser271, Ser282, Glu257, Glu268 and Glu279 that are proposed to contribute to the lipid selectivity of membrane-binding. AI motif, auto- inhibitory motif. Adapted from (Cornell and Ridgway, 2015). (B) Structure of M-domain in its inactive state. The disordered leash segment is positively charged and connects the AI motif with the catalytic domain. The structure of the AI motif is derived from rat CCT-312(Δ252–269). PDB codes 4MVC. Adapted from (Cornell, 2016). (C) Structure of M-domain in its active state. The image is based on two overlapping PDB structures 1PEH and 1PEI, both are M-domain peptides from rat CCTin complex with SDS Residues 242-288 are shown. The three serines 29    (magenta) mentioned in (A) interrupt the polar face. The three glutamates (green sticks with red oxygens) mentioned in (A) populate the interface between the polar and non-polar region. The colour code is as follows: apolar, yellow; basic, blue; acidic, salmon. Adapted from (Cornell, 2016). The C-terminus of CCT is the phosphorylation region, which is poorly conserved and disordered. In mammalian CCT, this region contains 16 serine residues and all can be phosphorylated (MacDonald and Kent, 1994). Many kinases are responsible for CCT phosphorylation. CCT is primarily phosphorylated by proline-directed kinases at 7 of these 16 serines (Ser315, Ser319, Ser329, Ser323, Ser331, Ser343, and Ser347 in rat CCT). Proline-directed kinases include the mitogen-activated protein (MAP) kinases and cyclin-dependent protein kinases. Other putative phosphorylation sites are recognised by protein kinase C, casein kinase II or glycogen synthase kinase III (MacDonald and Kent, 1994), although immunoprecipitation analyses found that the phosphorylation status of CCT was not changed by 12-O-tetradecanoyl phorbol-13-acetate which activates protein kinase C (Watkins and Kent, 1990). CCT phosphorylation has been observed to be correlated with its membrane association. The phosphorylation region is dephosphorylated in the membrane-bound active form of CCT and largely phosphorylated in the inactive soluble form of CCT (Dennis et al., 2011; Wang et al., 1993; Watkins and Kent, 1991). Phosphorylated CCT required more negatively-charged phospholipids for binding and activation compared with dephosphorylated CCT (Arnold et al., 1997), which is probably because the phosphorylated CCT tail competes with the negatively-charged phospholipids for the same basic residues on the M- domain. In human CCT, CCT2 was able to be highly phosphorylated like CCTCCT2 was even more sensitive to membrane affinity in different phosphorylation states (Dennis et al., 2011). Although evidence suggests that phosphorylation affects CCT membrane association, manipulation of the phosphorylation status including truncation of the phosphorylation region, substitution of serine with other amino acids, and treatment with inhibitors of proline-directed kinases, all have little or no effect on CCT activity or PC synthesis (Cornell et al., 1995; Wang and Kent, 1995a; Wieprecht et al., 1996). Thus, the exact function of the phosphorylation region needs further clarification. 30    1.5.2 Regulation of CCT 1.5.2.1 Transcriptional regulation CCT is transcriptionally upregulated during the S phase of the cell cycle (Golfman et al., 2001) and cell growth (Houweling et al., 1997; Tessner et al., 1991). There is evidence for transcriptional regulation of CCT; in general, the transcription factors Sp1 and Sp3 are respectively phosphorylated by cyclin-dependent kinase 2 and Ras/p42/44 MAP kinase, and the phosphorylated Sp1/Sp3 activates CCT transcription by binding to its promoter (Bakovic et al., 2003; Banchio et al., 2004), while the Est1-related protein Net acts as a transcriptional repressor (Sugimoto et al., 2005).  Est-1 is a co-activator of Sp1 for activating CCT transcription. The transcription of CCT is increased by the Ras/p42/44 MAP kinase pathway (Bakovic et al., 2003) and is induced during nerve formation and branching (Strakova et al., 2011). The promoter element of CCThas not yet been fully characterised and the relevant transcription factors have not been identified (Marcucci et al., 2008). 1.5.2.2 Post-translational regulation Post-translational regulation of CCT expression/turnover and activity is particularly important. 1.5.2.2.1 CCT protein stability CCTα has an estimated half-life of 8 hours in murine lung epithelial (MLE-12) cells, and its half- life is reduced by 56% in response to tumor necrosis factor-α (TNFα), which stimulates the ubiquitin-proteasome and calcium-activated neutral proteolysis (calpain) pathways (Mallampalli et al., 2000). Under TNF exposure, the total CCT pool decreased but CCT ubiquitination was estimated to increase as a result of the use of an anti-ubiquitin antibody. There is a mono- ubiquitination site at murine CCT Lys57 juxtaposing its NLS motifs (Chen and Mallampalli, 2009). The mono-ubiquitinated enzyme was excluded from the nucleus and subsequently degraded in the lysosome, whereas the Lys57Arg mutant was nuclear and displayed proteolytic resistance. When the calpain pathway was inhibited, this abrogated the TNF-induced lower molecular weight degradation products. For example, calmodulin antagonises calpain at the calpain cleavage sites and stabilises CCTexpression (Chen and Mallampalli, 2007). Interestingly, TNFα activates Ras/p42/p44 MAP kinase and protein kinase C (Mallampalli et al., 31    1999; Saklatvala, 1995; Wieprecht et al., 1996) and to some extent synchronously with an elevation in CCTα phosphorylation. The phosphorylated CCTα is relatively stable and protected from degradation in response to cholecystokinin in pancreatic acinar cells (Groblewski et al., 1995). 1.5.2.2.2 Phosphorylation status As mentioned in section 1.5.1.2, phosphorylation of the disordered P-region fine-tunes the affinity of CCT for membranes. Dephosphorylation promotes CCT membrane translocation and vice versa. Substitution of phosphoserine with alanine increased membrane association ten- fold but only increased PC synthesis by about 75%, suggesting that phosphorylation regulated membrane association does not necessarily cause enzyme activation (Wang and Kent, 1995a). It has also been shown that the negative charge of the CCT tail region decreases membrane-binding affinity and simultaneously increases dependence on membrane curvature/packing defects (Chong et al., 2014). These in vitro liposome data support the idea that if the P-region was highly charged/phosphorylated, the phosphorylated residues will probably weaken the electrostatic force between the positively-charged M-domain and the anionic membrane. Specifically, the phosphorylated residues might antagonise the positively-charged residues of the M-domain, resulting in CCT being more dependent on membrane-packing voids to strengthen binding (Chong et al., 2014). When each potential phosphoserine site was substituted with phosphomimetic  glutamate, this CCT variant also decreased membrane affinity and maintained enzymatic activity in a similar way to wild-type CCT in CHO cells (Wang and Kent, 1995a). This suggests that a negative charge in the P-region does affect the membrane association but does not directly influence enzyme function. 1.5.2.2.3 Regulation by membrane lipid compositional modifications The hypothesis that CCT recognises membrane lipid compositional changes is supported by various cellular and in vitro studies. This section will introduce the findings of these studies. 32    1.5.2.2.3.1 CCT translocation and activation in response to agents that change membrane lipid composition CCT localisation and catalytic function in a cell is mainly modulated by the conversion between a soluble inactive form and a membrane-bound form. Some agents that increase small PC catabolites or decrease the relative PC content in the membrane are reported to trigger CCT membrane partitioning and promote PC synthesis, while some agents that increase lysolipids have been shown to blunt CCT activation (Table 1.5). Table 1.5 CCT membrane partitioning correlate with the lipid compositional change Conditions Membrane partitioning Lipid compositional change PC synthesis Phospholipase C ↑ ↓ PC; ↑ DAG ↑ Phorbol ester ↑ ↑ DAG ↑ Unsaturated fatty acid ↑ ↑ FA ↑ Oleic acid feeding/lipogenesis ↑ ↑ FA; ↑ TAG ↑ Diacylglycerol ↑ ↑ DAG ↑ Choline deficiency* ↑ ↓ PC; ↑ PE ↓ Farnesol ↑ ↑ Farnesol ↑ Serum/G0 to G1 transition ↑ ↑ DAG; ↑ FA ↑ Angiotensin II ↑ ↑ DAG; ↑ FA ↑ Alkyl phosphocholine and related compounds ↓ ↑ Lyso-lipids ↓ *PC synthesis decreased because of deficient in choline substrate. Data are adapted from (Cornell, 2016). Small PC catabolites such as DAG, fatty acids and PA are hydrolytic products of phospholipases. For example, the treatment of cultured cells with phospholipase C (PLC) triggers PC breakdown into DAG and phosphocholine (Figure 1.10), which induces CCT translocation and activation at the cell membrane (Slack et al., 1991; Sleight and Kent, 1980; Sleight and Kent, 1983). Furthermore, it has been proposed that phorbal ester treatment will activate phospholipase, which will lead to elevated cellular DAG levels and CCT membrane association (Pelech et al., 1984; Utal et al., 1991). Phospholipase D (PLD) converts PC into PA and choline (Figure 1.10). In mammalian cells, induced PLD and PA phosphatase activity causes CCT translocation and activation at membranes (Morash et al., 1988; Tronchere et al., 1995). Phospholipase A1 (PLA1) and A2 (PLA2) respectively hydrolyse PC into 1-lysoPC and 2-lysoPC (Figure 1.10). When PLA2 and PA phosphatase were inhibited by bromoenol lactone in fibroblasts, a reduction in PC 33    synthesis and membrane affinity of CCT was observed. This could be due to the lack of cellular fatty acids and DAG (Ng et al., 2004). Other work has shown that direct treatment of cultured cells with lipid metabolites, like unsaturated fatty acids or DAG, augments CCT membrane partitioning and CCT activity (Cornell and Vance, 1987; Kolesnick and Hemer, 1990; Pelech et al., 1983; Wang et al., 1993), implying an important role for unsaturated fatty acids and DAG in the regulation of CCT. When oleic acid, an unsaturated fatty acid, was added to the membrane of Drosophila S2 or 3T3-L1 cells, activation of CCT activity and CCT membrane translocation occurred (Aitchison et al., 2015; Krahmer et al., 2011). Although lipid droplets were enlarged during oleic acid treatment, CCTα translocated from the nucleoplasm to the nuclear envelope but did not associate with LDs in 3T3-L1 cells. In contrast, CCT1 translocated to the monolayer of lipid droplets after oleic acid treatment in Drosophila S2 cells (Krahmer et al., 2011). This might be related to the relatively high PE content of Drosophila cells (Jones et al., 1992). This is consistent with CCT membrane activation in choline-deprived cells, in which the PC content was lower and PE levels higher (Jamil et al., 1990; Weinhold et al., 1994), suggesting that CCT responds to the paucity of PC. Figure 1.10 Specificity of phospholipases in the hydrolysis of phosphatidylcholine. PLA1 and PLA2 release free FAs by hydrolysing the sn-1 and sn-2 fatty acyl ester bonds resulting in 1-lysoPC and 2-lysoPC, respectively. PLB hydrolyzes both fatty acyl chains to form 34    glycerophoshpocholine and free FAs. PLC cleaves the glycerophosphoester bond to generate DAG and the phosphocholine. PLD hydrolyses off the head-group choline to release PA. PLA1, phospholipase A1; PLA2, phospholipase A2; FA, fatty acid; 1-lysoPC, 1-lysophosphatidycholine; 2-lysoPC, 2-lysophosphatidylcholine; PLB, phospholipase B; PC, phosphatidylcholine; PLC, phospholipase C; DAG, diacylglycerol; PLD, phospholipase D; PA, phosphatidic acid. The apoptosis inducer, farnesol, also led to transient activation of CCT nuclear membrane translocation and PC synthesis, whereas this agent promoted caspase cleavage and subsequent inhibition of PC synthesis at later time points (Lagace and Ridgway, 2005a). Since other apoptotic triggers such as UV exposure or the protein kinase inhibitor chelerytherine did not trigger CCT translocation (Lagace et al., 2002), farnesol appears to alter the membrane properties of the nuclear envelope in order to favour the association of CCTα.  Farnesol is a 15- carbon isoprenoid alcohol that is naturally produced from 5-carbon isoprene compounds in both plants and animals. Experimental attempts were made to identify whether cell cycle re-entry or growth-factor triggered relocalisation and activation of CCT in cells was mediated by increases in enzyme activators such as DAG and fatty acids. The addition of serum to quiescent (G0) fibroblasts resulted in an elevated DAG content and a subsequent increase in PC synthesis during the G0 to G1 transition (Ng et al., 2004; Northwood et al., 1999). While the serum-dependent CCT activity and membrane affinity was blunted by inhibitors of fatty acids and DAG, the effect was partially reversed by supplementation of free fatty acids and DAG. Restricting the production of PA had no effect on PC synthesis although PA is also one of the known lipid activators in liposomes (Ng et al., 2004). Likewise, treatment of CHO cells with angiotensin stimulated cellular DAG and fatty acid generation, in parallel with augmented CCT activity and PC synthesis (Kitos et al., 2006a; Kitos et al., 2006b). The manipulation of the intracellular supply of DAG affected the rate of PC synthesis but this influence was not observed following the manipulation of fatty acid levels (Kitos et al., 2006b), suggesting that DAG is likely the most important lipid regulator of CCT. In mammalian cells, the addition of alkyl phosphocholine and related compounds inhibited PC synthesis (Boggs et al., 1995; Geilen et al., 1992). There was an accumulation of choline and phosphocholine with the absence of CDP-choline (Boggs et al., 1995), implying that CCT activity was impaired. In addition, the activity impairment was not due to the lack of cellular 35    DAG (Geilen et al., 1992). Therefore, lyso-lipids are probably negative regulators of CCT via the creation of a membrane property unsuitable for CCT-binding. 1.5.2.2.3.2 Liposome-binding analysis: CCT preferentially binds to membranes enriched in anionic and cone-shaped lipids In vitro, liposome-binding assays have been employed to assess the specificity and affinity of phospholipid-CCT interactions. Lipid vesicles (liposomes) of different compositions can be produced by drying and then rehydration of the lipid mixture. The multilamellar vesicles can further be downsized by sonication and extrusion into unilamellar vesicles, which are composed of a lipid bilayer (Zhao and Lappalainen, 2012). Domain M of CCT preferentially binds to vesicles with a relatively high content of either anionic or cone-shaped phospholipids in a ‘common’ or ‘standard’ PC background. The anionic phospholipids include PG, PI, PS and PA (Arnold and Cornell, 1996; Krahmer et al., 2011; Taneva et al., 2005). Cone-shaped lipids are exemplified by DAG and PE (Attard et al., 2000; Krahmer et al., 2011; Taneva et al., 2005). CCT membrane affinity increased when negative curvature strain was proportionally increased by increasing the proportion of cone-shaped lipids; in contrast, alkyl phosphocholine and lysoPC (inverse-cone-shaped lipids, with a relative larger head-group than tail-group) generated positive membrane curvature and prevented CCT-binding (Boggs et al., 1995; Davies et al., 2001). CCT activation by negative curvature strain triggered by dioleoylglycerol was reversed by lysoPC with opposing curvature strain (Davies et al., 2001). CCT-binding promoted PG vesicle aggregation but this did not occur on PC vesicles with cone-shaped lipids, probably due to a stronger binding- affinity between CCT and anionic lipids (Taneva et al., 2005). A combination of these two classes of lipids (PA and DAG) has a synergistic effect on CCT membrane association (Arnold and Cornell, 1996). Purified rat CCT has also been reported to be activated by vesicles containing unsaturated PE, oxidised PC and unsaturated fatty acids (Cornell, 1991; Davies et al., 2001; Drobnies et al., 1999; Sohal and Cornell, 1990). The activation of CCT by these lipids could be not only related to their non-bilayer properties but also correlated with the reduction in the packing of acyl chains due to unsaturation. CCT activation in the presence of oleic acid is an example: PC/oleic acid vesicles were antagonised by sphingosine because it increased the packing of acyl chains (Sohal and 36    Cornell, 1990). Similarly, multilamellar vesicles containing more than 5 mol % oxidized PC activated CCT but sphingomyelin suppressed this activation (Drobnies et al., 1999). These studies demonstrated that CCT responds to changes in the physiochemical parameters of the membrane, such as negative surface potential, negative curvature and unsaturated acyl chains. PC is zwiterionic and its cylindrical shape imparts well-ordered bilayer packing. When the membrane is lacking in PC, for example where it is characterised by negative charge, loose head- group packing and high fluidity, CCT may sense the need to make PC and translocate it to the membrane. In agreement with this concept, the membrane-bound form has a Kcat that is elevated up to 80-fold (Friesen et al., 1999). The ‘sensing’ process can be driven by physical properties. For instance, decreased PC levels destabilise the membrane bilayer because this environment increases the interfacial packing defects and negative curvature strain. This non-ideal lipid- packing can be alleviated by insertion of the amphipathic helix of CCT. When CCT inserts its membrane-binding helix into the lipid bilayer, the energetic tension is reduced.  This is due to insertion of the CCT amphipathic helix which increases the lateral pressure and restores the negative curvature to normal. In other words, CCT-binding releases the membrane curvature stress (Attard et al., 2000). 37    1.5.2.2.4 Relocalisation   X-ray structure, molecular dynamic simulation and biochemical approaches have collectively informed a sophisticated regulatory mechanism for CCT activation in response to the translocation of CCT from a soluble to a membrane-bound conformation. It is proposed that the membrane-inducible amphipathic helix of CCT has the conformational plasticity that enables it to bind to membranes and initiate its enzyme activity or to become detached from membranes and inhibit catalysis (Cornell, 2016; Ding et al., 2012; Ramezanpour et al., 2018). In detail, the binding mechanism involves two steps (Figure 1.11B):   1) Electrostatic attraction between the positively-charged residues in the M-domain (the leash part, figure 1.11A) and a negatively-charged membrane. In the soluble/inactive form, the auto-inhibitory (AI) helices silence CCT by immobilising helix E and ‘capturing’ loop 2 (L2). The catalytic Lys122 in L2 is moved away from CTP because Lys122 interacts with the AI helix. Because CCT is a homodimer, two AI helices and two E helices form a four- helix bundle. 2) Partitioning the M-domain into the interfacial region of a lipid bilayer through hydrophobic interactions. Once the M-domain leash is bound to a membrane, the unstructured M-domain converts into an amphipathic -helix. This conformational change to the enzyme affects the catalytic site, at which the AI constraint is removed. The αE helices become flexible and bend onto a platform near to the membrane surface. This results in activation of the catalytic domain and synthesis of CDP-choline.  Ultimately, a membrane rich in PC prompts CCT dissociation and returns CCT to the inactive soluble form. 38      39    Figure 1.11 Structural domains of mammalian CCT isoforms and the proposed mechanism for membrane dependent activation of the enzyme. (A) Domain organisation of CCT. There are two key elements, loop2 (L2) and the E helix, in the catalytic domain. The membrane-binding domain has a disordered leash (when CCT is inactive) followed by the auto-inhibitory (AI) helix. (B) Model of the CCT homodimer in its soluble and membrane-bound forms (residues 40-367). Partitioning of the membrane-binding domain (M-domain) into the lipid bilayer disrupts the AI interaction with L2 and E, resulting in activation of the catalytic domain. In the membrane-bound form, the conformation of the M- domain turns into a long -helix. It is proposed that the bending of the αE helices in this stage would bring the active site closer to the membrane surface.  The structure of the linker between the E and M-domain is unknown so is represented as a trapezoid. Adapted from (Ramezanpour et al., 2018). 1.5.3 Membrane-binding amphipathic helices 1.5.3.1 Classical amphipathic helices and ALPS motifs The key feature of classical amphipathic helices (AH) is the segregation of nonpolar and polar residues onto opposite faces along the long axis of the helix which allows the AH to be inserted in parallel with the membrane surface. The nonpolar residues fit between fatty acyl-chains of the membrane lipids while the polar residues align with the lipid polar head groups (Hristova et al., 1999). Some AHs reflect adaptation to general or specific membrane features, such as the ALPS motif. The ALPS motif was first characterised in ArfGAP1 and its yeast homolog Gcs1p because of its peculiar membrane curvature-sensing properties (Drin and Antonny, 2010). In recent years, researchers’ awareness of ALPS-related AH structures has risen and thus, more proteins have been found to sense membrane curvature by relying on ALPS motifs. ALPS motifs are defined as amphipathic lipid-packing sensor motifs (Vanni et al., 2013), which are sequences of 20-40 amino acids with low sequence identity but showing similarities in their physiochemical properties (Drin et al., 2007). They differ from traditional AHs in their distribution of bulky hydrophobic amino acids (e.g. Phe, Leu and Trp) every three or four residues. It is believed that the bulky hydrophobic residues can detect large lipid-packing defects generated, for example, by curvature in small vesicles (Antonny, 2011). Additionally, the lack of basic charged residues in the polar face can distinguish ALPS motifs from classic AHs. Taking magainin 2 as an example, it displays four positively-charged lysines in the polar face of its classical AH and, as a result, it is sensitive to the presence of negatively-charged lipids (Wieprecht et al., 1997). In contrast to 40    magainin 2, ALPS motifs are weakly sensitive to negatively-charged liposomes but strongly favour small liposomes which manifest high membrane curvature (Drin and Antonny, 2010; Wieprecht et al., 2000). Although the lack of electrostatic interactions between an ALPS motif and lipid membranes implies that the ALPS motif is extremely sensitive to membrane curvature because of its strong hydrophobic face, the presence of charged residues on the polar face adds additional biophysical properties. For example, -synuclein is a presynaptic protein that binds to small synaptic vesicles. Its amphipathic region contains numerous lysines at the polar/nonpolar interface and large numbers of glutamates at the top of the polar face (Jao et al., 2008). These features enable the AH to sense lipid-packing defects and to respond to electrostatic interactions with membrane lipid head groups. It has been proposed that lysines interact with negative lipid head groups, whereas glutamates may promote interactions with positively-charged head groups like choline (Mishra and Palgunachari, 1996). This suggests that charged residues on the polar face, in addition to the hydrophobic face, also help to anchor curvature-sensing amphipathic helices in the lipid bilayer. The yeast lipin, Pah1, a phosphatase that dephosphorylates PA to yield DAG, is likewise considered to have an ALPS motif in its amphipathic helices. Pah1 may sense curvature changes under PA-enriched conditions on nuclear or ER membranes that are able to deform lipid bilayers. The deforming of PA-enriched lipid bilayers might be detected by the ALPS motif that would then mediate the binding of Pah1 and produce DAG for lipid biosynthesis (Karanasios et al., 2010). There is also a functional ALPS-like motif of 38 amino acids in the golgin GMAP-210 which tethers transport vesicles at the Golgi apparatus (Cardenas et al., 2009; Drin et al., 2008). GMAP-210 acts in the early secretory pathway: it is required for anterograde ER-to-Golgi trafficking, and is also involved in retrograde Golgi-to-ER trafficking (Roboti et al., 2015). An important feature of GMAP-210 tethering transport vesicles is their preference for small liposomes with a radius <40 nm, a unique characteristic both in vitro and in vivo via the N- terminal ALPS-like motif-sensing highly-curved vesicles (Magdeleine et al., 2016). 1.5.3.2 CCT has ALPS like motifs involved in membrane remodelling Proteins with a wide range of functions have been reported to have similar unstructured motifs which can fold into amphipathic helices upon encountering membranes deficient in PC (Antonny, 2011; Drin et al., 2007; Karanasios et al., 2010; Magdeleine et al., 2016)- often referred to as 41    nascent AHs. Mammalian CCT has a conserved ALPS motif in the M-domain as a membrane- inducible amphipathic helix (m-AH)  that responds to changes in the physical properties of PC- deficient membranes (Cornell, 2016; Payne et al., 2014). The amphipathic helix of CCTconsists of a developed hydrophobic face and a developed polar face (Cornell and Ridgway, 2015; Dunne et al., 1996) (Figure 1.12A). The hydrophobic face predominantly consists of bulky hydrophobic residues separated by three Serine residues (Dunne et al., 1996) (Figure 1.9C). The interspersed Serine residues limit the hydrophobicity/membrane insertion and increase the specificity of the -helix for membrane phospholipids (Johnson et al., 1998). The polar face is populated with multiple charged residues where the positively-charged residues modulate CCT interacting with negatively-charged phospholipids on the membrane; additionally, the negatively-charged residues facilitate lipid-selective membrane-binding (Johnson and Cornell, 1994; Johnson et al., 1998; Johnson et al., 2003) (Figure 1.12B). Intriguingly, when a negatively- charged residue Glu280 was deleted from the polar face of the human CCT, the Glu280del mutant had a faster membrane ‘off-rate’ than the wild-type enzyme, suggesting altered membrane-protein association (Payne et al., 2014).  At the polar/non-polar interface, the positively-charged residues (Lys) probably assist in CCT positioning into anionic lipids (Cornell and Ridgway, 2015); the three glutamates Glu257, 268 and 279 (Figure 1.9C) in rat CCTα would direct CCT binding to membranes with appropriate anionic content where the probability of glutamate protonation is higher. This is due to the surface anionic attraction of protons. Protonation of interfacial glutamates would eliminate electronic repulsion with the PL charge, thus increasing membrane affinity (Johnson et al., 2003).   Within domain M vertebrate CCTα and CCTβ have a set of 11-mer tandem repeats with a consensus resembling m-AH segments of -synuclein (Ding et al., 2012; Johnson et al., 2003), that is, the VEEKS-repeat peptide structure. Whereas the hydrophobic face of the CCT m-AH is rich in bulky amino acids like Phe, Leu and Ile, that of -synuclein has smaller residues such as Val and Ala (Cornell, 2016). Invertebrate CCT does not have obvious VEEKS repeats but does have the amphipathic sequence. Heliquest prediction of yeast CCT (Figure 1.13), named Pct1, residues 261 to 282 indicates amphipathic properties, with an asymmetrical distribution of hydrophobic vs. polar residues. The conserved -helical domain is predicted to have a strong propensity for hydrophobicity on the non-polar face, which contains key hydrophobic residues such as  Phe, Leu, Trp, Ile and Tyr. These hydrophobic residues might drive hydrophobic 42    interactions with the lipid bilayer and promote membrane intercalation (Kalmar et al., 1990). Another feature of this helical diagram is the distribution of positively-charged amino acids like Lys and Arg on the polar face, which probably interact with anionic phospholipid head groups of the target membrane. The partitioning of the helix is energetically more favourable for membranes with more negative curvature (Attard et al., 2000). Figure 1.12 Helical wheel plot of human CCT amphipathic -helix and the topological model of how the -helix interacts with membranes (related to Figure 1.11). 43    (A) Helical wheel plot was generated using HeliQuest (Gautier et al., 2008) and describe the amphipathic -helix of human CCT. Only the helix-forming residues 241–294 are displayed. The arrow indicates the direction of the hydrophobic moment. Residues VVLV constitute the hydrophobic phase. Colours indicate the properties of each amino acid: yellow, hydrophobic; red, negatively-charged; blue, positively-charged; purple, polar uncharged with hydroxyl side chain; pink, polar uncharged with an amide side chain; grey, small non-polar. (B) Topological model of CCT interaction with the membrane. Model showing how the amphipathic -helix of CCT in its membrane-binding form could insert wihin one of the leaflets of the membrane. The helix would be stabilized by hydrophobic interactions with the fatty acid core of the membrane, and ionic interactions between positively charged residues (blue) and negative charges of phospholipids. The negatively charged residues on the helix (red) form a crest facing the membrane surface. Adapted from (Lincoln Taiz, 2014). Figure 1.13 Helical wheel plot of Pct1 membrane-inducible amphipathic helix. (A) Schematic illustration of the Pct1 domain organisation, highlighting the amphipathic M- domain. m-AH, membrane-inducible amphipathic helix; AI, autoinhibitory segment; N-region, amino end region, C-domain, catalytic domain; M-domain, membrane binding domain; P-region, phosphorylated region. (B) Helical wheel plot was generated using HeliQuest (Gautier et al., 2008). Only the helix-forming residues 261–282 are displayed. The arrow indicates the direction of the hydrophobic moment. Residues FFLI constitute the hydrophobic phase. Colours indicate the properties of each amino acid: yellow, hydrophobic; red, negatively-charged; blue, positively- charged; purple, polar uncharged with hydroxyl side chain; pink, polar uncharged with an amide side chain. 44    1.6 Evidence for the physiological importance of PCYT1A PC is crucial for membrane biogenesis and, as discussed above, elaborately regulated by CCT. The physiological importance of CCT/PCYT1A/CCT1 is apparent from the reported lethality of Pcyt1a-null mice (Wang et al., 2005) and flies (Gupta and Schupbach, 2003). As far as humans are concerned, mutations in the PCYT1A genes have recently been associated with three different phenotypes: one characterised by congenital lipodystrophy with fatty liver disease (Payne et al., 2014), and another by spondylometaphyseal dysplasia with cone-rod dystrophy (Hoover-Fong et al., 2014; Yamamoto et al., 2014) and another by Leber congenital amaurosis without any metabolic or skeletal involvement (Nash et al., 2015). The sequence variants were identified by next generation sequencing. The detailed mutation sites will be discussed in Section 5.2.2. This section briefly describes the basic features of these genetic disorders. 1.6.1 Congenital lipodystrophy and fatty liver disease Clinical research studies in humans have shown that PCYT1A mutations caused lipid-related metabolic diseases, such as congenital lipodystrophy with severe insulin resistance and non- alcoholic fatty liver disease (Payne et al., 2014). The two affected patients were compound heterozygotes for different PCYT1A mutations. PC synthesis was decreased in patient cells with no alteration in PCYT1B or PEMT mRNA levels. Lipodystrophy is a rare genetic disorder (less than one in a million) characterised by a lack of functional adipose tissue and ectopic fat deposition in other tissues (Robbins and Savage, 2015). Since adipose tissue is responsible for fat storage, the failure of appropriate adipose tissue energy storage can result in fat accumulating in other tissues such as the liver, pancreas and skeletal muscle. Ectopic fat deposition in liver cells leads to fatty liver disease, termed nonalcoholic fatty liver disease. According to the pattern of fat loss, inherited lipodystrophy can be categorised into congenital generalized lipodystrophy (CGL) and familial partial lipodystrophy (FPL). CGL patients lack almost all body fat and have this phenotype from birth, while FPL patients lack regional fat and present later in life. Both CGL and FPL have serious complications including diabetes/insulin resistance and dyslipidemia. The severity of the symptoms generally depends on the degree of fat loss, that is, CGL patients usually suffer more serious metabolic consequences. To date there are at least eleven genes associated with inherited lipodystrophy (Robbins and Savage, 2015). The CGL-related genes, such as AGPAT2, CAV1, PTRF, PCYT1A and BSCL2, are broadly involved in lipid uptake and 45    adipogenesis. The FPL-related genes are involved in lipid droplet metabolism (PLN1 and CIDEC), adipogenesis (PPARG) and nuclear functions (LMNA, ZMPSTE24 and POLD1). Despite this progress, the underlying genetic causes of lipodystrophy are not fully known. PCYT1A may play a role in maintaining functional fat tissue. 1.6.2 Spondylometaphyseal dysplasia with cone-rod dystrophy Different PCYT1A mutations were associated with spondylometaphyseal dysplasia with cone-rod dystrophy (SMD-CRD; MIM 608940) in nine affected patients (Hoover-Fong et al., 2014; Yamamoto et al., 2014). Patients with SMD-CRD do not have obvious lipid-related metabolic disorders. SMD-CRD is an autosomal recessive disorder characterised by an abnormal skeletal phenotype and visual impairment due to cone-rod dystrophy. SMD-CRD is a rare disease. The bone dysplasia of SMD-CRD manifests short stature, rhizomelia with bowing of the lower limbs, platyspondylyl and metaphyseal (growth plate) irregularities (Wong, 2014). Patients had varying ages of disease onset between 2 and 51 years. CRD typically reduces the function of cone cells and then later destroys cone and rod cells (Nash et al., 2015), and this pathogenicity pattern was observed in most SMD-CRD patients. There have not yet been suggestions of genotype- phenotype patterns of SMD-CRD appearance. 1.6.3 Leber congenital amaurosis Leber congenital amaurosis (LCA) is classified as non-syndromic generalised retinal dystrophy (Nash et al., 2015). Recently, mutations in the PCYT1A gene have been reported in patients with isolated LCA but without any skeletal dysplasia, or metabolic or hepatic phenotypes (Testa et al., 2017). Three patients from two independent families were analysed and all the patients had compound heterozygous mutations.  LCA manifests early onset severe visual impairment including poor vision, nystagmus and no measurable light response on electroretinography testing (Nash et al., 2015; Testa et al., 2017). 46    1.7 Overview: experimental aims The overall aim of this study is to understand how CCT senses the need for PC synthesis in cells in vivo. This aim is underpinned by the fact that: 1) the importance of CCT has already been shown in patients with loss-of-function PCYT1A gene mutations; 2) in vitro studies have demonstrated that membrane-binding and catalytic activation of purified CCT is induced by conically-shaped or negatively-charged lipids, yet exactly how this occurs in vivo remains uncertain; and 3) cell and organelle membranes are composed of a complex mixture of phospholipids that determine their size, shape, and function. PC is the most abundant phospholipid in eukaryotic membranes. Although the conformational change of CCT and the biochemical pathways that generate PC have been well described, the relation between cellular CCT distribution and PC demand/regulation remains unclear. The specific objectives of the studies presented in this thesis were as follows: I. To establish a genetically tractable system to monitor the CCT homolog, Pct1, in yeast. To verify how Pct1 regulates PC synthesis in accordance with phospholipid compositional changes in vivo by aligning imaging with lipidomic analysis and data- driven modelling. These results are presented in Chapter Three. II. To determine if Pct1 senses and regulates the demand of PC exclusively within the nucleus. These results are presented in Chapter Four. III. To characterise the role of PCYT1A in adipogenesis in mammalian cells and the effects of the lipodystrophy-associated PCYT1A mutants on PC synthesis. These results are presented in Chapter Five.     47    Chapter 2 Materials and Methods 2.1 Saccharomyces cerevisiae cell culture 2.1.1 Saccharomyces cerevisiae strains Yeast strains used in this study are listed in Table 2.1. Gene deletions and epitope tagging at the endogenous locus were achieved using a one-step polymerase chain reaction (PCR)-based method (Janke et al., 2004; Longtine et al., 1998) and confirmed by PCR. The one-step PCR method requires a set of primers to amplify a module which includes a KanMX6, NatMX6, HISMX6, or hphNT1 selective marker and 60 bp of genomic sequence that flanks either the 5’ or 3’ end of the targeting sequences used for recombination. The correct replacement of the gene with the selective marker was confirmed by PCR using a primer for the cassette and an external primer that spans the left or right junction of the deletion module within the genome. All strains were stored in 20% glycerol at −80°C. Table 2.1 S. cerevisiae strains used in this study Strain Genotype Source/reference RS453 MATα ade2-1 his3-11,15 ura3-52 leu2-3112 trp1-1 (Wimmer et al., 1992) BY4742 MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0 Open Biosystems BY4741 MATa his3∆0 leu2∆0 met15∆0 ura3∆0 Open Biosystems pct1∆ BY4742 pct1::NatMX6 This study cho2∆opi3∆ BY4742 cho2::KanMX opi3::LEU2 Laboratory of George M. Carmen, Rutgers University, New Brunswick, New Jersey United States. cho2∆opi3∆pct1∆ pURA-OPI3 BY4742 cho2::KanMX opi3::LEU2 pct1::NatMX6 YCplac33-OPI3 This study cpt1∆ BY4742 cpt1:: KanMX Open Biosystems cpt1∆ept1∆ BY4742 cpt1:: KanMX ept1::hphNT1 This study cho2∆opi3∆pct1Δect1∆ pURA-OPI3 BY4742 cho2::KanMX opi3::LEU2 pct1::NatMX6 ect1:: hphNT1 YCplac33- OPI3 This study snf1Δ BY4741 snf1::KanMX4 Open Biosystems reg1∆ BY4742 reg1::KanMX4 Open Biosystems hxk2∆ BY4742 hxk2::KanMX4 Open Biosystems 48    mig1∆ BY4742 mig1::KanMX4 Open Biosystems sch9∆ BY4742 sch9::KanMX4 Open Biosystems SEY6210.1 MATα leu2-3,112 ura3-52 his3-Δ200 trp1- Δ901 suc2-Δ9 lys2-801 (Robinson et al., 1988) ANDY198 (∆Tether) SEY6210.1 ist2::HISMX6 scs2::TRP1 scs22::HISMX6 tcb1::KanMX6 tcb2::KanMX6 tcb3::HISMX6 (Manford et al., 2012) W303-1B MATα leu2-3112 ura3-1 his3-11,15 trp1-1 ade2-1 can1-100 (Thomas and Rothstein, 1989) Y3527 W303-1B tpk1M164Gtpk2M147Gtpk3M165G (Yorimitsu et al., 2007) BY4741 ole1∆ BY4741 ole1:: KanMX4 This study RS453 ole1∆ RS453 ole1:: KanMX4 This study HHY168 W303 MATα tor1-1 fpr1::NAT RPL13A- 2×FKBP12::TRP1 EUROSCARF SS1725 HHY168 URA3::YIplac211-NUP84- mCherry KAP60-FRB::KanMX Laboratory of Symeon Siniossoglou, Cambridge Institute for Medical Research, United Kingdom. SS2473 SS1725 PCT1-GFP::HISMX6 Laboratory of Symeon Siniossoglou, Cambridge Institute for Medical Research, United Kingdom   2.1.2 Saccharomyces cerevisiae growth All reagents were purchased from Sigma-Aldrich, unless otherwise specified. Yeast cells were grown at 30°C in a gyratory shaker (at 210 rpm) in synthetic complete (SC) medium containing 2% glucose, 0.2% yeast nitrogen base (Difco, BD, Franklin Lakes, NJ), 0.6% ammonium sulfate and amino acid drop-out solutions (60 mg/L leucine, 55 mg/L adenine, 55 mg/L uracil, 55 mg/L tyrosine, 20 mg/L of arginine, 10 mg/L histidine, 60 mg/L isoleucine, 40 mg/L lysine, 60 mg/L phenylalanine, 50 mg/L threronine, 10 mg/L methionine, 40 mg/L tryptophan) lacking the appropriate amino acids for plasmid selection. Glucose starvation experiments were performed in a synthetic medium lacking 2% glucose (glucose-free). Nitrogen starvation experiments were performed in a synthetic medium lacking nitrogen containing 0.17% yeast nitrogen base without amino acids, 0.5% sodium sulphate and 2% glucose. Where indicated in the figures, 1 mM choline or 1 g/L 5-fluoroorotic acid (5-FOA) was supplemented to the culture medium. YPD 49    media consisted of 2% glucose, 2% bactopeptone (BD, Franklin Lakes, NJ) and 1% yeast extract (BD, Franklin Lakes, NJ). Agar plates were made by adding 2% w/v agar (Biogene, Cambridge, UK) to the corresponding liquid medium. Fatty acid requiring ole1∆ strains was grown on medium supplemented with 1 mM oleic acid with 1% tergitol in order to solubilise the fatty acids. For time-course choline supplementation experiments, cells were grown in fresh media without choline for 24 hr. The cells were then supplemented with 1 mM choline and incubated for the times indicated in figures. 2.1.3 Lithium acetate transformation of yeast All reagents were purchased from Sigma-Aldrich, unless otherwise specified. DNA fragments or purified DNA plasmids were transformed into yeast cells by a modified lithium method (Ito et al., 1983). Yeast cells were grown to OD600 0.5–1.0 and a 10 mL culture was harvested by centrifugation at 3,000 rpm for 3 min. The pellets were washed once in 10 mL sterile water and gently mixed in 10 mL LiT buffer (100 mM lithium acetate in 10 mM Tris-HCl pH 7.4) with 10 mM dithiothreitol at room temperature for 45 min. After centrifugation at 3,000 rpm for 3 min, the cells were resuspened in LiT buffer at 10 ODs/mL. Each transformation reaction consisted of 100 L yeast cell suspension, 50 L LiT buffer, 5 L single-stranded deoxyribonucleic acid carriers (D7656), and 250–300 ng individual plasmid DNA. The reaction mixture was incubated on a Bio RS-24 mini-rotator (Biosan) at room temperature for 10 min. 300 L of 40% w/v Polyethylene glycol 4000 was added to the mixture, and vortexed briefly to mix. After incubating at room temperature for 10 min on a tube rotator, 15 L of dimethyl sulfoxide solution was added and the cells were heat-shocked at 42oC in a water bath for 4 min. Pellets were harvested immediately by centrifugation at 3,000 rpm for 3 min and gently resuspended in 1 mL YPD media. The suspension was incubated on a tube rotator at room temperature for 1 hr, and centrifuged at 3,000 rpm for 3 min. The cells resuspended in 200 L YPD media and plated onto the YPD media or selective SC medium, if plasmid selection was required. If KanMX, hygromycin (hphNT1) or NatMX module served as identifiers for the deletion, 200 mg/L Geneticin Selective Reagent (G418; Invitrogen) was added to select for KanMX; 300 mg/L hygromycin B (InvivoGen) was added to select for hphNT1; and 60 mg/L nourseothricin 50    (ClonNAT) was added to select for NatMX. Transformants were selected after 2–3 days incubation at 30oC. 2.2 Cloning strategies 2.2.1 Extraction of yeast genomic DNA All reagents were purchased from Sigma-Aldrich, unless otherwise specified. 10 mL of yeast cells were grown overnight (OD600>1) and centrifuged at 3,000 rpm for 3 min. The pellet was washed once with sterile water and then suspended in 500 L of sterile water and 200 L of genomic DNA breaking buffer (2% Triton X-100, 1% SDS, 100 mM NaCl, 10 mM Tris pH 8.0 and 1 mM EDTA). Purification was conducted with an equal volume of phenol:chloroform: isoamyl alcohol pH 8.0 (25: 24: 1) and one scoop of glass beads (BioSpec products). The mixture was mixed thoroughly for 3 min using a vortex-genie2 (Scientific Industries) and centrifuged at 13,200 rpm for 30 min. Precipitation with 1.2 mL of ice-cold pure ethanol was conducted for 10 min at −20°C in the aqueous phase. The DNA was spun down at 13,200 rpm for 30 min at 4°C and dried completely in a vacuum concentrator (Eppendorf, Hamburg, Germany) for 15 min at 45°C. Finally, the DNA was dissolved in 50 L of sterile water and stored at −20°C until use.   2.2.2 Yeast DNA plasmids Yeast plasmids expressing wild-type yeast genes used in this study are listed in Table 2.2. All plasmids were generated by a PCR-based cloning method and were verified by diagnostic restriction digest and sequencing. For expression of the indicated carboxyl-terminal GFP tag fusions, the gene of interest was cloned from yeast genomic DNA in frame with GFP in YCplac111 or pRS313 plasmids. PCT1-GFP or PCT1-mCherry was cloned into the YCplac111 vector in four steps. First, the PCT1 promoter and the PCT1 open reading frame (ORF) were amplified from BY4741 genomic DNA with primers YC1 and YC2 (Table 2.3), which added N- terminus SphI and C-terminus BamHI sites. The resulting product was digested with SphI and BamHI. Next, the fragment which starts from the STOP codon to PCT1 terminator was amplified with primers YC3 and YC4 (Table 2.3), which added N-terminus BamHI and C-terminus SacI sites. The resulting product was digested with BamHI and SacI. The two DNA fragments were combined in a three-way ligation with SphI and SacI digested YCplac111. Finally, a GFP or an mCherry tag was inserted in the BamHI site following the same ligation process. PCT1-GFP or - 51    mCherry were subcloned into a centromeric plasmid pRS313 that carries the HIS3 gene as an auxotrophic marker. The PCT1-GFP or -mCherry cassettes, containing the promoter and terminator of the Pct1, were amplified by PCR from plasmid YCplac111-PCT1-GFP or -mCherry using two primers (primer YC5 and primer YC4; Table 2.3) introducing an N-terminus XbaI site and a C-terminus SacI site. The PCR product was digested by XbaI and SacI, and then subcloned into the XbaI and SacI sites of the pRS313 vector. For the expression of the PCT1 gene in the cho2Δopi3Δpct1Δ (3∆) cells, pRS313 plasmids containing PCT1 were first transformed in the cho2Δopi3Δpct1Δ cells carrying the YCplac33-URA3-OPI3 plasmid, and the transformants were subsequently streaked out twice on 5-FOA plates to remove the YCplac33-URA3-OPI3 plasmid. To change the PCT1 internal promoter to an NOP1 promoter, a fragment containing start codon- PCT1-BamHI was amplified from YCplac111-PCT1-GFP with primers YC6 and YC8 (Table 2.3), which added N-terminus NotI and C-terminus BamHI sites. The DNA segments were digested with restriction enzymes and combined in a two-part ligation with NotI and BamHI digested YCplac111-NOP-GFP plasmid carrying the NOP1 promoter. The resulting plasmid was called YCplac111-NOP-PCT1. Similarly, a fragment containing start codon-PCT1-BamHI was amplified from YCplac111-PCT1-GFP with primers YC6 and YC7 (Table 2.3), which adds NotI sites to both the 3’-end and 5’-end. The DNA segments were digested with NotI and ligated into a NotI digested YCplac111-NOP-GFP plasmid carrying the NOP1 promoter. The resulting plasmid was called YCplac111-NOP-PCT1-GFP. The cloned plasmids were confirmed by DNA sequencing as described in section 2.3.10. All plasmids were stored in nuclease-free water in a −20°C freezer. Various cloning strategies for producing PCT1 mutations will be discussed in more detail in later sections, 2.2.3 to 2.2.6. 52    Table 2.2 S. cerevisiae plasmids used in this study Plasmid Description Source/reference YCplac111-NOP-GFP GFP under control of NOP1 promoter in CEN/LEU2 vector Laboratory of Symeon Siniossoglou YCplac111-NOP-PCT1 PCT1 under control of NOP1 promoter in CEN/LEU2 vector This study YCplac111-NOP-PCT1-GFP PCT1-GFP under control of NOP1 promoter in CEN/LEU2 vector This study YCplac111-PCT1 PCT1 under control of PCT1 promoter in CEN/LEU2 vector This study YCplac111-PCT1-GFP PCT1-GFP under control of PCT1 promoter in CEN/LEU2 vector This study YCplac111-PCT1-mCherry PCT1-mCherry under control of PCT1 promoter in CEN/LEU2 vector This study YCplac111-ADH-NLS-GFP NLS-GFP under control of ADH1 promoter in CEN/LEU2 vector Laboratory of Symeon Siniossoglou pA5211-ADH-NLS-GFP NLS-GFP under control of ADH1 promoter in CEN/ADE2 vector Laboratory of Symeon Siniossoglou pRS313-PCT1 PCT1 under control of PCT1 promoter in CEN/HIS3 vector This study pRS313-PCT1-GFP PCT1-GFP under control of PCT1 promoter in CEN/HIS3 vector This study pRS313-PCT1-mCherry PCT1-mCherry under control of PCT1 promoter in CEN/HIS3 vector This study YCplac33-SEC63-mCherry SEC63-mCherry under control of SEC63 promoter in CEN/URA3 vector (Barbosa et al., 2015) YCplac111-ATG8-mCherry ATG8-mCherry under control of ATG8 promoter in CEN/LEU2 vector Laboratory of Symeon Siniossoglou YCplac33-OPI3 OPI3 under control of OPI3 promoter in CEN/URA3 vector Laboratory of Symeon Siniossoglou YCplac33-MSN2-mCherry MSN2-mCherry under control of MSN2 promoter in CEN/URA3 vector Laboratory of Symeon Siniossoglou 53    Table 2.3 Sequences of primers used to generate S. cerevisiae Pct1 expression constructs Primers Product name Orientation Sequences (5'→3') Note YC1 PCT1 Sense TTTGCATGCTGCTTCTATCTGGCTGCAC AG SphI YC2 PCT2 Anti-sense TTTGGATCCGTTCGCTGATTGTTTCTTC BamHI YC3 PCT3 Sense TTTGGATCCTGAGCCCCTCCTCCAATCT CT BamHI YC4 PCT4 Anti-sense TTTGAGCTCTCGTAAAGTTGCTGAGCG TCA SacI YC5 PCT5 Sense TTTTCTAGATGCTTCTATCTGGCTGCAC AG XbaI YC6 NotI_Pct1_F Sense TTTGCGGCCGCGGCAAACCCAACAAC NotI YC7 NotI_Pct1_R Anti-sense TTTGCGGCCGCGTTCGCTGATTGTTTCT TCTTC NotI YC8 BamHI_Pct1_R Anti-sense TTTGGATCCGTTCGCTGATTGTTTCTTC TTCTGTGTT BamHI YC9 PCTNLS2F Sense TAACTCGAGCCTATCAAACCTATTTGCT GCTAATGCTAATGCTAGA XhoI YC10 PCTNLS2R Anti-Sense TTTCTCGAGTTAGAAAGCTTAGCCCTA XhoI YC11 AvrII_for Sense TTTCCTAGGCGTTACACCAACGAGTTG CCC AvrII YC12 AvrII_rev Anti-sense TTTCCTAGGCATTTTTCTTTGTGTCTGC TGTGCTTTG AvrII YC13 AvrIIRev Sense TTTCCTAGGTGTGTCACCCCAGAATTTC TAC AvrII YC14 AvrIIFwd Anti-sense TTTCCTAGGGATTTTGTCGCTAGGCACA CC AvrII YC15 15bpXbaI_fwd Sense GGGGGATCCACTAGTTCTAGATGCTTCT ATCTGGC XbaI YC16 GPd_rev Anti-sense GCCTATTAACTTGGAGGTAGCACCCCTT GCAAAGTTTC YC19 GPd_fwd Sense GCAAGGGGTGCTACCTCCAAGTTAATA GGCAATGAATTAAAGAAACAAAATCAA C YC20 NcoI15bp_rev Anti-sense AGTGACAAGTGTTGGCCATGGAACAG GTAGTTTTC NcoI 54    YC21 M13R Sense CAGGAAACAGCTATGACC YC22 M13F Anti-sense TGTAAAACGACGGCCAG YC23 PCTNLSF Sense TTTGGACACACAGCTCACTCCGCGGGC TGCTGCTGCTTTGACGAAGGAG SacII YC24 PCTNLSR Anti-sense TTTCCGCGGAGTGAGCTGTGTGTCCTTA TTTTC SacII   2.2.3 PCR-based PCT1 gene modification A traditional restriction-ligation-based cloning method was used to construct certain PCT1 mutations, as listed in Table 2.4. To generate the catalytic domain deletion [145-178]Δ (Cdel) construct, two pairs of primers were used to amplify the entire sequence of the pRS313-PCT1- GFP except for the specific region that was to be deleted. Two PCR fragments were obtained with the primer pairs YC5/YC14 and YC13/YC4 (Table 2.3). YC13 adds an AvrII site to the 5’ end of PCR fragment1; YC14 adds an AvrII site to the 3’ end of PCR fragment2. The two DNA segments were digested with XbaI, AvrII and SacI and combined in a three-part ligation with XbaI and SacI digested pRS313-PCT1-GFP. The product is named pRS313-PCT1Cdel-GFP. The nuclear localisation signal (NLS) truncation, NLS [2-75]Δ (NLStruncation), was obtained with the primer pairs YC5/YC12 and YC11/YC4 (Table 2.3). The two DNA segments were digested with restriction enzymes and cloned into the corresponding sites of pRS313-PCT1-GFP. In order to characterise the NLS on the PCT1 gene further, the NLS K62A/R63A/R64A/R65A modification (NLSm1) was cloned in frame into YCplac111-PCT-GFP between SphI and PstI. The NLSm1 mutant was PCR amplified with two pairs of primers YC21/YC24 and YC23/YC22 (Table 2.3), which were designed to add a unique SacII site at P60-R61 by a silent mutation in these two codons (CCC was altered to become CCG in codon60; CGC was altered to become CGG in codon61) and direct the internal mismatch mutations. 55      2.2.4 GeneArt cloning The sequences coded for the Pct1 proteins of interest were modified for expression of potential NLS mutants in yeast cells using the GeneART gene synthesis method (Thermo Fisher Scientific). Table 2.5 presents the mutants made by this method. A Pct1 DNA fragment was designed to contain the NLS K25A/K26A/K28A/K30A modification (NLSm2) and NLS K25A/K26A/K28A/K30A/K62A/R63A/R64A/R65A modification (NLSm1m2) (Table 2.6). The DNA fragment was PCR amplified and cloned in-frame into the PstI-XhoI pre-digested and gel purified YCplac111-PCT1-GFP vector, in which a silent XhoI restriction site was introduced by PCR primers YC9/YC10 (Table 2.3). Transformation was performed on XL1-Blue competent cells and there followed the typical cloning methodology described in section 2.3. To construct the NLS [60-66]Δ (NLS1del), a GeneArt DNA fragment was designed between XbaI and PsyI restriction sites (Table 2.7). The YCplac111 plasmid DNA was further sub-cloned into the sequence-verified RS313 plasmid and confirmed again by DNA sequencing. Table 2.4 S. cerevisiae plasmids used in this study (PCR-based PCT1 gene modification) Plasmid Description Source/reference pRS313-PCT1Cdel-GFP PCT1-[145-178]Δ-GFP under control of PCT1 promoter in CEN/HIS3 vector. This study YCplac111-PCT1NLSm1- GFP PCT1-K62A/R63A/R64A/R65A-GFP under control of PCT1 promoter in CEN/LEU2 vector. This study pRS313-PCT1NLSm1-GFP PCT1-K62A/R63A/R64A/R65A-GFP under control of PCT1 promoter in CEN/HIS3 vector. This study pRS313- PCT1NLStruncation-GFP PCT1-[2-75]Δ-GFP under control of PCT1 promoter in CEN/HIS3 vector. This study 56    Table 2.5 S. cerevisiae plasmids used in this study (GeneArt cloning) Plasmid Description Source/reference YCplac111-PCT1NLSm2- GFP PCT1-K25A/K26A/K28A /K30A-GFP under control of PCT1 promoter in CEN/LEU2 vector. This study YCplac111- PCT1NLSm1m2-GFP PCT1- K25A/K26A/K28A/K30A/K62A/R63A/R64A/ R65A -GFP under control of PCT1 promoter in CEN/LEU2 vector. This study pRS313-PCT1NLSm2-GFP PCT1-K25A/K26A/K28A /K30A-GFP under control of PCT1 promoter in CEN/HIS3 vector. This study. pRS313-PCT1NLSm1m2- GFP PCT1- K25A/K26A/K28A/K30A/K62A/R63A/R64A/ R65A -GFP under control of PCT1 promoter in CEN/HIS3 vector. This study. pRS313-PCT1NLS1del- GFP PCT1-[60-66]Δ-GFP under control of PCT1 promoter in CEN/HIS3 vector. This study.                                     57    Table 2.6 GeneArt fragments used in NLSm1 and NLSm1m2 construct GeneArt fragment Sequences (5'→3') NLSm2 ATTAGGGCTAAGCTTTCTAACTCGAGCCTATCAAACCTATTTGCTG CTAATGCTAATGCTAGACAGCGTGAGGAAACGGAAGAGCAGGAC AATGAGGATAAGGATGAGAGTAAGAACCAGGATGAAAATAAGGA CACACAGCTCACTCCCCGCAAGCGTCGCCGGTTGACGAAGGAGT TTGAAGAGAAGGAGGCTCGTTACACCAACGAGTTGCCCAAGGAA CTGCGCAAGTATCGTCCTAAAGGTTTCAGATTCAATTTGCCTCCAA CGGATAGACCCATCAGGATATATGCAGATGGTGTTTTTGATCTTTTC CATCTTGGCCACATGAAGCAACTGGAACAGTGTAAGAAGGCTTTC CCCAATGTAACACTGATAGTTGGTGTGCCTAGCGACAAAATCACTC ACAAACTAAAAGGTTTGACTGTGCTGACCGATAAGCAGCGTTGTG AAACTTTAACGCACTGCAGATGGGTTGACGAAGTC NLSm1m2 ATTAGGGCTAAGCTTTCTAACTCGAGCCTATCAAACCTATTTGCTG CTAATGCTAATGCTAGACAGCGTGAGGAAACGGAAGAGCAGGAC AATGAGGATAAGGATGAGAGTAAGAACCAGGATGAAAATAAGGA CACACAGCTCACTCCGCGGGCTGCTGCTGCTTTGACGAAGGAGTT TGAAGAGAAGGAGGCTCGTTACACCAACGAGTTGCCCAAGGAAC TGCGCAAGTATCGTCCTAAAGGTTTCAGATTCAATTTGCCTCCAAC GGATAGACCCATCAGGATATATGCAGATGGTGTTTTTGATCTTTTCC ATCTTGGCCACATGAAGCAACTGGAACAGTGTAAGAAGGCTTTCC CCAATGTAACACTGATAGTTGGTGTGCCTAGCGACAAAATCACTC ACAAACTAAAAGGTTTGACTGTGCTGACCGATAAGCAGCGTTGTG AAACTTTAACGCACTGCAGATGGGTTGACGAAGTC XhoI; Silent mutation TCATCG to TCGAGC; NLSm2 AAAAAAAATAAAAATAAAAGA to GCTGCTAATGCTAATGCTAGA (KKNKNKR to AANANAR); NLSm1 CCCCGCAAGCGTCGCCGGTTG to CCGCGGGCTGCTGCTGCTTTG (PRKRRRL to PRAAAAL); PstI            58    Table 2.7 GeneArt fragments used in NLS1del construct GeneArt fragment Sequences (5'→3') NLS1del TCTAGATGCTTCTATCTGGCTGCACAGCTATCTATACTGTGCTGA AAACTGTATGTGCGGAGGAAGTAGAAAGGCTAGCGACAGCCAG AAAGAAAAAAATGGAAAAAAGGAAAGGGAAGAACCCAATAAG CGAAATGTGCATTACCCGTATTTGTCGTTTGCATGTAGGAAGAA AAAAATGAGTCGAAGACCTACCGTTATATAGTCATTCTTTTTCTT ACGTTTTTACGCTTCGTTTCTTAACGCTTTTTAATAAGTATGAAAT ATATATCTGGTATTGGTTTTAGTACAGTTACCATTGTTTGTTTCTC CTGTTTCTTTTATTTTTTCTGCCGTTCTTGTATTTTTTATACCCATT CCCCTTTCTTGCATACGCCAGTCACTTTTCTTTATTGTATTGTTTA ACTTTCAAATTTCACTCTTGCTATTCTCTCTTGGTAGTACAAAGC ACAGCAGACACAAAGAAAAATGGCAAACCCAACAACAGGGAA GTCCTCGATTAGGGCTAAGCTTTCTAACTCATCGCTATCAAACCT ATTTAAAAAAAATAAAAATAAAAGACAGCGTGAGGAAACGGAA GAGCAGGACAATGAGGATAAGGATGAGAGTAAGAACCAGGATG AAAATAAGGACACACAGCTCACTACGAAGGAGTTTGAAGAGA AGGAGGCTCGTTACACCAACGAGTTGCCCAAGGAACTGCGCAA GTATCGTCCTAAAGGTTTCAGATTCAATTTGCCTCCAACGGATAG ACCCATCAGGATATATGCAGATGGTGTTTTTGATCTTTTCCATCTT GGCCACATGAAGCAACTGGAACAGTGTAAGAAGGCTTTCCCCA ATGTAACACTGATAGTTGGTGTGCCTAGCGACAAAATCACTCAC AAACTAAAAGGTTTGACTGTGCTGACCGATAAGCAGCGTTGTG AAACTTTAACGCACTGCAGATGGGTTGACGAAGTC XbaI; PsyI   2.2.5 Site-directed mutagenesis The QuickChange Primer Design Program was employed to construct mutagenic primers (Sigma- Aldrich) that were used in a PCR reaction using 125 ng upstream primer and 125 ng downstream primer in a final 50 L reaction, containing, 10 ng DNA template, 1 L deoxynucleotide triphosphate mix, 1 L PFU Ultra HF DNA polymerase and 3 L QuickSolution from QuickChange II Site-Directed Mutagenesis Kits (Angilent Technologies). The PCR program comprised an initial denaturation at 95°C for 1 min, followed by 18 cycles of 95°C for 50 sec, 60°C for 50 sec and 1 min per kb at 68°C and finally 68°C for 7 min. 1 L of DpnI restriction enzyme was added to the reaction and incubated for 1hr at 37°C. Per successful mutagenesis, 2 L or the needed volume was transformed into XL10-Gold Ultracompetent cells as described in 59    section 2.3.7. Colonies were picked, amplified and then subjected to mini-preparation DNA extraction as described in 2.3.9. The extracted DNA was sub-cloned into the sequence-verified plasmid and confirmed again by DNA sequencing. The following Table 2.8 presents the mutants made by this method. Table 2.8 S. cerevisiae plasmids used in this study (Site-directed mutagenesis) Plasmid Description Source/reference pRS313-PCT1Cmut-GFP PCT1-V169M/H195A/Y200A-GFP under control of PCT1 promoter in CEN/HIS3 vector. This study. pRS313-PCT14m-GFP PCT1-L262D/I273D/ F276D /F280D-GFP under control of PCT1 promoter in CEN/HIS3 vector. This study. pRS313-PCT1NLSm4m- GFP PCT1- K62A/R63A/R64A/ R65A/L262D/I273D/ F276D /F280D-GFP under control of PCT1 promoter in CEN/HIS3 vector. This study.   2.2.6 In-Fusion cloning To study membrane-binding domain deletion, In-Fusion cloning was used to generate region disruption constructs. This method depends on linearised vector mixing with two PCR products that have overlapping ends, and requires only two steps of PCR to obtain a sufficient amount of the gene disruption construct for one transformation experiment. Four PCR primers were designed in a way that they generate 5′- and 3′-regions of the target gene containing 15 bp extension ends that are homologous to the ends of the linearised vector; in addition, the two PCR products are homologous to each other. pRS313-PCT1-GFP is the template for building DNA transplant plasmids. For PCT1GPd [254-312]Δ, two PCR fragments were obtained with the primer pairs YC15/YC16 and YC19/YC20 (Table 2.3). Once amplified and spin-column purified, the PCR inserts were introduced into the vector by employing the In-Fusion Cloning kit (Takara Clontech). The in-fusion cloning reaction was performed on 100 ng linearised pRS313 vector (digested by XbaI and NcoI) in a final 10 L reaction, containing 200 ng purified upstream PCR product, 200 ng purified downstream PCR product and 2 L 5X In-Fusion HD Enzyme Premix. The reaction was incubated for 15 min at 50°C, and then placed on ice. Transformation was performed on Stellar Competent cells as described in 2.3.7. Positive colonies from the overnight 60    incubation at 37°C were prepared for mini-preparation DNA extraction as described in 2.3.9. The extracted DNA was further confirmed by DNA sequencing. Table 2.9, below, presents the mutants made using this method. Table 2.9 S. cerevisiae plasmids used in this study Plasmid Description Source/reference pRS313-PCT1GPd-GFP PCT1-[254-312]Δ-GFP under control of PCT1 promoter in CEN/HIS3 vector. This study   2.2.7 Amplification and expression of the PCYT1A gene from cDNA constructs Table 2.10 shows the list of plasmids expressing human PCYT1A used in this study. For expression human PCYT1A in yeast, the human PCYT1A gene was PCR amplified from pcDNA 3.1 constructs: pcDNA 3.1-mCherry-myc-PCYT1A; pcDNA 3.1-mCherry-myc-PCYT1A p.V142M; pcDNA 3.1-mCherry-myc-PCYT1A p.E280del and pcDNA 3.1-mCherry-myc- PCYT1A p.S333Lfs*164. The amplified cDNAs were cloned in frame into YCplac111-NOP- GFP at NotI restriction sites to generate N-terminal GFP tagged constructs under control of the NOP1 promoter (Table 2.10 and Table 2.11). The amplified cDNAs were also cloned in frame into YCplac111-NOP-GFP at NotI and BamHI restriction sites to generate a non-GFP tagged construct under control of the NOP1 promoter (Table 2.10 and Table 2.11). These GFP-tagged or non GFP-tagged constructs were further sub-cloned into an RS313 vector between XbaI and SacI restriction sites which were incorporated in PCR YC33/YC34 primers (Table 2.11). To conduct a more careful investigation of PCYT1A localisation, the NOP1 promoter was replaced with the Pct1 endogenous promoter in the wild-type PCYT1A constructs. It was created by PCR amplification of the Pct1 promoter segment from pRS313-PCT1 with primers YC21 and YC32 (Table 2.3 and Table 2.11), which included an XhoI site at the N-terminus and added a NotI site at the C-terminus. The PCYT1A gene segment resulted from the cutting of YCplac111-NOP- PCYT1A by NotI and BamHI. The two DNA segments were digested with restriction enzymes and combined in a three-part ligation with XhoI and BamHI digested pRS313-PCT1-GFP. The cloning product is named pRS313-PCYT1A. GFP coding sequences were further inserted in the BamHI site to create pRS313-PCYT1A-GFP. 61    Table 2.10 S. cerevisiae plasmids expressing human PCYT1A used in this study Plasmid Description Source/reference YCplac111-NOP-PCYT1A- GFP PCYT1A-GFP under control of NOP1 promoter in CEN/LEU2 vector. This study. By primers YC27/YC29 (Table2.11). YCplac111-NOP- PCYT1AV142M-GFP PCYT1A-V142M-GFP under control of NOP1 promoter in CEN/LEU2 vector. YCplac111-NOP- PCYT1AE280del-GFP PCYT1A-280Δ-GFP under control of NOP1 promoter in CEN/LEU2 vector. YCplac111-NOP- PCYT1A333fs-GFP PCYT1A-GFP under control of NOP1 promoter in CEN/LEU2 vector. PCYT1A ORF with S333L (a single- nucleotide deletion c.996delC), resulting in a frame shift leading to a considerably elongated C terminus missing several phosphorylation sites. This study. By primers YC27/YC30 (Table2.11). YCplac111-NOP-PCYT1A PCYT1A under control of NOP1 promoter in CEN/LEU2 vector This study. By primers YC27/YC28 (Table2.11). YCplac111-NOP- PCYT1AV142M PCYT1A-V142M under control of NOP1 promoter in CEN/LEU2 vector. Ycplac111-NOP- PCYT1AE280del PCYT1A-280Δ under control of NOP1 promoter in CEN/LEU2 vector. YCplac111-NOP- PCYT1A333fs PCYT1A under control of NOP1 promoter in CEN/LEU2 vector. PCYT1A ORF with S333L (a single-nucleotide deletion c.996delC), resulting in a frame shift leading to a considerably elongated C terminus missing several phosphorylation sites. This study. By primers YC27/YC31 (Table2.11). pRS313-NOP-PCYT1A-GFP PCYT1A-GFP under control of NOP1 promoter in CEN/HIS3 vector This study pRS313-NOP- PCYT1AV142M-GFP PCYT1A-V142M-GFP under control of NOP1 promoter in CEN/HIS3 vector. This study pRS313-NOP- PCYT1AE280del-GFP PCYT1A-280Δ-GFP under control of NOP1 promoter in CEN/HIS3 vector. This study pRS313-NOP-PCYT1A333fs- GFP PCYT1A-GFP under control of NOP1 promoter in CEN/HIS3 vector. PCYT1A ORF with S333L (a single- nucleotide deletion c.996delC), resulting in a frame shift leading to a considerably elongated C terminus missing several phosphorylation sites. This study 62    pRS313-NOP-PCYT1A PCYT1A under control of NOP1 promoter in CEN/HIS3 vector This study pRS313-NOP- PCYT1AV142M PCYT1A-V142M under control of NOP1 promoter in CEN/HIS3 vector. This study pRS313-NOP- PCYT1AE280del PCYT1A-280Δ under control of NOP1 promoter in CEN/HIS3 vector. This study pRS313-NOP-PCYT1A333fs PCYT1A under control of NOP1 promoter in CEN/HIS3 vector. PCYT1A ORF with S333L (a single-nucleotide deletion c.996delC), resulting in a frame shift leading to a considerably elongated C terminus missing several phosphorylation sites. This study pRS313-PCYT1A PCYT1A under control of PCT1 promoter in CEN/HIS3 vector This study pRS313-PCYT1A-GFP PCYT1A-GFP under control of PCT1 promoter in CEN/HIS3 vector This study   Table 2.11 Sequences of primers used to generate human PCYT1A expression constructs Primers Product name Orientation Sequences (5'→3') Note YC27 NotI_hpcyt1a_F Sense TTTGCGGCCGCGGGAGGTGA TGCACAGTGT NotI YC28 NotI_hpcyt1a_R Anti-sense TTTGCGGCCGCGTCTTCTTC ATCCTCACTGAT NotI YC29 BamHI-hpcyt1a_R Anti-sense TTTGGATCCGTCTTCTTCATC CTCACTGAT BamHI YC30 BamHI_hpcyt1a_333fs_R Anti-sense TTTGGATCCCACTACCTCCTT CTTCTG BamHI YC31 NotI_hpcyt1a_333fs_R Anti-sense TTTGCGGCCGCCACTACCTC CTTCTTCTG NotI YC32 PCT1PROM Anti-sense TTTCGCGGCCGCCATTTTTCT TTGTGTCTGCTGTGCT NotI YC33 Pcyt1a_HIS_fwd Sense TTTTCTAGACTGCAGGTCGA GGGGAGATCCATT XbaI YC34 Pcyt1a_HIS_rev Anti-sense TTTGAGCTCGGTACCCTAATA TTGGTCGTGTTAT SacI   63    2.3 Cloning techniques overview 2.3.1 Polymerase chain reaction Polymerase chain reaction (PCR) was performed on 50 ng of template DNA in a 50 L reaction containing 1x Phusion High-Fidelity or GC rich buffer (Thermo Fisher Scientific), 200 M dNTP mix (Promega), 0.5 L Phusion High-Fidelity DNA polymerase (Thermo Fisher Scientific) and 0.2 M of forward and reverse primers (Sigma-Aldrich). Cycling reactions were 30 sec at 98°C, followed by 25–30 cycles of 10 sec at 98°C, 30 sec per kb at 3°C above the Tm of the lower Tm primer and 20–60 sec at 72°C. A final extension of 10 min at 72°C was used. QIAquick PCR Purification Kit (Qiagen) was used to isolate and purify PCR products from amplification reactions according to the manufacturer’s instructions. PCR products were eluted in 30–50 L nuclease free water and the expected size was checked by agarose gel electrophoresis as described in section 2.3.2. 2.3.2 Agarose gel electrophoresis Agarose gels of various concentrations (0.5–2.0%) were prepared depending upon the size of DNA bands to be separated. Ultrapure agarose powder (Sigma-Aldrich) was mixed with the appropriate amount of 1x TAE buffer (CIMR media kitchen) and microwaved for 1–3 min until the agarose was completely dissolved. After the agarose solution was cooled down to room temperature, 0.01% SYBR Safe (Life Technologies) was added to the solution to allow DNA visualisation and was poured into a gel tray with the well comb in place. The gel was left to solidify at room temperature for 20–30 min and then loaded with samples and appropriate DNA ladder (Invitrogen). Electrophoresis was performed at 90 V for 30 min in 1x TAE buffer. Gels were visualised under ultra-violet light using the G:Box Chemi XL Gel Documentation System (Syngene). 2.3.3 Agarose gel extraction To obtain purified DNA, DNA products were separated by agarose gel electrophoresis and the band of interest was excised from the gel to remove excess primers and DNA fragments. The following clean-up was performed using Gel extraction kit (Qiagen) according to the manufacturer’s protocol. In general, the gel-encased DNA was dissolved in the optimal pH buffer 64    with the assistance of gentle heat and then DNA was purified by precipitation in an isopropanol solution onto the silica membrane of the spin column. Contaminants were removed by two spin- wash steps. The DNA fragment of interest was finally eluted in 30–50 μL of nuclease free water. 2.3.4 Restriction enzyme DNA digestion Vector and PCR products were digested using FastDigest enzymes (Thermo Fisher Scientific) or restriction enzymes (New England BioLabs) to generate compatible ends capable of being ligated. A standard reaction contained 1.5–2.0 L of restriction enzyme, 1–3 g DNA, 1x digest buffer, and nuclease free water to a final volume of 50 L. The reaction was incubated at digestion temperature (typically 37°C) for 1 hr, stopped by heat inactivation at 65°C for 15 min and held at 4°C. Samples were then loaded onto an agarose gel for electrophoresis. 2.3.5 Alkaline phosphatase treatment To prevent recircularisation of cloning vector DNA during ligation, digested vector was dephosphorylated by Antarctic Phosphatase (New England BioLabs) treatment. A typical reaction contained 30L digested vector, 1L of alkaline phosphatase and 1x Antarctic phosphatase buffer. The mixture was incubated for 30 min at 37°C, followed by 5 min at 80°C to inactivate the enzyme. The dephosphorylated vector was purified using agarose gel extraction as described in section 2.3.3. 2.3.6 DNA Ligation Typical DNA ligation was performed using a Rapid DNA Ligation Kit (Roche) according to the manufacturer’s instructions. A standard ligation reaction used a molar ratio of 1:3 linearised and dephosphorylated vector DNA to insert DNA. Ligations were performed as 20 L reactions. Vector DNA and insert DNA were dissolved thoroughly and mixed in 1x conc. DNA dilution buffer to a final volume of 10 µL, and then 10 L of T4 DNA ligation buffer and 1L of T4 DNA ligase were added. Reactions were performed at room temperature for 15 min before transformation into the appropriate bacterial cell line for cloning purposes. 2.3.7 Bacterial transformation XL-1 Blue Supercompetent cells (Life Technologies), XL10-Gold Ultracompetent cells (Agilent Genomics) or Stellar Competent cells (Takara Clotech) were transformed with the appropriate 65    amount of ligation mixture according to the manufactures’ instructions. Briefly, approximately 45 L of competent cells were incubated on ice for 30 min with DNA prior to heat shock at 42°C for 45 sec. Cells were incubated on ice for a further 2 min before 300–500 L of SOC media was added. Cells were incubated at 37°C with shaking at 225 rpm for 1 hr, before being streaked onto LB-Agar plates containing the appropriate selection of antibiotics. Plates were incubated at 37°C for approximately 17 hr. 2.3.8 Colony PCR Colonies were picked using a 10 L pipette tip, suspended in 20 L sterile water and 1 L of the suspension was used as the DNA template for screening PCR reactions. The reaction mixture (10 L) consists of the DNA template, 1x GoTaq Green Master Mix (Promega) and 1.25 M of suitable forward and reverse primers. The reaction conditions were 95°C 5 min, followed by 30 cycles of 95°C for 30 sec, 55–65°C for 45 sec, 72°C for 2 min and a final extension of 72°C for 10 min. PCR products were separated on an agarose gel as described in section 2.3.2. Candidate colonies were re-streaked for single colonies and plasmids were purified. 2.3.9 Plasmid DNA purification Colonies were grown for 16–18 hr in 3 mL of LB containing 100 g/mL ampicillin or other appropriate antibiotics. Cells were pelleted and plasmid DNA prepared using the Wizard Plus SV Miniprep DNA Purification System (Promega) as per the manufacturer guidelines. In brief, cells were resuspended, lysed and alkaline phosphatase added to inactivate endonucleases. The lysate was neutralised, and the cell debris removed before the cleared lysate was loaded on a spin column. Contaminants were then removed by two spin-wash steps. The purified plasmid DNA was eluted from the column in 100 L of nuclease free water. DNA concentration and quality were assessed by a Nanodrop 1000 Spectrophotometer (Thermo Fisher Scientific). The mutation of interest and sequence fidelity of the gene was confirmed through sequencing as described in section 2.3.10. In order to increase the yield and purity of plasmid DNA, 200 mL of bacterial culture supplemented with appropriate antibiotics was grown for 16–18 hr at 37°C by shaking at 225 rpm. Plasmid DNA was extracted using the HiSpeed Plasmid Maxi Kit (Qiagen) as per the manufacturers’ guidelines and resuspended in 1mL of nuclease free water. 66    2.3.10 Plasmid DNA sequencing Plasmids were sequenced to verify fidelity against the reference sequence. 50–100 ng of plasmid were subjected to a cycling reaction in the following total 10 L mixtures: 0.5 L BigDye Terminator Mix (Applied Biosystems), 1.75 L ABI 5x buffer (Applied Biosystems), 0.5 M sequencing primers (Sigma-Aldrich), and MilliQ water to make a final reaction volume of 10 L. Reaction conditions were 25 cycles of 96°C for 10 sec, 50°C for 5 sec and 60°C for 2–5 min. Dye terminator removal was achieved using Agencourt CleanSeq beads (Beackman Coulter) according to the manufactures’ instructions. Cleaned up products were run on a 3130 or 3730 DNA Analyser (Applied Biosystems). Sequence analysis was performed using Sequencher DNA sequence analysis software (Gene Codes) to confirm overall sequence fidelity, placement of start and stop codons, and any generated mutations or tags. 2.4 Fluorescence microscopy Cells were grown to either an exponential or a post-diauxic phase at 30°C in synthetic medium, pelleted at the appropriate density and mounted on slides (Thermo Fisher Scientific). Images were acquired on a Zeiss AxioImager epifluorescence upright microscope with a 100× Plan- Apochromatic 1.4 numerical aperture (NA) objective lens (Carl Zeiss) and recorded using a large chip sCMOS mono camera for sensitive fluorescence imaging (ORCA Flash 4, version 2). Raw image files were saved by Zeiss ZEN blue software and exported to Photoshop (Adobe). In a sequential scan, GFP was excited at 488 nm and the emission signal was collected between 494 and 562 nm; mCherry was excited at 587 nm and the emission was collected between 594 and 719 nm. 2.5 Confocal microscopy 2.5.1 Confocal microscopy of Saccharomyces cerevisiae Yeast cells were grown to the indicated growth phase at 30°C in selective SC medium, pelleted and mounted on slides (Thermo Fisher Scientific). Cells were immediately imaged at room temperature using a Leica TCS SP8 confocal microscope with 63x 1.4 numerical aperture oil- immersion objective lens. In a sequential scan using white light laser, GFP was excited at 488 nm and the emission signal was collected between 494 and 562 nm; mCherry was excited at 587 nm 67    and the emission was collected between 594 and 719 nm. For lipid droplet labelling, yeast cells were stained for 10 min with 10 μM monodansylpentane (MDH; SM1000a; Abgent). MDH signal was acquired using diode laser at 405 nm excitation and emission was collected between 420 and 480 nm. Images were analysed using Leica LAS AF Lite software and ImageJ. 2.5.2 Quantification of lipid droplet area Quantification of the lipid droplet area was performed using ImageJ. Lipid droplet structures larger than 20 pixels and smaller than 3 pixels were filtered by an FFT Bandpass Filter, with threshold adjustment used to assist the calculation of the particles. 2.5.3 Photobleaching assay Photobleaching experiments were performed on a Leica TCS SP8 confocal microscope with the optional FRAP Booster enabled. Yeast cells expressing Pct1-GFP and Sec63-mCherry were pelleted and imaged at room temperature. Images were acquired using a 63x oil immersion objective lens. After acquiring two images at 5-sec intervals, selected regions of interest were photobleached with 3 iterations of 100% laser power (white light laser) at 488 nm. The fluorescence intensity of GFP at the regions of interest was recorded for another 8–15 frames at 15–30 sec intervals. The photobleaching cycle was repeated 2–4 times. For FRAP and FLIP analyses, the fluorescence intensity of GFP was corrected by the loss of GFP fluorescence intensity obtained from cells under identical conditions but without a photobleaching event. The distance between the selected cells was greater than 3.5 m. 2.5.4 Quantification of GFP tagged protein localisation Quantification of Pct1-GFP localisation was performed using ImageJ by drawing a line of 5-pixel average width across the cells and plotted pixel intensities were used to determine Pct1-GFP localisation. Pct1-GFP was scored as intranuclear when the Pct1-GFP pixel intensity increased between the maximum pixel intensity of the Sec63-mCherry (an ER marker surrounding nucleus), and scored as nuclear membrane-bound when the maximum pixel intensity for both GFP and mCherry overlapped. To detect Pct1-GFP association with LDs, 3D images were acquired and analysed in IMARIS software to determine their co-localisation. 68    2.6 Immunoblotting of yeast protein 2.6.1 Preparation of yeast lysate for western blotting Yeast cells were grown in selective SC medium from OD600 0.05 to OD600 0.5–1.0 or to stationary phase OD600 5.0–7.0. Around a total of 15 OD600 (for example, 3.0 mL of OD 600 5.0) of yeast cells were harvested by centrifugation at 3,000 rpm for 3 min and these cells were washed once with 1 mL sterile water. Yeast cells were pelleted again by centrifugation at 3,000 rpm for 3 min and resuspended in the appropriate amount of LDS buffer containing 33% v/v of NuPAGE LDS sample buffer 4x (Thermo Fisher Scientific), 57% v/v of sterile water and 10% v/v of 1M dithiothreitol (Sigma-Aldrich). The amount of LDS buffer was adjusted to 32 L per OD600 yeast cells or 12.5 L per OD600 for stationary phase cells. 100 mg of glass beads (BioSpec products) was added to the yeast suspension which underwent two rounds of vigorous vortexing for 30 sec followed by 2 min incubations at 95°C. These beads were then spun down at 13,000 rpm for 15 min. The supernatant was directly loaded into the SDS-PAGE protein gels or stored at −80 °C for future use. 2.6.2 SDS-PAGE electrophoresis NuPAGE 4-12% gradient Bis-Tris gels (Invitrogen) were used and the appropriate amount of protein sample was loaded into each well in the gels. The gels were run using the Xcell system (Thermo Fisher Scientific) for 150 min at 120 V in 1x MOPS running buffer. Once the dye front had reached the positive end of the gel and the unit had been disassembled, protein was transferred onto a nitrocellulose membrane at 20 V for 7 min using the iBlot dry blotting system (Invitrogen). 2.6.3 Western blotting Antibodies used in the yeast study are listed in Table 2.12. Nitrocellulose membranes were blocked in 3% (w/v) bovine serum albumin (BSA; Sigma-Aldrich) or 5% (w/v) powdered Marvel dried skimmed milk (Premier International Foods Ltd) in Tris-buffered saline pH 7.6 (TBST) containing 50 mM Tris, 150 mM NaCl, 0.1% TWEEN 20 for 1 hr. Membranes were then incubated with an appropriate amount of primary antibody in 3% BSA/TBST or 5% milk/TBST for up to 16 hr at 4°C. Membranes were rinsed 3–5 times in TBST for 5 min before being incubated in the appropriate HRP-conjugated secondary antibody in 3% BSA or 5% millk for 1 69    hr at room temperature. Membranes were then washed 3–5 times in TBST for 5 min. Protein signals were developed using enhanced chemiluminescence (ECL) reagents (GE Healthcare or Millipore) and detected using the Image Quant LAS4000 (GE Healthcare) or ChemiDoc MP Imaging system (Bio-Rad). Table 2.12 Antibodies used in yeast studies Catalog No. Blocking (w/v) Primary antibody (w/v) Secondary antibody (w/v) Anti-Myc Millipore; 4A6 3% BSA/TBST 1:1000 in 3% BSA/TBST Anti-mouse 1:5000 in 3% milk/TBST Anti-mCherry GeneTex; GTX128508 5% milk/TBST 1:3000 in 5% milk/TBST Anti-rabbit 1:5000 in 5% milk/TBST Anti-Actin Abcam; ab8227 5% milk/TBST 1:5000 in 3% BSA/TBST Anti-rabbit 1:5000 in 3% milk/TBST Anti-GFP Roche; 11814460001 5% milk/TBST 1:1000 in 3% BSA/TBST Anti-mouse 1:5000 in 5% milk/TBST Anti-GAPDH GeneTex; GTX100118 5% milk/TBST 1:30000 in 3% BSA/TBST Anti-rabbit 1:5000 in 5% milk/TBST Anti-DPM1 Abcam; ab113686 5% milk/TBST 4 g/mL in 3% BSA/TBST Anti-mouse 1:5000 in 5% milk/TBST Anti-ERG7 Dr. Günther Daum, TU-Graz, Austria. 5% milk/TBST 1:5000 in 3% BSA/TBST Anti-rabbit 1:5000 in 5% milk/TBST Anti-AYR1 Dr. Günther Daum, TU-Graz, Austria. 5% milk/TBST 1:5000 in 3% BSA/TBST Anti-rabbit 1:5000 in 5% milk/TBST   2.7 Saccharomyces cerevisiae screening assay 2.7.1 5-Fluoroorotic acid assay For yeast survival and growth assays, cho2Δopi3Δpct1Δ cells carrying YCplac33-URA3-OPI3 were transformed with CEN/HIS plasmids expressing the indicated constructs. Cells were grown in the exponential phase in a selective SC medium, with the indicated supplementation when required, and 10 μL of standard 10-fold serial dilutions were spotted onto the appropriate SC plates with 1 g/L 5-Fluoroorotic acid (5-FOA) and incubated at 30°C for 2–4 days. 70    2.7.2 Phleomycin assay 2.5 g/mL or 5.0 g/mL phleomycin from aqueous-concentrated Streptomyces verticillus solutions (2.0 mg/mL) was freshly dissolved in YPD agar medium and stored in the dark. Yeast cells were grown in the appropriate selective SC medium to early logarithmic phase at 30°C. 10 μL from a 5-fold dilution was spotted onto YPD plates or phleomycin plates. The plates were incubated at 30°C for 2–4 days. 2.8 Saccharomyces cerevisiae lipidomic profiling analysis Yeast cells were grown to the indicated growth phases and prepared for mass spectrometry analysis as previously described (Folch et al., 1957) with minor modifications. Briefly, 50 mg of whole yeast cells were homogenised with 1 mL 2:1 chloroform-methanol mixture (v/v) and 100 mg 0.5 mm diameter glass beads (BioSpec products) in a FastPrep-24 instrument (MP biomedicals), using five short pulses of 5 m/s for 1 min with 1 min on ice between each pulse. Mixtures were sonicated in an ultrasonic bath (Elmasonic S10, Elma) at room temperature for 5 min. The homogenised cells were emulsified with 400 μL sterile water and mixed thoroughly for 1 min using Vortex-Genie 2 (Scientific Industries). The samples were centrifuged at 13,200 rpm for 10 min. The organic layer was collected and the remaining mixture was used in a second lipid extraction by following the same procedures. The organic layers from both extractions were combined in an amber 2 mL glass container (Agilent Technologies) and allowed to air-dry in a fume hood overnight. 2.8.1 Lipid profiling analysis For the lipid analysis, we collaborated with Dr Albert Koulman by using a LCMS method similar to that described previously (Koulman et al., 2009; Lu et al., 2016). This method given below was written by Dr Koulman. After addition of 60 μL internal standard in methanol (containing: Cholesteryl-2,2,3,4,4,6-d6-octadecanoate CE(18:0-d6), Myristoylphosphocholine-d42 LPC(C14:0)-d42, 1-palmitoyl(d31)-2-oleoyl-glycero-3-phosphocholine PC(C16:0-d31/C18:1), 1- palmitoyl(d31)-2-oleoyl-glycero-3-phosphoethanolamine PE(C16:0-d31/C18:1), N- palmitoyl(d31)-d-erythro-sphingosylphosphorylcholine (16:0-d31 SM) M(C16:0-d31), Glyceryl- tri(hexadecanoate-d31) (48:0-d93 TAG) all at 10 µg/mL), samples were vortexed 740 µL of 4:1 mix of 2-propanol and acetonitrile were added. 71    Using an UltiMate 3000 Rapid Separation LC System (Thermo Fisher Scientific) full chromatographic separation of intact lipids was achieved: briefly, injection of 5 µL (partial loop) onto an Acquity UPLC® BEH, stationary phase C18, 130 Å, 1.7 µm, I.D. 2.1 mm X 50 mm reversed phase UPLC-column maintained at 55°C. Mobile phase A was made up of 6:4, acetonitrile and water with 10 mM ammonium formate solution. Mobile phase B was made up of 9:1, 2-propanol and acetonitrile with 10 mM ammonium formate solution. The flow was maintained at 1,000 µL per minute through the following gradient changes. Time (min) % Mobile phase A: 0.00 min/60% A; 0.40 min/57% A; 0.45 min/50% A; 2.40 min/46% A; 2.45 min/30% A; 2.90 min/19% A; 4.00 min/1.0% A; 5.00 min/1.0% A; 7.00 min/60% A; 8.00 min/60% A. Total run time per sample was 8 min with a cycle time of 10 min during which the column re- equilibrated and the injection needle was washed (washing solution is 9:1, IPA and ACN with 0.2 % formic acid). The mass spectrometer used was the Thermo Scientific Exactive Orbitrap (Thermo Fisher Scientific) with a heated electrospray ionisation source (HESI). The mass spectrometer was tuned routinely using different positive ionisation calibration solution (recommended by Thermo Scientific). Additionally, the HESI is optimised at 50:50 mobile phase A to mobile phase B for spray stability (Capillary temperature: 380°C; source heater temperature: 420°C; sheath gas flow: 60 (arbitrary); auxiliary gas flow: 20 (arbitrary); sweep gas: 5 (arbitrary); source voltage: 3.5 kV). The mass spectrometer resolution was set to 50,000 with a full-scan range of m/z 110 to 1200 Da, cycling between positive and negative mode. For each lipid the area under the curve was determined using high resolution extracted ion chromatograms with a window of 5 ppm and expressed relative to the internal standard for that lipid class. 2.8.2 Estimates of the lipid contribution to membrane stored curvature elastic (SCE) stress of cells We collaborated with Dr Marcus Dymond for the mathematical modelling in each lipid profiling analysis, and this method given below, was written by Dr Dymond. To a first order approximation, the two material parameters C0 (the spontaneous curvature) and KM (the bending modulus) in combination govern the membrane SCE stress () within a lipid bilayer, Equation 1.  = - 2KMC0 Equation 1 72    Estimates of curvature elastic  can be made using a linear mixing rule, where C0 for a mixture of lipids (C0mix) is determined from C0 values for the individual lipid components of the mixture, using a principle of ideal additive mixing (Dymond et al., 2016). Thus C0mix for a mixture of two lipids (A, B) in a bilayer is given by Equation 2, xB is the mole fraction of lipid B and C0A and C0B are the spontaneous curvatures of lipids A and B respectively. 𝐶଴௠௜௫ ൌ ሺ1 െ 𝑥஻ሻ𝐶଴஺ ൅ 𝑥஻𝐶଴஻ Equation 2 KM varies by a relative small amount in comparison to C0 for lipids and an average value (5.0 x 10-20 J) was used for estimating. Lipid spontaneous curvature is defined as the C0 = 1/R0 where R0 is the radius of curvature of an unstressed monolayer cylinder of lipids in an inverse hexagonal lyotropic liquid crystal phase, when measured at the pivotal plane. The pivotal plane is that point along the length of the lipid molecule where the area occupied does not change on bending the lipid monolayer. Increases in lipid unsaturation, chain branching and to a lesser extent chain length make R0 tighter and thus increase SCE stress in membranes. Changes to lipid headgroups are less straight forward to predict, for neutral headgroups the size of the headgroup is the dominating feature, assuming hydrocarbon chain structure is unchanged. Thus, smaller headgroups make R0 tighter and increase SCE stress. Charged headgroups introduce the complication of greater headgroup- headgroup repulsion, which favours less tight values of R0 and can decrease SCE stress. Estimates of the SCE stress for membranes composed of the lipidomes reported were made, using Equation 1 and 2 above. When C0 values of lipids with the correct chain length were not available in the literature, C0 for a lipid with the same headgroup and with the same distribution of unsaturation in each chain, with overall similar chain length was used. Estimates of lipid contribution to stored curvature elastic stress were calculated by including lipid compositions for PC, PE, PS, PA and PG lipids. PI lipids were omitted due to their being no C0 values in the literature. A sensitivity analysis showed the inclusion of PI lipids, with speculative C0 values, had no significant effect on the trends reported, due to their relatively low abundance in the membrane. DAG and TAG lipids were omitted from this analysis, since these are predominantly associated with lipid droplets (Kuerschner et al., 2008). This analysis was also repeated with the DAG levels included, with very similar patterns of the results, likely reflecting that DAG levels did not change significantly over time. 73    2.8.3 Estimate of the total membrane stored curvature elastic (SCE) stress of cells (referred to herein as PSCE) PSCE is a data-driven metric of the total curvature elastic stress in a cellular system i.e. a proxy of lipid and protein contributions to SCE stress. It is determined from the absolute amounts of lipid using previously published methodology (Dymond, 2016), using the results of a data-driven modelling approach. Where PSCE is defined by a ratio control function, Equation 3, which calculates the balance of lipids that increase stored curvature elastic stress (type II lipids) and lipids that decrease stored curvature elastic stress (type 0 lipids). PSCE is calculated from the lipidome for each population of cells. , 0 , 0 0 [ ] 1 [ ] n II n m m b n SCE a m w L p L w     Equation 3 Where [LII, n] denotes the concentration of type II lipid n and (w)n is a weighting factor for lipid n. Similarly [L0, m] is the concentration of type 0 lipid m and (w)m is the weighting factor for lipid m. The variables a and b are the total numbers of type 0 and type II lipids respectively. Values of the weighting factor w were obtained from a data driven modelling approach and taken from previously published studies (Dymond, 2016), where cells were cultured under conditions that altered their membrane composition. w was determined from a coarse-grained model of lipid structure based on well-established trends, of how the number of unsaturation in each lipid chain and the headgroup structure contribute to SCE stress. The final w value selected gave the lowest variance in PSCE across a range of asynochronous cell populations and thus reflect, in a course- grained way, both protein and lipid contributions to total membrane SCE stress. The two sets of lipids (type II and type 0) are determined from their w values compared to w for the pivot lipid (wLp). Lipids with w > wLp are denoted type II, lipids with w ≤ wLp are denoted type 0. Previous work shows that this pivot lipid species is PE (34:1). Calculations of PSCE were performed using PC, PE, PS, PA, PI and DAG lipids. TAG lipids were omitted due their location in lipid droplets rather than the within cellular membranes. PG was omitted from the analyses, due their being currently no parameters available to incorporate it. A sensitivity analysis showed that the low abundance of PG lipids means that their inclusion in the models has no effect on the data and trends reported. 74    2.8.4 Brief summary of data driven model Data driven modelling studies are one of the few to provide any quantitative insight into the mechanism of regulation of the total cellular stored curvature elastic (SCE) stress in cells by calculating a data driven metric (PSCE). The experimental methodology to develop this approach utilised lipidomics to determine a set of parameters (w values in Equation 3) from a variety of cell types under culture conditions that cause compositional changes to occur in their membranes (Dymond, 2015; Dymond, 2016; Dymond et al., 2013; Hague et al., 2013). In a very coarse- grained way these w values take into account both protein and lipid contributions to SCE stress in cells, which we summarise below. The parameters w values in Equation 3 determined in previous studies (Dymond, 2015; Dymond, 2016; Dymond et al., 2013; Hague et al., 2013) were applied to the lipidomic data we obtained and this is why we can state the PSCE takes into account both protein and lipid contributions to SCE. It is well established that non-bilayer preferring lipids constrained to bilayer topologies cause significant SCE stress in lipid bilayers. It is also well established that some proteins can relieve and contribute to SCE stress in bilayers – notably CCT (Attard et al., 2000; Davies et al., 2001). Thus to a first order approximation the total SCE stress in a cellular membrane will be the sum of the lipid and protein contributions to it, Equation 4. CSCE = LSCE + ESCE Equation 4 There are currently no methods available to disentangle and quantify the varying magnitudes of LSCE and ESCE to CSCE. The data driven modelling methodology states that total cell stored elastic stress (CSCE) is equal to lipid stored elastic stress (LSCE) plus protein stored elastic stress (ESCE), Equation 4, which is summarised by the ratio control function Equation 3. To develop the PSCE metric we state that CSCE = , 0, 0 0 [ ] 1 [ ] n II n m m b n SCE a m w L p L w     75    where L describes the individual concentrations of lipids in the lipidome and w is a factor that effectively takes into account both the individual lipid contributions to CSCE and the magnitude ESCE. In the original work, individual populations of cells, under experimental conditions where biomembrane composition changed but CSCE was anticipated to be constant, lipidomics was used to determine [L]. An iterative, data-driven approach was then used to find values of w that gave the lowest variance in the calculated value of PSCE. This was repeated using the lipidomes of multiple cell types and populations to find a set of ‘best’ w values that could be used in subsequent studies to estimate PSCE as published (Dymond, 2015; Dymond, 2016; Dymond et al., 2013; Hague et al., 2013). Thus, the w values we have used for the calculation of PSCE reflect in a very basic way the contribution of ESCE to CSCE (Equation 4) and thus take some account of protein contribution. It must be stressed that this is a very coarse-grained approach. However, it is a step up compared to metrics like the unsaturation index and PE/PC ratio, which either ignore headgroup contributions or chain contributions to SCE and for that matter the contribution of proteins to SCE. 2.9 Isolation of Saccharomyces cerevisiae lipid droplets LD purification was performed as previously described (Fei et al., 2008; Leber et al., 1994) with minor modifications. All reagents were purchased from Sigma-Aldrich unless otherwise specified. In brief, cells were grown in selective medium to about 600 OD600. Cells were centrifuged at 3,300 rpm for 4 min (Allegra X-15R centrifuge, Beachman Coulter), washed in sterile water, preincubated in 100 mM Tris-HCl pH 9.4, and 10 mM DTT for 15 min at room temperature, washed, and resuspended in 30 mL spheroplasting buffer (1.2 M sorbitol and 20 mM K2HPO4/KH2PO4, pH 7.4). For spheroplast preparation, Zymolase 20T (Amsbio) was added (15 mg/g cell pellet) followed by incubation at 30oC and mixed gently at 25 rpm. Cells were incubated until a 10-fold drop in OD600 was observed (60–90 min). Spheroplasts were recovered by centrifugation (2,000 g, 4oC, 5 min), washed with spheroplasting buffer, and resuspended in breaking buffer (BB; 10 mM Tris-HCl, pH 6.9, 12% [wt/wt] Ficoll400, and 0.2 mM EDTA) at a final concentration of 0.15 g cells (wet weight)/mL. PMSF (1 mM) and Complete EDTA-free protease inhibitor (Roche) were added before homogenisation (loose-fitting pestle, 20 strokes) in a Dounce homogeniser on ice. The homogenate was transferred into 13.5 mL Ultra-Clear 76    centrifuge tubes (Beckman Coulter) and overlaid with an equal volume of BB and centrifuged (60 min, 28,000 rpm) in an optima LE-80K centrifuge (Beckman Coulter) with an SW-40 swinging bucket rotor. The floating layer was collected from the top of the gradient and the LDs were further purified. The floating layer was gently resuspended in BB using a Dounce homogeniser with loose-fitting pestle (5 strokes), transferred to a 13.5 mL Ultra-Clear tube, and overlaid with an equal volume of 10 mM Tris-HCl pH 7.4, 8% [wt/wt] Ficoll400, and 0.2 mM EDTA. Centrifugation was repeated as before (60 min, 28,000 rpm). The floating layer was gently collected and resuspended in 10 mM Tris-HCl, pH 7.4, 0.6 M sorbitol, 8% [wt/wt] Ficoll400, and 0.2 mM EDTA, transferred to 13.5 mL Ultra-Clear tubes, overlaid with equal volume of 10 mM Tris-HCl, pH 7.4, 0.25 M sorbitol, and 0.2 mM EDTA, and centrifuged once more (30 min, 28,000 rpm). The recovered high-purity top LDs fraction was snap frozen, and stored at −80°C. 2.10 Mammalian cell culture Human primary skin fibroblasts from PCYT1A mutant patients and healthy controls were cultured in Dulbecco’s modified eagle’s media (DMEM; Sigma-Aldrich) supplemented with 10% (v/v) fetal bovine serum (FBS, Hyclone), 1% L-glutamine, 1% non-essential amino acids, 1% sodium pyruvate and 1% penicillin/streptomycin. For TLC analysis, cells were grown in Dulbecco’s modified eagle’s media (DMEM; Sigma-Aldrich) supplemented with 10% or 1.0% (v/v) lipid- depleted fetal bovine serum (Biowest), 1% L-glutamine, 1% non-essential amino acids, 1% sodium pyruvate and 1% penicillin/streptomycin for 48 hr after cells were allowed to become confluent in the initial cell growth media. All cells were incubated at 37ºC in a 5% CO2 humidified chamber and passaged every 3–4 days. Five skin fibroblasts cell lines were established from the following patients, P0616 (p.Val142Met/p.Gllu280del), P0689 (p.Glu280del/p.Ser333Leufs*164), P2233 (homozygous p.Ala99Val mutation), P8069 (homozygous p.Glu129Lys mutation) and P8123 (homozygous p.Ser323Argfs*38 mutation); and five unrelated control fibroblast cell lines, C100, C101, C103, C104 and C105. Murine Ear Mesenchymal Stem Cells (EMSC) were cultured to <70% confluency in DMEM-F12 Ham media supplemented with 15% FBS, 2.5 mM L-Glutamine, 10 ng/mL recombinant murine FGF-basic (Peprotech) and 100 μg/mL Primocin (InvivoGen). To differentiate EMSC cells into adipocytes, cells were grown to 100% confluence (day 0) and incubated for 2 days after the 77    medium was substituted for DMEM-F12 Ham media supplemented with 15% FBS, 2.5 mM L- Glutamine, 100 μg/mL Primocin (InvivoGen), 1 μM insulin, 0.5 mM IBMX, 1 μM dexamethasone and 5 μM troglitazone (Cayman Chemical). On day 2 the medium was changed to DMEM-F12 Ham media containing 15% FBS, 2.5 mM L-Glutamine, 100 μg/mL Primocin (InvivoGen), 1 μM insulin and 5 μM troglitazone. On day 4 and every 2 days thereafter the media was replaced with new DMEM-F12 Ham media containing 15% FBS, 2.5 mM L-Glutamine and 100 μg/mL Primocin (InvivoGen). 2.11 siRNA transfection and knockdown of target genes To induce knockdown of endogenous PCYT1A expression in EMSC adipocytes, cells seeded in 6-well plates were either transfected with 30 nM Silencer Negative Control siRNA (Life Technologies) or 30 nM Pcyt1a siRNA (Life Technologies; s64594) using Lipofectamine RNAiMAX (Life Technologies) every 2 days after the cells reached confluency (day 0 for EMSC) and protein lysates were collected on day 10 of differentiation and Oil Red O staining was performed at day 10 on mature adipocytes. 2.12 Lipid (Oil Red O) staining and quantification To assess adipogenic differentiation of EMSC cells qualitatively, Oil Red O (ORO) staining was performed to measure accumulated neutral lipids in cells. Differentiated cells were initially washed with warm PBS, and fixed with 3.7% neutral-buffered formaldehyde (Formalin solution neutral buffered 10%, Sigma-Aldrich) for 15 min at room temperature. After fixation, cells were washed twice with PBS, and rinsed briefly with 60% isopropanol (2-Propanol ACS reagent, ≥99.5%, Sigma-Aldrich). The cells were then stained for 30 min with a working solution, which contained Oil Red O stock solution (2.5 mg/mL, dissolved in 100% isopropanol)/60% isopropanol/sterile water (6:1:3, v/v/v). Cells were then rinsed with 60% isopropanol and washed again with PBS to remove excess unbound dye. For a quantitative measurement of the intensity of Oil Red O stained LDs, dimethyl sulfoxide (DMSO) was added to each sample to dissolve the bound dye and left shaking at room temperature for 5 min. The absorbance of the samples was read at 492 nm using a TECAN spectrophotometer. 78    2.13 Immunoblotting from cell extracts Antibodies used in the mammalian study are listed in Table 2.13. To extract total protein, cells were washed twice with ice-cold PBS and lysed on ice using RIPA buffer (Sigma-Aldrich) containing Complete-Mini Protease Inhibitor cocktail (Roche) and PhosSTOP Phosphatase inhibitors (Roche). The cell lysate was clarified by centrifugation at 14,000 g for 15 min at 4°C and the protein concentration was determined using the colorimetric DC Protein Assay (BioRad) and analysed using the Asys Expert Plus microplate reader, wavelength 595 nm (Biochrom). For immunoblotting PCYT1A, Calnexin, Perilipin1 and aP2, cell extracts containing 10–30 μg total protein were diluted with NuPAGE 4x LDS sample buffer (Invitrogen), boiled for 7 min at 95°C, and then subjected to SDS-PAGE using precast NuPAGE 4–12% gradient Bis-Tris mini- gels (Invitrogen). Western blot procedures as described in section 2.6.2 and 2.6.3 were then conducted. Table 2.13 Antibodies used in mammalian study Catalog No. Blocking (w/v) Primary antibody (w/v) Secondary antibody (w/v) Anti-PCYT1A Epitomics; 3403-1 5% milk/TBST 1:1000 in 5% milk/TBST Anti-rabbit 1:5000 in 3% milk/TBST Anti-Perilipin Vala Sciences; 4854 3% BSA/TBST 1:2000 in 5% BSA/TBST Anti-mouse 1:5000 in 5% milk/TBST Anti- aP2 Santa Cruz; sc-18661 3% BSA/TBST 1:5000 in 5% milk/TBST Anti-goat 1:10000 in 5% milk Anti-Calnexin Abcam; ab75801 5% milk/TBST 1:5000 in 3% BSA/TBST Anti-rabbit 1:5000 in 5% milk/TBST               79    2.14 PEMT activity measurement PEMT activity was measured in human fibroblasts and mouse liver homogenates as a positive assay control. Samples were assayed for PEMT activity by measuring PC production from phosphatidylethanolamine (PE) and S-adenosyl-[methyl-3H] methionine (SAM[3H]). In brief, 68.7 μL samples (30 μg proteins in RIPA Lysis and Extraction Buffer) and 51.3 μL PEMT assay buffer were mixed in a 15 mL glassed capped tube (Thermo Fisher Scientific). The PEMT assay buffer consisted of 18.3 μL 1.25 M Tris-HCl buffer (pH 9.2), 18.3 μL of DTT (50 mM) and 14.7 μL phosphatidylethanolamine (PE) substrate mix (3.75 mg PE in 20 mM Tris-HCl, pH 9.1, 0.01% EDTA (w/v) and 1% Triton). The reaction was carried out in a water bath at 37°C with the addition of 30 μL SAM[3H] (2 mM, 0.50 μCi/mmol). After 30 min incubation, the reaction was stopped by adding 2 mL chloroform/methanol (2:1, v/v) and 2 mL 0.5% NaCl. The reaction tubes were briefly vortexed, centrifuged at 4,500 rpm for 10 min, and then the upper phase of the solution was discarded. The lower phase of the solution was washed three times with 2 mL of wash buffer (0.5% NaCl/methanol/chloroform, 25:25:2, v/v/v) and 200 μL transferred into a scintillation vial. The solution was then air-dried and solubilised with 4 mL scintillation fluid. After vortexing, samples were immediately analyzed for the level of radioactivity present. The specific activity of the SAM[3H] was also determined by measuring its radioactivity using a 30 μL aliquots. The PEMT activity was calculated as shown below: Activity = [(dpm/assay)*10] / [(specific activity)*(time)*(mg protein/assay)] 2.15 Lipid extraction from fibroblast cells Fibroblast cells were scraped from a confluent 10 cm dish in 1 mL ice-cold PBS buffer and transferred into a 15 mL glass tube (Thermo Fisher Scientific) and treated with 4.75 mL chloroform/methanol (2:1, v/v). 1.25 mL chloroform and 1.25 mL double-distilled water were added to facilitate phase separation and then the samples were vortexed for 2 min and centrifuged at 4,000 rpm for 15 min. The lower phase of the solution was vaporised under a stream of nitrogen. The extracted lipid was then stored at −20°C until further use. Prior to use, the lipid fraction was immediately dissolved in 30 L chloroform/methanol (2:1, v/v). 80    2.16 Thin-layer chromatography A silica gel-coated TLC plate (20 cm × 20 cm, no.1.05721.0001, Merck) was pre-dried by heating at 37 °C for 2 hr and then allowed to cool down gradually to room temperature. Typically, 30 L samples were pipetted onto a TLC plate using a Hirschmann micropipette (Z611247- 250EA, Sigma-Aldrich) with an interval of 2 cm between sample spots. After drying with a hair- dryer for 1 min, a chromatography tank (20 cm × 20 cm) was used to accommodate individual TLC plates throughout these studies. The TLC plate can be developed in a chloroform/methanol/water solvent system (65:25:4, v/v/v). After the solvent travelled up the plate until ~1 cm from the top, the plate was taken out from the solvent, air-dried for 30 sec, and then sprayed with 0.005% primuline in acetone/water (4:1, v/v). The pattern of the separated spots on the gel was scanned digitally with the BioRad Image Lab Software. The individual phospholipid intensity values were quantified using the Bio-Rad image analysis software. 2.17 RNA Isolation, reverse transcription and RT-PCR 2.17.1 RNA isolation and purification Cells grown in 12-well plates were initially lysed by adding 350 L RLT buffer (Qiagen) supplemented with 1.0% -mercaptoethanol (Sigma-Aldrich). RNA was purified using the Qiagen RNeasy kit following the manufacturer’s instructions. RNA concentration was determined using Nanodrop 1000 Spectrophotometer (Thermo Fisher Scientific). The absorbance ratio at 260/280 nm was used to assess RNA purity, with value of approximate 2.0 deemed acceptable. 2.17.2 Reverse transcription polymerisation chain reaction For each reaction, 500 ng total RNA extracted from cells was gently mixed with the RT Master Mix (2.0 L of 10x RT Buffer, 0.8 L of 100 mM dNTP Mix, 2.0 L of 10x RT Random Primers, 1.0 L of RNase Inhibitor, 1.0 uL of 50 U/uL MultiScribe Reverse Transcriptase, and 2.2 L of nuclease-free water, Life Technologies). Reverse transcription was carried out using a programmed thermal cycler as follows: 30 cycles of 25°C for 10 min, 37°C for 120 min and 85°C for 5 min, with a final 4°C hold step. 81    2.17.3 Quantitative-PCR cDNA samples from reverse transcription reactions described in section 2.17.2 were pooled to derive a standard control cDNA. Firstly, a cDNA dilution series 1:4, 1:16, 1:64, 1:256 and 1:1,024 was made. The primers and probes for PCYT1A and Glyceraldehyde-3-Phosphate Dehydrogenase (GAPDH) were purchased as a pre-mixed 20x stock from Applied Biosystems. For each qPCR reaction, 0.25 L of 20x primer/probe, 2.5 L of TaqMan 2x PCR Master Mix (Applied Biosystmes), 2.25 L of nuclease-free water and 2 L of diluted cDNA were mixed to a final 7 L volume. qPCR was performed on in a MicroAmp Optical 384-Well Reaction Plate (Applied Biosystems) using an Applied Biosystems 7900A Real Time PCR system. The cycling program was performed as follows: 40 cycles of 50°C for 2 min, 95°C for 10 min and 60°C for 1 min, with a final 4°C hold step. The annealing time was increased to 15 sec per cycle. The expression level of PCYT1A was normalised to the GAPDH mean expression level used as a reference housekeeping gene. 2.18 Statistical analysis Quantitative data are reported as mean ± SD. As indicated in the figure legends, differences between means were assessed using two-tailed Student's t tests or one-way ANOVA with a Bonferroni multiple comparisons test using GraphPad Prism software. Statistical significance was defined as p < 0.05.  82    Chapter 3 The Yeast PCYT1A Homolog, Pct1, Modulating Phosphatidylcholine Homeostasis by a Membrane Dependent on/off Switch Mechanism 3.1 Abstract The most abundant phospholipid in eukaryotic membranes is phophatidylcholine (PC), yet how cells detect and regulate its level in vivo remains poorly understood. Here I have assessed the expression and membrane targeting of yeast Pct1. This conserved enzyme, which defines the rate of PC synthesis, is intranuclear and re-locates to the inner nuclear membrane in response to the need for cell growth and PL synthesis. Extensive mutagenesis analysis, lipidomics and coarse grain modelling suggest that yeast Pct1 is associated with membranes using an amphipaphic helix that senses the stored curvature elastic stress of lipid bilayers. We conclude that nucleoplasmic Pct1 is activated in response to physical phospholipid composition changes in order to control PC homeostasis according to the physiological needs of the host cell. 3.2 Introduction 3.2.1 Evolutionary conservation of lipid metabolism pathways There are two conserved pathways of PC synthesis in eukaryotic cells: the Kennedy pathway and the PE methylation pathway (Figure 3.1). The yeast Saccharomyces cerevisiae provides a tractable model system that can be used to explore the CCT regulatory mechanism in detail, as the pathways have been particularly well characterised and, although the PE methylation pathway is the dominant route for PC synthesis in yeast (Chin and Bloch, 1988), genetic manipulation can circumvent this problem. As in mammalian cells, choline is an essential nutrient that primarily needs to be taken up from the environment to initiate the Kennedy pathway. Once choline has entered the cell, it is phosphorylated by choline kinase (Cki1) to form phosphocholine in the first step. The second step is catalysed by cholinephosphate cytidylyltransferase (Pct1) in the rate- limiting step of the pathway; in this step, cytidine triphosphate (CTP) is added and inorganic pyrophosphate (PPi) is eliminated to create CDP-choline. In the final stage of the reaction, diacylglycerol (DAG) is incorporated by transfer of a phosphobase from CDP-choline to form PC, with the generation of cytidine monophosphate (CMP) (Nikoloff and Henry, 1991). The 83    alternative PE methylation pathway is important in yeast cells. PE is converted to PC by three- sequential S-adenosyl methionine (AdoMet) methylation reactions, whereby the first step is catalysed by Cho2, and the following two steps are catalysed by Opi3. Both Cho2 and Opi3 are PE methyltransferases and are localised on the endoplasmic reticulum (ER) (Tavassoli et al., 2013; Zinser et al., 1991). Biosynthesis of PE in yeast is also similar to that in mammals. PE is generated through two pathways: the CDP ethanolamine-Kennedy pathway and the CDP-DAG pathway (Figure 3.1). The analogous enzymes of the CDP-ethanolamine branch of the Kennedy pathway include, ethanolamine kinase (Eki1), ethanolamine-phosphate cytidylyltransferase (Ect1) and sn-1,2- diacylglycerol ethanolamine- and cholinephosphotranferase (Ept1). Extracellular ethanolamine is the preferred substrate of Eki1 for which the specificity constant (Vmax/ Km) is 2.2-fold higher than that for choline (Kim et al., 1999). As a second possibility, ER-localised phosphatidylserine synthase (Cho1) may catalyse the formation of PS by displacing CMP from CDP-DAG with serine (Letts et al., 1983; Natter et al., 2005). PE can then be produced in the mitochondrial membrane by decarboxylation of PS via PS decarboxylase 1 (Psd1) (Clancey et al., 1993; Zinser et al., 1991) or in the Golgi and the vacuoles through  PS decarboxylase 2 (Psd2) (Trotter and Voelker, 1995). 84    Figure 3.1 Pathways for the synthesis of PC and PE in S. cerevisiae. The pathways shown for the synthesis of PC and PE include the relevant steps discussed in the text. Cds1,  CDP-diglyceride synthetase; Cho1, phosphatidylserine synthase; Psd1/2, phosphatidylserine decarboxylase 1/2; CDP-DAG, CDP-diacylglycerol; MMPE, monomethylphosphatidylethanolamine; DMPE, dimethylphosphatidylethanolamine; AdoMet, S- adenosyl methionine; AdoHCys, S-adenosyl homocysteine; Cho2, phosphatidylethanolamine methyltransferase; Opi3, phosphatidyl-N-methylethanolamine methyltransferase; Cki1, choline kinase; Pct1, cholinephosphate cytidylyltransferase; Cpt1, sn-1,2-diacylglycerol cholinephosphotransferase; Eki1, ethanolamine kinase; Ect1, ethanolamine-phosphate cytidylyltransferase; Ept1, sn-1,2-diacylglycerol ethanolamine- and cholinephosphotranferase. 3.2.2 Membrane curvature in cell biology: integration of molecular mechanisms It has been known for more than 20 years that stored curvature elastic stress (SCE) is linked to topological constraints, in particular the involvement of non-bilayer lipids in lipid bilayers, and with increased mismatch between the actual curvature of the membrane and the lipid geometry (Gruner, 1985). It is also known that some proteins with an AH (amphipathic helix) can be inserted into the membrane interface, which allows the chains of nearby lipids to splay and 85    relieve SCE stress. Notably, PCYT1A can alleviate SCE stress in vitro while SCE stress in the lipid membrane simultaneously modulates the activity of PCYT1A (Attard et al., 2000; Davies et al., 2001). This represents a shift from inactive PCYT1A to an active PCYT1A membrane-bound form in response to PCYT1A activators like DAG and PE, which causes SCE stress in membranes and introduces gaps in the packing of polar head groups. It is apparent, therefore, that at least in vitro, under conditions of sufficient PC in a lipid bilayer, PCYT1A becomes inactive and soluble. If PCYT1A can relieve SCE stress, other AHs relax membrane curvature and, in this way, stabilise the high curvature of the vesicles. Two examples are given by ArfGAP (Bigay et al., 2005), a GTPase activating protein for Arf, and α-synuclein (Kamp and Beyer, 2006), a protein whose folding disorder is linked to neurodegenerative diseases such as Parkinson’s. Both ArfGAP and α-synuclein bind selectively to highly curved membrane vesicles, and display affinity for packing defects in a bilayer membrane. By being inserted into surface defect sites, AH motifs can alleviate membrane SCE stress in vesicles with a small radius, such as synaptic vesicles and vesicles involved in inter-organelle traffic, and lead to a more stably-packed surface (Cornell and Taneva, 2006). In other cases, proteins with AHs can actually induce membrane curvature and are involved in vesicle formation in this manner. For instance, epsin is an adaptor of clathrin-mediated endocytosis (CME) (Hawryluk et al., 2006); in other words, epsin is indispensable for the formation of clathrin-coated vesicles on the plasma membrane (Chen et al., 1998; Messa et al., 2014). Their functions are attributed to a membrane-active ENTH domain (Epsin N-terminal Homology), which displays high affinity for phosphatidylinositol 4,5-bisphosphate (PIP2) (Yoon et al., 2010). PIP2-binding coordinates the folding of the first helix of the ENTH domain into an amphipathic helix, which exhibits a hydrophobic face and so is partially inserted into the inner membrane (Kweon et al., 2006), provoking the lipid membrane to bend towards the cytoplasm facilitating vesicle formation (Messa et al., 2014). This suggests a pivotal role for epsin in driving membrane curvature and CME. Another example is given by endophilinA1 and endophilinB1. These are both implicated in synaptic vesicle endocytosis through their BAR (Bin/Amphiphysin/Rvs) domain interacting peripherally with the surface of a curved bilayer, and then directly interacting with the membrane via their N-terminal AH (Jao et al., 2010). Moreover, 86    endophilin is sufficient to deform liposomes into narrow tubules of 20–100 nm diameter (Farsad et al., 2001). 3.3 Aims Membrane association rapidly facilitates CCT catalytic activity. Structural studies suggest a model whereby an unstructured loop folds into a helix resulting in the removal of a nearby helix which would otherwise prevent substrate from accessing the catalytic pocket of the dimeric enzyme (Huang et al., 2013). In vitro studies have shown that membrane association of CCT is induced by conically shaped lipids such as DAG or PE, or by negatively-charged phospholipids such as PA or PS (Arnold and Cornell, 1996; Cornell, 2016). This supports the hypothesis that CCT would sense a relative paucity of PC, resulting in its activation and alleviation of the membrane stress evoked by conically shaped lipids. Although the enzymology of CCT and the biochemical pathways that generate PC have been reported, how exactly cells detect the levels of PC within their membranes to maintain homeostasis in vivo remains unclear. The objectives of the work presented in this chapter were: I. To establish an in vivo model system to monitor the cellular localisation and activity of Pct1. For this, we took advantage of the genetic tractability of yeast cells. To block selectively either of the two pathways known to be involved in PC biosynthesis, the yeast genome was manipulated by homologous recombination to delete specific enzymes. Also, heterologous expression of fluorescently tagged versions of the wild-type or mutant versions of the enzyme Pct1 in yeast strains was carried out for cellular studies. II. To understand how cells sense and regulate PC levels in vivo. The role of Pct1 in PC-sensing has been elucidated through time-course studies. To aid in this investigation, whole cell lipidomic analyses of yeast cells expressing wild-type or mutant Pct1 was carried out at various growth stages to gauge relationships between Pct1 cellular localisation, activation and membrane lipid composition. Data driven modelling of the lipidomic data thus obtained was carried out to generate a possible model of the PC-sensing and regulating mechanism of Pct1/PCYT1A. 87    3.4 Results 3.4.1 Saccharomyces cerevisiae as a suitable model organism for studying the on/off switch mechanism of PCYT1A PCYT1A (CCT) is largely an intra-nuclear enzyme in most mammalian cells and is thought to be partitioned between a soluble inactive state and a membrane-bound activated state in this compartment (Huang et al., 2013; Lee et al., 2014). To understand how CCT nuclear membrane localisation and consequent activation are triggered at the molecular level in vivo, we used the yeast S. cerevisiae as a model system. As with higher eukaryotes, GFP-tagged Pct1, the yeast orthologue of PCYT1A, expressed from its endogenous promoter, localises to the nuclear envelope (NE) in rapidly proliferating cells in the exponential (Exp) phase, but ‘falls off’ the membrane during the post diauxic shift (PDS) phase, when glucose and other nutrients are exhausted (Figure 3.2A, (Barbosa et al., 2015)). A transient glucose starvation in the exponentially growing cells (Exp/GS) was also used to mimic the conditions of the PDS phase. When cells were deprived of glucose for 0.5 hr, Pct1 became nucleoplasmically localised (Figure 3.2A, (Barbosa et al., 2015)). The lipid droplet (LD) sizes were significantly enlarged in the PDS phase (Figure 3.2B), and this phenomenon suggests that a nutrient-sensing signal releases Pct1 from the nuclear membrane and deactivates Pct1, and then ensures that lipid intermediates are diverted from phospholipid synthesis into storage lipids. Interestingly, Pct1-GFP is released from the nuclear membrane into the nucleoplasm on choline supplementation (Exp + Cho), which increases the contribution of the Kennedy pathway for PC synthesis (Figure 3.2A). To confirm Pct1 on/off membrane translocation further, an mCherry-tagged Pct1 was generated, and similar experiments were conducted to detect the localisation of Pct1 (Figure 3.2D). The expression of Pct1 was checked by immunoblotting using either anti-GFP or anti-mCherry antibodies (Figure 3.2C). Pct1 nuclear envelope localisation could be acutely altered and visualised in both cases (Figures 3.2A and D), therefore, yeast appears to be an appropriate model organism for studying the membrane-binding mechanism of PCYT1A protein. 88    89    Figure 3.2 Binding of Pct1 to the nuclear membrane is associated with nutrient levels in S. cerevisiae. (A) pct1∆ cells expressing C-terminal GFP-tagged Pct1 were grown to exponential phase (Exp), exponential phase transferred to medium lacking glucose for 0.5 hr (Exp/GS), exponential phase supplemented with 1 mM choline (Exp + Cho) or post-diauxic shift phase (PDS), stained to visualize lipid droplets (LDs) and imaged as described in Methods section 2.5.1. Differential interference contrast (DIC) images are shown for context. Scale bar, 5 m. Images are representative of at least three independent experiments. (B) Measurement of LDs area. Data are mean ± SD from three independent experiments, with ~700 LDs counted per condition. One-way ANOVA with Bonferroni multiple comparison, **p < 0.01. (C) (i) Immunoblotting of lysates from pct1∆ cells expressing Pct1-GFP using GFP antibody; GAPDH was used as a loading control. (ii) Immunoblotting of lysates from pct1∆ cells expressing Pct1-mCherry using mCherry antibody; GAPDH was used as a loading control. (D) pct1∆ cells expressing C-terminal mCherry-tagged Pct1 were grown to exponential phase (Exp), exponential phase transferred to medium lacking glucose for 0.5 hr (Exp/GS), exponential phase supplemented with 1 mM choline (Exp + Cho) or post-diauxic shift phase (PDS). Images are representative of at least three independent experiments. Scale bar, 5 m. 3.4.2 Pct1 localisation to the nucleus and mobility within the nucleoplasm Considering that mammalian cells express a cytoplasmic isoform, PCYT1B, and PCYT1A was reportedly found on cytoplasmic LDs in some cells of Drosophila and mice (Krahmer et al., 2011), we next investigated whether Pct1 shuttles between the nucleus and cytoplasm. Pct1 might also transiently exit the nucleus and associate with other membranes. To test this possibility, we performed fluorescence loss in photobleaching (FLIP) and analyses with repeated bleaching of either an intranuclear or cytoplasmic region of yeast pct1∆ cells expressing either GFP as a control or Pct1-GFP. Repeated photobleaching of GFP in the nucleus of cells expressing GFP alone led to a prompt decrease in GFP signal in the cytosol, and a subsequent partial recovery of the GFP signal in the photobleached region within approximately 100 sec (Figure 3.3A). Photobleaching of a similar-sized region of the cytoplasm led to a more gradual reduction in the nuclear GFP signal (Figure 3.3B). In contrast to these data which suggest that GFP shuttles between the nucleus and cytoplasm, in cells expressing Pct1-GFP, repeated photobleaching of the cytosol did not alter the nuclear GFP signal (Figure 3.3C). We conclude that wild type Pct1 remains inside the nucleus in the conditions tested. In addition, fluorescence recovery after photobleaching (FRAP) experiments were performed in pct1∆ Pct1-GFP cells, in which a part of the nuclear signal was bleached and the signal recovery was measured. The Pct1-GFP signal was 90    recovered partially (Figure 3.3D), indicating that Pct1 is confined to the nucleus but is mobile within it. Figure 3.3 Pct1 is confined to the nucleus in yeast cells. (A and B) In wild-type (BY4742) yeast cells expressing GFP alone, repeated photobleaching of GFP in the nucleus (A) or cytosol (B) is rapidly followed by partial recovery of the fluorescence in the photobleached regions, with a concomitant progressive loss of fluorescence in the unbleached regions, indicating rapid movement of GFP between these two cellular compartments. (C) In pct1∆ cells expressing wild-type (WT) Pct1-GFP, repeated photobleaching of the signal in the cytosol does not result in a loss of fluorescence of WT Pct1-GFP within the nucleus, indicating little or no rapid shuttling of Pct1-GFP from the nucleus to the cytosol. (D) In pct1∆ cells expressing WT Pct1-GFP, repeated photobleaching of GFP in part of the nucleus is followed by partial recovery of the fluorescence in the photobleached regions with no change of nuclear fluorescence in other cells, indicating that Pct1 is mobile inside the nucleus. Data are mean ± SD from three independent experiments (5-8 cells each). Arrows on x axes indicate each bleaching event. The GFP fluorescence reported here were corrected for the loss of fluorescence during imaging as described in Methods section 2.5.3. 91    3.4.3 Pct1 localisation on the inner nuclear membrane in yeast Saccharomyces cerevisiae The inner and outer nuclear membranes (INM and ONM, respectively) of the nuclear envelope are enriched in distinct membrane proteins in eukaryotic cells (Katta et al., 2014), which play roles in maintaining nuclear architecture and genome stability. In the exponential phase, the specific perinuclear Pct1-GFP signal resembles the localisation of yeast INM proteins that are not a part of the peripheral ER network (Figures 3.2A and D, (Laba et al., 2014) ). However, we cannot rule out the possibility that Pct1 is associated with the ONM. It is worth noting that the membranes of ER are continuous with the ONM and the final step in the Kennedy pathway is catalysed by the ER, Golgi or mitochondria localised Cpt1, and the ER or NE localised Ept1 (Henneberry et al., 2002). Thus, the assembly of sequential pathway enzymes into the spatially confined subcellular localisation may offer advantages such as increasing catalytic efficiency. To assess whether or not Pct1-GFP is resident on the INM or ONM, we used the anchor-away method (Figure 3.4A, (Haruki et al., 2008)) to sequester a key -importin Kap60, which is essential for nuclear import and had previously been shown to bind Pct1 (MacKinnon et al., 2009), and investigated if this affects the accumulation of the intranuclear Pct1-GFP pool. A construct expressing the nuclear localisation signal of SV-40 large T antigen tagged with GFP (NLS-GFP) was used to make sure that the system is functionally operational. Figure 3.4B shows the NLS-GFP dominantly localised inside the nucleus due to normal import. Once rapamycin was added, the anchor-away system blocks the nuclear import causing NLS-GFP to become more cytoplasmic. Under these conditions, Pct1-GFP still shows a soluble intranuclear localisation when cells are starved of glucose, which promotes Pct1-GFP release from the membrane, whereas if it had been bound to the cytoplasmic face of the nuclear envelope it would be expected to become cytoplasmic. These data suggest that the site of Pct1 membrane association and activation is the inner nuclear membrane. Nevertheless, the anchor-away technique cannot entirely preclude  the possibility that lower levels of Pct1 localises to the ONM since a pool of Pct1 can still be detected in the cytoplasm. 92    Figure 3.4 Pct1 associates with inner nuclear membrane. (A) Schematic of the anchor-away technique (adapted from Haruki et al., 2008). Rapamycin promotes tethering of the karyopherin Kap60 (Kap60-FRB) to the ribosomal protein Rpl13A 93    (Rpl13A-FKBP12), due to heterodimerization of FKBP12 and FRB, causing depletion of Kap60 from the nucleus and would block the nuclear import of target proteins. (B) Anchor-away strain expressing Rpl13A-FKBP12, Kap60-FRB, Nup84-mCherry (nuclear membrane marker) and Pct1-GFP was grown to the exponential phase (Exp) and treated with rapamycin (+Rap) for 15 min. Glucose starvation (+Rap/GS) was induced after the initial rapamycin treatment by transferring the cells into rapamycin-medium lacking glucose for 0.5 hr. To confirm that depletion of Kap60 prevents nuclear import, GFP fused to a known nuclear localisation signal (NLS-GFP) was used as a positive control. NLS-GFP concentrates in the nucleus in untreated cells but it is increasingly redistributed to the cytosol following Kap60 inactivation demonstrating that nuclear import is successfully blocked. Differential interference contrast (DIC) images are shown for context. Cells were imaged as described in the Methods section 2.5.1. Images are representative of at least three independent experiments. Scale bar, 5 m. 3.4.4 Glucose depletion inducing dissociation of Pct1 from the nuclear membrane independently of the glucose signalling machinery Release of Pct1 from the nuclear membrane takes place during glucose deprivation (Exp/GS; Figures 3.2A and D). Given that this could be part of a general stress response that might allow yeast cells to coordinate physiological processes for optimal adaptation to changing environments, the possible impact of several additional stress factors that might affect Pct1 membrane translocation was also studied (Table 3.1, Figures 3.5.1A to E). Stresses used in this study included G1 phase cell cycle arrest, nutrient depletion, heat shock, agents inducing cell wall damage, oxidative stress, ER stress and hyperosmolarity. Of the various stress factors assessed, only glucose-starvation (GS) led to the release of Pct1 from the nuclear envelope (Table 3.1, Figure 3.5.1B). This suggests that glucose is a specific factor regulating Pct1 membrane association. 94    Table 3.1 Localisation of Pct1-GFP under physical or chemical stress Type of stress Stress condition Duration Pct1-GFP G1 phase cell cycle arrest -factor mating pheromone (10 mM) 0.5 hr Nuclear envelope 1.0 hr Nuclear envelope 2.0 hr Nuclear envelope Nutrient (glucose) depletion Glucose starvation 0.5 hr Nucleus 1.0 hr Nucleus >> Nuclear envelope Glucose starvation + 2-deoxy glucose (2 mg/mL) 0.5 hr Nucleus >> Nuclear envelope 1.0 hr Nucleus >> Nuclear envelope Cell wall damage SDS (0.02 %) 0.5 hr Fluorescence signal averaged over the whole cell (cell death) 1.0 hr Fluorescence signal averaged over the whole cell (cell death) 2.0 hr Fluorescence signal averaged over the whole cell (cell death) Oxidative stress H2O2 (1.5 mM) 0.5 hr Nuclear envelope 1.0 hr Nuclear envelope 2.0 hr Nuclear envelope ER stress (UPR activation) DTT (10 mM) 0.5 hr Nuclear envelope 1.0 hr Nuclear envelope 2.0 hr† Nucleus > Nuclear envelope Hyperosmolarity Sorbitol (1 M) 0.5 hr Nuclear envelope 1.0 hr Nuclear envelope 2.0 hr Nuclear envelope 95    Heat shock 37°C 0.5 hr Nuclear envelope 1.0 hr Nuclear envelope 2.0 hr Nuclear envelope Nutrition limitation (mTOR inhibition) Rapamycin (1 mg/mL) 15 min Nuclear envelope 1.0 hr Nuclear envelope 2.0 hr Nuclear envelope Nutrient (nitrogen) depletion Nitrogen starvation‡ 0 min Nuclear envelope / Atg8-mCherry whole- cell distribution 15 min Nuclear envelope/ - 30 min Nuclear envelope / Atg8-mCherry accumulates in the vacuole 60 min Nuclear envelope/ - 90 min Nuclear envelope / Atg8-mCherry accumulates in the vacuole 120 min Nuclear envelope/ - 180 min Nuclear envelope / Atg8-mCherry accumulates in the vacuole † Cell death observed. Pct1-GFP might be degraded within the nucleus. ‡Autophagy-related ubiquitin-like protein (Atg8) is known to be subject to strong expression after nitrogen starvation and is a marker of autophagy in yeast (Devenish and Prescott, 2015). The fluorescent signal releases and accumulates in the vacuole as autophagy proceeds (Nair et al., 2011). 96      Figure 3.5.1A and B Evaluation of stress factors and Pct1 translocation. (A) Wild-type yeast cells (BY4741) expressing Pct1-GFP and Sec63-mCherry (ER marker) were grown to exponential phase (Exp), post-diauxic shift phase (PDS) or to exponential phase followed by addition of -factor mating pheromone. The -factor pheromone arrests yeast in the G1 phase of their cell cycle (Naider and Becker, 2004). Time points after addition of -factor are indicated. Cells were collected at the indicated time points after -factor addition for imaging. Scale bar, 5 μm. (B)  Wild-type yeast cells (RS453) expressing Pct1-GFP and Sec63-mCherry (ER marker) were grown to exponential phase (Exp) or post-diauxic shift phase (PDS). Glucose 97    starvation (Exp/GS) was induced by transferring the exponentially growing cells into medium lacking glucose for 0.5 hr or 1.0 hr. Pct1 can be found in the nucleoplasm during glucose starvation. To confirm that depletion of glucose catabolism facilitates Pct1 translocation, glucose analogue 2-deoxy glucose was added to cells after glucose starvation. The main reason for this is the fact that 2-deoxy glucose is not metabolized but can mimic glucose in glucose signalling.  Time points after addition of 2-deoxy glucose are indicated. Cells were then collected at the indicated time point for imaging. The addition of 2-deoxy glucose to cells causes a glucose starvation-like response and inhibiting Pct1 relocation onto nuclear membrane. Cells take up 2- deoxy glucose and convert it to 2-deoxyglucose-6-phosphate; however, the absence of a hydroxyl group on C-2 prevents the further glucose catabolism (Jaspers and van Steveninck, 1975). Scale bar, 5 μm. Images are representative of at least three independent experiments.       98      Figure 3.5.1C Evaluation of stress factors and Pct1 translocation. (C) Wild-type yeast cells (BY4741) expressing Pct1-GFP and Sec63-mCherry (ER marker) were grown to exponential phase (Exp), post-diauxic shift phase (PDS) or to exponential phase followed by addition of the indicated stress factors. Time points after addition of the stress factors are indicated. Cells were collected at the indicated time points after respective stress factor 99    addition for imaging. Images are representative of three independent experiments. Scale bar, 5 μm.   100    Figure 3.5.1D and E Evaluation of stress factors and Pct1 translocation. (D) Wild-type yeast cells (RS453) expressing Pct1-GFP and Sec63-mCherry (ER marker) were grown to exponential phase (Exp), post-diauxic shift phase (PDS) or to exponential phase followed by addition of rapamycin. Rapamycin is an inducer of autophagy, as inhibition of TOR (target of rapamycin) protein mimics cellular starvation by blocking signals required for cell growth and proliferation (Fleischmann and Rocha, 2018). Cells were collected at the indicated time points after rapamycin addition for imaging. Scale bar, 5 μm. (E) (i) Nitrogen starvation was induced by transferring the exponentially growing cells into nitrogen-free medium for the indicated time. Cells were collected prior to (0.0 min) and at the indicated time points after nitrogen starvation for imaging. Scale bar, 5 μm. (ii) Wild-type yeast cells (RS453) expressing Pct1-GFP and Atg8-mCherry were grown to exponential phase followed by addition of rapamycin. Atg8 was used as a marker for nitrogen starvation (autophagy). Cells were collected at the indicated time points for imaging. Atg8 distributes to the cytosol in unstressed cells but it concentrated in the vacuoles following nitrogen starvation indicating that autophagy is successfully induced. Scale bar, 5 μm. Images are representative of three independent experiments. 101    Glucose is an essential primary messenger molecule for growth, fermentation capacity and stress resistance; therefore, several glucose-sensing and -signalling pathways are present in yeast (Rolland et al., 2002). The glucose-concentration-dependent pathways can be divided into glucose-repression pathways and glucose-induction pathways (Figure 3.5.2). The main glucose- repression pathway is regulated by its central components: the Snf1-protein kinase, the Mig1- Hxk2 transcriptional repressor complex and the Reg1 phosphatase 1 (Figure 3.5.2A). Snf1 kinase plays a central role in glucose metabolism and is required for derepression of many glucose- repressed genes, which are expressed only upon glucose starvation (Trumbly, 1992). Snf1 regulates the phosphorylation of Mig1, which is a DNA-binding protein (Smith et al., 1999). Hxk2 interacts with Mig1 to generate a repressor complex. In the presence of high levels of glucose, the Mig1-Hxk2 complex rapidly moves into the nucleus, where it binds to the promoters of glucose-repressible genes and exerts repression of these genes (Figure 3.5.2A(ii)). When the cells are deprived of glucose, the complex is rapidly transported back to the cytoplasm (Ahuatzi et al., 2007). Reg1 is phosphorylated by Snf1 in response to glucose limitation and dephosphorylated/activated by Glc7 when glucose is present. Activated Reg1 subsequently dephosphorylates/inhibits Snf1 during growth in the presence of glucose (Rolland et al., 2002) (Figure 3.5.2A(i)). The dominant glucose-induction pathways can be recapitulated by the activation of Ras-cAMP- PKA (Tpk) signalling (Figure 3.5.2B)  or by the induction of Sch9 (Figure 3.5.2C). The Ras- cAMP-PKA (Tpk) pathway is a major glucose-induced pathway involved in G-protein-coupled receptor signalling and cAMP synthesis (Cullen and Sprague, 2012; Santangelo, 2006). Glucose triggers the synthesis of cAMP from ATP through the enzyme activity of adenylate cyclase Cyr1. cAMP activates cAMP-dependent protein kinase A (Tpk in yeast) by binding to its regulatory subunits, thus activating the catalytic protein kinase subunits (encoded by TPK1, TPK2 and TPK3) and resulting in the activation of downstream target genes. Most of the downstream identified targets of Tpk are genes involved in growth and proliferation such as ribosomal protein genes, or in carbon metabolism and the stress response. The Sch9 signalling pathway may function in parallel with the PKA pathway.  Sch9 is a regulator of ribosomal biogenesis through the control of transcription (Lempiainen et al., 2009; Mirisola et al., 2014). 102    Figure 3.5.2 Glucose sensing- or -signalling pathways in S. cerevisiae. 103    (A) (i) Scheme of the Snf1-glucose signalling pathway (Conrad et al., 2014; Ronen and Botstein, 2006; Santangelo, 2006). Glucose is transported into the cell by Hxt transporters and converted to glucose-6-phosphate (G-6-P). When yeast grows in high glucose conditions, Reg1-Glc7 dephosphorylates Snf1 thereby deactivating Snf1 activity in the cytoplasm. Upon glucose exhaustion, Snf1 is phosphorylated in its active form. The phosphorylated Snf1 binds to Snf4. When the active complex Snf1-Snf4 is linked with Gal83, it enters the nucleus to trigger derepression of gluconeogenesis genes and non-fermentable carbon metabolic genes. Also, active Snf1 causes the translocation of the Mig1 and Hxk2 from the nucleus to the cytoplasm. In the nucleus, Mig1 and Hxk2 are repressors inhibiting the expression of genes controlled by glucose. (ii) Once glucose inhibition by Snf1 is released (under high-glucose conditions), both Mig1 and Hxk2 are bound to the promoter of glucose-related genes in the nucleus. Mig1 and Hxk2 repress transcription of their target genes, such as gluconeogenesis genes and respiratory genes (Gancedo and Flores, 2008; Ronen and Botstein, 2006). (B) Scheme of the Ras-cAMP-PKA (Tpk) pathway (Cullen and Sprague, 2012; Santangelo, 2006). Adenylate cyclase (Cyr1) in yeast is controlled by two G-protein coupled systems. Adenylate cyclase is activated by a glucose-sensing G-protein coupled receptor (GPCR) system, composed of the Gpr1 receptor and, the Gα protein Gpa2. Adenylate cyclase is also activated by the Ras1 proteins, which are controlled by the Cdc25. In response to glucose addition, either activated GTP-bound Ras proteins or activated GTP-bound Gpa2 can activate adenylyl cyclase to produce cAMP, which is a secondary messenger. cAMP binds to the regulatory subunit of Tpk, which normally suppresses the activation of the catalytic subunits of Tpk. Released and activated Tpks activate diverse target genes, for instance, ribosome biogenesis genes, stationary phase genes and stress response genes. (C) Simplified scheme of the TORC1-Sch9 signalling pathway. In the presence of sufficient nutrients, Sch9 is regulated by TORC1 and Pkh kinases via phosphorylation of two threonines and promote intracellular localisation in parallel with Sfp1. Within the nucleus, Sch9 and Sfp1 regulate transcriptional activation of ribosomal protein (RP) and ribosomal biogenesis (Ribi) gene (Lempiainen et al., 2009; Mirisola et al., 2014). 104    To probe the contributions of the glucose-responsive pathway to Pct1 translocation, Pct1-GFP firstly was expressed in five different mutant yeast strains, each lacking a glucose-sensing gene like snf1∆, mig1∆, hxk2∆, reg1∆ or sch9∆ (Table 3.2). These glucose-sensing mutants can be divided into two broad groups: one corresponds to mutant that fails to elicit a proper metabolic rearrangement in the PDS phase or during glucose starvation (Exp/GS), such as snf1∆; others are mutants that should mimic the rearrangement even during glucose-rich conditions, such as mig1∆, hxk2∆, reg1∆ and sch9∆. To test whether Pct1-GFP would stay on the membrane in the snf1∆ mutant at the PDS phase or glucose starvation (Exp/GS) state, we documented the localisation of Pct1-GFP in snf1∆ mutant by using fluorescence microscopy (Figure 3.5.3A and B). To test if Pct1-GFP was released from the membrane in the exponential phase in the mig1∆, hxk2∆, reg1∆ and sch9∆ mutants, these mutant cells were screened for Pct1 membrane on/off shifts by using fluorescence microscopy (Figure 3.5.3A). However, there were no apparent changes in Pct1 localisation compared to the wild-type strain, except for the snf1∆ strain. Here, snf1∆ cells expressing Pct1-GFP revealed that, while cell growth did not seem to be obviously affected by the absence of Snf1, the cells needed more time to release Pct1 from the membrane, when compared to the wild-type strain in the PDS phase and in the Exp/GS state (Figure 3.5.3B). This is the first evidence that Snf1, which regulates expression of glucose-repressed genes, could be related to Pct1 translocation, and that this could be a previously undescribed mechanism for responding to changes in glucose concentrations. 105    Table 3.2 Localisation of Pct1-GFP in wild type strains or mutant strains with a single glucose-related gene deletion Strain Growth phase Exponential Post-diauxic Exponential/ GS 30min Wild-type BY4741 Nuclear envelope Nucleus Nucleus Wild-type BY4742 Nuclear envelope Nucleus Nucleus snf1∆† Nuclear envelope Nucleus‡ Nuclear envelope > Nucleus mig1∆ Nuclear envelope Nucleus Nucleus hxk2∆ Nuclear envelope Nucleus Nucleus reg1∆ Nuclear envelope Nucleus Nucleus sch9∆ Nuclear envelope Nucleus Nucleus †snf1∆ shows Nuclear envelope > Nucleus in glucose deprivation medium for 30 min and 60 min. ‡snf1∆ OD600 = 2.20 Nuclear envelope; OD600 = 5.14 Nucleus. 106    Figure 3.5.3 Loss of Snf1 diminishes the Pct1 response. (A) Wild-type yeast cells (BY4741 or BY4742) or mutant cells with single glucose-signalling gene deletions as indicated express Pct1-GFP and Sec63-mCherry (ER marker). Cells were grown to exponential phase (Exp), post-diauxic shift phase (PDS) or to exponential phase followed by glucose starvation 30 min (Exp/ GS 30min). Cells were collected at the indicated time points for imaging. Only strains containing a deletion of SNF1 were unable to redistribute Pct1 away from the nuclear membrane, suggesting that the role of Snf1 in glucose signalling 107    controls the association of Pct1 with membranes. Scale bar, 5 μm. (B) The snf1∆ strain expressing Pct1-GFP and Sec63-mCherry  was grown to exponential phase followed by glucose starvation 30 min (Exp/ GS 30min), exponential phase followed by glucose starvation 60 min (Exp/ GS 60min), early post-diauxic shift phase (early PDS), or later post-diauxic shift phase (PDS). Cells were collected at the indicated time points for imaging. Both glucose starvation and early PDS are unable to shift Pct1 into the nucleoplasm  completely, suggesting that the SNF1 gene deletion can delay the cellular response to the absence of glucose. Scale bar, 5 μm. Images are representative of three independent experiments. To begin to evaluate the possible role of PKA signalling in Pct1 translocation, we first used a strain (Y3527) with mutations in the PKA encoded subunits Tpk1, Tpk2, and Tpk3 (Yorimitsu et al., 2007). Y3527 carries mutations in the ATP-binding pocket so it is sensitive to the cell- permeable ATP-analogue inhibitor C3-1’-naphthyl-methyl PP1 (1NM-PP1), which inhibits only the modified Tpk subunits. Msn2 was used as a reporter of PKA inactivation. Upon 1NM-PP1 treatment, the stress-responsive transcriptional activator Msn2 migrates from the cytoplasm to the nucleus where it promotes transcription of several stress-response genes in response to PKA inactivation, however no changes in Pct1 localisation were observed (Table 3.3, Figure 3.5.4). These data suggest that PKA has no role in governing Pct1 membrane-binding.   108    Table 3.3 Localisation of Pct1-GFP upon PKA inactivation in exponential phase Strain Y3527 Transformants + 5 M 1- NM-PP1 Ycplac111-PCT1-GFP and Ycplac33-SEC63-mCherry Ycplac111-PCT1-GFP and Ycplac33-MSN2-mCherry† 0 min Nuclear envelope Nuclear envelope/ Msn2 in the cytoplasm 5 min Nuclear envelope Nuclear envelope/ Msn2 in the nucleus and cytoplasm 10 min Nuclear envelope Nuclear envelope/ Msn2 in the nucleus and cytoplasm 15 min Nuclear envelope Nuclear envelope/ Msn2 in the nucleus and cytoplasm 20 min Nuclear envelope Nuclear envelope/ Msn2 in the nucleus and cytoplasm 30 min Nuclear envelope Nuclear envelope/ Msn2 in the nucleus > cytoplasm 90 min Nuclear envelope Nuclear envelope/ Msn2 in the nucleus 120 min Nuclear envelope Nuclear envelope/ Msn2 in the nucleus 120 min, following GS 30 min Nucleus Nucleus/ Msn2 in the nucleus †Msn2-mCherry translocates into the nucleus following the loss of PKA activity upon chemical inhibition. Msn2-mCherry is used as a control (Barbosa et al., 2015; Gorner et al., 2002). GS; Glucose starvation. 109    Figure 3.5.4 PKA inactivation does not affect Pct1 localisation. (A) Strain Y3527 (tpk1M164G, tpk2M147G, tpk3M165G) expressing Pct1-GFP and Sec63- mCherry (ER marker) were grown to exponential phase (Exp) and treated with 1NM-PP1 for the indicated time. Cells were also grown to exponential phase and treated with 1NM-PP1 for 120 min followed by glucose starvation for 30 min (120 + GS 30min). Cells were collected at the indicated time points for imaging. Only glucose starvation triggers Pct1 recruitment to the nucleoplasm, indicating that the lack of Ras-cAMP-Tpk signalling pathway does not affect Pct1 localisation. Scale bar, 5 μm. (B) Strain Y3527 expressing Pct1-GFP and Msn2-mCherry was grown to exponential phase (Exp) and treated with 1NM-PP1 for the indicated time. Cells were also grown to exponential phase and treated with 1NM-PP1 for 120 min followed by glucose starvation for 30 min (120 + GS 30min). Cells were collected at the indicated time points for imaging.  Msn2-mCherry obviously concentrates in the nucleus from time-point 30 min, indicating that 1NM-PP1 successfully inhibits PKA activity. Scale bar, 5 μm.  Images are representative of three independent experiments. 110    3.4.5 Manipulation of the localisation of Pct1 by supplementation of choline In yeast, both the methylation and Kennedy pathways contribute to PC synthesis, but the latter becomes dominant on choline supplementation. When excess choline (1 mM) is added to the medium of exponentially growing cells, Pct1-GFP is released from the nuclear membrane into the nucleoplasm (Figures 3.2A and D). In conditions that cause dissociation of Pct1-GFP from membranes, i.e. exponential in the presence of choline and PDS, the PC levels and the PC/PE ratio increase (Figure 3.6), suggesting that the nuclear membrane targeting of Pct1 is specifically associated with these phospholipid changes. Notably, enlarged LDs were found in the PDS phase because the lipid precursor DAG is redirected to TAG storage in this slow-growth phase. In contrast, DAG is mainly used for phospholipid synthesis in nutrient-rich conditions of the exponential phase in order to sustain rapid proliferation and leads to relatively smaller LDs. Forcing cells to increase PC synthesis by providing excess choline in the medium in the exponential phase (Exp + Cho), as shown in Figure 3.6A, triggers the redistribution of nucleoplasmic Pct1 but causes no change in LD size (Figure 3.2B). This suggests that other phospholipids are synthesised in the actively dividing cells, possibly indicating their pronounced DAG demands. Therefore, excess choline leads to membrane release of Pct1 but is not associated with lipid storage in exponentially growing yeast cells. 111      112    Figure 3.6 Binding of Pct1-GFP to the nuclear membrane is associated with PC levels in S. cerevisiae. (A) pct1∆ cells expressing Pct1-GFP were grown to exponential phase (Exp), post-diauxic shift phase (PDS), or to exponential phase supplemented with 1 mM choline (Exp + Cho) and imaged. Differential interference contrast (DIC) images are shown for context. Arrows indicate enlarged lipid droplets in the PDS. Scale bar, 5 m. Images are representative of at least three independent experiments. (B) Lipidomic analysis of PC and PE levels at the indicated growth phase. The PE and PC levels shown in nmoles (per 50 mg yeast) are relative to internal standards. (C) PC/PE ratio of pct1∆ cells expressing Pct1-GFP at the indicated growth phase. Data are mean ± SD from three independent experiments. One-way ANOVA with Bonferroni multiple comparison, *p < 0.05. 113    3.4.6 Implementation of PE methylation pathway- and Kennedy pathway- specific knockout yeast strains in the study of Pct1 function The fact that Pct1 nuclear membrane-binding can be visualised, quantified and rapidly reversed (e.g. with choline supplementation) provides a unique opportunity to evaluate how factors influence its localisation (and indirectly also its activity) in vivo. First, we sought to generate a conditional mutant in which PC synthesis can be confined to the Kennedy pathway in a temporally regulatable manner. In yeast, PC is synthesised either through the methylation of PE by Cho2 and Opi3 or through the Kennedy pathway (Figure 3.7A). Mutant yeast cho2Δopi3Δ cells cannot generate PC via the methylation pathway but can do so via the Kennedy pathway when choline is available. If choline is not provided in the growth medium, deletion of both CHO2 and OPI3 in yeast genes leads to slower cell growth and causes cell death after 4–5 generations (Boumann et al., 2006). To recapitulate the regulatory role that Pct1 plays in PC synthesis in vivo, we generated a cho2Δopi3Δpct1Δ triple deletion mutant (henceforth called 3Δ mutant), complemented by OPI3 expressed from a plasmid with the URA3 marker; Opi3 can partially compensate for the loss of Cho2 (Kanipes and Henry, 1997; Preitschopf et al., 1993; Thibault et al., 2012), which allows the 3Δ cells to grow by making PC through PE methylation. Transformation of the 3Δ cells with a plasmid expressing PCT1, allows the plasmid carrying URA3 to be counter selected by growing cells on 5-fluoroorotic acid (5-FOA). Otherwise nontoxic to yeast, 5-FOA is converted to the toxic form (i.e. 5-flurouracil) in strains expressing the functional URA3 gene coding for orotine-5-monophosphate decarboxylase that is involved in the synthesis of uracil (Boeke et al., 1984). URA3 converts 5-FOA into a toxic product, which prevents cell growth (3Δ pURA-opi3; Figure 3.7B). After counter-selection on 5-FOA, 3Δ cells employ Pct1 as the sole enzymatic source for PC in the presence of choline. Indeed, both Pct1 and Pct1-GFP can rescue the growth of the 3Δ cells in the presence of 5-FOA (Figure 3.7C). 3Δ cells expressing Pct1-GFP are viable when supplemented with choline but cell growth is arrested when the cells are deprived of choline over a period of 40 hr (Figure 3.7D, (Boumann et al., 2006)). Initial growth without choline may possibly be affected by the generation of choline by phospholipase D activity, which hydrolyses PC to choline and phosphatidic acid (Gibellini and Smith, 2010; Li and Vance, 2008; Rose et al., 1995), and might then help maintain a basal catalytic turnover from the Kennedy pathway even in the absence of exogenously supplied choline. 114    We also found that 3Δ cells show reduced viability on 5-FOA plates when expressing Pct1 mutants that significantly impair membrane-binding (GPd) or catalytic activity (Cmut and Cdel) (Figure 3.7C). The fact that 3Δ expressing these Pct1 mutants can still survive may be attributed to the Ect1 (Figure 3.7E) that may compensate for the loss of Pct1. Studies have shown that although Ect1 activates phosphoethanolamine to be further converted into PE, Pct1 is somewhat redundant in its substrate specificity and can use the same substrate to synthesise PE and this may happen in the opposite direction (Rockenfeller et al., 2015). Since the two proteins could be dispensable due to their substrate promiscuity, only the double knockout would entirely inhibit PC synthesis by the Kennedy pathway. To test this hypothesis, we generated a cho2Δopi3Δpct1Δect1Δ quadruple deletion mutant (henceforth called 4Δ mutant; Figure 3.7E), complemented by OPI3 expressed from a plasmid with the URA3 marker, and compared with the 3Δ mutant in a 5-FOA screening assay. In point of fact, 4Δ cells appeared to grow slightly less well than 3Δ cells in Pct1 loss-of-function genetic screen. (Figure 3.7F). This phenomenon suggests that PC synthesised via the Ect1 catalytic capability can compensate to some degree for the survival of the 3Δ mutant cells. Overall, the 5-FOA assay in 3Δ or 4Δ cells was established as a viable tool to test the activity of different Pct1 mutants. 115    Figure 3.7 Validation of the 3∆ (cho2∆opi3∆pct1∆) and 4∆ (cho2∆opi3∆pct1∆ect1∆) S. cerevisiae model. 116    (A) Schematic of the enzymes involved in PC biosynthesis. Deletion of Cho2 and Opi3 blocks the methylation pathway while deletion of Pct1 blocks the CDP-choline branch of the Kennedy pathway. PC, phosphatidylcholine; PE, phosphatidylethanolamine; Cho, choline; P-Cho, phosphocholine; CDP-Cho,  cytidinediphosphate choline; PMME, phosphatidyl- monomethylethanolamine; PDME, phosphatidyl-dimethylethanolamine. (B) Wild-type cells, cho2∆opi3∆ double deletion cells or 3Δ cells carrying Ycplac33-URA3-OPI3 were grown to exponential phase and serial dilutions of liquid cultures were spotted onto SC plates lacking histidine (-HIS) or plates supplemented with FOA and 1 mM choline (FOA + cho). (C) (i) 3Δ cells carrying Ycplac33-URA3-OPI3 were transformed with a CEN/HIS3 plasmid expressing wild-type Pct1, Pct1-GFP or mutant GPd-GFP, Cmut-GFP and Cdel-GFP. Empty CEN/HIS3 plasmid (EV) was used as a negative control. Cells were grown to exponential phase and serial dilutions of liquid cultures were spotted onto SC plates lacking histidine (-HIS) or plates supplemented with FOA and 1 mM choline (FOA + cho). (ii) Schematic illustration of the Pct1 domain organisation indicating the mutations in the GPd, Cmut and Cdel Pct1 mutants. C-domain, catalytic domain; M-domain, membrane-binding domain. Substitution mutations are indicated with red bars and targeted deletions are indicated with crossed boxes. (D) Growth of 3Δ cells expressing Pct1-GFP in medium supplemented with (+cho) or without (-cho) additional 1 mM choline. (E) Schematic illustration of PC synthesis routes through the methylation and Kennedy pathway. The enzymes deleted in the 4∆ mutant are cho2 and opi3 in the methylation pathway (blue cross) as well as pct1 and ect1 in the CDP-choline and CDP-ethanolamine pathway (red cross). PC, phosphatidylcholine; PE, phosphatidylethanolamine; Cho, choline; P-Cho, phosphocholine; CDP-Cho, cytidinediphosphate choline; PMME, phosphatidyl- monomethylethanolamine; PDME, phosphatidyl-dimethylethanolamine; Etn, ethanolamine; P- Etn, phosphoethanolamine; CDP-Etn, cytidinediphosphate ethanolamine. (F) 3Δ or 4Δ cells carrying Ycplac33-URA3-OPI3 were transformed with a CEN/HIS3 plasmid expressing Pct1- GFP or mutant GPd-GFP, Cmut-GFP and Cdel-GFP. Empty CEN/HIS3 plasmid (EV) was used as a negative control. Cells were grown to exponential phase and serial dilutions of liquid cultures were spotted onto SC plates lacking histidine (-HIS), plates supplemented with FOA (FOA), or plates supplemented with FOA and 1 mM choline (FOA + cho). Data are representative of at least two repeats. 117    3.4.7 Pct1 nuclear membrane localisation regulated by substrate availability If Pct1 is a sensor of PC levels, its membrane-binding should respond to specific changes in membrane properties e.g. PC levels and PC/PE ratios. To address this, we monitored how Pct1- GFP responds to choline addition after prolonged choline starvation and analysed the resulting phospholipid changes in 3Δ cells in time-course experiments. Under 24 hr choline starvation (0.0 hr in Figure 3.8A), Pct1-GFP is mostly bound to the nuclear membrane and PC levels are very low, which results in a decreased PC/PE ratio (Figure 3.8C and D). We also observed an increase in LD size in these cells with prolonged absence of choline (Figure 3.8E), which is consistent with previous data showing that PC deficiency induces supersized LDs (Fei et al., 2011). Addition of choline to these cells induces a prompt decrease of membrane-associated Pct1-GFP (Figures 3.8A and B) that correlates with an increase in PC levels and the PC/PE ratio (Figures 3.8C and D). As the cells synthesise PC, the LDs shrink in size during choline supplementation (Figure 3.8E). Anti-GFP antibody was used to detect the expression of Pct1-GFP on western blots, and the protein expression remains stable over the entire course of the assay (Figure 3.8F). Note that even in the choline-deprived state we do not detect Pct1-GFP staining around LDs, and this again suggests that it is not trafficking in and out of the nucleus. Taken together, these data suggest that substrate availability and consequent PC synthesis govern Pct1 membrane-binding. 118    Figure 3.8 Changes in the PC/PE ratio correspond to choline-induced WT Pct1-GFP translocation off the nuclear membrane. 119    3Δ yeast cells expressing Pct1-GFP were grown for 24 hr followed by addition of 1 mM choline. Cells were collected before (0.0 hr) and after choline supplementation at the indicated time points for imaging or lipidomics analysis. (A) Representative confocal microscopy images of Pct1-GFP localisation. Lipid droplets (LDs) were stained as described in Methods section 2.5.1. Differential interference contrast (DIC) images are shown for context. Scale bar, 10 m. Images are representative of at least three independent experiments. (B) Quantification of the Pct1-GFP localisation shown in (A). INM, inner nuclear membrane. Data are mean ± SD from three independent experiments; ~300 cells were counted from random fields at each time point per experiment. (C) PC and PE levels at the indicated time points are shown in nmoles (per 50 mg yeast) relative to internal standards. Data are mean ± SD from three independent experiments. (D) PC/PE ratio at the indicated time points. Data are mean ± SD from three independent experiments. One-way ANOVA, ***p < 0.001. (E) Measurement of LD area as described in Methods section 2.5.2 from cells imaged in (A). Data are mean ± SD from three independent experiments; ~600 LDs were counted in each condition. Two-tailed Student’s t-test, *p < 0.05. (F) Representative GFP immunoblotting of cell lysates from 3Δ yeast cells expressing Pct1-GFP at the indicated time points following choline addition. GAPDH was used as a loading control. 120    3.4.8 Phospholipid composition affecting Pct1 membrane affinity Next, we sought to determine how membrane phospholipid (PL) composition controls Pct1 localisation in vivo. The biophysical properties of membranes are influenced by both fatty acid (FA) and headgroup changes in the PLs, so mass spectrometry analysis was used to characterise the yeast lipidome comprehensively. By calculating the lipid headgroup composition, we found that exponential phase lipidome of the pct1Δ cells expressing Pct1-GFP mainly contained PC and PE (Figures 3.9.1A and B). Our analysis corresponds with earlier reports which have demonstrated that PC dominates the yeast glycerophospholipidome (Daum et al., 1999). We also observed that S. cerevisiae modulates its lipid composition in the PDS phase by (i) synthesising more TAG species (Figure 3.9.1A) and fewer PE species (Figures 3.9.1B); (ii) increasing saturated phospholipids and decreasing unsaturated phospholipids (Figures 3.9.1C and D); (iii)  PS enrichments primarily due to increases in PS (34:1) and PS (34:2) (Figure 3.9.1E). Overall cells in PDS phase have a significantly higher PC/PE ratio (1.7-fold) compared with the exponential phase (Figure 3.6C). Such  changes in lipid species composition occurring in the transition from exponential to the PDS growth phase have been reported before (Klose et al., 2012). Intriguingly, the TAG accumulation seen in the lipidomics data from PDS phase cells is deducible from the enlarged lipid droplets observed in vivo (Figure 3.2B). We note that in cells supplied with exogenous choline (Exp + Cho), although no TAG accumulation occurs, a similar pattern of changes in PE was observed (Figures 3.9.1B and D), along with a similar PC/PE ratio observed as in the PDS phase (Figure 3.6C). This suggests that a higher PC/PE ratio governs Pct1 translocation from nuclear membrane to the nucleoplasm. The effect of FA unsaturation on the membrane fluidity may be one of the factors regulating Pct1 membrane-binding (Attard et al., 2000). Thus, we looked at the different acyl chain species of PLs in the yeast lipidome. In pct1Δ cells expressing Pct1-GFP, the most abundant PC species are PC (32:2), PC (34:2) and PC (32:1) (Ejsing et al., 2009), which were present respectively in 36.6%, 34.5% and 13.2% of the total PC lipids in the exponential phase (Figure 3.9.1C). The unsaturated PE lipids, such as PE (34:2) (45.5%) and PE (32:2) (24.1%) (Figure 3.9.1D) were found to be the most abundant PE lipids, in agreement with previous studies showing that PE 16:1-16:1 and PE 16:1-18:1 are the dominant PE species (Ejsing et al., 2009). There was a decrease in PC (32:2) and PC (34:2) as well as an increase in PE (34:1), PC (32:1) and PC (34:1) in both the PDS and Exp + Cho condition, suggesting that the proportion of unsaturated acyl 121    chains in PLs could also affect Pct1 membrane-binding affinity, and Pct1 tends to bind to membranes rich in unsaturated phospholipids. In the 3Δyeast cells expressing Pct1-GFP, our lipidomic data show that depleted PC (at 0.0 hr) is compensated for by significant increases in PE, as well as small but significant decreases in PS lipid classes, whereas the amounts of other lipid classes present in biological membranes are unchanged (Figure 3.9.2A). In the choline-starved state (at 0.0 hr), we hypothesise that cells may increase decarboxylation of PS to PE to drive the CDP-DAG pathway (Figure 3.1), thus reducing PS levels; however, with the methylation pathway blocked by the deletion of Cho2 and Opi3, PE levels will tend to rise. TAG levels were higher without choline (at 0.0 hr) and fell on choline supplementation (Figure 3.9.2A (i)) concomitant with the reduction in LD size observed in Figure 3.8E. Changes in PC, PE and PS lipid fractions coincide with adaptive changes in the FA acylation patterns of these lipids (Figures 3.9.2B–D). In particular, PE (32:1) and PS (32:1) may be increased to about 30 mol% of total PE or PS (Figures 3.9.2C and D) at time 0.0 hr. After choline supplementation (at 24 hr), PE (32:1) and PS (32:1) decreased, but PE (34:2) and PS (34:2) rose to compensate for the loss. The reduced PC/PE ratio in the choline deficient state (at 0.0 hr) (Figure 3.8D) is expected to create high membrane elastic stress (Figure 3.9.2F), and so it appears that cells compensate for the stress by replacing mono-unsaturated FAs with saturated FAs predominantly in the PE lipids. 122    Figure 3.9.1 Data-driven modelling of stored curvature elastic (SCE) stress regulation from the lipidomes of pct1Δ cells expressing WT Pct1-GFP. 123    pct1Δ yeast cells expressing Pct1-GFP were grown to exponential phase (Exp), post-diauxic shift phase (PDS), or to exponential phase supplemented with 1 mM choline (Exp + Cho). Cells were collected at the indicated conditions for lipidomics analyses and the total (PSCE) and the lipid- driven stress in the cells was calculated as described in Methods section 2.8. These data correlate with imaging slides shown in Figure 3.2 and Figure 3.6. (A) Pie charts with the overall distribution of the main lipid headgroups. PC,   phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PI, phosphatidylinositol; PG, phosphatidylglycerol; PA, phophatidic acid; DAG, diacylglycerol; TAG, triacylglycerol. (B) Lipid composition by headgroup (mol% total membrane lipid) at the indicated conditions (see top middle inset for condition color codes). (C–E) Fatty acid composition of the PC (C), PE (D) and PS (E) fraction expressed as mol% of the respective total phospholipids. The phospholipid (32:1) or (34:1) represent species where the acyl groups at C-1 and C-2 contain a total of 32 or 34 carbon atoms with 1 double bond while phospholipid species (32:2) or (34:2) represent species with 2 double bonds. (F) Estimates of the total SCE stress (PSCE stress). (G) Estimates of the lipid contribution to SCE stress. All data are mean ± SD of 3-4 experimental repeats. One-way ANOVA, **p < 0.01. 124    Figure 3.9.2 Data-driven modelling of stored curvature elastic (SCE) stress regulation from the lipidomes of 3Δ cells expressing WT Pct1-GFP and undergoing PC recovery. 125    3Δ yeast cells expressing Pct1-GFP were grown for 24 hr followed by addition of 1 mM choline. Cells were collected before (0.0 hr) and after choline supplementation at the indicated time points (see top left inset for time point color codes) for lipidomics analyses and the total (PSCE) and the lipid-driven SCE stress in the cells was calculated as described in Methods section 2.8. These data correlate with imaging slides shown in Figure 3.8. (A) (i)  Lipid composition by headgroup (mol% total lipid); (ii) Lipid composition by headgroup (mol% total membrane lipid). PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PI, phosphatidylinositol; PA, phophatidic acid; DAG, diacylglycerol; TAG, triacylglycerol. (B-D) Fatty acid composition of the PC (B), PE (C) and PS (D) fraction expressed as mol% of the respective total phospholipids. The phospholipid (32:1) or (34:1) represent species where the acyl groups at C-1 and C-2 contain a total of 32 or 34 carbon atoms with 1 double bond while phospholipid species (32:2) or (34:2) represent species with 2 double bonds. (E) Estimates of the total SCE stress (PSCE stress). (F) Estimates of the lipid contribution to SCE stress. All data are mean ± SD of 8 experimental repeats. One-way ANOVA, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. To assess the specific role of unsaturated FAs in governing Pct1 membrane translocation in vivo, we generated a mutant yeast strain, oleΔ, with disruption in the OLE1 gene that codes for an essential FA desaturase (Resnick and Mortimer, 1966). A reduction of growth in the absence of oleic acid (OA) was observed in the ole1Δ cells (Figures 3.9.3A and B)  (Stukey et al., 1989)). This may correlate with the failure of unsaturated phospholipid formation, which was also documented by using a disrupted genomic fragment containing the OLEl gene to replace the wild-type OLEl gene in yeast and resulted in an unsaturated FA-requiring phenotype (Stukey et al., 1989). It is worth noting that this perturbation does not appear to affect cell survival and Pct1 localisation within 2–3 generations (Figure 3.9.3C), but it does highlight the evolutionarily conserved importance of unsaturated phospholipids in the development of membrane lipid structure. 126    Figure 3.9.3 Deletion of fatty acid desaturase Ole1 decreases cell viability. (A) Streak plates showing cell viability of Ole1 deletion strains (ole1 Δ) derived from RS453 or BY4741 on synthetic and YPD medium. Plates were incubated at 30⁰C for 3 days. Cells are unable to grow unless oleic acid (OA) is present on plates. (B) Growth of oleΔ cells derived from RS453 or BY4741 in medium supplemented with (+OA) or without (-OA) additional oleic acid. 127    (C) Wild-type or ole1∆ cells derived from RS453 or BY4741 expressing Pct1-GFP were grown to exponential phase (Exp) or post-diauxic shift phase (PDS) in medium supplemented with (+OA) or without (-OA) additional oleic acid. Cells were collected and imaged. Differential interference contrast (DIC) images are shown for context. Arrows indicate dead cells found in the PDS. Scale bar, 5 m. Images are representative of three independent experiments. Given the compelling in vitro evidence that DAG can also induce Pct1 membrane association (Arnold and Cornell, 1996), the higher DAG level in PDS phase cells (Figures 3.9.1A), though the magnitude is small and the significance is questionable, was surprising. It is because of the note that in the PDS phase Pct1 is nucleoplasmic and not membrane-bound. We hypothesised that this may be explained by the possiblity that DAG is present primarily in LDs in the PDS phase rather than in the nuclear membrane. Accordingly, both DAG and TAG levels are elevated in 3Δ cells at 0.0 hr when LDs are larger (Figure 3.9.2A (i)). To validate our hypothesis, we attempted to isolate a subcellular fraction enriched with LDs from yeast in the PDS phase in order to carry out further lipidomics analyses. The purification efficiency was tested at each of the sequential separation steps by western blotting with antibodies against known markers of the LDs and ER (Figure 3.9.4B). The LDs markers Ary1/Erg and the ER marker Dpm1 were selected based on published protocols (Athenstaedt, 2014; Mannik et al., 2014). Although LDs have reportedly been extracted from yeast cells, ER membranes have been a common contaminant that co- purifies with LDs (Bouchez et al., 2015; Vevea et al., 2015) (Figure 3.9.4B). Despite repeated optimisation, we were unable to effectively isolate LDs. However, several reports support our hypothesis that during LD formation, DAG might be present in considerable amounts on the LDs (Leikin et al., 1996; Thiam et al., 2013; Walther and Farese, 2012). Further studies are required to examine the purification steps, and to investigate the suitability of other ER markers, since the Dpm1 is currently accepted to be stabilized on the ER membrane (Ashida et al., 2006), yet it also appears to reside in LDs (Takeda and Nakano, 2008).  The purity of the isolated LDs is possibly higher than expected. 128    Figure 3.9.4 Isolation of lipid droplets from cultured yeast cells. (A) Scheme of the isolation and purification procedure for lipid droplets used in the present study based on (Fei et al., 2008; Leber et al., 1994). Cultured cells were converted into spheroplasts by zymolase and DTT. Cell material obtained was subjected to a three-step density gradient centrifugation separation. After several homogenization and ultra-centrifugation steps, the 3rd gradient separation results in the top floating layer of purified lipid droplets. (B) Evaluation of the purity of the lipid droplet isolation method. The purity of different fractions from (A) as indicated is analyzed by western blotting with antibodies recognizing lipid droplets or ER compartments. Note that Dpm1 is on the ER, but it also localises to LDs to some extent. A, Zymolase reaction buffer; B, Floating layer of 1st extraction; C, Floating layer of 3rd (final) extraction; P1, Pellet of 1st extraction; S1, Supernatant of 1st extraction; P2, Pellet of 2nd extraction; US2, Upper supernatant of 2nd extraction; I2, Interphase of 2nd extraction; LS2, Lower supernatant of 2nd 129    extraction; P3, Pellet of 3rd extraction; US3, Upper supernatant of 3rd extraction; I3, Interphase of 3rd extraction; LS3, Lower supernatant of 3rd extraction. 3.4.9 Membrane elastic stress determing Pct1 membrane localisation Stored curvature elastic (SCE) stress arises within membrane bilayers when lipids within the constituent monolayers are forced to adopt an unfavourable packing conformation, for instance, when ‘conically shaped’ (also known as type II) phospholipids accumulate in the membrane (McDonald et al., 2015). This stress results in hydrophobic ‘cavities’ in the membrane, also referred to as membrane-packing defects. Failure to alleviate the SCE stress can ultimately result in phase transition and pore formation. Mechanistically, cells can control curvature elastic stress in their membranes by changing lipid composition or by partitioning proteins that relieve curvature elastic stress to the membrane. To explore this in more detail, we collaborated with Dr Marcus Dymond (Division of Chemistry, School of Pharmacy and Biomolecular Sciences, University of Brighton), who calculated a proxy value of membrane curvature elastic energy, PSCE, derived from data-driven modelling studies (Dymond, 2016). This provides a metric for total cellular curvature elastic stress including both lipid and protein contributions. PSCE was calculated for each lipidome, and the individual concentrations of each lipid species for each population of cells were considered along with the relevant value of curvature elastic contribution (w) for that lipid species. The determined mean average of PSCE for each lipidome was found to be tightly maintained in the range of 0.05–0.15 in pct1Δ cells expressing Pct1-GFP (henceforth called pct1Δ Pct1 cells; Figure 3.9.1F). A similar trend in the PSCE data shows good evidence of overall regulation of the intrinsic curvature, suggesting that cells maintain membrane curvature elastic energy within tight boundaries. Since Pct1 translocates to membranes, we estimated the lipid contribution to stored curvature elastic stress (N) in membranes. This is different from PSCE and stored curvature elastic stress (N) is influenced by the molecular contribution of individual phospholipids. We estimate that the lipid contribution to membrane curvature elastic stress is in the order of 0.5-1.5 × 10−11 N in pct1Δ Pct1 cells (Figure 3.9.1G). Pct1 is membrane-bound when the curvature elastic stress is relatively high. Pct1 becomes soluble and inactivated in accordance with the decrease in the curvature elastic stress in PDS phase and Exp + Cho condition. This model suggests that sensing curvature elastic stress can be the dominant mechanism by which Pct1 on/off membrane-dependent activation is regulated. These results also suggest that yeast 130    cells might reduce elastic stress in response to the demand for PC by incorporating Pct1 into the inner nuclear membrane. In 3Δ cells expressing Pct1-GFP (henceforth called 3Δ Pct1 cells), we found a narrow range of PSCE values from 0.1 to 0.2 (Figure 3.9.2E) during the time-course choline supplementation experiment, which is evidence of preservation of membrane curvature elastic stress and membrane curvature properties throughout. However, the magnitudes of PSCE are slightly higher in 3Δ Pct1 cells than in pct1Δ Pct1 cells, indicating that the degree of curvature elastic stress energy is likely to be different in each of the different strains. The lipid contribution to membrane curvature elastic stress in 3Δ Pct1 cells is high, in the order of 2.8 x 10-11 N (Figure 3.9.2F), when PC levels are depleted and the Pct1 is localised on the NE (Figure 3.8A). The results of measurement suggest that changes in lipid FA acylation do not fully compensate for the stress induced by increases in PE at the early time points, and that Pct1 is recruited to the NE to alleviate the remaining elevated curvature elastic stress. Overall, the curvature elastic stress estimates show a good correlation with Pct1 nuclear membrane-partitioning during the time- course choline supplementation assay. It is noteworthy that a 1.4-fold increase of the curvature elastic stress is observed in 3Δ Pct1 cells after 24 hr of choline supplementation compared with pct1Δ Pct1 cells with excess of choline (Exp + Cho), which might be explained by the defective methylation pathway in the former; indeed, the total PC amount is lower in the 3Δ cells (Exp + Cho; Figure 3.9.1B vs. 24 hr; Figure 3.9.2A (ii)). These data above shed light on the importance of curvature elastic stress in Pct1 membrane-binding, with high levels of SCE causing Pct1 to be “on” and low levels causing Pct1 to come “off” the nuclear membrane. 3.4.10 Lack of dependence of Pct1 membrane localisation on the source of PC synthesis Previous studies have suggested that the FA composition of PC derived from the Kennedy pathway is different from the PC produced through methylation of PE (Boumann et al., 2003). Therefore, we were interested in whether or not Pct1 membrane association depends on the source of PC, particularly if Pct1 specifically sense PC generated through the Kennedy pathway. PC is produced exclusively from the Kennedy pathway in the cho2Δopi3Δ double mutant. In the cho2Δopi3Δ cells, Pct1 is associated with the nuclear membrane in the absence of its initial substrate choline, reflecting cellular deficiency of PC, but shows a gradual nucleoplasmic re- 131    localisation when the cells are supplemented with choline (Figures 3.10.1B and C). To force the cells to synthesise PC exclusively through the methylation of PE, we used a double mutant deleted for CPT1 and EPT1 genes (Figure 3.10.1A), which encode the two enzymes responsible for the last steps of the CDP-choline and CDP-ethanolamine branches of the Kennedy pathway respectively. (McMaster and Bell, 1994). We deleted both CPT1 and EPT1 as the latter is known to contribute a small percentage (5%) to PC synthesis in vivo, due to lack of substrate fidelity (Harrison et al., 2007; McMaster and Bell, 1994). We found that, on choline supplementation, while in the cpt1Δept1Δ mutant cells Pct1 appeared to localise primarily to the nuclear envelope, in the cpt1Δ cells Pct1 partitioned more within the nucleoplasm (Figure 3.10.1B and C). This indicates that the Kennedy pathway is still partially active in the cpt1Δ cells and implies that Ept1 truly accounts for some PC synthesis through this pathway. Therefore, only a double mutant cpt1Δept1Δ would block the Kennedy pathway completely. With regards to Pct1 membrane- targeting response resulting from the deprivation of PC biogenesis, cpt1Δept1Δ cells behaved similarly to that of cho2Δopi3Δ cells. When the demand of PC is not met i.e. the Exp phase in cpt1Δept1Δ or the –Cho Exp phase in cho2Δopi3Δ, Pct1-GFP mainly localises to the nuclear membrane (Figure 3.10.1B and C). When the demand of PC is met i.e. the PDS phase in cpt1Δept1Δ or the +Cho conditions in cho2Δopi3Δ, Pct1-GFP “falls off” into the nucleoplasm (Figure 3.10.1B and C). PE can either be made from the CDP-ethanolamine pathway or the decarboxylation of PS (Figure 3.1); thus, cpt1Δept1Δ cells are able to produce PE by the decarboxylation of PS and to generate PC via the PE methylation pathway. At the PDS growth phase, PC generated from either Kennedy or methylation pathway is sufficient to relieve nuclear membrane-packing defect or enough to maintain surface curvature stress in a slow-growing state. Overall, these data demonstrate that Pct1 membrane association/dissociation occurs regardless of the source of PC. 132    Figure 3.10.1 Pct1 localisation in response to PC levels is independent of the source of PC. 133    (A) Schematic illustration of PC synthesis through the methylation and Kennedy pathway. The enzymes deleted in the cho2Δopi3Δ mutant (blue cross) and in the cpt1Δ or cpt1Δept1Δ mutant (red cross) are indicated. PC, phosphatidylcholine; PE, phosphatidylethanolamine; Cho, choline; P-Cho, phosphocholine; CDP-Cho, cytidinediphosphate choline; PMME, phosphatidyl- monomethylethanolamine; PDME, phosphatidyl-dimethylethanolamine; Etn, ethanolamine; P- Etn, phosphoethanolamine; CDP-Etn, cytidinediphosphate ethanolamine. (B) Quantification of the Pct1-GFP localisation shown in (C). Data are mean ± SD based on random fields from three independent experiments; ~100 cells were counted at each time point in each independent experiment. (C) Pct1-GFP was expressed in a cho2Δopi3Δ mutant (defective in the methylation pathway), a cpt1Δ mutant or a cpt1Δept1Δ mutant (defective in the CDP-choline pathway). Cells were grown to the indicated growth phase with (+) or without (-) excess choline and imaged. DIC images are shown for context. Arrows indicate dead cells found in the PDS. Scale bar, 5 m. Images are representative of at least three independent experiments. 134    3.4.11 Overall PC composition is unaffected by the source of PC synthesis To study the specific contribution of either PC biosynthesis pathway to membrane PL composition, lipidomics analyses were carried out in the cho2Δopi3Δ and cpt1Δept1Δ mutant cells. The PC level in cho2Δopi3Δ cells was found to be comparatively lower than cpt1Δept1Δ cells at the PDS phase (Figure 3.10.2B), which is consistent with previous reports that proved PC is primarily synthesized in yeast by the Cho2 and Opi3 via the PE methylation pathway (Carman and Henry, 1999; Ko et al., 1994). Exponentially growing or PDS phase cho2Δopi3Δ mutant cells were stimulated to produce PC through the alternative Kennedy pathway when supplemented with choline. In the presence of choline, the overall PC content of the cho2Δopi3Δ cells was increased by about 15–20% when compared with cells grown in the absence of choline (Figures 3.10.2A and B). Similar to previous observations from 3Δ cells expressing Pct1-GFP, an increase in the PE content (1.5–2.5 fold) was seen as a means to make up for the loss of PC in the cho2Δopi3Δ cells after 24 hr of choline starvation (Figures 3.10.2A and B). These changes in PLs increase the lipid-driven SCE stress (Figure 3.10.3A) and thereby increase the membrane-bound active form of Pct1 (Figures 3.10.1B and C). A negative correlation was observed between the lipid-driven SCE stress and the PC/PE ratio (Figure 3.10.3B), indicating that increasing the amount of PE lipids in a PC bilayer can raise membrane elastic stress. Also, choline depletion in exponentially growing cho2Δopi3Δ cells correspondingly caused a decline in PS and DAG content, and enhanced TAG levels (Figure 3.10.2A). An overall shift in the cellular acyl chain composition occurs in the PE and PS species upon PC depletion (Figures3.10.2 E–H). Since PE has a slightly smaller head group than PC, it has a strong tendency to form non-lamellar phases; however, shortening and increased saturation of the PE acyl chains (Figures 3.10.2E and F; higher relative amount of PE (32:1)) may compensate for this non-bilayer formation propensity. Similarly, PS acyl chains had an increase in the degree of saturation with a higher relative amount of PS (32:1) and PS (34:1) observed (Figures 3.10.2G–H). We further compared the PC composition of the cho2Δopi3Δ cells with cpt1Δept1Δ cells. PC in cho2Δopi3Δ mutant was subjected to acyl chain remodelling during choline starvation, with saturated species replacing unsaturated species in both growth phases, as shown in Figure 3.10.2C and Figure 3.10.2D. The remodelling of PC was reversed by choline supplementation, where most of the PC is unsaturated PC (34:2) and PC (34:2) in the exponential phase. However,  the changes in the PC species profile were limited to the exponential phase while no changes 135    were observed in the PDS phase. Additionally, in cpt1Δept1Δ cells, PC (32:2) and PC (34:2) in the exponential phase were the dominant species (Figure 3.10.2C). We observed that in both mutants, there are higher amounts of saturated PC in the PDS phase (Figure 3.10.2D), with comparable levels of PC (32:1) and PC (34:1) found, when they were able to synthesize PC from either the methylation or the Kennedy pathway. These results suggested that the PC composition is maintained regardless of the source of PC synthesis. 136    137    Figure 3.10.2 Lipidomics analyses of cho2Δopi3Δ mutant and cpt1Δept1Δ mutant. cho2Δopi3Δ mutant and cpt1Δept1Δ mutant were grown to exponential phase (Exp) or post- diauxic shift phase (PDS) in medium supplemented with (+Cho) or without (-Cho) additional choline. Cells were collected at the indicated growth conditions (see top insets for condition color codes) for lipidomics analyses with one repeat. (A) Lipid composition by headgroup (mol% total lipid) in Exp. PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PI, phosphatidylinositol; PA, phophatidic acid; DAG, diacylglycerol; TAG, triacylglycerol. (B)  Lipid composition by headgroup (mol% total lipid) in PDS. (C–H) Fatty acid composition of the PC (C and D), PE (E and F) and PS (G and H) fraction expressed as mol% of the respective total phospholipids. The phospholipid (32:1) or (34:1) represent species where the acyl groups at C-1 and C-2 contain a total of 32 or 34 carbon atoms with 1 double bond while phospholipid species (32:2) or (34:2) represent species with 2 double bonds. Data are shown for n=1 experiment. Figure 3.10.3 Data-driven modelling of the lipid contribution to SCE stress regulation from the lipidomes of cho2Δopi3Δ mutant and cpt1Δept1Δ mutant. cho2Δopi3Δ mutant and cpt1Δept1Δ mutant were grown to exponential phase (Exp) or post- diauxic shift phase (PDS) in medium supplemented with (+Cho) or without (-Cho) additional choline. Cells were collected at the indicated growth conditions (see right insets for condition color codes) for lipidomics analyses  with one repeat. These data correlate with data shown in Figure 3.10.2. (A)  Estimates of the lipid contribution to SCE stress. (B) PC/PE ratio of cho2Δopi3Δ mutant and cpt1Δept1Δ mutant at the indicated growth condition. Data are shown for n=1 experiment. 138    3.4.12 Manipulating the catalytic domain of Pct1 has no effect on membrane association of Pct1 To investigate whether the catalytic activity of Pct1 has an effect on its ability to sense the lack of PC on the membrane, Pct1Cdel-GFP (C-domain deletion mutant; [145-178]Δ) and Pct1Cmut- GFP (C-domain catalytic mutant; V169M/H195A/Y200A) mutants were generated (Figure 3.7C(ii)). In order to identify the putative C-domain of Pct1 for deletion in the Cdel mutant, multiple protein sequence alignments were performed with homologs from different species with C-domain boundaries (Figure 3.11.1A). The three residues mutated to impair catalytic activity in the Cmut were chosen from studies by Payne et al., 2014 and Lee et al., 2009. The mutation V169M corresponds to the human disease mutation V142M equivalent, which is expected to destabilise the entire structural domain; H195A and Y200A are equivalent to rat CCTα active site residues H168 and Y173 that make major interactions with the beta-phosphate of CDP-choline in the crystal structure (Lee et al., 2009). As expected, the mutant Pct1Cdel-GFP is expressed at a very low level in 3Δ cells (Figure 3.11.1B) and this mutant failed to rescue the 3Δ strain, regardless of the presence of choline on 5-FOA plates (Figure 3.7C(i)). A diffuse Pct1Cdel-GFP signal was detected throughout the cell under confocal microscopy (Figure 3.11.2A), probably due to degraded forms of Pct1Cdel-GFP (Figure 3.11.1B). Interestingly, however, the Pct1Cmut- GFP mutant remains membrane-associated even in conditions that usually promote membrane release i.e. with 24 hr choline supplementation (Figure 3.11.2A). We also observed an increase in LD size in these cells with 24 hr absence of choline (Figure 3.11.1C; time 0.0 hr), which is consistent with previous wild-type Pct1 data. The lack of catalytic activity in Pct1Cmut-GFP causes the enzyme to be ‘trapped’ on the INM as expected due to the lack of PC synthesis compared with 3Δ Pct1 cells as seen in lipidomic analysis (Figure 3.11.1D). Even though Pct1Cmut remains on the INM following choline supplementation, it is unable to fully restore lipid based estimates of SCE stress (Figure 3.11.3F). This estimate suggests that Cmut is unable to restore PC levels to the same extent as WT Pct1; additionally, changes in lipid FA acylation (Figure 3.11.3B–D) do not fully compensate for the increase in PE that result from PC deficiency. We found a preserved range of PSCE values in 3Δ cells expressing Pct1Cmut although there were statistically significant differences between time-point means, the values are in the same range of PSCE values in 3Δ Pct1 cells (Figure 3.11.3E). This could imply that i) not only Pct1 but also other proteins are recruited to the nuclear membrane to alleviate the SCE stress; ii) 3Δ cells with 139    Pct1Cmut could not maintain steady SCE stress. Taken together, it is evident that catalytic activity plays no role in membrane-binding but is needed to relieve or escalate the membrane synthesis in response to changes in the PC level. Figure 3.11.1 Pct1 catalytic domain mutant impedes the PC synthesis. 140    (A) Multiple sequence alignments of homologous CCT proteins from three species: S. cerevisiae (yeast), H. sapience (human) and R. norvegicus (brown rat). Alignments were generated by the ClustalW program (McWilliam et al., 2013). Only the sequence of catalytic domain is shown. Color describes the properties of each amino acid: green, hydroxyl, sulfhydrl, or amine; red, small or hydrophobic; blue, acidic; magenta, basic. An * (asterisk) indicates positions which have a single and fully conserved residue; a : (colon) indicates conservation between groups of strongly similar properties; a . (period) indicates conservation between groups of weakly similar properties. Arrows indicate sites of mutations in Cmut Pct1-GFP. The〔 〕 (bracket) indicates deletion regions in Cdel Pct1-GFP. These data correlate with the scheme shown in Figure 3.7C(ii). (B) Representative GFP immunoblotting of cell lysates from 3Δ yeast cells expressing WT, Cumt or Cdel Pct1-GFP at the indicated growth phase. GAPDH was used as a loading control. Arrows indicate the expected size of the intact protein. (C) 3Δ cells expressing GFP- tagged Cmut or WT Pct1 were grown for 24 hr followed by addition of 1mM choline. Cells were collected at the indicated time-points for imaging and measuring LD area as described in Methods section 2.5.2. Data are mean ± SD from three independent experiments; ~600 LDs were counted in each condition. Two-tailed Student’s t-test, *p < 0.05, **p < 0.01. (D) 3Δ cells expressing GFP-tagged Cmut or WT Pct1 were grown for 24 hr followed by addition of 1 mM choline. Cells were collected before (0.0 hr) and after choline supplementation at the indicated time points for lipidomic analysis. (i) PC and (ii) PE levels at the indicated time pints are shown in nmoles relative to internal standards. (iii) PC/PE ratio at the indicated time points. Data are shown as mean ± SD from independent experimental repeats (n=8 for WT; n=3 for Cmut). Two- tailed Student’s t-test, *p < 0.05 at 24 hr. 141    Figure 3.11.2 Abrogation of the catalytic domain or membrane-binding domain alters Pct1 translocation. 142    (A) WT Pct1-GFP (WT), CmutPct1-GFP (Cmut) or CdelPct1-GFP (Cdel) was expressed in a 3Δ mutant (defective in both the methylation pathway and the CDP-choline pathway). Cells were grown to the indicated growth phase with (+) or without (-) excess choline and imaged. DIC images are shown for context. Arrows indicate dead cells found in the PDS. Scale bar, 5 m. (B) 4mPct1-GFP (4m) or GPdPct1-GFP (GPd) was expressed in a 3Δ mutant. Cells were grown to the indicated growth phase with (+) or without (-) excess choline and imaged. DIC images are shown for context. Scale bar, 5 m.  Images are representative of at least three independent experiments. 143    Figure 3.11.3 Data-driven modelling of stored curvature elastic (SCE) stress regulation from the lipidomes of 3Δ cells expressing CmutPct1-GFP or WT Pct1-GFP and undergoing PC recovery. 144    3Δ yeast cells expressing CmutPct1-GFP (Cmut) or WT Pct1-GFP (Pct1) were grown for 24 hr followed by addition of 1 mM choline. Cells were collected before (0.0 hr) and after choline supplementation at the indicated time points (see top left inset for time-point color codes) for lipidomics analyses and the total (PSCE) and the lipid-driven SCE stress in the cells was calculated as described in Methods section 2.8. These data correlate with imaging slides shown in Figure 3.11.2A. (A)  Lipid composition by headgroup (mol% total lipid). PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PI, phosphatidylinositol; PA, phophatidic acid; DAG, diacylglycerol; TAG, triacylglycerol. (B–D) Fatty acid composition of the PC (B), PE (C) and PS (D) fraction expressed as mol% of the respective total phospholipids. The phospholipid (32:1) or (34:1) represent species where the acyl groups at C-1 and C-2 contain a total of 32 or 34 carbon atoms with 1 double bond while phospholipid species (32:2) or (34:2) represent species with 2 double bonds. (E) Estimates of the total SCE stress (PSCE stress). (F) Estimates of the lipid contribution to SCE stress. Data are shown as mean ± SD from independent experimental repeats (n=8 for WT; n=3 for Cmut). One-way ANOVA for (A–E), *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Two-tailed Student’s t-test for (F), * p < 0.05. 3.4.13 Mutations disrupting the amphipathic helix of Pct1 interferes with the ability to synthesise PC The sequences of Pct1 membrane-binding domain (M-domain) from Metazoa and Saccharomyces diverged much further than the sequences of the catalytic domain. This makes the M-domain more variable. We performed sequence alignment of the M-domain from selected Metazoa with corresponding regions in two diverse yeasts (Figure 3.12.1A). Although the exact alignment of the individual residues may remain ambiguous, the global homology with high statistical significance is very likely correct. To test this prediction experimentally in vivo, two mutant constructs (Figures 3.12.1A and D) were designed: (i) Pct1GPd-GFP, a deletion of an M- domain predicted by global profile alignment Metazoa/Saccharomycetes, [254-312] Δ; (ii) Pct14m-GFP, four amino acid substitutions L262D/I273D/F276D/F280D in the lipid activation region  which were predicted to disrupt  the hydrophobic face of the amphiphilic helix in Pct1 using HeliQuest software (Figure 3.12.1B and C). In 3Δ cells, comparable expression of the wild- type and the three M-domain mutants were observed (Figure 3.12.1E). Pct1GPd-GFP designed significantly to impair membrane-binding show reduced association with the nuclear membrane under conditions that promote membrane association of the wild-type enzyme i.e. choline depletion, both in Exp and PDS phases (Figure 3.11.2B), whilst the Pct14m-GFP retain some membrane association properties in the same conditions (Figure 3.11.2B). Indeed, 3Δ Pct1GPd cells show reduced viability on the 5-FOA plate while Pct14m cells only slow the growth on the 145    same plate (Figure 3.12.1F). These results suggest that membrane-binding of Pct1 is important to fully activate the enzymatic activity. In keeping with the consistent nucleoplasm localisation of Pct1GPd-GFP, lipid analysis suggests that the GPd mutant was unable to restore PC levels to the same extent as WT Pct1, though GPd appeared to retain some catalytic activity (Figure 3.12.2A). This is consistent with previous data showing that the lack of catalytic activity in Pct1Cmut-GFP reduces the PC level and the PC/PE ratio (Figure 3.11.1D). Among both mutant Cmut and GPd, changes in PC, PE and PS lipid fractions coincide with adaptive changes in the FA acylation patterns of these lipids in 3Δ cells expressing WT Pct1 (Figures 3.11.3B–D; Figures 3.12.2C–E). Generally, PLs (32:1) and PLs (34:1) were increased at time 0.0 hr. After 24 hr choline supplementation, these elevated FA species decreased and PLs (32:2) and PLs (34:2) rose to compensate. In 3Δ cells expressing Pct1 GPd -GFP, we found a significant difference between time-point means of PSCE values, which is evidence of disorder of membrane SCE stress throughout (Figure 3.12.2F). Apart from this, GPd mutant was also unable to fully normalise lipid-derived SCE stress estimates (Figure 3.12.2G). It is consistent with the previous suggestion that WT Pct1 is recruited to the nuclear membrane to alleviate the remaining SCE stress. Taken together, membrane-binding is needed for Pct1 to synthesise PC and to fully restore membrane curvature stress. 146    Figure 3.12.1 Description of the -helix properties and related mutants in Pct1 membrane- binding domain. (A) Sequence alignment of the membrane binding domain from selected Metazoa with corresponding region in two diverse Saccharomycetes. Alignments were generated by the ClustalX program (Larkin et al., 2007). The used color scheme could be found in http://www.jalview.org/help/html/colourSchemes/clustal.html. The indicated region of the global profile alignment in Metazoa/ Saccharomycetes was chosen for Pct1 [254-312] Δ deletion testing, 147    as described in scheme (D) mutant GPd. H. sapiens, human; R. norvegicus, brown rat; C. intestinalis, sea squirt; D. melanogaster, fruit fly; T. adhaerens, representative of phylum Placozoa; C. elegans, roundworm; S. cerevisiae, yeast; H. polymorpha, yeast. (B) Helical wheel and 3D structure of an idealised -helix predicted to form membrane-binding domain in Pct1. Only the helix-forming residues 261–304 are displayed. The numbers of residues contributing to the hydrophobic face are displayed on the helical wheel. Arrows indicate the four residues (L262, I273, F276 and F280) were chosen for mutagenesis testing, as described in (D) mutant 4m. The 3D structure displays hydrophobic residues in blue and hydrophilic residues in red. (C) Helical wheel plot was generated using HeliQuest (Gautier et al., 2008) and describe the amphipathic helices of (i) WT Pct1 and (ii) 4m Pct1. Only the helix-forming residues 261–282 are displayed. The arrow indicates the direction and magnitude of the hydrophobic moment. The 4m mutant allowed reducing the hydrophobic moment by half. Color describes the properties of each amino acid: yellow, hydrophobic; red, negatively-charged; blue, positively-charged; purple, polar uncharged with hydroxyl side chain; pink, polar uncharged with an amide side chain; grey, small non-polar. (D) Schematic illustration of the Pct1 domain organisation indicating the mutations in the 4m and GPd mutants. Substitution mutations are indicated with purple bars and the targeted deletion is indicated with a crossed box. (E) Representative GFP immunoblotting of cell lysates from 3Δ yeast cells expressing WT Pct1-GFP (Pct1), 4mPct1-GFP (4m) or GPdPct1-GFP (GPd). GAPDH was used as a loading control. (F) 3Δ cells carrying Ycplac33-URA3-OPI3 were transformed with a CEN/HIS3 plasmid expressing wild-type Pct1-GFP or mutant 4m-GFP and GPd-GFP. Empty CEN/HIS3 plasmid (EV) was used as a negative control. Cells were grown to exponential phase and serial dilutions of liquid cultures were spotted onto SC plates lacking histidine (-HIS) or plates supplemented with FOA and 1 mM choline (FOA + cho). Data are representative of at least three independent experiments. 148    149    Figure 3.12.2 Data-driven modelling of stored curvature elastic (SCE) stress regulation from the lipidomes of 3Δ cells expressing GPdPct1-GFP or WT Pct1-GFP and undergoing PC recovery. 3Δ yeast cells expressing GPdPct1-GFP (GPd) or WT Pct1-GFP (Pct1) were grown for 24 hr followed by addition of 1 mM choline. Cells were collected before (0.0 hr) and after choline supplementation at the indicated time points (see top left inset for time-point color codes) for lipidomics analyses and the total (PSCE) and the lipid-driven SCE stress in the cells was calculated as described in Methods section 2.8. These data correlate with imaging slides shown in Figure 3.11.2B. (A) (i) PC and (ii) PE levels at the indicated time pints are shown in nmoles relative to internal standards. (iii) PC/PE ratio at the indicated time points. Data are shown as mean ± SD from independent experimental repeats (n=8 for WT; n=3 for GPd). Two-tailed Student’s t-test, * p < 0.05 at 24 hr. (B) Lipid composition by headgroup (mol% total lipid). PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PI, phosphatidylinositol; PA, phophatidic acid; DAG, diacylglycerol; TAG, triacylglycerol. (C–E) Fatty acid composition of the PC (C), PE (D) and PS (E) fraction expressed as mol% of the respective total phospholipids. The phospholipid (32:1) or (34:1) represent species where the acyl groups at C-1 and C-2 contain a total of 32 or 34 carbon atoms with 1 double bond while phospholipid species (32:2) or (34:2) represent species with 2 double bonds. (F) Estimates of the total SCE stress (PSCE stress). (G) Estimates of the lipid contribution to SCE stress. Data are shown as mean ± SD from independent experimental repeats (n=8 for WT; n=3 for GPd). One- way ANOVA, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. 150    3.4.14 DNA damage protection function of Pct1 in the nucleus The presence of Pct1 in the nucleus could couple its “sensing” function to another nuclear role such as transcriptional regulation of genes encoding PC biosynthetic enzymes or nuclear enevelope remodelling activity that leads to the formation of the nucleoplasmic reticulum (Gehrig et al., 2008; Lagace and Ridgway, 2005b). Nucleoplasmic reticulum has also been shown to function in cellular processes such as transcription, DNA repair and lipid metabolism (Drozdz and Vaux, 2017). We therefore assayed the sensitivity of the PC synthesis pathway-related mutants to phleomycin-induced DNA damage. Phleomycin is a glycopeptide antibiotic of the bleomycin family which binds and intercalates DNA thus destroying the integrity of the double helix (Sleigh, 1976). Intriguingly, pct1Δ cells displayed growth defects in the presence of phleomycin but cho2Δopi3Δ mutant cells do not show any effect (Figure 3.13A). In concert with pct1Δ, phleomycin has a strong impact on cells that lack the enzyme catalysing the final step in the Kennedy pathway, cpt1Δ (Figure 3.13A). This suggests that the DNA damage sensitivity is due to the lack of this PC pool. Wild-type Pct1, and not the partial loss of function mutants, could rescue pct1Δ DNA damage sensitivity (Figure 3.13B). Taken together these preliminary data suggest a possible ancillary role of Pct1 in DNA damage protection. 151    Figure 3.13 Evaluation the potential of Pct1 (or PC synthesised from the Kennedy pathway) to protect yeast cells against DNA damage. (A) Wild-type (BY4742), pct1Δ, cho2Δopi3Δ, or cpt1Δ yeast cells were grown to exponential phase and serial dilutions of liquid cultures were spotted onto YPD plates or YPD plates supplemented with phleomycin 2.5 g/mL. (B) Wild-type (BY4742), pct1Δ, or pct1Δ cells transformed with a CEN/HIS3 plasmid expressing Pct1 or mutant Cmut and 4m were used to investigate the cell viability. Cells were grown to exponential phase and serial dilutions of liquid cultures were spotted onto YPD plates or YPD plates supplemented with 2.5 g/mL phleomycin or 5.0 g/mL phleomycin. Data are representative of at least three independent experiments. 152    3.5 Discussion Pct1 is largely confined to the nucleus where it senses ‘surface packing defects’ generated primarily by the presence of glycerolipids with small uncharged headgroups, primarily PE and DAG, rather than PC itself. In other words, a relative paucity of PC is what is sensed. A relative increase in conical phospholipid levels within the membrane is thought to result in increased SCE stress and was shown to promote Pct1 membrane-binding and feedback into the regulation of PC synthesis. 3.5.1 Proposed membrane targeting mechanism of Pct1 The data in this thesis suggest a model in which PC synthesis is primarily regulated by intranuclear Pct1 which uses a nascent amphipathic helical domain to sense changes in the surface topology of the inner nuclear membrane. By associating with the membrane, Pct1 temporarily alleviates membrane SCE stress until adequate PC is generated to stabilise the membrane. This raises the question of how the properties of the Pct1 amphipathic helices ensure nuclear membrane binding. It is known that the membrane-induced amphipathic helix of CCT has conformational plasticity that enables it to bind membranes and initiate its enzyme activity, as well as to remove itself from membranes and inhibit catalysis (Ding et al., 2012). Here, this study did not explore the specific properties of the amphipathic helices in much detail. Instead, we studied a deletion mutant of the M-domain, Pct1GPd, which significantly impaired membrane- binding even in choline-depleted medium. 3Δ (cho2Δopi3Δpct1Δ) cells show reduced viability when expressing Pct1GPd. We also tried to prevent membrane-binding by changing four hydrophobic residues to negatively-charged residues, named Pct14m. Pct14m only partially disrupts membrane-binding and has a subtle effect on the FOA survival assay. These data are consistent with a substantial body of in vitro work (Cornell, 2016; Cornell and Ridgway, 2015; Ramezanpour et al., 2018) indicating that the binding of Pct1 to membranes through the M- domain is essential to activate its enzyme activity fully. 3.5.2 Conical phospholipids involved in PC regulation Conical phospholipids such as PE, DAG and PA impose negative curvature strain and packing defects in lipid bilayers. In vivo, PE and DAG serve as substrates for PC synthesis via the PE methylation or Kennedy pathways respectively. Although the enhancement of PC synthesis has 153    been reported to be the primary consequence of an increase in cellular DAG levels (Araki and Wurtman, 1997; Barbosa et al., 2015), DAG is present at low levels under normal circumstances. This could presumably be at least in part because their highly conical shape has a pronounced effect on membrane curvature stress. This study also found a defence mechanism against DAG accumulation by triggering its conversion into PC (Figure 3.9.2A (i); DAG decrease after choline supplementation) and probably against PA augmentation by rapidly converting PA to DAG or membrane glycerophospholipids as an alternative. This process is linked to PC synthesis and provides an exquisite membrane topological sensing mechanism which is consistent with in vitro data using liposomes with a broader spectrum of PCYT1A activating lipids (Cornell, 2016). In agreement with this observation, forcing cells to increase DAG levels by preventing the synthesis of TAG and the subsequent storage of TAG in lipid droplets, done in a yeast strain which lacks all enzymes required for neutral lipid synthesis (TAG and sterol ester), results in persistent membrane association of Pct1-GFP and increased PC synthesis (Barbosa et al., 2015). 3.5.3 Mammalian PCYT1A also associated with the nuclear envelope Not only yeast Pct1 but also mammalian PCYT1A associated with the nuclear envelope in response to cell growth signals for membrane biogenesis in several tissues. PC is essential in highly proliferating cells at key developmental stages, such as extensive ER membrane proliferation and expansion for immunoglobulin or hormone secretion (Fagone et al., 2007; Young, 1967). PC synthesis is increased for the synthesis of lipoproteins, bile and lung surfactant (Cole et al., 2012; Cornell and Ridgway, 2015; Exton, 1994; van der Veen et al., 2017). Recently, my host group has demonstrated that PCYT1A localises to the nuclear envelope in cells/tissues that are affected in PCYT1A-linked human diseases, namely bone growth plates, retina, liver and adipose tissue (Hoover-Fong et al., 2014; Payne et al., 2014; Testa et al., 2017; Wong, 2014; Yamamoto et al., 2014) where, accordingly, PCYT1A activity is likely to be physiologically important. In sections of femoral growth plates, PCYT1A immunostaining on the nuclear envelope was confined to the hypertrophic zone (Figure 3.14A) where the cellular cytoplasmic volume increases 5-fold (Brighton et al., 1973). In the eyes, nuclear rim staining was apparent in the nuclei of photoreceptors (Figure 3.14B). The multiple membrane layers of photoreceptors house light inducible rhodopsin, which ultimately leads to protein and membrane turnover (Ding et al., 2015; Molday and Moritz, 2015). PC synthesis is known to be highly upregulated in these cells, particularly during photoreceptor development. In the liver, where PC is consistently 154    needed for lipoprotein synthesis, PCYT1A localised to the nuclear envelope and furthermore its membrane localisation was negated in mice unable to synthesise lipoproteins (Figure 3.14C). This is consistent with a need for increased PC synthesis to sustain lipoprotein production. In adipocytes, PCYT1A has not been observed on the surface of lipid droplets in mouse tissues or in differentiating adipocytes, though a transient nuclear envelope localisation was seen in both 3T3- L1 (Aitchison et al., 2015) and in murine OP9 adipocytes (Figure 3.14D). These data suggest that PCYT1A is activated in the tissues where PC is required to maintain biomass production or during tissues of membrane remodelling. My host group has also observed that PCYT1A is intra- nuclear in lung epithelial A549 cells (data not shown), in which it was previously suggested that PCYT1A could translocate into the cytoplasm, so we suspect that PCYT1A may be intra-nuclear in all mammalian cell types. 155    Figure 3.14 Immunostaining of endogenous PCYT1A in selected mouse tissues. (A) (i) In the femoral growth plate of 15-day old mice, PCYT1A localises to the nuclear membrane of chondrocytes in the hypertrophic (HZ) but not in the resting (RZ) or proliferative (PZ) zone of growth plates; (ii) Zoomed-in images of the indicated field in (i). (B) (i) Immunostaining of adult mouse retina indicates that PCYT1A localises to the nuclear membrane in the outer nuclear layer. SC, sclera; CH, choroid; RPE, retinal pigmented epithelium; OS, outer segments of rods and cones; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner 156    nuclear layer; IPL, inner plexiform layer; GCL; ganaglion cell layer; (ii) Zoomed-in images of the indicated field in (i). (C) In ad libitum chow-fed adult mice, PCYT1A localises to the nuclear membrane in wild-type (WT) but not in Mtp knockout hepatocytes, which have impaired lipoproteins synthesis. (D) Upon adipogenic induction, PCYT1A translocates to the nuclear membrane in OP9 cells. Lipid droplets (LDs) stained with BODIPY (green). D0-3 indicate day after onset differentiation. Scale bar, 20 m. All results in this figure were the work of Dr KoiNi Lim. 3.5.4 Influence of glucose availability on Pct1 membrane-binding In vivo studies suggest that Pct1 is associated with membranes with low affinity, which presumably contributes to its relatively rapid fall-off as observed upon choline supplementation in pct1Δ, cho2opi3pct1Δ, or in glucose starved yeast. Exactly why glucose starvation results in Pct1-GFP translocation off the nuclear membrane remains unclear. My studies of severel putative mechanisms did not reveal a definitive answer. The fact that translocation was delayed in the snf1Δ strain suggests that glucose deprivation may be involved but the exact mechanism will require further characterisation. Snf1 is required for the adaptation of cells to glucose limitation and for growth of alternative carbon sources, such as ethanol, which is used by yeasts after the diauxic shift (Hedbacker and Carlson, 2008; O'Donnell et al., 2015). In other words, the snf1∆ strain causes problems of cell derepression, resulting in growth defects in non-fermentable carbon sources. Because metabolic adaptation taking place at the diauxic shift is compromised in snf1Δ, many events fail to take place like mitochondria activation, stress response induction, or cell cycle arrest. snf1Δ cells were indeed observed to cause elevated superoxide anions (inefficient respiratory activity coupled to deficient stress response) and a less frequently arrested G0/G1 stationary phase (Weinberger et al., 2010). In yeast, some genes seemed to be regulated by Snf1 and involved in PC biosynthesis. SNF1 shows negative genetic interactions with CHO2 and OPI3 (necessary for PC synthesis through the methylation pathway), but not the Kennedy pathway (Surma et al., 2013); additionally, Snf1 may regulate INO4 (part of a transcription factor required for phospholipid synthesis) at the transcription level (Chumnanpuen et al., 2012). Although the mechanism behind the link between glucose sensing and PC synthesis is yet to be clearly defined, the Snf1-mediated metabolic rearrangement during glucose deprivation may lead to changes in some gene expressions that affect PC levels, which in turn affects Pct1 activity/membrane detachment. One can also not rule out the impact of glucose-deprivation in inhibiting phospholipid synthesis and inducing neutral 157    lipid (TAG and steryl esters) accumulation. We have not yet documented the PL profile of glucose starved cells. This is an experiment that should be done in the future. 3.5.5 Scientific rationale for the intranuclear localisation of Pct1/PCYT1A Why is Pct1/PCYT1A an intranuclear enzyme? One possibility is that yeast and mammalian cells attempt to confine the enzymes used in the CDP-choline pathway to the nucleus, such as the rate- limiting enzyme Pct1/PCYT1A and the following terminal enzyme cholinephosphotransferase. Two cholinephosphotransferases have been identified as the final step catalysts  including CPT1 and EPT1 (CEPT1), the function of which is responsible for forming phosphatidylcholine from CDP-choline (Hjelmstad and Bell, 1990; Hjelmstad and Bell, 1991). Although CPT1 may localise to the Golgi, EPT1 (CEPT1) has been found on the ER as well as the nuclear envelope. Moreover, a pool of CEPT1 translocates to the nuclear membrane and reportedly co-localises with PCYT1A in response to exogenous fatty acid activation of PC synthesis (Henneberry et al., 2002). The lipin proteins, which catalyse the formation of DAG required for the terminal step of PC synthesis, also show a conserved nuclear pool from yeast to mammalian cells (Siniossoglou, 2013). The PC generated in the nucleus may then be distributed to the endomembrane system through vesicular or non-vesicular pathways since the outer nuclear membranes are continuous with the cytoplasmic ER network. The final step in PC synthesis may also occur in the cytoplasm since CDP-choline is water soluble. In this setting, the presence of Pct1/PCYT1A in the spatially separated nucleus may be responsible for detecting membrane-packing defects and regulating genes that encode PC biosynthesis enzymes on the transcriptional level. This process is likely to operate cooperatively with transmembrane sensors such as Mga2 and Sre1 (Burr et al., 2017; Covino et al., 2016), which are primarily involved in responding to changes in intramembranous fatty acid saturation. By sensing membrane properties such as fluidity and curvature elastic stress, which affect certain membrane protein activities and feedback into transcriptional regulation, cells can maintain these dynamic processes within relatively tight boundaries under conditions in which phospholipid composition is outside proper range. It is also possible that Pct1/PCYT1A may regulate the proliferation of nuclear envelope invaginations, also named the nucleoplasmic reticulum, which may support nuclear structure, affect the spatial organisation of chromatin, and participate in important cellular events such as 158    DNA replication and DNA damage repair (Broers et al., 1999; Drozdz and Vaux, 2017; Khadija et al., 2015; Lemaitre and Bickmore, 2015). Contemporary studies have demonstrated that PCYT1A may play a key role in promoting  the development of the nucleoplasmic reticulum (Lagace and Ridgway, 2005b).  To determine whether PC synthesis was also involved in the repair of DNA double strand breaks, we ectopically deleted enzymes of the Kennedy pathway enzymes or PE methylation pathway and incubated them with phleomycin, a DNA double-strand breakage-inducing reagent. Our earlier studies demonstrated that Pct1 or other Kennedy pathway deletion strains are incapable of proliferating on plates that contain phleomycin, but this concentration of phleomycin does not disturb PE methylation deletion strains. It would be of particular interest to understand if DNA repair is affected by Pct1 insertion into the INM, resulting in the appropriate change of membrane curvature to support tubulation of the nucleoplasmic reticulum, or another invagination structure more suitable for the DNA repair system. Future studies might include the use of super-resolution microscopy to examine the nucleoplasmic reticulum structure in cells affected by PCT1 mutations or clinically observed PCYT1A mutations. It will also be of interest to: i) test a broad range of DNA damage inducing reagents in our yeast lines (WT and mutants), and ii) see if PCYT1A in mammalian cells is needed for DNA damage repair e.g. we have used siRNA to KD Pcyt1a in various cell lines. 3.5.6 Conclusions In summary, this study proposes two models in which (i) membrane-bound Pct1 facilitates increased PC synthesis. This allows yeast cells to proliferate when nutrients are not restricted and (ii) PC synthesis is primarily regulated by intranuclear Pct1/PCYT1A which utilizss a nascent amphipathic helical domain to sense the surface topology of the inner nuclear membrane. These models are broadly consistent with the structural and in vitro biochemical properties of Pct1/PCYT1A; furthermore, this search for the regulatory mechanism facilitates future progress in understanding how cells detect and regulate membrane PL composition.     159    Chapter 4 The Importance of Nuclear Compartmentalisation of Pct1 4.1 Abstract The rate-limiting enzyme for Kennedy pathway, Pct1 in yeast cells or PCYT1A in mammalian cells, is specifically localised to the nucleus whereas the ER compartment is the major site of final PC synthesis. An N-terminal basic stretch of 60PRKRRRL66 has previously been identified in Pct1 as the putative nuclear localisation signal (NLS); however, the biological necessity of nuclear compartmentalisation is unknown. In this chapter, the role of Pct1 in the cytoplasm was investigated using NLS mutants. When four basic residues at amino acids 62–65 at the NLS locus were substituted with alanine, the nuclear targeting of Pct1 was abrogated and instead the NLS mutant localised to the plasma membrane and/or lipid droplets. Interestingly, the NLS mutant showed somewhat similar membrane on/off localisation to that of the wild-type enzyme in the 3∆ strain insofar as it associated with lipid droplet in choline starved conditions and then “fell off” when cells were supplemented with choline. Characterisation of the NLS mutant through lipidomic analysis and curvature stress prediction also revealed that the NLS mutant was enzymatically active and able to regulate membrane PC homeostasis comparably to wild-type Pct1. These results show that Pct1 nuclear localisation is not essential for its catalytic function, but as the nuclear compartmentalisation is conserved across evolutionarily diverse species, it is bound to play an important role. 4.2 Introduction 4.2.1 Classical nuclear localisation signals Nuclear localisation signals (NLS) are amino acid consensus regions in nuclear targeted proteins that are recognised by importin-α. Classical NLS can be classified as either monopartite NLS enriched in basic residues, or bipartite NLS comprised of two stretches of basic-rich residues separated by a 10–12-amino acid linker (Kosugi et al., 2009).  For example, the Simian Virus 40 (SV40) large T-antigen NLS is the first discovered classical monopartite signal (126PKKKRRV132) (Kalderon et al., 1984). The nucleoplasmin NLS is a prototype of the ubiquitous bipartite signal (155KRPAATKKAGQAKKKK170)  (Dingwall et al., 1988). A putative consensus sequence of the 160    monopartite NLS has been defined as K(K/R)X(K/R), suggesting monopartite NLS requires a lysine in the P1 position, followed by lysine or arginine in positions P2 and P4 (Hodel et al., 2001).  A putative sequence of the bipartite NLS is (K/R)(K/R)X10–12(K/R)3/5, where (K/R)3/5 indicates at least three of either lysine or arginine of five consecutive amino acids, in which X10– 12 represents the linker region that can be tolerant to amino acid substitution (Kosugi et al., 2009; Robbins et al., 1991). 4.2.2 Overview of nuclear import pathways Transport of macromolecules between the nucleus and the cytoplasm occurs through nuclear pore complexes (NPCs), cylindrical proteinaceous structures embedded within the nuclear envelope (Allen et al., 2000; Fahrenkrog and Aebi, 2003). The pore complex is built up from protein subunits termed nucleoporins (Nups), one third of which contain phenylalanine-glycine (FG)-rich repeats that form the central channel of the NPCs (Onischenko et al., 2017). The NPCs are about 120 nm in diameter and 70 nm thick (Stewart et al., 2001), and this channel is large enough to allow passive diffusion of ions or proteins smaller than 40 kDa (Christie et al., 2016; Paine et al., 1975). Thus, larger proteins with functions in the nucleus use an appropriate targeting signal for passage through the NPCs. Translocation of macromolecular cargo through NPCs requires additional carrier proteins. These carriers are collectively referred to as the karyopherin superfamily, which can be classified into alpha (α) and beta (β) subfamilies (Chook and Blobel, 2001). In classical nuclear import, the link between the β-karyopherin (importin-β) and the cargo is mediated by the adaptor protein importin-α (Gorlich et al., 1995).  Nuclear protein import proceeds in an energy-dependent translocation step, which is mediated by a small nuclear Ras-related GTP binding protein, Ran. Ran cycles between an active GTP-bound state and an inactive GDP-bound state (Quimby and Dasso, 2003).  Intriguingly, GTP-bound Ran is asymmetrically enriched in the nucleus while GDP-bound Ran is distributed in the cytoplasm (Joseph, 2006). This unbalanced pattern is modulated by compartmentalised regulatory proteins, mainly the Ran guanine nucleotide exchange factor (RanGEF) in the nucleus and the Ran GTPase-activating protein (RanGAP) in the cytoplasm (Smith et al., 2002). RanGEF triggers GDP exchange with GTP to form RanGTP in the nucleus, whereas in the cytoplasm RanGAP catalyses the hydrolysis of Ran from GTP to GDP.  The RanGTP/RanGDP gradient is thought to allow Ran to impart directionality on 161    nucleocytoplasmic transport (Lange et al., 2007)  as its binding regulates the loading and unloading of cargo from importin receptors.   In the classical nuclear import pathway, importin-α recognizes and binds cargo in the cytoplasm in the absence of RanGTP. Importin-β links with importin α-cargo and mediates the trimeric complex docking at the NPCs  via interaction with Nups (Strom and Weis, 2001). When the trimeric complex reaches the nucleus, it is dissociated by RanGTP-binding to the importin-β. This binding results in importin-β becoming locked in a conformation that prevents the flexibility required to bind to importin-α, thus enabling the release of importin-α and cargo complex in the nucleus (Christie et al., 2016). The cargo is then delivered into the nucleus with the action of an autoinhibitory region on the importin β-binding (IBB) domain of importin-α (Harreman et al., 2003). Finally, importin-α is recycled to the cytoplasm  in a complex with the RanGTP-bound export receptor CAS (Lange et al., 2007; Stewart, 2006). Also, the importin β-RanGTP complex is recycled in the cytoplasm, where RanGAP hydrolyses the GTP leading to the dissociation of RanGDP and freeing the importins for further rounds of import (Stewart, 2006). 4.2.3 Interaction of importin- with the NLS Structural studies (Christie et al., 2016; Conti et al., 1998) reveal that importin-α is composed of a flexible N-terminal auto-inhibitory (IBB) domain that also mediates binding to importin-β and a large C-terminal NLS-binding domain that contains ten armadillo (ARM) motifs. The continuous stacking of the regular ARM repeats generates a major and a minor classical NLS-binding site with a concave surface.  An in vitro binding assay (Kosugi et al., 2009) with GFP fusions with various NLS and recombinant importin-α variants showed that monopartite NLS was specifically bound to the major binding site of importin-α, whereas the bipartite NLS was bound to both sites: the larger stretch of basic residues bound to the major binding site and the smaller stretch of basic residues bound to the more C-terminal minor binding site. Notably, the IBB domain within importin-α consists of amino acid residues KRR that mimic the basic NLS (Harreman et al., 2003). Once importin-α is released by importin-β in the nucleus, the IBB region can fold over and bind to the major NLS-binding pocket on the importin-α (Kobe, 1999; Lange et al., 2007), as a result, facilitating the release of NLS cargo within the nucleus. 162    4.2.4 Evolutionary analysis of the PCYT1A and PCYT1B genes The CCT sequence is highly conserved in eukaryotes, especially in the catalytic domain (Cornell and Ridgway, 2015). The budding yeast expresses one CCT enzyme, Pct1, while higher eukaryotes express two isoforms: PCYT1A (also referred to as CCTα in mammals; CCT1 in Drosophila) which is ubiquitously expressed and PCYT1B (CCTβ in mammals; CCT2 in Drosophila) which displays restricted tissue distribution and lacks the N-terminal nuclear localisation signal (NLS) motif found in PCYT1A. The human PCYT1A and Drosophila CCT1 proteins are 53.4% identical and 69.8% similar to each other (protein alignment in EMBL-EBI, https://www.ebi.ac.uk/Tools/psa/emboss_needle/) and also share significant similarity with those of rats. As even the most primitive metazoan Trichoplax adherens contains two CCT genes, the initial split into two paralogues seems to have occurred very early in evolution. In Figure 4.1, the phylogenic tree of PCYT1A and PCYT1B indicates that the two paralogues evolved together and remain more similar to each than among species. To identify related homologues of either PCYT1A or PCYT1B, the Pfam database (http://pfam.xfam.org/family/PF01467) was used; many homologous proteins from Bacteria and Archaea were listed in the database but none of them have been annotated as CCT thus far. A detailed biochemical and structural study (Kwak et al., 2002) identified an enzyme with CCT activity in Streptococcus pneumonia that is evolutionarily unrelated to the eukaryotic ones and has close homologues in many Bacteria and Archaea. Figure 4.1 A CCT phylogenetic tree. The CCT phylogenetic tree based on the alignment of sequences comprising the catalytic and M- domain of CCT from selected Metazoan, fungal and plant representatives. UniProt sequence IDs 163    from top to bottom: B3RI62, B3RI63, P49585, Q9Y5K3, Q8IU09, P49583, Q7K4C7, Q9W0D9, Q9ZV56, F4JJE0, P13259; Labels A and B distinguish the paralogues in one species. Alignments were carried out with PROMALS (http://prodata.swmed.edu/promals/promals.php) and the tree was calculated with PhyMl algorithm with aLTR test as implemented in Phylogeny.fr package (http://www.phylogeny.fr). The branch length is proportional to the number of substitutions per site. This figure was generated by Dr Vladmir Saudek, University of Cambridge. 4.2.5 Prediction and alignment of classical nuclear localisation signals in human PCYT1A and yeast Pct1 proteins To pinpoint the presumptively classical NLS motif within human PCYT1A and S. cerevisiae Pct1, their protein sequences were scanned respectively through the online cNLS mapper (http://nls- mapper.iab.keio.ac.jp/cgi-bin/NLS_Mapper_form.cgi), which predicts the presence of importin-α dependent NLS. Both monopartite and bipartite NLS motifs were identified in the two-query sequences (Figure 4.2A). An SV40 T-antigen NLS-like monopartite motif was noticed in the N- terminal part of the PCYT1A protein (10NARKRRKEAP19) and PCT1 protein (59TPRKRRRLTKEF70) with high scores (Figure 4.2A (i)), suggesting strong NLS activities. A pairwise alignment of the predicted monopartite NLS motifs of human PCYT1A and yeast Pct1 in EMBL-EBI (https://www.ebi.ac.uk/Tools/psa/emboss_needle/) indicated that, with a 33.3% identity and 41.7% overall similarity, the four-residue RKRR is conserved among these two species (Figure 4.2B (i)). The predicted monopartite NLS motif in human PCYT1A overlaps with the one found in rats PCYT1A (8KVNSRKRRKEVPGPNGATEED28) (Wang et al., 1995); furthermore, the consensus four basic residues RKRR are also conserved. This suggests that the predicted monopartite motif is essential for nuclear localisation of the PCYT1A/Pct1. A bipartite signal (Figure 4.2A (ii)) was also noticed in the N-terminal part of human PCYT1A (10NARKRRKEAPGPNGATEEDGVPSKVQRC37) and S. cerevisiae Pct1 (25 KKNKNKRQREETEEQDNEDKDESKNQDE52). Although the bipartite NLS score in PCYT1A protein is up to 10.4, the predicted score in Pct1 is only 5.6, which implies it is a relativly weak NLS signal. The pairwise alignment of the predicted bipartite NLS motifs showed that they are 23.5% identical and 32.4 % similar to each other, suggesting that the highly conserved functional positions lie outside these contiguous regions. Nevertheless, the consecutive basic residues within PCYT1A 10NARKRRK16 and Pct1 25KKNKNKR31 are partially matched (Figure 4.2B (ii)), and one cannot rule out the possibility that they are monopartite NLS motifs. 164    A (i) Species/ Protein Position Predicted monopartite NLS Score Human PCYT1A 10-19 NARKRRKEAP 9.0 Yeast Pct1 59-70 TPRKRRRLTKEF 15 (ii) Species/ Protein Position Predicted bipartite NLS Score Human PCYT1A 10-37 NARKRRKEAPGPNGATEEDGVPSKVQRC 10.4 Yeast Pct1 25-52 KKNKNKRQREETEEQDNEDKDESKNQDE 5.6 B (i) Human PCYT1A 10 NARKRRKEAP-- 19 ..||||:... Yeast Pct1 59 TPRKRRRLTKEF 70 (ii) Human PCYT1A 10 --NARKRRKEAPGPNGATEEDGVPSKVQRC---- 37 |..||::| .|||.....|.:.. Yeast Pct1 25 KKNKNKRQRE------ETEEQDNEDKDESKNQDE 52 Figure 4.2 Pairwise sequence alignments of the presumptive NLS regions of human PCYT1A and yeast Pct1 proteins. (A) cNLS mapper results of the (i) predicted monopartite NLS and (ii) predicted bipartite NLS in human PCYT1A and yeast Pct1. The algorithm was applied with a cut-off score of 5.0. Higher scores suggest stronger NLS activities. Only the hit with the top score is shown. (B) Pairwise sequence alignment of the putative (i) monopartite NLS regions and (ii) bipartite NLS regions from human PCYT1A and yeast Pct1. Sequences were aligned with online EMBL-EBI software. '-' for a mismatch or a gap, '|' for an identity where both sequences have the same residues, '.' for any small positive similarity, and ':' for residues with similar side chains. 165    4.3 Aims Pct1 membrane-binding is required for the stimulation of its catalytic activity. It is known that Pct1 partitions between soluble and membrane-associated forms in the nucleus. However, its product CDP-choline is water-soluble and the localisation of its downstream enzyme remains unclear. As PCYT1B has a distinct catalytic function despite being cytosolic, it can be expected that yeast Pct1 will be able to retain its catalytic activity outside the cell nucleus. The aims of this chapter were: I. To characterise putative nuclear localisation signals and to determine the regions of Pct1 critical for nuclear localisation. II. To evaluate the effects of disrupting Pct1 nuclear targeting by imaging and lipidomic analysis. 4.4 Results 4.4.1 The N-terminal basic stretch 60PRKRRRL66 is required for Pct1 nuclear localisation Pct1 contains two highly basic regions at its N-terminus, 25KKNKNKR31 and 60PRKRRRL66, rich in lysine and arginine residues, which are thought to be highly homologous to consensus nuclear localisation signals (MacKinnon et al., 2009). To determine whether these basic stretches are required for Pct1 nuclear localisation, several strategies for mutagenesis were performed (Figure 4.3.1A); specifically, point mutations were introduced to replace the basic amino acids in either one or both of the NLSs, or different regions of the putative NLS were deleted. Consistent with the inner nuclear membrane targeting of Pct1, we found that mutating a stretch of four basic residues (60PRKRRRL66; henceforth called NLS mutant 1, NLSm1) to alanines led to its mislocalisation to the cell periphery in both exponential and PDS phase cells (Figure 4.3.1B). Point mutations within the second basic stretch (25KKNKNKR31; henceforth called NLS mutant 2, NLSm2) did not change Pct1 distribution and a double NLSm1m2 mutant localised similarly to the NLSm1 mutant (Figure 4.3.1B). As expected, a mutant in which 60PRKRRRL66 of Pct1 was deleted (henceforth called NLS1 deletion, NLS1del) showed the same localisation as NLSm1 (Figure 4.3.1B). The steady state expression of these NLS-modified proteins was determined by immunoblotting (Figure 4.3.2A). They were all stably expressed, except for a mutant in which 166    the whole N-region was deleted (henceforth called NLS truncation, NLStrun) which was expressed at a lower level (Figure 4.3.2B). A diffuse NLStrun-GFP signal was observed throughout the cells (Figure 4.3.1B) which we assume reflects expression and dispersion of degradation products (Figure 4.3.2B). To provide further confirmation that the mutation in NLSm1 blocks Pct1 nuclear import, a construct NLSm4m was generated where the NLSm1 was combined with the 4m (4m mutant partially disrupts the nuclear envelope-binding of Pct1; details in section 3.4.12; Figure 4.3.3A). The NLSm4m mutant was expected to be restricted to the cytoplasm and additionally, unable to bind to membranes in the cytoplasm. Indeed, this combination restricts Pct1-NLSm4m localisation to the cytosol without any obvious membrane- binding (Figure 4.3.3B). These data suggest that Pct1-NLSm1-binding to the cell periphery is through the membrane-binding domain. The expression of Pct1-NLSm4m was largely stable although a small pool of the degraded NLSm4m could be detected in immunoblots (Figure 4.3.3C). Together, these results suggest that the basic stretch 60PRKRRRL66 is the main signal responsible for Pct1 nuclear localisation. 167      168    Figure 4.3.1 Identification of specific nuclear localisation signal (NLS) sequences in Pct1. (A) Schematic illustration of the Pct1 domain organisation indicating the mutations in the NLSm1, NLSm2, NLSm1m2, NLS1del and NLStrun Pct1 mutants. N-region, N-terminal region; C-domain, catalytic domain; M-domain, membrane-binding domain;  P-region, C-terminal phosphorylated region. Position of substitution mutations are indicated with blue bars and basic residues of the putative NLS motifs were mutated to alanine (blue). The targeted deletions are indicated with crossed boxes. (B) pct1∆ cells expressing the GFP-tagged wild-type Pct1 (WT) or NLS mutants (NLSm1, NLSm2, NLSm1m2, NLS1del or NLStrun) were grown to exponential phase (Exp) or post-diauxic shift phase (PDS) and imaged as described in Methods section 2.5.1. Differential interference contrast (DIC) images are shown for context. Scale bar, 5 m. Images are representative of at least three independent experiments. (C) All panels are zoomed in images from the boxed regions of (B) NLSm1-GFP Exp or PDS. Scale bar, 5 m. 169      Figure 4.3.2 Immunoblotting of lysates from pct1∆ cells expressing GFP-tagged wild-type Pct1 or NLS mutants. (A) Representative immunoblotting of cell lysates from pct1∆ yeast cells expressing WT, NLSm1, NLSm2, NLSm1m2 or NLS1del Pct1-GFP at the indicated growth phase using anti-GFP antibody. GAPDH was used as a loading control. Arrow indicates the expected size of the intact protein. (B) Representative immunoblotting of cell lysates from pct1∆ yeast cells expressing WT Pct1-GFP (WT), NLStrun Pct1-GFP (NLStrun) and empty plasmid (EV) at the indicated growth phase using anti-GFP antibody. Two independent NLStrun plasmid clones #1 and #2 were used. GAPDH was used as a loading control. The upper arrow indicates the expected size of the WT Pct1-GFP, while the lower arrow indicates the expected size of the NLStrun Pct1-GFP. GFP alone typically runs at 27 kD so we assume that the prominent band at about 37 kD represent a partially degraded form of NLStrun Pct1-GFP. 170      171    Figure 4.3.3 Point mutations in the M-domain known to disrupt Pct1 membrane-binding ability induces cytosolic localisation of Pct1-NLSm1. (A) Schematic illustration of the Pct1 NLSm4m (NLSm4m) mutant. Basic residues of the putative NLS motif were mutated to alanine (blue). Substitution mutations which change hydrophobic to charged amino acids in the M-domain are indicated with purple bars. N-region, N-terminal region; C-domain, catalytic domain; M-domain, membrane-binding domain; P-region, C-terminal phosphorylated region. (B) 3∆ (cho2Δopi3Δpct1Δ) cells expressing the GFP-tagged wild-type Pct1 (WT), 4m or NLSm4m were grown to the indicated growth phase with (+) or without (-) excess choline and imaged as described in Methods section 2.5.1. Differential interference contrast (DIC) images are shown for context. Scale bar, 5 m. (C) Representative GFP immunoblotting of cell lysates from 3Δ yeast cells expressing WT Pct1-GFP (Pct1), Cmut- GFP (Cmut) and NLSm4m-GFP (NLSm4m) at the indicated growth phase with choline supplementation. GAPDH was used as a loading control. Arrow indicates the expected size of the intact protein. Data are representative of at least three independent experiments. 172    4.4.2 Mutation in the nuclear localisation sequence of Pct1 resulting in plasma membrane localisation The specific peripheral Pct1-NLSm1-GFP signal resembles the localisation of yeast cortical endoplasmic reticulum (cER) or plasma membrane (PM) proteins. In yeast, the cER is primarily aligned along the PM through three conserved protein families (Manford et al., 2012), which have been identified as ER-PM tethers: i) the three calcium and lipid-binding (TriCalBins; Tcb) proteins, ii) the suppressor of choline sensitivity 2 (Scs2) proteins and iii) the increased sodium tolerance (Ist2) protein. To assess whether NLSm1 is resident on the cER or the PM, a ∆tether strain was used (Figure 4.4A; (Manford et al., 2012)), in which six proteins among the three protein families that tether the cER to the PM were deleted, causing the collapse of the cER and its accumulation in the cytoplasm. The expression of Pct1-NLSm1-GFP in the Δtether strain was comparable to that seen in the corresponding wild type strain (Figure 4.4C). In the Δtether strain, Pct1-NLSm1 mutant still localised to the cell periphery despite the collapse of the cER (Figure 4.4B). This suggests that Pct1-NLSm1 is associated with the PM and not the cER. This may be due to the fact that the membrane-binding domain of Pct1 is particularly enriched with positively charged amino acids and this may facilitate its interaction with phosphatidylserine (PS), a PL with a net negative charge that is enriched at the inner leaflet of the PM (Daleke, 2003; Fairn et al., 2011). It is particularly worth noting that the PS content of the PM is similar to that of ER (Yang et al., 2018). However, the PS is specifically enriched on the cytosolic face of the PM, whereas in the ER it might be mostly on the luminal side (Fairn et al., 2011); thus, the cytosolic Pct1-NLSm1 tends to bind to the PM, not the ER. 173      174    Figure 4.4 Nuclear localisation signal mutant of Pct1 mislocalises to the plasma membrane. (A) Schematic of the PM-ER tethering in S. cerevisiae. (adapted from Manford et al., 2012). Three families of six integral ER proteins were identified as ER-PM tethers: 1) Tcb (tricalbins) family including proteins Tcb1, Tcb2 and Tcb3. The tricalbins contain transmembrane domains in their N-termini and C2 domains in their long cytoplasmic C-termini (Schulz and Creutz, 2004); 2) Scs2 and its homolog Scs22 in the VAP (vesicle-associated membrane protein-associated protein) family. The Scs2/22 contain a single transmembrane domain and a cytoplasmic major sperm protein (MSP) domain (Loewen et al., 2003; Loewen et al., 2007); 3) the putative member of the ANO/TMEM16 ion channel family Ist2. Ist2 protein is comprised of eight transmembrane domains and a polybasic domain in the C-terminus which associates strongly with the PM (Fischer et al., 2009; Manford et al., 2012). Loss of ER-PM tethers results in the separation of the ER and PM as well as the accumulation of internal ER structures. (B) GFP-tagged Pct1-NLSm1 and Sec63-mCherry (ER marker) were expressed in the ∆tether strain and the corresponding wild-type (WT) yeast strain, grown to exponential phase and imaged. White arrow marks the normal cortical ER in WT strain and arrowheads indicate the collapsed cortical ER in the ∆tether cells. Differential interference contrast (DIC) images are shown for context. Scale bar, 5 m. Images are representative of at least three independent experiments. (C) Representative GFP immunoblotting of cell lysates from Δtether yeast cells or corresponding wild-type yeast cells (WT yeast) expressing WT Pct1-GFP (WT) and Pct1-NLSm-GFP (NLSm1) in the exponential phase. GAPDH was used as a loading control. 4.4.3 The Pct1 NLS mutant is mobile at the plasma membrane but does not shuttle between the plasma membrane and the nucleus Considering that a small pool of the Pct1-NLSm1-GFP mutant can still be detected in the nucleus (Figure 4.3.1C), we next investigated whether Pct1-NLSm1 traffics between the PM and nucleus. Note that we cannot formally assess whether or not the intra-nuclear GFP signal is from a degraded product. To test the possibility that Pct1-NLSm1-GFP traffics into and out of the nucleus, FLIP analysis was performed and analysed with repeated bleaching of the whole nuclear region of yeast pct1Δ cells expressing Pct1-NLSm1-GFP. Repeated photobleaching of the nucleus did not change the PM GFP signal (Figure 4.5A), suggesting that significant amounts of Pct1-NLSm1 stably localised to the PM. Photobleaching of a similar sized region of the PM led to a prompt decrease of GFP signal in the PM, and a subsequent partial recovery of the GFP signal in the bleached region within about 400 sec (Figure 4.5B), indicating that Pct1-NLSm1 is mobile on the PM. Taken together, these data suggest that Pct1-NLSm1-GFP is mobile on the PM and that Pct1-NLSm1-GFP does not traffic into the nucleus. Considering that a small amount 175    of Pct1 targeted at the nucleus can still be observed, there may still be additional factors governing nucleocytoplasmic trafficking of Pct1 that need to be investigated.   Figure 4.5 FLIP analysis reveals that NLSm1-Pct1-GFP localises stably on the plasma membrane. (A) In pct1∆ cells expressing Pct1-NLSm1-GFP, repeated photobleaching of the signal in the nucleus does not result in a loss of fluorescence of Pct1-NLSm1-GFP on the plasma membrane, suggesting that little or no rapid trafficking of Pct1-NLSm1-GFP from the plasma membrane to the nucleus occurs. (B) In pct1∆ cells expressing Pct1-NLSm1-GFP, repeated photobleaching of GFP on part of the plasma membrane (PM) is followed by partial recovery of the fluorescence in the photobleached regions with no rapid change of PM fluorescence in other cells, indicating that Pct1 is mobile on the plasma membrane. Data are mean ± SD from three independent experiments (5-8 cells each). Arrows on x axes indicate time-points of each bleaching event. The GFP fluorescence reported here was corrected for the loss of fluorescence during imaging as described in the Methods section 2.5.3. 4.4.4 NLS mutant compensatation for the loss of the nuclear Pct1 in vivo We hypothesised that as long as NLS mutants have the ability to associate with membranes, they will be able to rescue 3Δ (cho2Δopi3Δpct1Δ) or 4Δ (cho2Δopi3Δpct1Δect1Δ) cells, which are unable to synthesise PC on their own. Figure 4.6.1A and Figure 4.6.1B show that various types of NLS mutants effectively rescued the 3Δ and the 4Δ cells like the wild-type Pct1, except for the NLStrun and NLSm4m mutants. The NLStrun mutant is unable to rescue PC deficient cells completely (Figure 4.6.1A), possibly because its expression is unstable as is evident from the significant amount of degradation in western blots (Figure 4.3.2B); NLSm4m has a relatively 176    weak rescue phenotype (Figures 4.6.1A and B) presumably because it cannot stably bind to the membrane and is therefore likely to be catalytically deprived. Pct1 catalytic function is tightly regulated by reversible membrane-binding by sensing PL levels, yet the necessity of nuclear localisation for regulating its activity remains unclear. To determine if this is an essential element in sensing and regulating PL levels, how 3Δ (cho2Δopi3Δpct1Δ) cells expressing Pct1-NLSm1-GFP respond to choline rescue after prolonged choline deprivation was monitored and the resulting PL changes in time-course experiments were analysed. Notably, 24 hr deprivation of choline leads to translocation of almost all Pct1-NLSm1-GFP to LDs (at 0.0 hr, Figure 4.6.2A). Since the cells lack choline at this time-point, a reasonable speculation is that Pct1-NLSm1 binds to the PC-deficient membranes but could not synthesise PC even on binding LDs due to the lack of substrate availability. Because of i) the failure of PC synthesis, and ii) more DAG diversion to TAG, the surface tension on LDs is expected to rise concomitantly, resulting in LD fusion and growth (Figure 4.6.2E; (Guo et al., 2008)); in addition, this might increase surface packing defects of LDs which may enhance Pct1 membrane insertion. Pct1 insertion may shield the neutral lipid core from the aqueous environment to some extent and help to alleviate surface tension. Consequently, exclusion from the nucleus facilitates the LD localisation of this mutant under PC-deficient conditions. Supplementation of choline in these cells leads to a gradual release of Pct1-NLSm1 from LD membranes (Figures 4.6.2A and B) that correlates with an increase in PC levels and the PC/PE ratio, and a decrease in mean LD size in this mutant (Figure 4.6.2C–E) with similar kinetics to that of wild-type Pct1 (Figure 3.8C–E). Pct1-NLSm1 also restores growth of the 3Δ cells to a similar extent to that of wild-type Pct1 (Figure 4.6.2F). To rule out any residual nuclear targeted NLSm1-Pct1 rescuing the survival of the 3Δ cells, the assay was repeated using the NLS1del mutant in which most of the amino acids comprising the NLS of Pct1 have been deleted. Pct1-NLS1del exhibited reduced nuclear localisation whereas Pct1-NLS1del also localised to the LDs in the absence of choline and localised to the PM in the presence of choline (Figure 4.6.2G), following a similar pattern to that of Pct1-NLSm1. Additionally, Pct1-NLS1del was still able to rescue the 3Δ phenotype as efficiently as the wild-type Pct1 and Pct1-NLSm1 (Figure 4.6.2H and Figure 4.6.1A). These data suggest that Pct1 can fully mediate PC synthesis not only from within the nucleus but also in the cytoplasm. 177    Figure 4.6.1 Pct1 NLS mutants have the same phenotype as wild-type Pct1 in yeast functional complementation assays. (A) 3Δ cells carrying Ycplac33-URA3-OPI3 were transformed with a CEN/HIS3 plasmid expressing either wild-type Pct1-GFP (WT) or mutant NLSm1, NLSm2, NLSm1m2, NLS1del, NLStrun, 4m, NLSm4m and Cdel (Pct1 catalytic domain deletion)-GFP. Cdel-GFP and empty CEN/HIS3 plasmid (EV) were used as negative controls. Cells were grown to exponential phase and serial dilutions of liquid cultures were spotted onto SC plates lacking histidine (-HIS) or plates supplemented with FOA and 1 mM choline (FOA + cho). (B) 4Δ cells carrying Ycplac33- URA3-OPI3 were transformed with a CEN/HIS3 plasmid expressing wild-type Pct1-GFP (WT) or mutant Cmut (three point mutations in the Pct1 catalytic domain.), NLSm1, NLSm1m2 and NLSm4m-GFP. Cmut-GFP and empty CEN/HIS3 plasmid (EV) were used as negative controls. Cells were grown to exponential phase and serial dilutions of liquid cultures were spotted onto SC plates lacking histidine (-HIS) or plates supplemented with FOA and 1 mM choline (FOA + cho). Data are representative of at least three independent experiments. 178      179    Figure 4.6.2 The Pct1 NLSm1 mutant is capable of synthesising PC and can compensate for the loss of the wild-type Pct1. (A-E) 3Δ yeast cells expressing Pct1-NLSm1-GFP were grown for 24 hr followed by addition of 1 mM choline. Cells were collected before (0.0 hr) and after choline supplementation at the indicated time points for imaging or lipidomics analyses. (A) Representative confocal microscopy images of Pct1-NLSm1-GFP (NLSm1) localisation. Lipid droplets (LDs) were stained as described in the Methods section 2.5.1. Merged fluorescent channels show co- localisation. Differential interference contrast (DIC) images are shown for context. Scale bar, 10 m. (B) Quantification of the Pct1-NLSm1 localisation shown in (A). LD, lipid droplets; PM, plasma membrane. Data are mean ± SD from three independent experiments; ~300 cells were counted from random fields at each time point per experiment. (C) PC and PE levels at the indicated time points are shown in nmoles (per 50 mg yeast) relative to internal standards. Data are mean ± SD from three independent experiments. (D) PC/PE ratio at the indicated time points. Data are mean ± SD from three independent experiments. One-way ANOVA, ***p < 0.001. (E) Measurement of LD area from cells imaged in (A) as described in Methods section 2.5.2. Data are mean ± SD from three independent experiments; ~600 LDs were counted in each condition. Two-tailed Student’s t-test, **p < 0.01. (F) Growth of 3Δ cells expressing WT Pct1 or Pct1- NLSm1 in medium supplemented with (+cho) or without (-cho) additional choline. (G) The NLS1 region (amino acids 60-66) of Pct1 was deleted in the Pct1-NLS1del-GFP mutant and expressed in 3Δ cells. Cells were grown without (-cho) or with (+cho) choline for 24 hr and stained to visualize lipid droplets (LDs), and imaged as described in Methods section 2.5.1. DIC images are shown for context. Scale bar, 5 m. (H) 3Δ cells carrying Ycplac33-URA3-OPI3 were transformed with a CEN/HIS3 plasmid expressing wild-type Pct1-GFP (WT), or mutant NLSm1- GFP and NLS1del-GFP. Empty CEN/HIS3 plasmid (EV) was used as a negative control. Cells were grown to exponential phase and serial dilutions of liquid cultures were spotted onto SC plates lacking histidine (-HIS) or plates supplemented with FOA and 1 mM choline (FOA + cho). Data are representative of at least three independent experiments. 180    4.4.5 Disruption of Pct1 nuclear localisation does not prevent phospholipid composition changes or membrane stored curvature stress (SCE) maintenance. To determine if changes in the phospholipid composition are affected by the NLSm1-Pct1 mutant, the lipidome in 3Δ cells expressing NLSm1-Pct1-GFP or WT-Pct1-GFP in choline-deprived conditions (at 0.0 hr) and at different time points following choline addition was analysed for comprehensive characterisation of fatty acid and headgroup changes in the PLs. As indicated in Figure 4.7A, both 3Δ NLSm1-Pct1 and 3Δ WT-Pct1 cells compensate for PC depletion (at 0.0 hr) by significantly increasing PE and also by decreasing PS lipid species. TAG levels were elevated in choline-deprived conditions and dropped on choline supplementation (Figure 4.7A), accompanied by corresponding changes in LD size (Figures 4.6.2E and 3.8E). 3Δ cells expressing either NLSm1-Pct1 or WT-Pct1 also compensate for the low PC by replacing monounsaturated FAs (PE (34:2) and PS (34:2)) with more saturated fatty acids (PE (32:1) and PS (32:1)) at time 0.0 hr (Figures 4.7C and D). Moreover, amounts of PC (34:2) increase when PC synthesis is restored (at 24 hr, Figure 4.7B). A narrow range of PSCE values from 0.1 to 0.2 was found in these two types of cells (Figure 4.7E) which is evidence of NLSm1-Pct1 contributing to preservation of membrane SCE stress throughout. Furthermore, 3Δ cells expressing NLSm1-Pct1-GFP show that equivalent processes regulate the lipid-driven SCE stress as in cells expressing WT-Pct1-GFP (Figure 4.7F). Taken together, 3Δ NLSm1-Pct1-GFP cells and 3Δ WT-Pct1-GFP cells manifest very similar behaviour in terms of lipidomic changes, PSCE and SCE stress, suggesting that Pct1 can regulate PC synthesis in response to changes in SCE stress whether sensed on the nuclear envelope, the lipid droplets or plasma membrane. Although the lipidomics and imaging data indicate that Pct1 is able to perceive changes in PLs on either membranes, its activation and PC synthesis still requires binding to the specific membranes. It was thus interesting to study the lipidomes of 3Δ cells expressing the Pct1-NLSm4m mutant (which is unable to translocate into the nucleus and partially loses membrane-binding ability) or the Pct1-4m mutant. Though the results shown here are from a single experiment, Figure 4.8A suggests that the range of PC/PE ratio of NLSm4m and 4m are both considerably lower than NLSm1 and WT, which normally reach values at 1.0–1.5 after 3.0 hr of choline supplementation 181    (Figure 4.6.2D and Figure 3.8D). This suggests that both the NLSm4m and 4m mutants cannot properly regulate PC synthesis presumably because of impaired membrane-binding. As NLSm4m and 4m still retain some catalytic activity, an increase in PC, increase in PS and also a decrease in TAG classes is observed on choline supplementation (Figure 4.8B). The adaptive changes in fatty acid saturation of PC, PE, and PS were concomitant with the generation of PC with a reduction in (32:1) and (34:1), and increase in (32:2) and (34:2) lipid species (Figure 4.8C–E). The lipid- driven SCE values after 24 hr choline supplementation (1.5–2.0 x 10-11 N, Figure 4.8G) were much higher than in 3Δ WT cells (~1 x 10-11N, Figure 4.7F), suggesting that NLSm4m and 4m mutants cannot completely restore the membrane stress. Overall, these data suggest that Pct1 interaction with membranes is necessary to regulate PC synthesis and drive the corresponding PL composition changes to maintain membrane SCE stress, though the result requires further validation. 182      Figure 4.7 Data-driven modelling of stored curvature elastic (SCE) stress regulation from the lipidomes of 3Δ cells expressing Pct1-NLSm1-GFP or WT Pct1-GFP and undergoing PC recovery. 183    3Δ yeast cells expressing either Pct1-NLSm1-GFP or WT Pct1-GFP were grown for 24 hr followed by addition of 1 mM choline. Cells were collected before (0.0 hr) and after choline supplementation at the indicated time points (see top left inset for time-point color codes) for lipidomics analyses and the total (PSCE) and the lipid-driven SCE stress in the cells was calculated as described in Methods section 2.8. These data correlate with imaging slides shown in Figure 4.6.2. (A)  Lipid composition by headgroup (mol% total lipid). PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PI, phosphatidylinositol; PA, phophatidic acid; DAG, diacylglycerol; TAG, triacylglycerol. (B–D) Fatty acid composition of the PC (B), PE (C) and PS (D) fraction expressed as mol% of the respective total phospholipids. The phospholipid (32:1) or (34:1) represent species where the acyl groups at C-1 and C-2 contain a total of 32 or 34 carbon atoms with 1 double bond while phospholipid species (32:2) or (34:2) represent species with 2 double bonds. (E) Estimates of the total SCE stress (PSCE stress). (F) Estimates of the lipid contribution to SCE stress. Data are shown as mean ± SD from independent experimental repeats (n=8 for WT-Pct1; n=3 for NLSm1-Pct1). One-way ANOVA, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.   184      185    Figure 4.8 Data-driven modelling of stored curvature elastic (SCE) stress regulation from the lipidomes of 3Δ cells expressing Pct1-NLSm4m-GFP or Pct1-4m-GFP and undergoing PC recovery. 3Δ yeast cells expressing Pct1-NLSm4m-GFP (NLSm4m) or Pct1-4m-GFP (4m) were grown for 24 hr followed by addition of 1 mM choline. Cells were collected before (0.0 hr) and after choline supplementation at the indicated time points (see top left inset for time-point color codes) for lipidomics analyses and the total (PSCE) and the lipid-driven SCE stress in the cells was calculated as described in Methods section 2.8. These data correlate with imaging slides shown in Figure 4.3.3. (A) (i) PC and (ii) PE levels at the indicated time pints are shown in nmoles (per 50 mg yeast) relative to internal standards. (iii) PC/PE ratio at the indicated time points. (B) Lipid composition by headgroup (mol% total lipid). PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PI, phosphatidylinositol; PA, phophatidic acid; DAG, diacylglycerol; TAG, triacylglycerol. (C–E) Fatty acid composition of the PC (C), PE (D) and PS (E) fractions expressed as mol% of the respective total phospholipids. The phospholipid (32:1) or (34:1) represent species where the acyl groups at C-1 and C-2 contain a total of 32 or 34 carbon atoms with 1 double bond while phospholipid species (32:2) or (34:2) represent species with 2 double bonds. (F) Estimates of the total SCE stress (PSCE stress). (G) Estimates of the lipid contribution to SCE stress. Data are shown for n=1 experiment. 186    4.4.6 Ability of NLS mutant of Pct1 to rescue DNA damage sensitivity As suggested in section 3.4.14, an initial chemogenetic screen suggests that Pct1 may be required for effective restoration of DNA damage induced by phleomycin. To test if nuclear localisation of Pct1 is necessary for DNA damage repair, the NLS mutants were transformed into pct1Δ cells where DNA damage was induced by phleomycin. Interestingly, both the NLSm1 and NLSm1m2 mutants rescued cell growth in a similar manner to that of wild-type Pct1 (Figure 4.9). The defective NLSm4m partially rescued cell growth in this setting (Figure 4.9). These data again verify that NLS mutants truly catalyse the production of PC. While maintenance of PC produced via the Kennedy pathway may have a role to play in DNA repair, the nuclear retention of Pct1 does not seem necessary for this activity. Clearly this will require further experimental confirmation in other DNA damage inducing reagents and in different cell types.   Figure 4.9 Evaluation of the potential of Pct1 NLS mutants to protect yeast cells against DNA damage. Pct1Δ cells expressing either wild-type Pct1-GFP (WT) or mutant Cdel, NLSm1, NLSm4m, NLSm1m2 and Cmut-GFP were grown to exponential phase and serial dilutions of liquid cultures were spotted onto YPD plates or YPD plates supplemented with 2.5 g/mL phleomycin (phleo). Empty plasmid (EV), Cdel-GFP and Cmut-GFP were used as negative controls. Data are representative of at least three independent experiments. 187    4.5 Discussion 4.5.1 Ability of CCT NLS mutant to rescue cells with wild-type CCT deficiency The yeast model suggests that nuclear compartmentalisation of Pct1 is not essential for its function. Pct1-NLSm1 can fully compensate for the loss of wild-type Pct1 in many aspects, e.g. PC synthesis, cell growth and DNA damage-resistance. This finding coincides with the results in a more complex metazoan model, Drosophila. Recent studies in my host lab have shown that a similar NLS mutant of the Drosophila CCT1 (CCT1-NLSm; K34A, R35A, K36A) is cytoplasmic, unlike Pct1-NLSm1 which localises to the PM. The CCT1-NLSm-GFP mutant localises to the periphery of lipid droplets in oleate loaded S2 cells (Figure 4.10A(i)). When expressed transgenically in fly photoreceptors (Figure 4.10A(ii)), this mutant was mainly cytoplasmic in the isolated ommatidia, which constitute the basic units of the compound eye of flies. Lack of CCT1 in their photoreceptors prevents the development of the usual light-sensing apical membrane of each photoreceptor (the structure referred as a rhabdomere) and consequently, retinal CCT1 deletion of flies cause blindness as analysed by electroretinogram. Notably, CCT1-NLSm expression could restore the light response to wild-type levels in CCT1179 (CCT1 deleted) or CCT299 (both CCT1 and CCT2 deleted) eyes (Figure 4.10B) and transmission electron micrographs of retinal cross-sections from these rescued flies indicated normal development of the rhabdomere (Figure 4.10C). The complementary results obtained from both yeast and fly models suggest the dispensability of Pct1/CCT1 confinement in the nucleus as the Pct1/CCT1 activity could be regulated by association with either the nuclear envelope or other membranes in the cytoplasm. Nevertheless, we could not entirely preclude other evolutionarily conserved ancillary roles in the intranuclear targeting of Pct1/CCT1/PCYT1A. There are at least two possible explanations for the ability of the NLS mutant to compensate for the lack of CCT: (i) Although Pct1-NLSm1, Pct1-NLS1del and CCT1-NLSm predominantly localised to the cytoplasm, we cannot exclude the possibility that a small fraction of the protein was intranuclear. In addition to the classical NLS transport pathway, there are many non-classical NLS transport routes being increasingly identified (Soniat and Chook, 2015), such as the karyopherin-β2 pathway, in which the proline-tyrosine (PY)-NLSs are recognised and transported into the nucleus by the receptor transportin-1/karyopherin-β2. There is also a growing 188    number of types of non-classical NLS sequences (McLane and Corbett, 2009). Hence, Pct1/CCT1 might employ as yet undetermined NLS motifs and enter the nucleus through its interaction with various transport receptors. (ii) Pct1/CCT1/PCYT1A is associated with the inner nuclear membrane of the nuclear envelope which in fact forms a membrane continuum with the outer nuclear membrane and the ER network in eukaryotic cells. The outgrowth of the ER network could presumably allow phospholipids to diffuse from the nucleus to the peripheral ER and vice versa (Bahmanyar, 2015), depending on the differential phospholipid distribution in membranes. Vesicle-independent movement of PC has been demonstrated between the ER and the plasma membrane (Kaplan and Simoni, 1985; van Meer et al., 2008); a similar transport mechanism may exist allowing PC transport from ER to other membranes. These findings may explain why NLS mutants of Pct1 in yeast and CCT1 in flies can compensate for the absence of the wild-type enzyme. Future work is required for a better understanding of other essential functions that require Pct1/CCT1/PCYT1A to be compartmentalised inside the nucleus. 189     Figure 4.10 CCT1-NLSm has a cytosolic localisation and can compensate for the loss of wild-type CCT in Drosophila photoreceptors. (A)(i) Drosophila S2 cells expressing C-terminal GFP-tagged wild-type CCT1 (CCT1) or CCT1- NLS mutant (CCT1-NLSm) were grown with (+OA) or without (-OA) oleic acid for 24 hr. Live cell imaging in cells stained for lipid droplet (LDs) and nuclear (DAPI) visualization was performed. Scale bar, 2.5 m. (ii) D. melanogaster eyes expressing CCT1-NLSm1-GFP. Individual ommatidia were isolated and imaged. All panels are zoomed in images from the boxed region. Scale bar, 10 m. (B) Response to light activation was assessed by electrotrtinograms and the mean response generated at the maximal intensity was plotted of each genotype. CCT1179, CCT1 deleted; CCT299, both CCT1 and CCT2 deleted. Data are reported as mean ± SD. One-way ANOVA with Bonferroni multiple comparison, ****p<0.0001. (C) Rhabdomere ultrastructure 190    was analyzed by TEM. (i) Control flies; (ii) CCT1179 or CCT299 flies; and (iii) flies with CCT null eyes rescued by transgenic expression of CCT1-NLSm mutant. Scale bar, 500 nm. All results in this figure were the work of Dr Afreen Haider. 4.5.2 Cytoplasmic Pct1 can replace the nuclear Pct1 but PCYT1B cannot replace the function of PCYT1A In most eukaryotic cells, there is a second cytosolic PCYT1B isoform (CCT2 in Drosophila) which presumably senses phospholipid changes in the ER and possibly in other membrane-bound organelles. In contrast to the ubiquitously expressed PCYT1A, its expression appears to be poor in many cell types (Karim et al., 2003; Marcucci et al., 2008). Pcyt1b null mice manifest subtle phenotypes including axonal branching reductions in sympathetic neurons and reproductive insufficiency mostly from defective ovarian follicle development and decreased spermatogenesis (Jackowski et al., 2004; Strakova et al., 2011) whereas Pcyt1a null mice die in utero (Wang et al., 2005). This suggests that despite the fact that the Pct1/CCT1 NLS mutant can compensate for the loss of wild-type CCT in yeast and fly models, in mice PCYT1B expression cannot compensate for the loss of PCYT1A expression during embryonic development. This result is also supported by the inability of endogenous CCT2 to rescue rhabdomere formation in CCT1 knockout fly  photoreceptors (Figure 4.10C (ii) CCT1179). While CCT1 knockdown/deletion results in flies with a blind phenotype, CCT2 knockdown does not cause any effect on its own (data not shown). Additionally, in Pcyt1a+/− heterozygous mice, the loss of PCYT1A did not appear to result in upregulation of PCYT1B expression in adult liver and other tissues (Wang et al., 2005). These findings are in agreement with later studies which show that, in human skin primary fibroblast cells from lipodystrophy patients, the low levels of PCYT1A expression do not lead to upregulation of PCYT1B (Payne et al., 2014). The expression of the PCYT1B isoform predominates only in brain, placenta, ovary, testis and all fetal tissues (Lykidis et al., 1998), which could account for the probability that PCYT1B serves as an additional enzyme to assist in the production of PC in certain tissues but it cannot substitute for PCYT1A. Therefore, the loss of PCYT1A may not be compensated for by endogenous PCYT1B because PCYT1A is more ubiquitously expressed than PCYT1B; additionally, PCYT1B expression level was not unregulated under Pcyt1a deletion conditions. However, this does not suggest that nuclear localisation is essential for CCT activity. The yeast Pct1-NLSm may support this idea as even though it is not in the cell nucleus, this had no effect on cell proliferation and survival. 191    4.5.3 Conclusions The results in this chapter have shown that yeast Pct1 in the absence of NLS region 60PRKRRRL66 localised to the plasma membrane and when cells were deprived of its substrate choline, the NLS mutant was associated with lipid droplets. Upon choline supplementation, the NLS mutant is gradually released from lipid droplet membranes, suggesting that it retained its catalytic ability and regulation by the on/off membrane-binding mechanism. Lipidomic analysis and SCE estimation show that the NLS mutant presumably senses phospholipid composition changes in terms of membrane stored elastic stress levels, which leads to its subsequent activation and downstream PC regulation, and it thus has the ability to restore SCE like wild-type Pct1. This was also obvious in the yeast survival assay, where the NLS mutant was able to rescue 3Δ cells from FOA induced toxicity just like wild-type Pct1. Furthermore, the NLS mutant was able to relieve phleomycin-derived DNA damage sensitivity in Pct1-null yeast cells. These results add new insights into Pct1/ PCYT1A function and its nuclear compartmentalisation.       192    Chapter 5 Initial Characterisation of Disease-Related Human PCYT1A Mutations in Cultured Cells 5.1 Abstract Human CTP: phosphate cytidylyltransferase A (CCT or PCYT1A; EC 2.7.7) mutations have recently been linked to rare genetic disorders whose clinical and biochemical manifestations vary greatly among affected individuals. In an effort to characterise the mutations in the human PCYT1A gene, both mammalian and yeast expression systems were used. i) In murine ear mesenchymal stem cells, Pcyt1a knockdown before the onset of adipocyte cell differentiation severely impairs adipogenesis. ii) In primary skin fibroblasts from human patients with compound heterozygous or homozygous bi-allelic mutations in PCYT1A, a small reduction in CDP-choline level was observed as compared to controls; however, the decrease was not sufficient to induce an increase in phosphatidylethanolamine N-methyltransferase activity. iii) In a yeast model, human PCYT1A can substitute for endogenous yeast Pct1 protein. Yeast complementation assays showed that the expression of mutant PCYT1A alleles  known to be associated with reduced enzyme activity in humans, V142M and 333fs, failed to complement cell growth in cho2∆opi3∆pct1∆ cells. These preliminary results demonstrate the role of PCYT1A in adipogenesis and in PC synthesis; in addition, I have successfully established the utility of a yeast-expression system for studying human PCYT1A variants. 5.2 Introduction 5.2.1 Animal models of PCYT1A deficiency CCT (PCYT1A) is encoded by the PCYT1A gene and regulates a major rate-limiting step in PC synthesis. The biochemical pathways involved in PC synthesis and the catalytic function of PCYT1A have been extensively investigated (Fagone and Jackowski, 2013; Gibellini and Smith, 2010). The physiological consequences of PCYT1A dysfunction have been examined through gene mutation and knockout strategies in several different transgenic animal models (Landis et al., 2003; Roggero et al., 2004; Wang et al., 2005; Weber et al., 2003; Yoshikawa et al., 2011). Table 5.1 summarises the PCYT1A mutant/knockout models and their corresponding phenotypes in mice, Drosophila melanogaster, Caenorhabditis elegans and Saccharomyces cerevisiae. The 193    complete knockout of PCYT1A results in reproductive failure in mice and Drosophila but only decreases the growth rate and chemical resistance in S. cerevisiae. The latter is presumably because the phosphatidylethanolamine N-methyltransferase (PEMT) pathway predominates in yeast (Chin and Bloch, 1988). Various PCYT1A mutations in Drosophila cause an abnormal body morphology and defects in eye development, which is consistent with observations made in human patients with certain PCYT1A mutations (Hoover-Fong et al., 2014) manifesting spondylometaphyseal dysplasias and visual cone-rod dystrophy (Walters et al., 2004). A PCYT1A knockdown model in C. elegans resulted in abnormal LD and triglyceride accumulation, which was also observed in the Drosophila S2 cell line when CCT1 was deleted (Krahmer et al., 2011). 5.2.2 Human PCYT1A mutations and the related clinical phenotype Apart from the abnormal phenotypes observed in animal models with gene mutations and knockout approaches, studies from clinical observations have very recently provided further insights into the relationship between the known PC synthetic pathways and PCYT1A. Our group firstly identified a novel lipodystrophy subtype in patients with bi-allelic mutations in the PCYT1A gene leading to early onset non-alcoholic fatty liver (FL) disease, short stature and congenital lipodystrophy (CLD) (Payne et al., 2014). Other research groups also identified different bi-allellic PCYT1A mutations in patients with spondylometaphyseal dysplasias (SMD) and visual cone-rod dystrophy (CRD) (Hoover-Fong et al., 2014; Kitoh et al., 2011; Wong, 2014; Yamamoto et al., 2014). In 2017, Testa et al. found some mutations in the PCYT1A gene that are responsible for isolated forms of retinal dystrophy, Leber congenital amaurosis (LCA). Although there is no clear explanation for the phenotypic heterogeneity, most of these mutations were found to reduce PCYT1A protein expression levels and impair PC synthesis. The locations of the genetic defects were mapped using whole exome sequencing and are illustrated in Figure 5.1. 194    5.2.3 Pathways for PC biosynthesis in humans PC is generated by three different biochemical pathways in humans: i) the CDP-choline pathway, which is also known as the Kennedy pathway (Kennedy and Weiss, 1956), ii) the phosphatidylethanolamine N-methyltransferase (PEMT) pathway (Horl et al., 2011) and iii) the Lands cycle (Moessinger et al., 2011). Figure 5.2 shows an overview of the main PC biosynthetic pathways. i) The Kennedy pathway is responsible for approximately 70% of PC biosynthesis taking place within rat primary hepatocytes (DeLong et al., 1999). Note that this varies in other cell types. Choline, which is primarily obtained from the diet, is initially phosphorylated by choline kinase (CK) to form phosphocholine. CCTα (i.e. the gene product of PCYT1A) then converts phosphocholine to CDP-choline. The final step of the Kennedy pathway involves transfer of the CDP-choline moiety to diacylglycerol (DAG) by cholinephosphotransferase (CPT) and completes PC synthesis. ii) The PEMT pathway is considered to be restricted to liver cells in mammals (Jacobs et al., 2010). In this pathway, the enzyme phosphatidylethanolamine N-methyltransferase (PEMT) catalyses three consequent methylations of phosphatidylethanolamine (PE) thereby converting PE to PC (Vance, 2013). iii) The third route for PC biosynthesis is via the Lands cycle. Phospholipids are first synthesised by the Kennedy pathway and their acyl-CoAs composition at the sn-2- position is altered in the Lands remodeling cycle. Fatty acids of phosphatidylcholine are liberated by phospholipase A2 (PLA2) and converted to lysophosphatidylcholine. Lysophosphatidylcholine is then converted to phosphatidylcholine in the presence of acyl- CoA by lysophosphatidylcholine acyltransferase (LPCAT). 19 5   Ta ble 5.1 O rth olo gs for th e h um an PC YT 1A ge ne an d r ela ted m uta tio n m od els . Or ga nis m Ge ne Mu tat ion m od els Re f. Mu tan t In for ma tio n Ab no rm al Ph eno typ es M us m us cu lu s Pc yt 1a Tar get ed (N ull /Kn ock out ) Em bry oni c le tha lity (W ang et al., 20 05 ) D ro so ph ila m el an og as te r C C T1 Nu ll Le tha l (W ebe r e t al ., 2 00 3) Mu tan ts (C ct 1B N 81 an d C ct 19 9 ho mo zyg ou s) De fec ts i n o og ene sis an d o var ian mo rph og ene sis (G up ta and Sch up bac h, 200 3) Ov ere xp res sio n Inc rea se in life sp an (La nd is e t al ., 2 003 ) Mu tan ts (dC CT 116 91 9/D f(3 L)e mc 5 o r dC CT EP 083 1/D f(3 L)e mc 5) Inc rea se in end ocy tos is; eff ect s o n eye de vel op me nt (de fec ts i n om ma tid ial po lar ity ); i ma gin al d isc - der ive d w ing m orp hog ene sis (W ebe r e t al ., 2 00 3) Ca en or ha bd iti s el eg an s CE LE _F 08 C6 .2 PC YT -1 siR NA (K no ckd ow n) Inc rea se in LD /tri acy lgl yce rol acc um ula tio n (W alk er et a l., 201 1) Sa cc ha ro m yc es ce re vi sia e S2 88 c PC T1 Nu ll De cre ase in ve get ativ e g row th rat e (Yo shi kaw a e t al ., 2 011 ) Nu ll Inc rea se in res ista nce to di bro mi de (Ro gg ero et al., 20 04) 196    Figure 5.1 Schematic illustration of the PCYT1A domain organisation indicating the location of mutations identified to date. The different PCYT1A phenotypes corresponding to the mutations shown in the diagram are highlighted with different colors. CLD-FL, congenital lipodystrophy and severe fatty liver disease (pink); SMD-CRD, spondylometaphyseal dysplasia with cone-rod dystrophy (black); LCA, Leber congenital amaurosis (brown). N-region, N-terminal region; C-domain, catalytic domain; M-domain, membrane-binding domain; P-region, C-terminal phosphorylated region. Figure 5.2 Biosynthetic pathways for the formation of phosphatidylcholine in humans. Schematic of the PC biosynthesis pathways include the Kennedy pathway, PEMT pathway and the Lands cycle. The known enzymes catalyzing individual steps in the pathway are also indicated. CK, choline kinase; DAG, diacylglycerol; CPT, cholinephosphotransferase; BHMT, betaine-homocysteine methyltransferase; MAT, methionine adenosyltransferase; SAM, S- adenosylmethionine; SAH, S-adenosyl-L-homocysteine; MS, methionine synthase; PEMT, phosphatidylethanolamine N-methyltransferase; PCP, phosphatidylcholine:ceramide cholinephosphotransferase; PLB, phospholipase B; LPCAT, lysophosphatidylcholine acyltransferase; PLA2, phospholipase A2. CCT  PLA2  PEMT pathway 197    5.3 Aims Human genetic data suggests that PCYT1A is particularly important in adipocytes, hepatocytes, retinal photoreceptor cells and chondrocytes. Why this is the case remains unclear and will require further studies. It is also elusive why mutations in the same gene are associated with different phenotypes. To begin to address some of these issues the aims of this chapter were: I. To identify and verify the role of PCYT1A in adipogenesis in a murine cell line. II. To characterise the consequences of selected loss-of-function PCYT1A mutations on PC synthesis in primary patient cell lines. III. To characterise selected patient-derived PCYT1A mutants using S. cerevisiae as a model system. 5.4 Results 5.4.1 The role of PCYT1A in mammalian cells 5.4.1.1 Effect of Pcyt1a knockdown on adipocyte differentiation As PCYT1A mutations have been linked to lipodystrophy, there is a strong suggestion of a role for PCYT1A in adipocyte metabolism. To analyse the role of the PCYT1A gene during adipogenesis, it would be interesting to evaluate the effect of Pcyt1a knockdown in early or later stages of cell differentiation in model systems. Two in vitro cell culture models have been widely used for the analysis of gene function during the course of adipogenic differentiation in my host laboratory: 1) 3T3-L1, an immortalised/established preadipocyte cell line derived from Swiss 3T3 mouse embryos; and 2) Ear mesenchymal stem cells (EMSC), adult stem cells derived from the external murine ear. We have previously observed that Pcyt1a knockdown in early adipocyte differentiation has a disruptive impact on 3T3-L1 198    differentiation (Payne et al., 2014). To confirm this phenomenon further, EMSC cells were used to investigate the importance of PCYT1A in adipogenesis. EMSCs were transfected with Pcyt1a-targeted siRNA and resulting mRNA expression levels were analysed by qRT-PCR. As shown in Figure 5.3, qRT-PCR analysis confirmed the effectiveness of the Pcyt1a-specific siRNA which was able to knockdown Pcyt1a to <50% of wild-type levels as compared to cells transfected with a non-targeting negative control siRNA. Pcyt1a can be efficiently knocked down 48 hr after transfection. Cells were therefore treated with siRNA every 48 hr in the following experiments to maintain knockdown throughout cell differentiation. Figure 5.3 Pcyt1a mRNA levels are decreased 48 hours after transfection using gene- specific siRNA in EMSC adipocytes.   The relative mRNA levels of Pcyt1a were compared with the corresponding negative controls (SiCtrl). Pcyt1a showed significantly lower expression after 48 and 96 hr post-transfection. Normalization was performed using the mRNA level of the Gapdh housekeeping gene. Data are presented as mean ± SD from three independent repeats. Two-tailed Student’s t-test, ***P < 0.001. 199    EMSC cells were grown to confluence and triggered to differentiate as described in the Methods section 2.10. LD biogenesis was examined at day 10 post-differentiation. When Pcyt1a was knocked down using siRNA prior to induction of adipocyte differentiation (2 days before cell differentiation), LD formation was significantly reduced (Figures 5.4A and 5.5) as judged by staining with a lysochrome diazo dye Oil Red O (ORO), which is typically used for neutral lipid staining (Ramirez-Zacarias et al., 1992). In contrast, in cells that were already undergoing differentiation, Pcyt1a knockdown from day 4 onwards had no effect on further differentiation and LD formation (Figures 5.4B and 5.5). This is consistent with previous results from 3T3-L1 cells (Payne et al., 2014) which show that LD formation was reduced when cells were subjected to Pcyt1a-specific siRNA treatment prior to but not during induction of adipocyte differentiation. Further, the protein expression levels of PCYT1A, perilipin and adipocyte protein 2 (aP2) were examined by immunoblotting from cell lysates (Figure 5.6). Perilipin is a protein associated with lipid droplets and aP2 is expressed specifically in mature adipocytes; therefore, both serve as good indicators of adipocyte differentiation. Pcyt1a knockdown at an early time point of differentiation decreases not only PCYT1A protein expression levels but also that of perilipin and aP2 suggesting that early knockdown of Pcyt1a inhibited LD formation and reduced expression of adipogenetic genes. Conversely, knockdown at later time points did not seem to impair either lipid accumulation or expression of aP2, whereas Perilipin expression seemed to be slightly reduced.   Figure 5.4 Pcyt1a knockdown at early time points of differentiation resulted in reduced lipid accumulation. 200    ORO dye specifically stains lipid droplets and exhibits a bright red color. EMSC pre- adipocytes were grown to a confluent state (day 0) before being induced to differentiate into mature adipocytes. The cells were stained at day 10 post-differentiation. Pcyt1a siRNA was transfected (A) 2 days before EMSC cell differentiation or (B) after the onset of EMSC cell differentiation at day 4. Cells were treated with siRNA every 2 days until day 10.   Figure 5.5 Pcyt1a knockdown in EMSC cells as quantified by ORO staining at day 10 post-differentiation. These data correlate with the images shown in Figure 5.4. The bound ORO dye was eluted with pure DMSO and quantified by measuring its absorption at 492 nm. Data are presented as mean ± SD from three independent repeats. Two-tailed Student’s t-test, ***P < 0.001.   Figure 5.6 Knockdown of PCYT1A expression from early time point inhibits differentiation of EMSC cells into mature adipocytes. 201    Representative immunoblotting of lysates from EMSC cells. The PCYT1A expression was knocked down (KD) at different time points (from day -2 or day 4) during cell differentiation, and cells were harvested at day 10. Perilipin1 and aP2 expression reduced slightly in the Pcyt1a knockdown cells at day -2 while no change was seen when Pcyt1a was knocked down at day 4. Immunoblots shown are representative of three separate experiments. 5.4.1.2 PCYT1A mutations impairing the Kennedy pathway Some of the CLD-FL patients with PCYT1A mutations, P0616 (p.Val142Met/p.Glu280del) and P0689 (p.Glu280del/p.Ser333Leufs*164) (the ‘P’ number refers to a database of cell lines in my host lab) have lower PCYT1A-expression levels and impaired PC synthesis, as previously shown from isolated fibroblast cell cultures (Payne et al., 2014). Since then we have obtained fibroblasts from additional patients with SMD-CRD, specifically from patients in the following mutations: P2233 (homozygous p.Ala99Val mutation), P8069 (homozygous p.Glu129Lys mutation) and P8123 (homozygous p.Ser323Argfs*38 mutation) (Hoover-Fong et al., 2014; Yamamoto et al., 2014). Table 5.2 summarises the above information. These mutations are present in different regions of the PCYT1A coding sequence, including the catalytic domain, membrane-binding domain and amongst C-terminal phosphorylation sites (Figure 5.1). To evaluate the functional implications of the PCYT1A mutations, a detailed mass spectrometry analysis of skin fibroblast cell lysates from patients and healthy controls was performed in collaboration with the mass spectrometry facility core at the University of North Carolina, United States. Table 5.2 Primary human skin fibroblasts from patients used in this study. Code Disease Mutation AA change in PCYT1A P0616 CLD-FL p.Val142Met/p.Glu280del P0689 CLD-FL p.Glu280del/p.Ser333Leufs*164 P2233 SMD-CRD homozygous p.Ala99Val mutation P8069 SMD-CRD homozygous p.Glu129Lys mutation P8123 SMD-CRD homozygous p.Ser323Argfs*38 mutation 202    These five patient fibroblasts and five control fibroblast lines, C100, C101, C103, C104 and C105 were analysed respectively in triplicate, and an independent t-test was used to compare the mean of each patient, the patient group and the control group. The quantification of phosphatidylcholine concentrations and other associated metabolites involved in the Kennedy pathway is shown in Figures 5.7.1 and 5.7.2. CDP-choline, the direct catalytic product of PCYT1A, exhibited a small but significant reduction in patient P0689 (Figure 5.7.1B) and P8123 (Figure 5.7.2C). The CDP-choline level had no significant reduction in the CLD-FL group as a whole (Figure 5.7.1C), while showed a significant reduction in the SMD-CRD group as a whole (Figure 5.7.2D). Furthermore, the level of phosphatidylcholine decreased significantly in patient P8123 (Figure 5.7.2C) but not in the other four patients. Although a remarkable decrease in the phosphatidylcholine level was observed in  the SMD-CRD group (Figure 5.7.2D), this change was not observed in the CLD-FL group (Figure 5.7.1C). Two possible explanations for the lack of a detectable change in phosphatidylcholine are that: i) alternative PC synthetic pathways may compensate for the impaired Kennedy pathway and restore the PC level in human cells, or ii) the medium used for cell culture may contain sufficiently high levels of PC to affect the results obtained. To test the second hypothesis, lipid-depleted FBS medium was utilised during fibroblast culture for 48 hr and the levels of different phospholipids were determined in lipid extracts by thin-layer chromatography (TLC). The TLC study revealed a decreasing amount of PC in cells treated with lipid-depleted 10% FBS medium or lipid-depleted 1.0% FBS medium respectively (Figure 5.8A). In addition, the PE level decreased (Figure 5.8A) while the PC/PE ratio seems to be maintained (Figure 5.8B). This suggests that the normal media may satisfy the requirement of phospholipids in cells and influence the data observed in the previously mentioned mass spectrometry analysis where PCYT1A mutations presumably indeed impair PC synthesis. We have not yet had an opportunity to repeat the lipidomic analysis in fibroblast cell lines after culture in lipid-depleted 1.0% FBS medium. 203    A B C Figure 5.7.1 Kennedy pathway metabolites in CLD-FL patients and healthy controls. Amounts of different metabolites from the CLD-FL patients (P0616 and P0689) and the control group (C100, C101, C103, C104 and C105) were quantified by mass spectrometry. The CDP-choline level was lower than the detection limit in patients P0689 where the threshold value of detection was used instead. (A) P0616 compared to the control group. (B) P0689 compared to the control group. (C) The CLD-FL patient group compared to the control group. Data are shown as mean ± SD from three independent repeats. Two-tailed Student’s t- test, *P < 0.05, **P < 0.01 and ***P < 0.001. 204    A B C D Figure 5.7.2 Kennedy pathway metabolites in SMD-CRD patients and healthy controls. Amounts of different metabolites from the SMD-CRD patients (P2233, P8069 and P8123) and the control group (C100, C101, C103, C104 and C105) were quantified by mass spectrometry. The CDP-choline level was lower than the detection limit in patients P8123 where the threshold value of detection was used instead. (A) P2233 compared to the control group. (B) P8069 compared to the control group. (C) P8123 compared to the control group. (D) The SMD-CRD patient group compared to the control group. Data are shown as mean ± SD from three independent repeats. Two-tailed Student’s t-test, *P < 0.05 and **P < 0.01. 205    Figure 5.8 Culture media compositions affect the levels of phospholipids in human primary fibroblast cells.   Fibroblasts from P0616 patient were cultured for 48 hours in three different media compositions: normal 10% FBS medium, lipid-depleted 10% FBS medium and lipid-depleted 1.0% FBS medium. The lipids were then extracted from cells and separated by thin-layer chromatography (TLC). The TLC plates were dried and the lipid spots were identified under a UV lamp after spraying with a primuline solution. The lipid species were identified by the comigration of standards. (A) Intensity of the separated spots for PC and PE after TLC separation. Two-tailed Student’s t-test, *P < 0.05, **P < 0.01. (B) PC/PE ratio in cells grown in the three medium. Data are shown as mean ± SD from three independent repeats. B A 206    5.4.1.3 PEMT does not compensate for deficits in the Kennedy pathway In the primary cells, the PC levels of most patients were similar to those of fibroblasts cells from healthy controls. (Figure 5.7.1 and Figure 5.7.2). Thus, it is possible that an alternative pathway for PC synthesis such as the PEMT pathway can compensate for the impaired Kennedy pathway and help to maintain PC levels. To test this possibility, PEMT enzyme activity was assayed; however, there was no apparent difference in PEMT activity between patients and healthy controls (Figure 5.9). This suggests that altered PCYT1A activity is not compensated for by the PEMT pathway. Future work is needed to examine potential compensation by other PL biosynthetic pathways. Figure 5.9 Analysis of the PEMT activity in human fibroblasts. PEMT activity was measured using isotope-labeled substrates as described in the Methods section 2.14. Mouse liver lysate was used as a positive control due to the high expression of PEMT in the liver. Data are normalized and represented as mean ± SD of 4 experimental repeats. One-way ANOVA. ns, no significant difference between patients (P0616, P0689, P2233, P8069 and P8123) and healthy controls (C104 and C105). ns 207    5.4.2 Characterisation of PCYT1A in yeast Saccharomyces cerevisiae 5.4.2.1 A yeast expression system for human mutant PCYT1A alleles As outlined in Chapter 3 (section 3.4.6), I have established a system in which S. cerevisiae cells entirely depend on transfected Pct1/PCYT1A for PC synthesis. This provides an opportunity to test human mutant allele function. To start with, the individual mutant alleles from CLD-FL patients (Figure 5.10A) were tested for their ability to promote growth of pct1Δ yeast that lacks an endogenous Pct1. Initially, cDNA of human wild-type or mutant PCYT1A was cloned downstream from the NOP1 promoter in a yeast plasmid that has an in-frame GFP coding region. The resulting plasmids coding for GFP tagged-PCYT1A were used to transform the pct1Δ strain. To determine if the patient-derived mutations affect protein expression, cell lysates from each transformant were prepared and immunoblotted with a PCYT1A antibody which targets an N-terminal epitope. The mutant V142M PCYT1A protein was barely detectable (Figure 5.10B). These findings are consistent with in silico predictions of its destabilising effect on the entire structural domain (Payne et al., 2014). The expression of the E280Δ mutant was similar to wild-type protein levels, whereas expression of the 333fs mutant was significantly reduced (Figure 5.10B). These observations were consistent with previous findings from immunoblotting of COS7 cell lysates, which were transfected with expression vectors for wild-type or mutant PCYT1A (Payne et al., 2014). As with yeast Pct1, wild-type PCYT1A localises to the nuclear envelope in the exponential phase and translocates into the nucleoplasm in the post-diauxic phase (Figure 5.10C). However, protein aggregates were also observed around the cell periphery. In order to rule out the possibility that the overexpression causes mislocalisation of PCYT1A, the construct was re-engineered such that the stronger NOP1 promoter was substituted for the weaker Pct1 endogenous one. As shown in Figure 5.10D, the plasmid expressing the human PCYT1A from the Pct1 endogenous promoter, results in a lower protein-expression level compared to the stronger NOP1 promoter as indicated by lower GFP fluorescence. A significant fraction of PCYT1A localised at the nuclear envelope but some still localised to the cell periphery. This is presumably related to the fact that the human PCYT1A nuclear localisation signal is not fully operational in yeast. These two PCYT1A orthologs do not share a large degree of sequence similarity. According to protein sequence alignment in EMBL-EBI (Li et al., 2015; McWilliam et al., 2013), the predicted human PCYT1A and yeast Pct1 proteins are 39.0% identical and 52.7% similar to each other, whereas the predicted monopartite NLS regions of 208    human PCYT1A (residues at amino acids 10–19) and yeast Pct1 (residues at amino acids 59– 70) are 33.3% identical and 41.7% similar to each other. The identified four-residue RKRR in the monopartite NLS motif is conserved among human and yeast (Figure 4.2B (i)); however, in EMBL-EBI alignment, the N-terminal regions of human PCYT1A (residues at amino acids 1–73) and yeast Pct1 (residues at amino acids 1–75) are only 6.2% identical/7.7% similar to each other, suggesting that PCYT1A and Pct1 can adopt different structures in the N-terminal region. Taken together, the non-significant similarity found in the N-terminal regions between human PCYT1A and yeast Pct1 may account, at least in part, for the apparent differences in PCYT1A and Pct1 localisation. The E280Δ mutant when expressed in pct1Δ cells, like wild-type PCYT1A, showed the expected localisation either to the nuclear envelope, plasma membrane or the nucleoplasm in the corresponding growth phase (Figure 5.10C). The cells expressing the V142M mutant showed diffuse GFP signal in the cytosol and nucleoplasm (Figure 5.10C). The 333fs mutant accumulated as patchy puncta within the cytosol (Figure 5.10C). These data suggest that 1) the V142M mutant is degraded; 2) the E280Δ mutant localises similarly to wild-type PCYT1A and 3) the 333fs mutant forms aggregates in the cytosol. 209    Figure 5.10 Expression of mutant human PCYT1A in S. cerevisiae. (A) Schematic illustration of the PCYT1A mutations associated with congenital lipodystrophy and severe fatty liver disease. (B) Representative PCYT1A immunoblotting of cell lysates from pct1∆ cells expressing GFP-tagged wild-type (WT), or mutant V142M, E280∆ or 333fs PCYT1A. GAPDH was used as a loading control. (C) pct1∆ cells expressing the GFP-tagged wild-type PCYT1A (WT), V142M, E280∆ and 333fs were grown to exponential phase (Exp) 210    or post-diauxic shift phase (PDS) and imaged as described in the Methods section 2.5.1. Differential interference contrast (DIC) images are shown for context. Scale bar, 5 m. Images are representative of three independent experiments. (D) pct1∆ cells expressing the GFP-tagged wild-type PCYT1A under the control of either the NOP1 promoter or the endogenous Pct1 promoter. Cells were grown to exponential phase and imaged as described in the Methods section 2.5.1. DIC images are shown for context. Scale bar, 5 m. Images are representative of three independent experiments. 5.4.2.2 PCYT1A mutant alleles reveal different phenotypes in the cho2∆opi3∆pct1∆ yeast strain For further in vivo characterisation of PCYT1A in a yeast model, where it is the sole enzymatic source for PC, the cho2∆opi3∆pct1∆ (3∆) mutant strain was used i) to examine the ability of wild-type or mutant PCYT1A functionally to complement the 3∆ mutant strain; and ii) to investigate the differences in wild-type or mutant PCYT1A protein translocation with or without excess choline. In Figure 5.11A, expressions of both Pct1 and PCYT1A from plasmids can rescue the growth of the 3∆ cells in the presence of 5-FOA. The E280∆ PCYT1A mutant also restores the growth of the 3∆ cells to a similar extent to that of the wild- type. The 3∆ cells show a slightly reduced but comparable viability when expressing the 333fs PCYT1A mutant, whereas they have significantly decreased viability when expressing the V142M mutant. These observations were further confirmed by the DNA damage sensitivity assay (Figure 5.11B) in which the V142M and 333fs PCYT1A mutants could not fully rescue the reduced viability of pct1∆ in the presence of phleomycin, while wild-type PCYT1A and the E280∆ PCYT1A mutant were able to protect the pct1∆ cells against DNA damage in a similar way to that of the wild-type Pct1. In the 3∆ cells, both wild-type PCYT1A-GFP and E280∆ mutant PCYT1A-GFP were associated with the nuclear membrane and cell periphery in the absence of choline (Figure 5.12), reflecting cellular deficiency of PC, but gradually re-localised to the nucleoplasm when the cells were supplemented with excess choline. As in pct1Δ cells, the V142M PCYT1A- GFP mutant displayed a diffuse nuclear and cytoplasmic signal (Figure 5.12). The mutant 333fs PCYT1A-GFP again showed punctate cytoplasmic-expression irrespective of the choline status of cells (Figure 5.12). Prolonged absence of choline leads to an increase in LD size, but when choline was added back into the culture medium, LD size was reduced in all cases (Figure 5.12). This implies that all the mutant cells still retain the capacity to utilise DAG for PC synthesis in the presence of choline, thus resulting in the observed LD size changes. 211    Overall, these data suggest that wild-type human PCYT1A and the E280∆ mutant can effectively compensate for 3∆ in regulating PC synthesis in yeast. The E280∆ mutant thus probably is not a severe mutant. Whereas the E280∆ mutant rescued 3∆ cells, the V142M mutant did not have any rescue effect. One may conclude that patient P0616 patient possibly relied entirely on the E280∆ allele to generate CDP-choline for PC synthesis. Notably, although the 333fs mutant rescued the phenotype in 3∆ yeast, the protein expression level of the 333fs mutant was almost undetectable in the mammalian COS7 cells and in primary cells from the patient (Payne et al., 2014). These unclear differences in expression make it difficult to draw too many conclusions from the yeast data. It is perhaps not too surprising that this mutant, when expressed, can restore catalytic activity as it retains the catalytic domain. However, in humans many frameshift mutations are degraded so this is likely to be a main factor in vivo. Figure 5.11 Human PCYT1A can compensate for the loss of wild-type Pct1 in yeast. (A) 3Δ cells carrying Ycplac33-URA3-OPI3 were transformed with a CEN/HIS3 plasmid expressing either wild-type yeast Pct1, wild-type human PCYT1A or the mutants V142M, E280Δ and 333fs. Empty CEN/HIS3 plasmid (EV) was used as a negative control. Cells were grown to exponential phase and serial dilutions of liquid cultures were spotted onto SC plates lacking histidine (-HIS) or plates supplemented with FOA and 1 mM choline (FOA + cho). Data are representative of at least three independent experiments. (B) pct1Δ cells transformed with a CEN/HIS3 plasmid expressing either wild-type yeast Pct1, wild-type human PCYT1A or the mutants V142M, E280Δ and 333fs. Cells were grown to exponential phase and serial dilutions of liquid cultures were spotted onto YPD plates or YPD plates supplemented with 2.5 g/mL phleomycin. Data are representative of at least three independent experiments. 212    Figure 5.12 Changes in PCYT1A-GFP localisation correspond to cellular choline induction studies. 213    3∆ cells expressing the GFP-tagged wild-type PCYT1A (WT), E280∆, V142M and 333fs were grown to the exponential phase with (+) or without (-) excess choline 1 mM. Lipid droplets (LDs) were stained and cells were imaged as described in the Methods section 2.5.1. Differential interference contrast (DIC) images are shown for context. Scale bar, 5 m. Images are representative of three independent experiments. 5.5 Discussion 5.5.1 The role of PCYT1A in adipogenesis Lipodystrophy was one of the key phenotypes recently reported in patients P0616 (V142M/E280∆) and P0689 (E280∆/333fs) with bi-allelic mutations in PCYT1A (Payne et al., 2014). The extreme lack of adipocyte in P0616 suggests that a defect in the Kennedy pathway of PC synthesis is sufficient to prevent the development of even partially functioning adipose tissue, though the partial lipodystrophy found in P0689 suggests that PCYT1A impairment is not enough to inhibit adipose development completely. These differences may be attributed, at least partly, to the fact that V142M is a more severe mutation than 333fs since in yeast assays and the V142M mutant does not show an efficient rescue effect in 3∆ cells. Although P0616 and P0689 manifest different levels of lipodystrophy, they both display defective adipocyte development. This study has demonstrated that a deficit in PCYT1A expression during early but not late adipogenesis inhibits lipid accumulation and adipocyte differentiation in EMSCs. Also, these data are consistent with previous observations in 3T3L-1 cell lines demonstrating that Pcyt1a knockdown early in differentiation impairs adipogenesis. These findings suggest that PCYT1A expression and PC synthesis play an important role in early adipocyte development. Early adipogenesis is a time for the entry of preadipocytes into the cell cycle and transcription of early adipogenic factors (Tang and Lane, 2012). Interestingly, when mesenchymal stem cells and 3T3-L1 cells were induced to differentiate, the pre-adipocytes firstly re-entered the cell cycle and underwent several rounds of cell division, a process known as mitotic clonal expansion (MCE) (Marquez et al., 2017; Tang et al., 2003). Cell cycle regulators in cell lines undergoing MCE are involved in the regulatory process governing adipogenesis (Rosen and MacDougald, 2006; Tang and Lane, 2012; Tang et al., 2003). Although the molecular mechanisms of MCE are not fully understood, it is required for completion of adipocyte differentiation in vitro. This study suggests that Pcyt1a knockdown prior to MCE in preadipocytes inhibits subsequent adipogenesis whereas knockdown in non-replicating differentiating cells does not do so. 214    Another possible explanation is that PCYT1A activity is particularly essential for lipid droplet biogenesis, presumably to supply sufficient PC for membrane synthesis. Relevant studies have been performed in Drosophila S2 cells and mammalian macrophages (Krahmer et al., 2011). When a lipid membrane has reduced PC content, membrane-packing defects and the accumulation of anionic lipids create a favorable environment for PCYT1A binding; thus membrane-binding activates the enzyme and allows for lipid droplet expansion (Brasaemle and Wolins, 2012; Krahmer et al., 2011; Xie et al., 2004). The lipid droplet-binding of PCYT1A in vivo, however, has been debated, as most cell lines may regulate lipid droplet formation without PCYT1A-binding to the lipid droplet surface; additionally, PCYT1A localises to the nuclear membrane when cells have a demand for PC. Yeast cells are one such model. Pct1 was nuclear membrane-bound when yeast cho2∆opi3∆pct1∆ cells expressing Pct1 grew in choline-deprived medium, leading to decreased PC synthesis and large lipid droplets. However, this phenotype was not seen in human fibroblasts so its relevance is unclear at this time. 5.5.2 The impact of PCYT1A mutations on PC and related metabolites in primary patient cells Given the dramatic impact of the PCYT1A mutations on decreasing PCYT1A expression and activity (Payne et al., 2014), it is somewhat surprising that the Kennedy pathway related- metabolite levels were not more significantly altered in patients with CLD-FL and SMD-CRD (Figure 5.7). Presumably this is related to the importance of maintaining PC levels in cellular membranes and hence the role of compensatory mechanisms. Similar surprisingly subtle perturbations in PC levels have been reported in liver-specific Pcyt1a mouse knockouts (Jacobs et al., 2004), which suggests an elevation of two-fold PEMT activity compared with controls. However, in agreement with the findings that the PEMT pathway cannot substitute for the Kennedy pathway (Houweling et al., 1995) and the fact that the PEMT enzyme activity showed no increase in patient fibroblasts (Figure 5.9), it is unlikely that the PEMT pathway compensates for the Kennedy pathway to provide sufficient levels of PC synthesis. Beyond the effect of exogenous PC being supplied in a lipid-containing medium (Figure 5.8), a remaining possibility is compensation by the Lands cycle. In developing Arabidopsis thaliana seeds, it has been proposed that lpcat1 and lpcat2 double mutants responsible for PC production in the Lands cycle can trigger activity of the Kennedy pathway as a compensatory response (Wang et al., 2012) and vice versa. The Lands cycle may compensate for normal PC levels in humans. The possible connection between these two related pathways needs to be 215    studied in more detail. 5.5.3 Implications of the mutations on PCYT1A: expression and function Of the five probands studied, P0616 and P0689 have a predominantly lipodystrophic phenotype, whereas the other three patients P2233, P8069 and P8123 have SMD-CRD. Previous findings from our group have suggested that PCYT1A expression is reduced in fibroblast lines from P0616, P0689 and P8069, is slightly decreased in P8123, and remains at a normal level for P2233 (Figure 5.13). Nevertheless, the P2233 (homozygous p.A99V mutation) is predicted to be dysfunctional. Structural prediction revealed that the PCYT1A A99V mutant will perturb packing of the hydrophobic core of the catalytic domain (analysis by Professor Rosemary Cornell, University of Alberta); in addition, A99 is adjacent to the Q98 residue, which is directly required for binding the CDP moiety (Hoover-Fong et al., 2014).   Figure 5.13 Expression of mutant PCYT1A in primary cells from patients. Fibroblast cell lysates from PCYT1A mutant patients (P0689, P0616, P2233, P8069 and P8123; see also Table 5.2) and healthy controls (C103 and C105) were harvested and immunoblotted for PCYT1A and calnexin. This result was contributed by Dr Amandine Girousse, a postdoctoral fellow in the National Centre for Scientific Research, France. The reduction of PC synthesis and PCYT1A expression in P8069 (homozygous p.E129K mutation) might result from the highly conserved E129 residue being substituted in the catalytic domain (Yamamoto et al., 2014). In vitro studies have demonstrated that site-directed mutagenesis in the conserved motif of the catalytic domain of PCYT1A disrupts the enzyme activity (Kent, 2005). The V142M mutation also lies in the catalytic domain of PCYT1A and thus is very poorly expressed both in yeast cells (Figure 5.10B) and in the P0616 (p.V142M/p.E280Δ) fibroblasts. The V142M may possibly destabilise the entire structural domain, since i) V142 is predicted to be involved in dimerisation of the catalytic domain (PCYT1A functions as a dimer) and ii) the side chain of V142 sustains the highly conserved catalytic core by forming hydrophobic interactions with other residues in the catalytic domain (Payne et al., 2014). Contrary to the V142M mutant, the E280Δ mutant behaves like the wild- 216    type when expressed in yeast systems. In FRAP experiments employing Drosophila S2 cells that express the human E280Δ mutant, the mutant, though bound on the LDs, could rapidly dissociate from the membrane compared to the more stably bound wild-type PCYT1A (Payne et al., 2014). According to these observations, deletion of E280 within an amphipathic helix might weaken the protein-membrane associations of PCYT1A but does not contribute to the significant reduction in expression of this mutant in vivo. The expression of E280Δ mutant also appeared normal in COS7 cells (Payne et al., 2014). These results are in agreement with studies on yeast expressing the E280Δ mutant, which shows normal protein-expression levels and the ability to rescue cho2Δopi3Δpct1Δ cells. Another proband with the E280Δ mutant P0689 (p.E280Δ/p.333fs), has low PC-related metabolites and almost undetectable PCYT1A. The 333 frameshift mutant in this patient (P0689) leads to an extended C-terminus but results in several missing phosphorylation sites in the protein (Payne et al., 2014). Although it is unclear how the phosphorylated region in the C-terminus exerts its effects, the 333fs mutant was mislocalised in yeast (Figure 5.10C) and its expression is very low in fibroblasts (Payne et al., 2014). Another phosphorylation-region mutation close to the 333fs is the 323fs. The patient P8123 with homozygous p.323fs mutation also expresses a very low level of truncated PCYT1A protein (Yamamoto et al., 2014) (Figure 5.13). It is very likely that the significant reduction in protein expression is also the result of missing phosphorylation sites in the C-terminus. So, at this point, we have not identified any clear differences between the mutations associated with lipodystrophy and those associated with SMD-CRD. If PCYT1A expression level in primary cells is not directly linked to the altered phenotypic attributes, there could be some other factors affecting gene expression or chromatin structure in the specific targeted cells which thereby result in different disease phenotypes. Possible factors include i) epigenetic processes generating complex genetic and environmental interactions, ii) allelic variation in gene expression, iii) differential gene expression mediated by transcription factors, cofactors and chromatin regulators and iv) other environmental factors. A better understanding of the factors that govern the penetrance and expressivity of PCYT1A will be needed for a full elucidation of the complexity of the PCYT1A mutations. 217    5.5.4 Conclusions To summarise, this preliminary study aimed to begin to characterise human PCYT1A and its disease-related mutations by using both mammalian and yeast systems. Based on results from mammalian and yeast cells, it is clear that i) Pcyt1a knockdown at the mRNA level before the onset of adipocyte cell differentiation severely impairs cell differentiation; ii) primary fibroblasts from CLD-FL and SMD-CRD patients with PCYT1A mutations manifest a small reduction in CDP-choline levels as compared to healthy controls. This decrease was not associated with an increase in PEMT activity; and iii) having established a tractable yeast model, three human PCYT1A mutant alleles from CLD-FL patients were analysed. These studies reveal that the V142M and 333fs mutants are likely to reduce PCYT1A expression/ function, whereas the E280Δ mutant was not deleterious in this system. The use of this yeast system may in the future be helpful in analysing other human mutant alleles involved in PCYT1A genetic diseases. The current results are very preliminary and will require additional work in other experimental systems.     218    Chapter 6 General Discussion 6.1 Pct1 is intranuclear and relocates to the nuclear membrane in response to the need for membrane PC synthesis Multiple studies have investigated the regulation of CCT (also called PCYT1A) as it is the rate-limiting enzyme of PC synthesis and PC is essential for cell growth and organelle function. PCYT1A in vitro was proposed to be a lipid compositional sensor through reversible membrane-binding; however, exactly how cells sense and adjust their membrane PC level in vivo remain to be elucidated. The data in this thesis suggest that yeast CCT, Pct1, is confined in the nucleus where it can interconvert between a soluble inactive form and a membrane- bound form, depending on the cellular PC/PE ratio. More specifically, Pct1 associates with the inner nuclear membranes by using a nascent amphipathic helix which detects membrane- packing defects generated mainly by the presence of glycerolipids with small uncharged headgroups, primarily PE and DAG, rather than PC itself. This is broadly consistent with a substantial body of in vitro work characterising the PCYT1A activating lipids (Cornell, 2016). Lipid packing defects cannot be quantified directly whereas stored curvature elastic stress can be estimated using lipidomics and coarse grain modeling. Increased membrane SCE stress triggered increased Pct1 membrane intercalation and vice versa, thus providing an exquisite membrane surface-sensing mechanism and a feedback loop of PC content. We hypothesise that shift in membrane partitioning temporarily alleviates membrane SCE until sufficient PC is made to stabilise the membrane. In concert with this regulation, the FA acyl patterns of PC, PE and PS changed adaptively with the ratio of PC/PE. Yeast cells compensate for the low PC and elevated PE through replacing monounsaturated FAs with more saturated FAs. The saturated FAs are prone to align, helping to reduce the SCE stress or the defects in the geometrical arrangement due to the lack of PC. Figure 6.1 shows a suggested model whereby PCYT1A senses changes in the surface topology of the inner nuclear membrane, and simultaneously regulates PC synthesis. The notion that eukaryotic PCYT1A functions mainly within the nucleus is supported by the findings that Saccharomyces cerevisiae only has a nuclear CCT isoform and the FLIP data which showed that it does not exit the nucleus; moreover, our group recently observed that PCYT1A immunostaining is exclusively nuclear in several mouse tissues, including the eyes, femoral growth plate and liver, as well as adipocytes and human alveolar basal epithelial cells. PCYT1A becomes perinuclear in cells which have a high requirement for membrane remodeling. The mechanism of yeast Pct1 219    relocation to the inner nuclear membrane in response to the demand for membrane PC synthesis could therefore apply to higher organisms as well.   Figure 6.1 PCYT1A translocate onto the inner nuclear membrane in response to increased PC need. PCYT1A re-localization is governed by membrane stored curvature elastic stress.  By associating with the membrane, the amphipathic helix of PCYT1A temporarily alleviates membrane surface curvature elastic stress until sufficient PC is generated to stabilize the membrane. Adapted from (Haider et al., 2018). Activation of PCYT1A involves membrane-sensing by the disordered region of domain M, which removes the auto-inhibitory helix of M-domain from the catalytic domain, Lys122 and the αE helix. The entire M-domain then folds into a continuous amphipathic helix and lies parallel to the plane of the membrane, with its hydrophobic face inserted into the non-polar core of the lipid bilayer (Ramezanpour et al., 2018). CDP-choline produced by the nuclear enzyme is the substrate for the final reaction to make PC. The terminal enzyme might be 220    choline phosphotransferase1 (Cpt1) or choline/ethanolamine phosphotransferase1 (Ept1) catalysing the addition of a phosphocholine head group to DAG. DAG content in membranes is scarce under normal circumstances most likely owing to its non-bilayer properties. In other words, its highly conical shape can induce severe negative curvature stress in membranes. Cells try to maintain PL composition so as to keep total SCE stress within tight boundaries. To achieve this purpose, the proposed model would provide a way to eliminate the excess DAG by rapidly incorporating it into PC. Stable bilayers are critical, but nevertheless this requirement is balanced by the need for subcellular organelles to retain their ability to undergo fission/fusion events, bud off or accept vesicles, and retain unique surface-identifying features. Hence, membrane physical properties should be regulated co-operatively with a set of homeostasis sensors, involved in detecting the headgroup composition as well as the acyl chain environment. The positive membrane curvature generation by proteins with an amphipathic helix insertion should also be considered. As shown in this thesis, the total SCE stress (PSCE stress) which is contributed to by both phospholipids and proteins are maintained within a narrow range even when the cells are deficient in PC. The current study in yeast Pct1 aimed to augment recent progress in understanding how cells sense and regulate their internal membrane composition. 6.2 Compensation by nuclear localisation signal (NLS) mutants for the loss of endogenous Pct1 The compensation assays performed using Pct1-NLS mutants suggest that cytosolic Pct1 can compensate for the absence of wild-type enzyme. These findings are supported by the results of PL composition changes. Based upon the lipidomic analysis, SCE stress was estimated to confirm the simultaneous occurrence of Pct1-NLS membrane translocation and PC synthesis. In the absence of choline, the NLS mutant localised to the lipid droplet surface. When choline was supplemented, PC synthesis and the entire cell PC/PE ratio increased in concert with the NLS mutant translocating to the plasma membrane, in keeping with the notion of the feedback regulation of Pct1 activity by PC. Pct1 should act as a general sensor of PC deficiency, either in a droplet monolayer or in a lipid bilayer. The lipid droplet monolayer surface contain mostly PC and PE, and rarely PS (Bartz et al., 2007), suggesting that the lipid droplet surface will harbour many packing defects due to the lack of PC. Additionally, more TAG accumulates in the expanding lipid droplet when cells stop making PC and start converting DAG to TAG. Pct1-NLS mutants were recruited to lipid droplets, thus allowing the 221    production of more PC molecules to surround the oil core. These factors may account for the Pct1-NLS mutants bound to the lipid droplet membrane, rather than other organelle membranes, during choline starvation.  After choline supplementation,  the Pct1-NLS mutants were forced off by a PC-rich membrane or crowded membrane proteins in shrinking droplets, and the mutants were then attracted by the PS-rich plasma membrane, where the negatively- charged PS could interact with the positively-charged ‘leash’ region of the M-domain. However, the electrostatic interaction may be affected by the phosphorylation status of the P- region and the mechanism by which the NLS mutants continuously bound to the plasma membranes (membrane-bound form) in the PC-rich cells remain largely unexplained. This apparent lack of requirement for nuclear Pct1 agrees with an earlier study in which expression of a partially extranuclear form of PCYT1A missing the NLS supported PC synthesis and growth in CCTα-deficient CHO58 cells (Wang et al., 1995); furthermore, a CCT1-NLS mutant in fruit flies was cytosolic and could maintain normal rhabdomere morphology in the CCT1 null eyes (Haider et al., 2018). In vivo in cells which express CCTthis may also contribute to total cellular PC synthesis. Another important fact is that the nuclear envelope is continuous with the ER. This continuity can probably facilitate PL diffusion from within the nucleus to the peripheral ER and, vice versa, thus the NLS mutants of Pct1 in yeast, PCYT1A in rat and CCT1 in flies can compensate for the absence of a wild- type/nuclear enzyme. In agreement with these findings, my host group recently demonstrated that overexpressed CCT2 (CCT form in fruit flies) can rescue rhabdomere formation in CCT1/2 knockout photoreceptors. A potential limitation of the Pct1-NLSm1 mutant and CCT2 studies was that they appeared to be detectable in the nucleus to some extent. It is thus possible that lower levels of nuclear Pct1-NSLm1 or CCT2 are sufficient to rescue the null phenotypes. However, Pct1-NLS1del in which most of the amino residues comprising the NLS of Pct1 has been deleted, was still able to rescue the 3Δ phenotype, and FLIP analysis does not suggest significant amounts of Pct1-NLSm1 traffic into the nucleus anyway, so we think that cytosolic CCT forms can rescue CCT knockout cells. 6.3 Disease-related human PCYT1A mutations impair PC synthesis in primary patient cells and adequate PC synthesis is essential for adipogenesis The investigation of PCYT1A function on lipid droplet growth and PC synthesis emerged from a previous study, which reported the association of PCYT1A mutations with human 222    lipodystrophy characterised by reduced PCYT1A expression and PC synthesis (Payne et al., 2014).  In this follow up study, I have used EMSCs to show that PCYT1A knockdown prior to adipocyte differentiation severely impairs adipogenesis.  The combined mass spectroscopy evidence and cell culture results revealed a reduction of CDP-choline in patient cells (P0616 and P0689, two patients with compound heterozygous mutations in PCYT1A and a diagnosis of CLD-FL; P2233, P8069 and P8123, another patient with a homozygous mutation in PCYT1A and a diagnosis of SMD-CRD) but the decrease was not compensated for by altered PEMT activity. Using a yeast complementation assay, the human disease alleles V142M and 333fs were found to have a marked reduction in enzyme function. Although the preliminary results support the role of PCYT1A in adipocyte maturation and PC synthesis, further studies are required to gain a more complete understanding of other mutant variants responsible for SMD-CRD and LCA.   The CLD-FL patients manifest not only adipose tissue failure but also non-alcoholic fatty liver disease (NAFLD). Due to the lack of appropriate fat storage sites, ectopic fat deposition in the liver appears to be a typical complication. Nevertheless, CLD-FL patients were diagnosed with severe steatosis. This suggests that PC might be directly involved in fatty acid metabolism. Previous studies did indeed find that SREBP1, which can bind to specific DNA sequences (sterol regulatory elements) in the genes that encode enzymes needed to make fatty acids, can form a regulatory circuit by which SREBP1 controls its own activation by reduced PC level or SAM in C. elegans and human liver cells (Walker et al., 2011). The reduction in the PC level resulted from the knockdown of PCYT1A expression. SAM is the methyl donor for PC synthesis in the PEMT pathway, and its production is controlled by SREBP1. Activated SREBP1c led to lipid droplet accumulation, in keeping with the steatosis seen in CLD-FL patients and in hepatic-specific Pcyt1a knockout mice (Jacobs et al., 2004). The regulatory role of SAM suggests that the PEMT pathway is also important for hepatic PC production, although the PEMT pathway only contributes to about 30% of hepatic PC (DeLong et al., 1999). Within the context of this thesis, the PEMT pathway shows no compensation effect for CLD-FL patients and has a comparable function in maintaining PC species to that of the Kennedy pathway in yeast. Studies in Pemt-deficient mice displayed a normal phenotype in hepatocyte morphology, plasma lipid levels and bile composition (Walkey et al., 1997); however, Pemt-deficient mice fed a choline-deficient diet suffered from severe liver damage with 50% reduction in PC synthesis, and the experiments had to be halted after 3 days to prevent pain or death (Walkey et al., 1998). Steatosis also occurred in choline- 223    deficient diet-fed mice that lacked both Pemt and the multiple drug-resistant protein 2 (Mdr2) (Li et al., 2006; Li et al., 2005). Mdr2 is distributed at the bile canalicular membrane of hepatocytes and is a PC flippase mediating PC secretion into bile (Smit et al., 1993). The hepatic PC content decreased by 50% in Pemt-/-/Mdr2-/- mice fed the choline-deficient diet for 21 days (Li et al., 2006; Li et al., 2005). It is surprising that, under same choline-deficient diet, the double knockout Pemt−/−/Mdr2−/− mice died after 90 days, whereas the Pemt−/−/Mdr2+/+ mice survived only 3 days (Li et al., 2006). The lack of Mdr2 responsible for transporting PC from hepatocyte to bile may account for the ability to preserve the hepatic PC in the double knockout mice; thus, the lower consumption of the hepatic PC content prolonged their survival or prolonged the maintenance of the PC/PE ratio. In general, the shortage of hepatic PC plays a role in the progression of liver failure. Liver PC needed to be supplied by the PEMT pathway when the Kennedy pathway was absent. Hepatic PC is also required for lipoprotein secretion. In hepatocytes from male Pemt-/-, compared with Pemt+/+ mice, VLDL secretion was reduced (Noga and Vance, 2003). Subsequent studies of rats fed with choline-deficient diets showed the inhibitory effect of PC deficiency on VLDL secretion (Yao and Vance, 1988; Yao and Vance, 1990). In the liver- specific Pcyt1a knockout mice, hepatic PCYT1A activity was decreased by 85%,  resulting in a reduction in the production of plasma PC, cholesterol, TAG, VLDL and HDL (Jacobs et al., 2008). This notion was reinforced by the re-introduction of hepatic PCYT1Athrough adenoviral delivery, which restored hepatic PC concentration as well as plasma VLDL and HDL levels (Jacobs et al., 2008). The need for PC synthesis to sustain lipoprotein production is supported by the observation that hepatic PCYT1A was localised to the NE in both fed and fasting mice, whereas hepatic PCYT1A was intranuclear in liver sections from liver-specific Mtp (microsomal triglyceride transfer protein) knockout mice (Haider et al., 2018), which had impaired lipoprotein biosynthesis (Khatun et al., 2012). 6.4 Future directions 6.4.1 Characterisation of the nuclear membrane: PL composition and stored curvature elastic stress An important limitation of the data in Chapter 3 of this thesis relates to the fact that lipidomic analysis was performed on whole cell lysates, whereas we now know that Pct1 is confined to the nucleus. It is, therefore, clear that lipidomic analysis of the nuclear membrane would provide an opportunity to refine the relationship between the phospholipid composition and 224    Pct1 distribution. The purification of nuclear membranes is feasible but challenging. Previous studies have shown that nuclei can be isolated from cell homogenates by saccharide-gradient centrifugation steps, and then the isolated nuclei can be the starting material to separate the nuclear membrane through subsequent sucrose-gradient steps (Dove et al., 1998; Humbert et al., 1996; Masmoudi et al., 1989). This purification process still has to be optimised for yeast; additionally, the compositional values given would be approximate because isolation of pure organelles is usually confounded by contamination of the enriched organelle fraction and other organelle membranes. Nevertheless, it would be valuable to attempt to investigate the curvature elastic stress and the electronic environment in the nuclear membrane. An interesting idea proposed by Cornell and Antonny was that localisation within the nucleus might enable Pct1/ PCYT1A to be a relatively ‘pure’ sensor of packing defects by conically shaped phospholipids, i.e. PE and DAG (Cornell and Antonny, 2018). The nuclear envelope is relatively flat compared with other smaller organelles. Furthermore, the nuclear membrane might be comparatively neutral relative to the plasma membrane since PS seems more abundant on the plasma membrane (Fairn et al., 2011). This hypothesis could be verified by lipidomic analysis or by modification of amino acid residues of the AH, for example, by substituting bulky hydrophobic amino acids with serine, and examining the Pct1-GFP binding difference in the nuclear envelope under microscopy. The AH region of Caenorhabditis elegans CCT has shown that when hydrophobic amino acids were changed to serine, lipid activation of this enzyme was reduced (Braker et al., 2009). Another potential direction is to modify the positively-charged residues at the interface between the polar and non-polar surfaces of the helix and detect the attractive effect of the negatively-charged membrane on Pct1-GFP and NLS-deleted Pct1-GFP. Previous studies have found that substitutions of lysines to glutamine residues in this region of the rat CCT reduce the electrostatic binding to anionic lipid vesicles (Johnson et al., 2003), and decrease membrane affinity in response to oleic acid (Gehrig et al., 2008). 6.4.2 Validation of the DNA damage protection function of yeast Pct1/human PCYT1A pct1Δ cells do not display any growth defect under normal conditions but they show enhanced sensitivity to phleomycin, a cytotoxic agent that causes DNA double strand breaks (DSB). Notably, this effect is not in cells lacking the yeast equivalent PEMT i.e. Cho2 and Opi3. Human PCYT1A, but not the lipodystrophy-associated V142M mutation, can rescue pct1Δ 225    DNA damage sensitivity, implying the conservation of its DNA damage protective function. Furthermore, cells lacking the enzyme catalysing the final step in the Kennedy pathway, cpt1Δ, behave like pct1Δ, suggesting that DNA damage sensitivity is due to the lack of this PC pool. It is unknown whether pct1Δ cells or other related strains are sensitive to other DSB agents, and if the DNA damage sensitivity is reproducible in other CCTα-null mammalian cell models. Dr Symeon Siniossoglou at the Cambridge Institute for Medical Research has recently found that the pct1Δ strain is not only sensitive to phleomycin but also to methyl methanesulfonate (MMS). MMS is an alkylation agent capable of inducing heat-labile DSB (Lundin et al., 2005). A range of other DNA-damaging agents (Blanpain et al., 2011), such as ionising radiation, ultraviolet, H2O2, and topoisomerase inhibitors like camptothecin or etoposide might be examined. Previous studies have revealed that persistent DSB or collapsed replication forks lead to the DNA lesion recruiting the nuclear envelope and nuclear pore complexes in order to facilitate repair (Bermejo et al., 2012; Seeber and Gasser, 2017). Through additional investigation of the combined sensitivity to MMS and hydroxyurea, which results in stalled replication forks that are recruited to the NE (Nagai et al., 2008), this may shed light on whether Pct1 has a role in the repair of DSB that are induced by replicative fork collapse. Fluorescent microscopy carried out in fission yeast would investigate whether Pct1 and NPCs co-localise during the time course of phleomycin treatment.  Nuclear pore components such as Nup84 (Lutzmann et al., 2002) could be used as an NPC marker. A key protein for DSB repair like Rad52 (Resnick and Martin, 1976) could be used as a DNA damage marker. If Pct1 is found to be redistributed in response to DNA damage, the localisation of the NLS-deleted Pct1 should next be examined, and this could determine whether or not Pct1 is directly involved in DNA repair. Strains with mutations in Rad52 or mutations in nucleoporins could be tested for DNA damage sensitivity alone or combined with pct1Δ to define if DNA repair is a joint process involving these components. Co- immunoprecipitation of Pct1 from yeast cells, followed by mass spectrometry analysis would be an alternative way to identify potential interacting partners. Interestingly, there is evidence that sustained DNA damage can cause metabolic disorders including lipodystrophy (Karakasilioti et al., 2013; Weedon et al., 2013). Therefore, a clear line of study could investigate whether primary patient fibroblasts carrying PCYT1A mutations are susceptible to DNA damage and whether CCT-deficient mammalian cells display DNA damage sensitivity. Previous attempts in our group have shown that RNA interference can be used for efficient knockdown of CCT in HeLa and Huh7 cells. Primary 226    patient fibroblasts and CCT-deficient mammalian cells could also be examined for their sensitivity to a series of DNA-damaging agents through the expression level of the DNA damage markers. For example, detecting the immunofluorescence of the phosphorylated histone H2AX (γH2AX) (Smart et al., 2008) as well as the phosphorylation and activation of 53BP1, ATM and CHK2 (Bartkova et al., 2006; Gupta et al., 2014; Zannini et al., 2014). Similar experiments could be performed with fibroblasts expressing SMD-CRD or LCA mutations; additionally, these mutant alleles could be studied in the context of the phleomycin-treated pct1Δ cells and the FOA-treated 3Δ cells. Previous studies have shown that the mammalian CCT and lamins regulate the proliferation of the nuclear reticulum in co-ordination (Gehrig et al., 2008; Lagace and Ridgway, 2005b), while the lack of lamin A/C disrupts repair factor recruitment and patients with lamin A mutations accumulate unrepaired DNA damage and genome instability (Dobrzynska et al., 2016; Singh et al., 2013). It is proposed that persistent unrepaired DSBs trigger changes to the lamin and cytoskeleton and/or to its nuclear envelope contact points, which would, in turn, regulate DSBs anchored to the nuclear pore complex, where DNA repair proteins are localised (Dellaire et al., 2009; Guenole and Legube, 2017). Through collaboration with a super-resolution electron microscopy team in the University of Oxford, our group has started to investigate the nuclear membrane morphology of patient fibroblasts carrying PCYT1A mutations. 6.4.3 Investigation of Cpt1 localisation Cpt1 is the final enzyme in the Kennedy pathway, i.e. it actually generates PC. However, it is elusive whether Cpt1 is an ER protein and whether Pct1 and Cpt1 co-localise or not.  Yeast Cpt1 has been described as localising to the mitochondria outer membrane or Golgi (Kuchler et al., 1986; Leber et al., 1995), whereas the GFP-tagged yeast Cpt1 has no co-localisation with any organelle markers (Huh et al., 2003). Trypanosoma brucei CEPT localises to the ER and the perinuclear region (Farine et al., 2015). In mammals, CPT1 was found in Golgi, while CEPT1 (EPT1) showed a dual distribution between the ER and the nuclear envelope (Henneberry et al., 2002). Immunogold electron microscopy staining carried out in yeasts could potentially demonstrate the exact localisation of Cpt1. Alternatively, Cpt1-GFP localisation could be examined under confocal microscopy using Sec63-mCherry as an ER marker. If necessary, the anchor-away system can assist effects to differentiate the ER and INM location, and the Δ tether strain can help to distinguish the PM from cortical ER. If the 227    above reveal Cpt1 is an ER-targeting protein, one could hypothesise that Pct1 is retained in the nucleus in order to facilitate spatial/temporal regulation of the rate-limiting step. The regulation of PC synthesis employs a notable feature of eukaryotic cells, namely subcellular compartmentalisation to ensure that synthesis occurs only when needed. Upon choline starvation, the lipid-packing properties would first be sensed by Pct1 at the inner nuclear membrane and this would trigger Pct1 to generate CDP-choline; secondly, the CDP-choline would translocate from the nucleus to the ER. At the ER membrane, Cpt1 catalyses the transfer of phosphocholine from CDP-choline to DAG, and thus forms PC.   6.4.4 Characterisation of the phenotype of cells affected by disease- associated PCYT1A mutations Despite the role of PCYT1A in early adipocyte differentiation has been confirmed, the potential tissue-specific phenotypes of PCYT1A mutations are unknown. To understand better how PCYT1A could affect retinal and bone development, i) long-term follow-up and a large cohort of patients with PCYT1A mutations would be invaluable; ii) characterisation of the gene and PC synthesis in representative cell models will be required. For example, we could investigate the photoreceptor degeneration in CCT1-deficient Drosophila eyes expressing the mutant PCYT1A gene. With the refinement of the photoreceptor model, the similar mutation sites but different phenotypes in the current results may be more clearly resolved. Another possibility is to use CRISPR/Cas9-based gene manipulation in human induced pluripotent stem cells (Angelos and Kaufman, 2015), where a gene could be knocked out or modified into disease mutants (Zhang et al., 2017), and this technique could be applied for PCYT1A studies. Additionally, induced pluripotent stem cells have the potential to differentiate into adipogenic (Guenantin et al., 2017; Hafner and Dani, 2014), chondrogenic (Nejadnik et al., 2015) and osteogenic (Jeon et al., 2016) lineages. Expression of disease-related PCYT1A mutations in those cell lines could feature the tissue-specific phenotypes as well as distinguishing the impact on different tissues. 228    6.5 Concluding remarks My thesis work has contributed to the discovery that yeast Pct1 (and probably CCT in most cell types), the rate-limiting enzyme of PC synthesis, translocate onto the inner nuclear membrane in response to increased PC synthesis needs in cells. This re-localisation of Pct1 is governed by membrane stored curvature elastic stress. PC deficiency increases membrane stored curvature elastic stress, and Pct1 uses a nascent amphipathic helix to sense these changes. This shift in membrane partitioning is due to the presence of glycerolipids with small uncharged headgroups, primarily PE and DAG. The process provides an exquisite membrane surface-sensing mechanism which operates cooperatively with sensors of the intramembranous fatty acyl chain environment. Budding yeast compensates for low PC and elevated PE by replacing monounsaturated fatty acids with saturated fatty acids predominantly in the PE lipids. Although Pct1 is largely confined in the nucleus, Pct1 mutants lacking the nuclear localisation signal can rescue Pct1 deficient cells as well as regulating PC synthesis in a similar way to wild-type Pct1. The rescue assay was also applied to examine some human mutant alleles responsible for PCYT1A genetic diseases,  pointing out which variants may impair PC synthesis. Through the investigation of CCT from yeast to humans, these data reflect the importance of this enzyme in maintaining phospholipid homeostasis.     229    References Ahmadian,  M.,  R.E.  Duncan,  K.  Jaworski,  E.  Sarkadi‐Nagy,  and  H.S.  Sul.  2007.  Triacylglycerol  metabolism in adipose tissue. Future lipidology. 2:229‐237.  Ahuatzi, D., A. Riera, R. Pelaez, P. Herrero, and F. Moreno. 2007. Hxk2 regulates the phosphorylation  state  of  Mig1  and  therefore  its  nucleocytoplasmic  distribution.  The  Journal  of  biological  chemistry. 282:4485‐4493.  Aitchison,  A.J.,  D.J.  Arsenault,  and  N.D.  Ridgway.  2015.  Nuclear‐localized  CTP:phosphocholine  cytidylyltransferase alpha  regulates phosphatidylcholine synthesis  required  for  lipid droplet  biogenesis. Molecular biology of the cell. 26:2927‐2938.  Alberts B, J.A., Lewis J, et al. . 2002. The Lipid Bilayer. In Molecular Biology of the Cell. 4th edition. .  New York: Garland Science.  Allen,  T.D.,  J.M.  Cronshaw,  S.  Bagley,  E.  Kiseleva,  and  M.W.  Goldberg.  2000.  The  nuclear  pore  complex: mediator of translocation between nucleus and cytoplasm. Journal of cell science.  113 ( Pt 10):1651‐1659.  Angelos, M.G., and D.S. Kaufman. 2015. Pluripotent stem cell applications for regenerative medicine.  Current opinion in organ transplantation. 20:663‐670.  Antonny,  B.  2011.  Mechanisms  of  membrane  curvature  sensing.  Annual  review  of  biochemistry.  80:101‐123.  Araki,  W.,  and  R.J.  Wurtman.  1997.  Control  of  membrane  phosphatidylcholine  biosynthesis  by  diacylglycerol  levels  in  neuronal  cells  undergoing  neurite  outgrowth.  Proceedings  of  the  National Academy of Sciences of the United States of America. 94:11946‐11950.  Arnold,  R.S.,  and  R.B.  Cornell.  1996.  Lipid  regulation  of  CTP:  phosphocholine  cytidylyltransferase:  electrostatic,  hydrophobic,  and  synergistic  interactions  of  anionic  phospholipids  and  diacylglycerol. Biochemistry. 35:9917‐9924.  Arnold,  R.S.,  A.A.  DePaoli‐Roach,  and  R.B.  Cornell.  1997.  Binding  of  CTP:phosphocholine  cytidylyltransferase  to  lipid  vesicles: diacylglycerol  and enzyme dephosphorylation  increase  the affinity for negatively charged membranes. Biochemistry. 36:6149‐6156.  Ashida,  H.,  Y.  Maeda,  and  T.  Kinoshita.  2006.  DPM1,  the  catalytic  subunit  of  dolichol‐phosphate  mannose synthase, is tethered to and stabilized on the endoplasmic reticulum membrane by  DPM3. The Journal of biological chemistry. 281:896‐904.  Athenstaedt, K.  2014.  Isolation and  characterization of  lipid droplets  from yeast. Hydrocarbon and  lipid microbiology protocols. 10:81‐91.  Attard,  G.S.,  R.H.  Templer,  W.S.  Smith,  A.N.  Hunt,  and  S.  Jackowski.  2000.  Modulation  of  CTP:phosphocholine cytidylyltransferase by membrane curvature elastic stress. Proceedings  of the National Academy of Sciences of the United States of America. 97:9032‐9036.  Bahmanyar, S. 2015. Spatial regulation of phospholipid synthesis within the nuclear envelope domain  of the endoplasmic reticulum. Nucleus. 6:102‐106.  230    Bakovic,  M.,  K.  Waite,  and  D.E.  Vance.  2003.  Oncogenic  Ha‐Ras  transformation  modulates  the  transcription  of  the  CTP:phosphocholine  cytidylyltransferase  alpha  gene  via  p42/44MAPK  and transcription factor Sp3. The Journal of biological chemistry. 278:14753‐14761.  Banchio, C., L.M. Schang, and D.E. Vance. 2004. Phosphorylation of Sp1 by cyclin‐dependent kinase 2  modulates the role of Sp1 in CTP:phosphocholine cytidylyltransferase alpha regulation during  the S phase of the cell cycle. The Journal of biological chemistry. 279:40220‐40226.  Barbosa, A.D., H. Sembongi, W.M. Su, S. Abreu, F. Reggiori, G.M. Carman, and S. Siniossoglou. 2015.  Lipid partitioning at the nuclear envelope controls membrane biogenesis. Molecular biology  of the cell. 26:3641‐3657.  Bartkova, J., N. Rezaei, M. Liontos, P. Karakaidos, D. Kletsas, N. Issaeva, L.V. Vassiliou, E. Kolettas, K.  Niforou,  V.C.  Zoumpourlis, M.  Takaoka,  H.  Nakagawa,  F.  Tort,  K.  Fugger,  F.  Johansson, M.  Sehested,  C.L.  Andersen,  L.  Dyrskjot,  T.  Orntoft,  J.  Lukas,  C.  Kittas,  T.  Helleday,  T.D.  Halazonetis, J. Bartek, and V.G. Gorgoulis. 2006. Oncogene‐induced senescence is part of the  tumorigenesis barrier imposed by DNA damage checkpoints. Nature. 444:633‐637.  Bartz,  R.,  W.H.  Li,  B.  Venables,  J.K.  Zehmer,  M.R.  Roth,  R.  Welti,  R.G.  Anderson,  P.  Liu,  and  K.D.  Chapman.  2007.  Lipidomics  reveals  that  adiposomes  store  ether  lipids  and  mediate  phospholipid traffic. Journal of lipid research. 48:837‐847.  Basu Ball, W., J.K. Neff, and V.M. Gohil. 2018. The role of nonbilayer phospholipids in mitochondrial  structure and function. FEBS letters. 592:1273‐1290.  Bermejo,  R.,  A.  Kumar,  and  M.  Foiani.  2012.  Preserving  the  genome  by  regulating  chromatin  association with the nuclear envelope. Trends in cell biology. 22:465‐473.  Bevers,  E.M.,  P.  Comfurius,  J.L.  van  Rijn,  H.C.  Hemker,  and  R.F.  Zwaal.  1982.  Generation  of  prothrombin‐converting activity and the exposure of phosphatidylserine at the outer surface  of platelets. European journal of biochemistry. 122:429‐436.  Bigay, J., and B. Antonny. 2012. Curvature, lipid packing, and electrostatics of membrane organelles:  defining cellular territories in determining specificity. Developmental cell. 23:886‐895.  Bigay,  J.,  J.F.  Casella,  G.  Drin,  B. Mesmin,  and  B.  Antonny.  2005.  ArfGAP1  responds  to membrane  curvature  through  the  folding of  a  lipid packing  sensor motif. The  EMBO  journal.  24:2244‐ 2253.  Blanpain, C., M. Mohrin, P.A. Sotiropoulou, and E. Passegue. 2011. DNA‐damage response in tissue‐ specific and cancer stem cells. Cell stem cell. 8:16‐29.  Boeke,  J.D.,  F.  LaCroute,  and G.R.  Fink. 1984. A positive  selection  for mutants  lacking orotidine‐5'‐ phosphate  decarboxylase  activity  in  yeast:  5‐fluoro‐orotic  acid  resistance.  Molecular  &  general genetics : MGG. 197:345‐346.  Bogan, M.J., G.R. Agnes, F. Pio, and R.B. Cornell. 2005.  Interdomain and membrane  interactions of  CTP:phosphocholine  cytidylyltransferase  revealed  via  limited  proteolysis  and  mass  spectrometry. The Journal of biological chemistry. 280:19613‐19624.  Boggs,  K.P.,  C.O.  Rock,  and  S.  Jackowski.  1995.  Lysophosphatidylcholine  and  1‐O‐octadecyl‐2‐O‐ methyl‐rac‐glycero‐3‐phosphocholine  inhibit  the  CDP‐choline  pathway  of  231    phosphatidylcholine  synthesis  at  the  CTP:phosphocholine  cytidylyltransferase  step.  The  Journal of biological chemistry. 270:7757‐7764.  Bork,  P.,  L.  Holm,  E.V.  Koonin,  and  C.  Sander.  1995.  The  cytidylyltransferase  superfamily:  identification of the nucleotide‐binding site and fold prediction. Proteins. 22:259‐266.  Bouchez,  I.,  M.  Pouteaux, M.  Canonge, M.  Genet,  T.  Chardot,  A.  Guillot,  and M.  Froissard.  2015.  Regulation of  lipid droplet dynamics  in Saccharomyces cerevisiae depends on the Rab7‐like  Ypt7p, HOPS complex and V1‐ATPase. Biology open. 4:764‐775.  Boumann, H.A., M.J. Damen, C. Versluis, A.J. Heck, B. de Kruijff,  and A.I.  de Kroon. 2003.  The  two  biosynthetic  routes  leading  to  phosphatidylcholine  in  yeast  produce  different  sets  of  molecular species. Evidence for lipid remodeling. Biochemistry. 42:3054‐3059.  Boumann, H.A.,  J. Gubbens, M.C.  Koorengevel,  C.S. Oh,  C.E. Martin,  A.J.  Heck,  J.  Patton‐Vogt,  S.A.  Henry,  B.  de  Kruijff,  and  A.I.  de  Kroon.  2006.  Depletion  of  phosphatidylcholine  in  yeast  induces shortening and increased saturation of the lipid acyl chains: evidence for regulation  of intrinsic membrane curvature in a eukaryote. Molecular biology of the cell. 17:1006‐1017.  Braker, J.D., K.J. Hodel, D.R. Mullins, and J.A. Friesen. 2009. Identification of hydrophobic amino acids  required for  lipid activation of C. elegans CTP:phosphocholine cytidylyltransferase. Archives  of biochemistry and biophysics. 492:10‐16.  Brasaemle, D.L., and N.E. Wolins. 2012. Packaging of fat: an evolving model of lipid droplet assembly  and expansion. The Journal of biological chemistry. 287:2273‐2279.  Brighton,  C.T.,  Y.  Sugioka,  and  R.M.  Hunt.  1973.  Cytoplasmic  structures  of  epiphyseal  plate  chondrocytes.  Quantitative  evaluation  using  electron  micrographs  of  rat  costochondral  junctions with  special  reference  to  the  fate of  hypertrophic  cells. The  Journal  of  bone and  joint surgery. American volume. 55:771‐784.  Broers, J.L., B.M. Machiels, G.J. van Eys, H.J. Kuijpers, E.M. Manders, R. van Driel, and F.C. Ramaekers.  1999. Dynamics of the nuclear lamina as monitored by GFP‐tagged A‐type lamins. Journal of  cell science. 112 ( Pt 20):3463‐3475.  Burr, R., E.V. Stewart, and P.J. Espenshade. 2017. Coordinate Regulation of Yeast Sterol Regulatory  Element‐binding Protein  (SREBP)  and Mga2 Transcription  Factors. The  Journal  of  biological  chemistry. 292:5311‐5324.  Cardenas,  J.,  S.  Rivero,  B.  Goud, M.  Bornens,  and  R.M.  Rios.  2009.  Golgi  localisation  of  GMAP210  requires two distinct cis‐membrane binding mechanisms. BMC biology. 7:56.  Carman,  G.M.,  and  S.A.  Henry.  1999.  Phospholipid  biosynthesis  in  the  yeast  Saccharomyces  cerevisiae  and  interrelationship with  other metabolic  processes. Progress  in  lipid  research.  38:361‐399.  Carter,  J.M.,  L.  Demizieux,  R.B.  Campenot,  D.E.  Vance,  and  J.E.  Vance.  2008.  Phosphatidylcholine  biosynthesis via CTP:phosphocholine cytidylyltransferase 2 facilitates neurite outgrowth and  branching. The Journal of biological chemistry. 283:202‐212.  Chen, B.B., and R.K. Mallampalli. 2007. Calmodulin binds and stabilizes the regulatory enzyme, CTP:  phosphocholine cytidylyltransferase. The Journal of biological chemistry. 282:33494‐33506.  232    Chen, B.B., and R.K. Mallampalli. 2009. Masking of a nuclear signal motif by monoubiquitination leads  to mislocalization and degradation of the regulatory enzyme cytidylyltransferase. Molecular  and cellular biology. 29:3062‐3075.  Chen, H., S. Fre, V.I. Slepnev, M.R. Capua, K. Takei, M.H. Butler, P.P. Di Fiore, and P. De Camilli. 1998.  Epsin  is an EH‐domain‐binding protein  implicated  in clathrin‐mediated endocytosis. Nature.  394:793‐797.  Chin, J., and K. Bloch. 1988. Phosphatidylcholine synthesis in yeast. Journal of lipid research. 29:9‐14.  Chong, S.S., S.G. Taneva, J.M. Lee, and R.B. Cornell. 2014. The curvature sensitivity of a membrane‐ binding  amphipathic  helix  can  be  modulated  by  the  charge  on  a  flanking  region.  Biochemistry. 53:450‐461.  Chook,  Y.M.,  and  G.  Blobel.  2001.  Karyopherins  and  nuclear  import.  Current  opinion  in  structural  biology. 11:703‐715.  Christie, M., C.W. Chang, G. Rona, K.M. Smith, A.G. Stewart, A.A. Takeda, M.R. Fontes, M. Stewart,  B.G. Vertessy, J.K. Forwood, and B. Kobe. 2016. Structural Biology and Regulation of Protein  Import into the Nucleus. Journal of molecular biology. 428:2060‐2090.  Chumnanpuen, P.,  J.  Zhang,  I. Nookaew, and  J. Nielsen. 2012.  Integrated analysis of  transcriptome  and  lipid  profiling  reveals  the  co‐influences  of  inositol‐choline  and  Snf1  in  controlling  lipid  biosynthesis in yeast. Molecular genetics and genomics : MGG. 287:541‐554.  Clancey,  C.J.,  S.C.  Chang,  and  W.  Dowhan.  1993.  Cloning  of  a  gene  (PSD1)  encoding  phosphatidylserine decarboxylase from Saccharomyces cerevisiae by complementation of an  Escherichia coli mutant. The Journal of biological chemistry. 268:24580‐24590.  Coderch,  L., O.  Lopez, A.  de  la Maza,  and  J.L.  Parra.  2003.  Ceramides  and  skin  function. American  journal of clinical dermatology. 4:107‐129.  Cole,  L.K.,  J.E.  Vance,  and  D.E.  Vance.  2012.  Phosphatidylcholine  biosynthesis  and  lipoprotein  metabolism. Biochimica et biophysica acta. 1821:754‐761.  Conrad, M.,  J.  Schothorst, H.N. Kankipati, G. Van Zeebroeck, M. Rubio‐Texeira, and  J.M. Thevelein.  2014.  Nutrient  sensing  and  signaling  in  the  yeast  Saccharomyces  cerevisiae.  FEMS  microbiology reviews. 38:254‐299.  Conti,  E.,  M.  Uy,  L.  Leighton,  G.  Blobel,  and  J.  Kuriyan.  1998.  Crystallographic  analysis  of  the  recognition of a nuclear  localization  signal by  the nuclear  import  factor karyopherin alpha.  Cell. 94:193‐204.  Cooley,  C.M.,  K.S.  Hettie,  J.L.  Klockow,  S.  Garrison,  and  T.E.  Glass.  2013.  A  selective  fluorescent  chemosensor for phosphoserine. Organic & biomolecular chemistry. 11:7387‐7392.  Cornell, R., and B. Antonny. 2018. CCTalpha Commands Phospholipid Homeostasis from the Nucleus.  Developmental cell. 45:419‐420.  Cornell,  R.,  and  D.E.  Vance.  1987.  Translocation  of  CTP:  phosphocholine  cytidylyltransferase  from  cytosol  to  membranes  in  HeLa  cells:  stimulation  by  fatty  acid,  fatty  alcohol,  mono‐  and  diacylglycerol. Biochimica et biophysica acta. 919:26‐36.  233    Cornell,  R.B.  1991.  Regulation  of  CTP:phosphocholine  cytidylyltransferase  by  lipids.  2.  Surface  curvature,  acyl  chain  length,  and  lipid‐phase  dependence  for  activation.  Biochemistry.  30:5881‐5888.  Cornell, R.B. 2016. Membrane lipid compositional sensing by the inducible amphipathic helix of CCT.  Biochimica et biophysica acta. 1861:847‐861.  Cornell, R.B., G.B. Kalmar, R.J. Kay, M.A. Johnson, J.S. Sanghera, and S.L. Pelech. 1995. Functions of  the  C‐terminal  domain  of  CTP:  phosphocholine  cytidylyltransferase.  Effects  of  C‐terminal  deletions  on  enzyme  activity,  intracellular  localization  and  phosphorylation  potential.  The  Biochemical journal. 310 ( Pt 2):699‐708.  Cornell,  R.B.,  and  I.C.  Northwood.  2000.  Regulation  of  CTP:phosphocholine  cytidylyltransferase  by  amphitropism and relocalization. Trends in biochemical sciences. 25:441‐447.  Cornell, R.B., and N.D. Ridgway. 2015. CTP:phosphocholine cytidylyltransferase: Function, regulation,  and structure of an amphitropic enzyme required for membrane biogenesis. Progress in lipid  research. 59:147‐171.  Cornell, R.B., and S.G. Taneva. 2006. Amphipathic helices as mediators of the membrane interaction  of amphitropic proteins, and as modulators of bilayer physical properties. Current protein &  peptide science. 7:539‐552.  Covino,  R.,  S.  Ballweg,  C.  Stordeur,  J.B.  Michaelis,  K.  Puth,  F.  Wernig,  A.  Bahrami,  A.M.  Ernst,  G.  Hummer, and R. Ernst. 2016. A Eukaryotic Sensor for Membrane Lipid Saturation. Molecular  cell. 63:49‐59.  Craig, L., J.E. Johnson, and R.B. Cornell. 1994. Identification of the membrane‐binding domain of rat  liver CTP:phosphocholine cytidylyltransferase using chymotrypsin proteolysis. The Journal of  biological chemistry. 269:3311‐3317.  Cullen,  P.J.,  and  G.F.  Sprague,  Jr.  2012.  The  regulation  of  filamentous  growth  in  yeast.  Genetics.  190:23‐49.  D'Angelo, G., S. Capasso, L. Sticco, and D. Russo. 2013. Glycosphingolipids: synthesis and functions.  The FEBS journal. 280:6338‐6353.  Daleke, D.L. 2003. Regulation of transbilayer plasma membrane phospholipid asymmetry. Journal of  lipid research. 44:233‐242.  Daum, G., G. Tuller, T. Nemec, C. Hrastnik, G. Balliano, L. Cattel, P. Milla, F. Rocco, A. Conzelmann, C.  Vionnet, D.E. Kelly,  S. Kelly, E.  Schweizer, H.J.  Schuller, U. Hojad, E. Greiner, and K.  Finger.  1999.  Systematic  analysis  of  yeast  strains with  possible  defects  in  lipid metabolism. Yeast.  15:601‐614.  Davies, S.M., R.M. Epand, R. Kraayenhof, and R.B. Cornell. 2001. Regulation of CTP: phosphocholine  cytidylyltransferase activity by the physical properties of lipid membranes: an important role  for stored curvature strain energy. Biochemistry. 40:10522‐10531.  De  Marothy,  M.T.,  and  A.  Elofsson.  2015.  Marginally  hydrophobic  transmembrane  alpha‐helices  shaping  membrane  protein  folding.  Protein  science  :  a  publication  of  the  Protein  Society.  24:1057‐1074.  234    de Meyer,  F.,  and  B.  Smit.  2009.  Effect  of  cholesterol  on  the  structure  of  a  phospholipid  bilayer.  Proceedings of the National Academy of Sciences of the United States of America. 106:3654‐ 3658.  Dellaire,  G.,  R.  Kepkay,  and  D.P.  Bazett‐Jones.  2009.  High  resolution  imaging  of  changes  in  the  structure and spatial organization of chromatin, gamma‐H2A.X and the MRN complex within  etoposide‐induced DNA repair foci. Cell cycle. 8:3750‐3769.  DeLong, C.J., Y.J. Shen, M.J. Thomas, and Z. Cui. 1999. Molecular distinction of phosphatidylcholine  synthesis  between  the  CDP‐choline  pathway  and  phosphatidylethanolamine  methylation  pathway. The Journal of biological chemistry. 274:29683‐29688.  Dennis, M.K.,  S.G.  Taneva,  and  R.B.  Cornell.  2011.  The  intrinsically  disordered  nuclear  localization  signal  and  phosphorylation  segments  distinguish  the  membrane  affinity  of  two  cytidylyltransferase isoforms. The Journal of biological chemistry. 286:12349‐12360.  Devaux,  P.F.,  and  R.  Morris.  2004.  Transmembrane  asymmetry  and  lateral  domains  in  biological  membranes. Traffic. 5:241‐246.  Devenish,  R.J.,  and M.  Prescott.  2015.  Autophagy:  starvation  relieves  transcriptional  repression  of  ATG genes. Current biology : CB. 25:R238‐240.  Dickey, A., and R. Faller. 2008. Examining the contributions of lipid shape and headgroup charge on  bilayer behavior. Biophysical journal. 95:2636‐2646.  Ding,  J.D.,  R.Y.  Salinas,  and  V.Y.  Arshavsky.  2015.  Discs  of  mammalian  rod  photoreceptors  form  through the membrane evagination mechanism. The Journal of cell biology. 211:495‐502.  Ding, Z., S.G. Taneva, H.K. Huang, S.A. Campbell, L. Semenec, N. Chen, and R.B. Cornell. 2012. A 22‐ mer segment in the structurally pliable regulatory domain of metazoan CTP: phosphocholine  cytidylyltransferase  facilitates  both  silencing  and  activating  functions.  The  Journal  of  biological chemistry. 287:38980‐38991.  Dingwall,  C.,  J.  Robbins,  S.M. Dilworth,  B.  Roberts,  and W.D.  Richardson.  1988.  The nucleoplasmin  nuclear location sequence is larger and more complex than that of SV‐40 large T antigen. The  Journal of cell biology. 107:841‐849.  Dobrzynska,  A.,  S.  Gonzalo,  C.  Shanahan,  and  P.  Askjaer.  2016.  The  nuclear  lamina  in  health  and  disease. Nucleus. 7:233‐248.  Dove,  J.E.,  J.S.  Brockenbrough,  and  J.P.  Aris.  1998.  Isolation  of  nuclei  and  nucleoli  from  the  yeast  Saccharomyces cerevisiae. Methods in cell biology. 53:33‐46.  Drin,  G.,  and  B.  Antonny.  2010.  Amphipathic  helices  and  membrane  curvature.  FEBS  letters.  584:1840‐1847.  Drin,  G.,  J.F.  Casella,  R.  Gautier,  T.  Boehmer,  T.U.  Schwartz,  and  B.  Antonny.  2007.  A  general  amphipathic  alpha‐helical  motif  for  sensing  membrane  curvature.  Nature  structural  &  molecular biology. 14:138‐146.  Drin, G., V. Morello, J.F. Casella, P. Gounon, and B. Antonny. 2008. Asymmetric tethering of flat and  curved lipid membranes by a golgin. Science. 320:670‐673.  235    Drobnies,  A.E.,  B.  van  Der  Ende,  J.L.  Thewalt,  and  R.B.  Cornell.  1999.  CTP:phosphocholine  cytidylyltransferase activation by oxidized phosphatidylcholines correlates with a decrease in  lipid order: a 2H NMR analysis. Biochemistry. 38:15606‐15614.  Drozdz,  M.M.,  and  D.J.  Vaux.  2017.  Shared  mechanisms  in  physiological  and  pathological  nucleoplasmic reticulum formation. Nucleus. 8:34‐45.  Dunne, S.J., R.B. Cornell, J.E. Johnson, N.R. Glover, and A.S. Tracey. 1996. Structure of the membrane  binding domain of CTP:phosphocholine cytidylyltransferase. Biochemistry. 35:11975‐11984.  Dymond,  M.K.  2015.  Mammalian  phospholipid  homeostasis:  homeoviscous  adaptation  deconstructed by  lipidomic data driven modelling. Chemistry and physics of  lipids. 191:136‐ 146.  Dymond,  M.K.  2016.  Mammalian  phospholipid  homeostasis:  evidence  that  membrane  curvature  elastic stress drives homeoviscous adaptation in vivo. Journal of the Royal Society, Interface.  13.  Dymond, M.K.,  R.J. Gillams, D.J.  Parker,  J.  Burrell, A.  Labrador,  T. Nylander,  and G.S. Attard.  2016.  Lipid  Spontaneous  Curvatures  Estimated  from Temperature‐Dependent  Changes  in  Inverse  Hexagonal Phase Lattice Parameters: Effects of Metal Cations. Langmuir : the ACS journal of  surfaces and colloids. 32:10083‐10092.  Dymond, M.K., C.V. Hague, A.D. Postle, and G.S. Attard. 2013. An in vivo ratio control mechanism for  phospholipid  homeostasis:  evidence  from  lipidomic  studies.  Journal  of  the  Royal  Society,  Interface. 10:20120854.  Ejsing, C.S.,  J.L. Sampaio, V. Surendranath, E. Duchoslav, K. Ekroos, R.W. Klemm, K. Simons, and A.  Shevchenko.  2009.  Global  analysis  of  the  yeast  lipidome  by  quantitative  shotgun  mass  spectrometry.  Proceedings  of  the  National  Academy  of  Sciences  of  the  United  States  of  America. 106:2136‐2141.  El Menyawi, I., M. Wogerbauer, H. Sigmund, H. Burgmann, and W. Graninger. 2000. Identification of  yeast  species  by  fatty  acid  profiling  as measured by  gas‐liquid  chromatography.  Journal  of  chromatography. B, Biomedical sciences and applications. 742:13‐24.  Esko,  J.D.,  M.  Nishijima,  and  C.R.  Raetz.  1982.  Animal  cells  dependent  on  exogenous  phosphatidylcholine  for  membrane  biogenesis.  Proceedings  of  the  National  Academy  of  Sciences of the United States of America. 79:1698‐1702.  Exton,  J.H. 1994. Phosphatidylcholine breakdown and signal  transduction. Biochimica et biophysica  acta. 1212:26‐42.  Fadeel, B., and D. Xue. 2009. The ins and outs of phospholipid asymmetry in the plasma membrane:  roles  in health and disease. Critical  reviews  in biochemistry and molecular biology.  44:264‐ 277.  Fagone, P., and S. Jackowski. 2013. Phosphatidylcholine and the CDP‐choline cycle. Biochim. Biophys.  Acta. 1831:523‐532.  Fagone, P., R. Sriburi, C. Ward‐Chapman, M. Frank, J. Wang, C. Gunter, J.W. Brewer, and S. Jackowski.  2007.  Phospholipid  biosynthesis  program  underlying  membrane  expansion  during  B‐ lymphocyte differentiation. The Journal of biological chemistry. 282:7591‐7605.  236    Fahrenkrog,  B.,  and  U.  Aebi.  2003.  The  nuclear  pore  complex:  nucleocytoplasmic  transport  and  beyond. Nature reviews. Molecular cell biology. 4:757‐766.  Fahy,  E.,  D.  Cotter,  M.  Sud,  and  S.  Subramaniam.  2011.  Lipid  classification,  structures  and  tools.  Biochimica et biophysica acta. 1811:637‐647.  Fairn,  G.D.,  N.L.  Schieber,  N.  Ariotti,  S.  Murphy,  L.  Kuerschner,  R.I.  Webb,  S.  Grinstein,  and  R.G.  Parton.  2011.  High‐resolution  mapping  reveals  topologically  distinct  cellular  pools  of  phosphatidylserine. The Journal of cell biology. 194:257‐275.  Falkenburger,  B.H.,  J.B.  Jensen,  E.J.  Dickson,  B.C.  Suh,  and  B.  Hille.  2010.  Phosphoinositides:  lipid  regulators of membrane proteins. The Journal of physiology. 588:3179‐3185.  Farine,  L.,  M.  Niemann,  A.  Schneider,  and  P.  Butikofer.  2015.  Phosphatidylethanolamine  and  phosphatidylcholine  biosynthesis  by  the  Kennedy  pathway  occurs  at  different  sites  in  Trypanosoma brucei. Scientific reports. 5:16787.  Farsad,  K.,  N.  Ringstad,  K.  Takei,  S.R.  Floyd,  K.  Rose,  and  P.  De  Camilli.  2001.  Generation  of  high  curvature membranes mediated by direct endophilin bilayer interactions. The Journal of cell  biology. 155:193‐200.  Fei, W., G. Shui, B. Gaeta, X. Du, L. Kuerschner, P. Li, A.J. Brown, M.R. Wenk, R.G. Parton, and H. Yang.  2008. Fld1p, a functional homologue of human seipin, regulates the size of  lipid droplets in  yeast. The Journal of cell biology. 180:473‐482.  Fischer,  M.A.,  K.  Temmerman,  E.  Ercan,  W.  Nickel,  and  M.  Seedorf.  2009.  Binding  of  plasma  membrane lipids recruits the yeast integral membrane protein Ist2 to the cortical ER. Traffic.  10:1084‐1097.  Fleischmann,  J.,  and M.A. Rocha.  2018. Nutrient depletion and  TOR  inhibition  induce 18S  and 25S  ribosomal RNAs resistant to a 5'‐phosphate‐dependent exonuclease in Candida albicans and  other yeasts. BMC molecular biology. 19:1.  Folch, J., M. Lees, and G.H. Sloane Stanley. 1957. A simple method for the isolation and purification  of total lipides from animal tissues. The Journal of biological chemistry. 226:497‐509.  Foster,  D.A.  2013.  Phosphatidic  acid  and  lipid‐sensing  by  mTOR.  Trends  in  endocrinology  and  metabolism: TEM. 24:272‐278.  Friesen, J.A., H.A. Campbell, and C. Kent. 1999. Enzymatic and cellular characterization of a catalytic  fragment  of  CTP:phosphocholine  cytidylyltransferase  alpha.  The  Journal  of  biological  chemistry. 274:13384‐13389.  Gancedo,  C.,  and  C.L.  Flores.  2008.  Moonlighting  proteins  in  yeasts. Microbiology  and  molecular  biology reviews : MMBR. 72:197‐210, table of contents.  Gautier, R., D. Douguet, B. Antonny, and G. Drin. 2008. HELIQUEST: a web server to screen sequences  with specific alpha‐helical properties. Bioinformatics. 24:2101‐2102.  Gehrig, K., R.B. Cornell, and N.D. Ridgway. 2008. Expansion of the nucleoplasmic reticulum requires  the  coordinated  activity  of  lamins  and  CTP:phosphocholine  cytidylyltransferase  alpha.  Molecular biology of the cell. 19:237‐247.  237    Geilen,  C.C.,  T.  Wieder,  and W.  Reutter.  1992.  Hexadecylphosphocholine  inhibits  translocation  of  CTP:choline‐phosphate  cytidylyltransferase  in Madin‐Darby canine kidney cells. The  Journal  of biological chemistry. 267:6719‐6724.  Gibellini,  F.,  and  T.K.  Smith.  2010.  The  Kennedy  pathway‐‐De  novo  synthesis  of  phosphatidylethanolamine and phosphatidylcholine. IUBMB life. 62:414‐428.  Golfman,  L.S.,  M.  Bakovic,  and  D.E.  Vance.  2001.  Transcription  of  the  CTP:phosphocholine  cytidylyltransferase alpha gene is enhanced during the S phase of the cell cycle. The Journal  of biological chemistry. 276:43688‐43692.  Gorlich, D., F. Vogel, A.D. Mills, E. Hartmann, and R.A.  Laskey. 1995. Distinct  functions  for  the  two  importin subunits in nuclear protein import. Nature. 377:246‐248.  Gorner, W.,  E. Durchschlag,  J. Wolf,  E.L. Brown, G. Ammerer, H. Ruis,  and C.  Schuller. 2002. Acute  glucose  starvation  activates  the  nuclear  localization  signal  of  a  stress‐specific  yeast  transcription factor. The EMBO journal. 21:135‐144.  Groblewski, G.E., Y. Wang, S.A. Ernst, C. Kent, and J.A. Williams. 1995. Cholecystokinin stimulates the  down‐regulation  of  CTP:phosphocholine  cytidylyltransferase  in  pancreatic  acinar  cells.  The  Journal of biological chemistry. 270:1437‐1442.  Gruner,  S.M.  1985.  Intrinsic  curvature  hypothesis  for  biomembrane  lipid  composition:  a  role  for  nonbilayer  lipids. Proceedings of  the National Academy of  Sciences of  the United  States  of  America. 82:3665‐3669.  Guenantin, A.C., N. Briand, E. Capel, F. Dumont, R. Morichon, C. Provost, F. Stillitano, D. Jeziorowska,  J.P.  Siffroi,  R.J.  Hajjar,  B.  Feve,  J.S.  Hulot,  P.  Collas,  J.  Capeau,  and  C.  Vigouroux.  2017.  Functional Human Beige Adipocytes From Induced Pluripotent Stem Cells. Diabetes. 66:1470‐ 1478.  Guenole, A., and G. Legube. 2017. A meeting at risk: Unrepaired DSBs go for broke. Nucleus. 8:589‐ 599.  Guo, Y., T.C. Walther, M. Rao, N. Stuurman, G. Goshima, K. Terayama, J.S. Wong, R.D. Vale, P. Walter,  and  R.V.  Farese.  2008.  Functional  genomic  screen  reveals  genes  involved  in  lipid‐droplet  formation and utilization. Nature. 453:657‐661.  Gupta, A., C.R. Hunt,  S. Chakraborty, R.K. Pandita,  J.  Yordy, D.B. Ramnarain, N. Horikoshi,  and T.K.  Pandita. 2014. Role of 53BP1  in the regulation of DNA double‐strand break repair pathway  choice. Radiation research. 181:1‐8.  Gupta, T., and T. Schupbach. 2003. Cct1, a phosphatidylcholine biosynthesis enzyme, is required for  Drosophila oogenesis and ovarian morphogenesis. Development. 130:6075‐6087.  Hafner, A.L., and C. Dani. 2014. Human induced pluripotent stem cells: A new source for brown and  white adipocytes. World journal of stem cells. 6:467‐472.  Hague,  C.V.,  A.D.  Postle,  G.S.  Attard,  and  M.K.  Dymond.  2013.  Cell  cycle  dependent  changes  in  membrane  stored  curvature  elastic  energy:  evidence  from  lipidomic  studies.  Faraday  discussions. 161:481‐497; discussion 563‐489.  Haider, A., Y.C. Wei, K. Lim, A.D. Barbosa, C.H. Liu, U. Weber, M. Mlodzik, K. Oras, S. Collier, M.M.  Hussain,  L.  Dong,  S.  Patel,  A.  Alvarez‐Guaita,  V.  Saudek,  B.J.  Jenkins,  A.  Koulman,  M.K.  238    Dymond,  R.C.  Hardie,  S.  Siniossoglou,  and  D.B.  Savage.  2018.  PCYT1A  Regulates  Phosphatidylcholine  Homeostasis  from  the  Inner  Nuclear  Membrane  in  Response  to  Membrane Stored Curvature Elastic Stress. Developmental cell. 45:481‐495 e488.  Hakomori, S. 1993. Structure and function of sphingoglycolipids in transmembrane signalling and cell‐ cell interactions. Biochemical Society transactions. 21 ( Pt 3):583‐595.  Hakomori,  S.,  and  Y.  Igarashi.  1995.  Functional  role  of  glycosphingolipids  in  cell  recognition  and  signaling. Journal of biochemistry. 118:1091‐1103.  Hannun,  Y.A.,  and  L.M.  Obeid.  2008.  Principles  of  bioactive  lipid  signalling:  lessons  from  sphingolipids. Nature reviews. Molecular cell biology. 9:139‐150.  Harreman,  M.T.,  M.R.  Hodel,  P.  Fanara,  A.E.  Hodel,  and  A.H.  Corbett.  2003.  The  auto‐inhibitory  function of importin alpha is essential in vivo. The Journal of biological chemistry. 278:5854‐ 5863.  Haruki,  H.,  J.  Nishikawa,  and  U.K.  Laemmli.  2008.  The  anchor‐away  technique:  rapid,  conditional  establishment of yeast mutant phenotypes. Molecular cell. 31:925‐932.  Hawryluk, M.J., P.A. Keyel, S.K. Mishra, S.C. Watkins, J.E. Heuser, and L.M. Traub. 2006. Epsin 1 is a  polyubiquitin‐selective clathrin‐associated sorting protein. Traffic. 7:262‐281.  Hedbacker,  K.,  and  M.  Carlson.  2008.  SNF1/AMPK  pathways  in  yeast.  Frontiers  in  bioscience  :  a  journal and virtual library. 13:2408‐2420.  Helmink, B.A.,  J.D. Braker, C. Kent, and  J.A.  Friesen. 2003.  Identification of  lysine 122 and arginine  196  as  important  functional  residues  of  rat  CTP:phosphocholine  cytidylyltransferase  alpha.  Biochemistry. 42:5043‐5051.  Henneberry, A.L., M.M. Wright,  and C.R. McMaster.  2002.  The major  sites of  cellular phospholipid  synthesis  and  molecular  determinants  of  Fatty  Acid  and  lipid  head  group  specificity.  Molecular biology of the cell. 13:3148‐3161.  Heo, W.D., T.  Inoue, W.S. Park, M.L. Kim, B.O. Park, T.J. Wandless, and T. Meyer. 2006. PI(3,4,5)P3  and PI(4,5)P2 lipids target proteins with polybasic clusters to the plasma membrane. Science.  314:1458‐1461.  Hjelmstad,  R.H.,  and  R.M.  Bell.  1990.  The  sn‐1,2‐diacylglycerol  cholinephosphotransferase  of  Saccharomyces cerevisiae. Nucleotide sequence, transcriptional mapping, and gene product  analysis of the CPT1 gene. The Journal of biological chemistry. 265:1755‐1764.  Hjelmstad,  R.H.,  and  R.M.  Bell.  1991.  sn‐1,2‐diacylglycerol  choline‐  and  ethanolaminephosphotransferases in Saccharomyces cerevisiae. Nucleotide sequence of the  EPT1 gene and  comparison of  the CPT1 and  EPT1  gene products. The  Journal  of  biological  chemistry. 266:5094‐5103.  Hodel,  M.R.,  A.H.  Corbett,  and  A.E.  Hodel.  2001.  Dissection  of  a  nuclear  localization  signal.  The  Journal of biological chemistry. 276:1317‐1325.  Hoover‐Fong, J., N. Sobreira, J. Jurgens, P. Modaff, C. Blout, A. Moser, O.H. Kim, T.J. Cho, S.Y. Cho, S.J.  Kim, D.K. Jin, H. Kitoh, W.Y. Park, H. Ling, K.N. Hetrick, K.F. Doheny, D. Valle, and R.M. Pauli.  2014. Mutations  in  PCYT1A,  encoding  a  key  regulator  of  phosphatidylcholine metabolism,  239    cause spondylometaphyseal dysplasia with cone‐rod dystrophy. American journal of human  genetics. 94:105‐112.  Horl, G., A. Wagner,  L.K. Cole, R. Malli, H. Reicher, P. Kotzbeck, H. Kofeler, G. Hofler, S. Frank,  J.G.  Bogner‐Strauss,  W.  Sattler,  D.E.  Vance,  and  E.  Steyrer.  2011.  Sequential  synthesis  and  methylation of phosphatidylethanolamine promote lipid droplet biosynthesis and stability in  tissue culture and in vivo. The Journal of biological chemistry. 286:17338‐17350.  Houweling, M., Z. Cui, L. Tessitore, and D.E. Vance. 1997. Induction of hepatocyte proliferation after  partial  hepatectomy  is  accompanied  by  a  markedly  reduced  expression  of  phosphatidylethanolamine N‐methyltransferase‐2. Biochimica et biophysica acta. 1346:1‐9.  Houweling,  M.,  Z.  Cui,  and  D.E.  Vance.  1995.  Expression  of  phosphatidylethanolamine  N‐ methyltransferase‐2  cannot  compensate  for  an  impaired  CDP‐choline  pathway  in  mutant  Chinese hamster ovary cells. The Journal of biological chemistry. 270:16277‐16282.  Hristova, K., W.C. Wimley, V.K. Mishra, G.M. Anantharamiah, J.P. Segrest, and S.H. White. 1999. An  amphipathic  alpha‐helix  at  a  membrane  interface:  a  structural  study  using  a  novel  X‐ray  diffraction method. Journal of molecular biology. 290:99‐117.  Huang, H.K.,  S.G. Taneva,  J.  Lee,  L.P. Silva, D.C. Schriemer, and R.B. Cornell. 2013. The membrane‐ binding  domain  of  an  amphitropic  enzyme  suppresses  catalysis  by  contact  with  an  amphipathic helix flanking its active site. Journal of molecular biology. 425:1546‐1564.  Huh, W.K.,  J.V.  Falvo,  L.C. Gerke, A.S. Carroll, R.W. Howson,  J.S. Weissman, and E.K. O'Shea. 2003.  Global analysis of protein localization in budding yeast. Nature. 425:686‐691.  Humbert, J.P., N. Matter, J.C. Artault, P. Koppler, and A.N. Malviya. 1996. Inositol 1,4,5‐trisphosphate  receptor is located to the inner nuclear membrane vindicating regulation of nuclear calcium  signaling by inositol 1,4,5‐trisphosphate. Discrete distribution of inositol phosphate receptors  to inner and outer nuclear membranes. The Journal of biological chemistry. 271:478‐485.  Ito, H., Y. Fukuda, K. Murata, and A. Kimura. 1983. Transformation of intact yeast cells treated with  alkali cations. Journal of bacteriology. 153:163‐168.  Jackowski, S., J.E. Rehg, Y.M. Zhang, J. Wang, K. Miller, P. Jackson, and M.A. Karim. 2004. Disruption  of  CCTbeta2  expression  leads  to  gonadal  dysfunction.  Molecular  and  cellular  biology.  24:4720‐4733.  Jacobs,  R.L.,  C.  Devlin,  I.  Tabas,  and  D.E.  Vance.  2004.  Targeted  deletion  of  hepatic  CTP:phosphocholine  cytidylyltransferase  alpha  in  mice  decreases  plasma  high  density  and  very low density lipoproteins. The Journal of biological chemistry. 279:47402‐47410.  Jacobs,  R.L.,  S.  Lingrell,  Y.  Zhao,  G.A.  Francis,  and  D.E.  Vance.  2008.  Hepatic  CTP:phosphocholine  cytidylyltransferase‐alpha  is a  critical predictor of plasma high density  lipoprotein and very  low density lipoprotein. The Journal of biological chemistry. 283:2147‐2155.  Jacobs,  R.L.,  Y.  Zhao, D.P.  Koonen,  T.  Sletten,  B.  Su,  S.  Lingrell,  G.  Cao, D.A.  Peake, M.S.  Kuo,  S.D.  Proctor, B.P. Kennedy,  J.R. Dyck, and D.E. Vance. 2010.  Impaired de novo choline synthesis  explains  why  phosphatidylethanolamine  N‐methyltransferase‐deficient  mice  are  protected  from diet‐induced obesity. The Journal of biological chemistry.. 285:22403‐22413.  240    Jamil,  H.,  Z.M.  Yao,  and  D.E.  Vance.  1990.  Feedback  regulation  of  CTP:phosphocholine  cytidylyltransferase  translocation  between  cytosol  and  endoplasmic  reticulum  by  phosphatidylcholine. The Journal of biological chemistry. 265:4332‐4339.  Janke,  C.,  M.M.  Magiera,  N.  Rathfelder,  C.  Taxis,  S.  Reber,  H.  Maekawa,  A.  Moreno‐Borchart,  G.  Doenges,  E.  Schwob,  E.  Schiebel,  and  M.  Knop.  2004.  A  versatile  toolbox  for  PCR‐based  tagging of yeast genes: new fluorescent proteins, more markers and promoter substitution  cassettes. Yeast. 21:947‐962.  Jao,  C.C.,  B.G.  Hegde,  J.  Chen,  I.S.  Haworth,  and  R.  Langen.  2008.  Structure  of  membrane‐bound  alpha‐synuclein from site‐directed spin labeling and computational refinement. Proceedings  of the National Academy of Sciences of the United States of America. 105:19666‐19671.  Jao, C.C., B.G. Hegde, J.L. Gallop, P.B. Hegde, H.T. McMahon, I.S. Haworth, and R. Langen. 2010. Roles  of  amphipathic  helices  and  the  bin/amphiphysin/rvs  (BAR)  domain  of  endophilin  in  membrane curvature generation. The Journal of biological chemistry. 285:20164‐20170.  Jaspers,  H.T.,  and  J.  van  Steveninck.  1975.  Transport‐associated  phosphorylation  of  2‐deoxy‐D‐ glucose in Saccharomyces fragilis. Biochimica et biophysica acta. 406:370‐385.  Jeon, O.H., L.M. Panicker, Q. Lu, J.J. Chae, R.A. Feldman, and J.H. Elisseeff. 2016. Human iPSC‐derived  osteoblasts  and  osteoclasts  together  promote  bone  regeneration  in  3D  biomaterials.  Scientific reports. 6:26761.  Johnson,  J.E., and R.B. Cornell. 1994. Membrane‐binding amphipathic alpha‐helical peptide derived  from CTP:phosphocholine cytidylyltransferase. Biochemistry. 33:4327‐4335.  Johnson, J.E., N.M. Rao, S.W. Hui, and R.B. Cornell. 1998. Conformation and lipid binding properties  of  four  peptides  derived  from  the  membrane‐binding  domain  of  CTP:phosphocholine  cytidylyltransferase. Biochemistry. 37:9509‐9519.  Johnson, J.E., M. Xie, L.M. Singh, R. Edge, and R.B. Cornell. 2003. Both acidic and basic amino acids in  an  amphitropic  enzyme,  CTP:phosphocholine  cytidylyltransferase,  dictate  its  selectivity  for  anionic membranes. The Journal of biological chemistry. 278:514‐522.  Jones,  H.E.,  J.L.  Harwood,  I.D.  Bowen,  and  G.  Griffiths.  1992.  Lipid  composition  of  subcellular  membranes from larvae and prepupae of Drosophila melanogaster. Lipids. 27:984‐987.  Joseph, J. 2006. Ran at a glance. Journal of cell science. 119:3481‐3484.  Jouhet,  J.  2013.  Importance  of  the  hexagonal  lipid  phase  in  biological  membrane  organization.  Frontiers in plant science. 4:494.  Kalderon, D., B.L. Roberts, W.D. Richardson, and A.E. Smith. 1984. A short amino acid sequence able  to specify nuclear location. Cell. 39:499‐509.  Kalmar, G.B., R.J. Kay, A. Lachance, R. Aebersold, and R.B. Cornell. 1990. Cloning and expression of rat  liver  CTP:  phosphocholine  cytidylyltransferase:  an  amphipathic  protein  that  controls  phosphatidylcholine  synthesis.  Proceedings  of  the  National  Academy  of  Sciences  of  the  United States of America. 87:6029‐6033.  Kalmar, G.B., R.J. Kay, A.C. LaChance, and R.B. Cornell. 1994. Primary structure and expression of a  human  CTP:phosphocholine  cytidylyltransferase.  Biochimica  et  biophysica  acta.  1219:328‐ 334.  241    Kamp, F., and K. Beyer. 2006. Binding of alpha‐synuclein affects the lipid packing in bilayers of small  vesicles. The Journal of biological chemistry. 281:9251‐9259.  Kanipes,  M.I.,  and  S.A.  Henry.  1997.  The  phospholipid  methyltransferases  in  yeast.  Biochimica  et  biophysica acta. 1348:134‐141.  Kaplan, M.R.,  and  R.D.  Simoni.  1985.  Intracellular  transport  of  phosphatidylcholine  to  the  plasma  membrane. The Journal of cell biology. 101:441‐445.  Karakasilioti,  I.,  I. Kamileri, G. Chatzinikolaou, T. Kosteas, E. Vergadi, A.R. Robinson,  I. Tsamardinos,  T.A. Rozgaja, S. Siakouli, C. Tsatsanis, L.J. Niedernhofer, and G.A. Garinis. 2013. DNA damage  triggers a chronic autoinflammatory response, leading to fat depletion in NER progeria. Cell  metabolism. 18:403‐415.  Karanasios, E., G.S. Han, Z. Xu, G.M. Carman, and S. Siniossoglou. 2010. A phosphorylation‐regulated  amphipathic  helix  controls  the  membrane  translocation  and  function  of  the  yeast  phosphatidate phosphatase. Proceedings of the National Academy of Sciences of the United  States of America. 107:17539‐17544.  Karim, M., P. Jackson, and S. Jackowski. 2003. Gene structure, expression and identification of a new  CTP:phosphocholine cytidylyltransferase beta isoform. Biochimica et biophysica acta. 1633:1‐ 12.  Katta, S.S., C.J. Smoyer, and S.L. Jaspersen. 2014. Destination: inner nuclear membrane. Trends in cell  biology. 24:221‐229.  Kennedy,  E.P.,  and  S.B.  Weiss.  1956.  The  function  of  cytidine  coenzymes  in  the  biosynthesis  of  phospholipides. The Journal of biological chemistry. 222:193‐214.  Kent,  C.  2005.  Regulatory  enzymes  of  phosphatidylcholine  biosynthesis:  a  personal  perspective.  Biochimica et biophysica acta. 1733:53‐66.  Khadija,  S.G.,  F.  Chen,  T. Hadden,  R.L.  Commissaris,  and A.  Kowluru.  2015. Biology  and Regulatory  Roles of Nuclear Lamins  in Cellular Function and Dysfunction. Recent patents on endocrine,  metabolic & immune drug discovery. 9:111‐120.  Khatun,  I.,  S.  Zeissig,  J.  Iqbal, M. Wang, D. Curiel, G.S.  Shelness, R.S. Blumberg, and M.M. Hussain.  2012.  Phospholipid  transfer  activity  of  microsomal  triglyceride  transfer  protein  produces  apolipoprotein B and  reduces hepatosteatosis while maintaining  low plasma  lipids  in mice.  Hepatology. 55:1356‐1368.  Kim, K., K.H. Kim, M.K. Storey, D.R. Voelker, and G.M. Carman. 1999. Isolation and characterization of  the  Saccharomyces  cerevisiae  EKI1  gene  encoding  ethanolamine  kinase.  The  Journal  of  biological chemistry. 274:14857‐14866.  Kitoh,  H.,  H.  Kaneko,  M.  Kondo,  T.  Yamamoto,  N.  Ishiguro,  and  G.  Nishimura.  2011.  Spondylometaphyseal  dysplasia  with  cone‐rod  dystrophy.  American  journal  of  medical  genetics. Part A. 155A:845‐849.  Kitos, T.E., C.M. Choi, and R.B. Cornell. 2006a. Angiotensin stimulates phosphatidylcholine synthesis  via  a  pathway  involving  diacylglycerol,  protein  kinase  C,  ERK1/2,  and  CTP:phosphocholine  cytidylyltransferase. Biochimica et biophysica acta. 1761:272‐279.  242    Kitos, T.E., A. Drobnies, M.N. Ng, Y. Wen, and R.B. Cornell. 2006b. Contribution of lipid mediators to  the  regulation  of  phosphatidylcholine  synthesis  by  angiotensin.  Biochimica  et  biophysica  acta. 1761:261‐271.  Klose, C., M.A. Surma, M.J. Gerl, F. Meyenhofer, A. Shevchenko, and K. Simons. 2012. Flexibility of a  eukaryotic lipidome‐‐insights from yeast lipidomics. PloS one. 7:e35063.  Klug, L., and G. Daum. 2014. Yeast lipid metabolism at a glance. FEMS yeast research. 14:369‐388.  Ko,  J.,  S. Cheah, and A.S.  Fischl. 1994. Regulation of phosphatidylinositol:ceramide phosphoinositol  transferase in Saccharomyces cerevisiae. Journal of bacteriology. 176:5181‐5183.  Kobe,  B.  1999.  Autoinhibition  by  an  internal  nuclear  localization  signal  revealed  by  the  crystal  structure of mammalian importin alpha. Nature structural biology. 6:388‐397.  Kolesnick, R.N., and M.R. Hemer. 1990. Physiologic 1,2‐diacylglycerol levels induce protein kinase C‐ independent  translocation  of  a  regulatory  enzyme.  The  Journal  of  biological  chemistry.  265:10900‐10904.  Korade,  Z.,  and A.K.  Kenworthy.  2008.  Lipid  rafts,  cholesterol,  and  the  brain. Neuropharmacology.  55:1265‐1273.  Kosugi, S., M. Hasebe, N. Matsumura, H. Takashima, E. Miyamoto‐Sato, M. Tomita, and H. Yanagawa.  2009.  Six  classes  of  nuclear  localization  signals  specific  to  different  binding  grooves  of  importin alpha. The Journal of biological chemistry. 284:478‐485.  Koulman,  A.,  G. Woffendin,  V.K.  Narayana,  H. Welchman,  C.  Crone,  and D.A.  Volmer.  2009.  High‐ resolution extracted ion chromatography, a new tool for metabolomics and lipidomics using  a  second‐generation  orbitrap  mass  spectrometer.  Rapid  communications  in  mass  spectrometry : RCM. 23:1411‐1418.  Krahmer, N., Y. Guo, F. Wilfling, M. Hilger, S. Lingrell, K. Heger, H.W. Newman, M. Schmidt‐Supprian,  D.E. Vance, M. Mann, R.V. Farese, Jr., and T.C. Walther. 2011. Phosphatidylcholine synthesis  for  lipid  droplet  expansion  is  mediated  by  localized  activation  of  CTP:phosphocholine  cytidylyltransferase. Cell metabolism. 14:504‐515.  Kuchler,  K.,  G.  Daum,  and  F.  Paltauf.  1986.  Subcellular  and  submitochondrial  localization  of  phospholipid‐synthesizing  enzymes  in  Saccharomyces  cerevisiae.  Journal  of  bacteriology.  165:901‐910.  Kuerschner, L., C. Moessinger, and C. Thiele. 2008. Imaging of lipid biosynthesis: how a neutral lipid  enters lipid droplets. Traffic. 9:338‐352.  Kwak, B.Y., Y.M. Zhang, M. Yun, R.J. Heath, C.O. Rock, S. Jackowski, and H.W. Park. 2002. Structure  and  mechanism  of  CTP:phosphocholine  cytidylyltransferase  (LicC)  from  Streptococcus  pneumoniae. The Journal of biological chemistry. 277:4343‐4350.  Kweon, D.H., Y.K. Shin, J.Y. Shin, J.H. Lee, J.B. Lee, J.H. Seo, and Y.S. Kim. 2006. Membrane topology  of helix 0 of the Epsin N‐terminal homology domain. Molecules and cells. 21:428‐435.  Laba, J.K., A. Steen, and L.M. Veenhoff. 2014. Traffic to the inner membrane of the nuclear envelope.  Current opinion in cell biology. 28:36‐45.  243    Lagace,  T.A.,  J.R.  Miller,  and  N.D.  Ridgway.  2002.  Caspase  processing  and  nuclear  export  of  CTP:phosphocholine cytidylyltransferase alpha during farnesol‐induced apoptosis. Molecular  and cellular biology. 22:4851‐4862.  Lagace,  T.A.,  and  N.D.  Ridgway.  2005a.  Induction  of  apoptosis  by  lipophilic  activators  of  CTP:phosphocholine cytidylyltransferase alpha (CCTalpha). The Biochemical journal. 392:449‐ 456.  Lagace,  T.A.,  and  N.D.  Ridgway.  2005b.  The  rate‐limiting  enzyme  in  phosphatidylcholine  synthesis  regulates proliferation of the nucleoplasmic reticulum. Molecular biology of the cell. 16:1120‐ 1130.  Landis,  G.N.,  D.  Bhole,  and  J.  Tower.  2003.  A  search  for  doxycycline‐dependent  mutations  that  increase Drosophila melanogaster  life  span  identifies  the VhaSFD,  Sugar baby,  filamin,  fwd  and Cctl genes. Genome biology. 4:R8.  Lange,  A.,  R.E. Mills,  C.J.  Lange, M.  Stewart,  S.E.  Devine,  and A.H.  Corbett.  2007.  Classical  nuclear  localization signals: definition,  function, and  interaction with  importin alpha. The Journal of  biological chemistry. 282:5101‐5105.  Larkin, M.A., G. Blackshields, N.P. Brown, R. Chenna, P.A. McGettigan, H. McWilliam, F. Valentin, I.M.  Wallace, A. Wilm, R. Lopez, J.D. Thompson, T.J. Gibson, and D.G. Higgins. 2007. Clustal W and  Clustal X version 2.0. Bioinformatics. 23:2947‐2948.  Lauber, K., E. Bohn, S.M. Krober, Y.J. Xiao, S.G. Blumenthal, R.K. Lindemann, P. Marini, C. Wiedig, A.  Zobywalski,  S.  Baksh,  Y.  Xu,  I.B. Autenrieth,  K.  Schulze‐Osthoff,  C.  Belka, G.  Stuhler,  and  S.  Wesselborg.  2003.  Apoptotic  cells  induce migration  of  phagocytes  via  caspase‐3‐mediated  release of a lipid attraction signal. Cell. 113:717‐730.  Leber, A., C. Hrastnik, and G. Daum. 1995. Phospholipid‐synthesizing enzymes in Golgi membranes of  the yeast, Saccharomyces cerevisiae. FEBS letters. 377:271‐274.  Leber, R., E. Zinser, G. Zellnig, F. Paltauf, and G. Daum. 1994. Characterization of lipid particles of the  yeast, Saccharomyces cerevisiae. Yeast. 10:1421‐1428.  Lee,  A.G.  2004.  How  lipids  affect  the  activities  of  integral  membrane  proteins.  Biochimica  et  biophysica acta. 1666:62‐87.  Lee, J., J. Johnson, Z. Ding, M. Paetzel, and R.B. Cornell. 2009. Crystal structure of a mammalian CTP:  phosphocholine cytidylyltransferase catalytic domain reveals novel active site residues within  a  highly  conserved  nucleotidyltransferase  fold.  The  Journal  of  biological  chemistry.  284:33535‐33548.  Lee,  J.,  S.G.  Taneva,  B.W.  Holland,  D.P.  Tieleman,  and  R.B.  Cornell.  2014.  Structural  basis  for  autoinhibition  of  CTP:phosphocholine  cytidylyltransferase  (CCT),  the  regulatory  enzyme  in  phosphatidylcholine  synthesis,  by  its  membrane‐binding  amphipathic  helix.  The  Journal  of  biological chemistry. 289:1742‐1755.  Leikin,  S.,  M.M.  Kozlov,  N.L.  Fuller,  and  R.P.  Rand.  1996.  Measured  effects  of  diacylglycerol  on  structural  and elastic  properties of  phospholipid membranes. Biophysical  journal.  71:2623‐ 2632.  244    Lemaitre,  C.,  and W.A.  Bickmore.  2015.  Chromatin  at  the  nuclear  periphery  and  the  regulation  of  genome functions. Histochemistry and cell biology. 144:111‐122.  Lempiainen,  H.,  A.  Uotila,  J.  Urban,  I.  Dohnal,  G.  Ammerer,  R.  Loewith,  and  D.  Shore.  2009.  Sfp1  interaction with  TORC1  and Mrs6  reveals  feedback  regulation on  TOR  signaling. Molecular  cell. 33:704‐716.  Letts, V.A., L.S. Klig, M. Bae‐Lee, G.M. Carman, and S.A. Henry. 1983. Isolation of the yeast structural  gene for the membrane‐associated enzyme phosphatidylserine synthase. Proceedings of the  National Academy of Sciences of the United States of America. 80:7279‐7283.  Li, C., B.K. Tan, J. Zhao, and Z. Guan. 2016. In Vivo and in Vitro Synthesis of Phosphatidylglycerol by an  Escherichia coli Cardiolipin Synthase. The Journal of biological chemistry. 291:25144‐25153.  Li, W., A. Cowley, M. Uludag, T. Gur, H. McWilliam, S. Squizzato, Y.M. Park, N. Buso, and R. Lopez.  2015.  The  EMBL‐EBI  bioinformatics web  and  programmatic  tools  framework. Nucleic  acids  research. 43:W580‐584.  Li,  Z.,  L.B.  Agellon,  T.M. Allen, M. Umeda,  L.  Jewell,  A. Mason,  and D.E.  Vance.  2006.  The  ratio  of  phosphatidylcholine  to  phosphatidylethanolamine  influences  membrane  integrity  and  steatohepatitis. Cell metabolism. 3:321‐331.  Li,  Z.,  L.B.  Agellon,  and  D.E.  Vance.  2005.  Phosphatidylcholine  homeostasis  and  liver  failure.  The  Journal of biological chemistry. 280:37798‐37802.  Li, Z., and D.E. Vance. 2008. Phosphatidylcholine and choline homeostasis. Journal of lipid research.  49:1187‐1194.  Lincoln Taiz, E.Z., Ian M. Møller, and Angus Murphy. 2014. Plant physiology and development, sixth  edition. Retrieved from http://6e.plantphys.net/appendices.html  Loewen, C.J., A. Roy, and T.P. Levine. 2003. A conserved ER targeting motif in three families of lipid  binding proteins and in Opi1p binds VAP. The EMBO journal. 22:2025‐2035.  Loewen,  C.J.,  B.P.  Young,  S.  Tavassoli,  and  T.P.  Levine.  2007.  Inheritance  of  cortical  ER  in  yeast  is  required for normal septin organization. The Journal of cell biology. 179:467‐483.  Longtine, M.S., A. McKenzie, 3rd, D.J. Demarini, N.G. Shah, A. Wach, A. Brachat, P. Philippsen, and  J.R. Pringle. 1998. Additional modules for versatile and economical PCR‐based gene deletion  and modification in Saccharomyces cerevisiae. Yeast. 14:953‐961.  Lu,  L., A. Koulman, C.J. Petry, B.  Jenkins,  L. Matthews,  I.A. Hughes, C.L. Acerini, K.K. Ong,  and D.B.  Dunger.  2016.  An  Unbiased  Lipidomics  Approach  Identifies  Early  Second  Trimester  Lipids  Predictive  of  Maternal  Glycemic  Traits  and  Gestational  Diabetes  Mellitus.  Diabetes  care.  39:2232‐2239.  Lundin, C., M. North, K. Erixon, K. Walters, D. Jenssen, A.S. Goldman, and T. Helleday. 2005. Methyl  methanesulfonate (MMS) produces heat‐labile DNA damage but no detectable  in vivo DNA  double‐strand breaks. Nucleic acids research. 33:3799‐3811.  Lutzmann, M., R. Kunze, A. Buerer, U. Aebi, and E. Hurt. 2002. Modular self‐assembly of a Y‐shaped  multiprotein complex from seven nucleoporins. The EMBO journal. 21:387‐397.  245    Lykidis,  A.,  I.  Baburina,  and  S.  Jackowski.  1999.  Distribution  of  CTP:phosphocholine  cytidylyltransferase  (CCT)  isoforms.  Identification  of  a  new  CCTbeta  splice  variant.  The  Journal of biological chemistry. 274:26992‐27001.  Lykidis,  A.,  K.G.  Murti,  and  S.  Jackowski.  1998.  Cloning  and  characterization  of  a  second  human  CTP:phosphocholine  cytidylyltransferase.  The  Journal  of  biological  chemistry.  273:14022‐ 14029.  MacDonald,  J.I.,  and  C.  Kent.  1994.  Identification  of  phosphorylation  sites  in  rat  liver  CTP:  phosphocholine cytidylyltransferase. The Journal of biological chemistry. 269:10529‐10537.  MacKinnon, M.A., A.J. Curwin, G.J. Gaspard, A.B. Suraci, J.P. Fernandez‐Murray, and C.R. McMaster.  2009.  The  Kap60‐Kap95  karyopherin  complex  directly  regulates  phosphatidylcholine  synthesis. The Journal of biological chemistry. 284:7376‐7384.  Magdeleine, M.,  R.  Gautier,  P.  Gounon,  H.  Barelli,  S.  Vanni,  and  B.  Antonny.  2016.  A  filter  at  the  entrance of the Golgi that selects vesicles according to size and bulk lipid composition. eLife.  5.  Mallampalli, R.K., E.J. Peterson, A.B. Carter, R.G. Salome, S.N. Mathur, and G.A. Koretzky. 1999. TNF‐ alpha  increases ceramide without  inducing apoptosis  in alveolar  type  II epithelial cells. The  American journal of physiology. 276:L481‐490.  Mallampalli, R.K., A.J. Ryan, R.G. Salome, and S. Jackowski. 2000. Tumor necrosis factor‐alpha inhibits  expression  of  CTP:phosphocholine  cytidylyltransferase.  The  Journal  of  biological  chemistry.  275:9699‐9708.  Manford,  A.G.,  C.J.  Stefan,  H.L.  Yuan,  J.A. Macgurn,  and  S.D.  Emr.  2012.  ER‐to‐plasma membrane  tethering proteins  regulate cell  signaling and ER morphology. Developmental  cell.  23:1129‐ 1140.  Mannik,  J.,  A. Meyers,  and P. Dalhaimer.  2014.  Isolation of  cellular  lipid  droplets:  two purification  techniques starting from yeast cells and human placentas. Journal of visualized experiments :  JoVE.  Marcucci,  H.,  C.  Elena,  P.  Gilardoni,  and  C.  Banchio.  2008.  Characterization  of  the  murine  CTP:phosphocholine cytidylyltransferase beta gene promoter. Biochimica et biophysica acta.  1781:254‐262.  Marquez, M.P., F. Alencastro, A. Madrigal, J.L. Jimenez, G. Blanco, A. Gureghian, L. Keagy, C. Lee, R.  Liu, L. Tan, K. Deignan, B. Armstrong, and Y. Zhao. 2017. The Role of Cellular Proliferation in  Adipogenic Differentiation of Human Adipose Tissue‐Derived Mesenchymal Stem Cells. Stem  cells and development. 26:1578‐1595.  Masmoudi, A., G. Labourdette, M. Mersel, F.L. Huang, K.P. Huang, G. Vincendon, and A.N. Malviya.  1989.  Protein  kinase  C  located  in  rat  liver  nuclei.  Partial  purification  and  biochemical  and  immunochemical characterization. The Journal of biological chemistry. 264:1172‐1179.  McDonald, C., G.  Jovanovic, O. Ces, and M. Buck. 2015. Membrane Stored Curvature Elastic Stress  Modulates Recruitment of Maintenance Proteins PspA and Vipp1. mBio. 6:e01188‐01115.  McLane,  L.M., and A.H. Corbett. 2009. Nuclear  localization  signals and human disease.  IUBMB  life.  61:697‐706.  246    McMahon,  H.T.,  and  E.  Boucrot.  2015.  Membrane  curvature  at  a  glance.  Journal  of  cell  science.  128:1065‐1070.  McMaster, C.R., and R.M. Bell. 1994. Phosphatidylcholine biosynthesis in Saccharomyces cerevisiae.  Regulatory  insights  from  studies  employing null  and  chimeric  sn‐1,2‐diacylglycerol  choline‐  and  ethanolaminephosphotransferases.  The  Journal  of  biological  chemistry.  269:28010‐ 28016.  McWilliam, H., W. Li, M. Uludag, S. Squizzato, Y.M. Park, N. Buso, A.P. Cowley, and R. Lopez. 2013.  Analysis Tool Web Services from the EMBL‐EBI. Nucleic acids research. 41:W597‐600.  Messa, M.,  R.  Fernandez‐Busnadiego,  E.W.  Sun, H.  Chen, H.  Czapla,  K. Wrasman,  Y. Wu, G.  Ko,  T.  Ross, B. Wendland, and P. De Camilli. 2014. Epsin deficiency impairs endocytosis by stalling  the actin‐dependent invagination of endocytic clathrin‐coated pits. eLife. 3:e03311.  Mirisola,  M.G.,  G.  Taormina,  P.  Fabrizio,  M.  Wei,  J.  Hu,  and  V.D.  Longo.  2014.  Serine‐  and  threonine/valine‐dependent  activation  of  PDK  and  Tor  orthologs  converge  on  Sch9  to  promote aging. PLoS genetics. 10:e1004113.  Mishra,  V.K.,  and  M.N.  Palgunachari.  1996.  Interaction  of  model  class  A1,  class  A2,  and  class  Y  amphipathic helical peptides with membranes. Biochemistry. 35:11210‐11220.  Moessinger,  C.,  L.  Kuerschner,  J.  Spandl,  A.  Shevchenko,  and  C.  Thiele.  2011.  Human  lysophosphatidylcholine  acyltransferases  1  and  2  are  located  in  lipid  droplets  where  they  catalyze  the  formation  of  phosphatidylcholine.  The  Journal  of  biological  chemistry.  286:21330‐21339.  Molday, R.S.,  and O.L. Moritz.  2015.  Photoreceptors  at  a  glance.  Journal  of  cell  science.  128:4039‐ 4045.  Monteiro,  N.,  A.  Martins,  R.L.  Reis,  and  N.M.  Neves.  2014.  Liposomes  in  tissue  engineering  and  regenerative medicine. Journal of the Royal Society, Interface. 11:20140459.  Morash, S.C., H.W. Cook, and M.W. Spence. 1988. Phosphatidylcholine metabolism in cultured cells:  catabolism via glycerophosphocholine. Biochimica et biophysica acta. 961:194‐202.  Nagai,  S.,  K.  Dubrana, M.  Tsai‐Pflugfelder, M.B. Davidson,  T.M.  Roberts, G.W.  Brown,  E.  Varela,  F.  Hediger,  S.M.  Gasser,  and  N.J.  Krogan.  2008.  Functional  targeting  of  DNA  damage  to  a  nuclear pore‐associated SUMO‐dependent ubiquitin ligase. Science. 322:597‐602.  Naider, F., and J.M. Becker. 2004. The alpha‐factor mating pheromone of Saccharomyces cerevisiae:  a model for studying the interaction of peptide hormones and G protein‐coupled receptors.  Peptides. 25:1441‐1463.  Nair,  U., M.  Thumm,  D.J.  Klionsky,  and  R.  Krick.  2011.  GFP‐Atg8  protease  protection  as  a  tool  to  monitor autophagosome biogenesis. Autophagy. 7:1546‐1550.  Nash,  B.M.,  D.C.  Wright,  J.R.  Grigg,  B.  Bennetts,  and  R.V.  Jamieson.  2015.  Retinal  dystrophies,  genomic applications in diagnosis and prospects for therapy. Translational pediatrics. 4:139‐ 163.  Natter, K., P. Leitner, A. Faschinger, H. Wolinski, S. McCraith, S. Fields, and S.D. Kohlwein. 2005. The  spatial  organization  of  lipid  synthesis  in  the  yeast  Saccharomyces  cerevisiae  derived  from  247    large  scale  green  fluorescent  protein  tagging  and high  resolution microscopy. Molecular &  cellular proteomics : MCP. 4:662‐672.  Nejadnik, H., S. Diecke, O.D. Lenkov, F. Chapelin, J. Donig, X. Tong, N. Derugin, R.C. Chan, A. Gaur, F.  Yang,  J.C.  Wu,  and  H.E.  Daldrup‐Link.  2015.  Improved  approach  for  chondrogenic  differentiation of human induced pluripotent stem cells. Stem cell reviews. 11:242‐253.  Ng, M.N., T.E. Kitos, and R.B. Cornell. 2004. Contribution of lipid second messengers to the regulation  of  phosphatidylcholine  synthesis  during  cell  cycle  re‐entry.  Biochimica  et  biophysica  acta.  1686:85‐99.  Niebergall, L.J., and D.E. Vance. 2012. The ratio of phosphatidylcholine to phosphatidylethanolamine  does  not  predict  integrity  of  growing  MT58  Chinese  hamster  ovary  cells.  Biochimica  et  biophysica acta. 1821:324‐334.  Nikoloff,  D.M.,  and  S.A.  Henry.  1991.  Genetic  analysis  of  yeast  phospholipid  biosynthesis.  Annual  review of genetics. 25:559‐583.  Noga,  A.A.,  and  D.E.  Vance.  2003.  A  gender‐specific  role  for  phosphatidylethanolamine  N‐ methyltransferase‐derived phosphatidylcholine in the regulation of plasma high density and  very low density lipoproteins in mice. The Journal of biological chemistry. 278:21851‐21859.  Northwood,  I.C.,  A.H.  Tong,  B.  Crawford,  A.E.  Drobnies,  and  R.B.  Cornell.  1999.  Shuttling  of  CTP:Phosphocholine  cytidylyltransferase  between  the  nucleus  and  endoplasmic  reticulum  accompanies  the wave of phosphatidylcholine synthesis during  the G(0)  ‐‐> G(1)  transition.  The Journal of biological chemistry. 274:26240‐26248.  O'Donnell, A.F., R.R. McCartney, D.G. Chandrashekarappa, B.B. Zhang, J. Thorner, and M.C. Schmidt.  2015.  2‐Deoxyglucose  impairs  Saccharomyces  cerevisiae  growth  by  stimulating  Snf1‐ regulated and alpha‐arrestin‐mediated trafficking of hexose transporters 1 and 3. Molecular  and cellular biology. 35:939‐955.  Onischenko, E.,  J.H. Tang, K.R. Andersen, K.E. Knockenhauer, P. Vallotton, C.P. Derrer, A. Kralt, C.F.  Mugler, L.Y. Chan, T.U. Schwartz, and K. Weis. 2017. Natively Unfolded FG Repeats Stabilize  the Structure of the Nuclear Pore Complex. Cell. 171:904‐917 e919.  Paine,  P.L.,  L.C. Moore,  and  S.B. Horowitz.  1975. Nuclear  envelope  permeability. Nature.  254:109‐ 114.  Paradies, G., V. Paradies, V. De Benedictis, F.M. Ruggiero, and G. Petrosillo. 2014. Functional role of  cardiolipin in mitochondrial bioenergetics. Biochimica et biophysica acta. 1837:408‐417.  Park,  Y.S.,  P.  Gee,  S.  Sanker,  E.J.  Schurter,  E.R.  Zuiderweg,  and  C.  Kent.  1997.  Identification  of  functional  conserved  residues  of  CTP:glycerol‐3‐phosphate  cytidylyltransferase.  Role  of  histidines in the conserved HXGH in catalysis. The Journal of biological chemistry. 272:15161‐ 15166.  Payne, F., K. Lim, A. Girousse, R.J. Brown, N. Kory, A. Robbins, Y. Xue, A. Sleigh, E. Cochran, C. Adams,  A. Dev Borman, D. Russel‐Jones, P. Gorden, R.K. Semple, V. Saudek, S. O'Rahilly, T.C. Walther,  I.  Barroso,  and  D.B.  Savage.  2014.  Mutations  disrupting  the  Kennedy  phosphatidylcholine  pathway in humans with congenital lipodystrophy and fatty liver disease. Proceedings of the  National Academy of Sciences of the United States of America. 111:8901‐8906.  248    Pelech,  S.L.,  H.B.  Paddon,  and  D.E.  Vance.  1984.  Phorbol  esters  stimulate  phosphatidylcholine  biosynthesis  by  translocation  of  CTP:phosphocholine  cytidylyltransferase  from  cytosol  to  microsomes. Biochimica et biophysica acta. 795:447‐451.  Pelech, S.L., P.H. Pritchard, D.N. Brindley, and D.E. Vance. 1983. Fatty acids promote translocation of  CTP:phosphocholine  cytidylyltransferase  to  the  endoplasmic  reticulum  and  stimulate  rat  hepatic phosphatidylcholine synthesis. The Journal of biological chemistry. 258:6782‐6788.  Pereira,  L.,  J.P.  Girardi,  and  M.  Bakovic.  2012.  Forms,  crosstalks,  and  the  role  of  phospholipid  biosynthesis in autophagy. International journal of cell biology. 2012:931956.  Pike, L.J. 2009. The challenge of lipid rafts. Journal of lipid research. 50 Suppl:S323‐328.  Preitschopf, W.,  H.  Luckl,  E.  Summers,  S.A.  Henry,  F.  Paltauf,  and  S.D.  Kohlwein.  1993. Molecular  cloning  of  the  yeast  OPI3  gene  as  a  high  copy  number  suppressor  of  the  cho2 mutation.  Current genetics. 23:95‐101.  Quimby, B.B., and M. Dasso. 2003. The small GTPase Ran:  interpreting the signs. Current opinion in  cell biology. 15:338‐344.  Ramezanpour, M., J. Lee, S.G. Taneva, D.P. Tieleman, and R.B. Cornell. 2018. An auto‐inhibitory helix  in  CTP:phosphocholine  cytidylyltransferase  hi‐jacks  the  catalytic  residue  and  constrains  a  pliable, domain‐bridging helix pair. The Journal of biological chemistry.  Ramirez‐Zacarias,  J.L.,  F.  Castro‐Munozledo,  and  W.  Kuri‐Harcuch.  1992.  Quantitation  of  adipose  conversion  and  triglycerides  by  staining  intracytoplasmic  lipids  with  Oil  red  O.  Histochemistry. 97:493‐497.  Resnick,  M.A.,  and  P.  Martin.  1976.  The  repair  of  double‐strand  breaks  in  the  nuclear  DNA  of  Saccharomyces  cerevisiae  and  its  genetic  control.  Molecular  &  general  genetics  :  MGG.  143:119‐129.  Resnick, M.A., and R.K. Mortimer. 1966. Unsaturated fatty acid mutants of Saccharomyces cerevisiae.  Journal of bacteriology. 92:597‐600.  Robbins, A.L., and D.B. Savage. 2015. The genetics of lipid storage and human lipodystrophies. Trends  in molecular medicine. 21:433‐438.  Robbins, J., S.M. Dilworth, R.A. Laskey, and C. Dingwall. 1991. Two interdependent basic domains in  nucleoplasmin  nuclear  targeting  sequence:  identification  of  a  class  of  bipartite  nuclear  targeting sequence. Cell. 64:615‐623.  Robinson,  J.S.,  D.J.  Klionsky,  L.M.  Banta,  and  S.D.  Emr.  1988.  Protein  sorting  in  Saccharomyces  cerevisiae: isolation of mutants defective in the delivery and processing of multiple vacuolar  hydrolases. Molecular and cellular biology. 8:4936‐4948.  Roboti,  P.,  K.  Sato,  and M.  Lowe.  2015.  The  golgin  GMAP‐210  is  required  for  efficient membrane  trafficking in the early secretory pathway. Journal of cell science. 128:1595‐1606.  Rockenfeller,  P., M.  Koska,  F.  Pietrocola, N. Minois, O.  Knittelfelder,  V.  Sica,  J.  Franz, D.  Carmona‐ Gutierrez, G. Kroemer, and F. Madeo. 2015. Phosphatidylethanolamine positively  regulates  autophagy and longevity. Cell death and differentiation. 22:499‐508.  249    Roggero,  R.,  R.  Zufferey,  M.  Minca,  E.  Richier,  M.  Calas,  H.  Vial,  and  C.  Ben  Mamoun.  2004.  Unraveling  the  mode  of  action  of  the  antimalarial  choline  analog  G25  in  Plasmodium  falciparum and Saccharomyces cerevisiae. Antimicrob. Agents Chemother. 48:2816‐2824.  Rolland, F.,  J. Winderickx, and  J.M. Thevelein. 2002. Glucose‐sensing and ‐signalling mechanisms  in  yeast. FEMS yeast research. 2:183‐201.  Ronen, M., and D. Botstein. 2006. Transcriptional response of steady‐state yeast cultures to transient  perturbations  in  carbon  source.  Proceedings  of  the  National  Academy  of  Sciences  of  the  United States of America. 103:389‐394.  Rose, K., S.A. Rudge, M.A. Frohman, A.J. Morris, and J. Engebrecht. 1995. Phospholipase D signaling is  essential for meiosis. Proceedings of the National Academy of Sciences of the United States of  America. 92:12151‐12155.  Rosen,  E.D.,  and  O.A.  MacDougald.  2006.  Adipocyte  differentiation  from  the  inside  out.  Nature  reviews. Molecular cell biology. 7:885‐896.  Ruaud, A.F., L. Nilsson, F. Richard, M.K. Larsen, J.L. Bessereau, and S. Tuck. 2009. The C. elegans P4‐ ATPase TAT‐1 regulates lysosome biogenesis and endocytosis. Traffic. 10:88‐100.  Saklatvala,  J. 1995.  Intracellular signalling mechanisms of  interleukin 1 and  tumour necrosis  factor:  possible targets for therapy. British medical bulletin. 51:402‐418.  Sando,  J.J.,  and  O.I.  Chertihin.  1996.  Activation  of  protein  kinase  C  by  lysophosphatidic  acid:  dependence  on  composition  of  phospholipid  vesicles.  The  Biochemical  journal.  317  (  Pt  2):583‐588.  Santangelo, G.M. 2006. Glucose signaling  in Saccharomyces cerevisiae. Microbiology and molecular  biology reviews : MMBR. 70:253‐282.  Sawai,  H.,  Y.  Okamoto,  C.  Luberto,  C.  Mao,  A.  Bielawska,  N.  Domae,  and  Y.A.  Hannun.  2000.  Identification  of  ISC1  (YER019w)  as  inositol  phosphosphingolipid  phospholipase  C  in  Saccharomyces cerevisiae. The Journal of biological chemistry. 275:39793‐39798.  Schulz,  T.A.,  and  C.E.  Creutz.  2004.  The  tricalbin  C2  domains:  lipid‐binding  properties  of  a  novel,  synaptotagmin‐like yeast protein family. Biochemistry. 43:3987‐3995.  Seeber,  A.,  and  S.M.  Gasser.  2017.  Chromatin  organization  and  dynamics  in  double‐strand  break  repair. Current opinion in genetics & development. 43:9‐16.  Shirai,  Y.,  and  N.  Saito.  2002.  Activation  mechanisms  of  protein  kinase  C:  maturation,  catalytic  activation, and targeting. Journal of biochemistry. 132:663‐668.  Simons,  K.,  and  R.  Ehehalt.  2002.  Cholesterol,  lipid  rafts,  and  disease.  The  Journal  of  clinical  investigation. 110:597‐603.  Singer,  S.J.,  and  G.L.  Nicolson.  1972.  The  fluid mosaic model  of  the  structure  of  cell  membranes.  Science. 175:720‐731.  Singh, M., C.R. Hunt, R.K. Pandita, R. Kumar, C.R. Yang, N. Horikoshi, R. Bachoo, S. Serag, M.D. Story,  J.W. Shay, S.N. Powell, A. Gupta, J. Jeffery, S. Pandita, B.P. Chen, D. Deckbar, M. Lobrich, Q.  Yang, K.K. Khanna, H.J. Worman, and T.K. Pandita. 2013. Lamin A/C depletion enhances DNA  damage‐induced stalled replication fork arrest. Molecular and cellular biology. 33:1210‐1222.  250    Siniossoglou,  S.  2013.  Phospholipid  metabolism  and  nuclear  function:  roles  of  the  lipin  family  of  phosphatidic acid phosphatases. Biochimica et biophysica acta. 1831:575‐581.  Slack, B.E., J. Breu, and R.J. Wurtman. 1991. Production of diacylglycerol by exogenous phospholipase  C  stimulates  CTP:phosphocholine  cytidylyltransferase  activity  and  phosphatidylcholine  synthesis  in  human  neuroblastoma  cells.  The  Journal  of  biological  chemistry.  266:24503‐ 24508.  Sleigh, M.J. 1976. The mechanism of DNA breakage by phleomycin  in vitro. Nucleic acids  research.  3:891‐901.  Sleight,  R.,  and  C.  Kent.  1980.  Regulation  of  phosphatidylcholine  biosynthesis  in  cultured  chick  embryonic  muscle  treated  with  phospholipase  C.  The  Journal  of  biological  chemistry.  255:10644‐10650.  Sleight, R., and C. Kent. 1983. Regulation of phosphatidylcholine biosynthesis in mammalian cells. II.  Effects  of  phospholipase  C  treatment  on  the  activity  and  subcellular  distribution  of  CTP:phosphocholine  cytidylyltransferase  in  Chinese  hamster  ovary  and  LM  cell  lines.  The  Journal of biological chemistry. 258:831‐835.  Small, D.M. 1984. Lateral chain packing in lipids and membranes. Journal of lipid research. 25:1490‐ 1500.  Smart,  D.J.,  H.D.  Halicka,  G.  Schmuck,  F.  Traganos,  Z.  Darzynkiewicz,  and  G.M.  Williams.  2008.  Assessment of DNA double‐strand breaks and gammaH2AX induced by the topoisomerase II  poisons etoposide and mitoxantrone. Mutation research. 641:43‐47.  Smit,  J.J., A.H.  Schinkel,  R.P. Oude Elferink, A.K. Groen, E. Wagenaar,  L.  van Deemter, C.A. Mol,  R.  Ottenhoff, N.M. van der Lugt, M.A. van Roon, and et al. 1993. Homozygous disruption of the  murine mdr2 P‐glycoprotein gene leads to a complete absence of phospholipid from bile and  to liver disease. Cell. 75:451‐462.  Smith, A.E., B.M. Slepchenko, J.C. Schaff, L.M. Loew, and I.G. Macara. 2002. Systems analysis of Ran  transport. Science. 295:488‐491.  Smith,  F.C.,  S.P.  Davies, W.A. Wilson, D.  Carling,  and D.G. Hardie.  1999.  The  SNF1  kinase  complex  from Saccharomyces cerevisiae phosphorylates  the  transcriptional  repressor protein Mig1p  in vitro at four sites within or near regulatory domain 1. FEBS letters. 453:219‐223.  Sohal, P.S., and R.B. Cornell. 1990. Sphingosine  inhibits  the activity of rat  liver CTP:phosphocholine  cytidylyltransferase. The Journal of biological chemistry. 265:11746‐11750.  Sohlenkamp,  C.,  and  O.  Geiger.  2016.  Bacterial  membrane  lipids:  diversity  in  structures  and  pathways. FEMS microbiology reviews. 40:133‐159.  Soniat,  M.,  and  Y.M.  Chook.  2015.  Nuclear  localization  signals  for  four  distinct  karyopherin‐beta  nuclear import systems. The Biochemical journal. 468:353‐362.  Stewart,  M.  2006.  Structural  basis  for  the  nuclear  protein  import  cycle.  Biochemical  Society  transactions. 34:701‐704.  Stewart,  M.,  R.P.  Baker,  R.  Bayliss,  L.  Clayton,  R.P.  Grant,  T.  Littlewood,  and  Y.  Matsuura.  2001.  Molecular  mechanism  of  translocation  through  nuclear  pore  complexes  during  nuclear  protein import. FEBS letters. 498:145‐149.  251    Strakova,  J.,  L.  Demizieux,  R.B.  Campenot,  D.E.  Vance,  and  J.E.  Vance.  2011.  Involvement  of  CTP:phosphocholine  cytidylyltransferase‐beta2  in  axonal  phosphatidylcholine  synthesis  and  branching of neurons. Biochimica et biophysica acta. 1811:617‐625.  Strom,  A.C.,  and  K.  Weis.  2001.  Importin‐beta‐like  nuclear  transport  receptors.  Genome  biology.  2:REVIEWS3008.  Stukey, J.E., V.M. McDonough, and C.E. Martin. 1989. Isolation and characterization of OLE1, a gene  affecting  fatty  acid  desaturation  from  Saccharomyces  cerevisiae.  The  Journal  of  biological  chemistry. 264:16537‐16544.  Sugimoto, H., K. Okamura, S. Sugimoto, M. Satou, T. Hattori, D.E. Vance, and T. Izumi. 2005. Sp1 is a  co‐activator  with  Ets‐1,  and  Net  is  an  important  repressor  of  the  transcription  of  CTP:phosphocholine  cytidylyltransferase  alpha.  The  Journal  of  biological  chemistry.  280:40857‐40866.  Surma, M.A., C. Klose, D. Peng, M. Shales, C. Mrejen, A. Stefanko, H. Braberg, D.E. Gordon, D. Vorkel,  C.S. Ejsing, R. Farese, Jr., K. Simons, N.J. Krogan, and R. Ernst. 2013. A lipid E‐MAP identifies  Ubx2 as a critical regulator of lipid saturation and lipid bilayer stress. Molecular cell. 51:519‐ 530.  Tafesse, F.G., P. Ternes, and J.C. Holthuis. 2006. The multigenic sphingomyelin synthase family. The  Journal of biological chemistry. 281:29421‐29425.  Takeda, Y., and A. Nakano. 2008. In vitro formation of a novel type of membrane vesicles containing  Dpm1p:  putative  transport  vesicles  for  lipid  droplets  in  budding  yeast.  Journal  of  biochemistry. 143:803‐811.  Taneva,  S.,  J.E.  Johnson,  and  R.B.  Cornell.  2003.  Lipid‐induced  conformational  switch  in  the  membrane binding domain of CTP:phosphocholine cytidylyltransferase: a circular dichroism  study. Biochemistry. 42:11768‐11776.  Taneva, S.G., P.J. Patty, B.J. Frisken, and R.B. Cornell. 2005. CTP:phosphocholine cytidylyltransferase  binds anionic phospholipid vesicles in a cross‐bridging mode. Biochemistry. 44:9382‐9393.  Tang,  Q.Q.,  and  M.D.  Lane.  2012.  Adipogenesis:  from  stem  cell  to  adipocyte.  Annual  review  of  biochemistry. 81:715‐736.  Tang, Q.Q., T.C. Otto, and M.D. Lane. 2003. Mitotic clonal expansion: a synchronous process required  for  adipogenesis. Proceedings of  the National Academy of  Sciences of  the United  States  of  America. 100:44‐49.  Tavassoli, S., J.T. Chao, B.P. Young, R.C. Cox, W.A. Prinz, A.I. de Kroon, and C.J. Loewen. 2013. Plasma  membrane‐‐endoplasmic  reticulum  contact  sites  regulate  phosphatidylcholine  synthesis.  EMBO reports. 14:434‐440.  Tessner, T.G., C.O. Rock, G.B. Kalmar, R.B. Cornell, and S. Jackowski. 1991. Colony‐stimulating factor 1  regulates  CTP:  phosphocholine  cytidylyltransferase  mRNA  levels.  The  Journal  of  biological  chemistry. 266:16261‐16264.  Testa, F., M. Filippelli, R. Brunetti‐Pierri, G. Di Fruscio, V. Di Iorio, M. Pizzo, A. Torella, M.R. Barillari, V.  Nigro, N. Brunetti‐Pierri, F. Simonelli, and S. Banfi. 2017. Mutations in the PCYT1A gene are  252    responsible  for  isolated  forms  of  retinal  dystrophy.  European  journal  of  human  genetics  :  EJHG. 25:651‐655.  Thiam, A.R., R.V. Farese, Jr., and T.C. Walther. 2013. The biophysics and cell biology of lipid droplets.  Nature reviews. Molecular cell biology. 14:775‐786.  Thibault, G., G. Shui, W. Kim, G.C. McAlister, N. Ismail, S.P. Gygi, M.R. Wenk, and D.T. Ng. 2012. The  membrane  stress  response  buffers  lethal  effects  of  lipid  disequilibrium  by  reprogramming  the protein homeostasis network. Molecular cell. 48:16‐27.  Thomas, B.J., and R. Rothstein. 1989. Elevated recombination rates  in  transcriptionally active DNA.  Cell. 56:619‐630.  Tian, H.F.,  J.M. Feng, and  J.F. Wen. 2012. The evolution of  cardiolipin biosynthesis and maturation  pathways  and  its  implications  for  the  evolution  of  eukaryotes.  BMC  evolutionary  biology.  12:32.  Tronchere,  H.,  V.  Planat,  M.  Record,  F.  Terce,  G.  Ribbes,  and  H.  Chap.  1995.  Phosphatidylcholine  turnover  in activated human neutrophils. Agonist‐induced cytidylyltransferase translocation  is subsequent to phospholipase D activation. The Journal of biological chemistry. 270:13138‐ 13146.  Trotter,  P.J.,  and  D.R.  Voelker.  1995.  Identification  of  a  non‐mitochondrial  phosphatidylserine  decarboxylase activity (PSD2) in the yeast Saccharomyces cerevisiae. The Journal of biological  chemistry. 270:6062‐6070.  Trumbly,  R.J.  1992.  Glucose  repression  in  the  yeast  Saccharomyces  cerevisiae.  Molecular  microbiology. 6:15‐21.  Utal,  A.K.,  H.  Jamil,  and  D.E.  Vance.  1991.  Diacylglycerol  signals  the  translocation  of  CTP:choline‐ phosphate  cytidylyltransferase  in  HeLa  cells  treated  with  12‐O‐tetradecanoylphorbol‐13‐ acetate. The Journal of biological chemistry. 266:24084‐24091.  van der Veen,  J.N.,  J.P. Kennelly, S. Wan,  J.E. Vance, D.E. Vance, and R.L.  Jacobs. 2017. The critical  role  of  phosphatidylcholine  and  phosphatidylethanolamine  metabolism  in  health  and  disease. Biochimica et biophysica acta. 1859:1558‐1572.  van Meer, G. 2011. Dynamic transbilayer lipid asymmetry. Cold Spring Harbor perspectives in biology.  3.  van Meer, G., D.R. Voelker, and G.W. Feigenson. 2008. Membrane  lipids: where  they are and how  they behave. Nature reviews. Molecular cell biology. 9:112‐124.  Vance,  D.E.  2013.  Physiological  roles  of  phosphatidylethanolamine  N‐methyltransferase.  Biochim.  Biophys. Acta. 1831:626‐632.  Vance, J.E. 2015. Phospholipid synthesis and transport in mammalian cells. Traffic. 16:1‐18.  Vanni,  S.,  L.  Vamparys,  R.  Gautier,  G.  Drin,  C.  Etchebest,  P.F.  Fuchs,  and  B.  Antonny.  2013.  Amphipathic lipid packing sensor motifs: probing bilayer defects with hydrophobic residues.  Biophysical journal. 104:575‐584.  Veitch,  D.P.,  and  R.B.  Cornell.  1996.  Substitution  of  serine  for  glycine‐91  in  the  HXGH  motif  of  CTP:phosphocholine  cytidylyltransferase  implicates  this motif  in  CTP binding. Biochemistry.  35:10743‐10750.  253    Veitch,  D.P.,  D.  Gilham,  and  R.B.  Cornell.  1998.  The  role  of  histidine  residues  in  the  HXGH  site  of  CTP:phosphocholine  cytidylyltransferase  in  CTP  binding  and  catalysis.  European  journal  of  biochemistry. 255:227‐234.  Vevea,  J.D.,  E.J.  Garcia,  R.B.  Chan,  B.  Zhou, M.  Schultz,  G. Di  Paolo,  J.M. McCaffery,  and  L.A.  Pon.  2015. Role for Lipid Droplet Biogenesis and Microlipophagy in Adaptation to Lipid Imbalance  in Yeast. Developmental cell. 35:584‐599.  Virtanen, J.A., K.H. Cheng, and P. Somerharju. 1998. Phospholipid composition of the mammalian red  cell  membrane  can  be  rationalized  by  a  superlattice  model.  Proceedings  of  the  National  Academy of Sciences of the United States of America. 95:4964‐4969.  Walker,  A.K.,  R.L.  Jacobs,  J.L. Watts,  V.  Rottiers,  K.  Jiang, D.M.  Finnegan,  T.  Shioda, M. Hansen,  F.  Yang,  L.J. Niebergall, D.E. Vance, M. Tzoneva, A.C. Hart, and A.M. Naar. 2011. A conserved  SREBP‐1/phosphatidylcholine  feedback  circuit  regulates  lipogenesis  in  metazoans.  Cell.  147:840‐852.  Walkey, C.J., L.R. Donohue, R. Bronson, L.B. Agellon, and D.E. Vance. 1997. Disruption of the murine  gene encoding phosphatidylethanolamine N‐methyltransferase. Proceedings of the National  Academy of Sciences of the United States of America. 94:12880‐12885.  Walkey, C.J., L. Yu, L.B. Agellon, and D.E. Vance. 1998. Biochemical and evolutionary significance of  phospholipid methylation. The Journal of biological chemistry. 273:27043‐27046.  Walters,  B.A., M.L.  Raff,  J.V. Hoeve,  R.  Tesser,  L.O.  Langer,  T.D.  France,  I.A.  Glass,  and R.M.  Pauli.  2004. Spondylometaphyseal dysplasia with cone‐rod dystrophy. American journal of medical  genetics. Part A. 129A:265‐276.  Walther, T.C., and R.V. Farese, Jr. 2012. Lipid droplets and cellular lipid metabolism. Annual review of  biochemistry. 81:687‐714.  Wang,  L.,  S.  Magdaleno,  I.  Tabas,  and  S.  Jackowski.  2005.  Early  embryonic  lethality  in  mice  with  targeted  deletion  of  the  CTP:phosphocholine  cytidylyltransferase  alpha  gene  (Pcyt1a).  Molecular and cellular biology. 25:3357‐3363.  Wang, L., W. Shen, M. Kazachkov, G. Chen, Q. Chen, A.S. Carlsson, S. Stymne, R.J. Weselake, and J.  Zou.  2012.  Metabolic  interactions  between  the  Lands  cycle  and  the  Kennedy  pathway  of  glycerolipid synthesis in Arabidopsis developing seeds. Plant Cell. 24:4652‐4669.  Wang,  Y.,  and  C.  Kent.  1995a.  Effects  of  altered  phosphorylation  sites  on  the  properties  of  CTP:phosphocholine  cytidylyltransferase.  The  Journal  of  biological  chemistry.  270:17843‐ 17849.  Wang,  Y.,  and  C.  Kent.  1995b.  Identification  of  an  inhibitory  domain  of  CTP:phosphocholine  cytidylyltransferase. The Journal of biological chemistry. 270:18948‐18952.  Wang, Y., J.I. MacDonald, and C. Kent. 1993. Regulation of CTP:phosphocholine cytidylyltransferase in  HeLa cells. Effect of oleate on phosphorylation and  intracellular  localization. The  Journal of  biological chemistry. 268:5512‐5518.  Wang,  Y.,  J.I. MacDonald,  and C.  Kent.  1995.  Identification of  the nuclear  localization  signal  of  rat  liver CTP:phosphocholine  cytidylyltransferase. The  Journal of biological  chemistry.  270:354‐ 360.  254    Watkins, J.D., and C. Kent. 1990. Phosphorylation of CTP:phosphocholine cytidylyltransferase in vivo.  Lack of effect of phorbol ester  treatment  in HeLa cells. The  Journal of biological chemistry.  265:2190‐2197.  Watkins, J.D., and C. Kent. 1991. Regulation of CTP:phosphocholine cytidylyltransferase activity and  subcellular  location  by  phosphorylation  in  Chinese  hamster  ovary  cells.  The  effect  of  phospholipase C treatment. The Journal of biological chemistry. 266:21113‐21117.  Weber, C.H., Y.S. Park, S. Sanker, C. Kent, and M.L. Ludwig. 1999. A prototypical cytidylyltransferase:  CTP:glycerol‐3‐phosphate cytidylyltransferase from bacillus subtilis. Structure. 7:1113‐1124.  Weber,  U.,  C.  Eroglu,  and  M.  Mlodzik.  2003.  Phospholipid  membrane  composition  affects  EGF  receptor and Notch signaling through effects on endocytosis during Drosophila development.  Developmental Cell. 5:559‐570.  Weedon, M.N., S. Ellard, M.J. Prindle, R. Caswell, H. Lango Allen, R. Oram, K. Godbole, C.S. Yajnik, P.  Sbraccia, G. Novelli, P. Turnpenny, E. McCann, K.J. Goh, Y. Wang, J. Fulford, L.J. McCulloch,  D.B.  Savage,  S.  O'Rahilly,  K.  Kos,  L.A.  Loeb,  R.K.  Semple,  and  A.T.  Hattersley.  2013.  An  in‐ frame deletion at  the polymerase active  site of POLD1 causes a multisystem disorder with  lipodystrophy. Nature genetics. 45:947‐950.  Weinberger, M., A. Mesquita, T. Caroll, L. Marks, H. Yang, Z. Zhang, P. Ludovico, and W.C. Burhans.  2010.  Growth  signaling  promotes  chronological  aging  in  budding  yeast  by  inducing  superoxide anions that inhibit quiescence. Aging. 2:709‐726.  Weinhold,  P.A.,  L.  Charles,  and  D.A.  Feldman.  1994.  Regulation  of  CTP:  phosphocholine  cytidylyltransferase  in  HepG2  cells:  effect  of  choline  depletion  on  phosphorylation,  translocation and phosphatidylcholine levels. Biochimica et biophysica acta. 1210:335‐347.  Wieprecht, M., T. Wieder, C. Paul, C.C. Geilen, and C.E. Orfanos. 1996. Evidence for phosphorylation  of CTP:phosphocholine cytidylyltransferase by multiple proline‐directed protein kinases. The  Journal of biological chemistry. 271:9955‐9961.  Wieprecht,  T.,  O.  Apostolov,  and  J.  Seelig.  2000.  Binding  of  the  antibacterial  peptide  magainin  2  amide to small and large unilamellar vesicles. Biophysical chemistry. 85:187‐198.  Wieprecht,  T., M.  Dathe, M.  Beyermann,  E.  Krause, W.L. Maloy,  D.L. MacDonald,  and M.  Bienert.  1997.  Peptide  hydrophobicity  controls  the  activity  and  selectivity  of  magainin  2  amide  in  interaction with membranes. Biochemistry. 36:6124‐6132.  Wimmer, C., V. Doye, P. Grandi, U. Nehrbass, and E.C. Hurt. 1992. A new subclass of nucleoporins  that functionally interact with nuclear pore protein NSP1. The EMBO journal. 11:5051‐5061.  Wong, C.K. 2014. Novel mutations in PCYT1A are responsible for spondylometaphyseal dysplasia with  cone‐rod dystrophy. Clinical genetics. 85:532‐533.  Xie,  M.,  J.L.  Smith,  Z.  Ding,  D.  Zhang,  and  R.B.  Cornell.  2004.  Membrane  binding  modulates  the  quaternary  structure  of  CTP:phosphocholine  cytidylyltransferase.  The  Journal  of  biological  chemistry. 279:28817‐28825.  Xu,  P.,  J. Okkeri,  S. Hanisch, R.Y. Hu, Q.  Xu,  T.G.  Pomorski,  and X.Y. Ding.  2009.  Identification of  a  novel mouse P4‐ATPase family member highly expressed during spermatogenesis. Journal of  cell science. 122:2866‐2876.  255    Yamamoto, G.L., W.A.  Baratela,  T.F.  Almeida, M.  Lazar,  C.L.  Afonso, M.K. Oyamada,  L.  Suzuki,  L.A.  Oliveira,  E.S.  Ramos,  C.A.  Kim,  M.R.  Passos‐Bueno,  and  D.R.  Bertola.  2014.  Mutations  in  PCYT1A cause spondylometaphyseal dysplasia with cone‐rod dystrophy. American journal of  human genetics. 94:113‐119.  Yang, W., K.P. Boggs, and S. Jackowski. 1995. The association of lipid activators with the amphipathic  helical domain of CTP:phosphocholine cytidylyltransferase accelerates catalysis by increasing  the affinity of the enzyme for CTP. The Journal of biological chemistry. 270:23951‐23957.  Yang,  Y.,  M.  Lee,  and  G.D.  Fairn.  2018.  Phospholipid  subcellular  localization  and  dynamics.  The  Journal of biological chemistry. 293:6230‐6240.  Yao, Z.M., and D.E. Vance. 1988. The active synthesis of phosphatidylcholine is required for very low  density  lipoprotein  secretion  from  rat  hepatocytes.  The  Journal  of  biological  chemistry.  263:2998‐3004.  Yao,  Z.M.,  and  D.E.  Vance.  1990.  Reduction  in  VLDL,  but  not  HDL,  in  plasma  of  rats  deficient  in  choline. Biochemistry and cell biology = Biochimie et biologie cellulaire. 68:552‐558.  Yoon, Y., J. Tong, P.J. Lee, A. Albanese, N. Bhardwaj, M. Kallberg, M.A. Digman, H. Lu, E. Gratton, Y.K.  Shin, and W. Cho. 2010. Molecular basis of the potent membrane‐remodeling activity of the  epsin 1 N‐terminal homology domain. The Journal of biological chemistry. 285:531‐540.  Yorimitsu, T., S. Zaman, J.R. Broach, and D.J. Klionsky. 2007. Protein kinase A and Sch9 cooperatively  regulate  induction of autophagy  in Saccharomyces cerevisiae. Molecular biology of the cell.  18:4180‐4189.  Yoshikawa,  K.,  T.  Tanaka,  Y.  Ida,  C.  Furusawa,  T.  Hirasawa,  and  H.  Shimizu.  2011.  Comprehensive  phenotypic  analysis  of  single‐gene  deletion  and  overexpression  strains  of  Saccharomyces  cerevisiae. Yeast. 28:349‐361.  Young,  R.W.  1967.  The  renewal  of  photoreceptor  cell  outer  segments. The  Journal  of  cell  biology.  33:61‐72.  Zannini, L., D. Delia, and G. Buscemi. 2014. CHK2 kinase in the DNA damage response and beyond.  Journal of molecular cell biology. 6:442‐457.  Zhang, Z.,  Y.  Zhang, F. Gao, S. Han, K.S. Cheah, H.F. Tse, and Q.  Lian. 2017. CRISPR/Cas9 Genome‐ Editing System in Human Stem Cells: Current Status and Future Prospects. Molecular therapy.  Nucleic acids. 9:230‐241.  Zhao, H., and P. Lappalainen. 2012. A simple guide to biochemical approaches for analyzing protein‐ lipid interactions. Molecular biology of the cell. 23:2823‐2830.  Zinser,  E.,  C.D.  Sperka‐Gottlieb,  E.V.  Fasch,  S.D.  Kohlwein,  F.  Paltauf,  and  G.  Daum.  1991.  Phospholipid  synthesis  and  lipid  composition  of  subcellular  membranes  in  the  unicellular  eukaryote Saccharomyces cerevisiae. Journal of bacteriology. 173:2026‐2034.          256    Appendix: Publication Afreen Haider*, Yu-Chen Wei*, Koini Lim*, Antonio D. Barbosa*, Che-Hsiung Liu, Ursula Weber, Marek Mlodzik, Kadri Oras, Simon Collier, M. Mahmood Hussain, Liang Dong, Satish Patel, Anna Alvarez-Guaita, Vladimir Saudek, Benjamin J. Jenkins, Albert Koulman, Marcus K. Dymond, Roger C. Hardie, Symeon Siniossoglou, and David B. Savage. PCYT1A regulates phosphatidylcholine homeostasis from the inner nuclear membrane in response to membrane stored curvature elastic stress. Developmental Cell. 2018;45(4):481-495.e8. * Joint first authorship