Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Structure-function correlates in human apolipoprotein A-1 McLeod, Roger Stephen 1992

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1992_fall_mcleod_roger_stephen.pdf [ 3.47MB ]
Metadata
JSON: 831-1.0086602.json
JSON-LD: 831-1.0086602-ld.json
RDF/XML (Pretty): 831-1.0086602-rdf.xml
RDF/JSON: 831-1.0086602-rdf.json
Turtle: 831-1.0086602-turtle.txt
N-Triples: 831-1.0086602-rdf-ntriples.txt
Original Record: 831-1.0086602-source.json
Full Text
831-1.0086602-fulltext.txt
Citation
831-1.0086602.ris

Full Text

STRUCTURE-FUNCTION CORRELATES IN HUMAN APOLIPOPROTEIN A-IbyROGER STEPHEN McLEODB.Sc.(Hons.), University of British Columbia, 1977A THESIS SUBMITTED IN PARTIAL FULFILMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESTHE(Department of Pathology)We accept this thesis as conformingto the requiredAugust 1992©Roger Stephen McLeod, 1992Signature(s) removed to protect privacyIn presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermISsiOn.(Signature)__________________________Department of /tiThe University of British ColumbiaVancouver, CanadaDate /_1DE-6 (2/88)Signature(s) removed to protect privacyABSTRACTApolipoprotein (apo) A-I is the major protein component of human high density lipoprotein(HDL). The plasma level of HDL or apo A-I is inversely related to the risk of developing atherosclerosisand the protective effect of HDL is believed to be related to its ability to promote the movement ofcholesterol from peripheral cells to the liver. Apo A-I is involved in the remodelling of HDL thataccompanies this reverse” cholesterol transport but the elements of its structure that are responsiblefor its function are not well understood.In this thesis, in vitro mutagenesis and eucaryotic expression were used to study the structureand function of recombinant human apo A-I. Four expression systems were developed and were utilizedto express the wild-type and mutant cDNA constructs. In vitro translation studies in rabbit reticulocytelysate established that the cDNA encodes the precursor, preproapo A-I. This precursor wasproteolytically processed to proapo A-I on addition of microsomal membranes, simulating in vivotranslocation and processing on the membrane of the endoplasmic reticulum (ER). Apo A-I wassynthesized and secreted constitutively from the precursor cDNA by three eucaryotic cell, types:transformed simian kidney (COS), baby hamster kidney (BHK) and chinese hamster ovary (CHO) cells.The apo A-I secreted by COS and BHK cells was proapo A-I, while CHO cells secreted mature apo A-I,indicating that the latter possess propeptide proteolytic activity. Low level propeptide hydrolysis was alsodetected during long term culture collections from BHK cells.In defined, serum free culture conditions, much of the apo A-I synthesized was eventuallydegraded. Long term collections of medium were found to contain the following levels of apo A-I: COSlOng/mI, BHK lOOng/ml and CHO l3Ong/ml per 24 hours.Proapo A-I secreted from COS cells was readily integrated into the HDL density fraction offetal bovine serum. The majority of the apo A-I secreted by CHO cells was lipid-free or lipid-poor butretained its ability to integrate into liposomes in vitro. A large portion of the apo A-I within COS cellsretained its signal peptide sequence following translation, indicating that ER processing of preproapo A-Iwas inefficient. It was concluded that COS cells were a poor model for large scale apo A-I expressionand CHO cells were the best model system for this purpose.The role of the apo A-I propeptide was investigated in BHK cells expressing apo A-I. The11results indicated that the propeptide was required for efficient cellular transport and secretion of apo AI. Removal of the propeptide from the cDNA sequence had no effect on the rate of apo A-I synthesis oron the fidelity of signal peptide hydrolysis, but the altered protein remained in the cells in large vesicularstructures which had some morphologic features of the ER. This change also appeared to reduce therate of apo A-I degradation. The observations suggested that non-hepatic cells expressing apo A-Idegrade a substantial portion of this protein. Furthermore, removing the propeptide caused much of theapo A-I to remain in the cell, perhaps by preventing the movement of the protein out of the ER.The functional roles of the middle and C-terminal regions of the apo A-I sequence were alsoinvestigated by generating mutants in these regions. Deletion of Lys107 (a naturally occurring mutationwith functional abnormalities) had minimal influence on the ability of these proteins to bind toliposomes, although the resulting complexes were more heterogeneous by density gradient centrifugation.Deletion of one amphipathic helix from the C-terminus of apo A-I altered its ability to form stable lipid-protein complexes when compared to wild-type. While recombinant wild-type apo A-I was approximately80% as effective a lecithin:cholesterol acyltransferase (LCAT) activator, the Lys’°7 and a -helix deletionmutants were extremely poor LCAT activators.In conclusion, the results indicate that the propeptide portion of apo A-I is involved in thecellular transport of apo A-I and might regulate the movement of proapo A-I between the ER and theGolgi apparatus. Furthermore, low affinity amphipathic helices in the middle hinge region and the Cterminal region of the apo A-I sequence may play a significant role in the LCAT activating mechanism.These elements do not appear to be major structural determinants of initial phospholipid binding butmay be involved in subsequent transformations of HDL. Extension of these studies will provideimportant insight into the mechanisms underlying the anti-atherogenic properties of HDL.111TABLE OF CONTENTSPageABSTRACT iiTABLE OF CONTENTS ivABBREVIATIONS viiiAMINO ACID DESIGNATIONS xiiLIST OF TABLES xiiiLIST OF FIGURES xivACKNOWLEDGEMENTS xviiCHAPTER 11. INTRODUCTION 1Li. High Density Lipoproteins and HDL Remodelling Processes 31.2. High Density Lipoprotein (HDL) and Atherosclerosis 41.3. HDL and Reverse Cholesterol Transport 51.4. Apolipoprotein A-I 71.4.1. Apo A-I Gene Structure 71.4.2. Apo A-I Gene Expression and Secretion 91.5. Functions of Apo A-I 121.5.1. Apo A-I and HDL Structure 121.5.2. Apo A-I and the Interaction of HDL with Cells 131.5.3. Apo A-I and Lecithin:cholesterol Acyltransferase (LCAT) 151.6. Structural Elements of Apo A-I 161.6.1. The Apo A-I Signal Peptide 161.6.2. The Apo A-I Propeptide 171.6.3. The Ampliipathic Helix Motif 191.6.4. Apo A-I Secondary Structure: Epitope Mapping 231.7. Naturally Occuring Structural Variants of Human Apo A-I 261.8. Expression of Recombinant Apo A-I 301.9. Scope of Thesis: Specific Aims 32CHAPTER 22. MATERIALS AND METHODS2.1. Materials 342.2. Growth and Transformation of E. Co1i 35iv2.3. Purification of DNA 362.3.1. Small Scale Plasmid Preparation 362.3.2. Large Scale Plasmid Preparation 362.3.3. Preparation of M13 Phage DNA 372.3.3.1. Preparation of Uracil-Containing ssDNA from M13 372.3.3.2. Preparation of M13 Phage DNA for Sequencing 382.4. Oligonucleotide-directed Mutagenesis 382.4.1. In vitro Mutagenesis 382.4.2. Identification of Putative Mutants 392.4.3. DNA Sequence Analysis 402.5. Construction of Expression Plasmids 412.5.1. Isolation of cDNA Fragments 412.5.2. Modification of Fragment Ends 412.5.3. Ligation into Expression Plasmids 422.6. In vitro Transcription 432.7. In vitro Translation 442.8. Eucaryotic Cell Culture 442.9. Transient Transfection of COS-1 Cells 452.10. Isolation of Stable Eucaryotic Cells Expressing Apo A-I 452.10.1. Calcium Phosphate Transfection 452.10.2. Selection of Stably Transfected Cells 462.10.2.1. BHK Cells 462. 10.2.2. CHO-K1 Cells 462.10.3. Screening of Clones for A-I Secretion 472.10.4. Amino-Terminal Amino Acid Sequence Analysis 472.11. Metabolic Labelling Studies 482.11.1. Determination of Synthesis Rate 482.11.2. Long Term Labelling Studies 482.11.3. Determination of Apo A-I Degradation and Secretion 492.12. Determination of S]Methionine Incorporated into Protein and Apo A-I 492.13. Isolation of Apo A-I by Immunoabsorption 492.14. Electrophoretic Analysis 502.14.1. DNA Fragment Separation on Agarose Gels 502.14.2. Protein Analysis 502.14.2. 1. SDS-Polyacrylamide Gels 502.14.2.2. Isoelectric Focusing Gel Electrophoresis 512.14.2.3. Two-dimensional Polyacrylamide Gel Electrophoresis 512.14.2.4. Immunoblot Analysis 512.15. Indirect Immunofluorescence Microscopy 522.16. Immunogold Electron Microscopy 53V2.17. Quantitation of Apo A-I by Competitive ELISA 532.18. Analysis of Apo A-I Function 542.18.1 Preparation of Single Bilayer Vesicles 542.18.2 Assessment of Lipid Binding Characteristics 542.18.3 Measurement of LCAT Cofactor Activity 55CHAPTER 33. DEVELOPMENT OF APO A-I EXPRESSION SYSTEMS3.1. Sequencing of the Full Length Apo A-I cDNA 563.2. In vitro Transcription and Translation of the Full Length cDNA 593.3. Expression of Wild-type Apo A-I (Apo A-Iwt) in COS Cells 623.3.1. Properties of Apo A-I Secreted by COS Cells 683.3.2. Immunofluorescence Localization of A-Iwt in COS Cells 733.4. Baby Hamster Kidney (BHK) Cell Expression of Apo A-Iwt 753.4.1. BHK-AIwt Cells Produce proapo A-I 753.4.2. Characterization of Apo A-I Synthesis and Secretion 773.5. CHO-Ki Cell Expression of Apo A-Iwt 823.6. DISCUSSION 85CHAPTER 44. THE APO A-I PROPEPTIDE AND INTRACELLULAR TRAFFICKING4.1 OVERVIEW 914.2 RESULTS 914.2.1 Transient Expression of Apo A-IdPRO in COS Cells 914.2.2. Stable Expression of Apo A-IdPRO in BHK cells 974.2.3. Comparison of Apo Al Synthesis and Secretion Rates 994.2.4. Intracellular Localization of Apo A-I Accumulations 1024.3. DISCUSSION 105CHAPTER 55. LYS’°7 AND THE C-TERMINAL AMPHIPATHIC HELIX IN LCAT ACTIVATION5.1 OVERVIEW 1115.2 RESULTS 1115.2.1. Production and Identification of Mutant cDNAs 1115.2.2. Transfected COS Cells Produce the Respective Mutant Proteins 1125.2.3. Development of CHO Cell Lines Expressing Apo A-I Mutants 1155.2.4. Role of Lys107 in Apo A-I Cellular Transport and Secretion 1175.2.5. Functional Characteristics of Apo A-I Mutants 118vi5.2.5.1. Lipid Binding Properties of Apo A-I Mutants 1185.2.5.2. LCAT Activation by Apo A-I Mutants 1255.3 DISCUSSION 128CHAPTER 66. CONCLUSIONS6.1 Summary of Major Findings 1306.2 Perspectives for Future Study 133REFERENCES 136viiABBREVIATIONSApo apolipoproteinATP adenosine triphosphateBHK baby hamster kidneybp base pairBSA bovine serum albuminCAD coronary artery diseasecDNA complementary DNACE cholesteryl esterCETP cholesteryl ester transfer proteinCHO chinese hamster ovaryChylo chylomicronCIP calf intestinal phosphataseCMV cytomegalovirusConA concanavalin ACOS transformed simian kidney cellDl or d(Pro220-As ?41) deletion of helix Prc?20-Asr41DAB 3,3-diaminobenzidinedAT? deoxyadenosine triphosphatedCTP deoxycytidine triphosphateddATP dideoxyadenosine triphosphateddCTP dideoxycytidine triphosphateddGTP dideoxyguanosine triphosphateddTTP dideoxythymidine triphosphateDEAE diethylaminoethylDEPC diethylpyrocarbonatedGTP deoxyguanosine triphosphateDHFR dihyrofolate reductaseviiidK107 ,dLys107 deletion of lysine residue at position 107DMEM Dulbeccos modified minimal essential mediumDMPC dimyristoyl phosphatidylcholineDMSO dimethylsulfoxideDNA deoxyribonucleic acidDNase deoxyribonucleasedNTP deoxynucleotide triphosphate mixturedPRO propeptide deletiondsDNA double-stranded DNADTT dithiothreitoldTTP deoxythymidine triphosphatedut deoxyuracil triphosphataseEDTA ethylenediamine tetra-acetic acidELISA enzyme-linked immunosorbent assayER endoplasmic reticulumFBS fetal bovine serumFITC fluorescein isothiocyanateG418 GeneticinGGE gradient gel electrophoresisHDL high density lipoproteinHDL-C high density lipoprotein cholesterolHEPES N-2-hydroxyethylpiperazine-N-2-ethanesulfonic acidhGH human growth hormoneHRP horse radish peroxidaseHTGL hepatic triacylglycerol lipaseIEF isoelectric focussingIgG immunoglobulin GIPTG isopropyl-8 -D-thiogalactopyranosideixIVS intervening sequence (intron)kD kilodáltonsLB Luria-BertaniLCAT lecithin:cholesterol acyltransferaseLDL low density lipoproteinLys107 lysine residue at position 107mMT-I mouse metallothionein-IMOl multiplicity of infectionmRNA messenger RNAO.D. optical densityPAGE polyacrylamide gel electrophoresisPBS phosphate buffered salinePEG polyethylene glycolp1 isoelectric pointPMSF phenylmethylsulfonyl fluoridePNK polynucleotide kinasepoly A polyadenylationrATP adenosine triphosphateRCT reverse cholesterol transportrCTP cytidine triphosphateRF replicative formRFLP restriction fragment length polymorphismrGTP guanosine triphosphateRNA ribonucleic acidRNase ribonucleaseRNAsin placental ribonuclease inhibitorrUTP uridine triphosphateSDS sodium dodecylsulfatexSSC saline sodium citratessDNA single-stranded DNAhalf-timeTAE Tris-acetate-EDTA bufferTBE Tris-borate-EDTA bufferTBS Tris buffered salineTCA trichloroacetic acidTE 10mM Tris-HC1, pH8.0/ 0.1mM EDTATG triacyiglycerol, triglycerideTris tris(hydroxymethyl)aminomethaneTRITC tetramethylrhodamine isothiocyanateUC unesterified cholesterolUFP universal forward primerung uracil N-glycosylaseURP universal reverse primerVLDL very low density lipoproteinWGA wheat germ agglutininwt wild-typeX-GAL 5-bromo-4,4-chloro-3-indoyl-3 -D-galactosideYT yeast tryptonexiAMINO ACID DESIGNATIONSAmino Acid Three Letter Single LetterGlycine Gly GAlanine Ala AThreonine Thr TSerine Ser STyrosine Tyr YTryptophan Trp WAspartic acid Asp DGlutamic acid Glu EGlutamine Gln QAsparagine Asp NPhenylalanine Phe FArginine Arg RLeucine Leu LProline Pro PMethionine Met MHistidine His HLysine Lys KValine Val VIsoleucine Ile ICysteine Cys CxiiLIST OF TABLESTable Description PageI Properties and composition of lipoprotein classes 2II The plasma apolipoproteins 3III Variant forms of apolipoprotein A-I 27IV Apo A-I mutagenic oligonucleotide primers and their 38propertiesV Apo A-I cDNA sequencing primers 41VI Amino terminal sequence analysis of apo A-I secreted 77from cell line BHK-AIwtB5VII Effect of growth conditions on apo Al synthesis and 79secretion in BHK-AIwtB5VIII Secretion of apo A-I recombinants by CHO cell lines 117XIIILIST OF FIGURESFigure Description Page1 Generalized structure of the plasma lipoproteins 12 Schematic diagram of the process of reverse cholesterol 6transport3 Location and structural organization of the human apo A-I 7gene4 Schematic diagram of the charge and size heterogeneity of 9apo A-I at different stages of proteolytic processing5 Diagramatic representation of the proteolytic processing of 11the apo A-I precursor6 Hydropathy plot of the apo A-I precursor 197 Helical wheel projection of the consensus sequence for the 2122 residue repeat structure of human apo A-I8 Schematic diagram of the hinged domain hypothesis for 24the interaction of apo A-I with HDL9 Proposed supersecondary structure of human apo A-I 2510 DNA sequence analysis of the apo A-I cDNA pBL13AI 5711 In vitro transcription and translation of the full length apo 60A-I cDNA12 Expression of the apo A-I cDNA in transiently transfected 64COS cells13 Immunoblot analysis of apo A-I expression in COS cells 6714 Apo A-I produced by transfected COS cells is associated 69with HDL15 Apo A-I secretion from transfected COS cells is 71stimulated by FBS16 Apo A-I produced in COS cells is subject to intracellular 72phosphorylation17 Immunofluorescence analysis of transfected COS cells 74indicates that apo A-I is retained in the ERxivFigure Description Page18 BHK cells expressing apo A-Iwt produce a single 76molecular species with the electrophoretic properties ofproapo A-I19 FBS stimulates the secretion of apo A-I by BHK-AIwtB5 7820 Long-term pulse-chase analysis suggests that degradation 80competes with secretion of apo A-I in BHK-AIwtB521 Zinc sulfate induction of apo A-I synthesis results in the 81accumulation of apo A-I immunofluorescence in BHKAIwtB522 CHO cells expressing apo A-Iwt secrete mature apo A-I 8323 Removal of the propeptide from the apo A-I sequence 92delays its secretion from transfected COS cells24 Apo A-IdPRO is retained in the ER of the transfected 96COS cell25 BHK clones contain proapo A-I (p-Al) or mature apo A-I 98(rn-Al)26 Apo A-I and protein synthetic rates in BHK lines p-Al 100and rn-Al27 Apo A-I secretion and accumulation in BHK lines p-Al 101and rn-Al28 Time course of apo A-I degradation and secretion in BHK 103cell clones29 Immunofluorescence localization of apo A-I in transfected 104BHK cell lines30 Immunogold localization of apo A-I in ultrathin 106cryosections of transfected BHK cell lines31 Transfected COS cells express apo A-I mutants with the 113appropriate molecular characteristics32 CHO cells express apo A-I mutants of the expected charge 116and size33 The apo A-IdK’°7 mutant protein is secreted and degraded 119more rapidly than apo A-Iwt in transfected COS cells34 Apo A-IdK107 incorporates into extracellular lipoprotein in 121cell culturexvFigure Description Page35 Cells expressing apo A-IdK107 contain apo A-I 122accumulations in both the ER and the Golgi apparatus36 Apo A-I and mutants are secreted from CHO cells into 123serum free medium in lipid-poor form37 Recombinant apo A-I produced by CHO cells activates 126LCAT to approximately the same extent as the purifiedplasma protein38 The ability of apo A-I mutants to activate LCAT is 127markedly diminished39 Proposed model for the role of the apo A-I propeptide in 132BHK cell transportxviACKNOWLEDGEMENTSThe invaluable assistance of members of the Hayden, Gillam and Pritchard laboratories was ofparamount importance at various phases of this work. In particular, I would like to thanl CarolynRobbins, Linda Peritz and Jeff Hewitt for teaching me the aspects of molecular biology which wererequired to initiate the work. Tom Hobman, Marita Lundstrom and Paul Sunga were instrumental in myintoduction to cell biology and occasionally provided a warm place to unwind. Alan Burns providedexpertise in cryoelectron microscopy and Dr. Ruedi Abersold performed the amino terminal sequenceanalysis. I would like to thank Karmin 0 and John Hill for their important contributions to my life, inscience and outside. I acknowledge all members of the Pritchard laboratory: Haydn, Jiri, Linda, Xingbo,Lida, Liz, Janine and Stan for providing the framework in which to operate. Most importantly, Iacknowledge the tolerance and personal support of my family: Shirley, Clinton arid Keith who haveprovided stability and sustenance through late nights and short weekends.This thesis is dedicated to the memory of my father, Gerald Wesley McLeod, and mygrandmother, Lena Springfeld.xvii1. INTRODUCTIONThe plasma lipoproteins are microemulsions composed of a non-polar lipid core surrounded bya polar surface monolayer which maintains the solubility of the complex in the aqueous environment ofthe plasma (Figure 1). The core components are mainly triglycerides (TG) and cholesteryl esters (CE),while the surface layer is composed of unesterified cholesterol (UC), phospholipids (PL) and associatedproteins, the apolipoproteins.Figure 1. Generalized structure of the plasma lipoproteins. PL = phospholipid, UC = unesterifiedcholesterol, TG = triacylglycerol, EC = esterified cholesterol, Apo = apolipoprotein.Lipoproteins differ in the quantity of the various components and have been classified on thebasis of their size and density. These properties form the basis for the separation techniques employed toisolate these macromolecules from the plasma (Table I). The major classes can also be separated byelectrophoresis in agarose gels (Noble, 1968). In this system, chylomicrons remain at the applicationpoint, VLDL migrates in the prqB region, LDL migrates in the fi region and HDL has x -mobility. Thelarger lipoproteins (chylo and VLDL) contain the bulk of the plasma TG, whereas the smaller classes(principally LDL and HDL) are relatively TG poor and carry the majority of the plasma cholesterol. Innormal individuals, LDL carries approximately 60-75% of the plasma cholesterol, and the remaining 25-40% is carried in the HDL density class. However, the distribution of components among thelipoproteins is subject to dynamic processes and the ultimate lipid composition of each lipoprotein isPLI iUG////i TG, ECApo1governed by its apolipoprotein content and by the action of enzymes which alter the lipid moiety of thesemacromolecular structures.Table I. Properties and composition of lipoprotein classes Size, hydrated density and mass compositionof the major lipoprotein classes from normal individuals. Adapted from Havel and Kane (1989).Surface CoreDensity DiameterClass Components Components(g/ml) (nm) (wt%) (%)UC PL Apo TG CEChylo 0.93 75-1200 2 7 2 86 3VLDL 0.93-1.006 30-80 7 18 8 55 12IDL 1.006-1.019 25-35 9 19 19 23 29LDL 1.019-1.063 18-25 8 22 22 6 42HDL2 1.063-1.125 9-12 5 33 40 5 17HDL3 1.125-1.210 5-9 4 25 55 3 13The apolipoproteins are a family of polypeptides characterized by their ability to solubilize lipidsspontaneously in an aqueous environment. Macheboeuf (1929) showed that plasma lipids were associatedwith proteins in water-soluble macromolecular complexes. These proteins have been obtained in pureform by extracting the lipid from the isolated lipoprotein and separating the individual protein species byprotein fractionation techniques. The apolipoproteins range in size from 8,000 to 550,000 Daltons (TableII) but share some structural characteristics and the ability to interact with lipids.The apolipoprotein component of each lipoprotein particle governs its interaction with cells andwith other proteins. Since protein structure is genetically determined while lipoprotein size andcomposition is largely governed by physicochemical considerations, the apolipoproteins provide animportant link to our understanding of the genetic control of lipoprotein metabolism. The elucidation ofspecific functional roles of the individual apoproteins in lipoprotein metabolism remains a majorchallenge of lipoprotein biochemistry.2Table II. The plasma apoilpoproteins. Plasma concentration (mean value for normolipidemic donors)and polypeptide molecular weight of the plasma apolipoproteins. Adapted from Havel and Kane (1989).Plasma Concentration Molecular WeightApolipoprotein (gIL) (kilodaltons)A-I 1.3 29.0A-lI 0.4 17.4A-IV 0.15 44.5B-48 0 241B-100 0.8 513C-I 0.06 6.6C-lI 0.03 8.9C-Ill 0.12 8.8D 0.10 19.0E 0.05 34.11.1. High Density Lipoproteins and HDL Remodelling ProcessesHDL is the most protein-rich of the plasma lipoproteins. Fifty percent of HDL mass isapoprotein, of which 70% is apolipoprotein (apo) A-I. HDL can be separated into two majorsubfractions by ultracentrifugation, HDL2 and HDI. While this distinction may not reflect truemetabolic pools of HDL, the two are sufficiently distinct to warrant individual consideration. HDL2 andHDL differ in density and in size as a result of differences in lipid and protein composition. However,additional properties of the HDL subfractions differ due to the relative quantities of the individualapolipoproteins. These differences may more accurately reflect the metabolic heterogeneity of HDLs.Thus, subpopulations of HDL can be separated by size criteria using nondenaturing gradient gelelectrophoresis (GGE) or by immunoaffinity chromatography on columns with coupled antibody to aspecific apolipoprotein. Using the latter technique, lipoproteins containing apo A-I only (Lp-AI) andlipoproteins containing apo A-I and apo A-Il (Lp-A-I!A-II) can be isolated (Cheung et al, 1987).The HDLs are modified during a series of enzymatic conversion and lipid transfer processes.The details of this complex remodelling have only recently been elucidated. High density lipoproteinsenter the circulation as precursors from two separate sources: the liver and the intestine. A nascent formof HDL is liver-derived and is protein-enriched and lipid-poor compared with the mature HDL.3Intestinally-derived chylomicrons are partially catabolized in the circulation and release surface remnants,containing PL, UC and apo A-I, which are also precursors of HDL. These two nascent species acquireadditional lipid and protein components by exchange processes in the circulation and can be isolated inthe HDL density range. Precursors from either source acquire UC from other lipoproteins or from cellmembranes. A small portion of the UC-enriched HDL has pre-3 electrophoretic mobility on agarose gelelectrophoresis and appears to be a major portion of the Lp-A-I isolated by immunoaffinity tecirniques.The cholesterol in this fraction is subsequently esterified by the enzyme lecithin:cholesterolacyltransferase (LCAT, EC 2.3.1.43) to produce lipoprotein of HDL3 density and electrophoreticmobility. Cholesterol esterification within the lipoprotein leaves the surface relatively deficient in UC asthe product CE moves from the particle surface to its core. The UC-depleted HDL then acquiresadditional UC from other lipoproteins or from cells by diffusion along the concentration gradient. Theparticle enlarges and becomes lighter as the lipid-to-protein ratio increases, eventually attaining HDL2density. The HDL remodelling cycle is completed by the action of hepatic triglyceride lipase (HTGL)and cholesterol ester transfer protein (CETP). These activities remove surface phospholipid and corecholesteryl ester (CE), respectively, to regenerate HDJ. The CE may be transferred to other plasmalipoproteins or may be taken up directly by hepatocytes and catabolized. Under most conditions CE istransferred to lower density lipoproteins in exchange for TG. HTGL possesses phospholipase activity,responsible for hydrolysis of HDL phospholipid, and TG hydrolase activity, contributing to a reduction inparticle diameter. HTGL and CETP may act sequentially or, alternatively, the modifications may occursimultaneously on the hepatocyte surface.1.2. High Density Lipoprotein (HDL) and AtherosclerosisAtherosclerosis is the progressive narrowing of large arteries due to intimal thickening and lipidaccumulation. It is the major cause of mortality and morbidity in developed nations. Atheroscleroticlesions develop as a result of complex and poorly understood interaction of genetic and environmentalinfluences (Sing and Moll, 1990). Hemodynamic, thrombogenic and metabolic factors contribute to theatherogenic process (Fuster et al, 1992) and the importance of each is supported by both epidemiologicand experimental data. In animal models of atherogenesis, high levels of plasma cholesterol play an4important role in the development and progression of this disease in response to an initiatingbiochemical or physical insult to the vessel wall (Ross, 1986). What is still poorly defined, however, is thesequence of molecular events that initiate the injury and the factors which might prevent or reverse theprogressive accumulation of lipid that characterizes the disease.Epidemiologic studies have attempted to identify the causal and modifying factors inatherogenesis. Studies of free-living populations and of populations selected for increased atheroscleroticrisk have indicated that the plasma cholesterol level, and specifically the portion associated with lowdensity lipoprotein (LDL), are directly correlated with atherogenic risk. However, the strength of thiscorrelation decreases with advancing age. Conversely, the portion of plasma cholesterol associated withHDL (HDL-C) is inversely correlated with atherosclerotic risk (Miller and Miller, 1975; Gordon et at,1977) and the strength of this relationship increases with age, ie. as we age our HDL-C level becomes amore powerful predictor of atherosclerotic risk. Prospective intervention trials (Manninen et at, 1988)have shown that raising the level of HDL-C, with pharmacologic agents or by lifestyle modification, has aprotective effect. Paradoxically, the analysis of genetic disorders affecting HDL level has indicated thatlow levels of this lipoprotein may or may not be associated with increased atherogenic risk (Schaeffer,1984) although derangements of cholesterol metabolism are observed.1.3. HDL and Reverse Cholesterol TransportThe molecular processes through which high levels of HDL reduce atherogenic risk are largelyspeculative. HDL may protect against vascular lipid deposition by mediating the transport of excesscellular cholesterol from peripheral tissues to the liver by the process termed “reverse cholesteroltransport”. While all cells are capable of cholesterol synthesis, only the hepatocyte is capable ofcholesterol degradation. There is, therefore, a need to move cholesterol from the peripheral cell to thehepatocyte and this movement is driven by the HDL remodelling process as indicated in Figure 2.Reverse cholesterol transport (RCT) was originally described by Glomset (1970) and iscomposed of three distinct elements:(i) Fluid phase transport of UC between cell membrane and lipoprotein surface bydiffusion along a concentration gradient. This process may be facilitated by high affinity5binding sites for HDL on the surface of peripheral cells (Oram et at, 1984).(ii) Development of a cholesterol concentration gradient by reduction of UC on theaccessible lipoprotein surface. This is achieved through esterification of UC to CE bythe enzyme LCAT (Fielding and Fielding, 1982).(iii) Transport of CE to the liver in lipoproteins for catabolism and excretion into bile. Thismay involve the direct interaction of HDL with liver cells or may requireinterlipoprotein lipid transfer processes (Barter et at, 1987) and hepatic uptake of theacceptor lipoproteins.Figure 2. Schematic diagram of the process of reverse cholesterol transport. FC = free (unesterified)cholesterol, CE = cholesteryl ester, LAT = lecithin: cholesterol acyliransferase, A-I apo A-I, nHDL =nascent high density lipoprotein, HDL = high density lipoprotein, LDL = low density lipoprotein,VLDL= very low density lipoprotein. After Fielding and Fielding (1982).Recent evidence has suggested that apolipoprotein A-I may be involved in all three aspects ofRCT. In addition, the expanding knowledge of the molecular genetics of this apoprotein has indicatedthat abnormalities of its gene locus may be associated with some forms of hyperlipidemia andatherosclerosis (Karathanasis et al, 1983; Rees et at, 1985; Ordovas et al, 1986).LIVERCELL MEMBRANECETPFCCEI LCATJ CEA-IEFFLUX -+ ESTERIFICATION - TRANSFER - CLEARANCE61.4. Apolipoprotein A-I1.4.1. Apo A-I Gene StructureThe mature plasma apo A-I protein is a single polypeptide chain of 243 amino acids with amolecular weight of approximately 28,000 daltons. The gene for apo A-I is one member of the solubleapolipoprotein gene family which includes apolipoproteins A-Il, A-IV, C-I, C-lI, C-Ill and E. The genesfor these apoproteins have regions of homology and similar genomic structure with similar distribution ofintron-exon borders, suggesting evolution from a common ancestral gene (Li et at, 1988).The human apo A-I gene has been cloned and sequenced (Law et at, 1983; Karathanasis et at,1983b; Cheung and Chan, 1983; Sharpe et at, 1984) and the cDNA sequence has verified the amino acidsequence obtained earlier from the purified plasma protein (Brewer et at, 1978).AA-I C-Ill A-IV5• I___13.BPstI PstIJ IVS-1 PJS-2 IVS-3 I50 100 150 200 Amino AcidI I I IV////,V////JV////t’////4’////4’////F/A’////,Exon3 Exon4Figure 3. Location and structural organization of the human apo A-I gene. A- Chromosomalorganization of the apo A-I/C-III/A-IV gene cluster on chromosome 11. B- Approximate positions ofthe intervening sequences (IVS) within the apo A-I gene. PstI indicates the location of the restrictionsites for this enzyme which flank the 2.2 Kbp gene. C- Organization of amphipathic helical repeats withinthe apo A-I coding region indicating the derivation from exon 3 or exon 4 of the gene. Large rectanglesindicate 22 amino acid residue helix, small rectangles are 11 residue repeats. (Redrawn from Segrest etat, 1992).cDNA clones have provided significant information on the structural and evolutionary aspects of7apo A-I. The cDNA sequence specifies a mRNA of 950 base pairs (bp) including 35 bp of 5 flankingsequence and 54 bp of 3 untranslated sequence preceding a poly A tail. The coding region spans 893 bp(see Figure 3) of which repeating sequences of 66 bp occupy the 3 half of the message and encode 22amino acid repeats of -helical structure in the apoprotein. The repetitive nature of this region has beenthe subject of intense speculation regarding evolutionary origins. Karathanasis and colleagues (1983b)have suggested that the sequence arose out of intragenic duplication events. Based on computer analysis,Fitch et at (1984) have proposed a smaller repeating sequence (33 bp) with duplication and unequalcrossover events leading to the observed 22 amino acid repeat structure. This structural motif is commonto many of the soluble apolipoproteins.Comparison of apo A-I genomic DNA and cDNA clones (Shoulders et at, 1982; Karathanasis etat, 1983b; Sharpe et al, 1984) has provided insight into the genomic organization. The apo A-I gene hasbeen mapped to the long arm (q) of chromosome 11 where it is clustered with apolipoproteins C-Ill(Bruns et at, 1984) and A-TV. The apo A-I gene is 1863 bp in length and contains three interveningsequences (IVS) which interupt coding region at sites which are similar in apos A-I, A-IT, C-Il, C-Ill andE (Li et at, 1988). In apo A-I, exon 1 contains the 5-untranslated region of the mRNA and is interruptedby the first intervening sequence (IVS-1, see Figure 3B). Exon 2 contains a small portion of 5-untranslated sequence and the majority of the signal peptide coding region. This is interrupted near thesignal peptide hydrolysis site by the second intervening sequence (IVS-2). Exon 3 encodes the signalpeptidase recognition site, and the N-terminus of the protein to residue 43 of the mature protein. Thethird intervening sequence (IVS-3) interrupts this codon and the remaining protein coding sequence ofexon 4 which comprises the remaining x -helical tandem repeats.A number of studies have attempted to identify a genetic marker for the presence ofatherosclerosis, and apo A-I has been a candidate gene. Restriction fragment length polymorphisms(RFLP) within the A-I/C-III/A-IV gene cluster have been identified in some populations, and theirrelationship to hyperlipidemia and/or atherosclerosis has been tested. In general, the results of thesestudies have been conflicting or misleading. Reliable markers in the vicinity of the apo A-I gene have notbeen consistently identified in the affected groups (Breslow, 1992). Most gene polymorphisms would notbe expected to affect apo A-I structure or function, since they are found more frequently in the regions8flanking the gene or within intervening sequences. Only rarely have mutations been described in whichapo A-I structure is altered (see Section 1.7).1.4.2. Apo A-I Gene Expression and SecretionIn humans, apo A-I mRNA is found only in the liver and the small intestine (Li et at, 1988).Both the hepatocyte and the enterocyte have been shown to synthesize and secrete apo A-I protein(Zannis et at, 1983). Intracellular processing of the primary translation product in these tissues has beendemonstrated and the successive stages can be monitored by changes in the p1 of the product (Figure 4).Zarmis and Breslow (1985) have established a nomenclature for cellular and plasma apo A-I speciesbased on isoelectric point (p1): A-I2, p1= 5.85; A-I4, p1= 5.74; A-I4, p1= 5.65; A-I4, p1= 5.52; and A- I4,p1 = 5.40. The predominant isoform in normal plasma is A-I4.. (). 0...... . . .Figure 4. Schematic diagram of the charge and size heterogeneity of apo A-I at different stages ofproteolytic processing. Solid circles represent the species found in the biological system indicated on theright. Dotted circles indicate the position of mature plasma apo A-I marker. Symbol size is proportionalto the quantity of each isomorphic form. Arrows at left indicate the direction of migration in eitherisoelectric focussing (IEF) or sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE).The numbers at the top of the diagram refer to the subscript of the apo A-I isoform as referred to in thetext. (Modified from Bojanovsky et al, 1985)123456SDS4 Preproapo A-IProapo A-IMature Apo A-ILymphPlasma9Apo A-I is synthesized as a preproapolipoprotein (Figure 5) containing an amino terminal 18residue prepeptide, or signal sequence, which directs the protein to the endoplasmic reticulum (ER). N-terminal prepeptides are a common structural feature of many proteins which are secreted fromeucaryotic cells (Nothwehr and Gordon, 1990). Other proteins contain signal sequences which are adistance from the N-terminus (internal signal sequences) and some secretory proteins lack anyprepeptide (Meusch et at, 1990). As preproapo A-I is synthesized on the ribosome, the emerging signalpeptide is recognized by a cytosolic factor, signal recognition particle (SRP). SRP binds the peptidechain, stopping further translation instil SRP anchors the ribosome to a specific ER membrane receptor.Translation then continues on the bound ribosome and the growing polypeptide is translocated across theER membrane into the lumen. Coincident with this translocation, the signal peptide is hydrolysed bysignal peptidase which is found on the luminal membrane of the ER. This processing has beendelineated in studies of N-terminal signal sequences based on the original work of Blobel andDobberstein (1975) who isolated translocation and proteolytic activity in the microsomal pellet ofsubcellular fractionations.The early secretory targetting of preproapo A-I has also been described in rabbit reticulocytes(Zannis et at, 1983). Preproapo A-I has p1>5.85 and is approximately 2000 Daltons larger than themature plasma protein. The proapo A-I (A-I2) product of signal peptide hydrolysis can be identifiedduring in vitro translation in the presence of microsomal membranes and in secretions from hepatocytes.The function of the apo A-I propeptide is not known although functions of propeptides in other plasmaproteins have been delineated (Peters, 1987). Some propeptides mediate the correct folding of thepolypeptide chain (eg., insulin). In other proteins, the propeptide serves to direct the post-translationalmodification of the primasy translation product (eg., y -carboxylation of clotting factors). Propeptides canalso be used to target proteins to a specific cellular compartment or as a control element which regulatesthe synthesis or secretion of the mature protein (eg., neuroendocrine hormones).A common feature of propeptides is the amino acid sequence at the hydrolysis site, which isoften preceeded by two basic amino acid residues, Arg-Arg or Lys-Arg. The enzymes which catalyse thehydrolysis are poorly characterized, although a recent report suggests that similar endopeptidases maycatalyse the hydrolysis of more than one proprotein (Wise et at, 1990). The apo A-I propeptide is10unusual, in that the hydrolysis site is preceded by two Gin residues (Figure 5).Preproapo A-I[MetLysAla SerGinAla ArgHisPheTrpGInGIn Asp Gin-24 -7 -6 -1 +1 243Signal Peptidase JrProapo A-I I ArgHisPheTrpGInGIn Asp Gin—6 -1 ÷1 243A-I Specific Propeptidase JrMature Apo A-I I Asp Gin I+1 243Figure 5. Diagramatic representation of the proteolytic processing of the apo A-I precursor showing theamino acid sequnece in the region of proteolysis. Positive and negative integer labels indicate the aminoacid sequence with respect to the first residue (+1) of the mature protein.The macromolecular nature of secreted apo A-I is not clear at present. Proapo A-I appears tobe secreted in a lipid-free or lipid-poor state and is then incorporated into HDL outside the cell (Vanceand Vance, 1990). Such a process would, by necessity, require remodelling of the lipoprotein toaccommodate the additional surface apoprotein. Although apo A-I appears to be secreted as theproprotein, this form accounts for less than 10% of circulating apo A-I. Propeptide hydrolysis may occurat or shortly after incorporation into HDL. An extracellular endopeptidase (propeptidase) catalysing thisreaction has been identified in lymph and plasma (Scanu, 1987) and it has been suggested that thismodification may regulate apo A-I metabolism or HDL biogenesis in the plasma compartment(Edeistein et al, 1983, Bojanovsky et al, 1985). A similar converting activity has been described for the11apolipoprotein A-IT precursor (Gordon ci at, 1984). The extracellular conversion of proapo A-I (A-I2) tomature apo Al (A-I4) is a slow process which has been demonstrated in vivo (Bojanovsky et at, 1985)and in vitro (Edeistein et at, 1983). Processing activity is Ca dependent (EDTA sensitive) and hasunusual proteolytic specificity for the Glrf1 -Asp’ peptide bond (Edelstein ci at, 1983). The enzyme is nota serine protease, based on serine-specific inhibitor studies. While some investigators have suggested thathydrolysis of the propeptide may be required for the secreted apo A-I to associate with HDL (Schmitz ciat, 1983), the timing of propeptide hydrolysis and lipoprotein integration is still largely speculative.Evidence from cell culture systems (Mallory ci at, 1987; Forte ci at, 1990) which do not normally expressapo A-I but do possess proapo A-I endopeptidase activity, suggests that the apo A-I propeptidase maynot be absolutely specific for proapo A-I.Additional post-traislational modifications of apo A-I have also been identified. Covalentphosphorylation (Beg ci at, 1989) and fatty acylation (Hoeg ci at, 1986) have been demonstrated,although the function of these modifications is unclear. It has been suggested that these additions may berequired for correct intracellular trafficking or efficient secretion. Apo A-I is not glycosylated, unlikemany secreted proteins, including apolipoproteins B, E and C-Ill. Isomorphic forms A-I4 and A-I4 onIEF gels probably represent deamidation products of the mature A-I4, which apparently arise over timein the circulation (Bojanovsky ci at, 1985). The origin and fate of apo A-I4, a minor component ofcirculating A-I, are unclear, but this peptide may be a deamidation product of proapo A-I (A-I4).1.5. Functions of Apo A-I1.5.1. Apo A-I and HDL StructureThe functions of apo A-I are tightly associated with the role of HDL in RCT. Apo A-I serves animportant structural function, maintaining the solubility of the polar lipids that are carried in the HDLdensity class. In addition, however, apo A-I performs metabolic functions which are central to theremodelling of HDL and RCT.The importance of apo A-I structure to HDL metabolism has been underscored by studies ofhuman apo A-I in transgenic mice. Mouse and human apo A-I are structurally distinct since antibodieshave been isolated which do not cross-react between the two species and cDNA probes for the two12genes do not cross-hybridize. Normal mouse plasma contains a single, homogeneous HDL population incontrast to human plasma which contains two major subpopulations (HDL2 and HDL) which differ inboth size and density. When transgenic mice were established which expressed human apo A-I, theplasma levels of HDL-C and apo A-I increased approximately two-fold (Walsh et at, 1989; Rubin et at,1991a). This increase was largely due to expression of the human transgene product. The quantity of theendogenous mouse apo A-I protein was markedly reduced despite normal mRNA levels. From thesestudies, it appears that the human gene product is more abundant than mouse apo A-I because of moreefficient mRNA translation or protein secretion in these animals. More importantly, the species of HDLin plasma become polydisperse, and the pattern resembles that found in human plasma.The transgenic human apo A-I mouse has also reinforced the notion of the protective role ofapo A-I in atherogenesis. Rubin et at (1991b) assessed the development of atherosclerotic lesions iii asusceptible mouse strain (C57BL/6) with or without the human apo A-I transgene. On an atherogenicdiet, these mice without the transgene rapidly developed fatty streak lesions, while those mice expressingthe human apo A-I developed significantly fewer lesions. The relative contributions of HDL level andHDL composition in this model have yet to be established. Chajek-Shaul et al (1991) have suggested,based on kinetic studies, that the presence of human apo A-I might prevent the selective hepatic uptakeof CE from HDL in these mice. They propose that elements of human apo A-I structure prohibit directHDL-CE removal by particle uptake, necessitating transfer to LDL and whole particle uptake as theroute of CE removal. Therefore, in the presence of human apo A-I, CE can only be removed fromcirculation if CETP is also present. Since mice do not have CETP, the esters accumulate in HDL in thetransgenic animal model. Thus, a more appropriate model for human HDL metabolism might be ananimal possessing human transgenes for both CETP and apo A-I.1.5.2. Apo A-I and the Interaction of HDL with CellsThe first component of reverse cholesterol transport is the desorption of excess cellularcholesterol onto HDL from cells. HDL has been shown to interact with cells and apo A-I has beenimplicated as the ligand for the interaction with an HDL binding protein. Biesbroeck el at (1983) andOram et at (1984) characterized high affinity, saturable binding of HDL to cultured cells which was13biochemically distinct fiom binding to the LDL receptor. Binding was increased by cholesterol loading ofthe cells and evidence was obtained that apo A-I was the ligand for binding (Fidge and Nestel, 1985;Brinton et at, 1985). Tyrosine residue modification (Brinton et at, 1985; Chacko, 1985) markedlydiminished binding, suggesting that this amino acid residue might mediate the interaction. Subsequently,the binding site was demonstrated in many cell types, including human liver (Schouten et at, 1990;Tozuka and Fidge, 1989; Hoeg et at, 1985), but the apoprotein ligand specificity is broader than initiallyanticipated, recognizing apo A-IT and apo A-TV in addition to apo A-I (Dvorin et at, 1986; Tozuka andFidge, 1989). The binding component appears to be a protein of 100-110 kilodaltons (Graham and Oram,1987; Tozuka and Fidge, 1989; Bond et at, 1991) which mediates the translocation of intracellularmembrane UC to the plasma membrane (Oram et at, 1991). Binding of HDL does not affect the rate ofplasma membrane UC desorption (Karlin et at, 1987). Therefore, while the putative receptor doesappear to mediate cellular cholesterol movement, it does not facilitate the direct movement ofmembrane cholesterol into HDL.Schmitz and colleagues (1985) have suggested that a retroendocytotic process is involved in theinteraction of HDL with cells, whereby the lipoprotein is bound, internalized, CE-depleted or UCenriched, and then resecreted without protein degradation. This is in contrast to the receptor-mediatedendocytosis of LDL, where the lipoprotein is degraded following internalization. A retroendocytoticmechanism for HDL uptake and resecretion has also received support from additional studies in otherlaboratories (Rahim et at, 1991; Kambouris et at, 1990; DeLamatre et at, 1990). While neither hypothesisis yet confirmed, elements of both may be involved in the actual mechanism.There is evidence that a plasma membrane protein is responsible for the interaction with HDL.This binding mediates changes in cellular cholesterol metabolism. Following HDL binding, cellular UCmay trarislocate from intracellular membranes to the cell membrane (Oram et at, 1991) or the CE of theHDL may be selectively taken up by the cell (Schmitz et at, 1985). The direction and type of lipidtransfer might depend on the cell type and its cholesterol status. The biologic activity of this bindingprotein (the 110 kD HDL binding protein) has been recently demonstrated by cloning and expression ofthe cDNA (Oram et at, 1992). However, the requirement for a distinct mechanism for interaction ofHDL with cells has not received universal support. Reichl and Miller (1989) have calculated that, in man,14the majority of the CE generated within HDL is delivered to the liver only after transfer to lipoproteinsof lower density. Therefore, the interaction of HDL with cells need not occur for either the peripheraluptake of UC or for the liver delivery of EC. However, even in the absence of a specific cell binding site,HDL is the best acceptor of excess cellular UC.1.5.3. Apo A-I and Lecithin:cholesterol Acyltransferase (LCAT)Lecithin:cholesterol acyltransferase is the plasma enzyme responsible for generating the majorityof plasma cholesterol esters via transfer of sn-2 fatty acid from phosphatidyicholine to the 3-positionhydroxyl group of cholesterol. Apo A-I has long been recognized as a potent activator of this enzymicreaction (Fielding et al, 1972; Soutar et at, 1975; Albers et at, 1979; Matz and Jonas, 1982). Despiteextensive investigation, the exact molecular mechanism of apoprotein activation of LCAT is stillincomplete.Several investigators (Albers et at, 1979; Fielding and Fielding, 1972; McLeod et at, 1986) haveestablished that apo A-I is the principal activator of cholesterol esterification by LCAT under in vitroconditions which resemble those in vivo. Indeed, when apo A-I containing lipoproteins are removed fromplasma by immunoaffmity chromatography, much of the cholesterol esterifying activity is also removed(Fielding and Fielding, 1980; Cheung et al, 1986). This implies a physical association between A-I andLCAT and reflects their functional relationship. While apo A-I and LCAT do not. appear to interactdirectly, LCAT associates largely with lipoproteins that contain apo A-I. In certain pathologic states(Pritchard et at, 1986) and with some in vitro manipulations, LCAT can be activated by otherapolipoproteins. Apo E, apo A-TV and apo C-I have been shown to activate LCAT in the absence of apoA-I (Soutar et at, 1975; Albers et at, 1979; Steinmetz and Utermann, 1985; Steinmetz et at, 1985; McLeodet al, 1986). In the presence of suboptimal levels of apo A-I, apo A-Il can also activate the enzymedespite being inhibitory at optimal apo A-I concentrations (Chen and Albers, 1986; Nishida et at, 1986)[The observation that other apoproteins can activate LCAT is of particular interest since several HDLdeficiency states have been described where cholesterol esterification proceeds normally, or at onlypartially reduced rates, despite extremely low levels of apo A-I (Schaeffer et at, 1984).Our understanding of LCAT activation by apo A-I is incomplete, reflecting our lack of complete15understanding of the mechanism of LCAT catalysis. The application of a molecular genetic approach tothe study of LCAT protein structure will no doubt provide insight into the mechanism of apo A-Iactivation. Studies using chemical modification techniques (Jauhiainen and Dolphin, 1986; Park et at,1987) and the application of predictive algorithms to the LCAT primary sequence (Yang et at, 1987)have provided preliminary evidence for the reaction mechanism, the active site and the secondarystructure of the enzyme. The active site contains a reactive serine residue at position 181 of the primarysequence. This region has considerable sequence homology with the serine esterase class of enzymesincluding the lipases. This residue hydrolyses the sn-2-position fatty acid of phosphatidylcholine to forman enzyme-oxyester intermediate. The proposed mechanism (Jauhiainen and Dolphin, 1986) indicatedthat intraenzyme transfer of the acyl group to a free sulthydryl group at Cys-31 and/or Cys-184 results ina thioester intermediate which ultimately transfers the acyl group to cholesterol. However, the proposedrole for the free sulfhydryl residues in the mechanism has been recently called into question bymolecular genetic studies in which the reactive Cys residues were replaced without affect ontransacylation (Francone and Fielding, 1991). In the absence of the cholesterol acceptor, the oxyesterintermediate formed in the first stage can be hydrolysed by water to form unesterified free fatty acid andregenerate enzyme-OH.Analysis of the cDNA clone for the human LCAT (McLean et at, 1986) has revealed that onlyone region of the enzyme has amphipathic character analogous to the soluble apolipoproteins. However,LCAT can associate with phospholipid surfaces. The question then arises: If LCAT can interact withphospholipid in the absence of apo A-I, what other role might the apolipoprotein play that makes itunique?1.6. Structural Elements of Apo A-I1.6.1. The Apo A-I Signal PeptideApo A-I, like all apolipoproteins, contains an N-terminal signal sequence which targets theprotein for translocation across the ER membrane and into the secretory pathway (see Section 1.4.2).The structure of the apo A-I signal peptide conforms to the general features documented for a numberof eucaryotic proteins (Nothwehr and Gordon, 1990). These include a positively charged N-terminus, a16hydrophobic interior segment (see Figure 6), and small amino acid side chains at positions -1 (Ala) and-3 (Ser) from the cleavage site. The apo A-I signal sequence is 18 residues in length, which is within therange described for many of these peptides (15 to 50 amino acids).1.6.2. The Apo A-I PropeptideThree of the soluble apolipoproteins (A-I, A-IT, and C-IT) have propeptide sequences, for whicha function has yet to be identified. These segments are retained following translocation across themembrane of the endoplasmic reticulum, but are removed as the protein attains its mature plasma form.The recognition sequences may be apolipoprotein-specific since proteolysis occurs at an unusual site andappears to require a specific endoprotease in each case. The apo A-I propeptide ends with a Gln-Glndipeptide and the propeptidase hydrolyses the Gin-Asp peptide bond between the propeptide and themature N-terminus. The propeptides of many other proproteins end in a pair of positively chargedresidues which appear to constitute a recognition site (see Section 1.4.2). The physiological role of apoA-I propeptidase activity (Edelstein et at, 1983) is not yet known, nor are the tissue and speciesdistribution of this hydrolytic activity. This enzyme has been found in plasma and in lymph where thehydrolysis is presumed to take place. Alternatively, the small amount of proapo A-I found in plasma mayrepresent a minor component of apo A-I that escapes a cleavage event which would mainly occur withinthe cell. This suggestion is supported by observations in hepatocyte-derived cultures (HepG2) whichsecrete partially processed (50%) apo A-I (Forte et at, 1987). In addition, the medium of CHO cellcultures expressing the human apo A-I cDNA contain as much as 90% mature apo A-I and only a minorportion as proapo A-I (Mallory et at, 1987). These experimental models suggest that the hydrolysis mightoccur within cells.In vivo, propeptide hydrolysis appears to be sensitive to secondary structure at the N-terminus ofthe protein. Von Eckardstein et at (1989) have studied natural mutations of the amino terminus ofmature apo A-I and found that some mutations altered the quantity of circulating proapo A-I. Theyfound that subjects with the Pro3 to Arg variant had normal plasma levels of HDL-C and apo A-I butthat the ratio of proapo A-I to mature apo A-I in plasma was increased. Similarly, the Pro3 to Hismutation also affected the rate of hydrolysis, but Pro4 to Arg did not. Analysis of the predicted17secondary structure in this region suggested that the apo A-I propeptide forms part of an -helix andthat Pro3 has high probability of being a 8 -turn residue (see Figure 6). It appears, therefore, that criticalpositioning of a,8-turn residue near the Glrf1 -Asp1 hydrolysis site may be an important element ofrecognition by the propeptidase.The presence or absence of the propeptide segment has no clear effect on the extracellularfunctions of apo A-I (Fennewald e at, 1988). While some evidence has suggested that proapo A-I mightbind to lipoprotein less avidly than the mature form (Zannis et at, 1983; Rosseneu et at, 1984), studies ofproapo A-I binding to HDL (Edeistein et a!, 1983) have indicated that the formation of the complex isrequired for the conversion to the mature form.The crucial function of the propeptide may be intracellular. Gordon et al (1986) have suggestedthat the propeptide may play a role in appropriate folding of the polypeptide chain, or may act as atargeting signal to compartmentalize apo A-I within the secretory pathway. Removal of the propeptidefrom the apo A-I cDNA delayed ER translocation during cell-free translation (Folz and Gordon, 1987;Stoffel and Binczek, 1988) but did not affect the fidelity of signal peptide hydrolysis.Two modes of secretion are generally recognized for eucaryotic proteins, and the selection isbased on cell type and elements of protein structure (Kelly, 1985). Endocrine and exocrine cells arecapable of regulated secretion in which protein products are stored in secretory granules for secretion ona specific stimulus. Most other cells secrete by a constitutive pathway in which no storage occurs;proteins are secreted immediately following their synthesis. Secretion of apo A-I from hepatocytes(Vance and Vance, 1990) and probably from enterocytes appears to occur by the constitutive pathway.When the apo A-I cDNA was expressed in the neuroendocrine cell line AtT-20 (Fennewald et at, 1988),cellular transport was not affected by the presence or absence of the propeptide. Neither the secretionrate nor the selection of regulated versus constitutive secretory pathway were altered. However, in thisexpression model apo A-I was secreted primarily via the regulated pathway, perhaps because themultiple amphipathic x -helices in apo A-I can target the protein to the secretory granules (Kizer andTropsha, 1991). Detailed analyses of the role of the apo A-I propeptide on constitutive secretion havenot been conducted.181.6.3. The Amphipathic Helix MotifPrimary sequence analysis and the application of structure prediction algorithms have indicatedthat many proteins contain regions of amphipathic structure which appear to mediate the interaction ofproteims with hydrophobic compounds (Segrest et at, 1990). The amphipathic helix motif is a prominantfeature of all of the apolipoproteins (Segrest et at, 1992). These structural elements are responsible forthe lipid binding properties of this protein class. Much of our understanding of apolipoprotein function isderived from studies of model peptides designed to duplicate their secondary structure, most notablyfrom study of the amphipathic helix (Segrest et at, 1974). In its present form (Segrest et at, 1990), themodel proposes that the hydrophobicity and the cx -helical potential of a given peptide, as defmed by itsprimary sequence, are independent determinants of its structure. Both determinants are required forinteraction with phospholipid surfaces, including lipoproteins. Experimental evidence from otherlaboratories (Ponsin et at, 1986) also supports this contention.Figure 6. Hydropathy plot of the apo A-I precursor, preproapo A-I. Hydropathic index was calculated bythe method of Kyte and Doolittle (1982) using Sequence (Delaney Software, Vancouver, B.C.) andplotted against residue number. Points below the line of average hydropathy indicate a residue likely toreside in an hydrophilic environment, residues above the line indicate probable location in a hydrophobicenvironment. Open bars indicate the positions of the pre- and pro- peptides of the N-terminus. Invertedtriangles indicate the positions of high -turn probability as predicted by Gamier et at (1978).Using the algorithm of Gamier et al (1978), apolipoprotein A-I contains many hydrophobicregions (Figure 6) with 66% of its residues in -helix. All members of the soluble apolipoprotein classhave similar structural features. A common property of the apolipoprotein class is the ability to bind toNATIVE APOLIPOPROTEIN Al><DzC-)I00>-IRESIDUE NUMBER19phospholipid emulsions. This is also a prerequisite for LCAT activation although not all apolipoproteinsactivate LCAT equally. The depth of penetration of the apolipoprotein into the phospholipid monolayer(Fukushima et al, 1980) may be a critical determining factor. Alternatively, apolipoproteins may promoteLCAT catalysis by enhancing the transfer of the enzyme between substrate lipoprotein particles followingtransacylation (Nishida et al, 1986), that is, to increase the enzyme “off-rate”. Apo A-I may promotetransacylation by LCAT more effectively than other apolipoproteins because it has regions whichenhance both binding to and release from phospholipid substrates. As yet, the mechanism of LCATactivation is not clearly established.Some aspects of LCAT activation were identified through analysis of fragments of the purifiedplasma apo A-I protein. Cyanogen bromide cleavage analysis was used to isolate an LCAT activatingregion from purified plasma apo A-I. Four fragments result from this treatment, only two of which arecapable of activating the enzyme (Soutar et al, 1975). The most effective fragment contained the majorityof the C-terminus of apo A-I. This region contains the majority of the -helix. Residues 99 to 230 specify6 repeating sequences of 22 amino acids each (22mer). The secondary structure of each repeat in thisregion is cx -helix with opposing poiar and nonpolar faces, ie an amphipathic helix (Segrest et al, 1974).Eleven amino acid residues complete three turns in an -helix and represents the monomeric unit (Fitchet al, 1984). The 22mer repeat structures found in the apolipoproteins are derived from this monomerunit.A wheel diagram of the predicted structure of a single 22mer (see Figure 7) indicates thatnegatively charged residues occupy one face of the protein and uncharged residues define an opposingface. The midline of this structure is occupied by predominantly positively charged residues. Functionally,these elements of the apo A-I sequence are believed to interact:(i) with the hydrophobic lipoprotein core via the nonpolar helical face;(ii) with the aqueous environment via the negatively charged polar face;(iii) with phospholipid headgroups via the positively charged helical midline residues (Sparrow andGotto, 1982). Proton magnetic resonance indicates that binding of apo A-I to lipoproteins and analogousstructures involves intercalation of -helical regions with the phospholipid monolayer (Brouillette et al,1982).20Figure 7. Helical wheel projection of the consensus sequence for the 22 residue repeat structure ofhuman apo A-I. (Modified from Segrest et cii, 1992).The consensus sequence of the repeating unit (see Figure 7) is not totally conserved among all22mers. Compared to the amino acids at most positions of the 22mer, the 13th residue of the consensussequence is least well conserved. This residue might, therefore, have additional functions in specific areasof the apo A-I molecule.The importance of the amphipathic helix in apo A-I function is underscored by studies ofsynthetic peptides containing this motif. Studies of peptides synthesized with the native apo A-I sequence(Sparrow et cii, 1980) have indicated that residues 148-185 were involved in LCAT activation (achieving20% of native apo A-I) and that residues 164-185 are involved in lipid binding. Residues 227-243 of theA-I sequence did not bind phospholipid and that polypeptides of multiple length spaiming residue 197 tothe carboxy terminus bound phospholipid but did not activate LCAT. Fukushima et at (1980) showedthat residues 121-164, which form two lipid binding domains, activated LAT up to 30% of the nativeprotein. Therefore, peptides spanning one helical repeat can activate LCAT but two repeat sequencesare more potent activators. Furthermore, while lipid binding is essential for activation by syntheticpeptides and native fragments, it is not sufficient for the full activation observed with the native protein.21These early conclusions must be viewed with caution, however, since recent evidence indicates that theDMPC substrates used in many studies do not accurately reflect activation of physiologic LCATsubstrates (Anantharamaiah et al, 1990).Synthetic peptides modelling the amphipathic helix but unrelated to apo A-I in sequence arealso capable of activating LCAT. The cofactor activity of these peptides correlates with their ability toform c -helix (Fukushima et at, 1980). Yokoyama et al (1980) described a 22 amino acid peptide withamphipathic potential that stimulated the phospholipase A2 activity of LCAT (to 50% of apo A-I) butwas substantially less effective at stimulating cholesterol esterification (18% of apo A-I). Another peptide(LAP-20) containing a single lipid-binding domain (Pownall et at, 1980) activated cholesterolesterification in DMPC substrates by up to 65% of the level obtained with native apo A-I. Both of thesestudies indicated that activation was due to interaction of the activator with the phospholipid substrate,rather than via an enzyme-activator interaction. Yokoyama suggested that in the absence of activatorphospholipid fatty acid side chains might be oriented toward the aqueous surface of the particle and thatreorientation of sn-2 groups into the hydrophobic region of the phospholipid monolayer might occurduring activation. Therefore, early synthetic peptide studies indicated that LCAT activation is mediatedby tx -helix interaction with substrate phospholipids, rather than with the enzyme.The separate contributions of hydrophobicily and -helix have been analysed by inserting aproline residue at various positions in the LAP-20 peptide (Ponsin et al, 1986). The introduction of thishelix-breaker” alters phospholipid binding and LCAT activation by the peptide without affecting theoverall hydrophobicity. The extent of binding and activation depended upon the site of substitution, withmost pronounced reduction observed when Pro was introduced near the middle of the sequence. Theauthors suggested, based on these observations, that helical segments of at least 15 uninterruptedresidues were required for lipid binding and LCAT activation.Segrest and colleagues have used synthetic peptides extensively to investigate the mechanism ofLDAT activation and have concluded that secondary structure is more important than amino acidsequence in LCAT activation. Synthetic peptides homologous to apo A-I sequence were no more potentLCAT activators than non-homologous amphipathic helical peptides. However, no synthetic peptidemodeling a single amphipathic helix (22iner) exceeded 30% of the activation of the isolated native22protein. Dimers of 22mer repeats approach the LAT activation properties of apo A-I more closely thanthe monomeric unit (Anantharamaiah et at, 1990). These studies also indicated that a single position inthe helix appeared to have an additional role. In dimer studies, Glu in the variable position 13 of eachmonomer unit was the most effective activator. In the apo A-I sequence, only amphipathic helical repeats2 and 3 (residues 66-121) contain Glu at position 13. They have proposed, therefore, that this region and,in particular, the Glu residues at positions 91/92 and/or 110/111 of apo A-I are involved in LCATactivation.Investigation of other specific residues in the A-I sequence has been somewhat limited. Jonas etat (1985) investigated the role of lysine residues in LCAT activation using chemical modificationtechniques. These authors concluded that while modification of lysine did reduce the ability to stimulatecholesterol esterification, the effect was evident only when modification altered the residue charge.However, since changes in the secondary structure of the apoprotein were also apparent, it wasimpossible to conclude that lysine, in itself, was involved in the activation process. Similar studies in ourlaboratory (R. McLeod and M. Bergseth, unpublished observations) have indicated that this type ofchemical modification reduces the ability of apo A-I to bind to DMPC vesicles. A recent report hassuggested that in vthv glycation of lysine residues also reduces its ability to activate LCAT (Gugliucciand Stahl, 1991). At present, however, chemical modification studies of the role of lysine residues in apoA-I activation must be regarded as inconclusive.In vitro mutagenesis techniques have recently been used to assess the role of the amphipathichelix in LCAT activation. Bruhn and Stoffel (1991) deleted two adjacent helices (d41, residues 146 to186) from apo A-I and showed that LCAT activation by the mutant protein was reduced by only 10%.They indicated, however, that this alteration might, in addition, destabilize the protein structure in vivo,since protein expression was reduced by40% in their E. coti system.1.6.4. Apo A-I Secondary Structure: Epitope MappingThe secondary and tertiary structures of proteins are critical determinants of their ultimatefunction. Several investigators have employed monoclonal antibodies to define structural epitopes of apoA-I. Continuous epitopes, defined as those that contain linear segments of primary sequence, are rare in23apo A-I and when present are near putative -turns in the sequence (Marcel et at, 1991). The N-terminus of apo A-I appears to exist in a compact conformation. Many of the epitope-specificmonoclonal antibodies which have been generated in a number of laboratories are directed towards themiddle of the molecule within a single -helix (residues 98 to 121). Since mobile, accessible domains areknown to be particularly antigenic, this region of apo A-I has been suggested to represent a mobile,“hinge” domain (Marcel et at, 1991; Cheung et at, 1987; Figure 8).Curtiss and colleagues (Curtiss and Edgington, 1985; Curtiss and Smith, 1988) recentlydeveloped monoclonal antibodies to apo A-I, three of which blocked activation of LCAT (Banka et at,1991). The extent of inhibition was related to the ability of the antibody to bind to a specific apo A-Iepitope. Other antibodies in this series bound avidly to distant epitopes but had no affect on activation.The results suggested that a specific region of apo A-I was involved in LAT activation and they wereable to identify residues 95-121 as a critical element for cofactor activity using peptide competitionstudies. These studies provided further evidence for the existence of a distinct mobile hinge domain inthe third helix of apo A-I (residues 99-121) and for involvement of this region in LCAT activation.Figure 8. Schematic diagram of the hinged domain hypothesis for the interaction of apo A-I with HDL.The hypothesis suggests that amphipathic helices of apo A-I may be in an exposed or buried conformation depending on the size of the HDL. From Cheung et at (1987), with permission of the publisher.TANDEM AMPt4IPAThIC HELIXESOF APOLIPOPROTEIN A-I AUPHI4’AThICHtI.IX-PHOSPHOL,pu,MONOLAYER SHELLTANDEM AMPHIPAThIC 45jj5ES HEUX-PHOSPHOLIPIDOF APOLIPOPROTEIN A-I MONOLAYSR SHELL :5 NEUTRAL 2HINGED LIPIDDOMAIN COREORESTANDEM AMPHIPATWC HELIXESOF APOLIPOPROTEIN A-ISHELL24Anantharamaiah et at (1990) have used peptide dimers to activate LCAT and have suggestedthat Glu residues within the hinge region might stabilize this domain via charge-charge interaction. Theyproposed that Glu pairs at positions 91/92 and at 110/111 might provide hinge stability. However, directassessment of this hypothesis in the intact protein has not been performed. Preliminary studies (Bruhnand Stoffel, 1991) have indicated that G1u1 1 has little influence on LCAT activation, since substitution ofthis residue reduced activation of the enzyme by only 15%. If ion pairing is involved in hinge stability,one might expect that Lys residues in the same region might play a role. The effect of altering Lysresidues in the hinge domain region has not yet been reported.Using an improved modelling technique, Segrest and colleagues (1992) have identified somedifferences among the amphipathic helices of apo A-I. The majority of the helices are of class A and arepredicted to penetrate the phospholipid layer to a greater depth than class Y helices. Class Y helices aremuch more prevalent in apo A-IV, and appear to penetrate the lipoprotein surface to a lesser extentthan class A helices. Only two regions of apo A-I contain helices of class Y, the C-terminus and the N-terminal side of the mobile hinge domain (see Figure 9). Therefore, these structural elements maycontribute to the unique functional properties of apo A-I.Figure 9. Proposed supersecondary structure of human apo A-I. The position of amphipathic helices(Segrest et at, 1992) have been superimposed upon the structural model derived from epitope mapping(Marcel et at, 1991). Rectangles represent amphipathic helices of 11 (small) or 22 (large) amino acidlength. Hatched regions are class Y, open regions are class A amphipathic helix. P indicates the Sei201,site of covalent phosphorylation. Arrow indicates Lys residue at position 107. The residues marking theends of helix repeats are indicated by residue number from amino (NH2) to carboxy terminus (COOH)of the mature protein. Triangles indicate the positions of proline residues. The putative hinge domain isbetween residues 99-143.COOH220251.7. Naturally Occuring Structural Variants of Human Apo A-IThe importance of apo A-I in cholesterol homeostasis has been demonstrated in pathologicstates in which the protein is absent or near absent. The combined deficiency of apo A-I and apo C-Ill(Karathanasis et at, 1984) is characterized by profound reductions in HDL and severe atherosclerosis.The defect has been elucidated at the molecular level (Karathanasis e a!, 1987) and is the result ofinversion of a 6 Kbp segment containing portions of both the apo A-I and apo C-Ill genes and intergenicsequence. Since these two genes are transcribed in opposite orientations, both produce abnormal mRNA.Neither apo A-I nor apo C-Ill is found in the plasma of subjects with this mutation.Tangier disease is an HDL deficiency state originally thought to be due to an apo A-I structuraldefect (Kay et at, 1982). Subsequent studies have shown that the apo A-I gene in Tangier diseasespecifies the normal protein sequence (Rees et at, 1984; Makrides et at, 1988), but the marked reductionin plasma apo A-I (1-5% of normal mass) in Tangier disease is associated with a larger than normalproportion ( 50%) of proapo A-I (Zaniis et al, 1982). The conversion of proapo A-I to the matureplasma protein occurs at the normal rate in Tangier plasma (Bojanovski et at, 1984) but the HDLparticle appears to be catabolized rapidly. Perhaps the most striking observation, yet to be explained, isthe persistence of the proapoprotein form in Tangier plasma despite the absence of the mature form.This might reflect the presence of separate apo A-I pools in vivo, with the proapo A-I pool largelyspared by the defect in this disease. The cause of Tangier disease remains undefined despite severaldecades of study.Structural variants of apo A-I have been identified in which the deletion, insertion, transpositionor substitution of a small segment of DNA results in the synthesis of a protein with altered structure.Several of these variants have been identified opportunistically or through population studies (Menzel etal, 1982; Menzel et al, 1984; Utermairn et at, 1982; von Eckardstein et a!, 1989). In many cases both theamino acid alterations and the nucleotide changes have been defined. The identification of these variantswas based on the isoelectric focussing pattern of the apo A-I, and therefore all variants involve deletionor substitution of a charged amino acid. The variants identified and characterized to date are shown inTable III. Mutations of Lys at position 107 and of Pro residues throughout the protein appear to be themost frequent.26Pro3—ArgPro3—’ HisPro4—’ ArgArgtO_ LeuGI96—ArgA1a37—÷ ThrGin84—’ StopAsp89—’ GluAsp103’ AsnLys107_, oLys107—’ MetGiu1 1O.._ LysGiut—* LysG1u1—* GlyPro143_, ArgdelGin146, P.ig16°G1u147—* ValAlat58, GiuPro165, ArgG1u169 GinArg173 CysArg177 HisGlu198—’ LysA-I202-’FSA-I FukuolcaA-I Norwayvon Eckardstein etal, 1989Menzel et ai, 1984Menzeletal, 1984; vonEckardstein etai, 1989Ladias et ai, 1990Nichols et ai, 1988Matsunagaetal, 1991Matsunagaetal, 1991von Eckardstein etai, 1990Menzel et al, 1984Utermann etal, 1982;Menzel et al, 1982; RaIl etal, 1984; von Eckardsteinet al, 1990von Eckardstein etai, 1990Takadaetal, 1991Mahleyetal, 1984; RaHetal 1986von Eckardstein etal, 1990Utermannetai, 1984Deeb et al, 1991von Eckardstein et ai, 1990Mahleyetal, 1984von Eckardstein etal, 1989von Eckardstein et ai, 1990Weisgraberetal, 1983Jabs et al, 1986Mahieyetal, 1984; vonEckardstein etal, 1990;Strobi et al, 1988Funke et al, 1991Mahieyetai, 1984Familial amyloidicpoiyneuropathy10% of Japanese controlsHDL deficiency, A-Ideficiency, atherosclerosisTable III. Variant forms of apolipoprotein A-I.DEFECT FAMILIES NAME REFERENCE NOTESAFFECTEDA-I Mueflster—3CA—I MueIlster—3BA—I BaltimoreA-I iowaA—I Muenster—3AA—I Marburçj’ A—I Muenster—2121Many1172121144A—I ciessenA-I SeattleA-I Mi 400Asp213 GlyDominant decreased A-IAbnormal HDLcatabolismHypoalphaFish-eye Disease variant,homozygous, low HDL,A-I with CysThe frequency of genetic variation in the human apo A-I protein (0.8/1 amino acids) is higher(von Eckardstein et cii, 1990) than in hemoglobin (0.2 per 1 residues) which suggests that variants inapo A-I protein sequence are less subject to negative selection. Within the N-terminus and the z -helicalregions of the apo A-I coding region, more charge-shift variants of the protein are found than would beexpected from random nucleotide substitution. The N-terminal mutants may arise due to variability atthe junction of exon 2 and intron 2. Inter-species comparison of apo A-I sequences has indicated that theamino acid sequence of the -helix region is poorly conserved across species and this region is also a27frequent site of human variation. However, few of the variants appear to alter substantially the secondarystructure in this region. Mutations which are predicted to alter the orientation of the amphipathic helix,for example those altering Lys107 and Pro165, appear to alter function also. Residues 66-98 are highlyconserved among species and no genetic variation in this region has been described in man. Despite theconsistency of the inter- and intra-species distribution of variation in the apo A-I sequence, thetechniques used for detection of mutants may have under-represented some regions of the protein wherethe alteration conserved charge or when the alteration is sufficiently deleterious to cause absolute apo AI deficiency.Apo A-I structural variants have rarely been associated with disorders of lipoprotein metabolism.One exception is the apo AJMiiano variant, in which affected subjects have reduced HDL levels (10-20%of normal). These individuals have no evidence of accelerated atherosclerosis. The abnormal proteinforms apo A-I homodimers and apo A-I/A-Il heterodimers via the variant Cys173 residue. Wild type apoA-I does not contain cysteine and apo AlMIIano is reported to have reduced LCAT activating ability(Mahley et at, 1984, Jonas et at, 1991).A second apo A-I variant containing Cys residues has been described (Funke et at, 1991) in asubject without personal or family history of CAD. The mutation affected plasma LCAT activity, whichwas reduced to 1/3 of normal. The defective apo A-I was distributed in LDL and d> 1.21 g/ml fractionsfollowing ultracentrifugation, indicating abnormal lipid binding properties. The defect was shown to be aframeshift mutation (apo A-1202-FS) caused by loss of C from codon 202 leading to prematuretermination after 229 amino acids. Codons 203-229 are altered from the wild-type sequence and shift thep1 by +7 charge units from the normal protein. The frameshift also results in the introduction of severalCys codons. The defect was detected in both homozygous and heterozygous individuals and affectedHDL and apo A-I levels in both. As with the apo Aliano the presence of Cys residues causedformation of stable homodimers and apo A-Il heterodimers. Since the region of apo A-I affected by themutation is highly conserved among species (von Eckardstein et at, 1990), the authors have suggestedthat this region may be linked to LCAT activation.Apo Al5eattie (Deeb et at, 1991) is unique in that reductions of HDL and apo A-I are dominantin the heterozygous state. It is the only apo A-I variant involving a large coding region deletion. The loss28of 45 bp from exon 4 removes 15 residues of a single -helix motif (Gin146 to Arg60). It was suggestedthat the mutant protein alters HDL structure and causes hypercatabolism. The changes also appear toreduce the affinity of LCAT and other apolipoproteins for the HDL.In all other subjects with variant forms of apo A-I, modest reductions in the level of HDL arethe most severe consequence of the abnormal protein. However, as previously noted, the overall impactof the mutation may not be expressed, since most of the variants have been identified only in theheterozygous form. One notable exception is the Norwegian variant, Glu136- Lys, which has beenidentified in the homozygous state (Mahley et al, 1984). The levels of LDL are markedly decreased inthis condition but HDL-C is near normal. The full influence of the remaining mutations on lipoproteinmetabolism awaits the identification and subsequent characterization of homozygous individuals, or thefunctional characterization of mutants generated in vitro.Functional studies of the purified mutant apo A-I proteins have been limited due to thedifficulties encountered isolating large quantities of pure mutant protein from plasma of heterozygoussubjects. Only two of the variant apo A-I have consistently abnormal ability to activate LCAT. Deletionof Lys107 reduced the activation by about 50% compared to native protein (Rail et al, 1984), an effectascribed to reduced lipid association properties (Ponsin et al, 1985). Models of secondary structurepredict that the deletion turns the polar face of the helix by 903, which may account for its reducedfunction. Apo AI,3iessen (Pro143-’A g) also demonstrated reduced ability to activate LCAT (Rall et al,1983). However, a more detailed analysis of Pro’43-.Arg, comparing the activation of the normal andvariant protein (isolated from the same plasma sample of heterozygous subjects), was inconclusive (Jonaset al, 1991). In this comprehensive study, only Lys107-’O function was significantly reduced from normal,forming HDL-like recombinants with altered structure which resisted normal transformation to largerspecies on incubation with LDL. LCAT activation by the variant was also reduced compared to normalbut the large variation among different normal preparations reduced the impact of the findings.Apo AJFukuoka (G1u110-.Lys) was shown to activate LAT normally (Takada et al, 1990). Thestructural changes predicted for this mutation are minimal. Despite the change of 2 charge units, theamino acid substitution is not predicted to alter the orientation of the polar and nonpolar faces of theamphipathic helix. The predicted structure is therefore consistent with the functional observations and29tends to support a more important functional role for Lys107. The abnormal apo AINorway (Glu36—Lys)has been identified in homozygous individuals, in which LCAT activity was normal (Rail et al, 1986)suggesting that this Glu residue is also not involved in the activation mechanism.While observations of apo A-I variants have provided valuable insight into apo A-I function,their scarcity has limited the amount of information acquired. The application of molecular genetictechniques to apolipoproteins in recent years has provided the basis on which structure-functioncorrelates can be more systematically investigated.1.8. Expression of Recombinant Apo A-IExpression of apo A-I in vitro has been the aim of many laboratories since the first cDNAs werecloned in the early 1980s. Zannis et a! (1983), using rabbit reticulocytes, demonstrated that the primarytranslation product of the apo A-I cDNA is preproapo A-I and that, in the presence of dog pancreaticmicrosomal membranes, preproapo A-I was processed to proapo A-I. Lamon-Fava et a! (1987) havesuggested that the ability to express apo A-I may be sufficient to permit a cell to produce lipoprotein andtested the hypothesis in 3T3 cells transfected with the apo A-I cDNA. They showed that these cellssecreted proapo A-I, some of which was isolated in HDL, with properties similar to HepG2 cellsecretions. Unlike HepG2 cells, the quantities of HDL produced were extremely low unless exogenouslipid was provided in the medium. In addition, HDL associated apo A-I was a minor component of thetotal product. The majority (75%) of the apo A-I produced in the absence of lipid was found in thelipoprotein-free (d> 1.25 g/ml) fraction. Similarly, C127 (Rhogani and Zannis, 1988) and L6E9 myogeniccells (Ruiz-Opazo and Zannis, 1988) transfected with the human apo A-I gene also secreted proapo A-I.Direct secretion of the mature protein was established in cells after deletion of the propeptide codingsequence from the gene (Rhogani and Zannis, 1988). This alteration had no apparent affect on the rateof secretion or on the fidelity of signal peptide processing. Mallory et a! (1987) described the mostpractical and efficient apo A-I expression in eucaryotic cells. They established CHO-Ki cell clones whichexpressed the human protein from a genomic DNA sequence under control of the humanmetallothionein-Il promoter. Many of the lines expressed as much as 1 mg of apo A-I per litre in serum-free culture. One clone produced as much as 10 mg/L. In addition, the apo A-I was fully processed to30the mature plasma isoform in all cases. They suggested that the apo A-I propeptidase was synthesizedand secreted from CHO-K1 cells and that the proteolysis occured extracellularly.For large scale preparation of recombinant proteins, procaryotic expression systems, in general,produce larger quantities of protein with much less effort and expense. However, given that these cellsdo not possess the ability for post-translational processing of mammalian gene products, these expressionsystems are applicable only to those mammalian proteins which do not require these modifications. ApoA-I would appear to be a candidate for procaryotic expression, since it is not glycosylated and it does notcontain cysteine residues, which can also limit expression levels in these systems. Attempts have beenmade to produce apo A-I as a recombinant in E. coli, with extremely limited success. Most authors havereported that the protein is rapidly degraded. A /9 -galactosidase fusion protein (Lorenzetti et at, 1986)has been described, but its half-life was only 45mins. The degradation appeared to affect specifically theapo A-I portion of the fusion product. No apo A-I could be detected without linkage to the bacterialproduct. Monaco et at (1987) described a Protein-A/apo A-I fusion protein which exhibited biologicalfunction. This product bound to mammalian cells and was selectively displaced by HDL. Large quantitescould be isolated from cell lysates (40-50 mg/L) but the protein formed pentamer aggregates whichproved difficult to disrupt. Isacchi et al (1989) have carefully investigated the role of the N-terminus ofapo A-I in its degradation by E. coli. They found that by inserting the codons for the first 8 amino acidsof mature apo A-I into the 5’ end of bacterial proteins, the quantity of protein expressed was markedlydiminished. Alterations were then made in the nucleotide sequence which retained the amino acidsequence. Some of these changes increased the level of expression of the bacterial protein and alsoimproved the yield of unfused apo A-I when incorporated into the original sequence. They also foundthat the propeptide segment of apo A-I was important for optimal expression in E. coli. In all of theirstudies, the level of apo A-I expression was related to the steady state level of the respective mRNA.Only one report has described the assessment of LCAT activation by apo A-I produced inbacteria. Bruhn and Stoffel (1991) expressed proapo A-I in E. coli by remoing the eucaryotic signalpeptide and attaching bacterial codons for Met-Gly onto the N-terminus of proapo A-I. The quantity ofapo A-I produced appeared to be substantial ( 5mg/L). LCAT activation by the recombinant proapo AI produced in this study was indistinguishable from the purified plasma protein.31Yeast expression of apo A-I has also been investigated by at least one laboratory (Segrest,personal communication 1989). This system has the advantages of procaryotic systems with respect tohigh level expression and low cost. In addition, since yeast are primitive eucaryotic cells, they have thecapacity to modify mammalian proteins following translation. However, there appears to be somequestion as to whether the recognition systems in yeast are the same as in higher eucaryotic cells.Regardless of these theoretical considerations, no reports of yeast expression of apo A-I have yetappeared. It appears that yeast may be incapable of apo A-I secretion and, like E. coli, degradeapolipoproteins rapidly.1.9. Scope of Thesis: Specific AimsIt is clear that while a great deal is known of the structure, metabolism and role of apo A-I inlipoprotein metabolism, surprisingly little information is available regarding the molecular mechanismsinvolved in its functions. Our knowledge at present is largely confined to LCAT activation properties andcontroversial studies of receptor interaction. The evidence has been derived mainly from studies of thepurified native protein and its fragments, or from the study of model peptides containing the amphipathichelix. Chemical modification of the native protein has provided only limited information due to theinability of this approach to affect a specific subset of amino acid residues and the potentialconformational changes induced by the modifying agents. Phenotypically expressed inborn errors ofmetabolism have provided information on functional domains of many other proteins. However, thisapproach has been limited by the rarity of mutations of the apo A-I gene in the homozygous state. Thishas made purification of the mutant protein in sufficient quantity an arduous task.Thus, the purpose of this thesis was to utilize recombinant DNA methodology to introducespecific changes within the apo A-I cDNA and to express the mutant proteins in cell culture. Thestructure-function relationships of apo A-I could then be addressed by analysing their function in vitro.Specifically, I set out to:(a) Develop expression systems for the production of human apo A-I in eucaryotic cellculture and to validate the expression of the wild-type protein.(b) Delete the propeptide of apo A-I and determine the functional consequences on cellular32processing and secretion.(c) Produce the apo A-I Lys107-.O mutant in vitro and determine its LCAT activatingproperties.(d) Modify the apo A-I cDNA by deletion of residues encompassing a complete c -helix inthe lipid binding domain and determine the effect of this deletion on LCAT activation.332. MATERIALS AND METHODS2.1. MaterialsThe cDNA encoding the apo A-I precursor (Seilhamer et al, 1984) was kindly provided byDr. Beatriz Ley-Wilson of the Gladstone Foundation Laboratories, San Francisco, Ca. Restriction andmodification enzymes for manipulation of DNA sequences were purchased from Bethesda ResearchLaboratories (BRL, Burlington, Ont.), Pharmacia-LKB Biotechnology (Baie dUrfe, Que.), or BoehringerMannheim Corporation (BMC, Laval, Que.). Enzymes and reagents for DNA sequencing were fromUnited States Biochemical (Cleveland, Ohio). Bacterial strains DH5a and DH5a -F’ were also purchasedfrom BRL, while the dut ung strain CJ236 was supplied by Biorad Laboratories (Mississauga, Ont.).Geneclean, for purification of double stranded DNA fragments, is produced by BiolOl (La Jolla, Ca). a -r5Sl-dATP,r5Sj-methionine, y-r2P]-ATP and rH]-cholesterol were obtained from Dupont Canada(Mississauga, Ont.) or Amersham Canada Ltd (Oakville, Ont.).Oligonucleotides were prepared in the OligonUcleotide Synthesis Laboratory, Department ofBiochemistry, UBC and purified by denaturing polyacrylarnide gel electrophoresis and reverse-phasechromatography (Sep-Pak C, Waters) as described (Atkinson and Smith, 1984). The expression vectorpCMV5 (Thomsen, 1984; Andersson, 1988) was a gift from Dr. David Russell, Department of MolecularGenetics, University of Texas Southwestern Medical Center, Dallas, Texas. The pNUT vector (Funk,1990; Palmiter, 1987) and recipient BHK cells were provided by Dr. Ross MacGillivray, Department ofBiochemistry, UBC. pSPT19 was purchased from Pharmacia. COS-1, an SV4O-transformed AfricanGreen monkey cell line (Gluzman, 1981; ATCC CRL-1650) and chinese hamster ovary cells (strainCHO-Ki, ATCC CCL61) were obtained from the American Type Culture Collection (Rockville, Md).Tissue culture reagents including fetal bovine serum (FBS) were supplied by Gibco-BRL (Mississauga,Ont.). Methotrexate (Cyanamid Canada, Inc., Montreal, Que,) for selection medium was obtained insterile saline from the University Hospital Pharmacy.Reagents for in vitro transcription and translation (SP6 polymerase, rabbit reticulocyte lysateand canine pancreatic microsomes) were purchased from Promega Corp (Madison, Wi). All fluorescenceand gold conjugates used for immunocytochemical studies were from Sigma Chemical Co (St. Louis,Mo). Monoclonal mouse anti-human apo A-I (6B8) was a gift of Drs. Ross Milne and Yves Marcel,34Clinical Research Institute, Montreal, Quebec. Polyclonal sheep antibodies to human apo A-I wereobtained from BMC. Recombinant Protein G reagents were produced by Genex Corp (Gaithersburg,Md).Electrophoresis grade reagents for gel elctrophoresis were obtained from Biorad or fromBRL. All other chemicals were of reagent grade or better and were purchased from Sigma or from BDHInc (Vancouver, BC).2.2. Growth and Transformation of E. ColiE. coli strains DH5ft and DH5x -F were maintained in LB (lOg/L tryptone, 5g/L yeastextract, lOg/L NaC1) and YT (8g/L tryptone, 5g/L yeast extract, 5g/L NaC1), respectively. Strain CJ236was maintained on agar plates containing minimal medium (Maniatis, 1982) and grown in YT broth. TheF-pillus was maintained by including chioramphenicol (3g/ml) as recommended by the supplier.Colonies maintained on the appropriate agar were viable for 2-4 weeks. Frozen bacterial stocks wereprepared in 20% glycerol from late log-phase broth cultures and were stored at -70 C.Competent E. coil were prepared from fresh lOOml cultures at mid-log phase (O.D.550 = 0.4-0.6) by the calcium chloride method (Maniatis et at, 1982). Bacteria were cooled on ice, pelleted by lowspeed centrifugation (5000 x g, 5mins) and gently resuspended in 25m1s of cold 10mM Tris-HC1 pH 8.0,50mM CaCl2 in an ice bath. After 30mins the cells were again pelleted by low speed centrifugation andresuspended in l0mls of ice-cold Tris/CaC12.The competent cells were used immediately or were mixedwith glycerol (to 20%, v/v), flash frozen in liquid nitrogen and placed at -70 C in 0.3m1 aliquots.Transformation was accomplished by gently mixing competent cells (0.3mls) with 5-4Ong ofplasmid DNA in an ice bath. After 30mins, the mixture was agitated briefly and heat shocked at 42 C for3mins. Transformants were diluted to 1.Oml with LB and incubated at 3 C for 1 hour with gentleagitation. Mixtures were plated onto LB-agar plates containing 10( g/ml ampicillin and incubatedinverted at 37 C for 16-24 hours. All expression plasmids used in these studies contained the -lactamasegene allowing growth of transformants in the presence of ampicillin.Recombinants in M13 were identified by blue-white color selection (Messing, 1983).Transformed DH5 -F were mixed with fresh DH5 -F’ overnight culture (2 1) and diluted to 2.5mls35with YT top agar (6g/L agar, 8g/L tryptone, 5g/L yeast extract, 5g/L NaC1) containing 3g/ml X-GAL(5-bromo-4,4-chloro-3-indoyl-3 -D-galactoside) and 0.7mM IPTG (isopropyl.9 -D-thiogalactopyranoside).This mixture was prepared at 45) C and was plated immediately onto YT agar. Once solidified, the plateswere inverted at 3? C for 12-16 hours at which time plaques were visible. Recombinant phage give riseto clear plaques on the bacterial lawn whereas those without DNA insertions are blue.2.3. Purification of DNA2.3.1. Small Scale Plasmid PreparationTwo ml aliquots of LB broth (containing antibiotic) were innoculated with a single bacterialcolony and were incubated at 3? C for 12-16 hours with vigorous agitation (250-280 rpm). Bacteria werepelleted by microcentrifugation (12,000 x g, 1mm) and resuspended in 10 1 of 25mM Tris-HCI pH 8.0,50mM glucose, 10mM EDTA. After 5mm at room temperature the cells were lysed with 20 1 of 0.2NNaOH, 1% SDS and placed on ice. After Smins the suspension was neutralized with 150i 1 of SMKAcetate pH 4.8 on ice. Cellular debris was removed by microcentrifugation (5mm) and the supernatant40 1) was extracted with an equal volume of phenol:chloroform (1:1, v:v). After brief centrifugationto separate phases, the upper aqueous layer was recovered and plasmid precipitated as the potassium saltwith 2 volumes of cold ethanol. The pellet after microcentrifugation was washed with 70% ethanol toremove coprecipitating salts. Finally, the pellet was dried in vacuo and dissolved in 5 1 of TE (10mMTris-HCI pH 8.0, 0.1 mM EDTA) containing 2g/ml DNase-free pancreatic ribonuclease.2.3.2. Large Scale Plasmid PreparationPlasmid preparations of suitable quality for eucaryotic cell transfection were purifiedaccording to a protocol published by Promega Biotec (Technical Bulletin 009). An aliquot of freshovernight culture (from a small scale plasmid preparation) was used to innoculate 10-l5mls of selectivebroth. Once an O.D.550 of approximately 0.6 was attained (4-6 hours incubation at 3? C), the entireculture was added to 2SOmls of warm selective broth in a 1L flask. This mixture was incubated overnightwith vigorous agitation. Cells were recovered by cold centrifugation (2800 x g, l5mins) and wereresuspended in 6mls of 25mM Tris-HCI pH 8.0, 10mM EDTA, 15% sucrose, 2mg/ml lysozyme on ice.36After 20mins, the cells were lysed with l2mls of 0.2N NaOH, 1% SDS and mixed by gentle inversion.The lysate was neutralized with 7.5mls cold 5M KAcetate and placed on ice for a further 20mins. Theviscous suspension was iransfered to silanized glass tubes (Corex) and cellular debris was removed bycentrifugation (25,000 x g, l5mins). The clear supernatant was recovered arid RNA in was digested with5Qu g ribonuclease for 20mins at 37 C. The DNA solution was extracted twice with equal volumes ofphenol:chloroform and total nucleic acid was precipitated from the upper phase with 2 volumes of 95%ethanol at -70 C. The pellet was recovered by centrifugation and dissolved in 1.6 mls water. 0.4mls of 4MNaC1 and 2mls of 13% polyethyleneglycol (PEG-8000) were added and plasmid precipitated on ice for atleast one hour. The pellet was recovered by centrifugation (12,000 x g, lOmins) and washed with 70%ethanol. The final product was dissolved in 20C I of TE and quantitated by ultraviolet absorbance(Maniatis, 1982).2.3.3. Preparation of M13 Bacteriophage DNA2.3.3.1. Preparation of Uracil-Conlaining ssDNA from M13Template ssDNA for mutagenesis (Kunkel, 1985) was prepared by infection of E. coli strainCJ236 (dut lung-; Mutagene, Biorad Laboratories) with M13 phage containing the apo A-I cDNA.Briefly, J236 were grown in 2 x YT to O.D.550 = 0.3 and infected with 2 x 1 p.f.u. of M13 phage (togive an M.O.I. = 0.2). After 6 hours growth, the cells were removed by centrifugation (5000 x g, l5mins)and the clear supernatant transfered to a fresh tube. The uracil-enriched bacteriophage were titred onboth CJ236 and DH5 -F, and routinely showed iO -1 higher titre on J236. This indicated that uracilwas incorporated into the phage with high efficiency. Phage particles were precipitated on ice for at leastone hour following addition of 0.25 volume of 15% PEG in 2.5M NaC1. The precipitate was recovered bycentrifugation and resolubilized in TE for 1 hour on ice. Insoluble debris was removed by centrifugationand the clear supernatant was extracted twice with phenol and once with phenol:chloroform. DNA wasprecipitated on ice from the upper phase by addition of 0.1 volume of 3M NaAcetate pH 5.2 and 2volumes ethanol. After one hour, the DNA was pelleted by centrifugation, washed with 70% ethanol anddried in air. The final residue was dissolved in 20Ci I of TE and quantitated by UV absorbance.372.3.3.2. Preparation of M13 Phage DNA for SequencingYT broth cultures were innoculated with fresh DH5 -F overnight culture and a single M13plaque. After incubation for 6 hours at 37) C, bacteria were removed by microcentrifugation and retainedfor purification of RF DNA (see Small Scale Plasmid Preparation). Phage were precipitated from 1.3mIsof culture supernate by addition of 0.3mls of 20% PEG, 2.5M NaCI and incubation at room temperature.After 15mm, phage were recovered by microcentrifugation and resuspended in 20.i I of 20mM Tris-HC1pH 7.5, 20mM NaCI, 1mM EDTA. The mixture was extracted with phenol:chloroform, and ssDNA wasprecipitated overnight at -7 C from 0.3M NaAcetate solution with two volumes of ethanol. The pelletwas recovered by centrifugation and ethanol precipitation was repeated. The final DNA pellet waswashed with 70% ethanol, dried in vacuo and dissolved in 2C 1 of TE.2.4. Oligonucleotide-directed Mutagenesis2.4.1. In vthv MutagenesisDirected mutagenesis was performed using a single oligonucleotide primer by the method ofKunkel (1987) as described by the supplier (Mutagene, Biorad Laboratories). Briefly, 200pmoles ofmutagenic oligonucleotide were 5’-phosphorylated with T4 polynucleotide kinase (PNK) for one hour at37) C. Residual PNK was inactivated at 68 C (lOmins) and l0pmoles of the phosphorylated primer wasmixed with ljt g of uracil-containing template ssDNA in 20mM Tris-HCI pH 7.4, 2mM MgC12, 50mMNaC1. The mixture was heated to the denaturation temperature (see Table IV) and primer and templatearinealled by slow cooling to 30 C or less.Table IV. Apo A-I Mutagenic Oligonucleotide Primers and Their Properties. DENAT = denaturationtemperature used prior to annealling M13 phage DNA with mutagenic primer. HYB temperature usedduring filter hybridization to screen putative mutant phage preparations.Mutagenic primers were extended on the uracil template with T4 DNA polymerase in thepresence of T4 DNA ligase in 23mM Tris-HCI pH 7.4, 5mM MgC12, 1.5mM DTT, 0.75mM ATP andDELETION PRIMER SEQUENCE DENAT HYBdPRO 5-ACGGGGAGCCAGGCTGATGAACCCCCCCAG-3 85> C 73> CdK107 5-TI’CCAGAAGTGGCAGGAG-3 7(PC N/DD1(dPro22°-Asn241) 5’-TCCGCCAAGGCCTGCFGACCCAGTGAGGCGCCC-3’ 100> C 75> C38400i M dNTPs. Reactions were assembled on ice and sequentially incubated for 5mins on ice, 5mins atroom temperature and finally for 2 hours at 37° C. Aliquots of the final reaction were used to transformcompetent DH5x -F, plated in YT top agar and placed at 37°C to allow plaques to form overnight.Control reactions were performed simultaneously including all reagents with the exception of themutagenic primer (unprimed). Experiments were evaluated according to plaque number in the primedand unprimed reactions. If the plaque ratio (primed: unprimed) was >5, phage were screened for thepresence of the mutation.2.4.2. Identification of Putative MutantsPlaques from mutagenesis experiments were recovered as agar plugs with a sterile pipet anddispersed in lml of TE. Fifty i 1 of the resulting suspension (including control wild-type phage) was usedto infect fresh DH5z -F’ overnight culture (2 1) and diluted to 2mls with YT. After 6 hour culture at370 C, phage and RF DNA were isolated as described above.The dPRO and Dl mutants were identified by selective hybridization of the mutagenicoligonucleotide to the bacteriophage DNA. Culture supernates containing the phage were applied tonylon membranes (Nytran) using a vaccuum manifold apparatus. The membrane was removed withforceps and placed in a plastic hybridization bag containing prehybridization solution: 6x SSC (saline-sodium citrate, lx SSC = 0.15M sodium chloride, 20mM sodium citrate, pH 7.0), 5x Denhardt’s reagent(lx Denhardt’s= 0.2 gIL Ficoll, 0.2 gIL polyvinylpyrrolidine, 0.2 gIL bovine serum albumin) and 0.1%sodium dodecylsulfate (SDS). After 2 hours at 550 C, this solution was replaced with hybridizationsolution (6x SSC, 5x Denhardt’s). Oligonucleotide probes were prepared by incorporation of t3 P1phosphate onto the 5’ end of the mutagenic oligonucleotide, using the T4 polynucleotide kinase reactionin the presence of [32 PIATP (Sambrook et al, 1989). The labelled probe was subsequently added to thehybridization bag and incubated for 6-16 hours at 55°C in a stationary water bath. Unboundoligonucleotide was removed by extensive washing in 6x SSC (2x lOmins each at room temperature andlx lOmins at 55° C) and an autoradiogram was exposed for 30-120 mins between enhancing screens. Thefilter was then washed in 6x SSC at increasing temperature (3-5°C increments) and an autoradiogramwas developed following each wash. The selective temperature was determined as that at which wild-type39phage no longer retained the labelled probe but where a strong signal was observed in slots of some ofthe putative mutant phage.Putative mutants for the Lys107 deletion were identified by restriction endonuclease analysisof the M13 RF DNA. The loss of the dK107 codon removed a single MboII site and altered therestriction fragment pattern generated by this enzyme.Putative mutants identified by either analysis were confirmed by DNA sequencing.2.4.3. DNA Sequence AiialysisThe sequence of M13 inserts were determined by enzymatic chain termination sequencing(Sanger, 1977) of the single stranded phage. Primers for sequencing (Table V) were selected to generatechain terminated fragments which could be readily separated on 6-8% polyacrylamide gels. Primers werespaced approximately 200 nucleotides apart so that useful sequence was obtained with shortelectrophoresis times. The enzyme used for sequencing was modified T7 polymerase (Sequenase, UnitedStates Biochemical). Reactions were performed according to the manufacturers recommendations.Briefly, 1 g of M13 phage DNA was mixed with 0.Spmol of sequencing primer in 40mM Tris-HCI pH7.5, 20mM MgC12, 50mM NaC1. The mixture was heated to 63 C for 2mins in a 200ml water bath andallowed to cool slowly to 2S C. The annealed primer was then simultaneously extended and labelled withSequenase in the presence of 1.M dGTP, dTTP, dCTP and Ci of-r5S]-dATP (Smins at roomtemperature). Chain termination was then achieved by removing 3. I of the labelling reaction mixtureinto 2. 1 of a prewarmed solution containing 8( 1 dNTPs and iM of one of the dideoxynucleotides:ddGTP (“ G” reaction), ddATP (“A” reaction), ddTTP (“T” reaction) or ddCTP (“C” reaction).Termination reactions were performed at 37°C for Smins and stopped by addition of Stop Solution (95%formamide, 20mM EDTA, 0.05% Bromophenol Blue, 0.05% xylene cyanol FF). To obtain sequenceinformation close to the primer, the dNTP concentration in the labelling reaction was reduced to 300nM.In some reactions, where GC base pairing was abundant, dITP was utilized in the reaction mixtureaccording to the manufacturers recommendations.Labelled products (2- 1) were separated on 6% or 8% polyacrylamide gels containing 1 xTBE (89mM Tris, 89mM boric acid, 2mM EDTA) and 8M urea at 32W constant power. At the40completion of the separation the gel was dried onto Whatman 3MM chromatography paper and exposedto autoradiographic film (Kodak XRP) for 16-96 hours.Table V. Apo A-I cDNA Sequencing PrimersPRIMER SEQUENCE Al cDNAUFP 5-GTAAAACGACGGCCAGT-3’URP 5-CAGGAAACAGCTATGAC-3Si 5-ATCGAGTGAAGGACCTGGCC-3’ 204-223S2 5-CCCAGGAGTTCfGGGATAAC-3’ 383-402S3 5-CACTGGGCGAGGAGATGCGC-3 606-625S4 5-CCGCGCTCGAGGACCFCCGC-3 804-8232.5. Construction of Expression Plasmids2.5.1. Isolation of cDNA FragmentsFragments of the original apo A-I cDNA were obtained by restriction enzyme digestion ofthe double stranded plasmid. Digestions were performed under the ionic conditions recommended by theenzyme supplier. The apo A-I cDNA and mutant cDNAs generated in M13 were cut from the RF DNAwith EcoRI. The resulting apo A-I cDNA fragment ( 1 Kbp) was separated from the larger M13 vectorDNA ( 7 Kbp) by agarose gel electrophoresis in 1 x TAE and gel containing the DNA was excised witha scalpel. The agarose was dissolved at 50 C in the presence of 65-75% saturated sodium iodide and theDNA was recovered by binding to Glassmilk (Geneclean, BiolOl) on ice. The suspension was pelleted bybrief microcentrifugation and washed three times with 50-fold volume excess of buffered ethanol solution(NEWwash BiolOl). After the last wash, all traces of the wash solution were removed and the DNAeluted into 5-1.i 1 of TE.2.5.2. Modification of Fragment EndsFor blunt-end ligations the purified cDNA fragments were treated with the Klenowfragment of DNA polymerase I to fill the 5’ protruding termini. Two units of Kienow polymerase weremixed with lg of DNA in 50mM Tris-FIC1 pH 7.2, 10mM MgSO4, 1QuM DTT, 5Qug/ml BSA, 125tM41dNTPs. The reaction was placed at 3? C for 30 mins, after which the DNA was purified on Geneclean.To reduce background recircularization of vectors during ligation reactions, the digestedvector was treated with calf intestinal phosphatase (CIP, Boehringer Mannheim). Linearized vector (1-2,u g) was brought to 5 1 in 50mM Tris-HC1 pH 9.0, 1mM MgC12,10i M ZnSO4, 1mM spermidine forthis reaction. For fragments with protruding 5’ termini, two successive incubations with CIP (0.01 unit)were performed (30mins each at 3? C). Fragments with blunt or recessed 5’ termini were incubated at3? C for l5mins and at 56 C for l5mins following each CIP addition. Reactions were terminated at 68 Cfor l5mins following addition of EDTA (to 1mM), NaC1 (to 100mM) and SDS (to 0.5%). The mixturewas extracted twice with phenol:chloroform and once with chloroform prior to fragment isolation onGeneclean.2.5.3. Ligation into Expression PlasmidsVector and insert concentrations were estimated by their ethidium bromide staining intensityon agarose electrophoresis gels and used to determine ligation requirements. For vector-insert pairs withoverlapping compatible ends (“ sticky’), the insert and vector were mixed at a molar ratio of 2:1. Forblunt end ligations the ratio was increased to 4:1. In general, 25-bOng of vector DNA was used perligation reaction. The final reaction volume was adjusted to 5p 1 in 50mM Tris-HC1 pH 7.6, 10mMMgC12, 1mM ATP, 1mM DTT, 5% PEG. T4 DNA ligase was added to 0.1 unit per p1 for sticky ends orto 1.0 unit/p 1 for blunt ends. All ligations were performed in a water bath at 12-15 C for 12-16 hoursand l-5p1 of the mixture was used to transform competent E. coil.In vitro transcripts were synthesized using pSPT19 (Pharmacia-LKB Bioteclmology). Thisvector contains promoters for the SP6 and T7 phage RNA polymerases in opposing orientations flankingthe polylinker region. Linearization of the plasmid outside of the polylinker region generates a templateDNA suitable for “run-off” transcription from either promoter using purified polymerase (Krieg andMelton, 1984). The resulting transcript is suitable for translation by rabbit reticulocyte lysate (Pelhamand Jackson, 1976), providing analytical quantities of protein from the cDNA. The apo A-I cDNA wasmost efficiently expressed in this system using the SP6 polymerase and cDNA insertion at the pSPT19Smal site. The appropriate orientation was established by restriction enzyme analysis.42Transient expression of apo A-I was studied with the eucaryotic expression vector pCMV5(Thomsen, 1984; Andersson, 1988). pCMV5 contains the promoter-regulatory region of the humancytomegalovirus (CMV) major intermediate early gene and the 3’ untranslated region of human growthhormone (hGH) which provides transcript termination and polyadenylation signals. This plasmid attainshigh copy number in SV4O transformed cells since the vector contains the SV4O origin of replication.Native and mutant apo A-I cDNA were ligated at the unique EcoRI restriction site, which placed thecDNA under the transcriptional control of the CMV promoter. Transformed DH were selected byampicillin resistance and the appropriate orientation for the CMV promoter established by restrictionanalysis.Stable cell lines expressing human apo A-I and its mutants were developed using theeucaryotic expression vector pNUT (Palmiter, 1987). This vector contains a mutant form of the gene fordihydrofolate reductase (DHFR) which allows for immediate selection of cells containing stablyintegrated plasmid DNA by their survival in high concentrations of methotrexate (Funk, 1990). Insertionat the Smal site of the vector placed the apo A-I cDNA under transcriptional control of the mousemetallothionein (mMT-I) promoter. Transformants in DH5 were identified on ampicillin plates and theappropriate orientation was established by restriction endonuclease analysis.2.6. In vitro TranscriptionPlasmid was linearized for “run-ofP’ transcription outside of the promoter-insert region bydigestion with HindIII and ethanol precipitation. All water used for transcription was deionized anddiethylpyrocarbonate (DEPC) treated prior to use. Transcription mixtures were assembled at roomtemperature and containing 2,u g of linearized pSPT19-AI plasmid. Each 100t 1 reaction contained 40mMTris-HC1 pH 7.5, 6mM MgC12,2mM spermidine, 10mM NaC1, 10mM DTT, 100 units RNAsin (placentalribonuclease inhibitor, Bethesda Research Laboratories) and 5OQu M each ribonucleoside triphosphate(rATP, rUTP, rCTP, rGTP). Transcription was initiated by addition of 20 units of purified SP6 RNApolymerase (Promega Corp) and incubation for 2 hours at 373 C. In some experiments transcripts werecapped during synthesis by including the cap analogue7mG(5’)ppp(5)G (50 M, Pharmacia LKBBiotechnology) and reducing the rGTP concentration to 50i M. Under these conditions 7mG was43preferentially incorporated at the 5 end of the in vitro transcript and the translation efficiency wasimproved (Alexander, 1987). Once the incubation was complete, total nucleic acid was isolated byphenol:chloroform extraction and ethanol precipitation from 0.3M sodium acetate solution. mRNA wasstored as the precipitate in 70% ethanol at -7O C. In some experiments template DNA was removedprior to phenol:chloroform extraction by digestion with DNase I (2 units, Pharmacia-LKB) for l5mins,permitting the quantitation of RNA by UV absorbance. This was not necessary, however, for efficienttranslation of the transcript.2.7. In vitro TranslalionRabbit reticulocyte lysate, prepared from phenyihydrazine treated New Zealand Whiterabbits (Promega Corp), was used to translate “run-ofP’ transcripts derived from the apo A-I cDNA.Immediately prior to the translation reactions, the RNA was recovered by centrifugation, dried in vacuoand solubilized in DEPC-treated water.Translation was initiated by addition of 0.5u g of in vitro transcript to 5( 1 lysate mixturecontaining 2Qu M mixed amino acids (without methionine) and 51’ Ci 1n S]methionine (Amersham CanadaLtd, Translation grade, SJ.204, 1.5mCi/1’ mol). Incubations were of 30mins duration at 3O C and wereterminated by freezing. Some incubations included canine pancreatic microsomes (Promega Corp),prepared by the method of Walter and Blobel (1983), at 0.5-1.0 g of microsomal protein per assay. Thispreparation contains the eucaryotic enzyme systems of the endoplasmic reticulum and provides the invitro capabilities for signal peptide hydrolysis and membrane translocation.2.8. Eucaryotic Cell CultureCOS-1 and BHK cells were maintained in T25 flasks in Dulbecco’s modified Eagle’smedium (DMEM) containing 5% fetal bovine serum (FBS) (Gibco-BRL). CHO-Ki cells were culturedin an equivolume mixture of DMEM and Ham’s F12 nutrient mixture (Gibco-BRL) containing 10%FBS. All FBS was mycoplasma and virus free as supplied and was heat inactivated at 56D C for 3Ominsbefore use as growth supplement. Cells were passaged as they reached confluence (every 3-4 days) bytrypsinization and 6-10 fold dilution in growth medium. All eucaryotic cells were maintained in a44humidified incubator with 5% carbon dioxide atmosphere. Media of stably transfected cells containedselection agents as indicated.2.9. Transient Transfection of COS-1 CellsTransient transfections were performed using the DEAE-dextran technique according topublished protocols (Kriegler, 1990). Briefly, COS-1 cells were seeded into 35mm dishes followingtrypsinization and 20-fold dilution in DMEM/5% FBS. Upon attaining approximately 80% confluence(24-36 hours) the medium was removed and the cells were washed twice with Transfection Buffer (25mM Tris-HC1, pH 7.4, 140 mM NaCI, 1 mM CaC12, 3 mM KC1, 0.5 mM MgC12,0.9 mM Na2HPO4).Plasmid DNA was diluted to 5 i g/ml in Transfection Buffer containing 500 j g/ml DEAE-dextran(M = 500,000; Pharmacia-LKB Biotechnology). 0.5m1 of this mixture was added to each monolayer andincubated for 30 minutes at 37) C. The DNA solution was then removed, replaced with DMEM/5% FBScontaining 80 JA M chioroquine, and incubated at 37) C for 3 hours. Chioroquine medium was removedand the cells were treated with DMEM/10% dimethylsulfoxide (DMSO) for 3 mins at roomtemperature. DMSO was removed with two washes of warm Transfection Buffer and growth mediumwas replaced. The cells were allowed to recover and initiate exogenous gene expression for 40-48 hoursprior to assessment of apo A-I expression.For endogenous labelling experiments, transfections were initiated in multiple 10 cm dishesand cells harvested 18-24 hours after the DMSO shock. These cells were then pooled, plated into 35mmexperimental dishes and allowed to adhere to the substratum for 20-24 hours.2.10. Isolation of Stable Eucaryotic Cells Expressing Apo A-I2.10.1. Calcium Phosphate TransfectionPlasmid DNA was introduced into recipient cells by the calcium phosphate coprecipitationmethod (Kriegler, 1990). Parent cells (in 10cm dishes) were grown to 90% confluence in the appropriategrowth medium prior to transfection. For transfection of BHK cells only pNUT-AI plasmids were used.However, since CHO-K1 cells were found to be relatively resistant to methotrexate, co-transfectionswere performed with Al-containing plasmid and pRc/CMV (Invitrogen, La Jolla, Ca) at a weight ratio45of 10:1. pRc/CMV provided the fl-lactamase gene for neomycin resistance. Plasmid DNA was diluted tolug/mi in HBS (5g/l HEPES, pH 7.05, 8g/l NaC1, 0.37g/l KC1, 0.lg/l Na2HPO4, lg/l glucose).Calcium phosphate was added to a concentration of 125mM to initiate coprecipitation of plasmid. Oneml of this mixture was added to lOmls of fresh growth medium and incubated overnight on the cellmonolayer. Transfection medium was subsequently removed and replaced with growth medium for a 24hour recovery period prior to initiating the selection procedure.2.10.2. Selection of Stably Transfected Cells2.10.2.1. BHK CellsBHK clones with integrated plasmid DNA sequences were selected during 10-14 daysculture with DMEM/5% FBS containing 500 jM methotrexate (Funk WD, 1990). Selection medium waschanged daily for the first 4 days and every 3-4 days, thereafter. Macroscopic colonies were visible at thattime and were then transferred with a 1 ml glass pipet to individual 20mm culture wells (Linbro, FlowLaboratories, Mississauga, Ont.). The colony was dispersed in 1 ml selecting medium to encouragemonolayer growth. Once the monolayer was established and confluent, selection medium was removedand replaced with serum-free maintenance medium (Optimem, Gibco-BRL). This was collected after 24-48 hours and clones secreting apo Al were identified (see below). Clones secreting maximal quantities ofAl were expanded to larger cultures in the presence of methotrexate. Pure cell populations secreting apoA-I were identified by immunofluorescence microscopy (see below) and were stored as frozen stocks in20% FBS, 10% DMSO in liquid nitrogen. One cell line derived from each of the transfectionexperiments was selected for further study. Studies of apo A-I processing and secretion were performedafter at least 48 hours culture in the absence of methotrexate.2.10.2.2. CHO-Ki CellsFollowing calcium phosphate transfection, the DNA solution was removed and replaced withgrowth medium for a 24 hour recovery period. The neomycin analogue Geneticin (G418, Gibco-BRL)was used for selection at 5OQu g/ml. This level was shown to reduce the survival of the parent CHO-Kicells to virtually zero after 14 days. Selection medium was replaced every 3-4 days and macroscopic46colonies were visible in transfected dishes after approximately 10 days. Colonies were harvested intomultiwell dishes and expanded under selection conditions as described above for BHK clones. However,since the level of apo A-I accumulation per cell was very low, the homogeneity of the cell populationscould not be assessed by immunofluorescence microscopy.2.10.3. Screening of Clones for apo A-I SecretionThe serum-free media from confluent monolayers were harvested and non-adherant cellsand debris were removed by centrifugation (12,000 x g, 10mm). Aliquots of media were applied tonitrocellulose membranes using a vaccuum manifold. Purified plasma apo A-I (McLeod et at, 1986) andmedium from cells transfected with empty pNUT vector were also applied as positive and negativecontrols, respectively. Umeacted binding sites on the nitrocellulose were blocked with 5% non-fat milkpowder in PBS (10 mM Tris-HC1, pH7.4, 150 mM NaCl). Apo A-I was identified by incubation withpolyclonal antibody to human apo A-I followed detection with protein G conjugated to horse-radishperoxidase (HRP). Membranes were washed extensively (4 x Smins) with 0.02% Tween-20 in PBSfollowing each incubation. The enzymatic color reaction used to visualize the immune complex contained0.05% 3,3-diaminobenzidine (DAB), 0.03% CoC12 and 0.006% hydrogen peroxide in PBS. The reactionwas stopped by extensive washing with tap water and relative quantities of apo A-I were determined byscanning densitometry of the dried membrane.2.10.4. Amino-Terminal Amino Acid Sequence AnalysisBHK cells expressing apo A-I were grown to 70% confluence in 20cm culture dishes inDMEM/10% FBS and subsequently depleted of serum by maintenance in Optimem (Gibco-BRLMississauga, Ont.) without FBS. After 24 hours the medium was removed to waste and replaced withfresh Optimem containing 50 M ZnSO4.This medium, which contained the secreted recombinant apoA-I, was collected after 24 hours and brought to 100 I.U./ml aprotinin, 0.1mM leupeptin and 1mMphenylmethylsulfonyl fluoride (PMSF). Phosphatidyicholine-cholesterol vesicles (4:1 molar ratio)(Batzriand Korn, 1973) were added and the mixture was incubated overnight at 373 C. During this period apo AI bound to the vesicles which were recovered by ultracentrifugation at d = 1.25 g/ml (40,000 rpm, 4847hours). The top 2mls were recovered from the ultracentrifuge tube by tube slicing, and proteins wereprecipitated from 15% (wlv) trichloroacetic acid. After centrifugation the protein pellet was delipidatedwith ethanol:ether (3:1) and then dissolved in SDS-PAGE Sample Buffer at 9fF C and resolved by SDSPAGE electrophoresis (see below). Proteins were blotted to Immobilon-P membranes (Millipore) andvisualized by staining with Coomassie Blue. The apo A-I band was cut from the membrane and subjectedto automated sequence analysis according to established techniques (Abersold et al, 1986; Matsudaira,1987).2.11. Metabolic Labelling Studies2.11.1. Determination of Synthesis RateProtein and apo A-I synthetic rates were measured by short-term incubation in the presenceof I S]methionine. Monolayer cultures (35mm Falcon dishes) were methionine depleted (DMEM minusmethionine, Gibco-BRL) for 20 mins at 37 C and subsequently labelled for 0-30 mills in the samemedium containing 6( Ci/mi [ Sjmethionine (700 Ci/mmol, New England Nuclear). At the timeindicated, cells were washed free of labelling medium and the monolayer was harvested in Cell LysisBuffer (50mM Tris-HC1 pH8.0, 62.5mM EDTA, 1% Nonidet P-40, 0.4% sodium deoxycholate, 1mMPMSF).2.11.2. Long Term Labelling StudiesCellular retention and secretion of apo A-I were measured by long term continuousincorporation of [ Simethionine. Near confluent monolayer cultures (35mm dishes, Falcon) wereequilibrated to lOQu M methionine by incubation for 20 mm in Optimem containing 10% FBS, lOQu Mzinc sulfate and 100 I.U./ml aprotinin. Equilibration medium was removed and [S]Methionine wasadded to a radiochemical concentration of 25-3Qu Ci/ml. After incubation for the indicated time, mediumwas recovered and cells harvested in Cell Lysis Buffer. Apo A-I secretion rates were determined byimmunoprecipitation from medium of these long term continuously labelled cultures. Immediately afterharvest, media and cells were cleared of intact cells or insoluble debris by 5 mm centrifugation at 12,000x g.482.11.3. Determination of Apo A-I Degradation and SecretionRates of intracellular degradation were measured utilizing an S]methionine pulse-chaseprotocol. Monolayers in 35mm dishes were washed free of growth medium and depleted of methionineby incubation in methionine-free DMEM (DMEM-Met, 20mm, 37’ C). The endogenous methionine poolwas then labelled with DMEM-Met containing 100-200 pCi/mir5Sjmethionine for 30 mins at 37J C.After removal of the labelling medium, the cells were rinsed with DMEM-Met, and the chase incubationinitiated in Optimem supplemented with 2mM methionine, 10 M zinc sulfate and 10% FBS. At theindicated time, medium and cells were recovered as described above and cleared of intact cells bycentrifugation for 5mm at 12,000 x g.2.12. Determination ofr5S]Methionine Incorporated into Protein and Apo A-IIncorporation of methionine into cellular protein was determined by precipitation ofradiolabelled cell lysate with 10% trichloroacetic acid (TCA) overnight on ice. The resulting pellet wasrecovered by centrifugation (12,000 x g, 10 mm) and washed twice with ice-cold 10% TCA and once withacetone. Pellets were then dried in air, dissolved in 2 x SDS sample buffer (0.1M Tris-HC1, pH 6.8, 2%SDS, 40% glycerol) and analysed by liquid scintillation spectrometry. Quantitations were performed ontriplicate dishes.Incorporation of methionine into apo A-I was determined following isolation from cell lysateby immunoprecipitation and SDS-PAGE as indicated below. l-2p g of purified apo A-I was added to theSjmethionine labelled immunoisolate for SDS-PAGE fractionation. Following electrophoresis the apoA-I band was located by Coomassie staining or by overlay of the autoradiogram on the correspondingdried gel. The gel fragment containing apo A-I was excised and dissolved by heating to 6ff C in 0.25mlsof 30% hydrogen peroxide. Radioactivity was the quantitated by liquid scintillation spectrometry in SmlsACS (Amersham Canada Ltd). Each quantitation was performed on triplicate 35mm dishes.2.13. Isolation of Apo A-I by ImmunoabsorptionApo A-I in media or cell lysates was concentrated and partially purified by solid phaseimmunoabsorption. Monospecific polyclonal antibodies to human apo A-I were pre-absorbed onto49agarose-immobilized protein G (GammaBind G Agarose, Genex Corp) at 40 C. After 30 minutes, asample of medium or cell lysate was added, and the suspension rotated end over end for 2 hours atroom temperature or overnight at 43 C. Agarose bound immune complexes were pelleted bycentrifugation (12,000 x g, 10 mins) and unabsorbed material removed by two washes with imi of TBS(10mM Tris-HCI, pH 7.4, 150mM NaC1). Antigen-antibody complexes were released by heating thebeads in 2 x SDS sample buffer at 90 C for 10 mins. Sepharose beads were removed by centrifugation (2mm, 12,000 x g) and the supernatant recovered for further analysis.2.14. Electrophoretic Analyses2.14.1. DNA Fragment Separation on Agarose GelsDNA fragments were separated in submersed horizontal agarose gels containing 1 x TAE(40mM Tris, 20mM NaAcetate, 1mM EDTA, pH 7.2) and l g/ml ethidium bromide. The agarosecontent varied between 0.8 and 2.0% (w/v) depending on the expected size of the fragment of interest.Samples were adjusted to 8% sucrose, 20mM EDTA, 0.05% bromophenol blue and 0.05% xylene cyanolto facilitate application to sample wells. The latter two components served as reference markers tomonitor the progress of the separation. Electrophoresis was performed under constant voltage conditions(80-100 volts).2.14.2. Protein Analysis2.14.2.1. SDS-Polyacrylamide GelsSodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) was performed indiscontinuous Tris-glycine buffer as described (Laemmli, 1970). Gels were cast in 0.75mm slabs (BioradMini Protean II) containing 12% acrylamide resolving gel and 4% acrylamide stacking gel. Samples wereadjusted to 50mM Tris-HC1 pH 6.8, 1% SDS, 20% glycerol, 0.005% bromophenol blue and 5% /3 -mercaptoethanol and were boiled for Smins prior to application to sample wells. Separation wasaccomplished at constant voltage (100V) until the tracking dye had migrated to the bottom of theresolving gel. Gels were stained (where indicated) with 0.25% Coomassie Blue R250 in 45% methanol,10% acetic acid and destained in the same solvent.50For autoradiography, gels were fixed in destaining solution, equilibrated with Amplify(Amersham Canada Ltd.) and dried onto Whatman 3MM chromatography paper. Autoradiograms wereexposed on X-Omat AR film (Eastman-Kodak) for 16-96 hours. [14 Cjmethylated proteins (AmershamCanada Ltd., Oakville, Ont.) were used as molecular mass markers for interpretation of autoradiograms.2.14.2.2 Isoelectric Focusing Gel ElectrophoresisIsoelectric focussing (IEF) was performed in 5% polyacrylamide slab gels containing 8Murea and 2.5% (w/v) Ampholines (Pharmacia-LKB, pH 4.0-5.5) as described by Hill and Pritchard(1991). For autoradiography, the gels were fixed for at least one hour and equilibrated with Amplify asdescribed above for SDS-PAGE slabs. For immunoblot analysis, proteins were electrophoreticallytransferred to nitrocellulose using 0.7% acetic acid as transfer buffer (see below).2.14.2.3. Two-dimensional Polyacrylaniide Gel ElectrophoresisTwo-dimensional gel electrophoresis was performed as described by O’Farrell (1975). Thefirst dimension was carried out in 1mm capillaries containing 8 M urea and 2.5% (w/v) Ampholines (pH4-5.5) for 4-6 hours at 250V. The gel was then extruded from the capillary and equilibrated with SDSsample buffer for 5-15 mins. Excess buffer was removed and the gel was placed on top of the seconddimension 12% SDS-PAGE slab gel described above. Separation in the second dimension was achievedat 100V until the tracking dye had reached the bottom of the gel. The gel was then processed forautoradiography.2.14.2.4. Immunoblot AnalysisProteins separated by polyacrylamide gel electrophoresis were transferred to nitrocellulosemembranes from slab gels by electroblotting (Towbin, 1979). Transfer was achieved at 100V in 1 hour ina cooled buffer chamber. Transfer buffer for SDS-PAGE was 20% methanol in 25mM Tris, 192mMglycine, pH 8.3. Transfer buffer for IEF gels was 0.7% acetic acid and utilized reverse polarity during thetransfer process. Unoccupied nitrocellulose sites were blocked with 5% non-fat powdered milk in PBS(30mins). Apo A-I was identified on the membrane by incubation with polyclonal antibodies (1:50051dilution in 0.5% milk powder/PBS) for 1 hour. The filters were then washed extensively with 0.02%Tween-20 in PBS and incubated with protein G HRP-conjugate for 30mins. After washing, the locationof the immune complex was determined with 0.5g/L DAB, 0.3g/L CoCI2, 0.006% (v/v) hydrogenperoxide in PBS. The color reaction was terminated in running tap water. All incubations wereperformed at ambient temperature.2.15. Indirect Immunofluorescence MicroscopyCell monolayers were harvested by trypsinization and transferred to poly-L-lysine coatedcover slips in appropriate growth medium. After 24 hours nonadherent cells were removed and themedium was replenished.Clonal cells expressing apo A-I under control of the mMT-I promoter were stimulated byaddition of zinc sulfate (10 M). After 4 hours, the cells were washed extensively with PBS and fixed.Transiently transfected COS cells were harvested 24 hours after the DMSO shock and plated ontocoverslips in growth medium. After 16-24 hours non-adherant cells were removed by PBS wash and thecoverslips were fixed.Fixation was achieved with 4% paraformaldehyde (lOmins, 37 C). The fixative was thenremoved by PBS wash and cells were permeabilized with 1% (w/v) Triton X-100. The latter treatmentallowed access of the antibody to intracellular structures. Cellular apo A-I was visualized by indirectfluorescence labelling. Murine monoclonal antibody to apo A-I (6B8) was used as the primary antibodyand FITC-labelled goat anti-mouse IgG (Sigma Chemical Co.) was utilized as detection antibody. Bothwere used at working concentration of 1% (v/v) in PBS. Incubations were 30mins duration at ambienttemperature, separated by PBS washes. Control preparations, omitting the primary antibody, were alsoperformed and displayed negligable fluorescence. Some coverslips were counterstained with TRITCcoupled wheat germ agglutinin (WGA) or concanavalin A (ConA) following the immunofluorescencedetection. This permitted localization of Golgi or endoplasmic reticulum, respectively (Kaariainen, 1983).Coverslips were mounted in veronal buffered (pH 8.6) 50% glycerol and viewed under a Zeiss 1M35fluorescence microscope equipped with filters for differential visualization of FITC and TRITC.522.16. Immunogold Electron MicroscopyBHK cells expressing apo A-I were harvested with a rubber policeman into PBS, 4 hoursafter addition of 10 M zinc sulfate. These samples were prepared for cryoultramicrotomy andimmunolabelling according to established procedures (Tokuyasu, 1973; 1978; 1983; 1984; Griffiths, 1984).Briefly, cells were washed with PBS by centrifugation and fixed for 30 mins in 1% glutaraldehyde in200mM HEPES, pH 7.4. The cell pellet was recovered by centrifugation and washed free of fixative with200mM HEPES containing 10% sucrose and 0.1% sodium azide. Samples were cryoprotected in 2.3Msucrose for 30 mins and then frozen in liquid nitrogen. Ultrathin (80-100 nm) cryosections were cut at-85’ C (using a diamond knife in an RMC MT6000-XL ultramicrotome with a CR2000 cryochamber),collected on 2.3M sucrose droplets and transferred to Formvar-coated grids. Cryosections wereimmunolabelled with antibody 6B8 (1:100 dilution) or non-immune mouse serum (bug/mi) for 30 mills,followed by secondary labelling with goat-anti-mouse IgG conjugated to lOnm colloidal gold(O.D.520 = 0.1; GAM-lO, Sigma Chemical Co., St. Louis, Mo. Cat. No. G-3641). Immunolabelledcryosections were embedded, contrasted with 1.8% methylcellulose and 0.3% uranyl acetate, andexamined on a Phillips EM400 transmission electron microscope.2.17. Quantitation of Apo A-I by Competitive ELISAApo A-I was quantitated in cell culture medium and in gradient ultracentrifugation fractionsby measuring the remaining unbound antibody to apo A-I after incubation with sample to equilibrium(Wong, L; personal communication). Apo A-I sample or HDL standard was mixed with antibody(diluted in PBS/0.05% Tween—20) and immune complexes formed overnight at ambient temperature.Simultaneously, wells of microtitre dishes were coated with 5Ong HDL in PBS. The coated plates werewashed with PBS/0.05% Tween—20 and lOQu I of the antigen-antibody mixture was added to each well.After 30mins, the solution was removed and the wells were washed. During this incubation, IgG whichhad not been bound by sample apo A-I was adsorbed to the coating HDL. Detecting antibody (donkeyanti-sheep IgG, HRP conjugate) was then added. After 30mins, additional washes were performed andsubstrate solution (0.04% o-phenylenediamine, 0.01% hydrogen peroxide in 0.1M sodium citrate pH 5.0)was added. Color development was complete within 5-l5mins (maximum absorbance had reached53O.D.490 = 1-L5). The reaction was stopped by adding sulfuric acid to 2N. Absorbance was measured at490nm in a Biorad Model 3550 microplate reader and apo A-I concentration determined by interpolationfrom the linear portion of the standard curve (generally between 25-150ng/ml).2.18. Analysis of Apo A-I Function2.18.1. Preparation of Single Bilayer VesiclesPhosphatidyicholine and cholesterol were mixed at 4:1 molar ratio for preparation ofvesicles (Batzri and Korn, 1973) for use as lipid binding and LCAT activation substrates. 1.3mg of eggyolk phosphatidyicholirie (Te III-E, Sigma Chemical Co., St. Louis, Mo.), 0.15mg cholesterol (CH-S,Sigma) and l2uCi of f7(n)-3HJcholesterol (Amershain, 5-15 Ci/mmol, lmCi/ml) were mixed and driedunder a stream of nitrogen gas at room temperature. The residue was redissolved in 12.i 1 of 99%ethanol and rapidly injected through a 25 guage needle into lOmis of TEN buffer (10mM Tris-HCI, pH7.4, 5mM EDTA, O.15M NaC1). The resulting preparation was concentrated to 2.5mls in an Amiconstirred cell ultrafiltration device (PM-30 membrane) and stored at 4°C for up to 5 days.2.18.2. Assessment of Lipid Binding CharacteristicsRecombinant apo A-I preparations were concentrated in dialysis membranes immersed incarboxymethylcellulose (Aquacide I, Calbiochem Corp., La Jolla, Ca.). Substrate lipid vesicles weremixed with recombinant apo A-I (molar ratio 200: 50: 0.005, PC: UC: A-I) in a final volume of 0.75mlsand incubated at 37 C for 30mins. The mixture was then layered onto the top of a 6m1 linear densitygradient of potassium bromide (KBr) in saline (1.006 to 1.250 g/ml). Centrifugation was performed at15C, 40,000rpm in an SW41Ti rotor (Beckman Instruments Inc., Palo Alto, Ca.) and was terminatedwithout braking after 44-48 hours. Gradients were fractionated into 0.Sml aliquots from the tubebottom. Apo A-I was measured in each fraction by competitive enzyme linked immunosorbent assay(ELISA) and the vesicle lipids were located within the gradient by liquid scintillation counting. Thedensity of each fraction was measured by diluting 10Oil aliquots to 5.Omls with water and measuring theconductivity of the resulting solution. Fraction densities were extrapolated from a standard curve derivedfrom density solutions which had been measured gravimetrically and by conductivity.542.18.3. Measurement of LCAT Cofactor ActivitySubstrate vesicles (3(u 1 containing 4.6mnoles of UC) were mixed with apo Al in a fmalvolume of 3OQu 1 and incubated at 37° C for 30mins to allow apo A-I to bind to the substrate. Fatty aciddepleted bovine serum albumin (Sigma) and j3 -mercaptoethanol were added to 2% (w/v) and 10mM,respectively. The reaction was initiated by addition of 2il 100 ng) of recombinant LCAT which wasexpressed in serum free culture medium by transfected BHK cells (J. Hill et al, in press). 100.i 1 aliquotsfrom each incubation were removed after 0, 30, 60, and l20mins at 37) C and were terminated by theaddition of 2mls of chloroform:methanol (2:1, v.v). Lipids were extracted by the method of Foich et at(1957), and the chloroform phase was recovered following low speed centrifugation. This was driedunder a stream of air and the residue was dissolved in 100l of chloroform containing 2(ig each FC andCE (as cholesterol oleate). Lipids were applied to the origin of silica gel G thin layer chromatographyplates and were developed in petroleum ether: ether: acetic acid (70:12:1). FC and CE bands wereidentified with iodine vapour, excised from the plate and quantitated by liquid scintillation spectrometryin toluene based scintillant (Omnifluor, NEN-Dupont, Mississauga, Ont.). LCAT activity was calculatedfrom the slope of the time course describing the conversion of FC to CE and was expressed as nmolesFC esterified per hour per ml of rLCAT.553. DEVELOPMENT OF APO A-I EXPRESSION SYSTEMSThe goal of the early phase of this work was to establish cell culture systems for the expressionof recombinant human apo A-I. Initially, nucleotide sequence analysis of the cDNA was performed toconfirm that the coding region for the apo A-I precursor was intact. In vitro transcription and translationstudies were then used to establish that the cDNA produced immunoreactive apo A-I. Eucaryoticexpression vectors were constructed and utilized to assess the molecular characteristics of the proteinproduced in transiently expressing cell cultures. Cell lines expressing apo A-I protein were establishedwhich had stably integrated the cDNA into their genome. These cell lines were compared for theircapacity to produce apo A-I under defined culture conditions. Finally, the biologic properties of therecombinant protein were assessed.3.1. Sequencing of the Full Length Apo Al cDNANucleotide sequence analysis was used to verify the published sequence of the apo A-I cDNAclone pBL13AI which contains lKbp of apo A-I sequence in the vector pBR329 (Sielhamer et al,1984). The strategy used for M13 subcloning and sequence analysis is illustrated in Figure bA in whichsubclones were selected to provide sufficient overlap to verify the entire coding region in the originalcDNA. Since this sequencing indicated concordance with the published sequence in the translated region,bidirectional sequencing over the entire length of the cDNA was not performed. However, the EcoRIcloning site was found at nucleotide 60 of the published sequence (Figure lOB). The same sequence wasobtained in three separate M13 clones derived from the original plasmid and did not, therefore, appearto be an artifact of subcloning performed in our laboratory. This cDNA lacks 13 bp of the 5-untranslated region of the apo A-I cDNA and was therefore less than full length although it containedthe complete translated sequence. It appears that the clone provided was actually pBL14AI, which wasalso described in the same laboratoiy (Seilhamer et al, 1984). The absence of this DNA segment in theflanking region had no deleterious effect on subsequent DNA manipulations, nor on the ability toexpress authentic apo A-I protein in vitro. Our sequence data predicted that insertion of the apo A-IcDNA at the Smal or EcoRI site of pSPT19 would produce transcript which contained the ATG codonof the apo A-I precursor as the only translation initiation site.56AEco RI Taq I SstI Sst I Taq I Taq I Eco RII I I I1 185 495531 790 856 961100 bpFigure 1OA. DNA sequence analysis of the apo A-I cDNA pBL13AI. Location of restriction enzyme sitesused to generate M13 subclones from the full length cDNA. Numbers represent nucleotide position inthe published sequence (Seilhamer et al, 1984). Arrows indicate the direction and quantity of sequenceinformation obtained from each subclone.57B1 A ATT OAA AAA AAA AAG AAA AAA AAA AAA AAA AAA AAA AAA AAA AAA49 GAG AGA CTG CGA GAA GGA GGT COO CCA CGG 000 TTC AGG ATG AAA GCTMet Lys Ala97 GCG GTG OTG ACO TTG GCC GTG CTC TTO CTG ACG GGG AGC CAG GOP CGGAla Val Leu Thr Lou Ala Val Lou Phe Leu Thr Gly Ser Gin Ala Arg145 CAT TTC TGG CAG CAA GAT GAA CCC 000 OAG AGC 000 TGG GAT CGA GTGHis Phe Trp Gin Gin Asp Glu Pro Pro Gin Ser Pro Trp Asp Arg Val193 AAG GAC CTG GCC ACT GTG TAC GTG GAT GTG CTC AAA GAC AGC GGO AGALys Asp Lou Ala Thr Val Tyr Val Asp Val Leu Lys Asp Ser Gly Arg241 GAO TAT GTG TCC CAG TTT GAA GGC TCC GCC TTG GGA AAA CAG CTA AAOAsp Tyr Val Ser Gin Phe Glu Gly Her Ala Lou Gly Lys Gin Leu Asn289 CTA AAG CTC CTT GAC AAC TGG GAC AGC GTG ACC TCC ACC TTC AGC AAGLGu Lys Leu Leu Asp Asn Trp Asp Ser Val Thr Her Thr Phe Ser Lys337 CTG CGC GAA CAG CTC GGC CCT GTG ACC CAG GAG TTC TGG GAT AAC OTGLou Arg Glu Gin Leu Gly Pro Vol Thr Gin Glu Phe Trp Asp Asn Lou385 GAA AAG GAG ACA GAG GGC OTG AGG CAG GAG ATG AGO AAG GAT CTG GAGGlu Lys Glu Thr Glu Gly Lou Arq Gin Glu Met Ser Lys Asp Leu Glu433 GAG GTG AAG GOC AAG GTG CAG 000 TAO OTG GAO GAO TTO OAG AAG AAGGlu Val Lys Ala Lys Val Gin Pro Tyr Lou Asp Asp Phe Gin Lys Lys481 TGG OAG GAG GAG ATG GAG OTO TAO OGO OAG AAG GTG GAG COG OTG OGOTrp Gin Glu Glu Met Giu Lou Tyr Arg Gin Lys Val Glu Pro Lou Arg529 GOA GAG OTO OAA GAG GGO GOG OGO CAG AAG OTG OAO GAG OTG OAA GAGAla Glu Lou Gin Glu Gly Aia Arg Gin Lys Lou His Glu Lou Gin Glu577 AAG OTG AGO OOA OTG GGO GAG GAG ATG OGO GAO OGO GOG OGO GOO OATLys Lou Her Pro Lou Gly Glu Glu Met Arg Asp Arg Aia Arg Aia His625 GTG GAO GOG OTG OGO AOG OAT OTG GOC 000 TAO AGO GAO GAG OTG OGOVal Asp Ala Lou Arg Thr His Lou Ala Pro Tyr Her Asp Glu Lou Arg673 CAG OGO TTG GOC GOG OGO OTT GAG GOT OTO AAG GAG AAO GGO GGO GOOGin Arg Lou Ala Ala Arg Lou Glu Aia Lou Lys Glu Asn Gly Gly Ala721 AGA OTG GOO GAG TAO OAO GOC AAG GOO AOO GAG OAT OTG AGO AOG OTOArg Lou Ala Glu Tyr His Ala Lys Ala Thr Glu His Lou Ser Thr Lou769 AGO GAG AAG GOO AAG 000 GOG OTO GAG GAO OTO OGO OAA GGO OTG OTGHer Glu Lys Ala Lys Pro Ala Lou Glu Asp Lou Arg Gin Gly Lou Lou817 000 GTG OTG GAG AGO TTO AAG GPO AGO TTO OTG AGO GOT OTO GAG GAGPro Val Lou Glu Ser Phe Lys Val Ser Phe Lou Ser Ala Lou Glu Glu865 TAO ACT AAG AAG OTO AAO ACO CAG TGA GGO GOO OGC OGO OGO 000 OCTTyr Thr Lys Lys Lou Asn Thr Gln9i3 TOO OGG TGO TOA GAA TAA AOG TTT OOA AAG TGT TAA AAA AAA AAA AAA961 GAA PrOFigure lOB. DNA sequence analysis of the apo A-I cDNA pBL13AI. Nucleotide and translated aminoacid sequence of pBL13AI. Numbers refer to the nucleotide position in the sequence of the originalpublication (Seilhamer et cii, 1984). Sequence data obtained in this study begins at position 60 (see textfor details).583.2. In vitro Transcription and Translation of the Full Length cDNAVectors for coupled transcription-translation were prepared to confirm that the apo A-I cDNAcontained the appropriate signals for eucaryotic translation and microsomal translocation. The cDNAwas cloned into the vector pSPT19 to facilitate in vitro transcription in either orientation. The cDNA wasinserted at either the EcoRI or the Smal restriction site of the vector and clones containing the apo A-Iin either orientation were isolated. In many instances low quantities of heterogeneous transcript wereobtained on addition of RNA polymerase. Large quantities of homogeneous transcript were synthesizedwhen the apo A-I cDNA was placed at the Smal site of pSPT19 and in the orientation for the SP6polymerase promoter. Approximately 1.5-2k g of DNase I resistant nucleic acid was obtained per g ofDNA template under run-off transcription conditions (Figure hA). Addition of this product to templatedependent rabbit reticulocyre lysate reactions generated a single protein species with apparent molecularmass of 31 kiodaltons (kD) on SDS-PAGE (Figure 11B, lane 2). This was consistent with thepredicted molecular mass for preproapo A-I. In some reactions the cap analogue7mG(5)ppp(5)G wasincluded during transcription which substantially reduced the quantity of transcript synthesized.Subsequent addition of this material to reticulocyte lysate indicated that the capped mRNA was moreefficiently translated than uncapped mRNA (data not shown). This observation was in agreement withprevious reports of more efficient translation of capped transcripts in the wheat germ translation system(Krieg and Melton, 1984). -Translation reactions were performed in the presence and the absence of dog pancreaticmicrosomes. In the presence of microsomes the primary translation product was partially converted to aspecies of lower molecular mass 30 kD, Figure 11B, lane 3). This observation was consistent withtranslocation of the preproapo A-I across the microsomal membrane and proteolysis of the signalpeptide, demonstrated using liver mRNA (Zannis et at, 1983). Both the larger and the smaller apo A-Ispecies were precipitated with monospecific antibodies to human plasma apo A-I (Figure 11B, lanes 5-8)indicating that both the primary translation product and the smaller translocated form were authenticapo A-I.The amino acid sequence of the apo A-I signal peptide (derived from the cDNA sequence)suggested that preproapo A-I and its translocated product could be differentiated by charge and size59mRNAM icrosomesImmunoprecipitate21 Kbp0.512345678— + + — + + + +—— + — — — + +— —— + + + + +Figure 11. In viiro transcription and translation of the full length apo A-I cDNA.A- Hindlil linearized pSPT19-AI (lane 1, 2i g) was incubated with SP6 RNA polymerase (lane 2, 20units) for 2 hours at 373 C. An aliquot of the reaction was removed and treated with DNase I to removetemplate DNA (lane 3). Aliquots of each reaction were fractionated by gel electrophoresis in 1% agaroseand visualized by ethidium bromide staining.B- Translation in template-dependent rabbit reticulocyte lysate. Each reaction (2. 1 final volume) wasincubated for 30mins at 311 C in the presence (+) or absence (-) of template mRNA (0.5 g) and dogpancreatic microsomes as indicated. In lanes 1 to 3, 0.u 1 of the reaction was applied directly to SDSPAGE gel containing 12.5% acrylamide. In lanes 4, 6 and 8, 1 1 of the reaction was subjected toimmunoabsorption with polyclonal anti-human apo A-I prior to SDS-PAGE fractionation. In lanes 5 and7, immunoisolates were prepared from 0.5 1 of each reaction. The arrow indicates the mobility ofpurified plasma apo A-I.A B1234+ M92.5-66.2-45-r 31-I2 1.5-14.4-—— — —60CIEF(+SDS92.5- 92.5-66.2 - 66.2 -45- 45-31-___31-U21.5- 21.5-14.4- •1WITHOUT MICROSOMES WITH MICROSOMESFigure 11C- Two dimensional electrophoretic separation of in vitro translation products. 1 of eachtranslation reaction was mixed with 1j g of purified plasma apo A-I and separated by two-dimensionalelectrophoresis according to O’Farrel (1975). The gels were stained with Coomassie Blue to identify theposition of mature apo A-I (dotted circle) and then processed for autoradiography to identify thetranslation products. The position of the stained molecular mass markers was determined by overlay ofthe autoradiographic film on the stained gel.61characteristics. The products of reticulocyte translation in the presence or absence of microsomes wereanalysed by two-dimensional gel electrophoresis according to O’Farrell (1975). Preproapo A-I was theonly species predicted from translation in the absence of microsomes. The product of this incubationdisplayed a more basic isoelectric point (p1) than purified plasma apo A-I, in addition to a highermolecular weight ( 31 kD vs 27 kD, Figure 11C). Additional species with p1 between preproapo A-Iand mature apo A-I were also evident. The origin of these charge-shifted forms is not clear. Two-dimensional electrophoretic analysis of products synthesized in the presence of microsomes showed thatthe smaller apo A-I species (M 30 kD) had p1 between preproapo A-I and mature plasma apo A-I.Precursor and product species were present in approximately equal proportions indicating that the signalpeptide was only partially hydrolysed. Complete hydrolysis of the signal sequence was not achieved evenwith additional microsomes (up to 2j.t g protein). Since the apo A-I signal peptide contains a singlecharged residue (Ly23), the charge differences we observed were consistent with the processing of thepreproapo A-I to proapo A-I on translocation across the ER membrane. Folz and Gordon (1987) haveshown previously that the proapo A-I product is sequestered within the microsome. The presence of anet charge on the propeptide (÷ 2) and on the signal peptide (+1) allows preproapo A-I, proapo A-I andmature apo A-I to be differentiated on the basis of charge and size criteria by two-dimensional gelelectrophoresis. In contrast to preproapo A-I, proapo A-I was not heterogeneous in the IEF dimension,which may indicate that amino acid residues in the signal peptide are more susceptible to in vitromodification.Our findings provide evidence that apo A-I cDNA expressed in vitro from its cDNA iscompartmentalized within the ER as proapo A-I and are consistent with those obtained with human livermRNA (Zannis et al, 1983). In order to produce quantities of recombinant protein for functionalanalysis, COS, BHK and CHO cells were evaluated as host cell cultures for the expression of human apoA-I.3.3. Expression of Wild-type Apo Al (Apo A.Iwt) in COS CellsObservations in the rabbit reticulocyte lysate established that apo A-I was synthesized from thecDNA as the preproapolipoprotein and was converted to proapo A-I during ER translocation. To62establish that, subsequent to ER translocation, the apo A-I protein was secreted from cultured eucaryoticcells which do not normally express apo A-I, trafficking and secretion were analysed following transienttransfection of COS cells. Forty hours after transfection, [ Simethionine was incorporated into cellularprotein in order to follow the cellular processing and secretion of newly synthesized apo A-I in asubsequent unlabelled methionine chase incubation. Apo A-I was isolated by immunoabsorption fromcells and medium and subjected to SDS-PAGE analysis (Figure 12A). A single apo A-I species(M 3OkD) was found in the medium, consistent with secretion of proapo A-I. However, cell lysatescontained a species of higher molecular mass ( 3lkD) in addition to proapo A-I. This larger apo A-Iwas observed in cell lysates throughout the chase interval, and the autoradiographic intensity of this banddecreased with time. The intensity of the intracellular proapo A-I signal also decreased, in parallel withthe larger form. The 31 kD band was identified as preproapo A-I on the basis of charge-size propertiesusing two-dimensional electrophoretic analysis (Figure 12B). The presence of preproapo A-I intransfected cell lysates was unexpected since signal peptide hydrolysis normally occurs as acotranslational event.Radiolabelled apo A-I was quantified at each time point by liquid scintillation counting of theimmunoabsorbed material. Approximately one-third of the cellular apo A-I, radiolabelled during thepulse period, was not recovered from either the cells or the medium after 24 hours chase, suggesting thatthis portion had been degraded. Apo A-I was a major product of the cell culture, approximately 3% oftotal cellular and 33% of total secreted protein (TCA precipitable) radioactivity during the chase period.The cellular half-life of apo A-I synthesized de novo was estimated to be 12 hours (Figure 12C) fromthis experiment.To determine if the cells produced sufficient quantity of apo A-I for structure-function analysis,the apo A-I mass in the medium and cell lysates of the transiently transfected cells was estimated bywestern blot analysis. Media and cell lysates were collected from 35mm culture dishes 40 hours aftertransfection with the apo A-I cDNA and the apo A-I product was concentrated by immunoabsorption.Immunodetection was used to estimate both the cellular and the secreted apo A-I concentrationfollowing SDS-PAGE separation and electroblotting (Figure 13). Control cultures transfected with emptypCMV5 vector contained neither cellular nor medium apo A-I, establishing that COS cells do not63AHOURSMEDIUM 0 2 4 6 8 2492.5-66.2-Mr 31 _21.5-14.4-CELLS92.5-66.2-Mr 31 - __ —21.5-14.4-Figure 12A. Expression of the apo A-I cDNA in transiently transfected COS cells. COS cells weretransfected with pCMV5-AIwt and the expressed product was identified in S]labelled cultures. A- ApoA-I species were isolated by immunoabsorption after the indicated chase interval, separated by SDSPAGE and visualized by autoradiography. The arrow indicates the position of plasma apo A-I which wasadded prior to electrophoresis as a marker.64BIEFSDSf31 Kd-’pCMV5-AlwtFigure 12B. COS cell lysate contains prepro- and pro- apo A-I. Immunoisolate from [ SI labelledtransfected cell lysate was mixed with purified plasma apo A-I (dotted circle), separated by two-dimensional gel electrophoresis and visualized by autoradiography. Horizontal and vertical arrowsindicate the direction of IEF and SDS-PAGE, respectively. The position of plasma apo A-I wasdetermined by Coomassie Blue staining.65UJLJO_ocLU0‘4-0-cCHASE TIME(hrs)Figure 12C. Quantitative pulse-chase analysis of apo A-I expression in transfected COS cells. Apo A-Iwas isolated by immunoabsorption after the indicated chase following pulse-labelling with[35 S]methionine. Radioactivity was quantitated as the proportion of the total apo A-I radiolabelrecovered per dish in CELLS or MEDIUM as determined by liquid scintillation counting. TOTALrepresents the sum of cell lysate and medium apo A-I radioactivity expressed as the percent of the lysatevalue at t=O.zII0661 5 10 50 100 C M C Mng MATURE Al +Figure 13. Immunoblot analysis of apo A-I expression in COS cells. Apo A-I was concentrated byimmunoabsorption from cells (C) and medium (M) following transfection with pCMV5 (-) or withpCMV5-AIwt (+). Samples were resolved by SDS-PAGE and visualized by immunoblotting as describedin MATERIALS AND METHODS. Purified plasma apo A-I was used as a standard as indicated. Mrindicates the relative mobility of molecular mass markers in kilodaltons.6792.5-66.2-45-Mr 31-2 1.5-14.4-express endogenous, immunoreactive apo A-I. Cells transfected with vector containing the apo A-IcDNA produced approximately 500ng of apo A-I per 35mm dish (2mls medium). However, only 5-10%of the apo A-I mass was secreted during the 40 hour collection period. Immunoreactive partialdegradation products were not evident on the blots.COS cells secreted only small quantities of apo A-I (approximately 5-10 ng/ml/24 hours),although analysis of cell lysates indicated that substantially more was synthesized than was secreted. Theintracellular pool contained both preproapo A-I and proapo A-I in approximately equal proportions.Retention of the signal peptide on a considerable portion of apo A-I suggested that ER translocationwas inefficient in the transfected cell. Since significant quantities of proapo A-I were also retained by thecells, later secretory events were also inefficient in these cells. The pulse-chase studies suggested that asubstantial portion of the apo A-I labelled during the pulse period was later degraded, although largeamounts of apo A-I protein were detectable at this time by immunoblot analysis. This suggested that atleast two pools of apo A-I exist in the transfected cell, with the apo A-I synthesized 40 hours aftertransfection subject to more rapid degradation than the apo A-I synthesized prior to that time.3.3.1. Properties of Apo A-I Secreted by COS CellsThe functional properties of secreted apo A-I were assessed in 1n S]methionine labelling studiesof transiently transfected COS cell cultures. The ability of the radiolabelled recombinant to integrate intoexisting lipoproteins was determined. Since apo A-I is believed to enter plasma as free apolipoprotein,rapid association with existing HDL is physiologically relevant. Radiolabelled apo A-I was chased fromthe cells into standard growth medium containing 10% FBS which was then fractionated by densitygradient ultracentrifugation. S]Methionine label was found in a symmetrical peak in fractions ofdensity 1.10-1.20 g/ml with maximum radioactivity at 1.15 g/ml (Figure 14A). Immunoabsorption andSDS-PAGE analysis of the gradient fractions showed that this label was associated exclusively with apoA-I (Figure 14B). No apo A-I was detected in the fractions of higher or lower density. Thus, therecombinant proapo A-I produced by transfected COS cells incorporates exclusively into lipoproteins ofthe HDL density class. Under these culture conditions, lipid-free apo A-I was not detected. Inpreliminary experiments, the secreted product was also analysed by non-denaturing gradient gel68a).9?Co0—LxpFigure 14. Apo A-I produced by transfected COS cells is associated with HDL in the culture medium.Cells transfected with pCMV5-AIwt were pulse-labeled withr5S]methionine and chased with mediumcontaining 10% FBS. The medium was concentrated by ultrafiltration and fractionated by densitygradient ultracentrifugation.A- Radioactivity (solid line) was found in gradient fractions (dotted line) with density 1.12-1.18 g/ml.B- Gradient fractions were immunoabsorbed with antibody to human apo A-I and resolved by SDSPAGE. Label was found in a single protein species with M 3OkD.4AB1.3001.2501.2001.1501.1001 .0502FRACTIONSDS-PAGE101.000169electrophoresis (not shown) which indicated that apo A-I was associated with lipoprotein with the sizecharacteristics of human plasma HDL. Therefore, transfected COS cells secrete recombinant humanapo A-I that associates readily with high density lipoproteins.We attempted to determine if such a complex could also form in cultures without addition ofserum. Very little apo A-I secretion could be detected by immunoabsorption of the medium under theseconditions arid it was not possible to characterize the complexes. However, the level of [ S]methionineincorporated into cellular apo A-I was not different in serum free medium, despite the lower level of apoA-I secretion. We hypothesized that FBS provided an essential component for apo A-I secretion andinvestigated this possibility further by assessing the quantitative effect of FBS on apo A-I secretion(Figure 15). Secreted apo A-I radiolabel increased in response to the concentration of FBS in themedium in a dose-dependent manner. Near maximal effect was achieved at very low levels of FBS, asten-fold stimulation of apo A-I secretion was observed at only 0.5% (v/v) FBS. Further increases in FBShad only modest stimulative effect. At extremely high levels of FBS (20%, v/v), preproapo A-I wasdetectable in the medium by SDS-PAGE.The individual lipoprotein components of human serum were tested for their influence on apoA-I secretion. Apo A-I secretion was stimulated by human LDL in a similar manner as by FBS, and theeffect was observed with as little as 7 g of LDL-C was added per ml medium. The stimulatory effect wasmaximal at approximately 700ig/m1. VLDL and the intravenous lipid emulsion Liposyn (AbbottLaboratories, Montreal, Que.) also stimulated apo A-I secretion but to a lesser extent. In contrast to thethe effect of other lipoproteins, HDL reduced apo A-I secretion by approximately 25% compared tocontrol dishes without additive.Post-translational phosphorylation of apo Al has been reported in primary human hepatocytesand in the hepatoma line HepG2 (Beg el at, 1989). The authors have suggested that a phosphorylationdephosphorylation process may be involved in secretion of apo A-I from these cells. Since we hadobserved impaired secretion of apo A-I from the COS cell, we investigated the possibility that these cellsmight be unable to phosphorylate or to dephosphorylate apo A-I. Endogenous labelling of transfectedcells withr2Pjorthophosphate resulted in time-dependent intracellular phosphorylation of apo A-I(Figure 16). The molecular mass on SDS-PAGE indicated that preproapo A-I was the labelled species.7015--(30LLJC(I)—_0<-5500 0.5 1 5 10 20FBS Content(Volume %)Figure .15. Apo A-I secretion from (ransfected COS cells is stimulated by FBS. COS cells weretransfected with pCMV5-AIwt and pulse-labeled with Simethionine. Secretion of apo A-I from thecells into medium containing the indicated concentration of FBS was determined after 4 hours chaseincubation. Apo A-I was recovered from the medium by immunoabsorption and quantitated by liquidscintillation counting. Results are expressed as the radioactivity ratio in the presence and absence ofadditive and represent single dishes from a representative experiment.71HOURSMEDIUM 0 2 4 6 2435S690-L4 3-Figure 16. Apo A-I produced in COS cells is subject to intracellular phosphorylation. COS cells weretransfected with pCMV5-AIwt and radiolabeled for 2 hours with r2 Pjorthophosphate in phosphate-freemedium, as described in MATERIALS AND METHODS. After replacing the medium withDMEM/1O% FBS and incubating for the indicated chase interval, apo A-I was concentrated byimmunoabsorption from medium or cell lysate and analysed by SDS-PAGE with autoradiography.Numbers along the top of the autoradiograms indicate the chase time in hours. Control lane S) is theimmunoabsorbed material from the medium of a transfected culture radiolabeled with [ S]methionine.This sample was used to demonstrate that apo A-I was secreted during the chase. The arrow indicatesthe position of mature plasma apo A-I.CELLSMrMr4 6.0-30.0-I14,3-69.0-4 6.0-30.0-I72Additional banding was detected at 46kD (and larger) molecular mass in the cells after 24 hours. Itwas not clear whether these species were phosphorylated aggregates of apo A-I, or were non-specificallyabsorbed phosphorylated proteins. Under conditions where we had previously demonstrated maximal apoA-I secretion (DMEM/1O% FBS), immunoreactive material from the medium did not containr2 Piphosphate-labelled apo A-I, although dishes labelled in parallel with [ S]methionine indicated thatapo A-I was secreted (see Figure 16). The observations in COS cells were consistent with theobservations of Beg et at, indicating that proapo A-I is subject to phosphorylation and dephosphorylationprocesses during secretion. However, these observations do not rule out the possibility that apo A-Iphosphorylation may play a role in processes other than secretion. Abnormal phosphate modification ofapo A-I in COS cells does not appear to explain the reduced rate of secretion.3.3.2. Immunofluorescence Localization of A-Iwt in Transfected COS CellsThe biochemical observations in COS cells expressing human apo A-I suggested that largeamounts of the recombinant protein were retained rather than secreted. The intracellular location of apoA-I in these cells was investigated by indirect immunofluorescence microscopy to gain insight into themechanism of retention. Intense apo A-I reactivity was found throughout the transfected cells,concentrated near the nucleus with diminishing intensity toward the cell membrane (Figure 17A). UsingTRITC-ConA as a marker for the high mannose sugars of the ER (Kaariainen et at, 1983), it wasobserved that the apo A-I immunoreactivity and the ER were similarly distributed in the transfected cells(Figure 17B). Both labelling procedures suggested that the ER had expanded considerably in many ofthe cells expressing apo A-I.In some experiments, however, the fluorescence intensity and its location were variable. The apoA-I fluorescence in the ER region was less intense and a juxtanuclear staining pattern was moreprominent, indicating that the Golgi apparatus was the site of apo A-I accumulation. Thus, cellulartransport of apo A-I in transiently transfected COS cells was heterogeneous, not only from oneexperiment to another, but also from cell to cell within the same experiment.73Figure 17. Inimunofluorescence analysis of transfected COS cells indicates that apo A-I is retained inthe ER. A- Cellular apo A-I was displayed in fixed, permeabilized cells by indirect immunofluorescenceusing mouse anti-human apo Al (6B8) and FITC-conjugated goat anti-mouse IgG. Intense fluorescencewas found extending from the nucleus to near the cell membrane. B- The extent of ER in the cells wasvisualized by counterstaining the same cell preparation with TRITC-conjugated ConA. Apo A-I andConA colocalize within the transfected cells. Bar indicates 1m in both panels.743.4. Baby Hamster Kidney (BHK) Cell Expression of Apo A-Iwt3.4.1. BHK-AJwt Cells Produce proapo A-ICOS cell expression studies indicated that low levels of apo A-I could be produced and secretedby transiently transfected non-hepatic mammalian cells in culture. Since the transfected cell populationoften appeared heterogeneous by immunofluorescence microscopy, a cell line which had the apo A-IcDNA stably integrated into its genome could provide a continuous source of apo A-I from ahomogeneous cell population. BHK cell lines expressing the apo A-I protein from the wild-type cDNA(apo A-Iwt) were established. Approximately 40 BHK clones were established following transfection andselection and were screened by immunoblot analysis for secretion of apo A-I. One third of the cell linessecreted detectable quantities of apo A-I when maintained at confluence in serum-free culture. A singleclone (BHK-AIwtB5), secreting the highest level of apo A-I under these conditions was chosen fordetailed analysis.The apo A-I produced by BHK-AIwtB5 was isolated by immunoabsoiption from cells labelledwithr5S]methionine. The autoradiogram in Figure 18 indicates that a single species of apo A-I waslabelled, with an apparent molecular mass of approximately 28,000 daltons. Lysates prepared from cellsof the parent BHK line or cells transfected with pNTJT vector alone contained no detectable apo A-I(data not shown). Based on its p1 with respect to the major isoform of plasma apo A-I, the apo A-Iproduced by BHK-AIwtB5 was identified as proapo A-I. In contrast to COS cell transfectants, hydrolysisof the signal peptide from preproapo A-I was complete in the BHK cell. Furthermore, since BHK cellsproduced proapo A-I from the cDNA encoding preproapo A-I, the electrophoretic analysis suggestedthat BHK cells lack apo A-I intracellular propeptidase activity. The apo A-I secreted by these cultureswas also analysed by sequential Edman degradation of the amino terminus which indicated that twopolypeptides were present in the medium. The major secreted apo A-I species was proapo A-I (TableVI). However, a portion of the secreted apo A-I (estimated to be 20%) was mature apo A-I following 24hour collections of serum-free medium.75IEFSDS+31 KdBHK-AlwtB5(p-Al)Figure 18. BHK cells expressing apo A-Iwt produce a single molecular species with the electrophoreticproperties of proapo A-I (p-Al). BHK cell line AIwtB5 was pulse-labeled with [ Simethionine andharvested by cell lysis. Apo A-I was recovered by immunoabsorption, mixed with purified plasma apo A-Iand resolved by two-dimensional gel electrophoresis. Horizontal and vertical arrows indicate thedirection of IEF and SDS-PAGE, respectively. A single molecular species was observed, with more basicp1 than mature plasma apo A-I used as marker (dotted circle). The electrophoretic properties of apo A-Iin this cell line are those of proapo A-I (p-Al).76Table VI. Amino terminal sequence analysis of apo A-I secreted from cell line BHK-AIwIB5. X = aminoacid detected but not identified. Major component is 80% of protein present, minor component is20%.Source N-Terminal Sequence AssignmentPlasma X-Glu-Pro-Pro-Gln-Ser Mature apo A-IBHK-AIwtB5 (major) X-X-Phe-X-Gln-Gln-Asp Proapo A-IBHK-AIwtB5 (minor) X-Glu-Pro-Pro-Gln-Ser Mature apo A-IThis indicates that BHK cells may secrete a small quantitity of propeptidase activity. Nonetheless, BHKcells secrete the major portion of apo A-I as proapo A-I when expressed from the preproapo A-I cDNA,similar to 3T3 and L6E9 cells (L,amon-Fava ci at, 1987; Ruiz-Opazo et at, 1988). Unlike CHO-Ki, thesecells do not rapidly and efficiently hydrolyse the propeptide (Mallory et at, 1987).3.4.2. Characterization of Apo A-I Synthesis and Secretion by BHK Clone AIwtB5The vector utilized for transfection and selection of BHK-AIwtB5 placed apo A-I expressionunder control of the mouse metallothionein promoter. By including divalent metal cations in the culturemedium, we could increase the transcription of the apo A-I gene. Northern blot analysis was performedand was corrected for nonspecific stimulation and mRNA recovery by normalizing to the fi -actin signalof the same nucleic acid preparation. Apo A-I mRNA levels increased 3.2-fold during 24 hour culture inthe presence of 10 M zinc sulphate (Zn504). However, pulse-chase analysis of the apo A-I protein(Figure 19) indicated that secretion of apo A-I into serum free medium (DMEM) was minimal in thepresence or absence of ZnSO4.Similar results were obtained using lower concentrations of zinc or with20-10t M cadmium chloride. The addition of FBS (10%, v/v), however, stimulated secretion of apo A-I,similar to the earlier observations in transfected COS cells.Further characteristics of this effect were obtained from continuous labelling studies (TableVII). Culture in the presence of 10 M ZnSO4 decreased the synthesis of apo A-I more than two-foldfrom synthesis in DMEM only. Addition of 10% FBS to DMEM increased apo A-I synthesisapproximately 1.5-fold. In the presence of both 10% FBS and 10 M ZnSO4,apo A-I synthesis increased77cJ)LiJJOLLLUTIME(mins)Figure 19. FBS stimulates the secretion of apo A-I by BHK-AIwtB5. BHK-AIwtB5 cells were pulselabeled with S]methionine and the secretion of apo A-I determined in the presence or absence of 10%FBS and l0Qu M zinc sulfate after chase incubation for the indicated time. Apo A-I in cell lysates andmedia were isolated by immunoabsorption and quantitated by liquid scintillation counting. Results areexpressed as medium apo A-I (cpm) divided by total dish apo A-I (cpm) as percent. Data points are themean of duplicate dishes from a single experiment.0 20 40 6078more than 5-fold, suggesting that ZnSO4 increased apo A-I synthesis only if FBS was present. ZnSO4 didnot appear to alter the amount of apoprotein secreted since the distribution between cells and mediumwas not affected by its addition to the medium. FBS, however, had an effect similar to that observedearlier in COS cells. When 10% FBS was included in the medium the portion of the apo A-I found inthe medium increased from 40% to 60%.Table VII. Effect of growth conditions on apo Al synthesis and secretion in BHK-AIwtB5. BHK cellsexpressing the apo A-Iwt cDNA were labelled with L S]methionine for 12 hours in the presence of theadditives indicated. Apo A-I was immunoabsorbed from medium and cell lysate and radioactivity wasmeasured by liquid scintillation counting. Values represent the mean of duplicate dishes, differing by lessthan 15%.GROWTH CONDITION TOTAL APO Al SECRETED APO AT(cpm/dish) (% of total)DMEM Only 5568 40+10tMZnSO4 2160 45+10% FBS 8384 62+10% FBS/10zM ZnSO4 30000 64The time course of apo A-I secretion was then investigated in long term radiotracer pulse-chaseincubations. In accord with our observations in continuous labelling studies, the highest level of apo A-Iexpression and secretion was observed in the presence of both ZnSO4 and FBS. Apo A-I secretion wasinitially rapid (Figure 20), as 25-30% of the pulse-labelled apo A-I was recovered in the medium after 2-4 hours. However, less than 10% of the remaining label was secreted thereafter. As much as 50% of theinitial apo A-I label was not recovered in the medium or the cell lysate, apparently degraded early in thetime course. As in the earlier COS cell studies, no evidence of partial degradation products was obtainedin the BHK cell experiments.Accumulation of apo A-I in the cells was monitored by immunofluorescence light microscopy.Apo A-I was barely detectable in BHK-AIwtB5 cells under standard growth conditions (Figure 21A).However, the addition of 10 M ZnSO4 caused a profound increase in immune fluorescence decorationof the cells, 4 hours (Figure 21B) and 8 hours (Figure 21C) after the addition. The fluorescence intensitywas greatest in the juxtanuclear region of the cells and apo A-I label colocalized with the Golgi79I(/)LLJ+00ckUJLJ0H-LiJ04-o_oCHASE TIME(hrs)Figure 20. Long-term pulse-chase analysis suggests that degradation competes with secretion of apo A-Iin BHK-AlwtB5. BHK-AIwtB5 cells were pulse labeled with Simethionine and the time course of apoA-I secretion was determined in the presence of 10% FBS and lOQu M zinc sulfate. Apo A-I was isolatedform cell lysates and media by immunoabsorption and quantitated by liquid scintillation counting. Resultsare expressed as the ratio of apo A-I radioactivity in CELLS or MEDIUM over total apo A-Iradioactivity recovered at that time point. In the top panel TOTAL apo A-I radioactivity at each timeduring the chase is expressed as percent of radioactivity at the initiation of the chase (t 0). Data wasobtained from single dishes.80Figure 21. Zinc sulfate induction of apo A-I synthesis results in the accumulation of apo A-Iimmunofluorescence in BHK-AlwtB5. BHK-AIwtB5 cells were grown on coverslips and incubated withgrowth medium containing 1Ot M zinc sulfate. After the induction period, cells were washed, fixed andprepared for indirect immunofluorescence localization of apo A-I. A- 0 hours induction, B- 4 hoursinduction, C- 8 hours induction. The cellular immunofluorescence colocalized with the Golgi apparatusmarker TRITC-WGA. Bar = 10tm.81apparatus marker WGA. This marker stains the complex carbohydrate moiety of glycoproteins of theGolgi (Kaariainen et al, 1983) but does not stain apo A-I.The biochemical and immunofluorescence evidence suggested that apo A-I was expressed andsecreted more efficiently from BHK cells than from COS cells. However, even with the BHK cell,optimal secretion required the presence of FBS in the medium. Since FBS contains apolipoproteins,these conditions were not acceptable for preparation of recombinant protein for structure-functionanalysis. Bovine apo A-I is structurally homologous to its human counterpart (Auboiron et at, 1990;Sparrow et at, 1992) and the antibodies used were found to have low level cross-reactivity between thetwo species. It was necessary, therefore, to exclude FBS from cultures during recombinant apo A-Icollections, at the expense of secretion efficiency. Furthermore, we found that BHK cells often adaptedpoorly tO these conditions, and were viable for only 24-48 hours. Twenty-four hour collections of serum-free medium (Optimem) contained approximately lOOng/ml of apo A-I. As an alternative to the BHKexpression system, stable recombinant lines using CHO-Ki cells were also established, since these cellshave been shown to adapt well to serum-free conditions. Since BHK cells lack efficient propeptidehydrolytic activity, BHK cells expressing apo A-I from the preproapo A-I cDNA are a useful model forinvestigation of the functional role of the propeptide segment in constitutive cellular transport (Chapter4).3.5. CHO-Ki Cell Expression of Apo A-Iw(CHO-K1 cells produce and secrete mature apo A-I when transfected with the cDNA encodingpreproapo A-I (Mallory, 1987). Therefore, this cell line provides a valuable model system for theproduction of mature wild-type apo A-I and its mutants as recombinants.Cells were transfected with pNUT-AI plasmid and pRc/CMV (which provided neomycinresistance) and were selected in medium containing the neomycin analogue Geneticin (G418). Drugresistent colonies expressing apo A-I were identified and expanded. Apo A-I produced by the cells wasisolated from cell lysates and medium by immunoabsorption following 12 hours labelling with( S]methionine. Apo A-I was not detected by autoradiography of extracts from cells transfected withempty pNUT vector (Figure 22A). Cells transfected with the apo A-I cDNA displayed apo A-I radiolabel82ApNUT AIwtCMCM- -6 6. 2 --I--IMr 31-p21 5-,___________•__ __ _ __ __Figure 22A. CHO cells expressing apo A-Iwt secrete mature apo A-I. CHO cells were transfected withpNUT vector with or without the apo A-Iwt cDNA and selected with Geneticin (G418) for chromosomalintegration. Cells expressing apo A-I were identified and cultured in the presence of [ Simethionine at1O M for 12 hours. Medium (M) and cell lysates (C) were recovered and absorbed with anti-apo A-I.Immunoisolates were then resolved by SDS-PAGE and visualized by autoradiography. Cells transfectedwithout the apo A-I cDNA (pNUT) contain no detectable apo A-I. The majority of the apo A-Iradiolabel in cells transfected with the apo A-I cDNA (Alwt) is extracellular.83BeFigure 22B. CHO cells expressing apo A-Iwt secrete mature apo A-I. Apo A-I was isolated byimmunoabsorption from CHO cells transfected with the apo A-Iwt cDNA after 24 hours growth inserum free medium. Isomorphic forms of apo A-I were resolved by IEF and detected by immunoblotanalysis. Proapo A-I (pro) is 25% and mature apo A-I (m) is 75% of the staining material as assessedby scanning densitometry. Cathode and anode are indicated by (-) and (+) symbols respectively.84I-pro—min both medium and cells. Approximately 2/3 of the apo A-I which was recovered from the dish wasextracellular, suggesting that secretion was more efficient than in our other expression systems.Cellular apo A-I species were analysed by isolelectric focusing with immunoblot detection whichconfirmed that the apo A-I produced by transfected CHO cells was fully processed to the mature protein(Figure 22B). Scanning densitometay of the immunostained filter indicated that approximately 75% ofthe intracellular apo A-I was fully processed to the mature apo A-I providing evidence that, in this modelsystem, propeptide proteolysis is an intracellular event. The remaining 25% of intracellular apo A-I wasproapo A-I and no preproapo A-I was evident.3.6. DISCUSSIONThe first major objective of this thesis was to establish eucaryotic expression systems for theproduction of recombinant human apo A-I in sufficient quantity for functional studies. My approach wasto evaluate each system for its ability to synthesize and secrete apo A-I and to assess the function of therecombinant as sufficient quantity became available. Four eucaryotic expression systems have beenemployed: in vitro translation in rabbit reticulocyte lysate, transient expression in COS cells, and stabletransfection of both BHK and CHO cells.Sequencing of the apo A-I cDNA established that this fragment encoded the entire apo A-Iprecursor and that it did not contain upstream sequences that could initiate translation. Subsequent invitro translation studies established that the construct contained appropriate information for translationinitiation and ER translocation of apo A-I in rabbit reticulocytes. This analysis also verified that thethree molecular forms of human apo A-I: preproapo A-I, proapo A-I and mature apo A-I could bedifferentiated by their gel electrophoretic mobility following translation and subsequent proteolyticprocessing.The cellular transport of apo A-I after ER translocation was initially investigated in transientlytransfected COS cells. Apo A-I secretion from these cells was very slow and the majority of the apo A-Isynthesized remained intracellular. Since only proapo A-I was found in the medium of these cultures, itwas concluded that signal peptide hydrolysis was necessary for apo A-I secretion. Evidence was obtainedfor a large intracellular pool of apo A-I following transfection and for rapid degradation of newly85synthesized apo A-I. This suggested that more than one kinetic pool of apo A-I was present in thetransfected COS cell. Furthermore, a portion of the cellular apo A-I was preproapo A-I, indicating thatsignal peptide processing was incomplete. Part of the explanation for this finding may be that the rate ofprocessing is affected by the level of apo A-I expression achieved in transient transfection. Stoffel andBinczek (1991) have recently described the expression of human apo A-I in COS cells using a minigeneexpression construct which retained the apo A-I promoter region under the control of the CMVpromoter-enhancer. A mutant construct which resisted signal peptidase action was also described. Theirstudies showed that, as indicated here, cleavage of the signal peptide was necessary for secretion of apoA-I. However, in contrast to the results presented here, the signal peptide of the wild type protein wascompletely hydrolysed when expressed in this construct. Incomplete signal peptide hydrolysis has beenpreviously demonstrated using apo A-I minigene constructs in C127 cells (Roghani and Zaimis, 1988).Therefore, the presence of an unprocessed signal sequence does not appear to be unique to the COS cellsystem. The evidence from this and other studies suggests that apo A-I signal peptide retention may berelated to the intracellular level of protein. A high level of apo A-I synthesis may saturate the capacity ofthe cell to efficiently hydrolyse the signal peptide and/or translocate the protein at the ER membrane.The slow rate of apo A-I secretion from COS cells may, in part, reflect slow transport out of theER. However, proapo A-I is still a large proportion of the total protein secreted by these cells (30%),suggesting that transport through other compartments might also be slowed as a consequence of highlevel apo A-I expression, or as a consequence of the transfection itself. Although apo A-I synthesis wasextensive, degradation was the eventual fate of much of the apo A-I produced. Previous reports of apoA-I expression have not addressed this alternate fate of the expressed apo A-I. Many of these expressionsystems did not provide sufficient apo A-I for functional analysis, perhaps because the majority wasdegraded intracellularly. Transient transfection could not provide sufficient material for structurefunction studies, since the secretion of the protein was so inefficient, especially in the absence of serum.This sytem did provide a rapid means of identifying cellular and secreted apo A-I species at theanalytical level.The assembly of HDL precursor lipoproteins has not been conclusively demonstrated withincells (Vance and Vance, 1990). Only apo B containing lipoproteins have been shown to assemble86intracellularly and only in hepatocyte systems. The remaining apolipoproteins are believed to be secretedin soluble form and assemble into lipoproteins outside of the cell. Since COS cells do not normallyproduce any of the soluble apolipoproteins, they represent a potential model for testing the ability of apoA-I to form lipoprotein in the absence of other apolipoproteins. Secreted proapo A-I associatedexclusively with HDL indicating high affinity for this lipoprotein fraction, in agreement with other studies(Dixon et a!, 1989; Hussain et a!, 1991). In contrast to Hussain et a! (1991), however, we did not find apoA-I associated with L,DL, although this may reflect the low levels of this lipoprotein in fetal bovineserum. Our inability to isolate apo A-I complexes from serum-free medium reflects both the low quantitysecreted and the large losses in our attempts to isolate them in concentrated form.The most striking finding in our attempts to stimulate apo A-I secretion from COS cells was theeffect of human lipoproteins. The observed effects of the lipoproteins might be explained by theiropposing influences on cellular lipid. LDL, which delivers cholesterol to cells, stimulated apo A-Isecretion, whereas HDL, which promotes cholesterol efflux, reduced apo A-I secretion by the transfectedcells. COS cells, in contrast to hepatocytes, appear to have an extremely limited capacity to synthesizethe lipid components required for lipoprotein assembly and vesicular transport. The inability to secretelarge quantities of apo A-I, particularly in serum free medium, may indicate that the cells ability toassimilate lipid has been exhausted by the high level of apo A-I synthesis. Once this has occurred, thecell may be incapable of further secretion and the apo A-I may then be degraded. Similar results mightbe expected from other extra-hepatic cells in culture and is supported by the observations of LamonFava et a! (1987) who noted that more apo A-I was associated with lipid if the cells were preincubatedwith a phospholipid emulsion.The covalent phosphorylation of apo A-I is of interest but its biological significance remainsunclear. We could demonstrate phosphorylation in COS cells in spite of the low level of secretion. It maybe that this process is related to events other than secretion, perhaps as a covalent marker forintracellular degradation (Stadtman, 1990).Cellular retention of the large majority of apo A-I radiolabel prompted an assessment of thesubcellular localization of apo A-I in the transfected cells. Since apo A-I does not undergo posttranslational glycosylation, lectin markers for these organelles will not reflect the location of the87overexpressed protein but rather the location of other, glycosylated proteins within the organelle. Apo AI accumulations were found in both the ER and in the Golgi apparatus of transfected COS cells.Localization to the ER can be accounted for by the impaired translocation. It was suspected, but notverified experimentally, that the differences in staining pattern represented cell-to-cell differences in thelevel of apo Al expression. In general, however, the immunotluorescence pattern was consistent with thebiochemical observations, suggesting marked expansion of the ER. The level of apo A-I accumulationwas sufficient to impair ER function markedly, since the hydrolysis of the signal peptide was incompleteand the rate of secretion was slow. We could not exclude the possibility that the observations reflectedthe net consequence of two distinct transfected cell populations: one which accumulates apo A-Iexclusively in the ER and is then incapable of further signal peptide hydrolysis, and a second whichdisplays apo A-I fluorescence in the Golgi apparatus and processes and secretes the protein efficiently.This cell to cell heterogeneity may reflect either heterogeneous expression of apo A-I or varyingtransfection efficiencies.BHK cells stably transfected with the apo A-I cDNA provided several advantages over thetransiently transfected COS cell. Selected BHK clones expressed apo A-I to high levels since geneamplification occurs in the region of the DHFR gene under the influence of methotrexate (Shimke et at,1987). The apo A-I cDNA will coamplify under these conditions. However, even with high levelexpression, signal peptide hydrolysis in these cells was complete. Intracellular apo A-I was proapo A-I,indicating that propeptidase activity was inefficient or not present within these cells and this observationprovided the opportunity to study the role of the propeptide on cellular transport and processing of apoA-I (see Chapter 4).Although apo A-I transcription was under the influence of the mMT-I promoter in BHK cells,induction conditions did not increase the quantity of apo A-I recovered in the medium unless FBS waspresent. This provided additional evidence for the role of serum factors, possibly lipids, in the regulationof apo A-I synthesis and secretion. Under optimal conditions, apo A-I was secreted rapidly for 2-4 hours,but slowed markedly thereafter, when degradation was increasingly apparent.Immunofluorescence studies indicated that the cellular apo A-I in BHK was Golgi-associated,reflecting more efficient secretion than in COS cells. These findings were supported by pulse-chase data.88Since the fluorescence was confined to the Golgi, we believe that the site of apo A-I degradation in thesecells is at or distal to this organelle in the constitutive secretory pathway.Despite the improvements in the efficiency of cellular processing and secretion in the BHKexpression system, these cells were less satisfactory for large scale production of recombinant proteinsince, under serum free conditions, secretion was essentially undetectable and cell viability was markedlyreduced.We turned to the CHO cell as a cultured cell with well documented growth characteristics indefined serum free medium. Stably transfected cells expressing apo A-I from the cDNA encodingpreproapo A-I contained predominantly mature apo A-I. Only a minor component of the intracellularapo A-I pool retained the propeptide and preproapo A-I was not detected. This is the firstdemonstration of intracellular apo A-I propeptide hydrolysis. Although it has previously been shown thatCHO cells do catalyse the conversion process (Mallory et al, 1987), it was presumed to be extracellular.The overall expression level in CHO was lower than in BHK since the selection conditions did notpromote gene amplification. CHO cells produced approximately 25-30% of the apo A-I mass of thecorresponding BHK clone. However, secretion was much more efficient in CHO since approximately 2/3of the apo A-I in each dish was found in the medium. The secretion efficiency was also evident in serumfree conditions, and suggested that intracellular propeptide hydrolysis may be an additional factorpromoting secretion of apo A-I from other cells (eg. hepatocytes) which are known to contain thisactivity. This expression system allowed for the production of apo A-I under serum free conditions for 5-7 days, without appreciable loss of secretion efficiency or cell viability. We therefore chose to producewild-type and mutant apo A-I in CHO cells for further functional studies (Chapter 5).In conclusion, these studies showed that apo A-I was expressed from the cDNA as preproapo AI and was translocated across the ER membrane coincident with the prepeptide hydrolysis to formproapo A-I. Signal peptide hydrolysis was required for secretion from cultured cells expressing the apoA-Iwt cDNA. Transiently expressing COS cells synthesized relatively large quantities of apo A-I(sufficient for western blot detection), but secreted only small amounts of the protein (5-10 ng/ml/24hours). In addition, signal peptide hydrolysis of the intracellular pooi of apo A-I was incomplete.Evidence was provided indicating that efficient secretion of apo A-I may be influenced by the provision89of lipoprotein to the cells. Interestingly, LDL was the most stimulatory of the lipoproreins, perhapssuggesting that lipoprotein internalization and provision of intracellular membrane lipid might enhanceapo A-I secretion. When the same cDNA construct was incorporated by stable transfection into BHKcells, a cell line expressing proapo A-I was obtained. The isolation of this cell line provided a modelsystem for investigation of the role of the apo A-I propeptide during intracellular trafficking andconstitutive secretion. Finally, CHO cells which had incorporated the apo A-Iwt cDNA were also isolatedand it was demonstrated that these cells, grown in serum free medium, expressed as much as 130ng/ml/24 hours of apo A-I. These cells hydrolysed the apo A-I propeptide as an intracellularmodification, thus secreting only mature apo A-I. This system was subsequently utilized for theproduction of mutant recombinant apo A-I (Chapter 5).904. THE APO A-I PROPEPTIDE AND INTRACELLULAR TRAFFICKING4.1. OVERVIEWThe functional role of the apo A-I propeptide in intracellular movement and secretion of theprotein was investigated by expressing the human cDNA with or without the propeptide segment ineucaryotic cells. Deletion of the propeptide coding segment from the cDNA encoding preproapo A-I wasachieved by oligonucleotide-directed mutagenesis and was confirmed by sequencing of the altered DNA.The modified segment was then subcloned into the eucaryotic expression vectors pCMV5 and pNUT fortransient and stable transfection studies, respectively. COS cells were used as host cultures for transienttransfection studies and BHK cells were used to establish stable lines expressing the respective cDNAconstructs.4.2. RESULTS4.2.1. Transient Expression of AIdPRO in COS CellsInitial investigations of the role of the apo A-I propeptide were performed in transientlytransfected COS cells. In these experiments, transfected cells were pulse-labelled with [35 S]methioninefor 30 minutes and the subsequent fate of the apo A-I radiolabel was observed during unlabelledmethionine chase incubations.Approximately one-half of the biosynthetically labelled apo A-I in cells transfected with apo AIdPRO cDNA was degraded within 12 hours of initiating the chase (Figure 23A, upper panel). In thesame experiment, the t112 of the wild type protein was approximately 8 hours. The quantity of apo A-Iradiolabel in cells transfected with the wild-type construct was lower at all time points than thecorresponding value in apo A-IdPRO. The differences, however, were not of sufficient magnitude toestablish convincingly that the propeptide had a significant effect on cellular expression of apo A-I.The intracellular apo A-I in AIdPRO transfectants represented a greater proportion of the totalapo A-I per dish than that produced by Alwt transfected cells (Figure 23A, bottom panel). Acorrespondingly smaller proportion of the total apo A-IdPRO in each dish was recovered in the mediumwhen compared with apo A-Iwt (Figure 23A, middle panel). This was the first indication that loss of thepropeptide might alter secretion of apo A-I.91A100— T:TALc— AIdPRO0-50- MEDIUM‘ I I ‘I I ‘I I!‘ 100::::::E:::::CELLSCt4C_.0— I I I0 8 16 24CHASE TIME(hrs)Figure 23A. Removal of the propeptide from the apo A-I sequence delays its secretion from transfectedCOS cells. COS cells were transfected with cDNA for apo A-Iwt (AIwt) or apo A-IdPRO (AIdPRO) andpulse labeled with [35 Slmethionine. Cells and media were recovered after the indicated chase interval andapo A-I was isolated by immunoabsorption. Apo A-I radiolabel was quantitated by liquid scintillationand expressed as percent of apo A-I radioactivity per dish (TOTAL) at the initiation of the chase (t = 0),or as percent of the apo A-I cpm per dish recovered in the MEDIUM or the CELLS. Apo A-IdPRO wassecreted more slowly than apo A-Iwt.92AIwt AIdPRO0 2 4 8 12 16 24 0 2 4 8 12 16 24 0 2 4 8 12 16 24 0 2 4 8 12 16 2466.2- 66.2-46-M 31 M 31-_____- ——----1.• 1 21.5-21.5-JLMEDIUM CELLS MEDIUM CELLSFigure 23B. Removal of the propeptide from the apo A-I sequence delays its secretion from transfecledCOS cells. Immunoisolates were analysed by SDS-PAGE with autoradiography and showed thatMEDIUM apo A-I was a single species in both transfectants. In both transfection experiments, CELLScontained an apo A-I doublet. Numbers along the top of each autoradiogram represent chase time inhours. Autoradiograms were overexposed to permit visualization of apo A-I in the medium. Thisexposure obscures the doublet present in cell lysates of AIdPRO, which were evident at shorter exposuretimes.93CIEF— —ø +SDS+31 KdAIwt31 KdW0AIdPROFigure 23C. Removal of the propeptide from the apo A-I sequence delays its secretion from transfectedCOS cells. Pulse-labelled cell lysates were mixed with purified plasma apo A-I, resolved by two-dimensional gel electrophoresis and visualized by Coomassie Blue staining and autoradiography.Horizontal and vertical arrows indicate the direction of IEF and SDS-PAGE, respectively. The positionof mature plasma apo A-I marker is indicated on each autoradiogram (dotted circle). The position of a3lkD molecular mass marker in the SDS-PAGE dimension is show to the left of each panel.94Immunoisolates from cell lysates and media were also analysed by SDS-PAGE andautoradiography. As indicated in Chapter 3, two apo A-I species, differing in molecular mass byapproximately 2000 daltons, were evident in both the AIwt and AIdPRO transfected cells. Only thesmaller species was detected in the medium of either transfectant (Figure 23B). The band intensity in thecell lysates was consistent with secretion or degradation of apo A-I from the cellular pool during thechase interval. However, since the intensity of the apo A-I band in the medium did not increasecorrespondingly with time, the results suggested that degradation was much more predominant thansecretion. This was also evident in the quatitative analysis (Figure 23A). Apo A-IdPRO accumulated inthe medium more slowly than did apo A-Iwt and suggested that apo A-I secretion was impaired by theabsence of the propeptide segment.The intracellular apo A-I species in these studies were further resolved by two-dimensional gelelectrophoretic analysis of radiolabelLed cell Lysate (Figure 23C). The smaller apo A-I species in cellstrarisfected with AIdPRO comigrated with mature plasma apo A-I used as electrophoretic marker. Thelarger, more basic form of apo A-IdPRO displayed the size and charge characteristics expected forpre(dPRO)apo A-I. This species had higher molecular mass and was more basic than the mature apo AI. This suggested that, like the AIwt expression product analysed in parallel transfections, AIdPROtransfected cells contained some primary translation product from which the signal peptide wasincompletely hydrolysed. Thus, in COS cells transfected with either apo A-I cDNA, a significant portionof the apo A-I produced retains its signal sequence, suggesting a functional abnormality of the ER in thetransfected cells.Immunofluorescence analysis was used to compare the cellular location of apo A-IdPRO andapo A-Iwt accumulations. As noted in earlier experiments (see Figure 17), cells expressing apo A-Iwtexhibited low level immunofluorescence throughout the cytoplasm, but greater intensity was found in thejuxtariuclear region (Figure 24A). The low level fluorescence pattern was coincident with the lectinmarker CoriA (Figure 24B), suggesting that at least a portion of the expressed apo A-I was retained inthe ER. Cells expressing apo A-IdPRO displayed much more intense fluorescence and a consistentlyreticular distribution (Figure 24C). The relative fluorescence intensity suggested that AIdPRO cellscontained more apo A-I than AIwt cells. The immunofluorescence in AIdPRO colocalized with the95Figure 24. Apo A-IdPRO is retained in the ER of the transfected COS cell. COS cells were transfectedwith pCMV5-AIwt (A,B) or with pCMV5-AIdPRO (C,D) and processed for intracellular apo A-Iimmunofluorescence (A,C). Preparations were counterstained with TRITC-labeled ConA (B,D) todemonstrate the extent of endoplasmic reticulum. Apo A-Iwt was found predominantly in thejuxtanuclear region (A) and did not colocalize with the ER marker. Apo A-IdPRO was distributedthroughout the cytoplasm (C) and colocalized with the ER marker. Bar represents lQu m.96ConA marker for the ER (Figure 24D). Therefore, the immunofluorescence analysis suggested that apoAIdPRO is retained to a greater extent in COS cells than apo AIwt and that the site of retention was theER. The findings were consistent with the biochemical data and indicated that secretion of apoAIdPROwas slower than apo A-Iwt. However, cells transfected with wild type or dPRO cDNA demonstratedconsiderable ER functional impairment. We could not ascertain whether the observed changes were dueto the loss of the propeptide, or due to the method of transfection.4.2.2. Stable expression of apo A-hIPRO in BHK cellsObservations in the transient transfections suggested that the loss of the apo A-I propeptidemight reduce the efficiency of its cellular transport and secretion. However, the results of expression ofthe wild type protein in this experimental model suggested seriously impaired cell function. To separatethe influences of the experimental model and the influences of apo A-I structure, we established BHKcell lines which stably expressed the human apo A-IdPRO sequence.Approximately 30 BilK clones were generated following calcium phosphate transfection withpNUT-A-IdPRO. These clones were screened for the ability to secrete apo A-I by immunoblot analysis.One third of all clones secreted detectable levels of apo A-I. The highest level of apo A-I secretion wasfound in line BHK-AJdPRO-C4 which was selected for detailed analysis. The expression of apo A-I bythis cell line was compared to BHK-AIwt-B5, which expresses apo A-I from the cDNA encodingpreproapo A-I.The molecular species of apo A-I produced by BHK-AIdPRO-C4 was initially established byelectrophoretic analysis of immunoabsorbed material from 1’ Simethionine pulse-labelled cells. Theautoradiogram in Figure 25A indicated that each cell line produced a single apo A-I species with M ofapproximately 28,000 daltons. This was in striking contrast to COS cell transient transfectants whichcontained two distinct molecular species. The apo A-I in BHK-AIdPRO-C4 cells (rn-AT in Figure 25)appeared slightly smaller than in line BHK-AIwt-B5 (p-Al in Figure 25). Two-dimensionalelectrophoretic analysis by the OFarrell technique (Figure 25B) indicated (hat the apo A-I produced ineach case was also homogeneous with respect to charge. As predicted from the cDNA sequence, the apoA-I expressed by clone BHK-AIdPRO-C4 (m-AI) co-migrated with the mature plasma apo A-I (Figure97A B IEFSDS+925662 —45M 31- 31- 3121.5-p-Al rn-Al p-Al rn-AlFigure 25. BHK clones contain proapo A-I (p-A!) or mature apo A-I (rn-Al). BHK clones expressing theprecursor cDNA with (pNUT-AIwt) or without (pNUT-AIdPRO) the propeptide segment were pulse-labelled with [ S] methionine. Apo A-I was immunoabsorbed from cell lysates, mixed with purifiedplasma apo A-I and fractionated by SDS-PAGE in 15% polyacrylamide (panel A) or by two dimensionalgel electrophoresis (panel B). Gels were stained with Coommassie Blue to identify the mature plasmaapo A-I (dotted circle) and processed for autoradiography to visualize the endogenously labelled BHKcell product. The apo A-I synthesized by the cells containing the full-length precursor cDNA (p-Al)migrates as proapo A-I. Cells expressing the apo AIdPRO cDNA synthesize apo A-I comigrating withthe mature form isolated from plasma (rn-Al). The position of the mature apo A-I marker was orientedby overlay of the autoradiogram on the stained gel.9825B). Amino-terminal sequence analysis was performed on the protein purified from serum-free culturemedium of this cell line. The first six residues of m-AI protein were identical to the apo A-I purifiedfrom human plasma. No other amino terminal sequences were detected. These results established thatremoving the propeptide coding sequence does not significantly affect the rate or fidelity of signalpeptide hydrolysis and that BHK-AIdPRO-C4 secretes only mature apo A-I. This cell line couldtherefore be used to investigate the consequences of propeptide deletion on apo A-I trafficking andsecretion in BHK cells. Northern blot analysis indicated that cells expressing the AIdPRO constructcontained 2-2.5 fold more apo A-I rnRNA than cells expressing the wild-type cDNA. For clarity, BHKcells expressing proapo A-I (clone BHK-AIwt-B5) were called p-AT, while BHK cells expressing matureapo A-I (clone BHK-AIdPRO-C4) were called m-AJ. This nomenclature was used to identify both thecell line and the apo A-I species it produced.4.2.3. Comparison of Apo Al Synthesis and Secretion Rates in BHK clones p-Al and rn-A!.Apo A-I and protein synthetic rates were measured in p-Al and rn-Al cells from short termlabelling withr5S]methionine. Radiolabel incorporation into apo A-I and TCA precipitable protein ofcell lysates was approximately linear over the 30 mm study period in the cell lines (Figure 26). The initialrates of apo A-I synthesis did not differ between p-Al and m-AI (Figure 26A), nor did total proteinsynthetic rate (Figure 26B). However, the mass of apo A-I in the cell lysates was much lower for p-Al(379 ± 28 ng/dish, n= 3) than for m-AI (645 ± 14 ng!dish, n 3). Therefore, the presence or absence ofthe propeptide segment affects the cellular level of apo A-I in the BHK clones without affecting the rateof synthesis. The quantity of apo A-I radiolabel secreted into the medium during these short termexperiments was too low for reliable estimation.To ascertain whether differences in the rate of apo A-I secretion might account for thedifferences in cellular apo A-I mass, we determined the levels of cellular and medium [Simethioninelabelled apo A-I during twelve hour labelling experiments. Apo A-I secretion into the medium was linearand the rate did not differ between p-Al and rn-Al (Figure 27A). Therefore, the observed differences incellular apo A-I mass were not due to marked differences in apo A-I secretion. However, as with themass measurements, more radiolabelled apo A-I accumulated in rn-Al cells than in p-AT cells (Figure99I(1)0cLC)TIME(mns)Figure 26. Apo A-I and protein synthetic rates in BilK lines p-Al and rn-Al. Cell lines rn-Al (•) or pAl O) were incubated for the indicated time with [35 Slmethionine and harvested by cell lysis.Incorporation of radiolabel into apo A-I (A) was measured following immunoabsorption and SDS-PAGEas described in MATERIALS AND METHODS. Incorporation of S into total protein (B) wasmeasured by TCA precipitation. Data points represent mean ± S.D. of triplicate dishes.64204bx00 10 20 3010032EDIUM,,,,/TI ME(hours)Figure 27. Apo A-I secretion and accumulation in BHK lines p-Al (0) and rn-Al (•). BHK cellsexpressing p-AT or rn-Al were continuously labelled with Simethionine (lOOuM). Apo A-I radioactivitysecreted into the medium (A) or retained by the cells (B) was determined by liquid scintillation countingafter immunoabsorption and SDS-PAGE resolution as described in MATERIALS AN]) METHODS.Radiolabel in total cell protein (C) was determined by TCA precipitation. Each data point representsmean ± S.D. of triplicate dishes.10127B). The difference between the cell lines was selective, since radiolabel in total cell protein followedthe same time course in the two cell lines (Figure 27C). Since the initial rates of synthesis were identicaland secretion rates were not different, the observed differences in cellular apo A-I might be accountedfor by selective degradation of proapo A-I in line p-Al.Degradation and secretion were examined in 12 hour chase incubations after 30 mm pulseincorporation of [3 S]methionine. Figure 28 shows the distribution of apo A-I radioactivity between cellsand medium during the chase period. Proapo A-I was released into the medium more rapidly thanmature apo A-I in the early portion of the chase (Figure 28A, upper panel). After two hours chase therewere minimal differences in the absolute quantity of [35 S]apo A-I recovered in the medium of the twocell lines, but since less radioactivity had incorporated into p-Al during the pulse, the portion secretedwas greater than in rn-Al (80% vs 20%). Medium apo A-I radiolabel decreased in both cell lines after 4hours chase. The cellular residence time of p-Al (30-40mins) was much shorter than m-AI(approximately 3 hours; Figure 28A, lower panel) due to more rapid secretion of p-Al early in the chase.However, the recovery of apo A-I as a percentage of total apo A-I radioactivity was lower (by 15-20%)from p-Al than from rn-Al during the first four hours, suggesting that p-Al was degraded more rapidly.At chase times longer than four hours, the unaccounted portion of radiolabel (degraded) did not differbetween the two cell lines. No partial degradation products were detected by autoradiographicassessment of cells or medium of either cell line following SDS-PAGE (Figure 28B). However,proteolysis may have released products which were too small for recognition by the polyclonal antibodyused. The more rapid disappearance of proapo A-I compared with mature apo A-I suggested that thesusceptibility to intracellular degradation might be related to the presence of the propeptide.Alternatively, changes in the distribution of apo A-I among subcellular compartments, as a consequenceof the propeptide deletion, may have delayed degradation of the rn-Al protein.4.2.4. Intracellular Localization of Apo AT Accumulations in BHK Clones.Immunofluorescence analysis was used to localize the apo A-I accumulations within p-Al and mAl cell lines. The distributions of proapo A-I (in cell line p-Al) and mature apo A-I (in cell line rn-AT)differed markedly (Figure 29). In p-Al cells, apo A-I fluorescence decorated the juxtanuclear region102Figure 28. Time course of apo A-I degradation and secretion in BHK cell clones expressing apo A-IcDNAs. Cells were pulse-labelled with [35 Sjmethionine and chase incubated with medium containing2mM methionine. After the indicated chase time, cells and medium were harvested and apo A-I wasisolated by immunoabsorption.A- Immunoisolates were mixed with purified plasma apo A-I, resolved by SDS-PAGE and the apo A-Iband excised from the gel and counted. Each point represents the mean ± S.D. of triplicate dishes.Where error bars are not visible, they are within the margins of the symbol.B- Aliquots were resolved by SDS-PAGE and detected by autoradiography. Chase time is indicatedabove or below the lane between the autoradiographic panels. S = Apparent molecular mass markers, inkilodaltons (kDa). p-Al O) and rn-Al •) indicate BHK cells expressing proapo A-I or mature apo A-I,respectively.A.17(-)2B.____p-Al rn-Al kDaMEDIUM-45-3121.5SCHASE 0 1 2 4 8 12 0 1 2 4 8 12(hrs)CELLSTIME(hours)—- — — -92.566.24531- 21.5S103Figure 29. Immunofluorescence localization of apo A-I in transfected BHK cell lines. Apo A-Iexpression was induced with 100 i M zinc sulfate for four hours prior to cell fixation. Cells werepermeabilized with Triton X-100 and incubated sequentially with mouse anti-human apo A-I (6B8) andFITC-goat anti-mouse IgG. Golgi zones were stained with TRITC-conjugated WGA (panel B) and ERwas visualized with TRITC-ConA (panel D). Preparations A and B were transfected with pNUT-AI (cellline p-Al). Preparations C and D were transfected with pNUT-AIdPRO (cell line rn-Al). A and C werephotographed for visualization of FITC (apo A-I), B and D represent TRITC-conjugated lectin. Barrepresents lQum.104(Figure 29A), and co-localized with the lectin marker WGA (Figure 29B). This lectin identifies theterminal carbohydrate residues which are added to glycoproteins in the Golgi apparatus. Conversely, inrn-Al cells, apo A-I fluorescence was found in more numerous, intense foci near the cell periphery(Figure 29C). These structures stained with the ER marker Con A (Figure 29D) which identifiesmarmose-rich structures. However, the characteristic reticular distribution of the ER was not observed inthese cells. The cytologic observations provided further evidence that removing the propeptide from apoA-I altered the rate of its intracellular transport, perhaps by slowing its movement from the ER to theGolgi.To delineate further the subcellular location of apo A-I in the BHK clones, the cells wereprocessed for immunogold electron microscopy. Cryo-ultramicrotomy with immunogold labellingdemonstrated that proapo A-I was present in the Golgi regions of line p-Al (Figure 30A). Gold label,indicating the presence of apo A-I, was found in the stacked membrane structures characteristic of thedictyosome. No unusual structures were observed in p-AT cells viewed at low magnification (Figure 30A,inset). In contrast, rn-AT cells contained numerous electron dense structures which were visible at lowmagnification (Figure 30B, inset) and were distributed in a pattern similar to the apo A-I depositsobserved at the light microscopic level by immunofluorescence. At higher magnification (Figure 30B),these structures were found to contain immunogold label and were surrounded by a membrane. Controlsections prepared from either cell line and incubated with non-immune serum showed negligable goldlabelling.While neither the immunofluorescence nor the immunogold studies alone are conclusive, takentogether, the light and electron microscopic findings suggest that the organelles containing apo A-I inline rn-AT may derive from the ER.4.3. DISCUSSIONThe synthesis and secretion of recombinant human apo A-I has been studied in a number ofeucaryotic cell systems (Mallory et at, 1987; Fennewald et al, 1988; Lamon-Fava et at, 1987; Ruiz-Opazoet at, 1988; Roghani and Zannis, 1988; Forte et at, 1990). Secretion levels have been low in many celltypes, particularly under serum free conditions. However, the intracellular location and mechanism of105Figure 30A. Immunogold localization of apo A-I in ultrathin cryosections of transfected BHK cell lines.Apo A-I expression was induced by 4 hour incubation in 1O M zinc sulfate prior to glutaraldehydefixation. Ultrathin cryosections were prepared and apo A-I was labelled by sequential incubation of thegrids with mouse monoclonal antibody 6B8 and goat-anti-mouse IgG conjugated with lOmn colloidalgold. Control sections incubated with normal mouse serum in place of 6B8 showed negligible goldlabeling in either p-AT or rn-Al cells. The micrograph indicates 10 mu gold accumulations in Golgi (G)zones of cell line p-A! (pNUT-A1). Inset shows low magnification of a single p-AT cell. Arrowheadsindicate the location of lOnm gold deposits. Arrow shows the stacked membrane structures of the Golgiapparatus. Bar corresponds to 0.5 m in high magnification view and u m in low magnification inset.106I.F-.ajFigure 308. Immunogold localization of apo A-I in ultrathin cryosections of transfected BHK cell lines.Large, membrane bound vesicles in the periphery of cells of line rn-A! (pNUT-AIdPRO) labelledextensively with lOimi gold (arrowheads). The lack of gold particles over the mitochondria (M) confirmsthe specificity of the immunolabelling for the unusual vesicular structures. Arrow indicates the membranesurrounding the apo A-I accumulations. Inset shows low magnification view of a single rn-AT cell. Barcorresponds to 0.5w m in high magnification view and i m in low magnification inset.107accumulation have not been addressed. Except in CHO cells (Mallory et a!, 1987; Forte et at, 1990), theeucaryotic cells expressing the full length cDNA secrete proapo A-I. This is consistent with the generallyaccepted view that the physiologic conversion of proapo A-I to mature apo A-I occurs in the plasma(Scanu, 1987). However, the function of the propeptide and its role in apolipoprotein metabolism havenot been established.In COS cells transfected with the apo AIdPRO cDNA, I found that the altered gene productwas secreted more slowly and was also somewhat more resistant to degradation. Apo A-IdPRO wasretained by the COS cells to a greater extent than apo A-Iwt, and the intracellular structures containingapo A-IdPRO resembled the ER by lectin staining criteria. However, these observations were equivocalsince even cells expressing the wild-type protein demonstrated ER abnormalities.I then demonstrated that BHK cells expressing apo A-I do not have evidence of the ERfunctional impairment found in COS cell transfectants. These cells completely hydiolysed the signalpeptide of apo A-IdPRO at the predicted site, releasing mature apo A-I into the ER lumencotranslationally. These cells did not hydrolyse the propeptide intracellularly as assessed in pulse-labellingstudies. However, serum free collection.s of the apo A-I produced by p-Al cells did contain detectablemature apo A-I (20%) by N-terminal sequence analysis. The BHK cell was used to model the synthesis,trafficking and secretion of proapo A-I by eucaryotic cells which constitutively secrete apo A-I.Trafficking and secretion of apo A-Iwt and apo A-IdPRO has been previously investigated inAtT-20 cells (Fennewald et at, 1988). However, the accumulation of mature apo A-I in cells expressingthe AIdPRO construct (rn-Al) was not reported. My data indicates that the absence of the propeptidealters the rate of cellular transport of apo A-I in BHK cells.The studies described here also demonstrated some differences in the level of degradation ofproapo A-I compared with apo AIdPRO in BHK cells. Cellular proapo A-I was found in the Golgiapparatus suggesting that processing by this organelle may precede degradation. Post-translationalmodifications of this apoprotein have been described (Hoeg et at, 1986; Beg et al, 1989) which may playa role in apo A-I secretion. Based on observations in BHK cells, secretion may be a minor fate of apoA-I expressed by non-hepatic cells. An important, and perhaps, principal fate may be degradation.The accumulation of apo A-IdPRO in cellular structures that differ morphologically from the108Golgi apparatus, and which retain some properties of the ER, suggests that the deletion of thepropeptide sequence might affect the rate of transport from the ER to the Golgi. Although I did not useultrastructural cytological markers to determine the subcellular origin of these membrane bound vesicles,the light microscopic findings suggest that the subcellular structures in which apo A-I accumulates in rnAl are derived from the ER. Therefore, I hypothesize that when apo A-I is synthesized as proapo A-I, itis transported through the ER into the Golgi apparatus along the constitutive secretory pathway, where itis secreted or is degraded. Engineered deletion of the propeptide, however, disrupts this transportcausing much of the apo A-I to remain in a pre-Golgi location and may slow the entry of apo A-I intothe cellular compartment in which degradation occurs. If the membrane bound structures that I observedin BHK line rn-Al are indeed derived from the ER, they may represent remnants of the organelle thathave been disrupted by accumulation of large quantities of apo A-I.The exact mechanism by which the propeptide sequence may direct intracellular trafficking ofapo A-I is not known. However, it is possible that the secondaiy structure of this region serves to mask alipid binding domain near the amino terminus, thus reducing the potential for the apo A-I to bind to themembranes of intracellular organelles. Alternatively, the propeptide may act as a recognition sequencefor a process which is sensitive to the lipid status of the cell. When cellular lipid is limiting, for instancein a non-hepatic cell grown in lipid poor medium, much of the protein may be degraded. Since theabsence of the propetide sequence appears to delay transport early in the secretory pathway, I predictthat the hydrolysis of the propeptide sequence might occur during the latter stages of secretion, but priorto exit from the cell. This is consistent with evidence from CHO cells where cellular processing has beendemonstrated (Mallory et a!, 1987; Forte et at, 1990). However, the specificity of this hydrolysis for theapo A-I propeptide must be questioned since CHO cells should have no need for such an activity. I havealso observed that BHK cells may also possess extremely low levels of propeptide hydrolytic activity,although as in other eucaryotic cells (Lamon-Fava et at, 1987; Ruiz-Opazo et at, 1988; Roghani andZannis, 1988) the majority of the apo A-I expressed form the precursor cDNA is secreted as proapo A-I.The role of the propetide in the processing of the apo A-I gene product has been previouslystudied by Gordon and coworkers. Initial cell-free translation studies (Folz and Gordon, 1987) suggestedthat removal of the apo A-I propeptide altered the interaction of preproapo A-I with eucaryotic signal109peptidase. However, subsequent studies in the AtT-20 cell (Fennewald el al, 1988) indicated that the apoA-I propeptide does not play a functional role in secretion of this apolipoprotein. This model system isdistinctly different from the BHK cells used in our investigations since the majority of the apo A-I inAtT-20 cells accumulated in post-Golgi dense core granules from which secretion could be stimulated bycAMP. The BilK cells described in our study are not capable of regulated secretion and thus secreteapo A-I constitutively.Recently, Kizer and Tropsha (1991) described a structural motif which directs proteins to thesecretory granules of AtT-20 cells. Their study appears to identify an amphipathic helix motif in the N-terminal sequence of proteins which are targetted to the dense core granules of this cell line. Thisstructure is extremely common in the apo A-I sequence, as predicted by the algorithms of Segrest andcolleagues (1990; 1992). If these structural elements direct proteins to the regulated secretory pathway ofthe AtT-20 cell, it is not surprising that regulated secretion would predominate in this model. Thepreponderance of amphipathic helices in apo A-I would direct apo A-I in a maimer independent of thepresence or absence of the propeptide. Conversely, our studies show that in the absence of a regulatedsecretory pathway, the apo A-I propeptide has a marked effect on cellular transport under conditionswhere apo A-I is overexpressed. Apo A-I movement from the ER to the Golgi is limited by deletion ofthis segment. This is physiologically relevant since secretion of apo A-I by hepatocytes and enterocytes isalso constitutive and expression level may increase in these cells with physiologic changes.In summary, I have established eucaryotic cell lines which synthesize human apo A-I with orwithout its propeptide segment. Removal of the propeptide did not affect the fidelity of signal peptidehydrolysis but did reduce the rate of apo A-I secretion from the cell, causing much of the protein toremain within vesicular structures which appear, by some histologic criteria, to be derived from the ER.This suggests that the propeptide may be required for efficient transport from the ER to the Golgiapparatus. Proapo A-I was degraded in BHK cell cultures and deletion of the propeptide segmentmarginally reduced its susceptibility to degradation. This may reflect direct involvement of the propeptidein the degradation mechanism or may result from limited entry of apo A-IdPRO into the compartmentin which degradation occurs. I conclude, therefore, that the apo A-I propeptide may regulate theintracellular transport and degradation of apo A-I in the constitutive secretory pathway.1105. LYS107 AJND THE C-TERMINAL AMPHIPATHIC HELIX IN LCAT ACTIVATION BY APO A-I5.1. OverviewSynthetic peptide studies have suggested that the amphipathic helices in apo A-I play animportant role in its function. In addition, the potential importance of the “mobile hinge domain” of theapo A-I polypeptide has been suggested on the basis of LCAT inhibition studies using monoclonalantibodies directed against this epitope. Chemical modification of Lys residues in purified normal apo AI alter LCAT activation if the charge on the side chain is altered, although the conditions of modificationmay profoundly affect protein structure. LCAT activation was diminished in only one naturally occuringapo A-I mutant. This mutant lacks a single lysine residue (Lys107), within the hinge domain. Theavailable evidence supports the hypothesis that the mobile hinge region, and perhaps Lys107, is involvedin the LCAT activation mechanism. However, this conclusion must be reserved, at the present time,because the most convincing studies, using the purified protein from plasma, may have utilizedpreparations contaminated with wild-type sequence.The approach chosen was to evaluate the functional importance of Lys107 deletion by producingthe protein as a recombinant in serum free eucaryotic cell culture (CHO cells) and assessing its ability toactivate recombinant LCAT in vitro. Since the alteration of this one residue is predicted to alter theorientation of single class Y amphipathic helix in a polypeptide that contains two of these structuralelements, the functional consequence of deleting an entire class Y amphipathic helix in a region of themolecule outside of the proposed hinge domain (mutant Dl, amino acid residues 220-241) was alsoinvestigated. Deletion mutagenesis of the helical region containing the hinge domain was also attempted.However, three separate mutagenic strategies have so far failed to generate this mutation.5.2. RESULTS5.2.1. Production and Identification of Mutant cDNAsMutagenesis was performed using the oligonucleotide primed site-directed approach. Potentialmutants for the Lys107 deletion were identified by restriction enzyme analysis. The deletion of the Lys107codon removes a single MboII restriction site from the apo A-I cDNA. Candidate plaques from themutagenesis experiment were selected for sequence analysis on the basis of an altered MboII restriction111fragment pattern when compared to the wild-type cDNA. DNA sequence analysis of the resulting dK107mutant confirmed the directed deletion.Candidate plaques for the Dl mutation (deletion of Pro2° to Asr?41 inclusive) were selected forsequencing based on high stringency hybridization with the mutagenic oligonucleotide. The presence ofthis deletion was also confirmed by DNA sequence analysis.Each cDNA was excised from the appropriate M13 RF DNA and subcloned into eucaryoticexpression vectors.5.2.2. Transfected COS Cells Produce the Respective Mutant ProteinsCOS cell transfection and pulse-labelling were used to identify the apo A-I produced from eachmutant cDNA. Apo A-I was concentrated by immunoabsorption from the labelled cell lysate and theresulting species were resolved electrophoretically and visualized by autoradiography. Cells transfectedwith expression vector without cDNA insert contained no apo A-I (Figure 31A, lane 1). Apo A-Iexpressed from the dK107 construct produced an apo A-I doublet (lane 3) which was indistinguishablefrom the wild-type product (lane 2). This was expected since the loss of a single amino acid should notalter SDS-PAGE mobility and since I had previously observed signal peptide retention in COStransfectants. In both apo A-IdK107 and apo A-Iwt, the larger polypeptide of the doublet represents apoA-I which has retained its signal peptide.The presence of the Lys107 deletion was evident by isoelectric focusing of the labelled product(Figure, 31B). The wild-type pattern contains two bands (lane 1), separated by one charge unit whichwas predicted from the amino acid composition of the signal sequence. Since the apo A-I signal peptidecontains a single basic residue (Lys at position -23), signal peptide hydrolysis results in a one charge unitshift to more acidic p1. Similarly, the molecular forms of apo A-I in the dK107 mutant were alsoseparated by one charge unit (lane 2). In addition both were shifted to more acidic p1 with respect to thecorresponding wild type polypeptide, reflecting the loss of positive charge. However, the difference inmobility between mutant and wild-type appeared to be somewhat less than one charge unit (based uponthe shift observed during signal peptide hydrolysis). Mixing the wild type and mutant preparations in onelane (Figure 31B, lane 3) confirmed that the loss of Ly°7 does not result in the equivalent change in p1.112A SDS/ ‘31 Kd- -123B JEF_—_-123 ±)Figure 31. Transfected COS cells express apo A-I mutants with the appropriate molecularcharacteristics. COS cells were transfected with the appropriate cDNA construct and were pulse labeledwith [35 S]methionine as described in IvL&TERIALS AND METHODS. Labelled protein was harvested bycell lysis and apo A-I concentrated by immunoabsorption. Labelled species were resolved bypolyacrylamide gel electrophoresis.A- SDS-PAGE separation of cells expressing apo A-IdK107.Apo A-I was not detected in cells transfectedwith empty vector (lane 1). Apo A-Iwt (lane 2) and apo A-IdK107 (lane 3) resolved as doublets indicatingincomplete signal peptide hydrolysis. The positions of a molecular mass standard (3lkD) and the anode(+) and cathode (-) are indicated.B- IEF analysis of apo A-IdK107.Isomorphic species of apo A-IdK107 (lane 2) are shifted cathodicallyfrom the respective apo A-Iwt species (lane 1), as expected due to the loss of a positively charged aminoacid. Mixing of apo A-Iwt and apo A-IdK107 (lane 3) suggests that Lys107 and Ly23 do not possessequivalent charge (see text for discussion). The arrow indicates the position of mature apo A-I in the pHgradient.113C123469-p46-M 30r14.3-CMFigure 31C. Transfected COS cells express apo A-I mutants with the appropriate molecularcharacteristics. SDS-PAGE analysis of mutant apo A-ID1. Apo A-ID1 (lane 2) and apo A-Iwt (lane 1)are both isolated as an intracellular doublet (C). Only the smaller form is found in the medium (M)after 2 hour chase incubation of either transfected cell population (lanes 3 and 4). Lane 3 = apo A-Iwt;lane 4 = apo A-ID1. Molecular mass (Mr) marker postions are shown at the left of the autoradiogram inkilodaltons.114This may reflect differences in secondary structure in the vicinity of Lyc23 and Lys107 affecting theultimate p1.The expression of apo A-ID 1 was also demonstrated in transfected COS cells. Biosyntheticallylabelled apo A-ID1 appeared as an intracellular doublet on SDS-PAGE (Figure 31C, lane 2). As with allother COS cell transfections, only the smaller form was found in the medium (lane 4), after 2 hour chaseincubation. By this criterion, the larger species is preproapo A-ID 1 and the smaller species is proapo AID1. Both preproapo A-ID1 and proapo A-ID1 were approximately 2 kD smaller than the correspondingwild-type species (lanes 1 and 3, respectively), reflecting the 22 amino acid deletion.Therefore, DNA sequencing of the mutant cDNAs and immunologic assessment of the proteinsexpressed in COS cells showed that the mutations had been introduced correctly.5.2.3. Development of CHO Cell Lines Expressing Apo A-IdK107 and Apo AID1The quantity of protein produced and the efficiency of secretion of apo A-I by COS cells wereinadequate for most functional studies, based on observations with apo A-Iwt. Therefore, CHO cell linesexpressing the apo A-IdK107 and apo A-ID1 proteins were established. This cell type efficiently secretedthe wild-type protein under serum free conditions (see Chapter 3.5) However, the conditions used forselection did not promote amplification of the apo A-I cDNA. Therefore, although secretion was moreefficient in CHO, the overall level of apo A-I expression was lower than in BHK. In addition, the CHOclones responded poorly to culture conditions designed to induce expression (1O M zinc sulfate). Cellsbecame rounded and refractory at approximately 5M zinc sulfate but were microscopically unaffectedwhen the concentration was reduced to 1z M. Unfortunately, the lower concentration did not improveapo A-I production.CHO cells lines expressing the dK107 and Dl mutants were selected with Geneticin and screenedfor secretion of apo A-I by ELISA. Cells expressing the double mutants dK107 /dPRO and Dl/dPROwere also generated. To establish the identity of the apo A-I species produced, medium and cell lysateswere isolated following 24 hours growth under serum free conditions. The samples were resolved by 1FFor SDS-PAGE electrophoresis and detected by western blot analysis (Figure 32).As indicated in the preliminary COS cell experiments, the cells expressing the mutant cDNA115A BdK’°7 DlpH 6 0 S wt dPRO dK107 dPRO wt Dl dPRC92.5-66.2-proAl- — .mAl- -PrOMdK 31-21.5—-- 14.4-pH4® CMC MC MCMFigure 32. CHO cells express apo A-I mutants of the expected charge and size. Selected CHO cellclones were grown to near confluence and then changed to serum free medium. After a 24 hour purgingincubation to remove bovine serum proteins, 48 hour medium collections were performed and cell lysateswere prepared. Apo A-I was concentrated by immunoabsorption and separated by gel electrophoresis.A- IEF analysis of cell lysate (C) and medium (M) preparations between pH 4 (+) and pH 6 (-). proAland proAldK107 indicate the positions of the apo A-I species with the propeptide. Note that these speciesare present only in the cell lysate (C) lanes, mAT and mAIdK107 indicate the postions of the respectivemature polypeptide forms. S = purified plasma apo A-I standard.B- SDS-PAGE analysis of medium preparations from cells expressing the apo A-ID1 (Pr20-Asi41)deletion mutant (Dl) and the apo A-ID1/dPRO double mutant (DIdPRO). Numbers to the left indicatethe mobility of molecular mass markers (in kilodaltons).116produced apo A-I protein with the charge shift appropriate for the loss of the single lysine residue.Proapo A-IdK’°7 was detected only within the cells and only mature apo A-IdI<’°7 was found in themedium of these cultures (Figure 32A). In cells transfected with the dK107 /dPRO construct, only matureapo A-IdK107 was found in cells or medium. The absence of the propeptide segment appeared to reducethe accumulation of apo A-I in the medium, but this was not a consistent finding with all mutant celllines (see Table VIII). Since only the mature form of apo A-IdK107 was found in the medium of CHOcells expressing the cDNA which included the propeptide, this cell line must also contain the apo A-Ipropeptidase activity.Western blot analysis of cells expressing the Dl and D1dPRO mutants were also consistent withthe analysis of COS transfectants. SDS-PAGE analysis of the secreted product showed that mutantprotein was of lower molecular mass than apo A-Iwt (Figure 32B). IEF analysis (data not shown)indicated that cells expressing apo A-ID 1 contained some proapo A-ID 1 but secreted only mature apoA-ID1. Thus, apo A-I propeptidase was also present in this cell line.In all mutant cell lines studied, the quantity of mutant apo A-I secreted was markedly lowerthan apo A-Iwt cell lines (see Table VIII).Table VIII. Secretion of apo A-I recombinants by CHO cell lines. Cell lines expressing apo A-I andmutants were grown to near confluence and changed to serum free medium which was then collectedafter 24 hours. The first 24 hour collection was discarded. After an additional 48 hours the medium wascollected and the apo A-I quantified by competitive ELISA as described in Materials and Methods.Values are mean ± S.D. of the number of determinations in parentheses.CHO Clone Apo A-I Secreted(ng/ml/24 hours)WI 137± 58 (5)dPRO 27± 18 (6)dK107 17± 6(6)dK107/dPRO 11± 3(5)Dl 8± 2(5)D1/dPRO 45± 1 (2)5.2.4. Role of Lys107 in Apo A-I Cellular Transport and SecretionThe consequences of the dK107 deletion on cellular transport and secretion of apo A-I wereinvestigated in transiently transfected COS cells. Quantitative analysis of immunoabsorbed radioactivity117from pulse-chase studies (Figure 33A) indicated that the dK107 mutant protein was degraded morerapidly in COS cells than the wild-type apo A-I. Less than half of the apo A-I labelled during the pulseperiod was recovered as apo A-I after 2 hours (TOTAL, upper panel). However, the portion of the apoA-I that reached the medium during the chase was greater than apo A-Iwt (Figure 33B). SDS-PAGEanalysis of the samples suggested that signal peptide hydrolysis was impaired in cells expressing dK107and occurred less efficiently in the mutant than in the wild-type transfected cells. Proapo A-IdK’°7appeared to leave the cells more rapidly than apo A-Iwt (Figure 33B). Degradation of apo A-I from thecultures was extensive, as observed in the previous transient expression studies.The lipoprotein binding properties of apo A-IdK107 were evaluated in the COS cell expressionsystem. Apo A-IdK107 was isolated from the medium in the HDL density range by gradientultracentrifugation (Figure 34). Lipoproteins containing apo A-Iwt or apo A-IdK107 were found in thesame density fractions. Thus, this mutant protein has a high affrnity for pre-existing HDL and this abilitydid not differ from the apo A-Iwt expressed in the same culture system. However, I was unable to obtainsufficient material to examine the secretion of lipoprotein particles containing apo A-IdK107 from serumfree cultures.Immunofluorescence analysis of COS cells expressing the dK107 mutant were consistent with theobservations in pulse-labelling. Apo A-I fluorescence was observed at low intensity in a reticular patternwith increasing intensity in the juxtanuclear region (Figure 35A). The latter colocalized with the Golgimarker WGA (Figure 35B). This suggested that apo A-IdK107 might reach the Golgi apparatus morerapidly than apo A-Iwt, and was consistent with the more rapid secretion ofr5Sjapo A-IdK107 observedduring the pulse-chase studies.5.2.5. Functional Characteristics of Apo A-I Mutants5.2.5.1. Lipid Binding Properties of Apo A-I MutantsRecombinant apo A-I and its mutants produced by CHO cells were used to analyse lipidbinding properties in cell culture and in vüro. Lipoprotein complexes containing apo A-I from serum freeculture media were concentrated by ultrafiltration and analysed by density gradient ultracentrifugation(Figure 36, left hand panels). A nadir of apo A-I concentration was observed in each gradient at118TOTAL•OA(wt0—0 AIdK107MEDIUM0_cCELLSUJO‘4-0_o4CHASE TIME(hrs)Figure 33A. The apo A-IdK107 mutant protein is secreted and degraded more rapidly than apo A-Iwt intransfected COS cells. COS cells were transfected with the appropriate cDNA construct and thetransfected cells subjected to pulse-chase analysis as indicated in MATERIALS AND METHODS. Atthe indicated time, medium was collected and cells were recovered by lysis. Apo A-I wasimmunoabsorbed from each fraction and analysed by liquid scintillation spectrometry. Each data pointrepresents a single dish from a typical experiment.1oo(1)IILLJ+aoe119Figure 33B. The apo A-IdK107 mutant protein is secreted and degraded more rapidly than apo A-Iwt intransfected COS cells. Immunoisolates from the pulse-chase experiment in Figure 33A were fractionatedby SDS-PAGE and visualized by autoradiography. Numbers along the top of each autoradiogramindicate the chase time in hours. Molecular mass (Mr) markers (in kilodaltons) are indicated on the left.Autoradiograms were overexposed to allow visualization of apo A-I in the medium. Intense bands in thecell lysate at this exposure obscure the doublet in some lanes.AIwt02 4 812 1624 02 4 8121624F56.2M ::21.5AidK10T0 2 4 8 12 18 24 0 2 4 8 12 16 24-___L JMEDIUM CELLS66.2-46-Mr 3121.5-MEDIUM CELLS1120O),— 4cCoQ)AIwtA1dK107a_’—-woFigure 34. Apo A-IdK107 incorporates into extracellular lipoprotein in cell culture. COS cells weretransfected with apo A-I cDNA and pulse-labelled as described in MATERIALS AND METHODS.Apo A-I was collected in DMEM/10% FBS during a 12 hour chase incubation and concentrated byultrafiltration. The concentrated samples were then applied to a linear salt gradient and separated byultracentrifugation (48 hours at 40,000 rpm in SW41Ti rotor). Centrifuge tubes were fractionated frombottom (fraction 1) to top and radiolabel quantitated in each fraction by liquid scintillation spectrometry.Aliquots of each fraction were subjected to immunoabsorption, SDS-PAGE and autoradiography toestablish the location of apo A-I radiolabel. Apo A-Iwt and apo A-IdK107 are both found in fractions ofHDL density.1211’1.3001 .2501 .200.1501.1001 .0501 .000FRACTIONFigure 35. Cells expressing apo A-IdK107 contain apo A-I accumulations in both the ER and the Golgiapparatus. COS cells expressing apo A-IdK107 were processed for apo A-I immunofluorescence asdescribed in MATERIALS AND METHODS. Apo A-I (A) was found in a reticular distribution butshowed increasing intensity in the juxtanuclear region of the cell. The most intense region ofimmunofluorescence colocalized with the TRITC-WGA marker for the Golgi apparatus (B). Barindicates 1C m.122WQc’-’Figure 36. Apo A-I and mutants are secreted from CHO cells into serum free medium in lipid-poorform but retain the ability to associate with exogenous liposomes. Recombinant apo A-I and its mutantswere collected from CHO cell cultures under serum free conditions and were concentrated byultrafiltration. Complexes formed under these conditions were fractionated by density gradientultracentrifugation (left panels, WITHOUT LPs). An equivalent portion of each collection was mixedwith lecithin:r H] cholesterol vesicles prior to ultracentrifugation (right panels, WITH LPs). Eachgradient was fractionated from bottom (fraction 1) to top and apo A-I was measured in individualfractions by competitive ELISA. r H] in each fraction was measured by liquid scintillation and densitydetermined by conductivity as described in MATERIALS AND METHODS. Each point in the lowertwo panels is the mean ± S.D. of six gradients. Apo A-I was measured in duplicate for each gradientfraction and data values represent the mean in ng per fraction.1231’WITHOUT LPs WITH LPs0UC0UCoLV-4-ioC.) II —,11FRACTIONapproximately 1.15 g/ml. 50-60% of the apo A-I mass was found in the higher density (d> 1.15 g/ml)region of the gradient indicating that most of the apo A-I was secreted in the lipid-poor or lipid-freestate. The wild-type and mutant proteins were similarly distributed among the gradient, fractions,indicating that all proteins were secreted from CHO cell culture in the lipid-poor or lipid-free state.The ability of the secreted recombinant apo A-I preparations to bind to exogenous liposomes(LPs) was determined under in vitro conditions. Medium was recovered from CHO cultures andconcentrated by ultrafiltration. Egg yolk phosphatidylcholine:cholesterol vesicles (4:1, molar ratio) wereadded and the mixture was incubated to allow apo A-I to form complexes which were then fractionatedby density gradient ultracentrifugation (Figure 36, right hand panels) for comparison with the originalculture concentrate (left hand panels). Liposomes incubated without culture supernate were found onlyin the lowest density fractions of the gradient (lower right panel). The addition of liposomes to apo AIwt sample shifted the position of wild-type apo A-I in the gradient from the highest to the lowestdensity fractions (wt, left panel compared to right panel). More than 85% of the apo A-Iwt wasrecovered with the liposomes at the lower density limit while less than 15% of apo A-Iwt remained athigh density following addition of liposomes. Therefore, apo A-I formed complexes with added lipid andthese complexes were stable under conditions of density gradient ultracentrifugation.A similar pattern was observed on addition of liposomes to apo A-IdK107.More than 80% of themutant apo A-I protein was found at lower density following incubation with LPs (dK107, right panel).However, the density of the lipid-protein complexes was not as uniform as with apo A-Iwt. Apo A-IdK107complexes were detected in the fractions ranging from 1.050-1.090 g/ml. 20% of the mutant apo A-Iremained in higher density fractions. The results suggested that dK107 formed lipid-protein complexeswith the added liposomes, but that these structures were more heterogeneous in density or were lessstable during ultracentrifugation, than those formed by apo A-Iwt.The apo A-ID1 mutant also retained the ability to form complexes with added liposomes.However, approximately 30% of the mutant protein was isolated at d> 1.15 g/ml following incubationwith liposomes (upper right panel). The remaining 70% was recovered as a homogeneous population atthe lower density limit. This suggested that apo A-ID 1 might have reduced ability to bind to theliposome or may have bound in a manner that was less stable to ultracentrifugation conditions.124In conclusion, deletion of Lys107 does not substantially alter the ability of the protein to interactwith phospholipid complexes but may produce more heterogeneous species than apo A-Iwt. Complexesformed with dK107 were more heterogeneous than those formed with either apo A-Iwt or apo A-ID 1.Furthermore, it appears that deletion of the C-terminal -helix from apo A-I (mutant Dl) might havereduced its ability to associate into stable lipid-protein complexes. While the apo A-I proteins secretedfrom CHO cells may or may not retain some cellular lipid, these proteins still possess the ability to bindto exogenous lipid substrates.5.2.5.2. LCAT Activation by Apo A-I MutantsThe functional consequences of apo A-I structural changes were tested in LCAT activationstudies using recombinant LCAT as enzyme source and lecithin:cholesterol vesicles as substrates. Theaddition of l g of purified plasma apo A-I (pAl) to a reaction containing enzyme and substrateincreased the initial rate of cholesterol esterification 95-fold (9.92 vs 0.10 nmoles/hr/ml, n=4; Figure 37).Similarly, recombinant apo A-I (rAl) increased the reaction rate 75-fold (to 7.69 nmoles/hr/ml, n=4).The interassay variation, however, was large (more than 30%), reflecting variability in liposome substratepreparations. If recombinant and plasma apo A-I preparations were analysed simultaneously and theactivation was expressed as the ratio of rAT to pAl (rAI:pAI, Figure 37 right panel), the interassaycoefficient of variation was approximately 12% (n = 4). Recombinant apo A-Iwt was 80% as efficient asan LAT activator as the purified plasma protein (rAI:pAI = 0.796 ± 0.099). This difference may reflectthe presence of a small quantity of cellular lipid in association with the recombinant protein, thusreducing the effective concentration of apo A-I on the LAT substrate particles during the assay.The ability of the dK107 and Dl mutants to activate cholesterol esterification were comparedwith apo A-Iwt and expressed as rAI:pAI ratio (Figure 38). Neither mutant protein was an effectiveLCAT activator. The mean activation ratio by dK107 (ratio = 0.040 ± 0.025, n=4) and by Dl (0.036 ±0.007, n=3) mutants were not significantly different from zero (ratio = 0.010 ± 0.012, n=6; p>O.OS inboth cases) in these experiments. The results suggested that alterations of the class Y amphipathichelices of human apo A-I seriously reduce the ability of the protein to activate LCAT.12515 1 .5>-—>U)C-)_jc:10 1 .0BL pAl rAl rAl:pAl•0.5Figure 37. Recombinant apo A-I produced by CHO cells activates LCAT to approximately the sameextent as the purified plasma protein. Activation of cholesterol esterification by apo A-I was measuredin assays containing recombinant enzyme and lecithin:cholesterol vesicles as described in MATERIALSAND METHODS. Recombinant (rAl) or purified plasma (pA!) apo A-I (1 g) was mixed with vesicles,incubated 3Omins to allow complex to formation and then incubated in the presence of enzyme. Activity(left panel) is expressed as nmoles CE formed per hr per ml rLAT. In the right panel, the activitydetermined in reactions containing recombinant apo A-Iwt (rAT) and plasma apo A-I (pAl) is expressedas a ratio. BL indicates the activity observed in the absence of apo A-I. Each histogram is mean ± S.D.(n=4).1260.8-0.6-F>----50E 0.4-FC)-J0.2-00BL WI dK107 DlFigure 38. The ability of apo A-I mutants to activate LCAT is markedly diminished. LCAT activation bythe recombinant apo A-I (1j g) was measured and expressed as the ratio of activation achieved bypurified plasma apo A-I (pAl) in parallel incubations. Histograms are mean ± S.D. of 3 different apo AI preparations measured in separate assays. Activation by the mutant proteins (dK107, Dl) was notsignificantly different from incubations without any source of apo A-I (BL).1275.3. DISCUSSIONSegrest and colleagues have recently (1990, 1992) reviewed the amphipathic helix as a structuralmotif in the plasma apolipoproteins. They have identified two distinct classes of amphipathic helices inapo A-I with different predicted properties based on computer modelling techniques (Segrest et at,1992). Class A helices are common to all of the apolipoproteins and appear to be responsible for lipidbinding. A second helix motif is also present in apo A-I and in apo A-IV (Class Y). This motif ispredicted by computer modelling to penetrate phospholipid surfaces to a lesser extent than Class A helix(Segrest et al, 1992) and may not participate in high affinity lipid association. The class Y helices in apoA-I are found in the vicinity of the putative hinge domain and near the C-terminus (see Figure 9). Theapo A-I mutations generated and expressed in vitro in my studies involved both of these regions.Only one naturally occurring apo A-I variant has been described in which a substantial portionof helix is lost (Gln146 to Arg’60, Deeb et at, 1991). Even in the heterozygous state, this abnormality wasassociated with abnormal HDL metabolism. This deletion affects a region of class A helix and suggeststhat loss of structural elements responsible for lipid binding might have a more profound influence onlipoprotein metabolism than any of the point mutations.Deletion of Lys107 is the most common mutation of the apo A-I protein (von Eckardstein et at,1990). Preliminary studies using mutant protein purified from patient’s plasma suggested that thisresidue, or the secondary structure of the protein which is maintained by this residue, are important forLCAT activation and perhaps for reverse cholesterol transport (Rall et at, 1984). The structuralalteration predicted in the mutant protein is a reorientation of the -helix in the hinge region. Such achange may be responsible for the changes observed in function (Ponsin et at, 1985). However, noconsistent lipoprotein abnormality is associated with dK107 in vivo. Further analysis of the mutant proteinisolated from plasma has been consistent with the preliminary studies (Jonas et at, 1991). However,possible contamination of the preparation with wild-type protein, and large variation among functionalproperties of purified wild-type preparations have been cited as factors hampering more definitiveanalysis of these natural mutations.The ability to produce mutant apo A-I as a recombinant has avoided the contamination withwild-type protein which occurs with plasma purified preparations and has allowed me to analyse128functional properties of two apo A-I mutants. The findings presented here indicate that the deletion ofLys107 may have even more pronounced consequences on LAT activation than previously demonstrated.Furthermore, analysis of a novel mutant protein (apo A-ID1), from which a single class Y amphipathichelix had been removed, indicated that changes in this region had a similar effect on apo A-I function.Neither alteration abolished the ability to bind to LCAT substrate particles, but both markedly decreasedthe ability to promote cholesterol esterification by LCAT. Due to limitations in the quantity ofrecombinant protein available to date, activation has been assessed only at a single concentration of apoA-I.There appear to be some differences in the stability of the complexes formed with the mutantproteins as assessed by gradient centrifugation. I cannot rule out that the reduced activation ofcholesterol esterification is due to the formation of unsuitable LCAT substrates. However, based on theobservations to date, I hypothesize that high lipid affinity (class A) amphipathic helices are important tomaintain the general structure of the lipoprotein, while class Y helices may play a more important role inHDL remodelling processes. The transformations of HDL composition will necessitate changes in apoA-I secondary structure at the lipoprotein surface. My studies suggest that deletion of Lys107 inducessufficient structural change to block LCAT activation, perhaps by preventing conformational changes inthe hinge region. This possibilty should be investigated in greater detail, including complete kineticanalysis of LCAT activation by the mutant proteins.1296. CONCLUSIONS6.1. Summary of Major FindingsThe studies described in this thesis were designed to investigate the role of structural elementsof apo A-I in the intracellular transport and extracellular functions of this apolipoprotein. Fourexpression systems were used in the course of this work to produce the wild-type protein. Three mutantshave been generated at the cDNA level and were expressed in at least two of these systems.In vitro translation sudies were used to establish that the cDNA encoded authentic apo A-Iprotein which was translocated and proteolytically processed at the endoplasmic reticulum membrane.Under the conditions employed, I was unable to demonstrate complete translocation and signal peptidehydrolysis in vitro.When the protein was expressed in non-hepatic eucaryotic cells, apo A-I was synthesizedefficiently during transient overexpression of the cDNA. Secretion of the protein was not directly relatedto the level of synthesis in standard growth conditions and was not detectable in serum free conditions. Asubstantial portion of the protein was degraded or retained by the cells in these cultures. COS cells andBHK cells trarisfected with the precursor cDNA secreted only proapo A-I, and signal peptide hydrolysiswas necessary for secretion in these systems. In COS cells, a portion of the intracellular apo A-I waspreproapo A-I, indicating that cotranslational proteolytic processing was incomplete. Proapo A-I wasreadily integrated into lipoproteins of high density (mean d = 1.15g/ml) in the culture medium of thetransfected cells.Mature apo A-I was secreted from CHO cells expressing the cDNA for the precursor protein,demonstrating that these cells possess apo A-I propeptidase activity. Approximately 75% of theintracellular apo A-I was the mature protein in serum free cultures, indicating that the hydrolysisoccurred within the cell. The apo A-I secreted from CHO cultures was found mostly in lipid-free orlipid-poor form (d> 1.15g/ml) under these culture conditions. However, the expressed protein had theability to bind avidly to exogenous lipid complexes in vitro.The level of apo A-I expression from CHO cells was lower than from BHK cells but secretionfrom these cells was far more efficient, In addition, BHK cells containing the precursor cDNA degradedapo A-I in culture. The production of apo A-I from the precursor cDNA was estimated in serum free130cultures of cells selected for maximal secretion under these conditions: COS = lOng/ml/24hr, BHK =lOOng/ml/24hr, CHO = l3Ong/mJ/24hr. Therefore, it was concluded that the CHO cell is the mostsuitable host cell for expression of apo A-I for structure-function analysis.In agreement with studies in liver cells, I observed intracellular phosphorylation of apo A-I inCOS cells. Secreted proapo A-I did not retain the [32 P]phosphate. I have provided no additionalinformation on the functional role of this post-translational modification.Perhaps the most important observation in this work was that the apo A-I propeptide wasrequired for efficient cellular transport in BHK cells. In the absence of this hexapeptide, the majority ofthe apo A-I was retained as the mature protein in the cell, although a portion of the protein was alsosecreted. The deletion had no effect on the fidelity of signal peptide hydrolysis. However, in its absence,the rate of apo A-I secretion and, to a lesser extent, degradation were reduced. Microscopic observationsindicated that apo A-IdPRO accumulated in vesicular structures which had some properties of the ER.Based on observations ii the BHK cell, it is proposed (Figure 39) that the normal translocated product(p-Al) is transported from the ER to the Golgi in transport vesicles where it is either secreted ordegraded. Degradation is a major fate of the apo A-I produced in BHK cells, but may be a minorpathway in cells capable of higher levels of lipid synthesis, eg. hepatocytes. The non-physiologic removalof the propeptide (in BHK clone rn-Al) caused apo A-I to accumulate to high levels in the ER whenexpression was stimulated. The potent lipid binding and self-association properties of apo A-I maypromote aggregation within the ER and segmentation of the reticulum into the vesicular structuresobserved in immunogold cryosections. Since only small amounts of the protein would reach the Golgiunder these conditions, the apo A-I accumulations in the ER would be relatively protected fromintracellular degradation.Studies of FBS and lipoprotein stimulation of apo A-I secretion from COS and BHK cellsexpressing the cDNA have suggested that the lipid status of the cell might determine the rate of apo A-Isecretion. At low levels of lipid, apo A-I may be degraded within the cell. Such a process has been shownto govern apo B secretion from hepatocytes (Boren et al, 1991) but has not been demonstrated for apoA-I. Clearly non-hepatic cells have lesser ability to assimilate lipid and may, therefore, be more markedlyaffected by culture in serum free conditions.131p-Al rn-AlERIO TransportVesiclesGolgi/NDegradationSecretion SecretionFigure 39. Proposed model for the role of the apo A-I propeptide in BHK cell transport. p-AI= BHKcells secreting proapo A-I, rn-A! = cells expressing mature apo A-I from the apo AIdPRO cDNA. Thethickness of the arrows in the figure indicates the approximate quantity of apo A-I committed to thattransport pathway.This thesis has also investigated the functional role of regions of the middle portion of apo A-Iand the C-terminus. Lysine107 lies within the proposed mobile hinge domain in the middle of the apo A-Isequence. Lysine107 is also the most frequent site of naturally occurring mutations of apo A-I. Thisregion is predicted by Segrest (1992) to be less tightly associated with phospholipid (Class Y aniphipathichelix) than the high affinity amphipathic helix of Class A.Only one other Class Y amphipathic helix is present in apo A-I, the most C-terminal 22mer(residues 220-241). This region is highly conserved (von Eckardstein et al, 1990) and the only naturalmutation lacking this region (apo A-1202- FS) has reduced endogenous LCAT activity (Funke et al, 1991).I have shown that deletion of Lys107, or of the complete C-terminal helix, causes some changesin the ability of the protein to assemble model lipoproteins in vitro although there is no apparent effecton the ability of the dK107 mutant to interact with natural lipoproteins in cell culture. The loss of the132entire C-terminal amphipathic helix had a greater effect on liposome binding than did the loss of Lys’°7.However, in preliminary studies, both the Lys107-.0 and the Dl mutant showed markedly reduced abilityto activate LCAT. Conversely, the recombinant wild-type protein was 80% as effective an LCATactivator as the purified plasma product. These results suggest that low affinity amphipathic helicaldomains of Class Y may play an important role in the LCAT activation mechanism. The sequence of apoA-I in the vicinity of Lys107 may govern the transformations of HDL involving the Class Y helix in thehinge domain.This study provides the first in vitro evidence for the involvement of a specific C-terminal helicalregion in LCAT activation. There is at present no evidence for direct interaction between apo A-I andLAT during cholesterol esterification. However, helix domains of apo A-I or LCAT might form ahydrophobic pocket to permit access of the enzyme to the hydrophobic core of the lipoprotein. This hasbeen suggested by Au-Young and Fielding (1992) for CETP and other proteins which require access tothe lipoprotein core for their action. Their work has suggested that the consensus sequence Phe-Leu-XLeu-X-X-X-N (where N is a hydrophobic amino acid) may be a binding site for CE or TG. In apo A-I,the sequence Phe229 -Leu-Ser-Ala-Leu-Glu-Glu-Tyr resembles this proposed consensus and is in theregion deleted in mutant DL Perhaps cooperation between apo A-I helices forms a CE binding site orhydrophobic pocket for transport of the CE product of the transacylation reaction to the lipoproteincore.6.2. Perspectives for Future StudySeveral lines of investigation should be followed to address some of the key issues raised in thisthesis. First, cell fractionation studies should be inititiated in BHK cell line rn-Al to establish the originand fate of the vesicular structures containing apo A-I. Further application of these techniques could alsoaddress the site of apo A-I degradation in this cell type.To assess adequately the functional consequences of the mutations described herein, a moreefficient system for the expression of apo A-I protein must be obtained. Eucaryotic expression systemsemploying liver-derived cell cultures (eg., the rat hepatoma cell, McA-RH7777) may be a usefulalternative to the non-hepatic cell cultures used to date, since in these cells the effects of lipid availability133can be assessed. The proposed model of propeptide function should be validated in this, morephysiologically relevant, model system. Additional studies of the apo A-I specific propeptidase should beused to determine the location of this enzyme activity in CHO cells and in HepG2 cells, where it isknown to be active.The BHK cell lines expressing apo A-I with and without its propeptide could be used to addressbasic features of the cell biolo’ of protein transport. The absence of the propeptide reduces the rate ofdegradation of apo A-I. This could be because the propeptide is a degradation signal, although thisseems unlikely since hepatocytes appear to secrete proapo A-I. In addition, proapo A-I does not havealtered extracellular function, so there does not appear to be a need to degrade this form if it could besecreted. A second possibility is that the propeptide provides a structural element required forintracellular transport of the protein. Chaperone proteins have been shown to be involved in proteinmovement between cellular compartments (Rothman, 1989). This is accomplished by preventingpremature or incorrect folding of the nascent polypeptide chain. The apo A-I propeptide may regulate aspecific protein-protein interaction which facilitates its transport from the ER to the Golgi apparatus. Inthe absence of the propeptide, apo A-I may not be transported efficiently and could aggregate in the ER,resulting in the large vesicular structures found in BHK line rn-Al.One aspect of this work, and of any studies involving mutagenesis, is the role of the structuralchanges on protein conformation. In the absence of definitive physical measurements (circular dichroism,X-ray crystallography, etc.), ii is not possible to exclude improper folding as the underlying mechanismfor any functional abnormality. These studies can only be carried out once large (milligram to gram)quantities of each mutant protein are available.Lectin markers were used in these studies to identify the subcellular structures containing apoA-I. These assignments have assumed that the lectins are organelle-specific (Virtanen et cii, 1980;Kaarianen et cii, 1983) and that apo A-I transport is vectorial. These studies should be extended toinclude enzyme markers for the ER in immunolluorescence studies or in subcellular fractionationexperiments. In addition, the possibility that apo A-IdPRO is subject to rapid retrograde transport(Rothman and Orci, 1992), from the Golgi to the ER, should be addressed. The latter trafficking processwould appear indistiguishable from slow ER exit in the studies I have performed to date.134An additional line of investigation which could be extended is the role of post-translationalmodification of apo A-I in its function. The site of apo A-I phosphorylation has been localized to Sei201(Beg et at, 1989), which is in the highly conserved region of the sequence. Studies in this thesis haveshown that phosphorylation also occurs in COS cells which transiently express the protein. Site-directedmutants at Sei201 could provide important information on the role of this modification in cellulartransport and secretion. In addition, the generation of the apo A-i202-’FS mutant in vitro and analysis ofphosphorylation in this variant might also provide important insight into the mechanism and role ofphosphorylation in cellular transport. Since this mutant also possesses altered C-terminal structure, therole of this region in LCAT activation and the pathophysiology of this natural mutation could also beobtained from analysis of the protein in vitro. Studies of the site and the function of covalent acylation ofapo A-I (Hoeg et ci, 1986) could also be investigated in this expression system.Finally, a more extensive analysis of the functional consequences of Lys107 deletion andsequential deletions of amphipathic helix must be completed. The absence of an adequate level ofexpression precluded more than preliminary analysis of these mutant proteins in this thesis. Completekinetic analysis should be performed on the mutant and wild-type proteins described in this work. Giventhe large variation in the kinetic parameters of purified plasma preparations (Jonas et ci, 1991), it isclear that the study of recombinant protein is the only approach which will provide conclusive evidencefor the role of specific residues in the apo A-I sequence.All of these studies will provide further insight into the mechanism and importance of apo A-Ito HDL metabolism.135REFERENCESAbersold RH, Teplow DB, Hood LE and Kent SBH. 1986. Electroblotting onto activated glass. Highefficiency preparation of proteins from analytical sodium dodecyl sulfate- polyacrylamide gels for directsequence analysis. J Biol Chem 2, 4229-38.Albers JJ, Lin JT and Roberts GP. 1979. Effect of human plasma apolipoproteins on the activity ofpurified LcAT. Artery 5: 61-75.Alexander D. 1987. An efficient vector-primer cDNA cloning system. Methods in Enzymology 154, PartE, 41-64.Anantharamaiah GM, Venkatachalapathi YV, Brouillette CG and Segrest JP. 1990. Use of syntheticpeptide analogues to localize lecithin: cholesterol acyltransferase activating domain in apolipoprotein A-I.Arteriosclerosis 10, 95-105.Andersson 5, Davis DN, Dahlback H, Jornvall H and Russell DW. 1989. Cloning, structure andexpression of the mitochondrial cytochrome P-450 sterol 26-hydroxylase, a bile acid biosynthetic enzyme.J Biol Chem 264, 8222-9.Atkinson T and Smith M. 1984. Solid phase synthesis of oligodeoxyribonucleotides by the phosphitetriester method. In MJ. Gait (Ed.), “Oligonucleotide Synthesis: A Practical Approach”. IRL Press,Oxford, pp 35-81.Auboiron 5, Sparrow DA, Beaubatie L, Bauchart D, Sparrow JT, Laplaud PM and Chapman MJ. 1990.Characterization and amino-terminal sequence of apolipoprotein A-I from high density lipoproteins inthe preruminant calf, BOS Spp. Biochem Biophys Res Commun 833-839.Au-Young J and Fielding CJ. 1992. Synthesis and secretion of wild-type and mutant human plasmacholesteryl ester transfer protein in baculovirus-transfected insect cells: The carboxyl-terminal region isrequired for both lipoprotein binding and catalysis of transfer. Proc Nat’l Acad Sci (USA) 9, 4094-98.Banka CL, Bonnet DJ, Black AS, Smith RS and Curtiss LK. 1991. Localization of an apolipoprotein A-Iepitope critical for activation of lecithin-cholesterol acyltransferase. I Biol Chem 23886-92.Barter PH, Hopkins GJ, Ha YC. 1987. The role of lipid transfer proteins in plasma lipoproteinmetabolism. Amer. Heart J. 113: 538-42.Batzri S and Korn ED. 1973. Single bilayer liposomes prepared without sonication. Biochim. Biophys.Acta 298: 1015-9.Beg ZH, Stonick JA, Hoeg IM, Demosky SJ Jr, Fairwell T and Brewer HB Jr. 1989. Humanapolipoprotein A-I. Post-translational modification by covalent phosphorylation. 1. Biol. Chem. 264, 6913-21.Biesbroeck R, Oram JF, Albers JJ and Bierman EL. 1983. Specific high affinity binding of higl densitylipoprotein to cultured human skin fibroblasts and arterial smooth muscle cells. I Clin Invest 71, 525-39.Blanche P1, Gong EL and Nichols AV. 1981. Characterization of human HDL by gradient gelelectrophoresis. Biochim. Biophys. Acta 665: 408-19.Blobel G and Dobberstein B. 1975. Transfer of proteins across membranes. I. Presence of proteolyticallyprocessed and unprocessed nascent immunoglobulin light chains on membrane-bound ribosomes ofmurine myeloma. I. Cell Biol. 67, 835-851.136Bojanovski D, Gregg RE, Zech LA, Meng MS, Ronan R and Brewer HB Jr. 1984. The in vivometabolism of proapolipoprotein A-I in Tangier disease. Clin. Res. 32: 390a.Bojanovski D, Gregg RE, Ghiselli G, Schaeffer El, Light IA and Brewer HB Jr. 1985. Humanapolipoprotein A-I isoprotein-2 metabolism: proapo A-I conversion to mature apo A-I. J. Lipid Res. 26:185-93.Bond HM, Morrone G, Venuta S and Howell KE. 1991. Characterization and purification of proteinswhich bind high-density lipoprotein. A putative cell-surface receptor. Biochem I 79, 633-41.Boren J, White A, Wettesten M, Scott J, Graham L, Olofsson S-O. 1991. The molecular mechanism forthe assembly and secretion of apo B-100-containing lipoproteins. Prog. Lipid Res. , 205-218.Breslow JL. 1986. Genetics of the human apolipoproteins. In AM Scanu and AA Spector (Eds),Biochemistry and Biology of Plasma Lipoproteins, Vol 11. Marcel Dekker, Inc., New York, pp 85-144.Breslow JL. 1992. Apolipoprotein genes and atherosclerosis. Clin Investig 7, 377-384.Brewer HB Jr, Fairwell T, La Rue A, Ronan R, Houser A and Bronzert TI. 1978. The amino acidsequence of human apo A-I, an apolipoprotein isolated from HDL. Biochem. Biophys. Res. Commun.80: 623-30.Brinton EA, Oram JF, Chen C-H, Albers II and Bierman EL. 1985. Binding of HDL to culturedfibroblasts after chemical alteration of apoprotein amino acid residues. J. Biol. Chem. 261: 495-503.Brouillette CG, Segrest JP, Ng T, Chung BH and Ragland JB. 1982. Dynamics ofapolipoprotein-phospholipid interactions. Biophys. 1. 37: 172-3.Bruhn H and Stoffel W. 1991. An approach to the functional analysis of lecithin: cholesterolacyltransferase. Activation by recombinant normal and mutagenized apolipoprotein A-I. Biol ChemHoppe-Seyler 225-34.Bruns GAP, Karathanasis SK and Breslow JL. 1984. Human apo A-I-C-ITT gene complex is located onchromosome 11. Arteriosclerosis 4: 97-102.Chacko GK. 1985. Modification of human HDL3 with tetranitromethane and the effect on its binding toisolated rat liver plasma membranes. I. Lipid Res. 26: 745-54.Chajek-Shaul T, Hayek T, Walsh A and Breslow IL. 1991. Expression of the human apolipoprotein A-Igene in transgenic mice alters high density lipoprotein (HDL) particle size distribution and diminishesselective uptake of HDL cholesteryl esters. Proc Nat’l Acad Sci (USA) 88, 6731-5.Chen C-H and Albers IJ. 1986. Stimulation of LCAT activity by apo A-TI in the presence of apo A-I.Eur. J. Biochemistryj: 589- 94.Chen C and Okayama H. 1987. High efficiency transformation of mammalian cells by plasmid DNA.Mol. Cell Biol. 7, 2745-52.Cheung MC, Wolf AC, Lum KD, Tollefson JH and Albers JJ. 1986. Distribution and localization ofLCAT and CETA in A-I- containing lipoproteins. I. Lipid Res. 27:1135-44.Cheung MC, Segrest JP, Albers JJ, Cone IT, Brouillette CG, Chung BH, Kashyap M, Glasscock MAand Anantharamaiah GM. 1987. Characterization of high density lipoprotein subspecies: structuralstudies by single vertical spin ultracentrifugation and immunoaffinity chromatography. J Lipid Res 28,9 13-929.137Cheung P and Chan L. 1983. Nucleotide sequence of cloned cDNA of human apo A-I. Nucl. Acids Res.11: 3703-15.Chou Y and Fasman GD. 1978. Prediction of the secondary structure of proteins from their amino acidsequence. Adv. Enzymol. Relat. Subj. Biochem. 47: 45-148.Clifton PM, MacKinnon AM and Barter PJ. 1985. Separation and characterization of HDLsubpopulations by gel permeation chromatography. J. Chrom. .4j4: 25-34.Curtiss LK and Edgington TS. 1985. Immunochemical heterogeneity of human plasma high densitylipoproteins. Identification with apolipoprotein A-I- and A-Il- specific monoclonal antibodies. 3 BiolChem 260, 2982-93.Curtiss LK and Smith RS. 1988. Localization of two epitopes of apolipoprotein A-I that are exposed onhuman high density lipoproteins using monoclonal antibodies and synthetic peptides. 3 Biol Chem 263,13779-85.Davis CG, van Driel IR, Russell DW, Brown MS and Goldstein JL. 1987. The LDL receptor.Identification of amino acids in cytoplasmic domain required for rapid endocytosis. J. Biol. Chem. 7:4075-82.Deeb SS, Cheung MC, Peng R, Wolf AC, Stern R, Albers JJ and Knopp RH. 1991. A mutation in thehuman apolipoprotein A-I gene. Dominant effect on the level and characteristics of plasma high densitylipoproteins. 3 Biol Chem 2, 13654-60.DeLamatre JG, Sarphie TG, Archibold RC and Hornick CA. 1990. Metabolism of apo E-free highdensity lipoproteins in rat hepatoma cells: evidence for a retroendocytotic pathway. J Lipid Res, 191-202.Dixon JL, Battini R, Ferrari S, Redman CM and Banerjee D. 1989. Expression and secretion of chickenapolipoprotein A-I in transfected COS cells. Biochim et Biophys Acta 111Q, 47-53.Dvorin E, Gorder NL, Benson DM and Gotto AM Jr. 1986. Apolipoprotein A-IV. A determinant forbinding and uptake of high density lipoproteins by rat hepatocytes. 3 Biol Chem 261, 15714-8.Edelstein C, Gordon JI, Toscas K, Sims HF, Strauss AW and Scanu AM. 1983. In vitro conversion ofproapo A-I to apo A-I. Partial characterization of an extracellular enzyme activity. J. Biol. Chem. 258:11430-3.Fennewald SM, Hamilton RL Jr and Gordon JI. 1988. Expression of human preproapo A-I andpre@ pro)apo A-I in a murine pituitary cell line (AtT-20). J. Biol. Chem. 7, 15568-77.Fidge NH and Nestel P3. 1985. Identification of apolipoproteins involved in the interaction of humanHDL3 with receptors on cultured cells. J. Biol. Chem. 260: 3570-5.Fielding CJ and Fielding PE. 1982. Cholesterol transport between cells and body fluids. Role of plasmalipoproteins and the plasma cholesterol esterification system. Medical Clinics of North America 66, 363-73.Fielding CJ, Shore VG and Fielding PE. 1972. A protein cofactor of lecithin: cholesterol acyltransferase.Biochem. Biophys. Res. Commun. 46: 1493-8.Fielding FE and Fielding CI. 1980. A cholesleryl ester transfer complex in human plasma. Proc. Natl.Acad. Sci. USA 77: 3327-30.138Fitch WM, Smith T and Breslow JL. 1984. Detecting internally repeated sequences and inferring thehistory of duplication. Methods in Enzymology 128, 773-88.Foich J, Lees M and Sloane-Stanley GH. 1957. A simple method for the isolation and purification oftotal lipids from animal tissues. J Biol Chem 226, 497-509.Folz RJ and Gordon JI. 1987. The effects of deleting the propeptide from human preproapolipoproteinA-I on cotranslational translocation and signal peptide processing. J. Biol. Chem., 17221-30.Forte TM, McCall MR, Amacher 5, Nordhausen RW, Vigne J-L and Mallory JB. 1990. Physical andchemical characteristics of apolipoprotein A-I-lipid complexes produced by Chinese hamster ovary cellstransfected with the human apolipoprotein A-I gene. Biochim. Biophys. Acta 1047, 11-18.Forte TM, Nichols AV, Selmek-Halsey J, Caylor L and Shore VG. 1987. Lipid-poor apolipoprotein A-Iin HepG2 cells: Formation of lipid-rich particles by incubation with dimyristoylphosphatidylcholine.Biochim. Biophys. Acta 920, 185-94.Francone OL and Fielding CJ. 1991. Effects of site-directed mutagenesis at residues Cy’ and Cys184 onlecithin-cholesterol acyltransferase activity. Proc Nat’l Acad Sci (USA), 1716-20.Fukushima D, Yokoyama S, Kroon DJ, Kezdy FJ and Kaiser ET. 1980. Chain length-function correlationof amphiphilic peptides: synthesis and surface properties of a tetratetracontapeptide segment of apo A-I.I. Biol. Chem. 255: 10651-7.Funk WD, MacGillivray RTA, Mason AB, Brown SA and Woodworth RC. 1990. Expression of theamino-terminal half-molecule of human serum transferrin in cultured cells and characterization of therecombinant protein. Biochemistry, 1654-60.Funke H, von Eckardstein A, Pritchard PH, Karas M, Albers JJ and Assmann G. 1991. A frameshiftmutation in the human apolipoprotein A-I gene causes high density lipoprotein deficiency, partiallecithin: cholesterol-acyltransferase deficiency, and corneal opacities. J Clin Invest 87, 371-6.Fuster V, Badimon L, Badimon JJ, Chesebro JH. 1992. The pathogenesis of coronary artery disease andthe acute coronary syndromes. New Engl. J. Med. 242-250.Gamier 3, Osguthorpe DJ and Robson B. 1978. Analysis of the accuracy of simple methods forpredicting the secondary structure of globular proteins. J Mol Biol 120, 97-120.Glomset JA. 1970. Physiological role of LCAT. Amer. 3. Clin. Nutr. 23: 1129-36.Gluzman Y. 1981. SV4O-Transformed Simian cells support the replication of early SV4O mutants. Cell23: 175-82.Gordon JI, Sims HF, Strauss AW, Scanu AM, Edelstein C and Byrne RE. 1986. Proteolytic processingand compartmentalization of the primary translation products of mammalian apolipoprotein mRNAs.CRC Critical Reviews in Biochemistry, 37-7 1.Gordon JI, Sims HF, Edelstein C, Scanu AM and Strauss AW. 1984. Human proapo A-TI is cleavedfollowing secretion from HepG2 cells by a thiol protease. J. Biol. Chem. 29: 15556-63.Gordon T, Castelli WP, Hjortland MC, Kaimel WB and Dawber TR. 1977. High density lipoprotein as aprotective factor against coronary heart disease. Am. J. Med. : 707-14.Graham DL and Oram JF. 1987. Identification and characterization of a high density lipoprotein-bindingprotein in cell membranes by ligand blotting. J Biol Chem 262, 7439-42.139Graham FL and van der Eb AJ. 1973. Transformation of rat cells by DNA of human adenovirus-5.Virology 52, 456-67.Griffiths G, McDowall A, Back R and Dubochet J. 1984. On the preparation of cryosections forimmunocytochemistry. J. Ultrastruct. Res..9, 65-78.Gugliucci A and Stahl AJC. 1991. In vitro glycation of human apolipoprotein A-I reduces its efficiency inlecithin: cholesterol acyltransferase activation. Clin Chim Acta.Q4, 37-42.Havel RJ and Kane JP. 1989. “Structure and metabolism of plasma lipoproteins” in The Metabolic Basisof Irtherited Disease, Scriver CR, Beaudet AL, Sly WS and Valle D (Eds), McGraw-Hill, Inc. pp 1129-1138.Hill iS and Pritchard PH. 1990. Improved phenotyping of apolipoprotein E: Application to populationfrequency distribution. Clin. Chem. , 1871-74.Hoeg TM, Meng MS, Ronan R, Fairwell T and Brewer HB Jr. 1986. Human apolipoprotein A-I: Post-translational modification by fatty acid acylation. J. Biol. Chem. 261, 3911-14.Hoeg JM, Demosky SJ, Edge SB, Gregg RE, Osborne JC Jr and Brewer HB Jr. 1985. Characterizationof a human hepatic receptor for HDL. Arteriosclerosis : 228- 37.Hussain MM, Roghani A, Cladaras C, Zanni EE and Zannis VI. 1991. Secretion of lipid-poor nascenthuman apolipoprotein (apo) A-I, apo C-Ill and apo E by cell clones expressing the corresponding genes.Electrophoresis,273-283.Jabs H-U, Assmann G, Greifendorf D and Benninghoven. 1986. High performance liquidchromatography and time-of -flight secondary ion mass spectrometry: a new dimension in structuralanalysis of apolipoproteins. J Lipid Res 27, 613-21.Jauhiainen M and Dolphin PJ. 1986. Human plasma LCAT. An elucidation of the catalytic mechanism.J. Biol. Chem. 261: 7032-43.Jonas A and McHugh HT. 1983. Reaction of LCAT with micellar complexes of apolipoprotein A-I andphosphatidyicholine, containing variable amounts of cholesterol. J. Biol. Chem. 258: 10335-40.Jonas A, Covinsky KE and Sweeney SA. 1985. Effects of amino group modification in discoidal apo AI-egg yolk phosphatidyicholine-cholesterol complexes on their reactions with LCAT. Biochemistry 24:3508-13.Jonas A, von Eckardstein A, Kezdy KE, Steinmetz A and Assman G. 1991. Structural and functionalproperties of reconstituted high density lipoprotein discs prepared with six apolipoprotein A-I variants. JLipid Res, 97-106.Kaariainen L, Virtanen I, Sareste I and Keranen 5. 1983. Transport of virus membrane glycoproteins,use of temperature-sensitive mutants and organelle-specific lectins. Methods in Enzymology96, 453-65.Kambouris AM, Roach PD, Calvert GD and Nestel PJ. 1990. Retroendocytosis of high densitylipoproteins by the human hepatoma cell line, HepG2. Arteriosclerosis 10, 582-90.Karathanasis SK, Zannis VI and Breslow JL. 1983. Isolation and characterization of the human apo A-Igene. Proc. NatI. Acad. Sci. USA Q: 6147-51.Karathanasis SK, Norum RA, Zannis VI and Breslow IL. 1983. An inherited polymorphism in thehuman apo A-I gene locus related to the development of atherosclerosis. Nature 301: 718-20.140Karathanasis SK, Zannis VI and Breslow JL. 1984. A DNA insertion in the apo A-I gene of patients withpremature atherosclerosis. Nature 305: 823-5.Karlin JB, Johnson WJ, Benedict CR, Chacko GK, Phillips MC and Rothblat GH. 1987. Cholesterol fluxbetween cells and high density lipoprotein. Lack of relationship to specific binding of the lipoprotein tothe cell surface. J Biol Chem 262, 12557-64.Kay LL, Ronan R, Schaeffer El, Brewer HB Jr. 1982. Tangier disease: a structural defect in apo A-I(apo A-I- Tangier). Proc. Nati. Acad. Sci. USA 79: 2485-9.Kelly RB. 1985. Pathways of protein secretion in eucaryotes. Science 230, 25-32.Kizer JS and Tropsha A. 1991. A motif found in propeptides and prohormones that may target them tosecretory vesicles. Biochem. Biophys. Res. Commun. 174, 586-92.Krieg PA and Mellon DA. 1984. Functional messenger RNAs are produced by SP6 in vitro transcriptionof cloned cDNAs. Nuci. Acids Res. 12, 7057-70.Kriegler M. 1990. Gene Transfer and Expression. A Laboratory Manual. Stockton Press, New York. pp96-101.Kunkel TA. 1985. Rapid and efficient site-specific mutagenesis without phenotypic selection. Proc. Nat!.Acad. Sci. (USA) 82, 488-92.Kunkel TA, Roberts ID and Zakour RA. 1987. Rapid and efficient site-specific mutagenesis withoutphenotypic selection. Methods in Enzymologyj4, 367-82.Kyte I and Doolittle RF. 1982. A simple method for displaying the hydropathic character of a protein. JMol Biol 157, 105-132.Ladias JAA, Kwiterovich P0 Jr, Smith HH, Karathanasis SK and Antonarakis SE. 1990. ApolipoproteinA-I Baltimore (Ar&° to Leu), a new apo A-I variant. Hum Genet 4, 439-45.Laemmli UK. 1970. Cleavage of structural proteins during the assembly of the head of bacteriophage T4.Nature 227, 680-5.Lainon-Fava S, Ordovas JM, Mandel G, Forte TM, Goodman RH and Schaefer El. 1987. Secretion ofapolipoprotein A-I in lipoprotein particles following transfection of the human apolipoprotein A-I geneinto 3T3 cells. I. Biol. Chem. 262, 8944-7.Lathe RF, Lecocq JP, and Everett R. 1983. DNA engineering: the use of enzymes, chemicals andoligonucleotides to restructure DNA sequences in vitro. In Genetic Engineering, R. Williamson (Ed.), 4,pp 1-56. Academic Press, New York.Law SW, Gray G and Brewer HB Jr. 1983. cDNA cloning of human apo A-I: Amino acid sequence ofpreproapo A-I. Biochem. Biophys. Res. Commun. jj2: 257-64.Li W-H, Tanimura M, Luo C-C, Datta S and Chan L. 1988. The apolipoprotein multigene family:biosynthesis, structure, structure-function relationships, and evolution. J Lipid Res 29 245-271.Luthman H and Magnusson G. 1983. High efficiency polyoma DNA transfection of chloroquine treatedcells. Nucl. Acids Res. 11, 1295-1308.Macheboeuf MA. 1929. Recherches sur les phosphoaminolipides et les sterides du serum et du plasmasanguins. Bull. Soc. Chim. Biol. , 268.141Mahley RW, Innerarity TL, Rail SC and Weisgraber KH. 1984. Plasma lipoproteins: apolipoproteinstructure and function. J Lipid Res, 1277-94.Maicrides SC, Ruiz-Opazo N, Hayden M, Nussbaum AL, Bresiow JL and Zannis VI. 1988. Sequence andexpression of Tangier apo A-I gene. Eur J Biochem i7, 465-71.Mallory JB, Kushner PJ, Protter AA, Cofer CL, Appleby VL, Lau K, Schiliing JW and Vigne J-L. 1987.Expression and characterization of human apo A-I in chinese hamster ovary cells. J. Biol. Chem. 262,424 1-7.Maniatis T, Fritch EF and Sambrook J. 1982. Molecular Cloning: A Laboratory Manual. Cold SpringHarbour Laboratory.Matsudaira P. 1987. Sequence from picomole quantities of proteins electroblotted onto polvinylidenedifluoride membranes. J. Biol. Chem. 262, 10035-10038.Matsunaga T, Hiasa Y, Yanagi H, Maeda T, Hattori N, Yamakawa K, Yamanouchi Y, Tanaka I, ObaraT and Hamaguchi H. 1991. Apolipoprotein A-I deficiency due to a codon 84 nonsense mutation of theapolipoprotein A-I gene. Proc Nat’l Acad Sci (USA) ., 2793-7.Marcel YL, Provost PR, Koa H, Raffai E, Vu Dac N Fruchart i-C and Rassart E. The epitopes ofapolipoprotein A-I define distinct structural domains including a mobile middle region. 1991. J BiolChem 266, 3644-53.Matz CE and Jonas A. 1982. Reaction of human LCAT with micellar complexes of apo A-I,phosphatidylcholine and cholesterol. J. Biol. Chem. .7: 4541-6.McLean J, Fielding Ci, Drayna D, Dieplinger H, Baer B, Kohr W, Henzel W and Lawn R. 1986.Cloning and expression of human LCAT cDNA. Proc Nat’l Acad Sci (USA) 83, 2235-9.McLeod R, Lacko AG, Pritchard PH and Frohlich J. 1986. Purification of biologically activeapolipoproteins by chromatofocussing. J. Chrom. 381, 271-83.McVicar JP, Kunitake ST, Hamilton RL and Kane JP. 1984. Characteristics of human lipoproteinsisolated by selected- affinity immunosorption of apolipoprotein A-I. Proc. Nati. Acad. Sci. USA1356-60.Menzel H-i, Kladetzky R-G and Assmann G. 1982. One-step screening method for the polymorphism ofapolipoproteins A-I, A-Il and A-IV. J. Lip. Res. 23: 915-22.Menzel Hi, Assmann G, Rail SC, Weisgraber KH and Mahley RW. 1984. Human apo A-Ipolymorphism. Identification of amino acid substitutions in three electrophoretic variants of theMuenster-3 type. J. Biol. Chem. 259:3070-6.Messing J. 1983. New M13 vectors for cloning. Methods in Enzymolo 20-78.Miller GJ and Miller NE. 1975. Plasma HDL concentration and development of ischaemic heart disease.Lancetl: 707-14.Muesch A, Hartmann E, Rohde K, Rubartelli A, Sitia R and Rapoport TA.. 1990. A novel pathway forscretory proteins? Trends in Biochem Sci j, 86-88.Mueller W, Weber H, Meyer F, and Weissmann C. 1978. Site directed mutagenesis in DNA: generationof point mutations in cloned 3-globin complementary DNA at the positions corresponding to amino acids121 to 123. J. Mol. Biol.124: 343-58.142Nichols WC, Dwulet WF, Liepnieks J and Benson MD. 1988. Variant apolipoprotein A-I as a majorconstituent of a human hereditary amyloid. Biochem Biophys Res Communj, 762-8.Nishida HI, Nakanishi T, Yen EA, Arai H, Yen FT and Nishida T. 1986. Nature of the enhancement ofthe LCAT reaction by various apolipoproteins. J. Biol. Chem. 12028-35.Noble RP. 1968. Electrophoretic separation of plasma lipoproteins in agarose gel. J. Lipid. Res. 9, 693-700.Nothwehr SF and Gordon JI. 1990. Targeting of proteins into the eucaryotic secretory pathway: signalpeptide structure/function relationships. Bioessays 12, 479-84.OFarrel PH. 1975. High resolution two dimensional electrophoresis of proteins. J. Biol. Chem. 25, 4007-21.Oram JF, Brinton EA and Bierman EL. 1984. Regulation of high density lipoprotein receptor activity inhuman cultured skin fibroblasts and human arterial smooth muscle cells. J Clin Invest 72, 1611-21.Oram JF, Mendez AJ, Slotte JP and Johnson TF. 1991. High density lipoprotein apolipoproteins mediateremoval of sterol from intracellular pools but not from plasma membranes of cholesterol-loadedfibroblasts. Arteriosclerosis and Thrombosis 11, 403-14.Ordovas JM, Schaefer EJ, Salem D, Ward RH, Glueck CJ, Vergani C, Wilson PW and Karathanasis SK.1986. Apo A-I gene polymorphism associated with premature coronary artery disease and familialhypoalphalipoproteinemia. N. EngI. J. Med. .j4: 671-7.Palmiter RD, Behringer RR, Quaife CJ, Maxwell IH and Brinster RL. 1987. Cell lineage ablation intransgenic mice by cell-specific expression of a toxin gene. Cell 50, 435-43.Park YB, Yuksel U, Gracy RW and Lacko AG. 1987. The catalytic center of LCAT: Isolation andsequence of diisopropylfluorophosphate-labeled peptides. Biochem. Biophys. Res. Commun. 143: 360-3.Pelham HRB and Jackson RJ. 1976. An efficient mRNA-dependent translation system from reticulocytelysates. Eur. J. Biochem. 7: 247-56.Peters T. 1987. Intracellular precursor forms of plasma proteins: Their functions and possible occurrencein plasma. Clin. Chem. 33, 1317-25.Ponsin G, Gotto AM, Utermann G and Pownall HJ. 1985. Abnormal interaction of the humanapolipoprotein A-I variant [Lys’°7-.0] with high density lipoproteins. Biochem Biophys Res Commun,856-62.Ponsin G, Hester L, Gotto AM Jr, Pownall HJ and Sparrow JT. 1986. Lipid-peptide association andactivation of LCAT. Effect of alpha helicity. J. Biol. Chem. 2: 9202-5.Pownall HJ, Massey JB and Gotto AM Jr. 1978. Kinetics of lipid-protein interactions: Interaction ofapolipoprotein A-I from human plasma HDL with phosphatidyicholines. Biochemistry II: 1183-7.Pownall HJ, Hu A, Gotto AM Jr, Albers JJ and Sparrow JT. 1980. Activation of LCAT by a syntheticmodel lipid associating peptide. Proc. Natl. Acad. Sci. USA 77:3154-8.Pritchard PH and Frohlich J. 1986. Apo A-I and LCAT in a patient with Tangier disease. In A Angeland J Frohlich (Eds), Lipoprotein Deficiency Syndromes, Advances in Experimental Medicine andBiology, Vol. 201. Plenum Press, New York. pp 105-10.143Rahim ATMA, Miyazaki A, Morino Y and Horiuchi S. 1991. Biochemical demonstration of endocytosjsand subsequent resecretion of high-density lipoprotein by rat peritoneal macrophages. Biochim BiophysActa 1082, 195-203.Rail SC, Menzel HJ, Assmann G, Utermann G, Haas J, Harris RJ, Weisgraber KH, Bersot TP andMahley RW. 1983. Identification of amino acid substitutions in five human apolipoprotein A-I variants.Arteriosclerosis 3: 515a.Rail SC Jr, Weisgraber KH, Mahley RW, Ogawa Y, Fielding Ci, Utermann G, Haas I, Steinmetz A,Menzel H-i and Assmarin G. 1984. Abnormal lecithin: cholesterol acyltransferase activation by a humanapolipoprotein A-I variant in which a single lysine residue is deleted. I Biol Chem 29 10063-70.Rail SC Jr, Weisgraber KH, Mahley RW, Enhoim C, Schamaun 0, Olaisen B, Blomhoff I? and TeisbergP. 1986. Identification of homozygosity for a human apolipoprotein A-I variant. J Lipid Res 27, 436-41.Rees A, Stocks J, Schoulders C, Carison LA, Baralle FE and Galton Di. 1984. Restriction enzymeanalysis of the apo A-I gene in Fish eye disease and Tangier disease. Acta Med. Scand. 215: 235-7.Rees A, Stocks J, Sharpe CR, Vella MA, Shoulders CC, Katz I, Jowett NI, Baralle FE and Galton DJ.1985. DNA polymorphism in the apo A-I-C-Ill cluster: association with hypertriglyceridemia. I ClinInvest 76, 1090-5.Reichi D and Miller NE. 1989. Pathophysiology of reverse cholesterol transport. Insights from inheriteddisorders of lipoprotein metabolism. Arteriosclerosis , 785-97.Roghani A and Zannis VI. 1988. Alterations of the glutamine residues of human apolipoprotein A-Ipropeptide by in vitro mutagenesis. Characterization of the normal and mutant protein forms.Biochemistry 27, 7428-35.Ross R. 1986. The pathogenesis of atherosclerosis- an update. New England Journal of Medicine.4,488-501.Rosseneu M, Assmann G, Taveirne MJ and Schmitz G. 1984. Lipid binding properties of the Tangierapo A-I and its isoproteins. J Lipid Res25, 111-20.Roth RI, Jackson RL and Gotto AM. 1977. Interaction of plasma “arginine-rich” apolipoprotein withdimyristoylphosphatidylcholine. Biochemistry j: 5030-36.Rothman JE. 1989. Polypeptide chain binding proteins: Catalysts of protein folding and related processesin cells. Cell 59, 59 1-601.Rothrnan JE and Orci L. 1992. Molecular dissection of the secretory pathway. Nature 355, 409-415.Rubin EM, Ishida BY, Clift SM and Krauss RM. 1991a. Expression of human apolipoprotein A-I intransgenic mice results in reduced plasma levels of murine apolipoprotein A-I and the appearance of twonew high density lipoprotein subclasses. Proc NatI Acad Sd (USA), 434-8.Rubin EM, Krauss RM, Spangler EA, Verstuyft JG and Clift SM. 1991b. Inhibition of earlyatherogenesis in transgenic mice by human apo A-I. Nature 265-7.Ruiz-Opazo N and Zannis VI. 1988. Expression of the human apolipoprotein A-I gene in the ratmyogenic L6E9 cells. J. Biol. Chem. 263, 1739-44.Sambrook I, Fritsch EF and Maniatis T. 1989. Molecular Cloning: A Laboratory Manual. 2nd Ed. ColdSpring Harbor Laboratory Press. Cold Spring Harbor, New York.144Sanger F, Nicklen S and Coulson AR. 1977. DNA sequencing with chain-terminating inhibitors. Proc.Nati. Acad. Sci. (USA) 74, 5463-9.Scanu AM. 1987. Proapolipoprotein-converting enzymes and high-density lipoprotein early events inbiogenesis. Am. Heart 1. 113, 527-32.Schaeffer EJ. 1984. Clinical, biochemical and genetic features in familial disorders of HDL deficiency.Arteriosclerosis 4: 303-22.Schimke RT, Roos DS and Brown PC. 1987. Amplification of genes in somatic mammalian cells.Methods in Enzymology 151, 85-104.Schmitz G, Assmann G, RaIl SC Jr and Mahley RW. 1983. Tangier disease: Defective recombination ofa specific Tangier apo A-I isoform (proapo A-I) with high density lipoproteins. Proc NatI Acad Sci(USA) 80, 6081-5.Schmitz G, Robenek H, Lohmann H and Assmann G. 1985. Interaction of high density lipoproteins withcholesteryl ester-laden macrophages: biochemical and morphological characterization of cell surfacereceptor binding, endocytosis and resecretion of high density lipoproteins by macrophages. EMBO 3 4,613-22.Schouten D, Kleinherenbrink-Stins M, Brouwer A, Knook DL, Kamps JAAM and van Berkel TIC. 1990.Characterization in vitro of interaction of human apolipoprotein E-free high density lipoprotein withhuman hepatocytes. Arteriosclerosis j, 1127-35.Segrest JP, Jackson RL, Morrisett JD and Gotto AM Jr. 1974. A molecular theory of lipid-proteininteractions in the plasma lipoproteins. FEBS Lett. : 247-53.Segrest JP, DeLoof H, Dohlman JG, Brouillette CG and Anantharamaiah GM. 1990. The amphipathichelix Motif: Classes and Properties. Proteins 8, 103-117.Segrest JP, Jones MK, De Loof H, Brouillette CG, Venlatachalapathi YV and Anantharamaiah GM.1992. The amphipathic helix in the exchangeable apolipoproteins: a review of secondary structure andfunction. J Lipid Res.3, 141-166.Seilhamer JJ, Protter AA, Frossard P and Levy-Wilson B. 1984. Isolation and DNA sequence of full-length cDNA and of the entire gene for human apolipoprotein A-I- Discovery of a new geneticpolymorphism in the apo A-I gene. DNA 3, 309-17.Sharpe CR, Sidoli A, Shelley S, Lucero MA, Shoulders CC and Baralle FE. 1984. Humanapolipoproteins A-I, A-lI, C-TI and C-Ill. cDNA sequences and mRNA abundance. Nuci. Acids Res. 12:3917-32.Shortle D, Grisafi P, Benkovic S and Botstein D. 1982. Gap misrepair mutagenesis: efficient site-directedinduction of transition, transversion and frameshift mutations in vitro. Proc. Nati. Acad. Sci. USA 79:1588-92.Shortle D and Botstein D. 1983. Directed mutagenesis with sodium bisulfite. Methods Enzymol. 100:457-68.Shoulders CC and Baralle FE. 1982. Isolation of the human HDL apo A-I gene. Nuci. Acids Res. 10:4873-82.Sing CF and Moll PP. 1990. Genetics of atherosclerosis. Annu. Rev. Genet. 24, 171-187.145Smith M. 1985. In vitro mutagenesis. Arm. Rev. Genetics 19: 423-62.Soutar AK, Garner CW, Baker HN, Sparrow JT, Jackson RL, Gotto AM and Smith LC. 1975. Effect ofhuman plasma apolipoproteins and phosphatidyicholine acyl donor on LCAT activity. Biochemistry 14,3057-64.Sparrow DA, Lee BR, Laplaud PM, Auboiron S, Bauchart D, Chapman MJ, Gotto AM Jr, Yang CYand Sparrow JT. 1992. Plasma lipid transport in the preruminant calf, Bos spp: primary structure ofbovine apolipoprotein A-I. Bichim et Biophys Acta, 145-150.Sparrow JT and Gotto AM Jr. 1980. Phospholipid binding studies with synthetic apolipoproteinfragments. AnnNY Acad Sci 187-211.Sparrow JT and Gotto AM Jr. 1982. Apolipoprotein/lipid interactions: Studies with syntheticpolypeptides. CRC Crit. Rev, in Biochem. 13: 87-107.Stadtman ER. 1990. Colaelnt modification reactions are marking steps in protein turnover. Biochemistry29, 6323-6331.Steinmetz A, Kaffarnik H and Utermann G. 1985. Activation of LCAT by human apo E isoforms. Eur. J.Biochem. 152: 747-51.Steinmetz A and Utermann G. 1985. Activation of LCAT by human apo A-TV. J. Biol. Chem. 260:2258-64.Stoffel W and Binczek E. 1988. Structural requirements of human preproapolipoprotein A-I fortranslocation and processing studied by site-directed mutagenesis in vitro. Biol Chem Hoppe-Seyler 368,1055-63.Stoffel W and Binczek E. 1991. Transient expression of wild type and mutant human apolipoprotein A-Iin COS cells. Biol Chem Hoppe-Seyler 481-488.Strobl W, Jabs H-U, Hayde M, Holzinger T, Assmann G and Widhalm K. 1988. Apolipoprotein A-I(Glu98-.Lys): A mutant of the major apolipoprotein of high-density lipoproteins occurring in a familywith dyslipoproteinemia. Pediatr Res 24 222-8.Takada Y, Sasaki J, Ogata 5, Nakanishi T, Ikehara Y and Arakawa K. 1990. Isolation andcharacterization of human apolipoprotein A-I Fukuoka (110 Glu to Lys). A novel apolipoprotein variant.Thomsen DR, Stenberg RM, Goins WF and Stinski MF. 1984. Promoter-regulatory region of the majorintermediate early gene of human cytomegalovirus. Proc Nat’l Acad Sci (USA) 81, 659-63.Tokuyasu KT. 1978. A study of positive staining of ultrathin frozen sections. J. Ultrastruct. Res. 63, 287-307.Tokuyasu KT. 1973. A teclmique for ultramicrotomy of cell suspensions and tissues. J. Cell Biol. 57, 551-65.Tokuyasu KT. 1983. Present state of immunocryoultramicrotomy. I. Histochem. Cytochem. 31, 164-7.Tokuyasu KT. 1984. Immuno-cryoultramicrotomy. Immunolabelling for Electron Microscopy. Polak andVarndell (Eds). Elsevier Science Publishers, Amsterdam. pp 71-82.Towbin H, Staehelin T arid Gordon J. 1979. Electrophoretic transfer of proteins from polyacrylamidegels to nitrocellulose sheets: Procedure and some applications. Proc Natl Acad Sci (USA) 76, 4350-4.146Tozuka M and Fidge N. 1989. Purification and characterization of two high-density-lipoprotein-bindingproteins from rat and human liver. Biochem J 239-44.Utermann G, Feussner G, Franceschini G, Haas J and Steinmetz A. 1982. Genetic variants of group Aapolipoproteins. Rapid methods for screening and characterization without ultracentrifugation. J. Biol.Chem. 257: 50 1-7.Utermann G, Haas J, Steinmetz A, Paetzold R, Rail SC and Weisgraber KH. 1984. Apolipoprotein Aiessen (Pro143—A g): a mutant that is defective in activating LCAT. Eur J Biochem j44, 326-31.Vance JE arid Vance DE. 1990. Lipoprotein assembly and secretion by hepatocytes. Ann Rev Nutrition10, 337-356.Virtanen I, Ekblom P, Laurila P. 1980. Subcellular compartmentalization of saccharide moieties incultured normal and malignant cells. J Cell Biol, 429-434.von Eckardstein A, Funke H, Henke A, Altiand K, Benriinghoven A and Assmann G. 1989.Apolipoprotein A-I variants. Naturally occurring substitutions of proline residues affect plasmaconcentration of apolipoprotein A-I. J Clin Invest 4, 1722-30.von Eckardstein A, Funke H, Walter M, Altland K, Benninghoven A and Assmann G. 1990. Structuralanalysis of human apolipoprotein A-I variants. Amino acid substitutions are nonrandomly distributedthroughout the apolipoprotein A-I primaiy structure. J Biol Chem, 8610-17.Walsh A, Ito Y and Breslow JL. 1989. High levels of human apolipoprotein A-I in transgenic mice resultin increased plasma levels of small high density lipoprotein (HDL) particles compared to human HDL.J Biol Chem 264, 6488-94.Walter P and Blobel G. 1983. Preparation of microsomal membranes for cotranslational proteintranslocation. Methods in Enzymology 96, 84-93.Weisgraber KH, Rail SC Jr, Bersot TP, Mahley RW, Franceschini G and Sirtori CR. 1983.Apolipoprotein A-I-Milano. Detection of normal A-I in affected subjects and evidence for a cysteine forarginine substitution in the variant A-I. J. Biol. Chem. 2508-13.Yang C-Y, Manoogian D, Pao Q, Knapp RD, Gotto AM Jr and Pownall HJ. 1987. LCAT. Functionalregions and a structural model of the enzyme. J. Biol. Chem. 262: 3086-91.Yokoyama 5, Fukushima D, Kupferberg JP, Kezdy FJ and Kaiser EJ. 1980. The mechanism of activationof LCAT by apo A-I and an amphiphilic peptide. J. Biol. Chem. 255: 7333-9.Zarmis VI, Lees AM, Lees RS and Breslow JL. 1982. Abnormal apo A-I isoprotein composition inpatients with Tangier disease. J. Biol. Chem. 257: 4978-86.Zannis VI and Breslow JL. 1985. Genetic mutations affecting human lipoprotein metabolism. In AHHarris and K Hirschhorn (Eds), Advances in Human Genetics, Vol. 14. Plenum, New York, pp 125-215.Zannis VI, Kurnit D and Breslow JL. 1982. Hepatic apo A-I and apo E and intestinal apo A-I aresynthesized in precursor isoprotein forms by organ cultures of human fetal tissues. J. Biol. Chem. 257,536-44.Zannis VI, Karathanasis SK, Keutmann HT, Goldberger G and Breslow JL. 1983. Intracellular andextracellular processing of human apo A-I: Secreted apo A-I isoprotein 2 is a propeptide. Proc. Natl.Acad. Sci. USA 80: 2574-8.147

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0086602/manifest

Comment

Related Items