Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Influence of osmotic forces and transbilayer membrane area imbalances on the stability and morphology… Mui, Barbara L.-S. 1993

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

Item Metadata

Download

Media
831-ubc_1993_fall_phd_mui_barbara.pdf [ 5.84MB ]
Metadata
JSON: 831-1.0098816.json
JSON-LD: 831-1.0098816-ld.json
RDF/XML (Pretty): 831-1.0098816-rdf.xml
RDF/JSON: 831-1.0098816-rdf.json
Turtle: 831-1.0098816-turtle.txt
N-Triples: 831-1.0098816-rdf-ntriples.txt
Original Record: 831-1.0098816-source.json
Full Text
831-1.0098816-fulltext.txt
Citation
831-1.0098816.ris

Full Text

INFLUENCE OF OSMOTIC FORCES AND TRANSBILAYER MEMBRANE AREA IMBALANCES ON THE STABILITY AND MORPHOLOGY OF MODEL MEMBRANE VESICLES by  BARBARA L.-S. MUI B.Sc. Biochemistry, The University of British Columbia, 1988  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF BIOCHEMISTRY  We accept thi  quired standard  THE UNIVERSITY OF BRITISH COLUM BIA  October, 1993 © Barbara L-S. Mui  In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission.  (Signature  Department of  Biochemistry  The University of British Columbia Vancouver, Canada  Date  ^October, 1993  DE-6 (2/88)  ABSTRACT This thesis is focused on the effects of osmotic pressure and transbilayer area asymmetry on the morphology and stability of model lipid vesicle systems. These forces have been implicated in many diverse biological functions including membrane fusion, maintenance of cellular shape and membrane trafficking. The osmotic stability of these large unilamellar vesicles (LUVs) in the present of plasma has also been investigated. This is of practical interest as these vesicles are currently being used as in vivo drug carriers. The LUVs used in this thesis are made by the extrusion procedure which involves repeatedly passing an aqueous lipid dispersion through small pore sized filters. The morphology of the resulting egg phosphatidylcholine:cholesterol (55:45, mol:mol) LUVs formed by the extrusion technique (LUVETs) is found to be predominantly non-spherical, a property which has important effects on their osmotic properties. In particular, the initial influx of water that results from exposure of these vesicles to a hypoosmotic solution is first accomodated by the vesicles "rounding up" to maximize their volume to surface area ratio. Further studies show that osmotically induced vesicle lysis is a very rapid event with most of the solute release occuring within the first 30 seconds. However. lysis results in only partial release of solute such that a residual osmotic gradient results. This residual gradient is similar to the gradient required to initiate lysis. The maximum residual osmotic gradients were measured for LUVET systems with different mean diameters (90 to 340 nin). These results indicate that the osmotic properties of LUVETs obey Laplace's law for a spherical vesicle, relating the pressure difference across a close elastic membrane to the membrane tension. Osmotic lysis studies were also conducted in the presence of plasma with palmitoyloleoylphosphatidylcholine:cholesterol (55:45. mol:mol) LUVETs. Plasma was  found to enhance solute release. However, both the residual and threshold osmotic gradients are reduced to the same extent. The plasma component responsible for the reduction in the membrane lysis tension is demonstrated to be the lipoproteins, with the high density lipoproteins exerting the greatest effect. The third area of investigation concerns the morphological consequences of imbalances between the surface areas of the vesicle's inner and outer monolayers as examined by cryo-electron microscopy techniques. Surface area imbalances are generated by inducing net transbilayer transport of DOPG in dioleoylphosphatidylcholine: dioleoylphosphatidylglycerol (DOPC:DOPG. 9:1. mol:mol) vesicles in response to transmembrane pH gradients. It is shown that when DOPG is transported from the inner monolayer to the outer monolayer. initially invaginated LUVETs are transformed to long narrow tubular structures. or spherical structures with one or more tubular extensions. Conversely. when DOPG is transported from the outer monolayer to the inner monolayer of non-invaginated LUVETs, a reversion to invaginated structures is observed. These results are consistent with proposals that factors leading to imbalances in monolayer surface areas could play important roles in intracellular membrane transport processes.  TABLE OF CONTENTS Abstract ^ Table of Contents ^  iv  List of Figures ^  vii  List of Tables ^  ix  Abbreviations ^ Acknowledgments ^ Dedication ^  Chapter 1 Introduction ^  xii xiii  1  1.1 Overview ^ 1 1 1.2 Structure of biological membranes ^ 3 1.3 Model membranes ^ 4 1.3.1 Monolayers ^ 5 1.3.2 Planar lipid membranes ^ 5 1.3.3 Liposomes ^ 6 1.3.3.1 Giant vesicles ^ 7 1.3.3.2 Large unilamellar vesicles ^ 8 1.3.3.3 Small unilamellar vesicles ^ 8 1.4 Chemical properties of membrane lipids ^ 9 1.4.1 Lipid structure ^ 12 1.4.2 Acid-base properties ^ 13 1.5 Lipid polymorphism ^ 16 1.6 Gel-liquid crystalline phase transition temperature ^ 19 1.7 Lateral and transbilayer lipid movement ^ 19 1.8 Mechanical properties of membranes ^ 20 1.8.1 Membrane deformation ^ 20 1.8.2 Membrane elasticity ^ 23 ^ 1.9 Membrane permeability 23 1.9.1 Water permeability ^ 25 1.9.2 Non-ionic solute permeability ^ 25 1.9.3 Ion permeability ^ ^ 26 1.10 Osmotic pressure 26 1.10.1 Theory ^ 27 1.10.2 Laplace's Law ^ 29 1.10.3 Biological systems ^ 29 1.10.4 Membrane rupture ^ 31 1.11 Vesicle morphology ^ 1.11.1 Vesicle shape as determined by the membrane's elastic 32 energy ^ 1.11.2 Influence of transbilayer area asymmetry between the 34 two monolayers ^ 34 1.11.3 Theoretical models ^ 37 1.11.4 Biological implications ^  38 1.12 Liposome-plasma interactions ^ 38 1.12.1 Influence of membrane properties ^ 39 1.12.2 Types of protein-membrane interactions ^ 1.12.3 Plasma proteins involved in liposome destabilization ^ 40 40 1.12.3.1 Lipoproteins ^ 1.12.3.2 Complement ^ 42 44 1.13 Liposomes as drug carriers ^ 45 1.14 Thesis outline ^ Chapter 2 Osmotic properties of LUVETs ^  46  2.1 Introduction ^ 2.2 Materials and Methods ^ 2.2.1 Lipids and chemicals ^ 2.2.2 Preparation of lipid vesicles ^ 2.2.3 Measurements of vesicle lysis ^ 2.2.4 Time course of CF release from vesicles ^ 2.2.5 Influence of solute molecular weight on osmoticallyinduced leakage ^ 2.2.6 Vesicle trapped volume measurements ^ 2.2.7 Solute equilibration of FATMLVs ^ 2.2.8 Cryo-electron microscopy ^ 2.2.9 Vesicle fusion analysis ^ 2.2.10 Vesicle size analysis ^ 2.2.11 Osmolarity measurements ^ 2.2.12 Analytical procedures ^ 2.3 Results ^ 2.3.1 Morphological and associated volume changes for LUVETs ^ 2.3.2 Quantitation of vesicle volume increase ^ 2.3.3 Source of the non-spherical morphology of LUVETs ^ 2.3.4 Osmotic lysis of large unilamellar vesicles ^ 2.3.5 Size of the membrane defect formed during osmotic lysis ^ 2.3.6 Influence of vesicle size on osmotic lysis ^ 2.4 Discussion ^  46 47 47 48 48 48  Chapter 3 Influence of plasma on the osmotic sensitivity of LUVETs ^  3.1 Introduction ^ 3.2 Materials and Methods ^ 3.2.1 Lipids and chemicals ^ 3.2.2 Preparation of LUVETs ^ 3.2.3 Blood collection ^ 3.2.4 Vesicle trapped volume measurements ^ 3.2.5 Determination of osmotically-induced solute release from LUVs ^ 3.2.6 Influence of solute molecular weight on osmoticallyinduced leakage ^ 3.2.7 Hemolytic assay of serum for complement activity ^ 3.2.8 Lipoprotein fractionation ^  49 49 50 50 51 51 52 52 52 52 55 56 60 69 69 72  76 76 77 77 78 78 78 78 79 79 80  3.3 Results ^ 3.3.1 Quantitation of vesicle volume increase ^ 3.3.2 Characterization of vesicle lysis in the presence of plasma ^ 3.3.3 Plasma component(s) responsible for lysis ^ 3.4 Discussion ^ Chapter 4 Influence of asymmetric transbilayer lipid distributions on the morphology of LUVs ^ 4.1 Introduction ^ 4.2 Materials and Methods ^ 4.2.1 Lipids and chemicals ^ 4.2.2 Preparation of LUVETs ^ 4.2.3 Generation of DOPG asymmetry ^ 4.2.4 Detection of DOPG asymmetry ^ 4.2.5 Cryo-electron microscopy ^ 4.3 Results ^ 4.3.1 Kinetics and quantification of DOPG transport ^ 4.3.2 Morphological changes resulting from the expansion of the vesicle's outer monolayer ^ 4.3.3 Morphological changes resulting from the expansion of the vesicle's inner monolayer ^ 4.3 Discussion ^ Chapter 5 Thesis summary ^  8I 81 83 92 96 100 100 102 102 102 102 103 103 104 104 106 108 115 119  LIST OF FIGURES Figure 1 The structure of a phospholipid and commonly occurring headgroups ^ 10 Figure 2 Polymorphic phases and corresponding dynamic molecular shapes of lipids ^ 15 Figure 3 Gel to liquid-crystalline phase transition ^ 18 Figure 4 Deformation of a small membrane element ^ 2I Figure 5 Derivation of Laplace's Law for a spherical vesicle ^ 28 Figure 6 Shapes adopted by erythrocytes ^ 33 Figure 7 Vesicle shapes predicted by the bilayer couple hypothesis ^ 36 Figure 8 Cryo-electron micrographs of extruded vesicles ^ 54 Figure 9 Trapped volume characterization of extruded vesicles ^ 58 Figure 10 Morphology of extruded vesicles revealed by cryo-electron microscopy ^ 59 Figure 11 Time course of carboxyfluorescein release from osmotically stressed vesicles ^ 61 Figure 12 Quench curve of carboxyfluorescein entrapped in EPC:Chol vesicles ^ .64 Figure 13 Influence of the osmotic differential on vesicle lysis ^ 68 Figure 14 Influence of vesicle size on osmotic lysis ^ 71 Figure 15 Influence of vesicle morphology on trapped volume ^ 82 Figure 16 Time course of carboxyfluorescein release from osmotically stressed vesicles ^ 84 Figure 17 Influence of the osmotic differential on vesicle lysis ^ 85 Figure 18 Calculated residual osmotic differential after vesicle lysis ^ 85 Figure 19 Quench curve of carboxyfluorescein entrapped in POPC:Chol vesicles ^ .89 Figure 20 Influence of plasma concentration on vesicle lysis ^ 91 Figure 21 Influence of albumin on vesicle lysis ^ 93 Figure 22 Influence of complement inactivated serum on vesicle lysis ^ 93  Figure 23 Influence of lipoprotein fractionated plasma on vesicle lysis ^ 95 Figure 24 Influence of VLDL and chylomicron. LDL. and HDL fractions on 95 vesicle lysis ^ Figure 25 Quantitation of DOPG asymmetry induced by a pH gradient 105 (pHo = 7.5. pHi = 4) ^ Figure 26 Morphological changes in DOPC:DOPG (9:1 ) vesicles generated 110 by the transport of DOPG to the outer inonolayer ^ Figure 27 Morphological changes in 100 nm diameter DOPC:Chol:DOPG (6:3:1) vesicles induced by the transport of DOPG to the outer 111 monolayer ^ Figure 28 Morphological changes of DOPC:DOPG (9:1) vesicles induced by the partitioning of monooleoyl PC into the outer monolayer 112 of 100 nm LUVETs ^ Figure 29 Morphological changes of DOPC:DOPG vesicles incubated with monooleoyl PC and subsequently exposed to a pH gradient (interior basic) to induce DOPG transport to the inner 113 monolayer ^ Figure 30 Influence of the aqueous medium on vesicle morphology ^ 114  LIST OF TABLES Table 1 Names and structures of some common fatty acids ^ Table 2 Lysis tension (-flys). area compressibility modulus (K) and the fractional increase in membrane area at lysis (ac) of membrane bilayers at 15°Ca or 25°Ct c ^ Table 3 Influence of re-extrusion on trapped volume of spherical vesicles ^ Table 4 Evaluation of theoretical models of solute release during lysis ^ Table 5 Influence of solute molecular weight on release during vesicle lysis ^ Table 6 Evaluation of theoretical models solute release during plasma enhanced lysis ^ Table 7 Influence of solute molecular weight on release during vesicle lysis in the presence of plasma ^  11  30 60 65 69 90 90  ABBREVIATIONS fractional change in surface area ( = (Af ac,^a at lysis Af^final surface area Ai^initial surface area bending modulus BSA^bovine serum albumin CF^carboxyfluorescein AA^difference between the outer and inner monolayer areas AA/As^relative difference in membrane areas where AA is the difference between the outer and inner monolayer areas and AAs is the corresponding value for a sphere AC^surface curvature change DGVB^dextrose gelatin veronal-buffered saline AP^osmotic pressure difference DSC^differential scanning calorimetry EGTA-DGVB ethylene glycol-bis-(b-amino ethyl ether) N. N'-tetraacetic acid dextrose gelatin veronal- buffered saline FATMLVs^frozen and thawed MLVs HDL^high density lipoprotein HEPES^N-(2-hydroxyethyl)piperazine-N'-2-ethanesulfonic acid H11^hexagonal area compressibility modulus ( = -r/a ) X^in-plane surface area extension ( = L1/L1 LDL^low density lipoprotein Lf^final surface length due to surface shear Li^initial surface length due to surface shear  lipids  Chol ^ cholesterol CL cardiolipin DOPC ^ dioleoylphosphatidylcholine DOPE ^ dioleoylphosphatidylethanolamine DOPG ^ dioleoylphosphatidylglycerol DPPC ^ dipalmitoylphosphatidylcholine EPC ^ egg phosphatidylcholine lyso PC ^ lysophosphatidylcholine ^ monooleoylphosphatidylcholine monooleoyl PC. NBD-PE ^ N-( 7-nitrobenz-2-oxa- 1 .3 -diazol- 4-y1)-DOPE ^ phosphatidic acid PA PC phosphaticlylcholine PE phosphatidylethanolamine PG phosphatidylglycerol PI phosphatidylinositol POPC ^ palmitoyloleoylphosphatidylcholine PS phosphatidylserine RHO-PE ^ N-(lissamine rhodamine b-sulfonyl) - DOPE SOPC ^ stearoyloleoylphosphatidylcholine  LUVETs^large unilamellar vesicles by extrusion techniques LUVs^large unilamellar vesicles bending moment MLVs^multilamellar vesicles NMR^nuclear magnetic resonance QELS^quasi-elastic light scattering vesicle radius RI, R2^principal (perpendicular) radius of curvature  a^shear force SRBs^sheep red blood cells SUVs^small unilamellar vesicles -c^membrane tension Te^gel to liquid-crystalline transition temperature at lysis TNS^2-(p-toluidinyl)naphthalene-6-sulfonic acid V/Vs^reduced vesicle volume. where V is the vesicle volume and Vs is the (maximum) volume of a sphere with the same membrane surface area  VLDL^very low density lipoprotein  ACKNOWLEDGMENTS  I'm very grateful for many things in my life and one of the pivotal ones is joining Pieter's lab. I wish to thank the many people in Pieter's lab who have made these past several years enjoyable and memorable. I would especially like to thank Tom Madden for his help throughout this thesis and Pieter for giving me the opportunity to grow  and learn.  TO MOM AND DAD AND GRANDMA  CHAPTER 1 INTRODUCTION  1.1 OVERVIEW  Biological membranes are not simply static barriers which separate an internal and external environment. Through the regulation of solutes which enter and leave the cell or cell organelle, biological membranes control intra- and extracellular activity. They also govern how the cell interacts with the external environment by the expression of integral membrane proteins and physical changes to the membrane itself. The dynamic complexity of biological membranes has lead to the use of simpler model lipid vesicle systems to study membrane properties. This thesis focuses on the effects of mechanical stress due to osmotic pressure (higher internal solute concentration) and transbilayer area asymmetry on model lipid vesicles. These forces may result in an increase in membrane tension, rupture (lysis), or deformation of vesicle shape and have been implicated in many diverse biological functions including membrane fusion, maintenance of cellular shape, and membrane trafficking. The ability of plasma proteins to interact with and lyse osmotically stressed vesicles is also examined. This is of interest both from the point of view of membrane-protein interactions as well as a more practical aspect as these model lipid vesicles are being utilized as in vivo drug carriers. This chapter reviews our current concept of biogical membranes and provides background knowledge on the characteristics of lipids and membranes which are relevant to the studies undertakened. Different model systems which have been used to study membrane properties will be reviewed. The force exerted on a vesicle membrane by a hypoosmotic gradient. resulting in an increase in membrane tension  1  and lysis, will be covered. Current knowledge of the destabilizing interactions of plasma proteins with membranes will be provided as this is the subject of Chapter 3. Subsequently, the rationalization of vesicle morphology based on the vesicle's membrane elasticity will be reviewed. The use of membrane vesicles as drug carriers as it relates to this present study will be discussed.  1.2 STRUCTURE OF BIOLOGICAL MEMBRANES  Biological membranes are composed mainly of lipids and proteins. The basic bilayer lipid structure was first proposed by Gorter and Grendel in 1925 when they observed that the area occupied by a monolayer dispersion of an erythrocyte lipid extract was approximately twice the original cell surface area. The involvement of proteins was later suggested by Danielli and Davson but it wasn't until 1972, when Singer and Nicholson developed the fluid mosaic model, that the present day model took form. In the fluid mosaic model, proteins are either embedded in a fluid lipid bilayer (integral membrane protein) or attached to the surface of the membrane by ionic interactions or hydrogen bonding (peripheral proteins). A general feature of biological membranes is the asymmetric distribution of different lipid and protein species. Protein asymmetry is important in the vectoral transport of solutes, cell structure via attachment of a cytoskeletal network, metabolism and many other enzymatic activities. Some of these proteins are glycosylated on the extracellular face and play important roles in cell attachment and as receptors in hormonal and immunoresponses. The function of lipid asymmetry is not fully understood but its importance is implicated in the observation that all biological membranes studied to date have shown an asymmetric distribution of their phospholipids and that this asymmetry is actively maintained (Devaux, 1991). Proposed functions of this asymmetry include regulation of the membrane's fusogenic  2  capacity (Cullis & Hope, 1988), maintenance of the erythrocyte discoid shape (Devaux, 1991), enzymatic activity of certain proteins (Houslay & Stanley, 1982), and intercellular recognition (Tanaka & Schroit, 1983: Schroit et al., 1985). Another general feature of biological membranes is the large diversity of lipid species. The erythrocyte membrane, for example, contains more than 100 different species (van Deenen et al.. 1974). Although the precise function of this diversity is not completely clear, different lipids are likely needed to modulate protein functions (Gennis, 1989), sterically stabilize membrane proteins (Sackman et al., 1984) and regulate membrane order (Lafeur et al., 1990). Both lipid diversity and membrane species asymmetry reinforce the idea that the formation of a semi-permeable barrier is not the sole function of membrane lipids but that many other subtle characteristics are needed for cellular viablity. The structural and functional characteristics of eukaryotic membranes are not dictated by the bilayer lipid/protein matrix alone. A complex protein network is attached to the intracellular membrane face and confers additional structural rigidity to the cell. Although the precise attachment of this cytoskeletal element is unknown, it can have a pronounced effect on the mechanical (Waugh, 1982; Evans et al., 1976) and morphological properties of the cell (Wortis et al., 1991). The cytoskeleton is also involved in other processes such as stabilizing asymmetric lipid distributions, cell motility, physical amplication of cellular signals. and movement of cellular organelles (for a review see Carraway & Carraway, 1989: Houslay & Stanley, 1983).  1.3 MODEL MEMBRANES  Due to the complexity of biological membranes, many different model systems have been developed to study membrane properties and protein-lipid interactions (for a review see Szoka & Papahadjopoulos. 1980: Hope et al., 1986). These systems are  easier to manipulate and information obtained from these studies is more defined. The purpose of model systems is not so much as to mimic biological situations but to help formulate basic concepts that are difficult to interpret from more complex biological systems. In addition, other practical applications have developed from these model systems (see Section 1.13). There are three main types of model membranes: monolayers, planar bilayers and liposomes.  1.3.1 Monolayers  Phospholipids and other amphiphilic molecules will spontaneously form an oriented monolayer if a small aliquot of a lipid solution, pre-dissolved in a volatile solvent, is applied onto an aqueous surface. The monolayer forms at the air-water interface with the lipid's polar domains in contact with the aqueous phase and their hydrophobic domains extended above. The major advantage of monolayer studies is that the lateral pressure can be externally varied and measured, thus providing information which is not available from bilayer experiments. Typically, these studies are carried out in a Langmuir film balance in which the monolayer surface area is measured at constant pressure or temperature. These pressure-area (isotherms) or temperature-area (isobar) curves can provide information about head-group interactions as a function of temperature, pressure, pH, salt concentrations and protein interactions. Several important findings have resulted from monolayer studies. One of the major observations is the "condensing" effect of cholesterol in which the area occupied by a cholesterol-phospholipid membrane is less than the sum of their areas in isolation (Demel & de Kruijff, 1976). A more spectulative proposal has been made by Georgallas et al. (1984) in which they modelled the degree of intermonolayer van der Waals interactions to explain differences in the phase transition temperature between monolayer films and bilayer membranes of dipalmitoylphosphatidylcholine.  4  -  The lateral membrane pressure measured with monolayer films is primarily due to the repulsive forces of the lipid headgroups. Typically, a 15 dyn/cm lateral pressure is needed to maintain the same headgroup area as in a bilayer vesicle. It should be noted however, that in the absence of osmotic forces, for example, a bilayer vesicle is in mechanical equilibrium where repulsive and attractive membrane forces are in balance, resulting in a net membrane tension of zero. Attractive interactions arise mainly from hydrophobic (interfacial) tension acting at the hydrocarbon-water interface while approximately equal repulsive forces are contributed from the acyl chain and head-group regions (for a review see Israelachvili et al., 1980).  1.3.2 Planar lipid membranes Planar lipid membranes are formed by dissolving lipid in an organic solvent and applying this to a thin aperture separating two aqueous compartments (Fettiplace et al., 1974). The solvent collects at the perimeter of the aperture, leaving behind a lipid bilayer. The main advantage of this technique is that the environment of the two compartments can be easily controlled and monitored. Planar lipid membranes have been used to study water permeability using tritiated water. However, a major limitation with these studies is that the effects of unstirred water layers at the membrane surface has to be taken into account and thus complicate interpretation of the results (Fettiplace & Haydon, 1980). Other disadvantages include possible permeability changes due to residual organic solvents, bilayer distortions created at the lipid-aperture interface and the relatively small membrane area involved in the measurements.  1.3.3 Liposomes The term liposomes refer to an aqueous dispersion of bilayer forming lipids that can differ in their size and larnellarity (single or multiple bilayers). Liposomes can be  5  arbitrarily divided into three catagories based on their size: giant, large and small vesicles with diameters of 1 - 30 jun, 0.5 - 0.05 jtm and <0.03 pm respectively.  1.3.3.1 Giant vesicles Giant vesicles are usually multilamellar and are formed by hydrating and dispersing (e.g. vortex mixing) dry lipid in an aqueous solution. The resulting multilamellar vesicles (MLVs) typically have diameters in excess of 1 jun and consist of concentric bilayers separated by narrow aqueous channels. The relatively large size and regular bilayer arrays of these MLVs have made them ideally suited for structural studies using X-ray diffraction. Interpretation of stuctural and motional studies by nuclear magnetic resonance are also more straightforward than in smaller systems due to a slower vesicle tumbling rate relative to the lipid's lateral diffusion and rotational times. MLVs have also been used to study water permeablity by monitoring the vesicle's swelling rate by light scattering in response to an osmotic gradient (Bangham et al., 1967; Blok et al., 1976). A mixture of giant unilamellar vesicles and MLV's can also be formed by slow hydration and dispersion of a thin lipid film in water or a low ionic strength solution (Reeves and Dowben, 1969). One currently used method involves coating a telfon disk with lipid solubilized in chloroform and allowing the solvent to evaporate to form a thin lipid film. The dried film is then prehydrated using a stream of argon saturated in water before exposure to an aqueous solution (Needham & Evans, 1988: Farge & Devaux, 1992). These vesicles can be visualized by light microscopy due to their large size and have been used to study the lateral diffusion of fluorescent probes by photobleaching (Fahey & Webb, 1978), vesicle morphology (e.g. Wortis et al., 1991) and membrane mechanical properties using micropipette aspiration techniques (e.g. Needham & Nunn, 1990). The major limitation with this technique is the rigorous conditions required for vesicle formation.  1.3.3.2 Large unilamellar vesicles  The techniques used to form large and small vesicles can result in unilainellar liposomes. Methods used for the formation of large unilarnellar vesicles (LUVs) include ethanol injection (Kremer et al., 1977), reverse phase evaporation (Szoka and Papahadjopoulas, 1978), detergent dialysis (for a review see Madden, 1986) and medium pressure extrusion (Hope et al., 1992). Lipids can be solubilized in organic solvents to form monomers and subsequently exposed or injected into an aqueous solution. The organic solvent can then be removed by heating the solution above the boiling point of the solvent, dialysis, gel filtration, or reduced pressure to form LUVs. LUVs can also be made by solubilizing lipid in detergent to form micelles then removing the detergent by gel filtration or dialysis. This technique is particularly useful in reconsituting membrane proteins into vesicles without protein denaturation. However, the production of LUVs from either organic solvents or detergent dialysis has limitations in that they are tedious and may contain residual solvents or detergents that could affect the physical properties of the membrane. An alternate technique used to form LUVs that has become very wide-spread is by repeatedly passing MLVs at intermediate pressures through polycarbonate filters of defined pore size. These vesicles result from shearing and fragmentation of the MLV membrane and are frequently called LUVETs (large unilamellar vesicles by extrusion techniques). LUVETs have been shown to be unilamellar by freeze-fracture, electron microscopy, nuclear magnetic resonance and quasi elastic light scattering techniques (Hope et al., 1992). Frequently, the MLVs used to produce LUVETs are repeatedly frozen in liquid nitrogen and thawed to promote solute equilibration and to increase the vesicle's trapped volume (Mayer et al., 1985). These frozen and thawed MLVs (FATMLVs) are morphologically different from MLVs in that the closely packed regular  7  bilayer arrays are transformed into other intravesicular structures such as vesicles inside vesicles and vesicles between lamellae (Mayer et al., 1985).  1.3.3.3 Small unilamellar vesicles  Small unilamellar vesicles can be formed either by sonication (Huang, 1969) or passage through a French press (Barenholz, 1979). Due to the high radius of curvature in these vesicles there is a higher percentage of phospholipid in the outer monolayer compared to the inner monolayer. In mixtures of lipids, this steric constraint can also lead to an asymmetric distribution of lipids as lipids with a more bulky polar headgroup will tend to favor the outer monolayer surface (for a discussion see Hope et al., 1989). Although SUVs are relatively simple to prepare they are unstable and tend to fuse to form larger structures. In addition, other physical properties of the lipid components are often perturbed (Culls et al., 1985).  1.4 CHEMICAL PROPERTIES OF MEMBRANE LIPIDS  The properties of a lipid system can be divided into their chemical properties, which describe the characteristics of individual molecules and their material properties, which depend on the cooperative interactions of the lipid population (Gruner 1987). These interactions are largely driven by the hydrophobic force which is the thermodynamic drive for the lipids to adopt a conformation such that contact between the nonpolar portions of the lipids and water is minimized. This "force" is entropic in origin and results from the unfavorable constraints placed on water as it packs around a nonpolar hydrocarbon. The purpose of this section is to provide background knowledge on the material properties of membranes discussed in subsequent sections. A more thorough review of the chemical properties of lipids is given by Small (1985).  8  1.4.1 Lipid Structure  The general definition of a lipid is a biological molecule that is soluble in organic solvents. This is a very broad definition and include a wide range of molecules including sterols, phospholipids. vitamins and hormones. However, the main lipid components in biological membranes are the phospholipids, sphingolipids, and cholesterol. The major lipid present in eukaryotic cells is phospholipid. These lipids are composed of a glycerol backbone with hydrophilic head group linked to the sn3 position via a phosphate ester and two acyl chains attached to the sni and sn2 positions. Phospholipids are further subdivided based on their polar headgroups. with the major types illustrated in Figure 1. The headgroup can influence the lipid's charge, polarity, size and hydrogen bonding capacity and thus would be expected to have a considerable influence on membrane property. For example, substitution of ethanolamine for choline produces drastic differences in the polymorphic phase preferences of these lipids (Section 1.5) and the gel-liquid crystalline phase transition (Section 1.6). The phospholipid's fatty acid constituents are variable between different types of membranes and according to organism or species. A representative list is given in Table 1. In erythrocyte membranes, the phospholipids contain mostly 16 and 18 carbon fatty acids with one chain fully saturated.  9  Figure 1 The Structure of a phospholipid and commonly occurring headgroups. Takened from Vance & Vance (1985).  Table 1 Names and structures of some common fatty acids # of carbon atoms  ^  structural formula  ^  name  saturated fatty acids 12 14 16 18 20  CH3(CH2)1 0002H CH3 (CH2)12CO2H CH3( CH2)140 02H  CH3(CH2)16CO2H CH3(CH2)18002H  lauric acid myristic acid palmitic acid stearic acid arachidic acid  unsaturated fatty acids ^ 16^CH3(C H2)6CH =CH(CH2) 7002H 18^CH3(CH2)7CH=CH(CH2)7CO2H 18^CH3( CH2)4( CH=CHCH2)2( CH2)6CO2H 20 CH3( CH2)4( CH:---CHCH2)4( CH2)2CO2H  palmitoleic acid oleic acid linoleic acid arachidonic acid  Another important class of lipids are the sphingolipids. The common feature of sphingolipids is a long-chain secondary amine, with the most common "long-chain base" being sphingosine. These bases occur as components of more complex lipids. When a fatty acid is acylated on the amide group of sphingolipids, a ceramide is formed. Modification of the C-1 hydroxyl of a ceramide by the addition of phosphocholine, carbohydrates. or carbohydrates linked to several sialic acid residues results in the formation of sphingomyelin. cerebrosides or gangliosides respectively. Sphingomylelin is a major component of nervous tissue and many plasma membranes. There are at least 50 different species of glycosphingolipids (cerebrosides and gangliosides). They are found on the outer surface of many plasma membranes and act as antigens and receptors for toxins, antibodies, and lectins. A third important class of lipids are the sterols. Cholesterol, the most common lipid of this class, consists of a rigid planar hydrophobic region and a relatively small hydroxyl group as its polar domain. Cholesterol is a major structural component of the plasma membrane of mammalian cells but is virtually absent from the endoplasmic reticulum and the mitochondria.  1.4.2 Acid-base properties Many lipids display weak acidic or basic properties due to the presence of various ionizable groups. At physiological pH, fatty acids. phosphatidylinositol (PI), phosphatidylserine (PS), phosphatidylglycerol (PG), cardiolipin (CL), and phosphatidic acid (PA) are negatively charged while phosphatidylcholine (PC) is zwitterionic. The pKa's of the lipid's ionizable groups are influenced by the membrane surface potential. giving rise to the concept of apparent and intrinsic pKa's. The intrinsic pl{a is the pH at the membrane surface when the group is 50 % charged. However, the pH at the membrane surface does not always correspond to the pH of the bulk solution. The charge on the membrane surface generates an electrostatic surface potential  which causes the redistribution of ions at the membrane/water interface. Thus a membrane containing negatively charged phospholipids will attract cations to the membrane surface. increasing the concentration of H+ ions at the interface, for example, resulting in a lower pH at the interface than in the bulk solution. This effect gives rise to an apparent pl{a, which is the pH of the bulk solution when the ionizable group is 50 % charged. For example, the piCa of various fatty acids in a lipid bilayer has been reported to be between 7.0 - 7.5 which is much higher than the piCa of 4.8 reported for free fatty acids (Tocanne & Tessie. 1990).  1.5 LIPID POLYMORPHISM  Lipid polymorphism refers to the fact that lipids can adopt many different structures upon hydration (for a review see Cullis et al., 1990). In addition to the familiar bilayer, lipids can form micelles and hexagonal structures (H11 phase). Lipid polymorphism has been rationalized in terms of the molecule's dynamic shape characteristics (Cullis and de Kruijff, 1979: Israelachvili et al., 1977). Lipids which have an inverted cone shape have a larger cross-sectional area at the polar domain compared to its hydrophobic domain and tend to form micelles. Hexagonal phase forming lipids have an conical shape whereas bilayer forming lipids are cylindrical. These concepts are illustrated in Figure 2. The lipid's phase preferences depend on both intrinsic and extrinsic factors. These include the nature of the lipid headgroup, the size and degree of unsaturation of the acyl chains, temperature, pH. the degree of hydration, the ionic strength of the medium, the presence of divalent cations and other lipids or proteins. Features, for example, which tend to favor a bilayer to H11 phase transition tend to increase the acyl chain cross sectional area relative to the headgroup area. Thus, the H11 phase will be favored by increases in temperature and  unsaturation of the acyl chains whereas increased size, ionization and hydration of the headgroup will inhibit the H11 phase formation. Although lipids capable of adopting non-bilayer phases in isolation are found in all biological membranes, their function is not fully understood. Perhaps the clearest example of their use however, is in membrane fusion where it is topologically impossible for fusion to occur between two bilayers without a departure from a bilayer structure. Consistent with this suggestion is the observation that the ability of erythrocyte phospholipids to fuse correlates with their ability to form non-bilayer phases (Culls & Hope, 1978). Visualization of the structural fusion intermediates, however, has not been successful. This is likely due to the short life time predicted for the fusion intermediates (Gruner, 1987: Siegel, 1986). In addition to membrane fusion, the function of non-bilayer forming lipids in biological membranes has been suggested to regulate membrane order (Lafleur et al., 1990), the transport of proteins across bilayers (Batenburg & de Kruijff, 1988) and the formation of tight junctions (Kachar & Reese, 1982).  LIPID  PHASE  -^MOLECULAR SHAPE  LYSOPHOSPHOLIPIDS  V"  —^—  DETERGENTS 4 17,j,-I °0 s. --• 0 .774-  ^-  MICELLAR  INVERTED CONE _  ,  "  I  -  PHOSPHATIDYLCHOLINE SPHINGOMYELIN PHOSPHATIDYLSER1NE PHOPHATIDYLINOSITOL PHOSPHATIDYLGLYCEROL PHOSPHATIDIC ACID CARDIOLIPIN DIGALACTOSYLDIGLYCER IDE -^-  71111!1/1!! a440.114 •  _  PHOSPHATIDICACID  CYLINDRICAL  BILAYER  PHOSPHATIDYLETHANOLAM1NE (UNSATURATED) CARDIOLIPIN - Ca 2+ PHOSPHATIDIC ACID- Ca 2+ (pH<6.0)  -^  /  ,  .,  'C, A  ,  ...  7  ......-----.. : •.  k)  /  -*-)b .^sii)‘4L'i •-i-^''' "•-,....^  IC_  (pH<3.0)  ,11-  A---  —_^-  PHOSPHATIDYLSE RIN E (pH <4.0) MONIOGALACTOSYLDIGLYCE R ID E HEXAGONAL (H11)  _CONE  Figure 2 Polymorphic phases and corresponding dynamic molecular shapes of lipids.  Takened from Vance & Vance (1985)  1.5 GEL-LIQUID CRYSTALLINE PHASE TRANSITION TEMPERATURE  Pure lipid membranes consisting of a single molecular species of lipid will undergo a sharp phase transition from a gel state to a fluid liquid-crystalline state with an increase in temperature. This transition involves an increase in the rotational freedom of the acyl chains as they depart from an all-trans configuration to the formation of gauche isomers (Figure 3A). A corresponding increase in the area per lipid molecule occurs. Differential scanning calorimetry (DSC) is the most common technique for characterizing the phase behavior of lipids in terms of the phase transition temperature (Tc). the enthalpy and width of the transition (Figure 3B). Tc is dependent upon the nature of the lipid head group, the acyl chains and the environment in which the lipid is dispersed (Silvius, 1982). In general, the Tc increases with acyl chain length and degree of saturation (Houslay & Stanley, 1982). The phospholipid's polar headgroup is also important. Charge repulsion between adjacent negatively charged species can cause lateral expansion of the lipid bilayer, favouring the liquid-crystalline state. For example, the Tc of dipalmitoylphosphatic acid is 66°C at pH 6 where it is a monovalent anion, but decreases to 43°C at pH 12 where it becomes divalent (Marsh, 1990). This implies that environmental factors, such as the presence of divalent cations or low pH. can influence the Tc of acidic phospholipids by altering their charge repulsion (Marsh, 1990). A characteristic of lipids at or approaching the transition temperature is that the membrane becomes very permeable due to segregation of crystalline domains. Cholesterol has an important modulatory effect on phospholipid phase behavior. Firstly, the presence of >30 mole percent cholesterol abolishes the phospholipid's sharp phase transition by disrupting the lipid's cooperativity. This is reflected in the broadening of the width of the phase transition in DSC scans (Figure  - 16 -  3B). Secondly, relative to the pure phospholipid membrane, the presence of cholesterol above the phospholipid's Te increases the membrane order due to cholesterol's condensing effect. Below the phospholipid's Tc however, cholesterol decreases the relative membrane order by preventing the acyl chains from adopting an all-trans configuration. The biological importance of the ability of lipids to adopt the gel phase has remained obscure as there is no evidence for the presence of gel-state lipid at physiological temperatures in eukaryotic membranes (Cullis & Hope, 1985). However, it has been suggested that protein activity, for example, can be modulated by the formation of gel-state domains within a liquid-crystalline bilayer (Sacktnan 1984). The main difficulty with this idea, however, lies with how such segregation can be regulated (Cullis & Hope. 1985).  ^ (A) ^ AU trans  (8) First-order Kink (2G1)  GTG  Endothermic Pure OPPC  +5mol% cholesterol  +12.5 mol% cholesterol + 20 mol% cholesterol  290^320^350 Average temperature (°K)  + 32 mol% cholesterol + 50 mol% cholesterol  Figure 3 Gel to liquid-crystalline phase transition. A. The configuration of the phospholipid acyl chains in the gel state is in an all-trans configuration but gauche isomer formation occurs in the liquid-crystalline state. B. A DSC curve for dipalmitoylphosphatidylcholine and cholesterol mixtures shows that cholesterol. broadens the phase transition by disrupting the phospholipid's cooperativity (from Houslay & Stanley. 1982).  1.7 LATERAL AND TRANSBILAYER LIPID MOVEMENT Phospholipids. fatty acids and cholesterol can migrate rapidly in the lateral plane of the bilayer when the phospholipid is in the liquid-crystalline state. The diffusion coefficients for such movement are in the order of 10-8 cm2 s-1 in both model and biological membranes (Houslay & Stanley, 1982). For a 100 nm vesicle, for example, this implies that a lipid can completely diffuse around the vesicle membrane in 30 msec. Lipid flip-flop, which is the movement of lipid molecules from one half of the bilayer to the other, is an extremely slow process for phospholipid molecules in model membranes with half-times on the order of days (Houslay & Stanley, 1982). Phospholipid flip-flop requires the movement of the polar phospholipid headgroup through the hydrocarbon domain and is thus, a thermodynamically unfavorable process. This applies particularly to charged molecules. In biological membranes, the rates of transbilayer movement are more variable. The half-times of phospholipids transbilayer diffusion range from a few hours to days (Devaux, 1991). Faster halftimes in the order of minutes can be obtained with protein-mediated facilitated or active transport (Devaux. 1991). Transbilayer movement of cholesterol, on the other hand, is relatively fast with half-times of a few minutes in both biological and model lipid membranes (for a review see Schroeder et al., 1991).  1.8 MECHANICAL PROPERTIES OF MEMBRANES  Membrane mechanics examines how the membrane responds to physical forces usually by modelling the membrane as a continuum (for a review see Evans & Hochmuth, 1978). A continuum specifies that the time and space over which membrane properties are measured must include a sufficient number of molecules such that the fluctuations due to the behavior of individual molecules are small.  Thus, biological membranes can only ideally be analysed as a 2-dimensional continuum in the plane of the membrane surface as the membrane retains molecular characteristics and thus discontinuities in the thickness dimension.  1.8.1 Membrane deformation  External forces applied as normal, tangent or moment resultants to the membrane surface can result in deformation of membrane shape. These shape changes for local surface regions (elements) can be independently quantified into: area change, a = (Af - Ai)/Ai: in-plane shear (i.e. stretch of the membrane at constant surface area). A, = L1/L1: and bending or curvature change. AC = A(1/121 + 1 /R2). Af and Ai are the final and initial membrane areas respectively and laf and Li are the corresponding membrane lengths (in any one direction). R1 and R2 are the principal (perpendicular) radii of curvature with A(1 /R1 + 1 /R2) denoting the change between the initial and final state. These deformations are shown in Figure 4. Osmotic swelling of spherical vesicles is an example of an external force that will result in a change in membrane area: micropipet aspiration of a red blood cell can result in membrane shearing ; and vesicle shape changes will result in curvature changes. 1.8.2 Membrane elasticity  Membranes can exhibit elastic (thermodynamically reversible), viscoelastic (irreversible), and viscoplastic (irreversible with permanent material changes) behavior. Viscoplastic flow can occur in erythrocytes when an extreme or prolonged force is applied as for example. by micropipet aspiration. Irrecoverable changes are thought to occur in the cytoskeletal protein network. Viscoelastic behavior, in which energy is lost in terms of heat or work, can result from both internal and external dissipative forces. However, for membrane deformations that are produced at a slow rate without plastic flow, elastic analysis of membrane behavior is a reasonable representation.  area change  surface extension at constant area (shear) Li  bending  Figure 4 Deformation of a small membrane element. Membrane deformation can be described by three independent elastic modulii corresponding to area change (K). shear at constant area (pt) and bending (B). a = (A1 - ANA; is the elemental area change. X = Lf/li is the shear ratio. and RI and R2 are the principal (perpendicular) curvatures (adapted from Evans & Needham. 1987).  For elastic material behavior there are three first-order relations which are expressed as proportionalities between intensive (independent of the amount of material) forces and static shape changes. (i) Firstly, the fractional change in surface area (a) and isotropic tension (T) is related by an area compressibility modulus. K: = Ka  The value of K is relatively large (102 - 103 dyn/cm). indicating that membranes are very resistant to area changes. Typically, a membrane will only increase its area by 25 % before the membrane ruptures (Evans & Needham. 1987; Needham & Nunn, 1990). (ii) Secondly. the in-plane surface area extension (k) is related to the shear force a by: 0 = tt(k2_k- 2)/2  wherell is defined as the surface shear modulus. Lipid membranes, whose mechanical behaviour is generally like that of a 2-dimensional liquid has practically no resistance to shear forces. Thus.tt is approximately 0 for lipid membranes. However, the same is not true for biological membranes, such as the erythrocyte, where the attachment of cytoskeletal elements confers shear rigidity to the membrane (Waugh & Evans. 1979: Evans, 1983). (iii) Lastly, changes in membrane curvature are related to bending moments M by: M = BA(1 /RI + 1/R2)  where B is the bending modulus. As reflected in the relatively small bending modulus (10-12 dyn /cm), the membrane's bending resistance is essentially negligible due to the  extreme thinness of the bilayer. However, it should be noted that the above relationship is a simplified representation. Biological membranes are composed of two opposing monolayers and thus. relative changes in free energies (e.g. area) between the two monolayers will change the bilayer membrane bending energy by inducing a bending moment (Evans, 1974: section 1.1 1).  1.9 MEMBRANE PERMEABILITY  The selective permeability property of a membrane is one of their basic aspects. There is a wide permeability range for different molecules. For example. the permeability coefficients range from 10-3 - 10-4 cm s-1 for compounds which rapidly cross membranes (e.g. water), to 10-10 cm s-1 for moderately permeable molecules (e.g. glucose), to 10-13 cm s-1 for "impermeable" ions such as sodium (Walter & Gutknecht, 1984; Deamer & Bramhall, 1986). For efflux out of a 100 nm vesicle, these values correspond to half-times in the range of milliseconds for water, a few hours for glucose and several days for sodium. In general. membrane permeability is usually greater in cells than in vesicles formed from the lipid extracts (Carruthers and Melchior, 1983). This may be due to transport through protein dependent pores or defects formed at the lipid/protein interface.  1.9.1 Water permeability  The osmotic sensitivity of most membrane systems (Carruthers & Melchior, 1983) results from the much higher permeability of water compared to other solutes. Water movement across pure lipid membranes has been postulated to occur by diffusion or through transient defects. In the solubility-diffusion model, individual water molecules dissolve in the non-polar region of the bilayer and cross by simple  diffusion. This model is supported by several observations including the high activation energy for water transport which would be consistent with the diffusion of water across a hydrophobic domain. In addition. Cass and Finkelstein (1967) observed a good agreement between a measured water permeability and a calculated value based on estimated water diffusion and partition coefficients in the hydrophobic region of the bilayer. Water transport has also been thought to occur through transient defects which can arise from thermal fluctuations in the lipid bilayer. This model is best supported by the observation that water permeability is only moderately influenced by the lipid hydrocarbon chain length or the degree of unsaturation whereas water solubility into the hydrocarbon region is markly affected (Carruthers and Melchior, 1983). The two models are not mutually exclusive and Deamer and Bramhall (1986) suggest that water permeation occurs mostly through monomer diffusion but that larger transient defects occasionally permit bulk water and solutes to cross the barrier. Water permeability can be measured either by monitoring the isoosmotic exchange of water (Pd) or in response to an osmotic gradient (Pf). Measurements of Pd are complicated by the presence of unstirred aqueous layers adjacent to the membrane which give rise to a lower apparent Pd (Fettiplace & Haydon, 1980). However, when corrections are made for this effect, ratios of Pf/Pd are generally near unity for pure lipid membranes but can be as great as 6 for the erythryocyte (Cass & Finkelstein, 1967) and many other biological membranes (Prescott and Zeuthen; Koefoed-Johnsen and Ussing. 1952). This greater Pf/Pd ratio reflects the presence of pores that allow for water movement. Pf is expected to be greater than Pd in these situations as Pd is a measure of water diffusion whereas Pf measures both diffusion and filtration. Water flow by filtration increases with pressure.  1.9.2 Nonionic solute permeability  The permeability of nonionic solutes has been well studied and the rate limiting step is thought to be the partitioning of the molecule into the hydrophobic domain (Jain, 1980). This has lead to the generalization known as "Overton's rule" which relates the hydrophobicity of a molecule to its permeability (Walter and Gutknecht, 1984). However, as for water permeability, solute transport has also been suggested to occur through small transient defects in the lipid acyl chain region as very small molecules can permeate faster than predicted by their partition and diffusion coefficients (Lieb and Stein, 1986). Other factors such hydrogen bonding capacity can also influence solute permeability (de Gier et al.. 1970: McElhaney, 1986).  1.9.3 Ion permeabiliy  Ion permeability is generally much lower than solute permeability. The major energetic barrier to electrolyte permeability is the Born energy barrier (Parsegian, 1969), which is the energy needed to move an ion from a high dielectric medium (buffer) to a medium with a lower dielectric constant (lipid hydrocarbon region). The Born energy can be decreased by decreasing the ionic charge. increasing the ionic radius, or increasing the dielectric constant of the hydrocarbon. This model is supported by the observation that raising the membrane dielectric constant with chlorodecane increases the permeability of thiocyanate (Dilger and McLaughlin, 1979). However, Born energy considerations do not account for greater anion permeabilities compared to cations. For example. the tetraphenylboron anion is much more permeable than the tetraphenylphosphonium cation even though these molecules have similar hydrophobicity and hydrogen bonding abilities (Flewelling and Hubbell, 1986a) These authors (1986b) proposed that a positive "internal dipole" is created at the membrane's hydrophilic/hydrophobic interface, favoring anion permeation. This dipole is suggested to be due primarily to the alignment the phospholipid's ester  oxygens along the plane of the bilayer. with the oxygens pointed towards the aqueous phase (Flewelling and Hubbell. 1986b). A "total potential" model of membranes, which includes Born energy. dipole potential. image energy (a function of the electrical interaction between the ion in the hydrocarbon and the hydrophilic interface), and neutral energy terms (all other nonelectrical interactions such as hydrophobic, van der Walls and steric factors) has been proposed to explain observed ionic permeabilities (Flewelling and Hubbell. 1986b).  1.10 OSMOTIC PRESSURE  1.10.1 Theory  An osmotic pressure difference will develop if two solutions, separated by a semi-permeable membrane, contain unequal concentrations of an impermeable solute. For example, imagine that compartment 1 contains a lower solute concentration than compartment 2, with temperature and pressure in both compartments being equal. The membrane barrier separating the two compartments is permeable to water but not to the solute. From a statistical perspective, the higher concentration of water molecules in compartment 1 will result in a net movement of water to compartment 2. The pressure required to maintain the initial volumes in both compartments is defined as the osmotic pressure difference between the two compartments. From a thermodynamic viewpoint. an osmotic pressure difference develops due to differences in chemical potential between the two solutions in isolation. For the situation described above, the chemical potential of water. u will be the same in both compartments at equilibrium:  tiw(1) = tt,.(2)  Note that a similar condition does not apply to the solute since, by definition, the solute does not have access to both compartments and thus cannot come to equilibrium. The chemical potential of a solution at constant temperature is given by:  p., =^+ RThiX„, + PV„,  where p* is the standard chemical potential of water. X„, is its mole fraction, V, is its partial molar volume. P is the hydrostatic pressure of the solution. R is the gas constant and T is the temperature. By combining the above two equations and simplifying the case for dilute solutions results in the van't Hoff expression for osmotic pressure:  AP = P( 1 ) - P(2) = RTAc  where AP and Ac are, respectively, the osmotic pressure and concentration difference between the two compartments.  1.10.2 Laplace's law Laplace's law relates the pressure difference across a closed elastic membrane to the tension in the membrane. Its specific form depends on the shape of the closed surface. For the studies described in this thesis, the most relevant shape to consider is a spherical object that is subjected to an higher internal pressure. The two forces which act in this system are (i) the pressure difference between the vesicle's interior (Pi) and exterior (P.) environment and (ii) the force per unit length exerted by the elastic membrane or the membrane tension (t). As a sphere is symmetrical. Laplace's Law can be derived by considering the forces required to maintain a hemisphere stationary  on a planar surface (Figure 5). The force acting to pull the hemisphere up off the plane due to the higher internal pressure is the product of the pressure difference. AP, times the contact area. nr2, where r is the radius of the hemisphere. The reacting force pulling the hemisphere down is due to the membrane tension and is the product of t and the circumference 2nr of the hemisphere. At equilibrium these two forces are in balanced: 21crt = Al3nr2 or AP =2t / r.  Figure 5 Derivation of Laplace's Law for a spherical vesicle. The forces acting on a spherical elastic membrane due to a higher internal pressure and the membrane tension can be simplified for the case of hemisphere. The downward arrows represent the forces exerted by the elastic membrane or the membrane tension. The upward arrows represent the forces due to the pressure difference.  1.10.3 Biological systems As mentioned in Section 1.9. model and biological membranes are much more permeable to water than to most other molecules and thus are sensitive to osmotic gradients. In biological systems. osmotic pressure has been suggested to be regulated by mechanosensitive ion channels which are found in a wide variety of different cells (for a review see Morris, 1990). For example, when kidney cells are exposed to a hypoosmotic gradient or mechanically stressed using a patch-clamp technique. an increase in ion channel activity is observed (Ubl et al.. 1988). Similarly, the kinetics of cation channel activity has be found to correlate with osmotic swelling and shrinking of the neuroblastoma cells (Falke & Misler, 1989). Osmotic gradients may also play a role in membrane fusion-dependent processes such as exocytosis by increasing the membrane tension (Cohen et al.. 1980; Hampton & Holz. 1983; Pollard et al., 1984).  1.10.4 Membrane rupture When a vesicle is exposed to a hyperosmotic gradient water will flow into the vesicle. This volume increase is first accommodated by the vesicle adopting a more spherical shape as it is much easier to change the membrane curvature than to increase the membrane area. Once the vesicle has maximized its volume through morphological changes. any further influx of water would result in an increase in membrane tension as predicted by Laplace's law and will ultimately lead to vesicle lysis. The membrane tension at which lysis occurs has been predomimently measured with giant vesicles using the micropipet aspiration technique. This lysis tension is dependent on the lipid composition and is typically in the range of 1 30 dyn/cm. Furthermore. membranes with high lysis tensions exhibit greater values in their area elasticity modulus. which is the force needed to stretch or compress the membrane a given length (see section 1.8). A representative list of the area elasticity  modulus (K), the critical area change (ac = a at lysis). and the membrane tension at lysis (Tlys) is given in Table 2.  Table 2 Lysis tension (Tiys), area compressibility modulus (K) and the fractional increase in membrane area at lysis (a„) of membrane bilayers at 15°C a or 25°C b,c  Lipid composition (mol:mol)  (dyn / cm)  SOPC SOPC:Chol (72:28) SOPC:Chol (50:50) bovine sphingomyelin:Chol (50:50) erythrocyte  190 240 780 1720 450b  ac  0.03 0.05 0.03 0.02 -0.03c  T lys (dyn CM)  6 13 20 23 -10c  a values from Needham & Nunn (1990) unless indicated otherwise b value from Evans (1977) C values from Evans & Waugh (1976)  The mechanism of membrane rupture is likely due to fluctuations in membrane density leading to pore formation and not the result of area dilation of individual molecules. This was initially calculated for erythrocytes by Evans et al. (1976) when they calculated the energy needed to rupture the lipid bilayer. Based on a measured 3 % increase in membrane area upon lysis (at) and an area elasticity modulus (K) of 95 dyn/cm. they estimated that the energy needed to expand the lipid bilayer to the point of rupture is approximately 3 cal/mol. This energy is considerably less than the calculated free energy of approximately 25 cal/mol needed to create waterhydrocarbon (hydrophobic) interactions corresponding to a 2 % area dilation (Tanford, 1973). This analysis also holds for pure lipid membranes based on measured values of ac and K (Evans & Needham. 1987: Needham and Nunn. 1990).  From the fluctuation theory of nucleation, the formation of these pores is believed to be a stochastic process (for a review see Dimitrov & Jain. 19841. Membrane rupture occurs when the pores formed are above a critical radius while pores formed below a critical radius tend to reseal. The critical size of the pores is proportional to the linear tension at the pore edges and inversely proportional to the membrane tension. This theory has been difficult to verify experimentally due to difficulties in estimating values for the line or edge tension (Dimitrov & Jain. 1984). In addition, the presence of impurities may also act to initiate and or stabilize the pores (Kashchiev & Exerowa. 1983: Bloom et al.. 1991). Once the critical size of the pores have been exceeded. Taupin et al. (1975) have suggested that lipid vesicles completely rupture due to indefinite expansion of the pore. This can occur when erythrocytes are osmotically lysed, for example, but is not the case for vesicle systems used in the present study (Chapters 2 & 3).  1.11 VESICLE MORPHOLOGY  Initial interest in cellular morphology was largely stimulated by the unique biconcave shape (discocyte) of the erythrocyte (Figure 6). Under physiological conditions, the cell volume is 60 % smaller than the (maximum) volume encompassed by a sphere of the same surface area. Hence, an erythrocyte could assume a variety of shapes. However, even after emerging from very small blood vesicles (Branemark & Lindstrom, 1963) or after being aspirated into a micropipet (Rand & Burton. 1964). erythrocytes resume their biconcave shape within seconds. Maintenance of the discocyte shape has been postulated to be due to an internal mechanical structure or due to the properties of the membrane itself. However. interculation of various amphipathic compounds into the membrane can change the erythrocyte biconcave shape to a stomatocyte or echinocyte (Deuticke. 1968: Sheetz & Singer. 1974). These  shapes have been modeled with some success by considering solely the energetics of the lipid bilayer (for a review see Wortis et al.. 1991). In addition to maintaining the biconcave erythrocyte shape. other dynamic membrane processes. such as endo and exocytosis. has been postulated to be energetically membrane driven.  1.11.1 Vesicle shape as determined by the membrane's elastic energy The shape formed by a given membrane is generally believed to correspond to the minimum value of the membrane elastic energy. This energy can be decomposed into the sum of the stretching. shear and bending energies. As mentioned in Section 1.4.3, it is much harder to stretch a membrane than to cause shear or bending deformations. Thus. vesicle shape or change in vesicle shape is determined by the membrane's shearing and bending energies with little change in membrane area. For lipid membranes which do not exhibit shear elasticity. only the bending energy is important. For biological membranes, in which an underlying cytoskeletal network and external glycocalyx can modify the physical properties of the membrane, the situation is more complex. For example. the bending elasticity of the erythrocyte membrane can be altered by chemical changes in the cytoskeletal network (Zeman et al., 1990) and as mentioned previously, this cytoskeletal network also confer shear elasticity to the membrane (Waugh. 1982: Evans. 1983). However, it is of interest to note that although biological systems are much more complex, the concept that the membrane's bending energy can stabilize cellular shape has been successful in rationalizing the biconcave erythrocyte shape for a given vesicle volume and surface area as first introduced by Can ham (1970).  Figure 6 Shapes adopted by erythrocytes. Intercalation of arnphipathic molecules into the erythrocyte's outer monolayer causes the cell to undergo a transition from a discocyte to an echinocyte while insertion of molecules into the membrane's inner monolayer causes the cell to undergo a discocyte to stomatocyte transition.  - 33 -  1.11.2 Influence of transbilayer area asymmetry between the two monolayers In 1974, Evans proposed that differences in surface tension between the two monolayers can produce membrane curvature (Evans. 1974). That is, since vesicle membranes are closed systems such that the monolayers cannot relieve differences in transbilayer membrane tension by sliding past each other, expansion of one monolayer relative to the other would cause the membrane to bend to relieve the strain. The influence of this area asymmetry on vesicle shape was qualitatively shown by Sheetz and Singer (1974) when they suggested a bilayer couple hypothesis to explain the different shape changes adopted by erythrocytes when various amphipathic molecules were interculated in the outer or inner monolayer of the cell membrane. Since the cytoplasmic face of the erythrocyte membrane is highly negative and the external face is neutral, permeable cationic amphipathic molecules will bind preferentially to the cytoplasmic half of the bilayer and thus expand the inner monolayer. These molecules were observed to cause the cells to invaginate inwards to form stomatocytes (Figure 6). Incubation of erythrocytes with anionic or impermeable amphipathic drugs has the opposite effect in that the cells formed outward invaginations or echinocytes, indicating that the drugs are partitioning into the outer monolayer.  1.11.3 Theoretical models The idea that vesicle morphology corresponds to the minimum membrane bending energy for a given vesicle volume, surface area and area difference between the two monolayers has been incorporated into more recent theoretical models. One diagramatic model developed by Svetina and Zeks is the bilayer couple hypothesis (1983, 1989). In their model. different vesicle shapes are explained in terms of two phenomenological parameters: (1) the reduced volume of the cell. V/Vs, where Vs is the maximum volume of a sphere of a given membrane area and (2) the relative difference in area AA/AAs between the outer and the inner leaflet of the membrane,  where AAs is the corresponding value for the sphere. By minimizing the total bending elastic energy for given values of V/Vs and AA/As a phase diagram of cell shapes can be formed. The vesicle shapes predicted by the bilayer couple hypothesis is diagrammed in Figure 8. A more recent proposal by Seifeit (1992) suggests that the membrane bending modulus can be variable in different membrane regions. These differences can result from domain structures formed by lateral redistribution of lipid in response to membrane curvature. A similar concept was proposed by Isrealachvili et al. (1980) when they considered the membrane's curvature elasticity in response to the lipid's dynamic shape. For example. in regions of high curvature, clustering of "wedgeshaped" lipids or proteins will tend to occur iii the outer monolayer while cone-shaped lipids will aggregate in the inner monolayer. Formation of these domains may also aid in membrane fission as the domain boundaries would occur at the neck of the bud (Lipowsky, 1992).  Figure 7 Vesicle shapes predicted by the bilayer couple hypothesis. Vesicle morphology predicted by the bilayer couple hypothesis correspond to a membrane bending energy minimum for a given vesicle volume, surface area, and transbilayer area difference. AA and V are the vesicle's inner and outer monolayer surface area difference and volume, respectively, while AA, and Vs are the corresponding values for a spherical vesicle of the same bilayer membrane area (average of the inner and outer membrane areas). Taken from Sacicman et al., 1986.  1.11.4 Biological implications Membrane curvature changes triggered by alterations in the relative areas of the two monolayers are suggested to be a mechanism for membrane budding leading to fission. exo- and endocytosis. These changes can be very slight as less than a 1 % change in relative surface areas can produce dramatic shape changes in erythrocytes (see Oster et al.. 1989) and giant lipid vesicles (Farge & Devaux. 1992). Transbilayer area asymmetry may arise from extrinsic factors similar to those which influence lipid polymorphism (Oster et al.. 1989: section 1.4). For example. binding of Ca+2 to the anionic cytoplasmic face of the erythrocyte membrane would decrease the inner monolayer's surface area. Area asymmetry may also develop from the actions of lipid pumps or lipid biogenesis. Lipid pumps specific for aminophospholipids have been demonstrated in the plasma membranes of many cells. mediating the transport of PS and PE into the inner monolayer (for a review see Devaux. 1991). This inward movement of lipid may facilitate or trigger endocytosis by inducing inward invagination of the membrane. A more intriguing situation occurs within the cell where a rapid turnover of lipid from the endoplasmic reticulum to the golgi to the plasma membrane occurs. It has been estimated that half the lipid from the endoplasmic reticulum leaves the organelle's surface every 10 minutes (Wieland et al.. 1987). Lipid biogensis occurs predominently on the cytoplasmic face of the endoplasmic reticulum membrane (for a review see Op den Kamp. 1979; Bishop & Bell, 1988) and is thus topologically consistent with budding of the membrane leading to the formation and subsequent exocytosis of transport vesicles. In addition to transport vesicles. narrow tubular lipid structures, referred to as "tubulovesicular processes". have been observed to occur between the cisternae of the endoplasmic reticulum and the trans-gogli network (Cooper et al.. 1990: Lee et al.. 1988: Dabora et al.. 1988: Lippincott-Schwartz et al., 1990). Transport of lipids and proteins may occur along these processes by diffusion.  vesiculation or a combination of both. Oster et al. (1989) further spectulates that a passive lipid and protein sorting mechanism may be involved in such transport as some lipids are more prone to vesiculation or stable in membranes of high curvature. The endoplasmic reticulum also contains a PC specific "flippase" which transports PC to the inner monolayer (Bishop and Bell. 1985). Thus, competing rates of lipid biogensis and PC transport may control the flux of material from the endoplasmic reticulum to the golgi.  1.12 LIPOSOME-PLASMA INTERACTIONS  Interest in liposome stability in the presence of plasma has been largely stimulated by the use of liposomes as carriers for drugs. proteins, and nucleic acids in vivo. Early work has shown that an enhanced release of liposomal solute can occur upon exposure of the vesicle to plasma (Gregoriadis. 1973: Scherphof et al.. 1978). This effect is attributed mainly to the interactions of plasma lipoproteins (for a review see Bonte & Juliano, 1986: Scherphof et al., 1981). In addition, plasma complement proteins may also have a disruptive effect on fiposomal membranes and thus will be discussed in this section (for a review see Aiving & Richards, 1983: Muller-Eberhard. 1986)  1.12.1 Influence of membrane properties  Protein binding can be greatly influenced by surface charge and hydrophobicity of the liposomal membrane. Surface charge can be conferred by the phospholipids themselves or lipid conjugated moieties and is an important factor in initial membrane-protein interactions. Sterically coating the liposome surface with large hydrophilic moieties such as gangliosides (Allen et al.. 1985) or polyethyleneglycol-PE conjugates (Blume & Cevc. 1990) can abolish the lytic effects of plasma, by preventing  membrane-protein binding (Chonn et al.. 1992). An increase in the hydrophobicity of the membrane surface through transient defects or by a reduction in packing density of the lipids will generally alter and enhance protein interactions in the hydrophobic membrane domain.  1.12.2 Types of protein membrane interactions -  Exposure of liposomes to plasma in vivo or in vitro results in the binding of a wide variety of proteins to the membrane (Chonn et al. 1992). As previously observed with single protein systems. the effects of these interactions can be variable. leading Papahadjopoulos and co-workers to group membrane-protein interactions into three categories based on phase transition, permeability and monolayer studies (Papahadjopoulos et al.. 1975). The first type of interaction involves only surface binding without penetration into the hydrocarbon region. This type of binding (e.g. ribonuclease, polylysine) is reduced by raising the ionic strength of the solution. The second type of interactions (e.g. apolipoproteins, cytochrome c) initially involves electrostatic forces but the protein is capable of undergoing conformational changes to expose hydrophobic regions for further interactions. The conformation of the inserted protein domain can cause gross changes in the membrane including expansion of the bilayer, a decrease in the phase transition temperature and alterations in the bilayer permeability properties. The third type interaction also involves protein insertion but it does not produce major changes in the phase transition temperature although membrane permeability is increased. This type of interaction has been suggested to be a mechanism by which intrinsic membrane proteins (e.g. myelin proteolipid) are inserted into the membrane (Widmer, 1979).  1.12.3 Plasma proteins involved in liposome destabilization  1.12.3.1 Lipoproteins Lipoproteins are a class of proteins encompassing a wide spectrum of lipidprotein complexes of varying size. composition and function (for a review see Morrisett et al., 1977). Lipoproteins are most commonly subdivided into four operational groups based on their densities which include the chylomicrons. very low density lipoproteins (VLDL). low density lipoproteins (LDL) and high density lipoproteins (HDL). The general lipoprotein structure consist of a core of neutral lipids (triglycerides and cholesterylesters) surrounded by a film of amphiphatic lipids and proteins. As reflected in their different densities. HDL has the highest protein to lipid ratio whereas the chylomicrons have the least. The protein components of lipoproteins (apoproteins) have many functions including maintaince of the lipoprotein structure, co-factors of lipolytic enzymes. cell receptors. and mediators of lipid transfer between lipoprotein particles. The major apoproteins associated with HDL are apoA-I and apoA-H. ApoB is the major protein in LDL and is also a principal constituent of chylomicrons and VLDL. The apoC proteins are major protein constituents of chylomicrons and VLDL and minor components of HDL. The main lipoprotein involved in liposomal destabilization is attributed by many groups to be HDL (Tall & Green. 1981: Chobanian et al.. 1979: Scherphof et al., 1978; Krupp et al., 1976). Solute release from cholesterol-poor SUVs (Damen et al., 1981; Scherphof & Morselt. 1984) or with high protein to vesicle phospholipid ratios (Klausner et al.. 1985) has been attributed to the loss of liposomal structure. However, enhanced solute release has also been observed to occur from liposomes which appear to retain their structure in the presence of plasma (Kirby & Gregoriadis. 1981: Yoshioka et al., 1984) or apoA-1 (Comiskey & Heath. 1990). Membrane  permeability changes as a result of lipoprotein interactions can result from lipid transfer or protein-membrane interactions. Transfer of lipoprotein-derived lipids to the liposome may form nucleation sites for membrane defects. However, in comparison to HDL similar or greater solute loss has been shown to occur from liposomes incubated with HDL-delipidated proteins. suggesting that lipid flux into the liposome membrane has little disruptive effect (Weinstein et al., 1981: Comiskey & Heath. 1990). Alternatively, net transfer of liposomal lipid out of the vesicle could increase the membrane tension. Specifically. Yoshioka et al. (1984) demonstrated that loss of liposomal lipid from egg PC LUVs corresponded to a reduction in vesicle trapped volume. However, this effect is more important in pure phospholipid liposomes as the presence of cholesterol greatly reduces the net transfer of phospholipid from the liposome (Damen et al.. 1981: Kirby et al., 1980). Thus, in cholesterol containing systems. the ability of HDL to increase membrane permeability is primarily considered to be the result of protein binding with penetration into the hydrophobic domain. Several different studies indicate that the ability of HDL to disrupt the membrane is sensitive to the membrane packing density. Serum-induced leakage, for example. is less pronounced for liposomal systems composed of cholesterol with either saturated PC or sphingomyelin (Gregoriadis & Senior. 1980; Damen et al., 1981: Allen & Everest, 1983). Secondly, the presence of bilayer defects, such as those arising at the phase boundaries between gel and liquid-crystalline domains, greatly favors protein insertion and solute release (Pownall et al.. 1979: Kamellis et al.. 1980: Klausner et al.. 1985: Epand et al.. 1989). Thirdly. liposomal systems containing high molar ratios of negatively charged phospholipids. such as PG or PI. show greater serum-induced leakage compared to equivalent neutral systems. This effect, which can be abolished by increasing the medium ionic strength. suggests that charge repulsion between adjacent phospholipids may reduce packing densities thereby  facilitating protein penetration (Cominsky & Heath. 1990). Finally. Allen & Cleland (1980) demonstrated that hypoosmotically stressed SUVs leak a greater amount of solute in the presence of plasma than isoosmotic vesicles. The role of other lipoproteins in liposomal destabilization has also been suggested. Comiskey & Heath (1990) demonstrated that both LDL arid HDL can enhance release of liposomal solute. Furthermore, compared to HDL. the effect of LDL is much greater with negatively charged liposomal systems than neutral vesicles. This suggests that electrostatic interactions involved in initial lipoprotein binding may be more important for LDL than HDL.  1.12.3.2 Complement proteins  Activation of the complement system results in the formation of membrane attack complexes which leads to cell lysis or liposomal membrane damage. The complement system can be activated either via the classical pathway (initiated by binding of Cl to immune complexes) or alternative pathway (initiated by C3 binding directly to cell surfaces in the absence of antibody). However, both pathways converge with the activation (cleavage) of C5 to C5b to form the membrane attack complex through a cascade of reactions. The complement proteins involved in the membrane attack complex are assembled in the following order:  C5b -C6 - C5b-6 -C7 - C5b-7 -C8 - C5b-8 -C9 - C5b-9n  where n denotes one or more C9 proteins. The pivotal step in terms of membrane damage is the membrane mediated association of C8. and subsequently C9. to the C5b-7 protein complex. Release of liposomal solute markers from PS containing vesicles initially occurs with the addition of C8 in the presence of C5b-7 and is further  enhanced by the addition of C9 (Malinski & Nelsestuen. 1989). This is reflected in the membrane penetration depth of the complement complex. Studies with spin label probes located near the membrane surface and in the hydrocarbon region indicates that C5b-7 interacts strongly with the bilayer surface but it is only after interaction with C8 and C9 does the complex penetrate deeper into the hydrophobic domain (Esser et al.. 1979). Two general mechanisms have been proposed to explain the increased permeability effect of the membrane attack complex. The "leaky patch" model suggests that the membrane attack complex produce local membrane lipid disorder (Esser et al., 1979) while the "transmembrane channel" theory proposes that porelike structures are formed (Tschopp. 1984: Malinski & Nelsestuen. 1989). Present consensus favor the latter mechanism (Alving & Richards. 1983: Malinski & Nelsestuen. 1989). The ability of complement proteins to cause membrane damage is strongly dependent on the initial membrane binding step. Most notably, the presence of either negative charge or positive charge on the membrane surface can activate complement via the classical or alternative pathways respectively, but uncharged PC or PC:Chol liposomes show little or no complement activity (Chonn et al.. 1992: Malinski & Nelsestuen, 1989). Generally. a decrease in membrane order by increasing the phospholipid unsaturation or decreasing the acyl chain length will enhance complement-mediated lysis (Shin et al., 1978). However, the situation can be more complex. For example, the presence of cholesterol can enhance or reduce lysis depending on the pathway of complement activation (Alving & Richards, 1983). In addition, there appears to be an optimal membrane order for complement activation via the alternative pathway (Cunningham et al.. 1979). This complexity likely reflects the strong dependence on the initial binding step as well as the multiple components involved in the assembly of the membrane attack complex.  1.13 LIPOSOMES AS DRUG DELIVERY SYSTEMS  A rapidly developing area of liposome research involves the use of liposomes as drug carriers (for a review see Cullis et al.. 1989: Ostro & Cullis, 1989). Initial goals were aimed at specifically targeting liposomes to disease sites. employing antibodies or other targeting agents linked to the vesicle exterior. Although this objective has not yet been achieved, the therapeutic potential of many drugs can be improved by administration in an encapsulated form. These improvements arise both from extending the release time of the drug. and thereby reducing the drug's toxicity. and alterations in the drug's biodistribution. The latter effect is primarily due to the removal of the drug-containing liposome from the circulation by the organs of the reticuloendothelial system (Mayer et al.. 1990: Rahman et al., 1980). This typically results in liposomal accumulation in the liver and spleen (Gregoriadis. 1988). In addition, there also appears to be preferential accumulation in tumours and at the sites of inflammation (Ogihara et al.. 1986: Morgan et al.. 1985). A partial list of liposomal drugs presently undergoing human clinical trials include: amphotericin B. an anti-fungal agent (Lopez-Berestein. 1988): muramyl tripeptide PE, an immunomodulator (Fidler et al.. 1986. 1980): and doxorubicin, an anticancer agent (Rahman et al.. 1986). Depending on their properties. these drugs can be covalently attached to the membrane lipids (e.g. muramyl tripeptide PE). entrapped in the membrane's hydrophobic domain (e.g. amphotericin B. doxorubicin) or encapsulated in the vesicles aqueous interior (doxorubicin). In the latter two situations. the presence of the drug in the membrane's hydrophobic domain or the liposome's aqueous compartment depends on the drug's partition coefficient (Madden et al.. 1990: Harrigan et al.. 1993). Amphotericin B is a very hydrophobic molecule mid is thus exclusively situated in the membrane. However other drugs. such as doxorubicin for example, distributes both in the liposome's aqueous interior as well as  in the membrane. This has two consequences relevant to the studies presented in this thesis. Firstly. like most weakly acidic or basic drugs. doxorubicin can be efficiently loaded into the vesicle employing a transmembrane pH gradient across the bilayer (Mayer et al., 1992). For example. weakly basic drugs can accumulate in vesicles with an acidic interior. As a result of the high internal buffering capacity needed to load and retain these drugs and the aqueous concentration of the drugs themselves, the liposomes are typically hypoosmotically stressed when injected into the blood stream. Secondly. the higher interior doxorubicin concentration in comparison to the external medium should result in an net expansion of the membrane's inner monolayer due to the hydrophobic partitioning of the drug.  1.14 THESIS OUTLINE This thesis examines the effects of mechanical stress due to osmotic forces and transbilayer area asymmetry on large unilamellar vesicles made by extrusion (LUVETs). These properties have been previously studied with giant vesicles by phase contrast microscopy and have provided information concerning membrane mechanical properties and vesicle morphology. However, these systems are harder to manipulate and analyse biochemically compared to a homogeneous vesicle population. In Chapter 2, the non-spherical morphology of LUVETs is visualized by cryoelectron microscopy and is shown to be the result of the extrusion procedure. The kinetics, magnitude and pore formation during vesicle lysis is subsequently characterized. The influence of plasma on the osmotic stability of LUVETs is the subject of Chapter 3. This topic is of practical interest due to the use of liposomes as in vivo drug carriers. Finally. in Chapter 4. the influence of transbilayer area asymmetry on vesicle morphology is examined.  CHAPTER 2 OSMOTIC PROPERTIES OF LUVETs  2.1 INTRODUCTION  Biological membranes display much higher perineabilities to water in comparison to other molecules and as a result are sensitive to osmotic gradients. Such gradients can have profound influences on cell structure and function and much recent work, therefore, has examined the mechanisms by which cells detect, and respond to. osmotic forces. Mechanosensitive ion channels have been reported in a variety of cell types and it has been suggested that these may control osmotic gradients by coupling bilayer permeability to membrane tension (Martinac et al., 1987; Miyamoto et al.. 1988: Yang and Sachs. 1989: Morris. 1990). Osmotic gradients have also been suggested to play a role in membrane fusion-dependent processes such as exocytosis (Cohen et al., 1980; Hampton and Holz. 1983: Pollard et al., 1984). Liposomes, which exhibit similar permeability properties to biological membranes (Bangham et al., 1967), represent a convenient model system with which to study osmotic stress and osmotically-induced lysis. In this context a number of studies have examined the osmotic sensitivity of large multilamellar vesicles (Bangham et al., 1967: Allianaty and Livne, 1974: Blok et al.. 1976). As model systems for osmotic studies. MLVs are superior to small unilamellar vesicles (SUVs) prepared by sonication, which are osmotically insensitive (Johnson and Buttress. 1973), however, they present two major difficulties. First, their onion-like structure with multiple internal aqueous compartments necessarily complicates studies involving lysis and solute release. Partial release of entrapped solute upon exposure of MLVs to an osmotic gradient. for example. might result from lysis of only the outer most lamellae.  The second disadvantage to the use of MLVs results from the non-equilibrium distribution of solutes within the various internal aqueous compartments (Gruner et al., 1985: Mayer et al.. 1985). Consequently when MLVs are placed in hypo- or hyperosmotic media different osmotic gradients will be experienced by different internal lamellae. These limitations can be overcome by the use of LUVs and we have therefore examined the osmotic properties of such vesicles prepared by the extrusion procedure. While earlier reports described the use of photon correlation spectroscopy to measure osmotic swelling and lysis in LUVs prepared by the pH-adjustment method (Li et al.. 1986: Li and Haines. 1986: Haines et al.. 1987). these data have recently been retracted (Rutkowski et al.. 1991).  2.2 MATERIALS AND METHODS  2.2.1 Lipids and chemicals Egg phosphatidylcholine (EPC). N-(lissamine rhodamine B-sulfony1)dioleoylphosphatidylethanolamine (RHO-PE). and N-(7-nitrobenz-2-oxa-1.3-diazol-4yl)dioleoylphosphatidylethanolamine (NBD-PE) were obtained from Avanti Polar Lipids (Alabaster, AL). Cholesterol (standard for chromatography) was from Sigma Chemical Co. (St.Louis, MO). [ '4C]-citrate, ['4C]-glucose and PM-glucose were purchased from NEN while [3I-11-dextran (avg. mol. wt. 70.000) and [HQdipahnitoylphosphatidylcholine ([14C1-DPPC) were obtained from Amersham. 5(6)carboxyfluorescein was purchased from Eastman Kodak and purified according to Weinstein et al. (1984). Gold 700 mesh bare EM grids were from Marivac Ltd. (Halifax, Canada).  2.2.2 Preparation of lipid vesicles Lipid mixtures were prepared by colyophilization from benzene:methanol (95:5 v/v) under high vacuum (<60 millitorr) for a minimum of 4 hours, protected from light. Unless otherwise stated, mixtures of phospholipid and cholesterol were prepared in a 55:45 molar ratio. FATMLVs were prepared by hydration of the dry lipid in an appropriate solution and the dispersion was then freeze-thawed five times employing liquid nitrogen to promote equilibrium transmembrane solute distributions (Mayer et al.. 1985). LUVETs were then prepared by extruding the FATMLVs ten times through two stacked polycarbonate filters (Nuclepore) using an Extruder (Lipex Biomembranes, Inc., Vancouver, Canada) as previously described (Hope et a).. 1985). Unless otherwise stated 100 nm pore size filters were used.  2.2.3 Measurements of vesicle lysis Osmotically-induced vesicle lysis was determined by following the release of either carboxyfluorescein or a radiolabeled solute initially entrapped within the lipid vesicles. These marker molecules were encapsulated by including them in the buffer solution used to prepare the vesicles. Unencapsulated carboxyfluorescein or radiolabel was then removed by gel exclusion chromatography. Details of individual experiments are given below.  2.2.4 Time course of carboxyfluorescein release from vesicles EPC:Chol vesicles (50 mg ml-1, 120 um average diameter) were prepared in 100 mM carboxyfluorescein, 600 mM NaCl. 60 inM HEPES. pH 7.4 and an aliquot (1 ml) passed down a Sephadex G-50 (medium) column (1.5 x 10 cm) pre-equilibrated with an isoosmotic solution. 750 niM NaCl. 75 itiM HEPES. pH 7.4. The peak lipid fraction (approximately 20 inM lipid) was collected. The high intravesicular concentration of carboxyfluorescein results in essentially complete fluorescence  quenching. Only following release of the fluorophore into the external solution can a fluorescent signal be detected (for review see Weinstein et al.. 1984). To determine the time course and extent of osmotically-induced lysis. vesicles were diluted into buffered saline solutions of various osmolarities at 22°C and the appearance of fluorescence monitored using a SLM Aminco SPF 500C spectrofluorimeter at 492 nm excitation (bandwidth 1 urn) and 520 nm emission (bandwidth 10 mil). Control vesicles were diluted into isoosmotic saline solutions. Complete release of the fluorophore was achieved by adding the detergent octylglucopyranoside (final concentration 25 mM) to the sample. Similar protocols were used in other studies in which carboxyfluorescein release was followed.  2.2.5 Influence of solute molecular weight on osmotically-induced leakage  Vesicles of EPC:Chol (50 mg m1-1) were prepared in either 300 mM sodium citrate, 30 mM HEPES, pH 7.4 or 800 inM sodium citrate. 80 mM HEPES, pH 7.4 containing 4 iCi m1-1 [3H1-dextran (average molecular weight 70,000) and 2 itCi m1-1 (14Q-glucose. Untrapped radiolabels were then removed by passing the vesicles down a Sepharose 2B-CL column (1.5 x 12 cm) pre-equilibrated in the same buffer used to prepare the vesicles. A hypoosmotic gradient was established by passing the vesicles down a similar 2B-CL column pre-equilibrated with 10 mM HEPES, pH 7.4. Following incubation at 22°C for 15 min the amount of radiolabel retained was determined following passage of an aliquot (500 pi) down a Sepharose 2B-CL column (1.5 x 10 cm).  2.2.6 Vesicle trapped volume measurements  Changes in trapped volumes of extruded vesicles were characterized using EPC:Chol (labeled with 9 x 10-4 tiCi fi4ci_j DPPC per mmol lipid) LUVETs prepared in 10 inM NaCl. 1 inM glucose (1.5 IACi nil-' PM-glucose). Glucose is slowly membrane  permeable (half-times for equilibration into 100 inn EPC:Chol vesicles are 14 h and 1 h at 30°C and 45°C respectively) and can therefore be used to follow changes in vesicle trapped volume over time. Vesicles were prepared by extrusion at 45°C and then diluted 5 fold into 1.5 [tCi m1-1 [31-1]-glucose spiked solutions of either 1 inM glucose or 10 inM NaCl. 1 mM glucose. Samples were filtered sterilized using 0.22 micron polycarbonate filters (Nuclepore Corp.) and incubated at 45°C . At various times. 500 td aliquots were withdrawn and vesicle trapped volumes determined after removal of the external radiolabel using pre-packed 9 ml Sephadex G-25 columns (Pharmacia) pre-equilibrated with 10 mM sodium citrate. 5 inM HEPES. pH 7.4.  2.2.7 Solute equilibration in FATMLV's To determine the influence of solute equilibration in FATMLVs on extruded vesicle morphology. EPC:Chol was hydrated at 50 mg m1-1 in 1 inM glucose (1.5 [A.Ci m1-1 (3HJ-glucose) and. following five freeze-thaw cycles, incubated at 30°C in sterilized Eppendorf tubes. At various times aliquots were taken and LUVETs were prepared through 100 rim filters at 30°C. These vesicles were then filter sterilized through 0.22 mm Nuclepore filters and incubated under sterile conditions at 30°C for 75 h. Initial and final (after 75 h) trapped volumes of the extruded vesicles were determined as described above.  2.2.8 Cryo-electron microscopy A drop of the liposomal suspension was placed on a bare 700 mesh gold EM grid held by tweezers mounted on a spring-loaded plunger. After removing excess sample by blotting the grid with filter paper. the sample was vitrified by plunging the grid into liquid propane cooled to -187°C. The blotting and vitrification step was rapidly done (<2 sec) to minimize evaporation from the sample. The grid was then transferred to a Gatan 126 cold stage at liquid nitrogen temperatures using a Reichart  Jung Universal Cryo-fixation system and the sample was visualized using a Zeiss EM 1 OC STEM.  2.2.9 Vesicle fusion analysis  A fluorescent assay based on resonance energy transfer between NBD-PE (the energy donor) and RHO-PE (the energy acceptor) was used to probe for vesicle fusion (Uster and Deamer, 1981). 1% NBD-PE or 1% RHO-PE labeled EPC:Chol vesicles containing 400 inM sodium citrate. 40 mM HEPES. pH 7.4 were initially mixed together in equal molar ratios and then osmotically shocked by dilution with 50 mM NaC1, 20 inM HEPES, pH 7.4 buffer. A decrease in fluorescence of NBD-PE (excitation 460 nm. emission 530 run) would be indicative of vesicle fusion. A standard curve was constructed by measuring NBD-PE fluorescence in vesicles containing various ratios of the two fluorophores and total fluorescence determined following addition of Triton X100 (2% final concentration). Fusion was probed under a variety of assay conditions including lipid concentration (0.5 to 10 mM EPC:Chol). lipid composition (EPC and EPC:Chol) and the magnitude of the applied osmotic gradient (870 - 1640 mOsm/kg).  2.2.10 Vesicle size analysis  The size distribution of extruded vesicles was determined by quasi-elastic light scattering (QELS) using a Nicomp 370 submicron particle sizer (Nicomp Instruments, Goleta. CA) operating at 632.8 nin and 5 inW. Aliquots of the vesicle suspension were diluted approximately 1:100 in 150 inM NaCI. 10 inM HEPES. pH 7.4. which had previously been filtered through a 0.22 micron polycarbonate filter. Measurements were made at 22°C.  2.2.11 Osmolarity measurements  Solution osmolarities were determined from freezing point depression using an Advanced Digimatic 3C2 Osmometer. Standards (100. 290 and 900 mOsin/kg) were analyzed prior to samples which were measured at least in duplicate.  2.2.12 Analytical procedures  In some experiments phospholipid concentrations were determined by phosphate assay (Fiske and Subbarow, 1925). Radiolabels were quantified by liquid scintillation counting using either a Packard Tricarb 2000 CA or Beckman LS3801 instrument.  2.3 RESULTS  While the assumption is often made that exposure of vesicles to a hypoosmotic solution will result in the creation of an osmotic pressure. this assumption is not necessarily valid. In the case, for example. of vesicles that are not initially spherical. the influx of water resulting from an applied osmotic differential will first be accommodated by a change in shape as the vesicles maximize their volume to surface area ratio. Only if the applied differential exceeds the volume increase resulting from vesicle shape changes will the lipid bilayer experience an osmotic pressure. Before examining the osmotic properties of LUVETs. therefore. it is appropriate to first characterize their morphology.  2.3.1 Morphological and associated volume changes for LUVETs.  Earlier studies employing the lipid dioleoylphosphatidylethanolamine (DOPE) at pH 9.9 indicated that vesicles prepared by extrusion are non-spherical (Seigel et al., 1989: Talinon et al..1990). We therefore examined EPC:Chol LUVETs prepared in  150 mM NaCl. 20 mM HEPES. pH 7.4 using the technique of cryo-electron microscopy. In agreement with these earlier studies we observe that LUVETs maintained under isoosmotic conditions exhibit a variety of morphologies (Figure 8A) ranging from tubular to spherical to invaginated vesicles. Exposure of extruded vesicles to a hypoosmotic solution should result in an influx of water causing the vesicles to adopt a more spherical shape to maximize its volume to surface area ratio. This behavior is illustrated in Figure 8B. While vesicles prepared in saline solutions can be non-spherical, similar systems made in distilled water and viewed by cryoelectron microscopy (Figure 8C) are spheres. This would indicate that in the absence of an impermeant solute membrane bending forces are sufficient to establish a spherical shape. These forces are relatively small. however (Evans and Hochmuth, 1978: Section 1.8.2). and for vesicles containing an impermeant solute are readily opposed by osmotic forces resulting from any volume change. It should be noted that earlier studies, which employed the technique of freeze-fracture electron microscopy to visualize extruded vesicles, did not report the presence of non-spherical systems (Hope et al. 1985; Mayer et al. 1986). This discrepancy may result from the use of high concentrations (approximately 3.4 M) of the membrane-permeable solute, glycerol, as a cryoprotectant in samples prepared for freeze-fracture.  t^4 lOr 4f-^ P-4(  **"'  '1'4> I. ‘V  -  A *4  4,•440:  k  - PIFTLFir44,  4*  Figure 8 Cryo-electron micrographs of extruded vesicles. Vesicles of EPC:Chol were made in 150 mM NaCI, 20 mM HEPES, pH 7.4 and exposed to an (A) isoosmotic or (B) hypoosmotic (75 mM NaC1, 20 mM HEPES, pH 7.4) solution. The lipid concentration is 20 mg m1-1. Vesicles of EPC:Chol prepared in distilled water are shown in (C) at a lipid concentration of 60 mg m1-1. The bar represents 100 inn.  2.3.2 Quantitation of vesicle volume increase Before characterizing the influence of osmotic gradients on extruded LUVETs it is important to first quantitate the change in intravesicular volume associated with "rounding up" of such systems. The influx of water associated with this process will dilute the intravesicular solution and to determine the osmotic gradient across the vesicle membrane it is necessary to correct for this effect. We therefore prepared EPC:Chol vesicles in 10 mM NaCland applied a small osmotic pressure by adjusting the external salt concentration to 2 rnM. Both solutions also contained 1 inM glucose (1.5 viCi m1-1 PHI-glucose). As glucose is slowly membrane permeable (t112 = 1 h. at 45°C), it can be used to measure any change in vesicle trapped volume following extrusion without impeding the process. The trapped volumes of 200 nm vesicles incubated for up to 18 hours under isoosmotic conditions or exposed to a small osmotic gradient are shown in Figure 9A. In the absence of a gradient (10 mM NaCl inside and outside) little change in trapped volume is observed with an average value of 1.3 1.11/timole lipid obtained over the time course followed. In contrast, a fairly rapid increase in trapped volume is shown by vesicles exposed to a small hypoosmotic gradient with a value of 2.0111/[tmole lipid obtained after 18 hours. It should be noted that this volume increase cannot be the result of osmotically-induced swelling of initially spherical vesicles. Based on an elastic area expansivity modulus of 1000 dyn cm-1 (Evans and Needham. 1987). an osmotic differential of 16 mOsm/kg would generate a volume increase of less than 0.2%. Similar changes in trapped volume upon "rounding up" were observed for vesicles prepared through 50, 100, and 400 nm pore size filters. To confirm that the measured increases in trapped volume reflected a change in vesicle shape. samples were examined by cryo-electron microscopy. As predicted. micrographs of vesicles examined shortly after extrusion show a variety of non-spherical vesicle morphologies (Figure 10A). In contrast. the same vesicles  exposed to a small osmotic differential (10 inM NaC1 inside: 2 inM NaC1 outside). while of similar size, are clearly spherical (Figure 10B).  2.3.3 Source of the non spherical morphology of LUVETs -  Two hypothesis can be advanced to account for the observation that extruded vesicles are non-spherical. First, it may be suggested that full solute equilibration is not achieved during preparation of FATMLVs and the extruded vesicles simply reflect this condition. Alternatively, it is possible that the extrusion process generates nonspherical systems and osmotic forces then prevent "rounding up". To determine which of these two explanations applies. EPC:Chol FATMLVs were prepared at 30°C in 1 triM glucose (1.5 tiCi^PHI-glucose) and incubated at 30°C for up to 75 hours. At various times aliquots were taken. 100 nm extruded vesicles prepared and vesicle trapped volumes determined immediately following extrusion or after incubation at 30°C for 75 'hours. Control experiments and cryo-electron microscopy verified that initially non-spherical extruded vesicles are able to "round up" in the presence of 1 mM glucose. The experimental rationale is that if the initial low trapped volume of extruded vesicles reflects a non-equilibrium solute distribution in the FATMLVs from which they are prepared then incubation of FATMLVs in the presence of a permeant solute (glucose) will alleviate this non-equilibrium situation. As shown in Figure 9B however, the discrepancy between initial and final trapped volumes of extruded vesicles is unaffected by pre-incubation of the FATMLVs. This indicates that it is the extrusion process itself which is responsible for the observed vesicle morphology. Additional support for this contention is provided by experiments in which LUVETs prepared in 1 mM glucose (containing 1.5 tiCi m1-I PHI-glucose). using 100 nm pore size filters, were allowed to "round up" and were then re-extruded through filters of the same pore size. As shown in Table 3. the trapped volume of vesicles allowed to "round up" following extrusion increases from 1.45 to 2.08 ul/pinole lipid in agreement with  previous experiments (c.f. Figure 9A). If these spherical vesicles are reextruded, however. their trapped volume is significantly reduced. to 1.74 vlitanole lipid. That this reduction is a consequence of a change in vesicle shape rather than a decrease in average diameter is illustrated by the fact that upon subsequent "rounding up" the trapped volume of these vesicles is the same as prior to reextrusion (Table 3). These experiments provide compelling evidence that passage through the filter pores is largely responsible for the non-spherical vesicle morphology observed.  2.4 CL  2.2 2.0  co  1.8 1.6  -0 co  Ct. CL CLS  1.4 1  .  2  1.0 4  ^ ^ ^ ^ 8 20 12 16 Time (hours)  2.0 1.8  •  1.6 1.4 1.2 1.0 0.8 0.6 0^10^20^30^40^50^60^70  ^  80  Incubation Time of FATMLVs (hours)  Figure 9 Trapped volume characterization of extruded vesicles. A. Vesicles of  EPC:Chol (50 ing m1-1, 190 nm diameter) were made in 10 mM NaCI, 1 inM glucose (1.5 [tCi m1-1 (3F11-glucose) and diluted 5 fold with either 1 inM glucose, • : or 10 mM NaC1, 1 InM glucose, 0 , and incubated at 45°C. Trapped volumes were measured at various times as described under Methods. B. An EPC:Cliol FATMLV suspension made in 1 mM glucose (1.5 faCi m1-1 E31-11-glucose) was incubated at 30°C. At various times, aliquots were extruded through 100 mit filters and vesicle trapped volumes measured immediately, 0 ; or after incubation at 30°C for 75 li. • .  Figure 10 Morphology of extruded vesicles revealed by cryo-electron microscopy. Vesicles of EPC:Chol (mean diameter 80 urn) prepared in 10 inM NaCI, 1 mM glucose were diluted into an isoosmotic medium (A) or a hypoosmotic solution (2 mM NaCI, 1 mM glucose), (B). The lipid concentration is 20 mg m1-1 and the bar represents 100 urn.  Table 3 Influence of Reextrusion on Trapped Volume of Spherical Vesicles  Trapped Volumes (t1/ [unole lipid)! Initial^ Final 2  First extrusion Reextrusion  1.45 + 0.04 1.74 + 0.04  2.08 + 0.05 2.04 + 0.05  1 Values given represent the mean and standard deviations of three samples. The POPC vesicles were allowed to fully "round up" by incubation at 30°C for 24 hours.  2  2.3.4 Osmotic lysis of large unilamellar vesicles  Having examined the morphological changes resulting from exposure of LUVETs to hypoosmotic solutions we next sought to characterize osmotic lysis of such systems. In experiments in which the release of a water soluble marker from osmotically-  stressed vesicles is measured, however, the first criterion to be established is whether the release represents a lytic event or is due to permeation. In Figure 11 is shown the time course of carboxyfluorescein release from EPC:Chol vesicles exposed to an osmotic gradient greatly in excess of that which could be accommodated by "rounding up" of the vesicles (see figure legend for details). It can be seen that most of the resulting loss of fluorophore occurs within about 30 seconds with little further release up to 10 minutes. Control vesicles diluted into an isoosmotic solution, as anticipated, show no release over this time period. It should be noted that the time course shown in Figure 11 cannot be assumed to represent the rate of lysis per se. Dequenching of carboxyfluorescein following vesicle rupture will require probe diffusion which may represent the rate limiting process. In similar experiments in which release of ("CIcitrate from osmotically-stressed EPC:Chol vesicles was followed for up to 60 min,  again essentially all of the loss occurred kvithin the earliest measurable time point (results not shown). These results indicate that the loss of vesicle contents, which occurs when they are subjected to a hypoosmotic gradient. is the result of lysis and is not due to permeation which would be expected to occur at a fairly constant rate.  50 a)^ cj^40 a)^o'  r^,4•••••  CC  Jo.* s-  ^.o-^0- -  o-  30  a) a)^20  x^10 o  ID  ,.. cc;  000 , ^o^ .^ .^ 0^  o  0  I  0^100^200^300^400^500  600  Time (seconds)  Figure 11 Time course of carboxyfluorescein release from osmotically stressed vesicles.  EPC:Chol vesicles containing 100 niM carboxyfluorescein. 600 mM NaC1, 60 mM HEPES. pH 7.4 were diluted (1:60) into 10 inM HEPES. pH 7.4 ( • ) or isoosmotic, 750 mM NaCl. 75 inM HEPES, pH 7.4 ( 0 ) at 23°C  Given the observation that, under the conditions described in Figure 11, an applied osmotic gradient of 1400 mOsin/kg results in the release of approximately 45 % of initially entrapped carboxyfluorescein. two generalized interpretations can be advanced. Either 45 % of the vesicles lyse releasing all of their entrapped fluorophore. or alternatively, all the vesicles lyse but lose only 45 % of their contents. To distinguish between these two hypotheses the experiment described below was performed. When carboxyfluorescein is entrapped within EPC:Chol vesicles. fluorescence quenching is seen at fluorophore concentrations above approximately 0.1 mM. This behavior is illustrated in Figure 12 which indicates a relatively steep concentrationdependent quenching between 0.2-10 inM carboxyfluorescein. We can take advantage of fluorophore quenching to distinguish between the two lysis models described above using a procedure similar to that described by Weinstein et al. (1981). The experimental rationale is that if lysis results in complete loss of intravesicular solute then any carboxyfluorescein remaining entrapped must be in unlysed vesicles at its original concentration and hence its initial level of quenching. Conversely, if all vesicles lyse, releasing a portion of their contents only. then the level of fluorophore quenching will be reduced in direct proportion to the percentage of entrapped solute released. Vesicles (210 nm, average diameter) were therefore prepared in 7.5 mM carboxyfluorescein, 2.5 M NaCl. 20 mM HEPES. pH 7.4 containing 2 RCi m1-1  114C  1_  citrate. Aliquots of the vesicle preparation were then osmotically shocked by passage down Sephadex G-50 columns preequilibrated with either 1.25 M NaCl. 20 mM HEPES, pH 7.4 or 20 mM HEPES. pH 7.4 only. Vesicles elute in the column void volume and are efficiently separated from any carboxyfluorescein or I I 4C1-citrate released. Following isolation, the vesicles were analyzed to determine the amount of carboxyfluorescein and ("CI-citrate retained and the level of fluorescent quenching of  entrapped carboxyfluorescein. As shown in Table 4. similar losses of both 114C1-citrate and carboxylluorescein are observed upon osmotic lysis. the extent of solute release being dependent upon the size of the applied osmotic differential. Interestingly, the level of carboxyfluorescein quenching for osmotically-shocked vesicles is dependent upon the extent of solute release (Table 4). Using the quench curve shown in Figure 12 we can calculate the average carboxyfluorescein concentration remaining within vesicles exposed to an osmotic gradient (Table 4). Correcting for the volume increase due to "rounding up" (53 %. c.f. Figure 9A) the average fluorophore concentration remaining entrapped is in good agreement with that predicted by a model in which only partial solute release occurs during lysis (Table 4) with the extent of release being dependent on the size of the osmotic gradient. Under lysis conditions producing solute losses of greater than 60 %. carboxyfluorescein quenching is actually lower than expected. This likely reflects a shift in the quench curve at lower intravesicular salt concentrations (see Figure 19).  100  0 0^2^4^6^8^10  Intravesioular Carboxyfluorescein (mM)  Figure 12 Quench curve of carboxyfluorescein entrapped in EPC:Chol vesicles. Carboxyfluorescein fluorescence was determined in EPC:Chol vesicles containing 20 mM HEPES, pH 7.4, 500 mM citrate, and 0.2 to 10 mM carboxyfluorescein. Total carboxyfluorescein fluorescence was obtained by solublizing the vesicles in 25 mM octylglucopyranoside.  Table 4 Evaluation of Theoretical Models of Solute Release During Lysis.  Applied osmotic^% Retained^% Fluorescence ^Intravesicular (CF) (mM) differential^Naci_ J citrate CF^Quenching^Theoretical^Measured3 (mOsm/kg) Totall Partial2 0  100  100  42  7.5  7.5  7.5  2400  82  81  67  5.0  4.0  3.9  4800  32  38  87  5.0  1.9  1.3  1 Theoretical intravesicular carboxvfluorescein concentration based on a model in which total solute release occurs from vesicles during lysis and hence any retained carboxyfluorescein is in unlysed vesicles. The calculation is based on an initial carboxyfluorescein concentration of 7.5 InM and assumes a 53 % maximum increase in vesicular volume due to "rounding up". 2 Theoretical intravesicular carboxyfluorescein concentration based on a model in which lysis results in partial solute release from all vesicles in the population. Again the calculation is based on an initial carboxyfluorescein concentration of 7.5 mM and assumes a maximum 53 % increase in vesicular volume due to "rounding up". 3  Determined from Figure 12 based on the measured fluorescence quenching.  We next examined in greater detail how the magnitude of the osmotic differential between the vesicle interior and the external solution affected the loss of contents during lysis. Vesicles (100 nm average diameter) composed of EPC:Chol were prepared in a solution of relatively high osmolarity (2.35 M NaC1, 100 mM carboxyfluorescein. 20 mM HEPES. pH 7.4) and then diluted into solutions of lower osmolarity. In Figure 13A the percentage loss of carboxyfluorescein that occurs is compared to the applied osmotic differential. It can be seen that little carboxyfluorescein release is observed until an osmotic differential of about 2000 mOsm/kg is applied. At differentials in excess of 2000 mOsin/kg progressively more of the entrapped fluorophore is lost. Clearly, however, volume changes associated with vesicles "rounding up" will result in the applied osmotic differential being considerably greater than the actual differential experienced by the vesicle membrane. In Figure 13B. therefore. we have corrected for both vesicle volume changes and, where applicable, for solute loss during lysis. in order to obtain the theoretical residual osmotic differential following lysis. The predicted relationship between applied and residual osmotic differentials (in the absence of lysis) for LUVETs is shown by the dashed line (Figure 13B). This indicates that as the applied gradient is increased to 1780 mOsm/kg the vesicles are able to accommodate the resulting influx of water by adopting a more spherical shape. consequently the residual osmotic gradient remains zero. At a differential of 1780 mOsm/kg. however, the vesicles are spherical and any additional increase in the applied gradient produces a corresponding increase in the residual differential. If we now compare the experimental data to this theoretical prediction a number of observations can be made. At applied differentials between 500 - 1500 mOsin/kg the experimental data points lie below the zero line. This results from a slight release of carboxyfluorescein (c.f. Figure 13A) and is to be expected given that all vesicles in the population will not  accommodate precisely the same volume change upon "rounding up". The fact that this deviation from predicted behavior is relatively small would suggest that LUVETs do not differ significantly with respect to their volume to surface area ratio. At higher applied osmotic differentials, the residual osmotic gradient approaches a limiting value of about 650 mOsm/kg. We would suggest that this value represents the average maximum osmotic gradient the vesicles can withstand. When LUVETs are exposed to differentials in excess of this value. lysis (or more likely a series of lytic events) will occur and sufficient intravesicular solute will be lost such that upon membrane resealing the vesicles can withstand the remaining osmotic gradient.  1 00  A  a)  cr,  ca  a) TD  fr  •  80 60  •  40  0 ZEs o  20  •  •  •  •  •  0 • ^ 5000 0^1000^2000^3000^4000 Applied Osmotic Differential (mOsm/kg)  1100 900 700 500 300  100 -100 -300 0  1000^2000^3000^4000  ^  5000  Applied Osmotic Differential (mOsm/kg)  Figure 13 Influence of the osmotic differential on vesicle lysis. A. EPC:Chol vesicles  containing 100 mM carboxyfluorescein. 2.35 M NaCl. 20 mM HEPES, pH 7.4 (4550 mOsm/kg) were diluted into hypoosmotic NaC1 buffers at 23°C and carboxyfluorescein release measured after 3 minutes. B. The residual osmotic differential was calculated for the EPC:Chol vesicles shown in (A) after taking into account a 53 % increase in vesicular volume. The dotted line represents the expected osmotic differential in the absense of lysis. Results shown have also been corrected for the fluorescence exhibited by the isoosmotic controls (3 % of total fluorescence).  - 68 -  2.3.5 Size of the membrane defect formed during osmotic lysis Given that osmotic lysis results in only a portion of the intravesicular solute being released such that the internal aqueous solution remains hyperosmotic with respect to the external medium, bilayer resealing must be fairly rapid and the defects generated by lysis may be relatively small. In an attempt to probe the size of the bilayer defects created during osmotic lysis aqueous markers of differing molecular weights were entrapped within EPC:Chol vesicles (100 urn average diameter) which were then subjected to an osmotic gradient. It can be seen from Table 5 that the percentage release of both glucose (mol. wt 180) and dextran (avg. mol. wt 70.000) is similar indicating that the membrane defects occurring during lysis are likely greater than about 12 nm diameter, the approximate average diameter of dextran.  Table 5 Influence of Solute Molecular Weight on Release During Vesicle Lysis  % Release 1 Osmotic differential ^ (mOsm/kg) ^ 800 ^ 2200  [14Q-glucose^131-1J-dextran 12^12 67^66  1 EPC:Chol vesicles were subjected to hypoosmotic lysis as described under Methods.  2.3.6 Influence of vesicle size on osmotic lysis When FATMLVs are extruded through polycarbonate filters, the mean diameter of the resulting vesicles is determined by the filter pore size. Before examining the influence of vesicle size on osmotic lysis. however, it was necessary to quantitate the  volume changes associated with "rounding up" for each system. The experiment described in Figure 9 was repeated therefore to measure trapped volume increases for vesicle preparations of mean diameters 90. 100. 190 and 340 nm (as measured by QELS). Similar volume increases (50. 53. 54 and 50%) were obtained for all four samples. Using conditions similar to those described in Figure 13, carboxyfluorescein release from these four vesicle systems was determined as a function of the applied osmotic differential. In Figure 14A are plotted the theoretical residual osmotic differentials for these systems after correcting for vesicle volume changes and any solute release. Clearly, there is a considerable size dependency with the smaller systems tolerating much greater residual osmotic differentials than the larger vesicles. This observation is consistent with previous studies (Sun et al., 1986) and is predicted by Laplace's Law which relates the pressure difference across a closed membrane to membrane tension. As mentioned in Section 1.10.2. Laplace's Law for a spherical vesicle can be written as: T  = AP r/2  where T is the membrane tension. AP is the pressure difference between the inside and outside. and r is the radius. Lysis will occur when the maximal membrane tension is exceeded and so for a given value of T. vesicle size and the osmotic pressure required to produce lysis should be inversely related. To confirm that this relationship holds true for the vesicle systems under study the data presented in Figure 14A are replotted in 14B. A linear relation between the maximum tolerated osmotic difference and the inverse of vesicle size is obtained, with an intercept close to zero. in agreement with theory. Further, from the slope of the line a value of 40 dyn cm-1 for the membrane tension at lysis can be calculated. in reasonable agreement with values reported by Needham and Nunn (1990) for giant bilayer vesicles composed of stearoyloleoylphosphatidylcholine and cholesterol.  800 600 400 200 0 -200 0  1800  1200  600  2400  Applied Osmotic Differential (mOsm/kg)  1000 800 600 400 200 I  0^5^10^15^20^25 1/radius (104 cm  30  -1)  Figure 14 Influence of vesicle size on osmotic lysis. A. EPC:Chol vesicles containing 100 mM carboxyfluorescein. 1.15 M NaC1, 20 mM HEPES, pH 7.4 were made by extrusion through 50 nm, 100 nm. 200 nm. or 400 nm pore sized filters and had mean diameters of 90 nm, 100 nm, 190 nm and 340 nm respectively as measured by QELS. Following dilution into various hypoosinotic NaC1 buffers the residual osmotic differential was calculated. Vesicular volume increases of 50. 53. 54. and 50 % respectively were also taken into account in the calculations. B. Relationship between vesicle radius and osmotic tolerance. From the data shown in (A) the maximum osmotic differential tolerated by different sized systems is plotted against the reciprocal of vesicle radius. Data points shown represent the mean +_one standard deviation. -71 -  While vesicles prepared by extrusion through 100 nm or smaller pore size filters are almost exclusively unilamellar (Hope et al.. 1985) systems prepared through larger filters contain some multilamellar structures. Using the 3 1P-NMR technique with manganese as a broadening reagent the percentage of unilamellar vesicles in systems prepared through 200 and 400 mil pore size filters has been shown to be 90 % and 66 % respectively (Mayer et al.. 1986). Multilamellar systems are not present. therefore, to an extent where they would compromise the data analysis presented above. An additional consideration. however, concerns the suggestion that osmotic gradients may play a role in membrane fusion. If fusion accompanied osmotic lysis of LUVETs then values for the mean vesicle diameter would need to be corrected accordingly. However, using a fluorescence assay based on resonance energy transfer between two lipid derivatives. NBD-PE and RHO-PE. no indication of fusion was obtained for vesicles exposed to an osmotic gradient or for control vesicles maintained under isoosmotic conditions.  2.4 DISCUSSION  Large unilamellar vesicles are widely used as model systems in membrane research. While they can be prepared by a variety of techniques, the extrusion procedure is perhaps the most commonly employed protocol. with advantages of generality, reproducibility and convenience. The results presented here are frequently counterintuitive and provide important insight into the morphology and osmotic properties of LUVETs prepared by extrusion. In addition, the characterization of osmotic lysis in this system has broad implications. A number of earlier studies have examined the morphology of giant unilamellar vesicles (diameter 20 - 30 mm) which can be conveniently visualized using phase contrast microscopy. In addition to characterizing the mechanical properties of such  systems (Section 1.8) these studies demonstrated that initially spherical vesicles undergo characteristic shape changes induced by changes in osmolarity. temperature. or transbilayer area asymmetry (Section 1.11 and references within). In these studies. the initial, unperturbed vesicle systems were generally spherical and it is not surprising, therefore. that LUVETs were also assumed to be spheres. Interestingly, it was noted by Hope eta!. (1985) that the trapped volume of LUVETs prepared through 100 nm pore size filters was appreciably smaller than predicted by theory. We show here that this discrepancy results from the non-spherical morphology of LUVETs and that if such systems are allowed to fully "round up" then trapped volumes in good agreement with theory are obtained. Further we have shown that it is passage through the filter pores that is responsible for the shape. and hence trapped volume. of extruded vesicles. The morphology of LUVETs has important implications concerning their osmotic properties. The volume changes associated with vesicles "rounding up" has the effect of diluting the intravesicular solute and hence. unless corrected for, leads to an overestimation of the osmotic gradient experienced by the vesicle membrane. In the present work we have quantitated these changes in intravesicular volume allowing an accurate determination of actual osmotic differentials. When we examine the effect of osmotic gradients of varying magnitude on LUVETs. we observe that vesicles are able to tolerate the osmotic pressure resulting from relatively small differentials. At a characteristic pressure. however. lysis occurs with the amount of intravesicular solute released depending on the magnitude of the osmotic differential. Of considerable interest is the observation that osmotic lysis is not an "all-or-nothing" event, that vesicles release only a portion of their contents during lysis and can in fact reseal while their intravesicular medium remains hyperosmotic with respect to the external solution.  Our observation that dextran and glucose are released to the same extent during lysis would suggest that for LUVETs of 100 um diameter a lower limit of about 12 nm diameter can be set for the bilayer defect created, based on the average hydrodynamic diameter of dextran. We can also derive a size estimate for this lysisinduced defect from theoretical considerations. When spherical vesicles are exposed to an osmotic differential the resulting hydrostatic pressure produces swelling and hence a small increase in membrane surface area. If we consider that at a critical pressure lysis occurs, the hydrostatic pressure is relieved, and the membrane returns to its original surface area then the area of the defect created will equal the area difference between the initial and fully swollen membrane surface areas. Based on a value for the critical areal strain (at) of 0.03 (Needham and Nunn, 1990: SOPC:Chol vesicles) we can calculate that for 100 urn diameter systems this area difference equals 9.4 x 10-12 cm2 corresponding to a spherical hole of about 17 urn diameter. While this calculation is instructive and yields a hole size consistent with our experimental data additional considerations such as the energy factor relating to exposure of the hydophobic bilayer interior ("edge tension") may constrain hole growth. Based on the observations reported herein it is our hypothesis that osmotic lysis involves the following series of events. Upon exposure of LUVETs to a relatively large osmotic differential, water influx generates an osmotic pressure resulting in bilayer rupture. The hydrostatic pressure is thereby released, a fraction of the intravesicular solute lost and the bilayer then reseals. The intravesicular solution remains hyperosmotic. however, resulting in further water influx, subsequent membrane rupture, additional solute loss, followed again by membrane resealing. This cycle continues until sufficient intravesicular solute has been released such that the lipid bilayer is able to withstand the osmotic pressure resulting from the residual osmotic differential. Based on this model the residual osmotic differential would closely approximate the maximum osmotic gradient the vesicle bilayer could withstand  without rupturing. We have measured the maximum residual osmotic differential for vesicle systems of different sizes and observe a linear relationship between vesicle diameter and lysis pressure. as predicted by Laplace's Law. Further, the slope of this line yields a value for the membrane tension at lysis (40 dyn cur') that is in reasonably good agreement with values obtained, using the micropipette technique, for giant unilamellar vesicles composed of stearoyloleoylphosphatidylcholine and cholesterol (Needham and Nunn. 1990). In passing. it should be noted that LUVETs have previously been employed to determine a membrane Young's modulus using photon correlation spectroscopy to follow osmotically induced changes in vesicle diameter (Rutkowski et al.. 1991). Our results would indicate that caution should be exercised in the interpretation of such data given the morphological changes that precede true vesicle swelling.  CHAPTER 3 INFLUENCE OF PLASMA ON THE OSMOTIC SENSITIVITY OF LlUVETs  3.1 INTRODUCTION  A number of earlier studies have examined the influence of plasma or serum on the physical properties of liposomes: research stimulated in part by the therapeutic potential of liposomal drug delivery systems. The incorporation of cholesterol into SUVs composed of phosphatidylcholine was shown to stabilize these vesicles (Kirby et al., 1980) preventing breakdown due to assimilation of liposomal phospholipid into high density lipoproteins (Scherphof et al.. 1978: Scherphof and Morselt, 1984). However, even liposomal systems that appear physically stable in plasma or serum often exhibit increased rates of solute leakage (Allen and Cleland. 1980; Comiskey and Heath, 1990), likely as the result of protein interactions with the liposome membrane (Weinstein et al., 1981; Allen et al., 1985). As mentioned in Section 1.12. these interactions involve binding and penetration of the lipid bilayer by a hydrophobic or amphipathic protein domain. While many studies have examined the structural stability and bilayer permeability of liposomes in the presence of serum or plasma. little attention has been focussed on how plasma proteins may influence the osmotic stability of liposomes. The only literature in this area is a brief report by Allen and Cleland (1980) indicating that serum-induced leakage is increased for vesicles exposed to an osmotic gradient. In Chapter 2, the osmotic properties of LUVETs were characterized. When such vesicles are placed in a solution that is hypoosmotic with respect to the intravesicular medium. the resulting influx of water first causes the vesicles to assume a more spherical shape before an osmotic pressure is created. This pressure results in an elastic expansion of the lipid bilayer (Evans and Needham. 1987: Needham and Nunn. 1990)  3.2 MATERIALS AND METHODS  3.2.1 Lipids and chemicals  1-palmitoyl. 2-oleoyl phosphatidylcholine (P0 PC). and monooleoylphosphatidylcholine (monooleoyl PC) were obtained from Avanti Polar Lipids (Alabaster. AL). Cholesterol (standard for chromatography). oleic acid, and fatty acid depleted bovine serum albumin and human serum albumin were purchased from Sigma Chemical Co. (St.Louis. MO). 114c]-citrate. PI]-dextran (avg. mol. wt. 70,000). PM-cholesteryl hexadecylether and PM-glucose were from NEN while 114C1dipalmitoylphosphatidylcholine ([14C]-DPPC) was obtained from Amersham. 5(6)carboxylfluorescein was purchased from Eastman Kodak and purified according to Weinstein et al. (1984). Rabbit anti-sheep polyclonal antibody was purchased from Cedar Lane (Ontario. Canada).  3.2.2 Preparation of LUVETs  LUVETs were prepared in the appropriate buffers through 100 urn diameter filters as described in Section 2.2.2. Unless otherwise stated LUVETs were composed of POPC and cholesterol (Choi) in a 55:45 molar ratio.  3.2.3 Blood collection  Blood was collected in EDTA or silicone-coated tubes for the isolation of plasma or serum respectively. The tubes were spun at 2000 x g to pellet the red cells. Sodium azide (0.03%) was included in the plasma prior to the lipoprotein fractionation protocol.  3.2.4 Vesicle Trapped Volume Measurements LUVETs are not spherical and when placed in a hypoosmotic solution will initially "round up" in order to maximize their volume to surface area ratio. The increase in trapped volume associated with this morphological change was determined for POPC:Chol vesicles as described in Section 2.2.6.  3.2.5 Determination of Osmotically-Induced Solute Release from LUVs Vesicles (50 mg total lipid m1-1) were prepared in 700 IIIM NaCl. 100 inM carboxyfluorescein. 20 mM HEPES. pH 7.4 (1700 mOsm/kg). To remove unencapsulated carboxyfluorescein. an aliquot (100 ml) was passed down a Sephadex G-50 (medium) column (1.5 x 10 cm) pre-equilibrated with an isoosmotic solution, 850 niM NaCl. 20 inM HEPES, pH 7.4 and the peak lipid fraction collected. As mentioned in Section 2.2.4. carboxyfluorescein is a convenient marker for solute release: at the high intravesicular concentrations employed fluorophore quenching is essentially complete, only following leakage and consequent dilution in the external medium can a fluorescent signal be detected. To determine the kinetics and extent of osmotically-induced lysis. LUVs (6 mM phospholipid) were diluted 1:100 into buffered glucose-NaC1 solutions containing the indicated concentrations of bovine or human serum albumin. plasma, or lipoprotein fractions and incubated at 22°C. Solution osmolarities were adjusted using glucose (0 to 1.4 M) with the NaCl concentration maintained at 150 mM. At various times. a 50 ml aliquot was diluted into 3 ml of the same osmolarity glucose-NaC1 buffer and carboxyfluorescein fluorescence determined. This dilution was performed to minimize protein-mediated quenching (Lelkes and Tandeter, 1982). Total fluorophore release was achieved by the addition of octylglycopyranoside (final concentration 25 inM) to the sample. Carboxyfluorescein fluorescence was measured using a Perkin Elmer LS 50 spectrofluorimeter at 492 inn (bandwidth 2.5 nin) excitation and 520 inn (bandwidth 5 urn) emission.  - 78 -  3.2.6 Influence of Solute Molecular Weight on Osmotically-Induced Leakage LUVETs of POPC:Chol (60 mg lipid m1-1) were prepared in 850 mM NaCI. 20 inM HEPES, pH 7.4 containing 4.4 IACif14CJ-citrate m1-1 and 10 [4.Ci in1-1 (31-11-dextran (avg. mol. wt. 70,000). Aliquots were then diluted 1:40 with either 850 mM NaC1 or 150 inM NaC1 in the presence or absence of 10% plasma. After a two minute incubation, the vesicles were passed down a Biogel A 15M column (1.5 x 20 cm) pre-equilibrated with either 850 mM NaCl. 20 mM HEPES. pH 7.4 or 150 mM NaC1, 20 inM HEPES. pH 7.4. Retention of ("CI-citrate and [3H]-dextran was determined by liquid scintillation counting using a dual radiolabel program. Phospholipid concentrations were quantitated by phosphate analysis (Fiske & Subbarow. 1925).  3.2.7 Hemolytic Assay of Serum for Complement Activity The hemolytic assay for complement activity was performed as described by Whaley (1985). Sheep red blood cells (SRBs) were incubated with rabbit anti-SRB polyclonal antibodies in DGVB (5 mM sodium barbital. pH 7.4. 75 mM NaC1, 2.5% glucose, 0.5 mM MaC12, 0.15 mM CaCl2 and 0.1% gelatin) at 50°C for 30 minutes. The antibody coated cells were then washed three times with EGTA-DGVB (DGVB containing 40 mM EGTA) by pelleting (2000 x g for 5 min) and resuspension of the cells. The cells were then resuspensed in ice cold DGVB and kept on ice. Aliquots (1001.t1) of the cells were then added 1:1 to DGVB solutions containing 0 to 100% serum and incubated at 37°C for 30 min. If all components in the classical complement cascade (which shares components in the last steps of the alternative cascade) are active. the F, portion of the antibodies initiates the cascade to form membrane lesions and ultimately lysis of the SRBs. The reaction was stopped by a 1 ml addition of EGTA-DGVB buffer. After pelleting the unlysed SRB cells, the degree  of lysis was determined by the amount of hemoglobin released (absorbance at 414 nm). The total hemoglobin content was assessed by lysing cells in distilled water.  3.2.8 Lipoprotein Fractionation The total lipoprotein fraction was separated from plasma by adjusting the plasma density to 1.25 g m1-1 before centrifuging at 114 000 x g using a Beckman 60Ti rotor at 15°C for 48 h. Lipoprotein subfractions were isolated by sequential density centrifugations as reported in Wills et al. (1984). Briefly. a 1.006 g in1-1 solution (195 inM NaCl. 1 inM EDTA, 0.03 % NaN3) was carefully laved over plasma and spun at 114 000 x gav using a Beckman 60 Ti rotor at 15°C for 1811. The upper layer containing a mixture of chylomicrons and very low density lipoproteins (VLDL) was isolated using a tube slicer. The clear zone beneath this layer was removed down to the plasma volume and the plasma density was adjusted to 1.063 g m1-1 by the addition of NaBr. After a second centrifugation at 114 000 x  gav  for 20 h. the low density lipoproteins (LDL) were  isolated from the upper layer. The plasma density was then adjusted to 1.21 g m1-1 with NaBr and the solution spun for 48 h at 114 000 x g. High density lipoproteins (HDL) were isolated from the upper layer. The fractions were dialysed against 200 volumes of 150 mM NaCl. 20 mM HEPES. pH 7.4 and diluted to approximately their normal plasma concentrations based on the initial plasma volume and volume of each fraction.  3.3 RESULTS  3.3.1 Quantitation of vesicle volume increase We have shown previously that LUVETs prepared and maintained under isoosmotic conditions are not spherical (Mui et al.. 1993). This morphology is likely a  result of passage through the filter pores during preparation. In consequence, when such vesicles are exposed to an osmotic gradient at least part of the resulting influx of water can be accommodated by the vesicles "rounding up" which maximizes their volume to surface area ratio before the membrane experience an osmotic pressure. The increase in vesicle volume resulting from this morphological transformation will have the effect of diluting the intravesicular solute. and in order to calculate the actual osmotic differential experienced by the vesicles this dilution effect must be taken into account. The increase in trapped volume resulting from this shape change for the POPC:Chol LUVETs was therefore examined. As described in Section 3.2.4. LUVETs were prepared in 10 inM NaCl. 1 niM glucose (1.5 mCi m1-1 PHI-glucose) and incubated at 45°C either under isoosmotic conditions or exposed to a small osmotic gradient by adjusting the external salt concentration to 2 mM. The external solution in both cases contained 1 mM glucose (1.5 ttCi m1-113H1-glucose). Glucose is slowly membrane permeable with a half time for equilibration at 45°C of 1 h. At various times trapped volumes were determined following passage of the vesicles down a Sephadex G-25 column to remove unencapsulated l3Fi1-glucose. As shown in Figure 15, control vesicles maintained in 10 mM NaCl show little change in trapped volume over the 45 h incubation period. In contrast. vesicles allowed to "round up" exhibit a 40% increase in internal volume over the same period. In subsequent experiments this value was used to determined the actual osmotic differential experienced by vesicles exposed to a given applied osmotic gradient.  Time (hours)  Figure 15 Influence of vesicle morphology on trapped volume. Vesicles of POPC:Chol (50 mg m1-1) were made in 10 mM NaC1, 1 mM glucose (1.5 liCi m1-1 PHI-glucose) and diluted five fold with either 1 mM glucose ( • ) or 10 mM NaC1, 1 mM glucose ( 0 ) and incubated at 45°C. Trapped volumes were measured at various times as described under Methods.  3.3.2 Characterization of vesicle lysis in the presence of plasma The kinetics of carboxyfluorescein release from the LUVETs in response to an osmotic shock are illustrated in Figure 16. When vesicles prepared with an internal osmolarity of 1700 mOsm/kg are diluted into 150 mM NaC1(300 mOsin/kg) there is a rapid release (<10 sec) of about 25 % of the intravesicular carboxyfluorescein with little further loss up to 400 s. While the presence of 10% plasma greatly enhances the extent of carboxyfluorescein release the kinetics are unchanged with essentially all of the loss occurring within the earliest measureable time point. This release profile is consistant with a transient rupture of the liposomal membrane. It should be noted that when vesicles are diluted into isoosmotic buffer (150 niM NaCl. 1.4 M glucose) containing 10 % plasma. no significant carboxyfluorescein release could be observed (Figure 16). In order to eliminate the possibility that high concentrations of glucose present in the isoosmotic buffer inhibited the plasma effect. vesicles prepared with an interior osmolarity of 300 mOsm/kg were diluted into 150 inM NaC1, 10 % plasma: again very little carboxyfluorescein release is observed over the time period followed.  The influence of the magnitude of the applied osmotic gradient on solute release was next examined in the presence and absence of plasma. LUVETs prepared with an intravesicular solute osmolarity of 1700 mOsin/kg were diluted into solutions of various osmolarities and the extent of carboxyfluorescein release monitored. As shown in Figure 17, in the absence of plasma little fluorophore release is observed until an osmotic differential in excess of about 1100 mOsin/kg is applied. As the differential is increased above this value proportionately more carboxyfluorescein is lost from the vesicles. In the presence of 10 % plasma. however, the threshold value for solute release is considerably reduced. to approximately 700 mOsm/kg. Based on an initial internal osmolarity of 1700 mOsin/kg and allowing for a 40 % increase in trapped volume, we can calculate that an osmotic gradient of about 500 mOsin/kg can  be accommodated by the vesicles "rounding up". Thus the actual minimum osmotic gradients the vesicles can withstand without lysis are 200 and 600 mOsm/kg in the presence and absence of plasma. respectively. The ability of plasma to lower the minimum osmotic gradient needed to initiate lysis suggests that a plasma component is able to interact with, and destabilize the liposome membrane prior to the formation of major bilayer defects such as those transiently created by lytic rupture. This interpretation is supported by the observation that if vesicles are diluted into a hypoosmotic buffer and plasma subsequently added, the extent of carboxyfluorescein release is the same as for LUVETs osmotically shocked in the presence of plasma (data not shown).  70 60 50 40 30 20 10 1111117-1  0  ^  100  ^  200  ^  300  ^  400  ^  500  Time (seconds) Figure 16 Time course of carboxyfluorescein release from osmotically stressed vesicles. POPC:Chol vesicles (1700 mOsm/kg internal osmolarity) were diluted 1:100 into: hypoosmotic (150 mM NaCl, 20 mM HEPES. pH 7.4) buffer in the presence ( • ) or absence ( • ) of 10 % plasma: or isoosmotic buffer (1.4 M glucose. 150 mM NaC1, 20 mM HEPES, pH 7.4) in the presence of 10 % plasma (0 ). POPC:Chol vesicles made in isotonic 300 mOsm/kg buffer (100 mM carboxyfluorescein, 20 inM HEPES, pH 7.4) were also similarly diluted into 150 rnM NaC1 buffer in the presence of 10 %  plasma (L J . - 84 -  70  •  60 50 40  •  30 20  •  • ^ ^• ,^ V •  10 0  500  ^  • •  1000  •  ^  1 500  Applied Osmotic Differential (mOsm/kg)  Figure 17 Influence of the osmotic differential on vesicle lysis. POPC:Chol vesicles  (1700 mOsm/kg internal osmolarity) were diluted 1:100 into solutions of varying osmolarity containing 150 niM NaCl. 20 niM HEPES, pH 7.4 and 0 to 1.4 M glucose, either in the presence ( • ) or absence ( • ) of 10 % plasma. After 2 minutes, the extent of carboxyfluorescien release was determined fluorometrcally following dilution of an aliquot of this mixture 1:60 with a saline/glucose solution of the same osmolarity.  1000 800 600 400  •  200  * 0  V. 1000  500  1500  Applied Osmotic Differential (mOsm/kg)  Figure 18 Calculated residual osmotic differential after vesicle lysis. The amount of  carboxyfluorescein released from POPC:Chol vesicles exposed to various hypoosmotic buffers in the presence ( • ) and absence ( • ) of 10 % plasma (Figure 16) was used to calculate the vesicles' residual osmotic differentials, taking into account an initial 40 % increase in trapped volume due to the vesicles "rounding up".  - 85 -  After vesicle lysis the theoretical residual osmotic gradient can be calculated taking into account both the amount of solute released and the increase in trapped volume associated with "rounding up". In Figure 18 we show the calculated residual differential as a function of the applied osmotic gradient using carboxyfluorescein release data taken from Figure 17. For applied osmotic gradients of less than about 500 mOsm/kg, influx of water can be accommodated by the vesicles adopting a more spherical shape: the residual differential therefore will be zero. For applied osmotic gradients greater than 500 mOsm/kg. however, influx of water into spherical LUVETs will create an osmotic pressure. The dashed line in Figure 18 represents the expected residual osmotic differential in the absence of any vesicle lysis. For vesicles exposed to osmotic gradients of varying magnitude in the absence of plasma. the residual differential approaches a limiting value of about 600 mOsm/kg (Figure 19). In the presence of 10 % plasma. however, the vesicle residual differential plateaus at about 200 mOsm/kg. It is notable that while the residual differentials in the presence and  absence of plasma are very different they are in each case in good agreement with corresponding values for the minimum osmotic gradient required to initiate lysis. In the case of vesicles osmotically shocked in the absence of plasma, we have shown previously that this correlation between the lysis threshold and residual differentials arises because vesicle lysis results in only partial release of intravesicular solute (Mui et al., 1993). Earlier studies have reported that SUVs composed of dipalmitoylphosphatidylcholine when exposed to either high density lipoprotein (HDL) or Apo-A-I lyse releasing all encapsulated solute (Weinstein et al.. 1981: Klausner et  al., 1985). We were therefore interested in determining whether osmotic lysis of LUVETs in the presence of plasma results in only partial carboxyfluorescein release or whether it constitutes an "all or nothing" response similar to that reported by Weinstein and colleagues (1981). The experiment described below was performed to resolve this issue.  When carboxyfluorescein is encapsulated within LUVETs it exhibits a steep concentration dependent self-quenching between 0.5 and 10 nuM (Figure 19). This behavior is also influenced by the ionic strength of the medium with the degree of selfquenching being less pronounced at lower salt concentrations. We can take advantage of fluorophore quenching to distinguish between partial and total solute release following lysis in the presence of plasma. using a procedure similar to that described in Section 2.3.4. (1981). The experimental rationale is that if lysis results in complete loss of intravesicular solute then any carboxyfluorescein remaining entrapped must be in unlysed vesicles at its original concentration and hence initial level of quenching. Conversely. if all vesicles lyse releasing a portion of their contents only. then the level of fluorophore quenching will be reduced in direct proportion to the percentage of solute lost. Vesicles composed of POPC:Chol containing 10 mM carboxyfluorescein, 835 mM NaC1, 20 mM HEPES pH 7.4 and ['4C1-citrate (5 !ICI m1-1) were hypoosmotically lysed in the presence of 10 % plasma. Carboxyfhtorescein and [HQcitrate released were then removed by passage of the vesicles down a Sephadex G-50 column. As shown in Table 6. two osmotic gradients of differing magnitude were examined both of which were in excess of that needed to trigger lysis. At applied osmotic gradients of 700 and 1400 mOsm/kg. ('4C1-citrate citrate and carboxyfluorescein are released to similar extents (Table 6). Based on the fluorescence quenching determined following lysis. we can calculate the intravesicular carboxyfluorescein concentration using the quench curves shown in Figure 19. This measured concentration can then be compared to concentrations predicted by the "partial release" or all or nothing" models after correcting for trapped volume changes associated with "rounding up". It is clear from Table 6 that the experimental data is consistent with the partial release model and implies that osmotic lysis of LUVETs even in the presence of plasma does not result in complete solute loss. It should be noted that in this experiment vesicle swelling resulting from the applied osmotic gradients will not significantly contribute  to the observed changes in fluorophore quenching. Based on a fractional increase in membrane area before failure (at) of 0.03 for stearoyloleoylphosphatidylcholine:cholesterol vesicles (Needham and Nunn. 1990) we can calculate a volume increase due to swelling of less than 5 % giving rise to a reduction in carboxyfluorescein quenching of less than 2 %. The size of the membrane defects created during osmotic lysis were probed by measuring the release of aqueous markers of differing size. In the absence of plasma. both 114Q-citrate (mol. wt. 180) and (31-1)-dextran (ay. mol. wt. 70 000. ay. hydrodynamic diameter 12 nm) are released to the same extent (Table 7). This suggests that plasma components do not significantly alter the size of the lysis defect or interfere with resealing of the vesicle membrane following rupture.  2^4^6^8  ^  10  Intravesicular Carboxyfluorescein (mM)  Figure 19 Quench curve of carboxyfluorescein entrapped in POPC:Chol vesicles.  Carboxyfluorescein fluorescence was determined in POPC:Chol vesicles containing 750 mM NaC1, 20 mM HEPES, pH 7.4 ( • ) or 300 mM NaC1, 20 mM HEPES pH 7.4 ( 0 ) and 0.2 to 10 rriM carboxyfluorescein. Total carboxyfluorescein fluorescence was determined by addition of octylglucopyranoside (25 mM final).  Table 6 Evaluation of Theoretical Models of Solute Release During Plasma Enhanced Lys is.  Applied osmotic^`Yo Retained^% Fluorescence ^Intravesicular [CF] (mM) differential^114Q-citrate CF^Quenching^Theoretical^Measured3 Total'^Partial2 (mOsm/kg)^ 0  100  100  43  10  10  10  700  87  92  58  7.1  6.4  6.5  1400  53  56  81  7.1  3.9  4.1  1 Theoretical intravesicular carboxyfluorescein concentration based on a model in which total solute release occurs from vesicles during lysis and hence any retained carboxyfluorescein is in unlysed vesicles. The calculation is based on an initial carboxyfluorescein concentration of 10 rnM and assumes a 40 % maximum increase in vesicular volume due to "rounding up". 2 Theoretical intravesicular carboxyfluorescein concentration based on a model in which lysis results in partial solute release from all vesicles in the population. Again the calculation is based on an initial carboxyfluorescein concentration of 10 mM and assumes a maximum 40 % increase in vesicular volume due to "rounding up". 3  Determined from Figure 19 based on the measured fluorescence quenching.  Table 7 Influence of Solute Molecular Weight on Release During Vesicle Lysis in the Presence of Plasma. % Release' ['4C]- citrate citrate^131-11-dextran  Buffer  ^  10 % plasma  ^  24+8^  20 + 10  39 + 8^  30 + 10  1 POPC:Chol vesicles were subjected to hypoosinotic lysis in the presence and absence of 10 % plasma as described under Methods. The measurements are from an average of 6 trials and the errors represent one standard deviation.  We next examined the effect of plasma concentration on vesicle osmotic lysis. Figure 19 shows the residual osmotic differential as a function of the applied osmotic gradient for LUVETs exposed to hypoosmotic solutions containing varying concentrations of plasma (0-10%). Again the dashed line represents the expected residual differential in the absence of lysis. There is clearly a plasma concentration dependent reduction in the residual differential and this effect is titrateable with little additional increase in osmotic sensitivity above 5 % plasma at the vesicle lipid concentration employed (60 mM).  800 700 600 500  0  0  •  400 300 200  A  100  A  A  • A  E  0 0  500  1000  1500  Applied Osmotic Differential (mOsm/kg)  Figure 20 Influence of plasma concentration on vesicle lysis. POPC:Chol vesicles (1700 mOsm/kg internal osmolarity) were exposed to various hypoosmotic buffers containing 0 % ( • ). 0.02 % (0 ). 0.1 % ( • ). 0.2 % ( 0 ). 5 % (  A ) and 10 % plasma  ( A). After a 2 minute incubation at 23°C the amount of carboxyfluorescein released was measured and the residual osmotic differential calculated.  3.3.3 Plasma component(s) responsible for lysis Having characterized the influence of plasma on the osmotic sensitivity of LUVETs we next sought to identify the component or components responsible for this effect. In the following series of experiments the influence of individual plasma constituents on vesicle lysis was examined. As the most abundant protein present in plasma. albumin was studied first. Figure 21 shows that the presence of 0.5 % fatty acid-free bovine serum albumin (BSA) has only a modest influence on vesicle lysis. Given that this BSA concentration is equivalent to that present in 10 % plasma it is clear that at lower plasma concentration. where the potentiation of vesicle lysis is still almost maximal, the contribution to this effect by albumin will be negligible. A second plasma constituent that might be anticipated to destabilize the liposome membrane would be the complement system. This system plays an important role in defense against foreign organisms. Once activated, complement proteins can insert into the target membrane creating pores and hence triggering lysis of the organism. To determine whether this system is involved in potentiating vesicle lysis, complement proteins were inactivated by heating serum at 56°C for 30 min. In this experiment serum was used instead of plasma due to the need to include Mg2+ in the hemolytic assay employed to determine complement activity. While incubation at 56°C eliminated complement activity, as shown in Figure 22. little difference was observed between heat inactivated and normal serum with respect to its ability to potentiate vesicle osmotic lysis.  800 E  700  •••  600 az,  500  /•0  •  400 0  300  E 0 0  200  °  100  00  ^A • A A AAA  ./‘  0  500  ^  1000  ^  1500  Applied Osmotic Differential (mOsm/kg)  Figure 21 Influence of albumin on vesicle lysis. POPC:Chol vesicles (1700 inOsm/kg  internal osmolarity) were exposed to various hypoosmotic buffers ( • ) or hypoosmotic buffers containing 10 % plasma ( A ) or 0.5 `)/0 BSA ( 0 ).  500 400 300  •  •••••  200 100  0  ^  500  ^  1000  ^  1500  Applied Osmotic Differential (mOsm/kg)  Figure 22 Influence of complement inactivated serum on vesicle lysis. POPC:Chol  vesicles (1700 mOsm/kg internal osmolarity) were exposed to hypoosmotic buffers in the presence of complement inactivated ( 0) or normal ( • ) serum. As described in Materials and Methods, the plasma proteins necessary for complement activity were denatured by incubating serum at 56°C for 30 minutes.  - 93 -  Several lipid species. including cholesterol. fatty acids and lysophospholipids. are readily able to exchange between membranes. Consequently incubation of POPC:Chol LUVETs with plasma might alter the lipid composition of the vesicles resulting in the observed enhanced susceptibility to osmotic lysis. The LUVETs used in this study were prepared with a cholesterol content (45 mole %) which should minimize any net sterol transfer upon incubation in plasma (Cooper et al.. 1975). Migration of fatty acid or lysophospholipid from chylomicrons. albumin or lipoproteins to the vesicles would. however, be expected and we therefore examined the influence of these lipids on the osmotic properties of POPC:Chol LUVETs. The incorporation of oleic acid (0 - 8 mole %) or monooleoyl PC (0 - 10 mole %) into POPC:Chol vesicles. however, did not increase their susceptibility to osmotic lysis (results not shown). We next turned our attention to the lipoproteins and using density gradient centrifugation separated a total lipoprotein fraction from whole plasma. When diluted to a concentration equivalent to that in plasma this lipoprotein fraction enhanced vesicle osmotic lysis to approximately the same extent as whole plasma (Figure 23). In contrast the lipoprotein-depleted plasma exhibited similar properties to those of albumin alone (c.f. Fig. 21). Further fractionation of the total lipoprotein pool using density gradient centrifugation yielded the chylomicrons plus very low density lipoproteins (VLDL). low density lipoproteins (LDL) and high density lipoproteins (HDL). Each fraction was diluted back to its normal plasma concentration and its influence on osmotic lysis was then assayed. While all three fractions enhance lysis. as shown in Figure 23. HDL had the most pronounced effect and was in fact comparable to whole plasma.  cy) E  800 700 600 500 400  0  cs  A  A  • •  300 200 100  -to _ co a)  0 0  ^  500  ^  1000  ^  1500  Applied Osmotic Differential (mOsm/kg)  Figure 23 Influence of lipoproteinfractionated plasma on vesicle lysis. Lipoproteins were fractionated from plasma by density centrifugation as described under Methods. The infranatent and lipoprotein containing supernatent were then diluted to their normal plasma concentrations based on the total plasma volume, and the volume of the two fractions. The ability of 0.2 % infranatent (0 ) and 0.2% lipoprotein (Li  )  fractions to osmotically lyse POPC:Chol vesicles were compared to 0.2 % plasma (A  ).  500 400 300 200  •  100  500  ^  1000  •  ^  1500  Applied osmotic gradient (mOsm/kg)  Figure 24 Influence of VLDL and chylomicron. LDL. and HDL fractions on vesicle lysis. The various lipoproteins were fractionated by sequential density centrifugation and diluted back to their normal plasma concentrations as described under Methods. The ability of these fractions to osmotically lyse vesicles were then determined by exposing POPC:Chol vesicles to hypoosmotic buffers containing the equivalent of 10 % VLDL and chylomicron ( 0 ), LDL ( • ), and HDL ( • ) fractions as well as 10 % whole plasma  ( • ).  - 95 -  3.3 DISCUSSION  The results presented here have important implications both with respect to the interaction of plasma components with lipid vesicles and relating to the application of liposomes as systemic delivery vehicles for therapeutic agents. These two areas will be discussed in turn. It is clear that plasma dramatically increases the osmotic sensitivity of POPC:Chol vesicles with the membrane tension at lysis being reduced from about 36 dyns/cm to about 12 dyns/cm. Despite this increased sensitivity, however, the characteristics of lysis in either the presence or absence of plasma remain similar. In both cases, for example, solute loss is relatively independent of molecular weight. at least for compounds of less than about 70 kilodaltons. Similarly. lysis is not an all-ornothing process but instead results in only partial loss of intravesicular solute with the percentage released depending upon the magnitude of the osmotic gradient. In both the presence and absence of plasma therefore, following bilayer resealing, the intravesicular solution remains hypertonic with respect to the external medium. As discussed in Chapter 2 the observation that the minimum osmotic differential required to trigger lysis is of similar magnitude to the residual gradient following lysis can be accounted for by a model that assumes solute release occurs over a series of lytic events. It is our belief that on exposure of large unilamellar vesicles to a relatively large osmotic differential. water influx generates an osmotic pressure resulting in bilayer rupture. This allows dissipation of the hydrostatic pressure and a fraction of the intravesicular solute is lost before the bilayer reseals. The intravesicular solution will remain hyperosmotic. however, resulting in further influx of water. subsequent membrane rupture, additional solute loss, followed again by membrane resealing. This cycle will continue until sufficient intravesicular solute has been released such that the lipid bilayer is able to withstand the osmotic pressure  resulting from the residual osmotic differential. Plasma. we contend, decreases the strain tolerance of the vesicle membrane necessitating additional cycles of swelling and rupture before the bilayer can tolerate the residual differential. While our observation that osmotic lysis in the presence of plasma results in only partial solute release appears to be at variance with earlier work (Weinstein et al.. 1981). the two experimental systems are very different. Weinstein and co-workers observed that interaction of serum lipoproteins or apolipoproteins with small unilamellar vesicles (SUVs) composed of dipahnitoylphosphatidylcholine could induce complete release of entrapped carboxyfluorescein in an "all-or-nothing" manner. In contrast to the present study. however. solute release was triggered by heating the vesicles through their gel to liquid-crystalline transition temperature. In this system protein interaction, and presumably solute release, is believed to occur at the lipid phase domain boundaries. Further, the authors noted that the characteristics of phase transition release were dependent on the type of liposome studied, large multilamellar. and large unilamellar vesicles exhibiting different behavior from that displayed by small unilamellar systems (Weinstein et al.. 1981).  The ability of plasma to increase vesicle osmotic sensitivity does not arise from a non-specific protein interaction. Albumin and other soluble plasma proteins, including those associated with the complement system. exert only a modest influence on the osmotic behavior of LUVETs. In contrast, all of the lipoprotein fractions examined lowered vesicle tolerance to osmotic pressure. with high density lipoproteins exhibiting the most pronounced effect. This observation is consistent with earlier studies which examined plasma-. or serum-induced solute leakage from liposomes and which identified lipoproteins as playing a major role in this process (Comiskey and Heath, 1990). As mentioned in the Introduction. lipoprotein-vesicle association appears to be influenced by the lipid packing density. This would be anticipated if protein interaction involved insertion of a hydrophobic or amphipathic domain into the  - 97 -  bilayer. In turn this provides a rationale for the increased osmotic sensitivity exhibited by vesicles in the presence of plasma. Upon exposure of vesicles to a hypoosmotic medium there will be a net influx of water creating a hydrostatic pressure. This outward pressure will cause elastic stretching of the membrane arid the resulting increase in area per lipid molecule will favour protein insertion. If this penetration disrupts bilayer cohesive interactions. vesicle rupture will occur. While this interpretation must be considered speculative, support for the general principle is provided by recent studies which examined the ability of certain amphiphiles to selectively interact with. and disrupt. vesicles subjected to osmotic stress (Naka et al.. 1992). This study compared the ability of several novel surfactants to elicit release of carboxyfluorescein from large unilamellar vesicles maintained under isoosmotic conditions or subjected to a hypotonic medium. Two of the compounds tested were lytic only when vesicles were osmotically stressed suggesting that bilayer penetration. and subsequent disruption, occurred only when the membrane was stretched and the lipid packing density therefore. reduced. Similar studies with synthetic amphipathic peptides may provide additional support for the mechanism we have proposed to account for lipoprotein-induced osmotic lysis. The present study also has practical implications with respect to the use of liposomes as drug delivery vehicles. A number of studies have demonstrated that. following intravenous administration. liposomes preferentially accumulate at sites of inflammation and disease, including tumor sites, and tend not to be deposited in healthy organs Such as the heart and kidney (for a review see Ostro and Cullis. 1989). As a consequence. liposomal encapsulation can reduce the toxic side effects of certain drugs on normal tissue. while maintaining or enhancing drug efficacy. Clearly. in order to derive the maximum benefit from liposomal delivery it is essential that the drug remains encapsulated until it's carrier either accumulates at the disease site or is cleared from the circulation. For a wide variety of pharmaceutical agents. efficient  liposomal uptake at high drug-to-lipid ratios can be achieved employing vesicles exhibiting a transmembrane pH gradient. interior acidic (Madden et al.. 1990). Drugs which are weak bases will redistribute between the intravesicular solution and the external medium in accordance with the Hendersen-Hasselbach equation. This uptake process. however, consumes intravesicular protons as accumulated drug is protonated at the vesicle interior. In order to maintain the pH gradient and maximize drug uptake therefore, it is important that the intravesicular medium provide a large buffering capacity. For such systems the final drug concentration encapsulated is (theoretically) dependent only on the buffer concentration of the intravesicular solution. We show here. however, that the osmotic sensitivity of liposomes. particularly in the presence of plasma. places constraints on the osmolarity of the intravesicular medium. Clearly given the objective identified above that a liposomal drug should remain associated with its carrier, the intravesicular buffering capacity. and hence osmolarity. should be selected so as to ensure that osmotic lysis does not occur upon exposure to plasma. For vesicles of 100 nm diameter composed of POPC:Chol we can set an upper limit to the osmotic differential they can tolerate of about 200 mOsm/kg. Given that plasma has an osmolarity of 290 mOsm/kg and taking into account the volume increase associated with "rounding up" of LUVETs the initial intravesicular osmolarity should ideally not exceed about 680 mOsm/kg.  CHAPTER 4 INFLUENCE OF ASYMMETRIC TRANSBILAYER LIPID DISTRIBUTIONS ON THE MORPHOLOGY OF LUVs 4.1 INTRODUCTION Differences between the surface areas of the two monolayers comprising a bilayer membrane can have dramatic effects on membrane morphology. This applies to biological membranes as well as lipid vesicle systems. In the case of the erythrocyte membrane. for example. the external addition of amphipathic compounds to increase the area of the outer monolayer results in a transition from discoid to echinocyte morphology (Sheetz & Singer. 1974). Alternatively, for giant lipid vesicles (1-10 [tin diameter), a larger inner monolayer surface area in comparison to the outer monolayer area can result in the formation of small internalized vesicles within the giant vesicle (Kas & Sackmann. 1991: Farge & Devaux. 1992). These small vesicles are joined to the external giant vesicle by narrow "tethers". As the area of the exterior surface monolayer increases with respect to the inner monolayer area. a variety of shapes progressing from the small vesicles inside giant vesicles. to discoid shapes and then to small vesicles outside the giant vesicle have been observed (Sackman et al.. 1986). This behaviour has been rationalized, for example. on the basis of the "bilayer couple" hypothesis (Svetina & Zeks. 1983. 1989) which assumes that phospholipids do not flip-flop in response to an imbalance in surface areas. but rather local invaginations or protrusions are generated. Invaginations oriented toward the interior of the vesicle compensate for excess inner monolayer area. whereas protrusions compensate for excess outer monolayer area. It has been noted for giant vesicle systems that these morphological changes occur for very small imbalances in surface area, in the range of 0.1% or less (Farge & Devaux. 1992). Further, it has been pointed out that the  - 100-  invagination or vesiculation events generated by imbalances in surface areas may play an important role in intracellular transport processes. Work in this area has been primarily confined to giant vesicles where the transbilayer area imbalances arise due to slightly different coefficients of expansion between internal and external monolayers. or the presence of impurities on one side of the membrane (Sackmann et al.. 1986: Berndl et al.. 1990: Kas & Sackmann.1991). However, previous reports from this laboratory have detailed an ability to generate asymmetric transbilayer distributions of lipids such as phosphatidylglycerol (PG) and phosphatidic acid (PA) in mixtures with phosphatidylcholine (PC) in response to transmembrane pH gradients in large unilamellar vesicles of 100 um diameter (Hope et al., 1989: Redelmeier et al.. 1990: Eastman et al.. 1991). Lipid transport arises due to the much higher permeability of the neutral (protonated) forms of PG or PA. and can result in the movement of as much as 10% of the lipid from one monolayer to the other monolayer. As there is no evidence for compensatory movement of PC in response to PG transport (Hope et al.. 1989), large changes in morphology would be expected. In this work we examine the morphological consequences of transbilayer transport of DOPG in DOPC:DOPG (9:1) LUVETs employing cryo-electron microscopy. It is shown that a progression of morphologies eventually leading to the formation of long tubules is observed as DOPG is transported to the outer monolayer. Alternatively. when PG is transported from the outer monolayer to the inner monolayer, the reverse behaviour leading to invaginated vesicles is observed. These studies establish LUV systems as useful models for the morphological consequences of transbilayer area imbalances and support the possibility that transbilaver lipid transport could play an important role in intracellular membrane transport processes.  4.2 MATERIALS AND METHODS  4.2.1 Lipids and chemicals Dioleoylphosphatidylcholine (DO PC). monooleoyl PC. and dioleoylphosphatidylglcerol (DOPG) were obtained from Avanti Polar Lipids. Inc. (Alabaster. AL). 13H]-DPPC was obtained from Ainersham Corp. (Arlington Hts.. IL). Cholesterol (standard for chromatography). TNS (2-(p-toluidinyl)naphthalene-6sulfonic acid) and all buffers were supplied by Sigma Chemical Co. (St. Louis. MO). Gold 700 mesh bare EM grids were obtained from Marivac Ltd. (Halifax. Canada).  4.2.2 Preparation of LUVETs LUVETs were prepared through 100 urn diameter filters as described in Section 2.2.2. MLVs (25 - 50 mM lipid) were formed in 530 inOsin/kg buffers composed of either 300 mM citrate.^4 or 300 inM HEPES. 40 inM Na2SO4. pH 7.5.  4.2.3 Generation of DOPG asymmetry LUVETs containing DOPC and DOPG (9:1) and radiolabeled with 2 uCif3MDPPC per mmol lipid were prepared in 300 mM citrate. pH 4 and passed down a Sepadex G-25 column pre-equilibrated with 150 inM Na2SO4. 1mM citrate. pH 4. The lipid concentration was then either adjusted to 10 inM if the sample was used for lipid asymmetry measurements or was adjusted to 20 mM if used for cryo-electron microscopy studies. In order to induce a net movement of DOPG to the outer monolayer. the vesicles (1 ml) were diluted with 0.4 ml of a buffer (pH 7.5) containing 300 mM HEPES. 40 inM Na2SO4 and 0.2 ml aliquots were incubated for various times at 60°C. DOPG transport to the outer monolayer was quenched by rapidly swirling the test tube in ice-water and the sample was then stored on ice. To induce a net transport of DOPG from the outer monolayer to the inner monolaver, a pH gradient  - 102 -  where the interior is basic is required. Vesicles were therefore prepared in 300 inM HEPES. 40 niM Na2SO4. pH 7.5 and passed down a Sephadex G-25 column preequilibrated with 150 inM Na2SO4. 1 inM HEPES. pH 7.5. A pH gradient. interior basic, was generated by diluting the vesicles with a 300 inM citrate. pH 4 buffer.  4.2.4 Detection of DOPG asymmetry using TNS Transbilayer distributions of DOPG were detected using the TNS assay described by Eastman et al. (1991). INS is a fluorescent lipophilic anion which exhibits enhanced fluorescence when associated with a lipid bilayer. Thus, the presence of a negatively charged lipid, such as PG. will decrease TNS absorption to the membrane, resulting in a decrease in fluorescence. In order to perform this assay. a 0.2 ml lipid sample was first diluted with 0.5 nil of 100 IIIM ammonium acetate. 100 mM citrate. pH 6 and 0.2 ml aliquots of this dispersion were diluted into 3 ml of 3 mM TNS. 5 inM ammonium acetate. 5 IIIM HEPES. pH 7. The fluorescence was measured using a Perkin Elmer LS-50 fluorometer at an excitation wavelength of 321 nm (bandwidth 5 urn) and emission wavelength of 445 nm (bandwidth 2.5 urn). Standard curves were constructed for DOPC and DOPC:Chol (6:3) vesicles containing various amounts of DOPG.  4.2.5 Cryo-electron microscopy As induced DOPG asymmetry was found to be stable for 12 h at 4°C. samples were kept on ice (maximum 7 h) before use. Lipid asymmetry was checked before and after each cyro- electron microscopy session. Visualization of the vesicle suspension by cryo-electron microscopy was performed as described in Section 2.2.8.  - 103 -  4.3 RESULTS  4.3.1 Kinetics and quantification of DOPG transport The first series of experiments were performed to characterize the morphological changes observed on transporting DOPG from the inner monolayer to the outer monolayer of 100 run diameter DOPC:DOPG (9:1) LUVETs. Previous studies with DOPC:DOPG (9:1) LUVETs have shown that DOPG is transported from the outer monolayer to the inner monolayer when a pH gradient. where the interior is basic (pH° = 4.0. pH; = 9.0). is applied across the vesicle membrane (Redelmeier et al.. 1990). This indicates that in order to transport DOPG from the inner monolayer to the outer monolayer at equivalent rates, a pH gradient where the interior is at pH 4.0 and the exterior is basic. should be ultized. An interior pH of 4 and an exterior pH of 7.5 were employed. The extent of DOPG transport observed in response to this ApH was then determined after an incubation at 60°C employing the TNS assay. The results are shown in Figure 25. For both DOPC:DOPG (9:1) and DOPC:Chol:DOPG (6:3:1) vesicles, nearly complete DOPG asymmetry can be obtained after a 20 min incubation at 60°C.  - 104-  CI  20 -T^.....^ ^ .0--^  CZ  .... 1^ 0^ ... 18 -^/^T -  T_-_-  --  .- -  T. ../^....0 ..- ''' 1 4--^'T T,^ ..- -. -  c':'-^  c 16^0  ir  a)^A ri +-. c^T/ 1-  o^o !  a  O  14 - ilil -4 -Ti •  a.^o/0 0^12 - ' ii  0  O  10 •^ 5^10^15^20^25^30  Time (min)  Figure 25 Quantitation of DOPG asymmetry induced by a pH gradient (pHo = 7-5,  pHi = 4). The amount of DOPG present in the outer monolayer of DOPC (9:1, mol:mol)  ( • ) or DOPC:Chol:DOPG (6:3:1, mol:mol:mol) ( 0 ) 100 nm diameter vesicles in the presence of the pH gradient and subsequently incubated at 60°C for various times was measured using the TNS assay defined in Methods. The outer monolayer DOPG concentration was determined from the INS fluorescence using a standard curve constructed from DOPC or DOPC:Chol (6:3) vesicles containing 10 to 20 mole % DOPG.  - 105 -  4.3.2 Morphological changes resulting from expansion of the vesicle's outer monolayer Morphological changes associated with the transport of DOPG to the outer monolayer were then investigated employing cryo-electron microscopy. A first objective was to obtain an accurate representation of the morphology prior to induction of lipid asymmetry. In this regard, whereas the transbilayer movement of DOPG in response to a pH gradient is relatively fast at 60°C. it is considerably slower at lower temperatures due to a high activation energy (Ea - 31 kcal/mol) for DOPG transport (Redelmeier et al.. 1990). It was found experimentally that no detectable transport of DOPG from the inner to the outer monolayer occurred in DOPC:DOPG (9:1) LUVETs with a pH gradient (pH; = 4.0. pH„ = 7.5) during a 2 h incubation at 4°C. The morphology of LUVs which are not exposed to higher temperatures should therefore provide an accurate depiction of the initial vesicle shapes. Representative shapes observed by cryo-electron microscopy are shown in Figure 26A. The most notable features are the vesicles' dimple-like structures, which could be interpreted as vesicles within vesicles. However, closer inspection of Figure 26A suggests that these structures represent invaginated LUVETs. where the vesicles indicated by the arrow represents an edge-on view. As indicated in Sections 4.1 and 1.11. a possible reason for invaginated morphology is an imbalance between the surface areas of the inner and outer monolayers. where the surface area of the inner monolayer is too large to be compatible with a spherical morphology. If this is the case, transport of DOPG to the outer monolayer should reduce the area of the inner monolayer and result in the disappearance of the dimple-like structures. This is what is observed as LUVs which have been incubated at 60°C for 3 and 20 minutes are transformed into long tubular shapes, as well as vesicles exhibiting one or more projections (Figures 26B. 26C). The presence of much smaller vesicles. which may have resulted from fragmentation of these tubular arms, are also evident in Figure 26C. These shape changes are not the  - 106 -  result of the 60°C incubation step as DOPG:DOPC (9:1) vesicles incubated for the same periods of time in the absence of a pH gradient (pH; = pH„ = 4.0) do not change their initial morphology (results not shown). Shape changes were also observed for LUVETs containing cholesterol. DOPC:Chol:DOPG (6:3:1) LUVETs were prepared with an internal pH of 4.0. transferred to a medium with a pH of 7.5 and incubated at 60°C for 20 min to induce essentially complete transport of DOPG to the outer monolayer. Before induction of DOPG transport. the morphology corresponds to spherical. invaginated vesicles (Figure 27A). After a 60°C incubation, long tubular structures are predominant (Figure 27B). These tubular structures appear much more stable than those observed with DOPC:DOPG (9:1) LUVETs in that the tubular projections are much longer and vesicle "fragments" were not observed. If the sequence of morphological changes in Figure 26 is simply due to a progressive increase in exterior surface area as DOPG is transported to the outer monolayer, the external addition of lipid to the outer monolayer of the LUVETs should also induce morphological changes which are similar to those caused by the transport of DOPG. The outer monolayer area can be increased by incubating the vesicles with lysophosphatidylcholine. which will readily partition into the outer monolayer. DOPC:DOPG (9:1) vesicles. in the absence of a pH gradient (pH; = pH. = 4.0), were therefore incubated with monooleoyl PC at 0 to 8 % mole ratios of monooleoyl PC to phospholipid. As shown in Figure 28. vesicles incubated with 0.5 % monooleoyl PC (Figure 28B) show fewer of the invaginated structures than do vesicles that have not been exposed to monooleoyl PC (Figure 28A). The presence of higher monooleoyl PC concentrations induces further morphological changes. LUVETs incubated with 5 and 8 % monooleoyl PC are shown to depart from a spherical shape to vesicles with tubelike projections (Figures 28C and 28D respectively). This sequence of morphological  - 107-  change is similar to that observed when PG is transported to the outer monolayer in response to a pH gradient (Figure 26).  4.3.3 Morphological changes resulting from expansion of the vesicle's inner monolayer If an increase in the outer monolayer surface area can cause invaginated vesicles to become tubular or form tubular projections. then increasing the inner monolayer surface area should reverse this morphology and cause tubular vesicles to become spherical and then invaginated. As indicated above, the area of the inner monolayer can be increased by transporting PG from the outer to the inner monolayer using a pH gradient. where the vesicle interior is basic (Hope et al.. 1989: Redelmeier et al., 1990). In order to clearly demonstrate the effect of this inward transport of lipid, it was necessary to start with LUVETs which did not already exhibit invaginated structures. This was accomplished by adding 2 mol % monooleoyl PC to the exterior of LUVETs composed of DOPC:DOPG (9:1) to reduce the proportion of LUVETs which exhibited invaginated morphology. These LUVETs were then incubated in the presence of a pH gradient (interior basic) to transport DOPG to the inner monolayer. Figure 29A illustrates DOPC:DOPG (9:1) LUVETs which have not been exposed to monooleoyl PC nor incubated at 60°C. It may be observed that in the absence of monooleoyl PC and DOPG asymmetry. the vesicles are predominantly invaginated. However, the vesicles become tubular after the addition of 2 % monooleoyl PC to the outer monolayer (Figure 29B). As shown in Figure 29C. this morphological change can be reversed by the subsequent transport of DOPG to the inner monolayer following a 10 min incubation at 60°C. It is of interest to note that the extent of inward DOPG transport was reduced in the absence of 2 % monooleoyl DOPC. An analysis of DOPG asymmetry induced in  DOPC:DOPG (9: 1) vesicles indicate that the maximum amount of DOPG that can be  - 108 -  transported into the inner monolayer is 70 % of the exterior DOPG whereas complete DOPG asymmetry can be obtained for vesicles which have had their outer monolayer area increased by a pre-incubation with 2 % monooleoyl PC (results not shown). 4.3.4 Influence of the aqueous medium on vesicle morphology DOPC:DOPG (9:1) vesicles under isoosmotic conditions (300 mM citrate pH 4 inside and out) show a variety of shapes ranging from invaginated to tubular vesicles (Figure 30A). From the results of our present study. such structures are indicative of vesicles containing more inner or outer monolayer membrane areas respectively than that needed to form a sphere. When these vesicles made in 300 mM citrate pH 4 buffer (530 mOsm/kg) are exposed to an external hypoosmotic pH 7.5 buffer consisting of 118 mM Na2SO4. 86 inM HEPES (330 mOsm/kg). invaginated vesicles are observed as shown in Figure 26A. It is unclear at present why these structures are seen. It is not a result of a change in the external pH as exposure of these vesicles to an external pH 4 buffer consisting of 107 inM Na2SO4. 86 mM citrate (330 mOsm/kg) also results in invaginated vesicle morphology (Figure 30B)  - 109 -  Figure 26 Morphological changes in DOPC:DOPG (9:1, mol:mol) vesicles generated by the transport of DOPG to the outer monolayer. Cyro-electron micrographs were taken of DOPC:DOPG (9:1) vesicles which have been exposed to a pH gradient (pH. = 7.5, pHi = 4.0) and incubated at 60°C for (A) 0 min: (B) 3 min: and (C) 20 min. The bar represents 200 nm.  - 110 -  Figure 27 Morphological changes in 100 nrn diameter DOPC:Chol:DOPG (6:3:1, mol:mol:mol) vesicles induced by the transport of DOPG to the outer monolayer. Cyroelectron micrographs were taken of DOPC:Chol:DOPG (6:3:1, mol:mol:mol) vesicles exposed to a pH gradient (pH. = 7.5, pHi = 4.0) before (A) and after (B) a 60°C incubation step for 15 min to generate DOPG asymmetry. The bar represents 200 inn.  Figure 28 Morphological changes of DOPC:DOPG (9:1) vesicles induced by the partitioning of monooleoyl PC into the outer monolayer of the 100 nm LUVs. DOPC:DOPG (9:1) vesicles prepared in 300 mM citrate, pH 4 were incubated for 5 min at 23°C with (A) 0 %, (B) 0.5 %, (C) 5 % and (D) 8 % mole ratios of monooleoy1PC to phospholipid in an external medium composed of 150 mM Na2SO4, 5 mM citrate pH 4. The bar represents 200 nm.  Figure 29 Morphological changes of DOPC:DOPG vesicles incubated with monooleoyl PC and subsequently exposed to a pH gradient (interior basic) to induce DOPG transport to the inner monolayer. DOPC:DOPG (9: 1 ) vesicles prepared in 300 mM HEPES,  40 mM Na2SO4 pH 7.5 were incubated for 5 min at 23°C in the absence (A) and presence (B) of a 2 % mole ratio of monooleoyl PC to phospholipid in 150 mM Na2SO4, 1 mM HEPES pH 7.5. Vesicles incubated with 2 % monooleoyl PC were then heated to 60°C for 10 min to generate DOPG asymmetry (C). The bar represents 200 nm.  Figure 30 Influence of the aqueous medium on vesicle morphology. The morphology of DOPC:DOPG (9:1) LUVs made and maintained in 300 mM sodium citrate pH 4 range from invaginated to tubular structures (A). When the external solution is changed to 107 inM Na2SO4, 86 mM citrate, pH 4 the vesicles are predominantly invaginated (B). The bar represents 200 nm.  - 114-  4.4 DISCUSSION The results presented here establish LUV systems as useful models for examining the morphological consequences of imbalances between the surface areas of inner and outer monolayers and provide interpretations for the morphological properties of LUVET systems. Further, the observations made here provide support for the proposal that transbilayer lipid transport processes could play an important role in the intracellular trafficking of membranes. The shapes of LUVETs, in the absence of transbilayer transport of PG are clearly influenced by the areas assumed by the inner and outer monolayers. These areas are not necessarily compatible with a spherical shape. Under isoosmotic conditions, extruded LUVs can display a variety of morphologies ranging from invaginated to tubular vesicles. These different morphologies are due to differences in surface areas between the inner and outer monolayers. In this regard. a 100 nm diameter spherical LUV should have approximately 10 % more area in the outer monolayer as compared to the inner monolayer. assuming a bilayer thickness of 4 nm. Tubular vesicles will occur if the relative outer monolayer area is greater than this value while invaginated shapes will occur if the outer monolayer area has less than a 10 % excess surface area. The shearing process involved in extruding an MLV through a 100 nm filter can influence the relative areas between the inner and outer monolayers by affecting the curvature of the membrane fragment just before it breaks off and re-anneals to form the LUV. Also. the degree of stretching or compression of the fragment's monolayers will also be a factor because when the fragment re-anneals to form a LUV, the lipids will tend to relax back to its unstressed state. These factors may be influenced by the lipid composition. and the temperature. speed, pressure. and filter pore size used in the extrusion process. In addition to the extrusion process. other environmental factors also appear to influence the relative inner and outer monolayer areas. Specifically. in the absence of  - 115 -  DOPG transbilayer asymmetry. the morphology of DOPC:DOPG (9:1) LUVs made in 300 mM sodium citrate, pH 4 and exposed to an external solution of 107 inM Na2SO4, 86 mM sodium citrate. pH 4 are invaginated (Figure 30B). At present. it is unclear why this morphology is observed. A more through study of the effects of different salt environments and lipid composition on vesicle shape may better clarify this phenomenon. The observation that LUVs can change shape to accommodate changes in relative monolayer areas would suggest that transbilayer tension is not the constraining factor in the redistribution of DOPG as suggested by Farge and Devaux (1993). The limiting factor is more likely the development of an osmotic pressure due to vesicle volume changes when invaginated or tubular structures are formed. In this regard, a 1 % change in vesicle volume, corresponding to an osmotic pressure difference of 5 mOsin/kg in our system. would require an energy input of 4 x 106 J/mol for a 100 nm diameter sphere. In comparison, the activation energy for transbilayer movement of PG is on the order of 1 x 105 J/mol (Redelmeier et at. 1990). An osmotic gradient can also be generated when DOPG is transported from the outer to the inner vesicle monolayer. Transport of DOPG occurs through its protonated neutral form (Redelmeier et al.. 1990). Once in the inner monolayer. where the aqueous environment is basic, the proton is transferred to the HEPES buffer as the pl{„ of the DOPG phosphate is approximately 2 (Watts et al.. 1978) compared to HEPES's pKa of 7.5. Protonation of HEPES results in the dissociation of the counter sodium cation and thus an increase in the intravesicular osmolarity. For example. transport of 10 % of the outer monolayer lipid to the inner monolayer would result in an increased intravesicular osmolarity of 50 mOsm/kg for a vesicle with a trapped volume of 1 ttl/ttmol lipid and 4 x 104 lipids/vesicle monolaver. Obviously, as the vesicle volume is decreased due to morphological changes. the intravesicular osmolarity will increase proportionally.  In contrast to smaller spherical vesicles seen in giant liposomal systems. the predominant structure observed when the outer monoler area is increased in LUV systems are tubular structures. These projections likely correspond to the "tethers" which connect smaller vesicles to the parent vesicle when surface area imbalances are induced in giant liposomal systems (Kas & Sackmann. 1991). The physical properties of the membrane in the tubular region may significantly be different from the rest of the vesicle membrane due to both curvature effects and lipid compositional differences. As suggested by Oster et at (1989). these compositional differences may arise from lateral lipid segregration. with lipids more stable in a highly curved membrane redistributing to the tubular domain. The LUV system described here provide a potentially useful model system to characterize the physical properties (e.g. lipid composition. fusogenic tendency) of these tubular structures. The morphological changes observed on inducing DOPG transport in cholesterol containing vesicles is an interesting observation in that there is strong evidence that cholesterol is able to undergo relatively rapid transbilayer movement (Schroeder et al., 1991). The fact that similar morphological changes are observed in the presence of cholesterol as in the absence of cholesterol suggests that it is energically much more favourable for the vesicle to undergo dramatic shape changes in response to imbalances between the areas of inner and outer monolayers. than to induce transbilayer redistributions of lipid. In turn, this would indicate that transbilayer packing differences do not play a sign ificant driving role in promoting the transbilayer transport of lipid. It is of obvious interest to relate these results to biological systems. It has been proposed that lipid pumps transporting lipid from one monolayer to the other could play an important role in modulating membrane transport processes (Oster et al., 1989: Devaux. 1991). The results presented here are clearly consistent with such a  - 117-  proposal. For example. the tubular projections formed when excess lipid is present in the outer monolayer of the membrane may provide a driving force for the transport of lipids and associated proteins from the endoplasmic reticulum to the plasma membrane. The formation of "tubulovesicular" structures have been observed to occur between the cisternae of both the endoplasmic reticulum and the trans-golgi network (Cooper et al.. 1990: Lee et al.. 1988: Dabora et al.. 1988: Lippincott-Schwartz et al., 1990). Tubulovesicle formation could be driven by phospholipid biogenesis. which occurs predominantly on the cytoplasmic face of the endoplasmic reticulum (Bishop & Bell, 1988). or by transbilayer lipid pumps transporting ltunenal lipid to the cytoplasmic monolayer. For example. a tube of 15 + 5 nin external diameter, which appears to be the limiting size in this study. would be expected to grow at a rate of 1 nm/sec for a transbilayer lipid transport rate of - 26 lipids/sec. assuming an area per lipid molecule of 0.6 nm2.  - 118 -  CHAPTER 5 SUMMARY  This thesis is focussed on the effects of mechanical stress due to osmotic forces and transbilayer area asymmetry on the morphology and stability of LUVETs. The morphology and osmotic properties of EPC:Chol (55:45) LUVETs were examined in Chapter 1. Contrary to expectations. such vesicles visualized by cryoelectron microscopy. under isoosinotic conditions. are non-spherical. This morphology is a consequence of vesicle passage through the, filter pores during extrusion. As a result when such LUVETs are placed in a hypoosmotic medium they are able to compensate, at least partially, for the resulting influx of water by "rounding up" and thereby increasing their volume with no change in surface area. The increase in vesicle trapped volume associated with these morphological changes was determined using the slowly membrane-permeable solute PM-glucose. This allowed calculation of the actual osmotic gradient experienced by the vesicle membrane for a given applied differential. When LUVETs were exposed to osmotic differentials of sufficient magnitude lysis occurred with the extent of solute release being dependent on the size of the osmotic gradient. Lysis was not an all-or-nothing event, but instead a residual osmotic differential remained after lysis. This differential value was comparable in magnitude to the minimum osmotic differential required to trigger lysis. Further, by comparing the release of solutes of differing molecular weights (glucose and dextran) a lower limit of about 12 mu diameter can be set for the bilayer defect created during lysis. Finally, the maximum residual osmotic differentials were compared for LUVETs varying in mean diameter from 90 to 340 mu. This comparison confirmed that these systems obey Laplace's law relating vesicle diameter and lysis pressure. This analysis also yielded a value for the membrane tension at lysis of 40 dyn /cm at 23°C. which is  - 119 -  in reasonable agreement with previously published values for giant tmilamellar vesicles. The influence of plasma on the osmotic sensitivity of LUVETs was examined in Chapter 2. In 100 nm diameter POPC:Chol (55:45) LUVETs at 23°C. the presence of plasma significantly decreased the membrane tension at rupture from 36 dyn /cm to 12 dyn/cm. Despite increasing vesicle sensitivity, however. plasma does not alter the characteristics of osmotically induced lysis. As in the absence of plasma. lysis is not an all-or-nothing event but instead results in only partial loss of intravesicular solute so that following membrane resealing the vesicle interior remains hyperosmotic with respect to the external medium. To identify the component responsible for the observed increase in vesicle osmotic sensitivity, plasma was fractionated by density centrifugation. Albumin and other soluble plasma proteins. including those associated with the complement system were found to exert only a modest influence on vesicle osmotic behavior. In contrast all of the lipoprotein fractions lowered vesicle tolerance to osmotic pressure with FIDL exerting an effect comparable to whole plasma. The morphological consequences of imbalances in the surface areas of the inner and outer monolayers of 100 nm LUVETs were examined by cryo-electron microscopy techniques in Chapter 4. Surface area imbalances were generated by inducing net transbilayer transport of DOPG in DOPC:DOPG (9:1) LUVETs in response to transmembrane pH gradients. It is shown that when DOPG is transported from the inner monolayer to the outer monolayer. initially invaginated LUVETs are transformed to long narrow tubular structures, or spherical structures with one or more tubular extensions. Similar morphological behavior is observed when monooleoyl PC is allowed to partition into the exterior monolayer of DOPC:DOPG (9:1) LUVETs in the absence of DOPG transport. Conversely. when DOPG is transported from the outer monolayer to the inner monolayer of non-invaginated LUVETs, a reversion to  - 120 -  invaginated structures is observed. The narrow tubular structures have a limiting diameter of approximately 15 nm and are likely related to the intervesicular tethers previously observed in giant vesicle systems. These results are consistent with proposals that factors leading to imbalances in monolayer surface areas could play important roles in intracellular membrane transport processes. The results presented in Chapters 2 and 4 indicate that the morphologies adopted by LUVs as a result of the extrusion process is the result of both osmotic forces and relative transbilayer membrane area differences. The results of Chapter 2 demonstrate that if a counteracting osmotic force is not allowed to develop. LUVETs will "round-up" to adopt a more spherical shape to minimize membrane bending energies. This is most clearly indicated by the observation that 100 nm diameter EPC:Chol (55:45) LUVETs made and maintained in 1 inM glucose (PM-glucose) will show an increase in measured trapped volume over time whereas vesicle's made and kept in 10 mM NaC1 maintain a constant trapped volume. Results in Chapter 4 indicate that osmotic forces are not the only consideration but relative differences in transbilayer area will also play an important role. That is. under constant osmotic conditions LUVETs will assume dramatically different morphologies in response to changes in relative areas between the inner and outer vesicle mcnolayer. There are several interesting aspects arising from these studies which would be important to pursue. The first concerns the extrusion procedure. The proportion of invaginated. spherical and tubular vesicles appear to depend on the lipid composition and the buffer conditions. For example. LUVETs prepared in low ionic strength solutions appear to contain a smaller proportion of invaginated structures than in higher ionic strength solutions (c.f. Figures 8A and IA). Similarly. LUVETs of DOPC:DOPG (9: 1) appear to be more invaginated than EPC:Chol (55:45) vesicles (c.f. Figures 8B and 26A). These observations must be made with caution however, as the temperature. pressure and speed of extrusion is not comparable or controlled in the  above examples. A more thorough and systematic study is needed to determine the relative effects of different factors on extruded vesicle morphology. These studies may provide a better insight into the molecular events involved in the extrusion process. For example. extrusion conditions which allow for rapid lipid flip-flop may be one explanation for the different degrees of invaginated vesicle shapes. A second aspect of interest involves the influence of osmotic pressure on the induction of transbilayer area asymmetry. An osmotic gradient may develop through a change in intravesicular pH when PG is transported to the inner monolayer or vesicle volume changes. A development of a 5 mOsm/kg osmotic gradient. for example. would theoretically inhibit DOPG transport. In the vesicle system decribed in Chapter 4, a 5 mOsm/kg osmotic gradient could be easily generated by a 1 `)/0 vesicle volume change or transport of 1 mol % PG to the vesicle's inner monolayer. However. transport of greater than 1 mol % DOPG to the inner monolayer does occur and vesicle volume changes does seem possible based on the cryo-electron micrographs of transbilayer DOPG asymmetric vesicles (e.g. Figure 26). The generated osmotic gradient may alternatively be dissipated through vesicle lysis. A comparative study on the degree of DOPG asymmetry that can be generated with vesicle solute release and volume changes would be helpful in this respect. Another interesting topic concerns the differences in vesicle morphology observed when transbilayer area asymmetry is induced in different lipid systems by transbilayer lipid transport or with the amphipathic drug doxorubicin. When transbilayer area asymmetry is generated in DOPC:DOPG (9:1) vesicles by the transport of DOPG to the vesicle's outer monolayer. invaginated vesicles are transformed into spherical vesicles and spherical vesicles with tubular projections. However, these tubular shapes are much more pronounced and elongated in DOPC:Chol:DOPG (6:3:1) systems (c.f. Figures 26D and 27B). One possible explanation could be that the pH gradient needed to induce and maintain DOPG  - 122 -  asymmetry is better maintained in the cholesterol containing LUVETs. Another possibility could be that the resulting higher concentration of conical-shaped cholesterol molecules present in the vesicle's inner monolayer is better able to stablize a highly curved tubular shape. However, if this is the case. one would similarly expect insertion of inverted-cone shaped lysophosphatidylcholine molecules into the vesicle's outer monolayer to have the same effect. However, as shown in Figures 26D and 28D. similar vesicle shapes occur with expansion of the vesicle's outer monolayer with either DOPG (cylindrical shaped) or monooleoyl PC. If vesicle morphology is influenced by lipid shape. a more comparative experiment would be to induce transbilayer area asymmetry with PG molecules of differing degrees of unsaturation. Lasic et al. (1992) recently demonstrated by cryo-electron microscopy that doxorubicin loaded vesicles are "coffee bean" shaped. That is. the vesicles are ellipsoidal with a darkened line extending down its middle. These authors attribute the dark line to the formation of a doxorubicin gel-state in the vesicle's aqueous interior. However, it is equally likely that partitioning of doxorubicin into the vesicle's inner monolayer results in expansion of the inner monolayer and thus, the bilayer buckles inward to compensate. If this is the case, then expansion of the outer monolayer of doxorubicin loaded vesicles by the external addition of lysophophatidylcholine. for example. would eliminate the invaginated morphology. Further areas of interest concerns the possibility that lipids are preferentially sorted or sequestered to spherical or tubular domains in LUVs with differences in transbilayer monolayer areas. This may lead to different membrane fusogenic properties. This area of research is biologically significant in that the tubular vesicles and projections observed here may be related to the formation of tubulovesicular structures observed in membrane trafficking processes.  - 123 -  REFERENCES Alhanaty. E.. & Livne. A. (1974) Biochim. Biophys. Acta. 339, 146-155 Allen, T.M., & Cleland. L.G. (1980) Biochim. Biophys. Acta 597. 418-426 Allen, T.M., & Everest. J.M. (1983) J. Pharinacol. Exp. Tiler 226. 539-544 Allen, T.M., Ryan. J.L., & Papahadjopoulos. D. (1985) Biochim. Biophys. Acta 818. 2055 -2 10 Alving, C.R., & Richards. R.L. (1983) in Liposomes (Ostro. M. Ed.) Marcel Dekker, New York pp 209-287 Bangham. A.D.. de Gier. J.. & Greville. G.D. (1967) Chem. Phys. Lipids 1. 225-246 Barnholz. Y.. Aniselem. S.. & Lichtenberg. D. (1979) FEBS Lett. 99. 210-215 Batenburg. A.M.. & de Kruijff, B. (1988) Bioscience Rep. 8. 299-307 Bernd!. K.. Kas. J.. Lipowsky, R.. Sackinann. E.. & Seifert. U. (1990) Europhys. Lett. 13. 659-664 Bishop, R.W.. & Bell, R.M. (1985) Cell 42. 51-60 Bishop. R.W.. & Bell. R.M. (1988) Ann. Rev. Cell Biol. 4. 579-610 Blok, M.C., van Deenen. L.L.M., & de Gier. J. (1976) Biochim. Biophys. Acta 433, 1-12 Bloom, M., Evans. E.. & Mouritsen. 0.0. (1991) Q. Rev. Biophys. 24. 293-397 Blume, G.. & Cevc. G. (1990) Biochim Biophvs Acta 1029. 91-97 Bonte, F.. & Juliano. R.L. (1986) Chem. Phys. Lipids 40. 359-372 Branemark. P.I., & Lindstrom, J. (1963) Biorheology 1. 139 Canham, P.B. (1970) J. Theoret. Biol. 26. 61-81 Carraway, K.L., & Carraway, C.A.C. (1989) Biochim Biophys Acta 988, 147-171 Carruthers. A., & Melchior. D.L. (1983) Biochemistry 22. 5797-5807 Cass, A., & Finkelstein, A. (1967) J. Gen. Physiol 50. 1765-1770 Chobanian. J.V., Tall. A.R.. & Brecher. P.I. (1979) Biochemisty 18. 180-187 Chonn, A.. Semple, S.C.. & Cullis. P.R. (1992) J. Biol. Chem. 267. 187559-18765 Cohen, F.S.. Akabas. M.H.. & Finkelstein. A. (1980) Science 217, 458-460 Comiskey. S.J.. & Heath. T.D. (1990) Biochim. Biophys. Acta 1024. 307-317 Cooper, M.S.. Cornell-Bell. A.H.. Chernjavsk■,:. A.. Dani. J.W.. & Smith, S.J. (1990) Cell 61. 135-145 Cullis, P.R. & Hope, M.J. (1988) in Molecular Mechanisms of Membrane Fusion (Ohki. S., Doyle. D.. Flanagan. T.. Hui. S.W. & Mayhew. E. Ed.) Cullis. P.R.. & de Kruijff. B. (1979) Biochim. Biophys. Acta 559. 399-420 Cullis. P.R.. & Hope. M.J. (1978) Nature 271. 672-674 Cullis. P.R.. & Hope. M.J. (1985) in Biochemistry of Lipids and Membranes (Vance. D.E.. & Vance. J.E. Ed.) The Benjamin/Cummings Publishing Co., Inc.. Menlo Park. California pp 25-72  CuHis, P.R.. Hope. M.J.. de Kruijff. B.. Verkleij. A.J.. & Tilcock. C.P.S. (1985) in Phospholipids and Cellular Regulation (Ktio. J.F. Ed.) CRC Press. Boca Raton. Florida 1-59 Cullis. P.R., Mayer. L.D.. Bally. M.B.. Madden. T.D.. & Hope. M.J. (1989) Advanced Drug Delivery Reviews 3. 267-282 Calls, P.R.. Tilcock. C.P.. & Hope. M.J. (1990) in Membrane Fusion (Wilschut, J., & Hoekstra, D. Eds.), Marcel Dekker. New York pp 35-64 Cunningham, C.M., Kingzette, M.. Richards. R.L.. Alving. C.R.. Lint. T.F. (1979) J. Immunol. 122. 1237-1242 Dabora, S.L.. & Sheetz. M.P. (1988) Cell 54. 27-35 Damen. J.. Regts. J.. & Scherphof. G. (1981) Biochim. Biophys. Acta 665. 538-545 de Gier. J.. Mandersloot. J.G.. Hupkes. J.V.. McEllianey. R.N.. & Van Beek. W.P. (1970) Biochim. Biophys. Acta 233. 610-618 Deamer, D.W.. & Braniliall. J. (1986) Chem. Phys. Lipids 40. 167-188 Demel, R.A.. & de Kruijff. B. (1976) Biochim. Biophy. Acta 457, 109-132 Deuling, H.J.. & Helfrich. W. (1976) Le Journal de Physique. 37. 1335-1345 Deuticke, B. (1968) Biochim. Biophys. Acta 163. 494-500 Devaux. P.F. (1991) Biochemistry 30. 1163-1174 Dilger, J.P.. & McLaughlin (1979) Science 206. 1196-1198 Dimitrov. D.S.. & Jain. R.K. (1984) Biochim. Biophys. Acta 779, 437-468 Epand, R.M.. Surewicz, W.K.. Hughes. D.W., Mantsch. H.. Segrest. J.P., Allen, T.M., & Anantharamaiah. G.M. (1989) J. Biol. Chem. 264. 4628-4635 Esser, A.F., Kolb, W.P., Podack, E.R., Muller-Eberhard. H.J. (1979) Proc. Natl. Acad. Sci. USA 76, 1410-1414 Evans, E.A. (1974) Biophys. J. 14. 923-931 Evans, E.A. (1983) Biophys. J. 43, 27-30 Evans. E.A.. & Hochmuth. R.M. (1978) Curr. Top. Membr. Transp. 10, 1-62 Evans, E.A.. & Needham. D. (1987) J. Phys. Chem. 91. 4219 Evans. E.A., & Waugh. R. (1977) Biophys. J. 20. 307-313 Evans, E.A., Waugh. R. & Melnik. L. (1976) Biophys J. 16. 585-595 Fahey, P.F.. Webb. W.W. (1978) Biophys. J. 21. Al24 Falke, L.C.. & Misler. S. (1989) Proc. Natl. Acad. Sci. USA 86. 3919-3923 Farge, E.. & Devaux. P.F. (1992) Biophys. J. 61. 347-357 Farge, E.. & Devaux. P.F. (1993) J Phys. Chem. 97. 2958-2961 Fettiplace. R.. & Haydon. D.A. (1980) Phys Rev. 60. 510-550 Fettiplace. R.. Gordon. L.G.M.. Hladky. S.B.. Requena. J.. Zingslien. H.P., & Haydon, D.A. (1974) in Methods in Membrane Biology. Vol. 4 (Korn. E.D. Ed) New York, pp 1-75 Fidler. I.J.. Jessup. J.M.. Fogler. W.E.. Staerkel. R.. & Mazumder. A. (1986) Cancer Res. 46. 994-998  Fidler. I.J., Raz, A.. Fogler. WE.. Bligelshi. P.. & Poste. G. (1980) Cancer Res. 40. 4460-447 Fiske, C.H.. & Subbarow. Y. (1925) J. Biol. Chem. 66. 375-400 Flewelling. R.F., & Hubbell. W.L. (1986a) Biophys. J. 49. 531-540 Flewelling. R.F.. & Hubbell. W.L. (1986b) Biophys. J. 49. 541-552 Gennis. R.B. (1989) in Biomembranes Molecular Structure and Function. SpringerVerlag, New York Georgatos, A.. Hunter, D.L., Lookman. T.. Zuckerinann. M.J., & Pink. D.A. (1984) Eur. Biophys. J. 11, 79-86 Gregoriadis, G. (1973) FEBS Lett. 36. 292-296 Gregoriadis. G. (1976) N. Engl. J. Med. 295. 704-707 Gregoriadis. G. (1988) in Liposomes as Drug Carriers. Wiley & Sons, Toronto. Ontario pp 3-19 Gregoriadis, G., & Senior. J. (1980) FEBS Lett. 119. 43-46 Gruner, S.M. (1987) in Liposomes: From Biophysics to Therapeutics (Ostro, M.J. Ed.) Marcel Dekker. New York pp 1-38 Gruner, S.M., Lenk, R.P.. Janoff. A.S.. & Ostro. M.J. (1985) Biochemistry 26, 54395447 Haines. T.H., Li, W.. Green. M.. & Cummins. H.Z. (1987) 26. 5439-5447 Hampton, R.Y., & Holz. R.W. (1983) J. Cell Biol. 96. 1082-1088 Harrigan. P.R., Wong. K.F., Redelmeier. T.E.. Wheeler. J.J.. & Cullis, P.R. (1993) Biochemistry, in press Hope, M.J., Bally. M.B.. Mayer. L.D.. Janoff. A.S.. & Cullis. P.R. (1986) Chem. Phys. Lipids 40, 89-108 Hope, M.J., Bally, M.B., Webb. G.. & Cullis. P.R. (1985) Biochim. Biophys. Acta. 812, 55-65 Hope, M.J., Nayar, R.. Mayer. L.D., & Cu Ills. P.R. (1992) in Liposome Technology Vol I (Gregoriadis, G. Ed.) pp 123-139 Hope, M.J., Redelmeier, T.E., Wong. KY.. Rodrigtieza. W.. & Cullis. P.R. (1989) Biochemistry 28. 4181-4187 Houslay. M.D. & Stanley K.K. (1982) Dynamics of Biological Membranes, John Wiley & Sons, Toronto Huang, C.H. (1969) Biochemistry 8, 344-352 J.N.. Marcelja, S.. & Horn. R.G. (1980) Q. Rev. Biophys. 13, 121-200 Israelachvili, J.N., Mitchell, D.J.. & Ninhain. B.W. (1977) Biochim. Biophys. Acta 470, 185-201 Jain, M.K. (1980) in Introduction to Biological Membranes (Jain, M.K.. & Wagner. R.C. Eds.) John Wiley & Sons. Toronto pp 117-142 Johnson. S.M.. & Buttress. N. (1973) Biochim. Biophys. Acta. 307. 20-26 Juliano. R.L. (1983) in Liposomes (Ostro. M.. Ed.) Marcel Dekker. New York pp 53-86 Kachar. B.. & Reese. T.S. (1982) Nature 296. 464-467  Kanellis. P., Romans, AN., Johnson. B.J., Kercret. H.. Chiovetti. R.. Allen. T.M.. & Segrest. J.P. (1980) J. Biol. Chem. 255. 11464-11472 Kas, L.. & Sackmann. E. (1991) Biophys. J. 60. 825-844 Kashchiev, D.. & Exerowa. D. (1983) Biochim. Biophys. Acta 732. 133-145 Kirby. C., & Gregoriadis. G. (1981) Biochein. J. 199. 251-254 Kirby, C., Clarke, J.. & Gregoriadis. G. (1980) Biochein. J. 186. 591-598 Klausner, R.D., Blumenthal. R., Innerarity. T.. & Weinstein. J.N. (1985) J. Biol. Chem. 260, 13719-13727 Koefoed-Johnsen. Valborg. & Ussing. H.H. (1953) Acta Physiol. Scand. 28. 60-76 Kremer. J.M.H., Esker. M.W.J., Pathinamanohovan. C.. & Wiersema. P.H. (1977) Biochemistry 16. 3932-3935 Krupp. L.. Chobanian. A.V.. & Brecher. I.P. (1976) Biochein. Biophys. Res. Commun. 72. 1251-1258 Lafleur, M.. Cullis. P.R.. Fine. B. & Bloom. M. (1990) Biochemistry 19. 8325-8332 Lasic, D.D.. Frederik. P.M.. Stuart. M.C.A.. Barenholz. Y.. & McIntosh. T.J. (1992) FEBS Lett. 312, 255-258 Lee, C., & Chen, L.B. (1988) Cell 554. 37-46 Lelkes, P.I.. & Tandeter. H.B. (1982) Biochim. Biophys. Acta 716, 410-419 Li, W., & Haines. T.H. (1986) Biochemistry 25, 7477-7483 Li, W., Aurora, S.. Haines, T.H.. Cummins. H.Z. (1986) Biochemistry 25. 8220-8229 Lieb, W.R.. & Stein. W.D. (1971) in Current Topics in Membranes and Transport Vol 2 (Bonner. F.. & Kleinzeller. A.) Academic Press. Orlando, Florida pp 1-39 Lipowsky, R. (1992) J. Phys. 11 2, 1825-1840 Lippincott-Schwartz, J., Donaldson. J.G.. Schweizer. A.. Berger, E.G.. Hauri, H.-P.. Yuan, L.C., & Klausner. R.D. (1990) Cell 60. 821-836 Lopez-Berestein, G. (1988) in Liposomes in the Therapy of Infections, Diseases and Cancer (Lopez-Berestein. G.. & Fidler. I.J. Eds.) Alan R. Liss. Inc., New York PP 317-321 Madden. T.D. (1986) Chem. Phys. Lipids 40. 207-226 Madden. T.D.. Harrigan, P.R., Tai. L.C.L.. Bally. M.B.. Mayer. L.D.. Redelmeier, T.E.. Loughrey. H.C.. Tilcock. C.P.S.. Reinish. & Cullis, P.R. (1990) Chem. Phys. Lipids 53. 37-46 Malinski. J.A.. & Nelsestuen. G.L. (1989) Biochemistry 28. 61-70 Marsh, D. (1990) in CRC Handbook of Lipid Bilayers CRC Press, Boca Raton. Rlorida Martinac. B., Buechner. M.. Delcour. A.H.. Adler. J.. & Kung. C. (1987) Proc. Natl. Acad. Sci. USA 84. 2297-2301 Mayer, L.D.. Hope. M.J., & Cullis. P.R. (1986) Biochim. Biophys. Acta. 858, 161-168 Mayer, L.D.. Hope. M.J., Cullis, P.R.. & Janoff. A.S. (1985) Biochim. Biophys. Acta 817, 193-196 Mayer, L.D.. Madden. T.D.. Bally, & Cullis. P.R. (1992) in Liposome Technology Vol. II (Gregoriadis. G. Ed.) pp 27-44  Mayer, L.M., Bally. M.B.. & CtiHis, P.R. (1990) J. Liposome Res. 1. 463-480 McElhaney, R.N. (1986) Biochein. Cell Biol. 64. 58-65 Miyamoto. S., Maeda. T.. & Fujime. S. (1988) Biophys. J. 53. 505-512 Morgan. J.R.. Williams. L.A.. & Howard. C.B. (1985) Br. J. Radio!. 58, 35-39 Morris. C.E. (1990) J. Membrane Biol. 113. 93-107 Morrisett. J.D.. Jackson. R.L., & Gotto. A.M. (1977) Biochim. Biophys. Acta 472, 93133 Mui, B. L.-S., Cullis, P.R., Evans, E.A.. & Madden. T.D. (1993) Biophys. J. 64. 443-453 Muller-Eberhard, H.J. (1986) Annu. Rev. linnitinol. 4. 503-528 Needham. D.. & Evans. E.A. (1988) Biophys. J. 27. 8261-8269 Needham. D.. & Nunn. R.S. (1990) Biophys. J. 58. 997-1009 Ogihara. I.. Kojimi. S.. & Jay. M. (1986) Eur. J. Nucl. Med. 11. 405-411 Op den Kamp. J.A.F. (1979) Ann. Rev. Biochein 48. 423-475 Oster, G.F.. Cheng. L.Y.. Moore. H.-P.H. & Perelson. A.S. (1989) 141, 463-504 Ostro, M.J., & Cullis. P.R. (1989) Am. J. Hosp. Marin. 46. 1576-1587 Papahadjopoulos. D., Moscarello. M.. Eyler. E.M.. & Isaac. T. (1975) Biochim. Biophys. Acta 401. 317-335 Parsegian. V.A. (1969) Nature 221. 844-846 Pollard, H.B., Pazoles, C.J., Creutz. C.E.. Scott. J.H.. Zinder. 0., & Hotchkiss (1984) J. Biol. Chem. 259. 1114-1121 Pownall, H.J., Massey, J.B., Kusserow. S.K.. & Gotto. A.M. (1979) Biochemistry 18, 574-579 Prescott. D.M., & Zeuthen, E. (1953) Acta Physiol. Scancl. 28. 77-94 Rahman. A., Carmichael. D.. Harris. M.. & Rob. J.K. (1986) Cancer Res. 46, 22952299 Rahman, A., Kessler, A., More. N.. Sikic. B.. Rowden, E.. Woolley, P., & Schein, P.S. (1980) Cancer Res. 40. 1532-1537 Rand, R.P.. & Burton. A.C. (1964) Biophys. J. 4. 115-135 Reeves, J.P.. & Dowben. R.M. (1969) Cell. Physiol. 73. 49-60 Rutkowski. C.A.. Williams. L.M.. Haines. T.H.. & Cummins. H.Z. (1991) Biochemistry 30. 5688-5696 Sackmann. E. (1984) in Biological Membranes Vol. 5 (Chapman. D. Ed.) pp 105-143 Sackmann. E., Duwe. H.P.. & Engellhardt. H. (1986) Faraday Discuss. Chem. Soc. 81, 281-290 Scherphof. G.. & Morselt. H. (1984) Biochem. J. 221. 423-429 Scherphof. G., Damen. J.. & Wilshut. J. (1984) in Liposome Technology, Vol. /// (Gregoriadis. G. Ed.) CRC Press Inc.. Boca Raton. Florida pp 205-224 Scherphof. G.. Roerdink, F.. Waite. M.. & Parks. J. (1978) Biochim. Biophys. Acta 542. 296-307  Schroeder, F.. Jefferson, J.R.. Kier. A.B.. Knittel, J.. Scallen. T.J.. Wood. W.G.. & Hapala. I. (1991) Proc. Soc. Exp. Biol. Med. 196, 235-252 Schroit, A.J., Madsen. J. & Tanaka. Y. (1985) J. Biol. Chem. 260. 5131-5138 Seifert. U. (1992) Phys. Rev. Lett. 70. 1335-1338 Seigel, D.P., Burns. J.L.. Chestnut. M.-I.. & Talmon. Y. (1989) Biophys. J. 56, 161-169 Sheetz, M.P.. & Singer. S.J. (1974) Proc. Nat. Acad. Sci. USA 71, 4457-4461 Shin, M.L., Paznekas. W.A.. & Mayer. M.M. (1978) J. linintinol. 120. 1996-2002 Siegel, D.P. (1986) in Membrane Fusion (Sowers. A.E. Ed.) Plenum Press. New York pp 181-207 Silvitts, J.R. (1982) in Lipid-Protein Interactions Vol. 2 (Jost. P.C., & Griffith. O.H. Eds.) Wiley. New York pp239-281 Small. D.M. (1985) in The Physical Chemistry of Lipids from Alkanes to Phospholipids, Plenum Press. New York Svetina, S.. & Zeks. B. (1983) Biomed. Biochint. Acta 42, 86-90 Svetina. S.. & Zeks, B. (1989) Eur. Biophys. J. 17. 101-111 Szoka. F. & Papahadjopoulos. D. (1980) Ann. Rev. Biophys. Bioeng. 9, 467-508 Szoka, F.. & Papahadjopoulos, D. (1978) Proc. Natl. Acad. Sci. 79. 4194-4198 Tall, A.R.. & Green. P.H.R. (1981) J. Biol. Chem. 256. 2035-2044 Talmon, Y.. Burns, J.L., Chestnut. M.-I., & Siegel. D.P. (1990) J. Electron Microsc. Tech. 14, 6-12 Tanaka. Y. & Schroit. A.J. (1983) J. Biol. Chem. 258. 11335-11343 Tanford, C. (1973) The Hydrophobic Effect John Wiley & Sons, New York Taupin, C., Dvolaitzky, M.. & Sauterey. C. (1975) Biochemistry 14. 4771-4775 Tocanne, J.-F.. & Teissie. J. (1990) Biochim. Biophys. Acta 1031, 111-142 Tschopp, J. (1984) J. Biol. Chem. 259. 7857-7863 Ubl, J., Murer. H.. & Kolb. H.-A. (1988) J. Membrane Biol. 104, 223-232 van Deenen, L.L.M., & de Gier. J. (1974) in The Red Blood Cell (Surgenor. D. Ed.) Academic Press. New York pp 147-213 Walter. A., & Gutknecht, J. (1984) J. Memb. Biol. 77. 255-264 Watts, A., Harlos. K.. Maschke. W.. & Marsh. D. (1978) Biochim. Biophys. Acta 510. 63-74 Waugh, R.E. (1982) Biophys J. 39, 273-278 Weinstein. J.N.. Klausner, R.D.. Innerarity. T.. Ralston. E.. & Blumenthal. R. (1981) Biochim. Biophys. Acta 647. 270-280 Weinstein, J.N., Ralston. E.. Leserman. L.D.. Klausner. R.D.. Dragsten, P.. Henkart. P.. & Blumenthal. R. (1984) in CRC Liposome Technology Vol. 3 (Gregoriadis. G. Ed.) CRC Press, Inc.. Boca Raton. Florida pp 183-204 Whaley. K. (1985) in Methods in Complementfor Clinical Immunologists (Whaley. K. Ed.) Churchill Livingstone. Edinburgh. New York. pp 77-139 Wicker. W. (1979) Ann. Rev. Biochem 48, 23-45  Wieland, F.T., Gleason, M.L., Serafini, ^& Rothman. WE. (1987) Cell 50, 289-300 Wills, G.L., Lane, P.A., & Weech, P.K. (1984) in A Guidebook to Lipoprotein Technique: Laboratory Techniques in Biochemistry and Molecular Biology Vol. 14 (Burdon, R.H., & van Krippenberg, P.H. Eds.) Elsevier, Amsterdam p 18 Wortis, M., Seifert, U., Berndl, K., Fourcade, B., Miao, L., Rao, M., & Zia, R.K.P. (1991) in Dynamical Phenomena at Interfaces, Surfaces and Membranes (???? Yang, X.-C., & Sachs, F. (1989) Science 243, 1068-1071 Yoshioka, S., Ishibashi, M., Shibazaki, T., & Uchiyama, M. (1984) J. Parent. Sci. & Technol. 38, 222-227 Zeman, K., Engelhard. H.. & Sackmann. E. (1990) Eur. Biophys. J. 18, 203-219  - 130-  

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.831.1-0098816/manifest

Comment

Related Items