Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Transmembrane pH gradients in liposomes: drug-vesicle interactions and proton flux Harrigan, Paul Richard 1992

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

Item Metadata


831-ubc_1992_fall_harrigan_richard_p.pdf [ 2.62MB ]
JSON: 831-1.0086538.json
JSON-LD: 831-1.0086538-ld.json
RDF/XML (Pretty): 831-1.0086538-rdf.xml
RDF/JSON: 831-1.0086538-rdf.json
Turtle: 831-1.0086538-turtle.txt
N-Triples: 831-1.0086538-rdf-ntriples.txt
Original Record: 831-1.0086538-source.json
Full Text

Full Text

TRANSMEMBRANE pH GRADIENTS IN LIPOSOMES: DRUG-VESICLE INTERACTIONS AND PROTON FLUX by P. RICHARD HARRIGAN B.Sc.(Hons) University of British Columbia, 1985 M.Sc. University of British Columbia, 1987 A THESIS SUBMITI’ED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES Department of Biochemistry We accept this thesis as conforming to the required standard  Signature(s) removed to protect privacy  THE  OF BRITISH COLUMBIA April 1992 © P.R. Harrigan, 1992  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(s) removed to protect privacy  (Signature  Department of  (51  oc/ ’ 1 eM (  The University of British Columbia Vancouver, Canada  Date  DE-6 (2/88)  AM1--(  /s /9L\  5’74  ABSTRACT This thesis examines the properties of large unilamellar lipid vesicles (LUVs) having transmembrane pH gradients (ApH). These pH gradients Induce proton movement across the membrane and can result in accumulation of amines into the vesicles. A major focus of the thesis is to develop a quantitative approach to describe the response of drugs which are lipophilic amines, as well as protons, to these pH gradients. Large, stable transmembrane pH gradients (ApH) of up to 3.5 units are shown to be detectable in LUVs by examining the transmembrane distribution of 14 This approach is subject to artefacts in situations where the [ ]C-methylamine. interior buffering capacity is low, where the interior vesicle volume changes due to osmotic effects, or where probe redistribution is too slow to be practical. These problems are generally easily overcome in liposomal systems. It is demonstrated that transmembrane pH gradients provide a practical method of entrapping drugs in lipid vesicles. It is shown that the anticancer drug doxorubicin accumulates into LUVs with an acidic interior via permeation of the neutral form of the drug. The critical dependence of translocation rates on pH, temperature and lipid composition allow manipulation of drug loading and release to achieve desired characteristics. A model incorporating vesicle volume, buffering capacity, drug partitioning and other factors is shown to describe the accumulation of doxorubicin in response to a zpH. This results in the conclusion that more than 95% of the encapsulated doxorubicin is partitioned into the inner monolayer for a 100 nm vesicle. Representative basic drugs from different drug classes also accumulate into vesicles in response to an acidic interior, although the extent of uptake varies considerably. Finally, the unique nature of the transbilayer movement of protons (or equivalents) was examined in well buffered lipid systems with large (3 unit) pH gradients. Development of a transmembrane electrical potential had a half-time of about 12 mm  in EPC LUVs at 25°C, with an activation energy near 11 kcal/mol,  11  near the activation energy of water transport. Further, model membrane systems were developed which exhibited stable membrane potentials without induced pH gradients. or stable pH gradients without induced membrane potentials.  in  TABLE OF CONTENTS  ABSTRACT  Ii  TABLE OF CONTENTS  iv  LIST OFTABLES  ix  LIST OF FIGURES ABBREVIATIONS USED  .xii  ACKNOWLEDGEMENTS  xiv  CHAPTER 1  INTRODUCTION  1.1 BACKGROUND  1  1.2 CHEMICAL AND MATERIALS PROPERTIES OF LIPID PREPARATIONS  3  1.2.1 STRUCTURE OF PHOSPHOLIPIDS AND CHOLESTEROL.  .  .  .  3  1.2.2 MATERIALS PROPERTIES OF LIPOSOMES  5  1.2.3 LIPID PHASES  5  1.2.4 EFFECTS OF CHOLESTEROL  6  1.3 PRODUCTION AND USE OF LIPOSOMES  6  1.3.1 MULTILAMELLARVESICLES (MLVs)  7  1.3.2 SMALL UNILAMELLAR VESICLES (SUVs)  9  1.3.3 LARGE UNILAMELLAR VESICLES (LLJVs)  10  1.3.4 OSMOTIC PROPERTIES OF EXTRUDED LIPOSOMES  12  1.3.5 TRAPPED SOLUTES AND TRAPPED VOLUME DETERMINATIONS  12  1.3.6 SOME RELEVANT PARAMETERS CONCERNING LIPOSOMES  13  1.4 SOLUTE PARTITIONING AND PERMEABILITY  14  1.4.1 PARTITION COEFFICIENTS iv  14  1.4.2 PERMEABILITY OF NON-ELECTROLYTES  16  1.4.3 PERMEABILITY OF WATER  19  1.4.4 PERMEABILITY OF IONS  20  1.4.5 PERMEABILITY OF PROTONS  22  1.5 TRANSMEMBRANE ION GRADIENTS  25  1.5.1 PANDzpH  25  1.5.2 MEASUREMENT OF ApH AND t’I’ IN LIPID VESICLES  26  1.5.3 SIGNIFICANCE OF ACID-BASE CHARACTERISTICS OF COMPOUNDS  28  1.6 DRUG TRAPPING IN LIPOSOMES  30  1.6.1 PASSIVE ENTRAPMENT TECHNIQUES  30  1.6.2 ACTIVE TRAPPING TECHNIQUES  31 32  1.7 THESIS OUTLINE  CHAPTER 2. MEASUREMENTS OF TRANSMEMBRANE pH GRADIENTS IN LUVS 2.1 INTRODUCTION  33  2.2 MATERIALS AND METHODS  34  2.2.1 MATERIALS  34  2.2.2 LIPID VESICLE PREPARATION  35  2.2.3 PARTICLE SIZE DETERMINATIONS  35  2.2.4 DETERMINATION OF ENTRAPPED SOLUTE  35  2.2.5 EFFECT OF INITIAL SOLUTE DISTRIBUTION ON ENTRAPPED SOLUTE  36  2.2.6 VESICLE VOLUME DETERMINATIONS  36  2.2.7 DETERMINATION OF LIPID CONCENTRATIONS  37  2.2.8 GENERATION AND MEASUREMENT OF TRANSMEMBRANE ION GRADIENTS  38  2.3 THEORETICAL CONSIDERATIONS  39  2.4 RESULTS  41 V  2.4.1 VESICLE CHARACTERISTICS  .  41 45  2.4.2 MEASUREMENTS OF EpH  60  2.5 DISCUSSION  CHAPTER 3. DRUG UPTAKE INTO LIPOSOMES IN RESPONSE TO pH GRADIENTS 3.1 INTRODUCTION  64  3.2 MATERIALS AND METHODS  66  3.2.1 MATERIALS  66  3.2.2 LIPID VESICLE PREPARATION  66  3.2.3 DETERMINATION OF DOXORUBICIN UPTAKE LEVELS..  .  .66  3.2.4 DOXORUBICIN FLUORESCENCE STUDIES  67  3.2.5 C]-NMR 13 STUDIES [  67  3.2.6 CRYO-ELECTRON MICROSCOPY  67  3.2.7 DRUG UPTAKE “SURVEY” EXPERIMENTS  68  3.2.8 OTHER ANALYTICAL PROCEDURES  68  3.2.9 KINETIC ANALYSIS  69  3.2.10 EQUILIBRIUM ANALYSIS  72 74  3.3 RESULTS 3.3.1 KINETICS OF DOXORUBICIN UPTAKE DETERMINED BY FLUOROMETRIC TECHNIQUES  74  3.3.2 pH DEPENDENCE AND ACTIVATION ENERGY OF DOXORUBICIN ACCUMULATION INTO LUVs  76  3.3.3 PARTITION COEFFICIENTS AND COUPLING CHARACTERISTICS ASSOCIATED WITH DOXORUBICIN UPTAKE  76  3.3.4 C1 13 NMR STUDIES ON DOXORUBICIN UPTAKE [  78  3.3.5 MORPHOLOGICAL FEATURES OF LUVs FOLLOWING DOXORUBICIN ACCUMULATION  81  3.3.6 DRUG UPTAKE “SURVEY”  83  3.3.7 CLASS 1. DRUGS WHICH EXHIBIT PARTIAL BUT STABLE vi  .83  UPTAKE  3.3.8 CLASS 2. DRUGS WHICH EXHIBIT PARTIAL UPTAKE AND 87 SUBSEQUENT RELEASE 3.3.9 CLASS 3. DRUGS WHICH EXHIBIT NO APPARENT RESPONSE  88  3.3.10 CLASS 4. DRUGS WHICH ARE TOTALLY ACCUMULATED. .88 3.3.11 PARTITION COEFFICIENTS AND COUPLING OF OTHER AMINES  88 91  3.4 DISCUSSION  3.4.1 DOXORUBICIN ENTRAPMENT IN RESPONSE TO A zpH... .91 3.4.2 ApH ACCUMULATION OF OTHER COMPOUNDS  94  CHAPTER 4. RATES AND ACTIVATION ENERGIES OF PROTON FLUX ACROSS LIPID BILAYERS 4.1 INTRODUCTION  96  4.2 MATERIALS AND METHODS  98  4.2.1 PREPARATION OF LIPID SAMPLES  98  4.2.2 GENERATION OF pH GRADIENTS  98  4.2.3 GENERATION OF MEMBRANE POTENTIALS  99  4.2.4 DETERMINATION OF PROBE UPTAKE INTO LUVs  99  4.2.5 CALCULATION OF MEMBRANE PQTENTIALS AND pH GRADIENTS  99  4.3 THEORETICAL CONSIDERATIONS  100  4.4 RESULTS  101  4.4.1 DETECTION OF PROTON FLUX  101  4.4.2 INFLUENCE OF LIPID COMPOSITION ON PROTON FLUX  103  4.4.3 ACTIVATION ENERGIES ASSOCIATED WITH PROTON FLUX  103  4.4.4 DETECTION OF PROTON FLUX IN RESPONSE TO MEMBRANE POTENTIALS  110  vii  4.4.5 LUVs EXHIBITING A 1xP OR pH, BUT NOT BOTH  111  4.5 DISCUSSION  CHAPTER 5.  110  SUMMARIZING DISCUSSION  114  117  REFERENCES  viii  LIST OF TABLES  1-I NAMES AND STRUCTURES OF SOME COMMON FATTY ACIDS  4  1-11 RELEVANT PARAMETERS CONCERNING UNILAMELLAR LIPOSOMES. .14 2-I. [ C1-GLUCOSE 1 4 AND C1-CITRATE 14 VOLUMES OF 100 NM EXTRUDED [ VESICLES 43 3-I. EXTENT AND STABILITY OF ACCUMULATION OF VARIOUS DRUGS BY VESICLES EXHIBITING A pH GRADIENT (ACIDIC INTERIOR). .84 ..  3-IT PARTITION COEFFICIENTS OF DRUGS EXAMINED  90  4-I. PROTON FLUX RATES AND ACTIVATION ENERGIES  106  ix  LIST OF FIGURES  1-1. ELECTRON MICROGRAPHS OF MLV AND FATMLV VESICLE PREPARATIONS  8  1-2. ELECTRON MICROGRAPHS OF EXTRUDED VESICLES OF DIFFERENT SIZE  11  1-3. SOLUBILITY-DIFFUSION MODEL OF MEMBRANE PERMEABILITY.... 18 1-4 REPRESENTATION OF PROTONS CROSSING LIPID BILAYERS BY A CARRIER MECHANISM  24  1-5. EFFECT OF TRANSMEMBRANE pH GRADIENTS ON THE EQUILIBRIUM DISTRIBUTION OF WEAK BASES  29  2-1 ENTRAPMENT OF 300 mM CITRATE BUFFER IN FATMLVs AS A FUNCTION OF FREEZING TIMES  42  2-2. DEPENDENCE OF SOLUTE TRAPPING ON INITIAL SOLUTE DISTRIBUTION  44  2-3 EFFECT OF LIPID COMPOSITION ON METHYLAMINE DISTRIBUTIONS. .47 2-4. METHYLAMINE RESPONSE IN VESICLES WITH DIFFERENT BUFFERING CAPACITIES AND OSMOTIC STRENGTH  49  2-5. EFFECT OF EXTERNAL METHYLAMINE CONCENTRATION ON METHYLAMINE UPTAKE  52  2-6. DETERMINATION OF pH GRADIENTS OVER A RANGE OF pH  53  2-7. MEASUREMENT OF LARGE pH GRADIENTS  55  2-8. TRANSMEMBRANE DISTRIBUTIONS OF RADIOLABELLED PROBES IN EPC LUVs FOR VESICLES WITH A BASIC INTERIOR  56  2-9. DETERMINATION OF pH GRADIENTS (INTERIOR BASIC) EMPLOYING 58  14 [ C]-ACETATE 2-10. THE RELATIONSHIP BETWEEN TRANSMEMBRANE TPP GRADIENTS AND MeNH 3 GRADIENTS  59  3-1. STRUCTURE OF DOXORUBICIN  65  3-2 MODEL OF INTERACTIONS OF DOXORUBICIN WITH LUVs IN THE x  PRESENCE OFAApH  71  .  3-3 DOXORUBICIN ACCUMULATION INTO LUVs IN RESPONSE TO TRANSMEMBRANE pH GRADIENTS  75  3-4. EFFECTS OF EXTERNAL pH AND TEMPERATURE ON KINETICS OF DOXORUBICIN ACCUMULATION  77  3-5. RELATIONSHIP BETWEEN RESIDUAL pH GRADIENT AND DOXORUBICIN ACCUMULATION  79  3-6. EFFECT OF TRANSMEMBRANE pH GRADIENTS ON THE C]-NMR 13 [ SPECTRA OF VESICLES INCUBATED WITH DOXORUBICIN  80  3-7. CRYOELECTRON MICROSCOPY OF DOXORUBICIN LOADED AND DOXORUBICIN FREE VESICLES  82  3-8. UPTAKE OF TIMOLOL BY EPC VESICLES  85  3-9. UPTAKE OF QUINIDINE BY EPC AND EPC CHOLESTEROL VESICLES  86  3-10. RELATIONSHIP BETWEEN RESIDUAL pH GRADIENT AND DRUG ACCUMULATION IN VESICLES WITH A TRANSMEMBRANE pH GRADIENT  89  4-1. GENERATION OF MEMBRANE POTENTIALS IN RESPONSE TO 3 UNIT ACIDIC OR BASIC PH GRADIENTS  102  4-2. COMPARISON OF THE EFFECTS OF CHOLESTEROL ON PROTON AND WATER FLUX  104  4-3. COMPARISON OF THE EFFECTS OF TEMPERATURE ON PROTON AND WATER FLUX  105  4-4. GENERATION AND STABILITY OF pH GRADIENTS IN RESPONSE TO TRANSMEMBRANE K GRADIENTS  108  4-5. VESICLES WITH STABLE MEMBRANE POTENTIALS BUT NO pH GRADIENTS AND VICE-VERSA  Xl  109  ABBREVIATIONS USED  ApH  Transmembrane pH gradient Transmembrane electrical potential  chol  Cholesterol  CHES  2-( N-cyclohexylamino )ethane-sulfonic acid  CCCP  Carbonyl cyanide m-chlorophenylhydrazone  DAPC  1 ,2-diarichinoyl-sn-glycero-3-phosphorylcholine  DMPC  1 ,2-dimyristoyl-sri-glycero-3-phosphorylcholine  DPPC  1 ,2-dipalmitoyl-sn-glycero-3-phosphorylcholine  DSPC  1 ,2-distearoyl-sri-glycero.-3 -phosphoryicholine  Ea  Activation energy  EPC  egg PC (from hen egg yolk)  EPPS  N-(2- Hydroxyethyl)piperazme- N’-3propanesulfonic acid  ESR  Electron spin resonance  FATMLVs  Frozen and thawed MLVs  3 HOC1O  Perchloric acid  HEPES  I 4-( 2-Hydoxyethyl)]-piperazlne ethanesulfonic acid  LUVs  Large unilamellar vesicles  3 MeNH  Methylamine  MES  2-(N-Morpholmo)ethanesulfonic acid  MLVs  Multilamellar vesicles  3 NaHSO  Sodium bisuffite  4 S 2 Na O  Sodium sulfate  NM R  Nuclear magnetic resonance  PC  Phosphatidyicholine  PL  Phospholipid  PLM  Planar lipid membrane  ppm  parts per million xii  psi  Pounds per square inch  QELS  Quasi-elastic light scattering  SA  Stearylamine  SCN  Thiocyanate  sn  stereospecific nomenclature  STEM  Scanning tunneling electron microscope  SUVs  Small unilamellar vesicles  TNBS  Trinitrobenzenesulfonic acid  TMS  Tetramethylsilanie  TPP  Tetraphenylphosphonium  TPB  Tetraphenylboron  Tris  2-Amino-2( hydroxymethyl)propane- 1,3 diol -  Gel to liquid-crystalline transition temperature 2 / 1 t  Half-time for transport  UV  Ultraviolet  xli’  ACKNOWLEDGEMENTS A large thank you must go to the many members of the Cullis lab (you know who you are), and to Pieter both for keeping me in food and clothing and for a display of super-human patience in the production of this thesis.  Also, I would like to thank God that I’m finally finished.  xiv  CHAPTER 1  INTRODUCTION  1.1 BACKGROUND  One of the earliest insights into the biochemical nature of membranes was the observation of Overton (1899) that the permeability of biological membranes to a variety of molecules correlated with the solubiity of the molecule in olive oil. This implied that the cellular permeability barrier was lipidic. After it was established that red blood cells contained enough lipid to cover the surface area of these cells twice (Gorter and Grendel, 1925). Danielli and Davson made the suggestion that the lipid could be organized as bimolecular leaflets, with the protein portion of biological membranes spread in monolayers on each side of the bilayer (Danielli and Davson, 1935; Robertson, 1957). Subsequent demonstrations of the ability of lipids and proteins to move within the bilayer and the existence of many proteins with transmembrane a-helices led to the ‘fluid mosaic” model of biological membranes (Singer and Nicholson, 1972). In this model, peripheral and integral proteins are supported in the fluid matrix of a lipid bilayer and proteins and lipids are able to diffuse laterally around the cell. The common picture of an inert lipid bilayer containing lumps of protein is, of course, oversimplified (see Bloom et aL. 1991). For example, the two faces of the bilayer are not equal. Lipids and proteins are generally asymmetrically distributed or oriented arid chemical modifications such as glycosylation are usually confined to one side of the membrane (Storch and Kleinfeld, 1985; Kleinfeld, 1987). Ionic composition, pH and electrical potential can also differ between the two sides (Kleinfeld, 1987). Other complexities of biological membranes not addressed in this model include the wide differences in the lipid composition between organisms, tissues and cells and organelles (for example, Op den Kamp, 1979; Schroeder et at 1991), and the astonishing variety of lipids and proteins found within even a single membrane (see Storch and Kleinfeld, 1985, van Deenen et al., 1974). 1  The inherent complexity of biological membranes has led many investigators to examine simpler model systems consisting of only a small number of defined chemical species. As mentioned above, the first model of the lipid portion of the membrane was olive oil. Later developments included planar lipid membranes consisting of a lipid bilayer in an organic solvent which separates two accessible compartments (see Muellar, 1962). Another model membrane system was established by Bangham (1965), based on the observation that when many phospholipids are hydrated they spontaneously form sealed systems (“liposomes”), which consist of lipid bilayers separated by water layers in an ‘onion-skin” arrangement. Simpler unilamellar systems can also be prepared. These systems are widely used to model the lipid portion of biological membranes. These lipid vesicles have also turned out to be useful as drug delivery vehicles. One goal in the design of liposomal delivery systems is to deliver drug specifically to diseased tissues, using targeted lipid-based systems. While such systems are not yet achievable, the therapeutic properties of many drugs can be improved by administration in an encapsulated form (for example, see Gregoriadis, 1976). Liposomes have several desirable properties, including a great flexibility with respect to size and biodistribution properties, biodegradability and a relative lack of toxicity, (Weinstein, 1984).  The therapeutic benefit likely results from the  altered pharmacokinetics and biodistribution of the entrapped drug (Ostro and Cullis, 1989; Mayer et al., 1990; Rahman et al., 1980) due to the removal of liposomes (containing the drug) from the circulation by the organs of the reticuloendothelial system. In particular, the liver and spleen accumulate much of a typical intravenously injected liposome preparation, resulting in a passive targeting to these organs (Gregoriadis. 1988). There also appears to be preferential accumulation of “untargeted liposomes” in tumours and at sites of inflammation (Ogihara et al., 1986; Morgan et al., 1985). There are several other applications of liposomal encapsulation (Weinstein, 1984), including extending the duration of drug action, for example, by reducing diffusion of local anaesthetics from the site of 2  interest (Gregoriadis, 1988). A partial list of liposomal drugs presently undergoing human clinical trials includes liposomal anti-fungal agents (Lopez-Berestein, 1988), liposomal anticancer agents (Rahman et al., 1986; Creaven et al., 1990) and liposomal immunomodulators (Sone et al., 1980). In the case of amphotericin B (an anti fungal agent) and doxorubicin (an anti-cancer agent), drug toxicity is reduced while efficacy is maintained or increased (Ostro and Cullis, 1989). A major limitation to the practical application of liposomes for drug delivery is the difficulty in entrapping sufficient material, due to the submicron diameters of these liposomes. One approach which can increase the efficiency of drug entrapment is to employ transmembrane pH gradients (ApH) across the bilayer. For example, weakly basic drugs can accumulate in vesicles with an acidic interior. The generation, measurement and stability of the ipH in liposomes, as well as the accumulation of some common pharmaceuticals into vesicles in response to a ApH are examined in this thesis. In addition, the transport of proton equivalents in response to these pH gradients is investigated. 1.2 CHEMICAL AND MATERIALS PROPERTIES OF LIPID PREPARATIONS  The characteristics of lipid systems can be conveniently divided into their chemical properties, which describe the characteristics of individual molecules, and their materials properties, which depend on the cooperative interactions of large numbers of molecules (Gruner, 1987). The properties under investigation in this thesis are generally the materials properties of mixtures of phospholipids, sterols (where present) and water rather than their respective chemical properties, though some of the chemical properties of these compounds should be considered, as indicated below. 1.2.1 STRUCTURE OF PHOSPHOLIPIDS AND CHOLESTEROL  This section will deal only with the lipids used in this thesis, which are mainly phosphatidyicholine (PC) and cholesterol. The chemical properties of other lipids 3  have been extensively reviewed elsewhere (Small. 1986). Phosphatidyicholine is the most common phospholipid in eukaryotic plasma membranes, a zwitterion composed of a glycerol-phosphate ester with a choline headgroup and two acyl chains esterified to the sn-i and sn-2 positions (Small, 1986). It is generally drawn as a circle (representing the headgroup) with two wavy lines (representing the acyl chains). Table i-I indicates the more common fatty acids found in eukaryotic PCs. In general, the sn-i position tends to be a saturated fatty acid, while the sn-2 tends to be unsaturated (Small. 1986). For example, the fatty acid side chains of egg derived PC (EPC) are 16:0(32% mol %), 18:0(15%); 18:1 (3 1%) and 18:2 (16%) (Blok at aL, 1974a), where the number preceding the colon refers to the number of carbons of the fatty acid and the number following it refers to the number of double bonds. Other fatty acids are present in EPC in trace amounts (Blok at aL, 1974a). it is striking that there is an extremely large number of structural possibilities even for a single phospholipid such as PC. Table 1-I Names and structures of some common fatty acids.  No of Carbon Atoms  Structural Formula  Name  Saturated fatty acids 12 14 16 18 20  0 CH C0 ( 3 CH 1 ) H 2 2 (CH 3 CH 1 ) H 2 C0 4 (CH 3 CH 1 ) H 2 C0 6 (CH 3 CH 1 ) H 2 C0 8 (CH 3 CH 1 ) H 2 C0  lauric acid myristic acid palmitic acid stearic acid arachidic acid  Unsaturated fatty acids 16 18 18 18 20  22 CH=CH(CH 2 5 ) CH CH ( 3 C0 H 7 ) O CH ( 3 CH C C 7 ) H 2 H=CH(CH O CH 3 CH ( 4 ( C 6 ) H 2 CH=CHCH 0 H CH C CH ( ( 3 C 5 ) H 2 CH=CHCH 2 )4( CH=CHCH 2 )4( 2 C0 ) CH H ( CH 3 CH  4  palmitoleic acid oleic acid linoleic acid linolenic acid arachidonic acid  Cholesterol is the major neutral lipid component of eukaryotic plasma membranes. Part of the molecule has a rigid steroid structure, with amphipathic characteristics due to the 3-f3-hydroxyl group on one end of the molecule. 1.2.2 MATERIALS PROPERTIES OF LIPOSOMES  Perhaps the most important materials property of phospholipid-water mixtures is the tendency of the lipid to form bilayers. This property is a result of the “hydrophobic effect” (Tanford, 1980), wherein the ordering of water molecules which would have to occur if the lipid acyl chains were exposed to the water prevents this exposure, driving the formation of micelles, bilayers, or other structures. An important point to note is that materials properties can be determined by the methods used to produce the final structures, so it is important to define the conditions used, and to emphasize the distinctions between apparently similar systems. For example, different preparations of the same lipid molecules often have radically different drug trapping properties, while preparations with different chemical compositions can behave similarly. 1.2.3 LIPID PHASES The organization of lipid molecules is described in terms of the lipid phase. The gel-liquid crystalline phase transition in the presence of excess water is one of the best characterized properties of phospholipids (Silvius, 1982; Marsh, 1991). Upon heating, aqueous dispersions of a saturated phospholipid such as DPPC can undergo transitions from a highly ordered Lf3 phase (gel-like), through a “ripple” phase, to a La liquid-crystalline phase, characterized by considerably less order in the hydrocarbon region (Chapman, 1975; Silvius, 1982). The gel to liquid crystalline transition is characterized by the transition temperature  generally  measured by differential scanning calorimetry (DSC) (Chapman, 1967; Chapman, 1975). T is influenced by the degree of hydration, lipid head group and acyl chain composition, the presence of cholesterol and a variety of other factors. As an example. the T of DPPC is 4 1 C (Silvius, 1982). Increased acyl chain length 0 5  increases T. DSPC, which has two more carbons per acyl chain than DPPC, has a phase transition temperature of 58° (Small, 1986). Unsaturation in the acyl chains reduces the transition temperature, to below 0°C for EPC, with an average of about one unsaturation per acyl chain (Chapman, 1975). Since living organisms are generally at a constant temperature, pressure and hydration state, it is unlikely that these phase changes have direct physiological relevance (Bloom et al., 1991). Other lipid phases which are perhaps less well known are the non-bilayer phases. For example, an unsaturated phosphatidic acid will adopt a non-bilayer hexagonal H 11 phase when exposed to divalent cations (Cullis, et al., 1983; Tate et at, 1991). These non-bilayer phases have been postulated to have biological significance, especially with regard to membrane fusion (Cullis and de Kruijff, 1979). Most biological membranes contain a significant proportion of non-bilayer forming lipids (Cullis and de Kruijff, 1979). 1.2.4 EFFECTS OF CHOLESTEROL  Cholesterol has varied and subtle effects on membrane properties. Addition of cholesterol to saturated PC progressively decreases the enthalpy of the gel-liquid crystalline phase transition until, at about 30 mol % cholesterol or higher. the transition can no longer be detected by DSC (Chapman, 1975). Cholesterol increases the order of the acyl chains of PCs which are above their phase transition temperature and decreases order for PCs which would otherwise be in the gel state (Gennis, 1989). The permeability of bilayers containing cholesterol generally decreases as acyl chain order increases (Jam, 1980). Cholesterol can also affect the bilayer to non-bilayer phase preferences of phospholipids (Bally et at., 1983). Finally, cholesterol can have a “condensing” effect on PC, such that the volume of a mixture of PC and cholesterol is less than the volume of the two components separately (Hyslop et at., 1990; Demel et at. 1968). 1.3 PRODUCTION AND USE OF LIPOSOMES  Lipid bilayers which have formed seated structures able to encapsulate water 6  soluble material are known as liposomes. There are basically three liposome types: multilamellar vesicles (MLVs); small unilamellar vesicles (SUVs); and large unilamellar vesicles (LUVs ) (see Hope et al.. 1986; Szoka and Papahadjopoulos, 1980). Both the vesicle type and the method of preparation influence the properties of the resulting systems. 1.3.1 MULTILAMELLAR VESICLES (MLVs)  MLVs spontaneously form when phospholipids are added to water (Bangham et al., 1965; Bangham, 1983). MLVs are very heterogeneous, with a wide distribution in both size and number of bilayers. They consist of concentric lipid bilayers and water layers in an onion-skin configuration and are typically greater than 1000 nm in diameter. MLVs have internal aqueous volumes of 2-3 L/mole phospholipid, but, because they have an unequal distribution of solute across their lamellae (Gruner et al., 1985; Perkins et al., 1988) the amount of solute entrapped is much lower than expected, corresponding to apparent internal volumes near 1 L per mol phospholipid. MLVs which exhibit higher solute entrapment may be prepared by ether evaporation from an ether-buffer-lipid mixture (Gruner et al.. 1985), or by repeated freeze-thawing of MLVs in liquid nitrogen (termed Frozen and Thawed MLVs, or F’ATMLVs) (Mayer et al., 1985a; Westman et al., 1982). FATMLVs are also heterogeneous in size and contain unique intravesicular structures such as vesicles within vesicles and vesicles between lamellae. Closely packed lamellae are rare in FATMLVs in comparison to “normal” MLVs (Hope et al., 1986). Freezing MLVs to form FATMLVs may involve more than a simple equilibration of solutes. Two reports have suggested that the internal solute concentrations achieved inside these FATMLVs can actually exceed the external concentrations in which the vesicles are prepared (Chapman, et al., 1990; Chapman et al, 1991) due to the freezing leading to locally high solute concentrations near the bilayer.  Conversely, one must be aware that 7  Figure 1-1. Electron micrographs of MLV and FATMLV vesicle preparations. EPC MLV (a) and FATMLV (b) lipid vesicles were prepared as in and examined by freeze-fracture electron Section 2.1 microscopy. The bar represents 100 nm and the arrowhead the direction of shadowing. Photographs courtesy of M.J. Hope 8  many agents act as cryoprotectants (Crowe and Crowe, 1988), preventing MLV disruption during freeze-thaw and hence reducing solute entrapment in FATMLVs. The size, heterogeneity and presence of multiple bilayers of MLV preparations complicate the interpretation of experiments designed to examine transbilayer transport or fusion (Hope et at, 1986). Their relatively large size also results in rapid clearance of these vesicles from the circulation upon intravenous injection (Gregoriadis, 1988), reducing the chemotherapeutic potential of these preparations. There have been a variety of procedures developed to make unilamellar lipid vesicles to overcome these disadvantages of MLVs. 1.3.2 SMALL UNILAMELLAR VESICLES (SUVs)  Small unilamellar vesicles (SUVs) can be produced by sonication of MLVs (Huang. 1969). The size of SUVs produced by sonication depends upon lipid composition, varying from 20-30 nm for EPC vesicles to 50 nm for cholesterolcontaining systems (Johnson, 1973). Vesicles in this size range may also be prepared using a “French press”, essentially a high pressure chamber with a narrow orifice (Szoka and Papahadjopoulos, 1980). SIJVs have such a high degree of membrane curvature that there is as much as 3-fold more surface area for the outer monolayer of the vesicle than for the inner monolayer (see Table 1-TI). This high curvature may contribute to the different properties of SUVs compared to LUVs. For example, SUVs prepared from lipid mixtures can have inherent asymmetric lipid distributions while larger vesicles do not (Lentz, et al., 1980). SUVs can also spontaneously fuse to form larger systems (Parente and Lentz, 1984). Differences in binding properties between the inner and outer faces of these vesicles may also occur (CafIso, 1989). Finally, permeability coefficients for ion transport across SUVs may be 1-2 orders of magnitude smaller than for corresponding larger systems (Deamer and Bramhall, 1986; Perkins and Cafiso, 1986). The mmiscule interior volumes (-0.2-0.5 tL/tmo1 lipid) and relative instability 9  of SUVs make them poor model membranes and also inappropriate for drug delivery. In order to overcome these inherent disadvantages of SUVs, several procedures have been developed to make unilamellar vesicles of a larger size. 1.3.3 LARGE UNILAMELLAR VESICLES (LUVs)  Large unilamellar vesicles (LUVs) can be made by ethanol injection (Kremer et al., 1977), ether infusion (Deamer and Bangham, 1976), reverse phase evaporation (Szoka and Papahadjopoulos, 1978), or detergent dialysis (for example, Mimms et al., 1981; Madden, 1986). In these methods, lipid is solubilized in organic solvent or detergent, followed by injection of the mixture into buffer. The solvent or detergent is removed by evaporation at the time of hydration (ether method), by reduced pressure (reverse phase method), by dilution (ethanol injection method) or by dialysis (detergent dialysis). LUVs prepared by these techniques have average diameters of 50 to 200 nm and trapped volumes of between 1-3 L per mol lipid (Szoka and Papahadjopoulos. 1980). These preparation techniques have the disadvantages that they are time-consuming. tend to result in heterogeneous vesicle populations, may not be applicable to all lipids and can contain residual detergent or organic solvents (Parente and Lentz, 1984). A more recent method of preparing LUVs without these disadvantages involves direct extrusion of multilamellar vesicles (Hope et al., 1985) through filters with a small pore size. Olson et al., (1979) demonstrated that vesicles produced by reversed-phase procedures were more homogeneous in size after low pressure extrusion through polycarbonate filters. Cullis and co-workers (Hope et al., 1985; Mayer et al., 1986) demonstrated that direct extrusion of MLVs or FATMLVs through polycarbonate filters (pore size 100 nm) results in vesicles with average diameters of 90-110 nm as judged by freeze fracture electron microscopy and quasi-elastic light scattering techniques. The vesicles are unilarnellar as judged by freeze fracture and NMR techniques (Hope et al., 1985). This method for producing vesicles has been discussed in detail elsewhere (Hope et al., 1986). It should be 10  Figure 1-2. Electron micrographs of extruded vesicles of different size.  Vesicles were prepared by extrusion of FATMLVs through (A) 400 nm (B) 200 nm (C) 100 nm, (D) 50 nm (E) 30 nm pore size polycarbonate filters and examined by freeze fracture electron microscopy. The bar represents 150 nm and the magnification is the same in all photos. Photographs courtesy of L.D. Mayer. 11  noted that vesicles extruded through filter pore sizes 200 nm or larger have increasing amounts of contaminating multilamellar systems (Mayer et al, 1986c). 1.3.4 OSMOTIC PROPERTIES OF EXTRUDED LIPOSOMES  LUVs are not necessarily spherical. Two lines of evidence suggest vesicles produced by extrusion under iso-osmotic conditions are nearly, but not quite, spherical. Cryo-electron microscopy (B. Mui, unpublished data) shows that extruded vesicles are slightly oblate. and the internal aqueous volumes of these vesicles increase slightly when the vesicles are incubated in a hypo-osmotic exterior (Chapter 2). This non-sphericity may be a result of vesicle distortion during the extrusion process. From simple geometrical constraints, an initially non-spherical vesicle cannot become spherical without increasing its internal volume. The “rounding up” would dilute the contents of the vesicle and generate an osmotic gradient which would oppose the volume increase. Transmembrane osmotic gradients (outside hypo-osmotic) below —800 mOsm, result in vesicle swelling to a maximum size, at which point the vesicles are spherical. At higher osmotic gradients a hydrostatic pressure gradient develops equal to the osmotic pressure difference (see Kedem and Katchaisky, 1958). Greater than -800 mOsm hypo-osmotic (outside) gradients results in vesicle rupture and resealing in 100 nm EPC vesicles (B. Mui, Personal Communication). Hyper-osinotic (outside) conditions would be expected to result in vesicle shrinkage in proportion to the osmotic gradient as a result of the permeability of the bilayer to water (Section 1.4.3). 1.3.5 TRAPPED SOLUTES AND TRAPPED VOLUME DETERMINATIONS The trapped” or internal aqueous volume is one of the most important parameters defining a lipid vesicle preparation. It is generally determined by preparing the liposomes in trace amounts of an impermeant water soluble compound such as I C} inulin and determining the amount of radiolabel 4 NaJ 2 2 or [‘ entrapped. The trapped volume can be calculated assuming that the proportion of 12  entrapped solute reflects the volume inside the vesicles. As noted earlier, this is not always the case, since MLVs prepared by hydratmg phospholipids in an aqueous medium exclude solute from their interior (Gruner, et al., 1985). This solute exclusion is a property of the process of liposome formation and does not depend on the specific solute used (if it is impermeant). Subsequent osmotic gradients across the membrane can result in vesicle volume changes, which can also introduce errors into vesicle volume determinations. An alternative method of determining vesicle volumes is to examine the volumes outside the vesicles and calculate the internal volumes from a knowledge of the lipid concentration. This approach to determining vesicle volume gives accurate measurements of the aqueous volume of liposomes (Perkins et al., 1988), but gives no indication of the amount of solute captured. Solute capture can be a more important parameter if it is the trapped molecule which is of interest, as for drug delivery, for example. 1.3.6 SOME RELEVANT PARAMETERS CONCERNING LIPOSOMES  Table 1-Il gives an indication of some of the parameters to consider when using unilamellar liposomes. These are based on calculations which assume the vesicles are idealized spheres with a bilayer thickness of 5 nm and a lipid surface area of 0.6 2 nm / molecule (Deamer and Bramhall, 1986). For example, a 100 nm diameter EPC unilarnellar vesicle (typically used in these studies) would be expected to contain on the order of 90,000 phospholipid molecules per vesicle, with a slightly higher number on the outer monolayer than the inner. It should be appreciated just how small “large” unilarnellar vesicles actually are. Each 100 nm vesicle has an interior volume of only about 10-19 L. Perhaps the most important considerations from Table 1-Il are the large number of vesicles produced by small amounts of phospholipids and the high surface area:volume ratio of liposomes. A milligram of PC has a surface area of approximately 2000 cm 2 and results in about 1013 100 nrn diameter vesicles with 13  an Internal volume of only about 2 giL. This high surface to volume ratio allows the detection of the flux of compounds with low permeability coefficients (Deamer and Bramhall, 1986).  Table 1-Il Relevant parameters concerning unilamellar liposomes  Vesicle Outer inner Diameter monolayer Monolayer (nm) (lipids/yes) (lipids/yes)  Outer: Inner Ratio  Total Lipids per Veslcle  # of Vesicles per 1mole PL  Aqueous Volume (1/mol)  % Volume which Is lipid  Surface: Volume /tLl 2 (m  20  1790  450  4.0  2240  14 2.68x10  0.14  87.5  3.00  40  7180  4040  1.7  11220  3 5.37x10’  0.76  57.8  0.56  60  16160  11220  1.4  27380  13 2.20x10  1.44  42.1  0.29  80  28720  21990  1.3  50710  3 1.19x10’  2.13  33.0  0.20  100  44880  36350  1.2  81230  2 7.41x10’  2.83  27.1  0.15  200  179520  162020  1.1  341540  2 1.76x10’  6.33  14.3  0.07  These parameters were calculated assuming that all vesicles are perfect spheres of the stated diameter with 5 urn bilayer thickness and a surface area of 0.60 2 / phospholipid molecule. nm It can also be seen that the lipid bilayer itself can be a significant proportion of the total volume of the liposome. Therefore, solute interactions with the lipid bilayer must be considered, as discussed in the next section. 1.4 SOLUTE PARTITIONING AND PERMEABILITY  1.4.1 PARTITION COEFFICIENTS Partitioning and binding of ligands to membranes can be analysed in several ways. In the simplest case, the membrane can be treated as a bulk “oil” phase, with a partition coefficient K* defined as  14  K*Cm/Cw  (1-1)  where Cm and C, are the concentrations of a given compound in the membrane and in the water phases respectively. More realistically, the bilayer cannot always be expected to behave as a thin isotropic oil, but can be treated as a twodimensional solvent. Hence, partitioning can also be expressed as a surface phenomenon  (1-2)  13=Nm/Cw  where the partition coefficient is 13 (in cm) and Nm is the surface concentration of bound ligand. If Nm is divided by d, the bilayer thickness, one obtains the value of the partition coefficient K* above. Alternatively, another partition coefficient, usually also called K (in mol/cm ), can be defined, wherein Nm is divided by 3  o,  the  depth of the membrane into which drug is bound. All three partition coefficients are commonly used (Gennis, 1989). A key parameter determining partition coefficients (as well as transport rates and activation energies) appears to be the number of intermolecular hydrogen bonds which a diffusing molecule can form in water, since these bonds cannot be re-formed in the bilayer (Cohen, 1975a,b). This relation breaks down above approximately 10 hydrogen bonds (Bangham and Hill, 1986). Each potential hydrogen bond changes the activation energy of transport by about 4 kcal (Cohen, 1975a), approximately the energy of formation of a hydrogen bond. Other “incremental changes in free energy” (which predict the effects of various substituents on partitioning) have been estimated for partitioning into bilayers (Diamond and Katz, 1975). There has been considerable effort put into finding the best oil model to mimic a membrane (Walter arid Gutknecht, 1984; Diamond and Katz, 1975), but essentially any low dielectric medium provides a reasonable membrane model to a first approximation. The standard oil for which partition coefficients are  15  determined is octanol. Octanol:water partition coefficients are widely available and appear to be surprisingly useful predictors of membrane:water partition coefficients for non-electrolytes (for example, see Walter and Gutknecht, 1984). Unfortunately, partition coefficients of acids and bases are generally not determined separately for the charged and uncharged species of these compounds. Most commonly, partition coefficients for acids and bases are reported under unbuffered or unspecified conditions, rendering it impossible to separate the relative contributions of two molecular species of the compounds (Strichartz, et al., 1990). 1.4.2 PERMEABILITY OF NON-ELECTROLYTES  Net solute flux (J, in units of moles 1) can be described by a permeability coefficient (P. which generally has units of cm 1 s ) , where J=PA(AC)  (1-3)  where A is the surface area and AC is the concentration gradient which drives solute flux. Typical values of P range from 1O-1O cm s for compounds which rapidly cross bilayers (e.g. water or urea), to 10-10 cm s’ for moderately permeable molecules (e.g. glucose) to 10-13 cm s’ for “impermeable’ ions such as sodium and potassium (Walter and Gutknecht, 1984: Deamer and Bramhall, 1986). Assuming a first order process for net diffusion of a compound out of a unilamellar liposome of constant interior volume it can be shown that to a first approximation  ln {C(t)/C(0)}  =  -kt  (1-4)  where C(t) is the internal solute concentration at time t. C(0) the initial concentration, t is time and k the first order rate constant. Considering the efflux of solute from an LUV of radius r, it follows that P=kr/3 16  (1-5)  The half-time of transport 1 (t /2 can be obtained easily from the rate constant: =.693/k 1 t 12  (1-6)  Half-times for efflux out of 100 nm diameter vesicles employing the permeability coefficients above are in the millisecond range for water, a few hours for glucose and several days for potassium. Permeability properties can be described by of a “solubility-diffusion” model. In this model the rate limiting step is diffusion across the bilayer and interfacial resistance is negligible (Gennis, 1989). Hence P=KD/d  (1-7)  where D is the diffusion constant within the membrane and d is the width of the bilayer. Both K (the partition coefficient) and D (the diffusion coefficient) are average values, which are not necessarily constant across the bilayer (Diamond and Katz, 1975). There is a good correlation between oil:water partition coefficients and lipid bilayer permeability over at least 7 orders of magnitude. a generalisation known as “Overtons rule’ (Walter and Gutknecht, 1984). Thus, if one knows the permeability of one or two small solutes to a given membrane, the permeability of other small solutes can be reasonably approximated (see Cohen, 1975; Walter and Gutknecht, 1984). It has been suggested that lipid solubility alone does not determine the permeability of all compounds, since some very small molecules may permeate faster than predicted by their partition and diffusion coefficients (Lieb and Stein, 1986). This behavior has been suggested to be due an ability of the small molecules to “fit” into small transient defects in the lipid acyl chain packing (Lieb and Stein, 1986). More glaring exceptions to Overton’s rule occur with larger molecules. If phospholipid membranes do deviate from the behavior expected of thin oil layers,  17  S S S  [o]<  S I I I  K  [mJ  >  S 4 4  ’ 0 Vm  4S  K  [I]  L’m,  Inside  Outside  Figure 1-3. permeability.  S 4 S S S S S I I I I I I I I I I I I S S * S S * S 4  Solubility-diffusion model of membrane  If the [nterfacial energy barrier is small, the interfacial concentration of a given compound can be assumed to be in equilibrium (as described by the partition coefficient. K) with the aqueous concentration. The rate limiting step of permeation is diffusion across the lipid acyl chain region of the bilayer. The depth of binding ô is not to scale. 18  one might expect the deviation to be more pronounced when the size of the diffusing molecule is significant compared to the dimensions of the bilayer. Phospholipids such as PC, for example, are relatively large molecules which have extremely high partition coefficients, but their rates of diffusion across the bilayer (“flip-flop”) are very slow (Hauser and Barratt, 1973). 1.4.3 PERMEABILITY OF WATER  Perhaps the most important solute to consider is water. Determinations of water permeability monitor the rate of exchange of water under isotonic conditions or in response to an osmotic gradient (F) (Finkeistein, 1987; Ye and Verkman, 1989; Benga et al., 1990). These values do not have to be the same. Different values of D arid F imply that water transport is occurring through aqueous channels (Fettiplace and Haydon, 1980; Dearner and Bramhall, 1986). After appropriate corrections for unstirred layer effects (wherein flux through “unstirred” water at the membrane:water interface can be slower than flux through the membrane itself see Finkeistein, 1987 for discussion and appropriate -  3 cm s* correction procedures) D and F in lipid bilayers are on the order of iOwith activation energies generally between 8 and 13 kcal/mol (Fettiplace and Haydon, 1980). These values are much higher than the activation energy of 4.6 kcal/mole expected for bulk diffusion of water through aqueous pores (Fettiplace and Haydon, 1980, Finkeistein, 1987). The same is not necessarily true for biological membranes. Red blood cell membranes, for example, appear to have protein dependent pores which can be poisoned by mercury containing compounds (Macey and Farmer, 1970). Water transport through red cells decreases about 5fold upon treatment with mercurials and the activation energy of water movement rises from 3-5 kcal/mol to 11.5 kcal/mole (Benga et al., 1990). There are two main models for water transport across protein-free model membranes, one based upon the solubility-diffusion model described above and the other based on water entering transient defects which spontaneously form in the 19  bilayer (Deamer and Bramhall, 1986). For example, water permeability increases dramatically in vesicles composed of lipids which are at the phase transition temperature, where these defects are expected to be most common (Deamer and Bramhall, 1986). These models are not mutually exclusive. 1.4.4 PERMEABILITY OF IONS The permeability of bilayers to small ions such as  or Na+ is very low in the  absence of lonophores, though proton/hydroxide flux appears to be an exception (see below). Permeability coefficients for these ions are on the order of 10-12 to 10-14 cm s (Bangham et al., 1965: Hauser et  a!.  1973). These slow rates are  generally attributed to the large energy barrier (‘Born’ energy) encountered upon entering the low dielectric constant medium of the bilayer, where the Born energy, w, of transferring an ion from water to a hydrocarbon can be approximated (in cgs units)  w  =  2 / e  Cr 2 h  (1-8)  where e is the ionic charge, r the ionic radius and hc the dielectric constant of the hydrocarbon (Parsegian, 1969). For example, a 0.2 nm radius monovalent ion passing into a 7 nm thick slab of hydrocarbon with a dielectric constant of 2 must pass an energy barrier of about 40 kcal/mol (Parsegian, 1969). One of the few tests of this hypothesis (Dilger and McLaughlin, 1979) showed that the presence of chiorodecane (expected to raise the membrane dielectric constant) increased the thiocyanate permeability of planar lipid membranes several thousand fold, consistent with Born energy considerations. While Born energy is a reasonable framework in which to view the low permeabifity of membranes to ions, the critical values of the membrane dielectric constant and the ionic radius used to calculate the energy barrier are not known with precision (D. Deamer, personal communication). For example, if the dielectric of a membrane and the ionic radius were both half of the values used in the 20  calculations used above, the calculated Born energy barrier would rise to above 160 kcal/mol and the predicted ion flux rate would be about 10100 slower. Thus, the observation that sodium has a permeability coefficient about three orders of magnitude greater than that predicted on the basis of Born energy considerations (Hauser at al., 1973) may not necessitate alternative mechanisms to describe ion transport across bilayers, due to the sensitivity of these calculations to these parameters. One observation which is not consistent with simple Born energy considerations is that membranes are generally much more permeable to anions than to cations. For example, tetraphenylboron anion (TPB-) is several orders of magnitude more permeable than the corresponding tetraphenyiphosphonium cation (TPP), even though these molecules have similar hydrophobicity and hydrogen bonding abilities (Flewellmg and Hubbell, 1 986a). An “internal dipole potential of about 250 mV, which favours anions, has been suggested to be the basis of the higher permeability of anions than cations (Fleweuing and Hubbell, 1 986b). A portion of the internal dipole is believed to be due to the alignment of the ester linkages of the phospholipids along the plane of the bilayer, while the rest may be a result of oriented water molecules (Flewelling and Hubbell, 1986a, 1986b). Adding the compound phloretm to the bilayers increases the permeability of cations and decreases the permeability of anions, apparently altering the membrane dipole potential (Perkins and Cafiso, 1987a). A “total potential” model of membranes, which includes Born energy, dipole potential, “image energy” (additional energy resulting from the finite thickness of the bilayer), and neutral energy terms (encompassing all other factors) has been proposed to explain observed ionic permeabilities (Flewelling and Hubbell, 1 986a,b). In vivo, ion permeation is controlled by proteins. In model membranes, ionic  selectivity is generally provided by ionophores, such as valinomycin. Valinomycin dramatically increases the permeability of unsaturated PC bilayers to K ions (Blok et a!., 1974b; Bangham, 1983). If there is an initial potassium ion gradient across 21  the bilayer, the result is a net movement of K+ ions across the bilayer. This ionophore is therefore “electrogenic”, since it promotes the net movement of charge across the bilayer. Other ionophores which are specific for different ions can also be used (see Section 1.4.5). An alternative way in which some ions may cross membranes is through the formation of neutral  “ion-pairs”.  In a low dielectric  medium, some ions can complex to form stable neutral species. For example, thiocyanate and a nitroxide analog of TPP+ have been shown to be much more permeable in the presence of agents which can form a neutral, hydrophobic ion-pair (Gutknecht and Walter, 1982). 1.4.5 PERMEABILITY OF PROTONS Because of the importance of the proton electrochemical gradient  in  biological  transport and energy transduction, proton transport across bilayers deserves special consideration. According to the “chemiosmotic hypothesis” (Mitchell, 1961) biological ATP-synthesis is coupled to the proton electrochemical potential, or protonmotive force (p.m.f.). This led to interest in determining the permeability of membranes to protons and to a considerable degree of controversy. It should first be noted that the term “protons” as used here does not refer to (H 0 +) ions. Like other ions, protons are expected to be solvated H+ or hydronium 3 in aqueous solution, although the exact number of solvating molecules is not known. There may be as many as 20 or 21 water molecules of hydration around a free proton and the protonic charge can likely delocalise around this water cluster (Yang et al., 1991). For most purposes it is convenient to define Hiaq) as “protons”. Consideration of proton flux across membranes forces a further broadening of the effective definition of “proton”. This is because it is difficult to distinguish between the movement of protons in one direction and the movement of hydroxyls in the opposite direction (Nichols and Deamer, 1980). Therefore the term net was developed to describe the undefined combination of these two events 1 (Nichols and Deamer, 1980) for small gradients around pH 7. It is also important to distinguish between the electrogenic movement (which 22  results in the generation of a transmembrane electrical potential) and non electrogenic movement of protons (which is electrically neutral) (Cafiso and Hubbell, 1983). If one has a probe which measures the change in the internal pH of vesicles, it is difficult to tell whether the change in internal pH arises from “proton” movement, or from the net redistribution of other compounds in response to a ApH (see Section 1.5.3). The system can be best described only if both electrogenic and non-electrogenic proton flows can be defined. Nichols and Deamer first noted that proton transport in bilayers was anomalous (Nichols and Dearner, 1978; Nichols and Deamer, 1980; Nichols et al., 1980; Deamer and Nichols, 1983). The key features of this anomaly are that proton permeability is greater than other small ions by orders of magnitude (for small gradients near pH 7), and that the permeability of protons appears to decrease as the size of the proton gradient increases. A wide range of proton permeabilities have been reported since, ranging from 1O to i0 cm s (Gutknecht, 1984; Gutknecht, 1987a; Elamrani and Blume, 1983; Cafiso and Hubbell, 1983, Perkins and Cafiso, 1986; Deamer, 1987, Gutknecht, 1987b, Nagle, 1987, Perkins and Cafiso, 1987b, and references therein). This suggests a million fold disagreement! However, much of this disagreement disappears when it is taken into account that calculated values of net cannot be compared unless experimental conditions are similar. This surprising conclusion is a result of the fact that initial proton flux across bilayers is relatively independent of the size of the proton gradient and is generally found to be about 10-13 to 10-15 mol cm’ s (for reviews, see Deamer, 1987, Gutknecht, 1987b, Nagle, 1987, Perkins and Cafiso, 1 987b). The available data are therefore not consistent with the simple diffusion mechanism of a charged species over an energy barrier (Perkins and Cafiso, 1987b; Deainer, 1987). The fact that proton conductance is relatively constant over a wide range of pH (i.e. of proton gradients) clearly implicates some sort of carrier mechanism as the rate determining step in proton flow (Deamer, 1987). 23  H+H  H  H H H  Figure 1-4 Representation of protons crossing lipid bilayers by a carrier mechanism. Protons (or equivalents) cross bilayers at a rate which is independant of the proton concentration gradient. The mechanism by which protons cross the bilayer (represented by the funnel) is unknown. 24  Several possible mechanisms have been put forward to explain the two key features of proton flux, which are that it is relatively independent of the size of the proton gradient and it is faster than other small cations. Many of these models depend on aggregates of intra-membrane water as key transporting agents, by  analogy with the high proton conductance of ice (Nagle and Morowitz, 1978; Nichols and Deamer, 1980; Deamer, 1987; Miller, 1987; Nagle, 1987 (three models!); Deamer and Nichols, 1989). These models share the common factor that protons can effectively “hop” down a chain of water molecules via a rearrangement of the water molecules, resulting in a higher mobility of protons than other ions. Others suggest that weak acid protonophores present in low levels in lipid preparations are responsible for the majority of proton flux across membranes (Gutknecht, 1984; Gutknecht, 1987b). 1.5 TRANSMEMBRANE ION GRADIENTS  1.5.1 AW AND zpH  A transmembrane electrical potential (z\’P) is established across a membrane by the uncompensated transmembrane movement of a charged species. If only one permeant monovalent ion is present, the equilibrium A’I can be estimated from the Nernst equation:  -  =  RT  ) 0 ln(C/C  (1-9)  F  where R is the gas constant, T is the temperature, F is the Faraday constant and C and C 0 are the internal and external concentrations of the permeant species, respectively. This equation makes the approximations that pressure, activity coefficient, and other differences between the internal and external compartments are small (Nobel, 1991). For the special case of protons, the protonmotive force (p.m.f.) is defined as  25  zero when, at equilibrium, at 25°C, bP=--60ApH  (1-10)  where iW is in mV. It should be emphasized that while the p.m.f. can decay to zero as protons leak out of vesicles to achieve their electrochemical equilibrium, a sizeable, stable M1’ can remain. Note also that membrane potentials and i\pH and an opposing 1 transmembrane pH gradients occur widely biologically, for instance in nerve cells, mitochondria, and chioroplasts. 1.5.2 MEASUREMENT OF ApH AND &P IN LIPID VESICLES The accurate measurement of transmembrane potentials and pH gradients is central to membrane bioenergetics (Rottenberg, 1979; Nicholls and Rial, 1989). From the viewpoint of this thesis, these techniques are important both in “accounting” for protons (Chapter 2 and 3) and in determining rates of proton movement (Chapter 4). Therefore, the conditions required for these determinations in lipid vesicles are discussed below. Many techniques to measure AW and ApH have been developed, including techniques which monitor the distribution of radiotracers (Rottenberg, 1979, 1984), ESR of spin-labelled probes (Cafiso and Hubbell, 1978a, 1978b, 1982), P]-NMR 31 [ (Redelmeier et aL, 1989, Moon and Richards, 1973), fluorescence of pH sensitive compounds (Nichols and Deamer, 1980, Bramhall, 1985), and other methods (Rottenberg, 1989) of determining the fate of the appropriate probes. Not all of these techniques have a strong theoretical basis (Rottenberg, 1989). In addition, these procedures generally require specialized equipment and are usually restricted to ApH values of 2 units or less. The equilibrium formalism shown below is pertinent to any probe which redistributes in response to membrane potentials and is based on a model developed by Cafiso and Hubbell (1978a), and Rottenberg (1984). A similar 26  formalism can be adapted for ipH probes (see Chapter 3). Consider the behavior of a probe of &P in response to a membrane potential as indicated in Fig. 1-3. At equilibrium, the electrochemical potential of the probe must be the same in all compartments. Thus, for a symmetric membrane where the probe partitions into narrow interfaces at the interior and exterior surfaces of a vesicle and where there are negligible surface potentials, the distribution of probe molecules in the four regions denoted in Figure 1-3 can be calculated as follows. If the number of unbound, unencapsulated molecules, N , is taken to be 1, then from the definition 0 of the partition coefficient, the relative number in the outer membrane region (Nm ) 0 is 0 Nm  0 mo /V  (1-11)  Since the equilibrium transmembrane concentrations of the ionic probe are determined by the presence of a AW (see Eq. 1-6), the relative number inside the vesicles, N, is: N  =  /V eW1lT 1 (V ) 0  (1-12)  and the number at the inner interface (Nm ), is therefore: 1  = 1 Nm  1 KVmi N  vi  =  KV 0 /V e’’/RT 1 (V )  (1-13)  1 v  The most important feature evident from these equations is that the distribution of molecules in all compartments will vary with the membrane potential. Thus, if the technique used to measure the equilibrium distributions of probe monitors  the aqueous population (for example, if the potential probe is  K+, which has a very small partition coefficient) or monitors  the membrane  bound population (as in ESR studies of lipophilic probes), the determination of zq’  27  is relatively simple. However, among the most commonly used AW (and z\pH) probes are hydrophobic fluorescent compounds which undergo changes in fluorescence in response to AM’ (or ApH) (Rottenberg, 1989). Only total fluorescence changes can be measured and these changes can arise from any or all compartments. It is often difficult to predict what fluorescence behavior to expect of these probes as a function of AM’. Furthermore, AM’ probes must not be directly affected by the presence of a pH (and vice-versa), vesicle volume changes, temperature and other factors unless appropriate corrections can be made. These considerations mean that caution must be employed when interpreting measures of AM’, tpH, or proton permeability. The continued successful use of changes in the fluorescent properties of these membrane potential probes is perhaps partly due to several self-cancelling errors (Rottenberg, 1989). 1.5.3 SIGNIFICANCE OF ACID-BASE CHARACTERISTICS OF COMPOUNDS  A great many biologically interesting compounds are weak acids or bases. The equilibrium constant (Ka) of a weak base is: KaHB/BH  (1-14)  where H is the proton activity, B is the activity of the unionized base and BH is the activity of protonated base and Ka is the dissociation constant. To a first approximation activity coefficients are not required and activities can be replaced by concentrations. The relative concentrations of the neutral and protonated weak base at a given pH are therefore described by the Henderson-Hasselbach equation:  pH  =  pK  +  log {[B]/1BHi}  (1-15)  If the dissociation constants (and activity coefficients) for a weak base are the same on both sides of a vesicle membrane  Ka  =  /[BHi 1 [H] LB]  28  =  /[BH1 [B] 0 0 [Hi  (146)  //  [D[rI=1O  H÷”118 [D]=1  ((  PK=81 [D]=1  pH4 Inside  pH? Outside  I  [Dl-r]=10000  L\PH  I  Effect of transmembrane pH gradients on the Figure 1-5 equilibrium transbilayer distribution of weak bases The neutral form of the weak base equilibrates across the bilayer As a result of reprotonation in the low pH environment of the vesicle interior a transmembrane gradient of the weak base develops At equilibrium the concentration gradient of the base equals the transmembrane proton gradient if pK>pH. The arrows reflect net movements under the conditions described in the numbers indicate the relative Chapters 2 and 3 concentrations of the species 29  where the subscripts  and  refer to inside and outside the vesicle, respectively. The  uncharged species of a given compound tends to be much more membrane permeable than the corresponding charged species (Crofts, 1967; Rottenberg, 1979), so if the compound resides in the aqueous space, the equilibrium concentration of the neutral species will be the same on both sides of the membrane (Figure 1-5). If pK  >>  0 pH  >>  pHi, the weak base distribution will therefore reflect the pH gradient: 0 [BH1/[BH]  =  (1-17)  The above analysis is included to emphasize that the equilibrium transmembrane distributions of all acids and bases which are not zwitterions and which are membrane permeable in the neutral form are expected to be influenced by the presence of a transmembrane pH gradient. This fact can be exploited for measuring pH gradients (Chapter 2) and for loading drugs into vesicles (Chapter 3). Since membrane potentials can induce transmembrane pH gradients and viceversa (Bramhafl, 1985; Redelmeier, 1989), the distribution of other ions can affect the distribution of a wide range of compounds either directly or by inducing a ApH. Finally, some weak acids are significantly permeable in both their charged  and uncharged forms (McLaughlin and Dilger, 1980). These compounds (for example, CCCP) are known as protonophores, since they can shuttle protons across the membrane (Kasianowicz et al., 1984). 1.6 DRUG TRAPPING IN LIPOSOMES  1.6.1 PASSIVE ENTRAPMENT TECHNIQUES  As mentioned earlier, liposomes have potential as drug delivery vehicles (for review, see Ostro and Cullis, 1989). Drugs can be “passively” encapsulated into liposomes by simply preparing vesicles in a solution containing the desired compound (Taylor et al., 1990). However, the resulting low efficiency of drug entrapment and low drug:lipid ratios could require unacceptably high levels of lipid  30  to be injected into patients (and could also result in waste of expensive drugs). For example, a preparation of 10 mg/mL EPC 100 nm diameter vesicles can encapsulate only about 1% of a given hydrophiic drug by passive entrapment. A “dehydration-rehydration” protocol can slightly improve the efficiency of drug entrapment (see Kirby and Gregoriadis, 1984a; 1984b). Leakage of encapsulated drugs during storage could also limit the usefulness of such passively loaded preparations. Not surprisingly, hydrophobic compounds show greater association with lipid vesicles than hydrophilic ones (Stamp and Juliano, 1979). A variety of parameters, such as the ratio of encapsulated drug to the lipid dose can have profound effects on liposomal drug disposition in vivo (Mayer, et al., 1990a). Passive drug loading protocols, however, do not permit simple, independent variation of these parameters. These problems can be overcome with “active” loading techniques as indicated in the next section. 1.6.2 ACTIVE TRAPPING TECHNIQUES To overcome the problems of inefficient drug entrapment, low drug:lipid ratios  and poor drug retention, other techniques have been developed for liposomal drug formulations. For example, the particularly high affinity of the drug doxorubicin for cardiolipin has been exploited to produce liposomal doxorubicin preparations with high drug trapping efficiency (Rahman et al., 1980). Alternatively, another liposomal doxorubicin preparation (detailed in Chapter 3) is based upon a procedure whereby the drug is encapsulated in liposomes in response to transmembrane pH gradients. This occurs essentially as outlined for weak bases in Section 1.5.3 (Mayer et al., 1990a; Mayer et a!., 1986a; Mayer et al., 1986b). A similar protocol was earlier shown to entrap catecholamines in liposomes (Nichols  and Deamer, 1976). This zpH loading technique has a key advantage in that it is potentially applicable to loading a wide range of compounds which are lipophilic weak acids or bases, including a large proportion of commonly used pharmaceuticals. zpH 31  loading allows independent variation of many liposomal parameters and can result In much higher drug:lipid ratios than can be achieved by passive drug entrapment (Mayer, 1986b). Drug leakage from the vesicles is also much slower (Mayer, 1 986b). Further, as the transmembrane distribution of the drug is determined by the proton gradient, it is possible to control the rate of drug release by altering the pH gradient. 1.7 THESIS OVERVIEW Chapter 2 presents an investigation of the accuracy of the techniques employed to measure zpH and z’I’ for LUVs with a variety of lipid compositions. It is shown that transhilayer distributions of the base methylamine accurately indicate ApH under a variety of conditions. In turn, if the vesicle volume, concentration and interior buffering capacity are known, these can be used to predict the level of accumulation of methylamine. The interactions of a variety of drugs with LUVs having transmembrane pH gradients are examined in Chapter 3 and the results compared with those predicted on the basis of the analysis of Chapter 2. The technique of ApH loading is shown to be applicable to a wide range of drugs, although drugs are accumulated to much different extents. An extension of the model derived in Section 2.1, incorporating the ability of drug to partition into the membrane can be used to explain the different levels of drug accumulation. Finally, in Chapter 4, proton flux and associated activation energy in response to a 3 unit transmembrane pH gradients are examined under various conditions. The activation energy of proton flux is found to be about 11 kcal/mol. Under appropriate conditions, proton flux is not exceedingly rapid and pH gradients do not decay quickly, allowing the generation of systems with a stable ApH and no AW, and vice-versa.  32  CHAPTER 2. MEASUREMENTS OF TRANSMEMBRANE pH GRADIENTS IN LUVs  2.1 INTRODUCTION  It is well known that weak acids and bases can undergo a net redistribution across bilayers in response to transmembrane pH gradients (Crofts 1966; Crofts, 1967; Rottenberg, 1979). Recent work has characterized the influence of transbilayer pH gradients in large unilamellar vesicle (LUV) systems on the transbilayer distributions of many of these compounds, including a variety of drugs (Nichols and Deamer, 1976; Bally et al., 1985; Bally et al., 1988; Madden et al., 1990), ions such as calcium (Viero and Cullis, 1990), modified peptides (Chakrabarti et al., 1992), fatty acids (Hope et al., 1987), and phospholipids (Redelmeier et a!., 1990, Eastman et al., 1991). An ability to measure the pH gradients (ApH) present across LUV membranes is clearly central to these and other investigations. A number of methods of measuring tpH in LUVs have been developed, based on early work demonstrating that the transmembrane pH gradients in organelles such as chloroplasts and mitochondria could be measured by determining the transmembrane distribution of weak bases such as ammonia (Deamer et a!., 1972; Rottenberg, 1979; Deamer, 1982). Similar procedures using lipophiic ions can be used to indicate iW (Rottenberg, 1989). As emphasized by Rottenberg, z\pH values determined by probes which can partition into the membrane can be misleading (Rottenberg, 1989; Section 1.5.2). The transmembrane distribution of trace amounts of C1-methylamine 14 can L be used to determine pH in LUVs (see Section 1.5.3). Due to the high permeability of the uncharged form of methylamine, rapid equilibration across the membrane occurs (Rottenberg, 1989). Protonation of the neutral form in the low pH environment of the vesicle interior results in a net accumulation of probe to achieve an equilibrium where the probe concentration gradient reflects the pH gradient. Separation of trapped probe from untrapped probe and the subsequent determination of entrapment allows the proton gradient to be measured. For  33  liposomes with acidic interiors, it is shown that protocols employing radiolabelled methylarnine in conjunction with gel filtration procedures to remove untrapped methylarnine provide accurate measures of ApH in most situations. However, there are situations in which the accuracy of this technique of ApH measurement can be compromised. First, as indicated above, to achieve equilibrium, the neutral form of the probe must readily permeate the vesicle bilayer, leading to possible errors for relatively 14 in [ 3 impermeable membranes. Second, the ratios of trapped to free C]-MeNH liposomes are usually determined by removing exterior (untrapped) probe employing centrifuged gel filtration mini-columns (Redelmeier et al., 1989). However, once the vesicles enter the gel matrix the system is not at equffibrium, resulting in possible efflux of the probe from the vesicles while the vesicles are on C]-MeNH on arrival in the vesicle 4 ‘ the column. Finally, the protonation of I 2 interior consumes a proton. Thus, the ApH is affected by the probe itself. The extent to which these and other factors can compromise the accuracy of ApH measurements are examined in this Chapter, procedures which avoid these difficulties are described. Further, techniques to determine the membrane potentials (zW’) induced in response to these pH gradients are also evaluated. Finally, it is of interest to determine the maximum transmembrane pH gradient which these vesicles can withstand. 2.2 MATERIALS AND METHODS 2.2.1 MATERIALS Buffers and other chemicals used were purchased from Sigma Chemical Co. (St. Louis, Missouri) unless otherwise stated. Radiolabels were supplied by New England Nuclear (Mississauga, Ontario). Benzene, methanol and other solvents were purchased from BDH (Vancouver, B.C.). All phospholipids were obtained from Avanti Polar Lipids, (Birmingham, Alabama).  34  2.2.2 LIPID VESICLE PREPARATION Cholesterol was incorporated (where noted) in lipid samples by co lyophilization from benzene:methanol (70:30, v/v). Multilamellar vesicles (MLVs) of the stated lipid composition were prepared by adding the indicated buffer (generally 300 mM citrate buffer, pH 4.0 or 300 mM CHES, pH 9.0) to the dry lipid powder at temperatures above the lipid gel-liquid crystalline phase transition temperature. Samples were vortexed for 5 mm, then subjected to 5 cycles of freezing (> 3 mm  in  liquid nitrogen) and thawing in order to produce “FATMLVs” (Mayer et al., 1985a), unless otherwise indicated. Large unilamellar vesicles were prepared by extruding these MLVs or FATMLVs 10 times through two stacked 100 nm filters as previously described (Hope et al., 1985). For saturated lipids, vesicle extrusion was performed above the gel-liquid crystalline phase transition temperature employing a thermally jacketed extrusion apparatus (Lipex Biomembranes, Vancouver, Canada). Vesicles prepared by this method are unilamellar, with diameters of approximately 100 nm (Hope et al., 1985; Nayar et al., 1989). 2.2.3 PARTICLE SIZE DETERMINATIONS Vesicle diameters of some samples were examined by freeze-fracture (Hope et al., 1985), or cryo-electron microscopy. Vesicle size was routinely determined by quasi-elastic light scattering (QELS) at about 0.1 mM lipid (Mayer et al., 1986a) with a Nicomp Model 200 Laser Submicron Particle Sizer (Nicomp Instruments, Goleta, CA) using a 5 nW Helium-Neon Laser at a wavelength of 632.8 nm. QELS analyses fluctuations in scattered light intensity generated by the diffusion of particles in solution. The measured diffusion coefficient is used to obtain the average hydrodynarnic radius and hence the mean diameter of the particles. 2.2.4 DETERMINATION OF ENTRAPPED SOLUTE In general, solute entrapment was determined by preparing vesicles in the presence of 2 iCi/mL of the indicated impermeable trap marker. Unencapsulated 35  L of the sample through a 5 mL Sephadex marker was removed by passing 250 1 14 marker), or Sepharose [ G-50 gel filtration column ([ C1-citrate or ]C-Histidine 14 H]-inulin). In the case of MLV or FATMLV preparations, unencapsulated 3 CL-4B ([ probe was removed by 5 cycles of washing and centrifugation in a Fischer microfuge at 9000 rpm for 3 mm. Trapped marker was quantified by liquid scintillation counting and phospholipid by phosphate assay (see below). The amount of trapped citrate buffer was established by adding 10 Ci/mL 14 to the 300 mM citrate buffer used for hydration and determining [ C]-citrate citrate entrapment by gel filtration after freeze-thaw and extrusion through 100 nm pore size filters. 2.2.5 EFFECT OF INITIAL SOLUTE DISTRIBUTION ON ENTRAPPED SOLUTE  EPC vesicles (extruded through the indicated filter pore size) derived from MLVs as well as from FATMLVs were prepared as above in 150 mM NaCl 20 mM H]-inulin. Vesicles were washed and 3 HEPES, pH 7.5, containing 5 iCi/mL [ centrifuged in a Fischer microfuge at 9000 rpm for 3 mm, the buffer removed and this cycle repeated 3 times to produces MLVs or FATMLVs with the solute marker present only inside the vesicles. This preparation was subsequently extruded and the levels of solute remaining inside the vesicles determined as above. Similar results were obtained using [‘ C]-Histidine as the trap marker. 4 Alternatively, the marker was added after the preparation of multilamellar vesicles (MLV or FATMLV) but before the extrusion procedure, resulting in systems where the marker is initially present only outside the vesicles. 2.2.6 VESICLE VOLUME DETERMINATIONS  The interior volume of extruded vesicles (in a hypo-osmotic medium) was determined by preparing vesicles in 300 mM citrate, pH 4.0, exchanging the external buffer for 150 mM NaCl, 20 mM HEPES, 0.2% azide containing 10 i.tCi/mL 14 and incubating the preparations for 24 h at 45°C. The half-time for [ C]-glucose glucose permeation in 100 nm EPC:cholesterol (55:45; mol:mol) LIJVs is 36  —  1 h at  45°C (B. Mui, unpublished data). Entrapped C 14 1-glucose was determined [ employing Sephadex G-50 gel filtration to separate free from entrapped material as above. The amount of entrapped glucose is expressed as an equivalent volume, assuming that the probe distribution reflects the volume distribution. In general, the experiments presented in this thesis are done under conditions where the interior of the vesicles are hyperosmotic with respect to the external medium (generally 600 mOsm inside, 300 mOsm outside), such that vesicles are at their maximum size. It should be noted that drug loading in response to pH gradients (see Chapter 3) results in an increase in the internal osmolarity of the vesicles. Under the conditions used, the osmotic gradient remains below 800 mOsm. 2.2.7 DETERMINATION OF LIPID CONCENTRATIONS Lipid concentrations were generally determined by analysis of lipid phosphorus as described previously (Fiske and Subbarow, 1925; Bottcher et al., 1961). Aliquots containing between 0.02 and 0.2 tmole phospholipid were digested 3 for at least 1 hr. After cooling, 7.0 mL of ammonium in 0.6 mL of 70% HC1O molybdate reagent (0.22%, w/v, ammonium molybdate in 2% 4 S0 w/v) and 0.6 2 H , mL of Fiske-Subbarrow reagent (30 g NaHSO , 1 g 3 3 SO and 0.5 g bis 1-amino2 Na 2-napthol-4-sulphonic acid in 200 mL water were added. Subsequently, the samples were heated for 20 mm  at 100°C and the absorbance at 815 nm was  determined after cooling. The amount of phospholipid was calculated by comparing the absorbance to a standard curve. On some samples, cholesterol assays (Rudel and Morris, 1973) were also performed. 1.0 mL of glacial acetic acid containing o-pthalaldehyde were added to lipid samples. After 10 min, 2.0 mL concentrated 4 S0 was added and the 2 H amount of cholesterol determined optically by reference to a standard curve. Cholesterol determined in the final vesicle preparations corresponded to the original proportions added, indicating that the neither component is preferentially 37  lost during vesicle preparation. This assay was also used when samples contained significant amounts of non-lipid phosphorous. 2.2.8 GENERATION AND MEASUREMENT OF TRANSMEMBRANE ION GRADIENTS The pH gradient was imposed by passing 300 iL of the vesicles down a 10 mL Sephadex G-50 column equilibrated with the appropriate external buffer, usually 150 mM NaC1, 20 mM HEPES, pH 7.0. These vesicles were quickly diluted into the same buffer containing 0.5 tCi C]-MeNH 1 [ / 3 mL 4 (or other radiolabel) to a final lipid concentration of 1-3 mM. The samples were incubated at the indicated temperatures and at appropriate times 100 tL aliquots were removed and passed down 1 mL Sephadex G-50 mini-columns which were centrifuged for 3 mm  at 2000  x g (Penefsky, 1977). Entrapped probe was determined using liquid scintillation counting and phospholipid concentrations determined using a modified phosphate assay described above. The ratios of entrapped to free concentrations of probe were determined employing the measured internal aqueous volumes indicated in Table 2-I for 100 nm vesicles. For some samples, transmembrane probe distributions were determined using a “centrifree” equilibrium binding analysis apparatus (Amicon, Danvers, MA, USA). After a 30 mm  incubation in the presence of the radiolabel under the  conditions indicated above, 1 mL samples were placed in the upper filter chamber  and the apparatus was centrifuged at 1500 x g for 5 mm. Probe distributions were then determined as indicated above. Measures of membrane potential using [ H1-TPP or C]-Iabelled 3 14 SCN were [ obtained in a similar manner as with 3 C]-MeNH For comparative purposes, 4 [‘ . electrical potentials are often expressed here as Iog(Iprobe1/[probe] ). No 0 corrections for probe binding were made to these values, since these were small (see Results).  38  2.3 THEORETICAL CONSIDERATIONS  The pH of the LUV interior can be influenced by the influx of probe (which consumes a proton) and by proton efflux to set up a AW in electrochemical equffibrium with the ApH. First consider the influence of amine accumulation on the interior proton concentration. For a buffer such as citrate with three acidic groups, it is straightforward to show that the concentration of the buffer in the neutral, fully protonated, form ([B]), in the singly deprotonated form ([WI) and so on are related to the total buffer concentration [B]tot via the relations [B]  [B-]  ] 2 [B  ] 3 [W  =  =  =  =  [BtOt] /fiHj  /[H] 1 K  x  /([H]) 1 K 2  (2-1)  [Btot] /fiH)  x  /([Hi) x 2 1 K 3  IBt0t] /fiH)  [Btot] /fiH)  (2-2)  (2-3)  (2-4)  where fiH)  =  (1  +  /[Hj 1 K  +  /[H] 1 K 2  +  /[H+I 2 1 K ) 3  (2-5)  and K , K 1 2 and K 3 are the dissociation constants of the titratable groups. The proton concentration is indicated by [Hf], and activity coefficients are omitted. Each (neutral) methylamine that moves across the vesicle membrane consumes a proton as it is reprotonated in the vesicle interior. By charge balance, the final concentration of charged amine in the vesicle interior ([AH] ) can thus be 1 expressed as 1 [AH]  =  (MW]) +2(A[B ]) +3(MW 2 ]) 3  39  (2-6)  where ALB1 represents the change in the internal concentration of the singly charged buffer (i.e. {LBjfmal  -  [Bjinitiai}) as a result of drug accumulation, and so  on. At equilibrium, assuming that the charged form of methylamine does not partition appreciably into the vesicle bilayer, the final inside:outside methylamine concentration gradient must obey the relation [AlI+eq  =  (2-7)  0 [AH+]eq  where the subscripts and  refer to the inner and outer environments,  0 and [H]>> Ka, where Ka is the dissociation respectively. Further, if [H] constant of methylamine (pK 10.6), then 0 [AH]  =  LA]tot  -  /V [AWl 1 (V ) 0 1  (2-8)  1 is the interior where [A]tot is the total (initial) concentration of methylamine, V (trapped) volume of the LUVs and V 0 is the external volume. Thus, combining equations 2-7 and 2-8, we obtain 1 0 [Al-l]q [H+1 ([A]tot  -  (2-9)  ) 1 /V [AH+]eq 1 (V ) 0  By substituting Equations 2-1 to 2-5 in Equation 2-6 the value of [H+]e can be calculated by an iterative process. A second factor which can reduce the internal proton concentration results from the initial efflux of protons from the acidic interior to the exterior environment to set up an opposing electrical potential (A’I’). This equilibrium is described by the Nernst relation A’P  =  -RT/F(ApH)  40  (2-10)  where R is the gas constant, T the temperature and F is the Faraday constant. The number of proton equivalents required to set up this equilibrium can be estimated from the membrane capacitance  (2-11)  Q=CAmAP  where  Q  is the charge in coulombs, Am is the area of the membrane and C is the  membrane capacitance. The number of protons released is given by N(Hi  =  Q/e  where e is the elementary unit of charge (1.6 x 10-19 coulomb). 2.4 RESULTS 2.4.1 VESICLE CHARACTERISTICS The vesicle systems used most commonly in this thesis are 100 nm diameter vesicles prepared by hydrating powdered lipid in 300 mM citrate, pH 4.0, (to make MLVs) followed by 5 freeze-thaw cycles in liquid nitrogen (to make FATMLVs) and extrusion through 100 nm pore size filters (to make LUVs). The amount of entrapped buffer is a critical parameter in these investigations, so the distribution of the citrate buffer was examined at each step of the vesicle preparation. Repeated cycles of freezing and thawing in liquid nitrogen increases the level of encapsulation of 300 mM citrate buffer in multilamellar vesicles (Fig 2-1) in a timedependent fashion. This increase in encapsulation is consistent with the observations of Gruner (1985) and Perkins et al. (1988) that MLVs do not have equilibrium solute distributions and with those of Mayer et al. (1985a) that freeze thawing can increase encapsulation efficiency. It can be seen that the vesicles must be maintained in liquid nitrogen for a surprisingly long time (>3 mm) in order for the citrate to becomes maximally entrapped in FATMLVs. This likely reflects the ability of these high concentrations of citrate buffer to act as a cryoprotectant (data not shown). Because of the higher solute entrapment of FATMLV-derived extruded vesicles,  41  0.60  _L  0.50  0.40  a.  0.30  I -  0.20  1  / -  I,  Co 0.10 C)  I  o.oo 1  4  3  2  5  Freezing Time (mm)  Figure 2-1 Entrapment of 300 mM FATMLVs as a function of freezing times  Citrate Buffer in  EPC MLVs were prepared m 300 mM citric acid pH 4 0 containing 10 tCi/mL []C-citrate as a solute marker. Vesicle 14 samples (in plastic “cryovials’) were plunged into liquid nitrogen and thawed in a 40° water bath for five cycles of the indicated freezing times. Unencapsulated buffer was removed by centrifugation and the amount of solute entrapped determined (see Methods)  42  Table 2-I. [ C1-g1ucose and C1-citrate 4 14 volumes of 100 nm [ diameter extruded vesicles. [14 1-Glucose Space (jiL/mol)  Lipid Composition  [1 4 C]-Citrate Space (iL/,mol)  Egg PC  1.78  ±  0.4 (n=8)  1.50  EPC:Cholesterol (55:45 mol%)  0.98  ±  .28 (9)  0.841  ±  0.1 (3)  ±  0.09 (3)  The amount of entrapped [‘ C1-glucose was determined after 4 incubating vesicles at 45° with 10 Ci/mL [‘ Ci-glucose. Citrate 4 buffer space was determined by preparing vesicles in 300 mM citrate, pH 4.0, freezing in liquid nitrogen for 5 five mm cycles and extruding through 100 nm filters. The number of experiments is given by n. vesicles were always subjected to freeze-thawing cycles (>3 mm  freezing time)  before extrusion. As seen in Table 2-I, the resulting 100 nm diameter vesicles have  maximal aqueous trapped volumes (as reflected by [‘ C1-glucose encapsulation) of 4 1.8 L/mol lipid for EPC vesicles and 1.0 L/mol for EPC/cholesterol (55:45, mol:mol). The amounts of citrate buffer entrapped in these vesicles corresponds to 1.5 and 0.84 L/mol lipid respectively. These values for aqueous volumes and buffer entrapment are used in the simulations presented in Chapter 2 and 3. It should be noted that the citrate buffer leaks only very slowly from these vesicles, even in the presence of the imposed pH gradient (greater than 80 % retention after 2 h at 60° for EPC vesicles incubated with a transmembrane pH gradient, results not shown). Other characteristics of these extruded vesicles are indicated in Hope et al., (1985). As indicated earlier, extruded lipid vesicles prepared from MLVs show greatly reduced solute entrapment compared to those prepared from FATMLVs. It is possible that these different degrees of solute entrapment result from the differences in the initial solute distribution of MLVs and FATMLVs, or from  43  2  0  E 0 0  I  I 0  a  I  0  0  0 0  200  400  800  -  800  Filter Pore Size (nm)  Figure 2-2. Dependence of Solute Trapping on Initial Solute Distribution. Solute entrapped per !mol lipid after extrusion, where 1 H]3 mulin was added to preformed FATMLVs (•) or MLVs (A). Alternatively, the marker was present only inside the FATMLVs (0) or MLVs (t.) before extrusion. Also indicated is the amount of solute entrapped when vesicles have an equal initial solute distribution (0). 44  physical differences in the resulting extruded vesicles themselves.. For example, MLV-derived extruded vesicles could contain internal lamellae which are not present in FATMLV-derived vesicles. To distinguish between these possibilities, the effects of initial solute distribution on the amount of encapsulated solutes were examined as a function of the filter pore size for both MLV-derived and FATMLV-derived extruded vesicles (Fig. 2-2). Vesicles prepared from MLVs and from FATMLVs behave similarly during the extrusion procedure. In both cases, solute equilibration between the intravesicular and external media does not occur as a result of the extrusion process (Fig. 2-2), despite a more than 10-fold reduction in vesicle size upon extrusion. These observations strongly suggest that the extremely low encapsulation efficiency of MLV-derived LUVs (Mayer et al., 1986a) results mainly from the initial solute exclusion of MLVs and that extruded vesicles prepared from MLVs or FATMLVs are otherwise similar. The majority of the solute lost during extrusion occurs on the first pass through the filters (results not shown). Once these smaller vesicles are formed, they appear to be able to pass through the polycarbonate filters (with a pore size equal to their diameter) without suffering further loss of contents. 2.4.2 MEASUREMENTS OF tpH Measurements of transmembrane pH gradients (zpH) are central to this thesis. Therefore, a first series of experiments were directed towards examining the influence of lipid composition and temperature on the zpH indicated by the transmembrane distribution of 1C-methylamine. 14 Vesicles composed of EPC, [ DPPC (16:0/16:0 PC), DSPC (18:0/18:0 PC) and DAPC (20:0/20:0 PC) alone and in combination with cholesterol (PC:cholesterol, 55:45 mol:mol) were therefore prepared by extrusion. A tpH of 3 units (inside acidic; pH 1  =  4.0, pH 0  =  7.0) was  imposed, as indicated in Methods. Note that the filter pore size used in the 45  experiments of Fig 2-3 was 200 nm, due to difficulty extruding the dispersions of the highly saturated lipids through filters with smaller pore size. Lipid composition can have a profound effect on the zpH reported. Specifically, whereas the zpH reported for the EPC vesicles accurately reflects the applied ApH, little or no apparent pH gradient is detected in the vesicles composed of the saturated lipids DPPC, DSPC or DAPC (Fig. 2-3). In the case of DAPC, the half-time for methylamine accumulation is  >  24 h at 20° C (results not shown). The  apparent lack of the methylamine response to the imposed zpH may initially be attributed to the impermeable, gel state nature of those vesicles at the incubation temperature (21°C), which would be expected to reduce the permeability of the neutral form of methylamine. This interpretation is supported by the results presented in Figs 2-3(b) and 2-3(c). The presence of 45 mol % cholesterol, which eliminates the gel-to-liquid crystalline transition and increases the motion available to previously gel state phospholipids, results in equilibration of methylamine which is nearly complete by 30 mm  for the DPPC, DSPC and DAPC systems (Fig. 2-3(b)).  Alternatively, heating the vesicles to temperatures above their gel-to-liquid crystalline transition temperature (‘Ps) should also result in rapid equilibration. As shown in Fig. 2-3(c), this is the case for the DPPC (Tc 41°C) and DSPC (Tc 58°C) vesicles when incubated at 60°C, where rapid transbilayer equilibration of methylamine is observed within 5 mm. It is interesting to note that the presence of gel state lipid per se does not prevent methylamine equilibration, as methylamine accurately reports the imposed ApH in the DAPC vesicles at 60°C, some 15°C below the T of this phospholipid (Nayar et al., 1989). Further, the C1-methylamine 14 [ distributions report a 3 unit pH gradient for all vesicle types incubated at 60° C and indicate that this ApH is stable for at least 1 h at this temperature in all lipid compositions tested. It is straightforward to show that the percentage of the probe that is accumulated for a given tpH obeys the relation:  46  a—  3  •  2 4.’  1  0 I-I C’,  I  z  0  —--  0  10  20  30  0  10  20  30  3 2 C.,  2:  z  0  C  3 2 1 0 0  10  20  30  Time (mm) Figure 2-3 Effect of lipid composition on methylamine distributions. (A) Methylamine distributions were determined as described In Methods for vesicles containing 300 mM citrate (pH 4.0) extruded through 200 nm ifiters and subsequently incubated in 150 mM NaC1, 20 mM HEPES, pH 7.0 at 21°C. Vesicles were composed of: EPC (.); DPPC (ê; DSPC (A); and DAPC (‘). (B) Methylamine response determined as In (A). but vesicles contained 45 mol % cholesterol. (C) Methylamine response for vesicles (in the absence of cholesterol) incubated at (open symbols) or 60° (filled symbols). The symbols are the same as In part (A). 370  47  0  /o  1 O1PH[PL]V.1  entrapped  1  +  xlOO%  (2-12)  1OAPH[PL]Vi  where [PLI is the lipid concentration and V 1 is aqueous volume per mol of lipid. Thus, under the experimental conditions employed here (2 mM lipid, 1 Ci/mL methylamine) detection of a 3 unit pH gradient involves the accumulation 1 of 67% of the probe, for a trap volume of 1 L/mol lipid. Given the initial external concentration of the radiolabelled MeNH 3 as 1 Ci/mL, or 21 tM (specific activity =  48 mCi/mmol) this indicates a final interior probe concentration of 7 mM. TNBS  assay of the radiolabel confirmed that the probe was of the indicated specific activity (results not shown). As each methylamine accumulated consumes a proton on arrival in the vesicle interior, it is clear that the vesicle interior must be reasonably well-buffered in order that radiolabelled methylamine provides an accurate measure of the initial ApH. This effect is illustrated in Fig. 2-4a, where it is found that at interior citrate concentrations below 50 mM (under iso-osmotic conditions) the accuracy of the zpH reported by methylamine is increasingly compromised for a pH gradient which was initially 3 units. As indicated under Methods, the presence of a pH gradient (inside acidic) across the vesicle bilayer will also induce a membrane potential (z\W; inside negative) due to the efflux of H+ ions. The effective interior concentration of protons [N(H+Hefc lost to establish electrochemical equilibrium can be written as [N(W1-)]eff  =  3.7 x 10 ApHid, for a membrane capacitance C  =  1 pF/cm , where d 2  is the vesicle diameter in cm and ApH is the equilibrium pH gradient. For a 100 nm diameter LUV with a ApH  3.0, this corresponds to a loss of 1.1 mM proton  equivalents of buffering capacity. Thus, the loss of buffering capacity to set up A’I’ would not be expected to compromise the measurement of ApH to the extent caused by the accumulation of the radiolabelled methylamine. This conclusion can  48  0  a)  I’  A  3  .0  0  _4-  1  0.  ---.--------  A  -  2  II  /  t  I,  a) .0 0  1  0. I-’  0) 0  0 10 20 30  0  40 50  300  Citrate Concentration (mM) 0 I—’  a)  3  .0  0  A-  A  /4’_  0.  2  B  C  a)  .0  0  1  :  —  0.  0 0  0  10 20 30 40 50  300  Citrate Concentration (mM)  Methylamine response in vesicles with Figure 2-4. different buffering capacities and osmotic strength. The effects of internal buffering capacity (A) and osmotic gradients (B) on the apparent transmembrane pH gradients and membrane potentials. EPC:Chol (55:45: mol:mol) vesicles (2 mM, 100 nm diameter) were prepared at the indicated concentration of citrate, pH 4.0. These vesicles were incubated for 30 mm with (A) an iso-osmotic NaCI-HBS buffer (pH 7.0) or (B) 150 mM NaC1, 20 mM HEPES, pH 7.0. Buffers contained 1.0 ) to determine ApH 3 ine (21 jtM MeNH 14 [ tCi/mL C]-methylam 1 (•) or 1.0 iCi/mL [‘H1-TPP (26 nM TPP) to determine ‘P (j. M CCCP to speed development The solutions also contained 5.0 1 of tW. The dotted lines represent theoretical curves derived from the model described in Methods. 49  be tested by measuring the induced z’I’ employing labelled tetraphenylphosphonium (( H1-TPP) as a probe of zS’I’. In this regard, as each TPP 3 moves into an LUV in response to i\M’, an H+ ion is released to re-establish AW, so there is a 1:1 stochiometry between TPP+ accumulation and internal protons MeNH + . However, 3 [ H ]TPP+ is available at specific activities “consumed’, as for 3 (39 Ci/mmol) which are nearly 1000-fold higher than those of 3 C]-MeNH As a 4 [‘ . result, under the standard initial conditions of 1 tCi/mL TPP to measure AW, the final interior TPP+ concentration (and thus the concentration of proton equivalents consumed) is only 8.5 tM. Thus, as shown in Fig. 2-4a, measurement of the z\pH induced M’ using [ H]-TPP can be a more accurate method of measuring zpH than 3 the methylamine procedure for low (< 50 mM citrate) internal buffering capacities. It should be noted that errors in determining the amount of entrapped + or in determining the aqueous trapped volume can result in large errors in 3 MeNH the estimates of ApH if zpH is small. An example of this is given by the ApH data of Fig 2-4(b), where the exterior aqueous buffer is maintained as 150 mM NaCl, 20 mM HEPES, while the interior citrate concentration is varied. At lower interior citrate concentrations the LUVs will shrink due to the osmotic imbalance to achieve an equilibrium volume  veq = v 0 /fl 1 (H )  (2-13)  where V and fl indicate the initial interior volume and osmolarity, respectively and 110 indicates the osmolarity of the exterior medium. As a result, less MeNH 3 will be accumulated to satis1,’ the relationship 3 0 ] [MeNH + /[MeNH ] =  /[W1 If no correction is made for the change in volume, an apparent ApH 1 [H+] . 0  (ApHaPP) will be measured which is less than the actual zpH according to the relation ApFIapp  zpH  50  -  log ) 1T ( 1 / 1  (2-14)  This volume correction would apply only to hyper-osmotic (outside) gradients since these vesicles do not swell beyond the maximal values seen in Table 2-I. It is interesting to note that the measured TPP distribution is less affected by vesicle shrinkage. This is likely related to the ability of this probe to partition in the vesicle membrane (see below). The influence of the internal citrate concentration and external (initial) 3 concentration on the levels of internalized MeNH MeNH 3 in 100 nm diameter EPC:cholesterol LUVs at equilibrium and the derived ApH values are illustrated in Fig. 2-5(a) and 2-5(b) respectively. At high initial concentrations of external 3 (10 mM), extremely high levels of internalized MeNH MeNH 3 can be achieved (300 nmol/i.tmol lipid) using internal citrate concentrations of 300 mM. The solid lines indicate the theoretical behavior expected on the basis of the analysis presented in Methods, employing the measured trapped buffer of 0.84 L/mol (Table 2-I) with no adjustable parameters. An important ramification of the observed agreement with theory is that the uptake of any simple weak base can be predicted on the basis of the buffering capacity, or alternatively, that the buffering capacity can be measured by determining methylamine distributions at higher methylamine concentrations. A final situation which would be expected to compromise the accuracy of the 14 technique for measuring zpH concerns the influence of higher [ 3 C]-MeNH internal and external pH values. The equilibrium relation 0 /[Hi 1 [H] 0 ii/IMeNH holds only for LHi [H] [MeNH 0 1 3  <<  =  K. where Ka is the  dissociation constant of the weak base. As the exterior and interior pH approach Ka, the apparent t\pH (zpHaPP) is related to the actual z\pH (pHal) by the relation zpHaPP  =  pHrea1  -  log (1  +  Ka/LH]o)  (2-15).  This would lead to a decrease in the measured ApH as [Hi 0 approaches Ka. More importantly, as the interior pH is raised, the proportion of internalized amine  51  3  A  A 0  0 .0 0 L. 0.  ô  %.  ‘ ..  2  —.  ..  ‘S  •s_. C  0 .0 0 C.  ‘.-::--—_.........  —-_  .  .—_._ .‘-..-  .•__‘---.__  U  A  A  :——  .  .  1• -  0) 0 I  0 2  0.  10  8  6  4  •  External Probe Concentration (mM)  0 C’ 0  I  •  A  B  300  A .-.— —  0  E E  — — — —  200  — — — — —  —  _.  —— ..———‘  0  ————  .  ___  .  _—__.  0. 0 C  -  100  •  ‘ii  ——  f4—.  E  - -  -  -  -  -- -  .  0 0  U  - - - - - - - - -  U  0  2  4  6  I  .  8  I  10  External Probe Concentration (mM)  • Figure 2-5. Effect of external methylamine concentration on methylamine uptake. EPC:cholesterol (55:45; mol:mol) LUVs (100 nm diameter, 4 nm lipid) were prepared in 50 mM, 100 mM (U). 200 mM (.) or 300 mM citrate (A), pH 4.0 (see Methods), and the vesicles incubated with the indicated concentration of methylamine, containing 0.5 Ci/mL CJ-.methylam 14 ( ine. Part (A) shows the effect of probe accumulation on the apparent pH: part (B) indicates the amount of accumulated MeNH . The dotted lines represent the 3 theoretical behavior predicted by the model described in Methods.  52  -  a)  3  ___—  A—__A__-—-—A-  A _-A  0  L.  a.  .  External pH  Figure 2-6 Determination of pH gradients over a range of pH EPC LUVs (100 nm) were prepared in 200 mM citrate, 200 mM MES or 200 mM HEPES. at pH 3.0. 4.0, 5.0, 6.0. or 7.0 (closed symbols). This buffer was exchanged for an external buffer of 150 mM NaC1. 20 mM MES. 20 mM HEPES and 20 mM CHES with the appropriate pH required to maintain a 3 pH unit difference between the internal and external buffers. Transmembrane distributions of I 3 14 C1-M eNH were determined after 30 mm Incubations at room temperature (.); transbi layer distributions of [ HJ-TPP (A) were determined after 30 mm in 3 the presence of 10 tM CCCP. 53  which is in the neutral, membrane permeable, form will be increased, leading to the probability of increased probe leakage during the spin column separation. This wifi also lead to lower measures of ApH. As shown in Fig. 2-6, the influence of these effects becomes noticeable at exterior pH values of 9.0 or higher for 100 nm EPC LUVs exhibiting a ApH of 3 units. As may be expected, detection of the induced AW employing [ H]-TPP is not subject to such limitations and provides an 3 accurate measure of eq (and thus, the equilibrium zpH) at exterior pH values up to 9.5. The results to this point indicate that the C1-MeNH 14 probe in combination L 3 with the spin column procedure provides a convenient and accurate measure of ApH (up to 3 units) assuming that conditions allowing the equilibrium transbilayer equilibration of the neutral form are observed and that the interior environment is sufficiently well-buffered. A further point of interest concerns the magnitude of the ApH which can be generated and measured. In this regard, it is experimentally convenient to work in a region where the maximum probe entrapment, at the maximum ApH, corresponds to 75% or less than the total amount of probe originally present in solution. For larger pH gradients this, in turn, limits the amount of phospholipid that can be employed. To detect a pH gradient of 5 units while only accumulating 75% of the probe would require using only 0.03 mM phospholipid, which could result in lipid phosphorus assay errors after the spin column. An alternative technique, which becomes progressively more accurate at high ApH values, involves equilibrium filtration to separate vesicles from the external buffer (see Methods). As shown in Fig. 2-7, both the equilibrium filtration  and the spin column procedure are in good agreement for imposed ApH gradients from 1-5 units. The maximum pH gradient which 100 nm diameter EPC:cholesterol (55:45 mol:mol) LUVs can maintain is approximately 3.7 units. This is also indicated by the transmembrane distributions of [ H1-TPP (Fig. 2-7). 3 The techniques discussed above relate to measurements of pH gradients in vesicles with an acidic interior. It is of interest to determine whether similar spin 54  5 0 I—’  o o  7  4  7  -  7 7  ‘I  C i  0  o I  C.  -.  i  3. 2 7  0  Applied pH Gradient  Figure 2-7. Measurement of large pH gradients. EPC:chol vesicles prepared in 300 mM citrate, pH 4.0 were incubated in 150 mM NaC1 and 20 mM each of citrate, MES, HEPES, EPPS and CHES buffers with pH values in the range 4.5 to 9.0. The ipH was determined by the transmembrane distribution of I Cj-methy1amine using spun mini-columns (.1 4 or “centrifree” filters (0) after a 30 minute incubation at room temperature. E H1-TPP distributions () werç determined in the 3 presence of 5 M CCCP. as indicated in Methods. The dotted line represents the size of the Imposed gradient. 55  -  0  0  o  -  1  P I  I  21  —  --  I— 1/  çw  C) -  o  I_•_ . . .. I.  .‘  _4  -  0  -1  0  1  2  3  4  Time (hrs)  Figure 2-8. Transmembrane distributions of radiolabelled probes in EPC LUVs for vesicles with a basic interior.  Apparent transmembrane distributions of radiolabelled benzoi c acid (A), acetylsalicylic acid (s), acetic acid (•). and mevalonic acid () were determined employing gel filtration in vesicle s containing 300 mM CHES. (pH 9.0) incubated in 150 mM NaCl, 20 mM MES (pH 6.0) with 0.5 iCi/mL of the indicated probe. Positive (interior) membrane potentials induced in response to these pH gradients were determined by the redistribution of I C 4 ]-thioc yanate in EPC vesicles using the gel filtration ’ separation procedure (v). 56  column procedures can be applied to determine zpH in vesicles with a basic interior employing radiolabelled weak acids as the ipH probes. Such studies were pursued for 100 nm EPC vesicles experiencing a 3 unit pH gradient (pHi =  =  0 9.0; p1-I  C]-labelled benzoic, acetylsalicylic, acetic and 4 6.0) utilizing as probes [-  mevalonic acid. As shown in Fig. 2-8, the measured transmembrane distributions of benzoic and acetylsalicylic acid do not reflect the imposed ApH, whereas the transbilayer distribution of acetic acid significantly underestimates the pH gradient. Mevalonic acid appears a useful indicator of ApH for vesicles with basic interiors, however, a long (2 hr) incubation time to achieve equilibrium is required at 20°C. The most rapid and accurate indication of ApH is given by the membrane potential indicator 11 4 C]-thiocyanate in the presence of CCCP, which gives a transmembrane distribution commensurate with the induced W (inside positive) expected for a 3 unit pH gradient. The response of [ C]-acetate was further examined in order to understand 4 1 the basis of the behavior exhibited in Fig. 2-8. As for MeNH , a logical possibility 3 is that entrapped acetate is released during the spin column procedure. The inclusion of cholesterol or the substitution of DSPC for EPC would be expected to reduce such leakage. As shown in Fig. 2-9(a), C]-acetate 14 provided a much [ improved measure of ApH for vesicles with these lipid compositions. Alternatively, as shown in Fig. 2-9(b), the equilibrium filtration procedure can be usefully applied to achieve accurate measures of ApH even for the EPC system. A final point of investigation concerned the relation between ipH as measured by probes such as methylamine and the induced AW measured by probes such as TPP+. Clearly, in the absence of other factors, the transbilayer concentration gradients detected by MeNH 3 and TPP resulting from a given ApH and induced  iW’,  respectively, should be the same at equilibrium. However, other workers have reported that TPP exhibits a significant membrane-water partition coefficient (Flewelling and Hubbell, 1986a,b). This would be expected to increase the inside outside concentration gradient for TPP+, for a given A’I, due to the small aqueous 57  Cu -D 0  3  I  0.  2 Cu .0  0 0.  1  C) 0  0 1  0  2  4  3  Applied pH Gradient Cu Cu  4-.  a,  ‘-  3  B  —-  C., Cu  2 Cu CU Cu U  —  -—  1  --N  Cu  0)  0  1  0  2  3  4  Applied pH Gradient  Figure 2-9. Determination of pH gradients (interior basic) C]-acetate. employing [1 4  (a) Transmembrane distributions of labelled acetate were determined as in Fig. 2-8 for 100 nm diameter LUVs composed or mol:mol) (a), EPC/cholesterol (55:45; rn); of EPC ne transmembra The (A). mol (b) (55:45; %) DSPC/cholesterol C]-acetate as determined by equilibrium 4 distribution of [‘ These vesicles were centrifugation in EPC (100 nm) LUVs. 20 mM HEPES and CHES, mM 20 NaCI, mM incubated in 150 between 9.0 and values to adjusted pH with the 20 mM MES, tCi/mL 1 0.5 of distribution ne the transrnernbra 6.0 and in Methods. indicated as filters “centrifree” by measured acetate 0.94. of a slope with regression, imear a is line The dotted 58  1500 1200 o  a  —  A  900  t ::  .  ss  External TPP Concentration (mM)  The relationship between transmembrane Figure 2-10. 3 gradients. TPP gradients and MeNH A 3.0 unit pH gradient (interior acidic) was created in EPC/cholesterol (55:45; mol:mol) LUVs (100 nm diameter) containing 300 mM citrate (pH 4.0) in the presence of 5 tM 3 (1) gradients were CCCP. Transmembrane TPP (s) and MeNH determined as a function of the external TPP concentration after a 30 mm incubation at 25°C as indicated in Methods.  59  volume to membrane volume ratio in the vesicle interior (see Section 1.5.2 of Introduction). As shown in Fig. 2-10, behavior corresponding to such effects can be observed in 100 nm EPC LUVs exhibiting a I\pH of 3 units (inside acidic). The measured ratios [TPPi/[TPP] 0 are consistently larger than the i/[MeNH ratios over a wide range of external TPP+ concentrations. [MeNH 0 i 3 However, the increased inside-outside ratios lead to a relatively small overestimate of AW  (—  7 mV), indicating that the effects of TPP partition can be neglected under  the conditions observed in this work. 2.5 DISCUSSION The results presented here give insight into factors influencing the encapsulation of compounds by lipid vesicles and into the validity of measurements of transmembrane pH gradients and membrane potentials in liposomal systems. The major factors considered in the measurement of ion gradients were lipid composition, interior buffering capacity, osmotic gradients, the absolute magnitude of the ApH, the measurement of tpH in vesicles with a basic interior and, with respect to measuring the induced membrane potential, the influence of probe partitioning into the lipid bilayer. These aspects are discussed in turn. The lipid composition can strongly influence the measured ApH determined by +. As illustrated here, liposomes composed of 3 weak bases such as MeNH saturated, gel state, lipids can exhibit apparent i\pH values which are substantially less than the actual gradient. Clearly, some minimum level of permeability of the neutral form of the probe through the membrane is required to allow equilibrium to be achieved within a reasonable time frame. As indicated earlier (see Introduction), amine uptake in response to zpH can be treated as a simple first order process, described by the relation IAH(t)1 1  =  EAH(eq)](1et) where (AH(t)] 1 is the interior  concentration of the amine at time t and [AH(eq)] is the equilibrium interior amine concentration at long incubation times. Uptake data of the type presented in Fig. 2-3 can be utilized to obtain the rate constant (k) associated with the uptake 60  process. Thus, an approximate measure of the mimmum permeability coefficient MeNH required can be determined from the rate constant determined from + for 3 the data of Fig. 2-3(b) for DPPC:cholesterol (55:45) at 20°C. This yields a value k  =  2 x i0 sec’. It is straightforward to show that for unilamellar vesicles, this rate constant is related to the permeability coefficient P of the neutral form of methylamine via the relation P  =  0 is the total external kVoLHio/(AmKa) where V  aqueous volume, Am is the membrane area and Ka is the dissociation constant of 3 MeNH  ). 6 (pK=lO.  Assuming an area per phospholipid of 0.60 nm , this 2  indicates a permeability coefficient for the neutral form of 7 x i0 cm/sec or larger is required. As indicated in Results, a brief incubation at an elevated temperature (e.g. 60°C) increases P for all the systems studied to the extent that equilibrium is achieved within 5 mm, without compromising the ipH. The accuracy of the zpH detected across LUV membranes by radiolabelled probes such as MeNH 3 is a sensitive function of both the interior buffering capacity of the vesicles and measures of the interior trapped volume. As detailed here, interior citrate concentrations of 20 mM or higher are necessary to accurately detect pH gradients of 2 units or higher for C]-labelled 14 MeNH [ 3 (specific activity 48 mCi/rn mol). The need for such high interior buffering capacities can be reduced by using probes with higher specific activity. In this regard, it is often convenient to employ probes of t\ P with higher specific activity (such as TPP), to 1 detect the iI induced in response to zpH as a more accurate measure of ipH. As indicated here, the ability of TPP to partition into the lipid bilayer does introduce a slight overestimate of zW and therefore tpH. This is relatively minor under the conditions employed here, approximately 0.1 pH units. An important general point is that for pH gradients of 3 units or more in 100 nm vesicle systems interior citrate buffering concentrations in excess  —  20 mM are required in order that the  pH gradient is not significantly dissipated by proton efflux required to form zW. Further, the presence of osmotic gradients which lead to vesicle shrinkage can cause significant underestimates of the tpH present. 61  The fourth point of discussion concerns the maximum pH gradients that can be achieved. A major thrust of this work has concerned the accurate measurement of relatively large pH gradients of 3 units or more. The results presented here for EPC/cholesterol (55:45) indicate a maximum ApH of  -  3.7 units, corresponding to a  AW of 220 mV. An inability to generate larger pH gradients and induced membrane potentials is likely due to electrical breakdown of the bilayer (El-Mashak and Tsong, 1985). The filtration centrifugation procedure is a sensitive technique for measuring very large tpH values and results give zpH values close to those using the spin column approach. With regard to the measurement of tpH in vesicles with a basic interior, a suitable probe for use with the gel filtration procedure is not readily identified. Most of the probes investigated (acetic, benzoic and acetylsalicylic acids) are poor indicators of the ApH in such situations, apparently because they leak from the vesicles during the separation procedure. Alternatively, while mevalonic acid does not leak from the vesicles during separation, long equilibration times are inconvenient. More accurate procedures are provided by the filtration centrifugation method or by measuring the A’P induced in response to ApH. The final point of discussion concerns the measurement of the AW induced in response to ApH employing cationic probes such as TPP+, and the influence of probe partitioning into the membrane. As indicated above, this partitioning does not introduce large errors (approximately 0.11 pH units, or 7 mV). However it is of interest to compare the value of the partition coefficient which may be calculated from this data with previous reports. Specifically, employing the formalism developed by Cafiso (Cafiso and Hubbell, 1978a) and Rottenberg (Rottenberg, 1984), a partition coefficient  (=  1 x 10-6 cm can be calculated for TPP in this  EPC: cholesterol LUV system. This is in reasonable agreement with previous values 7 cm in POPC MLVs (Altenbach and Seelig, 1985) and 4x10 of 6x10 6 cm for sonicated EPC vesicles (Flewelling and Hubbell, 1986a,b) using standard techniques. In passing, the underestimate of AW reported by Nakazato al. (1988) 62  employing TPP in cholesterol-containing systems is likely due to a kinetic effect. The small amounts of the protonophore CCCP present in the experiment shown in Fig. 2-4 ensure rapid development of the induced A’P.  63  CHAPTER 3.  DRUG UPTAKE INTO LIPOSOMES IN RESPONSE TO pH GRADIENTS  3.1 INTRODUCTION The technique of loading drugs into liposomes in response to a zpH (inside acidic) provides a dramatically improved method of drug entrapment compared to the conventional procedure of preparing the liposomes in a solution of drug. Drug trapping efficiencies approaching 100% can be readily achieved (Mayer et al., 1986b; Mayer et al., 1990b). Resulting drug:lipid ratios can be an order of magnitude higher than can be achieved using conventional procedures. The ipH loading procedure also markedly enhances drug retention properties (Mayer et al, 1990a). These considerations are of considerable importance in liposomal drug design (Mayer, 1986b). A ApH-loaded liposomal formulation of the anti-cancer drug doxorubicin (Fig. 3.1) is currently in advanced clinical trials (Creaven et al., 1990). The liposomally entrapped form of this drug shows reduced cardiac toxicity compared to free drug, while drug efficacy is maintained or enhanced (Mayer et al., 1989; Mayer, 1990a; Balazsovits et al., 1990). Certain aspects of zpH-dependent drug loading into vesicles remain unclear. For example, several catecholamines, antineoplastic agents and other drugs have been shown to accumulate inside lipid vesicles in response to pH gradients across the bilayer (Nichols and Deamer, 1980, Bally et al., 1988; Mayer et al., 1986b) and also in response to transmembrane potassium gradients (Bally et al, 1988; Mayer et al., 1985c). Since transmembrane potentials generate a ApH and vice-versa, it is not always clear which driving force is responsible for the uptake of a particular compound (see de Kroon et al., 1989; Chakrabarti et aL, 1992; Mayer et al., 1985c; Mayer et al., 1986b; Bally et al., 1988). Further, the relationships between the amount of drugs accumulated into LUVs in response to a ApH, the internal buffering capacity and the residual pH gradient have not been adequately examined. For example, it has been noted that the transbilayer concentration gradients of doxorubicin achieved considerably exceed the residual transbilayer 64  CH3  0  2 NH  Figure 3-1. Structure of doxorubicin. primary amine is 8.6 at 37°C.  -65  The PKa àf the -  proton gradient. (Mayer et al., 1990b). It is of interest to develop a more quantitative understanding of these and other factors which determine the extent of drug encapsulation in liposomes in response to ipH, which is developed in this Chapter for doxorubicin. In addition, it is of interest to establish the generality of this “active” zpH dependent method of loading drugs into lipid vesicles employing a variety of ammo  containing drugs. This was performed for representative examples from several different drug classes including antineoplastics, local anaesthetics and antihistamines. 3.2 MATERIALS AND METHODS 3.2.1 MATERIALS Doxorubicin and epirubicin were obtained from Adria Laboratories of Canada, Mississauga, Ont. while mitoxantrone was purchased from Cyanamid Canada Inc., Montreal, Que. Codeine and pethidine were supplied by Abbott Laboratories Ltd, Downsview, Ont. The radiolabels 17- 14 C]dopamine 17,81 4 C ]- imipramine were obtained from Amersham., while Ibenzene 3 ringH ]-chlorpromazine (23 Ci/mmol), H]-piocarpine, [43 [ H1-propranolol (19 Ci / mmol), Icarboxyl-’ 4 3 C]-lidocaine (48 mCi/mmol), C1-methylamine 14 (46 mCi/mmol) and [14C]-ethanolamine came [ 14 [ from NEN. The Liposome Company, Inc. N.J. provided C]-timolol. 3.2.2 LIPID VESICLE PREPARATION  Unless otherwise stated all experiments were performed using egg phosphatidyicholine (EPC) or EPC:cholesterol (55:45 mol%) vesicles prepared in 300 mM citrate pH 4.0 as described in Section 2.1, including five freeze-thaw cycles in liquid nitrogen. 3.2.3 DETERMINATION OF DOXORUBICIN UPTAKE LEVELS  Vesicle associated doxorubicin was determined (after separation of  66  unentrapped drug) by measuring the absorbance at 480 nm in a 1% Triton X-100 solution, which resulted in vesicle disruption and drug release. At extremely low drug concentrations, internal doxorubicin was measured using C]-doxorubicin 14 [ (0.25 mCi/mL) as a radiolabelled marker. 3.2.4 DOXORUBICIN FLUORESCENCE STUDIES Doxorubicin fluorimetry was performed at 480 nm (emission 590 nm) employing a Perkin-Ehner LS-50 fluorimeter. When doxorubicin is incubated with lipid vesicles with an acidic interior, the fluorescence intensity decreases in a time dependent manner which correlates with drug uptake (see Results). These effects likely reflect the accumulation of drug into the interior monolayer of the LUVs (see Discussion). LUVs which did not exhibit a transmembrane pH gradient had little effect on the fluorescence intensity of the doxorubicin, either at pH 4.0 or at pH 7.0 (data not shown). 3.2.5 [‘ C1-NMR STUDIES 3 Doxorubicin was made up in solutions of 300 mM citrate, pH 4.0, or in 150 mM NaC1, 20 mM HEPES, pH 7.0. EPC LUVs either with or without a pH gradient were added to these solutions to achieve a final concentration of 12 mM lipid and the final volume adjusted to 4.0 mLwith D 0. The proton decoupled [‘ 2 C]-NMR 3 spectra were obtained by employing a Bruker MSL 200 spectrometer operating at 50.3 MHz. Free induction decays corresponding to 62,000 transients was obtained by using a 10 s 62° pulse, a 1 s interpulse delay and a 220 ppm sweep width. An exponential multiplication corresponding to 5 Hz was applied to the free induction decay prior to Fourier transformation. The chemical shift is referenced to external TMS. 3.2.6 CRYO-ELECTRON MICROSCOPY Cryo-electron microscopy was performed as previously described (Frederik et al., 1991). Samples containing the LUVs were placed on a 700 mesh gold EM grid 67  and the excess blotted off with filter paper. The grid was plunged into liquid ethane cooled to -190 °C and transferred to a Gatan 126 cold stage at liquid nitrogen temperatures using a Reichart Jung Universal Cryo-Fixation system. The sample was visualised using a Zeiss EM1OC STEM. 3.2.7 DRUG UPTAKE ‘SURVEY” EXPERIMENTS Large unilamellar vesicles (1 mM lipid) were incubated with the drug (0.2 mM) in  300 mM NaCl, 20 mM HEPES pH 7.5 at 25°C unless otherwise stated. These  ratios were selected such that a redistribution similar to that seen with methylamine would result in approximately 50% of the drug being accumulated inside the vesicles. At various times up to 2 hours, aliquots (100 tL) of the mixture were taken and vesicles separated from unentrapped drug by centrifugation through a 1 mL “minicolumn” of Sephadex G-50. Lipid and drug were quantified as described below. Using radiolabelled methylamine as an indicator of bpH, at least a 3.0 pH unit gradient across the vesicle membrane was measured in the absence of drug for all preparations. 3.2.8 OTHER ANALYTICAL PROCEDURES Pilocarpine, chlorpromazine, timolol, proprariolol, imipramine, lidocaine, ethanolamine and dopamine were quantified using tracer quantities of the ( H]- or 3 14 [ C]-radiolabel. Physostigmine was assayed by fluorescence spectroscopy employing an SLM Aminco SPF 500C spectrofluorometer following solubilisation of the vesicles in 60% ethanol (v/v). The excitation and emission wavelengths used were 305 and 350 nm respectively. Quinacrine, chloroquine, quinine and quinidine were also quantified from their fluorescence using excitation and emission wavelengths of 420 nm and 505 nm; 335 nm and 375 nm; 335 nm and 365 nm; and 350 nm and 390 nm respectively. Vinblastine and vincristine were assayed using UV spectroscopy from their 68  absorbances at 262 nm and 297 nm, respectively, following solubilisation of the vesicles in 80% ethanol. Codeine was also measured by UV spectroscopy at 220 nm in this case after solubflisation in 40 mM octyl-13-D-glucopyranoside. Mitoxantrone was quantified from its absorbance at 670 nm following solubilisation of the vesicles in 2% Triton X-100. Diphenhydramine was assayed by gas-liquid chromatography using a HP 9850 gas chromatograph fitted with a Chromatographic Specialties DB-225 (25% cyanopropylphenyl) capillary column. The helium carrier flow rate was 1 mL min’ and detection was by flame ionization. An internal standard, methylpentadecanoate, was used to quantify diphenhydramine following its extraction from the aqueous sample in diethyl ether and its separation from egg phosphatidylcholine by thin layer chromatography. Trarisbilayer pH gradients were quantified employing the weak base methylamine C-labelled) 14 as described previously (see Chapter 2). ( 3.2.9 KINETIC ANALYSIS  Drug distributions in the presence of a transmembrane pH gradient are described using a four compartment model as indicated in Fig. 3-2, based upon a model developed by CafIso and Hubbell (1978a, 1978b) to describe the behavior (at low concentrations) of several spin-labelled ESR probes of t\pH and  iW’.  The  compartments will be denoted by the subscripts o (regions outside the vesicles), I 0 (in the outer monolayer into which drug partitions), and (the vesicle interior), m m (the inner monolayer into which drug partitions). Under the assumption that only the neutral form of the drug traverses the bilayer, it follows that d[D]tot 0  PA m  dt  0 V  0 ([D]  -  ) 1 [D1  (3-1)  [D1 t 0t is the total exterior concentration of drug (including charged, where 0 uncharged, free and membrane bound species), P is the permeability coefficient of 69  the neutral form, Am is the area of the membrane, V 0 is the external aqueous volume and [D1 0 and ID] 1 are the concentrations of the neutral form of the drug. Assuming that the drug dissociation constant Ka is similar for the free and membrane associated drug, and the membrane:water interface is at equilibrium, 0 can be expressed as: [Dim [D]tot 0  0 [Dim 1/K  +  0 [H]  +  (3-2) /V (Vm ) 0  +  0 [H]  /V (Vm ) 0  Ka  where K is the membrane:water partition coefficient for the charged form of the drug, [H+1 0 is the exterior proton concentration and Vm is the volume of the membrane. Using a solubiity-diffusion model of permeation (Section 1.4.2), under the assumption that [H] >> Ka (since the external pH is 7 and the pKa of 0 0 doxorubicin is 8.6), and that V  >>  Vm, it follows from Equation 3-2 that  d([D] tot)  KDAm  dt  Ka  t0t 0 [D]  dmVo  where D is the diffusion coefficient and dm is the width of the bilayer. This results in the relation [D(t)]otot  t0t ekt 0 [D(eq)]  =  where k is the rate constant associated with the process and k  (34)  =  KD/d Am  Ka/(Vo)[H+]o). t0t 0 As the interior drug concentration must obey the relation [D]  tot 0 ([Dl  -  t0t)Vo/V [D] , 1 we obtain [D(t)]t0t  =  t0t (1et) 1 [D(eq)1  70  (3..5)  DH  DH+  I  S  OW  ‘  S S  Fr Buffer LI  sb\  Vm,  InsIde  Outside  I  -  fpH  Figure 3-2 Model of the interaction of doxorubicin with phospholipid LUVs in the presence of a zpH.  Only the uncharged form of the drug is able to translocate the bilayer. The volumes Vm 1 and Vm 0 into which the charged drug partitions are not drawn to scale. For clarity, arrows indicating equilibria are omitted. 71  where [D(eq)1 is the equilibrium interior drug concentration. Thus, according to this analysis, the rate constant of drug accumulation would be expected to be inversely dependent on the external hydrogen-ion concentration. 3.2.10 EQUILIBRIUM ANALYSIS A model of equilibrium transmembrane drug distributions in the presence of a ApH requires calculation of the number of drug molecules in the outer and inner aqueous compartments (V 0 and V, respectively), and the number of drug molecules in the outer and inner monolayers of the membrane (Vmo and Vmj). Assuming that the membrane-water partition coefficient (K) of the charged form of —-the drug (dcnoted by the superscript ) is the same on the inside and outside of the vesicle K  =  N mo V 0  =  N.V. mi 1  V mo N o  (3-6)  V mi N i  Similarly, assuming that the drug dissociation constant (Ka) is the same in both the aqueous and membrane phase: =  Ka  / (N ) 0 V [H1  =  (Nmi/Vmi)[Hji  (Nmo/Vmo)[Hjo (37)  —  N’V 1 i ‘  N 0 ‘V ‘ 0  N mi ‘V ‘ mi  N mo +/\7 mo  By definition, Nt0t  Ntot 0  =  1 + Nmi + N +Nmi N N++N 0  mo  ++N  0  (3-8)  +N mo  N t 0t are the total number of internal and external drug molecules where N0t and 0  respectively. From relations 3-6 to 3-8 and making the well justified assumption that surface potentials due to the charged drug partitioning into the membrane are small (due to the high ionic strength of the interior buffer) 72  I  Nt0t  Jj  Ky )1  +  Ka  0 V t0t 0 N  \T  1 [Hj  +  +  1 V  KaVi  0 [H1  7 m o [Hi 0 Kao  +  Ka  KVmi  +  +  1  ‘mO) 0 V  Where [Hi >> [H1 1 >> Ka (the conditions usually employed are pH 0 =  7, pK  —  =  4, pH  8.6 for doxorubicin), and at low lipid concentrations, where Vm <<V 0  that the partition coefficient of the drug obeys the relation K  <<  SO  Vo/Vmo, it can be  readily shown that the ratio of vesicle internalized drug to external drug is approximated by: t0t 1 N  (1  =  +  ) 1 KV/V  1 IHi *  —  t0t 0 N  (VO)(  1  +  (39) 0 IFI1  ‘mo )  0 V  We define K* as the apparent bulk partition coefficient of the drug into the membrane where K*  (3-10)  KVmo/Vm  =  Eq. 3-9 simplifies to rni.tot I  Jo tot  (1  +  K*V m V.1  IH+1. *  (3-11) FH+1 1 Jo  [D] t 0t is the total free concentration of drug and 1 [D] t 0t the effective vesicle where 0 associated “concentration” of the drug (that is, the total amount of drug associated with the LUVs divided by the LUV volume). The proton concentration gradient 0 /[H1 can be readily determined by 1 [Hi measuring the transmembrane distribution of radiolabelled methylamine (see Chapter 2), and the amount of accumulated drug ([D1t0t) can be determined in a 73  t0t/[D] 1 [D] t ot to be calculated. In turn, this similar manner, allowing the ratio 0 allows the partition coefficient K* to be calculated using Eq. 3-11. At higher external drug or probe concentrations, the analysis by this model is complicated by the effects of drug accumulation on the pH gradient which causes the accumulation. Obviously, if sufficient amounts of an acid or base redistribute, the pH on at least one side of the membrane (and therefore the pH gradient) will be altered. These effects are discussed in more detail in Chapter 2 for methylamine. If drug accumulation is well “coupled” to the pH gradient (in the sense that non specific proton loss is minimal), both [DH] 1 and [H] 1 can be calculated for a given set of conditions with a knowledge of drug partition coefficient. In particular, as detailed in Chapter 2, the variation of internal pH expected for an internal buffer of 300 mM citrate as a function of methylamine uptake can be reasonably modelled using the equations of Section 2.1. 3.3 RESULTS 3.3.1 KINETICS OF DOXORUBICIN UPTAKE DETERMINED BY FLUOROMETRIC TECHNIQUES. In previous studies of drug accumulation into LUVs with acidic interiors, assays of entrapment have involved the separation of entrapped drug and subsequent assays of entrapped material (Mayer et aL, 1990b). For kinetic studies, particularly on systems exhibiting rapid uptake, a more convenient method of monitoring doxorubicin accumulation was to monitor doxorubicin fluorescence. This is illustrated in Fig. 3-3 for uptake into EPC:cholesterol (55:45) LUVs (100 nm diameter) in response to a transmembrane pH gradient (pH 4.0, pH t 7.0). A 0 small increase in fluorescence is observed on addition of LUVs to the doxorubicin solution (200 M doxorubicin) which is followed by an exponential time-dependent decrease in fluorescence to an equilibrium value. (The small initial fluorescence increase is not detectable in cases where the drug is accumulated more rapidly). The kinetics of fluorescence decrease correlate well with doxorubicin accumulation 74  130  o .  E  _  ‘_,  ./  o  150.  .-  -.  •_  D  --  105  /.  100 •.•  ::  ><  o  .  Ol 0  “  20  10  .-30 30  Time (mm)  Figure 3-3 Doxorubicin accumulation into LtJVs in response to transmembrane pH gradients  EPC:cholesterol vesicles (100 nm diameter) with an interior 300 mM citrate buffer, pH 4.0, were added tà a doxorubicin. containing solution (200 [IM final drug concentration) at pH 7.0 and the amount of vesicle assàciated doxorubicm was determined as in Methods (•). Doxorubicin association with vesicles without a transmembrane pH gradient is also shown at pH 4 (D) and at pH 7 (tJ. Also shown is the fluorescence decrease of doxorubicin incubated with vesicles in the absence of a transmembrane pH gradient (line). The temperature in the cuvette was 33.4°C, the lipid concentration was 1.5 mM and external pH was 7.0. .  75.  0 ii..  assayed by the chromatographic separation procedure. Both data sets can be fitted using the first-order kinetic analysis indicated in Methods, with a rate constant (k) of 3x10 3 3.3.2 pH DEPENDENCE AND ACTIVATION ENERGY OF DOXORUBICIN ACCUMULATION INTO LUVs  The kinetic model developed in Methods predicts that the rate constant of doxorubicin uptake should be proportional to the external proton concentration for the transbilayer movement of the neutral form. Thus a plot of log k vs the external pH should be a straight line with a slope of unity. The rates of doxorubicin uptake into EPC and EPC:cholesterol LUVs were determined over the pH range pH 0 to pH 0  =  =  5.7  8. As shown in Fig. 3-4(a), plots of the log of the rate constant derived  from this data show the expected linear dependence on pH, with slopes of 1.08 and 0.96 for EPC and EPC:cholesterol LUVs respectively. This provides strong evidence for the transbilayer movement of the neutral form of the drug. Rates of doxorubicin accumulation appear to be markedly temperature dependent, given the drastically enhanced rate of uptake at 60° compared to 20° (Mayer et al., 1990b). Rate constants derived from uptake data over the temperature range 5°C to 55°C demonstrate high activation energies as shown in the Arrhenius plots of Fig. 3-4(b). It is interesting to note that different activation energies are observed for different lipid systems, where Ea EPC:cholesterol (55:45) LUV system, whereas Ea  =  =  38 kcal/mol for the  28 kcal/mol for the EPC  systems. 3.3.3 PARTITION COEFFICIENTS AND COUPLING CHARACTERISTICS ASSOCIATED WITH DOXORUBICIN UPTAKE As indicated in Methods, an ability of doxorubicin to partition into the lipid bilayer can result in inside/outside drug concentration ratios which significantly exceed the inside/outside drug concentration ratios. In particular, a plot of 0 vs [H]th/[H]OUt should reveal a straight line with a slope of [drug]/[drug]  76  0 —1 07  0  -2  -3 -4 6.00  6.50  7.00  7.50  8.00  External pH —1  -3 C  -5 -7 -9 3.00  3.15  3.30  3.45  3.60  l/T (IC ) (xl000) 1  Figure 3-4. Effects of external pH and temperature on the kinetics of doxorubicin accumulation.  (A) EPC (.1 or EPC:cholesterol (.) vesicles prepared in 300 mM citrate pH 4.0 were incubated with 200 i.tM doxorubicin at pH 7.0 and the effects of pH on the rate constant of accumulation determined by monitoring the rate of fluorescence change as in Fig. 3-3. The temperature was 21°C with EPC vesicles, or 53°C with EPC:cholesterol vesicles. (B) The effect of temperature on the rate constant of doxorubicin accumulation were monitored by fluorescence changes (closed symbols) or spin columns (open symbols) for EPC (•) or EPC:cholesterol LUVs (•) as indicated in Methods. -  -  77  K*Vm/Vi. Drug uptake into 100 urn EPC:cholesterol LUVs (2 mM lipid) exhibiting a 3 unit pH gradient (pH 4 in, pH 7 out) containing 300 mM citrate was therefore examined over initial exterior doxorubicin concentrations up to 14 mM. Corresponding interior proton gradients were determined using 1C-rnethylamine. 14 L A plot of the interior/exterior drug concentration ratios vs the residual proton gradient is shown in Fig. 3-5(a), revealing a linear dependence with a slope of 24. A 100 nm diameter LUV is expected to have Vm/Vj  =  0.37, assuming a bilayer  thickness of 5 nm (see Section 1.3.6), leading to an estimated doxorubicin partition coefficient K* of 65. It is of interest to compare this value of K* with that determined by more classical procedures. This requires an estimate of the amount of drug binding to the vesicles with no applied ApH. A filter centrifugation technique was used to estimate the partition coefficient of doxorubicin for EPC:cholesterol LUVs at pH 4.0 for a range of lipid concentrations and yielded an estimated K* of 74 for EPC:cholesterol vesicles (results not shown). The data of Fig. 3-5 can also be employed to determine how well doxorubicin uptake is coupled to the interior buffering capacity, which provides a measure of the non-specific leakage which may be induced by drug accumulation. As indicated under Methods, for a well “coupled’ system, the values [H+] 1 and [drug1 after drug uptake can be calculated from a knowledge of the partition coefficient K*, the initial drug concentration arid the other vesicle parameters as described in Chapter 1. A plot of the theoretical doxorubicin and pH gradients vs the observed values is shown in Fig. 3-5(b). It may be noted that the theoretical and observed values agree reasonably well, suggesting that little non-specific leakage has occurred from these vesicles. 3.3.4 [‘ C] NMR STUDIES ON DOXORUBICIN UPTAKE 3 It is straightforward to calculate that for a K* value of 64, over 95% of the doxorubicin accumulated into 100 nm LUVs will be partitioned into the inner 78  1600 0  0  1200 BOO  I  400  0  0 0  40  20  60  [MeNHJJ[MeNHJt 4  4  3  3  0 ..  C  >< 0 C, 0  —  0.  —  2  21i.. 1 0  0  -  0  10  15  Drug. Concentration  Figure 3-5. RelatIonship between the residual pH gradients and doxorubicin accumulation Equilibrium doxorubicin and methylamine concentration gradients were determined by incubating EPC:cholesterol vesicles containing 300 mM citrate, pH 4.0 at 600 for 20 mm with the indicated concentration of doxorubicin. (A) The relationship between the equilibrium concentration gradient of doxorubicin and the residual pH gradient, as determined by the concentration. gradient of methylamine. The slope of the linear regression (solid line) is 24. (B) The dotted lines were generated using the model presented in Methods, using the parameters determined in Chapter 1 and the apparent partition coefficient determined in (A) above. Experimental results presented are equilibrium doxorubicin (.) or methylamine (.) concentration gradients. 79  aL b  -—  -  -.-  —  -  c.  LJJL  d  eILkL 200  -  i80  10  i40  020  000  pp  8C  60  4C  20  Figure 3-6. Effect of transmembrane pH gradients on the ”C]-NMR spectra of vesicles incubated with doxorubicin. 1 [ Natural abundance C1 13 NMR spectra of: a) free doxorubicin (4 [ mM) at pH 7.0: b) EPC vesicles; c) doxorubicin encapsulated in EPC vesicles in response to a transmembrane pH gradient: d) doxorubicin incubated with EPC vesicles without a pH gradient (pH 4 inside and outside the vesicles): and e) as in (d). but at pH 7 inside and out). Data collected with the assistance of K.F. Worig.  80  monolayer of the LUV bilayer. As a result, it would be expected that the motional properties of the accumulated drug would be restricted in comparison with the external free drug. This was tested by monitoring the [‘ C1 NMR behavior of 3 doxorubicin before and after uptake. In order to detect the natural abundance  C] NMR spectrum arising from doxorubicin within a reasonable time frame, 3 [relatively high drug concentrations (4 mM) must be employed. In turn, this requires higher LUV concentrations (12 mM) in order that 90% or more of the available drug is accumulated in response to the EpH. Therefore, in these samples there Is a much higher level of background binding (in the absence of a zpH) than in the previous results presented. Again, essentially all of the drug present is encapsulated in those vesicles with a transmembrane pH gradient. The I 13 C] NMR spectra arising from free doxorubicin and EPC alone are  shown in Figs 3-6a and 3-6b respectively. Additional resonances arise from the HEPES and citrate buffers. Interestingly, essentially no difference can be detected between the C]-NMR 13 spectra of EPC vesicles alone (3-6b) arid EPC vesicles which [ have accumulated doxorubicin in response to a bpH (3-6c). This is perhaps most clear for those resonances between 100 and 200 ppm, and may indicate that the doxorubicin is immobilized in the bilayer. Incubation of doxorubicin in the presence of LUVs without a pH gradient (Fig 3-6d and e) reveals a broadening of the doxorubicin resonances, particularly at pH 7but this broadening does not result in complete signal disappearance. It should be noted that the doxorubicin sample incubated at pH 7 (inside and out) was clumped and aggregated after 24 hours of signal accumulation, which may account for some of the signal reduction of this sample.  3.3.5 MORPHOLOGICAL FEATURES OF LUVs FOLLOWING DOXORUBICIN ACCUMULATION The experiments to this point show that doxorubicin can be accumulated into LUVs to high levels in response to a ApH and that this behavior can be understood  81  Figure 3-7. Cryoelectron microscopy of doxorubicin-free and doxorubicin loaded vesicles.  with a vesicles EPC:cholesterol diameter nm 100 transmembrane pH gradient were incubated at 600 for 15 mm in the (a) absence or (b) presence of 200 iM doxorubicin as detailed in Methods. The bar represents 150 nm. Photographs taken with the assistance of J.J. Wheeler. 82  on the basis of a model whereby the large majority of the encapsulated drug is associated with the inner monolayer of the vesicles. This raises questions concerning the inner monolayer. For example, assuming a doxorubicin cross2 (half the area of PC), this would still correspond to sectional area of only 0.3 nm an increase in surface area of 30%. The LUV morphology following doxorubicin uptake was investigated employing cryo-electron microscopy. An internal structure bisecting the length of the vesicles is commonly observed in vesicles which have accumulated doxorubicin, resulting in a “coffee-bean” appearance (Fig. 3-7(b)). This feature is absent in control vesicles which are not loaded with drug (Fig. 37(a)). 3.3.6 DRUG UPTAKE “SURVEY”  It is of interest to compare the tpH-dependent behavior of doxorubicin with other drugs. The ApH response of a variety of drugs examined under “standard” conditions (1 mM 100 nm diameter EPC vesicles, internal buffer 300 mM citrate, pH 4.0, 200 .tM drug, 20°) is summarized in Table 3-I. Four drug categories can be defined on the basis of their uptake characteristics: first, drugs which show partial but stable uptake; second, drugs which show partial uptake and then release; third, compounds which do not redistribute in response to a proton gradient; and finally, compounds which show essentially complete accumulation. While these categories can be expected to encompass a continuous spectrum of uptake behavior, in the majority of cases the assignment of a drug to a particular category was clear. Representative examples from these four classes of uptake will be discussed in turn. In the absence of pH gradients (pH 4 inside and outside the vesicles, or pH 7 inside and outside the vesicles), only low background levels of drugs were associated with the lipid for all compounds tested (results not shown). 3.3.7 CLASS 1. DRUGS WHICH EXHIBIT PARTIAL BUT STABLE UPTAKE. The uptake of timolol, an example of this class of drugs, is shown in Fig. 3-8. After 30 mm, timolol is taken up to about 100 nmoles/imo1e lipid (about 50% of 83  Table 3-I. Extent and stability of accumulation of various drugs by vesicles exhibiting a pH gradient (acidic interior)  Uptake at 15 mm (nmol /!Amol lipid)  Drug  Class  Mitoxantrone 4 4 Epirubicin 4 Daunorubicin Doxorubicin Vincristine Vinblastine  Antineoplastic  Uptake at 2 hr (nmol /imol lipid)  200 201 200 202 178 175 1  198 200 204 203 130 127  Lidocalne Local anaesthetics “ Chlorpromazine Dibucaine  87 98 194  87 96 176  Propranolol Timolol  Adrenergic antagonists  198 95  187 97  5 Quinidirie  Antiarrythmic agents  203  74  Cholinergic agents  <1 <2  <1 <1  1802 80  177 78  182  188  6 Antihistamine Diphenhydrainine  1761  87  Quinine 5 Chloroquine  Antimalarial  1481 1041  81 88  5 Quinacrine  Antiprotozoan  2  71  “ “ “ ‘I  It  Pilocarpine Physostigmine  “  Dopamine Serotonin  Biogenic amines  Imipramine  Antidepressant  Codeine  Footnotes:  “  Analgesic  <1  <1  1, maximum uptake taken at 5 mm; 2, maximum uptake taken  at 30 mm; 3, maximum uptake taken at 90 mins; 4, data collected by L.C.L. Tal, 5, data collected by M.B. Bally, 6, data collected by C.P.S. Tilcock.  84  -  200  C.  o E —  o E  •Q  150  100 I -  .  50  0  50  100  Time (mm)  Figure 3-8. Uptake of timolol by EPC vesicles. Timolol (200 M) was incubated with vesicles exhibiting a proton gradient (pH 4.0 in/pH 7.5 out) (•): or with control vesicles with no pE-I (pH 4.0 in/pH 4.0 out)(’. or (pH 7.5 in/pH 7.5 out)(D). -  -  85  0. 0  E  =  —.  4) C  0. 0,  C  4,  0• 0, C) 0  E  C  50  100  Time (mm) •0 0. 0  3  EPC:CHOL 200  o  o  E C) C •0 C  2  150  0. 0,  100  C) ‘1  0• 0,  50  4) 0  E  C  0— 0  0 50  100  Time (mm)  Figure 3-9. Uptake of qulnldlne by EPC and EPC:cholesterol vesicles. Quinidine (200 tM) was incubated with 1 mM EPC (a) or EPC:cholesterol (b) (55:45 mol:mol) 100 nm diameter vesicles. Drug uptake into the vesicles (0) and residual zpH () were determined as detailed in Methods.. -  86  -  available drug. This represents a 650 fold concentration gradient (interior/exterior), in reasonable agreement with the measured residual proton gradient (2.7 pH units). Other drugs which exhibit partial but stable uptake are quinacrine, chiorpromazine, lidocaine, ethanolamine, serotonin and chioroquine. 3.3.8 CLASS 2. DRUGS WHICH EXHIBIT PARTIAL UPTAKE AND SUBSEQUENT RELEASE. Qulnidme uptake into liposomes having a ApH is shown in Fig. 3-9. Initially, EPC vesicles rapidly accumulate virtually all available quinidine under the standard conditions employed. This uptake is unstable and within 30 mm  about  50% of the drug has been released from the vesicles (Fig. 3-9(a)). This release is not associated with any apparent structural change in the vesicles, such as fusion or aggregation, although the drug appears to cause a decay of the pH gradient. Prior to the addition of quinidine, 1 3 C]-MeNH distributions indicate a stable 4 pH gradient of greater than 3 units, which rapidly dissipates upon addition of the drug. It appears that the high internal quinidine concentrations generated increase the permeability of the membrane, leading to dissipation of the pH gradient. To test this hypothesis quinidine uptake was examined in vesicles composed of EPC:cholesterol. Cholesterol reduces the destabilizing effect of quinidine on the bilayer as shown in Fig. 3-9(b). Following quinidine uptake there is an initial decrease in the pH gradient (due to net proton binding by drug accumulated within the vesicles) but the level of drug accumulation and the residual pH gradient are then stable over the two hour incubation period. Other drugs which are released from egg phosphatidyicholine vesicles following uptake  are quinine, diphenhydramine, vinblastme and vincristine. The leakage rates vary considerably. Vesicles loaded with vincristine and vinbiastine lose about 27% of initially sequestered drug over two hours. As would be expected, this loss is associated with a corresponding reduction in residual ipH as determined using methylamine. A similar decrease in the proton gradient is observed as quinine and diphenhydramine are released from egg PC vesicles. 87  3.3.9 CLASS 3. DRUGS WHICH EXHIBIT NO APPARENT RESPONSE.  Drugs in this class include codeine, pilocarpine and physostigmine. The lack of uptake of these compounds may be explained if the compounds cause a major Increase in membrane permeability resulting in complete dissipation of the pH gradient. This is not the case, however, at least for physostigmine and codeine (results not shown). Under the conditions used to assess drug uptake (200 tM drug) only a small decrease in measured tpH was observed. At least in the case of physostigmine, the neutral form of the drug may simply not be sufficiently permeable that zpH accumulation can be detected. 3.3.10 CLASS 4. DRUGS WHICH ARE TOTALLY ACCUMULATED. A final set of drugs showed complete accumulation in response to the imposed ApH, including doxorubicin, dau norubicin, epirubicin, propranolol, dopamine, dibucaine and imipramine. Accumulation of these drugs in EPC vesicles was generally rapid and complete, with little or no release observed over two hours. In the absence of a pH gradient (vesicle interior and exterior both pH 4.0 or pH 7.5), only low levels of background binding are observed, representing less than 5% of the drug present (results not shown). 3.3.11 PARTITION COEFFICIENTS AND COUPLING OF OTHER AMINES It was of interest to ascertain whether the behavior of the different drugs seen  in Table 3-I reflects different values of the partition coefficient K*. The equilibrium levels of drug uptake and the residual pH gradients (indicated by methylamine distributions) were therefore examined and the partition coefficients (K*) calculated  employing the same procedures as used for doxorubicin. The partition coefficients obtained range from 0 to> 200 and are summarized in Table 3-Il, along with their octanol:water partition coefficients. For compounds which showed sufficiently slow accumulation, rates and activation energies of transport were also determined. Plots of the residual drug and pH gradients, as well as the theoretical behavior expected for a “well-coupled” system using these estimated K* values are shown in 88  Lidocaine  C  Tim ol 0)  a” • a” 2 •  0.  :  a a  —  —  •  4-  a  0  a  a  0  0 4  2  0  (  I  3  4  a  IS  0.  a  2  C  a I  0  4  2  I  4  Drug Concentration  Drug Concentration  Ethariolamine  Dopamine  IC  \ I  r.  a  —  x  “  0.  —  2  ‘b’•’  “  a  0.  —  1•  _2  a  ‘.-‘--.-.-----  “  __.-_____._ •  C  a  -“  -t  1 a  01  0  “-.___  0 0 0  4-  2  4  I  ...,..  0  IC  0  2  4  •1  I  I  10  Drug Concentration  Drug Concentration  I mip ramin e  Dibucaine  4  3  ‘  1.  o ‘2  a  a. a  ‘  a.’  C  —  —  2  —  a  a  •  0  a  —  C  ai  0  —  ‘  4-  I.  —  •  z  a  .-  a’  za.  £2  • a  0  2  4  I  8  10  0  Drug Concentration  -  Z  -  4-  1  I  IC  Drug.Concentratlon  FIgure 3-10. Relationship between residual pH gradIent and drug accumulation in vesicles with a transmembrane pH gradient. -  -  -  EPC:cholesterol (55:45 mol%) vesicles prepared in 300 mM citrate, pH 4.0 were incubated with the indicated drugs at, pH 7.0 and equffibrium pH gradients (.) and drug gradients J) were determined as described -in Methods. The dotted lines are simulations based on the models presented in Sections 2.1 and 3.1, using the partition coefficients shown in Table 3-Il, assuming no non-specific proton leakage. -  89  Table 3-11 Partition Coefficients of Drugs Examined  Drug  Lidocalne Ethanolamine  Drug Class  K*  Local Anaesthetic  0  0  5  0 0  -  Octanol: Water Partition Coefficient  Timolol  13-blocker  4  Dopamine  Biogenic Amine  50  800  Doxorubicin  Anti-cancer  64  13  Imipramine  Antidepressant  30  39,000  170  25,000  Dibucaine  Local Anaesthetic  90  Fig. 3-10. In general, drug uptake and residual pH appear to be reasonably well predicted by theory, indicating that the internal pH is coupled to drug uptake. Imipramine and dibucairie are exceptions, in that the uptake of these compounds are poorly modelled by these calculations. 3.4 DISCUSSION 3.4.1 DOXORUBICIN ENTRAPMENT IN RESPONSE TO A zpH The results presented for doxorubicin clarilr the mechanism of accumulation of the drug into LUVs with an acidic interior, demonstrate the high level of drug partitioning into the inner monolayer in the loaded systems and explore the consequences of high levels of drug accumulation into the membrane both in regard to permeability and morphological consequences. With regard to the mechanism of uptake, the kinetic studies demonstrating a linear dependence of the rate constant of accumulation on the external proton concentration unambiguously establish that the neutral form of doxorubicin is the transported species. This conclusion is in accord with Frezard and Gamier Suillerot (1991) who examined doxorubicin movement in the absence of transmembrane pH gradients. While consistent with the large body of literature showing that the neutral forms of weak acids and bases are generally the membrane permeable species (Rottenberg, 1979; Rottenberg, 1989), these conclusions are important for two reasons. First, results from at least two groups have been previously interpreted to implicate transbilayer diffusion of the charged (protonated) species of several weak bases (Bally et aL, 1988; Bally et a!., 1985; de Kroon et al., 1989), in part because the ApH apparently could not account for the amount of internalized drugs observed. (Mayer et al., 1986b; Mayer et al., 1985c). Secondly, in combination with the observed activation energies, the kinetics of drug accumulation and release properties can now be quantitatively described. Among other applications, this has utility in the design of liposomal drug delivery systems, particularly with regard to loading and regulated release properties. For example, 91  under the loading conditions employed to obtain the data of Fig. 3-3, the half-time of uptake is 3 mm; one can then obtain a generalized rate constant for doxorubicmn uptake as  k (T, pH , [PL]) 0  =  10 [PL]exp 62.3 (1-307/T) 51 2 x (pH7)  (3-12)  where T is in °K and [PLI is the lipid concentration in mM. From this relation it can be calculated that at 60°C, pHo  =  7, uptake is extremely rapid, as k  4 sec), while at 20°C uptake is much slower, k  112 (t  =  4 (t 1x10 112  =  =  4 2x10  110 mm).  It is interesting to note that after rapid entrapment at an elevated temperature, subsequent release at room temperature will be considerably slower, even though the effective lipid concentration for liposomally entrapped drug is considerably higher (—300 mM for a 100 nm LUV). Thus, a maximal rate constant of k  5 (t 3-8x10 112 18 hr) would be expected for doxorubicin efflux from  =  EPC:cholesterol LUVs at 20°C, pH 1 4.0 assuming that the pH gradient is dissipated  and that all drug which leaks out is immediately removed from the vicinity of the LUVs. The high activation energies associated with the transbilayer permeation of the neutral form of doxorubicin are also of interest. Briefly, previous studies on the ApH dependent transport of phospholipids which are weak acids (Redelmeier, et al, 1990; Eastman et aL, 1991) and (derivatized) peptides which are weak bases (Chakrabarti et aL, 1992) have demonstrated high activation energies, in the range  of 30 kcal/mol. Similar results are shown here for doxorubicin.  Activation  energies can be rationalised on the basis of the number of hydrogen bonds which must be broken and not reformed upon passing into the hydrocarbon (Lieb and Stein, 1971), multiplied by the energy associated with hydrogen bond breaking (—35 kcal/mol), or on the basis of molecular size. It is straightforward to see how activation energies in the range of 30 kcal/mol could be achieved for doxorubicin given its large size and its high potential for hydrogen bond forming. It should be  92  noted, however, that the observed activation energies are likely overestimates. This is because the PKa of doxorubicin Is strongly temperature dependent (Frezard and Garnier-Suilerot, 1991). If the pK of doxorubicin decreases by 0.5 pH units over the range 5° to 37° (Frezard and Garnier-Suilerot, 1991), the Ea determined would overestimate the actual activation energy by approximately 7 kcal. The increase in Ea observed in EPC:cholesterol vesicles implies that the presence of cholesterol in the bilayer adds a considerable barrier to doxorubicin movement. The results of the equilibrium uptake studies show that the extremely high levels of internalized doxorubicin achieved can be accounted for by the ability of doxorubicin to partition into the LUV membrane in a ApH-dependeril manner. Given the high ratio of membrane to aqueous volume of 100 nm vesicles (see Table 1-TI of the Introduction) any significant membrane partition would result in drug uptake levels considerably higher than those predicted if partitioning effects. It is somewhat surprising that the lipid bilayer can maintain a permeability barrier when one monolayer is exposed to effective drug concentrations in the range of 300 mM. Such concentrations could lead to detergent effects, or to transbilayer packing differentials, either of which could be expected to cause membrane disruption. As indicated in Methods, these levels of drug association would be expected to lead to at least a 30% expansion of the inner monolayer, which is difficult to reconcile with normal LUV bilayer structure. Doxorubicin encapsulated in LUVs in response to pH gradients results in a change in vesicle morphology, as determined by cryo-electron microscopy. It is not certain what the ‘line ’ running through the centre of vesicles which have t accumulated doxorubicin (see Fig. 3-7) represents. The fact that this morphology is not observed for LUVs experiencing a pH gradient in the absence of drugs suggests that it arises from accumulation of doxorubicin and its partition into the inner monolayer. As noted earlier, the inner monolayer is expected to be forced to undergo a significant increase in surface area, which may not be compatible with normal LUV structure. The observed morphology may therefore result from excess 93  lipid “blebbed” from the inner monolayer to reduce the surface area of the inner monolayer. Alternatively, since some of these compounds may be above their solubifity in the internal media (Madden et al., 1990), drug precipitation may occur in addition to partitioning. These possibilities are presently being examined. It is interesting to note that the value of K* obtained for doxorubicin here is much larger than would be expected on the basis of its octanol/water partition coefficient (—13), but it is similar to the value determined for DMPC bilayers (—40) (see Burke and Tritton, 1985). The model described in Methods can also be used to predict drug distributions for smaller or larger vesicles in a similar way. For spherical vesicles at low lipid concentrations (where K*Vm/Vo <<1), Eq. 3-11 can be written as 1 [DH] ] 0 [DH  =  1  +  K*  3 (r  -  ) 3 (r-d)  3 (r-d)  (3-13) [Hjo  where d is the bilayer thickness and r is the vesicle radius. Thus the drug concentration gradient will exceed the pH gradient by a factor related to the geometry of the vesicles for the range of LUV diameters commonly used (30-200 nm). In addition, reduced vesicle radius also reduces the internal buffering capacity. Thus, the observations of Mayer et al., (1990b) that there was apparently no relationship between the residual pH gradient and the amount of doxorubicin accumulation can be understood on the basis of equation 3-10 and the fact that the amount of internal buffer varies with vesicle size. 3.4.2 zpH ACCUMULATION OF OTHER COMPOUNDS It is clear from Table 3-Il that the ability of pharmaceutical agents to accumulate within lipid vesicles exhibiting a proton gradient is a fairly general phenomenon (Table 3-I). While the majority of drugs examined redistribute in response to a ApH it is interesting to note that the extent and stability of uptake varies considerably. Timolol, lidocaine, chlorpromazine, serotonin, chloroquine 94  and quinacrine redistribute in rough agreement with the pH gradient, while doxorubicin, dopamine. dibucaine and imipramine, among others, attain gradients considerably in excess of the proton gradient. The behavior of other drugs can also be rationalised on the basis of the partitioning model employed. Drugs such as lidocaine or timolol have small partition coefficients and dissipate the pH gradient at approximately the same external concentration as a comparable amount of the probe methylamine (see Chapter 2). The transmembrane concentration gradients therefore mirror the proton gradients. In contrast, drugs such as dopamine, imipramine or dibucaine which have larger partition coefficients are accumulated to a much greater extent (at the same initial external concentration). Thus their apparent transbilayer concentration gradients considerably exceed the residual proton gradient and the residual pH gradient of the vesicles is dissipated at a lower external concentration of drug than if the drug has a low partition coefficient. Finally, it is instructive to compare the behavior of the compounds studied here with that of compounds with extremely large partition coefficients, such as certain phospholipids and fatty acids. It has been previously shown that these molecules redistribute in accordance with the proton gradient (Redelmeier et al., 1989). This is consistent with Eq. 3-9, where the observable parameters are the number of molecules on each side of the bilayer, that is:  Nm  =  1 Vm  [H]  (3-14)  0 Nm  If [H 1 and IH 0 ]>>Ka for these molecules, their partition coefficients are 1 sufficiently high that the compounds reside entirely in the membrane and their transbilayer distributions again reflect ApH.  95  CHAPTER 4. RATES AND ACTIVATION ENERGIES OF PROTON FLUX ACROSS LIPID BILAYERS 4.1 INTRODUCTION As mentioned in Chapter 1, reported values of the proton-hydroxide permeability 1 net of unmodified lipid bilayers extend over a very wide range. This is mainly because the initial rate of proton flux is largely independent of the gradient which drives the flux (see Figure 4-1 and also Nichols and Deamer, 1989; Gutknecht, 1984; Perkins and Cafiso, 1986; Gutknecht, 1987b; Deamer, 1987; Perkins and Cafiso, 1 987b). Hence, there is an strong dependence of calculated net values on the size of the proton gradient imposed.  Much of the reported  discrepancies in the values reported for net can be accounted for by differences in the pH, and the pH gradient, as well as the lipid composition employed in the various studies (Perkins and Cafiso, 1986). Proton flux also appears to be sensitive to low levels of chloroform (Perkins and Cafiso, 1986), lipid oxidation (Perkins and Cafiso, 1986) anaesthetics (Barchfeld and Deamer, 1988; Rarnes and Cafiso, 1990),  and alcohols (Barchfeld and Dearner, 1985). Furthermore, there are reports that the direction of proton flux affects proton permeation (Norris and Powell, 1990),  and that the buffering capacity of the lipids must be included in any analysis of proton permeation which relies on the measurement of ApH (Grzesiek and Dencher, 1986). The independence of initial flux with respect to the size of the imposed proton gradient implicates some form of carrier mechanism in proton movement across lipid bilayers (Deamer, 1987). The physical nature of this carrier is unclear, though it appears to be uniquely available to protons (Deamer, 1987). In planar lipid bilayers, Gutknecht has produced considerable evidence that low levels of fatty acid contaminants are responsible for the majority of proton flux (Gutknecht, 1987b). In planar bilayers, the sensitivity of proton flux to albumin (which binds fatty acids), to phioretin (which is believed to alter the membrane dipole potential)  and to exogenous fatty acids suggests that most of the proton conductance is due 96  to low levels of fatty acids acting as protonophores (Gutknecht, 1 987b). In lipid vesicles, while electrogenic proton movement remains relatively independent of the size of the proton gradient, it is not altered by the addition of phioretin or long chain fatty acids (Perkins and Cafiso, 1 987a). Apparently two separate phenomena are being examined in the two systems (Perkins and Cafiso, 1987b). These arid other results seen in liposomal systems are not consistent with the rate limiting step of proton flux being the simple diffusion of a charged species (Perkins and Cafiso, 1987b; Deamer, 1987). One probable mechanism of proton flux across vesicle bilayers involves coupling to water movement (see Nagle, 1987; Nichols arid Deamer, 1989 for reviews). However, there is presently no definitive evidence to support a water-based mechanism. Structured water has not been found in membranes (Conrad and Strauss, 1985). Indeed, such evidence would be difficult to find, given that the structured water chains or “wires” would be expected to exist only transiently (Nagle, 1987). In addition, replacing H 0 with D 2 0 does 2 not appear to affect proton flux across the bilayer (Perkins and Cafiso, 1 987b). Theoretically, current-voltage relationships could shed light on the mechanism of proton permeation, (Nagle, 1987), but both linear and superlinear current-voltage relations have been described for liposomal systems (Perkins and Cafiso, 1986; O’Shea et al., 1984; Deamer, 1987; Krishnamoorthy and Hinkle, 1984). Activation energies often help to elucidate underlying mechanisms. Relatively few studies have examined the effects of temperature on rates of proton movement except at the lipid phase transition temperature, where there is a discontinuity in Arrhenius behavior (Bramhall, 1985; Elamrani and Blume, 1983). Kinetic measurements of proton movement in lipid vesicles are most often made by monitoring pH sensitive probes located in a poorly buffered vesicle interior. This  has the practical disadvantages that the interpretation of results is complicated by the titration of internal buffers and that non-electrogenic movement of protons cannot be distinguished from electrogenic flux (Cafiso and Hubbell, 1983). Additionally, these experiments may be poorly suited to situations in which the 97  temperature is varied, due to the extreme sensitivity of these experiments to temperature dependent changes in the pK of the reporter molecule. Since the pKa of many compounds decreases with increasing temperature, it is possible that the reported values of the activation energy of proton flux are overestimates. An alternative approach is to monitor the development of a membrane potential across vesicles in response to a ApH where protons are the only relatively permeable ion (Cafiso and Hubbell, 1983). In this case, only electrogenic proton movement is detected. Additionally, one can monitor the development of a pH gradient in response to a ‘I’, as utilised here. 4.2 MATERIALS AND METHODS 4.2.1 PREPARATION OF LIPID SAMPLES Cholesterol, cholesterol sulfate and phloretin were incorporated (where used) in lipid samples by co-lyophilization from benzene:methanol (70:30 v/v). Palmitic acid and stearylamine were added to the liposome mixture from a concentrated stock solution in ethanol. Vesicle preparation was otherwise conducted as indicated in Chapter 2.1.  4.2.2 GENERATION OF pH GRADIENTS Unencapsulated buffer was removed employing Sephadex 0-50 gel filtration columns equilibrated with 3 mM of the internal buffer at the initial pH. Transmembrane pH gradients were imposed by diluting the vesicles 8-fold (to a final lipid concentration of approximately 2 mM) in buffer solutions containing the appropriate ApH or zP probes and ionophores. The external buffer pH was not significantly altered upon the addition of the vesicles. Where employed, the potassium ionophore valinomycin was used at a concentration of 0.5 pg per mol lipid and the protonophore CCCP was used at a concentration of 10 tM. The  external buffer contained 0.5 Ci/mL of the appropriate membrane potential probe (TPP for negative interior potentials; SCN for positive interior potentials). Unless 98  otherwise stated, the external buffer was 100 mM 4 S0 20 mM HEPES (pH 7.0) 2 K , S0 20 mM MES (pH 6.0). 2 K , or 100 mM 4 4.2.3 GENERATION OF MEMBRANE POTENTIALS K diffusion potentials were created by entrapping 100 mM 4 S0 1 mM 2 K , HEPES, pH 7.0 in vesicles as indicated above. The external buffer was then exchanged for 100 mM 4 SO 20 mM HEPES, pH 7.0 containing 0.1 mM 4 2 Na , S0 2 K using Sephadex 0-50 columns. The membrane potential was generated by the 10fold dilution of the vesicles into an identical external medium which also contained 0.5 ig valinomycin per jtmol lipid. 4.2.4 DETERMINATION OF PROBE UPTAKE INTO LUVs Vesicle associated probe was separated from free probe by gel filtration chromatography employing 1 mL syringes which had been previously filled with Sephadex 0-50 (50-150) equilibrated in the appropriate external buffer as indicated in Section 2.1. Aliquots of vesicles (0.1 mL) were loaded and immediately eluted by centrifugation at 500g for at least 3 mm. The associated timing errors are estimated to be as high as 30 sec. 4.2.5 CALCULATION OF MEMBRANE POTENTIALS AND pH GRADIENTS Quantitation of vesicle associated probe was performed by liquid scintillation counting and phospholipid analysis as previously described (Section 2.1; Redelmeier et al., 1989). Briefly, membrane potentials can be calculated by membrane potential probe distributions according to:  z’IJ(mV)  =  RT -  ) 0 log(C/C  (4-1)  F  where C 1 and C 0 represents interior and exterior TPP+ or SCN- concentrations and R, T and F have the definitions used in Section 1.5. The pH gradients were estimated in an analogous manner employing methylamine (see Chapter 1). 99  4.3 THEORETICAL CONSIDERATIONS  If the membrane is described as a simple capacitor of capacitance C, and there is an approximately linear current-voltage relationship, as determined by Perkins and Cafiso (1986), the kinetics of development of a membrane potential of voltage V in response to an applied ApH obey the relation  Q  =  CV, where  Q  is the  charge on the capacitor (corresponding to the number of protons which have crossed the membrane). From the relation I dV  dQ/dt  =  =  V/R, it follows that  V  =  dt  (4-2)  RC  where R is the specific membrane resistance and C the specific membrane capacitance (approximately 1 iF/cm ). 2 This has the solution V  where k  =  =  Veq (1- ekt  (43)  1 /RC, where k is the rate constant associated with the buildup of  the voltage. The product of k, the first order rate constant and the membrane capacitance, yields the conductance. Thus, measures of EM’ following imposition of a ipH were fitted to the equation  1 A’P(t)  =  A’P(eq)i (1  -  et)  (44)  where zW(t) is the accumulated probe at time t and A’IJ(eq) is the equilibrium amount of accumulated probe (determined by adding 10 tM CCCP to allow equilibration of protons). A 7 mV correction factor was included to account for the small overestimate of z’I’ due to the partition coefficient of TPP (see Figure 2-10, Section 2.5). The generation of a z\pH in response to an imposed AP is slowed by the presence of internal buffers. A set of equations has been derived to describe the 100  effect of proton flux on the interior pH of vesicles containing buffers in the absence of membrane potentials (Whitmarsh, 1987). Such equations could potentially be used to model the systems in Figure 4-5b, but the solution is difficult for a triprotic buffer such as citrate. 4.4 RESULTS 4.4.1 DETECTION OF PROTON FLUX The movement of uncompensated charge across a lipid bilayer generates a transmembrane electrical potential. In Fig. 4-1, transmembrane pH gradients were created across the vesicles, and the subsequent development of a membrane potential can be measured by examining the transmembrane distributions of the lipophiic cations TPP (internal negative potential) or SCN- (internal positive) as indicated in Figure 1. In each case, a potential near 180 mV develops with a half time of 12 mm in EPC vesicles at 25°C. The interior of these LUVs are sufficiently well buffered that their internal pH does not change significantly due to proton influx or efflux to establish the membrane potential (see Chapter 2). There are several lines of evidence that the rate of development of the measured membrane potential in these systems is limited by proton movement rather than rates of equilibration of the TPP and SCN- probes. First, low (10 tM) concentrations of the proton ionophore CCCP dramatically increase the rate of development of zSI’ and the addition of 100 mM ammonium acetate (which depletes the transmembrane pH gradient) causes the rapid dissipation of the membrane potential (Figure 1). Second, TPP has a permeability coefficient of 10 to 10-8 cms’ and an activation energy of about 20 kcal/mol in EPC vesicles (Flewelling and Hubbell, 1986a). These rates are 2-3 orders of magnitude faster than the rates seen here, while the activation energy is about 8 kcal/mol higher than observed here (see below). Third, if valinomycin is used to induce a membrane potential in response to a transmembrane pH gradient, TPP+ distributions reflect a 180 mV  101  > E  0  30  60  -  90  120  Time (mm)  180 >  2 120  60  0 0  30  60  90  120  Time (mm)  Figure 4-1. Generation of membrane potentials in response to 3 unit acidic or basic pH gradients. 100 nm EPC vesicles containing 300 mM citrate, pH 4.0 (panel a) or 300 mM CHES pH 9.0 (panel b) were prepared as described in Methods. A 3 unit zpH was imposed across the vesicles by incubation in 100 mM 4 S0 20 mM HEPES, pH 7.0 2 K , containing 0.5 i.tCi/mL [ H]-TPP (panel a); or 100 mM 4 3 S0 2 K , 20 mM MES, pH 6.0 containing 0.5 iCi/mL C]-SCN 14 (panel [ 10 1 M CCCP was present in samples indicated by (s). b). Ammonium acetate (100 mM final concentration) at the external pH was added at the time indicated by the arrows. Also indicated is the development of membrane potential in EPC:cholesterol systems (•). Solid lines represent fits to the data using Equation 4-3. -  102  potential within 2 mm  or less (results not shown). Finally, micromolar  concentrations of the ion-pairing agent TPB- do not increase the apparent rate of potential development monitored by TPP (results not shown). 4.4.2 INFLUENCE OF LIPID COMPOSITION ON PROTON FLUX Proton flux rates across the membranes of EPC LUVs at 25°C can be fitted as a first order processes with rate constants of 9x10 4 1 s (half-time approximately , 12 mm) (Fig 4-la and 4-ib). This corresponds to an initial proton current of 100 , similar to that obtained by Perkins and Cafiso (1986) for large 2 picoampere/cm unilamellar vesicles. The rate of development of the  A’1’  is much slower in  EPC:cholesterol (55:45 mol%) LUVs (t 112 of 70 mm) (Figure 1). The effects of cholesterol content on proton flux in extruded POPC:cholesterol LUVs are indicated in Fig. 4-2. These results are compared with data determined by Koenig et al (1991) for water movement through similar vesicles. This provides a direct comparison of proton versus water flux for the same lipid preparations. Increasing membrane cholesterol content from 0 to 40 mol % decreases water and proton permeation rates approximately 8-fold (Fig. 4-2). The effects of lipid composition and exogenously added compounds such as fatty acids, stearylamine and phioretin on the rates and activation energies of proton flux are also indicated in Table 4-I. Briefly, the addition of fatty acid, stearylamine, or phloretin do not appear to influence the rate of electrogenic proton movement greatly. Higher concentrations of amines and fatty acids diminish the transmembrane ApH, as determined by the distribution of [1 4 C1-methylamine (results not shown). Inclusion of cholesterol and cholesterol sulfate in the bilayer decrease the rate of proton movement (Table 4-I). 4.4.3 ACTIVATION ENERGIES ASSOCIATED WITH PROTON FLUX A central objective of this investigation was to examine the effects of lipid composition and other factors on the activation energy (Ea) of proton flux (Figure 43; Table 4-I). Activation energies of 11±2 kcal/mol (40-50 kJ/mol) can be 103  4.’  .2  le-03  5e+03  8e-04  4e+03  6e-04  E 1  4)  :2  3e+03  -.  ‘c 5e-04  o  2e÷03  “S  A..  ‘.  20  Cholesterol (mol%)  Figure 4-2. ComparIson of the effects of cholesterol on proton and water flux. 100 nm vesicles containing the indicated amounts of cholesterol were prepared as in Figure 4-1(a). The rate constants of proton flux (•) are indicated as a function of cholesterol content at 25 °C. Also indicated are comparable data for water flux reported for these vesicles by Koenig et al. (1990) (A) 104  E 4.’  8  -7 U)  o o  -8  A  a  4-’  -  0..  7 1  -9 \S  3.00  3.50  3.25  5  3.75  4.00  ) 1 l/T x 1000 (K  Comparison of the effects of temperature on Figure 4-3 proton and water flux 100 nm POPC:chol (55:45 molô/o) extruded vesicles with a 3 unit ipH were prepared as in Methods and the rate of development of a membrane potential was determined as in Fig. 4-1, as a function of temperature (•). Also indIcated are comparable data for water flux reported for these vesicles by Koenig et al. (1990)  105  Table 4-I. Proton Flux Rates and Activation Energies.  Lipid System  zpH (in:out)  k (25°)  Ea (kcal/mol)  EPC  4/7  4 9.4x10  12.8  EPC  9/6  4 8.8x10  12.1  EPC:chol(55:45)  4/7  4 1.3x10  11.3  EPC:chol-Sulfate (55:45)  4/7  1.4x10  11.3  EPC:chol (55:45)  6/9  4 2.5x10  12.4  EPC:chol (55:45)  9/6  4 1.5x10  11.4  EPC:chol:Phloretin(50:45:5)  4/7  4 2.1x10  12.6  EPC:chol:(55:45) + 2 iM Palmitic Acid  6/9  4 2.6x10  11.  EPC:chol:(55:45) + 2 1 M Stearlyamine  9/6  4 2.6x10  11.8  DMPC:Chol (55:45)  4/7  1.1x1O  9.9  DPPC:Chol (55:45)  4/7  1.Ox1O  10.1  Transmembrane pH gradients (3 units) were imposed across well buffered vesicles of the indicated composition as indicated in Methods and the time course of development of a AP was examined using [ H1-TPP 3 14 (inside basic) as a function of [ (inside acidic) or C]-SCN temperature.  106  calculated from Arrhenius plots of the rate constants associated with proton flux (Figure 4-3). The Ea of proton flux determined for POPC vesicles containing different amounts of cholesterol (up to 40 mol%) ranged from 12 to 14 kcal/mol (data not shown). The Ea determined for water flux in these vesicles by Koenig et al (1991) was in the range of 11 to 16 kcal/mol. The activation energies determined were independent of the direction of the imposed pH gradient and the range of pH examined within the limits of accuracy of the technique (Table 4-I). While both cholesterol and cholesterol sulfate slow the rate of proton flux, they do not markedly affect the activation energy. 4.4.4 DETECTION OF PROTON FLUX IN RESPONSE TO MEMBRANE POTENTIALS  Proton movements can also be detected by creating an initial electrical potential by the addition of valinomycin to vesicles with a transmembrane K gradient and measuring the subsequent development of a transmembrane pH gradient (Redelmeier et at, 1989). The time course of the development of the pH gradient is complicated by the effects of internal buffers, which slow this equilibration. Early reports (Hope et al., 1985; Redelmeier, 1989) indicated that the resulting pH gradient does not appear to come to electrochemical equilibrium within the 4 hour time course of the experiments using EPC vesicles. In addition, the combination of valinomycin and CCCP resulted in the rapid (<5 mm) decay of both the tpH and zW due to an increase in the Na ion permeability of these membranes (Redelmeier et al., 1989). In contrast, the pH gradient and membrane potential are stable over several hours in the presence of both of these ionophores if the membrane is composed of EPC:cholesterol, rather than EPC alone (Fig.4-2). The Ea of proton movement measured by examining the initial rate of accumulation of j1 4 C1-methylamine in response to an electrical potential is 11 kcal, approximately the same as that determined for the development an electrical potential in response to a pH gradient (results not shown).  107  E 0 a  3  180  120  0  a.  C I-  60  .0  E S  0 0.00  0 0.50  1.00  1.50  2.00  Time (hours)  Generation and stability of pH gradients in Figure 4-4. response to transmembrane K+ gradients. A 1000-fold K gradient was imposed across 100 nm EPC:cholesterol (55:45 mol%) vesicles which contained 1 mM H1-TPP 3 HEPES buffer pH 7.0. The AW was monitored using [ using the measured was of the ApH development the while 4, [1 C1-methylamine (•). The open symbols refer to distribution of 4 samples which contain 10 iM (final concentration) CCCP. 108  180  A  --  3  180  3 -.  A4  120  2  ApH (units)  (my)  A’fr  2  ApH (units)  (my)  60  0 0.00  120  60  -_  0.50  1.00  1.50  0  0  U  2.00  0  1  Time (hours)  2  3  Time (hours)  C  180  2  120  120  ApH (units)  (mV)  D  180  3  2  .  (my)  -  .-----—  — —  000  I  I  I  0.50  1.00  1.50  n  0  -  0  2.00  Time (hours)  1  I 2  Time (hours)  Figure 4-5 Vesicles with stable membrane potentials but no pH gradients and vice-versa  Vesicles were prepared in 100 mM 4 S0 1 mM HEPES pH 7 0 2 K and incubated in 300 mM sucrose, 10 mM Tris, 10 mM HEPES, pH 7 0 contatnmg 1 iig valinomycm to mduce a membrane potential (panels (a) and (c)) or prepared in 300 mM citrate 10 mM 4 S0 pH 4 0 and incubated in 100 mM 4 2 K S0 20 mM 2 K HEPES pH 7 0 containing 1 .ig vahnomycm (panels (b) and (d) Membrane potential (.) and pH gradients I) were determined by the distribution of TPP and methylamine, respectively. Panels (c) and (d) also contained 10 uM (final concentration) CCCP. Data for (a) and (c) used with permission of T E Redelmeier 109  ApH (units)  60  60  3  I 3  — U  4.4.5 LUVs EXHIBITING A AW OR pH, BUT NOT BOTH A final area of investigation is the development of model membrane systems which have a stable membrane potential, yet exhibit no detectable transmembrane  pH gradient and vice-versa. LUVs with these (non-equilibrium) properties can be used to examine the relative contributions of iI1 and ApH to a variety of transport processes, such as those described in Chapter 3. Tris, a weak base, dissipates induced transmembrane pH gradients when present at 10 mM in the external buffer with LUVs exhibiting a K diffusion potential (Figure 4-5a). The nearly 170 mV (inside negative) AW which is established in the presence of valinomycin is stable (Figure 4-Sa) and close to that observed in Fig 4-3. However, no AW-induced zpH can be detected by MeNH 3 distributions. This observation has been previously documented by Sone et al., (1980), who also proposed that Tris is transported in the neutral form and dissipates the pH gradient upon reprotonation in the vesicle interior. Similarly, the presence of valinomycin/K+ for vesicles having an imposed pH gradient prevents the generation of a transmembrane electrical potential. If these vesicles are sufficiently well buffered, the transmembrane pH gradient can be maintained for several hours (Figure 4-5b). Even after 6 h, a gradient of more than 2 pH units can be detected, with no detectable z’I’ (results not shown). In both cases, the stability of the imposed gradients is expected to depend upon the net flux of proton equivalents. For example, about 225 nmol/iimol lipid of proton equivalents must move into the vesicles to negate the applied K gradient in Fig 45(a), and 900 nmol/jtmol lipid of proton equivalents are required to titrate the interior buffer in Fig 4-5(b) from pH 4.0 to 7.0. Since the initial proton flux is about 2x10’ 2 mol/sec per imol lipid, (corresponding to a current 100 pA/cm , 2 assuming a surface area of 2000 cm 2 / tmol lipid), the dissipation of these gradients can be expected to take as long as 100 h. Low levels of the proton ionophore CCCP cause the time dependent dissipation of the potential and the pH gradient (Figure 4-Sc and 4-5d). CCCP also 110  leads to electrochemical equilibrium for all time points.  4.5 DISCUSSION The uncompensated movement of protons across liposomal membranes generates a transmembrane potential, at a rate which depends upon the flux of protons or proton equivalents. As shown in Table 4-I, proton conductance as determined by lipophilic ion distributions is independent of pH (i.e. the size of the proton gradient), consistent with the observations of many previous workers that the value of the apparent proton permeability coefficient varies with the experimental conditions. Furthermore, in contrast to the conclusions of Norris and Powell, (1990) (who examined the effect of proton movement on the internal pH of lipid vesicles), electrogenic flux of protons does not depend upon the direction of the pH gradient. The results presented here employ LUV systems, whereas electrogenic proton flux has been well characterized by CafIso and coworkers for smaller (1-2 unit) pH gradients around pH 7 in sonicated vesicles (SUVs) of about 25 nm diameter. They have shown a linear current-voltage relationship (at least for 2 unit transmembrane pH gradients) and that there is little effect of phloretin and fatty acids on electrogenic proton flux in SUVs. It is possible, however, that proton flux  in SUVs is anomalous due to their extremely small size and highly curved nature. For example, proton movement in SUVs differs from that seen in planar lipid bilayers (Perkins and Cafiso, 1987b), and is about an order of magnitude slower in sonicated vesicles than LUVs of identical composition (Perkins and Cafiso, 1986). However, the experiments shown here indicate that SUVs and LUVs exhibit qualitatively similar characteristics with respect to proton movement, as indicated by the lack of effects of phioretin and fatty acids seen in Table 4-I, though the initial proton current is larger in the LIJVs. With regard to the activation energies associated with proton flux, there have been three previous reports measuring the Ea of proton movement through 111  unmodified lipid bilayers which are not at the phase transition temperature. One report determined an Ea of proton flux of 9.6 kcal/mol (Grzesiek and Dencher, 1986). In two other studies, the high activation energies reported (17-20 kcal) were interpreted as supporting the hypothesis that permeation of protons proceeds via hydrogen-bonded water molecules. However, reported activation energies of water flux through bilayers are generally somewhat lower (8-15 kcal/mol) (Finkelstein, 1987). The similarity of the activation energies of water and proton movement across bilayers suggests that the processes may be related. Indeed, the effects of cholesterol content on proton and water flux also appear to correlate well (Fig 43b). This provides supportive, but not conclusive, evidence of a relationship between water and proton flow. The lack of equilibration between zpH and z\’P previously reported (Redelmeier et at., 1989) for vesicles having transmembrane potassium gradients (in the presence of valinomycin) was puzzling in light of the high permeability coefficient of protons, generally given in the range of i0 3 to i0 cm/s. Even in the presence of internal buffers (pK between 4 and 7) at up to 100 mM, an induced ApH in equilibrium with the imposed &-P would be expected within seconds using these permeability coefficients. In fact, an internal buffer as poor as 1 mM HEPES is sufficient to retard the development of t\pH (Fig 4-3). These results emphasize the importance of the experimental conditions employed to monitor the apparent permeability of protons. Additionally, it should be noted that the shape of the ApH curve generated in response to A’I’ cannot be adequately modelled using an adaptation of the equations described in Whitmarsh (1987), describing the effect of internal buffers on the rate of change of pH on the internal pH of vesicles, regardless of the value of k (the rate constant) chosen. This is possibly due to the non-specific movement of other species (ions or buffers) during the long time course of these experiments, likely augmented by the presence of valinomycin. 112  The markedly different behavior of EPC and EPC:cholesterol vesicles in the presence of the combination of valinomycin and CCCP is also surprising (Fig.4-4). The combination of valinomycin and CCCP cause the rapid dissipation of AM’ and ApH in EPC vesicles by increasing membrane permeability to Na ions, but the presence of cholesterol prevents this decay. It can be calculated that the increase in Na permeability required to destabilise the AM’ can be quite small: from the data of Redelmeier et al. (1989), the Na permeability coefficient in the presence of valinomycm and CCCP is on the order of 10-12 cm/s in EPC LUVs. One of the ramifications of a significant proton flux is that vesicle preparations with a AM’ will generally develop a ApH and vice-versa, as protons reach their equilibrium. The results illustrated in Figure 4-5 demonstrate model membrane systems which exhibit a stable pH gradients and no observed membrane potentials and vice-versa. External concentrations of Tris rapidly redistribute across the bilayer in response to ApH in a similar manner as many other weak bases. This prevents the formation of a large pH gradient. In an analogous manner, the valinomycin/K+ combination prevents the development of an induced AM’ in vesicles with an imposed ApH, while the high buffering capacity of the vesicle interior allows the pH gradient to be relatively stable. The systems demonstrated in Fig.4-5a and b are also useful in distinguishing AMJ-dependent from ApH-dependent transport processes and are also potentially useful tools for determining whether agents can act as protonophores, since an increase in net proton flux leads to dissipation of the electrical potential and pH gradient. As an example, incubation of weakly basic drugs with a 180 mV A’P, but no detectable ApH (the vesicle systems shown in Fig. 4-5a) results in much lower levels of drug accumulation than seen in response to 3 unit ApH, consistent with drug accumulation being in response to ApH, as opposed to AM’ (T. Redelmeier, 1989b).  113  CHAPTER 5.  SUMMARIZING DISCUSSION  This thesis has been focussed on the measurement of ApH and  iM’  in LUV  systems, the development of realistic models to describe the uptake of these probes and lipophiic amines into LUVs exhibiting a zpH, and finally on the flux of protons through bilayers. With regard to the first area, for LUVs with an acidic interior, determination of the equilibrium transbilayer distributions of radiolabelled methylamine employing a gel filtration procedure provides a reliable procedure to measure ApH provided that transbilayer equilibration rates are sufficiently rapid, and that interior buffering capacities are sufficiently high. In situations where this accuracy is compromised, equilibrium centrifugation techniques or techniques to measure membrane potentials induced by the pH gradient provide straightforward alternatives. The expected intra-vesicular pH can be reasonably predicted after probes are accumulated utilizing several parameters: vesicle diameter and volume; vesicle and probe concentrations; amount of encapsulated buffer and buffer pKa(s); and the initial internal and external pH. In the second set of studies, the uptake of the anti-cancer drug doxorubicin into large unilamellar vesicles (LUVs) exhibiting a transmembrane pH gradient (inside acidic) was investigated in considerable detail, using both kinetic and equilibrium approaches. It was shown that doxorubicin accumulation into the vesicles proceeds via permeation of the neutral form of the drug. The critical dependence of translocation rates on pH and lipid composition suggest ways to manipulate drug loading and release in a predictive manner. The extent of drug accumulation at equilibrium was analysed in terms of a model which incorporates drug partitioning into the interior monolayer of the vesicles and also takes into account the influence of internalized drug on the interior buffer. The extent of accumulation can be rationalised on the basis of the partition of the drug in the vesicle interior, with a partition coefficient of 64 estimated for EPC:cholesterol bilayers. For a 100 nm vesicle, this indicates that more than 95% of the 114  encapsulated drug is partitioned into the inner monolayer. The accumulation of doxorubicin appears to be well coupled to the pH gradient.  The behavior of a  number of other compounds was examined in terms of the above model. Finally, the drug trapping efficiency and high drug:lipid ratios achieved using ApH liposome loading are of considerable practical value. The conductance and activation energy of the transbilayer movement of proton equivalents were examined for a variety of lipid systems with large (3 unit) pH gradients. Lipid vesicles (100 nm diameter) with well-buffered interiors generated transmembrane potentials which could be monitored using the lipophilic membrane potential probes TPP or SCN-. Development of the potential had a half-time of about 12 mm  in EPC LUVs at 25°C, with an activation energy near 11  kcal/mol. The rate and activation energy of proton movement were not affected by the addition of phloretin, fatty acids, or stearylamine, or by direction of the imposed pH gradient. The incorporation of cholesterol and cholesterol sulfate into the LUV bilayer decreased the rate of proton flux. Proton movements were also examined in vesicles with transmembrane potassium diffusion potentials, by monitoring the development of a ApH using methylamine. Finally, model membrane systems were developed which exhibited stable membrane potentials without induced pH gradients, or stable pH gradients without induced membrane potentials. There are several obvious directions for continued work in this area. The therapeutic benefits of liposomal encapsulation of drugs in response to a ApH, such as those listed in Table 3-Il, provides a rich research area. Since it appears that ApH-loaded loaded liposomal doxorubicin has clinical advantages of reduced toxicity over the free drug, similar benefits could be expected for doxorubicin analogs such as epirubicin and daunorubicin. The design of novel ApH-loaded drugs is also simplified by the considerations of Chapter 3. The desired amounts of entrapped drug and residual pH gradients can be controlled by altering the interior buffering capacity for a given drug, while 115  drug release could be inteffigently “tailored” by controlling lipid composition and vesicle pH. Another avenue of investigation is on the effects of drug accumulation on the lipid vesicles themselves. The incorporation of large amounts of drugs into the vesicles, particularly if localised in the vesicle inner monolayer, could lead to interesting drug-lipid interactions. Particularly intriguing is the nature of the “coffee-bean” vesicles seen by cryo-electron microscopy after doxorubicin accumulation. Finally, much of the work presented can be exploited as useful tools in other research. For example, the systems which show stable pH gradients without measurable induced membrane potentials could be used to differentiate ApH from AW-dependent phenomena.  116  REFERENCES Altenbach, C. and Seelig, J. (1985). Biochim. Biophys. Acta 818 410-4 15. Balazsovits, J., Mayer., L.D., Bally., M.B., Cullis., P.R., Ginsberg, R.G., and Falk, R. (1989) Cancer Chemotherapy and Pharmacology 23 81-86 Bally, M.B., Tilcock, C.P.S., Hope, M.J., and Cullis, P.R. (1983) Can. J. Biochem. 61, 346-352 Bally, M.B., Hope, M.J., van Echteld, C.J.A. and Cullis, P.R.(1985) Biochim. Biophys. Acta 812, 66-76. Bafly, M.B., Mayer, L.D., Loughrey, H., Redelmeier, T.E., Madden, T.D., Wong, K.M., Hope M.J., Harrigan, P.R. and Cullis, P.R (1988) Chem. Phys. Lipids, 47, 97-107. Bangham, A. (1983) in A.D. Bangham (ed.) Liposome Letters Academic Press, London 26 1-268 Bangham, A.D., Standish, M.M. and Watkins, J.D. (1965) J. Mol. Biol. 13, 238-244. Bangham, A.D. and Hill, M.W. (1986) Chem. Phys. Lipids 40, 189-206. Barchfeld, G.L. and Deamer, D.W. (1985) Biochim. Biophys. Acta 819, 161- 169. Barchfeld, G.L. and Deamer, D.W. (1988) Biochim. Biophys. Acta 944, 4048. Benga, G., Pop, V.1., Popescu, 0., and Borza, V., (1990)Bioscience Reports 10 31-36 Blok, M.C., de Gier, J. and van Deenen, L.L.M. (1974a) Biochim. Biophys. Acta 367, 202-209. Blok, M.C., de Gier, J. and van Deenen, L.L.M. (1974b) Biochim. Biophys. Acta 367, 2 10-224. Bloom, M., Evans, E., and Mouritsen, O.E. (1991) Quart. Rev. Biophys. 24(3) 293-397 Bottcher, C.J.F., van Gent, C.M. and Pries, C. (1961) Anal. Chim. Acta 24, 203-204. Burke, T.G., and Tritton, T.R., Biochemistry (1985)24 1768-1776) Cafiso, D.S. (1989) Methods in Enzymology 172 33 1-45 Cafiso, D.S. and Hubbell, W.L. (1978a) Biochemistry 17, 187-195. Cafiso, D.S. and Hubbell, W.L. (1978b) Biochemistry 17, 387 1-3788. Cafiso, D.S. and Hubbell, W.L. (1982)Biophys. J. 39, 263-272. Cafiso, D.S. and Hubbell, W.L. (1983) Biophys. J. 44, 49-57. Chakrabarti, A.C., Clark-Lewis, I., Harrigan., P.R. and Cullis, P.R., (1992) Biophys. J. 61 228-234 Chapman, D. (1975)  Q.  Rev. Biophys. 8, 185-235.  117  Chapman, D., Williams, R.M., Ladbrooke, B.D. (1967) Chem. Phys. Lip. 1, 445 Chapman, C.J., Erdahi, W.L., Taylor, R.W. and Pfeiffer, D.R., (1990) Chem. Phys. Lip. 55 73-83 Chapman, C.J., Erdahi, W.L., Taylor, R.W. and Pfeiffer, D.R., (1991) Chem. Phys. Lip. 57 201-208 Cohen, B.E. (1975a) J. Memb. Biol. 20, 205-234. Cohen, B.E. (1975b) J. Memb. Biol. 20, 235-268. Conrad, M.P., and Strauss, H.L. (1985) Biophys. J. 48, 117-124. Creaven, P.J., Cowens, J.W., Ginsburg, R., Ostro, M.J., and Browman, G. (1990) J. Liposome Research 1 48 1-490 Crofts, A.R. (1966) Bioc. Biophys. Res. Trans.24, 127-134. Crofts, A.R., (1967) J. Biol. Chem. 242 3352-3359 Crowe, J.H. and Crowe, L.H., (1988) Biochim. Biophys. Acta 939 327-334 Cullis, P.R. and de Kruijff, B. (1979) Biochim. Biophys. Acta 559, 399-420. Danielli, J.F. and Davson, H. (1935) J. Cell Comp. Physiol. 5, 495-502. de Kroon, A.I.P.M., de Gier., and de Kruijff (1989) Biochim. Biophys. Acta 981, 371-373 Deamer, D.W., Prince, R.C. and Croft, A.R. (1972) Biochim. Biophys. Acta 274, 323-335. Deamer, D.W. (1982) in Intracellular pH: Its Measurement, Regulation and Utilization in Cellular Functions, Alan R. Liss Ins., New York pp. 173-187. Deamer, D.W. and Nichols, J.W. (1983) Proc. Nat!. Acad. Sci. 80, 165168. Deamer, D.W. and Bramhall, J.(1986) Chem. Phys. Lipids 40, 167-188. Deamer, D.W. (1987) J. Bioenerg. Biom. 19, 457-468. Deamer, D.W. and Nichols, J.W. (1989) J. Membrane Biol. 107 9 1-103 Demel, R.A., Kinsky, Kinsky and van Deenen (1968) Biochim. Biophys. Acta 150, 655-665. Diamond, J. M. and Katz, Y. (1975) J. Mernb. Biol. 17, 12 1-154 Dilger, J.P. and McLaughlin (1979) Science 206, 1196-1198. Eastman, S.J., M.J. Hope and P.R. Cullis (1991). Biochemistry 30 17401745.  El-Mashaic, E.M. and Tsong, T.Y. (1985). Biochemistry 242884-2888. Elamrani, K. and Blume, A. (1983) Biochim. Biophys. Acta 727, 22-30. Finkeistein, A., (1987) Water Movement Through Lipid Bilayers, Pores and Plasma Membranes. J. Wiley and Sons, New York Fiske, C.H. and Subbarow, Y. (1925) J. Biol. Chem. 66, 375-400. Flewelling, R.F. and Hubbell, W.L. (1986a) Biophys. J. 49, 53 1-540. Flewelling, R.F. and Hubbell, W.L. (1986b) Biophys. J. 49, 541-552. 118  Frederik, P.M., Stuart, M.C.A., Bomans, P.H.H., Busing, W.M., Burger, K.N.J., and Verkleij, A.J. (1991) J. Microscopy 161(2) 253262 Frezard, F., and Garnier-Suilaerot, A. (1991) Biochemistry 30 5038-5043 Gennis, RB. (1989) Biomembranes: Molecular Structure and Function Springer-Verlag, New York Gorter, E. and Grendel, F. (1925) J. Exp. Med. 41, 439-443 Gregoriadis, G., (1988) in Liposomes as Drug Carriers, Wiley and Sons, Toronto, Ont., pp 3-19 Gregoriadis, G. (1976) N. Engl. J. Med. 295 704-707 Gruner, S.M., Lenk, R.P., Janoff, A.S and Ostro, M.J. (1985) Biochemistry 24, 2833-2842. Gruner, 5. (1987) in Liposomes: From Biophysics to Therapeutics (Ostro, M.J. ed) Marcel Dekker pp 1-38. Gutknecht, J. and Walter, A. (1982) Biochim. Biophys. Acta 685, 233-240. Gutknecht, J., (1984)J. Memb. Biol. 82, 105-112. Gutknecht, J., (1987a) J. Bioenerg. Biom. 19, 427-442. Gutknecht, J., (1987b) Biochim. Biophys. Acta 898, 97-108. Grzesiek, S., and Dencher, N.A. (1986) Biophys. J. 50265-276 Hauser, H. and Barratt, M.D. (1973) Biochem. Biophys. Res. Comm. 53, 339 Hope, M.J., Bally, M.B., Webb, G., and Cullis, P.R. (1985) Biochim. Biophys. Acta 812, 55-65. Hope, M.J., Bally, M.B., Mayer, L.D. Janoff, A.S. and Cullis, P.R. (1986) Chem. Phys. Lipids 40, 89-108. Hope, M.J. and Cullis, P.R. (1987)J. Biol. Chem. 262, 4360-4366. Huang, C.H. (1969)Biochemistry 8,344-352. Hyslop, P.A., Morel, B., and Sauerheber, R.D., (1990) Biochemistry 29 1025-1038 Jam, M.K. (1980) in Introduction to Biological Membranes,(Jain, M.K. and Wagner, R.C. ed) John Wiley and Sons, Toronto pp 117-142. Johnson, S.M. (1973) Biochim. Biophys. Acta 307, 27-41. Kasianowicz, J., Benz, R., and McLauglin, 5. (1984) J. Memb. Biol. 82 179- 190 Kedem, 0. and Katchaisky, A. (1958) Biochim Biophys. Acta 27229-246 Kirby, C. and Gregoriadis, G. (1984a) Biotechnology, Nov. 989-994 Kirby, C. and Gregoriadis, 0. (1984b) in Liposome Technology Vol 1,(ed. Gregoriadis, 0.) CRC Press pp 19-27. Kleinfeld, A.M., (1987) in Current topics in Membranes and Transport, Vol. 29, Academic Press Inc. pp 1-27.  119  Koenig, S.H., Ahkong, Q.F., Brown, R.D., Lafleur, M., Spfller, M., Unger, U., and Tilcock, C.P.S. (1991) Magnetic Resonance in Medicine 23 275-283 Kremer, R.M., Hasselbach, H. and Semenza, G. (1981) Biochim. Biophys. Acta 643, 233-242. Krishnamoorthy, G. and Hinkle, P.C. (1984) Biochemistry 23, 1640-1645. Lentz, B.R., Afford, D.R. and Dombrose, F.A. (1980) Biochemistry 19, 2555-2559. Lieb W.R. and Stein W.D.( 1986) J. Memb. Biol. 92, 111-119 Lieb, W.R. and Stein, W.D. (1971) in Current topics in membranes and transport Vol. 2 (ed Bonner F. and Klemzeller, A.) pp 1-39. Lopez-Berestein, G. (1988) in Liposomes in the therapy of infections, diseases and cancer (Lopez-Berestein G., Fidler, I.J. eds.). New York Alan R. Liss, Inc. 317-321 Macey, R.I. and Farmer, R.E.L. (1970) Biochim. Biophys. Acta 211, 104106. Marsh, D., (1991) Chem. Phys. Lip. 57, 2 109-121 Madden, T.D. (1986) Chem. Phys. Lipids 40, 207-226. Madden, T.D., Harrigan, P.R., Tal, L.C.L., Bally, M.B., Mayer, L.D., Redelmeler, T.D., Tilcock, C.P.S., Reinish, L.W., and Cullis, P.R. (1990) Chem. Phys. Lip. 53 37-46 Mayer, L.D., Bally, M.B., Hope, M.J. and Culls, P.R., (1985a) J. Biol. Chem. 260, 802-808. Mayer, L.D., Bally, M.B., Hope, M.J. and Cullis, P.R. (1985c) Biochim. Biophys. Acta 816, 294-302. Mayer, L.D., Hope, M.J., Culls, P.R. and Janoff, A.S.(1985a) Biochim. Biophys. Acta 817, 193-196. Mayer, L.D., Bally, M.B. and Culls, P.R., (1986a) Chem. Phys. Lip. 333-345.  40,  Mayer, L.D., Bally, M.B. and Cullis, P.R., (1986b) Biochim. Biophys. Acta 857, 123-126. Mayer, L.D., Bally, M.B., Hope, M.J. and Culls, P.R., (1986c) Biochim. Biophys. Acta 858, 161-168. Mayer, L.D., Wong, K., Menon, K., Chong, C., Harrigan, P.R. and Cullis, P.R. (1988) Biochemistry 27, 2053-2060. Mayer, L.D., Tal, L.C.L., Masin. D., Ginsberg, R., Cullis, P.R. and Bally, M.B. (1989) Cancer Res. 49 5922-5930 Mayer, L.D., Bally, M.B., and Cullis, P.R. (1990a) J. Liposome Res. 1(4) 463-480 Mayer, L.D., Tai, L.C.L., Bally, M.B., Mitilenes, G.N., Ginsberg, R.S., and Cullis., P.R. (1990b)Biochim. Biophys. Acta 1025 143-151 McLaughlin, S.G.A. and Dilger, J.P. (1980) Pharm. Rev. 60, 825-863. Mifier, I.R. (1987)Biophys. J. 52, 497-500.  120  Mimms, L.T., Zampighi, G., Nozaki, Y., Tanford, C., and Reynolds, J.A. (1981)Biochemistry 20833-840 Mitchell, P. (1961)Nature 191, 144-148. Morgan, J.R., Williams, L.A., Howard, C.B., (1985) Br. J. Radio!. 58, 35-39 Moon, R.B., and Richards, T.H. (1973) J. Biol. Chem. 248 7276-7279 Meuller, P. Rudin, D.O., Tien, H.T. and Wescott, W.C. (1962) Nature 194, 979-980. Nagle, J.F., and Morowitz, H.J. (1978) Proc NatI. Acad. Sc!. 75, 298-302 Nagle, J.F. (1987) J. Bioenerg. Biom. 19, 4 13-426. Nakazato, K., Murakarni, N., Konishi, T., and Hatano, Y. (1988) Biochim. Biophys. Acta 946 143-150 Nayar, R., Hope, M.J. and Cullis, P.R. (1989) Biochim. Biophys. Acta 986, 200-206. Nicholls, D.G., and R1a1,E. (1989) Methods in Ertzymology 174 85-95 Nichols, J.W. and Deamer, D.W. (1976) Biochim. Biophys. Acta 455, 269271. Nichols, J.W. and Deamer, D.W. (1978) Proc. Nat!. Acad. Sc!. 77, 2038. Nichols, J.W. and Deamer, D.W. (1980) in Frontiers of Bioenergetics. P.L. Dutton, J.S. Leigh and A. Scarpa, Eds., p1273-1283, Academic, New York Nichols, J.W. Hill, M.W. Bangham, A.D. and Deamer, D.W (1980) Biochim. Biophys. Acta 596, 393-403. Nobel, P. (1991) in Physiochemical and Environmental Plant Physiology Harcourt Brace Jovanovich, Toronto Norris, F.A. and Powell, G.L. (1990) Biochim. Biophys. Acta 1030 165-171 Ogthara, I., Kojimi, S., Jay, M., (1986) Eur. J. Nucl. Med. 11, 405-411 Olson, F., Hunt, C.A., Szoka,. E.C., Vail, W.J., and Papahadjopoulos, D. (1979) Biochim. Biophys. Acta 557, 9-23. Op den Kamp, J.A.F. (1979)Ann. Rev. Biochem. 48, 47-71. O’Shea, P.S., Thelen, S., Petrone, G. and Azzi, A. (1984) FEBS 172, 103108. Ostro, M., and Culls, P.R. (1989)Am. J. Hosp. Pharm. 46 1576-1587 Overton, E. (1899) Vieteljahresschr. Naturforsch. Ges. Zurich 40:159-201 (Cited in Walter, A. and Gutknecht, J. (1986) J. Memb. Biol 90, 207-217. Parente, R.A. and Lentz, B.R. (1984)Biochemistry 23 2353-2362 Parsegian, V.A. (1969) Nature 221, 844-846. Penefsky, H.S. (1977)J. Biol. Chem. 252 2891-2899 Perkins, W.R., and Cafiso, D.S. (1986) Biochemistry 25, 2270-2276. Perkins, W.R., and Cafiso, D.S. (1987a) J. Memb. Biol. 96, 165-173. Perkins, W.R., and Cafiso, D.S. (1987b) J. Bioenerg. Biom. 19, 443-455. 121  Perkins, W.R., Minchey, S.R., Ostro, M.J., Tarashi, T.F., and Janoff, A.J. (1988) Biochim. Biophys. Acta 943(1)103-107 Rahman, A., Kessler, A., More, N., Sikic, B., Rowden, E., Woolley P., and Schein, P.S. (1980) Cancer Res. 40 1532-1537 Rahman, A., Carmichael, D., Harris, M., and Roh, J.K. (1986) Cancer Research 46 2295-2299 Raines and Cafiso, D.S., (1990)Anesthesiology 70 57-63 Redelmeier, T.E., Mayer, L.D., Wong, K.F., Bally, M.B., and Cuffis, P.R. (1989a) Biophys. J. 56 385-393 Redelmeier,  T.E. (1989b) Vancouver, British Columbia, University of British Columbia; Ph.D. Thesis  Canada:  Redelmeier, T.E., Hope, M.J., and Cullis, P.R. (1990) Biochemistry 29 3046-3053. Robertson, J.D. (1957) J. Biophys. Biochem. Cytol. 3 1043-1047. Rottenberg, H. (1979) Methods in Enzymology 55, 547-569. Rottenberg, H., (1984)J. Membr. Biol. 81(2), 127-132 Rottenberg, H. (1989) Methods in Enzymology 172 63-84 Rudel, L. and M.D. Morris (1973). J. Lipids Res. 14 364-366. Schroeder, F., Jefferson, J.R., Kier, A.B., Knittel, J., Scallen, T.J., Wood, W.G., and Hapala, I. (1991) Proc. Soc. Exp. Biol. Med. 196, 235-252. Silvius, J.R. (1982) in Lipid-Protein Interactions, (Jost, P.C. and Griffith, O.H. eds.) John Wiley and Sons, pp. 239-280. Singer, S.J. and Nicholson, G.L. (1972) Science 175, 720-73 1. Small, D.M. (1986) in The Physical Chemistry of Lipids from Alkanes to Phospholipids, Plenum Press, New York. Sone, S., Poste, G., and Fidler, I.J. (1980) J. Immunol. 124 2197-2202 Stamp, D., and Juliano, R.L., (1979) Can. J. Physiol. Pharmacol. 57 535539 Strichartz, G.R., Sanchez, V., Arthur, R, Chafetz, R., and Martin, Dean (1990) Anesthesia and Analgesia 71 158-70 Storch, J., and Kleinfeld, A.M. (1985) Trends, Biol. Sci Nov. 418-421 Szoka, F. and Papahadjopoulos, D. (1978) Proc. Nati. Acad. Sci. 4 194-4 198.  79,  Szoka, F. and Papahadjopoulos, D. (1980) Ann. Rev. Biophys. Bioenerg. 9, 467-508. Tanford, C. (1980) in The Hydrophobic Effect: Formation of Micelles and Biological Membranes, John Wiley and Sons, New York. Tate,  M.W.,  Eikenberry, E.F., Turner, D.C., Shyamsunder, E., and Gruner, S.M., Chem. Phys. Lip. 57 147-164  Taylor, K.M.G., Taylor, C., Kellaway, I.W., arid Stevens, J. (1990) Journal of Pharmaceutics 58 49-55  122  mt.  Trauble, H. (1971) J. Memb. Biol. 4 193-209 van Deenen, LL.M. and de Gier, J. (1974) in The Red Blood Cell (D. Surgenor ed) Academic Press pp. 147-213. Viero, J.A., and P.R. Culls (1990). Biochim. Biophys. Acta 1025:109-115. Walter, A. and Gutknecht, J. (1984) J. Memb. Biol 77, 255-264. Whitmarsh, J. (1987) Photosynthesis Research 12 43-62 Weinstein, J.W. (1984) Cancer Treat. Rep. 68 127-135 Westman, J., Boulanger, Y., Ehrenberg, A. and Smith, I.C.P. (1982) Biochim. Biophys. Acta 685, 3 15-328. Yang, X. Castleman et al., (1991)J. Chem. Phys., 94, 3268 Ye, R. and Verkman, A.S. (1989) Biochemistry 28, 824-829  123  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items