Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Development of new tools to study drug-lipid interactions and their application to investigating amphotericin… Stoodley, Robin 2007

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2008_spring_stoodley_robin.pdf [ 13.22MB ]
Metadata
JSON: 24-1.0061817.json
JSON-LD: 24-1.0061817-ld.json
RDF/XML (Pretty): 24-1.0061817-rdf.xml
RDF/JSON: 24-1.0061817-rdf.json
Turtle: 24-1.0061817-turtle.txt
N-Triples: 24-1.0061817-rdf-ntriples.txt
Original Record: 24-1.0061817-source.json
Full Text
24-1.0061817-fulltext.txt
Citation
24-1.0061817.ris

Full Text

DEVELOPMENT OF NEW TOOLS TO STUDY DRUG-LIPID INTERACTIONS AND THEIR APPLICATION TO INVESTIGATING AMPHOTERICIN B'S ASSOCIATION WITH MODEL CELL MEMBRANES by ROBIN STOODLEY B.SC., THE UNIVERSITY OF BRITISH COLUMBIA, 2000 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Chemistry) THE UNIVERSITY OF BRITISH COLUMBIA NOVEMBER 2007 0 ROBIN STOODLEY 2007 ABSTRACT The interaction of different formulations of the antifungal drug amphotericin B (AmB) with model cell membranes was studied and new techniques of measuring this interaction using electrochemical and/or spectroscopic methods were developed. Two model cell membrane systems were used: sterol-free lipid monolayers adsorbed to a Hg electrode and sterol-free or sterol-containing floating lipid monolayers on a Langmuir trough. Electrochemical control over the adsorbed monolayer allowed the defectiveness of the layer to be varied and the interaction of AmB with both well- ordered and defective monolayers characterized. Measurements of monolayer capacitance and permeability were used to indicate the nature of the interaction. Capacitance provides a measure of the lipid organization, while permeability was measured via electroreduction of thallium (I) cation. The three AmB formulations and two control samples were examined and showed different interaction behaviour. The disruption of lipid order and permeabilization induced by the two commercial formulations correlated generally with in vivo studies of their toxicity. An experimental and possibly less toxic AmB formulation made monolayer significantly more permeable. In situ fluorescence microscopy of the monolayer on Hg was carried out after introducing a low concentration of fluorophore into the layer. Fluorescence intensity as a function of electrode potential was measured and was used to characterize the lipid on Hg model membrane system before we attempted to measure AmB's influence on the fluorescence. The fluorescence excitation and emission spectra of AmB itself were measured ex situ for two of the formulations. Using added surfactant to control AmB aggregation state, the relationship between AmB aggregation and its fluorescence properties was examined. We discovered AmB to have unusual dual fluorescence properties, the extent of which differed between formulations. We measured AmB's fluorescence in situ as the drug interacted with floating lipid monolayers on the Langmuir trough. Both the variation in fluorescence during compression of a mixed AmB/lipid monolayer and penetration of AmB into a phospholipid monolayer were measured. This experimental setup was configured to collect fluorescence only from AmB at the monolayer, and not from AmB in bulk solution. Fluorescence excitation was made using a laser diode extracted from a consumer electronics device. ii TABLE OF CONTENTS ABSTRACT^  ii TABLE OF CONTENTS ^  iii LIST OF TABLES  vii LIST OF FIGURES ^  viii LIST OF SYMBOLS AND ABBREVIATIONS^  xx ACKNOWLEDGMENTS ^  xxv DEDICATION^  xxvi 1^INTRODUCTION ^  1 1.1^Outline of the Thesis ^ 4 2^AMPHOTERICIN B ^ 6 2.1^Introduction  6 2.1.1 Use of Amphotericin B ^  7 2.1.2 Toxicity of Amphotericin B  9 2.1.2.1 Dose-Dependent Toxicity of FZ ^ 9 2.1.2.2 Mechanism of AmB Toxic Side Effects  10 2.1.3 Formulations of Amphotericin B  11 2.1.3.1 Heat-Treated Fungizone ^  12 2.1.4 Amphotericin B Mechanism of Action  12 2.1.4.1 Increased Permeability due to AmB-Sterol Pores ^ 15 2.1.4.2 Sterol-Independent Pore Formation ^  16 2.2^Summary of AmB Characteristics ^  17 3^THEORY AND BACKGROUND 19 3.1^Introduction ^  19 3.1.1 Electrochemistry of the Interfacial Double Layer ^ 19 3.1.1.1 The Helmholtz Model is akin to a Parallel-Plate Capacitor . . . ^ 19 3.1.1.2 Gouy-Chapman Theory introduces the Diffuse Layer Concept ^  21 3.1.1.3 Gouy-Chapman-Stern (GCS) Theory Describes the Solution Side  21 3.1.2 The Jellium Model of the Electrode Surface ^ 25 3.1.3 The Gibbs Adsorption Isotherm Describes the Interface Thermodynamically ^  29 iii 3.1.4 Extension of the Gibbs Adsorption Isotherm gives the Electrocapillary Equation ^  30 3.1.4.1 Surface Coverage can be Calculated from Surface Excess ^ 31 3.1.4.2 Adsorption of Neutral Species can also be Measured via Capacitance ^  31 3.1.5 Surface Coverage 0 is often Modeled with the Langmuir Isotherm . . . ^ 35 3.1.6 Potential Step Chronoamperometry involves Diffusion to the Electrode ^  37 3.1.6.1 The Cottrell Equation Predicts the Current-Time Transient . . . ^ 38 3.1.6.2 The Nature of Diffusion to a Partly-Blocked Electrode Varies with Time ^ 38 3.1.7 Structure and Properties of Monolayers at the GIS ^ 39 3.1.7.1 Mixed Monolayers ^ 47 3.1.8 The Biological Cell  48 3.1.8.1 Function and Structure of Cell Membranes ^ 48 3.1.8.2 Liposomes as a Model for the Cell Membrane 50 3.1.9 UV-Visible Spectroscopy ^ 52 3.1.9.1 Transition Probability and Intensity Depend on Transition Dipole Moment ^ 52 3.1.9.2 Selection Rules Further Limit Possible Transitions ^ 53 3.1.9.3 The Franck-Condon Principle Rationalizes Vertical Transitions ^  53 3.1.9.4 Absorption Spectra ^ 54 3.1.9.5 Molecular Exciton Theory Describes Aggregation-Induced Spectral Shift ^ 54 3.1.10 Fate of the Excited State ^  57 3.1.10.1^Fluorescence and Phosphorescence ^ 57 3.1.10.2^Non-Radiative Relaxation that does not give a Second Excited Species ^  59 3.1.10.3^Non-Radiative Relaxation that does give a Second Excited Species  60 3.1.11 Fluorescence Microscopy of Monolayers ^  61 3.1.11.1^The Epi-Fluorescence Microscope ^ 61 4^ELECTROCHEMICAL INVESTIGATION OF AmB-LIPID INTERACTION^ 66 4.1^Introduction ^ 66 4.2^Literature Review 69 4.2.1 Development of the DOPC Monolayer Adsorbed on Hg Electrode . . . ^ 69 4.2.2 Use of Electrochemistry to Probe Drug-Model Membrane Interactions ^  74 4.3^Summary of Electrode-Supported Biomimetic Systems ^ 77 4.4^Materials and Methods ^ 78 4.4.1 Introduction 78 4.4.2 Electrochemical Setup 78 4.4.3 Measurement Methodology ^ 82 iv 4.4.4 Electrochemistry in the Presence of Amphotericin B ^ 85 4.4.4.1 Thallium Electroreduction and Chronoamperometry ^ 87 4.5^Results and Discussion     95 ^ 4.5.1 Introduction     95 4.5.2 Capacitance Measurements   95 4.5.3 Tr Reduction Measurements ^  108 4.6^Conclusions ^  117 5^FLUORESCENCE MICROSCOPY OF Hg-SUPPORTED DOPC MONOLAYER ^ 118 5.1^Introduction ^  118 5.2^Literature Review  120 5.2.1 Spectroelectrochemistry ^  121 5.2.1.1 Spectroelectrochemistry of Biomimetic Systems ^ 122 5.2.2 Fluorescence Near a Metal Surface ^  123 5.2.3 Properties of Fluorescent Probes for Membrane Studies ^ 124 5.3^Materials and Methods ^  125 5.3.1 Introduction  125 5.3.2 Electrochemical Setup  125 5.3.3 Fluorescence Microscopy^  126 5.4^Results and Discussion ^  132 5.4.1 Electrochemistry of DOPC Monolayer Between -0.3 and -1.85 V . . . ^ 132 5.4.2 Electrochemistry of the 3 mol% Dye/DOPC Monolayer Between -0.3 and - 1.85 V ^  133 5.4.3 Electrochemistry of 3 mol% Dye/DOPC Monolayer Between -0.3 and -1.2 V  134 5.4.4 Fluorescence Microscopy of 3 mol% Dye/DOPC Monolayer Between -0.3 and -1.85 V ^  135 5.4.5 Fluorescence Microscopy of 3 mol% Dye/DOPC Monolayer Between -0.3 and -1.2 V  141 5.4.6 Fluorescence Microscopy of DOPC Monolayer with AmB in Electrolyte ^  142 5.5^Conclusions  145 6^FLUORESCENCE OF AmB IN FZ AND HTFZ ^ 147 6.1^Introduction ^  147 6.2^Literature Review  148 6.2.1 AmB Aggregation ^  148 6.2.2 Surfactant Control of Aggregation State ^  149 6.2.3 Spectroscopy of AmB 150 6.3^Materials and Methods ^  155 6.3.1 Sample Preparation  155 6.3.2 Instrumentation     156 6.4^Results and Discussion ^  157 6.4.1 Introduction  157 6.4.2 UV-Visible Absorbance Spectroscopy ^  157 6.4.3 Fluorescence Spectra ^  158 6.4.4 Fluorescence of Amphotericin A ^  162 6.4.5 Fluorescence Spectra with Added SDS 162 6.4.6 Dual Fluorescence of AmB 165 6.5^Conclusions ^  175 7^AmB INTERACTION WITH FLOATING LIPID MONOLAYERS ^ 176 7.1^Introduction ^  176 7.2^Literature Review  177 7.2.1 Monolayer Composition ^  178 7.2.2 Measurement of Biomolecule Interaction ^  179 7.2.3 Monolayer Studies of AmB 180 7.3^Materials and Methods ^  182 7.3.1 Introduction  182 7.3.2 Langmuir Trough  182 7.3.3 BODIPY FL DHPE ^ 185 7.3.4 Simultaneous Monolayer Compression and Epi-Fluorescence Microscopy ^  185 7.3.5 Simultaneous Monolayer Compression and Laser-Excited AmB Fluorescence  187 7.3.5.1 Characterization and Modification of the Blu-ray Laser ^ 188 7.3.5.2 Optical Arrangement and Signal Processing ^ 189 7.3.6 Experiments ^  192 7.4^Results and Discussion  193 7.4.1 Compression Isotherms and Fluorescence Imaging with Incorporated BODIPY^ 193 7.4.2 Excitation of Surface AmB ^ 197 7.5 7.4.2.1 Surface Pressure and Fluorescence of 20 mol% AmB/DOPC Mixed Monolayer ^ 201 7.4.2.2 AmB Penetration into a Dilute DOPC Monolayer^ 202 7.4.2.3 Gibbs Monolayer Formed from FZ ^ 203 7.4.2.4 AmB Penetration into a DOPC Monolayer at Biologically-Relevant Surface Pressure ^ 204 Conclusions ^ 210 8 CONCLUSIONS AND FUTURE WORK ^ 211 8.1 Opportunities for Immediate Application 215 8.2 Opportunities for Additional Method Development ^ 216 8.3 Use of New Fluorescence Tools ^ 217 8.4 Neutron Reflectometry Studies 218 9 REFERENCES ^ 219 vi LIST OF TABLES Table 1 Partial List of Toxic Side Effects of FZ ^  10 Table 2 Composition and Therapeutic Dose of AmB Formulations and Control Solutions Used. ^  86 Table 3 Composition of Deposited Monolayers ^  184 vii LIST OF FIGURES Figure 1 Chemical Structure of AmB. Partial numbering of carbon atoms on macrolide ring is given. AmB is characterized by an all-trans heptaene on one side of the macrolide ring and numerous hydroxyl groups on the other side.   8 Figure 2 Chemical structures of the fungal cell sterol ergosterol (left) and mammalian cell sterol cholesterol (right). Ergosterol has two additional double bonds, one in the ring structure and one on the side chain. Additionally, ergosterol has an extra methyl group on the side chain. 15 Figure 3 Schematic summary of possible AmB mechanism of action. FZ administration supplies AmB both as aggregates and as monomers. It is believed AmB monomer permeabilizes membrane only if ergosterol is present. AmB aggregates are thought to form pores if cholesterol, ergosterol, or no sterol is present. Pore-formation in cholesterol-containing membranes is the basis of AmB's toxicity. Interaction of AmB with lipid headgroups at membrane surface is described as a 'pre-pore'. The schematic representations used for lipid and sterol are given in the upper right corner.   18 Figure 4 Schematic structure of the charged electrode(solution interface. A negatively charged electrode will have solvent and possibly also have specifically adsorbed anions at the surface. The plane through the centre of specifically adsorbed anions defines the Inner Helmholtz Plane (IHP), while the plane of closest approach of hydrated cations defines the Outer Helmholtz Plane (OHP). The inner layer lies between electrode surface and IHP; the diffuse layer lies between IHP and bulk solution. Reprinted with permission from [90], Copyright 2001 John Wiley & Sons.   22 Figure 5 Potential drop across the double layer, calculated according to Gouy-Chapman-Stern theory for a 10' M 1:1 electrolyte at 25°C assuming a potential of 100 mV at the Outer Helmholtz Plane (x 2). The compact layer is an alternate name for the inner layer. Reprinted with permission from[90], Copyright 2001 John Wiley & Sons.  26 Figure 6 Differential capacitance near the potential of zero charge (pzc) according to the Gouy- Chapman-Stern model. pzc can be found experimentally by observing the dip in capacitance at low electrolyte concentrations. Adapted from [90].   27 Figure 7 Calculated electron density at a metal surface according to the jellium model. The background positive charge from the metal nuclei shows an abrupt transition at the metal surface. Within the metal, electron density shows Friedel oscillations. Electron 'spill-over' to the outside of the metal is present but decays rapidly with distance from the surface. r s represents the radius of the sphere which contains one free electron; two cases are shown. Distance is shown in Fermi wavelengths, which varies from metal to metal but is approximately 0.5 nm. Reprinted figure with permission from [92], Copyright 1970 American Physical Society.   28 viii Figure 8 Comparison of the interface between phases in real and model two-phase systems. In actuality, panel a), an interfacial region exists between phases a and P. The model system, panel b), shows the placement of the Gibbs dividing line separating the two pure phases. 30 Figure 9 Electrocapillary curves for mercury in contact with aqueous solutions of the salts indicated. The typical parabolic shape of electrocapillary curves is observed. Units of surface tension used here, dynes/cm, are equivalent to mN/m. The rational potential scale used on the x-axis has the PZC as the zero point. Reprinted with permission from [93], Copyright 1947 American Chemical Society.   33 Figure 10 The inter-relations between surface tension, charge density and differential capacity are shown for four cases of bulk surfactant activity (increasing 1) to 4)). States A and B represent when the electrode is uncovered by surfactant and when surfactant is adsorbed, respectively. The surface tension of the system as a function of potential will always follow the lowest possible path, shown here in solid line. Stepped changes in charge density occur for changes in adsorbate coverage (e.g. adsorption to and desorption from the electrode). As capacitance is the derivative of charge density with potential, peaks are observed in capacitance curve when coverage changes (shown schematically as vertical dashed line). Adapted from [95] 34 Figure 11 Calculated relation of surface pressure with bulk concentration of the adsorbate under Langmuir, Henry and Frumkin isotherms. In limit of low coverage, all converge to Henry isotherm linear dependence on concentration. Values of a are lateral interaction parameter for Frumkin isotherm; positive values indicate attraction. Reprinted with permission from [96], Copyright 1992 VCH Publishers  36 Figure 12 Schematic of relation between potential steps, resultant current-time transients and sampled-current voltammogram for a generalized 0 + ne R process. Left panel shows five separate steps, each to a more reductive potential. The step labelled '1' is to a potential at which reduction does not occur, while steps '4' and '5' are to potentials well beyond E. Steps '2' and '3' are intermediate. Zero on the time scale is set at the instant of the step. Centre panel shows no Faradaic current flows in response to the first step, but does for subsequent steps. Current transients have a t"'A dependence. Current response for steps '4' and '5' are identical because both are mass-transport (diffusion) limited. Value of current at time t is used to plot right panel voltammogram showing typical sigmoidal response curve. It can be clearly seen that the potential of step 1' is not reductive to 0 and steps '2' and '3' are in the kinetically-limited potential region with '3' being near the E A . The drawn curve joins the five points here but in practice would be made from many more points. Adapted from [90] .   40 ix Figure 13 Variation in diffusion geometry to a partly-blocked electrode as time after a reductive potential step increases. In this example, the spacing between active areas is -100 nm. Panel a) at very short times, diffusion is linear to individual active sites. Panel b) at intermediate times, the diffusion layer extends beyond the geometrical area of the active site, incorporating radial diffusion. Panel c) at long times, the diffusion layers of neighbouring individual sites have overlapped, giving linear diffusion over the whole surface. Reprinted with permission from [90], Copyright 2001 John Wiley & Sons.   41 Figure 14 Schematic diagram of Wilhemy plate used to measure surface tension. H is the submerged depth, H p the total plate height, W p the width of the plate, and 0 the contact angle between plate and surface. Note the scale of the organic layer has been grossly exaggerated for clarity  43 Figure 15 Composite compression isotherm giving the various states in which a monolayer may exist as it is increasingly compressed. Schematics of the molecular organization are given above the state. Not all species will show all possible states. At large mean molecular areas, the monolayer behaves as a two-dimensional gas. Compression changes the state to be liquid-like, before adopting an incompressible solid state at small mean molecular area. Further compression leads to monolayer collapse and possible bilayer formation. Surface pressure units of dyne/cm are equivalent to mN/m. Reprinted with permission from [100], Copyright 1982 John Wiley & Sons.   45 Figure 16 Diagram of the geometry involved in the balance of forces for wetting (spreading) of an organic droplet on a solid surface. ^ 46 Figure 17 Schematic diagram showing possible behaviour of a two component mixed monolayer system. Upper panel: components phase segregate into distinct regions. Middle Panel: mixing occurs but is not ideal. Lower Panel: Ideal mixing of the two components. . . . 49 Figure 18 Chemical structures of three lipids: octadecanol, dipalmitoylphosphatidylcholine (DPPC), and dioleoylphosphatidylcholine (DOPC). Octadecanol is a saturated fatty alcohol, DPPC is a phospholipid with saturated acyl chains, and DOPC features a cis-double bond in each chain. Both DPPC and DOPC are zwitterionic at neutral pH. DPPC and DOPC are both commonly used in models of biomembranes.   51 Figure 19 Diagram showing most intense E 0 -E 1 transitions are those with maximum vibrational wavefunction overlap, described under the Franck-Condon principle as being vertical transitions. E0 is the ground state, while E 1 is an excited state. Vibrational levels are numbered within each electronic level.   56 Figure 20 Diagram of energy level structure for molecular dimers with different orientation between transition dipoles. Aggregation with dipoles parallel to each other and perpendicular to the plane of aggregation is known as 'card-pack' and produces blue-shifted spectra. Solid arrows show allowed transitions, dashed arrows show forbidden transitions. Reprinted with permission from [112], Copyright 1963 Radiation Research Society. .. 58 Figure 21 Jablonski diagram showing fluorescence and phosphorescence pathways to excited state relaxation. States A and B have the same multiplicity, but that of state C is different. Vertical arrows indicate photon absorption or emission. Wavy line indicates a non-radiative process, while the diagonal line shows crossing to a state of different multiplicity (intersystem crossing).   59 Figure 22 The measured (data points) and calculated (line) radiative (intrinsic) fluorescence lifetime of the Eu3+ cation as a function of separation distance from a thick silver surface. Sharp decrease in radiative lifetime at short distances corresponds to fluorescence quenching, while oscillations at larger separations are due to a reflective-effect. The measured excited state lifetime in the absence of silver layer is 632 p.s. Chance et al. also studied the effect of fatty acid layers above the Eu3+ containing monolayer; S=0 in the figure indicates these data recorded when the fluorescent layer was topmost. Reprinted with permission from [117], Copyright 1975 American Institute of Physics.   63 Figure 23 Light path in the epi-fluorescence arrangement. White light from Xe arc lamp impinges excitation filter which transmits excitation wavelength. The dichroic mirror reflects excitation light up through the objective and onto the sample. Fluorescence from sample is collected through objective and transmitted by dichroic mirror. The dichroic mirror also reflects most of the backscattered excitation light while the emission filter provides additional discrimination against 'leaked' excitation light. A coupling lens ensures the size of the image corresponds to the size of the charge-coupled device (CCD) detector. . . . 64 Figure 24 Transmission spectra of the three components of the Olympus U-MWIBA filter cube. Light between 450 and 500 nm is transmitted by the excitation filter, chosen to match the absorbance spectrum of the fluorophore. The dichroic mirror reflects light below 505 nm and transmits above. Oriented at 45°, the dichroic mirror reflects excitation light up through the objective. Emission light impinging on the dichroic mirror is transmitted. Emission filter transmits in the range where fluorescence is expected, 505 to 580 nm. Fringing observed is a product of the interference nature of the filter construction.   65 Figure 25 The WK2 model hanging mercury drop electrode. The micrometer head at top drives a stainless steel piston into the reservoir of Hg held in the capillary. The capillary tip of our electrode was modified to have a conical point. Image courtesy of the Polish Academy of Sciences  80 Figure 26 Schematic diagram of electrochemical setup. In the cell at left, the hanging mercury drop electrode serves as working electrode (WE), a platinum coil is the counter electrode (CE) and argon gas can be supplied above or through the electrolyte. A saltbridge with stopcock connects the saturated calomel reference electrode (RE) to the rest of the cell. Potential control and current measurement are provided via the potentiostat and computer data acquisition system (DAQ). The lock-in amplifier is used in conjunction with the potentiostat when measuring capacitance  81 xi Figure 27 Process of formation of DOPC monolayer on mercury electrode. Left: DOPC dissolved in pentane is injected onto argonlelectrolyte interface and pentane allowed to evaporate, leaving DOPC monolayer. Centre: mercury electrode is slowly lowered through interface, coating lipid monolayer onto mercury surface. Right: monolayer forms with lipid alkyl tails towards mercury due to hydrophobic interaction. Note scale of lipids is grossly exaggerated with respect to electrode.   84 Figure 28 Sequence of potential applied to mercury electrode for Tl÷ step experiments. Initial and final potentials are -0.4 V. Prior to each step, the potential is held at -0.35 V for 2 s, allowing the monolayer to equilibrate. Increasingly negative step 'reductive' potentials are held for 0.25 s. This corresponds to increasing the driving force for Tr reduction at the electrode. Note that the first few 'reductive' potentials are actually oxidative. After each reductive step, the potential is returned to -0.2 V to allow any amalgamated Tl to oxidize to T1+(aq) and diffuse away from the electrode. The inset shows detail of the step to -0.5 V. The portion of the step for which the current transient is measured is shown in dashed line. Inset not to scale.   89 Figure 29 Sample Tr reduction transient. The large magnitude current immediately after time zero is the result of capacitive charging, while the non-zero current at longer times is the result of Tr reduction. This particular transient was recorded on the step to -1.10 V for an uncovered Hg surface. Currents recorded before time zero correspond to 'pre-points' used to correct for possible non-zero current offset. A small negative current offset is observed here. The (uncorrected) zero current line is sketched in for reference. The current at 50 ms was normalized and plotted in one method of analysis, while the entire transient was used for parameter-fitting purposes. Current was recorded every 29 p.S; the transient shown is truncated after 0.1 s while the recorded data actually extends beyond 0.2 s. Capacitive pickup of the 60 Hz line frequency is visible as spikes.   90 Figure 30 A simplex in 2-D parameter space and a representation of the operations available in the amoeba routine. Panel A) a simplex made of three points, with high and low as marked. One side of the simplex is thicker to aid in visualization of operations. Panels B-D show possible operations on the simplex of Panel A). Panel B) reflection operation: new test point chosen by reflection of the highest point across line of the other two points. Panel C) expansion operation: the simplex is lengthened in the direction of a favourable test point. Panel D) contraction operation: An unfavourable test point can be discarded in favour of contracting the simplex towards its centroid. When the simplex is over a minimum, the shrinking effect of the contraction operation gives refinement.  93 Figure 31 Sample result of fitting to recorded current transient. Top panel: current transient shown in black, fit result in grey. Bottom panel: residual between transient and fit. This particular transient was recorded on potential step to -0.875 V, with HTFZ present at 100% TD. ^ 94 xii Figure 32 Capacitance of Hg electrode as a function of potential. Labels refer to what is at the metallsolution interface. DOPC as deposited refers to the negative-going potential scan, DOPC reformed refers to the positive-going potential scan and 0.1 M KC1 is the designation for the uncoated electrode. Capacitance of DOPC layers in potential range -0.25 to -0.8 V is shown scaled 33x to demonstrate its invariance in this range. Two pseudo-capacitance peaks, representing molecular reorganization of the monolayer, are visible at -0.95 (`peak F) and -1.04 V (`peak 2').   101 Figure 33 Capacitance of DOPC monolayers in the presence of added FZ as a function of potential. Concentrations of added FZ are 0%, 30%, 75% and 100% TD. 100% TD for FZ corresponds to 0.541.1,M AmB. Capacitance in the potential range -0.3 to -0.8 V has been magnified 33 x for clarity. Full description of the AmB-induced changes is given in the text.   102 Figure 34 Capacitance of DOPC monolayer with different formulations of AmB or control added. The solid line denotes as-deposited monolayer, the dashed line represents the reformed monolayer, and the dotted line gives the capacitance of the uncoated Hg electrode. Concentration of the added formulation is 100% TD or equivalent for control samples. 100% TD for FZ and HTFZ is 0.54 p,M AmB; for Abelcet (ABLC) it is 5.4 p.M. Panel C gives the capacitance of DOPC monolayer in absence of any addition, and serves as a reference point. Panels A), B) and F) show the capacitance curves of formulations containing the AmB as active-ingredient. Panels D) and F) correspond to the effect of the control samples, deoxycholate and ABLC blank. Full description of the data is given in the text.   106 Figure 35 Summary of key features from capacitance of DOPC monolayer with added AmB formulation or control sample. Top panels represent the capacitance value of peak 1 for as- deposited (left) and reformed layers (right). Bottom panels represent value of capacitance averaged over the -0.35 to -0.65 V potential range. Representative error bars of one standard deviation are shown for the highest concentrations. Reproduced with permission from [3], Copyright 2002 Elsevier.   107 Figure 36 Normalized Tr reduction currents for DOPC-coated Hg drop in contact with subphase containing AmB formulation or control at 100% TD or equivalent concentration. The measured reduction currents in the absence of any monolayer are shown for reference. The capacitance of the DOPC monolayer is shown in the potential range -0.85 to -1.15 V to remind that the phase transitions observed result in a porous layer and are preceded by nucleation of defects. The impermeability of the DOPC layer alone is demonstrated by the large (0.4 V) overpotential required to measure Tr reduction. The presence of AmB formulations decreases the required overpotential for reduction. Full explanation of the data is given in the text. Reproduced with permission from [3], Copyright 2002 Elsevier. ^  115 Figure 37 Calculated pore radius (bottom), pore number density (middle), and lipid coverage 0 (top) for DOPC monolayer with added AmB formulation or control solution. Data points represent average of calculation results for individual runs, while error bars indicate one standard deviation for highest concentration datapoint given. Full explanation of the data is given in the text. Reproduced with permission from [3], Copyright 2002 Elsevier. 116 Figure 38 Absorption and emission spectra of fluorescein at pH 9. Data for this spectra from Invitrogen-Molecular Probes fluorescein reference standard (catalogue # F-1300). Dashed lines represent the limits of the spectral bandpass regions of the excitation (460-490 nm) and emission (515-550 nm) filters of the Olympus U-MWIBA filter cube. The peak wavelength and shape of the emission spectrum are largely independent of pH because the predominant dianion form at pH > 6.4 remains the dominant contributor to fluorescence even below pH 6.4 [219].   129 Figure 39 Spectroelectrochemical cell used for simultaneous fluorescence imaging and capacitance measurement. The optical window is a 0.17 mm thick coverglass. The filter cube shown here is the Olympus U-MWIBA cube suitable for use with fluorescein, but the cubes are mounted on a rotatable turret and could be interchanged if desired. The objective used here was 50x magnification, but 10x, and 20x objectives are interchangeable. An additional 1.5X magnification can be selected within the microscope if desired    130 Figure 40 Relationship between imaged portion of Hg electrode, the in-focus region, and the extent of the drop diameter. At left top: a brightfield image of the Hg electrode. The in-focus region is clearly visible. Left bottom: the size and position of the imaged area relative to the full extent of the drop. The circumference of the spherical drop is shown schematically by the black circle. The centre of the in-focus region was taken to be the bottom centre of the drop and the position of the drop's edge inferred from the drop size and the scale of the imaged region. At right, a bright field image of the electrode with added illumination from a flashlight at an oblique angle. The reflected light shows the Hg drop extends beyond the in-focus region.   131 Figure 41 Capacitance versus applied potential measurements of the Hg electrode in contact with 0.1 M KCl ( dotted line), coated with DOPC monolayer adsorbed from GIS (panel a) or coated with 3 mol% 5-octadecylaminofluorescein/DOPC monolayer adsorbed from GIS (panels b and c). Panel b gives measurement over the full potential range, while panel c shows the capacitance when the negative potential limit is restricted to -1.2 V. Solid line gives negative-going potential sweep, while dashed line gives positive-going sweep. A series RC circuit was taken to accurately represent the electrical character of the interface, allowing calculation of the interfacial capacitance.   136 xiv Figure 42 Spectroelectrochemical results for the 3 mol% 5-octadecylaminofluorescein/DOPC monolayer adsorbed on Hg electrode from GAS. The full potential range was examined, -0.3 to -1.85 V. Top panel: selected fluorescence images, 1 s exposure. White circle outlines the position of the in-focus region of the electrode. Middle panel: average fluorescence intensity calculated over the in-focus region. Alphabetical labels on the data points refer to fluorescence images above. Bottom panel: capacitance measurements made during fluorescence imaging (data points, only data from negative-going potential steps shown) and capacitance curve measured via potential sweep on a different adsorbed monolayer. Solid line: negative-going scan; dashed line: positive-going scan. Full interpretation of the data is given in the text. Reproduced with permission from [4], Copyright 2003 The Royal Society of Chemistry  137 Figure 43 Spectroelectrochemical results for the 3 mol% 5-octadecylaminofluorescein/DOPC monolayer adsorbed on Hg electrode from GAS. The potential range was limited to -0.3 to -1.2 V. Top: selected fluorescence images, 15 s exposure. White circle outlines the position of the in-focus region of the electrode. Middle panel: average fluorescence intensity calculated over the in-focus region. Alphabetical labels on the data points refer to fluorescence images above. Bottom panel: capacitance measurements made during fluorescence imaging (data points, only data from negative-going potential steps shown) and capacitance curve measured via potential sweep on a different adsorbed monolayer. Solid line: negative-going scan; dashed line: positive-going scan. Full interpretation of the data is given in the text. Reproduced with permission from [4], Copyright 2003 The Royal Society of Chemistry  143 Figure 44 Spectroelectrochemical results for the 3 mol% 5-octadecylaminofluorescein/DOPC monolayer adsorbed on Hg electrode from GIS in the presence of 0.54 AmB as FZ. The potential range was limited to -0.3 to -1.2 V. Top panel: selected fluorescence images, 12 s exposure. White circle outlines the position of the in-focus region of the electrode. Middle panel: average fluorescence intensity calculated over the in-focus region. Alphabetical labels on the data points refer to fluorescence images above. Bottom panel: capacitance measurements made during fluorescence imaging (data points, only data from negative- going potential steps shown) and capacitance curve measured via potential sweep on a different adsorbed monolayer. Solid line: negative-going scan; dashed line: positive-going scan.   144 Figure 45 UV-vis spectra of pure AmB in different aggregation states in water. Dot-dash line:10 -7 M AmB, corresponding to the monomeric state. Absorption maximum is 409 nm, with characteristic vibrational fine structure peaks at 385 and 365 nm. Solid line is 10 -4 M AmB, representing AmB in predominately aggregated form. The absorption maximum is a broad peak at 339 nm. Dashed line gives absorption spectrum of a 10' M AmB solution that was heated to 70 °C for 20 minutes. The absorption maximum is blue-shifted to 322 nm. Reprinted with permission from [10], Copyright 1997, Elsevier.   154 xv Figure 46 Energy level diagram for linear polyenes such as AmB, idealized as having C2h point group. Absorption process is labelled as 'A', while fluorescence is shown as 'F'. Possible internal conversion is not shown. The ground state, S o , is denoted with its symmetry as 1 l Ag . Similarly, the S 1 and S2 states are given as 2 1 Ag and 1 1 B o . The possibility of the dimer or aggregate having non-zero displacement energy is not considered. Reproduced with permission from [5], Copyright 2007 American Chemical Society.   155 Figure 47 UV-visible absorbance spectra of 5.16 x10 -5 M AmB as FZ with varying concentrations of added SDS as labelled. SDS concentrations are given as mol ratio with respect to AmB. FZ with no added SDS shows a strong absorption band at 326 nm assigned to aggregated AmB. When monomerized (209 SDS/AmB), the 411 nm band of monomeric AmB dominates, with related vibrational fine structure bands at 388, 367 and 350 nm. Very low absorbance is noted beyond 450 nm. Not all SDS concentrations measured are shown. Reproduced with permission from [5], Copyright 2007 American Chemical Society.  160 Figure 48 UV-visible absorption spectra of 5.16 x10 -5 M AmB as HTFZ with varying concentrations of added SDS as labelled. SDS concentrations are given as mol ratio with respect to AmB. The spectra are very similar to those of FZ; the subtle differences include the absorption band at 292 nm is increased for HTFZ and the symmetry around the 326 nm band with increasing SDS varies between FZ and HTFZ. Not all SDS concentrations measured are shown. Reproduced with permission from [5], Copyright 2007 American Chemical Society.   161 Figure 49 Fluorescence spectra of 5.16 xlCr 5 M AmB as FZ and HTFZ. Panel A) gives excitation and emission of AmB monomer. The excitation scan (Em 560 nm) shows peaks at 324, 339, 356, and 386 nm. Emission scan (Ex 408 nm) gives bands at 435, 471, 528, 562, and 608 nm. Panel B) gives results for AmB dimer. The excitation scan (Em 471 nm) shows peaks at 324, 338, 356 and 385 nm. The emission scan (Ex 350 nm) gives features at 452, 473, 494, and 525 nm. Peak positions are those of FZ; positions for HTFZ are similar. Units of fluorescence intensity are arbitrary. Reproduced with permission from [5], Copyright 2007 American Chemical Society.   163 Figure 50 Fluorescence emission spectra of 5.16x 10 -5 M AmB as FZ and HTFZ. 325 nm excitation excites higher aggregates of AmB. Units of fluorescence intensity are arbitrary. Reproduced with permission from [5], Copyright 2007 American Chemical Society.  164 Figure 51 Fluorescence excitation and emission spectra of 5.74 x10 -5 M nystatin-deoxycholate under similar conditions to those used acquiring the data of Figure 49. Excitation spectrum (Ex 408 nm) and excitation spectrum (Em 560 nm) both magnified 10x for clarity. Units of fluorescence intensity are arbitrary. Reproduced with permission from [5], Copyright 2007 American Chemical Society.   166 xvi Figure 52 Fluorescence excitation and emission scan of 5.16x10 -5 M AmB as FZ with varying concentrations of added SDS as labelled. Panel A) presents results for AmB monomer: emission scan (Ex 408 nm) and excitation scan (Em 560 nm). Panel B) shows results for AmB dimer: emission scan (Ex 350 nm) and excitation scan (Em 471 nm). Not all SDS concentrations measured are shown. Units of fluorescence intensity are arbitrary. Reproduced with permission from [5], Copyright 2007 American Chemical Society.  167 Figure 53 Fluorescence excitation and emission scan of 5.16x10 -5 M AmB as HTFZ with varying concentrations of added SDS as labelled. Panel A) presents results for AmB monomer: emission scan (Ex 408 nm) and excitation scan (Em 560 nm). Panel B) shows results for AmB dimer: emission scan (Ex 350 nm) and excitation scan (Em 471 nm). Not all SDS concentrations measured are shown. Units of fluorescence intensity are arbitrary. Reproduced with permission from [5], Copyright 2007 American Chemical Society.  168 Figure 54 Ratio of fluorescence emission intensity from S 2 (437 nm band) and S 1 (562 nm band) energy levels for 5.16 x10 -5 M AmB as FZ and HTFZ with increasing sodium dodecyl sulfate (SDS) concentration. Fluorescence excitation at 408 nm. Reproduced with permission from [5], Copyright 2007 American Chemical Society    171 Figure 55 Summarized fluorescence emission and absorbance of 5.16x 10 -5 M AmB as FZ and HTFZ with increasing surfactant concentration. Panels A) and B) give peak fluorescence emission from AmB monomer (Ex 408 nm) and higher aggregates (Ex 325 nm), respectively. Panels C) and D) give absorbance at 408 nm and 325 nm, respectively. Units of fluorescence intensity are arbitrary. Reproduced with permission from [5], Copyright 2007 American Chemical Society  172 Figure 56 Fluorescence excitation and emission spectra of monomeric BODIPY FL fluorophore in methanol. Data for this spectra from Invitrogen-Molecular Probes. Dashed lines represent the limits of the spectral bandpass regions of the excitation (460-490 nm) and emission (515- 550 nm) filters of the Olympus U-MWIBA filter cube.   186 Figure 57 Schematic of experimental setup for in situ fluorescence imaging of the monolayer on the Langmuir trough. A Labview program controls the barrier position and records the surface pressure at which fluorescence images were taken. The filter cube used varied depending if BODIPY monomer or dimer fluorescence was being measured. The 515-550 nm emission filter passes monomer fluorescence while the 590 nm long pass emission filter collects dimer fluorescence only. The microscope objective is located in the centre of the trough. The Wilhemy plate is located centrally between the barriers but close to the edge of the trough and does not interfere with spectroscopic measurement. The plate is shown off-centre here for clarity.   188 xvii Figure 58 Applied direct current-voltage characteristic for Blu-ray laser diode with simultaneous light output measurement. Solid curve represents potential, while dotted curve gives light output. Onset of lasing occurs near 22 mA applied current. Good linearity of light output with current is observed after lasing begins. Light intensity is measured as photodiode output and is in arbitrary units.   190 Figure 59 Laser and optics portion of Blu-ray DVD drive carriage, as used. The laser diode is located at left and the beam shines towards the upper right. A holographic grating is located immediately in front of the laser diode before the beamsplitters. The mirror reflects the light output in a direction into the page. The objective lens normally is mounted on the opposite side, below the mirror, but was removed before use.   191 Figure 60 Experimental setup for in situ measurement of AmB fluorescence from a monolayer on Langmuir trough. Laser beam strikes surface at 59° to surface normal, illuminating the focal point of the microscope objective. The height of the GIS interface is adjusted such that it is at the focal distance of the objective. The Wilhemy plate is hung off-centre with respect to the barriers to avoid blocking the path of the laser. The laser intensity is modulated within the linear region of the current-light output curve determined earlier. Collection of fluorescence is through a filter cube to a photomultiplier tube (PMT). A Labview program controlled barrier position and simultaneously recorded surface pressure and the amplified demodulated fluorescence signal. The inset at top left shows the geometry of the refracted laser beam relative to the acceptance cone of the objective.   193 Figure 61 Compression isotherms and fluorescence intensity for 3 mol% BODIPY FL DHPE/DOPC monolayer in the presence or absence of added AmB formulation in the subphase. Formulation concentration was 80 %TD, added after deposition of the monolayer. Fluorescence images were subjected to a background correction process before having the resultant mean intensity calculated. Measurements were made at room temperature. Calculation of area per deposited molecule includes both DOPC and fluorophore; a total of 6.5 x10 15 molecules were deposited to the surface.   198 Figure 62 Compression isotherms and fluorescence intensity for 3 mol% BODIPY FL DHPE, 9.6 mol% cholesterol/DOPC monolayer in the presence or absence of added AmB formulation in the subphase. Formulation concentration was 80 %TD, added after deposition of the monolayer. Fluorescence images were subjected to a background correction process before having the resultant mean intensity calculated. Measurements were made at room temperature. Calculation of area per deposited molecule includes DOPC, cholesterol and fluorophore; a total of 6.5 x 10 15 molecules were deposited to the surface.   199 xviii Figure 63 Compression isotherms and fluorescence intensity for 3 mol% BODIPY FL DHPE, 9.4 mol% ergosterol/DOPC monolayer in the presence or absence of added AmB formulation in the subphase. Formulation concentration was 80 %TD, added after deposition of the monolayer. Fluorescence images were subjected to a background correction process before having the resultant mean intensity calculated. Measurements were made at room temperature. Calculation of area per deposited molecule includes DOPC, ergosterol and fluorophore; a total of 6.5 x 10 15 molecules were deposited to the surface.   200 Figure 64 AmB fluorescence intensity and surface pressure of 20 mol% AmB/DOPC mixed monolayer on Langmuir trough. The monolayer was formed with the trough barriers in the fully open position (time zero), and allowed to equilibrate while measuring fluorescence and surface pressure. At time —2500 s, a compression isotherm was begun, and the trough area held fixed for —400 s once the collapse pressure reached. The trough barriers were then reopened, and an expansion isotherm measured. The barriers were then held fully open for the duration of the experiment. Units of fluorescence intensity are arbitrary.   205 Figure 65 Fluorescence intensity and surface pressure of DOPC monolayer after FZ was injected behind trough barriers in the fully-open position. The barriers were held open throughout the experiment. The injection was made just prior to time zero. Units of fluorescence intensity are arbitrary. Penetration is seen to be mostly complete within —400 s, as shown by the fluorescence intensity having almost reached its maximum.  206 Figure 66 AmB fluorescence intensity and surface pressure for Gibbs monolayers formed by injection of FZ to the subphase. The trough barriers were set to a partly-open position and held fixed for the duration of the experiment. Injections of AmB were 18 initially, then two aliquots of nine 111, and one final addition of two pi, all of 0.95 mg/ml AmB as FZ. The final aliquot was injected above the microscope objective, just below the laser illumination spot on the surface. Units of fluorescence intensity are arbitrary.   208 Figure 67 AmB fluorescence intensity and monolayer surface pressure for FZ penetration into a DOPC monolayer at a biologically-relevant surface pressure. A dilute DOPC monolayer was formed on the trough surface and allowed to equilibrate before compressing it to 30 mN/m. Once at this surface pressure, FZ was injected to the subphase behind the barriers. The surface pressure was maintained at 30 mN/m for the duration of the measurement. Units of fluorescence intensity are arbitrary.   209 Figure 68 Schematic representation of possible modes of AmB interaction with the model membrane systems used in this work. Top: Interaction with lipid monolayer adsorbed to Hg drop electrode. Bottom: Interaction with floating lipid monolayer. A legend is given at right. Interaction modes are A: pore formation from only AmB molecules; B: insertion of AmB monomer into lipid layer; C: peripheral contact of AmB aggregate with lipid headgroups; D: peripheral contact of AmB monomer with lipid headgroups; and E: pore formation from both AmB and sterol molecules. No specific aggregate structure, pore stoichiometry or structure should be inferred.   214 xix LIST OF SYMBOLS AND ABBREVIATIONS Chapter 1 FDA^Food and Drug Administration (of the USA) MOA Mechanism of Action AmB Amphotericin B AIDS^Acquired Immune Deficiency Syndrome HIV Human Immunodeficiency Virus AMP Antimicrobial protein DNA^Deoxyribonucleic Acid GAS The Gas-Solution Interface Chapter 2 CMC^Critical micelle concentration IUPAC International Union of Pure and Applied Chemistry FZ Fungizone LDL^Low density lipoprotein HDL High density lipoprotein ABLC Abelcet (Lipid-complexed AmB) DMPC^Dimyristoylphosphatidylcholine DMPG Dimyristoylphosphatidylgycerol HTFZ Heat-treated Fungizone Chapter 3 IPE^Ideally Polarizable Electrode d Separation distance between plates of a parallel plate capacitor Dielectric Constant E^ Voltage q Charge 80^ Permitivity of Free Space C Capacitance or capacitor IHP Inner Helmholtz Plane OHP^Outer Helmholtz Plane GCS Gouy Chapman Stern (Theory) p(x) Charge as a function of distance from the electrode z,^ The signed charge of an ion The population of an ion i in a reference layer e The charge on the electron (1)^ Electrostatic potential 1(13 Boltzmann's constant T Temperature 4)2^ Electrostatic potential at the OHP Inverse of the Debye Length RC Resistor Capacitor xx metal^ Charge on the metal° °solution Ionic charge in solution Number concentration of ions in a reference layer Cdi^ Capacitance of the double layer pzc Potential of zero charge Xm^Surface potential Alpha phase Beta phase Thermodynamic property of the dividing surface P^ Pressure G Gibbs free energy A Area Chemical potential of species i y^ Surface tension Surface excess concentration MX An M+ X- metal halide electrolyte L A neutral interfacially-active ligand X^ Mol fraction 0 Surface coverage AC Alternating current Frequency Cbulk^ Bulk concentration K Ratio of adsorption rate constant to the desorption rate constant kadsorption Adsorption rate constant kdesorption^ Desorption rate constant R Ideal gas constant 7C^ Surface pressure (film pressure) a Interaction parameter in Frumkin isotherm O Oxidized form of reactant Half-wave potential iF(t)^Time-dependent Faradaic current F Faraday's constant Do^Diffusion constant of oxidized species re, Radius of electrode Nd^ Number density of active sites R Reduced form of reactant Lp^Length of Wilhemy plate Wp Width of Wilhemy plate Tp^Thickness of Wilhemy plate D p^Density of Wilhemy plate D s^Density of subphase solution H Immersion depth of Wilhemy plate g Acceleration due to gravity 0^ Contact angle ESP Equilibrium spreading pressure xxi So Alw^Spreading coefficient for organic on air-water interface A, Average area per molecule for two species i and j Aexcess^ Excess surface area A Gexcess Excess free energy of mixing AE Transmembrane potential DPPC^Dipalmitoylphosphatidylcholine DOPC Dioleoylphosphatidylcholine UV Ultraviolet 1.1Ji^ Wavefunction describing an initial state tiff Wavefunction describing a final state Pfi^ Transition dipole moment Dipole moment operator P Probability S^ Total spin quantum number tyre^An electronic state A vibrational state r Unit distance in electronic coordinates R^ Unit distance in nuclear coordinates E' The lower energy level of splitting (molecular exciton coupling) E" The upper energy level of splitting (molecular exciton coupling) E0^Ground state energy level E 1^First excited state energy level Q Quantum yield kF^Rate constant for fluorescence kNR Rate constant for all non-radiative pathways combined Observed lifetime of the excited state do^Radiative (intrinsic) lifetime of excited state FRET Fluorescence resonance energy transfer r Separation distance between donor and acceptor CCD^Charge coupled device Chapter 4 SAM^Self-assembled monolayer HOPG Highly-oriented pyrolytic graphite DSPC Distearoylphosphatidylcholine DSPE^Distearoylphosphatidylethanolamine DDAB Didodecylammonium bromide ppb Part per billion ACS^American Chemical Society WE Working electrode CE Counter electrode RE^ Reference electrode DAQ Data acquisition system SCE Saturated calomel electrode DC^Direct current N^ Number of replicate measurements HPLC High pressure liquid chromatography UBC University of British Columbia TD^ Therapeutic dose ASA Adapted simulated annealing SA Simulated annealing T^ Global 'temperature' parameter in SA or ASA 2-D Two dimensional N Number of parameters and number of dimensions in SA or ASA tp^ Trial point C-E Chemical-electrochemical Chapter 5 IR^ Infrared ECL Electrochemical luminescence Ru(bpy) 3'^tris-2,2'-bipyridylruthenium (II) ssDNA single-stranded DNA FTIR Fourier transform infrared Wavelength of light DPH^ Diphenylhexatriene TMA-DPH Trimethylammonium - diphenylhexatriene NA Numerical aperture BODIPY^Dipyrrometheneboron difluoride Chapter 6 SDS^ Sodium dodecyl sulfate So^Ground state singlet energy level of AmB S I^First excited state singlet energy level of AmB S2 Second excited state singlet energy level of AmB A Absorption F^ Fluorescence F.., Fluorescence intensity corrected for inner filter effect Fobs Observed fluorescence intensity A,„^ Absorbance at the wavelength of excitation A. Absorbance at the wavelength of fluorescence emission AmA Amphotericin A NY-DOC^Nystatin-deoxycholate (nystatin analogue to FZ) Chapter 7 BAM^ Brewster angle microscopy FLIM Fluorescence lifetime imaging microscope FRAP Fluorescence recovery after photobleaching DPPE^ Dipalmitoylphosphatidylethanolamine SOPC 1-Stearoy1-2-oleoyl-phosphatidylcholine vi Virtual instrument (Labview programme) DVD^ Digital versatile disk n Refractive index 0 Angle of incidence with respect to surface normal PMT^ Photomultiplier tube xxiv ACKNOWLEDGMENTS There is only one name on the title page of this thesis, but this does not reflect the important contributions from many people who assisted me in this work. Without their guidance, input, and support, this thesis may have never come to fruition. Firstly, I thank my two co-supervisors, professors Kishor Wasan and Dan Bizzotto for guiding me through this project. Having two supervisors has given me a broader perspective on science than I would have otherwise had. Your teaching, suggestions and direction certainly helped this project along, but they have also given me a skill set that I can apply to future endeavours. Kish, your constant support and enthusiasm were an inspiration. Dan, I appreciated your near-infinite patience, support for attending conferences even early on, and your encouragement of my interest in teaching. My colleagues in the lab, both past and present, deserve my thanks for their help, discussions on science and life, and friendship. These include Dr. Jeff Shepherd, Dr. Eduard Guerra, Dr. Emily Chung, Yanguo Yang, John Agak, and Amanda Musgrove. Your friendship and support have been significant to me. Jeff and Yanguo were excellent role models early in my graduate studies and I still wish I shared their natural talent for science. To my current colleagues Aya Sode, Jeffrey Murphy, Jannu Casanova, and Sabrina Lorenz, I wish you success in your projects. Remember that graduate study about more than the science alone! The technical staff of the UBC Chemistry Department have been very supportive. Ken, Raz and Des in the mechanical shop have machined numerous parts for me and were always humourous and patient in explaining the mistakes in my drawings! The electronics shop, particularly Dave Tonkin and Mark Castro, worked hard to troubleshoot our problems and were always happy to oblige when my soldering skills proved insufficient for the job at hand. Lastly, I thank my parents for encouraging me to wonder why. xxv DEDICATION This thesis is dedicated to my wife Tran, for all her help, encouragement, and support. Cam dn. Per ardua ad astra xxvi 1 INTRODUCTION To the first recipients of antimicrobial therapy, the mechanism of the drug's action and the methodologies used to discern it would have been largely irrelevant. Viewed as miracle-drugs initially, antimicrobials, including antibiotics and antifungals, have come to be commonplace in our lives. Their continued development is the result of wide-ranging new knowledge in the fields of biology, biochemistry, and medicine. These advances have fundamentally changed our global outlook on health; serious infections and diseases 100 or even 50 years ago are now often regarded as being minor nuisances. One fundamental aspect of these advances has been the development of new methodologies that have improved our understanding of drug mechanisms. Understanding a drug's mechanism is becoming increasingly important. Historically, new drug protocols often began with empirical evidence of treatment, with determination of the mechanism being much less important. However, the requirements of the American Food and Drug Administration (FDA) for the drug development process now explicitly demand mechanistic studies as part of phase one results reporting. It is not only government regulations that are increasing the importance of mechanistic work; the trend towards 'designer' drugs necessitates using the knowledge of broad themes of drug-cell interaction as a step in the discovery process. Also known as 'rational design', this approach parallels the vaunted structure-function relationship in biology; new drug candidates are proposed on the basis of our understanding of existing related compounds. Simply put, a full understanding of how and why existing drugs work is advantageous to new drug development. Our interest is in the mechanism of action (MOA) of the membrane-active antifungal Amphotericin B (AmB) and in developing methodologies and tools so that its mechanism can be better studied. Indeed, the themes of this thesis are: 1) Understanding the interaction of different AmB formulations with lipids at the membrane level, and 2) the advancement of novel experimental approaches to studying this interaction. The need for additional, and new, studies on AmB is underscored by the fact that AmB's MOA is still not fully understood after decades of research. Incomplete 1 understanding of its MOA hinders efficient reformulation of AmB to reduce the toxicity that limits its usage. Improved versions of AmB would mean the drug's usage could be widened. AmB is off- patent and relatively inexpensive, thus the ability to use it to treat a wider range of patients would be particularly beneficial to those in the developing world. AmB is often used to treat the opportunistic fungal infections that are prevalent among patients with acquired immune deficiency syndrome (AIDS). Since 25 million people are living with HIV/AIDS in sub-Saharan Africa alone, it is easy to understand the importance of AmB in the developing world. Substantial quantities of AmB formulations are also prescribed in the first world. The importance of developing new drugs cannot be understated. Even though the need may appear self-evident, it is worthwhile to consider the implications of new antimicrobials as a subclass. Antimicrobials are ubiquitous in our lives; the USA alone produces some 23,000 tons annually [1], mostly of antibiotics, to be prescribed in milligram quantities. As is well-known, many organisms have developed drug-resistance, rendering many previously-successful drugs useless. Very few antimicrobials have received FDA approval since the 1970's, although a new class of antibiotic, cationic antimicrobial proteins (AMPs), may break this drought. Without new antimicrobials, our quality of life will decline. The second theme is the development of novel methods for examining drug-model membrane interaction. It has been recognized that a drug's interaction with the cell membrane occurs on a size scale for which there is a general gap in knowledge. We may understand the drug at the molecular or atomic scale, and the membrane on the cellular scale, but the processes that occur in between are less well known. Knowledge at the membrane level is important because the interplay between cell membrane components (lipids, proteins) and drugs forms the basis of numerous medical applications, such as anaesthesia and drug delivery technology. The development of new experimental techniques and theoretical approaches to the quantitative study of membrane self- assembly, lipid-protein or lipid-drug interaction, etc. have been identified as being key to the new scientific field of lipidomics [2]; accordingly, the work of this thesis falls into this field. 2 Our new methods were designed around assisting our mechanistic studies on AmB at the membrane level, but it was anticipated that ultimately their true value might lie in broader use with other drugs or biologically-interesting compounds. For example, both AmB and the AMPs discussed earlier form pores or ion-channels in membranes, so it may be reasonably expected that some of the methods developed here could also be used to study how AMPs interact with cell membranes. Additionally, the class of antifungals to which AmB belongs, the polyenes, is made up of about 200 molecular species, but only AmB has low enough toxicity to be clinically useful. If a better understanding of polyene MOA led to polyene formulations with better selectivity, these polyene antifungals would represent a huge reservoir of possible new drugs. The method development work we have done has implications for other branches of science as well. As an example, we further characterized one of our model membrane systems as a precursor to attempting to use variations in its properties as a method to monitor AmB-membrane interaction. The system was an electrode-supported lipid monolayer, and we contributed to the understanding of the behaviour of organic thin films under the influence of electric fields. As an example of the utility of such knowledge, the Bizzotto laboratory (and others) are working towards the goal of controlled-release of drugs from electrode surfaces. The gist of the idea is to incorporate drug molecules into a self-assembled organic layer which can be displaced from the electrode under electric potential control. Our studies of AmB-lipid membrane interaction can be considered from at least two different angles. Studying the effect of AmB formulations on model cell membranes provides information on AmB- cell wall interactions at the molecular level, which is useful to understanding the drug's MOA. On the other hand, AmB can be considered a probe species, and similar experiments can be thought to provide understanding of how the model cell membrane behaves in the presence of a foreign contaminant probe. Depending on perspective, such foreign contaminants could be drugs, DNA, sterols, or membrane proteins, to name a few. That is, we are not confined to being interested in the added-component, but instead may use it to learn more about the substrate. In these two respects, the work of this thesis is significant because it gives us an improved understanding about the 3 properties and function of the model cell membrane, from which inference may extend to real cellular systems. 1.1 Outline of the Thesis Chapter 2 gives an overview of the usage of AmB, including details on the most commonly administered formulations, its toxicity and what is known of its mechanism of action. Chapter 3 details the theory and background required for a full understanding of the concepts and techniques used in the research. Most of the chapters reporting our research results are based on two or more of the broad background concepts discussed; full understanding of the results requires some familiarity with the theory. In particular, the explanation of electrochemical concepts is given in detail, since it is expected that most readers will have less familiarity with this material. Chapter 4 discusses the interaction of different AmB formulations with a model lipid membrane supported on an electrode surface. This work represents the first use of this particular model cell membrane system to examine the interaction between the lipid and a widely-used, commercially- successful drug. Notably, different formulations of the same active drug ingredient (AmB) produced different results, testifying to the sensitivity of the technique. The bulk of the work described in Chapter 4 has been published as an article in the journal Biochimica et Biophysica Acta - Biomembranes [3]. Chapter 5 describes method development work based on the model membrane system introduced in Chapter 4. Here, in situ fluorescence microscopy was applied to examine the lipid monolayer supported on the electrode surface. To render the monolayer fluorescent, a low concentration of lipophilic fluorescent dye was incorporated. The monolayer's capacitance and simultaneous fluorescence behaviour as a function of potential were characterized first in the absence of AmB, and then with AmB present. In part because of the phenomenon of metal-induced fluorescence quenching, this work gives insight into the potential-dependent molecular organization processes that occur on the electrode. The results of this work provide improved understanding of the basic cell membrane model system we have used so far and additionally contribute to the knowledge about the behaviour of adsorbed organic species under applied electric fields. Nearly direct 4 comparison can be made between this work and allied studies using organic species with different physical properties adsorbed on solid electrodes. The work of Chapter 5 forms a part of our publication in the journal The Analyst [4]. Chapter 6 reports our studies examining the intrinsic fluorescence of AmB formulations. We mapped the fluorescent properties of different AmB aggregation states by using a powerful surfactant to control the solubilization (and hence aggregation state) of AmB. Measurements at different aggregation states are of interest because AmB aggregation has been related to its toxicity and efficacy. Two related formulations of AmB showed differing fluorescence characteristics, reflecting their different aggregation states. Other spectroscopic characterization methods do not distinguish these differences in aggregation. This work lays the basis for new methods of studying the interaction of AmB formulations with lipid membranes because of its sensitivity to aggregation being above and beyond existing techniques. This work has been published in the journal Langmuir [5 ]. Chapter 7 discusses our method development work of measuring AmB interaction with floating lipid monolayers. Characterization of the interaction with both sterol-containing and sterol-free layers at the gasI solution (GIS) interface was made with both Langmuir trough techniques and in situ fluorescence microscopy using an incorporated lipophilic fluorescent dye. This represents a mapping of the drug-monolayer interaction at different surface pressures. Additionally, we developed a technique to monitor fluorescence of AmB from a monolayer, and used this to follow AmB penetration into sterol-free lipid layers. This work ties together our efforts to examine the direct fluorescence of AmB with our abilities to probe its interaction with a monolayer of controlled properties. Broad, over-arching conclusions are presented in Chapter 8, emphasizing the inter-relations between the different parts of the work. Suggestions for the future directions the work could take are also given. 5 2 AMPHOTERICIN B 2.1 Introduction Amphotericin B 1 is a product of Streptomyces nodosus, an actinobacteria isolated from a Venezuelan soil sample, and its use as an antifungal was first described in 1956 [6]. AmB's macrocyclic structure as shown in Figure 1 was determined in 1970 [7]. The structure has a rigid hydrophobic all-trans heptaene fused to a hydrophilic polyol chain. AmB is thus an amphiphilic molecule; its amphiphilic behaviour dominates our understanding of its interaction with cell membranes. Additionally, AmB is characterized by a polar headgroup of a carboxylic acid and an amine on the sugar group. Together, these groups make AmB zwitterionic in neutral aqueous solution and give the molecule the amphoteric property of its name. The pKa of the carboxylic acid proton is 5.6 while that of the amine is 10; AmB's isoelectric point is then 7.8. AmB weighs 924.1 g/mole and its length (-21 A) is approximately equal to the length of a phospholipid, and thus similar to half a lipid-bilayer thickness [8]. AmB is poorly soluble in most solvents due to its amphiphilic nature. Pure AmB in aqueous solution has a critical micelle concentration (CMC) of 6x 10 -7 M [9] but is more soluble (mM range) in the polar aprotic solvents dimethylsulfoxide and dimethylformamide. Below the CMC, AmB exists as monomers, while above the CMC AmB forms aggregates. The mean number of AmB monomers per aggregate is unknown and is thought to correspond to a wide size distribution [10]. AmB absorbs blue light strongly, with an extinction coefficient >10 5 M -1 cm' at 410 nm. The absorption spectrum of AmB depends on the drug's aggregation state; peak absorption for the monomer is 410 nm, aggregated solutions show a blue-shift to — 340 nm. AmB was recently determined to be fluorescent, with the monomer fluorescing between 500 and 650 nm, and aggregates identified as dimers fluorescing between 400 and 550 nm [11]. According to IUPAC nomenclature, the chemical name of AmB is: (1R-(1R*,3S*,5R*,6R*,9R*,11R*,15S*,16R*,17R*,18S*,19E,21E,23E,25E,27E,29E,31E,33R*, 35 S * ,36R* ,37S*))-33-((3-Amino-3 ,6-dideoxy-beta-D-mannopyranosyl)oxy)-1,3,5,6,9,11,17,37- octahydroxy-15,16,18-trimethy1-13 -oxo-14,39-dioxabicyclo(33.3.1)nonatriaconta-19,21,23,25,2 7,29,31-heptaene-36-carboxylic acid. 6 2.1.1 Use of Amphotericin B AmB's clinical use is for the treatment of systemic fungal infections, most commonly of Candida, Cryptococcus or Asperigillus species. AmB has additionally been used clinically in the treatment of many other fungal infections 2 . Such fungi may form part of the natural flora found in the human body (e.g. in the gut or respiratory tract) that are normally kept in check by the body's immune system. As such, patients suffering systemic fungal infections are typically already immuno- compromised. Large increases in the number of recorded fungal infections accompanied the rapid spread of AIDS in the 1980's and 1990's, although organ-transplant anti-rejection drugs, cancer chemotherapy treatment, and malnutrition are also important causes of immunodeficiency. Intravenous administration of AmB is the backbone of treatment for severe and otherwise fatal fungal infections because of AmB's broad spectrum, potency, and rapidity of action. Additionally, the conventional formulation of AmB is relatively affordable compared to other anti-fungal drugs. Illustrating the magnitude of AmB's importance is that the American national demand for conventionally-formulated AmB is approximately one million 50 mg vials per year [12]. AmB's use as an antifungal is particularly important as it is fungicidal whereas many other treatments only limit fungal growth (fungistatic effect). Despite its use as a front-line antifungal for more than 50 years, resistant strains are rare [13]. AmB is also the treatment of choice for the parasite-caused disease Leishmaniasis. Approximately 12 million people globally are thought to be currently infected with Leishmaniasis, and it kills 500,000 annually. Approved indications for the use of AmB therapy vary according to the formulation. The conventional form, Fungizone® (FZ - Bristol Myers Squibb, New York, NY), has been approved by the American FDA for treatment of aspergillosis (due to Aspergillus fumigatus), torulosis (Cryptococcus neoformans), systemic candidiasis (Candida spp.), North American blastomycosis, coccidioidomycosis, certain susceptible forms of Zygomycetes, sporotrichosis (Sporothrix 2 Less common fungal infections for which AmB therapy has been used successfully include those due to Cladosporium, Phialophora, Histoplasma capsulatum, and Paracoccidioides brasiliensis fungi. 7 OH OH^OH^OH^OH^OH 33^31^29^27^25^23^21 Amphotericin B schenckii), and infections due to susceptible forms of Conidiobolus and Basidiobolus. FZ is also indicated for American mucocutaneous leishmaniasis. In addition to its approved indications, AmB has been found to act as a potentiator of existing anti- cancer drugs [14 and references 43 & 44 therein], as a mediator of prion disease progress (transmissible spongiform encephalopathies such as the bovine form: mad cow disease) [15, 16], as a possible treatment for Hodgkin's Lymphoma in combination with acyclovir [17], to have potent activity against malaria-causing parasite Plasmodium falciparum [18], as a first alternative to benzimadazoles for the treatment of Alveolar echinococcosis [19 and reference 26 therein] and to be a stimulant of the immune system [20]. Additionally, a 2006 editorial in Clinical Infectious Diseases hailed reports of promising antifungal activity on administration of fungal heat shock protein with Amphotericin B as possibly marking the arrival of a 'third age' of antimicrobial therapy [21]. This approach represents combination therapy where an antifungal is administered along with an immunomodulator (the fungal heat shock protein in this case) to boost the host immune system; `reviews' of the outlook have been published [e.g. 22]. CHI Figure 1 Chemical Structure of AmB. Partial numbering of carbon atoms on macrolide ring is given. AmB is characterized by an all-trans heptaene on one side of the macrolide ring and numerous hydroxyl groups on the other side. 8 2.1.2 Toxicity of Amphotericin B Although widespread, the use of AmB is limited by its dose-dependent toxicity. Toxicity in the context of this thesis means the adverse effects of the drug which are detrimental to human health. Conventional AmB, FZ, is the de facto standard of AmB toxicity against which other AmB formulations are compared. Therapeutic use of FZ provokes both acute intravenous infusion toxicities which are usually reversible, and chronic kidney toxicity which may be permanent. It's administration is associated with a long list of side effects (Table 1), but the most important toxic effect is FZ's kidney toxicity. Acute renal failure has been reported for between 49 and 65% of patients receiving AmB treatment [23]. Accordingly, the patient's renal function and history is taken into account when determining if FZ therapy is suitable. Those patients with renal disfunction or who otherwise cannot tolerate FZ may be prescribed an alternative AmB formulation or a non- AmB anti-fungal. 2.1.2.1 Dose-Dependent Toxicity of FZ The dose-dependent kidney toxicity of FZ is a cumulative effect resulting from chronic treatment. The renal disfunction may be characterized by decreased renal blood flow, decreased rate of glomerular filtration (a standard measure of kidney function) and impairment of electrolyte re- absorption. Irreversible renal toxicity is associated with cumulative doses of > 4 g AmB [23]. At lower cumulative doses (0.5-1 g) acidosis of the renal tubules is common, but this effect generally is reversed on cessation of FZ therapy. More generally, the local concentration of AmB is important. For a typical adult dose of 1 mg/kg/day, maximum AmB concentration in blood plasma is approximately in the range 1-3 µg/ml [24]. Chronic treatment results in AmB build-up to higher concentrations in the kidneys, liver and lungs. At concentrations above 5 µg/ml, AmB is non-specifically toxic to cells (cytotoxic). The therapeutic index of a drug is a way to measure its relative effectiveness. The therapeutic index is the efficacy-to-toxicity ratio, often defined as the maximum tolerated dose over the minimum curative dose. Clearly, AmB's therapeutic index is limited by its cytotoxicity, resulting in a low index. The aggregation state of AmB is related to its toxicity, with aggregated AmB found to induce 9 cytotoxicity and damage to DNA (genotoxicity)[25, 26]. Monomeric AmB showed no genotoxic effect and only slight cytotoxicity related to an oxidative-damage mechanism. Table 1 Partial List of Toxic Side Effects of FZ Kidney toxicity including acute failure Vertigo & Hearing Loss Cardiac Arrest Anemia Nausea & Vomiting Acute Liver Failure Fever, Rigors, Chills High or Low Blood Pressure Low Blood K+ or Mg' Concentration 2.1.2.2 Mechanism of AmB Toxic Side Effects Studies have indicated that multiple mechanisms may be responsible for the long list of AmB side effects, in addition to those caused by pore formation and increased cellular permeability [27 and reference 8 therein]. Specific cellular and molecular mechanisms producing the toxic effects have not been elucidated, but certain themes have been implicated. It has been well established that AmB degrades via an auto-oxidation mechanism and this may result in peroxidation of cell membrane lipids [28-30]. The products of lipid oxidation within both low-density lipoproteins (LDL) and high- density lipoproteins (HDL) themselves are highly toxic and may be expected to provoke a physiological response [31, 32]. Tancrêde's group have shown that FZ induces structural changes in LDL and increases LDL oxidation [33, 34]. The inhibitory effect of AmB on various cellular enzymatic systems has been documented in older Soviet biochemical literature and was later explored in more accessible western papers [e.g. 35, and 36, 37]. More recently, studies have linked them with increased production of cytokines involved in signalling of inflammation [38 and reference 10 therein]. 10 2.1.3 Formulations of Amphotericin B The first approved formulation of AmB was FZ, consisting of a 1:1.8 ratio of Amphotericin B and the bile salt sodium deoxycholate, and buffered with sodium phosphates. Addition of the deoxycholate surfactant is required to provide sufficient AmB solubility for intravenous administration and results in a colloidal dispersion of AmB. Rinnert and co-workers used light- scattering techniques to find that FZ aggregates held about 2000 molecules of AmB on average [39]. Van Etten et al. used cryo-transmission electron microscopy to find as reconstituted FZ has an average micelle size of about 4 nm, as well as thread-like structures of aggregated micelles [40]. At the concentration used for intravenous injection, FZ consists principally of aggregated AmB. Several other commercial formulations of AmB exist, each designed to limit its toxic side effects. Of these, Abelcet® (ABLC - Enzon Pharmaceuticals, Bridgewater, NJ) and Ambisome® (Astellas Pharma, Deerfield, IL) are the most commonly used. ABLC is a lipid-complexed form of AmB, consisting of drug and lipid in 1:1 molar ratio. Two lipids are used, dimyristoylphosphatidylcholine (DMPC) and dimyristoylphosphatidylgylcerol (DMPG) in 7:3 molar ratio. The resultant complex holds AmB in an aggregated state and forms ribbon-like structures 2-5 in size [41]. Despite requiring administration of an AmB dose that is five-fold that of FZ, the formulation is better tolerated and has a decreased nephrotoxic effect. The higher dosages are required to maintain the same efficacy as FZ because a smaller fraction of the AmB is free in the plasma and thus available for anti-fungal activity. Acute toxicity of ABLC is similar to the acute toxicity of FZ. Ambisome is a unilamellar liposomal version of the drug. The liposomes are smaller than 100 nm diameter and consist principally of saturated phospholipids, cholesterol and AmB. Like ABLC, Ambisome is also given in higher AmB dosages than FZ but is better tolerated. Lipid-associated formulations of AmB give a different tissue distribution of AmB compared to FZ. They favour higher concentrations in the liver, spleen and kidneys, where fungal infections are likely to reside [42]. This represents a desirable targeting of the drug. Both ABLC and Ambisome have been found to retain the broad spectrum of antibiotic action of the FZ formulation, but FDA approved indications are fewer. Generally speaking, they are used when FZ is unsuitable. The lipid based formulations of AmB all have improved therapeutic index 11 compared to FZ, but have a serious cost disadvantage. Daily costs for an average adult on lipid- based AmB are $300-$1000 USD whereas with FZ this cost is about $5 USD [43]. All formulations of AmB are administered intravenously and require hospitalization because close monitoring of its toxicity is required. 2.1.3.1 Heat-Treated Fungizone Mild heat treatment of Amphotericin B as FZ has been suggested as a possible way of reformulating AmB with the combined benefits of retaining FZ's low cost and decreasing its toxicity. As far back as 1979, Ernst et al. showed in light-scattering experiments that heating of FZ solutions leads to an increase in aggregate apparent mass by some 400-fold. Knowledge that AmB aggregation state modulates its therapeutic efficacy led Gaboriau et al. to characterize the physical-chemical properties and toxicity of heat-treated FZ in two seminal papers in the late 1990's [10, 44]. The structure of HTFZ aggregates from a 5 mg/ml solution was found to be pleiomorphic (varied) cobweb structures of approximately 300 nm in their largest dimension using a combination of cryo- transmission electron microscopy and laser-light scattering [40]. HTFZ therapy resulted in decreased toxicity to both in vitro human and animal kidney cell lines and in vivo to rats and rabbits [40, 44-48]. The anti-fungal and anti-parasitic activity of AmB has been found to be unchanged by the heating process; HTFZ maintains a low minimum inhibitory concentration and minimum lethal concentrations similar to FZ [18, 45, 48]. 2.1.4 Amphotericin B Mechanism of Action Immediately following intravenous administration of FZ, some 95% of AmB is released from the deoxycholate and is taken up by lipoproteins (both LDL and HDL) within the bloodstream, particularly those with high cholesterol content [49]. The remaining 5% is carried unbound in the blood. The bloodstream disperses the AmB throughout the body. AmB is cleared from the serum quickly, depositing in the cholesterol-rich cell membranes in the spleen, liver, kidney and lung. Cholesterol-bound AmB in HDL is taken up via reverse cholesterol transport mechanisms while cholesterol-bound AmB in LDL is taken up via the cell's LDL receptors [50, 51 and reference 33 therein]. The LDL-bound portion is associated with acute and nephrotoxicity; HDL-bound component is associated with decreased toxicity [52, 53]. Complexation of AmB with lipid 12 promotes its association with HDL, part of the reason the lipidic AmB formulations show reduced toxicity. Clearance of AmB occurs via macrophages in the liver and through radical-initiated auto- oxidation [30]. Despite being used clinically for many years and the subject of many studies, AmB's mechanism of action is still poorly understood. Part of what is known has been determined by analogy to the related polyene antibiotics nystatin and filipin. Filipin is a cyclic pentaene and shares the general hydrophilic polyol chain and conjugated polyene structure of AmB, but lacks the sugar moiety. Nystatin is a closer analogue of AmB, differing in the placement of two hydroxyl groups but with the primary difference being the heptaene of AmB is replaced by a diene-tetraene moiety in nystatin. All three are thought to bind sterols and form pores as their active antifungal action. It was recognized early on that AmB is active against cell types whose membranes include sterols (e.g. eukaryotic cells such as mammalian and fungal) but not active against species with sterol-free cell membranes (e.g. bacteria). Interaction with cell membrane sterols was thus postulated to be an integral part of AmB's mechanism [54]. At the cell membrane level, AmB was found to induce significant changes in water, mono- and divalent cation, and small non-electrolyte molecule permeability across sterol-containing lipid bilayers. Both liposome and cellular (Acholeplasma Laidlawii) models were studied [26 and references 72 & 90 therein, 55, 56]. The loss of these intracellular species leads to cell death [26 and reference 9 therein, 27, 54]. These results led to the hypothesis that specific AmB interaction with fungal cell membrane sterol (ergosterol) was the basis of its action and that interaction with cholesterol in mammalian cell membranes was the basis of its toxicity. This premise has become known as the sterol hypothesis. Three broad fields of evidence exist for this sterol hypothesis: 1) Fungal and mammalian cell membranes contain sterol and are known to be sensitive to AmB. 2) Free sterols were found to antagonize AmB-induced K+ leakage of Candida and Mycoplasma cells [57]. 3) AmB has been found to have a favourable interaction with sterols using physical-chemical methods. Both studies of AmB binding to sterol-containing phospholipid vesicles and solution-based binding experiments confirmed more specific binding to ergosterol than cholesterol [58-60]. The binding 13 constants for AmB with cholesterol and ergosterol were found to be 2.1 x10 5 and 1.66 x10 6 M -1 , respectively [61]. This similarity in binding constant is reflected in AmB's relatively low therapeutic index; the poor specificity of AmB binding means the toxic dose is only modestly higher than the minimum effective dose. Ergosterol and cholesterol share similar chemical structure as shown in Figure 2. Ergosterol has two additional double bonds, one within the ring structure and the other on the side chain, as well as an additional methyl substituent on the side chain. There is conflicting experimental evidence on the mode of AmB-sterol binding. The heptaene of AmB may interact hydrophobically with various portions of the sterol. Computer modelling as well as experimental studies have suggested both the nature of the sterol side chain and of the sterol ring structure to be the dominating governors of binding efficiency [62 and references 8, 10 & 22 therein]. The sterol specificity of AmB has been suggested to be in part due to differences in cell transmembrane potential [63]. Additionally, these differences are proposed to be the cause of AmB's non-uniform tissue distribution [64]. There is a potential-dependence to the opening and closing of AmB channel, but the mechanism of this is not certain. The mechanism might be related to an increase in antibiotic intramembrane or surface concentration. A third possible mechanism is a potential-sensitive lipid conformation which favours insertion of pre-formed AmB aggregates (`pre-pores') into the membrane from a surface-bound position [63]. Complicating study of AmB's interaction with sterols is the question of AmB aggregation state. Gruda et al. were the first to recognize the relation between AmB aggregation state and its therapeutic effectiveness [58]. Using absorbance spectroscopy of AmB binding to solvent-dissolved ergosterol, they noted monomeric AmB showed little binding to ergosterol, but that the presence of small aggregates (possibly dimers) resulted in near immediate binding. Further investigations confirmed aggregation state dependence of AmB activity, with AmB monomer active against ergosterol-containing membranes only and AmB aggregates active against both cholesterol- and ergosterol-containing membranes [65-68]. Since these studies, considerable effort has been made to develop AmB formulations that deliver the more selective monomer. This is challenging drug- delivery task; due to AmB's low CMC, direct administration of a normal lmg/kg/day dose would 14 Cholesterol HO HO Figure 2 Chemical structures of the fungal cell sterol ergosterol (left) and mammalian cell sterol cholesterol (right). Ergosterol has two additional double bonds, one in the ring structure and one on the side chain. Additionally, ergosterol has an extra methyl group on the side chain. require about 120 litres of AmB solution [65]. The trick then is to deliver the AmB in a relatively concentrated solution but to allow only monomeric AmB release from the formulation. 2.1.4.1 Increased Permeability due to AmB-Sterol Pores The increased permeability of sterol-containing cell membranes on exposure to AmB has been related to sterol-AmB complexes forming pores (ion channels) in the membrane. Two possible mechanisms have been proposed. The first relies not on direct AmB-sterol interaction, but postulates that sterol type and concentration modify the physical properties of the cell membrane such that AmB may become incorporated more readily and form pores [69-71]. The second mechanism is that of direct interaction between sterol and AmB in order to form pores. Such pores are thought to involve sterol and AmB in a 1:1 or 2:1 ratio with 5-9 AmB molecules per pore [62, 72, 73]. The structure of the pore is thought to be a collection of AmB molecules with their long axis oriented perpendicular to the membrane surface and mutually parallel. Arranged in a cylindrical shape, the polyol side of the ring is pointed towards the inside and the hydrophobic heptaene chain on the outside. The polar sugar moiety remains outside the membrane and is exposed to aqueous solution. In this model, sterol and AmB alternate around the cylinder. This is known as the barrel-stave model of AmB pore [56]. Theoretical pore diameter has been calculated to be 7-10 A [56]. Gruszecki et al. have recently used atomic force microscopy to image an AmB pore in DPPC monolayer, finding internal diameter of 6A [74]. 15 Using liposome models, the properties of the pore formed, and by inference the structure of the pore, have been found to depend on the nature of the sterol present, the lipid composition of the membrane, the AmB concentration in the membrane, the AmB aggregation state in solution, and the liposome size [59, 68, 75-77]. The average open time of the pore is dependent on the nature of the sterol; with ergosterol leading to longer average open times [63]. When AmB is added to one-side of a lipid bilayer model system, it has been proposed that permeability is due to a barrel-stave pore arrangement which may either span the bilayer (aided by bilayer thinning) or shuttle up and down within the bilayer [77, 78]. Cohen's work has suggested that two types of AmB pores may be formed [79]; at low concentrations non-aqueous channels allow permeation of monovalent cations resulting in collapse of the transmembrane potential but do not kill the cell. Higher concentrations results in larger aqueous pores which pass water, cations, anions and other species. These AmB/sterol mediated pores lead to a fatal increase in intracellular pH [80]. 2.1.4.2 Sterol-Independent Pore Formation The sterol-hypothesis has not gone without criticism. The fact that AmB reportedly binds to sterols is not definitive proof of the hypothesis [26]. The fact that bacteria, which are not susceptible to AmB, do not have sterols in their membranes is also not clear proof of the hypothesis. It has been shown that polyene-insensitive bacteria are unable to synthesize lecithins (phosphatidylcholine lipids), while polyene-sensitive organisms have in their membranes not only sterols but also lecithins [81]. This dated suggestion that the presence of lecithins in the cell membrane is important has received recent support, with Hac-Wydro et al. asserting that DPPC (di-C16:0) can be considered a key factor influencing the antifungal activity of both AmB and nystatin [82, 83]. Additionally, some bacteria (sterol-free cell membranes) have been found susceptible to various related pentaenic polyene antifungals [84], suggesting that, at least for some bacteria, sterol absence is not sufficient to prevent polyene antibiotic action. An additional, if somewhat anecdotal, questioning of the sterol hypothesis arises from the limited extent of red blood cell hemolysis observed upon AmB intravenous administration. This is in opposition to expected result when relatively concentrated AmB solution first meets cholesterol-containing cell membranes. 16 Numerous studies reveal pore formation in sterol-free liposome model systems [59, 61, 70, 85-87]. Direct interaction with the lipid is thought to be the mechanism of AmB insertion into the membrane. Such pores are thought to be more likely to form in osmotically-stressed cells [76, 87, 88]. Additionally, it is known that the presence of sterol is not required for AmB-induced permeability when the lipids of the membrane are in the gel-state [26]. According to Cohen's theory, only non-aqueous pores are formed in sterol-free vesicles [89]. The field of sterol- independent pore formation remains unresolved. A summary of the possible mechanisms of AmB action, both sterol-free and with sterol, is given in Figure 3. 2.2 Summary of AmB Characteristics AmB remains a drug of choice in the treatment of systemic fungal infections, but its use is limited by its toxicity, particularly that of the FZ formulation, which is the most widely used. Approved formulations with decreased toxicity are too expensive for widespread application. An improved understanding of AmB's mechanism of action is needed to aid the development of an inexpensive, effective AmB formulation. One possible advance in this direction is HTFZ, formed by simple heating of FZ, which shows decreased toxicity in preliminary studies. Minimal work on HTFZ's mechanism of action has been carried out. At the cell membrane level, AmB is known to increase membrane permeability, particularly when sterols are present in the membrane. AmB binding to sterols to form aqueous pores within the bilayer is the proposed rationale for the increased permeability and accordingly for AmB's mechanism of action towards both fungal and mammalian cells. This explanation, the sterol- hypothesis, has been challenged by studies that suggest pore formation even in sterol-free membranes. Probing the sterol dependence or independence of AmB's action was one goal of our studies of AmB interaction with model cell membranes. Chemically, study of AmB is complicated by its aggregation behaviour in most solvents. This is of prime importance, since the aggregation state of the drug has been correlated with its effectiveness and toxicity. All reports of AmB effectiveness within a particular biological model must be read with care to note the AmB aggregation state which produced these results. The 17 approach used in this thesis was to examine AmB formulations, rather than purified AmB, and to introduce them to model membranes at concentrations similar to those used clinically. UV-visible absorbance spectroscopy easily differentiates AmB monomer and aggregated state. Aggregated AmB displays blue-shifted spectra relative to the monomer. Additionally, AmB is weakly fluorescent and the fluorescence emission is different between monomer and aggregated forms. After characterizing the relatively new discovery of AmB fluorescence, we have shown that this property can be used for in situ monitoring of AmB within model membrane systems.  AmB Lipid^SterolFZ (AmB micellar dispersion) b Equilibrium? Aggregate Monomer Pre-pore' AmB channel formed when cholesterol, ergosterol or no sterol present AmB channel formed when ergosterol present Figure 3 Schematic summary of possible AmB mechanism of action. FZ administration supplies AmB both as aggregates and as monomers. It is believed AmB monomer permeabilizes membrane only if ergosterol is present. AmB aggregates are thought to form pores if cholesterol, ergosterol, or no sterol is present. Pore-formation in cholesterol- containing membranes is the basis of AmB's toxicity. Interaction of AmB with lipid headgroups at membrane surface is described as a 'pre-pore'. The schematic representations used for lipid and sterol are given in the upper right corner. 18 3 THEORY AND BACKGROUND 3.1 Introduction This chapter outlines the theory behind the techniques used in this thesis. The span of topics is quite broad, ranging from electrochemistry, to surface science and spectroscopy. The fundamental theories which describe the electrodelsolution interface and the behaviour of partly-blocked electrode surfaces are presented first. The structure and properties of monolayers at the gasIsolution interface are described, as well as discussion of models for cell membranes. Principles of absorption and fluorescence spectroscopy are given including the topics of fluorescence near metal surfaces and molecular exciton theory. Details of the epi-fluorescence microscopy arrangement used for in situ work then follow. 3.1.1 Electrochemistry of the Interfacial Double Layer Chemistry is largely about interactions between molecular or atomic species. Electrochemistry is no exception, though one or more species is charged and the interactions often occur at a phase boundary. Throughout this thesis, phase boundaries will be denoted with the I symbol. For example the gas-solution interface is thus the GIS. Our interest in events occurring at an interface brings with it special considerations. Consider a charged metal surface (an electrode) immersed in a conductive aqueous solution. In this case, electrostatic attraction or repulsion of ions in the region near the electrode creates an asymmetric environment. A positive metal surface will result in a localized increase in anion concentration and decrease in cation concentration. Further asymmetry may result from a polar solvent adopting a preferred dipole orientation in response to the electric field. The interfacial double layer region encompasses all the charged species and solvent extending from the electrode surface out to the bulk of solution. Double layer theory has been developed to aid our understanding of this interfacial region. 3.1.1.1 The Helmholtz Model is akin to a Parallel-Plate Capacitor Helmholtz proposed that as the charge on a metal electrode is to be found in a thin surface layer, the required counter-balancing charge found in solution would also be in a thin layer. This is analogous 19 to a parallel-plate capacitor, which approximates so-called ideally polarizable electrode (IPE) behaviour. An IPE is a idealized electrode for which no charge-transfer current flows regardless of the potential applied. Charge-transfer current is often called Faradaic current and is distinguished from capacitive current. Showing no Faradaic current flow the IPE thus behaves similarly to a parallel plate capacitor which shows no charge transfer with applied voltage. IPE behaviour in real systems is approximated when charge-transfer currents are thermodynamically or kinetically unfavourable. Mercury in contact with de-oxygenated potassium chloride solution approximates IPE behaviour over a wide potential range (-2 V). The idea that the electrical characteristics of the polarized electrodejsolution interface are similar to those of a capacitor is important in the understanding of the interface and thus the basics of capacitance are reviewed here. A parallel-plate capacitor is characterized by the separation distance between the plates (d) and the dielectric constant of the material between the plates (c). Application of a voltage (E) to the capacitor results in stored charge (q) according to equation (1), with e o being the permittivity of free space. ESOC =^- q- d E A capacitive current can be measured if the potential across the capacitor varies with time. dq c  dE dt dt A circuit consisting of some number j of capacitors in a series arrangement is mathematically equivalent to that of one capacitor according to the inverse sum: (1) (2) 1 1^l^1^n — ...+^= E —CI c2^CI^= CI (3)CEquivalent 20 3.1.1.2 Gouy-Chapman Theory introduces the Diffuse Layer Concept Helmholtz theory incorrectly predicts that double layer capacitance is constant with varying potential. Gouy and Chapman rationalized that since the solution conductance is much lower than that of metal, the charge built-up in solution would not be in a thin layer, but spread out in a diffuse layer. The diffuse layer exists due to the balance of the electrostatic forces attracting and repelling ions at the interface and thermal fluctuations disrupting this ordering. With greater charge on the electrode, the electrostatic forces are increased while the thermal fluctuations remain constant, causing the diffuse layer to thin. This is analogous to varying the thickness of the capacitor's dielectric (d in equation (1)), and rationalizes capacitance's dependency on potential. 3.1.1.3 Gouy-Chapman-Stern (GCS) Theory Describes the Solution Side In the Gouy-Chapman model, ions are treated as point charges, leading to the prediction of very high double layer capacitance at large potential due to unlimited thinning of the diffuse layer. Stern introduced restrictions on ion location in solution. As ions have finite size, the closest approach to the electrode surface must be the ionic radius. Solvated ions can only approach to within their solvated radius. The Inner Helmholtz Plane (IHP) is defined as the plane of closest approach for ions which are specifically adsorbed to (chemically-interacting with) the electrode. The Outer Helmholtz Plane (OHP) is defined as the plane of closest approach of solvated cations, sometimes denoted as distance x 2 from the electrode as is shown schematically in Figure 4. The region from the electrode surface to the IHP is known as the inner layer, while the region from the IHP to the bulk of solution is the diffuse layer. There is still charge balance between metal and solution, divided as: — a metal = asolution = Ginner layer + a diffuse layer^ (4) These two layers may be thought of as being separate capacitors arranged in series, leading to the total double layer capacitance according to 1^1^1 Cdouble layer - Ginner layer ▪ Cdiffuse layer ^ (5) 21 M ^ IHP OHP (om 42 Diffuse layer Solvated cation Metal Specifically adsorbed anion O = Solvent molecule q Figure 4 Schematic structure of the charged electrodelsolution interface. A negatively charged electrode will have solvent and possibly also have specifically adsorbed anions at the surface. The plane through the centre of specifically adsorbed anions defines the Inner Helmholtz Plane (IHP), while the plane of closest approach of hydrated cations defines the Outer Helmholtz Plane (OHP). The inner layer lies between electrode surface and IHP; the diffuse layer lies between IHP and bulk solution. Reprinted with permission from [90], Copyright 2001 John Wiley & Sons. 22 Examination of GCS theory in a quantitative fashion gives analytical solutions for the double layer capacitance and the variation of potential away from the electrode surface. The derivation is only traced here, a complete derivation is given in [90]. Consider first the diffuse layer; the region from the OHP to the bulk of solution may be divided up into layers each of infinitesimal width. The ionic concentration (hence charge) in any layer is related to that of a reference layer in bulk solution far from the electrode via the Boltzmann distribution, and the charge per volume in any layer is: _z,e0 p(x) = z i n ° e kBT (6) Where p(x) represents the total charge at distance x (x ^ x2) from the electrode, z, the signed charge on ion i, II,' the species i ion population in reference layer, e the charge on the electron, 43. the potential with respect to the bulk of solution, kB the Boltzmann constant, and T the temperature. The Poisson equation from electrostatics provides that the charge is related to the potential by: d 2 0 P(x) = Eeo dx 2 Equating equations (6) and (7), integrating, and applying the special case of only z:z electrolytes (e.g. KCl) yields that the potential gradient across the diffuse layer varies with the hyperbolic sine of potential. 1 CO " 81cE Tn° 2 " ze0 = — sinh (8) dX X> X2 Ego^) 2kBT) Integration of equation (8) between the limits (13, and (1)2 (potential at distance x and x 2 , respectively) relates the potential in the diffuse layer with distance from the electrode, giving the potential profile in equation (9). (7) 23  zecb 4kB T zecb2 ,4k B T tanh - e _K(x-x2) (9) tanh Where lc is given as: K - 2n Oz 2 e2 2^ (10) ego kB T The inverse of lc is known as the Debye length and is important in characterizing the diffuse layer thickness. Somewhat analogous to the RC circuit time constant T, one Debye length contains 63.2% of the diffuse layer charge, five Debye lengths contain 99.3% of the charge. For a 0.1 M solution of a 1:1 electrolyte at 25 °C, the Debye length may be worked out to be 9.6 A. Under the assumption that no specific adsorption occurs, the electric field across the inner layer is constant at (c14/dx) x=x2 and the potential drop across the inner layer is linear. The potential drop across the whole double layer region is shown in Figure 5. The linear portion across the inner layer is expressed as: 0o = 02 - X2 Consideration of a Gaussian enclosure extending from the OHP to the bulk and application of Gauss' law can be used to relate the charge on the electrode to the potential at the OHP, giving: anietal = °solution = —66o r dO" dx x=x2 = (8k B Teso n ° ) 2 sink 1 0 \ze 2 2kB Ti (12) 24 Differential capacitance of the double layer is found by substituting in (1) 2 using equation (11) then differentiating equation (12) with respect to .4) at constant chemical potential and rearranging: 1 2 1  Cyr^eg0 ( 2 EEO Z 2 e 2 n ° 2^( Zed ^ cosh^2 k BT j^,2k B T (13) Which can be seen to contain terms for the capacitance of inner and diffuse layers according to equation (5). Note that the capacitance of the inner layer is independent of potential, while that of the diffuse layer is dependent on potential. The overall double layer capacitance will always be dominated by the smaller of the inner and diffuse layer capacitance, Cf. equation (5). For low electrolyte concentration, Cd1 has a minimum near the potential of zero charge (PZC) because the capacitance of the diffuse layer is low, as in Figure 6. The minimum at the PZC (6m=0, (1)2=0) occurs because no excess charge exists in the diffuse layer to balance the zero charge on the metal. Note that GCS theory predicts the double layer capacitance to be constant for potentials away from the PZC. There, the diffuse layer capacitance is larger than that of the inner layer and so the inner layer capacitance dominates. Experimentally, the double layer capacitance does vary, and this is partially due to potential-dependent changes in the structure of the dielectric in the inner layer. Specific adsorption of ions (e.g. on Hg) has the effect of varying the potential gradient in the inner layer. Adsorbing ions with charge opposite to that on the electrode causes an increase in the gradient, while similarly charged ions causes a decrease. Put another way, specifically adsorbed anions will cause the potential at the OHP42 , to be more negative. 3.1.2 The Jellium Model of the Electrode Surface The GCS model effectively treats the metal as an inert holder for a surface layer of charge. However, this charged surface layer has properties of its own, as described by the self-consistent jellium model. In this model, the positive metal nuclei are described as a semi-infinite background of positive charge. The electron density profile from the bulk metal towards the metal surface is calculated considering nuclei-electron and electron-electron interactions. Figure 7 shows the 25 140 120 100 80 E 6 60 — as>' 9 E U 20 Linear profile to r2 0 2 = 100 mV Diffuse layer 1 10^20^30 x, A 40 Figure 5 Potential drop across the double layer, calculated according to Gouy-Chapman-Stern theory for a 10' M 1:1 electrolyte at 25°C assuming a potential of 100 mV at the Outer Helmholtz Plane (x 2). The compact layer is an alternate name for the inner layer. Reprinted with permission from[90], Copyright 2001 John Wiley & Sons. 26 4r ...„.• High [electrolyte] • Moderate [electrolyte] Contribution from inner layer capacitance^ *- Low [electrolyte] Dip caused by small diffuse layer capacitance dominating Rational Potential Scale (E-Epic) - Figure 6 Differential capacitance near the potential of zero charge (pzc) according to the Gouy-Chapman-Stern model. pzc can be found experimentally by observing the dip in capacitance at low electrolyte concentrations. Adapted from [90]. 27 -0.5^0^0 5 OISTANCE (FERMI WAVELENGTHS' 1.0 Figure 7 Calculated electron density at a metal surface according to the jellium model. The background positive charge from the metal nuclei shows an abrupt transition at the metal surface. Within the metal, electron density shows Friedel oscillations. Electron 'spill-over' to the outside of the metal is present but decays rapidly with distance from the surface. r s represents the radius of the sphere which contains one free electron; two cases are shown. Distance is shown in Fermi wavelengths, which varies from metal to metal but is approximately 0.5 nm. Reprinted figure with permission from [92], Copyright 1970 American Physical Society. 28 calculated profile. The electron density spills over the surface and shows a deficiency just inside the surface, creating a net surface dipole and surface potential, xm. The electron spill-over is thought to extend 1-2 A from the metal surface [91]. 3.1.3 The Gibbs Adsorption Isotherm Describes the Interface Thermodynamically The Gibbs adsorption isotherm relies on the consideration of the interfacial region between two phases a and 13 in relative terms. An imaginary dividing plane of unit area is set between the two phases as shown in Figure 8, and properties of the bulk phases assumed to extend to the dividing line. The concentration of chemical species at the dividing plane can be determined by comparison of the concentration in the whole system with the concentration in the two phases. Taking o to denote a property of the dividing plane, we can write: = n - a - (14) which states that the amount of species i in the dividing plane (the surface excess, IV) is equal to the amount of species i in the whole system minus the amounts to be found in pure phases a and 13, assuming they extend to the dividing line. Such a division between dividing plane and phases a and 13 may be made for any extensive property. Doing this for free energy at constant T and P results in: ( X \ dGa = ■ - 0A ) dA + E ,u, ^ydA + E ,u, n, (15) Where y represents surface tension. Integrating and taking the cross-derivatives (analogous to derivation of Gibbs-Duhem equation) yields the Gibbs Adsorption Isotherm, relating surface tension with surface concentration: nta - dy = E F^ A ; with F 1 a = ^(16) 1-7 is known as the surface excess concentration and represents the concentration of species i in the interfacial region. Notice that the value of n i° (and hence IV') will depend on the location of the 29 Gibbs dividing plane, meaning that an absolute value of P i° is not directly measurable. Relative surface excess is a measurable quantity and is relative to a reference component (often the solvent). a) Actual ^ b) Model a  a Interfacial region Hypothetical Gibbs dividing surface Figure 8 Comparison of the interface between phases in real and model two-phase systems. In actuality, panel a), an interfacial region exists between phases a and 3. The model system, panel b), shows the placement of the Gibbs dividing line separating the two pure phases. 3.1.4 Extension of the Gibbs Adsorption Isotherm gives the Electrocapillary Equation Consider a system of an ideally-polarized electrode in an aqueous metal-halide (MX) electrolyte with a neutral interfacially-active species L in solution. Writing the Gibbs Adsorption Equation for this system and simplifying substantially gives the electrocapillary equation: - dr = o-mclE_+ XMx  rF M+ -^H 0 2XH0 2 du,, + F ni - ^ F H 0 d,um^(17) ZH2o 2 With x being mol fraction and E. the potential of the metal electrode relative to the reference electrode, which is sensitive to anion X - concentration. The terms in brackets are relative surface excesses (relative to water). Under conditions of constant temperature, pressure and chemical potential this reduces to the Lippmann equation: am = — ( °Y\OE) T,P,A1 (18) The Lippman equation shows that the potential-dependent charge on the metal is given by the negative slope of the interfacial tension - potential curve (electrocapillary curve). Such curves are 30 typically approximate concave-down parabolae as shown in Figure 9. The maximum in the parabola is called the electrocapillary maximum and can be seen through equation (18) to correspond to the PZC since the slope at the maximum is zero. Electrocapillary curves have been used extensively in the study of adsorption of both ionic and neutral species [93, 94]. Other cross- derivatives of the electrocapillary equation provide expressions for the surface excess of the interfacially-active ligand L and the electrolyte cation M. 3.1.4.1 Surface Coverage can be Calculated from Surface Excess The electrocapillary equation, linking surface tension and relative surface excess, provides a way to determine the coverage of adsorbate on the electrode. Surface coverage, 0, may be expressed as Prmax• rmax is the maximum surface excess, determined by plotting interfacial tension against the natural logarithm of bulk concentration for the potential of maximum adsorption (typically the PZC). If the capacitance of the uncovered and fully covered electrode are known, the capacitance at any other coverage may be determined by assuming a parallel capacitor model of the surface, according to: C = C0=0 (1 — + C0=1(9) (19) 3.1.4.2 Adsorption of Neutral Species can also be Measured via Capacitance As functions of potential, both the capacitance and charge density can be determined from measured electrocapillary curves according to equation (20). The differential capacitance of the interface is the negative second derivative of the electrocapillary curve or the derivative of a charge density - potential curve. y C differential = \ OE 2 / T,P,pi ^p; ( aam  OE_ (20) Figure 10 shows schematically the relation between surface tension, charge density and capacitance for an electrode in contact with a surfactant solution. Columns 1 through 4 represent cases of increasing bulk surfactant concentration. In cases 1 and 2, no surfactant adsorption takes place, denoted as state A, while surfactant adsorbs on the electrode in cases 3 and 4, denoted state B. The 31 ' ocr^£5'a" ^Owe_  r+  _ , 5E, 0 \de' , OE pL (21) ( Oa' C — \OE) surface tension at the electrodelsolution interface will always be minimized, and the tension is decreased when surfactant adsorbs. In cases 1 and 2, the charge density curve is linear, in keeping with differentiation of the (idealized) parabolic electrocapillary curve. Cases 3 and 4 show sharp changes in charge density upon adsorption and desorption. Further differentiation gives the capacitance, shown here as constant in the absence of adsorption, with a lowering of capacitance noted upon adsorption. Importantly, the capacitance may be measured directly using AC potential perturbation techniques and this is a more sensitive method than measuring interfacial tension and double-differentiating. Capacitance measured directly with AC perturbation can be integrated to give charge density or interfacial tension if desired. Adsorption of neutral species to the electrode surface can be measured capacitively if the dielectric constant of the adsorbate differs from that of the electrolyte solution. Equation (1) shows that if the dielectric at the interface changes from that of water (-80) to that of an organic species (varies by species, but —1-10), the capacitance measured will decrease. Additionally, changes in the double layer thickness upon adsorption may also affect the capacitance. A capacitance plot for mercury in the presence of a neutral organic generally shows a depression roughly centred about the PZC and so-called pseudocapacitance peaks further positive and negative. The bottom panel (capacitance) of Figure 10 could, for example, represent the changes which occur on adsorption of straight-chain butyl-alcohol (C 4H9OH). The pseudocapacitance peaks arise from potential-dependant changes in molecular surface coverage 0, and contribute to measured capacitance via: Which shows that measured capacitance can have two sources, true electrical capacitance resulting from change in charge density with potential, and pseudocapacitance from changes in charge density with surface coverage induced by changing potential. Capacitance measurements can thus be used to monitor potential-induced changes in adsorbate surface coverage. The pseudocapacitance is 32 a ?360 z0 6 340zW 1.- Q-) 320 0 4 tct 300 41 H- z 280 260 QS^06^Q4^0.2^0^-0.2 -0.4 -Q6 -08^-10^-12^-1.4 RATIONAL POTENTIAL,C IN VOLTS Figure 9 Electrocapillary curves for mercury in contact with aqueous solutions of the salts indicated. The typical parabolic shape of electrocapillary curves is observed. Units of surface tension used here, dynes/cm, are equivalent to mN/m. The rational potential scale used on the x-axis has the PZC as the zero point. Reprinted with permission from [93], Copyright 1947 American Chemical Society. 33 2) B _0A +0 C A^A^A^A^A^A -I B r 1B r -E -E -E -E Figure 10 The inter-relations between surface tension, charge density and differential capacity are shown for four cases of bulk surfactant activity (increasing 1) to 4)). States A and B represent when the electrode is uncovered by surfactant and when surfactant is adsorbed, respectively. The surface tension of the system as a function of potential will always follow the lowest possible path, shown here in solid line. Stepped changes in charge density occur for changes in adsorbate coverage (e.g. adsorption to and desorption from the electrode). As capacitance is the derivative of charge density with potential, peaks are observed in capacitance curve when coverage changes (shown schematically as vertical dashed line). Adapted from [95]. 34 dependant on the frequency of the measuring AC perturbation, and decreases from an equilibrium value for w=0 to zero at infinite frequency. 3.1.5 Surface Coverage 6 is often Modeled with the Langmuir Isotherm A simple but often used theory describing the thermodynamics of adsorption is the Langmuir adsorption isotherm. The fractional surface coverage 0 depends on the bulk concentration of the adsorbate, CB uik, a constant describing the ratio of adsorption rate constant to desorption rate constant, K, and the portion of the surface which is unoccupied, 1-0: 0 = C Bull,^- (22) The value of K is related to the free energy of adsorption AG IDads through an Arrhenius type expression: K- ^ — e kadsorption k desorption ^RT ^ (23) A link can be drawn between this free energy of adsorption if the isotherm is stated in terms of surface excess concentration rather than coverage, by equating 0 to F/I'max in equation (22): ^F = C Bulk K(F max — F ^ (24) Combining this with the results of the Gibbs Adsorption Isotherm allows film pressure t (extent of interfacial-tension lowering) to be related to surface excess concentration. R is the universal gas constant. = 78=0 — = — RIF max in 1 — ^■^F max ) (25) If an isotherm other than Langmuir' s had been considered, this relationship between film pressure and surface excess concentration would be found to be different. Using the Henry isotherm, valid in the limit of low surface coverage or low bulk adsorbate concentration, the relation is simply: n- = — RTF^ (26) L‘Gadsorptinn 35 1a =^a = Henry = —2 15 12 E 9 z E 6 3 0 0 0.2 0.4^0.6 0.8 1 While use of the Frumkin isotherm, which accounts for attractive (interaction coefficient a positive) or repulsive (a negative) adsorbate-adsorbate lateral interaction, produces equation (27). = — RTT max ( ,F inax ) In I- + a 2 - (27) Figure 11 shows the calculated variation of film pressure with adsorbate concentration for the different isotherms. f3c/rmax Figure 11 Calculated relation of surface pressure with bulk concentration of the adsorbate under Langmuir, Henry and Frumkin isotherms. In limit of low coverage, all converge to Henry isotherm linear dependence on concentration. Values of a are lateral interaction parameter for Frumkin isotherm; positive values indicate attraction. Reprinted with permission from [96], Copyright 1992 VCH Publishers. 36 3.1.6 Potential Step Chronoamperometry involves Diffusion to the Electrode Potential step chronoamperometry is one of many electrochemical methods for determining the concentration of a redox-active species in solution. All these methods relate rate of charge transfer at the electrode surface to the bulk concentration of the redox active species. The current-potential characteristic for these techniques depends on the type of potential perturbation, on mass transport of the species, and on the species' reaction kinetics. We will consider here only the response of a species with facile (fast) kinetics to a large potential step, such that mass transport limitation dominates over kinetic limitations. Consider an electroactive species, 0, in solution in its oxidized state. The electrochemical behaviour of 0 is characterized by its half-wave potential E A, which allows definition of two regions of potential relevant to the electroreduction process. There is a region where reduction will not occur, and a region at substantially more negative potential where the kinetics of reduction become so fast that no 0 can exist at the electrode surface and the reaction rate depends only on the rate of diffusion of 0 from the bulk. If the potential of the electrode is rapidly set to the latter region, the current response has two features. Initially a large current flows due to instantaneous reduction of 0 at the electrode surface. Thereafter, the current depends on the rate of diffusion. With time, the region near the electrode becomes depleted of 0 and the current decreases because 0 must continually diffuse from further away. The current-time response of this diffusion-limited current is characterized by a r 1/4 dependence. If the potential is stepped to a region near E A , the reduction of 0 at the electrode surface will not be instantaneously complete, but rather partial. Nonetheless, the [0]surface will be less than [O] bulk and a diffusion gradient will be setup. The measured current transient will still reflect the r 1/4 dependence, but its magnitude will be smaller. Finally, if the potential is stepped to a region at which reduction will not occur, the measured current will be zero. If the current-time transient is sampled at some fixed time after the potential step and these values plotted against applied potential, a typical voltammetric sigmoidal response is observed, as detailed in Figure 12. 37 3.1.6.1 The Cottrell Equation Predicts the Current-Time Transient The current-time response for a step into the diffusion-limited potential region can be analytically derived to give the Cottrell equation, which varies in form according to electrode geometry. The variation is due to different geometry of the diffusion field. For a spherical electrode with resultant spherical diffusion, such as a mercury drop, the equation is: i F (t) = nFAD0C0* 1 1 + (28) ( 7rDot) 2 1.0 where iF(t) is the Faradaic current as a function of time, n is the number of electrons transferred in the reduction reaction, F is Faraday's constant, D o the diffusion constant of 0 species, Co* the initial concentration of 0, t the time since the potential step and r o the radius of the electrode. 3.1.6.2 The Nature of Diffusion to a Partly-Blocked Electrode Varies with Time A partly-blocked electrode is one for which the microscopic (electroactive) electrode area is smaller than that of the geometric area. Examples are arrays of microelectrodes forming a larger electrode, conducting particles embedded in insulating matrix, or imperfect film-covered electrodes. At a partly-blocked electrode, the geometry of the diffusion field lines evolves after a potential step occurs. It is customary to consider a microelectrode array of equally-sized and equally-spaced active areas as shown in Figure 13; analysis of irregular geometries and sizes follows the same principles. At very short times after the potential step, each active area has its own linear diffusion layer in a region characterized by the active site's geometric area. At intermediate times, the individual diffusion fields extend beyond the geometric area and the diffusion incorporates a radial contribution to become hemispherical. At longer times still, when the spacing between active areas is smaller than the diffusion layer thickness, the formerly hemispherical diffusion layers have overlapped to give a condition indistinguishable from linear diffusion to the whole surface. Such behaviour at long times may give the appearance that the whole surface is electrochemically active when in fact it is not. 38 Whether the diffusion at any given time is hemispherical or linear has an effect on the current-time transient. If all other parameters are held constant, hemispherical diffusion will result in a greater current because a greater volume of solution is being sampled for each active area. The number of active sites per area and the size of the electroactive sites will also influence the transient. Scharifker [97] has analytically derived the time-dependent current density for the case of randomly arranged electroactive sites as follows: 1 [-g(u+N r(2 )]}e (29) — 7-Nd ro nFD0 Co ICU Here i is the current, ro the radius of the active site, u a dimensionless scaled time and Nd the number density of active sites. u is itself given by Ndro(TcDot) 1/2, and Nd related to the average distance between nearest neighbour sites (d) through 1/4( d ) 2 . It is important to note that this model assumes the number of active areas and their size are time-invariant after the potential step. Additionally, the sites are assumed to all be uniform in size. Modelling of the parameters of Equation (29) to suit the recorded current transient allows determination of the average electroactive area radius and number density. 3.1.7 Structure and Properties of Monolayers at the GIS Organic monolayers formed at the gas-solution interface (GIS) have been used to study lipid-lipid and lipid-protein interactions in a simplified system. Monolayer nomenclature depends on whether the material at the surface is in equilibrium with dissolved material or not. Gibbs monolayers are the result of dissolved material segregating to the surface and the amount of material at the surface depends on bulk concentration according to the Gibbs adsorption isotherm. On the other hand, if the molecule is not significantly soluble in the aqueous phase, monolayers can be formed by spreading from an organic solvent. The monolayer material is dissolved in a volatile but insoluble solvent such as n-pentane or chloroform and a small amount of the mixture deposited to the aqueous surface which spontaneous spreads as the solvent evaporates, forming what is called an insoluble or Langmuir monolayer. 39 5 4 3 2 Time 4&5 3 2 ^ 1 Time Figure 12 Schematic of relation between potential steps, resultant current-time transients and sampled-current voltammogram for a generalized 0 + ne R process. Left panel shows five separate steps, each to a more reductive potential. The step labelled '1' is to a potential at which reduction does not occur, while steps '4' and '5' are to potentials well beyond E. Steps '2' and '3' are intermediate. Zero on the time scale is set at the instant of the step. Centre panel shows no Faradaic current flows in response to the first step, but does for subsequent steps. Current transients have a 0 dependence. Current response for steps '4' and '5' are identical because both are mass-transport (diffusion) limited. Value of current at time t is used to plot right panel voltammogram showing typical sigmoidal response curve. It can be clearly seen that the potential of step '1' is not reductive to 0 and steps '2' and '3' are in the kinetically-limited potential region with '3' being near the E y,. The drawn curve joins the five points here but in practice would be made from many more points. Adapted from [90]. 40 Solution phase Inactive surface Electronically ^Insulating matrix conducting phase (a) (b) (c) Figure 13 Variation in diffusion geometry to a partly-blocked electrode as time after a reductive potential step increases. In this example, the spacing between active areas is —100 nm. Panel a) at very short times, diffusion is linear to individual active sites. Panel b) at intermediate times, the diffusion layer extends beyond the geometrical area of the active site, incorporating radial diffusion. Panel c) at long times, the diffusion layers of neighbouring individual sites have overlapped, giving linear diffusion over the whole surface. Reprinted with permission from [90], Copyright 2001 John Wiley & Sons. 41 Pioneering work on such monomolecular films and their applications was done by Langmuir and Blodgett using stearic (octadecanoic) acid or barium stearate [98, 99]. Monolayers can be made from many amphiphiles due to their distinct hydrophobic and hydrophilic regions. Fatty acids, long- alkyl chain alcohols and phospholipids are all commonly used. The stability of the formed layer depends on the length of the hydrophobic tail and the polarity of the headgroup. Increased chain length provides stronger Van der Waals interaction while increased headgroup polarity gives more hydrogen-bonding, both increasing the stability of the layer. Studies on the nature of monolayers are often made using the Langmuir trough, an instrument that allows control of the mean area per surface molecule by using adjustable hydrophobic barriers (e.g. of Teflon). Approaching the two barriers towards each other decreases the surface area available to the monolayer and causes an increased surface pressure. Large troughs have maximum surface area about 1000 cm', while for smaller troughs it may be 100 cm'. Measurement of surface pressure is made using a film balance, a sensitive electrobalance measuring the forces upon a float of defined geometry. Small, rectangular sheets of filter paper (Wilhemy plate) are often used, as are sheets or rings of platinum. The forces on the Wilhemy plate are those of gravity, buoyancy from the subphase, and surface tension, shown in Figure 14. The net downward force measured by the electrobalance for a plate of length L p, width Wp, thickness Tp, and plate density Dp, immersed in a subphase of density D s to a depth H is: F = g(13 p )(I, p Tp Wp )d- 27(Tp Wp)(cos — g(Ds )(Wp Tp H)^(30) Where y is the surface tension, g is the acceleration due to gravity, and 6 the contact angle between the surface and the plate. If the plate is immersed to a constant height, and the plate assumed to be completely wetted (0=0), then the surface pressure of the film can be determined as follows: 7r = 7 subphase 7 film = — AF/[2(Tp + Wp )]^ (31) = — \F/(2W) (Wp » Tp ) 42 Liquid Subphase 0 Contact Angle To electrobalance Gas MA^g gaYer Figure 14 Schematic diagram of Wilhemy plate used to measure surface tension. H is the submerged depth, H p the total plate height, W, the width of the plate, and 0 the contact angle between plate and surface. Note the scale of the organic layer has been grossly exaggerated for clarity. A plot of surface pressure as a function of decreasing surface area is called a compression isotherm because of the analogy to the three dimensional compression of a gas. Further, during compression some amphiphiles will undergo phase changes, the two dimensional analogues of gas, liquid and solid states, shown in Figure 15. The gaseous state is characterized by a mean molecular area which is considerably larger than the dimensions of the molecule, so that neighbours are far apart and van der Waals interactions are negligible. However, subphase-headgroup interaction is sufficient to prevent amphiphile loss to the (bulk) gas phase and the subphase-alkyl tail hydrophobic repulsion prevents solubilization. In this state the molecules have average translational kinetic energy of VABT for each of two degrees of freedom. The equation of state governing a gaseous monolayer is thus: IrA = k BT (32) where A is an area per molecule and 71 the surface pressure. Strictly, this holds true only for very dilute surface concentrations. At slightly higher surface concentration, typically favourable interactions between neighbours results in TEA < kBT. Unfavourable interactions would give TEA > kBT. 43 In the liquid phase, the nearest neighbour distances are smaller and Van der Waals forces between neighbours become important. Some amphiphiles display two liquid states, the liquid expanded and the liquid condensed. The expanded state has intermolecular distances larger than that of the bulk liquid, while the condensed phase is established at higher compression. With further compression, a solid state is formed, where the area available per molecule is very close to the molecular area. The solid monolayer has its molecules organized in a fixed orientation to the surface and its incompressibility can be seen from the near vertical surface pressure - area curve. Continued compression of a solid monolayer leads to formation of bi- or multi-layers, termed monolayer collapse. The collapse pressure represents the highest possible pressure before formation of a new phase. Due to kinetics, the collapse pressure depends on the compression rate as well as the nature of the amphiphile and the cleanliness of the trough. Linear extrapolation of the steeply-rising portions of the compression isotherm back to zero surface pressure gives the limiting mean molecular area for the state. Saturated straight-chain fatty acids like stearic acid (described as C 18:0, this nomenclature giving the number of carbons in the acyl chain and the number of double bonds with stereochemistry and position if applicable) have surface area per molecule of about 20.5 A2, considered a good measurement of the actual dimensions. Substitution of one cis-double bond into the chain decreases intermolecular adhesion and limits chain flexibility, leading to a larger limiting area (32 A2 for oleic acid; C 18:1, cis-9). Trans-double bonds interrupt molecular packing less than cis-double bonds. The bulk melting temperature of the amphiphile is sometimes used as a measure of the interactions involved in molecular packing. Stearic acid packs readily, has strong inter-molecular interactions, and has a high melting point (70°C). The mono-unsaturated equivalent, elaidic acid (C18:1; trans-9) melts at 52°C while the cis- form, oleic acid (C 18:1 cis-9), melts at 14°C. A related measure of inter-chain interaction in a mono- or bilayer is the gel-to-liquid crystalline transition temperature, which can be considered a melting of the alkyl chains from a well ordered state (gel) to a liquid-like disordered state, and occurs below the bulk melting point. An important characteristic of spread monolayers described in this dissertation is the equilibrium spreading pressure (ESP). The ESP is defined as the pressure at which the monolayer is in 44 111141 !L . collapsed solid (condensedk fluid (liquid expanded, intermediat and liquid copdensed) 4-^ ^ff 20^30t40 40 -0 i 30 ti■ 20 10 0.01 gaseou 50 60 Area (42/molecule) —compression of barrier- Figure 15 Composite compression isotherm giving the various states in which a monolayer may exist as it is increasingly compressed. Schematics of the molecular organization are given above the state. Not all species will show all possible states. At large mean molecular areas, the monolayer behaves as a two-dimensional gas. Compression changes the state to be liquid-like, before adopting an incompressible solid state at small mean molecular area. Further compression leads to monolayer collapse and possible bilayer formation. Surface pressure units of dyne/cm are equivalent to mN/m. Reprinted with permission from [100], Copyright 1982 John Wiley & Sons. 45 100 200 equilibrium with its bulk phase. Much work on organic monolayers is done at the ESP because it is experimentally easily accessible (no compression required) and reproducible other than slight temperature-induced variations. Formation of a monolayer at the ESP can be achieved by introducing an excess of surfactant to the surface. As the deposition solvent droplet evaporates, the amphiphile is shed from the droplet to cover the surface and a reservoir of bulk surfactant is left. The extent of material spreading, and thus the ESP, depends on the relative balance of forces at the interface. Wetting of a solid surface proceeds similarly and is described by Young's equation: yLiv COS 6, = rslv rsIL (33) with y the tension at the interfaces defined by the subscripts L (liquid), V (vapour) and (S) solid, and with 0 the contact angle shown in Figure 16. Wetting occurs when the contact angle is less than 90°. On an aqueous surface, surface deformation occurs making Young's equation invalid. Instead, the spreading coefficient [101] is used as follows: SO, = 7 AI W (701W + 70!A) (34) Which states that the spreading coefficient for an organic species at the airjwater interface is given by the relative magnitude of the airlwater tension and the sum of the organiclwater and organiclair interfacial tension. Film spreading occurs when S is positive. Monolayers and monolayer techniques have a long-standing history as biomimetic models for understanding biological questions. For instance, determination of the structure of the cell membrane as a bilayer was on the basis of the area of monolayer formed by lipids extracted from YLV YS V YSL Figure 16 Diagram of the geometry involved in the balance of forces for wetting (spreading) of an organic droplet on a solid surface. 46 red blood cells being roughly equal to double the surface area of the cells themselves [102]. A monolayer can thus be considered analogous to one leaflet of the lipid bilayer in a cell membrane. Accordingly, phospholipid monolayers provide a physical system for studying drug-lipid, metal ion- lipid, protein-lipid, etc. interactions that may be akin to those at the cell surface. For example, monolayer studies have related phospholipid structure to the magnitude of condensing effect of cholesterol [103]. Lipid monolayers provided an ideal system in which to systematically examine cholesterol's interaction with many lipids of different chemical structure. For maximum utility as a model of the cell membrane, a monolayer needs to be at a surface pressure which holds the monolayer in a state similar to that of the bilayer. An equivalent surface pressure for lipids in the cell membrane has been studied by Ohki [104, 105]. The surface pressure required for a monolayer to be equivalent to a bilayer state is determined by finding the surface pressure at which a lipid monolayer shares similar properties to a cell bilayer. The surface pressure determined to be equivalent varies with the choice of equivalence criteria (e.g. molecular packing vs. transition enthalpy, etc.) but is often quoted in the range 30-35 mN/m [106]. Some theoretical studies have suggested the equivalence pressure might be considerably higher, up to 50 mN/m [107, 108]. 3.1.7.1 Mixed Monolayers In the context of this thesis, three mixed monolayers are important. One is those formed by the addition of fluorescent dye to lipid monolayer, the second is the lipid-sterol mixed system, and the last the mixed monolayer formed from AmB penetration into a lipid or lipid-sterol monolayer. These first two are Langmuir monolayers while the latter is a Langmuir-Gibbs layer, because AmB is also present in the subphase. Generally, the two components in the mixed monolayer may mix fully, partially or can be entirely immiscible, as shown in Figure 17. Characterization of mixed monolayers is made on the basis of comparison to the pure components. For a two-component system, the average area per molecule can be expressed as: Ail =^+ j A^ (35) where A is the mean molecular area at a fixed surface pressure, x the mole fraction and the 47 subscripts i and j refer to the two species respectively. The total surface pressure may be similarly given: = %; 71-1 Xj 71-1^ (36) A graphical representation of the extent of mixing is found by plotting A u or rc u as a function of x; a linear relation defines ideal mixing, while deviations represent non-ideal mixing. Variations from ideal mixing behaviour can be quantified by the excess function, equation (37), which is the separation between the ideal behaviour line and the experimental line on the graph at a given composition; positive values signify an expansion effect, while negative values result from condensing. A excess =xjA^ (37) The excess free energy of mixing (the free energy above that determined for an ideal mixed film) can be calculated from the integral of the excess area: A G excess -7-- f: {A — (Xi Al + X 1 ' 4 j)thr^ (38) 3.1.8 The Biological Cell Monolayers have been discussed as being relevant to the study of cellular systems because a monolayer may be thought of as one leaflet of a cell membrane bilayer structure. The use of model systems like monolayers or liposomes is principally due to the structural and molecular complexity of actual cell membranes. Artificial models allow simpler evaluation of specific effects due to changes from lipid composition, electrolyte nature, and so on, using a wide range of bio-analytical and spectroscopic methods. Accordingly, it is important to consider the cell itself. 3.1.8.1 Function and Structure of Cell Membranes Cells are often considered to be the basic building block of life, as they are the simple structural unit of all living organisms. One important feature of cells is that they are enclosed and separated from their environment by the cell membrane. The cell membrane plays an important role in maintaining 48 Completely Immiscible 1111 1 11111 Partially Miscible Fully Miscible 1 iolo io loiololoio io lol Figure 17 Schematic diagram showing possible behaviour of a two component mixed monolayer system. Upper panel: components phase segregate into distinct regions. Middle Panel: mixing occurs but is not ideal. Lower Panel: Ideal mixing of the two components. 49 the transmembrane potential of the cell, AE, in part by being semi-permeable to specific ionic species. At its simplest, this membrane is composed of a self-enclosing bilayer of lipid molecules. Lipids are a class of amphiphilic hydrocarbon organic compounds. Their defining feature is that they contain both polar and non-polar regions. In aqueous solution, these regions are hydrophilic and hydrophobic respectively, and these opposing behaviours lead to self-aggregation into a variety of forms, one of which is the lipid bilayer. All the self-aggregated forms adopt a structure that arranges the hydrophilic portions of the molecule in contact with water while minimizing water interaction with the hydrophobic portions. The driving force for aggregation is the minimization of free energy, which occurs via a decrease in enthalpy and an increase in entropy on aggregation. Hydrogen bonding to the polar headgroup of the lipid provides the enthalpic term, while the entropy change involves a disordering of water and ordering of lipid acyl chains. The disorder gained by the water more than compensates for the sequestration of the hydrophobic tails, giving a net positive change in entropy. The polar region of the lipid is typically denoted as the `headgroup', while the non-polar regions are the 'tails' owing to their often elongated nature. Structurally, lipids may appear quite distinct; Figure 18 shows three common lipids used in the work of this thesis. A realistic model of the cell membrane is that put forward by Singer and Nicholson in 1972 [109] that the membrane can be considered a collection of proteins, sterols and other membrane components solvated in a bilayer of phospholipid molecules, like a two-dimensional solution. High lateral mobility of the membrane constituents is a key feature of the model. Their model has been refined by introduction of the lipid raft concept, which suggests the cell membrane matrix is not homogenous, but instead micron-sized domains of differing lipid composition exist [110]. 3.1.8.2 Liposomes as a Model for the Cell Membrane Study of basic cellular behaviour and properties has been eased considerably through the use of cell membrane model systems. Liposomes, also called vesicles, have found much use owing to their ease of preparation, versatile composition and structural similarity to the real cell membrane. A liposome is simply a self-enclosed sphere of lipid bilayer, with the interior typically being an aqueous environment. Important cell features such as transmembrane potential, lateral mobility, and 50 !c Octadecanol ^ DPPC HO DOPC LoQ/ i Figure 18 Chemical structures of three lipids: octadecanol, dipalmitoylphosphatidylcholine (DPPC), and dioleoylphosphatidylcholine (DOPC). Octadecanol is a saturated fatty alcohol, DPPC is a phospholipid with saturated acyl chains, and DOPC features a cis-double bond in each chain. Both DPPC and DOPC are zwitterionic at neutral p1-1. DPPC and DOPC are both commonly used in models of biomembranes. 51 ability to incorporate proteins can be replicated. Liposomes can be formed in both uni-lamellar and multilamellar forms; the multilamellar forms have liposomes nested each inside another. Liposomes may be formed by a number of methods, common are sonication of an aqueous lipid dispersion, rapid injection of alcoholic lipid solution into water, or high-pressure extrusion of an aqueous lipid dispersion through small pores in polymer membranes. Unilamellar liposomes in particular serve as excellent models of the cell membrane. As it is relatively easy to form liposomes with different internal and external solution composition, studies of membrane permeability in response to external stimulus can be recorded. A typical example is the measurement of potassium release from liposomes upon introduction of an ionophoric molecule to the membrane. Relatively high concentrations of liposomes in solution can be obtained and such samples are often used in calorimetry experiments, probing the transition temperature of these model cell bilayers. pH sensitive fluorophores can be incorporated into the entrapped solution inside liposomes to report on intra-liposomal pH changes upon imposition of external stimulus. Virtually any technique used to probe interaction with or behaviour of a biological cell membrane can be equally applied to liposomal systems. 3.1.9 UV-Visible Spectroscopy Absorbance spectroscopy in the ultraviolet-visible portion of the electromagnetic spectrum is a ubiquitous tool in science. This section briefly describes the underlying principles, with a focus on the changes in UV-vis spectra that may be observed on molecular aggregation. This material will aid in fully understanding the spectroscopic studies of the AmB-surfactant systems described in Chapter 6. Additionally, the absorption process discussed here is equally applicable to the fluorescence technique. 3.1.9.1 Transition Probability and Intensity Depend on Transition Dipole Moment UV-visible absorbance spectroscopy measures the extent to which a sample absorbs light of different wavelengths. This absorption process has at its root the transition of an electron from a lower to a higher energy level. At the same time, the electron may change vibrational state, and the transitions are often described as vibronic (vibrational-electronic). The electrons of the sample atom 52 or molecule interact with the electric field of the incoming radiation and may create a change in dipole moment in the sample species. Electronic transitions do not proceed without the formation of this transition dipole moment. The electronic state of the species in its initial (ground) and final (excited) states are characterized in terms of their wavefunctions, 1p-, and 'Erb respectively. The strength (intensity) of a pure electronic transition is governed by the transition dipole moment, as follows: ) = Tifi^ (39) Where 1.1 is the dipole moment operator. The corresponding probability of the electronic transition occurring is proportional to the square of the transition dipole moment: 2 P^= (I-1j2 ^ (40) Forbidden transitions are those with ri fi equal to zero (no change in dipole moment of species) and understandably have zero probability. In practice, forbidden transitions may still be observable, but have weak intensity. Any transition with non-zero Il i, is termed an allowed transition. For all but the smallest molecules, these transitions are typically related to the orbital energy levels of a particular moiety of the molecule, called a chromophore. Extended conjugated double bond systems (e.g. cyclic or linear polyenes) are a classic example of a chromophore. 3.1.9.2 Selection Rules Further Limit Possible Transitions Quantum-mechanically derived rules specifying which transitions can occur are called selection rules. In the context of UV-visible (vibronic) transitions, an important rule is that allowed transitions have AS=0, where S is the total spin quantum number. This means that allowed transitions are those between states of the same multiplicity. Forbidden transitions (such as triplet- singlet) actually may occur, but have weak intensities and are improbable. 3.1.9.3 The Franck-Condon Principle Rationalizes Vertical Transitions Theoretical calculations of transition intensity may be carried out using the Franck-Condon principle. In essence the principle states that the electronic transition occurs with no change in 53 position of the atomic nuclei and is rationalized by the fact the electronic redistribution (the transition) happens faster than the redistribution of the heavier nuclei. That is, the nuclei are thought of as being fixed in position during the transition. Mathematically, this is expressed as follows: fi (Vi f ,e (r) it (r)X f ,v (R) tif i,v(R)) (41) with li e and Ili, refering to electronic and vibrational states respectively, and r and R describing each of electronic and nuclear coordinates. The first term gives the electric transition dipole moment as before and the second term is the overlap of the vibrational wavefunctions for initial and final states. The electronic and vibrational considerations thus separated, it can be seen that for a given electronic transition, the highest intensity occurs for the transition with greatest vibrational wavefunction overlap. Accordingly, absorption (and emission) are often stated as being vertical transitions, as shown in Figure 19. 3.1.9.4 Absorption Spectra A plot of the extent of light absorbance versus wavelength is known as an absorbance spectrum. For a system that does not have spectrally overlapping species, an absorbance spectrum can be used to determine concentration through the use of Beer's law. Additionally, since the relative shape of an absorbance spectrum is related to the spacing between electronic and vibrational energy levels of the absorbing species, an absorbance spectrum may also give characterization of changes in energy levels due to environmental effects. A simple example of this is the wavelength shift in absorbance when the polarity of the solvent is changed (solvatochroism). Different molecular packing (aggregation state) may lead to similar spectral changes, and the molecular exciton theory describing these changes is therefore of interest. 3.1.9.5 Molecular Exciton Theory Describes Aggregation-Induced Spectral Shift An exciton may be thought of as a delocalized excited state or as a travelling excitation. There are different variations of the exciton model, depending on the nature of the system (having atomic or molecular subunits) and the strength of the inter-unit interaction. The theory to describe the excited states of molecular aggregates (e.g. molecular crystals or colloidal aggregates) is called molecular exciton theory, pioneered by the work of Kasha [111, 112]. The theory relates the excited states of 54 a molecular aggregate with the excited states of the aggregates' constituents. To do so, quantum- mechanical intermolecular interaction integrals are calculated using a dipole-dipole intermolecular potential term. For a strongly-interacting molecular dimer, it is found that the change in energy levels is determined by the orientation between the monomers. Figure 20 gives the energy-level diagram for molecular dimers according to exciton theory. An arrangement of the monomers with transition dipoles parallel or head-to-tail results in similar splitting of energy level, but with different allowed and forbidden transitions. In the parallel arrangement, the dipoles are out-of-phase to give a lowering of energy (E' level) while the in-phase orientation raises the energy (E" level). Whether transitions to these levels are allowed or forbidden depends on the dimer transition dipole moment, the vector sum of the two molecular transition dipoles. Thus the out-of-phase orientation leads to a forbidden transition while the transition with in-phase dipoles is allowed. Corresponding logic may be applied to the head-to-tail packing with opposite effect. Additionally shown is the result for oblique transition dipoles which gives smaller energy level splitting with each transition moderately allowed. Extension of these results can be made to larger aggregates; in the case of linear molecular polymers, parallel arrangement of transition dipoles perpendicular to the plane of aggregation ("card-pack aggregation"; left panel of Figure 20) adds new equally-spaced energy levels between E" and E'. Transitions into most of these new energy levels are forbidden; the result is that the most intense transition remains the one into the highest energy state. Similarly, head-to-tail packing (middle panel) in a linear polymer adds equally-spaced energy levels, but the lowest-lying state is most allowed. Accordingly, card-pack aggregation may be deduced from blue-shifted absorption spectra, while head-to-tail packing is indicated by red-shifted absorption relative to the monomer. Other packing arrangements such as helical structures have different energy-level splitting. The root cause of these energy-level shifts between monomer and aggregated structures is change in orbital electron density caused by the proximity of neighbouring monomer at short intermolecular distance. Orbital electron density is commonly changed through extended conjugation of double 55  A $-bA■ a) Internuclear Distance Figure 19 Diagram showing most intense E 0 E, transitions are those with maximum vibrational wavefunction overlap, described under the Franck-Condon principle as being vertical transitions. E0 is the ground state, while E 1 is an excited state. Vibrational levels are numbered within each electronic level. 56 bond systems or through it bond - it bond interaction such as 7t-stacking in the direction normal to the double bonds. 3.1.10 Fate of the Excited State Once in an electronically excited state, a molecule may return to the ground state through emission of a fluorescence or phosphorescence photon or via radiation-less transition. 3.1.10.1^Fluorescence and Phosphorescence Consideration of possible pathways back to the ground state is facilitated by examining a Jablonski diagram, Figure 21. Here, the ground state A and electronic excited state B share the same multiplicity, while excited state C is of different multiplicity. Absorption results in the promotion of an electron into one of the vibrational levels of the excited electronic state. The electron undergoes a rapid relaxation into the ground vibrational state due to collisions with neighbouring molecules. From there, the energy decay may proceed back to the ground electronic state by emission of a fluorescence photon or the electron may be transferred to a lower state of differing multiplicity. If the photon is emitted, it will be of lower energy than the absorbed photon, with the energy difference known as the Stokes shift. Therefore fluorescence emission is red-shifted relative to the excitation wavelength. The other option, transferal of the electron to a different-multiplicity state is called intersystem-crossing and is typically a singlet-triplet transfer. Further relaxation down to the ground electronic state by photon emission is termed phosphorescence. A species' fluorescence can be mapped by measuring an excitation spectrum and an emission spectrum. The excitation spectrum measures fluorescence intensity at a fixed wavelength while the excitation wavelength is varied. Conversely, an emission spectrum holds the excitation wavelength fixed and measures the fluorescence emitted at varying wavelength. Typically, multiple bands are seen in both the excitation and emission spectra, corresponding to different vibronic transitions. These are described according to their initial and final vibrational levels as (v, -v f). For example, Figure 19 shows the (0-2) excitation, from the lowest vibrational level of the ground electronic state to the second excited vibrational level of the first excited state. The (0-2) emission is also shown. 57 PARALLEL HEAD-TO-TAIL^OBLIQUE F E" it E- N G ^G ^G MONOMER DIMER^MONOMER DIMER^MONOMER "DIMER ^ BLUE-SHIFT^RED-SHIFT^BAND -SPLITTING (Hypsochromic) (Bothochromic) Figure 20 Diagram of energy level structure for molecular dimers with different orientation between transition dipoles. Aggregation with dipoles parallel to each other and perpendicular to the plane of aggregation is known as 'card-pack' and produces blue-shifted spectra. Solid arrows show allowed transitions, dashed arrows show forbidden transitions. Reprinted with permission from [112], Copyright 1963 Radiation Research Society. The efficiency of fluorescence relative to other relaxation pathways defines the fluorescence quantum yield, Q. Q is given by the ratio of the rate constant for the fluorescence process, k F, to the sum of rate constants for all pathways to the ground state, as shown in equation (42). Here, we consider all other pathways under the one moniker of 'non-radiative' (NR). k, Q - k_^ir - F ' ''NR (42) Quantum yield may equally be defined in terms of the observed lifetime of the excited state, T, and the radiative (intrinsic) lifetime t o of the excited state, which is the lifetime of the excited state in the absence of all non-radiative pathways. 7-Q=^ (43) 0 58 The intrinsic lifetime can be calculated from the fluorophore's absorption spectrum, extinction coefficient, and emission spectrum. Any chemical interaction which leads to a decrease in ti is called a quench. Typical fluorescent quenchers are dissolved oxygen, increased temperature or concentration effects [113]. 3.1.10.2 Non-Radiative Relaxation that does not give a Second Excited Species Non-radiative relaxation of the excited state may end up with a total loss of the excitation energy to heat or a transferal of the energy to a second species. The principal mechanism for loss of the energy is intermolecular collisions. Molecules in an excited state have increased reactivity due to their higher energy and if the excited state is long-lived, chemical reactions may have enough time to proceed significantly. One important chemical reaction that may occur is the formation of excimers (excited-state dimers) when an excited monomer forms a complex with a ground state monomer. Excimers may also be formed by excitation of a ground state dimer. A second possible type of chemical reaction is photodecomposition. Photodecomposition leading to a non-fluorescent species is termed photobleaching and is a common complication in the use of fluorescence for analysis. Rel a xation • -A "IL Intersystem crossing C Excitation Fluorescence Phosphorescence Figure 21 Jablonski diagram showing fluorescence and phosphorescence pathways to excited state relaxation. States A and B have the same multiplicity, but that of state C is different. Vertical arrows indicate photon absorption or emission. Wavy line indicates a non-radiative process, while the diagonal line shows crossing to a state of different multiplicity (intersystem crossing). 59 3.1.10.3 Non-Radiative Relaxation that does give a Second Excited Species Relaxation of the excited molecule to give another excited species is a question of energy transfer. The nomenclature of the transfer defines the original excited species as the donor, and the final excited species as the acceptor. Fluorescence resonance energy transfer (FRET) is a well-known example of the phenomenon. Here, dipole-dipole coupling between donor and acceptor is the mechanism of the transfer, and the efficiency has a r 6 dependence, where r is the separation distance between donor and acceptor. The distance at which energy transfer is 50% efficient is called the FOrster distance and is typically on the order of tens of A. The efficiency of FRET between a donor and acceptor is also dependent on their relative orientation. A common rule of thumb for determining if a particular pair of fluorophores will exhibit FRET is that the probability of FRET varies with the overlap between donor emission spectrum and acceptor absorption spectrum. Significant overlap correlates with improved FRET efficiency. It is important to note however, that this energy transfer truly is non-radiative; the process does not involve photon emission from the donor that is absorbed by the acceptor. Scholes has pointed out the FRET rule-of-thumb need not apply for significant FRET in some cases, notably that of molecular aggregation of donor or acceptor [114]. Briefly, in FOrster's theory the dipole-dipole interaction between donor and acceptor acts as an approximation of the electronic coupling between them. When the centre-to-centre separation distance between donor and acceptor is comparable in size to the molecules themselves, the dipole-dipole interaction does not accurately represent the electronic coupling [115]. It is worth noting that the theory of FRET can be considered as the weak-coupling case under molecular exciton theory. A second form of energy transfer that is relevant is that of non-radiative transfer to a metal. If an excited species exists close to a metal surface, the presence of the metal results in a shortening of the excited state lifetime. At large separations, the metal has no effect on the lifetime, and at intermediate distances, the fluorescence lifetime as a function of separation shows oscillatory behaviour. A good example of this is the fluorescence of Eu 3+ near a silver surface, a system studied experimentally by Drexhage [116] and theoretically by Chance [117]. The experimental model used fatty acid monolayers as spacers between the metal surface and the fluorophore. Varying numbers of fatty acid monolayers were sequentially deposited onto a thick silver film and a monolayer of 60 Europium-containing species deposited as the top layer. The fluorescence lifetime of the Eu" was measured as a function of separation distance (# of fatty acid monolayers) from the silver surface, and is shown in Figure 22. Short fluorescence lifetimes at small separations are the result of efficient fluorescence quenching by the metal surface due to coupling of the oscillating dipole of the fluorophore with collective electron oscillations of the metal (surface plasmons). At larger separations, coupling to the surface plasmons is extremely weak. Instead, the interaction between the oscillating dipole of the fluorophore with its own image reflected in the metal surface results in constructive or destructive interference depending on the distance from the metal, and sets up a kind of standing wave of varying fluorescence lifetime. This explains the periodicity in observed lifetime. In the absence of metal surface, the observed lifetime of the Eu is 632 [is. At short separation distances, there is an inverse cubic relation between lifetime and distance for thick metal films [118]. Pineda's modeling of the effect of surface roughness on fluorescence lifetimes has shown that a fluorophore near a rough silver surface can be expected to have 2-4 orders of magnitude shorter lifetime than one near a smooth silver surface [119]. 3.1.11 Fluorescence Microscopy of Monolayers The technique of fluorescence microscopy was used extensively in the work detailed in this thesis. The principal advantages of fluorescence microscopy over more general fluorescence technique are the small sampling area and spatial resolution of the fluorescence signal. Lipid monolayer systems that are not intrinsically fluorescent can still be examined with fluorescence microscopy providing a suitable fluorescent dye can be introduced without compromising the integrity of the model system. Many fluorescent dye molecules have been derivatized to provide amphiphilic character and to help promote ideal mixing. Such lipid-tagged fluorescent probes exist with either polar lipid headgroups or hydrophobic alkyl chains attached. In this way, the position of the fluorophore within the layer can be controlled. Owing to their extensive use in cellular and molecular biochemistry, some fluorophores are extremely well-characterized. 3.1.11.1^The Epi-Fluorescence Microscope Fluorescence microscopy typically involves illuminating a sample from above with visible or UV light, collecting fluorescence photons through the objective and recording the signal with digital 61 camera or other detector. The arrangement used here differs slightly in that the illumination and collection are both done through the objective, and is known as the epi-arrangement. In our case, both are done from below the sample. To obtain good sensitivity, the emitted fluorescence must be discriminated from reflected or scattered excitation light. This is achieved through the use of a set of three optical filters, called a filter cube. The light path arrangement is shown diagrammatically in Figure 23. A white light source is directed onto an excitation filter which transmits only the wavelengths that will excite the fluorophore. This excitation light is redirected upwards to the objective and sample by a dichroic mirror. Emitted fluorescence photons and scattered excitation light will be collected by the objective. The fluorescence is transmitted by the dichroic mirror while most of the stray excitation light is reflected. An emission filter is used to further reject stray light. The fluorescence is finally focussed onto the charge coupled device (CCD) of the digital camera. The transmission spectra of the three components of a typical filter cube are shown in Figure 24. Each component has a different spectrum, in keeping with its role. Wavelengths suitable for excitation of the fluorophore are passed by the excitation filter while other wavelengths are blocked.. The reflectivity of the dichroic mirror varies strongly with wavelength, so that some wavelengths are reflected and others transmitted. The emission filter serves to block reflected or scattered excitation light while allowing fluorescence to be collected. 62 A g / Eu + 3,'A I R S= 0 10)00^2000^3000^4000^5000 d (^) 1000 I,' 800 La 600 w 400 200 Figure 22 The measured (data points) and calculated (line) radiative (intrinsic) fluorescence lifetime of the Eu' cation as a function of separation distance from a thick silver surface. Sharp decrease in radiative lifetime at short distances corresponds to fluorescence quenching, while oscillations at larger separations are due to a reflective-effect. The measured excited state lifetime in the absence of silver layer is 632 [Ls. Chance et al. also studied the effect of fatty acid layers above the Eu3+ containing monolayer; S=0 in the figure indicates these data recorded when the fluorescent layer was topmost. Reprinted with permission from [117], Copyright 1975 American Institute of Physics. 63 Excitation Filter Microscope Objective Dichroic Mirror Emission Filter Coupling Lens CCD Array Detector Fluorescent Monolayer Figure 23 Light path in the epi-fluorescence arrangement. White light from Xe arc lamp impinges excitation filter which transmits excitation wavelength. The dichroic mirror reflects excitation light up through the objective and onto the sample. Fluorescence from sample is collected through objective and transmitted by dichroic mirror. The dichroic mirror also reflects most of the backscattered excitation light while the emission filter provides additional discrimination against 'leaked' excitation light. A coupling lens ensures the size of the image corresponds to the size of the charge-coupled device (CCD) detector. 64 100 90 80 70 60 50 40 30 20 10 t —Dichroic Mirror * – Excitation Filter - - Emission Filter 0 350 500^550400 ^ 450 600 ^ 650 ^ 700 Wavelength, nm Figure 24 Transmission spectra of the three components of the Olympus U-MWIBA filter cube. Light between 450 and 500 nm is transmitted by the excitation filter, chosen to match the absorbance spectrum of the fluorophore. The dichroic mirror reflects light below 505 nm and transmits above. Oriented at 45°, the dichroic mirror reflects excitation light up through the objective. Emission light impinging on the dichroic mirror is transmitted. Emission filter transmits in the range where fluorescence is expected, 505 to 580 nm. Fringing observed is a product of the interference nature of the filter construction. 65 4 ELECTROCHEMICAL INVESTIGATION OF AmB- LIPID INTERACTION 4.1 Introduction In this work, the interaction of AmB formulations with model lipid membranes was examined using a novel biomimicking system. The model system was based on a lipid monolayer adsorbed to a Hg electrode and electrochemical measurements characterized the interaction. Three AmB formulations and two related control samples were tested. We measured the effects of these samples on lipid order and the permeability of the monolayer to a redox-active species. We found the three formulations of AmB each induced a different response. Measurements on our control samples allowed us to be sure these responses were due to the AmB content of the formulations. A general correlation was observed between known in vivo toxicity of the formulations and the extent of their effect on the lipid monolayer. This work had two objectives: 1) to compare the response of the related formulations FZ and HTFZ, and 2) to probe AmB pore formation in a sterol-free system in order to provide insight into the sterol hypothesis. These objectives will now be further explained. The conventional formulation of AmB, FZ, produces significant toxicity when administered. Preliminary testing of HTFZ in animal models suggests it has lower toxicity than FZ, but there is a paucity of physico-chemical measurements in the literature about the differences between FZ and HTFZ which might be at the root of their significantly different toxicity. The most significant knowledge is that HTFZ aggregates are larger than those of FZ. We formulated a research question: will we observe a differing response between FZ and HTFZ interaction with our novel sterol-free model system? If affirmative, the follow-up question to be pondered is then how can these different interactions be related to what is already known about the structure and make-up of the two formulations. The concept of comparing FZ and HTFZ was carried throughout the thesis work. 66 Secondly, we believed our unique electrochemical setup, in which we can measure monolayer porosity, would serve as a good testing ground in which to ask questions concerning AmB mechanism of action. As described in Chapter 2, considerable effort has been expended on studying AmB' s mechanism of action. One aspect of recent studies has been to question if the sterol hypothesis is correct, or at least if there may be alternative competing pathways of AmB action. The studies noting AmB-induced pore formation in sterol-free model systems challenged the accepted dogma of AmB interaction and showed a clear need for further studies of AmB interaction with sterol-free membranes. Our review of the literature suggested that the sterol hypothesis is not the `smoking-gun' explanation of AmB mechanism, and so we hypothesized that we would observe increased porosity in our sterol-free model membrane system upon AmB addition. A number of goals needed to be met to allow us to approach these objectives. We first characterized the behaviour of a previously studied model system, the DOPC monolayer adsorbed on Hg electrode, in order to define the baseline against which the influence of AmB formulations or controls was compared. We defined a set of measurements that would be used to quantify AmB interaction. The monolayer capacitance and the variation in monolayer permeability (porosity) with applied potential were chosen. The monolayer capacitance within two regions of applied potential were identified as being important. Furthermore, the characteristics of the monolayer in these regions differed depending on whether the potential was being varied negatively or positively, so these are considered additionally distinct, giving a total of four separate characteristic capacitance measurements. Variations in these values from layer to layer and from day to day were characterized, mapping out the 'normal' behaviour and variation in the absence of AmB. When studying the effects of a drug on a model system, it is important to introduce the drug in an appropriate nature and in appropriate concentrations. Our approach was to introduce AmB as formulated, rather than as the pure substance, because the patient always receives formulated AmB. We wanted to maintain this clinically-relevant approach in selecting the AmB concentration for our experiments. Given the unique nature of the model system, determination of a clinically-relevant concentration was a challenge. The experimental setup of the model has a floating lipid monolayer at the GAS interface; a portion of this layer is then adsorbed onto the Hg electrode. In comparison 67 to a biological environment, the ratio of aqueous solution to lipid is much higher in our setup. For maximum comparability to the clinical dose, it is desirable to have identical AmB concentration at the lipid monolayer as there is at the surface of a cell membrane. However, the fraction of added AmB formulation that interacts with the floating monolayer will depend on the strength of the formulation's surface activity, while the rest will reside in the subphase electrolyte solution. Without a priori knowledge of the surface activity of each formulation, the relation between bulk concentration and surface concentration is uncertain. This question of suitable AmB concentration is complicated by the fact that variation between clinical doses of AmB in ABLC and FZ is 10-fold and the HTFZ formulation has no defined clinical dose. We defined an analogy between the clinical dose and the concentration introduced to our model system to meet this challenge. Additionally, all the formulations have AmB mixed with excipients. Excipients are generally thought of as being pharmacologically inert, but this does not preclude them from producing an effect in our system. We defined appropriate control systems to ensure that the observed response related to the AmB and not the excipients. Two important control groups were therefore identified. Lastly, a procedure was developed to ensure the highest possible consistency between the different types of measurement. Because the system is very sensitive, it was important to conduct both the capacitance and porosity measurements under equal conditions. For each formulation under study, the measurements of capacitance or porosity each constituted a full day, and so we ran the two sets of experiments on back-to-back days. The characteristics of the uncoated and DOPC-coated electrode capacitance recorded each day were used as a reference point between them, and as reference between runs with different formulations. Lastly, it is important to recognize the limitations of the experiments. Fundamentally, AmB forms pores in membrane bilayers, but AmB itself is only long enough to span half a bilayer. Either the pores form through an individual pore structure in each leaflet of the bilayer giving a through-pore or they form via membrane thinning with a one AmB-long structure. In the former case it is clear that our use of a monolayer may mean AmB interacts differently with our model than with a cell bilayer, but it may also be true for the latter mechanism. Our monolayer architecture and the lack of incorporated sterols in our model system must be kept in mind when interpreting the experimental results. 68 With these limitations recognized and our goals met, we measured the effect of AmB formulations on monolayer capacitance and porosity, as described further in the rest of this chapter. 4.2 Literature Review To place this work in its proper context within the literature, the reader needs to be familiar with two areas of work. First, the development of the DOPC lipid adsorbed on Hg electrode system as a biomimicking model is covered. Second, prior art for the use of electrochemistry and model membrane systems for the purposes of examining drug-lipid interactions are detailed. 4.2.1 Development of the DOPC Monolayer Adsorbed on Hg Electrode The model membrane system used in this work owes its beginnings to studies of the adsorption of soluble surfactants (for example n-butanol) on the Hg electrode surface. These studies linked adsorption to changes in Hg surface tension through electrocapillarity and provided the groundwork for theories relating monolayer coverage with capacitance. Interested readers are directed to the works of Grahame and Gouy (e.g. [93, 120]) as a starting point. Of more direct relevance to our system, the group of Frumkin in the (then) Soviet Union extensively studied systems of insoluble surfactants adsorbed on Hg, and also on Ga and other non-noble metal electrodes. These works differ fundamentally from the earlier ones with soluble surfactants because the insoluble adsorbate is not in equilibrium with its counterparts in the bulk electrolyte. Typical surfactants studied were long-chain alcohols and fatty acids, such as cetyl alcohol (C16:0) or oleic acid (C18:1, cis-9) [121] and capacitance measurements were used to determine that the adsorbed layer had monomolecular thickness. Miller worked extensively with biologically-relevant species adsorbed to the Hg electrode. His interest was principally in the transport of ions across cell membranes and he studied the structure and properties of organic monolayers (e.g. of decylammonium or DOPC) on Hg as a base system. To this system were added DNA or pore-forming proteins such as alamethicin and melittin [122]. The surface concentration of lipid in the monolayer and its relation to the surface concentration at the GIS interface was determined using radioactively-labelled oleyl alcohol (C18:1, cis-9) [123]. At potentials near the PZC, the surface concentration on mercury was similar to that at the GAS, 69 demonstrating near unity transferal of the monolayer. Studies of phospholipid adsorption onto mercury revealed that they could coat the metal to form a well-ordered monolayer. On the basis of the hydrophobicity of both the Hg surface and the lipid alkyl tails, it was thought the monolayer was physi-sorbed with the lipid headgroups pointed toward solution. Miller noted that at very negative potentials, the capacitance of the lipid-modified electrode matched that of an uncovered electrode, suggesting the monolayer was desorbed from the Hg surface and replaced by electrolyte. Measurement of capacitance was shown to be a suitable technique to observe potential-induced changes in monolayer structure. Studies on ion transport behaviour of Ce, Ag+ and Tr led to determination that three energetic contributions were involved: 1) electrostatics between the ion and any charge on the monolayer (e.g. zwitterionic character of lipid), 2) compressive work on the monolayer to create a hole for the ion to pass and 3) line tension work due to the formation of a boundary line between monolayer and hole [124]. Lastly, but importantly, Miller's group noted effects of applied potential on the conformation of proteins (alamethicin and bacteriorhodopsin) embedded in the monolayer [125]. This latter point represents a general interest in the effect of electric potential on the conformation, and thus action, of proteins in the cell bilayer. It is known that electric potential can affect protein action, for example voltage-gated ion channels (for sodium, potassium, calcium, etc.). For many other proteins, the influence is largely unknown [126]. The voltage-gating effect is due to a change of protein conformation to open or close the channel. It is however, not entirely clear to what extent the change in conformation is due to direct electric potential effect on the protein and to what extent the change may be mediated by potential-induced changes in the lipid matrix, thus indirectly affecting protein conformation. Further development of the phosphatidylcholine lipid on Hg system as a cell membrane model was and continues to be done by Nelson. With colleagues, he used early computer modelling [127, 128], in order to better understand the potential-induced phase transitions of the DOPC monolayer. They found potential dependence of the transitions to be a function of the surface affinity between the lipid head groups and the Hg, as well as the extent of hydrophobicity of the lipid alkyl chains. Since the phase transitions depend on the interplay between lipid-water, lipid-Hg and lipid-lipid 70 interaction, they are sensitive to any change in organization of the monolayer. Accordingly, any compound with membrane-perturbing character can be examined for the changes they produce in the phase transitions. Disruption of the well-organized nature of the lipid layer leads to the phase transitions taking place over a wider range of applied potential, and is measurable using capacitance techniques. Nelson began to use the system to probe biologically-interesting species such as the ion-channel forming protein gramicidin, environmental contaminants such as polychlorinated biphenyls (PCB), pesticides and the ubiquitous membrane component cholesterol [129-131]. Nelson then made porosity measurements of the model membrane with and without gramicidin, showing that the model is indeed biologically relevant because the incorporated gramicidin serves as a monovalent cation selective ion channel, as it does in its native environment. Cd" was not passed through the gramicidin channel. Changes in the kinetics of Tr reduction were measured when various compounds were incorporated into the monolayer (e.g. the antipsychotic chlorpromazine, the negatively-charged lipid phosphatidylserine, the vitamin retinol) [131]. Incorporated negatively- charged species gave increased T1+ permeability, while positively-charged species gave the opposite effect. 4 Lindholm-Sethson has used this same DOPC model in combination with multivariate principal components analysis to investigate interaction of the anthracycline antibiotic and anticancer drug doxorubicin with the monolayer [132]. The drug hinders replication of cancer cells by intercalating between base pairs within DNA and is known to bind to negatively-charged phospholipids. Doxorubicin was chosen as a model candidate for demonstrating the use of multivariate principal components analysis of drug interaction with the lipid monolayer. In his studies of heavy metal uptake by marine organisms, Kozarac [133] used mixtures of lecithin (phosphatidylcholines) and albumin adsorbed on Hg electrode. He examined their permeability to Cd' in NaCl solution, sea water, and with added detergents. This was compared against earlier work based on monolayers of long-chain fatty-acids adsorbed on Hg [134]. In later work, he introduced a fluorescent species, pyrene, as a model compound representing organic pollutants. This work was done using lipid monolayers adsorbed on Hg to give Cd' permeability data, and with floating monolayers in a 71 Langmuir trough to give complimentary measurements from absorption, reflection and fluorescence spectroscopies as well as measurements of surface pressure and surface potential [135, 136]. Bizzotto and Nelson worked together to experimentally examine the physical processes underlying the potential-induced phase transition peaks of the DOPC monolayer and found that the first transition is preceded by pore or defect formation while the second represents a growth and coalescence of monolayer defects. In making these measurements, the similarity between potential- induced behaviour of lipid adsorbed on Hg and insoluble surfactants (e.g. octadecanol) on gold electrodes was noted. In particular, a similar physical mechanism was invoked for the desorption/re- adsorption process which occurs at very negative potentials. Later work by Agak et al. [137], using electrochemical impedance spectroscopy, noted that the presence of a DOPC monolayer resulted in decreased ion mobility near the electrode irregardless of whether the monolayer was adsorbed or had been desorbed. This highlights the unusual behaviour of the lipid monolayer when desorbed from the electrode surface. A related effect is that water molecules near the surface were determined to have increased adsorption relaxation times, close to the time scale expected for ice. The relaxation time for water is a measure of the time it takes for a molecule of water to rotate. Recently, Almaleck et al. [138] studied the electrochemistry of DMPC and mixed DMPC/DMPG monolayers on Hg, with a view to showing the relationship between physicochemical properties of the adsorbed monolayer and other experimental model systems of lipid membranes. They examined the as yet unstudied phase state and the corresponding membrane dipole potential. To this end, they recorded the electrochemical behaviour of the monolayer as a function of potential and temperature. For the DMPC monolayer, a sharp transition in the potential of a phase transition peak was noted to occur at 24 °C. This temperature corresponds to the gel-liquid crystalline phase transition temperature of DMPC liposomes, and of DMPC monolayers at the GIS interface. This points to lipid adsorbed on Hg behaving similarly to other cell membrane models, and insinuates that data collected on these different systems maybe inter-comparable. The lipid used in our experiments, DOPC, has its phase transition at --6 °C. 72 The use of gold electrodes in the study of adsorbed insoluble surfactant monolayers has been on- going for quite some time. Initially, studies used monolayers of non-biologically relevant species (e.g. octadecanol, or pyridine-derivatives), but increasingly work is being done to develop biologically-relevant models using gold electrodes. Recently, Lipkowski et al. have succeeded in forming lipid bilayers on single-crystal gold, although it is unclear to what extent this system is defect-free or impermeable. Using in situ infrared measurements [139], the change in lipid tilt- angle as a function of potential has been elucidated for systems formed by vesicle-spreading and through Langmuir-Blodgett (vertical dipping) and Langmuir-Schaefer (horizontal touching) methods. The physical state and molecular arrangement of lipid molecules in the bilayer are found to be similar to those in adsorbed monolayers. Additionally, the changes in the bilayer which occur on addition of cholesterol have been examined [140]. Similar spectroelectrochemical studies using monolayers or bilayers of 4-pentadecylpyridine on gold showed potential-induced structural changes in the layer. This molecule was likened to a hockey-stick in structure, and changes in rotation of the blade (pyridine moiety), bending of handle (hydrocarbon chain) and tilt angle of handle and blade were observed with varying applied potential. When applied to related systems of organic films adsorbed on electrode surfaces, this spectroelectrochemical approach is expected to significantly aid in understanding the nature of layer reorganizations which occur with changes in applied potential. Cell membrane mimics have been considered a fruitful basis for the development of biosensors. If built-up on electrodes, the systems offer the opportunity for electrochemical sensor readout. The field is quite broad and sensors have been developed or conceived for a wide variety of substances. One example that is of interest here is Yilma's work towards sterol-sensitive biosensors utilizing AmB embedded in lipid layer on nanoporous silicon. AmB/sterol ion channels are formed within the layer and ion transport measured electrochemically. Different areas of the sensor can be read individually by addressing the sensor with a laser beam; the photovoltaic property of the nanoporous silicon creates a potential driving the ion transport. Interaction of any sterol-binding species with the sensor is expected to separate the AmB from the sterol, collapsing the pore structure and decreasing the ion current. It has been envisaged that incorporating sterol-linked antibodies will broaden the variety of detectable species [141, 142]. 73 Interest in ion transfer across thin organic films is of interest to biology, chemically-selective electrodes, for characterizing self-assembled monolayers (SAM) and testing corrosion inhibitors. This interest may be in understanding biological ion transport processes, measuring the barrier quality of an adsorbed layer, or characterization of defect size or surface density. Electrochemistry provides an ideal method for characterizing ion transport. A suitable redox probe is introduced into solution and its reduction or oxidation measured at a nominally blocked electrode surface. The extent of measured current then depends on defects in the blocking organic layer. For systems involving low coverage of defects (small size or low density), it is understood that non- linear diffusion of electroactive species to the defects will factor into the electrochemical response, while for systems with high defect coverage, linear diffusion will characterize the response. Models have been developed for arrays of microelectrodes in random and specific geometries by Gueshi and Amatore [143, 144], with subsequent refinement by Scharifker [97]. The gist of the analytical expressions is that higher defect coverage or smaller defect-defect distance results in more rapid shift from non-linear diffusion to linear diffusion. Accordingly, if an appropriate geometry of defect sites can be assigned, it is possible to use the shape of the transient current response to determine average defect size and density. Various probe ions are used, principally depending on the potential- region of interest, but T1+ and Cd" are often used as stand-ins for the biologically-prevalent species I(+ and Ca'. For reference, the hydrated radius of the ions are: Ca" 4.12 A, Cd" 4.26 A, K+ 3.31 A, and Tr 3.30 A [145], showing the analogy extends beyond simple valency. 4.2.2 Use of Electrochemistry to Probe Drug-Model Membrane Interactions Electrochemical studies of cell membrane - drug (or other biologically-interesting species) interaction use modified electrodes to achieve biomimickry. Recognition of electrochemical experiments as suitable basis for probing drug-membrane interaction has increased in recent years. Model systems in use include lipid monolayers on mercury, self-assembled monolayers of thiols (thiol-SAM), lipid bilayers on highly oriented pyrolytic graphite (HOPG), and derivatives such as bilayers built atop supported monolayers on gold. Despite its liquidic behaviour, use of fig as a support is usually categorized in with the solid electrode supports. A significantly different approach is to use an organic monolayer at a liquid-liquid interface. 74 The nature of the biomimetic membrane at the electrode is typically a monolayer or bilayer, although hybrid systems incorporating liposomes have been made. Bilayer systems have a better innate analogy to cell membranes and can more readily incorporate species that span the width of the bilayer. Due to the generally hydrophobic nature of most electrodes and the hydrophilic nature of lipid headgroups, it can be difficult to form good quality defect-free bilayers. Monolayer systems, on the other hand, (typically) have the advantage of ease of control over layer properties (e.g. packing density, surface pressure). Irregardless of the approach, it is important that the behaviour of the cell membrane mimic be time-invariant in the absence of added species; this is not always the case. Additionally, layer-to-layer variations must be minimized. On these two counts, the system of the DOPC adsorbed on Hg excels. Existing studies which have used electrochemistry to characterize drug-membrane interactions are relatively few, and the field is clearly still under development. Maguire [146] used self-assembled oleate (C18:1 cis-9) and linoleate layers (C18:2 cis-9 cis-12) on Pt as the basis for examining the pore-forming abilities of nystatin and alamethicin. The formed pores were probed using azide ion (N3) and the surface coverage of azide within the pores calculated. On average, 15.6 azide ions were found to absorb to the Pt surface within a nystatin pore; using the estimated radius of the azide ion, the size of pores in this model system was found to be 0.78 nm diameter. This represents excellent agreement to the size of nystatin pores in cells, which are —0.8 nm in diameter [77]. Nystatin is structurally very closely related to AmB, differing only in the division of the heptaene backbone into a diene-tetraene. Partitioning of the hypnotic and 'date-rape' drug flunitrazepam into lipid monolayers of DSPC (distearoylphosphatidylcholine, di-C 18 :0) or DSPE (distearoylphosphatidylethanolamine, di-C 18:0) at a liquid-liquid interface was studied by Monzon [147]. Addition of the drug was found to decrease DSPE monolayer permeability to tetraethylammonium cation, but a similar effect was not observed for the DSPC monolayer, showing the versatility of this model for probing sensitive interactions. The considerably less-noxious drug ibuprofen was allowed to interact with lipid bilayers formed on glassy-carbon electrode and probed with ferri-ferrocyanide ion [148]. Redox reaction of the iron at the electrode was used to determine permeability of the bilayer. At low 75 ibuprofen concentration, a stabilizing effect was noted and thought to be due to decreased electrostatic repulsion between lipid headgroups after ibuprofen interstitial insertion. Higher concentrations led to defect-formation and ultimately desorption or displacement of the bilayer from the electrode surface. Dolfi et al. have incorporated bacteriorhodopsin-containing purple membrane fragments into the DOPC monolayer adsorbed to Hg electrode [149, 150]. Bacteriorhodopsin is a protein which acts as a proton-pump when illuminated, and the kinetics of the light-driven proton transport were investigated. The lipid-coated Hg electrode serves as a suitable biomimetic environment while simultaneously allowing ready electrochemical characterization. Related work in the group has focussed on the development of tethered-bilayers on the fig electrode and their application in probing the channel-forming characteristics of a variety of ionophores including valinomycin and melittin [151-153]. Often drug formulations contain more than just the drug itself. Excipients maybe added to assist drug uptake or help drug targeting. Electrochemical techniques have also been used to study the effects of excipients. For example, Yang et al. [154] studied low molecular weight chitosans interacting with a didodecylammonium bromide (DDAB, di-C12:0) bilayer formed by liposome fusion onto a mercaptopropionic acid-modified gold electrode. Higher molecular weight chitosans have been used as a non-viral gene carrier and for enhancing adsorption of drugs to cells through a membrane disruptive effect; Yang's study showed the lower molecular weight chitosans also disrupted the lipid bilayer and can likely be used to aid drug delivery. Carlsson's interest [155] was in determining the binding of an anti-sense oligonucleotide to DSPC/DPPC (dipalmitoylphosphatidylcholine, di-C 16:0) monolayer at a liquid-liquid interface when the cationic surfactant cetylpyridinium was present or absent. Measurements of cyclic voltammetry, capacitance and derivation of surface charge density characterized the binding. Among their more specific conclusions was that electrochemical measurements are a powerful method for understanding events of oligonucleotide insertion into the lipid monolayer. 76 Some studies of drug-model membrane interaction have largely been for the purposes of demonstrating the applicability of a new membrane model. Becucci et al. [156] built on their substantial knowledge and experience with lipid monolayers on Hg to develop a system better suited to protein incorporation (reconstitution). This model was a bilayer with distinctly asymmetric layers. A lipid-like monolayer presenting acyl chains into solution was first formed on Hg electrode. A hexapeptide chain was tethered to the Hg via a thiol linkage at the N-terminus, while DMPC lipid was covalently bound through its headgroup to the C-terminus. This modified electrode was then passed through a —5:1 DOPC:Cholesterol monolayer floating at the GIS interface, coating the lipid/sterol layer atop to form a bilayer. Use of pure DOPC to form the outer monolayer resulted in poorly organized bilayers. Overall, this unusual setup is said to be more stable when compared to other bilayer systems built up with hydrophilic spacers on Hg. Alamethicin, melittin and gramicidin were separately allowed to interact with the bilayer after addition to the subphase, and the electrochemical response probed with impedance spectroscopy. Biomimicking electrochemical models have been used in environmental detection applications. Khodari et al. [157] made phospholipid and fatty acid-modified carbon electrodes and determined detection limits for the antihistamine promethazine. Nanomolar concentrations were detectable, allowing detection of the antihistamine in serum and urine with no pretreatment step. The same group also examined the effect of pH and lipid charge on behaviour of the anti-cancer anthracyclines adriamycin and epirubicin at the bilayer-coated carbon electrode [158]. Increasing negative charge on the lipid was found to result in more drug penetration into the layer, seen by increased layer capacitance. They have additionally summarized the field of lipid- modified electrodes for environmental pollution control applications [159]. 4.3 Summary of Electrode-Supported Biomimetic Systems The literature on the development and use of electrochemical-based biomimetic systems shows them to be suitable for studying the interactions between a wide range of biologically-relevant species and the underlying lipid layer. Electrochemical systems provide a unique built-in ability to measure porosity or permeability, since measurement of redox reactions at the electrode surface are easily undertaken. Numerous electrode-supported biomimicking models have been developed, meaning 77 the choice of system can be tailored to the experimental requirements. If probing pore-forming species, a model with low intrinsic defect density is needed, in order to discriminate between formed pores and pre-existing defects. The DOPC monolayer adsorbed onto a Hg electrode surface gives excellent defect-free monolayers. 4.4 Materials and Methods 4.4.1 Introduction The electrochemical experiments described in this chapter required stringent experimental conditions. The materials and methods used were selected in part to ensure no organic contamination of the system. In this respect, cleanliness of glassware and purity of reagents were very important. Additionally, the necessary stability of the mercury drop electrode required careful filling and maintenance of the electrode. The general electrochemical setup is first described, including the instrumentation and data analysis components. Secondly, the measurement methods and the composition of the different AmB formulations used are detailed. 4.4.2 Electrochemical Setup The working electrode for all electrochemical experiments was a syringe-type hanging mercury drop electrode. Metrohm model EA 190 (Herisau, Switzerland) dispensed Hg drops with calculated surface area of 0.0088 cm 2 , determined by weighing a series of dispensed drops and assuming spherical geometry. A second electrode used, shown in Figure 25, model WK2 from the Polish Academy of Sciences (Warsaw, Poland), produced Hg drops with surface area 0.0096 cm 2 . The electrodes feature a micrometer screw-head which forces a stainless steel piston into a fixed reservoir of mercury. Prior to being filled with triply-distilled instrument grade mercury (Bethlehem Apparatus Co., PA), the glass capillary of the electrode was made hydrophobic through application of a silane coating. The capillaries were immersed in 4% (v:v) dichlorodimethylsilane (Fluka, Buchs, Switzerland; catalogue #40140, >99%) in trichloroethane (Sigma, St. Louis, MO; catalogue #402877, >99%) for 78 15 minutes, then removed and baked at 200° C for 4 hours. The heating helps to ensure formation of a covalent bond between the silane and the glass surface and evaporates any excess solution. The exterior tip of the glass capillary was made hydrophilic by removing the silane solution through immersion in 1 M aqueous sodium hydroxide solution after the capillary had been filled with mercury. Spare capillaries that were silanized but not yet required for use were stored in a desiccator. The glass electrochemical cell was cleaned in warm 1:1 (v:v) concentrated sulphuric and nitric acid for at least 2 hours. Glassware was thoroughly rinsed with Milli-Q water (Millipore, Billerica, MA). This Milli-Q system uses reverse osmosis and multiple ion exchange cartridges to de-ionize the water, with final UV light exposure to photolyze organic contaminants. The end product is >18.2 MO•cm resistivity water with total oxidizable carbon content of < 1 ppb and pll of 6. A set schedule for replacement of water system ion-exchange cartridges was followed to ensure high quality of the output water. The electrolyte solution was approximately 70 ml of 0.1 M KCI, made up after calcining the KCI (Fluka; catalogue# 60133 puriss p.a. ACS grade, low Br) at 500° C overnight. A platinum coil, cleaned by heating in a butane-air flame followed by water quenching, served as counter electrode. Through this thesis, all potentials quoted are with respect to the saturated calomel reference electrode (SCE). The reference electrode was connected to the cell via a saltbridge as shown in the diagram of the electrochemical cell, Figure 26. The cell side of the saltbridge was first filled with the 0.1 M KCI electrolyte, then the reference electrode side was filled with saturated KCI solution. A stopcock separating the two sides allowed ionic transport but eliminated solution mixing. The electrolyte was de-oxygenated by bubbling argon (Praxair) for 20 minutes, and a continuous flow of argon was maintained over the electrolyte during experimentation to prevent re-oxygenation. The argon was passed through an in-line activated charcoal filter (Supelco, St. Louis, MO) to absorb hydrocarbons. 79 Figure 25 The WK2 model hanging mercury drop electrode. The micrometer head at top drives a stainless steel piston into the reservoir of Hg held in the capillary. The capillary tip of our electrode was modified to have a conical point. Image courtesy of the Polish Academy of Sciences. 80 Computer and DAQ Figure 26 Schematic diagram of electrochemical setup. In the cell at left, the hanging mercury drop electrode serves as working electrode (WE), a platinum coil is the counter electrode (CE) and argon gas can be supplied above or through the electrolyte. A saltbridge with stopcock connects the saturated calomel reference electrode (RE) to the rest of the cell. Potential control and current measurement are provided via the potentiostat and computer data acquisition system (DAQ). The lock-in amplifier is used in conjunction with the potentiostat when measuring capacitance. 81 21+ (44)'C- ^ V„ 4.4.3 Measurement Methodology Each electrochemical experiment began with a check of the cleanliness of the system and verification that the surface area of the mercury drop was constant by measuring the interfacial capacitance across a wide potential range. The absence of organic contamination is indicated by the value of the capacitance, and the stability of the drop size indicated by the constancy of the capacitance on back-to-back measurements. Day to day variability was assessed by recording the capacitance scan of the uncoated and DOPC-coated electrode prior to adding any sample solutions. Results presented are either average values with N typically > 3, or are chosen as representative of a number of measurements. Experiments were carried out at room temperature, typically 21 °C. A potentiostat (Fritz Haber Institute ELAB) generated a continuous DC voltage cycle (i.e. triangular wave) between -0.3 and -1.2 V at a 5 mV/s rate. A lock-in amplifier (EG&G/PAR, Sunnyvale,CA; model 5210) provided a 2 mV rms AC voltage at 70 Hz which was superimposed on the DC voltage. The current resulting from the applied AC voltage was demodulated into in-phase (real) and 90°-out-of-phase (imaginary) current by the lock-in amplifier. A National Instruments data acquisition board (Austin, TX, model PCI-6070E, 1.25 MSamples/s, 12-bit) was used to digitize the current and interface with a custom software programme for calculating differential capacitance in a Labview (National Instruments) environment. Capacitance as a function of applied DC potential was calculated via: which assumes that the total impedance of the system can be modelled as that of a series RC circuit, and i t are the imaginary and real components of the current, Vac is the amplitude of the AC signal, and w its frequency. In this model, the resistance is that due to solution, while the capacitance originates at the mercurylelectrolyte interface. The zero point of the phase angle was set daily by taking as zero the angle at which the applied sinusoidal voltage across the cell was a maximum. 82 After recording the capacitance of the uncoated mercurylelectrolyte interface, a lipid monolayer was formed at the GIS interface. DOPC ( >99.0%, from Avanti Polar Lipids, Alabaster, AL or Northern Lipids, Vancouver, BC) was dissolved in n-pentane (Fisher Scientific, HPLC grade) to form a 2mg/m1DOPC solution. DOPC sourced from Sigma was found to be unsuitable for the formation of well-ordered defect-free monolayers. The stock DOPC solution in n-pentane was stored at -18 °C, protected from light. Eight of this solution was injected from a Hamilton (Reno, NV) microsyringe onto the GIS interface in a dropwise fashion. Each drop added to the surface was allowed to spread before adding the next drop. Five minutes was allowed for evaporation of the pentane, leaving the DOPC monolayer in equilibrium with bulk phase DOPC present as droplets. The eight added volume represents approximately four-times the amount of lipid needed to form a complete monolayer over the surface area of the argonIelectrolyte interface in the cell, thereby accounting for losses to the glass cell walls and providing an excess to ensure the monolayer reached the equilibrium spreading pressure (ESP) [160]. The ESP corresponds to the greatest achievable surface concentration while maintaining a thermodynamically stable phase; for DOPC the value is 50 mN/m [161]. With the electrode potential set to -0.4 V vs. SCE, the mercury drop at the capillary tip was dislodged while in the electrolyte, the electrode raised above the monolayer, and a new mercury drop extruded. The electrode was then lowered through the monolayer at the argonIelectrolyte interface, coating the mercury drop with the DOPC monolayer as shown in Figure 27. Due to the fact that the monolayer at the GIS is oriented with lipid headgroups in aqueous solution and the alkyl tails extend upwards away from solution, and to the fact that the alkyl tails and the Hg are both hydrophobic surfaces, the monolayer coated onto the Hg electrode is oriented with alkyl tails on Hg. A study by Agak et al. confirmed this orientation by measuring the influence of subphase Ca2+; calcium is known to associate with lipid headgroups [137]. Formation of the biomimetic lipid layer by touching the electrode to a floating monolayer at the ESP has the advantage of being very simple. No special technological approach is used to ensure a reproducible deposition. Rather, the structure of the adsorbed film relies on the interplay between lipid-water, lipid-Hg and lipid-lipid forces. Since the underlying Hg electrode is liquid and atomically-smooth (unlike solid electrodes), the adsorbed monolayer can be formed to be defect-free and very impermeable to ions. Being a 83 DOPC in n -pentane Electrolyte Figure 27 Process of formation of DOPC monolayer on mercury electrode. Left: DOPC dissolved in pentane is injected onto argonlelectrolyte interface and pentane allowed to evaporate, leaving DOPC monolayer. Centre: mercury electrode is slowly lowered through interface, coating lipid monolayer onto mercury surface. Right: monolayer forms with lipid alkyl tails towards mercury due to hydrophobic interaction. Note scale of lipids is grossly exaggerated with respect to electrode. conductive substrate, the Hg allows simultaneous control of potential across the biomimetic DOPC layer and measurement of current at, or capacitance of, the interface. Relative to other biomimetic model systems, this property makes the DOPC on Hg system rather unique. Since deposition of a defect-free monolayer depends on the development of good lipid seal at the lipid-capillary-Hg edge, the process of extruding a Hg drop and depositing a monolayer was repeated several times prior to making the first lipid-coated experimental measurements. This is simply a question of ensuring the condition of the electrode has reached a 'steady-state'. Monolayer capacitance is sensitive to both the nature of the adsorbed film and its structure (Cf equation (1)). The capacitance of the monolayer-coated mercury was measured in the potential range -0.3 to -1.2 V for several successive newly formed mercury drops. This formed the basis of all electrochemical experimental systems. At the end of the day, the capillary tip was cleaned of lipid by a combination of repeated cycles between immersion in chromic acid and washing with water, and extrusion of multiple mercury drops from the electrode. The capillary tip was dried using a methanol rinse and was kept filled with mercury to prevent fouling of the capillary bore. 84 4.4.4 Electrochemistry in the Presence of Amphotericin B The capacitance of the DOPC monolayer in the presence of different formulations of AmB or control preparations was measured. The AmB-containing systems studied were the commercially available formulations FZ and ABLC, and the non-commercial variant, HTFZ. FZ is supplied as a powder for reconstitution made up of 50 mg AmB, 41 mg sodium deoxycholate, and 20.2 mg sodium phosphate. FZ was reconstituted with 10 ml of water to form a stock solution; FZ working solution was 1 mg AmB per ml. HTFZ was prepared by heating the FZ working solution at 70 °C for 20 minutes in a water bath. ABLC is a lipid-complexed formulation of AmB, supplied as an aqueous solution and was used as received. Each ml contains 5 mg AmB, 3.4 mg DMPC (di-C 14:0), 1.5 mg DMPG (di-C14:0), and 9 mg sodium chloride. The two control systems were aqueous sodium deoxycholate and an aqueous dispersion of DMPC/DMPG. Sodium deoxycholate (Sigma ultra grade, >99%, D5670) solution was made to 0.85 mg/ml. The sodium deoxycholate and DMPC/DMPG (ABLC Blank) controls were made such that they represented FZ and ABLC respectively without AmB, as in Table 2. The DMPC/DMPG control was made with lipids from Avanti Polar Lipids (Alabaster, Al) and KC1(Fluka, puriss p.a. ACS grade, low Br) was used instead of NaCl. For completeness, the effect of heat-treated sodium deoxycholate (i.e. AmB-free analogue to HTFZ) was examined, but found to have identical effect to the unheated deoxycholate control (results not shown). All solutions were made with high-purity 18.2 MQ•cm Millipore water. FZ and ABLC were obtained from the pharmacy of the UBC hospital. For all systems, the relevant AmB or control solution was injected into the electrolyte beneath the monolayer. The electrolyte was gently stirred with a teflon-coated magnetic stirbar to mix the solution without disturbing the monolayer at the electrolyte surface. Different concentrations were examined: approximately 7, 15, 30, 45, 75 and 100% of a 'therapeutic dose'. In clinical administration, the typical recommended daily therapeutic dose for FZ is 0.5 mg AmB/kg body weight, and for ABLC it is 5 mg/kg. In an attempt to mimic physiological conditions, we assumed that the human body could reasonably be modelled as being all water, so that a 1 mg/kg dose in the clinic was taken to be equivalent to 1 mg/1 electrolyte in our system. The 100% therapeutic dose (TD) values for FZ and ABLC correspond to AmB concentrations of 0.54 p.M and 5.4 [1M in the electrolyte respectively. The TD of HTFZ was assumed to be identical to that of FZ, since there is 85 Table 2 Composition and Therapeutic Dose of AmB Formulations and Control Solutions Used. Formulation Name Composition of injected solution (per ml) Clinical Adult Therapeutic Dose Fungizone (FZ) 1 mg Amphotericin B 0.82 mg sodium deoxycholate 0.40 mg sodium phosphate 0.5 mg AmB per kg body weight Heat-Treated Fungizone (HTFZ) 1 mg Amphotericin B 0.82 mg sodium deoxycholate 0.40 mg sodium phosphate None determined; introduced at same dose as FZ Abelcet (ABLC) 5 mg Amphotericin B 3.4 mg DMPC 1.5 mg DMPG 9 mg sodium chloride 5 mg AmB per kg body weight Abelcet Blank (ABLC Blank) 3.4 mg DMPC 1.5 mg DMPG 9 mg potassium chloride N/A; introduced at same lipid concentration as ABLC Sodium Deoxycholate 0.85 mg sodium deoxycholate N/A; introduced at same deoxycholate concentration as FZ 86 no clinically established dosing information. The 0% TD concentration was represented by the DOPC/mercury system in the absence of added AmB or control. Sodium deoxycholate was introduced in concentrations equivalent to the deoxycholate concentration in each addition of FZ. Similarly, ABLC Blank, the lipid control for ABLC, was injected to give equal lipid concentration to the actual ABLC sample. 4.4.4.1 Thallium Electroreduction and Chronoamperometry A second set of electrochemical experiments concerned the measurement of monolayer porosity. Aliquots of stock thallium nitrate (Fluka, catalogue #88272, purum p.a. 99.0%) solution were added to the electrolyte to form a 10' M Tr solution (Ereversib I e —0.45 V vs. SCE). A custom Labview software programme was used to alternate the electrode potential between an oxidative potential (-0.2 V vs. SCE) and more reducing potentials. A step to a reducing potential allows Tr (aq) to diffuse to the electrode and become amalgamated as Tl(Hg) if the monolayer is sufficiently porous. The Tr (aq) /T1(Hg) couple is kinetically fast on the bare mercury electrode surface [162], meaning the current transient response is dominated by the mass transport process and the characteristics of the transient can be used to measure monolayer permeability. Prior to each reductive step, a 'safe' potential of -0.35 V was held for 2 s to allow the monolayer to equilibrate. The potential is returned to the oxidative 'base' potential for 10 s between steps to convert Tl(Hg) back to TP(aq) and allow it to diffuse away from the electrode. The step potentials started at -0.275 V and progressed negatively in 0.025 V steps, shown in Figure 28. The negative potential excursions were held for 0.25 s, during which the resultant current vs. time transient recorded. A typical current transient curve is shown in Figure 29. The current was sampled every 29 [IS until 7300 data points were collected. Additionally, the current before the potential step was measured (200 data points) to determine an initial baseline current. The typical baseline current was small, but an offset to make the baseline zero was applied to the data. The current response was passed through a 10 KHz low pass filter. These `Tr step experiments' were carried out with the uncoated mercury electrode, with the DOPC- coated electrode, and with the monolayer-coated electrode in the presence of AmB or control. 87 Measurement at the uncoated mercury surface determines the half-wave (reversible) potential for the T1+ (aq) + e - Tl(Hg) reaction, and the mass transport limited current recorded was used to calculate the [Tr] in the cell. This concentration was determined by back-calculation from the Cottrell equation using a least-squares regression to fit linearized current transients for potentials well beyond the reversible potential, where mass transport limitation of the current dominates. Typically, the Tr concentrations back-calculated from the steps to -1.2, -1.1, -1.0 and -0.9 V were averaged together to determine the Tr concentration in the cell. Using this concentration and the known surface area of the Hg drop, the limiting current, 'lim i t ing' at t=50 ms was determined. Measurement of the current 50 ms after the potential step allows enough time for any capacitive charging current to decay to a near zero value, meaning the measured current is entirely dominated by Faradiac charge transfer. Analysis of the current transients was carried out in two ways. In one, normalized reduction currents for the different systems were plotted together as a function of potential, providing direct comparison of the barrier property of the monolayer. In the other, the characteristic curvature of the transient was used with a fitting regime to determine average pore size and number density. To prepare the data for plotting, the magnitude of the current at t=50 ms was extracted for each potential step. 20 data points on either side of t=50 ms were averaged together, giving i @ t=50.00±0.58 ms. Normalized reduction current was taken as i / ilim ,t , ng• Values of the normalized reduction current were then plotted against electrode potential. Determination of average pore size and number density within the monolayer relies on the nature of diffusion to the pores. As described in Chapter 3, diffusion to a collection of electroactive areas (the pores) initially proceeds via hemispherical diffusion, until such time as the diffusional fields around the individual pores begin to overlap to give linear diffusion. An analytical solution for the time-dependent current transient expected for a system of random pores was given in equation (29). A fitting routine was used to vary the parameters of pore size and number-density in order to generate a simulated transient which best fit the measured transient. Two different algorithms were used in the fitting. Initial determination of the parameters was conducted with an Adapted Simulated Annealing (ASA) algorithm developed by Ingber [163]. This algorithm is known to be 88 ui -0.40 U (/) -0.60 - To E' O O 0- -0.80 - -0.20 -1.00 - To^To -0.2 V^-0.2 V -0.35 V -0.5 V -1.20 0^50 100 150 200 250 300 350 400 Time / s Figure 28 Sequence of potential applied to mercury electrode for Tr step experiments. Initial and final potentials are -0.4 V. Prior to each step, the potential is held at -0.35 V for 2 s, allowing the monolayer to equilibrate. Increasingly negative step 'reductive' potentials are held for 0.25 s. This corresponds to increasing the driving force for Tr reduction at the electrode. Note that the first few `reductive' potentials are actually oxidative. After each reductive step, the potential is returned to -0.2 V to allow any amalgamated Ti to oxidize to Tr (aq) and diffuse away from the electrode. The inset shows detail of the step to -0.5 V. The portion of the step for which the current transient is measured is shown in dashed line. Inset not to scale. 89 0.25 - 0.00 - -0.25 - -0.50 - -0.75 - cf) -1.00 - -1.25 - -1.50 - -1.75 ^ 0.000^0.025^0.050^0.075^0.100 Time / s Figure 29 Sample T1+ reduction transient. The large magnitude current immediately after time zero is the result of capacitive charging, while the non-zero current at longer times is the result of T1+ reduction. This particular transient was recorded on the step to -1.10 V for an uncovered Hg surface. Currents recorded before time zero correspond to 'pre-points' used to correct for possible non-zero current offset. A small negative current offset is observed here. The (uncorrected) zero current line is sketched in for reference. The current at 50 ms was normalized and plotted in one method of analysis, while the entire transient was used for parameter-fitting purposes. Current was recorded every 29 p.S; the transient shown is truncated after 0.1 s while the recorded data actually extends beyond 0.2 s. Capacitive pickup of the 60 Hz line frequency is visible as spikes. 90 particularly well suited for global optimization problems of complex systems. In the context of this thesis, the problem at hand is to find the global minimum in the variation between the recorded current transient and the fit transient using the parameters r (pore radius) and N (number density of pores). It is important to avoid determination of r and N based on a local minimum; we seek the overall minimum. ASA is an offshoot of the simulated annealing (SA [164]) algorithm. As its name suggests, it has an analogy to the physical process of metallurgical annealing. The physical process of annealing involves heating followed by slow, controlled cooling of a metal to increase crystal size and reduce the density of defects. The heating gives the atoms higher energy and increases their rate of diffusion. The atoms may then diffuse between different energy states. Slow cooling limits the number of states accessible and increases the chance of the atom ending in a low energy state, while rapid cooling sets the atoms 'in-place' in higher-energy local minima. Finding the global minimum in the fitting routine is analogous to the atom ending in the lowest energy state. In SA, an initial solution (seed) of parameters is compared to a random 'nearby' solution. If the new solution is an improvement (a 'downhill' move), it replaces the seed and the process is iterated. If the new solution is not an improvement (an 'uphill' move), then the solution may be either accepted or rejected before iterating. The probability of the unfavorable uphill move being accepted depends on the difference between the function (actual current transient) and the fit as compared to a global parameter T. T is called the temperature and is slowly decreased throughout the fitting process. Near the beginning of the process, when T is high, uphill moves are relatively common. This effectively allows a coarse-sampling random walk over the entire parameter space early on, before settling down to increasingly lower minima as T is decreased. If the parameter space is considered to be a random 2-D topology of mountains and valleys, the effect of allowing the uphill moves is to allow movement up over the mountains into neighbouring valleys; it avoids getting stuck in one valley (local minimum). It is this characteristic behaviour which makes SA a suitable algorithm for finding globally-optimized solutions. 91 SA allows the rate of temperature cooling to be specified during the optimization. A low rate of cooling will give a higher chance of the random walk escaping from high valleys during the early stages of the process, and can be expected to lead to the global minimum. A high rate of cooling increases the likelihood of becoming stuck in a high valley. Naturally, a slow rate of cooling requires more iterations and increases the computational time of the algorithm. Adapted SA (ASA) is a refinement which automatically adjusts the rate of cooling according to algorithm process, thus simplifying the fitting process for the user. One drawback to the SA and ASA algorithms is that once near a global minimum solution, refinement is inefficient. For this reason, the second step in our fitting process was the application of the amoeba (Nelder-Mead Simplex) routine. This algorithm is much better at refining a solution, but used alone, it is prone to reporting a local rather than global minimum. It is for this reason that it was used after the ASA routine, to explore the parameter space close around the ASA-reported global minimum. The amoeba routine uses a 'simplex', a geometrical figure of N+1 points, where N is the is number of parameters of the problem. For three dimensions, the simplex is a tetrahedron, while in two dimensions it is a triangle. The following simplified description supposes a 2-D simplex with 3 points. The parameters of each point are then evaluated to determine their 'goodness of fit' to the function, in our case, the measured transient. The points are ordered according to goodness of fit (deviation from measured transient), giving f(1) < f(2) < f(3). The worst point, f(3), will be substituted with a replacement point called a trial point (tp). A tp located by reflection across the line between point 1 and point 2 is evaluated to give f(tp). If f(1) < f(tp) < f(2), that is if f(tp) is neither the new best or worst point, then f(tp) replaces f(3). If f(1) and f(2) < f(tp), then f(tp) is the new best point and the trial point is discarded in favour of one further in that direction. The simplex is thus expanded in a favourable direction. On the other hand, if f(tp) < f(2) and f(1), then it is the new worst point. This would signify movement up a valley, and the trial point is discarded in favour of one closer to the centroid, a contraction of the simplex. These operations are shown in Figure 30. Once the combination of the ASA and amoeba routines was complete, the parameters r and N were noted and the whole process restarted with a new random seed location in parameter space. 92 Low Point Successive convergence to the same minimum indicated greater likelihood of the minimum being global and increased confidence in the fitting process. A sample fit to a current transient is given as Figure 31. A) High Point B) C) Figure 30 A simplex in 2-D parameter space and a representation of the operations available in the amoeba routine. Panel A) a simplex made of three points, with high and low as marked. One side of the simplex is thicker to aid in visualization of operations. Panels B-D show possible operations on the simplex of Panel A). Panel B) reflection operation: new test point chosen by reflection of the highest point across line of the other two points. Panel C) expansion operation: the simplex is lengthened in the direction of a favourable test point. Panel D) contraction operation: An unfavourable test point can be discarded in favour of contracting the simplex towards its centroid. When the simplex is over a minimum, the shrinking effect of the contraction operation gives refinement. 93 0 1 0.1 Time / s 0.2 0.1 Time / s 0.0 0.2 0.10 0.05 - -c-rim  0.00 - 0 cc -0.05 - -0.10 Figure 31 Sample result of fitting to recorded current transient. Top panel: current transient shown in black, fit result in grey. Bottom panel: residual between transient and fit. This particular transient was recorded on potential step to -0.875 V, with HTFZ present at 100% TD. 94 4.5 Results and Discussion 4.5.1 Introduction Electrochemical measurements of AmB-lipid membrane interaction were carried out using two complimentary methods, one based on monolayer capacitance and the other on measurement of membrane porosity. Taken together, they provide a consistent measure of the disruptive effects AmB formulations have on the model membrane. First, the capacitance results will be presented, followed by the results of the membrane porosity study. 4.5.2 Capacitance Measurements Characterization of the unadulterated DOPC monolayer adsorbed on the Hg electrode was carried out first, in order to provide a benchmark for the changes wrought by later addition of AmB. The principal characterization method was to measure interfacial capacitance as a function of potential. As detailed in equation (1), capacitance depends on the thickness of the interfacial dielectric layer and its dielectric constant. The capacitance of a monolayer of DOPC depends primarily on the dielectric constant, since the thickness may be taken as constant. Water present in the monolayer due to defects in uniform molecular packing will result in higher recorded capacitance because the dielectric constant of water is —80 20.c while that of organic much lower (Cf. 1.9 200c for n-hexane). Using a measured monolayer thickness of 1.3 nm, the dielectric constant of DOPC has been determined to be —2.2 [122]. It should be noted that the dielectric constants here for water and n- hexane refer to bulk material, and not surface layers specifically. It is unclear how dielectric constant changes for surface layers. Capacitance measurements are shown in Figure 32 (page 102) for both the DOPC-coated and the uncoated Hg electrode surface. The capacitance of the uncoated Hg electrode is near its maximum of 40 4F/cm2 at the PZC (-0.45 V), and decreases to —18 p,F/cm 2 as the potential is made increasingly negative. This decrease in capacitance is related to the reorientation of water dipoles at the electrode surface under the effect of electric field. Measurement of the electrode capacitance from -0.3 to -1.2 V and back to -0.3 V constituted one capacitance scan. A capacitance scan of an uncoated Hg electrode which had the positive-going scan perfectly trace the negative-going scan 95 signalled a stable and clean Hg drop. The capacitance measurement of the uncoated Hg is included in all capacitance graphs to serve as a reference point. The DOPC layer was deposited onto the Hg electrode at -0.4 V and a capacitance scan measured. The negative-going scan is shown in solid line and is characteristic of the 'as-deposited' layer. In other words, the DOPC layer deposited onto the Hg drop is representative of the layer at the GIS interface. The floating DOPC monolayer is at its ESP, and is well organized. As such, the as- deposited DOPC layer will serve as the baseline measurement against which to compare the interaction of AmB formulation or control with a model cell membrane which is free of defects or pores. The capacitance scans show an initial low and potential-invariant capacitance of-1.8 p.F/cm 2 in the range of -0.3 to about -0.8 V, known as the minimum capacitance region. The decrease in capacitance as compared to the uncoated electrode is due to the much lower dielectric constant of the organic film. The value of 1.8 g/cm 2 represents formation of a well-ordered, defect-free, cohesive DOPC film. The instrumental settings used to record the capacitance in this potential window were adjusted to result in a 33 x gain factor, to better be able to observe the low and invariant capacitance. Any increase in the capacitance in this potential window indicates a decrease in monolayer molecular order and the introduction of defects. The remainder of the capacitance scan, in the range -0.8 to -1.2 V, was recorded with unity gain. As the applied potential is scanned negative of -0.8 V, defects are formed in the film, as indicated by the slight increase in capacitance leading up to the peak at -0.965 V. There, a tall, sharp peak is observed in the capacitance plot, with a second such peak at -1.025 V. These peaks represent potential-induced phase transitions of the monolayer and are sensitive to monolayer molecular order. The peaks are known as pseudo-capacitance peaks. The 'pseudo-capacitance' portion of the name derives from the fact that although this phenomena is observed in measurements of capacitance vs. potential, the underlying cause is not capacitive. That is, the mechanism is not a variation in charge density with potential (do/dE), but rather from a variation in surface coverage with potential (de/dE) as described in Chapter 3. The variation in surface coverage can be thought of as a molecular 96 reorganization of the surface layer. The precise nature of these changes are not known. Computer modelling of the phase transitions by Leermakers and Nelson [127] has suggested that the first peak represents conversion of the monolayer into two phases: a thick and thin form of monolayer, while the second peak is said to represent a change from this inhomogeneous monolayer into a porous bilayer. Work by Bizzotto and Nelson [165] showed that there is an increase in monolayer porosity just prior to peak 1 and that the process underlying peak 2 is one of nucleation of defects followed by their growth and coalescence. Whatever the true nature of the transitions, it is clear from the increased capacitance after peak 1 that the layer is no longer defect-free. The potential range through which the phase change occurs will depend upon the energetics of the growth of the new phase from the potential-induced defects. If all defects are energetically homogeneous, which would be expected for a well-ordered phase, the phase change will occur at a single value of the potential. If the monolayer is disordered, the potential-induced defects will be energetically heterogeneous and the phase change will occur over a range of potentials, causing the pseudo-capacitance peaks to be broad and less intense. The height of the peaks are therefore a sensitive indicator of the lipid monolayer organization. When a potential of -1.2 V was reached, the potential scan direction was reversed to be positive- going. Shown in dashed line in Figure 32, the positive-going scan is representative of the `reformed' layer. During the negative-going scan, the applied potential acts as a stress on the system, resulting in the layer becoming porated and strongly defective. Any additive in solution can interact with these holes or defects in a different manner to the interaction with the as-deposited layer. The nomenclature of the reformed layer stems from the fact that the potential-induced defects are closed up or annealed during the positive-going scan. We are distinguishing, then, between the layer in the state prior to being made defective under potential control and afterward; these are the as-deposited and reformed layers. The differences between the as-deposited and reformed layers will be seen to become more important upon addition of AmB formulations. The capacitance of the reformed DOPC monolayer traces that of the as-deposited layer without hysteresis from -1.2 V until peak 2 is reached. Peak 2 is more intense and shifted to slightly more 97 positive potential. This indicates the nature of the phase transition is different depending on scan direction. On the other hand, the capacitance curve traces peak 1 nearly identically to the negative- going scan before returning to the original minimum capacitance value in the potential region -0.8 to -0.3 V. This return to the original minimum capacitance value indicates the reformation of the well-ordered, defect free DOPC film. The overall capacitance scan of DOPC monolayer in Figure 32 can be summarized as showing the monolayer to be in a well-ordered, defect-free arrangement near the PZC, with the layer becoming porated or defective at the phase transition peaks. The reverse scan closes-up or anneals the defects as it returns through the phase transitions, and the well- ordered layer is re-achieved. The fact that the height of peak 1 and the minimum capacitance are identically achieved regardless of the potential scan direction illustrates the ideal character of this monolayer and helps make it a suitable choice for investigations of drug/lipid interaction. These characteristics were always achieved before addition of any drug to the subphase, and therefore represents a convenient control to ensure a consistent starting point. Aliquots of AmB solutions were added in stepwise fashion to the electrolyte subphase. The process of addition, measurement, and further addition allowed characterization of the lipid monolayer in the presence of increasing concentrations of additive. The concentrations used were approximately 7, 15, 30, 45, 75 and 100% of the therapeutic dose or equivalent as defined in Table 2. The effect of the added AmB or control solution was interrogated through measurement of the interfacial capacitance. We hypothesized that interaction with the phosphatidylcholine headgroups of the lipid layer would result in a decrease of the phase transition peak heights but produce little effect on the minimum capacitance of the layer. On the other hand, intercalation or direct insertion into the DOPC layer was expected to increase the minimum capacitance and decrease the peak heights. These changes were anticipated to be more prominent for the reformed layer capacitance because the annealing of defects will be hampered by incorporation or interaction with added formulation. Figure 33 shows the measured capacitance of the DOPC monolayer alone and in the presence of increasing concentrations of FZ. Panel A) shows the capacitance of the DOPC monolayer in the absence of FZ and of the uncovered mercury electrode, both as reference points. The capacitance of the uncovered electrode (i.e. of the HgI0.1M KC1 interface) is repeated in each panel. In panel 98 B), FZ was added to a concentration of 30% TD. The minimum capacitance of the as-deposited layer is increased slightly, to 1.8611F/cm 2 . A small but noticeable up-curvature near -0.8 V shows that the layer is more defective than in the absence of AmB. Peak 1 reaches a height of approximately 58[LF/cm 2 , a substantial drop from the 101 p,F/cm 2 observed for DOPC alone. Peak 2 is shortened and broadened, but the capacitance value at more negative potentials remains unchanged compared to that measured for the DOPC alone. On the return-scan, the changes are more dramatic. Peak 2 is no longer observed, while peak 1 is shifted to more positive potential, broadened asymmetrically, and shortened considerably to 2011F/cm 2 . In the minimum capacitance region, the capacitance of the reformed layer is higher than the as-deposited layer, indicating a more defective monolayer. The value of the capacitance here is 2.08 p.F/cm 2 . Panel C) shows the effect of 75% TD of FZ on the capacitance of the DOPC monolayer. The minimum capacitance of the as-deposited layer is similar to that observed at 30% TD concentration, 1.82 [LF/cm2 . The height of peak 1 is reduced to about 50 FIF/cm 2 and the shoulder is no longer prevalent. Peak 2 is completely eliminated, but again the capacitance in range -1.1 to -1.2 V is unchanged. The reformed layer capacitance shows no peak 2 and peak 1 is divided into a small, sharp step at -0.95 V and a small broad bump near -0.9 V. The minimum capacitance region shows very clear curvature near -0.7 V and never recovers the capacitance value of the as-deposited layer. The minimum capacitance is 2.17 [I.F/cm2 . The capacitance curve in the presence of 100% TD of FZ, shown in panel D), follows similar trends. The capacitance of the as-deposited layer shows a minimum of 1.88 I.LF/cm2 , peak 1 is reduced to 25 [LF/cm 2 and peak 2 is no longer present. The reverse-scan of the reformed layer does not display the peak 2 phase transition, and peak 1 is nearly eliminated. The minimum capacitance region is no-longer potential-invariant, being significantly concave up, both near -0.8 and -0.3 V. The capacitance value at the minimum is 2.33 FIF/cm 2 . The capacitance curves for 7, 15 and 45% TD of FZ are not shown but the effect of FZ at those concentrations can generally be interpolated from the other concentrations. In fact the general characteristics observed in the capacitance measurements of all AmB formulations and control solutions are similar to those noted in Figure 33. The disruptive effects of these additives are, in 99 each case, noted as a decrease in the heights of the phase change peaks 1 and 2 as well as increases in the minimum capacitance. Figure 34 presents the capacitance of the DOPC monolayer in the presence of 100% TD or equivalent of each AmB and control sample tested. Panel C) represents the DOPC monolayer alone, whose characteristics were described earlier. It is included here as a reference. Panel B) shows the interfacial capacitance when FZ was present in the subphase. The changes relative to the DOPC layer alone have been described above, but we can note here that relative to the other formulations examined FZ showed a strong influence on the monolayer. FZ shows the largest increase in as- deposited layer minimum capacitance of all the formulations. The disruption of the DOPC monolayer maybe due to a strong interaction of non-conductive pre-pore complexes of AmB with the membrane [63, 89, 166]. Brutyan [63] suggests these pre-pore complexes are loose aggregations of AmB molecules which lie atop the lipid, interacting with the lipid headgroups in a flat orientation. Addition of HTFZ produced a similar interaction to that of FZ both in its characteristics and its magnitude (Panel A). Considering the significant structural differences between aggregates of AmB in FZ and HTFZ, it is interesting that FZ and HTFZ provoke a similar response in the DOPC layer capacitance. FZ and HTFZ are both a —2:1 mixture of deoxycholate and AmB. Therefore a deoxycholate control was also measured (Panel D). The minimum capacitance of the as-deposited layer is the same as the DOPC monolayer, although peak 1 intensity was reduced to 65 µF/cm 2 . The reformed layer demonstrated a similar decrease in peak height to —52 11F/cm 2 , but the minimum capacitance remained at 1.77 uF/ cm 2 . Deoxycholate is an anionic surface active agent that can be expected to interact with the DOPC head groups. It is evident from the capacitance measurements that deoxycholate only interacts with the lipid in a peripheral way, slightly influencing the potential- induced lipid phase transition, but not increasing the minimum capacitance. The AmB/deoxycholate complex (FZ or HTFZ) had a significant disrupting effect on the layer which must be due to the AmB and not the deoxycholate. Initially AmB strongly interacts with the lipid monolayer. Once defects or pores are created in the lipid layer through scanning the potential negatively to the phase transition peaks, AmB can insert itself into the lipid matrix causing a 100 120 100 - U- a)0 80 0 60 #40440044004141410$0664 X33 0 DOPC as deposited DOPC reformed 40 ^ 0.1M KCI 20 .....^...... •••"•.......................... •• ........ 0 -1.20^-1.00^-0.80^-0.60^-0.40^-0.20 Potential / V vs SCE Figure 32 Capacitance of Hg electrode as a function of potential. Labels refer to what is at the metalisolution interface. DOPC as deposited refers to the negative-going potential scan, DOPC reformed refers to the positive-going potential scan and 0.1 M KC1 is the designation for the uncoated electrode. Capacitance of DOPC layers in potential range -0.25 to -0.8 V is shown scaled 33 x to demonstrate its invariance in this range. Two pseudo-capacitance peaks, representing molecular reorganization of the monolayer, are visible at -0.95 (`peak 1') and -1.04 V (`peak 2'). 101 D) " • 444,44 120 - 100 - 80 - C) 120 - E 40 — DOPC 75% TD FZ as-deposited DOPC 75% TD FZ reformed, 1.‘"4"aavvAivit..440.4,0,0-«•..ww.A.......v...../...4---• X33 20 -....................................................... as 80 )(33 — DOPC 30% TD FZ as-deposited DOPC 30% TD FZ reformed......... ................ ...............20 - 0 . • • . • • 100 - 80 - 60 - B) 60 - 40 - ^%"1'4,19-••••^^........wko•••••••"4`4.^X33 60 - — DOPC 100% TD FZ as-deposited ^ 40 - ^ DOPC 100%^TD FZ reformed .... ................................. ^. • - ....... • 120 100 80 #1141■14410414401001100191101.4414004,4011.1001,4***1011 X33 - DOPC as-deposited DOPC reformed ^ 0.1M KCI^ ......... ................................ 0 -1.20^-1.00^-0.80^-0.60^-0.40^-0.20 E / V vs SCE Figure 33 Capacitance of DOPC monolayers in the presence of added FZ as a function of potential. Concentrations of added FZ are 0%, 30%, 75% and 100% TD. 100% TD for FZ corresponds to 0.54 p,M AmB. Capacitance in the potential range -0.3 to -0.8 V has been magnified 33x for clarity. Full description of the AmB-induced changes is given in the text. 102 20 - 0- 60 40 20 significant decrease in lipid order. The intercalation of AmB into the layer at these potentials hinders the reformation process by providing defect sites which cannot be annealed. It is clear from the increased minimum capacitance of the reformed layer that once AmB has had the opportunity to intercalate into the monolayer, the reformation process does not expel the drug. ABLC, the lipid-complexed formulation of AmB which clinically provides decreased toxicity even though it is administered at 10-fold the dose of FZ, was also examined for its effect on DOPC monolayer capacitance. Panel E) shows that the addition of 100% TD of ABLC in the subphase elicits a small decrease in the peak height for both the as-deposited and reformed layers. The values reached are —70 and 33 iff/cm2 , respectively. The observed decrease is considerably smaller than that of FZ and HTFZ despite the fact the AmB concentration here is 10x greater. The as-deposited layer minimum capacitance was not strongly perturbed (1.75 p.F/cm2) but a slight increase in the monolayer capacitance was measured for the reformed layer (1.86 [IF/cm 2). This increase is indicative of AmB insertion into the layer and creation of permanent defect sites. ABLC is composed of a lipid complex of AmB/DMPC/DMPG, and the appropriate control, ABLC Blank, made up of DMPC/DMPG was also examined. The ABLC blank measurements (Panel F) were virtually-identical with the baseline DOPC results, indicating that the lipid complex in solution had little influence on the lipid monolayer. The changes observed for ABLC are thus due to the presence of AmB which enhances the interaction with the lipid monolayer. The lipid-complexed AmB and the ABLC blank seem to interact only with the DOPC head groups, thereby influencing the phase change but not introducing defects in the DOPC layer in the minimum capacitance region of the as-deposited layer. ABLC has been shown to have low in vivo toxicity though remaining effective at the recommended TD. These results illustrate the limited influence of this formulation on the disruption of the pure DOPC monolayer, in contrast to the FZ results presented above. Figure 35 compiles capacitance peak data and minimum capacitance data as a function of concentration (% TD or equivalent) for the five solutions used. Representative error bars are given at the highest concentrations and are one standard deviation from the measurement of at least three 103 different lipid monolayer depositions. Variability in the measurements was greatest at highest concentrations, so as an upper-limit, the estimated error bars for the remaining data points can be considered to be proportional. The left-hand column, representing the as-deposited layer will be considered first. The minimum capacitance for the as-deposited layer increases with subphase concentration for only the FZ and HTFZ formulations. The peak capacitance changes in a more complex manner. The largest changes observed are again for the FZ and HTFZ with a sharp decrease from 100 to —35 1.1.F/cm 2 . Deoxycholate, the FZ control, also influences the peak capacitance, but in a less dramatic manner, decreasing it to about 6011F/cm2 . This again indicates that the largest changes to the DOPC monolayer were observed in the presence of AmB. In contrast to these large changes, the ABLC blank only causes a decrease in the peak capacitance at the lower concentrations, but no significant decrease was observed. The decrease in peak capacitance for ABLC was larger than for the ABLC blank, similar to that observed for deoxycholate. These results reveal that ABLC induces smaller changes or disorganization in the as-deposited DOPC monolayer as compared to FZ or HTFZ. The matrix in which the AmB exists also has a significant influence. Comparing the ABLC blank and deoxycholate results clearly demonstrates this effect. Deoxycholate influences the DOPC layer in a manner that appears similar to ABLC itself, while the ABLC blank does not interact or disrupt the lipid monolayer. These observations suggest that the solubilizing agent may also cause damage to the membrane in addition to the effect of AmB. Scanning the potential to the negative limit introduces pores or defects into the adsorbed lipid monolayer. Once porated, reformation of the adsorbed layer provides further information on the interaction of the drug with defective regions of the DOPC monolayer. The results for the reformed layer can be used to estimate the effect of the drug on sterol-free cell membranes which are under stress, creating weak areas within the membrane. An example of this kind of stress is osmotic stress; the mechanical stress on the membrane as a result of cell volume swelling or shrinking in response to differences in osmolality between intracellular and extracellular solution. Osmotic stress has been shown to be a sensitizing condition for AmB ion channel formation [76, 87]. AmB's most important toxic side effects occur in the kidney, where osmotically stressed cells are prevalent. Conversely, 104 the results of the as-deposited layers can be considered to model the effect of the drug on sterol-free cell membranes which are not under stress. The capacitance peaks measured for the reformed layer show that FZ and HTFZ significantly affect the organization of the layer. The capacitance peak disappears at concentrations above 40% of the TD. The value of —811F/cm 2 for the height of peak 1 achieved when FZ or HTFZ are present above 40% TD represents the baseline value of capacitance in this potential region. Deoxycholate also decreases the peak heights but less efficiently than FZ or HTFZ. ABLC strongly interacts with the defective layer but the response is saturated at 35 [IF/cm 2 , illustrating that the lipid-complexed AmB has a limited influence on the defective monolayer. The ABLC blank has no influence on the reformation of the DOPC monolayer. The excess lipid present in the ABLC may act to mitigate the disruptive influence of AmB, or the lipid may shield the sterol-free DOPC monolayer from interaction with the complexed AmB. Support for the general discussion above was also observed with the minimum capacitance for the reformed layer. The only significant increases are observed for FZ and HTFZ, while the other formulations did not modify the capacitance of the reformed monolayer. This suggests that FZ and HTFZ have components that are incorporated within the lipid monolayer, while ABLC, ABLC blank, and deoxycholate are displaced from the monolayer through the defect-annealing process of the positive potential sweep through the phase transitions, resulting in a reformed DOPC layer similar to the as-deposited layer. These indirect observations of the disruptive effect of AmB, in its various formulations, on the organization and molecular order of a one component lipid layer are generally correlated to the in-vivo toxicity results [167]. That is, FZ has a strong toxic effect on renal cells which limits the amount of drug that can be administered. Complexation of AmB with lipid (ABLC) demonstrates a lower toxicity but maintains its efficacy though requiring a larger dose. The lowered toxicity may result from the buffering or healing effect of the excess lipid. Other AmB formulations not examined here such as Ambisome (liposomal AmB) and Amphotec (AmB/cholesteryl sulfate complex) are also known to have lower toxicity. 105 Deoxycholate ..... ........................ 120 100 80 60 40 20 0 Abelcet ....... ........................ • • - • • • 120 100 80 60 40 20 0 120 100 80 60 40 20 0 120  - 100 80  - 60  - 40  - 20 - ^ A) HTFZ ...... ..... • • ............ 0 - 120 -E 100 - 80 - 60 -c.) 40 - a_ c.) 20 - cc; 0 0 120 100 80 60 40 20 0 B) ......... • ................ FZ -1.20 -1.00 -0.80 -0.60 -0.40 -0.20^-1.20 -1.00 -0.80 -0.60 -0.40 -0.20 Potential / V vs SCE^ Potential / V vs SCE Figure 34 Capacitance of DOPC monolayer with different formulations of AmB or control added. The solid line denotes as-deposited monolayer, the dashed line represents the reformed monolayer, and the dotted line gives the capacitance of the uncoated Hg electrode. Concentration of the added formulation is 100% TD or equivalent for control samples. 100% TD for FZ and HTFZ is 0.54 11M AmB; for Abelcet (ABLC) it is 5.4 p.M. Panel C gives the capacitance of DOPC monolayer in absence of any addition, and serves as a reference point. Panels A), B) and F) show the capacitance curves of formulations containing the AmB as active-ingredient. Panels D) and F) correspond to the effect of the control samples, deoxycholate and ABLC blank. Full description of the data is given in the text. 106 2.4 - 2.2 -E LL --- 2.0 -E E E 0 1.8 1.6 • Deoxycholate • FZ • HTFZ • Abelcet Blank • Abelcet 100 80 E0 60 0 -4Q. 0 20 - 0 ^ I 0 20 40 60 80 100 II^I 0 20 40 60 80 100 % Therapeutic Dose or Equivalent As-Deposited Layer 0 20 40 60 80 100 % Therapeutic Dose or Equivalent Reformed Layer 0 20 40 60 80 100 Figure 35 Summary of key features from capacitance of DOPC monolayer with added AmB formulation or control sample. Top panels represent the capacitance value of peak 1 for as- deposited (left) and reformed layers (right). Bottom panels represent value of capacitance averaged over the -0.35 to -0.65 V potential range. Representative error bars of one standard deviation are shown for the highest concentrations. Reproduced with permission from [3], Copyright 2002 Elsevier. 107 4.5.3 Tr Reduction Measurements AmB's mechanism of action, at least in part, revolves around the formation of pores or ion channels in cell membranes. These pores allow leakage of intracellular media, leading to cell death. As detailed in Chapter 2, studies on liposomes and bilayer lipid membranes have shown increased permeability of mono- and divalent cations as well as small non-electrolytes (e.g. glucose). Indeed, measurement of IC - release is a de facto standard technique in determining activity of new AmB formulations. It is accordingly desirable to examine the influence of AmB formulations on the porosity of our supported DOPC monolayer. These measurements will provide a complement to the capacitance measurements since the latter can investigate the effect of the drug on the model membrane only through an indirect albeit sensitive manner. The conventional wisdom of AmB pore-formation is that initial AmB interaction with the cell membrane is the formation of pre-pore complexes on the membrane surface, then interaction with membrane sterol (cholesterol or ergosterol) to insert into the membrane and form a pore. However, our model system is sterol-free. Since AmB-induced pore formation has also been noted in sterol- free systems [68, 86, 168] challenging ideas about AmB mechanism, our system provides a unique opportunity to further study this behaviour. The formation of pores in sterol-free models has often been associated with osmotic stress. As mentioned above, the process of potential-induced defect formation can be considered to be an analogous stress. The porosity of the lipid monolayer can be characterized by measuring metal ion reduction through the monolayer, assessing the barrier properties of the layer. This data can be investigated in two ways. The electrode potential required to generate a certain reduction current is related to the porosity of the film because a larger overpotential will be required if the film is less porous. Variations in measured current due to metal ion concentration or electrode area can be dealt with by normalizing the current with respect to the diffusion-limited current. The plot of current vs. potential resembles a polarogram for the electroactive species. While this approach allows comparison between the pore-forming ability of different membrane-interacting species, no quantification of the formed pores is possible. The second method allows such quantification. In this case, the shape of the measured current transient can be analyzed to give an average pore size 108 and pore number density, based on a model which treats the pores as being part of a random array of active microelectrodes in a blocking (redox inactive) matrix. From this, the fraction of the surface which is electroactive (available to redox reaction) can be determined. Before examining the data in each of the two ways, it is important to first consider the nature of the ion reduction process. As described fully in Chapter 3, diffusion to a pore occurs in a hemispherical fashion. This results in measured currents that are larger than what would be expected for linear diffusion to an active area of equal size. For an array of pores, the diffusion behaviour is more complicated. At short times after the reductive potential step, diffusion proceeds to each pore individually through hemispherical diffusion, but when neighbouring hemispherical diffusion fields overlap, the nature of the diffusion evolves to become linear diffusion. At long times after the step, measurement of the current for an array of microelectrodes gives the same value as would be achieved for a fully-active, un-blocked electrode surface. Although still primarily covered by monolayer, the electrode can appear uncovered. An additional contribution to this apparent low coverage can occur if the blocking layer is non-uniform. For example, thin regions still provide coverage but may allow ion transport and thus appear uncovered. Figure 36 plots the normalized T1+ reduction current 50 ms after the potential step as a function of electrode potential when the DOPC monolayer is in the presence of the AmB formulations or controls. The results of the same measurement in the absence of any monolayer and for the DOPC monolayer alone are given as reference points. These measurements allowed for investigation of the effect of potential on the integrity (porosity) of the adsorbed lipid layer and the influence the additives had on the monolayers. Panel a) compares the results for the uncoated Hg electrode, the DOPC monolayer alone, DOPC with ABLC, and DOPC with ABLC blank. The reduction current for the uncoated electrode rises sharply beginning at a potential of -0.4 V. The limiting current is reached for potentials more negative than -0.55 V. The value of potential where the current is half the diffusion-limited value can be taken as the reversible potential for the Tr/T1(Hg) reduction reaction, and is about -0.45 V. The standard reduction potential for this reaction is -0.575 V vs. SCE. With the DOPC monolayer coating the Hg electrode, the potential required to reach half the diffusion-limited current, V2( is shifted to about -0.88 V. The potential at which 1/4(i-.limiting) is 109 reached will be called the break-through potential. The overpotential required to drive the Tr through the monolayer is then 0.43 V, showing the DOPC monolayer represents an excellent barrier to ion transport, characteristic of a highly-organized defect-free monolayer. Figure 36 includes the measured capacitance of the DOPC monolayer in the region of the phase transitions to remind that there is a potential-induced process of defect formation which occurs in the —0.1 V positive of the first peak. In the absence of additives, it is this potential-induced defect formation which rationalizes the increase in reduction current observed at -0.88 V. That this is indeed due to changes in the monolayer and not simply to sufficient driving force for Tr transport, has been proved by the observation that Cd2+ transport begins in the same potential region [165]. When AmB formulation or control is added, any increase in the reduction current at a given potential directly indicates increased porosity of the lipid monolayer. ABLC blank, being a simple mixture of lipids and containing no AmB, was expected to show little effect on Tr transport, and this was observed. Only a very slight decrease in overpotential was required for ion transport relative to the DOPC layer. With ABLC added at 100% TD concentration, it is observed that the overpotential required for Tr transport across the monolayer and reduction is decreased by about 35 mV relative to the pure DOPC monolayer. The effect of ABLC is to introduce some defects which allow increased ion transport. Nonetheless, the required overpotential remains quite large, indicating the created defects are few and the monolayer remains a significant barrier to transport. That no significant changes were observed for ABLC and ABLC blank supports the minimum capacity measurements which indicated an undisturbed DOPC layer. Panel b) presents the analogous results for FZ, HTFZ and deoxycholate each at 100% TD or equivalent. The measured current in the presence of FZ is greater than that for DOPC for potentials from -0.75 V until the diffusion limit is reached. The break-through potential is 50 mV less negative than that for the DOPC alone. Additionally, the reduction current vs. potential curve for FZ shows a larger slope in the important potential range -0.75 to -0.85 V. FZ interaction with the DOPC monolayer can be said to result in an increase in porosity, although the overall barrier property of the monolayer remains intact. Underscoring the fact that the changes observed for FZ are due to AmB, the measurement with deoxycholate has an overpotential intermediate between FZ and DOPC 110 alone at the break-through potential. The effect of added deoxycholate is similar to that observed for ABLC. HTFZ produced the most dramatic changes in the reduction profile. Significant reduction current was measured beginning just negative of the reversible potential. This current reached a plateau of about 30% of li mi t ing until the onset of potential-induced defects near -0.8 V. A constant, small number of HTFZ-induced pores or defects would explain this plateau behaviour. The number must be small, meaning the average inter-defect distance is large. If the defects were spaced close together on average, the diffusion-fields to the individual pores would have overlapped by the 50 ms time of our measurement and we would measure the diffusion-limited current. This type of behaviour has been noted previously when the ion channel forming protein gramicidin was incorporated into a DOPC monolayer [169]. The electrode potential needed to reach the break- through potential with HTFZ was more than 0.1 V more positive than that of the DOPC monolayer alone. Clearly, HTFZ interaction with the monolayer drastically increases the porosity. The second method of analyzing the T1+ ion transport measurements was to fit the expected current transient for a random array of microelectrodes model to the measured current transient. This fitting process resulted in values for the mean radius of the pore (r) and the number density of pores (N). The fraction of the electrode surface covered with lipid, the coverage 0, can be calculated by determining the total area occupied by pores. This value of 0 must be considered to be an estimate rather than an exact value because metal ions are known to be able to penetrate through the intact monolayer if it is thin or when the overpotential is suitably large [170-172]. The range of pore radius estimated through this current transient fitting approach is limited by experimental factors. The minimum determinable size is about 0.1 p,m because the measured current transient is always a convolution of capacitive and faradaic current. Determination of small pore sizes relies on using the portion of the transient closer to the early times (< 5 ms) after the potential step, where the capacitive current dominates. The maximum determinable size is limited by the time at which the individual hemispherical diffusion fields overlap to give linear diffusion. For a very porous system, either with many closely-spaced pores or fewer large pores, the transition 111 occurs very quickly after the potential step. In this case, after the capacitive current has dissipated, it may be possible only to observe linear diffusion and no size determination can be made. When not limited by experimental constraints, the determined pore size must still be interpreted with caution. It is known that the current transient for diffusion to a random array of microelectrodes (i.e. a process involving initial non-linear diffusion) appears similar to the transient expected for a simple linear diffusion process in which the electron transfer reduction reaction is preceded by a chemical reaction (i.e. complexation) step. Such a so-called C-E' mechanism has been invoked in the understanding of Tr ion transport through gramicidin ion channels incorporated in a DOPC monolayer [132, 169]. Gramicidin is known to be a selective ion channel, allowing through only monovalent cations. This selectiveness suggests a chemical complexation step, likely through a binding site at the mouth of the channel, and justifies the choice of examining the transient as a C-E process. AmB pores have been found to be cation-selective when AmB added to one-side of a model cell membrane, but allow passage of mono- and di-valent cations [173]. Khutorsky has used computer modelling to identify an energy minimum about half-way through one side of a double- sided pore [174], but no specific cation-binding sites for AmB pores have been identified. Accordingly, our analysis did not deal with the transient possibly being characteristic of a C-E process. Lastly, it is important to recognize that within the random-array of microelectrodes model, a patch of small pores within the monolayer maybe indistinguishable from a patch-sized pore. The results of the fitting procedure are shown in Figure 37 as a function of additive concentration for the potential step to -0.85 V. This potential is in the region where potential-induced defect formation occurs, and was chosen because it represents the potential where the greatest changes due to additive interaction occur. Since the current transients are due to a reduction process and the potential steps were made in the negative direction, these results are characteristic of interaction with the as-deposited monolayer. The data points represent the average result over each additive's set of measurements (1\1 - ^ 3). The error bars represent one standard deviation about the average for the highest concentration data point shown. The data are divided into two groups based on their similarity to ABLC or FZ. The 0% TD value of each panel represents the values calculated from the DOPC monolayer alone. The value of 0% TD in each column represents a different 112 measurement, based on Tr reduction current transients before the first addition of respective additive. The Abelcet group consists of the ABLC and the AmB-free lipid analogue ABLC blank. ABLC displays slightly larger holes and density of holes than does the blank. In both cases, the size of the holes increase with % TD, and the density slightly decreases. The calculated coverage 0 of lipid is close to one for the blank, but a consistently smaller coverage was measured for ABLC. In fact, a decrease in lipid coverage was observed for ABLC at the 30% TD concentration. This may be related to the changes observed for the as-deposited capacitive peak maximum in Figure 35, where a sharp decrease in peak height was observed until 30% TD ABLC was reached. This suggests that ABLC/DOPC may interact strongly at this concentration, introducing holes or defects in the monolayer. The calculated active size of the ABLC-induced pores was determined to be a radius of 5-12 p.m, varying with concentration. The number density is on the order of 100,000 per cm 2 , meaning that there are on the order of 1000 pores on the electrode. The FZ group contains the fitting results for deoxycholate, FZ and HTFZ. For this group, the data presented in Figure 37 does not cover the full % TD range. FZ and HTFZ at high TD ( > 50%) produce very porous lipid monolayers. Linear diffusion behaviour thus occurs very rapidly after the potential step, preventing an estimate of the coverage or hole size and density. For coverage values below —0.5, large deviations in the fitting parameters were observed and therefore the standard deviation for HTFZ is not shown. Measurements of deoxycholate-induced permeability were only made up to 80% TD. Deoxycholate interacted with the DOPC monolayer yielding holes of radius and number density similar to that observed for ABLC. An increase in hole size was observed for TD values greater than 40%, and this is reflected in the calculated coverage. FZ gave values very similar to deoxycholate, though the layer was less porous (higher 0) at 40% TD. Above 50% TD concentration the FZ presence produced a lipid layer that was very porous, giving a calculated lipid coverage of about 0.5. The largest disturbance in the integrity of the monolayer was observed for HTFZ. The average hole size sharply increased for values above 20% TD, reaching 15 1.tm. The hole density induced by HTFZ shows a maximum at —10% TD, and then decreased while the pore size was increasing. This may reflect an in-monolayer aggregation process, although the coverage 113 simultaneously declined rapidly, showing an overall increase in pore area. Numerous studies in literature have raised the possibility of pore-formation requiring a threshold number of AmB molecules to be embedded in the monolayer. This data illustrates the disruptive effect of HTFZ on a pure DOPC monolayer, disrupting and porating the DOPC monolayer at moderate % TD. FZ is heat-treated to reduce toxicity, but our HTFZ results demonstrate a larger disruptive influence on pure DOPC monolayer as compared to FZ. In contrast to published in vitro and in vivo toxicity studies on HTFZ [45, 46, 48], our measurements were performed on sterol-free lipid monolayers. The results might differ significantly with incorporated sterols. For all additives, the calculated pore radius was between 5 and 15 pm. Considering that AmB pores are said to have an internal diameter of 4-6 A and external diameter 17 A [73, 175], it is important to comment on our results. Structures of micron-size dimension have been observed in floating lipid monolayers at the GIS interface [176, 177], suggesting the pores we measure may represent porous patches within the lipid monolayer. Milhaud et al. used atomic force microscopy to probe AmB interaction with sterol-free DPPC lipid bilayer supported on mica and found 0.2-0.5 pm structures covering nearly the complete surface within patches —10 p,m 2 [178]. The fact that many of these structures exist close together fits well with our idea that we may be measuring a porous patch of monolayer as if it were an individual pore. The results of the T1+ reduction experiments correlates well with the capacitance observations described previously. In both tests of control solutions, only minor changes were observed; the lipid monolayer was essentially unchanged for ABLC blank, and moderately disrupted in the presence of deoxycholate. ABLC does interact with the DOPC monolayer at intermediate TD values, but generally the DOPC monolayer remained intact. FZ and HTFZ had the largest destructive effects on the lipid monolayer, and for values > 40% TD, the barrier property of the lipid layer was eliminated, offering no impediment to metal ion reduction. In all cases, the formulations containing AmB were significantly more disruptive to the DOPC monolayer than the corresponding blank 114 1.0 0.8 tz g- 0.6 .E 0.4 0.2 0.0 ^ Uncoated Hg o DOPC • Abelcet • Abelcet Blank 100 80 60 40 20 0 - 100 - 80 - 60 - 40 - 20 0 -1.2 -1.1 -1.0 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 E / V vs. SCE Figure 36 Normalized T1+ reduction currents for DOPC-coated Hg drop in contact with subphase containing AmB formulation or control at 100% TD or equivalent concentration. The measured reduction currents in the absence of any monolayer are shown for reference. The capacitance of the DOPC monolayer is shown in the potential range -0.85 to -1.15 V to remind that the phase transitions observed result in a porous layer and are preceded by nucleation of defects. The impermeability of the DOPC layer alone is demonstrated by the large (0.4 V) overpotential required to measure Tr reduction. The presence of AmB formulations decreases the required overpotential for reduction. Full explanation of the data is given in the text. Reproduced with permission from [3], Copyright 2002 Elsevier. 115 Abelcet Group^ FZ Group HFZ liE FZ Deoxycholate 5.8 cCr" 5.6 E Q 5.4 ,.,-- 5.2 0) 2 5.0 4.8 15 E 10 t. # Abelcet blank Abelcet i Abelcet • Abelcet blank HFZ FZ i Deoxycholate , (i)^ , HFZ ti i FZ # Deoxycholate 0 20 40 60 80 100 0 20 40 60 80 100 %Therapeutic or Equivalent Dosage Figure 37 Calculated pore radius (bottom), pore number density (middle), and lipid coverage 0 (top) for DOPC monolayer with added AmB formulation or control solution. Data points represent average of calculation results for individual runs, while error bars indicate one standard deviation for highest concentration datapoint given. Full explanation of the data is given in the text. Reproduced with permission from [3], Copyright 2002 Elsevier. 116 experiments, illustrating that AmB is the major contributor to the disorganization and destruction of the lipid layer. 4.6 Conclusions A model system consisting of a DOPC monolayer adsorbed on a Hg electrode was used to study the interaction of three formulations of AmB with the lipid monolayer. The character of the interaction was interrogated utilizing electrochemical techniques sensitive to the organizational state and porosity of the lipid monolayer. The model system was sensitive to changes in the lipid layer due to the interaction with various formulations of AmB. The largest disruptive interactions were observed for the FZ and HTFZ formulations, with HTFZ-induced porosity significantly different than that of FZ . Deoxycholate (control solution for FZ and HTFZ) weakly interacted with the lipid layer, significantly less than when present with AmB. The lipid AmB complex, ABLC, demonstrated a much lower disruptive effect towards the lipid monolayer, even though the amount of AmB in the subphase was ten times larger when compared to FZ. The ABLC blank did not significantly influence the character of the DOPC monolayer. The lipid present in the ABLC may have a shielding, annealing and/or healing effect that buffers the disruption due to the presence of AmB. 117 5 FLUORESCENCE MICROSCOPY OF Hg-SUPPORTED DOPC MONOLAYER 5.1 Introduction In the previous chapter, the biomimetic system of a DOPC monolayer adsorbed on a Hg electrode was introduced and used to study AmB's interaction with a lipid layer. In this chapter, the application of in situ fluorescence microscopy of the DOPC layer at the Hg electrode is described. These spectroelectrochemical experiments were designed to probe the nature of the adsorbed DOPC layer itself and provide insight into the behaviour of the lipid layer under the influence of electric field. The variation in this behaviour in the presence of AmB was proposed to give an indirect measure of AmB interaction. The spectroscopic measurement of, and the observation of potential- controlled fluorescence from, a lipid-coated Hg electrode have not been reported previously. We incorporated a fluorescent dye into the lipid monolayer at low concentration, and measured . fluorescence from the monolayer as the electrode potential was varied. We first worked with AmB- free systems to benchmark the behaviour of our model membrane system. We correlated changes in fluorescence intensity with changes measured in the capacitance signal to show conclusively that at extremely negative potentials the lipid monolayer is desorbed from the electrode surface. Additionally, the potential-induced phase transitions of the DOPC monolayer were preceded by a decrease in fluorescence intensity, allowing us to comment on the nature of lipid behaviour in this potential region. Finally, we applied the in situ fluorescence microscopy technique to the DOPC monolayer when AmB was present in the subphase. This was ultimately unsuccessful due to the variation of in situ fluorescence intensity being dominated by potential-induced changes of the host lipid layer rather than due to drug interaction. The results of these experiments are important because they increase our understanding of the biomimetic system of the DOPC monolayer on Hg. Although mentioned previously, we reiterate here that the characterization of lipid-coated electrodes offers a unique way to investigate the effect of electrical variables (lipid surface charge density, transmembrane potential) on model cell 118 membrane behaviour, as well as being useful for possible biosensor applications. These electrical variables have important roles in controlling or moderating a whole host of biological processes, including cell fusion, adhesion, and division, ion transport, antigen/antibody recognition, proton and electron transport and redox reactions. Understanding the electrical properties of the membrane surface is a step toward having a full description of physiological behaviour of biomembranes; studies on model systems are an increment within this step. Although our specific biomembrane model system has been well characterized electrochemically, the information obtained from in situ spectroscopy is typically complimentary in nature. Electrochemistry can measure processes occurring at the electrode surface and does so in a spatially-averaging way. That is, the response is across the whole electrode area. By contrast, the fluorescence microscopy technique we use here is most sensitive to the region near, but not on, the electrode surface and the use of microscopy provides spatial resolution. The research described in this chapter had several objectives. We theorized that spectroscopic characterization of the electrode surface would increase our understanding of lipid behaviour under an electric field. The potential-induced phase transitions that occur in the DOPC monolayer adsorbed on Hg were introduced in the previous chapter. The nature of these transitions is not fully understood; numerous questions exist about the physical changes in the monolayer at the transitions. It is known that the pseudocapacitance peaks manifested by the transitions are related to changing surface coverage of lipid, but it is not known how this coverage changes. Additionally, electrochemical evidence has suggested the DOPC monolayer can be desorbed or displaced from the electrode surface if sufficiently negative potential is applied, with re-adsorption of the monolayer if the potential is then made more positive. The form of the displaced layer and its physical location were unclear. The natural questions is, Where does the organic layer go when desorbed? and a follow-up question, Why it is possible to reabsorb the layer with no observable loss of material? By providing sensitivity to the lipid when it is no longer on the electrode surface, fluorescence microscopy was expected to provide insight into these questions. The work needed to reach these objectives was simplified by a colleague's inaugural use of our fluorescence microscope for in situ spectroelectrochemical studies of octadecanol adsorbed onto a 119 gold electrode [179]. The results from that system form an important basis of comparison for this work. Specific points that needed to be addressed for success with the DOPC on Hg system were nonetheless numerous. Selection of a suitable, lipophilic fluorophore was required, along with characterization of its effect on the electrochemistry. An ideal dye would be one combining excellent spectroscopic properties with a minimal disruptive effect on the organization of the lipid monolayer, and displaying no susceptibility to phase segregation. Additionally, we learnt how to deal with the fact the electrode surface is spherical rather than planar, and determined reasonable exposure times for capture of fluorescence images with the digital camera. At times this was a balancing act because the monolayer would fluoresce dimly at some potentials and very intensely at others. The fluorescence images displayed unexpected spatial variation, and the intensity varied with potential in a manner which we had not anticipated. Accordingly, customized methods of data analysis were developed, and the reproducibility of the results characterized. This work, and a related fluorescence characterization of a supported lipid bilayer structure formed on the Hg electrode have been published in the journal The Analyst [4]. 5.2 Literature Review To place the work of this chapter within the greater body of scientific work, relevant studies in the fields of spectroelectrochemistry and surface-modified fluorescence are discussed. The review of spectroelectrochemistry focuses on the biological applications ofthe technique and provides specific references to work on electrode-supported biomimetic systems. For the spectroscopy, the presence of a metal surface near a fluorophore can invoke dramatic changes in the properties of the fluorophore. These changes and their mechanisms are given. Lastly, the properties of fluorescent probes are described very briefly, simply to introduce the reader to the complexities of the field and to provide basic literature sources. The base system used in this chapter is the monolayer of DOPC adsorbed on Hg electrode, which was described and its development cited in the previous chapter. Agak's electrochemical impedance spectroscopy measurements [137] of the DOPC monolayer desorbed from the electrode at very 120 negative potentials showed the interface to have unusually low ion mobility, interpreted to mean the desorbed lipid must reside very close to the electrode surface. 5.2.1 Spectroelectrochemistry The first uses of spectroelectrochemistry were in monitoring redox reactions at optically-transparent metal thin film or semiconductor electrodes [180, 181]. The redox species examined were visible- light absorbing organic compounds such as methyl viologen, and non-aqueous solvents were used. Generalization of the technique to aqueous solutions, to reflective (rather than transparent) electrodes, and to UV and IR wavelengths followed [182, 183]. The Encyclopedia of Electrochemistry series gives recent reviews on UV-visible [184], IR [185], and Raman [186] spectroelectrochemistry. A differential technique is commonly used, subtracting the spectroscopic response at one potential from the response at a second potential. If the system is robust to potential-cycling, electromodulation can be used. Electromodulated UV-vis spectroscopy probes changes in the absorption properties of molecules at the electrode surface. Electron-transfer reactions can be studied if the absorption is different in reduced and oxidized forms of the analyte. Redox inactive species may be studied if the external electric field interacts with the static dipole moment of the adsorbate, splitting the adsorbate's energy levels; an effect called the inverse Stark shift [187]. The inverse Stark shift can be used to estimate molecular orientation as a function of potential; for example, Sagara examined the orientation of hemin (iron porphyrin and pharmaceutical drug) on HOPG [188]. The electromodulation technique can also be applied to other spectroscopies. Despite long and widespread use of Hg electrodes in electrochemical investigations, few instances of spectroelectrochemistry of the Hglelectrolyte interface exist. Previous in situ optical studies include Bewick' s monitoring of water adsorption to the Hg electrode [189], configuration as an optically transparent electrode [190], or as supported on Nafion [191] to study heptylviologen reduction. In addition, the nucleation of Hg droplets on a polished Pt electrode has been studied using light scattering [192, 193]. 121 Early biological applications were to redox-active enzymes [194] and proteins, notably cytochrome c [195]. Studies of protein redox chemistry are often complicated by simultaneous non-faradaic processes such as reorientation, conformational transition or acid-base equilibrium. The advantage that the combination of spectroscopy and electrochemistry provide is a separate measure of the structure and reaction dynamics of the adsorbed species, unfettered by the background non-faradaic behaviour. Spectroelectrochemistry of proteins has also been reviewed in the Encyclopedia of Electrochemistry series [196]. 5.2.1.1 Spectroelectrochemistry of Biomimetic Systems Groups interested in the effect of applied potential on adsorbed lipid and lipid-like layers have used a variety of in situ spectroscopies to study the interface. Sagara and coworkers studied 4- pentadecylpyridine adsorption on Au by absorption and fluorescence spectroscopy after incorporating an anthracene-derivative dye into the monolayer [197]. The desorption and re- adsorption of the 4-pentadecylpyridine was found to involve micelle formation and re-spreading respectively. Similar results were found for fluorescently-tagged stearic acid mono- and multilayers on Au [198, 199]. Later, the octadecanol on Au system was probed with spatial resolution by using confocal fluorescence imaging of a monolayer doped with a carbocyanine dye [200]. The fluorescence intensity, size and structure of desorbed aggregates differed with varying initial deposition conditions, underlining the correspondence of the monolayer state at GIS and that adsorbed to the electrode from GAS. The adsorption of octadecanol or 4-pentadecylpyridine has also been studied with neutron reflectivity measurements [201, 202]. In situ fluorescence microscopy has been used to further examine octadecanol and oleyl alcohol adsorption on gold. For these systems, fluorescence is fully quenched when the fluorophore-containing monolayer is adsorbed to the Au surface. When the electrode potential was made very negative, the organic layer was desorbed from the electrode surface resulting in a decrease of efficiency of the quenching process, and an increase in fluorescence emission. The spatial distribution of the desorbed fatty alcohol layer was interrogated. These studies found that the desorbed monolayer resides within 40 nm of the electrode surface, and that the incorporated carbocyanine dye in the alcoholic monolayers aggregates on desorption and disaggregates on re-adsorption [179, 203, 204]. The adsorption of these fatty alcohols on gold is considered to be a complimentary system to the DOPC on Hg arrangement used 122 in this work. Similar study of 2-(2'-thienyl)pyridine on gold electrode made use of fluorescence spectroscopy to observe surface electrochemical dimerization at positive potentials [205, 206]. In situ fluorescence has also been used to study the conformation and structure of DNA layers on gold and the co-adsorption of mercaptohexanol. Reorientation and subsequent desorption of fluorescent-tagged DNA induced by mercaptohexanol adsorption was observed [207, 208]. The group of Lipkowski has applied in situ FTIR spectroscopy to adsorbates on Au. Lipid bilayers of DMPC formed by liposome fusion, mono- and multilayers of octadecanol, and monolayers of 4- pentadecylpyridine have each been studied [139, 209-211]. Changes in the molecular organization of the adsorbate, measured as the tilt-angle with respect to the surface normal, were correlated to changes in the measured interfacial capacitance. Desorption of the DMPC layer resulted in a 1 nm `cushion' of water between the lipid and the electrode, determined by neutron reflectivity [201]. Data collection in neutron and FTIR studies is slow and thus requires very stable experimental systems. Additionally, one drawback of these approaches is that homogeneity within and between the bi- or multi-layers is assumed. The biological application of spectroelectrochemistry was recently taken to a new level by Busalmen and coworkers, who studied the growth of Pseudomonas fluorescens bacteria cells on polarized Au electrodes [212]. This bacteria takes its name from the fact that it secretes the fluorescent dye fluorescein, so fluorescence imaging of the influence of potential on cell growth required no additional fluorophore. Another interesting application is that of Mali, who has immobilized the biomolecule avidin or avidin-complexed nanoparticles on micropatterned Au electrodes via a thiol-linkage [213]. Fluorescence microscopy was used to follow the controlled release of the avidin or nanoparticles from the surface after reductive desorption. Potential applications in drug delivery and lab-on-a-chip devices were envisaged. 5.2.2 Fluorescence Near a Metal Surface In situ fluorescence spectroscopy of the electrode surface has proved to be an effective way of probing the interface, and has been reviewed by Dias [214]. Nonetheless, the proximity of a metal electrode near the fluorophore can change the properties of the fluorophore from those determined in bulk solvent. Fluorescence intensity, excited state lifetime and quantum yield may all be affected because the spontaneous emission of light is not an intrinsic property of the emitter but depends on 123 the local optical density of states into which the emitted photons are released [116, 215, 216]. The local optical density of states represents the ability of a localized portion of sample to support an electromagnetic eigenmode. The magnitude ofthe metal-induced changes in fluorescence properties is related in a non-trivial way to the separation distance between metal and fluorophore. Kuhn and Chance [e.g. 117, 215, 216] investigated energy transfer of fluorescent molecules near silver surfaces. Variable numbers of fatty acid monolayers were built-up as spacers between the silver surface and the fluorophore. For small silver-fluorophore separation (<50 nm), the excited state lifetime is sharply decreased, approaching zero as the separation decreases. This is due to efficient non-radiative energy transfer from the fluorophore to the metal. For a thick metal film (>> 4 the distance dependence of the excited state lifetime is to the inverse third power. At larger separations (50-500 nm), the excited state lifetime increases and decreases in an oscillatory manner. The roughness of the metal surface is also important in determining the efficiency of fluorescence quenching; Pineda and Ronis calculated that fluorescence lifetimes at rough surfaces are 2-4 orders of magnitude shorter than the lifetime at an equal separation from a smooth metal surface [119]. Variation in the surface morphology creates conditions more favourable for non-radiative energy transfer. Engineering fluorophore-metal separation distance for improved fluorescence properties has become known as radiative decay engineering; Lakowicz has summarized the field periodically in recent years in books and book chapters [217, 218]. Typically, nano-scale spherical or obloid particles are used to give metal-enhanced fluorescence, with the ratio of measured emission rate to natural emission rate increasing up to 2-3 orders of magnitude. The quantum yield increases as the excited state lifetime decreases. The most dramatic effects are for low quantum yield fluorophores. 5.2.3 Properties of Fluorescent Probes for Membrane Studies The range of fluorescent probes developed for biological research is vast. No attempt to survey this field is made here, rather it is simply important for the reader to recognize that fluorescent probes have been developed for numerous specialized applications, and each has its own unique properties. It is desirable for the probe to have the least possible effect on the properties of its host. Many have been extremely well characterized, so that properties such as the preferred lipid phase or the location 124 of the probe after incorporation into a lipid layer are known (e.g. diphenyhexatriene, DPH, and its derivative tetramethylammonium-diphenylhexatriene, TMA-DPH). Some are sensitive to transmembrane potential (e.g. rhodamine), to specific metal ions (e.g. Fura-2 sensitive to calcium), to pH (e.g. fluorescein), and probes have been designed for studying antibodies, enzyme substrates, receptors, ion channels, cell viability and so on. The choice of fluorescent probe is complicated and the handbook produced by one of the principal probe vendors is an excellent resource [219]. Fluorescein, a derivative of which was used in this work, is known to adopt a shallow position in lipid layers, residing in the polar headgroup region [220]. Coupled with the smoothness of the Hg surface, this should limit the extent of fluorescence quenching in our setup. 5.3 Materials and Methods 5.3.1 Introduction The experimental setup used in the work of this chapter is in part similar to that used to study AmB- lipid interaction and described earlier. The electrochemical measurement of the capacitance of the electrodejsolution interface was substantially the same, as was the manner of preparing the lipid- coated electrode. The minor variations made for adaptation to an arrangement suitable for in situ fluorescence microscopy will be noted, but the reader is referred to the previous chapter for a full description. The fluorescence microscopy setup has not yet been detailed and will be described completely. 5.3.2 Electrochemical Setup To accommodate the requirements of simultaneously measuring fluorescence intensity and capacitance, the electrochemical cell was modified. From the perspective of electrochemistry, the principal difference is that the 'microscope cell' has a much larger surface area. Accordingly, a greater amount of DOPC was deposited to the GIS to form a monolayer of DOPC at the equilibrium spreading pressure (ESP). A 25 ill aliquot of 2 mg/ml DOPC solution was injected onto the surface of the deaerated electrolyte. The DOPC solution was made up of DOPC and 3 mol% of 5- octadecylaminofluorescein (a lipophilic fluorescent dye, catalogue # 0-322, Invitrogen-Molecular Probes, Eugene, OR). The required amount of dye stock solution in chloroform was evaporated to 125 dryness under argon, then DOPC in n-pentane added. Final DOPC concentration was 2 mg/ml in n-pentane. Prior to simultaneous electrochemical and spectroscopic study, the adsorbed DOPC/dye monolayer was characterized by measuring its capacitance as a function of potential. This served to examine the influence of the incorporated dye on the electrochemical behaviour of the monolayer. These measurements were carried out using the same measurement methodology as in the previous chapter. During the spectroelectrochemical experiments, a modified procedure was used. The time scale of the fluorescence imaging (0.5-15 s per image) meant that a simple linear potential sweep could not be used since it is desirable to capture the image when the electrode is at a fixed potential. Instead of a potential sweep, potential steps were made first increasingly more negative and then reversed at a potential limit and made increasingly more positive back to the starting potential. Two different potential limits were used: -1.2 and -1.85 V, representing defect formation and desorption regimes respectively. Measurements were made at least every 0.1 V, but in regions of interest the step spacing was reduced. Collection of the fluorescence images and measurement of capacitance were triggered 500 ms after the potential step and the next step made after the fluorescence image saved to the computer. A custom Labview (National Instruments, Austin, TX) programme was written in house to control the electrode potential, record the potential and capacitance and note when fluorescence images were captured. 5.3.3 Fluorescence Microscopy The addition of the 5-octadecylaminofluorescein fluorophore to the monolayer rendered it fluorescent. Fluorescein has absorption maximum at 490 nm and fluorescence emission maximum at 514 nm as shown in Figure 38. The fluorescein chromophore is characterized by relatively high absorptivity, excellent quantum yield (0.93) and good water solubility. The C18 alkyl tail of the fluorescein-derivative used here helps to decrease the dye's solubility. Furthermore, within a lipid bilayer environment, the alkyl tail segregates into the hydrophobic core; similar segregation into the alkyl tail region of our lipid monolayer was expected to occur. The chromophore fluorescein remains in the aqueous environment of the headgroup region of the monolayer [220]. The fluorescence of fluorescein is known to be pH dependent, with several relevant equilibria. The most 126 important of these is between monoanion and dianion forms with pK a 6.43 [221]. The peak wavelength and shape of the fluorescence emission spectrum is largely pH independent because the dianion form dominates, with only small contributions from the monoanion [219]. The modified DOPC monolayer at the GAS was examined with fluorescence microscopy to gauge possible segregation of the dye from the lipid. The observed images showed uniform fluorescence with occasional bright spots representing concentrated dye. These aggregates could represent slight segregation of the dye into aggregates or could represent the excess of lipid material at the surface required to maintain the ESP. The modified DOPC monolayer was deposited onto the Hg electrode and characterized with fluorescence and electrochemical measurements. In situ fluorescence imaging was done using a specially-designed electrochemical cell, shown in Figure 39. An inverted epi-fluorescence microscope (model IX70, Olympus, Central Valley, PA), was mounted in a light-proof housing on a large anti-vibration table. The microscope was equipped with a 75 W Xe short-arc light source (Olympus UXL-S75XE) operating in DC mode. Lamp output was directed through an Olympus filter cube (U-MWIBA; bandpass 460-490 nm, dichroic mirror 505 nm, and acceptance band 515-550 nm) to select a band of wavelengths suitable for exciting fluorescein. This excitation light was focussed through the microscope objective (Olympus LMPlanFl 50x objective with numerical aperture (NA) of 0.5 and working distance 10 mm) up onto the Hg electrode surface. The optical window was a cover glass slip 0.17 mm thick sealed into the glass body of the electrochemical cell. This thin window minimized aberration and helped retain excellent image quality. Emitted fluorescent light was collected back through the objective, and traversed the dichroic mirror and emission filter to reject scattered light prior to being directed to the monochromatic digital camera (SPOT RT, Diagnostic Instruments, Sterling Heights, MI). Control over image collection was via the camera's own software. This camera collects the image on a Kodak interline transfer CCD (model KAI-2092) which was Peltier-cooled to 37° C below ambient temperature to reduce dark noise. The CCD is 1080 by 1520 pixels, with each square pixel having area 5511m 2 . An optical reducer was used to couple the camera to the microscope, ensuring that the image size matched the CCD size. Images were recorded using 12-bit resolution, varying exposure times, and 2X2 binning to give images of 540 x 760 pixels. The images covered an area 127 of roughly 0.2 x 0.3 mm on the electrode surface, with each pixel representing an area of 0.148 p.m 2 on the electrode surface. The diffraction-limited resolution is 0.627 p,m for fluorescein emission (514 nm) when using the 50x objective and expresses the smallest separation at which two features can still be resolved. Imaging of the highly curved, reflective surface of the Hg drop presented special challenges as compared to earlier experiments in our laboratory that used a planar Au(111) electrode. A brightfield image of the Hg drop is given in Figure 40. Only the bottom part of the electrode surface is represented by the in-focus region, seen as a central bright disc 0.1 mm in diameter. The actual extent of the Hg drop is greater than the image size; the relative circumference of the spherical drop is shown schematically as the larger black circle. It can be convincingly shown that the electrode does indeed extend beyond the in-focus disc region by shining a flashlight onto the Hg surface at an oblique angle, as shown on the right in Figure 40. The image shows the reflection of the flashlight illumination off the Hg drop and into our collection optics. Given the curved nature of the Hg drop, we observed fluorescence not only from the in-focus region of the drop, but also from the whole imaged region. As the outer edges of the imaged drop could reflect fluorescence from the GIS, analysis of the fluorescence images concentrated on the in-focus region only. Image analysis to output the average fluorescence intensity within the in-focus region was performed using Image Pro Plus (version 4.0, Media Cybernetics, Bethesda, MD) or the DlPimage software toolbox [222] for MATLAB (The Mathworks, Natick, MA). 128 0.9 0.8 0 . 7 2 0 0.6 0.5 0.4 to. 0.3 .c1) 0.2 0.1 0 1 0.9^—Absorption -- Emission 0.8 - 0.7 0.6 0 0.5 0.4 ac • 0.3 0.2 0.1 0 380 400 420 440 460 480 500 520 540 560 580 600 620 Wavelength (nm) Figure 38 Absorption and emission spectra of fluorescein at pH 9. Data for this spectra from Invitrogen-Molecular Probes fluorescein reference standard (catalogue # F-1300). Dashed lines represent the limits of the spectral bandpass regions of the excitation (460-490 nm) and emission (515-550 nm) filters of the Olympus U-MWIBA filter cube. The peak wavelength and shape of the emission spectrum are largely independent of pH because the predominant dianion form at pH > 6.4 remains the dominant contributor to fluorescence even below pH 6.4 [219]. 129 CE Potentiostat & Lock-in Amplifier Computer and DAQ Filter Cube Dichroic mirror 505nm Abs 460-490nm Em 515-550nm Spot RT CCD Figure 39 Spectroelectrochemical cell used for simultaneous fluorescence imaging and capacitance measurement. The optical window is a 0.17 mm thick coverglass. The filter cube shown here is the Olympus U-MWIBA cube suitable for use with fluorescein, but the cubes are mounted on a rotatable turret and could be interchanged if desired. The objective used here was 50x magnification, but 10x, and 20x objectives are interchangeable. An additional 1.5 x magnification can be selected within the microscope if desired. 130 With Added Side Illumination Figure 40 Relationship between imaged portion of Hg electrode, the in-focus region, and the extent of the drop diameter. At left top: a brightfield image of the Hg electrode. The in-focus region is clearly visible. Left bottom: the size and position of the imaged area relative to the full extent of the drop. The circumference of the spherical drop is shown schematically by the black circle. The centre of the in-focus region was taken to be the bottom centre of the drop and the position of the drop's edge inferred from the drop size and the scale of the imaged region. At right, a bright field image of the electrode with added illumination from a flashlight at an oblique angle. The reflected light shows the Hg drop extends beyond the in-focus region. 131 , 5.4 Results and Discussion The DOPC monolayer on Hg has been previously characterized electrochemically; part of this work was to characterize it spectroelectrochemically. An important aspect of this is the question of what happens to the lipid layer upon potential-induced desorption from the electrode surface. Desorption requires substantially negative potential, -1.85 V. Our earlier discussion of the DOPC on Hg system concentrated on its behaviour in the potential range -0.3 to -1.2 V, so we will first consider here its electrochemistry when the potential is swept further negative. A second point is that the incorporation of a fluorescent probe into the monolayer was expected to disrupt the well-ordered nature of the monolayer, and to produce an effect on the phase transitions, resulting in measurable changes in the capacitance vs. potential curve. Accordingly, the influence of the added dye was characterized by electrochemical measurements in the potential ranges -0.3 to -1.2 V and -0.3 to -1.85 V. The 5-octadecylaminofluorescein fluorophore was found to mix with DOPC in a non-ideal manner; it did induce disruption, but the overall characteristics of the DOPC layer were maintained. After the electrochemical characterizations, the results of the in situ fluorescence microscopy of the DOPC monolayer are presented next. The origin of the signal measured with fluorescence and capacitance are significantly different and give complimentary information. The capacitance measurements indicate potential-induced changes at the surface of the Hg electrode, while the fluorescence measurements preferentially characterize the behaviour of the organic layer when it is separated from the electrode. The spectroelectrochemical behaviour is considered both in the potential region to -1.2 V and to desorption. The changes in capacitance and fluorescence measurements together presented a consistent picture of the changes in the interface as potential was varied. Comparison is made to a colleague's earlier results examining octadecanol adsorbed on a gold electrode [179, 203, 204]. It was expected that the behaviour of the DOPC on Hg system would follow closely that observed for octadecanol on gold, but this was determined to be unfounded. 5.4.1 Electrochemistry of DOPC Monolayer Between -0.3 and -1.85 V The measured capacitance of the adsorbed DOPC monolayer in the limited potential range of -0.3 to -1.2 V was presented in the previous chapter. Recall that the potential was swept first negatively, 132 then reversed at -1.2 V to return back to the starting potential. The capacitance when sweeping the potential to desorption (-1.85 V), is likewise begun with increasingly negative potentials. Identical to the limited sweep, there is an initial minimum capacitance of 1.85 p.,F/cm 2 . The first pseudocapacitance peak at -0.95 V is tall and sharp, and its underlying mechanism involves an increase in monolayer permeability. The second peak at -1.05 V is much smaller and represents a defect nucleation and growth process. Beyond -1.2 V, there is a third peak at ca. -1.35 V of intermediate size and more broad than peaks 1 and 2, shown in Figure 41 panel a. This peak has been described in literature as possibly representing an adsorption/desorption process due to its similarity to features seen for insoluble amphiphiles adsorbed on gold electrodes [198, 223]. Further potential excursion negatively results in the capacitance increasing slowly to match that observed for the uncoated Hg electrode in contact with 0.1 M KC1 electrolyte. This equivalence of measured capacitance suggests the lipid layer is desorbed from the electrode surface. It was believed that the desorbed lipid resides in a region near, but not on the electrode surface. On reversal of the potential sweep to the positive-going direction, peak 3 is seen to be shifted almost 0.2 V positive, and has a much broader shape. This hysteresis points to different mechanisms underlying the phase change as the potential is made more negative or more positive. On the positive-going scan, peak 2 is shifted slightly negative in potential and is considerably increased in peak height, while peak 1 is absent. The minimum capacitance region shows a value of about 2.2 .tF/cm 2 , representing a more defective, less well-ordered lipid organization in comparison to the starting condition. 5.4.2 Electrochemistry of the 3 mol% Dye/DOPC Monolayer Between -0.3 and -1.85 V When the 5-octadecylammoniumfluorescein dye was incorporated into the lipid monolayer at a concentration of 3 mol%, the capacitance curve was altered, signifying dye-induced changes in lipid order. Since the goal was to use fluorescence microscopy to characterize the behaviour of the DOPC monolayer itself and not the dye-influenced monolayer, it was desirable for the disruptive effect of the dye to be minimized. It is important that the basic behaviour of the DOPC monolayer under variable potential be retained. Figure 41 panel b shows the recorded capacitance of the 3mol% dye/DOPC monolayer over the full potential range. The initial value of capacitance in the minimum region was about 2.1 p,F/cm 2 , greater than that observed for the DOPC only monolayer. This indicates the inclusion of the fluorescein dye disordered the arrangement of lipid molecules in 133 the film. At -0.75 V a small bump in capacitance was observed. This feature was only present when the 5-octadecylaminofluorescein dye was incorporated and could represent a change in the dipolar orientation of the dye's fluorescein moiety [224]. Further negative, peak 1 is shortened and broadened while peak 2 is broadened. Peak 3 is relatively unchanged, with the shoulder near -1.35 V becoming more prominent. Near -1.5 V, the capacitance differs from that of the monolayer without incorporated dye, and remains higher until -1.65 V, suggesting the structural changes in the layer just before desorption may be energetically different when dye is incorporated. The return potential sweep shows a depression in the capacitance before peak 3 and the peak itself is broadened. As before, peak 2 is considerably taller and sharper than the negative-going scan and peak 1 no longer exists as a clear transition. The recovered minimum capacitance is similar to that observed at the start of the measurement cycle. Overall, the disruptive effect of the fluorescein dye is greater than desired, but the principal features of the DOPC monolayer are retained. If comparing the disruptive effects of the fluorescent dye with those induced by Amphotericin B formulations in the previous chapter, it is important to recognize the dye was integral to the spreading solution deposited at the GIS and was designed to fully intercalate in the monolayer, whereas AmB had to penetrate the lipid layer from the subphase. From this perspective, the disruptive effect is actually quite slight. Experiments were conducted with lower concentrations of incorporated dye to further reduce the disruptive effect, but these dye concentrations did not produce sufficient fluorescence for the in situ fluorescence imaging experiments. 5.4.3 Electrochemistry of 3 mol% Dye/DOPC Monolayer Between -0.3 and -1.2 V The influence of the incorporated fluororphore on the capacitance of the monolayer within the restricted potential range -0.3 to -1.2 V is given in Figure 41 panel c. The initial minimum capacitance is 2.111F/cm 2 , a small increase in capacitance is observed near -0.7 V and peaks 1 and 2 are shortened and broadened on the negative-going potential sweep as noted previously. When the potential sweep is reversed at -1.2 V and swept positively, the measured capacitance curve differs from that observed with a more negative potential limit. Peaks 1 and 2 conserve the same height as on the negative-going scan, with peak 1 narrowed asymmetrically. Nearing the positive potential limit, the minimum capacitance recovers the 2.1 p.F/cm 2 value. 134 5.4.4 Fluorescence Microscopy of 3 mol% Dye/DOPC Monolayer Between -0.3 and -1.85 V As mentioned above, comparison was made between the spectroelectrochemical results of this system and that of a monolayer of octadecanol adsorbed on a planar single-crystal gold electrode. For this reason, a brief summary of the behaviour of the fluorescently-doped octadecanol layer is warranted. No fluorescence was observed from the adsorbed octadecanol layer. Only as the potential was varied to values inducing octadecanol desorption did the fluorescence become measureable. The fluorescence images were reminiscent of a starry night sky, with a moderate fluorescence background punctuated by many small bright spots. For sequential fluorescence images taken at desorption potential, the structure and pattern of fluorescent features was unchanged. The desorption/re-adsorption process displayed significant hysteresis, but once the octadecanol was re-adsorbed, the fluorescence was again quenched. It was expected that the fluorescence of the DOPC monolayer would follow a similar trend, being quenched at adsorption potentials and increasing upon desorption. The results of the combined fluorescence microscopy and capacitance measurements of the fluorophore-doped lipid monolayer over the full potential region are shown in Figure 42. The data in the figure represent one run, but the trends observed are typical of those observed in each run. The exact magnitude of fluorescence changes was variable between runs. The top portion of the figure gives actual fluorescence images at potentials of interest. The exposure time here was 1 s, with a 1.5 s interval between images. The alphabetical labels of these images correspond to the marked potentials shown in the middle panel. The location of the bright in-focus region of the electrode is shown schematically with a white circle. The middle portion of the panel gives the average fluorescence intensity as a function of potential, as calculated from the grey scale of the pixels within the area of the white circle. The lower portion of the figure gives a plot of capacitance versus potential, recorded immediately prior to the spectroelectrochemical experiment but on a different monolayer-coated Hg drop. The curve shown is typical of all those recorded in this potential range for the fluorescently-doped DOPC monolayer. Actual capacitance values measured during the negative-going portion of the spectroelectrochemical experiment (i.e. originating from same monolayer as fluorescence data) are super-imposed as individual data points. 135 b) .,,..............L......0"(20 . • • ' " " A ..... .......... - • ' • . Ir./4 .^.^I^.^.^.^.^I " I^• • I • • • ' I ' • " I ' • x20^.: WafAr1431PAw.s......AMVX : . - • •^I c) 100 N .E 75 LL • 50 0 25 0 100 N'E IL' 50 0 25 0 100 ci E 75 C.) I'L 50 0 25 0 -1.75^-1.50^-1.25^-1.00^-0.75^-0.50^-0.25 E / V vs SCE Figure 41 Capacitance versus applied potential measurements of the Hg electrode in contact with 0.1 M KC1 ( dotted line), coated with DOPC monolayer adsorbed from GAS (panel a) or coated with 3 mol% 5-octadecylaminofluorescein/DOPC monolayer adsorbed from GIS (panels b and c). Panel b gives measurement over the full potential range, while panel c shows the capacitance when the negative potential limit is restricted to -1.2 V. Solid line gives negative-going potential sweep, while dashed line gives positive-going sweep. A series RC circuit was taken to accurately represent the electrical character of the interface, allowing calculation of the interfacial capacitance. 136 F E D B • 24 20 8 16 - c 8 12 - u) a) 8 -0 LL 4 - GI^ \ \^I_...--..■.) / \^,^-.• Kr \ / .... __^LT CB^-......--..:6- ..- °- r_ : AF4,:_ .E.,^---tiD 1 \ .7. 11.1,iiliTiiiii i ii-li 60 "40 0 20 - 0 x20 ■ -1.75^-1.50^-1.25^-1.00^-0.75^-0.50^-0.25 Potential / V vs SCE Figure 42 Spectroelectrochemical results for the 3 mol% 5- octadecylaminofluorescein/DOPC monolayer adsorbed on Hg electrode from GIS. The full potential range was examined, -0.3 to -1.85 V. Top panel: selected fluorescence images, 1 s exposure. White circle outlines the position of the in-focus region of the electrode. Middle panel: average fluorescence intensity calculated over the in-focus region. Alphabetical labels on the data points refer to fluorescence images above. Bottom panel: capacitance measurements made during fluorescence imaging (data points, only data from negative- going potential steps shown) and capacitance curve measured via potential sweep on a different adsorbed monolayer. Solid line: negative-going scan; dashed line: positive-going scan. Full interpretation of the data is given in the text. Reproduced with permission from [4], Copyright 2003 The Royal Society of Chemistry. 137 The fluorescence intensity of the as-deposited layer is low. Through the range -0.4 to -1.75 V, the fluorescence intensity remains at a background level (images/data points A through E of Figure 42). The fluorescence intensity rises slightly beginning at -1.8 V and has doubled upon reaching the most negative potential of -1.85 V (image/data point F). Examination of fluorescence image F shows a diffuse region of moderate brightness located primarily outside the in-focus region, towards the bottom of the recorded image. When the potential is stepped back positively, the observed changes are much more dramatic. The fluorescence intensity increases substantially, reaching a maximum at potential -1.7 V which is —6x more intense than that observed at the negative limit. The fluorescence images at -1.75 V (G) and -1.7 V (H) show that the feature observed previously has extended into the in-focus region of the Hg drop image. Image H in particular shows very intense fluorescence appearing as a near vertical band in the recorded image. At these very negative potentials, correspondence of measured capacitance in the presence and absence of the DOPC layer was previously suggested to imply desorption of the lipid layer from the electrode surface. This hypothesis is confirmed here by the fluorescence measurements. The large increase in fluorescence intensity at very negative potentials must be the result of decreased fluorescence quenching as desorption increases the separation between the lipid layer (including fluorophore) and the metal surface. The majority of runs on this system showed that the maximum of fluorescence was observed to originate near the in-focus part of the electrode surface (overlayed circle) and move towards the bottom of the image. The optical arrangement is such that the drop was observed from below, so movement towards any edge of the image is a movement upwards toward the GAS interface. We hypothesized that this movement could be related to the positioning of the counter electrode and thus be related to the electric field distribution, but reorientation of the counter electrode did not change the direction of observed movement. In addition, the illumination of the interface was not uniform across the image and was more intense for the bottom half of the image. The images show only diffuse features with dimensions in the tens of i_tm or more. No small features are observed. Often images collected showed similar broad features shifted in position relative to the prior image. The desorbed lipid layer appears very fluid. This is in contrast to results 138 observed for the octadecanol/Au system. In that case, a significant increase in fluorescence was also observed at desorption potentials, but the character of the desorbed film was quite different. The images observed were very static, showing the same features throughout the potential range of desorption and re-adsorption. This static nature allowed analysis of feature size and distribution; images with up to 800 individual features of size -20 p.m 2 were typically found. No such analysis was attempted for the DOPC/Hg system since the results were reasonably expected to be ill-defined given the dynamics of the fluorescent features. A plausible explanation for the fluidity of the surface is convective flow due to surface tension gradients (the Marangoni effect). If differences in surface tension exist on a liquid surface, the liquid will flow toward the region of higher surface tension. Changes in surface tension at the Hg surface are to be expected as the potential is changed (Cf. Figure 9). The observed motion might be holdover of flow induced by the preceding potential step. Another possible source of the gradient is non-uniform organic coverage of the Hg surface. Surface tension is decreased as the surface concentration of lipid increases; regions of lower coverage would then have higher surface tension, producing a gradient. This type of effect has been examined in relationship to higher than expected currents in polarography. The so-called polarographic maximum of the third kind is related to the Marangoni effect and has been discussed extensively in Soviet electrochemical literature. An accessible paper is [225]. The trend of fluorescence features moving towards the bottom of the image is probably also related to the Marangoni effect. As the potential is stepped further positive, the observed fluorescence intensity decreases, reaching a minimum between -1.4 and -1.3 V which almost reaches the background intensity. Simultaneously, the capacitance shows a slight lowering just negative of peak 3. At the desorption potential, the lipid layer does not diffuse away, but remains near the electrode surface, so upon reversal of the potential the lipid moves closer to the electrode surface to form a pre-adsorbed state, effectively quenching the fluorescence. This type of behaviour was also noted in the octadecanol/Au system. This pre-adsorbed state undergoes a phase transition (peak 3) to form a lower capacitance layer. 139 Unexpectedly, beginning at peak 3 the fluorescence intensity increased again and persisted broadly as the potential was made more positive. The maximum occurs at -1.1 V, but is broad relative to the potential spacing of fluorescence images, so this must be taken as an approximate potential of maximum fluorescence. The corresponding fluorescence image (I) shows relatively homogenous fluorescence intensity extending over most of the image area; darker regions are observed at top left and bottom right. The next image in the series, J (-1.0 V) shows a roughly similar pattern but shifted towards the bottom of the image. The increase in fluorescence beginning at peak 3 is surprising since it deviates from the behaviour seen in the octadecanol/Au model. This suggests that the process of reforming the DOPC monolayer requires reorganization of the lipid such that some of the lipid/dye moves far enough away from the electrode surface for fluorescence quenching to be lessened and a signal observed. More positive than peak 2 the fluorescence intensity decreases slightly with each additional potential step. At the potential of -0.4 V, the fluorescence intensity observed for the reformed layer is low, but remains higher than the as-deposited layer at the same potential. This, along with capacitance measurements that show the minimum capacitance is slightly higher for the reformed layer, means that the reformed layer differs in organization from the as-deposited layer, and some lipid/dye must exist outside the defective monolayer. This may be especially noted in the fluorescence images since the chromophoric portion of the dye is known to reside in the lipid headgroup region that is the portion of the monolayer which exists furthest from the electrode surface. Given the inverse cubic dependance of quenching on separation distance, it is possible that fluorescence from lipid/dye that is excluded from the monolayer is highly sensitive to changes in the lipid layer as the phase transitions are traversed. The measured capacitance at each potential step follows that observed during the analogous capacitance measured using the potential sweep method. Some discrepancy is seen in the minimum capacitance region where the values recorded when potential stepping are lower than that measured with the potential sweep for the negative-going scan. These differences are quite small. 140 5.4.5 Fluorescence Microscopy of 3 mol% Dye/DOPC Monolayer Between -0.3 and -1.2 V Reformation of a defect-free lipid layer was possible if the extent of applied negative potential did not exceed -1.2 V. The capacitance, fluorescence images and average fluorescence intensity of the image's in-focus region are shown in Figure 43 for this limited potential region. The image exposure time was 15 s, considerably longer than those used before since the fluorescence intensity observed in this region was low due to fluorescence quenching. By way of comparison, no measureable fluorescence intensity was recorded from the octadecanol/Au system when the monolayer was adsorbed on the gold. The major difference between these two metals is not their respective optical constants in the visible wavelengths, but the scale of surface roughness. The mercury electrode, being liquid, is characterized by a smooth surface on the atomic scale, while the gold electrode was a hand-polished planar single crystal with p.m scale roughness. The smoothness of the Hg surface results in reduced fluorescence quenching, a fortuitous effect here as it allows in situ fluorescence characterization of potential-induced changes in the lipid monolayer over a wide potential range. Unexpectedly, the fluorescence in the minimum capacitance region was larger than the fluorescence negative of the phase transitions. The fluorescence decreased smoothly as the potential was scanned negatively from the minimum capacitance region. This behaviour was in complete contrast to what was expected given the known results from the Au/octadecanol system, where fluorescence increased negative of the phase transition in the octadecanol layer. The decrease in fluorescence in the DOPC/Hg system occurs over the same region of applied potential that is known to result in the formation of defects in the monolayer. When the potential is stepped back positively, the minimum capacitance returns to a value similar to that of the as- deposited layer and the fluorescence intensity increases back to near its original intensity. Since the fluorescein moiety of the fluorophore resides in the lipid headgroups, it is possible that the formation of defects allows the dye to interact with the Hg surface more freely than if covered by the defect- free lipid monolayer, thereby decreasing the observed fluorescence due to more efficient quenching. The reformation of a well-ordered, compact low-capacitance lipid film would tend to 'squeeze out' 141 the dye and thus a greater fluorescence would be measured because the dye would be further from the metal surface. This is the same kind of mechanism invoked in the previous chapter to explain why a well-ordered defect-free monolayer could be reformed after interaction with deoxycholate or the Amphotericin B formulation ABLC. The fluorescence images collected throughout the minimum capacitance region show large diffuse features. In some images (e.g. A in Figure 43), distinct localized regions of high fluorescence are observed outside of the in-focus region. These are thought to represent reflections of small regions of concentrated fluorophore at the GIS, described before as being due to either slight dye segregation out of the lipid matrix or regions of excess lipid as needed to maintain the ESP. Such features were always observed towards the edge of the fluorescence images and never in the in-focus region. The average fluorescence intensities recorded in the experiments of this chapter give the average over the in-focus region only. 5.4.6 Fluorescence Microscopy of DOPC Monolayer with AmB in Electrolyte In situ fluorescence measurements of the DOPC monolayer identified that the measured intensity is sensitive to the position of the fluorophore in the layer. Potential-induced defect formation allowed the dye to move closer to the Hg surface and be more efficiently quenched, while annealing those potential-induced defects resulted in the dye being forced further from the electrode where it fluoresced more intensely. In order to couple this newly developed tool of in situ fluorescence microscopy with our interest in AmB-lipid interaction described in the previous chapter, spectroelectrochemical experiments were conducted with AmB formulation added to the electrolyte. Monolayers of 3 mol% 5-octadecylaminofluorescein/DOPC were formed and deposited on the Hg electrode, while FZ was added to the electrolyte at a concentration equivalent to 100% TD (0.54 .tm AmB). The potential was stepped through the -0.4 to -1.2 V range and simultaneous capacitance and fluorescence measured. ? presents the fluorescence intensity as a function of potential, showing a reversible decrease in intensity in the region of the phase transitions. That is to say, the same pattern as that observed in Figure 43 for the AmB-free measurement, suggesting the dominant 142 x20 •■^■ I^ i F ^H E ^ D ^ C ^ B 18 60 ED_0= Q 0 160ca)(.) (") 14 LI 12 qi E 0 40Li_t. 0 20 0 -1.25^-1.00^-0.75^-0.50^-0.25 Potential / V vs SCE Figure 43 Spectroelectrochemical results for the 3 mol% 5- octadecylaminofluorescein/DOPC monolayer adsorbed on Hg electrode from GIS. The potential range was limited to -0.3 to -1.2 V. Top: selected fluorescence images, 15 s exposure. White circle outlines the position of the in-focus region of the electrode. Middle panel: average fluorescence intensity calculated over the in-focus region. Alphabetical labels on the data points refer to fluorescence images above. Bottom panel: capacitance measurements made during fluorescence imaging (data points, only data from negative- going potential steps shown) and capacitance curve measured via potential sweep on a different adsorbed monolayer. Solid line: negative-going scan; dashed line: positive-going scan. Full interpretation of the data is given in the text. Reproduced with permission from [4], Copyright 2003 The Royal Society of Chemistry. 143 G ^ H ^ L F • 15 • 14 C) 13 8 • 12 a) o 11 LT. 10 60 (E) 40 0 20 0 E D B x20 -1.25^-1.00^-0.75^-0.50^-0.25 Potential / V vs SCE Figure 44 Spectroelectrochemical results for the 3 mol% 5- octadecylaminofluorescein/DOPC monolayer adsorbed on Hg electrode from GIS in the presence of 0.54 p,M AmB as FZ. The potential range was limited to -0.3 to -1.2 V. Top panel: selected fluorescence images, 12 s exposure. White circle outlines the position of the in-focus region of the electrode. Middle panel: average fluorescence intensity calculated over the in-focus region. Alphabetical labels on the data points refer to fluorescence images above. Bottom panel: capacitance measurements made during fluorescence imaging (data points, only data from negative-going potential steps shown) and capacitance curve measured via potential sweep on a different adsorbed monolayer. Solid line: negative-going scan; dashed line: positive-going scan. 144 influence on the fluorophore is potential rather than the incorporated antifungal drug. It was postulated that changes in the fluorescence vs. potential curve due to AmB might be observable using a different fluorophore that resides in a different position in the layer (e.g. in the lipid tails) or a fluorophore that behaved differently upon interaction with the potential-induced defects. Accordingly, fluorescence spectroelectrochemistry was undertaken with the fluorophores DPH (Invitrogen-Molecular Probes catalogue # D-202), lipid-tagged DPH (catalogue # D-476), a dioctadecyl carbocyanine dye (catalogue # D-307), or a dipyrrometheneboron difluoride-tagged lipid (`BODIPY-tagged', catalogue # D-3800). DPH and lipid-tagged DPH are known to intercalate into the alkyl chains, while the other two dyes label the headgroup region. In particular the DPH-based fluorophores were thought to be suitable since AmB is known to produce a quenching effect on DPH fluorescence [226]. Experiments were first conducted in the absence of AmB to benchmark the fluorescence vs. potential measurements. Experiments with DPH and the carbocyanine dye gave no meaningful fluorescence signal whatsoever, possibly due to efficient quenching by the metal electrode. Measurements with lipid-tagged DPH and BODIPY-tagged lipid in the absence of AmB showed behaviour similar to the fluorescein probe. That is, in the potential range -0.4 to -1.2 V, the fluorescence was initially high, decreased through the phase transition region and returned to a high value when the potential was stepped back to the positive limit. While this served to support our earlier rationalization of the phase transition process allowing the fluorescein dye to move close to the electrode, the trend was unchanged upon AmB addition, and the use of in situ fluorescence microscopy to follow AmB-induced changes in the adsorbed lipid layer was discontinued. 5.5 Conclusions Characterization of the potential-induced changes of a lipid-coated Hg interface was accomplished using electrochemistry and in situ fluorescence microscopy. We examined the lipid behaviour of the monolayer within two different potential windows. Between -0.3 and -1.2 V, we characterized the phase transition and minimum capacitance regions, while between -0.3 and -1.85 V we characterized the process of desorption of the lipid film from the electrode surface. In the limited potential window, the monolayer adsorbed from the GIS was characterized by a strong decrease in fluorescence as defects were induced in the monolayer, explained as a change in 145 structure of the adsorbed lipid layer such that the fluorophore moved closer to the electrode surface. This movement towards the electrode produced more efficient fluorescence quenching and thus lower intensity. Reformation of the layer by reversing the potential returned the more intense fluorescent signal. The very ability to measure such fluorescence behaviour from the layer adsorbed on the Hg electrode was rationalized as being the result of the smooth Hg surface; other rougher electrode surfaces have shown highly efficient quenching at every potential for which the organic layer is adsorbed. Similarity between the DOPC/Hg system and the octadecanol/Au system was noted, occurring principally in the potential region near desorption. In both cases, the capacitance increased upon desorption and the fluorescence intensity increased significantly. Re-adsorption of the organic layer to the electrode surface showed significant hysteresis in the fluorescence signal. The so-called pre-adsorbed state preceding peak 3 on the positive-going scan gave a weak fluorescence signal, indicating the pre-adsorbed state of the monolayer must be close to the electrode. The fluorescence was more intense for the reformed layer than the originally-deposited layer suggesting a more defective re-adsorbed lipid layer. This was supported by electrochemical observation of higher capacitance in the minimum capacitance region. We examined the changes in fluorescence vs. potential profile upon addition of AmB formulation to the electrolyte. Interaction of AmB formulation with the fluorophore-doped monolayer produced no discernable changes in fluorescence attributable to AmB influence despite our use of several different fluorophores. Those fluorophores which gave measurable fluorescence all showed the same pattern of intensity vs. potential, suggesting that from the point of view of fluorophore-metal separation, the potential-induced changes in lipid organization within the monolayer dominate over changes due to incorporated AmB. 146 6 FLUORESCENCE OF AmB IN FZ AND HTFZ 6.1 Introduction The intrinsic fluorescence of AmB formulated as FZ and HTFZ will be described. In 2003, Gruzsecki et al. [11] determined that highly purified AmB is weakly fluorescent. The fluorescent properties of AmB suggests new possibilities for further studies of its mechanism of action since many well-founded fluorescence techniques exist to study drug-cell and drug-model membrane interactions [227, 228]. Our desire was to exploit the intrinsic fluorescence of AmB to study its interaction with model cell membranes, but since we are examining AmB formulations rather than the purified form, we first examined its fluorescence ex situ. ABLC, being a lipid-complex of the drug, is a turbid solution and light-scattering prevents meaningful measurements. Gruszecki identified differing fluorescence excitation and emission bands that depended on the state of AmB (aggregated or monomeric). We measured fluorescence from FZ and HTFZ preparations using added sodium dodecyl sulfate (SDS) or deoxycholate to control the aggregation state. In the course of our experiments, we observed unusual fluorescence emission not reported by Gruszecki. This emission was found to originate from a higher energy level than the fluorescence observed by Gruszecki. Our primary objective was to study fluorescence from AmB as FZ, with aggregation state varying between that of FZ as-reconstituted and fully monomeric. For comparison, the fluorescence of HTFZ was measured, but based on the strong similarity of their absorption spectra, we expected similar fluorescence. Additionally, the use of SDS to break-up AmB aggregates allowed comparison of the stability of FZ and HTFZ towards disaggregation. The comparison of their spectra and stability is of interest because HTFZ is touted as a possible new formulation of AmB. Little is known about the differences between FZ and HTFZ; the most significant difference is that HTFZ and FZ differ in aggregate size [10, 44, 229]. Since AmB fluorescence was shown to be sensitive to aggregation, it was thought appropriate to analyze both FZ and HTFZ. 147 To reach these objectives, a suitable AmB concentration for the experiments was first defined. We chose a concentration that was relatively high from a spectroscopic standpoint, and relatively low in comparison to AmB's clinical use. Without added SDS, the concentration used has most AmB in aggregated form. Previous spectroscopic study of AmB and heat-treated AmB has suggested that the composition of the samples (monomer vs. aggregate) may change slightly over the time span of 20 minutes [10]. To complete spectroscopic measurements on our many samples within such a short time would be impossible. Instead, the experiments were completed as fast as possible, with at least one hour elapsed before the first measurement. This allows the steady-state behaviour of the samples to be measured. Full details of the protocols to ensure sample robustness are given in the materials and methods section. Portions of this chapter have been published in an article in the journal Langmuir, and are reproduced in part with permission from [5], Copyright 2007 American Chemical Society. 6.2 Literature Review The work of this chapter is significantly different from that described in earlier chapters and thus the extent of literature to be reviewed is substantial. Principally, the important points to be covered are studies on AmB aggregation state, the use of surfactant to control aggregation, and the spectroscopy of AmB. 6.2.1 AmB Aggregation The aggregation state of AmB is a key factor in determining both the drug's toxicity and its spectroscopy. AmB's sterol-binding selectivity depends on its aggregation state [58, 65-67, 230]. AmB may exist as monomer, dimer, aggregate or super-aggregate [10, 11], but the dimer form has not been isolated. The monomer is thought to be very active towards ergosterol-containing membranes [67, 231], while the aggregated form acts on both sterol-free and ergosterol- or cholesterol-containing membranes [58, 232]. The equilibrium between aggregation states is sensitive to concentration, solvent, temperature, ionic strength and ionic character [233, 234]. Interest in the relation between AmB aggregation and toxicity has been renewed by studies on HTFZ. Numerous theoretical studies and calculations based on spectra have been carried out on the 148 structure of AmB aggregates, either as pure AmB in water or as aqueous FZ. AmB dimer is a commonly recurring motif [11, 72, 235-238]. AmB dimer might also be considered as a building- block for larger aggregates. 6.2.2 Surfactant Control of Aggregation State As discussed above, interest in AmB's aggregation state has been primarily due to its influence on pharmacological behaviour, but similar questions of molecular aggregation are often considered in relation to dyes for photographic emulsions, photovoltaic and photosynthesis processes [239, 240]. To examine these questions, surfactants are commonly used to control the aggregation state while the system is probed with fluorescence and/or absorbance measurements [e.g. 241, 242]. The separate hydrophobic and hydrophilic regions of surfactants often leads to well-known self- association behaviour such as micelle formation at the CMC. The anionic surfactant SDS is widely used in dye aggregation studies as well as being fully characterized by its widespread use to influence protein structure. Dye-surfactant interactions are generally governed by the interplay of hydrophobic and electrostatic contributions. Gruda and co-workers used light-scattering and absorbance measurements to characterize AmB aggregation in the presence of sodium deoxycholate, Triton X-100, or lauryl sucrose in the course of studying the effect of aggregation state on sterol-binding and efficacy [9, 66, 230, 243]. With the concentration of lauryl sucrose below the CMC, a potentiative effect on AmB fungal cell toxicity was observed, while above the CMC, when AmB was monomerized in lauryl sucrose micelles, AmB action on fungal cells was hindered [230]. The surfactant concentration which gives the maximal selectivity to ergosterol vs. cholesterol-binding is that of the CMC, at which monomerization of the AmB is induced [9]. Lamy-Freund used static and dynamic light-scattering experiments to study AmB-deoxycholate aggregates and their stability at different deoxycholate concentrations [233], while Shervani has examined the self-association of AmB in the presence of non-ionic surfactant Triton X-100 [244]. The deoxycholate concentration required to monomerize AmB is large, while Triton X-100 readily monomerizes the drug but is too toxic for in vivo use. 149 6.2.3 Spectroscopy of AmB The absorbance spectra of AmB, both pure and as FZ, are well known. For pure AmB in water, the aggregated form gives an absorption maximum at 339 nm, while the monomeric form's maximum is at —410 nm. The absorption of AmB monomer is characterized by vibrational fine structure peaks at 385 and 365 nm, as shown in Figure 45 (results of Gaboriau et al. [10]). The presence of this fine structure is considered diagnostic of AmB monomer. For FZ, the presence of deoxycholate shifts the aggregate maximum to —328 nm, while the monomer absorption is unchanged [39, 235]. Both aggregated and monomeric forms absorb light strongly, seen by their large extinction coefficients (> 80,000 M -1 cm-1 ). The heat-treatment of AmB solutions induces a blue-shift in absorption. For example, pure aggregated AmB in water has an absorption maximum of 322 nm after heating to 70 °C for 20 minutes. HTFZ has a similar absorption maximum; Baas et al. noted it at 320 nm [232], although it shifts slightly with concentration. The heated AmB formulations are also strongly absorptive. Circular dichroism spectroscopy has also been applied to the study of AmB aggregation [e.g. 235, 245, 246]. Monomeric AmB shows no differential absorption of right and left-circularly polarized light, but aggregated AmB does, giving an intense doublet. Unfortunately, interpretation is not straight-forward, with some authors [247] suggesting that the intense doublet is due to AmB dimer and others [235, 248] asserting AmB dimers have no differential absorption and that the doublet is due to larger aggregates. Studies of AmB fluorescence are limited. Bittman and Fischkoff [249] used fluorescence to measure binding of a variety of polyene antibiotics to cholesterol and cholesterol-derivatives in lipid liposomes. Excitation of AmB was made at 340 nm, but no spectra were published. In 1979, Khmel'nitskii and Cherenkevich recorded fluorescence spectra for a series of heptaene antibiotics in liposomes, and examined the changes in spectra on binding to cholesterol [250]. They observed little change of spectra upon binding. This paper appears to have been largely overlooked subsequently. Two years later, Petersen et al. [251] measured fluorescence from AmB samples which had been purified by high pressure liquid chromatography (HPLC), finding that after removal of a possible tetraenic fluorescent contaminant, purified AmB did not fluoresce. Their data does 150 suggest removal of a fluorescent contaminant, but their assertion that AmB does not fluoresce is weak. They plot absorbance at 408 nm and fluorescence emission intensity (excitation 335 nm, emission measured at 465 nm) as a function of retention time, finding the fraction with maximum absorbance is different from that of maximum fluorescence. Although this points to a fluorescent contaminant, it does not definitively mean AmB is non-fluorescent. In 2003, Gruszecki et al. [11], unaware of Petersen's work, studied the fluorescence of HPLC-purified AmB, finding it to be weakly fluorescent. Additionally, they found that AmB monomer and aggregates (identified as dimers) have different fluorescence excitation and emission spectra. The monomers were excited at 408 nm and fluoresced between 500-650 nm, with the peak intensity at 563 nm. Dimers were excited at 350 nm and fluoresced between 400 and 550 nm, with maximum at 472 nm. Excitation of 'higher aggregates' has been suggested to occur at 325 nm [252], in agreement with UV absorbance of purified AmB. The quantum yield of monomers was determined to be 0.06% and that of the dimers 4.47%. Bolard [248] has disputed the assignment of dimer fluorescence, arguing that it originates from a contaminant. If Bolard' s assertion is to be correct, the contaminant must display nearly identical aggregation behaviour to AmB when the solvent system or surfactant concentration is changed, and ' dimer' fluorescence thus can still be exploited as a probe of AmB regardless of its true identity. The assignment of AmB dimer as a fluorescent species was made based on two principal arguments: the presence of fine vibrational structure, and a van't Hoff plot of aggregate to monomer fluorescence intensity. We note the van't Hoff analysis fits trimer or tetramer models nearly as well as dimer, and the presence of fine structure is indicative of the fluorescent species being a small aggregate, not specifically indicative of dimer. Accordingly, the assignment of dimer fluorescence may be uncertain, though it seems likely that the fluorescence originates from small AmB aggregates. For clarity, we will retain here the usage of the term dimer, with the understanding that `various low-N aggregates' might be a more accurate term. The differences in absorption and fluorescence spectra between monomeric, dimeric or higher- aggregates of AmB may be rationalized via exciton theory [111, 112]. van Amerongen's book provides an accessible introduction to the mathematics of molecular exciton theory [253]. Briefly, 151 different packing patterns of monomers within aggregates leads to a red- or blue-shift of spectral features relative to the monomer alone. One model which accounts for the spectral blue-shift of AmB aggregates is card-pack aggregation; helical structures have also been proposed in agreement with the blue-shift [237, 254]. The helical model is attractive because it results in the hydrophobic conjugated portion of the AmB molecules being shielded from aqueous solution. Additionally, the helical model seems intuitively more able to form a membrane pore structure than the card-pack (linear) aggregate. On the other hand, the card-pack model is easier to conceptualize. In dimer card- pack aggregation, parallel orientation of AmB electronic transition dipole moment in a direction perpendicular to the plane of aggregation results in splitting of the excited state energy level. Absorption into these two excited state energy levels is determined by how the transition dipole moment of the two monomers interacts. The transition dipole for a linear polyene such as AmB is modelled to lie approximately 13° off the polyene axis [255]. The transition probability depends on the square of the vector sum of the dipoles; attraction between out-of-phase dipoles lowers the energy level but results in a forbidden transition. The characteristic blue-shifted absorption of aggregated AmB is due to in-phase dipolar arrangement, resulting in an allowed transition to the high-lying exciton state. The large Stokes shift measured for AmB dimer fluorescence (> 100 nm) shows that the energy-levels involved in excitation and emission are different. Excitation of AmB dimer is an allowed-transition into the high-lying exciton state, while emission results from the low- lying exciton state, as shown in Figure 46. For a greater extent of aggregation, the energy-level diagram becomes more complicated. Assuming strictly card-pack aggregation (i.e. a linear aggregate), the splitting between the exciton bands enlarges with increasing number of molecules in the aggregate, doubling in separation from dimer to infinite aggregate. As before, absorption into the highest-lying exciton state is allowed, while those levels below are less-allowed according to the dipole sum. If the aggregation extends in two or three dimensions instead of only one (e.g. mixed card-pack and head-to-tail aggregation) then the energy-level diagram becomes very complicated. Additionally, any packing disorder within aggregates will further increase the complexity of the diagram, and variation in the size of aggregates within a sample will result in overlap of the energy-levels, both leading to broad, 152 continuum-type features in measured spectra. In Figure 46, we show the uncertainty of aggregate energy levels as a continuum of states, illustrated as a shaded box. The properties of excited states of linear polyenes is a significant field itself, since carotenoids vital to photosynthesis fall into this category. Despite much study, fundamental questions remain, including determination of the energies of low-lying excited states, and understanding how conjugated chain length or substitution vary the energy-levels [256]. Carotenoids typically have -10 conjugated double bonds, but shorter, simpler molecules such as 2,4,6,8,10,12-hexadecaheptaene are often used as models. For all of these, a standard energy-level diagram is used to rationalize the principal spectral features, and is based on the idealized C2h point group (i.e. having symmetry about 180° rotation and a horizontal mirror plane). The ground state, S o, is assigned Ag symmetry, as is the lower of two excited states, S. The higher excited state, S2, has B u symmetry, as shown in Figure 46. Although neither AmB nor most carotenoids have true C2h symmetry, it has been found that their spectroscopic characteristics are very similar to unsubstituted polyenes which do have C2h symmetry [256]. Accordingly, the symmetries of the electronic states are given by the C2h set of irreducible representations. Since a transition between states of the same symmetry is forbidden, the S o to S I transition is not observed. On the other hand, the S o to S2 transition involves states of differing symmetry and is allowed. This transition gives rise to AmB's absorption spectrum. Fluorescence has been observed to originate from either the S I or S2 states of linear polyenes [257]. The energy gap between the two excited states increases with longer polyene conjugation length, resulting in shorter polyenes fluorescing from the 2'A g level and longer polyenes fluorescing from 1 'B u due to decreased internal conversion [258, 259]. Most fluorophores emit from the lowest excited state that has the same multiplicity as the ground state (most often S I -S 0), known as Kasha's rule. S 2-S o fluorescence, an exception to Kasha's rule, may occur when there is a large difference between S 2 and S I energy levels. Some polyenes exhibit both S I -S o and S 2 - S o emission, termed dual fluorescence. Whether a polyene emits from S 2 or S I depends on the rate of internal conversion between S 2 and S I , which is very sensitive to the energy difference between the states and thus sensitive to chemical structure of the molecule. For example, the sole structural difference between 153 (1.mor i .cm -1 ) 120000 100000 80000 60000 40000 20000 0 300^350^400 ^ 450 WAVELENGTH (rim) Figure 45 UV-vis spectra of pure AmB in different aggregation states in water. Dot-dash line is 10 -7 M AmB, corresponding to the monomeric state. Absorption maximum is 409 nm, with characteristic vibrational fine structure peaks at 385 and 365 nm. Solid line is 10 -4 M AmB, representing AmB in predominately aggregated form. The absorption maximum is a broad peak at 339 nm. Dashed line gives absorption spectrum of a 10 -4 M AmB solution that was heated to 70 °C for 20 minutes. The absorption maximum is blue- shifted to 322 nm. Reprinted with permission from [10], Copyright 1997, Elsevier. 154 Monomer 1 1 13,..r.... 2 1 A g A F F 1'A Dipole Dimer stacking Aggregates tiN A F 1,3,5,7,9,11,13-tetradecaheptaene and 2,4,6,8,10,12-hexadecaheptaene lies in the terminal methyls on the latter, but the tetradecaheptaene has dual fluorescence with S 2-S o bands at 430, 440 and 460 nm, while 2,4,6,8,10,12-hexadecaheptaene shows only S I - S o fluorescence [259]. 6.3 Materials and Methods The fluorescence and absorbance of samples of FZ or HTFZ, both with varying concentration of SDS or deoxycholate were studied. The sample preparation and instrumentation used were both relatively simple. 6.3.1 Sample Preparation A stock solution of AmB (1.45 x10' M) in the form of commercial micellar dispersion FZ was prepared in Milli-Q water. 2.50 ml aliquots of stock solution were diluted to 7.00 ml volume with Figure 46 Energy level diagram for linear polyenes such as AmB, idealized as having C2h point group. Absorption process is labelled as 'A', while fluorescence is shown as 'F'. Possible internal conversion is not shown. The ground state, S o , is denoted with its symmetry as 1 1 Ag . Similarly, the S, and S2 states are given as 2'A g and 1 'B u . The possibility of the dimer or aggregate having non-zero displacement energy is not considered. Reproduced with permission from [5], Copyright 2007 American Chemical Society. 155 Milli-Q water to form twelve FZ working solutions. Similarly, twelve working solutions of HTFZ were prepared after heating half the FZ stock solution at 70°C for 20 minutes. This ensured the AmB concentration in the FZ and HTFZ stock solutions was identical. To eleven of each of these sets of solutions was added SDS or deoxycholate, with one solution remaining free of added surfactant. The added volume of 0.6635 M SDS (Sigma, St. Louis, MO; catalogue# L3771) or 0.5672 M deoxycholate (Sigma catalogue #D5670) stock solution varied between 0 and 109 Ill volume, resulting in maximum dilution of the AmB solution by 1.6%. A solution analogous to FZ was prepared with the antifungal nystatin (Sigma catalogue #N3503), deoxycholate, sodium phosphate (Fisher Scientific Canada, Ottawa, ON; ACS grade), and Milli-Q water. All solutions were made the same day as measurement and were protected from exposure to light. At least one hour elapsed between each sample's preparation and measurement. To further limit possible measurement bias due to differences in standing time, measurements were first made on every second sample of increasing surfactant concentration, then the remaining samples measured. 6.3.2 Instrumentation Fluorescence spectra were collected on an IS S K2 fluorometer (Champaign, IL, USA) equipped with 300 W Xe arc lamp and 0.1 mm entrance and exit slits to give 8 nm excitation and emission bandwidth. The fluorescence intensity was taken as proportional to the photomultiplier tube current output averaged over a 2.5 s interval, and was measured every 1 nm. The fluorescence intensity was normalized for variations in excitation intensity by using a concentrated rhodamine 6G quantum counter. Samples were held in a quartz (Starna, Atascadero, CA) dual path-length cuvette (2 & 10 mm) with the long axis perpendicular to the excitation beam. UV-vis absorption spectra were collected on a Varian Cary 4000 spectrometer (Palo Alto, CA), using the 2 mm pathlength of the cuvette. Operating conditions used were 2 nm bandwidth, absorbance measurement every 0.5 nm, integration time of 0.1 s at each data point and scan rate of 300 nm/min. The Cary 4000 light source was switched automatically from tungsten halogen (visible) to deuterium arc lamp (UV) at 345 nm. Recorded fluorescence intensities and absorbances were corrected for the dilution effect ofthe added surfactant solution. Timer filter effect corrections for absorption of both excitation and emitted radiation were made to fluorescence intensities using the formula of Lakowicz [113]: 156 F o„ = F „10R Aex +A—V21  (45) Where F or, is the fluorescence corrected for the inner filter effect, Fobs the observed intensity, A ex the absorbance at the excitation wavelength and Aen, the absorbance at the emission wavelength. The absorbance values used in the correction took into account the different path length for excitation and emission. Generally, correction due to absorbance at the excitation wavelength dominated. 6.4 Results and Discussion 6.4.1 Introduction The results of absorbance and fluorescence measurements of AmB as FZ or HTFZ in different aggregation states are presented. Our findings provide the groundwork for future fluorescence- based characterization of FZ or HTFZ interaction with cellular or model membranes. The results show that fluorescence from formulated AmB follows that of purified AmB, and confirm Gruszecki's assertion that monomer and dimer states have different fluorescence spectra [11]. Predictably, increasing surfactant concentration acts to monomerize AmB. For AmB monomer, the maximum fluorescence intensity corresponds with the highest surfactant concentration. Interestingly, however, the maximum fluorescence intensity for higher-aggregates of AmB does not correspond to the most aggregated state. Additionally, unexpected differences in the fluorescence emission of FZ and HTFZ were observed, which point to subtle differences in the chemical environment of AmB within aggregates and superaggregates. We begin by examining the results of the standard method of characterizing AmB aggregation, UV-visible absorbance spectroscopy. 6.4.2 UV-Visible Absorbance Spectroscopy UV-visible absorption spectra for solutions of 5.16 x10 -5 M AmB as FZ with varying SDS concentration are presented in Figure 47. SDS concentrations are expressed with respect to AmB and vary between 0 and 209 mole SDS/mole AmB. The UV-Vis absorbance of FZ in the absence of added SDS agrees well with that given in literature [235]. The principal absorption is at 326 nm, 157 with less intense bands at 292, 363, 386, 406 and 418 nm. In the AmB literature, there are two cases for which absorption peaks near 326 nm are observed: aggregated AmB complexed with deoxycholate [260] and super-aggregates formed by heat-treating AmB alone [10]. Pure AmB aggregates absorb at 340 nm; deoxycholate is thought to interact with AmB aggregates such that the monomers are moved closer to each other [72], blue-shifting the absorbance to 326 nm. The FZ absorption spectrum thus corresponds to AmB in the aggregated state. As expected, addition of SDS shifts the main absorption from 326 nm of aggregated AmB to the 410 nm band diagnostic of monomeric AmB [10] along with appearance of vibrational fine structure peaks at 388, 367 and 350 nm. UV-visible absorbance spectrum of HTFZ, shown in Figure 48 is nearly identical and undergoes similar changes upon addition of SDS, although subtle differences do exist. Formation of super- aggregates does not lead to further blue-shift of absorbance. This similarity between the FZ and HTFZ absorption spectra suggests that the deoxycholate present in FZ acts to maximize the extent of exciton coupling that causes blue-shift. 6.4.3 Fluorescence Spectra Figure 49 shows the fluorescence excitation and emission spectra for AmB as FZ. In panel A, monomer excitation at 408 nm shows weak fluorescence with emission maximum at 562 nm. This large energy difference between absorption and emission is typical of linearly conjugated polyenes and represents emission from an electronic energy level below that involved in the excitation. For AmB, emission is from the 2'A g state, which is lower energy than the 1 'B,, state to which absorption occurs [258]. The excitation spectrum (Em 560 nm) has three strong bands between 324 and 356 nm. Panel B gives corresponding spectra under conditions of dimer excitation and emission. When excited at 350 nm, FZ has broad fluorescence emission with maximum at 473 nm. The excitation spectrum (Em 471 nm) again has three strong bands between 324 and 356 nm. The fluorescence excitation and emission spectra of FZ are roughly similar to those observed for purified AmB [11], but with some differences. Our monomer emission spectrum agrees with Gruszecki's, with the (0- 0), (0-1), and (0-2) transitions at 528, 562 and 608 nm respectively. The fluorescence excitation spectrum measured at 560 nm differs. We record weak (0-0) and (0-1) transitions at 410 and 386 158 nm, but Gruszecki's (0-2) transition at 371 nm is absent. This presumably weak fluorescence peak may simply be hidden by more intense neighbouring peaks. We additionally record transitions at 324, 339 and 356 nm that Gruszecki observed only in dimeric AmB excitation spectrum. Our excitation spectrum (Em 471 nm) agrees with Gruszecki's results, with transitions at 324, 338, and 356 nm. These bands may originate from excitation of AmB dimer or higher-aggregates and thus the vibrational transitions have not been assigned. The corresponding emission spectrum (Ex 350 nm - dimer excitation) also compares favourably, with transitions observed near 452, 473 and 494 nm. Instrumental resolution limits exact determination of these band positions. The weak monomer emission intensity is the combined result of low monomer quantum yield and of AmB molecules in FZ being substantially aggregated. Since AmB aggregation state affects the selectivity of its sterol binding, characterization of the aggregation state in HTFZ is important to understanding the mechanism of its decreased toxicity. AmB fluorescence is sensitive to differing aggregation, and it is thus of interest to examine the fluorescence spectra of HTFZ. Figure 49 shows that the fluorescence excitation and emission spectra of HTFZ are very similar to those of FZ, but with increased intensity. Increased emission intensity upon excitation of AmB monomer in HTFZ was unexpected, but agrees with Gaboriau's finding of slightly increased monomeric AmB content after heating [10]. Increased emission intensity for dimeric AmB excitation is likely not due to increased dimer concentration, but may be related to a change in quantum yield for the fluorescence process. Figure 50 shows the fluorescence emission of AmB as FZ and HTFZ when exciting higher-aggregates of AmB at 325 nm. The emission scan shows identically those bands which were observed for FZ and HTFZ fluorescence under 350 nm excitation. Fluorescence emission intensity for FZ is increased 26% relative to that of 350 nm excitation and this increase is nearly constant across the spectrum. HTFZ excited at 325 nm shows an approximate 10% increase relative to that of 350 nm, but this increase is only in the range 450-540 nm. 159 ---- 0 SDS/AmB -- 94 SDS/AmB — 105 SDS/AmB --- 116 SDS/AmB ^ 209 SDS/AmB 300^350^400 ^ 450 ^ 500 Wavelength / nm Figure 47 UV-visible absorbance spectra of 5.16x10 -5 M AmB as FZ with varying concentrations of added SDS as labelled. SDS concentrations are given as mol ratio with respect to AmB. FZ with no added SDS shows a strong absorption band at 326 nm assigned to aggregated AmB. When monomerized (209 SDS/AmB), the 411 nm band of monomeric AmB dominates, with related vibrational fine structure bands at 388, 367 and 350 nm. Very low absorbance is noted beyond 450 nm. Not all SDS concentrations measured are shown. Reproduced with permission from [5], Copyright 2007 American Chemical Society. 160 1.6 - ---- 0 SDS/AmB -- 94 SDS/AmB — 105 SDS/AmB --- 116 SDS/AmB   209 SDS/AmB 0.4 - I^ I 300 350^400 Wavelength / nm 0 500450 Figure 48 UV-visible absorption spectra of 5.16X10 -5 M AmB as HTFZ with varying concentrations of added SDS as labelled. SDS concentrations are given as mol ratio with respect to AmB. The spectra are very similar to those of FZ; the subtle differences include the absorption band at 292 nm is increased for HTFZ and the symmetry around the 326 nm band with increasing SDS varies between FZ and HTFZ. Not all SDS concentrations measured are shown. Reproduced with permission from [5], Copyright 2007 American Chemical Society. 161 6.4.4 Fluorescence of Amphotericin A Under most countries' pharmacopeia, FZ preparations are permitted to contain up to 5% of the tetraene Amphotericin A (AmA), although usual AmA concentrations are lower. AmA differs from AmB only in that a single bond replaces the C28-C29 double bond of AmB. In order to gauge possible fluorescence contribution from AmA, we recorded fluorescence excitation and emission spectra for a nystatin solution prepared analogously to FZ, denoted nystatin-deoxycholate (NY- DOC). Nystatin shares a similar structure to AmA except the C8 and C9 hydroxyls are shifted to C7 and C10 respectively (Cf. Figure 1). The diene-tetraene moiety responsible for fluorescence is identical in both. Given the similarities in nystatin and AmA structures, their fluorescence characteristics are expected to be comparable. The fluorescence spectra of 5.74 x10' M nystatin as NY-DOC (1:1.8 nystatin:deoxycholate) are presented in Figure 51. Fluorescence excitation and emission wavelengths are identical to those used to record the data of Figure 49; reported intensities for excitation at 408 nm (emission scan) and emission at 560nm (excitation scan) have been magnified ten-fold for clarity. When excited at 408 nm, NY-DOC produces weak fluorescence that tails off gradually. The excitation scan (Em 560 nm) reveals a broad peak centred at 350 nm. Excitation at 350 nm produces more intense fluorescence emission with maximum around 440 nm. Measurement of fluorescence intensity at 471 nm while scanning the excitation wavelength shows a maximum at 336 nm and lesser peaks at 324 and 306 nm. Comparison of NY-DOC spectra to that of FZ shows they are clearly different although there is overlap between the two. Under conditions of monomer AmB excitation and emission there is no meaningful contribution from AmA. When excited analogously to AmB dimer, the intensities of NY-DOC are similar to those of FZ. Since FZ contains maximally 5% AmA, the contribution of AmA to FZ fluorescence is minimal. Accordingly, it is clear that the fluorescence we observe from FZ is not solely that of AmA. 6.4.5 Fluorescence Spectra with Added SDS Addition of SDS to the system was made with the parallel goals of modulating the AmB monomer/aggregate content and to evaluate the relative stability of HTFZ compared to FZ. The fluorescence excitation and emission spectra of FZ in the presence of varying concentrations of SDS are presented in Figure 52. The equivalent figure for HTFZ fluorescence is Figure 53. Panel A of 162 800 - 600 - ▪ • c-/-) a) a) 2 400 - a)0 a) (,) 0 LI 200 0 300 ^ 350 300^350^400^450^500^550^600^650^700 300 • 250 (C5 200 -a) 8 150 - 0 O 100- LT_ 50 Emission Scan (408nm Ex) FZ HTFZ Excitation Scan (560nm Em) FZ •• • HTFZ  0 Emission Scan (350nm Ex) ••A.^-- FZ I*1 , 1^---- HTFZ I^1 14.!^lit^Excitation Scan (471 nm Em)II— FZ I^-1^1 ^ Î • • • • . HTFZI I^..‘ %11,0%.,i r• ,^ \ 1•I /  , ^" A I 1^%■ 1% ;1 1,,^.... ^)II^N<.%., / / \\,. .. ..../^ ....,••,„:..x... ....., ^40 ^ 450^500^550 ^ 600^650^700 Wavelength I nm • Figure 49 Fluorescence spectra of 5.16x10 -5 M AmB as FZ and HTFZ. Panel A) gives excitation and emission of AmB monomer. The excitation scan (Em 560 nm) shows peaks at 324, 339, 356, and 386 nm. Emission scan (Ex 408 nm) gives bands at 435, 471, 528, 562, and 608 nm. Panel B) gives results for AmB dimer. The excitation scan (Em 471 nm) shows peaks at 324, 338, 356 and 385 nm. The emission scan (Ex 350 nm) gives features at 452, 473, 494, and 525 nm. Peak positions are those of FZ; positions for HTFZ are similar. Units of fluorescence intensity are arbitrary. Reproduced with permission from [5], Copyright 2007 American Chemical Society. 163 1000 Emission Scan (325nm Ex)/^4.1 i^ 1^-- FZI ---- HTFZi ..--•^% P/ i./ \ • i\\r *\ tc^\ ...\ i^ N.. \--.. ......■ ^ ..„^‘ -\..% it N.,..,if^"N. \......,‘ /id 444..,..1., -.:,`,...%,.. Q 800 - U) 600 - a)0 a) 400 - U)a) 0 LT_ 200 - 0 • 400^425^450^475^500^525 ^ 550 ^ 575 ^ 600 Wavelength / nm Figure 50 Fluorescence emission spectra of 5.16x10' M AmB as FZ and HTFZ. 325 nm excitation excites higher aggregates of AmB. Units of fluorescence intensity are arbitrary. Reproduced with permission from [5], Copyright 2007 American Chemical Society. 164 Figure 52 shows monomer emission scan (Ex 408 nm) and monomer excitation spectrum (Em 560 nm). Moderate increases in SDS concentration lead to increased fluorescence intensity for bands at 339 and 356 nm of the excitation scan. Higher SDS concentration shows decreased intensity for these bands with a small red shift. SDS addition dramatically and monotonically increases fluorescence intensity for the 388 and 410 nm monomer excitation bands suggesting increasing AmB monomer concentration. The corresponding emission scan shows large increases in AmB monomer fluorescence intensity at 528, 562 and 608 nm peaks and unexpected similar increases at 439 and 460 nm. Panel B presents the dimer emission scan (Ex 350 nm) and excitation scan (Em 471 nm) showing that increasing SDS/AmB ratio leads to an initial increase in fluorescence emission intensity while higher SDS concentrations reverse this trend. The largest increase occurs at the 473 nm band. At high SDS concentration, monomer emission bands at 562 and 608 nm become apparent. The fluorescence excitation scan shows that the 325, 338 and 356 nm bands increase with moderate SDS concentration, but then begin to decrease at high SDS concentration. Monomer excitation bands at 386 and 410 nm grow in monotonically with increasing SDS concentration. The fluorescence of HTFZ in the presence of SDS is very similar to that of FZ. Figure 53 shows HTFZ has increased dimer fluorescence intensity, and the peak intensities occur at slightly lower SDS concentrations. 6.4.6 Dual Fluorescence of AmB The emission at 437 and 460 nm after 408 nm monomer excitation that is observed in Figure 52 panel A can be rationalized in terms of S 2 - S o fluorescence emission from the l'B u state to the l'Ag ground state. By comparison, the S 2 -S 0 fluorescence of 1,3,5,7,9,11,13-tetradecaheptaene appears at 430, 440 and 460 nm. In panel B, the monomer excitation bands which appear for high SDS concentrations at 386 and 410 nm are a further expression of S 2 -S 0 fluorescence; the bandwidth of our 471 nm emission measurement allows detection of the edge of the 460 nm S 2-S 0 transition. Examination of the ratio of S 2 -S 0 to S,-S 0 fluorescence can give detail about the rate of internal conversion. Figure 54 shows the ratio of S 2 -S o to S,-S 0 emission intensity for FZ and HTFZ as a function of SDS concentration. S2 fluorescence was taken as the peak intensity for the 437 nm band, 165 •••. •., . v. • 1000 D 800 - cn c 600 -a) "e" a)Uc 400 - 08 0 1.1  200 - 0 300 ...4*-1^Emission Scan d I -- 350nm Ex ---- 408nm Ex (x10) Excitation Scan •• • •. 471nm Em — 560nm Em (x10) V l•\..^ . Ni 1.4\• dift\t\wk,Auseti $40A440Amp....144,:A.A.44e"..,.•-' t.'^.350^400^450^500^550 600^6 50 Wavelength / nm ... Figure 51 Fluorescence excitation and emission spectra of 5.74 x10 -5 M nystatin- deoxycholate under similar conditions to those used acquiring the data of Figure 49. Excitation spectrum (Ex 408 nm) and excitation spectrum (Em 560 nm) both magnified 10 x for clarity. Units of fluorescence intensity are arbitrary. Reproduced with permission from [5], Copyright 2007 American Chemical Society. 166 300 800 ^ A) 700 - < 600 - 350^400^450^500^550^600^650^700 0 SDS/AmB ^ 105 SDS/AmB ---- 116 SDS/AmB — 209 SDS/AmB '-,'^PI/ L500--^• ' I a) i t'1^ ' ; • 400 -^11 ' O i a)^sw'. •%,, X 300  -^I. 92^c: o r • 200 - cu_^r.: 100 -‘41 0 B) 1600 - t`♦ .^h^ dr • ..... ‘t it ... ... ...... ......... „ .••^• ....... ... i • I^% r Ai ..1 i',^Ali -- 0 SDS/AmB / A   105 SDS/AmB A A ^---- 116 SDS/AmB1 /^♦^— 209 SDS/AmBI .., I ♦ 1^ S.I % I .0 1 :^ . / : ♦ /^f'N..^ ., \ ♦ A / ..- -/^\ . .....^. /.'...,,,..^ .......... %^-.-..., ...-.......--...... 300^350^400^450^500^550^600^650^700 Wavelength / nm Figure 52 Fluorescence excitation and emission scan of 5.16x10 -5 M AmB as FZ with varying concentrations of added SDS as labelled. Panel A) presents results for AmB monomer: emission scan (Ex 408 nm) and excitation scan (Em 560 nm). Panel B) shows results for AmB dimer: emission scan (Ex 350 nm) and excitation scan (Em 471 nm). Not all SDS concentrations measured are shown. Units of fluorescence intensity are arbitrary. Reproduced with permission from [5], Copyright 2007 American Chemical Society. •  I 1200 - ca) a) 800 - a) a) 0 400 - iv 0 `itaya I I I 167 300 800 ^ A) 700 - < 600 - ...s co 500 - a)^It. .70 400 -^It 0^ Ai %c % N 300--^ I (no 1 It92 1 I 0^ i 12 200 -u_ \ I yr .... 2500 -- 0 SDS/AmB ^ 105 SDS/AmB ---- 116 SDS/AmB — 209 SDS/AmB . I•• 40 0• 10: 300^350^400^450^500^550^600^650^700 Wavelength / nm Figure 53 Fluorescence excitation and emission scan of 5.16x10 -5 M AmB as HTFZ with varying concentrations of added SDS as labelled. Panel A) presents results for AmB monomer: emission scan (Ex 408 nm) and excitation scan (Em 560 nm). Panel B) shows results for AmB dimer: emission scan (Ex 350 nm) and excitation scan (Em 471 nm). Not all SDS concentrations measured are shown. Units of fluorescence intensity are arbitrary. Reproduced with permission from [5], Copyright 2007 American Chemical Society. 2000 - a) .73 c 1500 - — CD 0 1000CO0 z LT_ 500 - 100 i.^ ... r."%^I ^s^...r^it r %., ..s 1^t I-.^.^"..I I / t i I r ■^... ^1 . ... ^ rit^.., \\^ i^ n.,,^ .... .. . .. ....^: ,,•.„, , ... %,.... ....... ...,, ,. / • It \̂ `-'^ •■. • . ./N II • •• O. "I'Vrabwk_ ■•••••,„„.. W. 0.. • •■.rw,;zb..r"...,..:Z: 350^400^450^500^550^600^650^700 -- 0 SDS/AmB ^ 105 SDS/AmB ---- 116 SDS/AmB — 209 SDS/AmB 168 while S I fluorescence was taken as the peak intensity of the 562 nm band. For zero added SDS, the S 2/S, emission ratio for FZ is close to unity. A slight decrease in the ratio is observed at low SDS concentration, but the overall trend is a gradual increase with increasing SDS. For this increase to occur as AmB is forced to the monomeric state, the quantum yield of the S 2 - So transition must be increasing, likely due to a decrease in the rate of S2 to Si internal conversion. The corresponding ratio for HTFZ shows significantly more S 2 than S, emission at zero SDS concentration, followed by a rapid decrease at moderate SDS concentrations. At 109 SDS/AmB mol ratio the emission ratio for HTFZ matches that of FZ and this remains so for higher SDS concentrations. The environment of AmB in the 'super-aggregates' of HTFZ must be such that the S 2 - S o transition is more favoured. It is known that solvent stabilization preferentially effects the l'B u (S 2) state more than the 2 1 /kg (S I ) due to the larger transition dipole [261], and any change in S 2-S, energy gap will be reflected in the emission ratio. The controlling factor for S 2-S 0 vs. S I -So fluorescence from AmB is not the aggregation state of the drug itself, but the 'solvating' influence of neighbouring AmB or SDS. Accordingly the ratio of S 2 - S o to S I -S o fluorescence may serve as a useful tool for measuring the super-aggregated nature of AmB solutions; we note the strong similarity in FZ and HTFZ electronic absorption spectra precludes the use of the UV-Vis absorption technique for this purpose. It is interesting to note there are instances of monomer emission features appearing after dimer excitation (dimer to monomer "cross-over"). Figure 49 shows emission from AmB monomer (560 nm) can occur after stimulation of the excitation bands of 324, 339 and 356 nm. Additionally, dimer excitation at 350 nm leads to broad monomer emission near 528 nm, presumably the (0-0) transition. Clearly, dimer excitation may lead to monomer emission. Energy transfer from dimer to monomer may occur by either transfer from the dimer's upper exciton energy level into the l'B u monomer level or from the lower exciton level into the 2'A g monomer level. Identical AmB emission bands when excited at 325 and 350 nm suggests there is a further energy transfer from higher-aggregate upper exciton level to dimer lower exciton level. Excitation of either higher-aggregates or dimers results in dimer emission. The most likely explanation of these energy transfer processes is resonance energy transfer via Coulombic coupling. The separation distance which provides 50% efficiency of energy transfer (the FOrster distance) may be in the tens of A. The distance between 169 AmB monomers within aggregates has been variously calculated to be 7.8 A [74], 2.9 A [237], 4.9 A [254] and 4.6 A [11], all of which suggest that resonance energy transfer is a feasible mechanism. Figure 55 summarizes fluorescence emission intensity and UV-Vis absorbance data as a function of added surfactant/AmB ratio. The effect of added SDS to FZ and HTFZ is shown along with the effect of deoxycholate added to FZ for comparison. Panels A and B show the fluorescence emission intensity after excitation at 408 nm (monomer - left panel) and 325 nm (aggregate - right panel). Panels C and D show the corresponding absorbance data. Addition of SDS has little effect on the FZ and HTFZ monomer absorbance and fluorescence emission until about 100 surfactant/AmB ratio where a sigmoidal increase is observed. The rate of increase in the FZ and HTFZ monomeric absorption slows above 127 surfactant/AmB, but still continues to rise at the highest SDS concentration suggesting that AmB is largely but not completely monomeric above 127 surfactant/AmB. Mixtures of 16:1 SDS:AmB have previously been reported to hold AmB in monomeric form, if SDS concentration is above its CMC of 8 mM [262]. Forcement of monomeric AmB is determined by SDS concentration more than surfactant/AmB ratio, with the onset of monomerization at the CMC. The higher-aggregate absorbance of FZ and HTFZ shows a sigmoidal trend with high initial absorbance at low SDS concentration, then a rapid decrease between 100 and 127 surfactant/AmB ratio. Further increases in SDS concentration do not lead to lower absorbance suggesting the higher-aggregates are entirely broken-up. Addition of deoxycholate also gives increased monomer fluorescence emission and absorbance, but the changes are smaller and more gradual compared to added SDS. Even at 209 deoxycholate/AmB ratio, FZ is still partially in aggregated form. Addition of deoxycholate to FZ also leads to a decreased absorbance at 325 nm but it is more gradual than the decrease with SDS. The absorbance at the highest surfactant/AmB ratio is higher for deoxycholate than SDS suggesting incomplete break-up of higher-aggregates by deoxycholate. The fluorescence emission intensity of FZ and HTFZ higher-aggregates with added SDS (panel B) show a different trend than that predicted from absorbance measurements. At low surfactant concentration, the intensity is low and decreases on initial surfactant addition. At 85 SDS/AmB ratio, fluorescence intensity undergoes rapid increase, peaking near 107 SDS/AmB ratio. Further increases of SDS result in decreased fluorescence intensity to level similar to the zero-added 170 *al. ....^ •• ...I. ...^ ••• ••■ ..^ .0 ft, ..., . ••• 2.5 c 0 .ctT̀' ffw (6- 1.5 - 0 (.6\j .0^1 rts cc 0.5 - --o— HTFZ - -11. - FZ A - • 0^50^100^150^200 SDS/AmB Mol Ratio Figure 54 Ratio of fluorescence emission intensity from S2 (437 nm band) and S I (562 nm band) energy levels for 5.16 x10 -5 M AmB as FZ and HTFZ with increasing sodium dodecyl sulfate (SDS) concentration. Fluorescence excitation at 408 nm. Reproduced with permission from [5], Copyright 2007 American Chemical Society. 171 600 Â) • FZ/SDS >, 450 - • HTFZ/SDS FZ/DOCcu)a) 0 ▪ 300- U 150 -0 U- >, 1500 a) 7:3) ▪ 1000 0 0 2 500 -0 LT_ *1-.4 - -^• I- - - 2000 0^50^100^150^200 Surfactant/AmB Mol Ratio 1.2 1 •- a) 0.8 -0 C •ca 42 0.6 - U_ct < 0.4 - 0.2 - ^ 0 , ^ 0^50^100^150^200 Surfactant/AmB Mol Ratio 0^50^100^150^200 Surfactant/AmB Mol Ratio 1.2 C ) 0 0.8 -0 420.6 - co < 0.4 - 0.2 - 0 0 50^100^150^200 Surfactant/AmB Mol Ratio Figure 55 Summarized fluorescence emission and absorbance of 5.16 x10 -5 M AmB as FZ and HTFZ with increasing surfactant concentration. Panels A) and B) give peak fluorescence emission from AmB monomer (Ex 408 nm) and higher aggregates (Ex 325 nm), respectively. Panels C) and D) give absorbance at 408 nm and 325 nm, respectively. Units of fluorescence intensity are arbitrary. Reproduced with permission from [5], Copyright 2007 American Chemical Society. 172 surfactant case. FZ with added deoxycholate does not show a peak in fluorescence intensity. Rather, an initial decrease then a moderate step-wise increase is observed between 85 and 100 deoxycholate/AmB ratio, followed by an unchanging fluorescence with continued deoxycholate addition. These results may be rationalized in part by considering the behaviour commonly observed in dye- aggregation studies using surfactants to modulate aggregation. Premicellar surfactant-dye complexes are formed, and with continued surfactant addition the aggregate is broken-up to give micelle-incorporated dye monomer [263, 264]. Tancrêde et al. [9] found that surfactants added to solutions above AmB's CMC of about 10 -7 M first yielded an increase in scattered light intensity as surfactant penetrated AmB micelles and changed their size or shape. Further addition of surfactant gave a decrease in scattering, indicative of micelle-break-up and dispersal of AmB in solution. Above the CMC of the added surfactant, scattering increased again due to formation of surfactant- only micelles. Accordingly, we may expect strong absorbance and fluorescence from AmB higher- aggregates at zero added surfactant concentration, followed by decreases for aggregates near the CMC and a corresponding large increase in monomer absorbance and fluorescence. The CMC of our SDS/FZ system was found using surface tension measurements to be 6.4 mM SDS (7.4 mM total surfactant concentration including SDS, AmB and deoxycholate). This corresponds to an SDS/AmB ratio of 124. The CMC of the deoxycholate/FZ system is expected to be near 4.5 mM [9], or 78 added deoxycholate/AmB ratio on our scale. We observed dramatic increases in monomer fluorescence and absorbance slightly below the CMC when SDS is added, indicative of SDS intercalation into AmB micelle leading to eventual break-up of the micelle, which was confirmed by the corresponding drop in higher-aggregate absorbance. The weak fluorescence emission from higher-aggregates at zero added surfactant concentration does not fit our expectations. The 325 nm excitation is into a strongly absorbing band and we expected to see strong dimer emission at 471 nm. Instead the emission is relatively weak and this is presumably due to fluorescence quenching of the excited higher-aggregate or excited dimer. Any process which decreases the quantum yield by favouring alternate relaxation pathway(s) can be considered a quench. The source of this quenching is unclear, though self-quenching seems a likely 173 possibility. Self-quenching occurs by energy transfer to a non-fluorescent state on a like molecule either by collisional orbital interaction (Dexter transfer) or by Coulombic coupling between molecules (Forster transfer). Higher-aggregates are likely to have AmB molecules in a variety of microstates, raising the possibility of successful Coulombic coupling and resultant FOrster energy transfer. We hypothesize that further surfactant intercalation into the higher-aggregate relieves the quench, resulting in increased fluorescence near 100 surfactant/AmB ratio (Figure 55 Panel B). Decreased quenching may be related to the rigid conformation AmB was found to adopt in SDS micelles [262], which may restrict the modes available for FOrster energy transfer. The peak-shape of the fluorescence vs. added surfactant curve is the result of a competition between decreased quenching and aggregate break-up as SDS concentration is increased. Pre-CMC complex formation between surfactant and AmB which changes the shape or structure of the AmB micelle such that AmB is more closely packed is consistent with initial increases in absorbance and initial decreases in higher-aggregate fluorescence due to quenching. The constant fluorescence emission for deoxycholate/FZ above 100 deoxycholate/AmB ratio reflects deoxycholate being less effective than SDS at breaking-up the higher-aggregates. The SDS/AmB ratio required to instigate break-up of AmB aggregates and form monomer can be compared for FZ and HTFZ to determine their relative stabilities to disaggregation by surfactant. Decrease in absorbance at 325 nm measures break-up of aggregates, while increases in monomer absorption at 408 nm and increases in fluorescence emission at 560 nm (Ex 408 nm) measure the rise in AmB monomer concentration. The trend of increasing monomer content with increasing SDS/AmB is virtually identical for FZ and HTFZ, with the highest rate of change at 115 SDS/AmB The corresponding decrease in higher-aggregate content shows similar trend for FZ and HTFZ absorbance, while the fluorescence intensity peaks at slightly lower SDS/AmB ratio for HTFZ. The slight difference in peak fluorescence intensity is likely not significant, so we conclude that FZ and HTFZ have similar stability towards disaggregation by SDS. This result is relevant to understanding how AmB interacts with the cell membrane since AmB interaction with lipid may require an analogous disaggregation of FZ and HTFZ. 174 6.5 Conclusions We have measured the fluorescence spectra of FZ and HTFZ and found that the fluorescence is indeed that of AmB and not of a contaminant. The fluorescence spectra of monomeric and dimeric AmB are distinct, but FOrster energy transfer is observed to occur from excited dimer to monomer. Additionally, excitation of AmB monomer gives both S 2 - S o as well as the previously observed S I - S o fluorescence emission. Application of AmB fluorescence measurements to the study of monomeric or dimeric AmB interaction with membranes will thus require careful selection of the excitation and emission wavelengths to ensure species-specific response. FZ and HTFZ have similar S I - So fluorescence spectra. The ratio of S 2-S o to S I - So fluorescence differs for FZ and HTFZ at zero and low surfactant concentration because the solvating environment around the AmB molecules is different between FZ and HTFZ. Use of this ratio may therefore prove to be an efficient way to characterize the super-aggregation of AmB solutions. SDS forms complexes with micelles of FZ or HTFZ at concentrations below the CMC of SDS. Complexes at low surfactant concentration featured decreased higher-aggregate fluorescence due to increased fluorescence quenching, while addition of further surfactant decreased the quenching and increased fluorescence. At surfactant concentrations close to the CMC, higher-aggregates are broken-up to form AmB monomer. FZ and HTFZ higher-aggregates are equally stable towards SDS-induced break-up, with similar surfactant concentration required for disaggregation. 175 7 AmB INTERACTION WITH FLOATING LIPID MONOLAYERS 7.1 Introduction This chapter describes measurements of AmB interaction with lipid monolayers at the GIS interface. These experiments are complimentary to our earlier electrochemical investigations of AmB- influence on the electrode-supported monolayer in Chapter 4. Although the state of the monolayer at the GIS interface cannot be controlled by electric potential, the advantage to working at the GIS is that the surface packing density and thus surface pressure can be readily varied. Experiments were made with monolayers formed by co-deposition of AmB and lipid and with AmB allowed to interact with DOPC monolayers that were loosely or tightly-packed. Although similar, these different approaches significantly change the nature of AmB interaction with the film. All the experiments of this chapter combined measurement of monolayer properties (e.g. surface pressure) with in situ collection of fluorescence from the monolayer. Two approaches were used. In the first, a fluorophore-tagged lipid was doped into the DOPC layer. In the second, fluorescence emission was measured from AmB in the monolayer. This work had four objectives: 1) We wanted to characterize AmB's influence on the floating lipid monolayer, in an effort to have better understanding of the properties of the AmB-influenced monolayers that were adsorbed to the Hg electrode in our earlier studies. Although it has been shown that the state of the monolayer adsorbed to an electrode surface is characteristic of its state at the GIS interface prior to adsorption, we had not examined our AmB-influenced monolayers at the GIS earlier. 2) We wished to measure AmB-induced structural changes such as possible pore formation. The earlier electrochemical work suggested ion-permeable regions of micron-sized dimensions. This is large enough to potentially visualize using microscopy. Much as a fluorophore may segregate to a particular lipid domain, variation in local fluorophore probe concentration (high or low) in the porous region was expected to provide contrast relative to the non-porous parts of the monolayer. 3) We wanted to examine the influence of sterol incorporation on AmB-induced changes in monolayer structure. 4) We desired to be able to detect AmB within a monolayer, using 176 its intrinsic fluorescence. Measurement of AmB fluorescence means that the drug is simultaneously the probe, and the model system is thus simplified by elimination of possible effects of a separate fluorescent probe. These latter experiments were made late in the overall progress of this work and only initial experiments have been completed. Nonetheless, these experiments tie together several of the otherwise disparate parts of this thesis and they hold significant promise in their own right for the future study of AmB-lipid monolayer interaction. To reach these goals, a number of challenges were overcome. The earlier electrochemically-based experiments of AmB-monolayer interaction showed the largest changes after potential-induced defect formation within the monolayer. With no potential-control available at the GAS interface to put the monolayer into a defective state, conditions were developed to simulate conditions of easy AmB insertion. Although this approach gives our lipid monolayer different properties than a biological membrane, it was thought that a greater surface concentration of AmB would allow detection more readily. The most significant challenges, however, related to the direct measurement of AmB fluorescence from a monolayer. Here, an unconventional laser source used to excite AmB proved to be useful. Further, careful adjustments were made to match the laser illumination spot on the monolayer with the microscope focal point and the vertical position of the monolayer. As the amount of AmB within a monolayer is low, and AmB is only weakly fluorescent, care was taken to maximize the collected AmB fluorescence signal. 7.2 Literature Review Systematic studies of the characteristics of floating monolayers date to work of Pockels and Lord Rayleigh at the end of the 19' century, with our understanding and techniques today largely due to Langmuir [99]. Blodgett, Langmuir' s assistant, pioneered the deposition of floating monolayers onto solid supports to form so-called Langmuir-Blodgett films [265]. The use of lipid monolayers as a cell membrane model with which to study interaction of biomolecules also began in the 1930s. For example, Rideal [266] studied the penetration of fatty acids into cholesterol monolayers, finding facile penetration. Recent developments in the use of floating monolayers as membrane models have been the subject of several reviews [267, 268]. In comparison with other models, floating monolayers on the Langmuir trough have the advantage of easily varied lipid composition. 177 Furthermore, the use of floating monolayers does not rely on what may be rough or ill-defined supporting surfaces, such as those of adsorbed liposomes or Langmuir-Blodgett films. 7.2.1 Monolayer Composition Since the monolayer technique allows the lipid composition to be easily varied, an important question is: What lipid composition should be used to best mimic the cell membrane? A wide variety of synthetic and natural phospholipids, glycolipids and steroids are available, and the specific species included in the monolayer depends on the nature of the membrane to be modelled or the goals of the study. As phosphatidylcholines are a major component of mammalian cell membranes, simple models often use DPPC (di-C16:0) or DOPC. DPPC has saturated acyl chains, while those of DOPC are unsaturated. Accordingly, DPPC monolayers show a liquid-expanded to liquid- condensed phase transition upon compression that is not seen with DOPC. Studies based on a penetrant's influence on phase behaviour would likely use DPPC. On the other hand, the more fluid DOPC monolayer presents a liquid phase similar to that found in the hydrophobic core of a membrane [75] and thus the DOPC monolayer has found extensive use, in particularly when mixed sterol-phospholipid films are used [269-271]. Indeed, DOPC has been suggested to be the best model phospholipid for the cell membrane [272]. A second question concerns the concentration of incorporated sterol. Although different cells have different sterol loading, a typical cholesterol concentration is approximately 30 mol% [273, 274]. Within the kidney, where AmB damage occurs, proximal tubule cholesterol content is 20 mol% [275]. Fungal cells typically contain about 13 mol% ergosterol [75]. Nonetheless, lower sterol concentrations, in the range 5-13 mol% have been used in several studies of polyene-sterol interactions [75, 276-278, 279 and reference 4c therein]. Mixed monolayers of phospholipids and sterols have been studied extensively, as reviewed in [280, 281]. Cholesterol is known to interact with phospholipids such that the hydrocarbon chains of the lipid layer become more fluid below the phase transition temperature and more rigid above [282]. Additionally, cholesterol condenses the average cross-sectional area of phosphatidylcholines above the transition temperature [283, 284]. Ergosterol induces a similar condensing effect [283, 285]. 178 7.2.2 Measurement of Biomolecule Interaction The interaction of biomolecules with lipid monolayers may be studied by forming mixed monolayers or through examining the penetration of a soluble biomolecule into the floating layer. The application of both methods to the study of drugs, enzymes, proteins and hormones is reviewed in [286]. Penetration can be measured through changes in surface area or surface pressure. Simple measurement of surface pressure is not without drawbacks, however, because large changes in surface pressure are not necessarily proportional to penetration [287, 288]. Small changes are usually proportional. Other ways to measure penetration may involve complicated methods for separating the surface layer from the bulk for further analysis, such as aspiration [289], slicing, or adsorption [290]. Fortunately, methods have been developed for observing the monolayer surface, applicable to studying both co-deposited layers and penetration of layers. Some of the methods, such as reflective vibrational or UV-vis spectroscopy, or synchrotron X-ray diffraction, probe the properties of the monolayer but without spatial resolution. Two commonly used techniques which do provide spatial resolution are Brewster angle and fluorescence microscopy, reviewed in [291, 292] and [293], respectively. Brewster angle microscopy (BAM) uses the reflection of polarized light from the interface to provide information on the thickness of the monolayer and its optical anisotropy. Fluorescence microscopy of monolayers usually involves incorporation of a lipophilic fluorescent probe molecule at low concentration. The probe molecule may partition into a particular lipid phase. Unequal distribution of the probe between phases is the basis of the contrast in fluorescence images [294]. The incorporation or penetration of a biomolecule into the layer may result in a change in phase behaviour which can then be imaged. For example, Bringezu et al. used epi-fluorescence microscopy to examine the effect of environmental tobacco smoke on monolayers of lung surfactant monolayers, finding the smoke altered the distribution and fraction of crystalline domains in the monolayer [295]. Their Texas Red fluorophore-tagged lipid partitioned to the more disordered liquid-expanded phase and out of the liquid-condensed phase. Biswas et al. used epi-fluorescence microscopy of BODIPY-doped lipid monolayers on a Langmuir trough to study the effect of a synthetic variant of lung surfactant protein B [296]. The BODIPY fluorophore segregated to the liquid-expanded phase. Analyzing the effect of the protein on the phase behaviour of the lipid, they 179 determined that near-physiological concentrations of the variant protein induced morphological changes that were analogous to those of the actual protein. Although fluorescence microscopy of monolayers is common, other fluorescence methods without spatial resolution can also be used. For example, Momsen et al. developed a fluorescence-based method for real-time measurement of solute partitioning using two fiber optic cables mounted above the monolayer. Measurement of fluorescence intensity and spectra was possible, and they used the setup to measure the kinetics of BODIPY-labelled protein binding to the monolayer [297]. They observed facile penetration into a fatty acid monolayer, but not into a phospholipid monolayer. Many other forms of monolayer fluorescence measurement are also possible, such as fluorescence anisotropy, fluorescence resonance energy transfer (FRET), fluorescence lifetime imaging (FLIM), and fluorescence recovery after photobleaching (FRAP). For example, Yapoudj ian et al. used FRET to study adsorption of lipase to a cis-parinaric acid (C18:4; cis-9,15, trans-11,13) monolayer, exploiting the intrinsic fluorescence of the protein's tryptophan residue and the parinaric acid itself [298]. 7.2.3 Monolayer Studies of AmB Monolayer-based studies of AmB are numerous in the literature, and can be grouped generally according to the theme of the work. Common themes are: the stoichiometry of AmB/sterol or AmB/phospholipid complexes; the state of AmB within a layer (aggregation state or molecular orientation); penetration of AmB into monolayers; and effects of spreading method or solvent. Complexes of AmB/cholesterol with 2:1 stoichiometry have been reported [e.g. 142, 299-302], while similar 2:1 complexes have also been reported for AmB/ergosterol [e.g. 301, 303]. Furthermore, 2:1 complexes of AmB/phospholipid have been noted, with the AmB thought to be oriented vertically and the lipid horizontally within the monolayer [304, 305, 306 and references 17 & 18 therein]. Ockman [307] used polarized reflection spectra of AmB and AmB/cholesterol monolayers to study molecular orientation of AmB. The presence of cholesterol in the monolayer changes the orientation 180 of AmB; increasing cholesterol concentration causes the polyene to orient more vertically. Compression isotherms of pure AmB monolayers show a phase transition which corresponds to a horizontal to vertical reconfiguration of the layer [306 and references 17 & 24 therein, 308]. BAM of AmB monolayers suggests this transition results in a three-fold increase in monolayer thickness [309], in keeping with the proposed reorganization. Gagos et al. [310] used linear dichroism FTIR of monolayers deposited to a solid support to examine AmB orientation with and without added sterol. Incorporation of cholesterol changed the fraction of AmB in horizontal orientation from 59% to 85% and vertical from 41% to 15%. The possibility of AmB orientation in the layers being intermediate between horizontal and vertical was discounted. Measurements of subphase AmB penetration of lipid monolayers are few compared to studies that use a co-deposition approach. The studies that have been done used pure AmB rather than the more soluble formulations, thus limiting the maximum subphase concentration. Nonetheless, Saint- Pierre-Chazalet and co-workers [301] characterized penetration of radioactively-labelled AmB into phosphatidylcholine/sterol monolayers by measurement of the increase in surface pressure and radioactivity. The groups of Miriones and Dynarowicz-lAtka have published several studies of AmB penetration into DPPC, sterol, or mixed DPPC/sterol monolayers [272, 311, 312]. Their study of penetration into mixed DOPC/sterol films showed that the aggregation state of AmB in the subphase is important; monomeric AmB penetrated cholesterol-containing layers more readily than ergosterol-containing ones. On the other hand, aggregated AmB penetrated more easily into ergosterol-containing monolayers. BAM showed minimal and slow AmB insertion into the DOPC monolayer when the subphase contained monomeric AmB, and much more rapid and extensive insertion when aggregated AmB was present [272]. Features in the BAM images representing insertion were —7 p,m across for monomeric AmB in the subphase and slightly smaller when AmB in the subphase was aggregated. The effect of spreading solvent on AmB monolayers has been studied by Sykora [313 and references 11 & 15 therein]. Deposition of organic monolayers is usually done with a volatile, water- immiscible spreading solvent, but AmB is poorly soluble in these solvents. Various solvent systems have been used successfully, but with differing results. When spread from methanol/chloroform, 181 AmB's compression isotherm shows a transition between liquid-expanded and liquid-condensed states, but this transition is not observed for spreading from 2-propanol/water. Miliones studied mixed AmB/phospholipid films formed either by co-deposition of the two components or sequential deposition [314]. Sequentially-spread monolayers showed desorption of AmB into solution upon compression. Desorption of AmB from monolayers had been observed previously and found [302] to obey the Ter Minassian-Saraga model [315] of initial dissolution followed by diffusion. 7.3 Materials and Methods 7.3.1 Introduction The experiments described in this chapter all centre on the manipulation of floating lipid monolayers and simultaneous measurement of fluorescence from the layer. Our use of the fluorescence microscope here is largely similar that described in Chapter 5, but two different light sources were used in this work. Additionally, the technique used here to deposit the lipid monolayer to the trough surface was the same as described in Chapters 4 and 5. We will thus begin discussion with the setup of the Langmuir trough and the method of surface pressure measurements. The properties of the BODIPY fluorophore used to render the monolayer fluorescent in these experiments is then described, with a focus towards its suitability for lipid membrane studies and its spectral characteristics. Modifications made to the previously described fluorescence imaging methods are next noted. Lastly, the characterization and use of an external fluorescence excitation source for the purposes of collecting AmB fluorescence directly are discussed. 7.3.2 Langmuir Trough As described in Chapter 3 on theoretical background, the Langmuir trough is a simple instrument used to manipulate the properties of monolayer films. The trough provides a way to control the available surface area of the film. Measurement of surface pressure as a function of area produces a compression isotherm. Compression isotherms were recorded using a model 102A Langmuir trough (Nima Technology, Coventry, UK). The teflon trough features a 100 cm 2 surface area, and an approximate 50 ml volume. The trough was modified to provide a well in the bottom, allowing a microscope objective to be brought close to the surface. A quartz coverglass slip (0.17 mm thick) served as the window. The simultaneous in situ fluorescence measurements were made with the 182 trough mounted above the microscope, and the whole apparatus contained within a light proof box mounted on an anti-vibration table. A Nima PS4 film balance measured the force on the Wilhemy plate. As described in the theory section, a component of this force is due to surface tension. The measured surface pressure, 'Tr, is the difference between interfacial tension at a bare subphase surface and the monolayer-covered surface. The balance was calibrated using a 100 mg weight according to the manufacturer's protocol. Plates supplied by Nima were made from Whatman (Middlesex, UK) #1 chromatography paper, with dimensions 10.25 mm wide, 0.25 mm thick and thus have perimeter 21.00 mm. Use of a paper Wilhemy plate provides excellent wetting and allows the contact angle to be taken as zero. The trough was enclosed in a plexiglass box with a continuous argon gas purge. The trough was cleaned prior to each compression isotherm. If applicable, the prior monolayer was first removed from the surface by aspiration. Two ml of alkaline Extran 300 anionic detergent (EMD chemicals, NJ) was added and the trough filled with warm water. The teflon barriers were cycled open and closed twice, then the soapy solution aspirated out. To fully rinse the trough, it was filled with warm water and aspirated empty, a procedure repeated ten times. An eleventh rinse with room temperature water ensured the trough was cooled. All water used was 18.2 MQ•cm from a Milli-Q system. After filling the trough with 0.1 M KC1 subphase, a pre-wet Wilhemy plate was hung into solution. Two tests were used to check trough cleanliness: a rapid compression isotherm yielding a surface tension increase of <0.2 mN/m and determination of the absolute surface tension of the solution to be 73.0 ± 0.1 mN/m, the known surface tension of 0.1 M KCl [316]. Additionally, separate compression isotherms of DOPC and cholesterol were periodically measured and compared to literature results. Measured collapse pressure and limiting mean molecular area were compared to literature values. Prior to addition of the lipid solution to the surface, the force on the film balance was zeroed, so that the film balance would provide measurement of surface pressure after deposition of the organic 183 layer. With the barriers fully open, an aliquot of the organic solution was injected from a Hamilton microsyringe onto the GIS interface. As an example, 17 p,1 of DOPC/fluorophore solution in n- pentane solution (0.5 mg/ml DOPC, 3 mol% BODIPY FL-DHPE) was added, spotting drops around on the surface. Ten minutes was allowed to elapse for the pentane to evaporate. For the case of compression isotherms recorded with drug in the subphase, an aliquot of AmB-containing drug solution or control was injected behind each barrier in a lengthwise fashion. No stirring was used, so the system was allowed to equilibrate for 20 minutes. This may not achieve homogeneous AmB concentration in subphase, but did allow the surface pressure increase to reach equilibrium. If homogeneity is assumed, AmB concentration was approximately 0.411M when FZ or HTFZ added, and 4 when ABLC added. These concentrations are roughly equivalent to 80% TD on the concentration scale used earlier in electrochemical characterization of AmB-monolayer interaction. The monolayer was compressed at a rate of 3 cm 2/min and the film pressure measured. To avoid cross-contamination, different microsyringes were used for depositing the lipid monolayer and adding AmB formulations. Three different monolayer compositions were studied, each in the presence and absence of subphase AmB. The systems were that of DOPC, mixed cholesterol-DOPC, and mixed ergosterol-DOPC. Each of these systems also contained BODIPY FL DHPE fluorophore. For each system, a total of 6.5 x 10 15 molecules were deposited to the trough surface. The detailed composition of these systems is given in Table 3. The DOPC monolayer was examined when the subphase was only 0.1 M KC1 and also when the subphase contained FZ, HTFZ, ABLC or deoxycholate. We did not examine the effect of ABLC or deoxycholate on the sterol-containing layers. Table 3 Composition of Deposited Monolayers Monolayer name Mol% DOPC Mol% BODIPY FL Mol% Sterol DOPC 97 3 - DOPC/Cholesterol 87.4 3 9.6 DOPC/Ergosterol 87.6 3 9.4 184 7.3.3 BODIPY FL DHPE The fluorophore used in these experiments was a lipid-tagged version of dipyrrometheneboron difluoride (BODIPY, 4,4 -difluoro-4-borata-3a-azonia-4a-aza-s-indacene). Dipalmitoylphosphatidylethanolamine (DPPE, di-C 16:0) headgroup-tagged with BODIPY was purchased from Invitrogen-Molecular Probes (catalogue # D-3800; Eugene, OR). BODIPY fluorophores are recognized as having superior chemical and spectroscopic properties for lipid membrane studies. BODIPY has high quantum yield (> 0.9 in lipidic environment), is resistant to photobleaching, is relatively insensitive to environment (pH, membrane potential, medium polarity and oxygen), has high molar absorptivity (typically > 80,000 M - ' cm'), and has concentration- dependent spectrum due to dimer formation. Two ground-state dimer structures have been determined, one has the transition dipoles arranged head-to-tail, producing red-shifted absorption (maximum at 570 nm) and is fluorescent (maximum at 630 nm). The other has card-pack aggregation, giving a blue-shifted absorption spectrum, and is non-fluorescent [317]. Energy- transfer from excited monomer to the non-fluorescent dimer is characterized by a Forster distance of 57 A. The fluorescence excitation and emission spectra of BODIPY FL are given in Figure 56. Studies of a related BODIPY-labelled lipid in monolayers of 1-stearoy1-2-oleoyl- phosphatidylcholine (SOPC, 18:0, 18:1 cis-9) indicated that BODIPY does not self-aggregate. That is, the tagged-lipid did not segregate from the SOPC matrix. Accordingly, the concentration- dependent spectral shift of BODIPY-tagged lipid has found use in studying cellular uptake and lipid transport processes. BODIPY-tagged lipid has also been used to determine two-dimensional lipid concentration in liposomes after calibration in lipid monolayers [318]. Partitioning and aggregation of gangliosides (cell signaling modulators) within monolayers has been characterized by monitoring monomer/dimer emission from BODIPY-tagged ganglioside [319]. 7.3.4 Simultaneous Monolayer Compression and Epi-Fluorescence Microscopy Fluorescence imaging of the monolayer surface were made using the same Olympus IX70 inverted epi-fluorescence microscope described in the Chapter 4. The output of a 75W Xe arc lamp was directed to a filter cube to select a band of wavelengths specific to BODIPY fluorophore's absorption spectrum. Different emission filters were used depending on whether monomer or dimer 185 1 - 0.9 - 0.8 - 0.7 - c 0.6 - if( • 0.5 - 0.4 -cc 0.3 - 0.2 - 0.1 - 0 —Absorption — Emission 1 - 0.9 - 0.8 - 0.7 - 0.6 F, - 0.5 al - 0.4 Ce) - 0.3 0.2 0.1 0 400 420 440 460 480 500 520 540 560 580 600 620 Wavelength (nm) Figure 56 Fluorescence excitation and emission spectra of monomeric BODIPY FL fluorophore in methanol. Data for this spectra from Invitrogen-Molecular Probes. Dashed lines represent the limits of the spectral bandpass regions of the excitation (460-490 nm) and emission (515-550 nm) filters of the Olympus U-MWIBA filter cube. emission was collected. Monomer emission measurements used an Olympus U-MWIBA cube, with excitation band 460-490 nm, dichroic filter 505 nm and emission band 515-550 nm. For collection of red-shifted BODIPY dimer fluorescence we substituted this emission filter for a 590 nm long pass filter. The excitation light was focussed through the microscope objective onto the monolayer at the GIS, as detailed in Figure 57. As before, emitted fluorescent light was collected back through the objective, traversing the dichroic mirror and emission filter before reaching the digital camera. Imaging during Langmuir trough experiments used an Olympus LMPlanFl 20x objective with 0.4 numerical aperture and 12 mm working distance. Images were collected with 2x2 binning, 12-bit resolution, and encompassed an area of about 0.5 x 0.7 mm. Each pixel represents an area of 0.93 lim 2 , while diffraction-limited resolution is 0.79 I.Lni at 520 nm. Fluorescence images were typically collected every 5-10 s, with 100 ms exposure time. Simple image analysis to output the average fluorescence intensity over the whole image was performed using the DlPimage set of routines [222] 186 for MATLAB. A dark current image was subtracted from each fluorescence image captured, and the resultant image then divided by a long-exposure time fluorescence image of the GIS when no lipid layer was present. This division operation corrects for non-uniform illumination of the surface. The mean grey scale of the resultant images was taken to represent the average fluorescence intensity. A custom Labview program (vi) was used to control the trough area, measure the surface pressure, and interface with collection of fluorescence images. The trough manufacturer, NIMA, kindly provided their source code for Labview control of the trough; this was modified to facilitate in situ fluorescence imaging. As before, control of the digital camera was via dedicated SPOT RT software; the number of images in a sequence, exposure time, interval time, and other imaging parameters were set here. Each time an image file was saved to the computer, the Labview program would note the filename and most recent values of surface pressure, trough area and time in an output file. Surface pressure, area and time were also recorded in a second datafile, with data sampling at a faster rate (-10 Hz). 7.3.5 Simultaneous Monolayer Compression and Laser-Excited AmB Fluorescence The second series of in situ fluorescence experiments was made exploiting the fluorescence of AmB itself These results represent initial experiments on this system. The nature of our trough-based experiments, where AmB is present both at the surface and in the subphase, means that excitation via the epi-fluorescence arrangement cannot be used. To avoid excitation of AmB within the fluorescence collection path, a separate light source was used, shone onto the surface from above. The source used was a Blu-rayTM laser diode from the DVD drive of a Sony PlayStation 3 videogame console. The entire optical carriage of the DVD drive (part# KES-400A) is available through internet auction websites for —$100. These parts may represent units which fail quality- control protocols at the factory; the wavelength specification for Blu-ray devices is 405 nm, while the laser received showed peak output at 412 nm (5 nm full width at half maximum). The laser assembly also contains red and IR laser diodes as well as a photodiode for monitoring output power; these were not used. 187 0 1 M KCl  Xe Arc Lamp INIAARPoweagN',Uio • Surface Balance ••7 to Labview vi Barrier/Area Control Filter Cube Dichroic mirror 505 nm Abs 460-490 nm Em 515-550 nm or 590 nm LP ComputerSpot RT CCD Figure 57 Schematic of experimental setup for in situ fluorescence imaging of the monolayer on the Langmuir trough. A Labview program controls the barrier position and records the surface pressure at which fluorescence images were taken. The filter cube used varied depending if BODIPY monomer or dimer fluorescence was being measured. The 515-550 nm emission filter passes monomer fluorescence while the 590 nm long pass emission filter collects dimer fluorescence only. The microscope objective is located in the centre of the trough. The Wilhemy plate is located centrally between the barriers but close to the edge of the trough and does not interfere with spectroscopic measurement. The plate is shown off-centre here for clarity. 7.3.5.1 Characterization and Modification of the Blu-ray Laser Prior to use in experiments, the properties of the laser were investigated. A current-voltage relationship was recorded for the laser with simultaneous measurement of light output using a external photodiode detector. As seen in Figure 58, the on-set of lasing occurs when 22 mA current is applied. No engineering specifications were available, so we used the diode laser rule-of-thumb that the working current limit is —10 mA above the lasing threshold. Accordingly, the maximum current applied was 30 mA in order to avoid damaging the laser diode. The laser was kept attached to its heat-sink for ease of handling and thermal protection. A hand-held laser power meter (Coherent LaserCheck, Santa Clara, CA) showed 475 [LW at 26.5 mA and 850 p,W at 30.0 mA 188 applied DC. The frequency response of the laser output to an oscillating drive current was examined through the 0.1 Hz to 100 KHz range. The response curve showed the highest fidelity within the region of 5 Hz to 50 KHz, with rapid decrease of signal strength outside this range. Long term (multi-hour) stability tests of the laser output showed minimal variation in signal intensity. Laser diodes output a raw beam which is typically wedge-shaped with —10° x 30° divergence. To simplify the optics necessary to produce a useful beam, the laser diode was kept attached to its optical sled, composed of a holographic grating, two beamsplitters and a lens, as shown in Figure 59. The holographic grating and beamsplitters are remnants of the DVD drive application and served no useful purpose here. The DVD drive objective lens was removed. A detailed break-down of all the components of the optical carriage is available online [320]. Three holes were drilled through the aluminum body of the optical carriage to allow the carriage to be mounted in a Thorlabs (Newton, NJ) 1" optical cage system. A quartz piano-convex lens (catalogue #A410020, ESCO, Oak Ridge, NJ) was used to converge the laser output. The laser output was noted to form a concentrated spot, with four less-bright beams located away from the central spot. A pinhole was introduced into the beam path and adjusted to eliminate the outer beams. 7.3.5.2 Optical Arrangement and Signal Processing Experiments were made with the laser operating in AC mode to allow a frequency lock-in technique to be used on the detected signal, improving signal-to-noise ratio. A DC bias of 26.5 mA was applied to the laser diode, supplemented with an AC signal of 3.4 mA rms amplitude at 1 KHz frequency. A lock-in amplifier (SRS model 830, Sunnyvale, CA) generated the AC signal, which was fed to a precision current source (ILX Lightwave 3207B, Bozeman, MT) which served as laser diode driver. As before, the Langmuir trough was enclosed in a plexiglass box above the microscope, but for these experiments, a hole was made in the back of the box, allowing the laser beam to illuminate the surface without distortion. The Wilhemy plate was positioned off-centre in the trough to avoid blocking the beam. The beam was directed onto the GAS surface of the Langmuir trough such that the beam area coincided with the collection area of the microscope objective at a fixed height, shown schematically in Figure 60. In order to have these to spots coincide with the height of the GAS surface, the trough was filled completely with the subphase solution and then 189 4.5 - 4.0 - 3.5 - 3.0 - > Es. 2.5 - CD 2.0 -0a_ 1.5 - 1.0 - 0.5 - 0.0 0 10^20 Current / mA 30 Figure 58 Applied direct current-voltage characteristic for Blu-ray laser diode with simultaneous light output measurement. Solid curve represents potential, while dotted curve gives light output. Onset of lasing occurs near 22 mA applied current. Good linearity of light output with current is observed after lasing begins. Light intensity is measured as photodiode output and is in arbitrary units. 190 Beams • litters^Mirror Laser Diode Collimating lenses Figure 59 Laser and optics portion of Blu-ray DVD drive carriage, as used. The laser diode is located at left and the beam shines towards the upper right. A holographic grating is located immediately in front of the laser diode before the beamsplitters. The mirror reflects the light output in a direction into the page. The objective lens normally is mounted on the opposite side, below the mirror, but was removed before use. solution removed until the microscope focal point was at the surface. The lipid layer was deposited after the solution level was brought to the correct height. The laser beam was oriented at 59° to the surface normal. At this angle of incidence, the transmitted beam is approximately 40° from the surface normal. Although the light is bent towards the microscope objective, the numerical aperture of the objective. (10x, NA 0.3) is such that the transmitted laser beam is not collected. Numerical aperture and the half-angle of the objective's acceptance cone, 0, are related by: NA = n sin 8 (46) where n is the refractive index of the medium. Solution of this equation using the refractive index of water shows that the light collected must originate within a cone of half-angle 13° to the surface normal, and the refracted laser beam at 40° misses the microscope objective. 191 Fluorescence emitted from the surface layer was collected through the microscope as before. A filter cube consisting of a 505 nm dichroic mirror and a 590 nm long pass emission filter was used. AmB monomer fluoresces in the range of 500 - 650 nm, with peak emission near 560 nm. The emission filter used was thus not ideal, but was used because it was readily available. The fluorescence light was directed to a photomultiplier tube (PMT; Model 77348, Newport-Oriel, Irvine, CA). The collimated output from the microscope was focussed onto the PMT area using a 5 cm focal length lens. The PMT was biased with a 600 V DC supply (SRS model PS310, Sunnyvale, CA). The PMT signal was fed to a current-to-voltage amplifier (EG&G PAR model 5182, Oak Ridge, TN) operating at a gain of 10 -' A/V. The AC output of the amplifier was passed to the lock-in amplifier, and the magnitude of the AC response at 1 KHz frequency was the measured signal. 7.3.6 Experiments Four different experiments were made measuring AmB fluorescence within a monolayer. 1) A mixed monolayer of AmB and DOPC was formed on the surface and fluorescence intensity measured as the layer was compressed and expanded. This experiment proves our measurement is sensitive to the small amount of AmB present in a monolayer. 2) FZ was repetitively injected into the subphase with no lipid present at the surface. Gibbs monolayers of AmB and deoxycholate can be expected to form at the interface. Fluorescence intensity was measured with increasing bulk FZ concentration. These experiments probed the surface sensitivity of the measured fluorescence when AmB was also present in solution. 3) A dilute DOPC monolayer was deposited on the interface and FZ injected behind the trough barriers, after which fluorescence intensity was measured with time. This replicates the procedure used in the BODIPY-doped monolayer experiments, and gauges AmB insertion into a dilute lipid film. 4) A lipid monolayer was deposited to the GAS and compressed to a film pressure of 30 mN/m before injecting FZ into the subphase. Fluorescence intensity was monitored vs. time to examine AmB insertion into the layer. This film pressure coincides with the effective pressure thought to exist in cell membranes, and thus these experiments correlate best with biological conditions of AmB administration. The subphase for all experiments was 0.1 M KC1 solution. 192 Barrier/Area Control Precision Current Source Filter Cube Dichroic mirror 505 nm Em 590 nm long pass Amplified Pia output Demodulated signal Computer and DAQ 590 Figure 60 Experimental setup for in situ measurement of AmB fluorescence from a monolayer on Langmuir trough. Laser beam strikes surface at 59° to surface normal, illuminating the focal point of the microscope objective. The height of the GAS interface is adjusted such that it is at the focal distance of the objective. The Wilhemy plate is hung off-centre with respect to the barriers to avoid blocking the path of the laser. The laser intensity is modulated within the linear region of the current-light output curve determined earlier. Collection of fluorescence is through a filter cube to a photomultiplier tube (PMT). A Labview program controlled barrier position and simultaneously recorded surface pressure and the amplified demodulated fluorescence signal. The inset at top left shows the geometry of the refracted laser beam relative to the acceptance cone of the objective. 7.4 Results and Discussion 7.4.1 Compression Isotherms and Fluorescence Imaging with Incorporated BODIPY A compression isotherm for the 3 mol% BODIPY FL DHPE/DOPC monolayer is presented in the lower panel of Figure 61. The isotherm shows a smooth, gradual increase in surface pressure as the trough area is decreased, characteristic of a liquid-expanded monolayer state. At room temperature, 193 the unsaturated bond in DOPC's acyl chains results in a fluid monolayer. The initial surface pressure was 2 mN/m. We measured the collapse pressure of our mixed monolayer to be 45 mN/m, in agreement with the 45-50 mN/m range cited in literature for pure DOPC [321, 322], suggesting the incorporated dye does not strongly modify intermolecular interaction. The measured collapse pressure is the maximum pressure to which the monolayer can be compressed without detectable expulsion of molecules from the layer. As such it is not a thermodynamic quantity and recorded values vary, principally with the rate of compression. Shown in the upper panel of Figure 61 is the mean grey scale of the fluorescence images of the monolayer during compression. In all cases, the images themselves had very uniform fluorescence and showed no discernable features. Accordingly, the images are not shown. As the monolayer is compressed, the fluorescence intensity decreases. This decrease is due to increasing ground state dimer formation as the BODIPY molecules are moved closer together. The non-fluorescent form of dimer will clearly not be detected, but the fluorescent dimer is also not detected because its emission is red-shifted outside the wavelength band accepted by our emission filter. Thus both forms appear as non-fluorescent in our setup. Measurements were also made collecting dimer rather than monomer fluorescence, and showed increasing dimer fluorescence as the monolayer was compressed, thus confirming our explanation. At the collapse pressure, the fluorescence intensity is approximately one-third its pre-compression value. Compression isotherms of the fluorophore-doped DOPC monolayer in contact with AmB formulations or control solution are also given in Figure 61. The compression isotherm with deoxycholate is similar to that of the mixed fluorophore/DOPC monolayer alone. The collapse pressure is decreased slightly, to 43 mN/m. When AmB is present, irrespective of formulation, more significant changes in the isotherm are observed. ABLC, the lipid-complexed form of AmB, does not produce any significant change in the shape of the isotherm, and the collapse pressure is only slightly lower at 43 mN/m. An increase in the initial surface pressure to —6 mN/m is observed. This may be understood as an increase in surface pressure at a fixed area per molecule on the surface, or as a decrease in area per molecule at a fixed surface pressure, or as a combination of the two. Generally-speaking, interaction with the lipid headgroups just below the surface leads to a change 194 in surface pressure but not area, while intercalation into the lipid monolayer leads to significant changes in area but not necessarily in surface pressure. It is important to note here that the x-axis values are in A2 per deposited molecule. This is the available area of the trough divided by the number of molecules in the spread Langmuir monolayer. Any increase in the number of surface molecules due to the surface activity of ABLC is not reflected in the x-axis values, meaning this value will be different from the true surface area available per molecule. Additionally, it may not only be AmB influencing the surface; any of the components of ABLC (AmB, DMPC, DMPG) or complexes of AmB/lipid can be expected to segregate to the surface. FZ and HTFZ produce more dramatic changes in the isotherms. Both have high initial surface pressure, above 7 mN/m. Again, this may be understood as a change in surface pressure or a change in true surface area per molecule. With these formulations, any of AmB, deoxycholate, or complexes of the two could segregate to the surface. The recorded 7 mN/m 'overpressure' is intermediate between the An values of 4 and 11 mN/m determined by Miriones [272] for pure AmB penetration into a DOPC monolayer when present in the subphase at concentrations 10r 7 M and 2 x10' M, respectively. The AmB concentration here is 4 x10' M. Both isotherms show a greater slope at low and moderate surface pressures than the isotherm of the dye/DOPC monolayer alone. The collapse pressures remain near 45 mN/m. For all monolayers examined, the fluorescence intensity decreases during compression. In the presence of ABLC or deoxycholate the layers have intensity vs. compression curves nearly identical to the DOPC layer alone. In each case, the minimum fluorescence is observed on maximum compression. With deoxycholate, a sharp dip in intensity is observed near 110 A2 area per deposited molecule. At this point, the surface pressure of the layer is still quite low and this dip likely represents a monolayer-free region of surface passing above the microscope objective. The fluorescence intensity measured with FZ or HTFZ in the subphase show steeper decreases as the layers are compressed. With HTFZ, the minimum intensity reached is similar to that with the other formulations, but with FZ, the lowest recorded intensity is both lower and does not occur at maximum compression. The significance of these changes is uncertain. 195 Similar measurements were undertaken with sterol-containing monolayers. Figure 62 presents the compression isotherms and corresponding fluorescence intensity measurements for the dye/cholesterol/DOPC monolayer. Comparison of the isotherm with the sterol-free analogue in Figure 61 shows clearly the well-known condensing effect of cholesterol. At large area per deposited molecule the surface pressure is zero, while the collapse pressure of 45 mN/m is reached at a lower area per deposited molecule. Addition of FZ or HTFZ into the subphase results in an initial increased surface pressure, with FZ slightly higher than HTFZ at about 7 mN/m. These increases can be compared to the —10 and 33 mN/m increases that Miriones [272] observed for pure AmB (10 -7 M and 2 x10 -5 M, respectively) injected below a 50 mol% cholesterol/DOPC monolayer. The presence of deoxycholate in the FZ and HTFZ formulations acts to increase AmB solubility and its binding to AmB must act to decrease the surface activity relative to AmB alone. The isotherm measured with FZ present is surprisingly similar to that measured without FZ, appearing roughly parallel, and maintaining a similar collapse pressure. When HTFZ is in the subphase, the isotherm has a greater slope initially and reaches a lower collapse pressure of about 41 mN/m. The fluorescence intensity during compression is broadly similar for cholesterol-containing monolayer with and without FZ or HTFZ. The minimum intensity occurs with FZ. Figure 63 shows the corresponding results when the monolayer contained ergosterol instead of cholesterol. The compression isotherm shows the expected condensing effect of ergosterol, and reaches a surface pressure of 45 mN/m before collapse. The increase in initial surface pressure when FZ or HTFZ was present in the subphase is less than that observed with cholesterol. This result is in keeping with the results of Miriones [272], who noted the largest increases in surface pressure for cholesterol-containing monolayers. The FZ and HTFZ isotherms are nearly identical at low surface pressure, with HTFZ diverging to a steeper slope before reaching a lower collapse pressure of 40 mN/m. The collapse pressure with FZ present is just under 45 mN/m. The fluorescence intensity from these monolayers decreases upon compression, as before. The three curves are broadly similar. The FZ and HTFZ influenced monolayers show lower intensity when surface pressure is moderate, but the intensity at the end of compression is in any case similar. The initial intensity of the monolayer without added AmB formulation shows very low fluorescence, 196 corresponding to an 'empty' region of surface. This intensity increases rapidly upon the initial compressive action which brings together any 'islands' of lipid on the surface. The compression isotherm data collected here on the influence of AmB formulations on sterol- containing and sterol-free lipid monolayers is important with respect to the earlier electrochemical investigations of AmB interaction with the lipid monolayer adsorbed on the Hg electrode. In those experiments, a lipid monolayer at the ESP was deposited to the electrode. When the influence of AmB formulations was being examined, the monolayer was deposited onto the electrode after AmB injection to the subphase. The influence of the AmB formulation on the ESP of the monolayer was unknown. In this work, we observed that the collapse pressure of the DOPC monolayer in the presence of AmB formulations maintained the same value as when they were absent. Consideration of the images and intensity of BODIPY fluorescence emission during monolayer compression, whether with incorporated sterol or subphase AmB formulation, suggests this technique is not well-suited for probing AmB-lipid layer interactions. The fluorescence images were always very uniform, with no features that could be identified as possible regions of AmB penetration. The measured intensity was sensitive to BODIPY dimerization induced by compression, but this dimerization process appears insensitive to AmB influence. 7.4.2 Excitation of Surface AmB Our method of examining AmB-lipid layer interaction by monitoring BODIPY fluorescence was unsuccessful, since the BODIPY did not report on AmB penetration directly or indirectly. Nonetheless, we believed that in situ spectroscopy could be a very useful tool for studying AmB- lipid interaction. In seeking a probe to faithfully report on AmB penetration, we attempted to make use of AmB's intrinsic, albeit weak, fluorescence. We used laser illumination from above the Langmuir trough, and used the microscope's collection optics to direct the fluorescence emission to a sensitive detector. As this was an initial attempt to measure surface AmB fluorescence, we conducted a series of four experiments to survey the possibilities of the technique. 197 1.25 1 Cl) 0.75 c 0.50its 2 0.25 - DOPC +Deoxycholate — - — +FZ +HTFZ ^ +ABLC 0^1111 ^ 40 50 60 70 80 50 45 - 90^100 110 120 130 s'\ ^ 40 -^ \ '‘‘^\\E-2 35 - \ --- 30 -^ \ X \ \.2 25 - \ \\N• 20 — DOPC +Deoxycholate 'c • 15 - — - — +FZ cn -- +HTFZ 10 ^ +ABLC 5 0 40^50^60^70^80^90 100 110 120 130 Area (A2) per Deposited Molecule Figure 61 Compression isotherms and fluorescence intensity for 3 mol% BODIPY FL DHPE/DOPC monolayer in the presence or absence of added AmB formulation in the subphase. Formulation concentration was 80 %TD, added after deposition of the monolayer. Fluorescence images were subjected to a background correction process before having the resultant mean intensity calculated. Measurements were made at room temperature. Calculation of area per deposited molecule includes both DOPC and fluorophore; a total of 6.5 x 10 15 molecules were deposited to the surface. 198 IDOPC/Cholesterol +FZ 1 - — • — +HTFZ a) ov's (/) 0.75 coc 0.5 -a) 2 0.25 - 0 40^50^60^70^80^90 100 110 120 130 1.25 50 ^ 45 40 - E^- -2 35 - E -F13 30 - v 25 - a) 20 (.) 'S 15 - U) 10 - DOPC/Cholesterol +FZ — • — +HTFZ N N 0 ^ 40^50^60^70^80^90 100 110 120 130 Area (A2) per Deposited Molecule Figure 62 Compression isotherms and fluorescence intensity for 3 mol% BODIPY FL DHPE, 9.6 mol% cholesterol/DOPC monolayer in the presence or absence of added AmB formulation in the subphase. Formulation concentration was 80 %TD, added after deposition of the monolayer. Fluorescence images were subjected to a background correction process before having the resultant mean intensity calculated. Measurements were made at room temperature. Calculation of area per deposited molecule includes DOPC, cholesterol and fluorophore; a total of 6.5 x10' 5 molecules were deposited to the surface. 199 I^,^I DOPC/Ergosterol - +FZ a) Cu a0.75 a) (.D co 0.5 -a) 0.25 - - — +HTFZ 1.25 0^'^1^- 40 50 60 70^80^90 100 110 120 130 • I^I 50 45 - 40 HE • 35 30 n 25 -m • 20 - (.) • 15 - (1) 10 - 5- 0 40 DOPC/Ergosterol +FZ — - — +HTFZ 50^60 ^ 70^80^90 100 110 120 130 Area (A2) per Deposited Molecule Figure 63 Compression isotherms and fluorescence intensity for 3 mol% BODIPY FL DHPE, 9.4 mol% ergosterol/DOPC monolayer in the presence or absence of added AmB formulation in the subphase. Formulation concentration was 80 %TD, added after deposition of the monolayer. Fluorescence images were subjected to a background correction process before having the resultant mean intensity calculated. Measurements were made at room temperature. Calculation of area per deposited molecule includes DOPC, ergosterol and fluorophore; a total of 6.5 x 10 15 molecules were deposited to the surface. 200 7.4.2.1 Surface Pressure and Fluorescence of 20 mol% AmB/DOPC Mixed Monolayer A mixed monolayer of AmB and DOPC deposited to the surface of the Langmuir trough was selected as a test system since it has an AmB-free subphase. Figure 64 presents the AmB fluorescence emission intensity, monolayer surface pressure and the area available to the mixed monolayer during the measurements. The AmB/DOPC solution was deposited to the surface shortly before time zero. Spreading solvent was allowed to evaporate and the monolayer to equilibrate for —2500 s with the trough barriers held in the fully-open position. During this time, the surface pressure decreased slightly, suggesting either reorganization into a slightly different state or some loss of material from the interface. The AmB fluorescence shows an initial small peak-like feature, which was observed even when an AmB-free DOPC layer was examined. We believe that this is a result of a transient change in light scattering from the surface, related to initial monolayer spreading. In each case, the measured fluorescence intensity returned to its original value. The fluorescence intensity then climbed slowly to reach a value that is approximately 5x the background intensity. When the compression isotherm was started at —2500 s, the surface pressure began to increase, and ultimately reached a collapse pressure of 45 mN/m. Simultaneously, the fluorescence emission from AmB increased, peaking about 13 x higher than the background. When the monolayer reached the collapse pressure, movement of the barriers was stopped, and the trough area held fixed. The surface pressure decreased as the monolayer relaxed. During this time, a rapid decrease and then increase in fluorescence intensity was observed. With high packing density, small variations in the state of AmB in the monolayer may lead to relatively large changes in fluorescence. The exact nature of this decrease and increase is not yet known. When the trough barriers were opened to expand the monolayer, the surface pressure dropped to —2 mN/m, below its starting value. The fluorescence intensity measured during the expansion also showed a decrease, in keeping with fewer AmB molecules in the excitation beam as the surface packing density decreased. With the barriers kept fully open, the fluorescence intensity returned to a value similar to that measured before compression. Based on the correlation of increased surface pressure and increased fluorescence intensity upon compression and the analogous decreases on monolayer expansion, there can be no question of the surface sensitive nature of our AmB fluorescence measurements. This serves as a powerful proof- 201 of-concept experiment, showing that in situ monitoring of AmB fluorescence within a monolayer is possible. This method provides selective measurement of AmB monomers since monomer and aggregates are excited with different wavelengths of light. If the energetics of pore formation or other forms of in-layer self-aggregation can be varied, perhaps through temperature changes, it might be possible to use this technique to monitor the process. 7.4.2.2 AmB Penetration into a Dilute DOPC Monolayer We applied this new method of measuring AmB fluorescence to study its penetration into a dilute lipid layer, according to a protocol similar to that used with the BODIPY-labelled monolayer. With the barriers of the Langmuir trough in the fully open position, a monolayer of DOPC was formed on the surface. This monolayer was characterized by a low surface pressure (-2 mN/m) and a large mean area per molecule, thus characterizing it as dilute. A nine aliquot of FZ solution was injected behind each barrier to give subphase AmB concentration 80% TD (0.41.1M), and the surface pressure and fluorescence measured over time, as presented in Figure 65. Immediately after injection, the surface pressure increased to —6 mN/m, and reached a maximum of about 7 mN/m within a minute. The fluorescence of AmB increases more slowly, reaching a plateau after a few minutes. The difference between the time scales of these two increases raises an important point. The fluorescence intensity is measured at a point in the middle of the Langmuir trough, and the signal is sensitive to local AmB concentration in the interfacial region. On the other hand, the surface pressure is measured slightly off-centre, but records a value representative of the monolayer as a whole, regardless if it is DOPC, AmB, or deoxycholate molecules present on the surface near the Wilhemy plate. Accordingly, the surface pressure may show an increase due to AmB insertion to the layer before this insertion is detected with fluorescence. With continued measurement beyond a few minutes, the fluorescence intensity holds approximately steady, while the surface pressure declines slightly. The fact that the fluorescence intensity approaches its final value within a few minutes suggests that the 20 minute equilibration period used in our studies of AmB influence on BODIPY-containing lipid layers was a sufficient waiting time. 202 7.4.2.3 Gibbs Monolayer Formed from FZ FZ was injected into the subphase of the Langmuir trough and allowed to form a Gibbs monolayer at the surface. A Gibbs monolayer is characterized by thermodynamic equilibrium of the surfactant in the monolayer with surfactant in the bulk subphase. In this case, we may expect the monolayer to be composed of both AmB and deoxycholate, in proportion related to the strength of their surface activity. With equilibrium between bulk and surface, increased bulk concentration leads to higher surface concentration. As the surface concentration increases, the surface tension decreases according to the Gibbs isotherm, and surface pressure increases. Figure 66 shows the changes in surface pressure and AmB fluorescence as the Gibbs monolayer is formed and the subphase concentration increased. With the Langmuir trough barriers set to a partly-open position, initial measurement of fluorescence and surface pressure was made with only 0.1 M KC1 in the trough. The fluorescence intensity is low and constant, while the surface pressure was also low. At time —700 s, an aliquot of FZ was added to the subphase, with portions of the aliquot distributed around in the trough but not in the immediate vicinity of the laser excitation spot. When the FZ was injected, the surface pressure rapidly increased, peaking at about 10 mN/m before relaxing to 6.5 mN/m. Measurement of fluorescence was suspended momentarily to make the injection, but immediately afterwards, the fluorescence intensity increased over a period of a few minutes before reaching a rough plateau. At the 1600 s mark, a second smaller aliquot of FZ was injected to the subphase, again distributed around in the trough area. The surface pressure increased immediately after injection and decayed back to 7.2 mN/m. This slight increase in surface pressure with a relatively large change in bulk concentration is in agreement with the Gibbs isotherm, which shows an asymptotic approach to a limiting surface pressure. Interestingly, the fluorescence intensity was slightly decreased from the value recorded after the first injection. At about 2300 s, another small aliquot of FZ was added. The surface pressure followed the now familiar pattern of a rapid increase and slow decay, finally reaching 7.2 mN/m. On the other hand, the AmB fluorescence shows unexpected behaviour, a nearly linear increase with time. The cause of this increase is not clear. 203 On the hypothesis that this increase might be due to fluorescence from subphase AmB diffusing in the region of laser excitation and fluorescence collection, a final small aliquot of FZ solution was added to the subphase. This addition of FZ was made just under the surface at a spot above the microscope objective but below the laser spot. If our fluorescence measurement was sensitive to bulk and surface AmB in this region, it was expected that a locally high AmB concentration in the subphase would give a significant increase in fluorescence intensity. After this last injection at 3400 s, the surface pressure increased slightly and then decayed back to 7.2 mN/m. The fluorescence intensity, however, showed no significant change. The insensitivity ofthe fluorescence signal to high AmB concentration in the subphase near the excitation region suggests that our setup has good surface sensitivity and that the signals we observe must truly be related to changes in the monolayer. 7.4.2.4 AmB Penetration into a DOPC Monolayer at Biologically-Relevant Surface Pressure The final experiment undertaken with our new technique of in situ excitation of AmB was to probe AmB's penetration into a phospholipid monolayer held at the biologically-relevant surface pressure of 30 mN/m; a monolayer at 30 mN/m is generally regarded as having properties similar to that of one leaflet of a cell bilayer. Due to the non-zero depth of field of the microscope objective, the fluorescence signal measured for AmB penetration may originate from AmB intercalating to the monolayer, or from AmB associated with the polar lipid headgroups in a shallow sub-surface position. Both modes of interaction are consistent with the concept of penetration as used in the literature. Figure 67 shows the interaction of FZ with a DOPC monolayer under these conditions. Initially, a dilute DOPC monolayer was formed on the Langmuir trough with the barriers in the fully-open position and the layer allowed to equilibrate for almost 800 s. During this time, a low background fluorescence intensity was recorded. After equilibration, the DOPC monolayer was compressed rapidly to 30 mN/m surface pressure, and then FZ solution injected to the subphase on the outside of the barriers. For the duration of the measurement, the surface pressure was held at 30 mN/m by adjusting the area between the barriers. The trough area available to the monolayer is shown in the lower panel of Figure 67. The fluorescence intensity remains low for the first 600 s after injection before increasing in a bimodal manner over the next —1500 s. At long times after 204 .cn c 15 - 12 - --- 9 _ 6 - a) t 3 - 0 0 z^- E 40 - 30 - (eki 20 - 0 10 - co 'c 0 ^ w 0 90 - 80 - o70 -70 ET) 60 - ¢ 50- 40 0 1000 ^ 2000 ^ 3000^4000 ^ 5000 ^ 6000 Time / s 1000^2000^3000^4000^5000^6000 Time / s 1 000 ^ 2000 ^ 3000^4000 ^ 5000 ^ 6000 Time / s Figure 64 AmB fluorescence intensity and surface pressure of 20 mol% AmB/DOPC mixed monolayer on Langmuir trough. The monolayer was formed with the trough barriers in the fully open position (time zero), and allowed to equilibrate while measuring fluorescence and surface pressure. At time —2500 s, a compression isotherm was begun, and the trough area held fixed for —400 s once the collapse pressure reached. The trough barriers were then reopened, and an expansion isotherm measured. The barriers were then held fully open for the duration of the experiment. Units of fluorescence intensity are arbitrary. 205 250 500^750 Time / s 100 0 1 1250 1i 250 500^750 Time / s 1000 1250 250 I 500^750 1000 1 1250 U) C80 - < 0- 6 - 40-U) 20 0 LL^0 Ez8 ,- E 90 - N 88 - E 86 - 84 - < 82 - 80 0 Time / s Figure 65 Fluorescence intensity and surface pressure of DOPC monolayer after FZ was injected behind trough barriers in the fully-open position. The barriers were held open throughout the experiment. The injection was made just prior to time zero. Units of fluorescence intensity are arbitrary. Penetration is seen to be mostly complete within —400 s, as shown by the fluorescence intensity having almost reached its maximum. 206 AmB injection, a rough plateau in fluorescence intensity is observed, being about an order of magnitude stronger than the background signal. The time scale of the changes we observe here may be compared to the results of Tancrêde and Barwicz, who studied the surface pressure increase of lipid and lipid/sterol monolayers as a function of time after AmB solution was injected to the subphase [75]. Their measurements when pure AmB was introduced below a DOPC monolayer at 30 mN/m showed that after 3000 s the surface pressure is at or near a steady-state value if the AmB subphase concentration remains 6.5 LIM or less. The concentration of AmB in our experiment was 0.4 MM, and our fluorescence intensity reaches an approximate steady-state value after a similar length of time. Tighter lipid packing means these times are naturally longer than the < 20 minutes it took for AmB penetration into our dilute DOPC film. An additional point of interest is their calculation based on spectroscopic measurements of AmB subphase concentration, that approximately 15% of the dissolved AmB left the bulk subphase to interact with the DOPC monolayer. While we do not have an analogous measure to compare, and variations due to the differing form of AmB in the subphase (pure vs. FZ) are likely to exist, their measure nonetheless provides context for our experiment. 207 U) c200—— < 150 - 100 -u) N 0 LL^0 10 - E 8 - Fo 6 - to 4 -a_ 2 -U 0 ^ co^0 64 - 50 - 1.9X10-8mol AmB injected to subphase 56 0.9X 1 0 -8 mol^0.9X10-8 mol AmB injected^AmB injected to subphase to subphase viv^1.3 0.2X 1 0 -8 mol AmB injected to subphase 2000 Time /s 3000 4000 2000 Time /s 3000 4000 1 i I 1000 1000 NE 62 - 60 .4e. 58 - 1 0^1000^2000^3000^4000 Time /s Figure 66 AmB fluorescence intensity and surface pressure for Gibbs monolayers formed by injection of FZ to the subphase. The trough barriers were set to a partly-open position and held fixed for the duration of the experiment. Injections of AmB were 18 p.1 initially, then two aliquots of nine and one final addition of two pi, all of 0.95 mg/ml AmB as FZ. The final aliquot was injected above the microscope objective, just below the laser illumination spot on the surface. Units of fluorescence intensity are arbitrary. 208 z E 30 a) .7) 20u) °- 10a)U 90 80 E 70 al E12 60 50 u)^1.9X108mo1.-i-_- AmB injected= 13 60 -^to subphase •=t — cc f)— a) 20 - C --,, 40 - 6 0m U- E 0 _ - 1000 2000^3000 Time / s 4000 5000 - 0 - 1000 2000^3000 Time / s 4000 1 5000 - - - - 0 f 1000 2000^3000 4000 I 5000 Time / s Figure 67 AmB fluorescence intensity and monolayer surface pressure for FZ penetration into a DOPC monolayer at a biologically-relevant surface pressure. A dilute DOPC monolayer was formed on the trough surface and allowed to equilibrate before compressing it to 30 mN/m. Once at this surface pressure, FZ was injected to the subphase behind the barriers. The surface pressure was maintained at 30 mN/m for the duration of the measurement. Units of fluorescence intensity are arbitrary. 209 7.5 Conclusions The Langmuir trough based experiments of AmB-lipid interaction described in this chapter allow us to conclude two principal points. First, we may conclude that the influence of AmB on floating BODIPY/DOPC monolayers does not produce visually distinguishable regions in the layer which could be interpreted as AmB pores or other form of AmB penetration. Additionally, the variation in BODIPY fluorescence emission intensity during monolayer compression does not vary significantly with AmB presence and this method does not hold promise for monitoring AmB-lipid layer interaction. Second, our initial attempts at directly exciting AmB fluorescence within a monolayer as a means for monitoring penetration and interaction were successful. To our knowledge, this represents the initial use of AmB fluorescence to study its interaction with a lipid matrix; previous studies simply documented AmB's fluorescent properties. We demonstrated the use of the technique for AmB in the surface region both as a mixed monolayer with DOPC and as mixed Langmuir-Gibbs monolayers where AmB was also present in the subphase. The changes in fluorescence and surface pressure as AmB penetrated dilute or well-packed lipid monolayers were generally correlated. In the application documented here, we excited AmB monomer, but it should in principle be equally possible to excite and monitor AmB aggregates. The higher quantum yield of AmB dimer should help offset the lesser intensity of their presumably lower concentration. It is anticipated that monitoring of AmB penetration into sterol-containing layers could be done without any changes to the setup. Although more work needs to be done to fully commission this technique, we believe that this lays the groundwork for a new and highly-sensitive method of studying AmB - model membrane interaction. 210 8 CONCLUSIONS AND FUTURE WORK This work has been built around two themes, those of studying the mechanism of the antifungal drug AmB and the development of new tools for the study of its mechanism. These two themes are linked. In the first case, we are fundamentally attempting to measure the interaction between the drug and the lipid membrane under varying conditions. For the second theme, we are fundamentally building new methodologies of measuring lipid disruption or variations in lipid behaviour in model membrane systems. Since interaction of the drug must involve changes in membrane order or organization, the two themes converge into the question, How can changes in lipid behaviour within the membrane be characterized? This is a very important question, given that so many drugs have as their site of action the cell membrane, for example through physical modification of the membrane, interaction with receptors, or blocking ion channels. Accordingly, the techniques and methods of this thesis may be applicable not only to the study of AmB but also to the study of other drugs. With regard to AmB specifically, some of the methods we used to probe the interaction between AmB formulations and the DOPC monolayer were successful and some were not. The successful methods have given us new information and characterization of the interactions, but it has been difficult to draw specific conclusions about AmB's mechanism of action. This is reasonable as it is still relatively early-on in our exploratory study of this very complicated drug. We believe that further studies using our techniques will allow more detailed understanding of AmB's mechanism. This conclusion begins with a short summary of our results from each chapter, with an emphasis on the significance of the findings. The broad implications of our work to the study of AmB's mechanism of action are then considered. Suggestions for future work to further extend our mechanistic studies and to further develop tools for studying AmB are given last. In Chapter 4 we used a lipid monolayer supported on a Hg electrode to probe the AmB-membrane interaction of different formulations of AmB. We found differing behaviour for three AmB formulations, in general correlation to their in vivo toxicity. Importantly, this establishes our 211 supported monolayer model as suitable for the study of differences between AmB formulations. Upon introduction of AmB to the model layer, monolayer permeability increased. The finding of increased porosity in our sterol-free system has a direct implication to the sterol-hypothesis of AmB action, although we cannot be sure that the increased porosity is due to pores with specific channel- like structure. We observed significantly different permeabilizing behaviour between FZ and HTFZ, which is significant because it marks an additional piece of the puzzle of why their toxicities differ when their compositions are similar. At this point in the study of HTFZ, any differentiation between it and FZ is important. The in situ fluorescence microscopy work described in Chapter 5 was so far unsuccessful as a method for following AmB interaction, but did advance our knowledge of the Hg-supported monolayer system. The inclusion of a low concentration of fluorophore into the monolayer allowed monitoring of changes in lipid layer organization. We found that the nature of the potential-induced phase transitions is such that the fluorophore reversibly moves closer to the electrode surface. This finding gives us better understanding of our underlying model system. We attempted to use fluorescence microscopy to monitor the interaction of AmB with the monolayer, but found the fluorescence response was dominated by changes in potential rather than any influence of AmB. In Chapter 6 we determined that the fluorescence of formulated AmB is similar to that of pure AmB. It was confirmed that AmB fluorescence varies with aggregation state. Together, these results mean that we may proceed with fluorescence studies of formulated AmB in confidence and that fluorescence provides a method for selective monitoring of AmB according to aggregation state. Additionally, our finding that FZ and HTFZ have different dual fluorescence properties means that they must hold AmB in different chemical environments. As mentioned above, determination of the differences between these two formulations is needed before their in vivo toxicity can be rationalized. We exploited our newfound knowledge of AmB fluorescence in Chapter 7, to measure fluorescence from monolayer AmB, using the laser source of a consumer electronics product as an inexpensive external light source. Our success in detecting monolayer-localized AmB, discriminating against 212 fluorescence from subphase AmB, and following AmB penetration into the lipid monolayer is significant in that AmB's fluorescence has not previously been exploited in studies of its mechanism. Additionally, it shows our technique to be suitable for studies of formulated AmB, where the drug is present not only in the layer but also in the subphase and where species other than AmB may also segregate to the surface. We measured AmB penetration to monolayers of different state, establishing our technique viz-a-viz existing studies of penetration and documenting the versatility of our combined Langmuir trough-fluorescence technique. We have had some success in developing methods to study AmB, but we have yet to consider the implications of this work in the context of AmB's mechanism of action. We will begin with a very brief summary of the current understanding around AmB at the beginning of this project. In Figure 3 of Chapter 2, a schematic of AmB's different modes of action was given. It shows that each of the different aggregation states of AmB may interact with the membrane. This interaction may lead to the formation of AmB-complexes in the headgroup region of the membrane (`pre-pores'), or may lead to pores in the membrane. Depending on the aggregation state of the AmB present, this pore formation may occur in membranes containing cholesterol, ergosterol or in sterol-free membranes. Also in Chapter 2, we introduced the monomer, aggregated and super-aggregated forms of AmB; differentiation among the first two is possible with UV-vis spectroscopy, while we have shown in Chapter 6 that differentiation between the latter two can be made with fluorescence spectroscopy. The less toxic HTFZ, formed by heat-treatment of FZ to super-aggregate the AmB, represents a possible break-through in the development of an inexpensive, low-toxicity formulation of AmB. Much like the schematic of Figure 3, we may consider the nature of AmB interactions with the two model membrane systems around which this thesis work has centred, the electrode-supported DOPC monolayer and the floating DOPC monolayer on the Langmuir trough. The analogous picture is given in Figure 68. In this figure, we envisage five possible modes of AmB interaction with the monolayer: A) AmB may form pore structures within the layer that are made up of only AmB; B) AmB monomer may intercalate into the monolayer; C) Aggregates of AmB may interact with the lipid headgroups at the surface of the monolayer; D) AmB monomers may interact peripherally with the surface; and E) AmB may form pore structures that include sterols (if incorporated). This list 213 Floating mono layer Legend AmB  Sterol Figure 68 Schematic representation of possible modes of AmB interaction with the model membrane systems used in this work. Top: Interaction with lipid monolayer adsorbed to Hg drop electrode. Bottom: Interaction with floating lipid monolayer. A legend is given at right. Interaction modes are A: pore formation from only AmB molecules; B: insertion of AmB monomer into lipid layer; C: peripheral contact of AmB aggregate with lipid headgroups; D: peripheral contact of AmB monomer with lipid headgroups; and E: pore formation from both AmB and sterol molecules. No specific aggregate structure, pore stoichiometry or structure should be inferred. is not exhaustive, and other modes and variations on these modes are possible. These five modes apply equally to our supported or the floating monolayer models, although possibly differing slightly in detail. As the monolayer composition can be easily varied to incorporate cholesterol or ergosterol in either of these approaches, we include here the possible measurement of AmB/sterol complexes. We now consider which of these modes of action can be monitored or detected using either the electrochemical or fluorescence approaches described in this thesis. In the electrochemical experiments of AmB-induced changes in the sterol-free DOPC monolayer on Hg, we used measurements of both monolayer porosity and capacitance to characterize the interaction of different AmB formulations. The ability to measure permeability and calculate pore size (within limitations) is one of the highlights of this technique. Our method is sensitive to those AmB modes of action which involve incorporation to the monolayer such as modes A and B in 214 Figure 68. With cholesterol or ergosterol incorporated into the electrode-supported monolayer, the Tr reduction measurements of porosity should allow study of the mode of action shown as E. The capacitance measurements are also sensitive to intercalated AmB but are in addition sensitive to the peripheral modes of interaction shown as C and D. On the other hand, our Langmuir trough and fluorescence measurements of AmB interaction with floating monolayers have differing sensitivity to these modes. AmB fluorescence after excitation with our Blu-ray laser diode allows detection of AmB modes of action B and D in Figure 68. In overview then, it can be seen that our approaches offer the possibility to monitor AmB interaction in any of these modes. Since AmB's mechanism of action against fungal cells and the mechanism of its toxicity must be related to one or more of these modes, this work has direct implications to the further study of AmB mechanism. 8.1 Opportunities for Immediate Application To maximize the value of our sterol-free studies of AmB-membrane interaction, .a systematic series of experiments could be made using both the Langmuir trough-fluorescence setup and the supported-monolayer electrochemical approach. On the trough, the work would need to begin with a minor modification to include a mechanism for gentle stirring of the subphase. Then, the penetration of different AmB formulations to monolayers of varying composition (sterol-free, with cholesterol, or with ergosterol) could be undertaken in a dose-dependent manner. If high purity AmB is available, then fluorescence and trough studies of mixed AmB/lipid monolayers could also be done. To extend the electrochemical work to better be able to comment on the sterol-hypothesis, sterol-containing monolayers should be adsorbed to the Hg electrode and characterization of AmB interaction made as before. Indeed, it was always our intention to conduct these electrochemical experiments, but technical difficulties with our Hg electrodes prevented us from doing so. Accordingly, the first step towards that work would have to be the commissioning of a new hanging Hg drop electrode. Some preliminary work incorporating cholesterol in the lipid layer on Hg was done, and it appears likely that the method of using capacitance and porosity to characterize AmB interaction will be equally viable when sterol incorporated. 215 8.2 Opportunities for Additional Method Development With additional method development work, the power of the Langmuir trough-fluorescence setup might be increased to include detection of AmB dimer or aggregate localized at the monolayer. Our setup uses a Blu-ray laser diode to excite monomeric AmB, but it should in principle be possible to use a different light source to excite fluorescence from AmB dimer or higher-aggregates. To excite AmB dimer, a XeF exciplex laser at 351 nm would be suitable, while a 325 nm HeCd laser could be used to excite higher-aggregates. Both these lasers are commonly used for industrial applications and acquiring an affordable, low-power model might be difficult. However, mW power laser diodes at the 351 nm wavelength of dimer excitation are just beginning to be produced and may shortly be a viable option. Although the surface concentration of dimers or higher-aggregates could be lower than that of AmB monomer, higher quantum yield (at least for dimers, —66x that of monomer) suggests detection could still be possible. A rather new avenue of work could be the study of AmB interaction with lipid monolayers in the gel state as compared to in the liquid-crystalline state. It has been suggested that sterols need not be present in a membrane for AmB to induce permeability if the lipids of the membrane are in the gel-state [26]. We propose variable temperature experiments using our electrode-supported monolayer system and the floating monolayer in the Langmuir trough-fluorescence setup. A lipid with a gel-to-liquid crystalline transition higher than that of DOPC would need to be used; DMPC or SOPC are likely options. Temperature variation studies of DMPC monolayers adsorbed on Hg have a literature precedent [138]. The Langmuir trough has a built-in reservoir for temperature control purposes and our laboratory has a jacketed electrochemical cell, so the experiments are feasible from a technical standpoint. Since sterols are known to rigidify cell membranes above the transition temperature, this approach might be an excellent way to probe structural effects on AmB interaction. It has been suggested that the presence of sterol in a membrane acts to sensitize AmB action not by chemical interaction between AmB and sterol, but by sterol-induced modification of membrane properties. Measuring the interaction of AmB with gel-phase monolayers could represent the study of sterol-free layers in a state similar to the sterol-containing layers. This approach may also provide a perspective on the question of AmB influence on stressed vs. unstressed lipid layers. The monolayer in the liquid crystalline state is likely to be defect-free, while in the gel state defects 216 between different 'grains' or lipid patches may exist. The defective monolayer may mimic a stressed membrane. 8.3 Use of New Fluorescence Tools Measurements of fluorescence may represent a fruitful way to continue this work; fluorescence lifetime is sensitive to the environment around the fluorophore. The ISS K2 fluorometer used in Chapter 6 has the capability to measure fluorescence lifetimes as short as 1 ns. If AmB's fluorescence lifetime is longer than this, studies of AmB lifetime in FZ and HTFZ as a function of aggregation state could be carried out, and would be complimentary to our work in Chapter 6. AmB's binding to sterols and the relationship between this binding and osmotic stress could be studied by measuring the fluorescence lifetime of AmB in liposomes. The composition of liposomes is easily varied, and the effect of osmotic stress could be examined by varying the ionic strength of the bathing solution after liposome formation. It would be necessary to use a dilute solution of liposomes to limit scattering. Additionally, UBC is likely to acquire a fluorescence lifetime imaging microscope in the near future. This new instrument may be useful in studying either AmB- membrane interaction or for further characterizing the model membrane systems themselves. For AmB, a distinct advantage of the FLIM is that it operates in single-photon counting mode. Unlike intensity-based measurements, this means that using a low quantum yield fluorophore is not a disadvantage. We note that FLIM is often implemented using an inverted microscope arrangement, meaning our electrochemical and Langmuir trough setups should be largely transferable to the new microscope. Lifetime-based measurements of AmB interaction with the floating monolayer or of the fluorescein-doped DOPC monolayer on Hg could be attempted. The relative insensitivity of lifetime to fluorophore concentration might provide a different view of the processes occurring on the Hg drop. If the lifetime of AmB is too short to be measurable, indirect approaches to lifetime-based measurements of AmB interaction could still be viable. The fluorophore NBD (nitrobenzoxadiazole) has measurable lifetime, and Invitrogen-Molecular Probes sells NBD-tagged cholesterol. AmB binding to NBD-tagged cholesterol might result in a measurable change in NBD lifetime that could be used to probe the interaction of different formulations with liposomes or 217 monolayer containing the tagged sterol. A further indirect approach worth attempting is lifetime- based measurement of the FRET process, to monitor changes in the lifetime of the FRET donor. A FRET acceptor represents an unusually efficient quench for the donor; the nearby presence of an acceptor shortens the fluorescence lifetime. Penetration of AmB into monolayers or liposomes could be followed by measuring the lifetime of DPH, a